Social and psychological determinants of the misuse of antibiotics and tranquilizers in the Galician and Lebanese populations

DOCTORAL THESIS SOCIAL AND PSYCHOLOGICAL DETERMINANTS OF THE MISUSE OF ANTIBIOTICS AND TRANQUILIZERS IN THE GALICIAN AND LEBANESE POPULATIONS Narmeen Mallah INTERNATIONAL DOCTORAL SCHOOL OF UNIVERSITY OF SANTIAGO DE COMPOSTELA DOCTORAL PROGRAM IN EPIDEMIOLOGY AND PUBLIC HEALTH SANTIAGO DE COMPOSTELA 2021 Author Contribution Article: Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Design, reliability and construct validity of a Knowledge, Attitude and Practice questionnaire on personal use of antibiotics in Spain. Sci Rep. 2020;10(1):20668. doi: 10.1038/s41598-020-77769-6 Narmeen Mallah performed the literature review, designed the questionnaire, collected, analyzed, and interpreted the data, and conceived and wrote the manuscript Article: Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Development and validation of a knowledge, attitude and practice questionnaire of personal use of tranquilizers. Drug Alcohol Depend. 2021;224:108730. doi: 10.1016/j.drugalcdep.2021.108730. Narmeen Mallah carried out the literature review, designed the questionnaire, collected, analyzed, and interpreted the data, and conceived and wrote the manuscript. Article: Mallah N, Badro DA, Figueiras A, Takkouche B. Association of knowledge and beliefs with the misuse of antibiotics in parents: A study in Beirut (Lebanon). PLoS One. 2020;15(7):e0232464. doi: 10.1371/journal.pone.0232464. Narmeen Mallah: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, writingoriginal draft, and writingreview and editing. Article: Mallah N, Figueiras A, Takkouche B. Comparison of Longitudinal and CrossSectional Approaches in Studies About Knowledge, Attitude, and Practices Related to Antibiotic Misuse. Drug Saf. 2021. doi: 10.1007/s40264-021-01075-x. Narmeen Mallah carried out the literature review, formulated the research idea, collected, analyzed, and interpreted the data, and conceived and wrote the manuscript Article: Mallah N, Badro DA, Figueiras A, Takkouche B. Association of knowledge and attitudes with the misuse of tranquilizers in parents: a study in Beirut (Lebanon). Psychol Health. 2021. doi: 10.1080/08870446.2021.1912341. Narmeen Mallah: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, writingoriginal draft, and writingreview and editing. Article: Mallah N, Figueiras A, Heidarian Miri H, Takkouche B. Association of knowledge and attitudes with practices of misuse of tranquilizers: A cohort study in Spain. Drug and Alcohol Dependence. 2021;225:108793. doi: 10.1016/j.drugalcdep.2021.108793 Narmeen Mallah conceived the research idea, designed the study, collected, analyzed, and interpreted the data, and designed and wrote the manuscript. Article: Mallah N, Tohidinik HR, Etminan M, Figueiras A, Takkouche B. Prenatal Exposure to Macrolides and Risk of Congenital Malformations: A Meta-Analysis. Drug Saf. 2020;43(3):211-221. doi: 10.1007/s40264-019-00884-5. Narmeen Mallah: Conceptualization of the manuscript, review and synthesis of the literature, and data extraction and analysis. Autorización de las revistas Las revistas donde se publicaron los artículos reproducidos en anexo no necesitan autorización específica para reproducirlos en tesis doctorales ya que los derechos pertenecen al autor. En particular: La revista Scientific Reports donde se ha publicado el articulo “Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Design, reliability and construct validity of a Knowledge, Attitude and Practice questionnaire on personal use of antibiotics in Spain. Sci Rep. 2020;10(1):20668. doi: 10.1038/s41598-020-77769-6” dice: The Author acknowledges first and reference publication in the journal. The author retains the following non-exclusive rights: To reproduce the contribution in whole or in part in any printed volume (book or thesis) of which they are the author(s). La revista Drug Safety donde se han publicado los artículos: “Mallah N, Figueiras A, Takkouche B. Comparison of Longitudinal and Cross-Sectional Approaches in Studies About Knowledge, Attitude, and Practices Related to Antibiotic Misuse. Drug Saf. 2021. doi: 10.1007/s40264-021-01075-x” y “Mallah N, Tohidinik HR, Etminan M, Figueiras A, Takkouche B. Prenatal Exposure to Macrolides and Risk of Congenital Malformations: A Meta-Analysis. Drug Saf. 2020;43(3):211-221. doi: 10.1007/s40264-019-00884-5” dice: Authors have the right to reuse their article’s Version of Record, in whole or in part, in their own thesis. La revista Drug and Alcohol Dependence donde se han publicado los artículos: “Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Development and validation of a knowledge, attitude and practice questionnaire of personal use of tranquilizers. Drug Alcohol Depend. 2021;224:108730. doi: 10.1016/j.drugalcdep.2021.108730” y “Mallah N, Figueiras A, Heidarian Miri H, Takkouche B. Association of knowledge and attitudes with practices of misuse of tranquilizers: A cohort study in Spain. Drug and Alcohol Dependence. 2021;225:108793. doi: 10.1016/j.drugalcdep.2021.108793” dice: Authors can include their articles in full or in part in a thesis or dissertation for non-commercial purposes. La revista Plos One donde se ha publicado el articulo “Mallah N, Badro DA, Figueiras A, Takkouche B. Association of knowledge and beliefs with the misuse of antibiotics in parents: A study in Beirut (Lebanon). PLoS One. 2020;15(7):e0232464. doi: 10.1371/journal.pone.0232464” dice: All PLOS content is at the highest possible level of Open Access, meaning that scientific articles are immediately and freely available to anyone, anywhere, to be downloaded, printed, distributed, read, reused and remixed (including commercially). ACKNOWLEDGMENTS My sincere gratitude goes to my thesis director and mentor, Professor Bahi Takkouche, for giving me the opportunity to carry out my doctoral thesis under his supervision and learn from his expertise, for being receptive to any research idea and a continuous support to accomplish the research aims, for his encouragement and proper guidance as well as for his meticulous suggestions, constructive criticism and advice throughout the whole doctoral period. Appreciation continues to my thesis co-director, Professor Adolfo Figueiras, for his support in the research undertaken in the present thesis and for his constructive criticism and advice. Thanks for Doctor Danielle A. Badro for her support in the studies carried out in Lebanon. My sincere thanks to the Department of Preventive Medicine at USC, for facilitating the environment to carry out this research. A special thanks to Professor Francisco Caamaño Isorna and Professor Agustin Montes Martinez for being always there to answer any inquiry related to the doctoral activities. DEDICATION هذه يدهأيتاناكمإ يف اوقثو ،اًمئاد يلجأ نم اوحض نيذلا ،ةمطافو دمحأ ،ّيدلاو ،ىلعلأا يلثمو يتودق ىلإ ةحورطلأا يف ينومعدو و ،يتايح نم ةلحرم لكينومّلع فعضلا نم ةوقلا عنصأ نأ. ف يل اهتياعر ىلع ،دبلأا ىلإ يلخاد يف شيعتس يتلا ،)الله اهمحر( ماستبإ يتلاخ ىلإ قيمعلا ينانتمإ برعأ امك .يرغص ي ّبح ىلع ،)الله اهمحر( تيناجو يلع ،يدادجلأو لصيف يلاخلو ،رونو نيفين ،يلع ،يئاقشلأ ينانتمإ نع.مهمعدو مه Dedico esta tesis a Erich, mi alma gemela y compañero en la vida, con mi profundo agradecimiento, por ser él y por su apoyo continuo durante los años de preparación de esta tesis. La dedico también a mi segunda familia, Don Agustín y Doña Nelly, Rocío y Rous por todo el cariño y el apoyo. Gracias a Inma, Antia y Teresa, por su amistad y cariño y por ser fuente de energía positiva en los momentos difíciles. “There are no limits of what you can accomplish, except the limits you place on your own thinking” Brian Tracy cada ano de educación asóciase cun 4% menos de probabilidades de calquera aspecto do mal uso de antibióticos nos países de nivel de riqueza media-baixa [OR: 0,96 (95% IC: 0,92-1,00)] e en Oriente Medio [OR: 0,96 (95% IC: 0,93, 1,00)]. Pola contra, asóciase cun 3% máis de probabilidades de almacenamento de antibióticos, un tipo específico de mal uso [OR: 1,03 (95% IC: 1,01-1,06)]. Metaanálise da asociación de macrólidos con malformacións conxénitas Métodos Realizáronse procuras en MEDLINE, EMBASE e outras bases de datos ata xuño de 2019. Avaliouse a calidade dos estudos e verificouse a heteroxeneidade e o rumbo de publicación. Realizamos tres análises diferentes e comparamos o efecto dos macrólidos con cada unha das seguintes poboacións non expostas: Grupo 1: fetos non expostos a ningún medicamento antes do nacemento, Grupo 2: fetos expostos a antibióticos non macrólidos / non teratógenos, e Grupo 3: poboación mixta do primeiro e dos segundo grupos de comparación. Resultados Observouse unha asociación débil entre os macrólidos e as malformacións conxénitas de calquera tipo cando se compararon os macrólidos coa poboación mixta [ORgrupo3: 1,06 (95% IC: 1,01-1,10)]. A análise de subgrupos mostra que esta asociación débil limítase á exposición do feto no primeiro trimestre do embarazo [OR: 1,06 (95% IC: 1,01-1,11)] e aos estudos de cohortes [OR: 1,07 (95% IC: 1,02-1,13). Atopouse que as malformacións do sistema dixestivo están levemente asociadas coa exposición prenatal a macrólidos [ORgrupo3: 1,14 (95% IC: 1,02, 1,26)]. Tamén se atopou que o sistema musculoesquelético está potencialmente afectado [ORgrupo3: 1,21 (95% IC: 1,08-1,35) e ORgrupo3: 1,15 (95% IC: 1,05-1,26)]. Os estudos europeos mostran unha asociación lixeiramente máis forte que os estudos estadounidenses nestas dúas comparacións. 6. Conclusions 1. Os instrumentos validados para medir a asociación de Coñecementos e Actitudes con Prácticas de mal uso de antibióticos e tranquilizantes xa están dispoñibles por primeira vez. 2. Os cuestionarios foron validados en tres idiomas: galego, árabe e francés, e probados en dous contextos socioeconómicos diferentes: Galicia e Líbano. Os cuestionarios foron moi aceptados pola poboación galega e libanesa. 3. O nivel baixo de Coñecemento e os conceptos erróneos sobre o papel dos antibióticos e os perigos dos tranquilizantes están asociados co Prácticas de mal uso destes medicamentos. 4. As Actitudes médicamente inadecuadas, incluída a tendencia para reducir ou estender a duración do tratamento doutro xeito que a prescrita, a tendencia de obter antibióticos e tranquilizantes sen receita e a aceptación de usar tranquilizantes para mellorar o soño, o traballo ou o estado de ánimo, asócianse con maiores probabilidades de Prácticas de mal uso de antibióticos e tranquilizantes. 5. As Actitudes negativas cara ao médico asócianse con maiores probabilidades de Prácticas de uso indebido de antibióticos e tranquilizantes. 6. A dispoñibilidade e accesibilidade aos medicamentos identificáronse como factores de risco do mal uso de medicamentos 7. O mal uso de antibióticos e tranquilizantes prodúcese en contornas sen un sistema sanitario centralizado, como no Líbano, así como en contornas con sistemas centralizados como Galicia. 8. Os enfoques transversais e lonxitudinais lanzan resultados distintos con respecto á asociación de Coñecementos e Actitudes con Prácticas de mal uso de antibióticos e tranquilizantes. Ningún está libre de rumbos e os resultados de ambos os enfoques deben analizarse coidadosamente para obter conclusións sólidas nos estudos CAP. 9. O metaanálise de dose-resposta revelou que o uso indebido de antibióticos ocorre por todas as persoas independentemente do seu nivel de ingresos e o seu nivel de educación. 10. Un metaanálise mostrou unha asociación marxinal entre a exposición prenatal a macrólidos e a malformación conxénita. Necesítase unha investigación de risco-beneficio para abordar a cuestión de se a prescrición de macrólidos debe restrinxirse durante o embarazo. Abstract The misuse of antibiotics and tranquilizers contributes to the development and exacerbation of antibiotic resistance, and dependence and addiction to tranquilizers, two international public health problems with heavy burdens. Antibiotic resistance has long been recognized as a global problem, yet fully validated tools to measure the psychosocial determinants, i.e., Knowledge Attitudes, and Practices of antibiotic misuse are unavailable. Furthermore, no longitudinal epidemiologic studies have been carried out so far. Dependence on tranquilizers is associated with alarming public health indicators, nonetheless related investigation has been limited to sociodemographic and psychiatric determinants. Drugs are deemed misused when taken without medical prescription or with medical advice but without adherence to the treatment regimen in terms of timing, dosage, and duration. The present doctoral project aimed at uncovering the psychosocial and socioeconomic determinants of antibiotic and tranquilizer misuse. The project was carried out in Galicia and Lebanon and involved three stages: 1) Development and validation of questionnaires on Knowledge, Attitudes and Practices (KAP) of antibiotic and tranquilizer misuse in Galician, Arabic and French; 2) Application of the validated questionnaires in Galicia and Lebanon in various epidemiological studies to determine the association of Knowledge and Attitudes with Practices of antibiotic and tranquilizer misuse. This stage also encompassed comparison of findings from cross-sectional and longitudinal approaches related to KAP studies, using data collected from the same individuals, in order to assess and discuss biases associated with each design; and 3) a: Dose-response meta-analyses to summarize inconsistent findings in the literature about the association of income level and education level with antibiotic misuse. b: Meta-analysis on prenatal exposure to macrolide antibiotics as a possible determinant of congenital malformation. For the first time, the present doctoral project provided validated KAP instruments in three languages to be used in studies of antibiotic and tranquilizer misuse. These instruments are the backbone for assessing, designing, and evaluating prevention programs to improve the rationale use of these drugs. This doctoral project also uncovered culture-specific misconceptions and Attitudes towards these drugs in two settings, Galicia and Lebanon, with different socioeconomic characteristics and public health systems. Inconsistencies about the association of income and education with antibiotic misuse were also resolved. Understanding the psychosocial and socioeconomic determinants of antibiotic and tranquilizer misuse is crucial to control the progression of antibiotic resistance and tranquilizer dependence. Keywords: Arabic; antibiotics; antibiotic resistance, cohort study; cross-sectional study; dependence; dose-response; French; Galicia; Galician; knowledge-attitude-practice; Lebanon; meta-analysis; misuse; questionnaire; tranquilizers; validation SCIENTIFIC PRODUCTION OF THE THESIS Thirteen articles have been derived from the present Doctoral Dissertation. Seven articles have already been accepted for publication: 6 in journals of the first quartile of the JCR Ranking and one in the second quartile, due to the downranking of the journal between submission and publication of the article. Three articles are currently under review by first quartile ranking journals. Three articles are being prepared for their submission in the next few days. 1. Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Development and validation of a knowledge, attitude and practice questionnaire of personal use of tranquilizers. Drug and Alcohol Dependence 2021:108730 (first JCR quartile) 2. Mallah N, Figueiras A, Heidarian Miri H, Takkouche B. Association of knowledge and attitudes with practices of misuse of tranquilizers: A cohort study in Spain. Drug and Alcohol Dependence 2021 (in print) (first JCR quartile) 3. Mallah N, Badro DA, Figueiras A, Takkouche B. Association of knowledge and attitudes with the misuse of tranquilizers in parents: A study in Beirut (Lebanon). Psychology and Health 2021 (in print) (first JCR quartile) 4. Mallah N, Figueiras A, Takkouche B. Comparison of longitudinal and cross-sectional approaches in studies about knowledge, attitude and practices related to antibiotic misuse. Drug Safety 2021 (in print) (first JCR quartile) 5. Mallah N, Rodríguez-Cano R, Figueiras A, Takkouche B. Design, reliability and construct validity of a knowledge, attitude and practice questionnaire on personal use of antibiotics in Spain. Scientific Reports 2020, 10(1):20668. (first JCR quartile) 6. Mallah N, Tohidinik T, Etminan M, Figueiras A, Takkouche B. Prenatal exposure to macrolides and risk of congenital malformations: A meta-analysis. Drug Safety 2020, 43(3), 211-221. (first JCR quartile) 7. Mallah N, Badro DA, Figueiras A, Takkouche B. Association of knowledge and beliefs with the misuse of antibiotics in parents: A study in Beirut (Lebanon). Plos One 2020, 15(7): e0232464. (second JCR quartile) 8. Mallah N, Orsini N, Figueiras A, Takkouche B. Income level and antibiotic misuse: A systematic review and dose-response meta-analysis. (Manuscript under review by European Journal of Health Economics, first JCR quartile) 9. Mallah N, Orsini N, Figueiras A, Takkouche B. Education level and misuse of antibiotics. A systematic review and dose-response meta-analysis. (Manuscript under review by Antimicrobial Resistance and Infection Control, first JCR quartile) 10. Mallah N, Battaljia J, Figueiras F, Takkouche B. Comparison of longitudinal and crosssectional approaches in studies about knowledge, attitude and practices related to tranquilizer misuse. (Manuscript under review by Journal of Clinical Medicine, first JCR quartile) 11. Mallah N, Rodríguez-Cano R, Badro DA, Figueiras A, Takkouche B. Development and validation of a knowledge, attitude and practice questionnaire on misuse of antibiotics in French language in Lebanon. (Manuscript under preparation for submission to International Journal of Antimicrobial Agents, first JCR quartile) 12. Mallah N, Rodríguez-Cano R, Badro DA, Figueiras A, Takkouche B. Development and validation of a knowledge, attitude and practice questionnaire on misuse of tranquilizers in French language in Lebanon. (Manuscript under preparation for submission to Journal of Clinical Medicine, first JCR quartile) 13. Mallah N, Figueiras A, Takkouche B. Epidemiology of antibiotic and tranquilizer misuse: a review of the evidence. (Manuscript under preparation for submission to Drug Safety, first JCR quartile) TABLE OF CONTENTS 1. INTRODUCTION .................................................................................................................................. 1 1.1. ANTIBIOTIC RESISTANCE: A MULTIFACETED CHALLENGE .................................................................. 2 Public health impact ...................................................................................................................................... 2 Clinical impact ............................................................................................................................................... 3 Socio-economic impact .................................................................................................................................. 3 1.1.2. DETERMINANTS OF ANTIBIOTIC RESISTANCE .................................................................................. 4 1.1.3. PREVALENCE OF ANTIBIOTIC MISUSE .............................................................................................. 4 1.1.4. DETERMINANTS OF ANTIBIOTIC MISUSE .......................................................................................... 5 Lack of Knowledge and awareness about antibiotic .................................................................................... 5 Attitudes of patients towards antibiotics and towards their healthcare provider......................................... 5 Sociodemographic and socioeconomic determinants of antibiotic misuse .................................................. 6 1.1.5. SOURCES OF ANTIBIOTICS WITHOUT PRESCRIPTION....................................................................... 6 1.1.6. ANTIBIOTIC RESISTANCE LIMITS TREATMENT OPTIONS IN CRITICAL SITUATIONS LIKE PREGNANCY ........................................................................................................................................................ 6 1.2. LEBANON AND SPAIN: EXAMPLES OF COUNTRIES WITH DIFFERENT SOCIOECONOMIC STATUS AND PUBLIC HEALTH SYSTEMS .................................................................................................................................. 7 1.2.1. Lebanon: Socioeconomic and sociodemographic indicators ..................................................... 7 1.2.2. Lebanon: Healthcare system ....................................................................................................... 7 1.2.3. Antibiotic misuse and resistance in Lebanon.............................................................................. 8 1.2.4. Spain: An example from the developed world ............................................................................. 9 1.2.5. Actions Taken in Lebanon and Spain to tackle antibiotic resistance......................................... 9 Surveillance system ....................................................................................................................................... 9 Awareness campaigns ................................................................................................................................... 9 Dispensing of antibiotics ............................................................................................................................... 9 1.3. TRANQUILIZER MISUSE, AN EXPONENTIALLY GROWING BUT INSUFFICIENTLY RECOGNIZED PUBLIC HEALTH ISSUE .................................................................................................................................................. 10 1.3.1. Prevalence of tranquilizer misuse ............................................................................................. 10 1.3.2. Public health impact of tranquilizer misuse.............................................................................. 11 1.3.3. Determinants of tranquilizer misuse ......................................................................................... 11 1.4. THE KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE MODEL ........................................... 11 2. HYPOTHESES ..................................................................................................................................... 13 3. OBJECTIVES ...................................................................................................................................... 15 3.3. MAIN OBJECTIVES .............................................................................................................................. 16 3.4. SPECIFIC OBJECTIVES ......................................................................................................................... 16 4. METHODS ........................................................................................................................................... 17 4.1. GLOBAL STUDY SETTING AND POPULATION ............................................................................................ 18 Lebanon ....................................................................................................................................................... 18 Spain ............................................................................................................................................................ 18 4.2. ETHICS ....................................................................................................................................................... 18 Lebanon ....................................................................................................................................................... 18 Spain ............................................................................................................................................................ 19 4.3. DEVELOPMENT AND VALIDATION OF SIX KNOWLEDGE, ATTITUDES AND PRACTICE (KAP) QUESTIONNAIRES ABOUT ANTIBIOTICS AND TRANQUILIZERS ........................................................................ 19 4.4. APPLICATION OF THE VALIDATED KAP QUESTIONNAIRES ..................................................................... 26 4.4.1. Association of Knowledge and Attitudes with antibiotic misuse Practices: two studies in Galicia and Lebanon ................................................................................................................................................ 26 4.4.2. Association of Knowledge and Attitudes with tranquilizer misuse Practices: two studies in Galicia and Lebanon ................................................................................................................................................ 28 4.4.3. DOSE-RESPONSE META-ANALYSES OF THE SOCIOECONOMIC DETERMINANTS OF ANTIBIOTIC MISUSE .......................................................................................................................................................................... 29 4.4.3.1. Dose-response meta-analysis of the association of income with antibiotic misuse Practices ...... 29 4.4.3.2. Dose-response meta-analysis of the association of education with antibiotic misuse Practices .. 32 4.4.3.3. META-ANALYSIS OF THE ASSOCIATION OF MACROLIDES WITH CONGENITAL MALFORMATION ..... 33 5. RESULTS ............................................................................................................................................. 36 5.3. DEVELOPMENT AND VALIDATION OF SIX KNOWLEDGE, ATTITUDE AND PRACTICE (KAP) QUESTIONNAIRES ABOUT ANTIBIOTICS AND TRANQUILIZERS ........................................................................ 37 5.3.1. Content validity .......................................................................................................................... 37 5.3.2. Face validity and pilot testing .................................................................................................... 37 5.3.3. Test–retest reliability .................................................................................................................. 38 5.3.4. Construct validity of KAP questionnaire about antibiotic use in Galicia ................................ 38 5.3.5. Construct validity of KAP questionnaire about antibiotics in French in Lebanon ................. 39 5.3.6. Construct validity of KAP questionnaire about tranquilizers in Galicia.................................. 40 5.3.7. Construct validity of KAP questionnaire about tranquilizers in French in Lebanon ............. 41 5.3.8. Questionnaire and item acceptability ........................................................................................ 41 5.3.9. Questionnaire overall reliability ................................................................................................ 42 5.4. ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF ANTIBIOTIC MISUSE .............. 57 5.4.1. Cross-sectional study in Lebanon .............................................................................................. 57 5.4.2. Comparision study of cross-sectional and longitudinal approaches in studies about Knowledge, Attitude and antibiotic misuse Practices in Galicia ................................................................ 62 5.5. ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF TRANQUILIZER MISUSE ........ 72 5.5.1. Cross-sectional study in Lebanon .............................................................................................. 72 5.5.2. Cohort study in Spain ................................................................................................................ 77 5.6. DOSE-RESPONSE META-ANALYSIS OF THE ASSOCIATION OF INCOME WITH ANTIBIOTIC MISUSE PRACTICES ........................................................................................................................................................ 89 5.7. DOSE-RESPONSE META-ANALYSIS OF THE ASSOCIATION OF EDUCATION WITH ANTIBIOTIC MISUSE PRACTICES ...................................................................................................................................................... 101 5.8. META-ANALYSIS OF THE ASSOCIATION OF MACROLIDES WITH CONGENITAL MALFORMATION ... 119 6. DISCUSSION ..................................................................................................................................... 131 6.1. DEVELOPMENT AND VALIDATION OF KAP QUESTIONNAIRES ABOUT ANTIBIOTIC AND TRANQUILIZER USE ........................................................................................................................................ 132 6.2. APPLICATION OF THE DEVELOPED QUESTIONNAIRES IN COUNTRIES WITH DIFFERENT SOCIOECONOMIC AND PUBLIC HEALTH INDICATORS: GALICIA AND LEBANON .......................................... 135 6.3. META-ANALYSES OF SOCIOECONOMIC DETERMINANTS OF ANTIBIOTIC MISUSE PRACTICES ........ 138 6.4. MACROLIDES, AN ALTERNATIVE TREATMENT TO PENICILLIN RESISTANT INFECTIONS. A METAANALYSIS OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATION ............................. 141 7. CONCLUSIONS ................................................................................................................................. 143 8. REFERENCES ................................................................................................................................... 145 9. ANNEXES .......................................................................................................................................... 167 ANNEX I. VARIOUS LANGUAGES OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRES ABOUT ANTIBIOTIC AND TRANQUILIZERS .................................................. 168 ENGLISH VERSION OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE ABOUT ANTIBIOTIC USE . 168 ENGLISH VERSION OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE ABOUT TRANQUILIZER USE ........................................................................................................................................................................ 171 GALICIAN VERSION OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRES ABOUT ANTIBIOTIC AND TRANQUILIZER USE IN GALICIA .................................................................................................................... 174 ARABIC VERSION OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRES ABOUT ANTIBIOTIC AND TRANQUILIZER USE IN LEBANON ................................................................................................................... 178 FRENCH VERSION OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRES ABOUT ANTIBIOTIC AND TRANQUILIZER USE IN LEBANON ................................................................................................................... 184 ANNEX II. PUBLICATIONS DERIVED FROM THE THESIS .............................................................. 190 MALLAH N, RODRÍGUEZ-CANO R, FIGUEIRAS A, TAKKOUCHE B. DESIGN, RELIABILITY AND CONSTRUCT VALIDITY OF A KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE ON PERSONAL USE OF ANTIBIOTICS IN SPAIN. SCIENTIFIC REPORTS 2020, 10(1):20668. (JOURNAL RANKING: FIRST QUARTILE) ............................... 190 MALLAH N, RODRÍGUEZ-CANO R, FIGUEIRAS A, TAKKOUCHE B. DEVELOPMENT AND VALIDATION OF A KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE OF PERSONAL USE OF TRANQUILIZERS. DRUG AND ALCOHOL DEPENDENCE 2021: 108730 (JOURNAL RANKING: FIRST QUARTILE) ............................................. 200 MALLAH N, BADRO DA, FIGUEIRAS A, TAKKOUCHE B. ASSOCIATION OF KNOWLEDGE AND BELIEFS WITH THE MISUSE OF ANTIBIOTICS IN PARENTS: A STUDY IN BEIRUT (LEBANON). PLOS ONE 2020, 15(7): E0232464. (JOURNAL RANKING: SECOND QUARTILE) ...................................................................................................... 209 MALLAH N, FIGUEIRAS A, TAKKOUCHE B. COMPARISON OF LONGITUDINAL AND CROSS-SECTIONAL APPROACHES IN STUDIES ABOUT KNOWLEDGE, ATTITUDE AND PRACTICES RELATED TO ANTIBIOTIC MISUSE. DRUG SAFETY 2021 (IN PRINT) (JOURNAL RANKING: FIRST QUARTILE) ......................................................... 221 MALLAH N, FIGUEIRAS A, HEIDARIAN MIRI H, TAKKOUCHE B. ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF MISUSE OF TRANQUILIZERS: A COHORT STUDY IN SPAIN. DRUG AND ALCOHOL DEPENDENCE 2021 (IN PRINT) (JOURNAL RANKING: FIRST QUARTILE) .......................................................... 234 MALLAH N, TOHIDINIK T, ETMINAN M, FIGUEIRAS A, TAKKOUCHE B. PRENATAL EXPOSURE TO MACROLIDES AND RISK OF CONGENITAL MALFORMATIONS: A META-ANALYSIS. DRUG SAFETY 2020, 43(3), 211-221. (JOURNAL RANKING: FIRST QUARTILE) .......................................................................................................... 235 ANNEX III. INTERNATIONAL MENTION OF PHD DEGREE ............................................................ 246 AUTHORIZATION OF THE ACADEMIC COMMISSION OF DOCTORAL PROGRAM IN EPIDEMIOLOGY AND PUBLIC HEALTH FOR A THREE-MONTH RESEARCH STAY AT KAROLINSKA INSTITUTET, SWEDEN ............ 246 REPORT ABOUT THREE-MONTH RESEARCH STAY FROM THE SUPERVISOR AT KAROLINSKA INSTITUTET, SWEDEN .......................................................................................................................................................... 247 LIST OF TABLES TABLE 1 INTRACLASS CORRELATION COEFFICIENTS (ICCS) AND 95% CONFIDENCE INTERVALS (CIS) OF THE KNOWLEDGE AND ATTITUDE ITEMS OF THE GALICIAN, ARABIC AND FRENCH VERSIONS OF THE QUESTIONNAIRE ON ANTIBIOTIC USE .......................................................................................................... 43 TABLE 2 INTRACLASS CORRELATION COEFFICIENTS (ICCS) AND 95% CONFIDENCE INTERVALS (CIS) OF KNOWLEDGE AND ATTITUDE ITEMS OF THE GALICIAN, ARABIC AND FRENCH VERSIONS OF THE QUESTIONNAIRE ON TRANQUILIZER USE .................................................................................................... 44 TABLE 3 COMPARISON OF THE GOODNESS OF FIT PARAMETERS BETWEEN MODELS OF KNOWLEDGE, ATTITUDES AND PRACTICE QUESTIONNAIRE ON ANTIBIOTIC USE IN THE GALICIAN ADULT POPULATION ...................................................................................................................................................................... 45 TABLE 4 FACTOR LOADINGS AND STANDARD ERRORS FROM THE THREE-FACTOR MODEL (MODEL 3) OF THE KNOWLEDGE, ATTITUDES AND PRACTICE QUESTIONNAIRE ON ANTIBIOTIC USE IN GALICIA ................. 46 TABLE 5 COMPARISON OF THE GOODNESS OF FIT PARAMETERS BETWEEN MODELS OF KNOWLEDGE, ATTITUDES AND PRACTICE QUESTIONNAIRE ON ANTIBIOTIC USE IN FRENCH LANGUAGE TESTED IN THE LEBANESE ADULT POPULATION .................................................................................................................. 48 TABLE 6 FACTOR LOADINGS AND STANDARD ERRORS FROM THE THREE-FACTORS MODEL (MODEL 5) OF THE KNOWLEDGE, ATTITUDES AND PRACTICE QUESTIONNAIRE IN FRENCH ON ANTIBIOTIC USE IN LEBANON ...................................................................................................................................................................... 49 TABLE 7 COMPARISON OF THE GOODNESS OF FIT PARAMETERS BETWEEN MODELS OF KAP QUESTIONNAIRE ON TRANQUILIZER USE IN GALICIA ............................................................................................................. 51 TABLE 8 STANDARD LOADING ESTIMATES OF THE KNOWLEDGE AND ATTITUDE ITEMS OF THE TRANQUILIZER QUESTIONNAIRE IN GALICIAN ON THEIR RESPECTIVE FACTORS (MODEL 4) ............................................ 52 TABLE 9 COMPARISON OF THE GOODNESS OF FIT PARAMETERS BETWEEN MODELS OF KNOWLEDGE, ATTITUDE AND PRACTICE QUESTIONNAIRE ON TRANQUILIZER USE IN FRENCH LANGUAGE TESTED IN THE LEBANESE ADULT POPULATION ........................................................................................................... 54 TABLE 10 STANDARD LOADING ESTIMATES OF THE KNOWLEDGE AND ATTITUDE ITEMS OF THE TRANQUILIZER QUESTIONNAIRE (MODEL 3) IN FRENCH ON THEIR RESPECTIVE FACTORS ..................... 55 TABLE 11 DEMOGRAPHIC CHARACTERISTICS OF INDIVIDUALS PARTICIPATED IN THE CROSS-SECTIONAL STUDY ABOUT THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF ANTIBIOTIC MISUSE IN LEBANON .................................................................................................................................... 59 TABLE 12 ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH THE OUTCOME “ANY MISUSE” PRACTICE OF ANTIBIOTICS IN 1392 LEBANESE PARENTS OF SCHOOLCHILDREN ............................................................. 60 TABLE 13 ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH SPECIFIC ASPECTS OF ANTIBIOTIC MISUSE PRACTICES IN LEBANESE PARENTS OF SCHOOLCHILDREN ........................................................................ 61 TABLE 14 GENERAL CHARACTERISTIC OF INDIVIDUALS WHO PARTICIPATED IN THE STUDY ABOUT THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF ANTIBIOTIC MISUSE IN GALICIA 64 TABLE 15 ASSOCIATIONS OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF ANTIBIOTIC MISUSE IN GALICIA ASSESSED BY COHORT AND CROSS-SECTIONAL APPROACHES ..................................................... 65 TABLE 16 IRRS AND 95%CI OF MISUSE PRACTICES OF ANTIBIOTICS IN GALICIA AFTER IMPUTATION OF MISSING DATA (MICE PROCEDURE) ........................................................................................................... 68 TABLE 17 IRRS AND 95%CI FROM THE TWO SENSITIVITY ANALYSES OF THE LONGITUDINAL DATA COLLECTED IN GALICIA ABOUT MISUSE PRACTICES OF ANTIBIOTICS ...................................................... 70 TABLE 18 DEMOGRAPHIC CHARACTERISTICS OF INDIVIDUALS PARTICIPATED IN THE CROSS-SECTIONAL STUDY ABOUT THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF TRANQUILIZER MISUSE IN LEBANON .................................................................................................................................... 74 TABLE 19 ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH ANY PATTERN OF MISUSE PRACTICES OF TRANQUILIZERS BY PARENTS OF SCHOOLCHILDREN IN LEBANON ............................................................ 75 TABLE 20 ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH SPECIFIC PATTERNS OF MISUSE PRACTICES OF TRANQUILIZERS BY PARENTS OF SCHOOLCHILDREN IN LEBANON ............................................................ 76 TABLE 21 GENERAL CHARACTERISTIC OF INDIVIDUALS WHO PARTICIPATED IN THE STUDY ABOUT THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES WITH PRACTICES OF TRANQUILIZER MISUSE IN GALICIA ...................................................................................................................................................................... 79 TABLE 22 INCIDENCE RATE RATIO (IRR) OF TRANQUILIZER MISUSE PRACTICES ACCORDING TO QUANTILES OF LEVELS OF AGREEMENT OF THE GALICIAN POPULATION ON KNOWLEDGE AND ATTITUDES STATEMENTS ................................................................................................................................................ 80 TABLE 23 ESTIMATES OF THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES OF THE GALICIAN POPULATION WITH ANY MISUSE PRACTICE OF TRANQUILIZERS USING CROSS-SECTIONAL AND LONGITUDINAL APPROACHES ................................................................................................................................................ 85 TABLE 24 ESTIMATES OF THE ASSOCIATION OF KNOWLEDGE AND ATTITUDES OF THE GALICIAN POPULATION WITH SPECIFIC ASPECTS OF MISUSE PRACTICES OF TRANQUILIZERS USING CROSS-SECTIONAL AND LONGITUDINAL APPROACHES ...................................................................................................................... 87 TABLE 25 GENERAL CHARACTERISTICS OF STUDIES INCLUDED THE META-ANALYSIS OF INCOME LEVEL AND ANTIBIOTIC MISUSE ..................................................................................................................................... 92 TABLE 26 META-ANALYSIS OF THE ASSOCIATION OF INCOME LEVEL REPRESENTED AS UNITS OF GROWTH DOMESTIC PRODUCT (GDP) PER CAPITA BASED ON PURCHASING POWER PARITY (PPP) WITH ANTIBIOTIC MISUSE ....................................................................................................................................................... 100 TABLE 27 GENERAL CHARACTERISTICS OF STUDIES INCLUDED IN THE DOSE-RESPONSE META-ANALYSIS ABOUT EDUCATION LEVEL AND MISUSE OF ANTIBIOTICS ......................................................................... 106 TABLE 28 SUMMARY ODDS RATIOS (OR) AND THEIR 95% CONFIDENCE INTERVAL (CI) ESTIMATED BY CATEGORICAL AND CONTINUOUS APPROACHES OF DOSE-RESPONSE META-ANALYSIS ON EDUCATION AND ANTIBIOTIC MISUSE ................................................................................................................................... 116 TABLE 29 CHARACTERISTICS OF COHORT STUDIES OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATIONS ...................................................................................................................................... 122 TABLE 30 CHARACTERISTICS OF CASE-CONTROL STUDIES OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATIONS ................................................................................................................ 125 TABLE 31 POOLED ODDS RATIOS (ORS) AND 95% CONFIDENCE INTERVALS (CIS) OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATIONS (COMPARISON GROUP 1: FETUSES NOT EXPOSED TO ANY DRUG) ............................................................................................................................ 127 TABLE 32 POOLED ODDS RATIOS (ORS) AND 95% CONFIDENCE INTERVALS (CIS) OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATIONS. (COMPARISON GROUP 2: FETUSES EXPOSED TO NON-TERATOGENIC DRUGS/NON-MACROLIDES).................................................................................. 128 TABLE 33 POOLED ODDS RATIOS (ORS) AND 95% CONFIDENCE INTERVALS (CIS) OF MACROLIDES PRENATAL EXPOSURE AND CONGENITAL MALFORMATIONS. (COMPARISON GROUP 3: MIXED POPULATION OF UNEXPOSED FETUSES) ..................................................................................................... 129 Narmeen Mallah Introduction 6 The trust between patients and their healthcare providers was also identified as a factor to improve adherence to antibiotic therapy [44]. Sociodemographic and socioeconomic determinants of antibiotic misuse The misuse of antibiotics was associated with age (young adults and elderly > 65 years), female gender, unemployment, difficult access to health care, unaffordable cost of clinical visits, long waiting periods in clinics and lack of time or transportation problems [2, 26-28, 35, 45]. Literature review regarding the association of education and income with antibiotic misuse showed controversial findings. Some studies reported up to six fold increased odds of antibiotic misuse in high income individuals [46, 47], while other studies did not find any association between income and antibiotic misuse [48-50], or detected lower odds of antibiotic misuse in high income than in low income individuals [51, 52]. Likewise, several studies reported a negative association between low education and antibiotic misuse [53-55]. Inversely, other studies reported that high education level is associated with a greater risk of antibiotic misuse [56-60]. Some studies found no association between education and antibiotic misuse [61-63]. Furthermore, it is unclear whether the association of income and education with antibiotic misuse maintains in settings with various social classes, different levels of access to healthcare and diverse regulations of antibiotic dispensing. 1.1.5. Sources of antibiotics without prescription Friends, families and use of leftover antibiotics were identified as common sources of unprescribed antibiotics in the low-, middleand high-wealth countries [2, 26, 36, 45, 64]. In developing countries, where many of the available antibiotics are of poor quality and regulations related to the dispensing of these drugs are lacking or insufficiently enforced, pharmacies represent the major source of unprescribed antibiotics [2, 65, 66]. Even in those countries where the dispensing of antibiotics is strictly regulated, such as in the European Union, access to antibiotics without a prescription still takes place and differences between countries are observed when looking at the national picture. For instance, where overall 7% of the Europeans declared using antibiotics obtained without a prescription from a pharmacy or leftover from a previous course of antibiotics, 14% of Latvian and Bulgarian people obtained antibiotics without a prescription in comparison to only 1% of Dutch citizens [27]. Citizens of countries with antibiotic regulations may obtain antibiotics without a prescription from pharmacies, buy them from e-pharmacies via the internet or from other countries during a holiday, use antibiotic leftover or share these drugs with someone else [64]. 1.1.6. Antibiotic resistance limits treatment options in critical situations like pregnancy Appropriate medication during pregnancy is crucial to the safety of the foetus and the pregnant woman. It is estimated that 8% of pregnant women experience urinary tract infections which Narmeen Mallah Introduction 7 accounts for 10% of clinical visits during pregnancy [67]. As a result, antibiotics are prescribed to prevent the negative consequences of those infections such as pyelonephritis and preterm labour [67]. Women from deprived areas are more likely to receive an antibiotic prescription than those women from wealthier regions [68]. Due to antibiotic resistance, penicillin and other cost-effective antibiotics are no longer recommended to treat asymptomatic bacteriuria [67]. Macrolides are wide-spectrum antibiotics that are prescribed with high frequency during pregnancy [68]. Some studies showed that the use of macrolides (excluding erythromycin) during early pregnancy increases the risk of spontaneous abortion between 1.65-fold (for azithromycin) and 2.72-fold (for quinolones) [69]. Nonetheless, the literature does not provide clear and consistent evidence about the association of macrolide prenatal exposure with birth defects [70-74]. Birth defects are principal contributors to infants’ morbidity and mortality [75, 76]. Though it is a global problem, lowand middle-income countries are the most affected ones as more than 94% of births with serious congenital malformations and 95% of related deaths take place in these countries [75]. Congenital malformations of non-genetic origin can be prevented through adequate antenatal care and medication [76]. 1.2. Lebanon and Spain: Examples of countries with different socioeconomic status and public health systems 1.2.1. Lebanon: Socioeconomic and sociodemographic indicators The Lebanese Republic (commonly known as Lebanon) is a Middle Eastern country of 10,452 km2 [77]. The population size in Lebanon is 6.85 million individuals, 1.79 million (26.14%) of whom are 14 years or younger, and 1.21 million (17.71%) are young adults (15-24 years) [78]. The literacy rate in Lebanon is 95.1% for individuals 15 years or older [79]. According to 2018 data, the life expectancy of the Lebanese at birth is 79 years and the mortality rate of children under five is 7% [78]. One-tenth of the Lebanese population lives in rural areas [78]. Lebanon is classified as an upper-middle income country by the World Bank [77], however, the country frequently experiences fragilities due to its vulnerability to political tensions and external financing instability. Wars in neighbouring countries and high immigration rates to Lebanon also affected its economy and health indicators. Since 2017, the economic situation in the country is on a downward spiral, where the Lebanese currency (LBP) to American dollars (USD) exchange rate has drastically decreased and left a huge consequence on the purchasing power. 1.2.2. Lebanon: Healthcare system The healthcare system in Lebanon relies on private and public healthcare providers, with a predominance of private healthcare [80, 81]. It is considered among the most expensive systems in the world [81], with healthcare costs that cannot be sustained by many households [82]. Three-quarters of private healthcare expenditures are out-of-pocket [82]. The available hospital beds are concentrated in the capital (Beirut) and its suburbs at a ratio of 70 beds per 10,000 individuals, in comparison to 20 beds per 10,000 in poorer and rural regions [82]. Narmeen Mallah Introduction 8 The Lebanese healthcare system encompasses the following schemes [80]: National Social Security fund (NSSF) under the tutelage of the Ministry of Labour: it covers the employees in the private sector and their dependents, and the population employed in the public sector but who do not benefit from the Civil Service Cooperative (CSC) scheme. The population covered by NSSF is relatively young, as individuals are expulsed upon retirement. Only half of the Lebanese are supported by NSSF, and this healthcare assistance scheme covers 90% of hospital care through direct payment to the hospitals and 85% of ambulatory care via reimbursement to the patient. NSSF is funded by contributions from employers, employees and the government. Civil Servants Cooperative (CSC) under the tutelage of the presidency of the Council of Ministers: it provides healthcare assistance for the regular government staff and their dependents. CSC covers hospital care through direct payment to the hospitals (90% for the employee and 75% for the dependents) and ambulatory care via reimbursement to the patient (75% for the employee and 50% for the dependents). It is funded by the government. Funds for the military and security forces under the tutelage of the Ministry of Defense: it provides healthcare assistance for uniformed staff members and their dependents. It covers hospital and ambulatory care through direct payment (100% for the member, 75% for the spouse and children and 50% for dependent parents). The funds for the military and security forces are supported by the government. Ministry of Public Health (MoPH): it covers Lebanese individuals who neither have private insurance nor benefit from the above-mentioned schemes, upon request. MoPH covers hospital care through direct payment, provides expensive drugs for catastrophic illnesses and vaccines and essential drugs. It is funded by the government. Private insurance under the tutelage of the Ministry of Economy and Trade: individuals are voluntary enrolled under this scheme. The covered expenses vary according to the health insurance company. It is supported by households and employers and employees for complementary insurance. Mutual fund under the tutelage of the Ministry of Agriculture: individuals are voluntary enrolled under this scheme. The covered expenses vary according to the contracted company. It is supported by households and government subsidies. 1.2.3. Antibiotic misuse and resistance in Lebanon Previous studies in Lebanon reported that antibiotic misuse is a frequent Practice in the country [83, 84], most of the surveyed individuals thought that antibiotics should be taken for the common cold [85], and did not know that antibiotics are not anti-viral medicines [83]. Use of antibiotics without prescription in Lebanon was associated with lower educational level [85], lack of time, gender and sore throat symptoms [84]. Knowledge and Attitudes of the Lebanese population towards antibiotics is influenced by income, education, place of residence, and access to healthcare [83]. Nonetheless, none of these studies examined the association of Knowledge and Attitudes with Practices antibiotic misuse through effect measures. Besides, all previous studies undertaken in the country relied on a limited sample recruited by convenience. Narmeen Mallah Introduction 9 1.2.4. Spain: An example from the developed world Spain has a national health system that provides free access to all Spanish citizens. The health system in Spain is regionalized, i.e., each of the 17 autonomous communities is responsible to provide equal access to healthcare to all the population. In Spain, public healthcare predominates over private healthcare services. The literacy rate in Spain is 98.1% [79]. In Galicia, the autonomous community where an important part of the present doctoral project took place, the public healthcare system is centralized, and medical prescriptions are provided electronically. The National Social Security supports hospital and ambulatory care. 1.2.5. Actions Taken in Lebanon and Spain to tackle antibiotic resistance Surveillance system Lebanon designed a national surveillance system for antibiotic use that includes monitoring the sales or the consumption of antibiotics in health services, but that system has not been implemented yet [86]. Spain regularly collects data (annually or biannually) and reports on antibiotic sales or consumption at the national level for human use as well as on the appropriateness of antibiotic prescriptions in public and private health facilities [86]. Awareness campaigns Lebanon carried out a limited or small-scale antibiotic resistance awareness campaign that target some relevant stakeholders [86]. Spain carries out government-supported activities at the national scale which aim at improving the behaviour towards antibiotics in target groups in human health from public and private sectors. It evaluates the change in awareness and behaviour every five years. [86]. Dispensing of antibiotics In Lebanon, dispensing antibiotics without a prescription is an unlawful action (Decision number 1/2636) [87], however, the related regulations are not enforced. A recent study reported that 31% of the surveyed pharmacies in Lebanon distribute antibiotics without a prescription and only 6% of the pharmacists referred their patients to a doctor [88]. In Spain, guidelines on optimizing antibiotic use are implemented for all major syndromes and data on the use of these drugs are systematically fed back to prescribers. Antibiotic dispensing without a medical prescription is prohibited by law in Spain (Real Decreto 1718/2010) [89], nonetheless, a study in 2017 showed that antibiotics are still supplied without a prescription in one-fifth of the surveyed pharmacies, especially in rural settings and under patient’s pressure [90]. Narmeen Mallah Introduction 10 1.3. Tranquilizer misuse, an exponentially growing but insufficiently recognized public health issue Besides antibiotics, the other class of drugs which are prone to worldwide misuse is represented by tranquilizers. The term “tranquilizers” is a generic label for drugs that are used to reduce anxiety, pain, tension and agitation, relax muscles, calm or sedate, and help get a better sleep. Tranquilizer use is safe and well-tolerated, but it may negatively affect the central nervous system and lead to dependence, especially when these drugs are misused or consumed in high doses [91]. Almost half of the young adults who had already been prescribed tranquilizers but misused them experienced a non-fatal overdose [92]. In the United States, in 2019, one-quarter of opioid overdose deaths also involved benzodiazepines [93]. Tranquilizers, especially benzodiazepines, rank among the topmost misused psychoactive drugs, with a higher prevalence among women than men [94]. Tranquilizers are the second most seized drugs after opioids, with the difference that most tranquilizers are derived from legal sources [94]. 1.3.1. Prevalence of tranquilizer misuse The simultaneous use of tranquilizers with other illicit drugs is an internationally salient issue that is associated with heavy public health, sociologic and economic burdens. In the United States, 4.6% of individuals 12 years of age or older who used illicit drugs in 2010 had started with misusing tranquilizers, a percentage that is higher than that of cocaine, heroin and hallucinogens combined [95]. In 2015, 1.4 million Americans of the same age were recent initiates of tranquilizer misuse [96]. In the same year, 15% of the Americans who reported pastyear use of tranquilizers misused these drugs to relax and relieve tension (45%) and get help with sleep (20%) [96]. In 2017, the prevalence of tranquilizer misuse among Americans of the same age was 2.2%, and a higher prevalence was estimated among people of 18 to 25 years of age [94]. According to the 2019 World Drug Report, tranquilizer misuse in Europe is more frequent than that of cannabis with a prevalence ranging from as low as 1% in Portugal to around 20% in Czechia [94]. In 2014, the overall prevalence of misuse of sedatives, including tranquilizers, among Europeans from Denmark, Germany, Great Britain, Spain and Sweden aged 12 to 49 years was 6% for past-year and 11% for lifetime misuse [97]. This reveals an upward trend in tranquilizer misuse in Europe as a study involving 31 European countries had reported that the lifetime misuse of tranquilizers was practiced by 5.6% of European adolescents in 2003 [98]. Spain ranked the first among the European countries in misusing tranquilizers with a prevalence of 9% for past-year and 18% for lifetime misuse [97]. In general, the consumption of tranquilizers has increased in Spain in the last decade [99]. The prevalence of tranquilizer misuse among Spanish adolescents was 2.94% in 2014, with a higher prevalence among girls than boys [100]. The concurrent use of cannabis and tranquilizers was also reported among young Spanish adults (15-34 years) [101]. Tranquilizer misuse is also reported in countries of South America, Africa, North Africa and Asia, though survey data are limited [94, 102-106]. Narmeen Mallah Introduction 11 In Lebanon, tranquilizers are the most misused medicinal drugs among high school and university students according to data collected between 1999 and 2001 [107]. About 13% of the university students and 3% of school students reported ever misusing tranquilizers and declared that access to these drugs was easy [107]. To the best of our knowledge, no other studies were conducted in the country. The reported sources of the misused tranquilizers are doctor prescription and social sources, i.e., sharing with family and friends who had been prescribed these drugs. 1.3.2. Public health impact of tranquilizer misuse Tranquilizer misuse is associated with heavy consequences on public health, though the available information is limited to North America [108-110]. In the United States, deaths from drug overdose including tranquilizers have increased by ten folds during the last decade, and emergency room visits attributed to benzodiazepine overdose have doubled [108, 109]. 1.3.3. Determinants of tranquilizer misuse The determinants of tranquilizer misuse are understudied. Only a few studies have examined the association of sociodemographic and psychiatric attributes with tranquilizer misuse [100102, 111-113]. Identified factors included age (adolescents and young adults), low educational level, unemployment, alcohol intake, tobacco smoking, use of illicit drugs, receipt of a tranquilizer prescription, dissatisfaction with relationships, presence of chronic diseases, lack of perception of health risks related to these drugs, and psychiatric disorders [98, 100-102, 111114]. The odds of tranquilizer and sedative misuse is higher among adolescents who: received a prescription for tranquilizers (7 folds for girls, and 10 folds for boys), smoke tobacco (1.3 folds), drink alcohol (1.3 folds), consume illicit drugs (3.5 folds), skip classes (1.4 folds), have friends using tranquilizers (6 folds) and have siblings using tranquilizers (2.7 folds) [98]. The association of the Knowledge of the individuals and their Attitudes towards tranquilizer misuse Practices has not been investigated earlier. 1.4. The Knowledge, Attitude and Practice Questionnaire Model Knowledge, Attitude and Practice (KAP) questionnaires are largely applied instruments in the assessment of psychosocial properties related to a wide variety of health-related issues such as pain [115], vaccination [116], diet [117], and misuse of medicines such as antibiotics [118]. KAP questionnaires consist of a series of declarative statements (items) that are designed to collect information about what the population knows (i.e., Knowledge) and what the population thinks about (i.e., Attitudes) a specific topic [119]. KAP model questionnaires also entail a set of questions about the respondents’ performance toward the studied topic (i.e., Practice) [119]. Examples of Practice questions include whether the patient adheres to the treatment regimen. Conclusions obtained from KAP-based studies can help determine the need for educational intervention programs, design the interventions and assess their efficiency [120]. Narmeen Mallah Introduction 12 Until the present work, validated questionnaires to measure the use of antibiotics and/or tranquilizers were not available. Many KAP questionnaires on antibiotic use by the general adult population are available in the literature. Nevertheless, none of these questionnaires have undergone a full validation process. A recent review report about Knowledge and/or Attitudes towards antibiotic use by the general population concluded that 11 studies had tested the validity and the reliability of the used questionnaire [118], but the undertaken validity steps in those studies were limited to face and/or content validity, i.e., other essential validation steps such as construct validity and questionnaire acceptability were not examined. It is worth mentioning that Alumran and colleagues validated the construct of a questionnaire that intended to measure the perceptions of the parents on the use of antibiotics for their children with upper respiratory tract infections [121], however, this instrument cannot be applied to assess Knowledge and Attitudes of adults concerning their own use of antibiotics because: 1) the questions in that questionnaire are oriented toward children such as “Children with common colds get better faster when antibiotics are given” and “when I visit the doctor for my child’s common cold, I expect prescription for medication including antibiotics”, 2) parents are usually more judicious with antibiotic use for children than for themselves [122]. Concerning studies about the association of Knowledge and Attitudes of the general population with Practices of tranquilizer use, to the best of our knowledge, no studies were available in the literature before this thesis, with the exception of two studies from the past century that measured limited attitudes towards tranquilizers [123, 124] 13 2. HYPOTHESES Narmeen Mallah Hypotheses 14 Hypotheses a) Knowledge about antibiotics and tranquilizers is associated with their misuse Practices. b) Personal Attitudes towards the use of antibiotics and tranquilizers are associated with their misuse Practices. c) Attitudes towards health-care providers are associated with the misuse Practices of antibiotics and tranquilizers. d) The socioeconomic level of the individuals is associated with antibiotic misuse Practices. e) The use of antibiotics such as macrolides may be associated with negative health consequences. 15 3. OBJECTIVES Narmeen Mallah Methods 22 of the questions, the ease of answering as well as the time taken to completely answer the questionnaire. 4.3.8. Test-retest reliability The reliability testing of the questionnaire aims at measuring the stability of the instrument over time [127, 130]. In other words, it examines the ability of the questionnaire to generate reproducible results. The stability or reproducibility of a questionnaire is explored using testretest procedure. We tested the reproducibility of each of the six questionnaires in their corresponding populations, i.e., Galician in Galicia and Arabic and French in Lebanon. We carried out the test-retest for Knowledge and Attitude questions as these parameters, unlike Practice, are known to be stable over time [130]. The questionnaires were administered on two-time occasions to the same participants within a four-week time interval. Participants were informed about the aim of the study and accepted to answer the questionnaire on two occasions. Intraclass Correlation Coefficients (ICC) reflected the test-retest reliability index of each item of the questionnaire. Using data collected from the two test rounds, ICCs were calculated with their 95% Confidence Intervals (CIs) relative to the average measure of the two-way mixedeffects model [131]. Items with ICC values ≥ 0.4 were deemed reliable [132]. 4.3.9. Confirmatory Factorial Analysis Construct validity is an evaluation of the extent to which a questionnaire measures a target construct. In a questionnaire, the construct is defined as the set of items that represent the subject to be measured, such as Knowledge and Attitudes about antibiotics. The set of items that are designed to measure a specific parameter such as Knowledge are denoted factor (trait or dimension) [127]. A construct might encompass one or more factors. An empirical assessment of the validity of the construct is carried out using factorial analysis. It is undertaken when the construct of the questionnaire is designed to measure more than one factor, such as the case of our questionnaires that were developed to measure Knowledge about the drug (antibiotics or tranquilizers) and Attitudes towards that drug. For each of the six questionnaires, we carried out a Confirmatory Factorial Analysis (CFA) to examine the validity of the Knowledge and Attitude construct. CFA explores the relationships between the items and their corresponding factor. Items are allocated at first to a specific factor using theoretical and logical grounds. A matrix of correlations between the items is also useful to identify those items that might belong to the same factor or share content; however, the theoretical concepts should always prevail while assigning the items to their factors. In the first stage, we assigned the items to their corresponding factors. Items that tested the Knowledge about the drug were assigned to the Knowledge factor, and those that explored the agreement of the participants about certain Attitudes towards the use of the drug and towards health care provider were attributed to the Attitude factor. In the second stage, the normal distribution of the data of each item was checked and then CFA was undertaken using the Maximum Likelihood Robust Estimation method. Missing data were Narmeen Mallah Methods 23 handled by applying Full Information Maximum Likelihood (FIML). Factors were standardized by constraining them to a mean of 0 and a variance of 1. We assessed the standardized residual correlations between items and applied the modification indexes method to better select the items to be added to a factor, and consequently enhance the fit of the model [133, 134]. Standardized factor loading estimates represent the correlation between the items with their related factors. For each questionnaire, we tested several models and examined the goodness of fit of each model to select the most appropriate one. We explored the goodness of fit of the models using the following statistics: Root Mean Squared Error Approximation (RSMEA, acceptable if <0.08), Comparative Fit Index (CFI, acceptable if ≥0.90), Tucker-Lewis Index (TLI, acceptable if ≥0.90) and Standardized Root Mean Square Residual (SRMR, acceptable if <0.08) [135]. In addition to these indicators, we compared Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and sample-size adjusted BIC (aBIC). AIC reveals the relative amount of information lost by a model. The lower AIC values the better the quality of the models is. BIC is an indicator similar to AIC, however it is more conservative (penalizes the model more) than AIC [136]. 4.3.10. Questionnaire overall reliability We examined the internal reliability of each of the six questionnaires by calculating Cronbach’s coefficient alpha [130]. A questionnaire was considered to have acceptable internal reliability when an index ≥ 0.6 was obtained [137, 138]. 4.3.11. Questionnaire acceptability We tested the acceptability of each questionnaire by calculating the percentage of individuals who accepted answering the questionnaire. The item-response rate was also computed from the percentage of missing data for each item [139-142] Statistical analyses of questionnaire validation were carried out using SPSS (SPSS Inc. Released 2011. SPSS for Windows, Version 20.0. Chicago), and R (version 4.0.0) and R package: lavaan (version 0.6–6). Narmeen Mallah Methods 24 Figure 1 Flow diagram describing the steps followed to develop and validate the KAP questionnaire on antibiotic use in Galicia. Narmeen Mallah Methods 25 Figure 2 Flow diagram describing the steps followed to develop and validate the KAP questionnaire on tranquilizer use in Galicia. Narmeen Mallah Methods 26 4.4. Application of the validated KAP questionnaires 4.4.1. Association of Knowledge and Attitudes with antibiotic misuse Practices: two studies in Galicia and Lebanon The two KAP studies about antibiotics in Galicia and Lebanon have been published (Drug Safety 2021 (in print) and Plos One 2020, 15(7): e0232464). The corresponding publications are available in Annex II page 209 and 221. 4.4.1.1. Study Design and Measures We used the validated questionnaires to measure the association of Knowledge and Attitudes with Practices of antibiotic misuse in Galicia and Lebanon. Lebanon: A cross-sectional study involving 1421 parents of schoolchildren was carried out between November 2018 and January 2019 to determine any misuse of antibiotics in the past month. Spain: A cohort study involving 847 attendants of primary care clinics was undertaken between January and December 2019. Individuals who answered the baseline questionnaire were followed-up by phone bimonthly to ask about their use of antibiotics in the past two months. Participant who reported using antibiotics were asked the questions of the Practice block of the questionnaire to ascertain the occurrence of misuse (Q19 to Q24 of the English version of the questionnaire available in Annex I page 168). The questionnaires used in Galicia and Lebanon in their respective languages are also available in Annex I. Individuals who did not participate in any follow-up questionnaire and who could not be reached after at least four attempts were excluded from the study. 4.4.1.2. Exposure The exposure was defined as low level of Knowledge, medically inappropriate Attitudes towards antibiotics, or negative Attitudes towards healthcare provider. The exposure was ascertained using the items of the Knowledge and Attitudes of the questionnaire. Each of those items measured an independent exposure that was analyzed in a separate model. 4.4.1.3. Outcome The outcome was the misuse of antibiotics. A misuse event was defined as the occurrence of any of the following five practices: 1) use of antibiotics without prescription (Q19), 2) shortening the course of treatment (Q20), 3) storing antibiotic leftover or sharing them with someone else (Q21), 4) not adhering to the treatment regimen regarding the timing and the dosage, i.e. skipping doses (Q22), doubling the dose when forgetting to take the previous dose or taking it when remembered (Q23), or 5) changing the prescribed dose without medical advice (Q24). Each of those five Practices of misuse represented a separate outcome. A sixth composite outcome “any misuse of antibiotics” was generated if one or more of those five Narmeen Mallah Methods 27 Practices were presented. The cited question numbers correspond to the English version of the questionnaire available in Annex I page 168. 4.4.1.4. Potential confounding variables We explored the effect of potentially confounding sociodemographic variables. These variables included age, gender, educational level, employment, family size, frequency of medical consultations in case of sickness and receiving a medical prescription over the phone. Besides these covariables, we additionally controlled for the following factors in the study carried out in Lebanon: income, having health insurance, availability of a healthcare facility close to the dwelling, marital status, spouse educational level and spouse employment status. 4.4.1.5. Statistical analysis Cross-sectional study in Lebanon Associations of Knowledge and Attitudes with antibiotic misuse were modelled using multiple logistic regression. Adjusted Interquartile Odds Ratios (aIqOR) and their 95% CIs were calculated to measure the effect of exposure change from the 25th to the 75th percentile. Cohort study in Spain The data of the cohort study were analysed using the following two approaches: Person-time approach using Poisson regression: Each participant contributed to the study by a person-time value since the beginning of the study until the occurrence of the first antibiotic misuse Practice, drop-out, loss to follow-up or end of the study, whichever took place first. We estimated adjusted Incidence Rate Ratios (aIRRs) of antibiotic misuse and their 95% CIs using Poisson Regression models. Longitudinal Approach using Generalized Linear Mixed Models (GLMM): GLMM analysis allows considering repeated observations contributed by the same individual during the followup periods. Using GLMM models fitted with the binomial family, adjusted ORs (aORs) of antibiotic misuse Practices and their 95% CIs were estimated. In Poisson and GLMM analyses, the data was categorized per quantiles of the distribution of the exposure (Knowledge and Attitude items). The category the represented the highest level of Knowledge or the most medically approved Attitude was used as a referent. Comparison of Cross-sectional and Longitudinal approaches To compare findings from cross-sectional and longitudinal approaches of analysis of data collected from the same individuals, we analysed the data collected at the baseline of the cohort study. Only individuals who answered the baseline questionnaire and at least one follow-up assessment were included. aORs and 95%CI were estimated using multivariate logistic regression models. For all statistical models, adjustment for potential confounders was carried out using the change-in-estimate method [143]. Covariables with p-value < 0.2 were introduced consecutively into the model, and those that modified the value of the measure of effect (OR or IRR) by at least 10% were retained in the final model. Narmeen Mallah Methods 28 Robustness Analyses We undertook two robustness analyses to examine the effect of attrition bias due to loss to follow-up: 1) we used Multiple Imputations by Chained Equations (MICE) with 20 imputed data sets. We followed Rubin’s rules to combine the results across the 20 imputed datasets [144, 145], and then re-estimated aIRRs and their 95%CIs from the imputed data using Poisson regression models. 2) we recalculated the originally estimated aIRRs and their 95%CIs under two extreme conditions. In the first assumption, we assumed that all individuals who were lost to follow-up had misused antibiotics (presented the outcome). In the second scenario, we supposed that none of the participants who were lost to follow-up had misused antibiotics (did not present the outcome). aIRRs and their 95%CIs were recalculated using Poisson regression models in both scenarios. Statistical analyses were carried out using STATA v12 (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP.), and lme4 package of R Statistical Software (version 4.0.1). 4.4.2. Association of Knowledge and Attitudes with tranquilizer misuse Practices: two studies in Galicia and Lebanon As in the study about antibiotics, we used the validated KAP questionnaires about tranquilizers to measure the association of Knowledge and Attitudes with Practices of tranquilizer misuse in Galicia and Lebanon. We carried out a cross-sectional study in Lebanon and a cohort study in Galicia. We followed the same protocol as that of the study on antibiotics. We also compared cross-sectional and longitudinal approaches with respect to KAP studies on tranquilizer misuse. In addition, for the cohort study in Galicia we carried out a validation sub-study to examine the robustness of our results to exposure misclassification [146]. We calculated the specificity and the sensitivity of the questionnaire using the data collected for the test-retest analysis during the questionnaire validation step, as no other superior instrument (gold standard) is available. We compared the ascertainment of the exposure (low level of Knowledge and medically disapproved Attitudes) by using the answers of the first round of the test-retest as a reference. We then applied the specificity and the sensitivity estimates to correct the previously computed relative risk measures. We compared ORs before and after correction assuming that there were no dropouts. Statistical analyses were carried out using SPSS (SPSS Inc. Released 2011. SPSS for Windows, Version 20.0. Chicago), and mgcv package of R Statistical Software (version 4.0.5) [147]. The two KAP studies about tranquilizers in Galicia and Lebanon have been published [(Psychology and Health 2021 (in print) & Drug and Alcohol Dependence 2021 (in print)]. The corresponding publication of Drug and alcohol dependence is available in Annex II page 234. The article of Psychology and Health is not reproduced. Another manuscript about the comparison of cross-sectional and longitudinal approaches is currently under review by Journal of Clinical Medicine. Narmeen Mallah Methods 29 4.4.3. Dose-response meta-analyses of the socioeconomic determinants of antibiotic misuse This part of the thesis was carried out during a three-month internship at the Department of Global Public Health, Karolinska Institutet, Stockholm, Sweden (Annex III page 246). Two systematic reviews and dose-response meta-analyses were undertaken to quantify the association of income and education with antibiotic misuse Practices. We defined antibiotic misuse as the use or purchase of these drugs without medical prescription to treat oneself or another person, storage of antibiotics, and non-adherence to physicians’ instructions in terms of dosage, timing, and duration of treatment. 4.4.3.1. Dose-response meta-analysis of the association of income with antibiotic misuse Practices To date, there is no clear evidence of the association of income with antibiotic misuse, where divergent results were reported across studies. Therefore, we aimed to carry out a dose-response meta-analysis of income levels and antibiotic misuse. The protocol of this meta-analysis is registered in PROSPERO (Protocol ID: CRD42021233075). Literature Search and Study Selection We searched each of the following databases since their inception until January 19th, 2021 without any language or date restrictions: Medline, EMBASE, Conference Proceedings Citation Index-Science, the Open Access Theses and Dissertations, and the five regional bibliographic databases of the World Health Organization: African Index Medicus, Latin American and Caribbean Health Science Literature Database, Index Medicus for the Eastern Mediterranean Region, Index Medicus for South-East Asia Region and Western Pacific Region Index Medicus. We used this search syntax in Medline (Socioeconomic Factors OR income) AND (antibiotic*) AND ((drug storage [MeSH]) OR (compliance) OR (adherence) OR (Nonprescription Drugs / administration & dosage* [MeSH]) OR (misuse) OR (irrational use) OR (left-over)) and adapted it for the other databases. To ensure the comprehensiveness of the search, we additionally reviewed the reference lists of related reviews and those of included studies. We included studies that met the following criteria: 1) providing at least two levels of monthly income as an exposure, with defined boundaries, 2) measuring the association between income and antibiotic misuse by the general population, and 3) reporting OR or Risk Ratio (RR) and their 95% CI, or data for their calculation. When more than one report corresponded to the same study population, we included the most complete one. Authors of related studies were contacted to inquire about any lacking information before deeming a variable missing or excluding any study due to lack of data. Narmeen Mallah Methods 30 Data Extraction and Synthesis From each included study, we extracted: 1) General study characteristics: first author last name and year of publication, study period, participants՚ characteristics (age and gender), and country where the study took place. 2) Exposure (dose): levels of monthly income. We defined the dose as the midpoint of the upper and lower boundaries of a specific income category. When the lowest boundary was open-ended, it was set to zero, and when the upper boundary of a specific income level was not reported, we assumed that this level has the same amplitude as the adjacent level. Income that was reported on weekly or yearly basis was converted to monthly units. Standardizing the income level: to control for socioeconomic differences across countries, we standardized the income using the following two indicators as proxies of socioeconomic status: A) Gross domestic product (GDP) per capita based on purchasing power parity (PPP). GDP approximates the total value of all goods and services produced within a country for a defined period [148]. PPP is a currency conversion rate that equalises the purchasing power of different monetary units and thus permits comparing living standards and economic productivity across countries. [149]. B) Adjusted net national income per capita (expressed in US dollars): is an indicator that complements gross national income (GNI) in evaluating economic progress by providing a broader measure of national income that accounts for the depletion of natural resources such as forest, energy, and mineral depletion. Depletion of natural resources reflects the decrease in asset values (i.e. stock) related to the extraction and harvesting of natural resources [148]. We extracted the historical values of PPP, GDP per capita based on PPP, and adjusted net national income per capita that corresponds to the year of data collection in the studies included in the meta-analysis from their specific portals in World Bank [149-151]. 3) Measure of effect: for each income level, we extracted the number of subjects who practiced antibiotic misuse, the total sample size, i.e., the total of subjects who practiced and who did not practice antibiotic misuse, adjusted OR or RR and 95% CIs, and restriction, adjustment, or matching variables. We used the OR or RR adjusted for the largest number of variables when possible. 4) Multiple outcomes: When data was provided for more than one outcome of antibiotic misuse Practices such as use without prescription and non-adherence to treatment regimen, we extracted the data of all Practices. In the dose-response analysis, each outcome was treated as a separate study. 5) Countries classification by economy: to stratify the studies according to countries wellness, we extracted the corresponding classification from the World Bank [152]. Country economies are categorized as Low income, Lower middle income, Upper middle income and High income. To avoid confusion throughout the text, we adopted the term “wellness” such as low wellness countries, instead of “income”, when referring to a country. 6) Countries classification by WHO region: Countries are classified geographically by WHO into six regions [153]: African region, Eastern Mediterranean region, European region, Region of the Americas, South-East Asian region and Western Pacific region. Narmeen Mallah Methods 31 Quality Assessment The methodological quality of studies included in the meta-analysis was evaluated using an adapted version of New Castle Ottawa Scale for cross-sectional studies [154]. We gave 1 point for the fulfilment of each of the following aspects: 1) well-defined population (1 point), else (0 point); 2) reported response rate (1 point), else (0 point); 3) well described and appropriate statistical analysis (1 point), else (0 point); 4) justified sample size (1 point), else (0 point); 5) studies adjusted, matched or restricted for age, sex, educational level and household size (1 point), else (0 point); 6) previously tested or validated questionnaire (1 point), else (0 point); and 7) external assessment such as examination of prescription in addition to self-reporting was carried out to determine the outcome (1 point), else such as only self-report (0 point). When details on a specific item were not provided, we graded this item with 0 point. Finally, the points across the items were added to obtain a global quality score of maximum seven points. Statistical Analysis We undertook dose-response meta-analysis of income standardized to proxies of socioeconomic indicators: 1) GDP per capita PPP and 2) adjusted net national income per capita. We carried out the dose-response meta-analysis using a one-stage mixed-effects model taking into account heterogeneity across studies [155, 156]. We first used a linear function to estimate a summary OR of antibiotic misuse associated with an increase of 1 unit in income. We next flexibly modelled the dose, standardized income, using restricted cubic splines with 3 knots fixed at 10th, 50th and 90th percentiles of its distribution. We tested the departure of the second spline from non-linearity using Wald type test. Subsequently, we categorized the income into tertiles using the lowest as comparator to facilitate a tabular presentation of the summary ORs. We performed a stratified analysis by: 1) type of antibiotic misuse Practice (use without prescription, non-adherence to treatment, or storage/sharing of antibiotic leftover), 2) WHO geographic classification, 3) country economy, 4) country literacy rate (≥90%, <90%); 5) publication year (≤2015, >2015. 2015 is the year of WHO release to its global action plan to defeat antibiotic resistance [31]), 6) method of exposure ascertainment (use of pretested/validated questionnaire; untested questionnaire or not reported); 7) comparability (control for age, sex, educational level and household size; incomplete control); and 8) Quality score (≤ 3 points, >3 points). We checked for publication bias visually using funnel plot and formally through Egger’s test [157], and the trim and fill method [158]. The manuscript of the meta-analysis of income and antibiotic misuse was submitted for publication and is currently under review by European Journal of Health Economics. Narmeen Mallah Results 38 5.3.3. Test–retest reliability In Galicia One hundred forty individuals answered the KAP questionnaire about antibiotics on the twooccasions. Table 1 summarizes ICC values for the Knowledge and Attitude items towards antibiotics. All items showed acceptable reliability index (ICC ≥ 0.4), except Q5, Q10 and Q11. Low ICC values could be caused by the lack of sample heterogeneity for those items. Accordingly, to examine further the reliability of Q5, Q10 and Q11, we calculated their Spearman’s correlation coefficient (rs). The answers on Q5 on the two occasions were weakly correlated (rs = 0.193; p = 0.022), and those on Q10 and Q11 were moderately correlated: Q10 (rs = 0.433; p < 0.0001) and Q11 (rs = 0.405; p < 0.0001). One-hundred thirty-seven adults completed twice the KAP questionnaire about tranquilizers. Table 2 represents ICC estimates for the Knowledge and Attitude statements about tranquilizers. All ICC values were >0.5, proving the reliability (reproducibility) of these items. In Lebanon Sixty Lebanese adults answered the two rounds of the KAP questionnaire about antibiotics and tranquilizers in Arabic, and other 31 individuals answered twice the questionnaires in French. Table 1 and Table 2 represent ICC estimates for the Knowledge and Attitude statements in the Arabic and French questionnaires about antibiotics and tranquilizers, respectively. In the four questionnaires, ICC values were >0.5, demonstrating the reproducibility of the Knowledge and Attitude items. 5.3.4. Construct validity of KAP questionnaire about antibiotic use in Galicia Model 1: According to theoretical grounds, the 17 items of the Knowledge and Attitude construct were distributed into two factors: Knowledge about antibiotics (Knowledge), and Attitude towards antibiotics (Attitude). The structure of this initial 2-factor model (Model 1) was tested by CFA. As shown in Table 3, the goodness of fit indicators of Model 1 did not present an acceptable fit. Model 2: In Model 1, Q8 “Antibiotics can kill the bacteria that normally live on the skin and in the gut” did not load significantly in the Knowledge factor. Besides, the items “Q10, Q13, Q15-Q17” loaded negatively in the Attitude factor. The items “Q10, Q13, Q15-Q17” examined the patient-health care provider relationship with respect to antibiotics, hence they were assigned to a third factor (Attitude-Healthcare provider) in a new model named Model 2. Model 2 consisted of three factors: 1) Knowledge: “Q1, Q2, Q4, Q6-Q8 and Q11”, 2) AttitudePersonal: “Q3, Q5, Q9, Q12, Q14” and 3) Attitude-Healthcare provider: “Q10, Q13, Q15Q17”. The goodness of fit indicators of Model 2 are represented in Table 3. They were better than those of Model 1, yet the model was not acceptable. Model 3: To improve the structure of Model 2, various iterations were made based on crossloadings and modification indexes. Correlations between item residuals were also analyzed in order to enhance further the goodness of fit. As a result, Q8 “Antibiotics can kill the bacteria that normally live on the skin and in the gut” was removed from Model 2, and the modified Narmeen Mallah Results 39 model was called Model 3. Table 4 summarizes the loadings of the 16 preserved KAP items on their respective factors. As shown in Figure 3, there was significant correlations between the residuals of Q4 and Q7 (r= 0.12, p-value= 0.001), Q10 and Q13 (r= 0.18, p < 0.0001), Q11 and Q10 (r= 0.20, p-value < 0.0001) and Q13 and Q14 (r= − 0.13, p-value= 0.001). Model 3 represented an adequate fit as concluded from indicators of the goodness of fit assessment: RMSEA= 0.044; CFI= 0.92; TLI= 0.90; SRMR= 0.047 (Table 3). There was a statistically significant difference between Model 2 and Model 3 as revealed by χ2 test (Δ χ2= 331.97, Δdf= 22, p= 0.0001). As compared to the other models, Model 3 showed the lowest AIC, BIC and aBICR values. Therefore, Model 3 was adopted for the questionnaire in the Galician adult population. There was a statistically significant positive correlation between the two factors: Knowledge and Attitude-Personal (r= 0.56, p < 0.0001). Knowledge and Attitude-Healthcare provider were not correlated (r= 0.04, p= 0.417). A weak negative correlation existed between AttitudePersonal and Attitude-Healthcare provider (r= − 0.11, p= 0.023) (Figure 3). 5.3.5. Construct validity of KAP questionnaire about antibiotics in French in Lebanon Model 1: In the initial model (Model 1) we followed the structure of the model validated earlier in the Galician population to allocate the Knowledge and Attitudes items into three factors: Knowledge (Q1, Q2, Q4, Q6-8 and Q11), Attitude-Personal (Q3, Q5, Q9, Q12 and Q14) and Attitude-Healthcare provider (Q10, Q13, Q15-17). The fit indicators of Model 1 were poor (Table 5). Model 2: To improve the model fit, we applied modification indexes which suggested the inclusion of Q10 in the Attitude-Personal (Model 2). In this model, items Q7 “Each type of infection needs a different antibiotic” and Q9 “If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible” did not significantly load on their respective factors. The fit indicators improved but they remained outside the acceptable range. Model 3: Q7 and Q9 were removed from the model and the residuals of the items Q10 and Q13 were correlated as it was suggested by the analysis of the correlations between item residuals. The new model (Model 3) yielded better fit indicators than that of Model 2, but the model was not adequately fit yet. Model 4: Accordingly, in Model 4, we additionally correlated items Q11 with Q10, and Q1 with Q3 and Q8. The fit indicators of the model enhanced, but they could be improved further. Model 5: Based on various iterations, we moved item Q6 “When I have a sore throat, I prefer to use an antibiotic” to the factor Attitude-Personal and correlated Q3 with Q2 and Q12. There was a statistically significant difference between Model 5 and Model 4 as revealed by χ2 test (Δ χ2= 70.15, Δdf= 3, p < 0.0001). AIC, BIC and aBICR were the lowest for Model 5 as compared to the other models. In Model 5, all items loaded significantly on their corresponding factors (Table 6), and significant correlations were observed between Q10 and Q13 (r= 0.196; p-value< 0.0001), Q11 and Q10 (r= 0.192; p-value< 0.0001); Q1 and Q3 (r= 0.236, p-value< 0.0001); Q1 and Q8 (r= -0.186, p-value< 0.0001); Q3 and Q12 (r= -0.135, p-value< 0.0001); and Q2 and Q3 (r= 0.247, p-value< 0.0001). Model 5 showed adequate fit: RMSEA= 0.043; CFI= 0.926; TLI= 0.901; SRMR= 0.042. Therefore, Model 5 was adopted for the French questionnaire about antibiotics in the Lebanese population (Figure 4). Knowledge factor was Narmeen Mallah Results 40 positively correlated with Attitude-Personal (r= 0.434, p-value<0.0001), but not with AttitudeHealthcare provider (r= 0.057, p-value= 0.181). The two factors Attitude-Personal and Attitude-Healthcare provider also did not correlate (r= -0.026, p-value= 0.574). 5.3.6. Construct validity of KAP questionnaire about tranquilizers in Galicia Model 1: The 16 Knowledge and Attitude items about tranquilizers were initially assigned to two factors: Knowledge and Attitudes. The Attitude factor englobed items that were designed to measure attitudes of the patients towards tranquilizers as well as towards the healthcare provider. Table 7 shows that Model 1 had poor fit indicators. The inspection of the factor loadings of this model indicated that items Q9 and Q14-Q16 did not load significantly on the Attitude factor. These items were specifically designed to measure attitudes of the patients towards the healthcare provider. A further evaluation of the item content and the correlation matrix suggested the addition of a third factor to Model 1. Model 2: Accordingly, the Attitude items were allocated into two factors: Attitude-Personal which encompassed Q1-Q4, Q11 and Q13 and Attitude-Healthcare provider that included Q9, Q12 and Q14-Q16 (Model 2). The items loaded significantly on their corresponding factors (Knowledge, Attitude-Personal, and Attitude-Healthcare provider), and the fit indicators of Model 2 were better than that of Model 1, but they were not acceptable yet (Table 7). Therefore, we explored the modification statistics which suggested correlating the residuals of certain items within the factors to enhance the specification of the model. Model 3: The new model (Model 3), has the same structure as Model 2 but the residuals of the following items were correlated: Q3-Q4 (rQ14-Q15= 0.76, p-value< 0.001) and Q14-Q16 (rQ14Q16= 0.34, and rQ15-Q16= 0.35, both p-value< 0.001). The magnitude of χ2 statistic importantly reduced and Model 3 showed better fit indicators than Model 2 (Table 7), yet the model required to be better specified. Model 4: An additional analysis relying on item content and modification statistics suggested the creation of a new model (Model 4) with a cross-load of item Q13 on the two factors: Attitude-Personal and Attitude-Healthcare provider. This modification to the model structure was appropriate since the content of Q13 “If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription” includes information related to both Attitude factors. Model 4 showed acceptable fit indicators CFI= 0.93, TLI= 0.92, RSMEA= 0.054 and SRMR= 0.05) (Table 7). The χ2 Difference Test showed a significant decrease in the magnitude of χ2 with respect to Model 3 (Δ χ2= 67.66, Δdf= 1, p-value< 0.001). Model 4 also had the lowest AIC, BIC and aBIC values as compared to the other models (Table 7). In Model 4, all items loaded significantly on their corresponding factors (Table 8). Therefore, Model 4 was selected for the Galician adult population (Figure 5). There was a negative relation between Knowledge and Attitude-Personal factors (r= -0.60, pvalue< 0.0001), and a positive relation between Attitude-Personal and Attitude-Healthcare provider (r= 0.24, p-value< 0.038). Narmeen Mallah Results 41 5.3.7. Construct validity of KAP questionnaire about tranquilizers in French in Lebanon Model 1: Following the structure of the model adopted in Galicia, we assigned the 16 Knowledge and Attitude items to three factors: Knowledge (Q5-Q8 and Q10); Attitude-Personal (Q1-4 and Q11); Attitude-Healthcare provider (Q9, and Q12-Q16). Model 1 had poor fit indicators (Table 9) and item Q13 “If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription” did not significantly load on the factor “Attitude-Healthcare provider”. Model 2: Based on modification indexes, item Q13 was moved to factor Attitude-Personal (Model 2). All items loaded significantly on their respective factors and the model fit indicators improved but they were still not in the acceptable range (Table 9). Model 3: As suggested by the modification indexes, we correlated the residuals of items Q3 and Q4 and those of Q11 and Q13. Model 3 showed an adequate fit; CFI= 0.93, TLI= 0.91 RSMEA= 0.059 and SRMR= 0.055 (Table 9). The items loaded significantly on their corresponding factors (Table 10). The magnitude of χ2 statistic, AIC, BIC and aBIC was the lowest for Model 3 (Table 9). Therefore, Model 3 was adopted for the Lebanese population (Figure 6). The factors Knowledge and Attitude-Personal showed a negative correlation (r= -0.261, p< 0.0001). There was a weak but positive relation between Knowledge and Attitude-Healthcare provider (r= 0.151, p< 0.001). No relation was observed between Attitude-Personal and Attitude-Healthcare provider (r= 0.046, p< 0.218). 5.3.8. Questionnaire and item acceptability In Galicia, out of 879 individuals, 844 (96%) and 836 (95%) completed the KAP questionnaires on antibiotics and tranquilizers, respectively, revealing a high acceptability of these questionnaires by the Galician population. The proportion of the unanswered questions in the KAP questionnaire on antibiotics was between 0.4% and 2.7%, and that in the KAP questionnaire on tranquilizers ranged between 1% and 4%, indicating an elevated item-response rate. In Lebanon, out of 1046 individuals, 1016 (97.1%) and 1015 (97.0%) completed the KAP questionnaires in French about antibiotics and tranquilizers, respectively. The proportion of unanswered questions in the antibiotic questionnaire was between 0.1% and 4.4%, and that in the tranquilizer questionnaire ranged between 0.1% and 3.1%, indicating an elevated itemresponse rate. Accordingly, the four questionnaires were deemed highly accepted by the general adult population in Galicia and Lebanon. Narmeen Mallah Results 42 5.3.9. Questionnaire overall reliability Using data of the returned questionnaires, the reliability of the questionnaires in Galician and French languages about antibiotics and tranquilizers were assessed. Concerning the questionnaires in Galician, the KAP questionnaire about antibiotics had a Cronbach’s alpha= 0.62. For KAP questionnaire about tranquilizers, the Cronbach’s alpha was 0.65 for the Knowledge factor, 0.72 for Attitude-Personal factor, and 0.65 for AttitudeHealthcare provider factor. The two instruments validated in Galicia showed an acceptable reliability index. Regarding the questionnaires in French, the Cronbach’s alpha was low (0.54) for the KAP questionnaire on antibiotics, but acceptable for KAP questionnaire about tranquilizers (0.67). Narmeen Mallah Results 43 Table 1 Intraclass Correlation Coefficients (ICCs) and 95% Confidence Intervals (CIs) of the Knowledge and Attitude items of the Galician, Arabic and French versions of the questionnaire on antibiotic use №: number of participants who answered the KAP questionnaire on antibiotics on two occasions. The questionnaire in Galician was tested in Galicia, while those in Arabic and French were tested in Lebanon. Item ICC (95%CI) Galician (№=140) Arabic (№=60) French (№=31) Q1. Antibiotics are effective against viruses 0.82 (0.75, 0.87) 0.74 (0.57, 0.85) 0.94 (0.86, 0.97) Q2. When I get a cold, I take antibiotics to help me feel better faster 0.62 (0.48, 0.73) 0.72 (0.53, 0.83) 0.94 (0.88, 0.97) Q3. If I feel better after a few days, I sometimes stop taking my antibiotics before completing the course of treatment 0.40 (0.17, 0.57) 0.83 (0.72, 9.00) 0.79 (0.56, 0.90) Q4. I expect my doctor to prescribe antibiotics if I suffer from common cold or flu symptoms 0.68 (0.55, 0.77) 0.60 (0.33, 0.76) 0.82 (0.62, 0.91) Q5. It is good to be able to get antibiotics from relatives or friends without having to see a medical doctor 0.06 (-0.32, 0.32) 0.81 (0.69, 0.89) 0.84 (0.67, 0.92) Q6. When I have a sore throat, I prefer to use an antibiotic 0.62 (0.46, 0.73) 0.84 (0.73, 0.90) 0.96 (0.91, 0.98) Q7. Each type of infection needs a different antibiotic 0.67 (0.54, 0.77) 0.75 (0.60, 0.85) 0.78 (0.52, 0.89) Q8. Antibiotics can kill the bacteria that normally live on the skin and in the gut 0.67 (0.54, 0.77) 0.66 (0.42, 0.79) 0.84 (0.67, 0.92) Q9. If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible 0.60 (0.44, 0.72) 0.73 (0.54, 0.84) 0.95 (0.90, 0.98) Q10. I take the antibiotics according to the doctor’s instructions 0.36 (0.10, 0.54) 0.65 (0.41, 0.79) 0.84 (0.67, 0.92) Q11. If antibiotics are consumed in excess, they will not work when they are really needed 0.19 (0.14, 0.42) 0.71 (0.51, 0.83) 0.67 (0.31, 0.84) Q12. I prefer to keep antibiotics at home in case there is a need for them later 0.58 (0.41, 0.70) 0.76 (0.60, 0.86) 0.87 (0.73, 0.94) Q13. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics 0.50 (0.29, 0.64) 0.64 (0.39, 0.78) 0.52 (0.01, 0.77) Q14. If I believe that I need an antibiotic and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0.50 (0.29, 0.64) 0.67 (0.45, 0.80) 0.85 (0.68, 0.93) Q15. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing antibiotics 0.70 (0.58, 0.79) 0.77 (0.62, 0.86) 0.78 (0.55, 0.90) Q16. Doctors often explain clearly to the patient the instructions for the use of antibiotics 0.65 (0.51, 0.75) 0.82 (0.70, 0.89) 0.76 (0.49, 0.89) Q17. When you buy antibiotics, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 0.63 (0.48, 0.74) 0.90 (0.83, 0.94) 0.67 (0.38, 0.86) Narmeen Mallah Results 44 Table 2 Intraclass Correlation Coefficients (ICCs) and 95% Confidence Intervals (CIs) of Knowledge and Attitude items of the Galician, Arabic and French versions of the questionnaire on tranquilizer use №: number of participants who answered the KAP questionnaire on tranquilizers on two occasions. The questionnaire in Galician was tested in Galicia, while those in Arabic and French were tested in Lebanon. Item ICC (95%CI) Galician (№= 137) Arabic (№= 60) French (№= 31) Q1. I would agree to take tranquilizers in order to sleep better 0.84 (0.77, 0.88) 0.74 (0.57, 0.85) 0.86 (0.71, 0.93) Q2. If I feel better after a few days, I will keep taking my tranquilizers even after completing the prescribed course of treatment 0.72 (0.60, 0.80) 0.78 (0.62, 0.87) 0.69 (0.35, 0.85) Q3. I would take tranquilizers in order to enjoy myself with my family 0.77 (0.67, 0.83) 0.76 (0.60, 0.89) 0.76 (0.50, 0.88) Q4. I would agree to take tranquilizers when I feel down and sad in order to work better 0.73 (0.62, 0.81) 0.77 (0.61, 0.86) 0.77 (0.52, 0.89) Q5. Tranquilizers reduce people’s control over what they do 0.68 (0.55, 0.78) 0.63 (0.38, 0.78) 0.68 (0.34, 0.85) Q6. People taking tranquilizers are at increased risk of traffic accidents 0.71 (0.59, 0.79) 0.54 (0.23, 0.73) 0.77 (0.52, 0.89) Q7. Psychotropic drugs (such as tranquilizers) may affect children’s learning abilities when prescribed to them 0.71 (0.59, 0.80) 0.70 (0.50, 0.82) 0.88 (0.75, 0.94) Q8. If I feel side effects during a course of treatment of tranquilizers, I should stop taking it as soon as possible 0.60 (0.43, 0.71) 0.57 (0.28, 0.74) 0.94 (0.87, 0.97) Q9. I would take the tranquilizers according to the doctor’s instructions 0.77 (0.67, 0.84) 0.66 (0.43, 0.80) 0.88 (0.75, 0.94) Q10. If tranquilizers are consumed in excess, they won´t work when they are really needed 0.59 (0.42, 0.71) 0.75 (0.59, 0.85) 0.82 (0.63, 0.91) Q11. I prefer to keep tranquilizers at home in case there is a need for them later 0.69 (0.56, 0.78) 0.65 (0.42, 0.79) 0.92 (0.83, 0.96) Q12. I will trust the doctor’s decision if s/he decides to prescribe or not prescribe tranquilizers 0.82 (0.74, 0.87) 0.66 (0.43, 0.70) 0.75 (0.47, 0.88) Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0.53 (0.34, 0.67) 0.68 (0.47, 0.81) 0.88 (0.75, 0.94) Q14. I think that doctors often explain clearly to the patients the reasons for prescribing or not prescribing tranquilizers 0.65 (0.51, 0.75) 0.67 (0.45, 0.81) 0.72 (0.43, 0.87) Q15. I think that doctors often explain clearly to the patients the instructions for the use of tranquilizers 0.63 (0.47, 0.74) 0.82 (0.69, 0.89) 0.88 (0.75, 0.94) Q16. I think that, when dispensing tranquilizers, the pharmacists tell the customers about the importance of correct therapeutic compliance/adherence 0.65 (0.51, 0.76) 0.76 (0.59, 0.86) 0.78 (0.55, 0.90) Narmeen Mallah Results 45 Table 3 Comparison of the goodness of fit parameters between models of Knowledge, Attitudes and Practice questionnaire on antibiotic use in the Galician adult population Model 1 encompassed two factors (Knowledge and Attitude), Model 2 consisted of three factors (Knowledge, Attitude-Personal and Attitude-Healthcare provider) and Model 3 involved the same factors as Model 2 but excluding Q8. χ 2: Chi-square value; df: Degree of Freedom; p: p-value (Chi-square); RSMEA: Root Mean Squared Error Approximation; CFI: Comparative Fit Index; TLI: Tucker-Lewis Index; AIC: Akaike Information Criterion, BIC: Bayesian Information Criterion; aBIC: sample-size adjusted BIC; SRMR: Standardized Root Mean Square Residual Indicator Model 1 Model 2 Model 3 χ 2 1037.074 580.45 248.49 df 118 116 94 p <0.0001 <0.0001 <0.0001 RSMEA (90% CI) 0.096 (0.091, 0.102) 0.069 (0.063, 0.075) 0.044 (0.038, 0.051) CFI 0.55 0.77 0.92 TLI 0.49 0.74 0.90 AIC 67594.035 67141.414 62873.317 BIC 67840.296 67397.146 63147.993 aBIC 67675.160 67225.659 62963.803 SRMR 0.088 0.073 0.047 Narmeen Mallah Results 46 Table 4 Factor loadings and standard errors from the three-factor model (Model 3) of the Knowledge, Attitudes and Practice questionnaire on antibiotic use in Galicia Item Loading estimate Standard Error P-value Standard loading estimate Knowledge Q1. Antibiotics are effective against viruses 1.65 0.09 <0.0001 0.60 Q2. When I get a cold, I take antibiotics to help me feel better faster 1.46 0.08 <0.0001 0.81 Q4. I expect my doctor to prescribe antibiotics if I suffer from common cold or flu symptoms 1.22 0.09 <0.0001 0.40 Q6. When I have a sore throat, I prefer to use an antibiotic 0.49 0.09 <0.0001 0.26 Q7. Each type of infection needs a different antibiotic 0.12 0.08 0.014 0.09 Q11. If antibiotics are consumed in excess, they will not work when they are really needed -0.38 0.06 <0.0001 -0.25 Attitude-Personal Q3. If I feel better after a few days, I sometimes stop taking my antibiotics before completing the course of treatment 1.55 0.09 <0.0001 0.55 Q5. It is good to be able to get antibiotics from relatives or friends without having to see a medical doctor 0.66 0.05 <0.0001 0.56 Q6. When I have a sore throat, I prefer to use an antibiotic 0.84 0.12 <0.0001 0.37 Q9. If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible 0.62 0.11 <0.0001 0.21 Q10. I take the antibiotics according to the doctor’s instructions -0.40 0.05 <0.0001 -0.39 Q12. I prefer to keep antibiotics at home in case there is a need for them later 1.28 0.09 <0.0001 0.48 Q13. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics -0.34 0.08 <0.0001 -0.18 Q14. If I believe that I need an antibiotic and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0.90 0.06 <0.0001 0.54 Attitude-Healthcare provider Q10. I take the antibiotics according to the doctor’s instructions 0.12 0.04 0.004 0.11 Q13. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics 0.39 0.06 <0.0001 0.23 Q15. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing antibiotics 1.76 0.08 <0.0001 0.77 Q16. Doctors often explain clearly to the patient the instructions for the use of antibiotics 1.69 0.08 <0.0001 0.82 Q17. When you buy antibiotics, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 1.05 0.07 <0.0001 0.41 Narmeen Mallah Results 47 Figure 3 Representation of the model selected by CFA analysis (Model 3) for the Knowledge and Attitude construct of the KAP questionnaire on antibiotic use in the Galician adult population. Each of the three factors (Knowledge, Attitude-Personal and Attitude-Healthcare provider) is represented with its corresponding standardized items loadings and their residuals. Knowledge includes items that explore the knowledge of the adults towards antibiotics. Attitude-Personal encompasses statements about attitudes towards the personal use of antibiotics. Attitude-Healthcare provider involves phrases about the patient-healthcare provider relationship with respect to antibiotics. The double-sided arrows represent correlations between the variables. Q1Q7 and Q9-Q17 are items of the Knowledge and Attitude construct (Q8 was deleted in a previous step). The single headed arrows represent the correlation of the items and their respective factors. Narmeen Mallah Results 54 Table 9 Comparison of the goodness of fit parameters between models of Knowledge, Attitude and Practice questionnaire on tranquilizer use in French language tested in the Lebanese adult population Model 1 consisted of three factors (Knowledge, Attitude-Personal and Attitude-Healthcare provider) and items were allocated based on the model adopted for Galicia. Model 2 differed from Model 1 by the inclusion of Q13 in the factor Attitude-Personal. Model 3 had the same structure as Model 2 but residuals of similar items were correlated. χ 2: Chi-square value; df: Degree of freedom; CFI: Comparative Fit Index; TLI: Tucker-Lewis Index; RSMEA: Root Mean Square Error Approximation; CI: Confidence Interval; SRMR: Standardized Root Mean Square Residual; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; aBIC: Sample-size adjusted BIC Indicator Model 1 Model 2 Model 3 χ 2 801.956 610.044 451.471 df 101 101 99 p <0.0001 <0.0001 <0.0001 RSMEA (90% CI) 0.083 (0.077, 0.088) 0.070 (0.065, 0.076) 0.059 (0.054, 0.065) CFI 0.86 0.90 0.927 TLI 0.83 0.88 0.912 AIC 71168.874 70976.962 70822.388 BIC 71419.929 71228.017 71083.288 aBIC 71257.948 71066.036 70914.956 SRMR 0.087 0.059 0.055 Narmeen Mallah Results 55 Table 10 Standard loading estimates of the Knowledge and Attitude items of the tranquilizer questionnaire (Model 3) in French on their respective factors Item Loading estimate Standard Error P-value Standard loading estimate Knowledge Q5. Tranquilizers reduce people’s control over what they do 1.435 0.052 <0.0001 0.713 Q6. People taking tranquilizers are at increased risk of traffic accidents 1.481 0.053 <0.0001 0.836 Q7. Psychotropic drugs (such as tranquilizers) may affect children’s learning abilities when prescribed to them 1.015 0.045 <0.0001 0.612 Q8. If I feel side effects during a course of treatment of tranquilizers, I should stop taking it as soon as possible 0.502 0.065 <0.0001 0.243 Q10. If tranquilizers are consumed in excess, they won´t work when they are really needed 0.568 0.050 <0.0001 0.350 Attitude-Personal Q1. I would agree to take tranquilizers in order to sleep better 1.190 0.043 <0.0001 0.719 Q2. If I feel better after a few days, I will keep taking my tranquilizers even after completing the prescribed course of treatment 0.755 0.032 <0.0001 0.664 Q3. I would take tranquilizers in order to enjoy myself with my family 1.045 0.040 <0.0001 0.671 Q4. I would agree to take tranquilizers when I feel down and sad in order to work better 1.210 0.042 <0.0001 0.708 Q11. I prefer to keep tranquilizers at home in case there is a need for them later 1.195 0.052 <0.0001 0.584 Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0.604 0.037 <0.0001 0.482 Attitude-Healthcare provider Q9. I would take the tranquilizers according to the doctor’s instructions 0.429 0.042 <0.0001 0.305 Q12. I will trust the doctor’s decision if s/he decides to prescribe or not prescribe tranquilizers 0.763 0.048 <0.0001 0.419 Q14. I think that doctors often explain clearly to the patients the reasons for prescribing or not prescribing tranquilizers 1.396 0.034 <0.0001 0.870 Q15. I think that doctors often explain clearly to the patients the instructions for the use of tranquilizers 1.445 0.035 <0.0001 0.918 Q16. I think that, when dispensing tranquilizers, the pharmacists tell the customers about the importance of correct therapeutic compliance/adherence 0.967 0.041 <0.0001 0.568 Narmeen Mallah Results 56 Figure 6 Confirmatory Factorial Analysis model (Model 3) of the Knowledge and Attitudes construct of the KAP questionnaire in French on tranquilizer use in the Lebanese adult population. The model includes three factors (Knowledge, Attitude-personal and Attitude-Healthcare provider) that are represented with their corresponding items, the standardized loading of these items and their residuals. The correlation between an item and its corresponding factor(s) is represented by a single headed arrow. The correlation between factors is indicated by double-sided arrows. *** corresponds to p-values < 0.0001. Narmeen Mallah Results 57 5.4. Association of Knowledge and Attitudes with Practices of antibiotic misuse The two KAP studies about antibiotics in Galicia and Lebanon have been published (Drug Safety 2021 (in print) and Plos One 2020, 15(7): e0232464). The corresponding publications are available in Annex II page 209 and 221. 5.4.1. Cross-sectional study in Lebanon Participants՚ socio-demographic characteristics One-thousand four-hundred twenty-one individuals completed the KAP questionnaire on antibiotics in Lebanon and were included in the analysis. The demographic characteristics of the participants are represented in Table 11. Almost two-thirds (64.1%) of the participants reported not always visiting a physician in case of illness; 34.3% due to lack of time, 27% declared that it was unnecessary to always consult a physician, and 8.9% because of financial reasons. A similar fraction (60%) mentioned having consulted a doctor over the phone. Patterns of antibiotic misuse In the month preceding the study, more than one-quarter of study participants (16%, N=277) had used antibiotics and 41.4% of them (N=94) showed signs of at least one antibiotic misuse Practice. The most common misuse Practices involved non-prescribed use of antibiotics (22.5%), storing antibiotic leftover for future need and/or sharing the leftover with someone else (22%), and doubling the prescribed dose of antibiotics or taking the skipped dose when recalled, in case of forgetting to take the previous dose (10.6%). Association of Knowledge about antibiotics with the misuse Practices of these drugs Individuals who did not know that antibiotics are not effective against viruses had twice the odds to present any antibiotic misuse Practice [Q1 aIqOR: 2.08 (95% CI: 1.32, 3.19)] (Table 12). In specific, they had higher chance of using antibiotics without medical prescription [aIqOR: 2.83 (95% CI: 1.61, 5.33)], seizing the course of treatment before completion [aIqOR: 2.66 (95% CI: 1.07, 6.27)], and changing the prescribed dose without medical advice [aIqOR: 5.94 (95% CI: 1.23, 28.04)] (Table 13). Individuals who ignored that antibiotics do not treat colds had substantial higher odds of the outcome “any misuse Practice” of antibiotics than those who had better knowledge [Q5 aIqOR: 1.81 (95% CI: 1.41, 2.29)] (Table 12). Specifically, they had more than two-fold higher chance to use antibiotics without prescription [Q5 aIqOR: 2.29 (95% CI: 1.69, 3.22)], shorten the course of antibiotic treatment [Q5 aIqOR: 2.36 (95% CI: 1.52, 3.73)], store or share antibiotic leftover [Q5 aIqOR: 2.22 (95% CI: 1.57, 3.04)] and modify the prescribed dose of antibiotics without medical approval [Q5 aIqOR: 4.42 (95% CI: 2.14, 9.38)]. These individuals also had considerable higher odds of doubling the subsequent dose of antibiotics or taking it when remembered when a previous dose was forgotten [Q5 aIqOR: 1.94 (95% CI: 1.26, 3.13)] (Table 13). Narmeen Mallah Results 58 Association of Attitudes towards antibiotics with the misuse Practices of these drugs Individuals who preferred to use antibiotics to treat sore throat had more than twice higher odds of misusing antibiotics than those who did not prefer to take antibiotics for this purpose [Q7 aIqOR: 2.19 (95% CI: 1.61, 2.93)] (Table 12). In concrete, they had three times the chance to self-prescribe antibiotics [Q7 aIqOR: 3.30 (95% CI: 2.29, 5.00)], were more likely to curtail the treatment course [Q7 aIqOR: 2.19 (95% CI: 1.22, 3.86)], store or share antibiotic leftover [Q7 aIqOR: 2.10 (95% CI: 1.40, 3.05)], and modify the prescribed dose without medical advice [Q7 aIqOR: 4.01 (95% CI: 1.69, 9.54)] (Table 13). Individuals who preferred to store antibiotics at home for potential need in the future had substantial odds of misusing these drugs [Q11 aIqOR: 2.44 (95% CI: 1.68, 3.46)] (Table 12). In particular, they had more than three-fold the odds of using unprescribed antibiotics [Q11 aIqOR: 3.64 (95% CI: 2.19, 6.05)], and five-fold higher chance of changing the prescribed dose of antibiotics without medical approval [Q11 aIqOR: 5.05 (95% CI: 1.59, 16.78)] (Table 13). Similar findings were obtained for individuals who showed their agreement on statements about buying unprescribed antibiotics in case they think that they need an antibiotic, but the physician did not prescribe it (Q13) (Table 12 and Table 13). Association of Attitudes toward healthcare provider with the misuse Practices of antibiotics Individuals who agreed that the physician clearly explains to the patient the reasons for prescribing or not prescribing antibiotics had lower odds to use antibiotics without medical prescription [Q14 aIqOR: 0.63 (95% CI: 0.43, 0.92)] (Table 13). Similar findings were observed for individuals who agreed that the physician clearly explains the instructions of use of antibiotics (Q15). These individuals had reduced odds of the outcome “any antibiotic misuse practice” [Q15 aIqOR: 0.66 (95% CI: 0.47, 0.90)] and especially, they are less likely to use unprescribed antibiotics [Q15 aIqOR: 0.59 (95% CI: 0.37, 0.90)] (Table 12 and Table 13). Narmeen Mallah Results 59 Table 11 Demographic characteristics of individuals participated in the cross-sectional study about the association of Knowledge and Attitudes with Practices of antibiotic misuse in Lebanon *: the outcome “any misuse” encompasses one or more of the following Practices: use without prescription, shortened treatment, modified dose, doubling a skipped dose, or taking it when remembered. Characteristic Total (№= 1421) Any Misuse in the last month (№= 94)* Gender Male 276 (19.4%) 22 (23.4%) Female 1092 (76.8%) 69 (73.4%) Missing 53 (3.7%) 3 (3.2%) Age <38 years 355 (25.0%) 37 (39.4%) 39 -42 years 227 (16.0%) 14 (14.9%) 43 - 47 years 313 (22.0%) 13 (13.8%) >= 48 years 259 (18.2%) 14 (14.9%) Missing 267 (18.8%) 16 (17.0%) Marital status Married 1327 (93.4%) 87 (92.6%) Other 74 (5.2%) 6 (6.4%) Missing 20 (1.4%) 1 (1.1%) Educational level Until high school 252 (17.7%) 31 (33.0%) University 1157 (81.4%) 62 (66.0%) Missing 12 (0.8%) 1 (1.0%) Spouse educational level Until high school 313 (22.0%) 36 (38.3%) University 1028 (72.3%) 53 (56.4%) Missing 80 (5.6%) 5 (5.3%) Number of family members 2-4 555 (39.1%) 36 (38.3%) 5-6 715 (50.3%) 50 (53.2%) >6 129 (9.1%) 6 (6.4%) Missing 22 (1.5%) 2 (2.1%) Family income <500$ 21 (1.5%) 2 (2.1%) 500$ - 1499$ 260 (18.3%) 34 (36.2%) 1500 - 2500$ 223 (15.7%) 17 (18.1%) >2500$ 808 (56.9%) 36 (38.3%) Missing 109 (7.7%) 5 (5.3%) Consulting a doctor Rarely or never 155 (10.9%) 8 (8.5%) Sometimes 756 (53.2%) 63 (67.0%) Always 493 (34.7%) 22 (23.4%) Missing 17 (1.2%) 1 (1.1%) Reasons for not always consulting a doctor No need 390 (27.4%) 24 (25.5%) Fear 29 (2.0%) 2 (2.1%) No money 127 (8.9%) 20 (21.3%) No time 321 (22.6%) 26 (27.7%) Long waiting time 166 (11.7%) 15 (16%) No near clinic 53 (3.7%) 4 (4.3%) Ever received medical consultation over the phone No 491 (34.6%) 30 (31.9%) Yes 892 (62.8%) 62 (66%) Missing 38 (2.7%) 2 (2.1%) Narmeen Mallah Results 60 Table 12 Association of Knowledge and Attitudes with the outcome “any misuse” Practice of antibiotics in 1392 Lebanese parents of schoolchildren Knowledge or attitude statements Percentile of responses No misuse № Any Misuse № alqOR** (95%CI) 25% 50% 75% Q1. Antibiotics are effective against viruses 0 4 7 1171 86 2.08 (1.32, 3.19) Q2. Each type of infection needs a different antibiotic 8 10 10 1174 87 1.17 (0.94, 1.44) Q3. Antibiotics can kill the bacteria that normally live on the skin and in the gut 5 8 10 1138 85 1.16 (0.77, 1.61) Q4. If antibiotics are consumed in excess, they won’t work when they are really needed 8 10 10 1168 85 0.96 (0.81, 1.14) Q5. When I get a cold, I take antibiotics to help me feel better faster 0 0 4 1187 88 1.81 (1.41, 2.29) Q6. I expect my doctor to prescribe antibiotics if I suffer from common cold or flu symptoms 0 5 9 1180 88 1.42 (0.91, 2.36) Q7. When I have a sore throat, I prefer to use an antibiotic 0 0 5 1187 88 2.19 (1.61, 2.93) Q8. If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible 5 9 10 1170 86 1.40 (0.95, 2.01) Q9. It is good to be able to get antibiotics from relatives or friends without having to see a medical doctor 0 0 0 1186 88 1.00 (1.00, 1.00) Q10. I take the antibiotics according to the doctor’s instructions 10 10 10 1186 86 1.00 (1.00, 1.00) Q11. I prefer to keep antibiotics at home in case there is a need for them later 0 2 6 1181 86 2.44 (1.68, 3.46) Q12. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics 8 9 10 1183 87 0.88 (0.76, 1.04) Q13. If I believe that I need an antibiotic and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0 0 2 1180 87 1.51 (1.35, 1.72) Q14. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing antibiotics 5 7 9 1180 86 0.75 (0.57, 1.00) Q15. Doctors often explain clearly to the patient the instructions for the use of antibiotics 5 8 10 1179 86 0.66 (0.47, 0.90) Q16. When you buy antibiotics, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 3 6 9 1181 87 1.27 (0.83, 1.87) №: number of events; aIqOR: adjusted interquartile Odds Ratio; *: the outcome “any misuse” encompasses one or more of the following Practices: use without prescription, shortened treatment, modified dose, doubling a skipped dose, or taking it when remembered; **: adjusted for gender and age Narmeen Mallah Results 61 Table 13 Association of Knowledge and Attitudes with specific aspects of antibiotic misuse Practices in Lebanese parents of schoolchildren Knowledge and Attitude statements Use without prescription (№=1361) Shortened treatment (№=1376) Changed dose (№= 1293) Improper action when skipping a dose (№= 1295) Stored or shared leftovers (№= 1298) № alqOR* (95%CI) № alqOR* (95%CI) № alqOR* (95%CI) № alqOR* (95%CI) № alqOR* (95%CI) Q1. Antibiotics are effective against viruses 47 2.83 (1.61, 5.33) 23 2.66 (1.07, 6.27) 10 5.94 (1.23, 28.04) 24 1.71 (0.75, 3.58) 46 1.71 (1.00, 3.19) Q2. Each type of infection needs a different antibiotic 49 1.17 (0.86, 1.59) 23 1.28 (0.77, 2.07) 11 0.77 (0.52, 1.17) 24 1.14 (0.76, 1.77) 46 1.06 (0.81, 1.42) Q3. Antibiotics can kill the bacteria that normally live on the skin and in the gut 49 0.86 (0.53, 1.40) 22 0.86 (0.73, 1.76) 11 0.70 (0.25, 1.84) 24 1.16 (0.56, 2.49) 45 1.40 (0.82, 2.49) Q4. If antibiotics are consumed in excess, they won’t work when they are really needed 45 0.94 (0.76, 1.17) 21 0.86 (0.64, 1.17) 9 0.72 (0.50, 1.04) 22 1.08 (0.72, 1.59) 43 0.98 (0.76, 1.21) Q5. When I get a cold, I take antibiotics to help me feel better faster 49 2.29 (1.69, 3.22) 23 2.36 (1.52, 3.73) 11 4.42 (2.14, 9.38) 24 1.94 (1.26, 3.13) 46 2.22 (1.57, 3.04) Q6. I expect my doctor to prescribe antibiotics if I suffer from common cold or flu symptoms 49 1.42 (0.76, 2.77) 23 3.52 (1.30, 10.60) 11 1.20 (0.29, 4.79) 24 1.20 (0.52, 3.00) 47 1.42 (0.76, 2.77) Q7. When I have a sore throat, I prefer to use an antibiotic 49 3.30 (2.29, 5.00) 23 2.19 (1.22, 3.86) 11 4.01 (1.69, 9.54) 24 1.34 (0.73, 2.39) 47 2.10 (1.40, 3.05) Q8. If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible 47 2.49 (1.28, 4.83) 23 2.70 (1.00, 7.59) 10 1.28 (0.42, 3.86) 24 1.05 (0.56, 2.10) 46 1.16 (0.70, 1.84) Q9. It is good to be able to get antibiotics from relatives or friends without having to see a medical doctor 49 1.00 (1.00, 1.00) 23 1.00 (1.00, 1.00) 11 1.00 (1.00, 1.00) 24 1.00 (1.00, 1.00) 47 1.00 (1.00, 1.00) Q10. I take the antibiotics according to the doctor’s instructions 47 1.00 (1.00, 1.00) 23 1.00 (1.00, 1.00) 10 1.00 (1.00, 1.00) 24 1.00 (1.00, 1.00) 50 1.00 (1.00, 1.00) Q11. I prefer to keep antibiotics at home in case there is a need for them later 47 3.64 (2.19, 6.05) 23 2.08 (1.06, 4.00) 10 5.05 (1.59, 16.78) 24 1.06 (0.53, 2.08) 46 3.30 (1.97, 5.29) Q12. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics 48 0.86 (0.71, 1.08) 23 0.83 (0.61, 1.12) 11 0.59 (0.41, 0.85) 24 0.89 (0.71, 1.37) 47 0.88 (0.71, 1.10) Q13. If I believe that I need an antibiotic and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 48 1.64 (1.39, 1.90) 23 1.49 (1.19, 1.88) 11 1.64 (1.21, 2.25) 24 1.51 (1.21, 1.88) 47 1.54 (1.32, 1.82) Q14. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing antibiotics 47 0.63 (0.43, 0.92) 23 0.66 (0.39, 1.13) 11 0.60 (0.27, 1.26) 23 1.22 (0.69, 2.22) 46 0.89 (0.60, 1.36) Q15. Doctors often explain clearly to the patient the instructions for the use of antibiotics 47 0.59 (0.37, 0.90) 23 0.59 (0.31, 1.16) 11 0.44 (0.17, 1.10) 23 0.77 (0.39, 1.54) 46 0.62 (0.39, 1.00) Q16. When you buy antibiotics, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 48 2.44 (1.27, 4.61) 23 1.50 (0.65, 3.46) 11 3.81 (0.89, 16.16) 24 1.27 (0.57, 2.84) 47 1.42 (0.78, 2.57) №: number of events; aIqOR: adjusted interquartile Odds Ratio; *: adjusted for gender and age Narmeen Mallah Results 62 5.4.2. Comparision study of cross-sectional and longitudinal approaches in studies about Knowledge, Attitude and antibiotic misuse Practices in Galicia Participants՚ socio-demographic characteristics Eight hundred forty-seven out of 890 individuals completed the baseline KAP questionnaire on antibiotics, of whom 115 showed signs of antibiotic misuse Practices. Forty-three approached individuals refused to participate in the study because of lack of time. One hundred (11.81%) of the 847 participants dropout from the study; 76 did not have time to answer the follow-up questions because of work or family duties and 24 had given incorrect phone number. Therefore, data collected from the remaining 747 individuals who answered the baseline questionnaire and at least one follow-up questionnaire were included in the analysis. Table 14 represents the general demographic characteristics of the 747 individuals. Most of them were females (N= 560, 75.0%), university graduates (N= 462, 61.8%), employed (N= 561, 75.1%), and living in a household with ≤ 4 members (N= 609, 81.5%). Around half of the participants (N= 381, 51.0%) reported not always visiting the doctor in case of sickness and 40.0% (N= 299) declared ever receiving a medical prescription over the phone. Regarding those 100 individuals who dropped out from the study, their level of agreement on Knowledge and Attitude statements was not notably different from that of those who were followed-up. Besides, the two groups did not differ with respect to antibiotic misuse practices. 5.4.2.1. Cross-Sectional data analysis Seventy-eight individuals showed signs of antibiotic misuse Practices. In this study, antibiotic misuse was defined as the occurrence of one or more of the following Practices: use without prescription, shortened treatment, modified dose, doubling a skipped dose, or taking it when remembered. As shown in Table 15, low level of Knowledge and medically inappropriate Attitudes were substantially associated with antibiotic misuse Practices. Association of Knowledge about antibiotics with the misuse Practices of these drugs Individuals who highly agreed on using antibiotics to alleviate common cold symptoms had three-fold higher odds of antibiotic misuse than those who totally rejected using antibiotics for these purposes [Q2 aOR: 2.48 (95%CI: 1.48, 4.15)] (Table 15). Individuals who showed agreement on storing antibiotics at home for a potential need in the future had almost six times higher odds of antibiotic misuse than those who refused keeping antibiotics at home [Q11 aOR: 5.66 (95%CI: 3.14, 10.22)] (Table 15). Association of Attitudes towards antibiotics with the misuse Practices of these drugs Individuals with a high level of agreement on the statement about quitting antibiotic treatment earlier than prescribed had considerable higher odds of antibiotic misuse than those who completely disagreed on abandoning antibiotic treatment before completing the recommended duration [Q3 aOR: 4.44 (95%CI: 2.69, 7.32)] (Table 15). Individuals who did not show a complete disagreement with the statement about tendency to get antibiotics from relatives or friends without consulting a doctor had five times higher odds of misusing antibiotics [Q5 aOR:5.12 (95%CI: 2.83, 9.26)] than those who fully rejected this statement (Table 15). Narmeen Mallah Results 63 Individuals who preferred using antibiotics to treat sore throat infections had substantial higher odds of antibiotic misuse than those who did not prefer to use antibiotic for these conditions [Q6 aOR:2.68 (95%CI: 1.60, 4.49)] (Table 15). High odds of antibiotic misuse were observed for those who mentioned that they do not usually adhere to the prescribed treatment regimen [Q9 aOR: 2.88 (95%CI 1.09, 7.60)] (Table 15). Likewise, individuals who do not know that excessive intake of antibiotics renders them ineffective were more likely to misuse antibiotics than their counterparts [Q10 aOR:1.94 (95%CI: 1.19, 3.18)] (Table 15). Association of Attitudes towards healthcare-provider with the misuse Practices of these drugs Higher odds of antibiotic misuse were observed for individuals who would accept buying unprescribed antibiotics even when the doctor does not prescribe them [Q13 aOR: 2.47 (95%CI: 1.45, 4.20)] (Table 15). 5.4.2.2. Cohort data analysis Person-time approach using Poisson regression: Practices of antibiotic misuse was reported by 46 individuals during the follow-up assessment. Unlike the findings of cross-sectional approach, Poisson regression analysis showed associations with antibiotic misuse Practices for only three of the 16 Knowledge and Attitude statements: 1) agreement on the statement about the willingness to store antibiotics for a potential need in the future [Q11 aIRR6-10: 2.33 (95%CI: 1.20, 4.54)], 2) agreement on the statement about using antibiotics in case of cold symptoms [Q2 aIRR3-10: 2.28 (95%CI: 1.24, 4.21)] and 3) incomplete agreement on adhering to the prescribed treatment guidelines [Q9 aIRR: 3.02 (95%CI: 1.04, 8.70) (Table 15). Longitudinal approach using GLMM: The follow-up data yielded fifty events of antibiotic misuse that were included in GLMM analysis. There was no association between the Knowledge and Attitude statements and antibiotic misuse Practices, except for the statement about the willingness to store antibiotics for a potential need in the future [Q11 aOR6-10: 2.64 (1.39, 5.02)] (Table 15). Robustness analysis of the cohort approach Table 16 about MICE (missing data) analysis demonstrates that there was no substantial difference between aIRRs computed from data collected in this study and those calculated from imputed data. In addition, MICE confirmed that Knowledge and Attitudes were not associated with Practices of misuse of antibiotics. The two sensitivity analyses under extreme assumptions, i.e., 1) all individuals who dropped out had misused antibiotics and 2) all individuals who dropped out had not misused antibiotics, yielded similar findings to that of MICE analysis (Table 17). Narmeen Mallah Results 70 Table 17 IRRs and 95%CI from the two sensitivity analyses of the longitudinal data collected in Galicia about misuse Practices of antibiotics Item Lost individuals considered as non cases (№= 847) Lost individuals considered as cases (№= 847) Misuse aIRR (95% CI) Misuse aIRR (95% CI) No Yes No Yes Q1. Antibiotics are effective against viruses 0 354 17 1.00 15 356 1.00 1-2 46 3 1.45 (0.42, 5.00) 3 46 1.02 (0.74, 1.39) 3-8 256 15 1.25 (0.62, 2.51) 9 262 1.10 (0.93, 1.29) 9-10 135 11 1.70 (0.79, 3.65) 7 139 1.08 (0.88, 1.32) Q2. When I get a cold, I take antibiotics to help me feel better faster 0 553 25 1.00 24 554 1.00 1-2 66 3 0.95 (0.29, 3.15) 1 68 1.02 (0.79, 1.32) 3-10 179 18 2.19 (1.19, 4.04) 9 188 1.12 (0.94, 1.33) Q3. If I feel better after a few days, I sometimes stop taking my antibiotics before completing the course of treatment 0 570 30 1.00 23 577 1.00 1 40 1 0.45 (0.62, 3.33) 2 39 0.97 (0.70, 1.35) 2-10 185 15 1.64 (0.87, 3.09) 9 191 1.08 (0.91, 1.28) Q4. I expect my doctor to prescribe antibiotics if I suffer from common cold or flu symptoms 0 375 18 1.00 15 378 1.00 1-2 47 2 0.86 (0.20, 3.72) 2 47 1.00 (0.73, 1.35) 3-10 374 26 1.43 (0.78, 2.62) 17 383 1.00 (0.96, 1.15) Q5. It is good to be able to get antibiotics from relatives or friends without having to see a medical doctor 0 725 42 1.00 33 734 1.00 1-10 73 4 0.94 (0.34, 2.63) 1 76 1.14 (0.90, 1.46) Q6. When I have a sore throat, I prefer to use an antibiotic 0 540 35 1.00 24 551 1.00 1-2 67 0 N.A. 2 65 1.07 (0.82, 1.40) 3-10 188 11 0.90 (0.46, 1.80) 8 191 1.12 (0.94, 1.33) Q7. Each type of infection needs a different antibiotic 0-5 220 12 0.86 (0.43, 1.72) 6 226 1.04 (0.88, 1.23) 6-8 190 9 0.76 (0.35, 1.64) 10 189 0.99 (0.83, 1.19) 9-10 383 24 1.00 18 389 1.00 Q8. If I feel side effects during a course of treatment of antibiotics, I should stop taking them as soon as possible 0-4 311 16 1.00 16 311 1.00 5 96 8 1.58 (0.68, 3.69) 5 99 0.90 (0.72, 1.14) 6-10 376 21 1.06 (0.55, 2.04) 13 384 0.99 (0.85, 1.15) Narmeen Mallah Results 71 Item Lost individuals considered as non cases (№= 847) Lost individuals considered as cases (№= 847) Misuse aIRR (95% CI) Misuse aIRR (95% CI) No Yes No Yes Q9. I take the antibiotics according to the doctor’s instructions 0-6 24 4 3.07 (1.08, 8.72) 3 25 0.99 (0.66, 1.48) 7-9 149 6 0.66 (0.28, 1.57) 6 149 1.03 (0.86, 1.24) 10 615 36 1.00 25 626 1.00 Q10. If antibiotics are consumed in excess, they will not work when they are really needed 1-9 254 15 1.00 (0.53, 1.86) 14 255 0.99 (0.85, 1.15) 10 530 31 1.00 20 541 1.00 Q11. I prefer to keep antibiotics at home in case there is a need for them later 0 479 23 1.00 19 483 1.00 1-5 177 9 1.02 (0.47, 2.21) 7 179 1.01 (0.85, 1.21) 6-10 131 14 2.39 (1.25, 4.65) 8 137 1.09 (0.90, 1.33) Q12. I trust the doctor’s decision if s/he decides to prescribe or not prescribe antibiotics 0-8 207 14 1.27 (0.67, 2.40) 13 208 0.97 (0.83, 1.15) 9-10 569 31 1.00 21 579 1.00 Q13. If I believe that I need an antibiotic and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0 652 37 1.00 30 659 1.00 1-10 134 9 1.23 (0.59, 2.56) 4 139 1.10 (0.92, 1.33) Q14. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing antibiotics 0-5 356 15 0.70 (0.32, 1.57) 17 354 0.99 (0.83, 1.20) 6 63 4 1.09 (0.34, 3.50) 2 65 1.06 (0.79, 1.42) 7-9 192 16 1.47 (0.66, 3.24) 6 202 1.03 (0.84, 1.27) 10 172 10 1.00 9 173 1.00 Q15. Doctors often explain clearly to the patient the instructions for the use of antibiotics 0-5 226 12 0.73 (0.36, 1.48) 8 230 0.96 (0.81, 1.14) 6-8 248 12 0.68 (0.34, 1.37) 8 252 0.97 (0.82, 1.15) 9-10 307 22 1.00 16 313 1.00 Q16. When you buy antibiotics, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 0-3 220 11 0.77 (0.33, 1.78) 13 218 0.96 (0.83, 1.24) 4-6 182 12 0.99 (0.43, 2.62) 6 188 1.03 (0.84, 1.28) 7-9 212 12 0.88 (0.39, 2.01) 6 218 1.01 (0.83, 1.24) 10 173 11 1.00 9 175 1.00 aIRR: Incidence Rate Ratio adjusted for age and gender; CI: Confidence Interval; №: Number of participants. Misuse is defined as the occurrence of any of the following Practices: use without prescription, shortened treatment, modified dose, doubling a skipped dose, or taking it when remembered. Narmeen Mallah Results 72 5.5. Association of Knowledge and Attitudes with Practices of tranquilizer misuse The two KAP studies about tranquilizers in Galicia and Lebanon have been published [Psychology and Health 2021 (in print) & Drug and Alcohol Dependence 2021 (in print)]. The corresponding publication of Drug and alcohol dependence is available in Annex II page 234. The article of Psychology and Health is not reproduced. Another manuscript about the comparison of cross-sectional and longitudinal approaches is currently under review by Journal of Clinical Medicine. 5.5.1. Cross-sectional study in Lebanon Participants՚ socio-demographic characteristics One thousand three-hundred ninety-six individuals completed the questionnaire. The sociodemographic characteristics of the study population is represented in Table 18. Ninety-one (6.5%) participants reported using tranquilizers in the month preceding the study, and 63 (62.2%) of them presented one or more patterns of misuse. Most individuals who misused tranquilizers were females (N=51, 81%), married (N= 57, 90.5%) and had a professionally active spouse (N=55, 87.3%). Three quarters of tranquilizer misusers reported not always visiting a doctor in case of illness (N=47, 74.6%), two thirds were university graduates (N=52, 66.7%), a similar fraction declared ever receiving a phone medical prescription (N=40, 63.5%) and never consuming alcohol (N=39, 61.9%). Almost half of the misusers had upper-intermediate household income (N= 30, 47.6%). Patterns of tranquilizer misuse The major pattern of misuse was storing or sharing tranquilizer leftover (N=46, 73%), followed by use of tranquilizers without prescription (N=28, 44.4%), curtailing tranquilizer treatment course (N=26, 41.3%), changing the dose of tranquilizer without medical advice (N=18, 28.6%), and doubling the dose or taking it when remembered in case of forgetting to take a previous dose (N=12, 19.0%). Association of knowledge about tranquilizers with the misuse Practices of these drugs Awareness about the consequences of tranquilizers was associated with lower odds of misuse of these drugs (Table 19). Substantial lower odds of tranquilizer misuse were observed for individuals who knew that tranquilizers: reduce people’s control over their actions had [Q5 aIqOR: 0.50 (95%CI: 0.33, 0.77)], could increase the risk of traffic accidents [Q6 aIqOR: 0.62 (95%CI: 0.37, 1.00)] or affect children’s learning abilities [Q7 aIqOR: 0.69 (95%CI: 0.45, 1.00)] (Table 19). Narmeen Mallah Results 73 Association of attitudes towards tranquilizers with the misuse Practices of these drugs Substantial associations were observed between medically in appropriate and misuse of tranquilizers. There were five-fold higher odds of misusing tranquilizers by individuals who concur on storing tranquilizers for potential need in the future as compared to those who don´t keep these medicines [Q11 aIqOR: 5.00 (95%CI: 3.30, 7.59)] (Table 19). Higher odds were also observed for these specific misuse patterns: curtailing tranquilizers treatment course [Q11 aIqOR: 8.66 (95%CI: 3.71, 19.97)], unprescribed use of tranquilizers [Q11 aIqOR: 5.77 (95%CI: 3.18, 10.49)], and storing or sharing tranquilizer leftovers [Q11 aIqOR: 5.77 (95%CI: 3.57, 9.24)] (Table 20). Higher odds of any misuse pattern of tranquilizers were also obtained for individuals who would agree on taking tranquilizers to improve sleeping, work better or enjoy themselves [Q1 aIqORsleep: 2.35 (95%CI: 1.82, 2.99), Q4 aIqORwork: 1.99 (95%CI: 1.69, 2.31) and Q3 aIqORenjoy: 1.30 (95%CI: 1.20, 1.41)] (Table 19). As for the specific patterns of misuse, these attitudes were associated with higher odds of using unprescribed tranquilizers (1.26 ≤ aIqOR ≤ 2.10), truncating the tranquilizer treatment course (1.34 ≤ aIqOR ≤ 2.00), and storing or sharing tranquilizer leftover (1.31 ≤ aIqOR ≤ 2.41) (Table 20). Individuals who would accept prolonging the course of treatment with tranquilizers beyond the prescribed duration were more likely to misuse tranquilizers in general [Q2 aIqOR: 1.26 (95%CI: 1.16, 1.36)], and to store the leftover tranquilizers [Q2 aIqOR: 1.25 (95%CI: 1.15, 1.37)] in specific (Table 19 and Table 20). Similar findings were observed for individuals who expressed their willingness to get tranquilizers without a prescription if they believe that they need these drugs, but the doctor did not prescribe them (Table 19 and Table 20). On the contrary, lower odds of tranquilizer misuse were found for individuals who declared their willingness to follow the physician’s instructions (Q9). They were less likely to take tranquilizers without prescription [Q9 aIqOR: 0.72 (95%CI: 0.59, 0.90)], or to cut-down the course of treatment [Q9 aIqOR: 0.69 (95%CI: 0.53, 0.86)] (Table 20). Association of attitudes toward healthcare provider with the misuse Practices of tranquilizers Individuals who thought the physician clearly explains the reasons of tranquilizers prescription or non-prescription were more likely to shorten the course of treatment [Q14 aIqOR: 3.46 (95%CI: 1.19, 9.69)] (Table 20). Narmeen Mallah Results 74 Table 18 Demographic characteristics of individuals participated in the cross-sectional study about the association of Knowledge and Attitudes with Practices of tranquilizer misuse in Lebanon №: Number of participants Characteristic Total (№= 1396) Any misuse in the last month (№= 63) Gender Male 271 (19.4%) 8 (12.7%) Female 1075 (77.0%) 51 (81.0%) Missing 50 (3.6) 4 (6.3%) Age <38 years 348 (24.9%) 20 (31.7%) 39 -42 years 225 (16.1%) 8 (12.7%) 43 - 47 years 304 (21.8%) 10 (15.9%) >= 48 years 257 (18.4%) 12 (19.0) Missing 262 (18.8%) 13 (20.6%) Marital status Married 1307 (93.6%) 57 (90.5%) Others 72 (5.2%) 6 (9.5%) Missing 17 (1.2%) 0 Educational level Until high school 242 (17.3%) 21 (38.1%) University 1144 (81.9%) 42 (66.7%) Missing 10 (0.7%) 0 Spouse educational level Until high school 301 (21.6%) 24 (38.1%) University 1020 (73.1%) 35 (55.6%) Missing 75 (5.4%) 4 (6.3%) Family size 2-4 543 (38.9%) 28 (44.4%) 5-6 706 (50.6%) 27 (42.9%) >6 127 (9.1%) 8 (12.7%) Missing 20 (1.4%) 0 Monthly Family income <500 US$ 19 (1.4%) 0 500 US$ - 1499 US$ 253 (18.1%) 17 (27.0%) 1500 US$ - 2500 US$ 218 (15.6%) 15 (23.8%) >2500 US$ 802 (57.4%) 30 (47.6%) Missing 104 (7.4%) 1 (1.6%) Consulting a doctor Rarely or never 151 (10.8%) 5 (7.9%) Sometimes 743 (53.2%) 42 (66.7%) Always 490 (35.1%) 16 (25.4%) Missing 12 (0.9%) 0 Reasons for not consulting a doctor No need 386 (27.7) 17 (27.0%) Fear 28 (2.0%) 1 (1.6%) No money 123 (8.8%) 9 (14.3%) No time 317 (22.7%) 24 (38.1%) Long waiting time 164 (11.7%) 12 (19.0%) No near clinic 52 (3.7%) 5 (7.9%) Medical consultation over the phone No 478 (34.2%) 22 (34.9%) Yes 887 (63.5%) 40 (63.5%) Missing 21 (2.2%) 1 (1.6%) Employment status Employed 1039 (74.4%) 36 (57.1%) Unemployed 336 (24.1%) 27 (42.9%) Missing 21 (1.5%) 0 Spouse Employment Status Employed 1164 (83.4%) 55 (87.3%) Unemployed 150 (10.7%) 3 (4.8%) Missing 82 (5.9%) 5 (7.9%) Alcohol Intake Never 625 (44.8%) 39 (61.9%) In special occasions 421 (30.2%) 14 (22.2%) Others 319 (22.9%) 10 (15.9%) Missing 31 (2.2%) 0 Narmeen Mallah Results 75 Table 19 Association of Knowledge and Attitudes with any pattern of misuse Practices of tranquilizers by parents of schoolchildren in Lebanon Knowledge and Attitude Statements Q1 Q2 Q3 No misuse № Misuse № aOR* (95%CI) aIqOR (95%CI) Q1. I would agree to take tranquilizers in order to sleep better 0 0 3 1147 49 1.33 (1.22, 1.44) 2.35 (1.82, 2.99) Q2. If I feel better after a few days, I sometimes keep taking my tranquilizers even after completing the prescribed course of treatment 0 0 1 1137 49 1.26 (1.16, 1.36) 1.26 (1.16, 1.36) Q3. I would take tranquilizers in order to enjoy myself with my family 0 0 1 1137 49 1.30 (1.20, 1.41) 1.30 (1.20, 1.41) Q4. I would agree to take tranquilizers when I feel down and sad in order to work better 0 0 2 1140 50 1.41 (1.30, 1.52) 1.99 (1.69, 2.31) Q5. Tranquilizers reduce people’s control over what they do 5 8 10 1131 50 0.87 (0.80, 0.95) 0.50 (0.33, 0.77) Q6. People taking tranquilizers are at increased risk of traffic accidents 5 8 10 1131 49 0.91 (0.82, 1.00) 0.62 (0.37, 1.00) Q7. Psychotropic drugs (such as tranquilizers) may affect children’s learning abilities when prescribed to them 6 8 10 1121 50 0.91 (0.82, 1.00) 0.69 (0.45, 1.00) Q8. If I feel side effects during a course of treatment of tranquilizers, I should stop taking it as soon as possible 6 9 10 1123 49 1.02 (0.93, 1.13) 1.08 (0.75, 1.63) Q9. I take the tranquilizers according to the doctor’s instructions 8 10 10 1116 48 0.93 (0.84, 1.02) 0.86 (0.71, 1.04) Q10. If tranquilizers are consumed in excess, they won’t work when they are really needed 8 10 10 1115 48 0.96 (0.87, 1.07) 0.92 (0.76, 1.14) Q11. I prefer to keep tranquilizers at home in case there is a need for them later 0 0 5 1131 48 1.38 (1.27, 1.50) 5.00 (3.30, 7.59) Q12. I trust the doctor’s decision if s/he decides to prescribe or not prescribe tranquilizers 5 8 10 1124 49 1.07 (0.96, 1.18) 1.40 (0.82, 2.29) Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0 0 1 1131 49 1.30 (1.20, 1.41) 1.30 (1.20, 1.41) Q14. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing tranquilizers 2 6 8 1123 48 1.04 (0.95, 1.13) 1.27 (0.74, 2.08) Q15. Doctors often explain clearly to the patient the instructions for the use of tranquilizers 5 7 9 1116 48 0.98 (0.89, 1.08) 0.92 (0.63, 1.36) Q16. When you buy tranquilizers, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 4 6 9 1116 49 1.06 (0.96, 1.17) 1.34 (0.82, 2.19) aOR: adjusted odds ratio; aIqOR: adjusted interquartile OR; Q1: 1st quartile; Q2: 2nd quartile; Q3: 3rd quartile; №: Number of participants; *: OR adjusted for age, gender and spouse employment status Narmeen Mallah Results 76 Table 20 Association of Knowledge and Attitudes with specific patterns of misuse Practices of tranquilizers by parents of schoolchildren in Lebanon aOR: adjusted odds ratio; aIqOR: adjusted interquartile OR; №: Number of participants; *: OR adjusted for age, gender and spouse employment Knowledge and Attitude Statements Unprescribed use Shortening treatment Storing leftover Misuse № aOR* (95%CI) aIqOR (95%CI) Misuse № aOR* (95%CI) aIqOR (95%CI) Misuse № aOR* (95%CI) aIqOR (95%CI) No Yes No Yes No Yes Q1. I would agree to take tranquilizers in order to sleep better 1158 24 1.28 (1.14, 1.43) 2.10 (1.48, 2.92) 1167 17 1.26 (1.11, 1.44) 2.00 (1.37, 2.99) 1130 41 1.34 (1.22, 1.46) 2.41 (1.82, 3.11) Q2. If I feel better after a few days, I sometimes keep taking my tranquilizers even after completing the prescribed course of treatment 1148 24 1.1 (0.96, 1.25) 1.10 (0.96, 1.25) 1157 17 1.15 (0.99, 1.33) 1.15 (0.99, 1.33) 1120 41 1.25 (1.15, 1.37) 1.25 (1.15, 1.37) Q3. I would take tranquilizers in order to enjoy myself with my family 1148 24 1.26 (1.13, 1.41) 1.26 (1.13, 1.41) 1157 17 1.34 (1.18, 1.53) 1.34 (1.18, 1.53) 1121 41 1.31 (1.20, 1.43) 1.31 (1.20, 1.43) Q4. I would agree to take tranquilizers when I feel down and sad in order to work better 1152 24 1.34 (1.2, 1.49) 1.80 (1.44, 2.22) 1160 17 1.40 (1.23, 1.6) 1.96 (1.51, 2.56) 1126 42 1.41 (1.29, 1.55) 1.99 (1.66, 2.40) Q5. Tranquilizers reduce people’s control over what they do 1143 24 0.87 (0.77, 0.98) 0.50 (0.27, 0.90) 1151 17 0.90 (0.78, 1.03) 0.59 (0.29, 1.16) 1116 42 0.88 (0.81, 0.96) 0.53 (0.35, 0.82) Q6. People taking tranquilizers are at increased risk of traffic accidents 1142 24 0.88 (0.77, 1.00) 0.53 (0.27, 1.00) 1150 17 0.98 (0.82, 1.17) 0.90 (0.37, 2.19) 1116 41 0.91 (0.82,1.01) 0.62 (0.37, 1.05) Q7. Psychotropic drugs (such as tranquilizers) may affect children’s learning abilities when prescribed to them 1133 24 0.91 (0.79, 1.05) 0.69 (0.39, 1.22) 1141 17 1.10 (0.88, 1.37) 1.46 (0.60, 3.52) 1106 42 0.91 (0.82, 1.02) 0.69 (0.45, 1.08) Q8. If I feel side effects during a course of treatment of tranquilizers, I should stop taking it as soon as possible 1134 24 1.09 (0.93, 1.28) 1.41 (0.75, 2.68) 1142 17 1.21 (0.94, 1.55) 2.14 (0.78, 5.77) 1107 41 1.03 (0.92, 1.15) 1.13 (0.72, 1.75) Q9. I take the tranquilizers according to the doctor’s instructions 1126 24 0.85 (0.77, 0.95) 0.72 (0.59, 0.90) 1134 17 0.83 (0.73, 0.93) 0.69 (0.53, 0.86) 1100 40 0.95 (0.85, 1.06) 0.90 (0.72, 1.12) Q10. If tranquilizers are consumed in excess, they won’t work when they are really needed 1125 24 0.97 (0.84, 1.12) 0.94 (0.71, 1.25) 1133 17 1.28 (0.92, 1.77) 1.64 (0.85, 3.13) 1099 40 0.97 (0.87, 1.10) 0.94 (0.76, 1.21) Q11. I prefer to keep tranquilizers at home in case there is a need for them later 1141 24 1.42 (1.26, 1.6) 5.77 (3.18, 10.49) 1149 17 1.54 (1.30, 1.82) 8.66 (3.71, 19.97) 1115 40 1.42 (1.29, 1.56) 5.77 (3.57, 9.24) Q12. I trust the doctor’s decision if s/he decides to prescribe or not prescribe tranquilizers 1135 24 0.96 (0.85, 1.09) 0.82 (0.44, 1.54) 1143 17 1.04 (0.88, 1.22) 1.22 (0.53, 2.70) 1109 41 1.12 (0.99, 1.26) 1.76 (0.95, 3.18) Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 1142 24 1.41 (1.27, 1.56) 1.41 (1.27, 1.56) 1150 17 1.53 (1.35, 1.74) 1.53 (1.35, 1.74) 1115 41 1.26 (1.15, 1.38) 1.26 (1.15, 1.38) Q14. Doctors often explain clearly to the patient the reasons for prescribing or not prescribing tranquilizers 1134 23 1.09 (0.97, 1.24) 1.68 (0.83, 3.64) 1142 16 1.23 (1.03, 1.46) 3.46 (1.19, 9.69) 1107 40 1.04 (0.94, 1.14) 1.27 (0.69, 2.19) Q15. Doctors often explain clearly to the patient the instructions for the use of tranquilizers 1127 23 0.95 (0.83, 1.09) 0.81 (0.47, 1.41) 1135 16 1.05 (0.88, 1.26) 1.22 (0.60, 2.52) 1100 40 1.00 (0.89, 1.11) 1.00 (0.63, 1.52) Q16. When you buy tranquilizers, the pharmacist tells you about the importance of correct therapeutic compliance/adherence 1127 24 1.11 (0.96, 1.28) 1.69 (0.82, 3.44) 1135 17 1.11 (0.94, 1.32) 1.69 (0.73, 4.01) 1100 41 1.05 (0.93, 1.16) 1.28 (0.70, 2.10) Narmeen Mallah Results 77 5.5.2. Cohort study in Spain Participants՚ socio-demographic characteristics Seven hundred forty-seven (88.19%) out of 847 recruited participants were included in the main analysis. Table 21 represents the demographic characteristics of the participants. Most of the participants were females (N=506, 74.97%), employed (N = 561, 75.10%) and living in a household of a maximum of four members (N= 609, 81.53%). More than half of them were university graduates (N= 462, 61.85%), and had never consumed alcohol (N= 421, 56.36%). Around half of the participants had between 36 and 45 years (N=373, 49.93%) and a similar fraction declared not always consulting the doctor in case of illness (N=381, 51.00%). Two hundred ninety-nine (40.02%) participants reported ever receiving a phone medical prescription over the phone. Fifty-eight tranquilizer misuse events were identified during the follow-up producing an overall incidence rate of 0.17 year-1. Association of knowledge about tranquilizers with the misuse Practices of these drugs The level of knowledge about the consequences of tranquilizer was not associated with the misuse Practices of these drugs (Table 22). Individuals who were not aware that tranquilizers decrease people’s control over their actions [Q5 1st tertile aIRR: 1.29 (95%CI: 0.65, 2.59)], influence the learning capacity of children [Q7 1st tertile: aIRR: 1.17 (95%CI: 0.58, 2.33)] or become ineffective when taken in excess [Q10 aIRR: 1.31 (95%CI: 0.74, 2.32)] were not at a higher risk of misusing tranquilizers as compared to those who had a better level of knowledge (Table 22). Association of attitudes towards tranquilizers with the misuse Practices of these drugs There were strong associations between medically inappropriate attitudes towards tranquilizers with the misuse Practices of these drugs (Table 22). Individuals who showed agreement on statements about using tranquilizers to sleep or work better are at considerable higher risk of tranquilizer misuse than those who disagreed on taking tranquilizers for these purposes [Q1 3rd tertile: aIRR: 5.10 (95%CI: 2.74, 9.48) and Q4 aIRR: 2.04 (95%CI: 1.05, 3.99)] (Table 22). Individuals who did not totally disagree on extending tranquilizers treatment course without medical advice had two-fold higher risk of misusing these drugs than those who totally disagreed on this statement [Q2 aIRR: 2.45 (95%CI: 1.46, 4.13)] (Table 22). Individuals who do not completely refuse the intake of tranquilizers for recreational purposes are at increased risk of misusing these drugs as compared to those who totally disagreed on using tranquilizers for this motive [Q3 2nd tertile: aIRR: 1.85 (95%CI: 1.04, 3.32)] (Table 22). Individuals who showed a preference to storing tranquilizers at home for a future need had five folds higher risk than those who totally disagree on storing tranquilizers [Q11 aIRR: 5.07 (95%CI: 2.73, 9.40)] (Table 22). Narmeen Mallah Results 78 Association of attitudes toward healthcare provider with the misuse Practices of tranquilizers There was an association between negative patient-healthcare provider relationship and misuse of tranquilizers. Individuals who declared incompletely trusting the physician’s decision about tranquilizers prescription had double the risk of misusing these drugs than those who fully trust their physician [Q12 aIRR: 1.92 (95%CI: 1.12, 3.30)] (Table 22). Individuals who thought that the physician do not always explain to the patient the reasons of prescribing/non-prescribing tranquilizers with enough clarity are at increased risk of tranquilizer misuse [Q14 2nd tertile: aIRR: 1.98 (95%CI: 0.97, 4.06)] (Table 22). Robustness Analysis We re-estimated the aIRR under two extreme assumptions; 1) when supposing that all individuals who dropped had not experienced the event (tranquilizer misuse) similar results to those of the main analysis were obtained, but 2) when assuming that all individuals who dropped out had misused tranquilizers, the magnitude of the association was meaningfully smaller. As shown in Table 22, MICE analysis confirmed the association of Knowledge and Attitudes with Practices of tranquilizer misuse, but the magnitude of association was smaller. The greatest differences were seen for aIRR estimates of Q1 (from 5.10, 95%CI: 2.74-9.48 to 2.16, 95%CI: 1.46-3.19), Q2 (from 2.45, 95%CI: 1.46-4.13 to 1.58, 95%CI: 1.11-2.26) and Q11 (from 5.07, 95%CI: 2.73-9.40 to 1.70, 95%CI: 1.13-2.56). Following the correction for non-differential exposure misclassification using the sensitivity and specificity estimates, a higher magnitude of ORs was observed, particularly high for Q2, Q3, Q6, Q13, Q14 and Q16 (data not shown). The results we observed were then conservative. Narmeen Mallah Results 79 Table 21 General characteristic of individuals who participated in the study about the association of Knowledge and Attitudes with Practices of tranquilizer misuse in Galicia №: Number of participants Characteristic Total (№= 747) Tranquilizer Misusers (№= 58) Sex Male 187 (25.03%) 8 (13.79%) Female 560 (74.97%) 50 (86.21%) Missing 0 0 Age <35 years 131 (17.54%) 8 (13.79%) 36 - 45 years 373 (49.93%) 27 (46.55%) >= 46 years 243 (32.53%) 23 (39.66%) Missing 0 0 Educational level Until high school 258 (34.54%) 22 (37.93%) University 462 (61.85%) 36 (62.07%) Missing 27 (3.61%) 0 Family size ≤4 609 (81.53%) 52 (89.66%) >4 111 (14.86%) 6 (10.34%) Missing 27 (3.61%) 0 Consulting a doctor Not always 381 (51.00%) 37 (63.79%) Always 338 (45.25%) 21 (36.21%) Missing 28 (3.75%) 0 Medical consultation over the phone No 419 (56.09%) 29 (50.00%) Yes 299 (40.03%) 28 (48.28%) Missing 29 (3.88%) 1 (1.72%) Employment status Employed 561 (75.10%) 39 (67.24%) Unemployed 160 (21.42%) 19 (32.76) Missing 26 (3.48%) 0 Alcohol Intake Never/less than once per month 421 (56.36%) 35 (60.34%) Others 298 (39.89%) 23 (39.66%) Missing 28 (3.75%) 0 Narmeen Mallah Results 86 Knowledge or Attitude Statement, level of agreement Cross-sectional approach (№= 847) Longitudinal approach (№= 1343) No misuse (№) Misuse (№) aOR (95%CI)* No misuse (№) Misuse (№) aOR (95%CI)** Q10. If tranquilizers are consumed in excess, they won’t work when they are really needed 0-6 216 15 0.66 (0.35 – 1.25) 328 40 1.49 (0.73 - 3.03) 7-9 147 17 1.21 (0.66 – 2.23) 224 27 1.51 (0.66 - 3.44) 10 380 41 1.00 627 51 1.00 Q11. I prefer to keep tranquilizers at home in case there was a need for them later 0-3 485 23 1.00 780 36 1.00 4-8 202 27 3.00 (1.66 – 5.43) 296 44 2.7 (1.28 - 5.70) 9-10 62 23 6.75 (3.49–13.08) 102 41 9.49 (4.05 - 22.25) Q12. I will trust the doctor’s decision if s/he decided to prescribe or not prescribe tranquilizers 0-6 166 19 1.35 (0.75 – 2.46) 268 30 1.35 (0.61 - 3.01) 7-9 202 18 1.05 (0.57 – 1.93) 302 43 2.23 (1.08 - 4.6) 10 387 37 1.00 626 48 1.00 Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0 579 47 1.00 828 84 1.00 1-5 138 14 1.31 (0.68 – 2.52) 207 13 0.65 (0.25 - 1.66) 6-10 36 12 4.20 (1.99 – 8.87) 57 23 5.5 (2.03 - 14.87) Q14. I think that doctors often explain clearly to the patients the reasons for prescribing or not prescribing tranquilizers 0-5 334 28 0.93 (0.52 – 1.67) 516 50 1.04 (0.5 - 2.16) 6-8 175 22 1.45 (0.78 – 2.71) 254 34 1.57 (0.7 - 3.54) 9-10 242 24 1.00 418 37 1.00 Q15. I think that doctors often explain clearly to the patients the instructions for the use of tranquilizers 0-5 291 27 1.19 (0.66 – 2.14) 453 42 0.95 (0.46 - 1.96) 6-8 181 21 1.50 (0.80 – 2.81) 261 34 1.49 (0.68 - 3.25) 910 280 24 1.00 473 45 1.00 Q16. I think that, when dispensing tranquilizers, the pharmacists tell the customers about the importance of correct therapeutic compliance/adherence 0-4 182 24 0.89 (0.48 – 1.65) 298 43 1.24 (0.55 - 2.77) 5-8 388 24 0.46 (0.25 – 0.83) 578 42 0.59 (0.27 - 1.28) 9-10 178 26 1.00 298 36 1.00 CI: Confidence interval; №: Number of Participants; OR: Odds Ratio; *: OR adjusted for age, gender, alcohol intake and medical consultation over the phone; **: OR adjusted for age, gender, employment status and frequency of doctor consultation in case of illness. Misuse is defined as the occurrence of any of the following Practices: use without prescription, shortened treatment, modified dose, doubling a skipped dose, or taking it when remembered. Narmeen Mallah Results 87 Table 24 Estimates of the association of Knowledge and Attitudes of the Galician population with specific aspects of misuse Practices of tranquilizers using crosssectional and longitudinal approaches Knowledge or Attitude Statement Cross-sectional approach (№= 847) Longitudinal approach (№= 1343) Shortened treatment Stored leftover Changed dose Stored leftover Changed dose Level of agreement № aOR (95%CI)* № aOR (95%CI)* № aOR (95%CI)* № aOR (95%CI)** № aOR (95%CI)** Q1. I would agree to take tranquilizers in order to sleep better 0-5 7 1.00 15 1.00 14 1.00 20 1.00 7 1.00 6-8 9 3.93 (1.41, 10.95) 15 3.29 (1.55, 6.97) 7 1.52 (0.59, 3.90) 19 3.36 (1.24, 9.13) 7 4.31 (1.03, 17.93) 9-10 8 5.93 (2.04, 17.19) 17 5.84 (2.76, 12.36) 13 4.13 (1.81, 9.45) 18 4.55 (1.51, 13.74) 25 16.98 (4.67, 61.77) Q2. If I feel better after a few days, I will keep taking my tranquilizers even after completing the prescribed course of treatment 0 10 1.00 22 1.00 13 1.00 27 1.00 19 1.00 1-5 8 2.42 (0.94, 6.28) 15 2.19 (1.10, 4.36) 14 3.47 (1.59, 7.59) 20 2.66 (1.04, 6.78) 9 1.3 (0.39, 4.32) 6-10 6 5.78 (2.01, 16.69) 10 4.54 (2.02, 10.22) 7 4.33 (1.57, 11.95) 10 3.59 (0.98, 13.22) 11 4.93 (1.35, 17.94) Q3. I would take tranquilizers in order to enjoy myself with my family 0 11 1.00 19 1.00 13 1.00 30 1.00 18 1.00 1-5 6 1.97 (0.71, 5.47) 17 1.92 (0.97, 3.80) 10 1.59 (0.68, 3.70) 10 0.88 (0.31, 2.5) 11 1.1 (0.35, 3.44) 6-10 7 1.34 (0.51, 3.55) 11 2.48 (1.13, 5.44) 11 3.27 (1.37, 7.76) 17 2.63 (0.93, 7.45) 10 1.8 (0.52, 6.16) Q4. I would agree to take tranquilizers when I feel down and sad in order to work better 0 9 1.00 15 1.00 11 1.00 21 1.00 22 1.00 1-5 8 2.25 (0.85, 5.96) 14 1.37 (0.65, 2.91) 12 1.62 (0.70, 3.74) 13 1.16 (0.42, 3.21) 7 0.52 (0.16, 1.74) 6-10 6 1.02 (0.36, 2.91) 17 3.45 (1.65, 7.21) 10 2.33 (0.93, 5.82) 23 3.54 (1.27, 9.85) 10 1.22 (0.36, 4.09) Q5. Tranquilizers reduce people’s control over what they do 0-4 5 0.84 (0.28, 2.52) 12 1.11 (0.52, 2.37) 11 1.85 (0.78, 4.42) 26 3.24 (1.12, 9.38) 16 2.4 (0.71, 8.05) 5-7 9 0.84 (0.33, 2.11) 16 0.81 (0.41, 1.62) 12 1.00 (0.42, 2.36) 16 0.97 (0.33, 2.89) 14 1.24 (0.37, 4.17) 8-10 10 1.00 19 1.00 11 1.00 14 1.00 9 1.00 Q6. People taking tranquilizers are at increased risk of traffic accidents 0-5 6 0.77 (0.26, 2.28) 18 1.39 (0.67, 2.89) 12 0.98 (0.42, 2.29) 20 1.43 (0.5, 4.12) 9 0.98 (0.28, 3.4) 6-9 10 1.03 (0.40, 2.68) 15 0.90 (0.42, 1.91) 9 0.62 (0.26, 1.51) 20 1.13 (0.39, 3.24) 17 1.32 (0.42, 4.15) 10 8 1.00 14 1.00 12 1.00 16 1.00 13 1.00 Q7. Psychotropic drugs (such as tranquilizers) may affect children’s learning abilities when prescribed to them 0-5 4 2.38 (0.71, 7.92) 8 2.00 (0.85, 4.72) 13 0.90 (0.39, 2.04) 30 1.17 (0.47, 2.92) 18 1.14 (0.35, 3.66) 6-8 10 1.78 (0.73, 4.36) 17 1.41 (0.73, 2.73) 8 0.96 (0.38, 2.45) 4 0.26 (0.06, 1.21) 12 1.51 (0.43, 5.31) 9-10 10 1.00 22 1.00 12 1.00 20 1.00 9 1.00 Q8. If I feel side effects during a course of treatment of tranquilizers, I should stop taking it as soon as possible 0-5 9 0.46 (0.18, 1.16) 16 0.55 (0.27, 1.12) 14 0.83 (0.37, 1.87) 18 0.53 (0.2, 1.42) 13 0.69 (0.23, 2.11) 6-9 5 0.61 (0.20, 1.80) 13 1.09 (0.51, 2.32) 9 1.09 (0.42, 2.80) 18 1.27 (0.44, 3.65) 10 1.25 (0.35, 4.4) 10 11 1.00 18 1.00 11 1.00 21 1.00 16 1.00 Q9. I would take the tranquilizers according to the doctor’s instructions 0-7 6 2.31 (0.90, 5.91) 10 1.11 (0.52, 2.35) 9 1.57 (0.69, 3.57) 17 2.03 (0.74, 5.55) 4 0.56 (0.13, 2.42) 8-9 7 2.22 (0.74, 6.60) 8 1.45 (0.64, 3.32) 6 1.34 (0.48, 3.73) 15 3.45 (1.17, 10.15) 9 2.17 (0.66, 7.14) 10 11 1.00 29 1.00 19 1.00 25 1.00 26 1.00 Narmeen Mallah Results 88 Knowledge or Attitude Statement Cross-sectional approach (№= 847) Longitudinal approach (№= 1343) Shortened treatment Stored leftover Changed dose Stored leftover Changed dose Level of agreement № aOR (95%CI)* № aOR (95%CI)* № aOR (95%CI)* № aOR (95%CI)** № aOR (95%CI)** Q10. If tranquilizers are consumed in excess, they won’t work when they are really needed 0-6 5 0.58 (0.21, 1.65) 11 0.85 (0.41, 1.78) 10 1.18 (0.50, 2.75) 26 4.12 (1.5, 11.33) 12 1.72 (0.52, 5.66) 7-9 5 1.21 (0.42, 3.47) 11 1.22 (0.58, 2.57) 7 1.48 (0.61, 3.59) 16 3.09 (0.96, 9.93) 16 3.72 (1.16, 11.94) 10 14 1.00 25 1.00 15 1.00 12 1.00 11 1.00 Q11. I prefer to keep tranquilizers at home in case there was a need for them later 0-3 7 1.00 14 1.00 11 1.00 20 1.00 8 1.00 4-8 9 2.71 (0.98, 7.48) 14 6.86 (3.11, 15.11) 10 2.05 (0.85, 4.95) 15 1.38 (0.47, 4.01) 13 3.91 (1.11, 13.83) 9-10 8 6.86 (2.37, 19.86) 19 11.78(5.54, 25.02) 12 6.08 (2.50, 14.83) 22 6.72 (2.34, 19.3) 18 13.94 (3.82, 50.85) Q12. I will trust the doctor’s decision if s/he decided to prescribe or not prescribe tranquilizers 0-6 10 3.11 (1.20, 8.07) 11 1.25 (0.59, 2.65) 13 2.45 (1.12, 5.39) 19 2.33 (0.81, 6.69) 7 0.66(0.16, 2.75) 7-9 6 1.55 (0.53, 4.57) 14 1.46 (0.72, 2.95) 7 0.93 (0.35, 2.48) 23 2.97 (1.1, 8.03) 13 1.78(0.6, 5.24) 10 8 1.00 22 1.00 14 1.00 15 1.00 19 1.00 Q13. If I believe that I need a tranquilizer and the doctor did not prescribe it, I will get it at the pharmacy without a prescription 0 15 1.00 27 1.00 17 1.00 38 1.00 28 1.00 1-5 4 1.05 (0.34, 3.25) 10 1.66 (0.78, 3.57) 9 2.15 (0.90, 5.14) 9 1.1 (0.34, 3.51) 4 0.66(0.15, 2.9) 6-10 5 4.40 (1.50, 12.93) 9 5.05 (2.17, 11.74) 7 6.15 (2.35, 16.05) 9 3.1 (0.81, 11.84) 7 3.2(0.76, 13.52) Q14. I think that doctors often explain clearly to the patients the reasons for prescribing or not prescribing tranquilizers 0-5 8 2.41 (0.85, 6.84) 17 1.06 (0.50, 2.25) 15 1.14 (0.49, 2.62) 28 4.11 (1.11, 15.24) 15 0.88(0.29, 2.64) 6-8 9 0.81 (0.30, 2.24) 17 2.07 (0.97, 4.41) 9 1.33 (0.53, 3.37) 25 7.87 (2.04, 30.36) 8 1.04(0.3, 3.67) 9-10 7 1.00 13 1.00 10 1.00 4 1.00 16 1.00 Q15. I think that doctors often explain clearly to the patients the instructions for the use of tranquilizers 0-5 6 2.78 (0.91, 8.56) 16 1.24 (0.59, 2.60) 15 1.53 (0.66, 3.52) 23 2.26 (0.75, 6.83) 11 0.74(0.22, 2.45) 6-8 11 1.24 (0.47, 3.27) 16 2.01 (0.95, 4.26) 8 1.32 (0.51, 3.44) 25 4.65 (1.52, 14.23) 12 1.78(0.57, 5.59) 910 7 1.00 14 1.00 10 1.00 9 1.00 16 1.00 Q16. I think that, when dispensing tranquilizers, the pharmacists tell the customers about the importance of correct therapeutic compliance/adherence 0-4 11 2.96 (1.00, 8.74) 13 0.72 (0.34, 1.53) 12 1.95 (0.82, 4.66) 19 1.4 (0.47, 4.2) 14 0.83(0.27, 2.59) 5-8 8 1.23 (0.39, 3.84) 16 0.43 (0.21, 088) 11 0.69 (0.29, 1.63) 23 0.8 (0.28, 2.25) 7 0.33(0.1, 1.09) 9-10 5 1.00 18 1.00 11 1.00 15 1.00 18 1.00 CI: Confidence interval; №: Number of participants; OR: Odds Ratio; *: OR adjusted for age, gender, alcohol intake and medical consultation over the phone **: OR adjusted for age, gender, employment status and frequency of doctor consultation in case of illness. Narmeen Mallah Results 89 5.6. Dose-response meta-analysis of the association of income with antibiotic misuse Practices The manuscript of the meta-analysis of income and antibiotic misuse was submitted for publication and is currently under review by European Journal of Health Economics. Literature Search and Study Selection The selection process of studies about the association of income with antibiotic misuse Practices is represented in Figure 7. Out of 1,453 identified studies, 314 were selected for full text review. Fifty-one studies involving 51,008 individuals and 18,094 antibiotic misuse events were included in the meta-analysis (Table 25). These studies originated from 22 countries and were published between 2001 and 2021. All studies were of cross-sectional design and published in English, except for one in Spanish [165] and another in Croatian [166]. Continuous approach of dose-response meta-analysis The data the 51 studies included in the meta-analysis were compatible with a flat linear association between income standardized to GDP per capita based on PPP and antibiotic misuse (OR= 1.00; p-value= 0.954, p-value non-linearity= 0.452). Figure 8 represents the dose-response trend of summary ORs of antibiotic misuse conferred by income standardized to GDP per capita based on PPP. Likewise, the association of income standardized to adjusted net national income per capita and antibiotic misuse followed a flat linear trend (OR= 1.00; p-value= 0.940). Categorical approach of dose-response meta-analysis In the analysis of income standardized to GDP per capita based on PPP, overall, in reference to low income (1st tertile), there is no association between income and general antibiotic misuse [medium income2nd tertile: OR= 1.04 (95%CI: 0.89, 1.20), and high income3rd tertile: OR= 1.03 (95%CI: 0.82, 1.29)] (Table 26). However, in the subgroup analysis, certain associations between income and antibiotic misuse were observed. As compared with low income individuals, those with medium income are at 19% higher odds of the misuse Practice “storage of antibiotic leftover” [OR=1.19 (95%CI: 1.07, 1.32)]; but high income individuals have 51% lower odds of the misuse Practice “non-adherence” to antibiotic treatment [OR= 0.49, (95%CI: 0.34, 0.70)] (Table 26). In upper-middle wealth countries, high income people have 11% higher odds of any Practice of antibiotic misuse than those with low income (OR= 1.11; 95%CI: 1.00, 1.22) (Table 26). An association of medium income level with antibiotic misuse was also suggested in African countries (OR= 1.18; 95%CI: 1.00, 1.39) (Table 26). Comparing to studies undertaken until 2015, there were higher odds of antibiotic misuse in medium income [OR until 2015= 0.95 (95%CI: 0.75, 1.20) and OR after 2015= 1.12 (95%CI: 0.99, 1.26)] and high income individuals [ORuntil 2015= 0.91 (95%CI: 0.62, 1.35) and ORafter 2015= 1.15 (95%CI: 0.93, 1.41)] after 2015 (Table 26). There was no meaningful difference in the odds of antibiotic misuse by income level (medium and high) when countries were classified according to the literacy rate of their population (Table 26). Narmeen Mallah Results 90 Similar findings were obtained for the categorical approach of income standardized to net national income per capita to that of income standardized to GDP per capita based on PPP (data not shown). Methodological characteristics of studies included in the dose-response meta-analysis of income and antibiotic misuse Practices Studies were sub-grouped based on their use of a pretested or validated questionnaire. No meaningful change in the pooled OR estimates were observed for medium and high income when the analysis was restricted to those studies that used pretested or validated questionnaires [ORmedium= 1.06 (95%CI: 0.91, 1.24) and ORhigh= 1.04 (95%CI: 0.85, 1.27) (Table 26). Stratifying the studies according to their comparability for age, gender, education and family size, showed that studies that incompletely controlled for those variables generated higher ORs than those with complete control for the four variables. This observation applies for both medium income [ORincomplete= 1.09; (95%CI: 0.95, 1.24) and ORcomplete= 0.90 (95%CI: 0.71, 1.15)] and high income categories [ORincomplete= 1.05 (95%CI: 0.84, 1.31) and ORcomplete= 0.60 (95%CI: 0.30, 1.23)] (Table 26). There was no notable difference in the summary estimates upon stratifying the analysis according to the quality score of studies (Table 26). The shape of the funnel plot (Figure 9) and the results of Egger’s test (p-value = 0.39) and that of the trim-and-fill analysis which did not suggest the addition of any study, confirmed the absence of publication bias. Narmeen Mallah Results 91 Figure 7 Flow diagram of the selection of studies about income level and misuse of antibiotics Narmeen Mallah Results 92 Table 25 General characteristics of studies included the meta-analysis of income level and antibiotic misuse Author, Year Country Setting Age (Years) Sex Outcome Mean Income (USD) Total N/leve l Outcome/ level Odds Ratio (95%CI) Adjustment, restriction or matching variables Moktan 2021 [167] India Attendants of public hospital 18-90 M: 309 F: 195 Use without prescription 37.50 137 41 Reference category age, gender, educational level, marital status, public and private clinics, frequency of doctors’ consultation, family/friend influence (other family members self-medicating with antibiotics), symptoms (minor illness) 112.51 185 59 1.10 (0.68 - 1.77) 225.01 129 52 1.58 (0.95 - 2.63) 375.01 53 19 1.31 (0.67 - 2.56) Chen 2020 [168] Mali Medical university students Mean (SD) 21.3 (2.4) M:310 F:136 Storage of antibiotics 82.95 290 168 Reference category age 506.50 114 77 1.51 (0.96 - 2.38) 1181.60 42 27 1.31 (0.67 - 2.56) Use without prescription 82.95 290 73 Reference category 506.50 114 29 1.01 (0.62 - 1.67) 1181.60 42 19 2.46 (1.27 - 4.77) Elmahi 2020 [169] Sudan General population ≥18 M: 130 F: 116 Use without prescription 49.50 182 110 1.05 (0.59 - 1.87) age, pregnancy, current antibiotic use 149.50 64 38 Reference category Mallah 2020 [170] Lebanon Children´s caregivers ≥18 M:276 F:1092 Any misuse practice 249.50 21 2 Reference category age, sex, educational level, area of residence, alcohol consumption, access to medical care facilities, and frequency of telephone medical consultation., 999.50 260 34 1.43 (0.32 - 6.41) 2000.00 223 17 0.78 (0.17 - 3.65) 3000.50 808 36 0.44 (0.10 - 1.98) Nusair 2020 [171] Jordan General population 0 to >65 M: 674 F: 1169 Use without prescription 88.75 175 61 Reference category past month antibiotic use 266.61 659 253 1.16 (0.82 - 1.65) 444.11 1042 458 1.47 (1.05 - 2.05) Rathish 2020 [48] Sri Lanka General population Mean (SD): 36 (21) M: 181 F: 203 Use without prescription 150.00 267 263 Reference category NA 450.00 117 111 2.15 (0.37 - 12.54) Xu 2020 [172] China Children´s caregivers Parents with children <13 years old M: 1344 F: 4935 Use without prescription 377.50 --- --- Reference category age, gender, educational level, medical background, residential location 1132.58 --- --- 0.76 (0.57 - 1.03) 1887.58 --- --- 0.81 (0.54 - 1.21) Storage of antibiotics 377.50 --- --- Reference category 1132.58 --- --- 1.03 (0.91 - 1.17) 1887.58 --- --- 1.16 (0.99 - 1.36) Ateshim 2019 [173] Eritrea General population Median (IQR): 37 (24) M: 238 F: 339 Use without prescription 0.00 291 --- Reference category age, gender, educational level, marital status, occupational status, knowledge about antibiotics, attitudes towards antibiotics 32.53 92 --- 0.92 (0.54 - 1.56) 113.78 136 --- 1.22 (0.78 - 1.19) 211.28 58 --- 1.43 (0.75 - 2.73) Narmeen Mallah Results 93 Author, Year Country Setting Age (Years) Sex Outcome Mean Income (USD) Total N/leve l Outcome/ level Odds Ratio (95%CI) Adjustment, restriction or matching variables Benameur 2019 [174] Saudi Arabia University students Mean (SD): 20.96 (0.148) M:166 F:69 Use without prescription 133.37 164 95 Reference category age, gender, educational level, marital status, speciality (medical vs nonmedical), residential location, health insurance 667.50 50 26 0.79 (0.42 - 1.49) 1468.63 18 14 2.54 (0.80 - 8.06) Bogale 2019 [49] Ethiopia General population 18 to > 60 M: 246 F: 349 Use without prescription 10.75 --- 46 Reference category age, gender, educational level, marital status, residential location, occupational status, healthcare profession 32.27 --- 74 2.55 (1.18 - 5.50) 64.52 --- 42 1.08 (0.47 - 2.46) 107.52 --- 92 1.42 (0.62 - 3.25) Bulabula 2019 [47] South Africa Pregnant women attending public hospital Mean (SD): 29 (6.1) F: 301 Use without prescription 49.50 --- --- Reference category age, gender, educational level, residential location, knowledge about antibiotics, attitudes towards antibiotics 174.50 --- --- 5.40 (0.90 - 29.90) 375.00 --- --- 4.10 (0.80 - 19.40) 625.00 --- --- 6.40 (1.20 - 35.20) Mate 2019 [175] Mozambi que General population Median (IQR): 33 (25-47) M:294 F:797 Use without prescription 21.24 528 108 Reference category age 63.75 224 45 0.98 (0.66 - 1.44) 127.51 183 40 1.09 (0.72 - 1.64) 212.51 117 26 1.11 (0.68 - 1.80) Incomplete course of treatment 21.24 506 150 Reference category 63.75 215 68 1.10 (0.78 - 1.55) 127.51 175 60 1.24 (0.86 - 1.79) 212.51 114 21 0.54 (0.32 - 0.89) Mukattash 2019 [176] Jordan Children´s caregivers 20 to ≥ 50 M: 134 F: 712 Use without prescription 352.50 94 41 Reference category age 1058.21 325 141 0.99 (0.62 - 1.57) 1763.21 427 150 0.70 (0.44 - 1.10) Sun 2019 [59] China Children´s caregivers Parents with children <13 years old M: 2243 F: 7283 Storage of antibiotics 230.50 2102 874 Reference category age, gender of the parents, gender of the child, educational level, socioeconomic characteristics (residential location and GDP per capita), health insurance, specialty (medical vs non-medical) 615.50 2889 1434 1.22 (1.08 - 1.38) 1154.00 2749 1355 1.17 (1.02 - 1.33) 1923.00 1786 917 1.36 (1.16 - 1.60) Hu 2018 [177] China Medical university students Mean (SD): 22 (1.5) M: 661 F: 1158 Use without prescription 768.50 1565 59 Reference category age, gender, educational level, parents’ educational level, parents medical background, residential location, knowledge - attitudes - and practice score, center of recruitment 2306.50 254 18 1.95 (1.13 - 3.36) Narmeen Mallah Results 94 Author, Year Country Setting Age (Years) Sex Outcome Mean Income (USD) Total N/leve l Outcome/ level Odds Ratio (95%CI) Adjustment, restriction or matching variables Tong 2018 [52] China Attendants of primary care clinics <45 to >60 M:340 F:374 Nonadherence 153.20 162 150 Reference category age, gender, educational level, residential location, occupation, employment status, knowledge about antibiotics 344.78 180 163 0.72 (0.33 - 1.57) 651.18 187 158 0.40 (0.20 - 0.82) 880.98 185 150 0.33 (0.16 - 0.66) Peng 2018 [50] China University students Guizhou Mean (SD): 21.3 (2.1) Zhejiang Mean (SD): 19.7 (2.6) M: 2035 F: 1960 Use without prescription 230.92 --- --- Reference category age, socioeconomic characteristics (GDP per capita and residential location) 1001.00 --- --- 0.65 (0.39 - 1.09) 2079.08 --- --- 0.66 (0.33 - 1.31) Storage of antibiotics 230.92 --- --- Reference category 1001.00 --- --- 1.30 (1.10 - 1.53) 2079.08 --- --- 1.14 (0.90 - 1.43) Buying without prescription 230.92 --- --- Reference category 1001.00 --- --- 1.14 (0.90 - 1.44) 2079.08 --- --- 1.05 (0.76 - 1.46) Redzick 2018 [166] Croatia Attendants of primary care clinics --- M: 142 F: 402 Use without prescription 84.62 88 5 Reference category age 226.12 55 13 5.14 (1.72 - 15.38) 339.32 97 4 0.71 (0.19 - 2.75) 452.52 100 15 2.93 (1.02 - 8.42) 594.02 199 25 2.39 (0.88 - 6.45) Wang 2018 [178] China University students Mean (SD): 20.7 (2.7) M: 5515 F: 5677 Storage of antibiotics 230.92 3417 --- Reference category age, gender, educational level, parents’ educational level, parents medical background, residential location, speciality (medical vs non-medical) 1001.00 5823 --- 1.15 (1.04 - 1.27) 2310.08 1435 --- 1.02 (0.88 - 1.19) 3850.08 517 --- 1.00 (0.81 - 1.23) Use without prescription 230.92 3417 --- Reference category 1001.00 5823 --- 0.89 (0.67 - 1.19) 2310.08 1435 --- 1.13 (0.75 - 1.71) 3850.08 517 --- 0.93 (0.53 - 1.63) Abdelrahm an 2017 [179] Saudi Arabia General population <18 to >65 M: 735 F: 293 Use without prescription 200.12 368 112 Reference category age 867.62 146 60 1.59 (1.07 - 2.37) 2002.50 198 72 1.31 (0.91 - 1.88) 3337.63 316 146 1.96 (1.43 - 2.69) Albawani 2017 [180] Yemen Attendants of pharmacies Mean (SD): 28.6 (7.7) M: 204 F: 159 Use without prescription 116.80 268 229 Reference category age 352.80 51 46 1.57 (0.59 - 4.19) 581.90 44 41 2.33 (0.69 - 7.89) Narmeen Mallah Results 95 Author, Year Country Setting Age (Years) Sex Outcome Mean Income (USD) Total N/leve l Outcome/ level Odds Ratio (95%CI) Adjustment, restriction or matching variables Erku 2017 [54] Ethiopia General population Mean (SD): 33.19 (10.82) M: 163 F: 487 Any misuse practice 50.00 331 282 Reference category age, gender, educational level, marital status, household size, employment status, frequency of visiting health care institutions, satisfaction about healthcare service 125.50 201 170 0.95 (0.58 - 1.55) 175.50 118 83 0.41 (0.25 - 0.68) Gebrekirst os 2017 [181] Ethiopia Attendants of pharmacies Median (IQR): 30 (16) M: 473 F: 307 Use without prescription 3.26 130 76 1.67 (1.13 - 2.48) age, gender, educational status, marital status, employment status, household size, residential location, type of illness, healthcare insurance, previous experience with antibiotics, access to healthcare 13.00 92 41 0.96 (0.61 - 1.50) 26.00 81 32 0.78 (0.48 - 1.26) 39.02 477 218 Reference category Gillani 2017 [182] Pakistan Nonmedical university students Mean (SD): 23.0 (3.4) M:352 F:375 Use without prescription 75.00 245 110 Reference category age, specialty (non-medical) 225.00 180 80 0.98 (0.67 - 1.45) 400.00 136 54 0.81 (0.53 - 1.24) 600.01 166 82 1.20 (0.81 - 1.78) Hassali 2017 [183] Malaysia General population Mean (SD): 28.7 (7.4) M: 171 F: 229 Any misuse practice 124.88 231 82 Reference category age, gender, educational level, marital status, race, healthcare related occupation, employment status, health insurance 499.88 94 29 0.51 (0.27 - 0.98) 1000.00 47 13 0.40 (0.16 - 0.78) 1500.13 28 7 0.42 (0.13 - 1.34) Jamhour 2017 [85] Lebanon General population >18 M: 182 F: 218 Use without prescription 499.50 88 36 Reference category age, gender, educational level, specialty (unrelated to health care) 1500.00 97 54 1.81 (1.01 - 3.25) Kajeguka 2017 [61] Tanzania General population Mean (SD): 35.4 (13.4) M:144 F:156 Use without prescription 49.50 162 70 2.82 (0.47 - 16.68) age, gender, educational level, marital status, employment status, self-treated condition 300.50 102 74 1.02 (0.22 - 4.76) 700.50 36 23 Reference category Kurniawan 2017 [184] Indonesia Attendants of primary care clinics Median (IQR): 45 (18-49) M: 137 F: 263 Use without prescription 87.50 186 146 Reference category age, gender, educational level, marital status, employment status, health insurance 262.50 54 34 0.52 (0.24, 1.12) Nuñez 2017 [185] Perú University students Mean: 19.82 M: 492 F: 508 Use without prescription 462.00 321 204 Reference category age 1386.62 322 211 1.09 (0.79 - 1.51) 2772.62 178 119 1.16 (0.79 - 1.70) 4620.62 179 120 1.17 (0.79 - 1.72) Senadheer a 2017 [53] Sri Lanka General population ≥18 M: 190 F: 174 Use without prescription 87.50 292 15 Reference category age, gender, educational level, employment status, health insurance, household size, receiving medical treatment in the last three months, knowledge of antibiotic name 262.51 288 26 1.83 (0.95 - 3.54) Narmeen Mallah Results 102 education [OR= 0.70 (95%CI; 0.44, 1.13)], and each 1-year increment in education is associated with 3% lower odds of antibiotic misuse [OR= 0.97 (95%CI: 0.93, 1.01)] (Table 28). Stratifying by WHO regions revealed that in the Eastern Mediterranean region, individuals with high education is associated with 36% lower odds of any Practice of antibiotic misuse than those with low education level [OR= 0.64 (95%CI: 0.42, 1.00)]. Inversely, in the European region, individuals with high education have 25% higher odds of antibiotic misuse than those with low education [OR= 1.25 (95%CI: 1.00, 1.58)]. Likewise, in the continuous approach, every 1-year increase in education is associated with 4% lower odds of antibiotic misuse in the Eastern Mediterranean region [OR= 0.96 (95%CI: 0.93, 1.00)] and 2% higher odds of misuse in the European region [OR= 1.02 (95%CI: 1.00, 1.04)] (Table 28). Summary estimates from studies published after 2015 indicated 19% lower odds of antibiotic misuse by individuals with high education level than those with low education level [OR= 0.81; (95%CI: 0.64, 1.02)]. Each 1-year increment in education is associated with 2% reduced odds of antibiotic misuse after 2015 [OR after 2015= 0.98 (95%CI: 0.96, 1.00)] (Table 28). Methodological characteristics of included studies Pooled estimate from the 68 studies that used a pretested or validated questionnaire did not notably differ from that estimated from all studies for medium education [OR= 1.01 (95%CI: 0.89, 1.15)] as well as for 1-year increment [OR= 0.99 (95%CI: 0.98, 1.01)]. Nevertheless, the association of high education with antibiotic misuse was less strong among studies that applied validated questionnaire than among all studies analyzed together (OR= 0.95; 95%CI: 0.80, 1.13) (Table 28). No association between education and antibiotic misuse was observed in the subgroup of studies that controlled for age and gender or in the subgroup of studies with higher quality score (>3 points) (Table 28). The funnel plot of studies reporting medium education level was slightly skewed to the left (Figure 14). However, the absence of publication bias was confirmed by both Eggers’s test (pvalue= 0.065) and trim-and-fill analysis that did not suggest the addition of any study. The funnel plot of studies that assessed high education level was also slightly skewed to the left (Figure 14). Egger’s test indicated the presence of publication bias (p-value= 0.001), but no additional studies were suggested by trim-and-fill analysis. Narmeen Mallah Results 103 Figure 10 Flow diagram of the selection of studies about the association of education level with misuse of antibiotics Narmeen Mallah Results 104 Figure 11 Forest plot of studies examining the association between medium education and antibiotic misuse Narmeen Mallah Results 105 Figure 12 Forest plot of studies examining the association between high education and antibiotic misuse Narmeen Mallah Results 106 Table 27 General characteristics of studies included in the dose-response meta-analysis about education level and misuse of antibiotics Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Cohort studies Afari-Asiedu 2020 [209] Ghana Adults (<20 to >60) M: 36.0% F: 64% General population Any misuse 0.0 283 27 1.00 Age, gender, marital status, occupation, place of residence, health insurance, type of drug supplier 3.5 176 19 0.90 (0.40, 1.70) 8.0 138 22 0.60 (0.30, 1.10) 13.0 164 34 0.40 (0.20, 0.70) Ho 2010 [210] Australia Adults (≥ 18) ≤ 30 to > 45 M: 43% F: 57% Emergency department attendants Nonadherence 3.0 16 4 1.00 Age, gender, marital status, employment status, having a regular general practitioner, number of usual medications, place of birth, use of herbal medicine 9.0 70 16 0.79 (0.46, 4.55) 12.0 35 10 0.39 (0.06, 2.78) 15.5 71 9 0.33 (0.08, 1.43) Cross-sectional studies Moktan 2021 [167] India Adults (>18) M: 61.3% F: 38.7% Attendants of primary care centers Use without prescription 0.0 28 12 1.00 Age, gender, marital status, income, public and private clinics, frequency of doctors’ consultation, family/friend influence (other family members self-medicating with antibiotics), symptoms (minor illness) 6.5 206 65 0.61 (0.28, 1.37) 14.5 158 59 0.79 (0.35, 1.80) 19.0 112 35 0.61 (0.26, 1.42) Bianco 2020 [211] Italy Adults mean ± SD: 47.8 ± 16.7 M: 41.2% F: 58.8% Attendants of primary care centers Use without prescription 4.0 --- --- 1.00 Age, gender, marital status, employment status, nationality, having a kid who had used antibiotics in the last year 11.0 --- --- 1.38 (0.86, 2.21) 16.0 --- --- 1.34 (0.78, 2.29) Elmahi 2020 [169] Sudan Adults (>18) M: 52.8% F: 47.2% General population Use without prescription 6.0 35 23 1.00 Age 14.5 211 125 0.76 (0.36, 1.61) Hallit 2020 [212] Lebanon Adults mean ± SD: bad practice 33.33 ± 8.80 good practice 31.49 ± 6.68 M: 48.5% F: 51.5% Children caregivers Any misuse 3.0 48 40 1.00 Age 11.0 40 22 0.24 (0.09, 0.65) 14.5 116 41 0.11 (0.05, 0.26) Mallah 2020 [170] Lebanon Adults (>18) M: 19.4% F: 76.8% Children caregivers Any misuse 6.0 252 31 1.00 Age, gender, area of residence, access to medical care facilities, frequency of telephone medical consultation 14.5 1157 62 0.40 (0.27, 0.64) Nusair 2020 [171] Jordan Unrestricted age group M: 36.6% F: 63.4% General population Use without prescription 0.0 51 18 1.00 --- 5.5 52 20 1.15 (0.51, 2.55) 11.5 273 104 1.13 (0.60, 2.11) 15.5 1505 632 1.33 (0.74, 2.38) Narmeen Mallah Results 107 Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Rathish 2020 [48] Sri Lanka Unrestricted age group Mean ± SD: 36.0 ± 21.0 M: 47% F: 53% General population Use without prescription 11.0 252 247 1.00 --- 13.5 132 127 3.95 (0.74, 21.10) Shah 2020 [213] Pakistan ≥ 15 M: 33% F: 67% General population Use without prescription 0.0 45 25 1.00 Age 4.5 231 30 0.12 (0.06, 0.24) 10.5 198 54 0.30 (0.15, 0.58) 14.5 221 71 0.38 (0.20, 0.73) Xu 2020 [172] China Parents with children <13 years old M: 21.5% F: 78.5% Children caregivers Use without prescription 4.5 1344 --- 1.00 Age, gender, income, medical background, residential location 11.0 1771 --- 0.85 (0.58, 1.25) 14.5 3164 --- 0.77 (0.53, 1.14) Antibiotics Storage 4.5 1344 --- 1.00 11.0 1771 --- 1.38 (1.19, 1.60) 14.5 3164 --- 1.55 (1.33, 1.81) Ateshim 2019 [173] Eritrea Median (IQR): 37 (24) M: 41.2% F: 58.8% General population Use without prescription 0.0 32 11 1.00 Age, gender, income, marital status, occupational status, knowledge about antibiotics, attitudes towards antibiotics 3.0 64 22 1.96 (0.56, 6.88) 6.5 95 33 1.62 (0.46, 5.72) 9.5 246 111 1.92 (0.55, 6.69) 14.0 140 81 2.8 (0.74, 10.63) Bogale 2019 [49] Ethiopia 18 to > 60 M: 41.3% F: 58.7% General population Use without prescription 0.0 --- 26 6.39 (1.45, 28.19) Age, gender, income, marital status, residential location, occupational status, healthcare profession 4.5 --- 72 1.50 (0.82, 2.74) 10.5 --- 75 1.46 (0.74, 2.87) 14.5 81 1.00 Bulabula 2019 [47] South Africa Mean ± SD: 29 ± 6.1 F: 100% Pregnant women attending public hospital Use without prescription 5.0 14 --- 1.00 Age, gender, income, residential location, , knowledge about antibiotics, attitudes towards antibiotics 11.0 228 --- 1.25 (0.15, 10.10) 15.5 5 --- 1.60 (1.03, 2.55) Ekambi 2019 [214] Cameroon ≥ 15 Mean ± SD: 35.02 ± 10.5 M: 52% F: 48% Attendants of pharmacies Use without prescription 3.0 25 8 1.00 Age, gender, marital status, occupation 10.0 86 28 1.02 (0.94, 2.66) 15.0 66 33 2.07 (1.10, 4.00) Mate 2019 [175] Mozambique Adults (>18) Median (IQR): 33 (25-47) M: 26.9% F: 73.1% General population Use without prescription 0.0 69 17 1.00 Age 4.0 344 63 0.69 (0.37, 1.26) 11.5 555 119 0.83 (0.47, 1.50) 14.5 114 28 1.00 (0.50, 1.99) Nonadherence 0.0 65 15 1.00 4.0 334 88 1.19 (0.64, 2.23) 11.5 531 177 1.67 (0.91, 3.05) 14.5 110 33 1.43 (0.70, 2.90) Narmeen Mallah Results 108 Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Mukattash 2019 [176] Jordan 20 to ≥ 50 M: 15.8% F: 84.2% Children caregivers Use without prescription 6.0 205 106 1.00 Age 18.0 1487 554 0.56 (0.42, 0.75) Rajendran 2019 [215] India Adults (>18) M: 51% F: 49% General population Use without prescription 9.5 151 2 1.00 Age 11.0 60 2 2.57 (0.35, 18.67) 14.0 540 21 3.01 (0.70, 13.00) Sun 2019 [59] China Parents with children <13 years old M: 23.5% F: 76.5% Children caregivers Antibiotics storage 5.0 2519 1026 1.00 Age, gender of the parents, gender of the child, socioeconomic characteristics (residential location and GDP per capita), health insurance, specialty (medical vs nonmedical) 11.0 2765 1365 1.34 (1.20, 1.51) 14.5 4242 2189 1.50 (1.33, 1.70) Voidăzan 2019 [216] Romania Adults (20-80) Mean ± SD: 45 ± 12.5 M: 36.5% F: 63.5% Attendants of primary care centers Use without prescription 2.5 200 11 1.00 --- 9.0 564 94 3.43 (1.80, 6.56) 15.0 438 102 5.22 (2.73, 9.96) 22.5 790 77 1.86 (0.97, 3.56) Adisa 2018 [217] Nigeria Mothers of children under five F: 100% Mothers of children under five attending primary care centers Use without prescription 3.0 19 8 1.00 Gender 9.5 148 89 2.07 (0.79, 5.46) 14.5 160 73 1.15 (0.44, 3.02) Al-Qahtani 2018 [218] Saudi Arabia Adults (≥18) M: 43.4% F: 56.6% Attendants to primary care centers Use without prescription 0.0 13 4 1.00 Age, gender 3.5 39 13 1.13 (0.29, 4.35) 10.5 108 39 1.27 (0.37, 4.40) 15.5 281 119 1.65 (0.50, 5.49) 21.0 48 23 2.07 (0.56, 7.65) Chang 2018 [219] China Caregivers of children under seven M: 33.3% F: 66.7% Caregivers of children under seven Use without prescription 4.5 336 --- 1.00 Gender, child gender and age, number of children, health insurance, residential location 11.0 1428 --- 0.75 (0.57, 0.98) 14.5 1407 --- 0.82 (0.62, 1.08) 19.0 187 --- 0.86 (0.56, 1.32) Cheng 2018 [220] China 30 to ≥ 71 M: 41.4% F: 68.3% General rural population Use without prescription 0.0 135* 83* 1.00 Age, gender, household size, health insurance 3.5 165* 112* 1.19 (0.69, 2.05) 8.0 227* 139* 0.89 (0.50, 1.57) 13.0 97* 57* 0.84 (0.40, 1.74) Dar-Odeh 2018 [221] Saudi Arabia 15-64 Mean ± SD: 29.08 ± 9.32 M: 40.7% F: 59.3% Attendants of primary care centers Use without prescription 6.0 138 47 1.00 Age 15.5 333 81 0.63 (0.41, 0.97) 21.0 29 7 0.62 (0.25, 1.56) El-Sherbiny 2018 [55] Egypt Adults (≥ 18) M: 50.3% F: 49.7% Attendants of primary care centers Any misuse 6.0 419 --- 1.00 Age, gender, income, occupation, residential location 14.5 181 --- 0.52 (0.34, 0.79) Narmeen Mallah Results 109 Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Horumpende 2018 [222] Tanzania Median (IQR): 23 (20.5 - 36.5) M: 53.33% F: 46.67% General population Use without prescription 0.0 17 9 1.00 Age, gender, income, marital status, occupation 4.0 111 70 1.45 (0.46, 4.51) 10.0 172 95 1.02 (0.32, 3.25) Kamata 2018 [63] Japan 20 - 69 M: 51.2% F: 48.8% General population Antibiotics storage 5.0 111 18 1.00 Age 11.0 1265 115 0.52 (0.30, 0.89) 14.5 1932 254 0.78 (0.46, 1.32) Nonadherence 5.0 111 26 1.00 11.0 1265 282 0.94 (0.59, 1.48) 14.5 1932 479 1.08 (0.69, 1.69) Ngu 2018 [223] Cameroon ≥ 21 Median (IQR): 35 (27 - 49) M: 44.8% F: 55.2% Attendants of primary care centers Use without prescription 0.0 29 14 1.00 Age 6.5 279 115 0.75 (0.35, 1.62) Redzick 2018 [166] Croatia --- M: 26.1% F: 73.9% Attendants of primary care centers Use without prescription 4.5 32 8 1.00 Age 10.5 312 38 0.42 (0.17, 0.99) 14.5 60 4 0.21 (0.06, 0.78) 19.0 140 12 0.28 (0.10, 0.76) Tong 2018 [52] China <45 to >60 M: 47.6% F: 52.4% Attendants of primary care centers Nonadherence 6.5 323 286 1.00 Age, gender, income, residential location, occupation, employment status, knowledge about antibiotics 14.5 391 335 0.77 (0.50, 1.21) Abdelrahman 2017 [179] Saudi Arabia <18 to >65 M: 71.5% F: 28.5% General population Use without prescription 0.0 8 4 1.00 Age 5.0 58 21 0.57 (0.13, 2.51) 11.0 360 117 0.48 (0.12, 1.96) 15.5 602 248 0.70 (0.17, 2.83) Albawani 2017 [180] Yemen ≥ 18 Mean ± SD: 28.6 ± 7.7 M: 56.2% F: 43.8% Attendants of pharmacies Use without prescription 5.0 40 36 1.00 Age 11.0 71 68 2.52 (0.53, 11.87) 15.0 229 191 0.56 (0.19, 1.66) Narmeen Mallah Results 110 Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Akici 2017 [205] Turkey ≥ 15 M: 42.6% F: 57.4% Attendants of primary care centers Use without prescription 0.0 18 8 1.00 --- 4.5 18 6 0.63 (0.16, 2.41) 10.5 18 8 1.11 (0.29, 4.20) 14.5 18 9 1.25 (0.34, 4.64) Nonadherence 0.0 9 6 1.00 4.5 9 6 1.00 (0.14, 7.10) 10.5 9 7 1.75 (0.22, 14.22) 14.5 9 7 1.75 (0.22, 14.22) Antibiotics storage 0.0 8 3 1 4.5 9 4 1.67 (0.23, 12.22) 10.5 8 5 2.78 (0.37, 21.03) 14.5 9 6 3.33 (0.45, 24.44) Barber 2017 [62] Philippines Adults (≥ 18) Median (IQR): 32 (20) M: 37.2% F: 56.7% General population Use without prescription 3.5 83 69 1.00 Age, gender, household size 10.5 186 143 0.77 (0.30, 1.67) Erku 2017 [54] Ethiopia <29 to > 60 Mean ± SD: 33.19 ± 10.82 M: 25.9% F: 74.9% General population Use without prescription 0.0 182 125 5.01 (2.62, 9.34) Age, gender, income, marital status, household size, employment status, frequency of visiting health care institutions, satisfaction about healthcare service 4.5 201 180 2.81 (1.32, 6.15) 10.5 179 159 1.96 (0.91, 4.51) 14.5 88 71 1.00 Gebrekirstos 2017 [181] Ethiopia Adults (≥ 18) Median (IQR): 30 (16) M: 60.6% F: 39.4% Attendants of pharmacies Use without prescription 0.0 152 75 1.00 Age, gender, income, marital status, employment status, household size, residential location, type of illness, healthcare insurance, previous experience with antibiotics, access to healthcare 4.5 149 71 0.93 (0.59, 1.47) 10.5 260 121 0.89 (0.60, 1.33) 14.5 219 100 0.86 (0.60, 1.31) Hassali 2017 [183] Malaysia Adults (≥ 18) Mean ± SD: 28.7 ± 7.4 M: 42.75% F: 57.25% General population Any misuse 8.5 61 20 1.00 Age, gender, income, marital status, race, healthcare related occupation, employment status, health insurance 14.0 339 111 0.83 (0.53, 1.31) Jamhour 2017 [85] Lebanon Adults (≥ 18) M: 45.5% F: 54.5% General population Use without prescription 4.5 34 14 1.00 Age, gender, educational level, specialty (unrelated to health care) 11.0 151 76 1.45 (0.68, 3.08) Kajeguka 2017 [61] Tanzania Adults (≥ 18) Mean ± SD: 35.4 ± 13.4 M: 48.0% F: 52.0% General population Use without prescription 0.0 26 16 1.00 Age, gender, income, marital status, employment status, self-treated condition 4.0 87 33 0.38 (0.12, 0.94) 11.0 74 46 1.03 (0.41, 2.57) 16.0 133 72 1.10 (0.46, 2.64) Kurniawan 2017 [184] Indonesia Adults (≥ 18) Median (IQR): 45 (18-49) M: 34.3% F: 65.8% Attendants of primary care centers Use without prescription 3.5 26 24 1.00 Age, gender, income, marital status, employment status, health insurance 8.0 44 37 0.44 (0.08, 2.30) 11.0 139 101 0.22 (0.05, 0.98) 14.5 31 18 0.12 (0.02, 0.58) Narmeen Mallah Results 111 Source Country Age (years) Gender Population Type of misuse Dose N n OR (95% CI) Adjustment, restriction or matching factors Senadheera 2017 [53] Sri Lanka Adults (≥ 18) M: 31.30% F: 68.70% General population Use without prescription 11.0 362 14 1.00 Age, gender, income, employment status, health insurance, household size, receiving medical treatment in the last three months, knowledge of antibiotic name 13.5 245 37 0.32 (0.17, 0.63) Torres 2017 [165] Ecuador Adults (≥ 18) range: 18 - 64 M: 45.6% F: 54.4% General population Use without prescription 3.5 51 22 1.00 Age 10.5 173 83 1.22 (0.65, 2.28) 14.5 195 102 1.45 (0.78, 2.69) Abdulraheem 2016 [224] Nigeria Adults (≥ 18) Median (range): 25 (19-68) M: 61.1% F: 38.9% Attendants of primary care centers Use without prescription 3.5 623 --- 1.00 Age, gender, symptoms, occupation 9.5 390 --- 1.24 (1.13, 1.87) 14.5 137 --- 1.32 (1.18, 1.96) Aleem 2016 [186] Saudi Arabia <25 to ≥55 M: 39.5% F: 60.5% Children caregivers Use without prescription 6.0 102 29 1.00 Age, gender, income, household size 15.0 508 42 0.23 (0.13, 0.39) Al Rasheed 2016 [225] Saudi Arabia Adults (> 18) M: 23.2% F: 76.8% Attendants of primary care centers Use without prescription 0.0 100 13 1.00 Age, gender, marital status, employment status, occupation, symptoms 3.5 186 16 1.59 (0.73, 3.45) 10.5 145 13 1.52 (0.67, 3.43) 15.5 145 50 0.28 (0.14, 0.59) Bilal 2016 [187] Pakistan Mean ± SD: 48.6 ± 4.4 M: 65.8% F: 34.2% Attendants of primary care centers Use without prescription 0.0 161 159 1.00 Age, residential location, specialty (nonmedical related participants) 9.5 86 81 0.20 (0.04, 1.07) 11.5 65 49 0.04 (0.01, 0.17) 14.5 50 26 0.01 (0.003, 0.06) 19.5 38 10 0.004 (0.001, 0.02) Nigigi 2016 [226] Kenya Adults (≥ 18) M: 32.0% F: 68.0% Attendants of primary care centers Use without prescription 0.0 48 12 1.00 --- 4.0 115 35 1.13 (0.90, 1.97) 9.5 179 61 1.31 (0.92, 2.40) 13.5 20 9 1.57 (1.20, 2.50) Ding 2015 [189] China Adults ≤ 29 to > 50 M: 9.7% F: 90.3% Children caregivers Nonadherence 3.0 198 68 1.00 Age, access to healthcare (number of clinics) 8.0 304 87 0.77 (0.52, 1.13) 13.0 120 47 1.23 (0.77, 1.97) Gebeyehu 2015 [190] Ethiopia Mean ± SD: urban: 34.1 ± 12.9 rural: 34.5 ± 11.5 M: 24.3% F: 75.7% General population Use without prescription 0.0 137 56 4.21 (1.47, 12.07) Age, gender, educational level, marital status, employment status, residential location, household size, level of healthcare service satisfaction, knowledge on antibiotics use 4.5 145 40 2.01 (0.93, 4.34) 10.5 89 14 1.01 (0.34, 2.94) 14.5 35 10 1 Kusturica 2015 [227] Serbia Adults (≥ 18) M: 20.1% F: 79.9% General population Antibiotics storage 4.5 27 9 1.00 Age 10.5 169 75 1.60 (0.68, 3.76) 14.0 25 10 2.53 (0.89, 7.23) 16.5 162 84 2.15 (0.91, 5.08) Narmeen Mallah Results 118 A) represents the subgroup of studies that assessed the association of medium education with antibiotic misuse. B) represents the subgroup of studies that assessed the association of high education with antibiotic misuse. Figure 3. (A) (B) Figure 3. (A) (B) Figure 14 Funnel plot of the studies about education and antibiotic misuse. Narmeen Mallah Results 119 5.8. Meta-analysis of the association of macrolides with congenital malformation The meta-analysis of macrolides and congenital malformation has been published (Drug Safety 2020, 43(3), 211-221) and the corresponding publication is available in Annex II page 235. Literature Search and Study Selection Figure 15 represents the process of the selection of studies included in the meta-analysis of the association of prenatal exposure to macrolides with congenital malformation. Twenty-one studies published between 1998 and 2017 fulfilled the inclusion criteria (Table 29, Table 30 and Figure 16). The studies were mainly conducted in Europe and North America. They included 17 cohort and four population-based case control studies with an average of 10,483 cases and 12,937 controls. Two studies examined the effect of two different macrolides but involved the same population were pooled and analyzed as one study [242, 243]. The included studies applied as comparator fetuses who were not exposed to any drug before birth (comparison group 1: 14 studies) and/or fetuses who were prenatally exposed to drugs that are not macrolide antibiotics or non-teratogens (comparison group 2: 13 studies). Three analytical approached that involved comparison group 1, comparison group 2 and a mixture of the two comparison groups (comparison group 3) were carried out and yielded similar results (Table 31, Table 32 and Table 33). In comparison with fetuses who were not exposed to any drug (comparison group 1), overall, there is no association between prenatal exposure to macrolides and congenital malformation [OR: 1.05 (95%CI: 0.99, 1.12)] (Table 31). Studies that applied comparison group 2 and those studies that used mixed controls of groups 1 and 2 (comparison group 3), revealed a borderline positive association between macrolides prenatal exposure and congenital malformation [OR: 1.06 (95%CI: 1.00, 1.12) and 1.06 (95%CI: 1.01, 1.10), respectively] (Table 32 and Table 33). The association between macrolides and congenital malformation was not observed in any of the subgroups when compared to fetuses unexposed to any drug before birth (comparison group 1). The heterogeneity between studies was absent-to-marginal (Table 31). In the analysis that involved comparison groups 2 and 3, a substantial amount of heterogeneity across studies of several subgroups was present (Table 32 and Table 33). A slightly higher odds ratio of congenital malformation was observed for studies carried out in Europe [ORgroup2: 1.15 (95%CI: 1.01, 1.31) and ORgroup3: 1.12 (1.03, 1.22)] than that undertaken in North America [OR group 2: 1.04 (95%CI: 0.98, 1.11) and ORgroup3:1.03 (95%CI: 0.98, 1.09)] (Table 32 and Table 33). Higher odds were also observed for musculoskeletal system [ORgroup2: 1.21 (95%CI: 1.08, 1.35) and ORgroup3: 1.15 (95%CI: 1.05, 1.26)] and digestive system malformations [ORgroup3: 1.14 (95%CI: 1.02, 1.26)] (Table 32 and Table 33). No association was found between macrolides prenatal exposure and cardiovascular congenital malformations when compared with fetuses unexposed to any drug (comparison group 1) (Table 31) and to the mixed comparison group (comparison group 3) (Table 33). However, a negative association existed when compared to fetuses prenatally exposed to other antibiotics [OR group 2: 0.87 (95%CI: 0.81, 0.95)] (Table 32). When group 3 was used a comparator, a marginal association was also observed for the subgroups: cohort studies [ORgroup3: 1.07 (95%CI: 1.02, 1.13)], and macrolides exposure during the first trimester of pregnancy [ORgroup3: 1.06 (95%CI: 1.01, 1.12)] (Table 33). Narmeen Mallah Results 120 Studies included in the meta-analysis provided data about several classes of macrolides. The classes were mainly: erythromycin, azithromycin, clarithromycin and roxithromycin. Analyzing the effect of each type of macrolides separately revealed a substantial but statistically non-significant association of roxithromycin prenatal exposure with congenital malformation in the three approaches: [ORgroup1: 2.03 (95%CI: 0.75, 5.48), ORgroup2: 1.50 (95%CI: 0.81, 2.77), and ORgroup3: 1.63 (95%CI: 0.96, 2.75)] (Table 31, Table 32 and Table 33). Methodological assessment A median quality score of 4 points was obtained from the quality assessment. In the studies that used fetuses unexposed to any medicine as a reference (comparison group 1), no difference was found between higher (≥ 4 points) and lower quality (< 4 points) studies [ORgroup1: 1.04 (95%CI: 0.95, 1.14) versus ORgroup1: 1.06 (95%CI: 0.98, 1.16), respectively] (Table 31). However, when comparison groups 2 and 3 were used, higher quality studies yielded a statistically significant association that was similar to the one estimated from all studies [OR group 2: 1.08 (95%CI: 1.01, 1.15) and ORgroup3: 1.07 (95%CI: 1.01, 1.13)] (Table 32 and Table 33). The association was not statistically significant in the group of lower quality studies [ORgroup2: 1.02 (95%CI: 0.92, 1.13) and ORgroup3: 1.04 (95%CI: 0.97, 1.12)] (Table 32 and Table 33). There was no meaningful difference between studies with full adjustment and those with incomplete adjustment in the three approaches (Table 31, Table 32 and Table 33). The sensitivity analysis under the extreme assumptions showed no increase in the odds of congenital malformations from prenatal exposure to macrolides [OR: 1.00 (95%CI: 0.98, 1.03)]. Publication bias The funnel plot showed some degree of asymmetry with a larger number of studies favouring the presence of association between macrolides prenatal exposure and congenital malformation (right hand side of the figure, OR>1) (Figure 17), yet the negative result of Egger’s regression test did not confirm this asymmetry (p-value= 0.074). The trim-and-fill analysis suggested the imputation of six additional studies and a corrected OR of 1.02 (95%CI: 1.00, 1.04). Narmeen Mallah Results 121 Figure 15 Flow diagram of the selection of studies about macrolides prenatal exposure and congenital malformation Narmeen Mallah Results 122 Table 29 Characteristics of cohort studies of macrolides prenatal exposure and congenital malformations Source Type of Macrolide Country Type of malformation Exposure Period (months of pregnancy) Comparison Type and OR (95% CI) Cases/ Cohort Size General OR (95% CI) Adjustment, Matching, and Restriction Factors Muanda 2017 [244] azithromycin, erythromycin, clarithromycin Quebec Cardiac, digestive, head and neck, musculoskeletal, nervous, respiratory, urogenital 1 – 3 Unexposed to any antibiotic: 1.08 (0.95, 1.23) Exposed to nonmacrolide antibiotics: 1.04 (0.95, 1.14) 13, 852/139,938 1627/15469 1.05 (0.98, 1.14) Maternal age, urinary tract infections, socio-demographic variables, chronic maternal illness, endometriosis and other maternal infections, healthcare utilization, year of delivery, infant’s gender Lê Nguyên 2017 [245] macrolides France Congenital malformations 1 – 3 Unexposed to any antibiotic: 1.02 (0.71, 1.46) Treated with penicillin: 0.93 (0.63, 1.37) 47/62,846 47/12,193 0.98 (0.75, 1.27) Maternal age, long-term maternal illness, gestity, parity, multiple pregnancy Källén. 2014 [246] erythromycin Sweden Any, cardiac 1 – 3 Unexposed to any antibiotic: 1.14 (0.96, 1.36) 70339/1575847 1.14 (0.96, 1.36) Maternal age, year of delivery, parity, smoking, obesity Lund 2014 [72] azithromycin, clarithromycin, erythromycin, roxithromycin, spiramycin, Denmark Infantile hypertrophic pyloric stenosis • 1 – 6 • 6 – 9 Unexposed to any antibiotic: 1.23 (0.85, 1.76) Exposed to nonmacrolide antibiotics: 1.23 (0.83, 1.83) 30/315569 159/60732 1.23 (0.94, 1.60) Birth order, infant’s sex, calendar period, current age of the infant Andersen 2013 [74] clarithromycin Denmark Cardiac, musculoskeletal, urogenital 1 – 3 Unexposed to any antibiotic: 1.03 (0.53, 2.00) 24817/705837 1.03 (0.53, 2.00) Maternal age, education, number of previous births, economic status Dinur 2013 [247] azithromycin, clarithromycin, erythromycin, roxithromycin Israel Cardiac, digestive 1 – 3 Unexposed to any antibiotic: 1.07 (0.84, 1.38) ---/105492 1.07 (0.84, 1.38) Maternal age, year of delivery, parity, ethnicity, chronic maternal illness Narmeen Mallah Results 123 Source Type of Macrolide Country Type of malformation Exposure Period (months of pregnancy) Comparison Type and OR (95% CI) Cases/ Cohort Size General OR (95% CI) Adjustment, Matching, and Restriction Factors Bar-Oz 2012 [248] azithromycin, roxythromycin, clarithromycin • Czech Republic • Germany • Israel • Italy • Netherlands Cardiac 1 – 3 Exposed to nonteratogenic agents: 1.42 (0.70, 2.88) 32/1146 1.42 (0.70, 2.88) Maternal age, smoking, alcohol consumption, previous abortions, previous child with structural anomaly, macrolide exposure Romøren 2012 [249] erythromycin, azithromycin, clarithromycin, spiramycin Norway Any, Cardiac 1 – 3 Unexposed to any antibiotic: 1.02 (0.86, 1.23) Exposed to nonmacrolide antibiotics: 1.15 (0.96, 1.38) 8865/178142 413/9069 1.08 (0.95, 1.24) Maternal age, urinary tract infections, chronic maternal illness, parity, marital status, smoking, pregnancy supplement, previous abortions Cooper 2009 [250] azithromycin, erythromycin United States Any, digestive, head and neck, nervous, musculoskeletal, urogenital • 1 – 3 • 1 – 9 Unexposed to any antibiotic: 0.92 (0.73, 1.16) Exposed to nonmacrolide antibiotics: 1.03 (0.88, 1.21) 869/30,049 589/7471 0.99 (0.87, 1.13) Maternal age, year of delivery, race, rural residence, economic status, chronic maternal illness, filling of prescriptions of other known teratogens Bar-Oz 2008 [251] azithromycin, roxythromycin, clarithromycin • Croatia • Israel Cardiac 1 – 3 Exposed to nonmacrolide antibiotics: 1.62 (0.68, 3.86) 32/1066 1.62 (0.68, 3.86) Unadjusted Chun 2006 [252] roxythromycin South Korea Major 1 – 3 Unexposed to any teratogenic agent: 1.37 (0.07, 27.57) 3/187 1.37 (0.07, 27.57) Maternal age, gravity Sarkar 2006 [253] azithromycin Canada Major 1 – 3 Exposed to any nonteratogen: 1.01 (0.20, 5.11) 6/227 1.01 (0.20, 5.11) Maternal age, gestational age at call, smoking, alcohol consumption Wolfgang 2005 [254] roxithromycin Hungary Congenital anomalies 1 – 3 Unexposed to any antibiotic: 2.13 (0.75, 6.1) 15/275 2.13 (0.75, 6.1) Maternal age, gestational age at call Narmeen Mallah Results 124 Source Type of Macrolide Country Type of malformation Exposure Period (months of pregnancy) Comparison Type and OR (95% CI) Cases/ Cohort Size General OR (95% CI) Adjustment, Matching, and Restriction Factors Cooper 2002 [255] erythromycin, nonerythromycin, lincomycin, clindamycin, clarithromycin, azithromycin, dirithromycin United States Digestive • 6 – 9 • 1 – 9 Exposed to nonmacrolide antibiotics: 1.28 (0.96, 1.70) 679/260,799 1.28 (0.96, 1.70) Maternal age, education, geographic residence, use of other antibiotics, infant´s gender, infant´s race, birth order, year of delivery, infant´s postnatal prescriptions for erythromycin Mahon 2001 [256] macrolides United States Digestive 1 – 9 Unexposed to any antibiotic: 1.19 (0.6, 2.3) 43/14,876 1.19 (0.6, 2.3) Birth weights, gestational age Einarson 1998 [257] clarithromycin Canada Major 1 – 3 Exposed to nonteratogenic antibiotics: 1.1 (0.44, 2.78) 19/266 1.1 (0.44, 2.78) Maternal age, smoking, alcohol consumption Wilton 1998 [258] azithromycin United Kingdom Congenital anomalies 1 – 3 Exposed to nonmacrolide antibiotics: 1.75 (0.01, 31.23) 14/556 1.75 (0.01, 31.23) Unadjusted Narmeen Mallah Results 125 Table 30 Characteristics of case-control studies of macrolides prenatal exposure and congenital malformations Source Type of Macrolide Country Type of malformation Exposure Period (months of pregnancy) Comparison Type and OR (95% CI) № Cases/Controls General OR (95% CI) Adjustment, Matching, and Restriction Factors Lin KJ, et al. 2013 [259] Any macrolide, erythromycin, non-erythromycin • Canada • United States Cardiac, digestive, head and neck, musculoskeletal, nervous, respiratory, urogenital • 1 – 3 • 3 – 6 • 6 – 9 Unexposed to any antibiotic: 0.99 (0.85, 1.16) 4867/6,952 0.99 (0.85, 1.16) Maternal age, calendar year when they were ascertained, race, education, geographic residence, obesity, family history of congenital malformations or diabetes mellitus, smoking, pregnancy supplement, multiple pregnancy, urinary tract infections, maternal chronic illness Crider KS, et al. 2009 [70] erythromycin United States Cardiac, digestive, head and neck, musculoskeletal, nervous, urogenital 1 – 3 Unexposed to any antibiotic: 1.12 (0.96, 1.32) Exposed to nonmacrolide antibiotics: 1.01 (0.91, 1.13) 13155/4941 1384/516 1.04 (0.95, 1.15) Maternal age, race, education, obesity, gestational age at call, pregnancy supplements, smoking, alcohol consumption Louik C, et al. 2002 erythromycin • Canada • United States Digestive • 1 – 6 • 6 – 9 Unexposed to any antibiotic: 0.81 (0.57, 1.14) 1,044/1704 0.81 (0.57, 1.14) Maternal age, geographic region, study period, parity, infant´s gender, gestational age Czeizel AE, et al. 2000 & Czeizel AE, et al. 1999 [242, 243] erythromycin, spiramyicn, roxithromycin, oleandomycin, josamycin, Hungary Cardiac, head and neck, musculoskeletal, urogenital, nervous, others • 1 – 3 • 1 – 9 Exposed to other agents (not macrolides): 1.19 (0.92, 1.54) 22,865/38,151 1.19 (0.92, 1.54) Maternal age, urogenital disorders, birth order, maternal chronic illness, other drug uses Narmeen Mallah Results 126 Figure 16 Forest plot of studies included in the meta-analysis about macrolides prenatal exposure and congenital malformations Narmeen Mallah Results 127 Table 31 Pooled Odds Ratios (ORs) and 95% Confidence Intervals (CIs) of macrolides prenatal exposure and congenital malformations (comparison group 1: fetuses not exposed to any drug) CI: Confidence Interval; OR: Odds Ratio; *Ri: Proportion of total variance due to between-study variance. № of studies OR (95% CI) Fixed effects OR (95% CI) Random effects Ri* Q test (p value) Any congenital malformation All studies 14 1.05 (0.99, 1.12) 1.05 (0.99, 1.12) 0.00 0.81 Study Design Cohort 11 1.07 (0.99, 1.15) 1.07 (0.99, 1.15) 0.00 0.90 Case-control studies 3 1.02 (0.92, 1.14) 1.01 (0.88, 1.17) 0.39 0.21 Anatomic location Cardiovascular 8 1.03 (0.93, 1.14) 1.05 (0.90, 1.22) 0.48 0.07 Head and Neck 4 1.27 (0.94, 1.72) 1.27 (0.94, 1.72) 0.00 1.00 Musculoskeletal system 5 1.06 (0.91, 1.24) 1.06 (0.85, 1.31) 0.33 0.27 Digestive system 8 1.13 (0.97, 1.31) 1.12 (0.9, 1.33) 0.18 0.30 Urogenital system 5 1.00 (0.80, 1.24) 0.98 (0.71, 1.35) 0.43 0.17 Nervous system 4 1.14 (0.86, 1.52) 1.12 (0.81, 1.55) 0.19 0.28 Adjustment Full 2 1.04 (0.94, 1.15) 1.04 (0.94, 1.15) 0.00 0.50 Incomplete 12 1.06 (0.98, 1.15) 1.06 (0.98, 1.15) 0.00 0.75 Quality score ≥ 4 7 1.04 (0.95, 1.14) 1.04 (0.95, 1.14) 0.00 0.78 < 4 7 1.06 (0.98, 1.16) 1.06 (0.98, 1.16) 0.00 0.56 Macrolide exposure period First trimester 11 1.07 (1.00, 1.14) 1.07 (1.00, 1.14) 0.00 0.93 Third trimester 3 1.00 (0.79, 1.28) 1.08 (0.73, 1.59) 0.60 0.11 Geographic location Europe 6 1.10 (0.98, 1.23) 1.10 (0.98, 1.23) 0.00 0.71 North America 6 1.03 (0.96, 1.12) 1.03 (0.96, 1.12) 0.00 0.45 Type of treatment Erythromycin 7 1.05 (0.97, 1.15) 1.05 (0.97, 1.15) 0.00 0.46 Azithromycin 2 1.15 (0.97, 1.36) 1.15 (0.97, 1.36) 0.00 0.37 Clarithromycin 2 1.10 (0.88, 1.38) 1.10 (0.88, 1.38) 0.00 0.84 Roxithromycin 2 2.03 (0.75, 5.48) 2.03 (0.75, 5.48) 0.00 0.79 Narmeen Mallah Annexes 230 Narmeen Mallah Annexes 231 Narmeen Mallah Annexes 232 Narmeen Mallah Annexes 233 Narmeen Mallah Annexes 234 Mallah N, Figueiras A, Heidarian Miri H, Takkouche B. Association of knowledge and attitudes with practices of misuse of tranquilizers: A cohort study in Spain. Drug and Alcohol Dependence 2021 (In print) (Journal Ranking: First Quartile) Narmeen Mallah Annexes 235 Mallah N, Tohidinik T, Etminan M, Figueiras A, Takkouche B. Prenatal exposure to macrolides and risk of congenital malformations: A meta-analysis. Drug Safety 2020, 43(3), 211-221. (Journal Ranking: First Quartile) Narmeen Mallah Annexes 236 Narmeen Mallah Annexes 237 Narmeen Mallah Annexes 238 Narmeen Mallah Annexes 239 Narmeen Mallah Annexes 246 Annex III. International Mention of PhD Degree Authorization of the Academic Commission of Doctoral Program in Epidemiology and Public Health for a three-month research stay at Karolinska Institutet, Sweden Narmeen Mallah Annexes 247 Report about three-month research stay from the supervisor at Karolinska Institutet, Sweden Narmeen Mallah Annexes 248 Report about activities performed during three-month research stay at Karolinska Institutet, Sweden Narmeen Mallah Annexes 249