Exploring the influence of medical staffing and birth volume on observed-to-expected cesarean deliveries: a panel data analysis of integrated obstetric and gynecological departments in Germany

Stöcker, Arno; Pfaff, Holger; Scholten, Nadine; Kuntz, Ludwig Article — Published Version Exploring the influence of medical staffing and birth volume on observed-to-expected cesarean deliveries: a panel data analysis of integrated obstetric and gynecological departments in Germany The European Journal of Health Economics Provided in Cooperation with: Springer Nature Suggested Citation: Stöcker, Arno; Pfaff, Holger; Scholten, Nadine; Kuntz, Ludwig (2025) : Exploring the influence of medical staffing and birth volume on observed-to-expected cesarean deliveries: a panel data analysis of integrated obstetric and gynecological departments in Germany, The European Journal of Health Economics, ISSN 1618-7601, Springer, Berlin, Heidelberg, Vol. 26, Iss. 6, pp. 987-1022, https://doi.org/10.1007/s10198-024-01749-0 This Version is available at: https://hdl.handle.net/10419/330433 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Vol.:(0123456789) The European Journal of Health Economics (2025) 26:987–1022 https://doi.org/10.1007/s10198-024-01749-0 ORIGINAL PAPER Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean deliveries: apanel data analysis ofintegrated obstetric andgynecological departments inGermany ArnoStöcker1,4,5 · HolgerPfaff1,4· NadineScholten2,4,5· LudwigKuntz3,4 Received: 18 September 2024 / Accepted: 3 December 2024 / Published online: 21 January 2025 © The Author(s) 2025 Abstract Introduction Cesarean deliveries account for approximately one-third of all births in Germany, prompting ongoing discussions on cesarean section rates and their connection to medical staffing and birth volume. In Germany, the majority of departments integrate obstetric and gynecological care within a single department. Methods The analysis utilized quality reports from German hospitals spanning 2015 to 2019. The outcome variable was the annual risk-adjusted cesarean section ratio—a metric comparing expected to observed cesarean sections. Explanatory variables included annual counts of physicians, midwives, and births. To account for case number-related staffing variations, full-time equivalent midwife and physician staff positions were normalized by the number of deliveries. Uniand multivariate panel models were applied, complemented by multiple instrument variable analyses, including two-stage least square and generalized method of moments models. Results Incorporating data from 509 integrated obstetric departments and 2089 observations, representing 2,335,839 deliveries with 720,795 cesarean sections (over 60% of all inpatient births in Germany), multivariate model with fixed effects revealed a statistically significant positive association between the number of physicians per birth and the risk-adjusted cesarean section ratio (0.004, p = 0.004). Two-stage least square instrument variable analysis (0.020, p < 0.001) and a system GMM estimator models (0.004, p < 0.001) validated these results, providing compelling evidence for a causal relationship. Conclusion The study established a robust connection between the number of physicians per birth and the risk-adjusted cesarean section ratio in integrated obstetric and gynecological departments in Germany. While the cause of the effect remains unclear, one possible explanation is a lack of specialization within these departments due to the combined provision of both obstetric and gynecological care. Keywords Longitudinal design· Obstetric care· Inpatient sector· Hospital· Organizational management· Medical staffing JEL Classification C23· I10· N34· P46 * Arno Stöcker arno.stoeck[email protected] 1 Faculty ofHuman Sciences andFaculty ofMedicine andUniversity Hospital Cologne, Institute ofMedical Sociology, Health Services Research andRehabilitation Science, Chair ofQuality Development andEvaluation inRehabilitation, University ofCologne, Cologne, Germany 2 Faculty ofMedicine andUniversity Hospital Cologne, Institute ofMedical Sociology, Health Services Research andRehabilitation Science, Chair ofHealth Services Research, University ofCologne, Cologne, Germany 3 Department ofBusiness Administration andHealth Care Management, Faculty ofManagement, Economics andSocial Sciences, University ofCologne, Cologne, Germany 4 Center forHealth Services Research Cologne, Interfaculty Institution oftheUniversity ofCologne, Cologne, Germany 5 Center forHealth Communication andHealth Services Research, Department forPsychosomatic Medicine andPsychotherapy, Faculty ofMedicine, University Hospital Bonn, Bonn, Germany 988 A.Stöcker et al. Introduction Over the past few decades, the global prevalence of cesarean section (C-section) births has steadily increased [1]. Some experts and researchers go so far as to characterize this trend as an endemic or pandemic of C-sections [2, 3]. This is noteworthy given that vaginal delivery is generally regarded as the preferred method compared to cesarean birth [4]. In Germany, the C-section rate has doubled since 1991 and has stabilized at approximately 30% since the mid-2000s [5]. Thereby, indications for a cesarean section can be divided into absolute and relative indications. This increase is predominantly attributed to relative indications, such as breech presentation, birth arrest, impending fetal hypoxia, and post-section conditions [6], with 90% of C-sections lacking absolute medical indications [7]. Thus, understanding non-medical factors, including organizational influences, is imperative for a comprehensive study of delivery methods. Given the imperative for health care providers to align their actions with medical necessities and patient preferences, comprehending the factors influencing medical practice decisions is crucial. In the case of C-sections, obviously maternal desire is a key factor [6, 8–10]. From the provider's perspective, numerous factors influence the choice of birth method. These encompass individual factors of the physician [3, 8, 11–14] or midwife [15, 16], as well as organizational factors at the hospital or obstetric department level [13, 17–22]. Additionally, context-related factors, such as the complexities of caseload [23], contribute to the decision-making process. Studies on C-section rates and the quality of obstetric care reveal organizationspecific differences linked to factors such as hospital ownership, number of beds, and teaching activity, both in Germany [17–19, 24, 25] and internationally [26–30]. While variations between departments with different organizational factors have been extensively studied, understanding relationships and interactions within obstetric departments is equally vital. These inter-organizational influences can potentially impact all departments uniformly, independent of specific organizational factors such as culture, region, socio-demographics, and patient population effects. From an organizational standpoint, comprehending these influences is crucial for making informed decisions about medical practices. Organizations can enhance their understanding of the factors shaping medical practice within their purview and implement adjustments accordingly [31]. This analysis focuses on two organizational factors, namely, medical staffing and caseload volume, and explores their interplay. Both factors have been acknowledged for their influence on the quality and quantity of obstetric care in general [27, 32] and specifically on C-sections (e.g., volume of births [33, 34] or the number of physicians/deliveries per physician per year [35–37]). Overall, medical staffing levels have been less frequently investigated than birth volume as an explanatory variable. Our study enhances current understanding through a longitudinal investigation that encompasses a significant portion of German obstetric departments and deliveries. Importantly, by employing an externally evaluated riskadjusted C-section ratio, our analysis addresses methodological limitations by accounting for medical indications on the part of both mother and child. Consequently, the ratio and the analysis remain robust, mitigating biases induced by underlying medical reasons that may prompt a C-section [38]. Our research question delved into how the number of full-time equivalent physicians and midwives per birth, coupled with the volume of births, impact the risk-adjusted C-section ratio in German combined obstetric and gynecological departments. In doing so, we address methodological shortcomings by focusing on differences within hospital departments, contributing to the understanding of factors independent of variances between hospital departments. Utilizing a panel model data set with individual and time effects, we control for departmentand patientpopulationspecific characteristics, ensuring a comprehensive analysis [39]. Methods Data source This analysis was based on the structured quality reports of German hospitals, which serve as a cornerstone of transparency in the inpatient sector in Germany [40–44]. These reports are compiled and disclosed in accordance with Sects. 136 and 137of the German Social Security Code (Sozialgesetzbuch) V. They offer a comprehensive overview of the structures, services, and quality of the respective hospital and its specialist departments. Annual reports include documentation of performance, organizational metrics, and various quality indicators and scores. Details are delineated to the department level. Although this data set has been utilized for cross-sectional analyses of various indicators [18, 40–42, 45, 46], its application in the context of longitudinal analyses with a specific focus on obstetrics is novel. During the study period, approximately 30 quality indicators related to obstetric care were published (Appendix Table7). Our analysis centers on the quality indicator for risk-adjusted C-section rates per hospital. In preparation for this analysis, a panel data set was compiled using annually published quality reports. In these 989 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… reports each hospital is assigned a unique identification code, and each site within a hospital is designated a site number. Aspreviously outlined [47, 48], identification codes and site numbers may undergo changes over the years. Therefore, automated linking without content verification could yield a success rate ranging from 80 to 90%. Recognizing potential changes in identification codes and site numbers over the years, a manual matching process was employed to ensure accuracy, supplementing a machine linkage via identification codes and site numbers. The linkage incorporated hospital and site addresses, bed numbers, and the names of responsible department and general hospital managers. Two independent researchers performed the manual linkage, resolving disputed assignments through consensus. Each hospital and site received a master identification number, forming the foundation for the panel study. As this analysis employs secondary data, and the data are publicly available, no ethical committee vote was deemed necessary. Study population In Germany, the vast majority of deliveries occur in hospitals, with less than two percent taking place at home or in birthing centers. Obstetric care is predominantly administered in departments specializing in both obstetrics and gynecology simultaneously. Consequently, two closely related but increasingly distinct medical services are offered within the same departments. Given the inherent organizational disparities between those integrated obstetrics departments and departments primarily dedicated on obstetric care, we excluded obstetric departments with a primary focus on obstetric care from our analysis. Importantly, there are no systematic differences between these two types of departments in terms of patient population or medical standards. The cut-off value was determined through a data-driven approach, as information on organizational focus was not consistently or continuously included in the quality reports. Additionally, departments with attending physicians were excluded due to their typically higher C-section rates in Germany [17, 18, 25]. Ensuring data integrity, we identified potential outliers through Cook's distance and subsequently verified them manually. Closed departments were confirmed through cross-referencing with press reports. The exclusion was deemed necessary to avoid inaccuracies resulting from scenarios where a department, operational for only six months, reported staffing numbers analogously for the entire year. Consequently, staffing figures were not proportionately reported in the quality report, and the number of deliveries was only accounted for during the specified 6-month period. This meticulous approach was adopted to maintain the accuracy and reliability of the dataset. Study period The observation period spans from 2015 to 2019, as the relevant quality indicator on the cesarean section ratio was introduced in 2015, and 2019 marks the last year before the onset of the COVID-19 pandemic. Recognizing the potential impact of measures associated with the pandemic on daily hospital practices [49, 50], data from 2020 onwards was excluded to minimize bias. Moreover, the introduction of the Robson indicator to the quality indicator in 2020 [51] further complicates longitudinal comparisons. Measures Outcome variable The dependent variable in the study was the risk-adjusted cesarean section ratio at the department level (quality indicator no. 52249). This ratio is calculated as the observed number of C-sections divided by the expected number of C-sections. The declared quality objective is the minimization of cesarean births [51]. The numerator includes all observed cesarean deliveries within a department, while the denominator comprises risk-adjusted expected cesarean deliveries. In 2015 and 2016, all mothers with at least one child born after 24weeks were included, while from 2017 onwards, only mothers delivering between weeks 24 and 42 were considered. Risk adjustment is performed by the Institute for Quality Assurance and Transparency in Healthcare (Institut für Qualitätssicherung und Transparenz im Gesundheitswesen [IQTIG]). While individual patient data are considered in this process, only aggregated, annual departmentlevel data were accessible for this analysis open access. Notably, the researchers involved in this study were not directly engaged in the risk adjustment process. The selection of risk factors was guided by Becker and Eissler [52] in collaboration with the Federal Perinatal Medicine Group. Risk factors considered encompassed variables such as maternal age, comprehensive data on infant and maternal health status, and information on previous deliveries (Appendix Table22). Each year, these risk factors were adjusted based on their respective regression coefficients. Consequently, the risk adjustment undertaken compensates for the divergent patient structures across different facilities, offering a more equitable basis for facility comparisons. This adjustment proves pivotal in ensuring a fair assessment, as patients bring individual risk factors, including concomitant diseases, that could systematically influence the quality outcome. Through risk adjustment, institutions with a higher 990 A.Stöcker et al. prevalence of high-risk cases can be statistically juxtaposed more equitably with those handling a larger proportion of low-risk cases, thereby facilitating an unbiased analysis [53]. In the context of the risk-adjusted C-section ratio, values below 1 indicate that a department is performing fewer cesarean sections than expected. Conversely, a ratio above 1 signifies that more risk-adjusted cesarean sections are being performed than anticipated. For instance, a C-section ratio of 1.1 would imply that 10% more cesarean sections were performed than expected. Furthermore, a reference category is defined for each year, designating departments above the 90th percentile as conspicuous. The corresponding value for the review period ranged from 1.23 to 1.27. Explanatory variables Three independent variables were utilized: (1) the number of deliveries per department (mothers giving birth) per 1000, (2) the number of full-time equivalent physicians in the department per 1000 deliveries, (3) and the number of fulltime equivalent midwives in the department per 1000 deliveries. Staffing levels have been divided by 1000 to account for volume related differences in the departments. The data on deliveries were extracted from the denominator of the quality indicator for each department. The data on both staffing levels were presented separately for each department within a hospital. Staffing data for physicians were aggregated for residents and specialists, collectively referred to as physicians in subsequent discussions. Physician assistants were excluded. As these departments integrate obstetrics and gynecology, physicians represent both specialties. In Germany, the ‘Facharztstandard’ (specialist standard) ensures high-quality medical treatments and procedures, typically carried out or supervised by a specialist. However, sufficiently trained residents or assistant physicians may also perform treatments and procedures independently, including cesarean sections. Since there is no strict threshold for determining when a physician has gained sufficient experience and knowledge, we opted to include the total number of physicians in our variable. In addition, a midwife has to be present during labor in Germany (‘Hinzuziehungspflicht’; midwife's obligation to consult). Control variables Three organizational variables served as control variables in the analyses: (1) hospital ownership (non-profit, private, public), (2) teaching status (no teaching assignment, academic teaching hospital, university hospital), and (3) perinatal care level (regular obstetric departments [care level 4], perinatal focus [care level 3], perinatal center level II [care level 2], perinatal center level I [care level 1]). Regular obstetric departments provide standard perinatal care from 36 + 0weeks of gestation without anticipated complications. Departments with a perinatal focus cater to pregnant women expecting premature infants with an estimated birth weight of at least 1500g or with a gestational age from 32 + 0 to less than or equal to 35 + 6weeks. Level II perinatal centers serve pregnant women with anticipated premature infants weighing between 1250 and 1499g or with a gestational age from 32 + 0 to less than or equal to 35 + 6weeks. Level I perinatal centers offer the highest level of obstetric care for pregnant women expecting premature infants with an estimated birth weight under 1250g or with a gestational age from 29 + 0 to less than weeks. All three organizational variables were frequently cited as influential factors in C-section rates [54, 55]. Model description We developed various static and dynamic panel models featuring timeand individual-specific effects (two-way effects), employing cluster-robust estimators for each department: (a) Fixed effects estimator models (b) Correlated random effects estimator models (c) Two stage least square estimator models (d) System generalized method of moments estimator (Blundell/Bond [56] and Arellano/Bover [57]) models where yit = vector of dependent variable, xit = vector of independent variables, xi = vector of cluster means independent variables, xit = vector of estimated independent variables from the first stage least square regression, zit = vector of (time constant) independent yit = 𝛽0 + 𝛽1xit + 𝛾t + 𝛼i + eit yit = 𝛽0 + 𝛽1xit + 𝛽2xi + 𝛽3zit + 𝛾t + 𝛼i + eit yit = 𝛽0 + 𝛽1 x it + 𝛽2 z it + 𝛾t + 𝛼i +e it xit = 𝛿0 + 𝛿1 i it + 𝛿2 z it + 𝜓t + 𝜙i + 𝜐it Model:yit =𝛽0+𝛽1xit−1+𝛽2yit−1+𝛽3zit +𝛾t+𝛼i+eit Difference:y it −y it−1 = 𝛽1 (x it−1 −x it−2 ) +𝛽2(yit−1−yit−2) + 𝛽3( z it −z it−1) + ( e it −e it−1) Level:y it =𝛽 0 +𝛽 1 x it−1 +𝛽 2 y it−1 + 𝛽3 z it−1 + 𝛾t + 𝛼i +e it 991 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… control variables, 𝛽 = vector of model coefficients, 𝛾t = unobserved time-specific effect, 𝛼i = unobserved department-specific effect, eit = individual error term, 𝛿 = vector of parameters to be estimated in the first stage least square regression, iit = instrument for independent variable, 𝜓t = unobserved time-specific effects in the first stage, 𝜙i = unobserved department-specific effects in the first stage, 𝜐it = error term in the first stage. Where entities (departments) are denoted as i = 1, … ,n and observation periods (years) t = 2015, … , 2019 . The term 𝛾t incorporated the time effect on the dependent variable, independent of observable or unobservable differences between individual observation units. This temporal effect captured the influence of changes over time on the dependent variable and served as a global temporal component. 𝛼i was the fixed unobserved heterogeneity of each hospital department, and eit signified the error term for each hospital department over time. Standard errors robust for group-wise heteroscedasticity and serial correlation were used. To choose between fixed effects or random effects models, we utilized a robust Hausmanlike test. Organizational characteristics in the German hospital sector exhibit minimal variation and remain relatively time-persistent. Including them in a fixed effect model would render the results valid only in the rare event of a shift in one of the categories, making it unsuitable for a true comparison between different organizational factors. Consequently, alongside a model with fixed effect estimators, we adopted a correlated random effects (CRE) modeling approach [58] to incorporate and control for other organizational variables, enhancing model sensitivity and specificity. Acknowledging potential unmeasured confounders and endogeneity with the explanatory variables in our panel regression models, two models with instrument variables (IV) were constructed and analyzed to address possible endogeneity. While the panel structure of the data already accounted for some aspects of endogeneity [59], models with instrumental variable estimation were employed to establish potential causal effects. Recommended best practices advocate for the inclusion of additional data to address endogeneity before resorting to IV estimation [58–61], our data source had limitations in providing meaningful potential additional variables. The first model employed a static IV approach with a two-stage least squares (2SLS) estimator, using the number of nursing staff per 1000 deliveries as an external IV. The second model employed a dynamic approach with generalized method of moments (GMM) estimators, incorporating the lagged ratios of observed to expected rates of cesarean births as an internal IV [56, 57, 62]. Additionally, to minimize data loss due to the unbalanced dataset and use more instruments for more efficient estimators, we referred to the system GMM estimator instead a difference GMM estimator [56]. Two-step GMM estimators were chosen for their robustness to autocorrelation and heteroscedasticity [63]. Statistical analysis Data preparation (tidyverse package [2.0.0]) and analysis (plm package [2.6–3], lmtest package [0.9–40], gtsummary package [1.7.2], modelsummary package [1.4.3]) were performed in R (version 4.2.2) and R Studio (version 2023.06.1 + 524). Results Data inclusion Following the exclusion of duplicate entries, we identified data from 912 departments performing obstetric care, providing 3627 observations from the quality reports (Fig.1). Regulatory authorities censored 48 observations due to data privacy concerns (less than four deliveries in the reporting year), these were excluded from further analysis. Additionally, 166 observations lacked data on the C-section ratio quality indicator for the respective year, leading to their exclusion. Of the remaining data, 360 reported having attending physicians in the respective year, and these departments were excluded. Similarly, departments primarily offering obstetric services were excluded, with a data-driven cut-off value set at 1.4 times the number of full inpatient cases compared to births (Appendix Fig.2). 377 observations fell below the threshold and were categorized as solo obstetric departments, subsequently excluded. A manual check using Cook's distance method revealed 60 entries with conspicuous data, confirmed through cross-checking with media reports indicating closures within the reporting year. These departments were consequently excluded. Lastly, 527 observations lacked data on the variable for the number of midwives, resulting in their exclusion. The final analysis included 2089 observations from 519 obstetric departments. In total, the study population with 2089 observations represented 2,335,839 mothers giving birth and 720,795 C-section deliveries (Table1). The panel is unbalanced. As the study population no longer corresponded to a full survey of German obstetric departments, we validated the study population against data from IQTIG for the overall numbers on births and cesarean deliveries in obstetric departments in Germany. Over the period 2015 to 2019, the total number of mothers giving births increased from 713,563 to 745,941, with a slight decrease in the C-section rate from 31.42 to 30.85%. These trends were mirrored in the smaller study 992 A.Stöcker et al. Fig. 1 Flow chart of study population 993 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… population (444,555–479,176 births). The study population covered 62.8% of all delivering mothers and 62.4% of all cesarean deliveries in Germany. While the year-by-year C-section rate in the study population was slightly lower than the national rate for all years except 2015, the differences were not significant (t test, p = 0.408). Notably, the deviation in 2016 was attributed to missing data in the quality reports, discussed in more detail in the limitations section. Additionally, solo obstetric departments reported on average cesarean rates (30.3%) similar to those of integrated departments (Appendix Table8). Characteristics ofstudy population The observed risk-adjusted C-sections, on average, were marginally below the expected number, indicating a ratio of 0.98 (median 1.00) (Table2). The number of deliveries per department averaged 1118.2 births. Notably, from 2015 to 2019, the mean number of deliveries exhibited an increase of 123.1, rising from 1031.5 to 1154.6 births, with an intermediate spike observed from 2015 to 2016. However, the median, standing at 913, consistently trailed the mean. Regarding cesarean births, the annual average per department was 345, showing a modest increase of 21.7 from 2015 to 2016. Yet again, the median, at 240.3 cesarean births, lagged behind the mean. The average number of physicians per department over the observation period was 12.5, with a median of 10.9. This represents an increase of more than one full-time equivalent from 2015 to 2019. As outlined earlier, the total number of full-time equivalent physicians was divided by the department's number of births to ensure comparability across departments. The average number of full-time equivalent physicians per 1000 deliveries per department remained constant around 12.5, with a median of 11.8, showing no significant change from 2015 to 2019. In terms of midwives, the average number per department hovered around 11 during the observation period, with a median of 9.2. The average number of full-time equivalent midwives per 1000 deliveries per department was 10.7 (median 10.3). However, from 2015 to 2019, average number of full-time equivalent midwives per 1000 deliveries decreased by nearly one. The average number of full-time equivalent nursing staff (excluding midwives) per department was 24, with a median of 19.6. Per 1000 deliveries, there were 23.3 full-time equivalent nursing staff, with a median of 21.1. Notably, there was a decrease in the average number of full-time equivalent nursing staff per 1000 births from 24.7 in 2015 to 22.3 in 2019. Private hospitals constituted the minority in the study population at 16.6%, while non-profit hospitals (36.3%) and public hospitals (47.1%) comprised the majority. Regarding academic teaching, 70.5% of hospitals were listed as academic teaching hospitals, 7.6% were university hospitals, and 22% were not engaged in academic teaching. Beyond regular obstetric departments, the landscape encompasses facilities specializing in perinatal care, including level 1 and 2 perinatal centers. Additionally, 45% categorized as departments with the regular level of perinatal care. 19.4% reported a perinatal focus, 8.9% were a level II perinatal center, and 26.8% were a level I perinatal center—the highest form of obstetric health services for high intense care. Compared to integrated obstetric departments, solo departments, on average, have a significantly lower riskadjusted cesarean section ratio (0.95; t test, p = 0.001), more deliveries (1563.5; t test, p < 0.001), and more cesarean sections (473.7; t test, p < 0.001), but fewer full-time equivalent physicians (9.7; 6.8 per 1000 deliveries; t test, p < 0.001) and more midwives (13.4; 9.1 per 1000 deliveries; t test, p < 0.001). The crude cesarean section rate did not differ significantly from each other (mean solo department: 30.04%, mean integrated department: 30.41%; t test: p value = 0.371). Solo departments were more likely to be involved in academic teaching (80.1%) and to have a higher perinatal care level (care level 4 = 33.1%; care level 1 = 41.2%) (Appendix Table9 for more details). Table 1 Comparison of births and cesarean sections between total German hospital population and study population IQTIG Study population Deliveries (all mothers who have had at least one birth of a child) Cesarean deliveries C-section rate Deliveries (all mothers who have had at least one birth of a child) Cesarean deliveries C-section rate 2015 713,563 224,197 31.42 444,555 139,940 31.48 2016 753,289 235,096 31.21 420,158 130,253 31.00 2017 756,146 235,765 31.18 492,692 151,074 30.66 2018 749,024 229,676 30.66 499,258 152,461 30.54 2019 745,941 230,105 30.85 479,176 147,067 30.69 Σ/Ø 3,717,963 1,154,839 31.06 2,335,839 720,795 30.87 994 A.Stöcker et al. Table 2 Descriptive description study population a n (%); mean/median (SD) Characteristic Overall, N = 208912015, N = 43112016, N = 37612017, N = 43012018, N = 43712019, N = 4151 Risk-adjusted cesarean ratio 0.98/1.00 (0.18) 0.97/0.97 (0.19) 0.98/0.99 (0.18) 0.98/1.00 (0.19) 0.99/1.00 (0.18) 0.99/1.01 (0.18) Number of deliveries 1118.16/913.00 (695.53) 1031.45/828.00 (641.20) 1117.44/905.50 (701.29) 1145.80/934.00 (708.85) 1142.47/936.00 (702.03) 1154.64/950.00 (719.08) Number of C-sections 345.04/263.00 (240.27) 324.69/250.00 (229.90) 346.42/267.50 (240.73) 351.33/266.00 (241.12) 348.88/265.00 (242.04) 354.38/265.00 (247.48) Number of fulltime equivalent physicians 12.48/10.86 (7.10) 11.75/10.33 (6.57) 12.35/10.73 (7.30) 12.56/11.09 (7.12) 12.74/11.00 (7.17) 13.01/11.14 (7.32) Number of fulltime equivalent physicians per 1000 deliveries 12.46/11.75 (4.45) 12.85/11.84 (4.90) 12.32/11.64 (4.54) 12.22/11.50 (4.36) 12.40/11.84 (4.21) 12.51/11.81 (4.17) Number of fulltime equivalent midwives 11.03/9.17 (7.56) 10.80/9.00 (6.80) 11.11/9.40 (7.56) 11.15/9.16 (7.59) 10.95/9.04 (7.71) 11.16/9.48 (8.13) Number of fulltime equivalent midwives per 1000 deliveries 10.64/10.30 (4.53) 11.38/10.67 (4.35) 10.60/10.14 (4.09) 10.53/10.22 (4.86) 10.25/10.16 (4.31) 10.45/10.37 (4.88) Number of fulltime equivalent nursing staff 23.97/19.60 (17.03) 23.33/19.80 (15.86) 24.60/19.70 (18.54) 24.30/19.65 (16.97) 24.00/19.68 (17.12) 23.64/19.30 (16.68) Missing 49 40 2 2 3 2 Number of fulltime equivalent nursing staff per 1000 deliveries 23.34/21.06 (11.04) 24.69/22.25 (12.18) 23.99/21.90 (11.63) 23.19/21.17 (10.75) 22.76/20.54 (10.20) 22.27/20.12 (10.38) Missing 49 40 1 2 3 2 Ownership Non-profit 759 (36.33%) 166 (38.52%) 123 (32.71%) 165 (38.37%) 159 (36.38%) 146 (35.18%) Private 346 (16.56%) 72 (16.71%) 67 (17.82%) 67 (15.58%) 72 (16.48%) 68 (16.39%) Public 984 (47.10%) 193 (44.78%) 186 (49.47%) 198 (46.05%) 206 (47.14%) 201 (48.43%) Teaching status No teaching assignment 459 (21.97%) 97 (22.51%) 90 (23.94%) 98 (22.79%) 89 (20.37%) 85 (20.48%) Academic teaching hospital 1472 (70.46%) 309 (71.69%) 256 (68.09%) 298 (69.30%) 313 (71.62%) 296 (71.33%) University hospital 158 (7.56%) 25 (5.80%) 30 (7.98%) 34 (7.91%) 35 (8.01%) 34 (8.19%) Perinatal care level Regular obstetric department (care level 4) 939 (44.95%) 201 (46.64%) 164 (43.62%) 193 (44.88%) 195 (44.62%) 186 (44.82%) Perinatal focus (care level 3) 405 (19.39%) 73 (16.94%) 73 (19.41%) 87 (20.23%) 90 (20.59%) 82 (19.76%) Perinatal centers level II (care level 2) 185 (8.86%) 44 (10.21%) 36 (9.57%) 36 (8.37%) 36 (8.24%) 33 (7.95%) Perinatal centers level I (care level 1) 560 (26.81%) 113 (26.22%) 103 (27.39%) 114 (26.51%) 116 (26.54%) 114 (27.47%) 1001 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… the quality reports proved unfeasible. Consequently, the data for the year 2016 contain a bias due to the absence of these departments. Remarkably, a considerable number of missing values pertaining to midwives were noted, despite regulatory mandates requiring their presence at all hospital births in Germany. The imperative for a midwife's attendance extends to all hospital births, including cesarean sections (§ 4 Abs 1.-Hebammengesetz). However, certain departments reported the presence of attending midwives, leading to the exclusion from the study. A notable percentage of hospitals indicated either a complete absence of midwives or reported them outside obstetric departments. A recurring observation was that departments slated for closure in the subsequent year failed to furnish any data on medical staff for the preceding year. This phenomenon can be attributed to the time lag of over a year in the data collection and compilation process. The overall data quality in this aspect is suboptimal, potentially serving as a source of bias. Finally, it is crucial to recognize that healthcare, particularly obstetric care, is significantly influenced by cultural, policy, and local factors [8, 33, 89, 90]. The complex and divergent nature of the relationship between these factors underscores the challenge of drawing general and crossnational conclusions from the study results [37]. Implications forpractice The findings underscore that within integrated gynecological and obstetric departments, a higher number of physicians per delivery significantly correlated with more observed riskadjusted C-sections than expected. The dual provision of gynecological and obstetric care in one department poses a challenge, potentially contributing to an elevated C-section ratio with an increased number of physicians per birth as each physician may be less acquainted with obstetric care. As a practical approach, directing attention towards focus and specialization within the medical team could yield a reduction in the C-section ratio, potentially involving intradepartment staff reallocation allowing for a more specialized workforce between the two areas of responsibility. This approach does not necessarily imply reducing the total number of physicians but rather enhancing specialization between obstetrics and gynecology care. An insightful analysis of departments surpassing the reference value (90th percentile) reveals that these departments (n = 118) exhibit an average of 15.1 physicians per 1000 deliveries, accompanied by 872.5 births and 358.9 C-sections (41.1%). In contrast, inconspicuous departments (n = 1971) exhibit an average of 12.3 physicians per 1000 deliveries, along with 1132.9 births and 344.2 C-sections (30.4%). A targeted reduction of physicians per 1000 births by 3, aligning with the inconspicuous department average, could anticipate a modest reduction of 0.012 in the observed expected C-sections, based on our results. On the other hand, departments in the 10th percentile (n = 217) deployed on average 10.9 physicians per 1000 deliveries, accompanied by 1216.5 births and 259.7 C-sections (21.3%). Importantly, as we estimated within-effects the effect size is smaller as a between comparison would suggest as they incorporate department individual effects. As for departments that already have low C-section rates (and a potential underperformance of C-sections) it is unclear whether a further increase in the number of deliveries per physician would lead to improved quality of care. These implications suggest that a nuanced approach to physician staffing, coupled with a strategic focus on specialization and a potential realignment of resources, could contribute to achieve the objective of the quality indicator of minimizing C-sections within integrated obstetric and gynecological departments. If the goal, as articulated in the quality indicator, is to minimize the occurrence of cesarean section births, it is essential to consider additional factors beyond the number of physicians per birth presented in this analysis. The assessment must also consider the preferences of the woman giving birth and relevant medical considerations, ensuring that these crucial aspects are not disregarded. It is vital that a narrow focus on reducing the cesarean section rate does not compromise the health of women and infants. Conclusion Our examination of integrated German obstetric and gynecological departments revealed a noteworthy and robust positive association between the number of physicians per birth and an elevated ratio of observed-to-expected cesarean sections. Additionally, instrumental variable analysis indicated a potential causal effect. However, given the reliance on annual, aggregated averages and the inherent uncertainties despite the instrumental analysis employed, a cautious approach is essential when interpreting causality. Preliminary interpretations suggest that specialization, as indicated by the number of deliveries per physician, may influences the cesarean section ratio within integrated obstetric and gynecological departments. Appendix See Tables6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 and Fig.2. 1002 A.Stöcker et al. Table 6 Dynamic IV models with system generalized method of moment estimators Ratio of observed to expected (O/E) cesarean births (2015– 2019) (two-way fixed effects model) Univariate model Multivariate model Multivariate model with control variables Lagged ratio of observed to expected ratio (O/E) of cesarean births 0.485*** 0.503*** 0.489*** p value (< 0.001) (< 0.001) (< 0.001) 95% CI [0.355, 0.616] [0.374, 0.633] [0.383, 0.596] SE (0.068) (0.069) (0.054) Number of physicians per 1000 deliveries 0.004*** 0.004*** 0.004** p value (< 0.001) (< 0.001) (0.002) 95% CI [0.002, 0.006] [0.002, 0.007] [0.001, 0.006] SE (0.001) (0.001) (0.001) Number of midwives per 1000 deliveries − 0.002 − 0.002 p value (0.164) (0.105) 95% CI [− 0.004, 0.001] [− 0.004, 0.000] SE (0.001) (0.001) Number of deliveries per 1000 − 0.010+− 0.037*** p value (0.095) (< 0.001) 95% CI [− 0.022, 0.002] [− 0.054, − 0.020] SE (0.006) (0.009) Ownership: public (ref. category: ownership private) − 0.015 p value (0.225) 95% CI [− 0.039, 0.009] SE (0.013) Ownership: non-profit (ref. category: ownership private) − 0.002 p value (0.906) 95% CI [− 0.027, 0.024] SE (0.014) Perinatal centers level I (care level 1) (ref. category: regular obstetric department (care level 4)) 0.051*** p value (< 0.001) 95% CI [0.024, 0.078] SE (0.014) Perinatal centers level II (care level 2) (ref. category: regular obstetric department (care level 4) 0.016 p value (0.304) 95% CI [− 0.014, 0.045] SE (0.016) Perinatal focus (care level 3) (ref. category: regular obstetric department (care level 4) − 0.010 p value (0.355) 95% CI [− 0.033, 0.012] SE (0.012) Teaching status: academic teaching hospital (ref. category: teaching status: no teaching assignment) 0.013 p value (0.224) 95% CI [− 0.008, 0.034] SE (0.011) Teaching status: University Hospital (ref. category: teaching status: no teaching assignment) 0.014 p value (0.263) 1003 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Table 6 (continued) Ratio of observed to expected (O/E) cesarean births (2015– 2019) (two-way fixed effects model) Univariate model Multivariate model Multivariate model with control variables 95% CI [− 0.010, 0.037] SE (0.013) Num. obs 2520 2520 2520 Hansen–Sargan test/J test (p value) 6.983 (0.639) 11.725 (0.385) 12.310 (0.831) Arellano–bond test/autocorrelation test (1) (p value) − 6.094 (< 0.001) − 6.058 (< 0.001) − 6.321 (< 0.001) Arellano–bond test/autocorrelation test (2) (p value) − 0.101 (0.919) − 0.094 (0.925) − 0.253 (0.800) Wald test for coefficients (p value) 110.125 (< 0.001) 148.589 (< 0.001) 321.429 (< 0.001) Wald test for time dummies (p value) 3.130 (0.372) 3.084 (0.379) 4.702 (0.195) Std. errors HC1 HC1 HC1 + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 1004 A.Stöcker et al. Table 7 Obstetric quality indicators for the period 2015–2019 QI-ID Indicator description Years Quality goal 330 Antenatal corticosteroid therapy for preterm births with a prepartum inpatient stay of at least two calendar days 2015–2019 Frequent initiation of antenatal corticosteroid therapy (lung maturation induction) in births with a gestational age of 24 + 0 to under 34 + 0weeks, excluding stillbirths, and with a prepartum inpatient stay of at least two calendar days 50,046 Administration of antibiotics in cases of premature rupture of membranes 2015 Not listed in 2015 50,045 Perioperative antibiotic prophylaxis in cesarean section deliveries 2015–2019 High rate of perioperative antibiotic prophylaxis in cesarean section deliveries 52,243 Cesarean births 2015 Not listed in 2015 52,249 Ratio of observed to expected rate (O/E) of cesarean births 2015–2019 Low rate of cesarean births 1058 Decision-to-delivery interval over 20min in emergency cesarean sections 2015–2019 Rarely a decision-to-delivery interval of more than 20min in emergency cesarean sections 319 Measurement of umbilical artery pH in singleton live births 2015 Not listed in 2015 321 Acidosis in mature singletons with umbilical artery pH measurement 2015–2019 Low rate of acidosis in singleton live births with umbilical artery pH measurement 51,397 Ratio of observed to expected rate (O/E) of acidosis in mature singletons with umbilical artery pH measurement 2015–2019 Low rate of acidosis in singleton live births with umbilical artery pH measurement 51,826 Acidosis in preterm singletons with umbilical artery pH measurement 2015 Not listed in 2015 51,831 Ratio of observed to expected rate (O/E) of acidosis in preterm singletons with umbilical artery pH measurement 2015–2019 Low rate of acidosis in singleton live births with umbilical artery pH measurement 318 Presence of a pediatrician at preterm births 2015–2019 Frequent presence of a pediatrician at the birth of preterm live births with a gestational age of 24 + 0 to under 35 + 0weeks 1059 Critical outcome in mature newborns 2015 Not listed in 2015 51,803 Quality index for critical outcome in mature newborns 2015–2019 Rare occurrences of child deaths, 5-min Apgar score below 5, pH below 7, and Base Excess < − 16 in mature newborns 51808_51803 Level 1: Ratio of observed to expected rate (O/E) of child deaths 2018, 2019 Not specified 51813_51803 Level 2: Ratio of observed to expected rate (O/E) of children with a 5-min Apgar score below 5 2018, 2019 Not specified 51818_51803 Level 3: Ratio of observed to expected rate (O/E) of children with Base Excess below − 16 2018, 2019 Not specified 51823_51803 Level 4: Ratio of observed to expected rate (O/E) of children with acidosis (pH < 7.00) 2018, 2019 Not specified 322 Thirdor fourth-degree perineal tear in spontaneous singleton deliveries 2015 Not specified 51,181 Ratio of observed to expected rate (O/E) of thirdor fourth-degree perineal tears in spontaneous singleton deliveries 2015–2017 Low number of mothers with thirdor fourth-degree perineal tears in spontaneous singleton deliveries 323 Thirdor fourth-degree perineal tear in spontaneous singleton deliveries without episiotomy 2015 Not listed in 2015 324 Thirdor fourth-degree perineal tear in spontaneous singleton deliveries with episiotomy 2015 Not listed in 2015 52,244 Mothers and children discharged home together 2015 Not listed in 2015 52,254 Ratio of observed to expected rate (O/E) of mothers and children discharged home together 2015 Not listed in 2015 1005 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Table 7 (continued) QI-ID Indicator description Years Quality goal 331 Maternal mortality in the context of perinatal surveys 2015–2019 Rare occurrences of maternal deaths 181,800 Quality index for fourth-degree perineal tears in singleton deliveries 2018, 2019 Low number of mothers with fourth-degree perineal tears in spontaneous or vaginal-assisted singleton deliveries 181801_181800 Level 1: Ratio of observed to expected rate (O/E) of fourth-degree perineal tears in spontaneous singleton deliveries 2018, 2019 Not specified 181802_181800 Level 2: Ratio of observed to expected rate (O/E) of fourth-degree perineal tears in vaginal-assisted singleton deliveries 2018, 2019 Not specified 1006 A.Stöcker et al. Table 8 Comparison of births and cesarean sections between integrated and solo obstetric departments Study population (integrated departments) Solo departments Deliveries (all mothers who have had at least one birth of a child) Cesarean deliveries C-section rate Deliveries (all mothers who have had at least one birth of a child) Cesarean deliveries C-section rate 2015 444,555 139,940 31.48 90,690 28,866 31.83 2016 420,158 130,253 31.00 93,824 27,489 29.30 2017 492,692 151,074 30.66 91,709 28,149 30.69 2018 499,258 152,461 30.54 98,282 28,996 29.50 2019 479,176 147,067 30.69 107,058 32,398 30.26 Σ/Ø 2,335,839 720,795 30.87 481,563 145,898 30.32 1007 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Table 9 Descriptive description solo obstetric departments a n (%); mean/median (SD) Characteristic Overall N = 308a2015 N = 61a2016 N = 57a2017 N = 61a2018 N = 62a2019 N = 67a Risk-adjusted cesarean ratio 0.95/0.95 (0.18) 0.97/0.99 (0.17) 0.92/0.92 (0.19) 0.95/0.96 (0.18) 0.94/0.94 (0.18) 0.95/0.96 (0.19) Number of deliveries 1563.52/1474.50 (709.62) 1486.72/1293.00 (716.81) 1646.04/1603.00 (785.32) 1503.43/1464.00 (634.92) 1585.19/1523.00 (681.82) 1597.88/1520.00 (734.48) Number of C-sections 473.69/437.00 (249.22) 473.21/436.00 (253.57) 482.26/437.00 (282.01) 461.46/433.00 (225.98) 467.68/426.50 (229.55) 483.55/448.00 (259.85) Number of fulltime equivalent physicians 9.66/8.34 (4.96) 9.28/8.25 (5.03) 9.73/8.60 (5.59) 9.19/8.04 (4.22) 9.74/8.56 (4.56) https://doi. org/10.32/8.34 (5.36) Number of fulltime equivalent physicians per 1000 deliveries 6.82/5.82 (3.31) 7.03/6.08 (3.68) 6.72/5.79 (3.52) 6.79/5.48 (3.33) 6.68/5.70 (3.03) 6.89/6.16 (3.09) Number of fulltime equivalent midwives 13.39/12.79 (7.38) 13.22/12.30 (8.20) 13.72/12.80 (8.02) 12.61/12.64 (6.77) 13.24/12.89 (6.85) 14.12/13.60 (7.17) Number of fulltime equivalent midwives per 1000 deliveries 9.07/9.20 (3.80) 9.57/9.19 (4.84) 8.95/8.86 (4.14) 8.70/9.29 (3.24) 8.79/9.08 (3.23) 9.32/9.52 (3.41) Number of fulltime equivalent nursing staff 17.65/15.88 (9.93) 18.05/16.00 (9.89) 18.81/15.80 (12.93) 16.39/15.14 (7.99) 17.97/16.66 (10.31) 17.13/16.19 (8.26) missing 1 0 0 0 0 1 Number of fulltime equivalent nursing staff per 1000 deliveries 12.17/10.93 (5.92) 13.51/12.29 (7.67) 12.23/10.54 (6.33) 11.41/10.94 (4.42) 12.25/10.58 (6.17) 11.53/11.09 (4.45) Missing 1 0 0 0 0 1 Ownership Non-profit 176 (57.14%) 34 (55.74%) 29 (50.88%) 34 (55.74%) 38 (61.29%) 41 (61.19%) Private 31 (10.06%) 6 (9.84%) 8 (14.04%) 5 (8.20%) 5 (8.06%) 7 (10.45%) Public 101 (32.79%) 21 (34.43%) 20 (35.09%) 22 (36.07%) 19 (30.65%) 19 (28.36%) Teaching status No teaching assignment 39 (12.66%) 8 (13.11%) 6 (10.53%) 10 (16.39%) 9 (14.52%) 6 (8.96%) Academic teaching hospital 247 (80.19%) 51 (83.61%) 47 (82.46%) 47 (77.05%) 47 (75.81%) 55 (82.09%) University hospital 22 (7.14%) 2 (3.28%) 4 (7.02%) 4 (6.56%) 6 (9.68%) 6 (8.96%) Perinatal care level Regular obstetric department (care level 4) 102 (33.12%) 18 (29.51%) 23 (40.35%) 19 (31.15%) 20 (32.26%) 22 (32.84%) Perinatal focus (care level 3) 31 (10.06%) 6 (9.84%) 4 (7.02%) 7 (11.48%) 7 (11.29%) 7 (10.45%) Perinatal centers level II (care level 2) 48 (15.58%) 11 (18.03%) 8 (14.04%) 10 (16.39%) 9 (14.52%) 10 (14.93%) Perinatal centers level I (care level 1) 127 (41.23%) 26 (42.62%) 22 (38.60%) 25 (40.98%) 26 (41.94%) 28 (41.79%) 1008 A.Stöcker et al. Table 10 Uniand multivariate panel models with two-ways random effects on risk-adjusted C-section ratio + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 Ratio of observed to expected (O/E) cesarean births (2015–2019) (two-way random effects model) Univariate models Multivariate model Intercept 0.893*** 0.964*** 1.022*** 0.918*** p value (< 0.001) (< 0.001) (< 0.001) (< 0.001) 95% CI [0.852, 0.934] [0.924, 1.004] [0.994, 1.051] [0.852, 0.984] SE (0.021) (0.020) (0.014) (0.034) Number of physicians per 1000 deliveries 0.007*** 0.007*** p value (< 0.001) (< 0.001) 95% CI [0.004, 0.010] [0.003, 0.010] SE (0.002) (0.002) Number of midwives per 1000 deliveries 0.002 0.000 p value (0.327) (0.912) 95% CI [− 0.002, 0.005] [− 0.004, 0.004] SE (0.002) (0.002) Number of deliveries per 1000 − 0.036*** − 0.017 p value (< 0.001) (0.135) 95% CI [− 0.056, − 0.015] [− 0.039, 0.005] SE (0.010) (0.011) Num. obs 2089 2089 2089 2089 R20.061 0.000 0.019 0.059 R2 adj 0.061 0.000 0.019 0.057 AIC − 1282.3 − 1158.3 − 1201.5 − 1279.1 BIC − 1265.3 − 1141.3 − 1184.5 − 1250.9 Std. errors HC1 HC1 HC1 HC1 Table 11 First stage least square regression for number of nursing staff per 1000 deliveries as an instrument variable for number of physicians per 1000 deliveries + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001 First stage regression (fixed effects) Number of nursing staff per 1000. deliveries 0.239*** p value (< 0.001) 95% CI [0.208, 0.269] SE (0.016) Num. obs 2037 R20.329 R2 adj 0.106 AIC 9101.9 BIC 9113.1 F-statistic 237.206 Std. errors HC1 1009 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Table 12 Endogeneity check for instrument variable on riskadjusted C-section ratio + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 Ratio of observed to expected (O/E) cesarean births (2015–2019) (two-way fixed effects model) univariate model multivariate model Number of physicians per 1000 deliveries 0.014*** 0.014*** p value (< 0.001) (< 0.001) 95% CI [0.009, 0.019] [0.009, 0.019] SE (0.003) (0.003) Number of midwives per 1000 deliveries − 0.001 p value (0.627) 95% CI [− 0.004, 0.002] SE (0.002) Number of deliveries per 1000 0.001 p value (0.933) 95% CI [− 0.024, 0.026] SE (0.013) Endogenous part from the IV − 0.006* − 0.006* p value (0.023) (0.021) 95% CI [− 0.012, − 0.001] [− 0.012, − 0.001] SE (0.003) (0.003) Num. obs 2037 2037 R20.066 0.067 R2 adj − 0.246 − 0.247 AIC − 3436.6 − 3433.6 BIC − 3419.7 − 3405.5 Std. errors HC1 HC1 1010 A.Stöcker et al. Table 13 Comparison of effect size lagged variable for OLS, fixed effect and difference GMM estimator on riskadjusted C-section ratio + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 Ratio of observed to expected (O/E) cesarean births (2015–2019) (two-way fixed effects model) OLS Panel model with fixed effects difference GMM Intercept 0.144*** p value (< 0.001) 95% CI [0.111, 0.177] SE (0.017) Lagged ratio of observed to expected ratio (O/E) of cesarean births 0.864*** 0.046 0.510*** p value (< 0.001) (0.241) (< 0.001) 95% CI [0.837, 0.891] [− 0.031, 0.123] [0.265, 0.754] SE (0.014) (0.039) (0.125) Number of physicians per 1000 deliveries 0.000 0.004* 0.003 p value (0.769) (0.045) (0.239) 95% CI [− 0.001, 0.001] [0.000, 0.007] [− 0.002, 0.007] SE (0.001) (0.002) (0.002) Number of midwives per 1000 deliveries 0.000 0.002 0.003 p value (0.884) (0.336) (0.150) 95% CI [− 0.001, 0.001] [− 0.002, 0.006] [− 0.001, 0.006] SE (0.001) (0.002) (0.002) Number of deliveries per 1000 − 0.006* − 0.013 − 0.008 p value (0.037) (0.670) (0.844) 95% CI [− 0.012, 0.000] [− 0.074, 0.048] [− 0.089, 0.072] SE (0.003) (0.031) (0.041) Num. obs 1479 1479 1024 R20.753 0.012 R2 adj 0.752 − 0.427 AIC − 2916.6 − 4015.8 BIC − 2884.8 − 3989.3 Std. errors HC1 HC1 1017 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Table 21 Uniand multivariate panel models with two-ways difference generalized method of moment estimators on crude C-section/birth ratio Ratio of C-section to all births (2015–2019) (two-way fixed effects model) Univariate model Multivariate model Multivariate model with control variables Lagged C-section to delivery ratio 0.522*** 0.503*** 0.511*** p value (< 0.001) (< 0.001) (< 0.001) 95% CI [0.310, 0.734] [0.289, 0.716] [0.295, 0.727] SE (0.108) (0.109) (0.110) Number of physicians per 1000 deliveries 0.001 0.001 0.001 p value (0.117) (0.413) (0.433) 95% CI [0.000, 0.003] [− 0.001, 0.002] [− 0.001, 0.002] SE (0.001) (0.001) (0.001) Number of midwives per 1000 deliveries 0.001 0.001 p value (0.336) (0.310) 95% CI [− 0.001, 0.002] [− 0.001, 0.002] SE (0.001) (0.001) Number of deliveries per 1000 − 0.014 − 0.013 p value (0.447) (0.471) 95% CI [− 0.050, 0.022] [− 0.050, 0.023] SE (0.018) (0.019) Ownership: public (ref. category: ownership private) 0.000 p value (0.995) 95% CI [− 0.051, 0.051] SE (0.026) Ownership: non-profit (ref. category: ownership private) 0.003 p value (0.899) 95% CI [− 0.038, 0.043] SE (0.021) Perinatal centers level I (care level 1) (ref. category: regular obstetric department (care level 4)) 0.009 p value (0.367) 95% CI [− 0.011, 0.029] SE (0.010) Perinatal centers level II (care level 2) (ref. category: regular obstetric department (care level 4) 0.008 p value (0.594) 95% CI [− 0.020, 0.035] SE (0.014) Perinatal focus (care level 3) (ref. category: regular obstetric department (care level 4) − 0.011 p value (0.334) 95% CI [− 0.032, 0.011] SE (0.011) Teaching status: academic teaching hospital (ref. category: teaching status: no teaching assignment) − 0.006 p value (0.509) 95% CI [− 0.022, 0.011] SE (0.008) Teaching status: University Hospital (ref. category: teaching status: no teaching assignment) − 0.004 p value (0.675) 1018 A.Stöcker et al. + p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001 Table 21 (continued) Ratio of C-section to all births (2015–2019) (two-way fixed effects model) Univariate model Multivariate model Multivariate model with control variables 95% CI [− 0.021, 0.013] SE (0.009) Num. obs 1031 1031 1031 Hansen–Sargan test/J test (p value) 1.937 (0.858) 1.704 (0.888) 1.685 (0.891) Arellano–Bond test/autocorrelation test (1) (p value) − 6.098 (< 0.001) − 6.225 (< 0.001) − 6.240 (< 0.001) Arellano–Bond test/autocorrelation test (2) (p value) − 0.365 (0.715) − 0.494 (0.622) − 0.405 (0.686) Wald test for coefficients (p value) 28.950 (< 0.001) 30.083 (< 0.001) 56.313 (< 0.001) Wald test for time dummies (p value) 9.975 (0.019) 9.845 (0.020) 9.890 (0.020) Std. errors HC1 HC1 HC1 Table 22 Example of a depiction for calculating the expected risk-adjusted cesarean section rate by IQTIG for 2018. Own translation. (Source: https:// iqtig. org/ downl oads/ auswe rtung/ auswe rtung/ 2018/ 16n1g ebh/ QSKH_ 16n1GEBH_ 2018_ QIDB_ V02_ 20190411. pdf, p. 19) Reference probability: 14.205% (odds: 0.165) Risk factor Regression coefficient Standard error Z value Odds ratio 95% confidence interval Constant − 1.798342673800550 0.004 − 408.407 – – Age 35–38years 0.034231814566637 0.008 4.439 1.035 1.019–1.051 Age > 38 0.286484745408656 0.011 24.971 1.332 1.302–1.362 Birth risk: amnion infection syndrome (suspected) 2.647566531445380 0.040 66.591 14.120 13.061–15.264 Birth risk: diabetes mellitus 0.354402849335064 0.014 24.645 1.425 1.386–1.466 Birth risk: premature birth 0.356693940042668 0.018 20.341 1.429 1.380–1.479 Birth risk: hypertensive Pregnancy disorder or HELLP Syndrome 1.471801217071380 0.018 81.161 4.357 4.205–4.515 Birth risk: pathological CTG. poor fetal heart sounds. or acidosis during birth (detected by FBS) 0.922892533667843 0.007 124.693 2.517 2.480–2.553 Birth risk: placenta praevia 3.414601549901360 0.061 55.693 30.405 26.962–34.287 Birth risk: breech position 3.585447507902870 0.018 199.288 36.069 34.820–37.364 Birth risk: face/forehead Presentation 1.942056761249630 0.063 30.713 6.973 6.160–7.893 Birth risk: transverse/oblique position 6.515521815847840 0.268 24.293 675.546 399.356–1142.746 Birth risk: previous cesarean section or other uterus operations 1.994488346490800 0.016 122.135 7.348 7.117–7.587 Multiple pregnancy 1.453148908289270 0.024 59.963 4.277 4.078–4.485 Mother's record: hypertension or proteinuria 0.241446328929522 0.025 9.548 1.273 1.212–1.338 Mother's record: placental insufficiency 0.735298957679869 0.031 23.853 2.086 1.964–2.216 Mother's record: previous cesarean section or uterus operations 0.294514404293640 0.016 17.888 1.342 1.300–1.387 1019 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Author contributions Conceptualization: Arno Stöcker, Ludwig Kuntz, Nadine Scholten; Methodology: Arno Stöcker; Formal analysis, Investigation, Data Curation: Arno Stöcker; Writing—original draft preparation: Arno Stöcker; Writing—review and editing: Ludwig Kuntz, Nadine Scholten, Holger Pfaff; Resources: Holger Pfaff, Nadine Scholten; Supervision: Ludwig Kuntz. Funding Open Access funding enabled and organized by Projekt DEAL. Data availability The analysis provided relied upon the hospital quality reports from German acute care hospitals. All the data utilized in this analysis is accessible to the public and can be found at www.gba. de/ quali taets beric hte. Declarations Conflict of interest The authors have no relevant financial or non-financial interests to disclose. Ethics approval Not applicable. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. References 1. Boerma, T., Ronsmans, C., Melesse, D.Y., etal.: Global epidemiology of use of and disparities in caesarean sections. Lancet 392, 1341–1348 (2018). https:// doi. org/ 10. 1016/ S01406736(18) 31928-7 2. Fait, T., Šťastná, A., Kocourková, J., etal.: Has the cesarean epidemic in Czechia been reversed despite fertility postponement? BMC Pregnancy Childbirth 22, 469 (2022). https:// doi. org/ 10. 1186/ s1288402204781-1 3. Owens, K.: When less is more: shifting risk management in American childbirth. Adv. Med. Sociol. 20, 45–62 (2019). https:// doi. org/ 10. 1108/ S105762902 01900 00020 008 4. Sandall, J., Tribe, R.M., Avery, L., etal.: Short-term and longterm effects of caesarean section on the health of women and children. Lancet 392, 1349–1357 (2018). https:// doi. org/ 10. 1016/ S01406736(18) 31930-5 5. Statistisches Bundesamt (2022) Kranken hausent bindungen in Deutschland. https:// www. desta tis. de/ DE/ Themen/ Gesel lscha ftUmwelt/ Gesun dheit/ Krank enhae user/ Tabel len/ krank enhau sentb indun genkaise rschn itt. html. Accessed 20 Oct 2022 6. Schneider, H.: Risiko-Nutzen-Verhältnis bei natürlicher Geburt und elektiver Sectio. Gynäkologe 46, 709–714 (2013). https:// doi. org/ 10. 1007/ s001290133179-x 7. Deutsche Gesellschaft für Gynäkologie und Geburtshilfe (DGGG) (Ed.). (2020). S3-Leitlinie. Die Sectio caesarea [S3 Guideline. The caesarean section]. https:// www. awmf. org/ uploa ds/ tx_ szlei tlini en/ 015084l_ S3_ Sectiocaesa rea_ 202006_1_ 02. pdf. 8. Betrán, A.P., Temmerman, M., Kingdon, C., etal.: Interventions to reduce unnecessary caesarean sections in healthy women and babies. Lancet 392, 1358–1368 (2018). https:// doi. org/ 10. 1016/ S01406736(18) 31927-5 9. Kirchengast, S., Hartmann, B.: Recent lifestyle parameters are associated with increasing caesarean section rates among singleton term births in Austria. Int. J. Environ. Res. Public Health 16, 14 (2018). https:// doi. org/ 10. 3390/ ijerp h1601 0014 10. Loke, A.Y., Davies, L., Li, S.: Factors influencing the decision that women make on their mode of delivery: the Health Belief Model. Fig. 2 Ratio deliveries to full inpatient cases 1020 A.Stöcker et al. BMC Health Serv. Res. 15, 274 (2015). https:// doi. org/ 10. 1186/ s129130150931-z 11. Wolf, J.H.: Risk and reputation: obstetricians, cesareans, and consent. J. Hist. Med. Allied Sci. 73, 7–28 (2018). https:// doi. org/ 10. 1093/ jhmas/ jrx053 12. Mulchandani, R., Power, H.S., Cavallaro, F.L.: The influence of individual provider characteristics and attitudes on caesarean section decision-making: a global review. J. Obstet. Gynaecol. 40, 1–9 (2020). https:// doi. org/ 10. 1080/ 01443 615. 2019. 15876 03 13. Panda, S., Begley, C., Daly, D.: Clinicians’ views of factors influencing decision-making for caesarean section: a systematic review and metasynthesis of qualitative, quantitative and mixed methods studies. PLoSOne 13, e0200941 (2018). https:// doi. org/ 10. 1371/ journ al. pone. 02009 41 14. Facchini, G.: Forgetting-by-not-doing: the case of surgeons and cesarean sections. Health Econ. 31, 481–495 (2022). https:// doi. org/ 10. 1002/ hec. 4460 15. Monari, F., Di Mario, S., Facchinetti, F., etal.: Obstetricians’ and midwives’ attitudes toward cesarean section. Birth 35, 129–135 (2008). https:// doi. org/ 10. 1111/j. 1523536X. 2008. 00226.x 16. Josefsson, A., Gunnervik, C., Sydsjö, A., etal.: A comparison between Swedish midwives’ and obstetricians’ & gynecologists’ opinions on cesarean section. Matern. Child Health J. 15, 555–560 (2011). https:// doi. org/ 10. 1007/ s109950100630-7 17. Science Media Center Germany (2020) Erste S3-Leitlinie Kaiserschnitt: Hintergrund und regionale Datenanalyse. https:// www. scien cemed iacen ter. de/ alleangeb ote/ inves tigat ive/ detai ls/ news/ erstes3leitl iniekaise rschn itthinte rgr undundregio naledaten analy se/. Accessed 21 Oct 2022 18. Zipfel, L., Weidmann, C.: Einflussfaktoren auf die Kaiserschnittraten in deutschen Krankenhäusern in den Jahren 2015– 2017. Eine ökologische Studie (Factors Influencing the Caesarean Section Rates in German Hospitals in the Years 2015–2017: an Ecological Study). Gesundheitswesen 84, 944–951 (2022). https:// doi. org/ 10. 1055/a15314998 19. Gross, M.M., Matterne, A., Berlage, S., etal.: Interinstitutional variations in mode of birth after a previous caesarean section: a cross-sectional study in six German hospitals. J. Perinat. Med. 43, 177–184 (2015). https:// doi. org/ 10. 1515/ jpm20140108 20. Hemminki, E., Heino, A., Gissler, M.: Should births be centralised in higher level hospitals? Experiences from regionalised health care in Finland. BJOG Int. J. Obstet. Gynaecol. 118, 1186–1195 (2011). https:// doi. org/ 10. 1111/j. 14710528. 2011. 02977.x 21. Biro, M.A., Knight, M., Wallace, E., etal.: Is place of birth associated with mode of birth? The effect of hospital on caesarean section rates in a public metropolitan health service. Aust. N. Z. J. Obstet. Gynaecol. 54, 64–70 (2014). https:// doi. org/ 10. 1111/ ajo. 12147 22. Joaquim, I., Pereira, L.N., Nunes, C., etal.: C-sections and hospital characteristics: a long term analysis on low-risk deliveries. Res. Health Serv. Reg. 1, 1–9 (2022). https:// doi. org/ 10. 1007/ s4399902200014-2 23. Perrotta, C., Romero, M., Sguassero, Y., etal.: Caesarean birth in public maternities in Argentina: a formative research study on the views of obstetricians, midwives and trainees. BMJ Open 12, e053419 (2022). https:// doi. org/ 10. 1136/ bmjop en2021053419 24. Mikolajczyk, R.T., Schmedt, N., Zhang, J., etal.: Regional variation in caesarean deliveries in Germany and its causes. BMC Pregnancy Childbirth 13, 99 (2013). https:// doi. org/ 10. 1186/ 147123931399 25. Kolip, P., Nolting, H.-D., Zich, K.: Kaiserschnittgeburten— Entwicklung und regionale Verteilung. Bertelsmann Stiftung, Gütersloh (2012) 26. Hanley, G.E., Janssen, P.A., Greyson, D.: Regional variation in the cesarean delivery and assisted vaginal delivery rates. Obstet. Gynecol. 115, 1201–1208 (2010). https:// doi. org/ 10. 1097/ AOG. 0b013 e3181 dd918c 27. Guglielminotti, J., Deneux-Tharaux, C., Wong, C.A., etal.: Hospital-level factors associated with anesthesia-related adverse events in cesarean deliveries, New York State, 2009–2011. Anesth. Analg. 122, 1947–1956 (2016). https:// doi. org/ 10. 1213/ ANE. 00000 00000 001341 28. Walther, F., Kuester, D., Bieber, A., etal.: Are birth outcomes in low risk birth cohorts related to hospital birth volumes? A systematic review. BMC Pregnancy Childbirth 21, 531 (2021). https:// doi. org/ 10. 1186/ s1288402103988-y 29. Kyser, K.L., Lu, X., Santillan, D.A., etal.: The association between hospital obstetrical volume and maternal postpartum complications. Am. J. Obstet. Gynecol. 207, 42.e1–17 (2012). https:// doi. org/ 10. 1016/j. ajog. 2012. 05. 010 30. Epstein, A.J., Nicholson, S.: The formation and evolution of physician treatment styles: an application to cesarean sections. J. Health Econ. 28, 1126–1140 (2009). https:// doi. org/ 10. 1016/j. jheal eco. 2009. 08. 003 31. Plough, A., Henrich, N., Galvin, G., etal.: Common challenges managing bed and staff availability on labor and delivery units in the United States: a qualitative analysis. Birth 45, 303–310 (2018). https:// doi. org/ 10. 1111/ birt. 12342 32. Leonard, P.S.J., Crouse, D.L., Boudreau, J.G., etal.: Provider volume and maternal complications after caesarean section: results from a population-based study. BMC Pregnancy Childbirth 20, 37 (2020). https:// doi. org/ 10. 1186/ s128840192709-5 33. Lee, K.-S., Kwak, J.-M.: Effect of patient risk on the volumeoutcome relationship in obstetric delivery services. Health Policy (Amsterdam, Netherlands) 118, 407–412 (2014). https:// doi. org/ 10. 1016/j. healt hpol. 2014. 05. 007 34. Bozzuto, L., Passarella, M., Lorch, S., etal.: Effects of delivery volume and high-risk condition volume on maternal morbidity among high-risk obstetric patients. Obstet. Gynecol. 133, 261–268 (2019). https:// doi. org/ 10. 1097/ AOG. 00000 00000 003080 35. Clapp, M.A., Melamed, A., Robinson, J.N., etal.: Obstetrician volume as a potentially modifiable risk factor for cesarean delivery. Obstet. Gynecol. 124, 697–703 (2014). https:// doi. org/ 10. 1097/ AOG. 00000 00000 000473 36. Zbiri, S., Rozenberg, P., Goffinet, F., etal.: Cesarean delivery rate and staffing levels of the maternity unit. PLoSOne 13, e0207379 (2018). https:// doi. org/ 10. 1371/ journ al. pone. 02073 79 37. Gombolay, M., Golen, T., Shah, N., etal.: Queueing theoretic analysis of labor and delivery: understanding management styles and C-section rates. Health Care Manag. Sci. 22, 16–33 (2019). https:// doi. org/ 10. 1007/ s107290179418-2 38. Ramani, S., Halpern, T.A., Akerman, M., etal.: A new index for obstetrics safety and quality of care: integrating cesarean delivery rates with maternal and neonatal outcomes. Am. J. Obstet. Gynecol. 226, 556.e1-556.e9 (2022). https:// doi. org/ 10. 1016/j. ajog. 2021. 10. 005 39. Curry, L.A., Brault, M.A., Linnander, E.L., etal.: Influencing organisational culture to improve hospital performance in care of patients with acute myocardial infarction: a mixed-methods intervention study. BMJ Qual. Saf. 27, 207–217 (2018). https:// doi. org/ 10. 1136/ bmjqs2017006989 40. de Cruppé, W., Geraedts, M.: Wie konstant halten Krankenhäuser die Mindestmengenvorgaben ein? Eine retrospektive, längsschnittliche Datenanalyse der Jahre 2006, 2008 und 2010 (How Steady are Hospitals in Complying with Minimum Volume Standards? A Retrospective Longitudinal Data Analysis of the Years 2006, 2008, and 2010). Zentralbl. Chir. 141, 425–432 (2016). https:// doi. org/ 10. 1055/s003413833 71 41. de Cruppé, W., Geraedts, M.: Mindestmengen unterschreiten, Ausnahmetatbestände und ihre Konsequenzen ab 2018. Komplexe Eingriffe am Ösophagus und Pankreas in deutschen 1021 Exploring theinfluence ofmedical staffing andbirth volume onobserved‑to‑expected cesarean… Krankenhäusern im Zeitverlauf von 2006 bis 2014 (Falling Short of Minimum Volume Standards, Exemptions and Their Consequences from 2018 Onwards. Complex Procedures on Oesophagus and Pancreas in German Hospitals from 2006 to 2014). Zentralbl. Chir. 143, 250–258 (2018). https:// doi. org/ 10. 1055/a05732625 42. Scholten, N., Pfaff, H., Lehmann, H.C., etal.: Who does it first? The uptake of medical innovations in the performance of thrombolysis on ischemic stroke patients in Germany: a study based on hospital quality data. Implement. Sci. 10, 10 (2015). https:// doi. org/ 10. 1186/ s130120140196-7 43. Vorbeck, L., Naumoska, D., Geraedts, M.: Assoziation von Strukturvariablen mit der Versorgungsqualität der Krankenhäuser in Deutschland (Association of Structural Variables with Quality of Care in German Hospitals). Gesundheitswesen 84, 242–249 (2022). https:// doi. org/ 10. 1055/a13411246 44. Messer, M., Reilley, J.T.: Qualitätsberichte als Vermittlungsinstanz im Wettbewerb zwischen Krankenhäusern: Patienten als rationale Akteure. Berl. J. Soziol. 25, 61–81 (2015). https:// doi. org/ 10. 1007/ s116090150279-6 45. Volkert, A., Pfaff, H., Scholten, N.: What really matters? Organizational versus regional determinants of hospitals providing medical service centres. Health Policy (Amsterdam, Netherlands) 124, 1354–1362 (2020). https:// doi. org/ 10. 1016/j. healt hpol. 2020. 07. 011 46. Volkert, A., Stöcker, A., Pfaff, H., etal.: What organisational and regional factors influence the outpatient provision of curettages in Germany? A longitudinal secondary data analysis using hospital quality reports data from 2013 to 2019. BMJ Open 13, e072887 (2023). https:// doi. org/ 10. 1136/ bmjop en2023072887 47. Kraska, R.A., de Cruppe, W., Geraedts, M.: Probleme bei der Verwendung von Qualitätsberichtsdaten für die Versorgungsforschung (Problems with Using Hospital Quality Reports as a Secondary Data Source for Health Services Research in Germany). Gesundheitswesen 79, 542–547 (2017). https:// doi. org/ 10. 1055/s003515559 53 48. Ji, L., Geraedts, M. & de Cruppé, W.: A theoretical framework for linking hospitalslongitudinally: demonstrated using German Hospital Quality Reports 2016–2020. BMC Med Res Methodol 24, 212 (2024).https:// doi. org/ 10. 1186/ s1287402402317-zh 49. Woeber, K., Vanderlaan, J., Long, M.H., etal.: Midwifery autonomy and employment changes during the early COVID-19 pandemic. J. Midwifery Womens Health 67, 608–617 (2022). https:// doi. org/ 10. 1111/ jmwh. 13400 50. Schmiedhofer, M., Derksen, C., Dietl, J.E., etal.: Birthing under the condition of the COVID-19 pandemic in Germany: interviews with mothers, partners, and obstetric health care workers. Int. J. Environ. Res. Public Health 19, 1486 (2022). https:// doi. org/ 10. 3390/ ijerp h1903 1486 51. Beschreibung der Qualitätsindikatoren und Kennzahlen nach QSKH-RL. Geburtshilfe. Endgültige Rechenregeln für das Erfassungsjahr 2020: Beschreibung der Qualitätsindikatoren und Kennzahlen nach QSKH-RL Endgültige Rechenregeln, Veröffentlichungsversion (2021) 52. Becker, A., Eissler, U.: Die standardisierte primäre Sectiorate (SPSR) und ihre Anwendung im Qualitätsmanagement und für Krankenhausvergleiche. Prädiktoren der primären Sectio als Beitrag zur Versachlichung einer komplexen Diskussion (Standardised primary Caesarian section rate (SPSR) and how it can be applied in Quality Management and for Hospital). Interdisciplinary Contributions to Hospital Management: Medicine, Patient Safety and Economics (2013) 53. Chen, N., Pan, J.: The causal effect of delivery volume on severe maternal morbidity: an instrumental variable analysis in Sichuan, China. BMJ Glob. Health 7, e008428 (2022). https:// doi. org/ 10. 1136/ bmjgh2022008428 54. Brazier, E., Borrell, L.N., Huynh, M., etal.: Variation and racial/ ethnic disparities in Cesarean delivery at New York City hospitals: the contribution of hospital-level factors. Ann. Epidemiol. 73, 1–8 (2022). https:// doi. org/ 10. 1016/j. annep idem. 2022. 06. 003 55. Hoxha, I., Zhubi, E., Grezda, K., etal.: Caesarean sections in teaching hospitals: systematic review and meta-analysis of hospitals in 22 countries. BMJ Open 11, e042076 (2021). https:// doi. org/ 10. 1136/ bmjop en2020042076 56. Blundell, R., Bond, S.: Initial conditions and moment restrictions in dynamic panel data models. J. Econom. 87, 115–143 (1998). https:// doi. org/ 10. 1016/ S03044076(98) 00009-8 57. Arellano, M., Bover, O.: Another look at the instrumental variable estimation of error-components models. J. Econom. 68, 29–51 (1995). https:// doi. org/ 10. 1016/ 03044076(94) 01642-D 58. Antonakis, J., Bastardoz, N., Rönkkö, M.: On ignoring the random effects assumption in multilevel models: review, critique, and recommendations. Organ. Res. Methods 24, 443–483 (2021). https:// doi. org/ 10. 1177/ 10944 28119 877457 59. Papies, D., Ebbes, P., Van Heerde, H.J. (2017). Addressing Endogeneity in Marketing Models. In: Leeflang, P., Wieringa, J., Bijmolt, T., Pauwels, K. (eds) Advanced Methods for Modeling Markets. International Series in Quantitative Marketing. Springer, Cham. https:// doi. org/ 10. 1007/ 978-331953469-5_ 18 60. Germann, F., Ebbes, P., Grewal, R.: The chief marketing officer matters! J. Mark. 79, 1–22 (2015). https:// doi. org/ 10. 1509/ jm. 14. 0244 61. Rossi, P.E.: Invited paper—even the rich can make themselves poor: a critical examination of IV methods in marketing applications. Mark. Sci. 33, 655–672 (2014). https:// doi. org/ 10. 1287/ mksc. 2014. 0860 62. Arellano, M., Bond, S.: Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Rev. Econ. Stud. 58, 277 (1991). https:// doi. org/ 10. 2307/ 22979 68 63. Roodman, D.: How to do Xtabond2: An introduction to difference and system GMM in Stata. Stand. Genom. Sci. 9, 86–136 (2009). https:// doi. org/ 10. 1177/ 15368 67X09 00900 106 64. Lee, D.S., McCrary, J., Moreira, M.J., etal.: Valid t-ratio inference for IV. Am. Econ. Rev. 112, 3260–3290 (2022). https:// doi. org/ 10. 1257/ aer. 20211 063 65. Turner, L., Culliford, D., Ball, J., etal.: The association between midwifery staffing levels and the experiences of mothers on postnatal wards: cross sectional analysis of routine data. Women Birth J. Aust. Coll. Midwives 35, e583–e589 (2022). https:// doi. org/ 10. 1016/j. wombi. 2022. 02. 005 66. Gallagher, L., Smith, V., Carroll, M., etal.: What would reduce caesarean section rates? Views from pregnant women and clinicians in Ireland. PLoSOne 17, e0267465 (2022). https:// doi. org/ 10. 1371/ journ al. pone. 02674 65 67. Bohren, M.A., Hofmeyr, G.J., Sakala, C., etal.: Continuous support for women during childbirth. Cochrane Database Syst. Rev. 7, CD003766 (2017). https:// doi. org/ 10. 1002/ 14651 858. CD003 766. pub6 68. Han, K.-T., Kim, S.J., Ju, Y.J., etal.: Do hospital characteristics influence Cesarean delivery? Analysis of National Health Insurance claim data. Eur. J. Public Health 27, 801–807 (2017). https:// doi. org/ 10. 1093/ eurpub/ ckx061 69. Ghafari-Saravi, A., Chaiken, S.R., Packer, C.H., etal.: Cesarean delivery rates by hospital type among nulliparous and multiparous patients. J. Matern. Fetal. Neonatal. Med. 35, 8631–8639 (2022). https:// doi. org/ 10. 1080/ 14767 058. 2021. 19908 84 70. Clapp, M.A., James, K.E., Melamed, A., etal.: Hospital volume and cesarean delivery among low-risk women in a nationwide sample. J. Perinatol. 38, 127–131 (2018). https:// doi. org/ 10. 1038/ jp. 2017. 173 1022 A.Stöcker et al. 71. Hentschker, C., Mennicken, R.: The volume-outcome relationship revisited: practice indeed makes perfect. Health Serv. Res. 53, 15–34 (2018). https:// doi. org/ 10. 1111/ 14756773. 12696 72. Huguet, M.: Centralization of care in high volume hospitals and inequalities in access to care. Soc Sci Med 260, 113177 (2020). https:// doi. org/ 10. 1016/j. socsc imed. 2020. 113177 73. Martin-Arribas, A., Escuriet, R., Borràs-Santos, A., etal.: A comparison between midwifery and obstetric care at birth in Spain: across-sectional study of perinatal outcomes. Int. J. Nurs. Stud. 126, 104129 (2022). https:// doi. org/ 10. 1016/j. ijnur stu. 2021. 104129 74. Thiessen, K., Nickel, N., Prior, H.J., etal.: Maternity outcomes in manitoba women: a comparison between midwifery-led care and physician-led care at birth. Birth 43, 108–115 (2016). https:// doi. org/ 10. 1111/ birt. 12225 75. Carlson, N.S., Corwin, E.J., Hernandez, T.L., etal.: Association between provider type and cesarean birth in healthy nulliparous laboring women: a retrospective cohort study. Birth 45, 159–168 (2018). https:// doi. org/ 10. 1111/ birt. 12334 76. Hamlin, L., Grunwald, L., Sturdivant, R.X., etal.: Comparison of nurse-midwife and physician birth outcomes in the military health system. Policy Polit. Nurs. Pract. 22, 105–113 (2021). https:// doi. org/ 10. 1177/ 15271 54421 994071 77. Souter, V., Nethery, E., Kopas, M.L., etal.: Comparison of Midwifery and obstetric care in low-risk hospital births. Obstet. Gynecol. 134, 1056–1065 (2019). https:// doi. org/ 10. 1097/ aog. 00000 00000 003521 78. Hoxha, I., Fejza, A., Aliu, M., etal.: Health system factors and caesarean sections in Kosovo: a cross-sectional study. BMJ Open 9, e026702 (2019). https:// doi. or g/ 10. 1136/ bmjop en2018026702 79. Sandall, J., Soltani, H., Gates, S., etal.: Midwife-led continuity models versus other models of care for childbearing women. Cochrane Database Syst. Rev. 4, CD004667 (2016). https:// doi. org/ 10. 1002/ 14651 858. CD004 667. pub5 80. Hanahoe, M.: Midwifery-led care can lower caesarean section rates according to the Robson ten group classification system. Eur. J. Midwifery 4, 7 (2020). https:// doi. org/ 10. 18332/ ejm/ 119164 81. Rivo, J.C., Amyx, M., Pingray, V., etal.: Obstetrical providers’ preferred mode of delivery and attitude towards non-medically indicated caesarean sections: a cross-sectional study. BJOG Int. J. Obstet. Gynaecol. 125, 1294–1302 (2018). https:// doi. org/ 10. 1111/ 14710528. 15122 82. Kearney, L., Kynn, M., Craswell, A., etal.: The relationship between midwife-led group-based versus conventional antenatal care and mode of birth: a matched cohort study. BMC Pregnancy Childbirth 17, 39 (2017). https:// doi. org/ 10. 1186/ s128840161216-1 83. Andraczek, T., Magister, S., Bautzmann, S., etal.: Geburt im hebammengeleiteten Kreißsaal eines Perinatalzentrums— Lernkurve, Ergebnisse und Benchmark (Birth in the Midwife-Led Delivery Room of a Perinatal Center - Learning Curve, Outcomes and Benchmark). Z. Geburtshilfe Neonatol. (2023). https:// doi. org/ 10. 1055/a20822176 84. Berger, C.: Mütterliches und neonatales geburtshilfliches Outcome von geplanten Entbindungen in einem Hebammenkreißsaal: Kohortenstudie in einem Zentrum der Maximalversorgung in Deutschland. Hebamme 33, 12 (2020). https:// doi. org/ 10. 1055/a12922055 85. Ullah, S., Zaefarian, G., Ullah, F.: How to use instrumental variables in addressing endogeneity? A step-by-step procedure for non-specialists. Ind. Mark. Manage. 96, A1–A6 (2021). https:// doi. org/ 10. 1016/j. indma rman. 2020. 03. 006 86. Hoxha, I., Sadiku, F., Lama, A., etal.: Cesarean delivery and gender of delivering physicians: a systematic review and metaanalysis. Obstet. Gynecol. 136, 1170–1178 (2020). https:// doi. org/ 10. 1097/ AOG. 00000 00000 004172 87. Freeman, M., Savva, N., Scholtes, S.: Gatekeepers at work: an empirical analysis of a maternity unit. Manag. Sci. 63, 3147–3167 (2017). https:// doi. org/ 10. 1287/ mnsc. 2016. 2512 88. Friedman, A.M., Ananth, C.V., Huang, Y., etal.: Hospital delivery volume, severe obstetrical morbidity, and failure to rescue. Am. J. Obstet. Gynecol. 215, 795.e1-795.e14 (2016). https:// doi. org/ 10. 1016/j. ajog. 2016. 07. 039 89. Smith, V., Hannon, K., Begley, C.: Clinician’s attitudes towards caesarean section: a cross-sectional survey in two tertiary level maternity units in Ireland. Women Birth J. Aust. Coll. Midwives 35, 423–428 (2022). https:// doi. or g/ 10. 1016/j. wombi. 2021. 08. 004 90. Kingdon, C., Downe, S., Betran, A.P.: Non-clinical interventions to reduce unnecessary caesarean section targeted at organisations, facilities and systems: systematic review of qualitative studies. PLoSOne 13, e0203274 (2018). https:// doi. org/ 10. 1371/ journ al. pone. 02032 74 Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.