Are long hospitalizations substituting primary and long-term care? Evidence from Brazil and Mexico

Aranco, Natalia; Bauhoff, Sebastian; Schwarz, Natalie; Stampini, Marco Working Paper Are long hospitalizations substituting primary and longterm care? Evidence from Brazil and Mexico IDB Working Paper Series, No. IDB-WP-1632 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Aranco, Natalia; Bauhoff, Sebastian; Schwarz, Natalie; Stampini, Marco (2024) : Are long hospitalizations substituting primary and long-term care? Evidence from Brazil and Mexico, IDB Working Paper Series, No. IDB-WP-1632, Inter-American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0013126 This Version is available at: https://hdl.handle.net/10419/302210 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/3.0/igo/ Are Long Hospitalizations substituting Primary and Long-term Care? Evidence from Brazil and Mexico Natalia Aranco Sebastian Bauhoff Natalie Schwarz Marco Stampini WORKING PAPER No IDB-WP-1632 Inter-American Development Bank Social Protection and Health Division August 2024 Are Long Hospitalizations substituting Primary and Long-term Care? Evidence from Brazil and Mexico Natalia Aranco Sebastian Bauhoff Natalie Schwarz Marco Stampini Inter-American Development Bank Social Protection and Health Division August 2024 Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Are long hospitalizations substituting primary and long-term care?: evidence from Brazil and Mexico / Natalia Aranco, Sebastian Bauhoff, Natalie Schwarz, Marco Stampini. p. cm. — (IDB Working Paper Series ; 1632) Includes bibliographical references. 1. Primary health care-Brazil. 2. Primary health care-Mexico. 3. Long-term care facilities-Brazil. 4. Long-term care facilities-Mexico. 5. Population agingBrazil. 6. Population aging-Mexico. 7. Medical care-Brazil. 8. Medical careMexico. 9. Medical policy-Brazil. 10. Medical policy-Mexico. I. Aranco, Natalia. II. Bauhoff, Sebastian. III. Schwarz, Natalie. IV. Stampini, Marco. V. Inter-American Development Bank. Social Protection and Health Division. VI. Series. IDB-WP-1632 http://www.iadb.org Copyright © 2024 Inter-American Development Bank ("IDB"). This work is subject to a Creative Commons license CC BY 3.0 IGO (https://creativecommons.org/licenses/by/3.0/igo/legalcode). The terms and conditions indicated in the URL link must be met and the respective recognition must be granted to the IDB. Further to section 8 of the above license, any mediation relating to disputes arising under such license shall be conducted in accordance with the WIPO Mediation Rules. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the United Nations Commission on International Trade Law (UNCITRAL) rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this license. Note that the URL link includes terms and conditions that are an integral part of this license. The opinions expressed in this work are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. 43 Abstract 1 Prolonged hospital stays, or hospital stays that are longer than medically necessary, are a major concern for patients, payers, and providers. We conceptualize and empirically estimate the prevalence and cost of prolonged stays among elderly hospital patients (65 years and older) in Brazil and Mexico. We develop a continuum-of-care conceptual framework based on prior literature and insights obtained through interviews and focus group discussions with experts from Mexico, Argentina, and Colombia. In this framework, hospitals are part of a wider system. This system involves both pre-admission and post-discharge medical and social care services. There are three main sources of prolonged stays: (i) lack of appropriate primary healthcare that leads to more complex admissions; (ii) hospital inefficiency; and (iii) lack of rehabilitation, social, and longterm care at discharge. We estimate the count and share of inappropriate hospital days due to prolonged stays overall and for each source. This estimation is based on administrative records on discharges from public sector hospitals in 2019. Our results show that hospital days due to prolonged stays account for approximately half of all hospital days. Although most of the inappropriate days can be attributed to hospital inefficiency (36% in Brazil and 49% in Mexico), an important share is linked to the lack of rehabilitation, social, and long-term care. Lack of these services accounts for 12% of total hospital days in Brazil and 7% in Mexico. In a back-of-theenvelope calculation, we estimate that providing six weeks of long-term care services to address the care needs brought about by only thirteen causes of admission would generate annual net savings of approximately US$174 million in Brazil and US$45 million in Mexico. Keywords: healthcare costs; prolonged hospitalizations; primary health care; long-term care; medical care; population aging; older persons; public policy; social care; rehabilitation care; Latin America and the Caribbean; Mexico; Brazil. JEL classification: I10, J14, H5, J18 1 All authors are with the Social Protection and Health Division of the Inter-American Development Bank (IDB). Email: [email protected]; [email protected]; [email protected]; ms[email protected]. This study was elaborated with funding from IDB’s Economic and Sector Work RG-E1871 “Can long-term care services reduce healthcare costs through shorter hospitalizations?”. We thank Ricardo Pérez-Cuevas, Ignacio Astorga, and Hugo Godoy for suggestions and guidance. We are also grateful to Pablo Ibarrarán, David Evans, Agustin Filippo and an anonymous reviewer for their useful comments; Nadin Medellin and Diego Wachs for their support in the data processing at the early stages of this research; Rocío Aguilera for her support in the qualitative analysis. Finally, we thankfully acknowledge the contributions from experts who kindly participated in interviews and focus group discussions that informed the construction of the conceptual framework. The document was professionally edited by Guillermo Rubens. Remaining errors are ours only. The content and findings of this paper reflect the opinions of the authors and not necessarily those of the IDB, its Board of Directors, or the countries they represent. 43 Table of Contents Are long hospitalizations substituting primary and long-term care? Evidence from Brazil and Mexico ........................................................................................ Error! Bookmark not defined. 1. Introduction ......................................................................................................................... 3 2. Conceptual framework ........................................................................................................ 4 3. Evidence from the existing literature ................................................................................... 7 3.1. Determinants of length of stay ...................................................................................... 7 3.2. Costs of long hospitalizations ....................................................................................... 8 3.3. Definitions of prolonged hospital stays ......................................................................... 9 4. Data and methodology ......................................................................................................10 4.1. Data sources ...............................................................................................................10 4.2. Definition of Prolonged Hospitalizations and Decomposition of Length of Stay ...........12 5. Evidence from Brazil and Mexico .......................................................................................16 5.1. Prolonged hospitalizations account for approximately half of all hospital days ............16 5.2. Which conditions account for most excessive days? ...................................................17 5.3. How much can be saved by providing rehabilitation, social, and long-term care services? ..........................................................................................................................................25 6. Discussion .........................................................................................................................26 7. Conclusions and policy recommendations .........................................................................28 References ...........................................................................................................................30 Annex 1. Findings from interviews and focus group discussions ............................................35 Annex 2. Decomposition of hospital days in Brazilian and Mexican states .............................40 Annex 3. Sensitivity analysis .................................................................................................42 43 1. Introduction In the Latin American and Caribbean region, the convergence of an aging population and technological advancements is expected to significantly raise healthcare spending (Rao et al., 2022). This increase is further compounded by the rising prevalence of chronic diseases and dependence among older adults. Over 85% of those aged 70 and above have at least one chronic condition, and 14% of those over 65 require assistance with activities of daily living (IHME, 2020; Aranco, Ibarrarán, and Stampini, 2022). Furthermore, there is a notable lack of robust primary health, social, post-operative, and long-term care systems, as well as support for family caregivers (Aranco et al., 2022). Rationalizing hospital use is a key strategy for controlling rising health expenditures. Hospitals account for about one-third of total health spending in the region and are central to the adoption of costly medical technology. They also bear the consequences of inadequate primary care and social systems, which can lead to hospital stays extending beyond what is medically necessary. A prolonged hospitalization occurs when "a medically fit patient is needlessly kept in hospital due to internal organizational/operational factors or where a patient is flagged as in need of alternate level of care and is delayed because of deferred transition of care and/or lack of external transferof-care arrangements" (Micallef et al., 2020, p. 105). Prolonged hospitalizations can be driven by several factors. First, preventable comorbidities or patient frailty may extend hospital stays. Second, inefficiencies within the hospital may lead to longer stays. Third, a lack of appropriate discharge destinations that offer rehabilitative care or social support can also lead to prolonged hospitalizations. Additionally, hospitals face admissions and readmissions that are entirely avoidable with effective primary and social care. Older persons are particularly at risk of prolonged hospitalizations due to their more complex health conditions, frailty at admission, and the need for safe discharge arrangements (Picone et al., 2003; Lenzi et al., 2014). Prolonged hospital stays are common, costly, and risky for patients. A meta-analysis of 64 studies conducted in Europe and North America found that such stays account for an average of 22.8% of all bed days. The figures range from 1.6% in England to 91.3% in Canada, depending on the methodologies, data sources, and populations studied (Landeiro et al., 2019). There is also substantial within-country variation. These prolonged stays contribute to increased healthcare costs and can worsen access and wait times when hospital capacity is limited (Falcone et al., n.d.; Landeiro et al., 2019). Additionally, prolonged hospitalizations can be potentially unsafe for patients (Lingsma et al., 2018; Landeiro et al., 2019; Rojas‐García et al., 2018). Existing research has identified several driving factors of prolonged hospitalizations, including a lack of adequate care structures outside the hospital, as well as hospital and inter-hospital processes from admission to discharge, such as early admission to reserve a bed for a scheduled procedure or administrative delays (Landeiro et al., 2019; Siddique et al., 2021; Micallef et al., 2020). In this study, we conceptualize and empirically measure hospital days due to prolonged stays among older people in Brazil and Mexico. First, we propose a continuum-of-care conceptual framework for prolonged stays based on existing literature and qualitative insights from interviews and focus group discussions with experts from Mexico, Argentina, and Colombia. The framework categorizes drivers of prolonged stays into three parts: (i) lack of appropriate primary healthcare, leading to more complex admissions; (ii) hospital inefficiency; and (iii) lack of rehabilitation, social, and long-term care at discharge. Second, we estimate the prevalence of prolonged stays and the contribution of these three factors using administrative records on discharges for patients aged 65 and older from public sector hospitals in Brazil and Mexico from 2019. Finally, we estimate the total cost of inappropriate days, by multiplying their number by the average cost of one day of 43 hospitalization (including infrastructure and equipment amortization, procedures and human resources). Our results suggest that prolonged stays are highly prevalent and costly, primarily driven by hospital inefficiencies and the lack of discharge destinations that provide post-operative, rehabilitative, and social support. Specifically, we estimate that inappropriate hospital days account for 48.1% of hospital days in Brazil and 56.2% in Mexico. The scarce supply of rehabilitation, social, and long-term care services accounts for 12.1% of all hospital days in Brazil and 6.9% in Mexico. To the best of our knowledge, this is the first paper to provide such a decomposition of the causes of prolonged hospitalizations, allowing for an initial estimation of the savings that could be achieved through the provision of long-term care. The remainder of the paper is organized as follows. Section 2 outlines the conceptual framework that situates hospitals within a broader health and social care system. We illustrate how primary healthcare, hospital inefficiency, and rehabilitation, social, and long-term care can affect the length of stay. Section 3 reviews the literature on the determinants, definition, and costs of long hospital stays. Section 4 describes the data and explains the methodology used to define prolonged hospitalizations, and decompose the length of stay into its components, following the logic of our conceptual model. Section 5 presents the results on the magnitude of inappropriate hospital days and information on the conditions that contribute most to these excessive days, as well as the potential savings from improved post-discharge services. In Section 6, we further discuss our findings. Section 7 concludes and provides policy recommendations. 2. Conceptual framework We developed a continuum-of-care conceptual framework for prolonged hospital stays based on existing literature and qualitative insights from interviews and focus group discussions with experts from Mexico, Argentina, and Colombia. These discussions and interviews, which included 6 participants from focus groups and 3 medical doctors, were conducted virtually between June and October 2023. The data was analyzed using thematic analysis, an inductive approach that helped us identify key themes and patterns. This analysis allowed us to conceptualize the information into three stages of care: pre-hospital, in-hospital, and post-hospital. Box 1 presents selected quotes that informed our model, and Annex 1 provides a more detailed summary of the findings. Our framework views hospitals as part of a broader health and social care system that includes primary care and services for rehabilitation, social support, and long-term care (Falcone et al., n.d.). Prolonged hospital stays can result from inefficiencies and bottlenecks at any stage of this care continuum, leading to avoidable admissions or longer hospitalizations than medically necessary. For instance, patients may be admitted too early or too late, experience delays in becoming clinically fit for discharge once admitted, or face discharge delays due to a lack of posthospital care support. Figure 1 illustrates our conceptual framework and identifies three potential sources of inappropriate hospital days: 1. Intake Issues: Both inadequate primary healthcare (A) and insufficient rehabilitation, social, and long-term care services (B) can lead to avoidable admissions or readmissions and increase patient frailty, contributing to prolonged hospitalizations (Component 1) (Freitas et al., 2012; Lenzi et al., 2014; Bo et al., 2016; Toh et al., 2017). For example, an older person hospitalized for a femur fracture may require a longer stay if they have poorly 43 managed chronic conditions that need stabilization or additional care needs that complicate treatment. The same considerations apply to readmissions. Lack of primary healthcare (A) and insufficient rehabilitation, social, and long-term care (B) increase the probability of rehospitalization for people who have been previously discharged and may complicate the clinical picture, extending the duration of these readmissions (Misky et al., 2010). Additionally, too-early discharges due to internal hospital issues may also increase the likelihood of readmission. 2. In-Hospital Inefficiencies: Prolonged stays during hospitalization (C) may arise from inefficiencies such as lack of resources, delays in procedures, or poor planning and management (Component 2) (Holmås, Kamrul Islam, et al., 2013). For instance, hospitals with a lower physician-to-patient ratio may experience longer stays as patients wait longer for consultations or test results (Marfil-Garza et al., 2018; Carey et al., 2005). 3. Discharge Delays: Discharges for clinically fit patients may be delayed due to a lack of available rehabilitation, social, and long-term care services (B) (Component 3) (Toh et al., 2017; Landeiro et al., 2016; Moore et al., 2015; Carey et al., 2005). For example, older patients may remain hospitalized if they need care or rehabilitation that cannot be provided at home due to a lack of family support or public home care services. Alternatively, they may face delays if suitable institutional arrangements (e.g., rehabilitation centers or longterm care facilities) are unavailable. Such needs may arise from the hospitalization itself or preexist but become more pronounced post-hospitalization. Additionally, families might use the hospital stay as an opportunity to obtain care from public services and may seek to delay or prevent the patient's discharge. Below, we operationalize this framework to estimate the contribution of the three components – intake, in-hospital, and discharge – to the overall count and prevalence of inappropriate hospital days (prolonged stays) (Figure 1). In practice, distinguishing between hospitalizations and readmissions is challenging because each hospitalization episode is recorded separately and cannot be linked to previous stays. Moreover, we cannot determine whether readmissions result from issues with the initial stay (e.g., premature discharge) or deficiencies in non-hospital support services. Thus, the effect on readmissions is also considered under Component 1. Additionally, we cannot determine whether a stay is prolonged due to family refusal to discharge the patient. 43 Table 1. Main characteristics of the sample Brazil Mexico Number of hospitalizations 2,166,900 768,173 Number of days of hospitalization 15.006.168 4,639,140 Average length of stay (days) (Standard deviation) 6.9 (10.6) 6.0 (30.4) Age 65-69 (%) 26.8 27.9 Age 70-74 (%) 23.2 24.0 Age 75-79 (%) 19.6 19.7 Age 80-84 (%) 15.0 14.4 Age 85+ (%) 15.3 14.0 Average age (years) 75.7 75.4 Females (%) 49.7 51.8 Males (%) 50.3 48.2 % with comorbidities 22.1 - Charlson comorbidity index 0.27 Type of admission: Elective (%) 17.3 - Type of admission: Emergency (%) 82.7 - Procedure complexity: Medium (%) 89.6 - Procedure complexity: High (%) 10.4 - In-hospital mortality (%) 12.8 10.6 Source: Authors’ elaboration based on DATASUS hospitalization database, 2019 and Mexico's Health Sector Hospital Discharge Database, 2019. 4.2. Definition of Prolonged Hospitalizations and Decomposition of Length of Stay Following the conceptual framework, we decompose a hospital length of stay (LOS) into four parts: (i) the medically appropriate stay (T); (ii) excessive days due to the lack of appropriate primary healthcare that leads to more complex admissions (ED1); (iii) excessive days due to hospital inefficiency (ED2); (iv) excessive days due the lack of rehabilitation, social, and long-term care at discharge (ED3). We include all primary diagnoses in the analysis, even those conditions that should have been prevented at the primary care level. Extensive evidence suggests that a robust primary care system can reduce hospitalizations. However, this paper does not aim to quantify the potential savings from such reductions; instead, it focuses on understanding the factors contributing to prolonged hospital stays once admission has occurred. Box 2 illustrates the decomposition using four stereotypical examples. For each condition, we define the medically appropriate duration of stay as the average length of stay in the most efficient state (the “benchmark state”) or, in other words, in the state with the lowest average length of stay for that condition. To calculate this average, we restrict the sample to patients without comorbidities (secondary diagnoses) that could have been prevented at the primary level. That is, only patients without Ambulatory-Care-Sensitive Conditions (ACSC) comorbidities are considered to determine the benchmark state. By restricting the sample in this way, we remove the portion of the stay that is attributed to clinical complications linked to comorbidities that could have been managed through appropriate primary healthcare. 3 Implicitly, 3 For hospitalizations due to ACSC (as primary condition), both LOS and ED1 could be avoided through appropriate primary healthcare. 43 we assume that in the benchmark state, ACSC comorbidities are properly managed at the primary level. We proceed as follows. For each condition, we calculate the average duration of hospitalizations for patients without ACSC comorbidities by state (𝐿𝑂𝑆      𝑐,𝑠). We then take the lowest average value as the threshold that defines the medically appropriate stay 𝑇𝑐. The calculations are given by equations [1] and [2]. 𝐿𝑂𝑆      𝑐,𝑠 =∑𝐿𝑂𝑆𝑖,𝑐,𝑠 ⬚ 𝑖,𝑐,𝑠 ∑𝐼𝑖,𝑐,𝑠 ⬚ 𝑖,𝑐,𝑠 for 𝑖 without ACSC − comorbidities [1] 𝑇𝑐=min𝑠( 𝐿𝑂𝑆      𝑐,𝑠) [2] Where 𝐿𝑂𝑆𝑖,𝑐,𝑠 is the length of stay of patient i, for condition c in state s. I is an indicator equal to 1 that counts the hospitalizations for the purpose of calculating the average length of stay. Equation [2] identifies T and the corresponding benchmark state (BS), for each condition c. For each hospitalization, the number of excessive days 𝐸𝐷 is the difference between the actual length of stay and the threshold, as shown in equation [3]. 𝐸𝐷 is also equal to the sum of its three components, as shown in equation [4]. 𝐸𝐷𝑖,𝑐 = 𝐿𝑂𝑆𝑖,𝑐 − 𝑇𝑐 𝑖𝑓 𝐿𝑂𝑆𝑖,𝑐 > 𝑇𝑐; 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 [3] 𝐸𝐷 =𝐸𝐷1 + 𝐸𝐷2 + 𝐸𝐷3 [4] To calculate 𝐸𝐷1, i.e., the excessive days that could have been avoided through appropriate management of ACSC comorbidities at the primary level, we calculate the average length of stay (𝑍𝑐) among patients with ACSC comorbidities hospitalized for condition c in the benchmark state BS identified in equation [2]. The formula is shown in equation [5]. 𝑍𝑐=∑𝐿𝑂𝑆𝑖,𝑐,𝐵𝑆 ⬚ 𝑖,𝑐,𝐵𝑆 ∑𝐼𝑖,𝑐,𝐵𝑆 ⬚ 𝑖,𝑐,𝐵𝑆 for 𝑖 with ACSC − comorbidities [5] Only patients reporting ACSC comorbidities are considered in equation [5]. The assumption is that poor management of ACSC comorbidities would lead to longer stays even in the most efficient state, because it increases the complexity of a patient’s clinical picture at admission. This implies that Z is larger than T. 4 For patients with ACSC comorbidities in all states, 𝐸𝐷1 is computed as the difference between 𝑍𝑐 and 𝑇𝑐, or the difference between the actual length of stay and 𝑇𝑐 if the length of stay is shorter than 𝑍𝑐. For patients without ACSC comorbidities, 𝐸𝐷1 is zero by definition. This is summarized in equation [6]. 4 In few cases where Z<T, we set Z=T. Also, in very few outlying cases in which Z exceeds 2T, we set Z=2T. 43 𝐸𝐷1𝑖,𝑐 = 0 for 𝑖 without ACSC − comorbidities [6] 𝐸𝐷1𝑖,𝑐 = 𝑍𝑐− 𝑇𝑐 if 𝑖 has ACSC − comorbidities and 𝐿𝑂𝑆𝑖,𝑐 > 𝑍 𝐸𝐷1𝑖,𝑐 = 𝐿𝑂𝑆𝑖,𝑐 − 𝑇𝑐 if 𝑖 has ACSC − comorbidities and 𝑍𝑐>𝐿𝑂𝑆𝑖,,𝑐 > 𝑇𝑐 𝐸𝐷1𝑖,𝑐 = 0 if 𝑖 has ACSC − comorbidities and 𝐿𝑂𝑆𝑖,𝑐 < 𝑇 This calculation of 𝐸𝐷1 can only be done for Brazil, as the Mexican data do not include information on comorbidities. Consequently, for Mexico 𝐸𝐷1 is included partly in T and partly in 𝐸𝐷2. It should also be noted that a strong primary healthcare would avoid admissions due to ACSC altogether. That is, the 𝑇𝑐 part of admissions due to ACSC as primary conditions can also be considered excessive days (more specifically, ED1). However, for this analysis, we aim to identify the contribution of ACSC comorbidities to prolonged hospitalizations, even in cases where a person has been hospitalized due to a primary ACSC. After accounting for ED1, the decomposition of the remaining excessive days depends on whether the condition that caused admission generates new post-discharge care needs. For instance, people with musculoskeletal conditions will require rehabilitation care after leaving the hospital. Similarly, neurological conditions and systemic diseases, mainly respiratory and cardiac failures, affect a patient's mobility and thus generate new post-discharge rehabilitation needs. In contrast, patients with more generic conditions, such as diabetes, can recover without rehabilitation care. We identify the conditions that generate new care needs (CN) through expert opinions who assessed a set of 35 diagnoses that most contribute to excessive hospital days in our analysis. 5 We assume that conditions not assessed belong to the no-new-care-needs (NCN) group. If a condition does not create new care needs (NCN), we assume that all remaining excessive days are due to hospital inefficiency (ED2). If a condition generates post-discharge care needs (CN), the remaining excessive days are further disaggregated into ED2 and lack of rehabilitation, social, and long-term care (ED3). For conditions in the NCN group, the component due to hospital inefficiency is defined by equation [7]. 𝐸𝐷2𝑖,𝑐 = 𝐿𝑂𝑆𝑖,𝑐 − 𝑇𝑐−𝐸𝐷1𝑖,𝑐 𝑖𝑓 𝑐 ∈ 𝑁𝐶𝑁 𝑎𝑛𝑑 𝐿𝑂𝑆𝑖,𝑐 >(Tc+𝐸𝐷1𝑖,𝑐); 0 otherwise [7] For conditions in the CN group, we assume that hospital inefficiency is equal to the average inefficiency observed for the NCN group, 𝐸𝐷2       , which is defined as: 5 The following conditions were classified as generating medium to high care needs: Angina pectoris; Bacterial infection of unspecified site; Bacterial pneumonia, not elsewhere classified; Cerebral infarction; Fracture of the femur; Fracture of lower leg, including ankle; Heart failure; Other chronic obstructive pulmonary disease; Other degenerative diseases of nervous system, not elsewhere classified; Other sepsis; Pneumonia, organism unspecified; Sequelae of cerebrovascular disease; Shock, not elsewhere classified; Stroke, not specified as hemorrhage or infarction. The following conditions were assessed as generating no or low care needs: Acute myocardial infarction, Cholecystitis; Cholelithiasis; Chronic ischemic heart disease; Chronic kidney disease; Epilepsy; Essential (primary) hypertension; Intracranial injury; Malignant neoplasm of colon; Other bacterial diseases, not elsewhere classified; Other cerebrovascular diseases; Other diseases of digestive system; Other disorders of fluid, electrolyte and acid-base balance; Other disorders of skin and subcutaneous tissue, not elsewhere classified; Other disorders of urinary system; Other peripheral vascular diseases; Paralytic ileus and intestinal obstruction without hernia; Respiratory failure, not elsewhere classified; Type 2 diabetes mellitus; Unknown and unspecified causes of morbidity; Unspecified diabetes mellitus. See also Table 4. 43 𝐸𝐷2       =∑𝐸𝐷2𝑖,𝑐 ⬚ 𝑖,𝑐 ∑𝐼𝑖,𝑐 ⬚ 𝑖,𝑐 ,for 𝑐 ∈ 𝑁𝐶𝑁 [8] 𝐸𝐷2 is then defined by equation [9]. 𝐸𝐷2𝑖,𝑐 =𝐸𝐷2       𝑖𝑓 𝑐 ∈ 𝐶𝑁 𝑎𝑛𝑑 𝐿𝑂𝑆𝑖,𝑐 >(Tc+𝐸𝐷1𝑖,𝑐 +𝐸𝐷2       ) [9] 𝐸𝐷2𝑖,𝐶 = 𝐿𝑂𝑆𝑖,𝑐 − Tc−𝐸𝐷1𝑖,𝑐 if 𝑐 ∈ 𝐶𝑁 𝑎𝑛𝑑 (Tc+𝐸𝐷1𝑖,𝑐 +𝐸𝐷2       )> LOSi,c >(Tc+𝐸𝐷1𝑖,𝑐) 𝐸𝐷2𝑖,𝐶 = 0 otherwise Finally, 𝐸𝐷3, the number of excessive days that are due to the lack of rehabilitation, social and long-term care services for conditions that generate new care needs is defined by equation [10]. For conditions that generate no care needs, ED3 is zero by definition. 𝐸𝐷3𝑖,𝑐 = 0 for 𝑐 ∈ 𝑁𝐶𝑁 [10] 𝐸𝐷3𝑖,𝑐 = 𝐿𝑂𝑆𝑖,𝑐 − 𝑇𝑐−𝐸𝐷1𝑖,𝑐 −𝐸𝐷2𝑖,𝑐 if 𝑐 ∈ 𝐶𝑁 𝑎𝑛𝑑 𝐿𝑂𝑆𝑖,𝑐 >(Tc+𝐸𝐷1𝑖,𝑐 +𝐸𝐷2𝑖,𝑐), 𝐸𝐷3𝑖,𝐶 = 0 otherwise Box 2 illustrates this decomposition analysis for different examples of conditions and types of patients. Box 2. Examples of decomposition calculations, Brazilian database Example 1: Patient with acute myocardial infarction, no ACSC comorbidities. LOS Tc Zc ACSC New care needs ED ED1 ED2 ED3 9 5.1 6.2 No No 3.9 0 3.9 0 Example 2: Patient with acute myocardial infarction, with ACSC comorbidities. LOS Tc Zc ACSC New care needs ED ED1 ED2 ED3 10.2 5.1 6.2 Yes No 5.1 1.1 4.0 0 Example 3: Patient with fracture of the femur, no ACSC comorbidities. LOS Tc Zc ACSC New care needs ED ED1 ED2 ED3 12 6.7 7.0 No Yes 5.3 0 3.1 2.2 Example 4: Patient with fracture of the femur, with ACSC comorbidities. LOS Tc Zc ACSC New care needs ED ED1 ED2 ED3 12 6.7 7.0 Yes Yes 5.3 0.2 3.1 2.0 43 5. Evidence from Brazil and Mexico 5.1. Prolonged hospitalizations account for approximately half of all hospital days The decomposition analysis shows that excessive days represent 48.1% of total hospital days in Brazil and 56.2% in Mexico (Figure 3). These excessive days come from 1 million hospitalizations in Brazil and 440,000 hospitalizations in Mexico classified as prolonged, representing 46% and 57% of the total number of hospitalizations, respectively. In Brazil, we estimate that 0.5% of total hospital days result from the increased fragility and clinical complexity of patients with secondary conditions that could have been managed at the primary care level (referred to as ACSC comorbidities). Additionally, 35.5% of hospital days are attributable to inefficiencies within hospitals, such as lack of resources and management models. Finally, 12.1% of hospital days could be avoided by providing better rehabilitation, social, and long-term care services. In Mexico, hospital inefficiency is the predominant factor, accounting for 49.3% of total hospital days. The percentage of hospital days that could be saved through the provision of rehabilitation, social and long-term care services is lower than in Brazil, at a 6.9%. As discussed in the methodology section, the data from Mexico does not allow for the estimation of the share of excessive days due to mismanagement of ACSC comorbidities. Therefore, these excessive days are partly included in the estimated medically appropriate stay and partly in the share of excessive days due to hospital inefficiency. Our analysis considers all main causes of hospitalizations, including those for conditions that could have been avoided with better primary care (i.e., hospitalizations with a primary diagnosis of ACSC). From the hospital’s perspective, these are valid admissions. However, from the broader perspective of a health system, these admissions should not have occurred and therefore the appropriate length of stay should be zero days, making all days for these conditions excessive. Expected days of stay from ACSC constitute 14.4% and 12.5% of all hospital days in Brazil and Mexico, respectively. If we count all days from admissions with primary diagnosis of ACSC as medically inappropriate, excessive days account for 64.7% and 68.7% of total hospital days in Brazil and Mexico, respectively. 43 Figure 3. Decomposition of hospital days in Brazil and Mexico, 2019 Source: Authors’ elaboration based on DATASUS hospitalization database, 2019 and Mexico's Health Sector Hospital Discharge Database, 2019. Note: Component ED1 cannot be estimated for Mexico, due to lack of information on ACSC comorbidities. The share of excessive days that can be attributed to the different components varies by state (Annex 2, Tables A2.1 and A2.2), especially for the part related to the lack of rehabilitation, social, and long-term care services. In Brazil, this component accounts for a share of total hospital days ranging between 3.6% in Mato Grosso do Sul and 19% in Amapá. In Mexico, it accounts for a share of total hospital days ranging from 5% in Michoacán de Ocampo and Tabasco to 10.7% in Mexico City. Hospital inefficiency accounts for 24.8% in Paraná and up to 46.0% in Rio de Janeiro. In Mexico, these figures range from 42.4% in Colima to 52.9% in Baja California. 5.2. Which conditions account for most excessive days? 43 Table 2 shows that in both Brazil and Mexico, the expected stay 𝑇𝑐 varies substantially across conditions. Considering the 20 conditions responsible for most of the excessive days, in Brazil, this parameter ranges between just over 2 days for hypertension to nearly 10 days for sequalae of stroke. In Mexico, it varies from 2.6 days for unknown causes of morbidity to slightly more than 6 days for a fracture of the femur. Similarly, the observed average length of stay varies greatly across conditions. For the same 20 conditions in both countries, the average length of stay in Brazil ranges from 4.3 days for cholelithiasis to 32.6 days for sequelae of cerebrovascular disease. In Mexico, it varies between 4.7 days in the case of hypertension and 9.7 days for fracture of the femur. 19 Table 2 also shows that just 20 conditions account for 51% of excessive days in Brazil (Panel A) and 41% in Mexico (Panel B). In Brazil, two infection-related diseases, pneumonia and sepsis, are at the top of the chart, jointly accounting for more than 11% of excessive days (Table 2, Panel A). If we add bacterial pneumonia, bacterial infection, erysipelas, and other bacterial diseases, the share due to infectious diseases reaches 18% of excessive days. Cardiovascular diseases also rank high, with 5.5% of excessive days due to heart failure, 4.3% to stroke, 3.1% to acute myocardial infarction, 2.2% to angina pectoris, and 1.3% to sequelae of cerebrovascular diseases. Taken together, these cardiovascular conditions account for 16.6% of excessive days. In Mexico, the fracture of the femur is the condition that accounts for most excessive days, 5.4% of the total (Table 2, Panel B). Chronic kidney disease, pneumonia, and diabetes mellitus (type 2) jointly account for an additional 11.4%. Table 3 indicates key demographic and hospitalization characteristics for the conditions reported in Table 2, comparing the full sample of hospital stays with the subsample of prolonged hospitalizations by country. Overall, there are no clear patterns differentiating prolonged hospitalizations from all hospital stays. In both Brazil (Table 3, Panel A) and Mexico (Table 3, Panel B), demographic and health characteristics are similar among subsamples, except for conditions like cholelithiasis, unspecified diabetes mellitus, and cholecystitis that show slightly lower percentages for females among prolonged stays in Mexico (Table 3, Panel B). Additionally, in Brazil, the type of admission (elective versus emergency) and the complexity of the procedure undertaken do not differ substantially between the two samples. In addition, in both countries, inhospital mortality varies slightly across subsamples, but without a clear pattern, and the magnitude of the differences is relatively small. In Brazil (Table 3, Panel A), for heart failure, stroke, and bacterial diseases, excessive days appear correlated with a higher mortality rate. In contrast, for conditions like sepsis and sequelae of stroke excessive days seem associated with lower levels of mortality. In Mexico (Table 3, Panel B), prolonged hospitalizations show lower inhospital mortality rates for acute myocardial infarction and sepsis, and higher rates for diseases classified as unknown and unspecified causes of morbidity. From Table 3, it seems that the quantity of excessive days is primarily attributed to the cause of hospitalization rather than the patient's characteristics. This aligns with our approach of using an index that varies by condition to identify the component of excessive days due to lack of rehabilitation, social and long-term care services. 20 Table 2. Top 20 conditions responsible for excessive days of hospitalization - Panel A. Brazil Thresholds All stays Long stays T1 Z2 av. days3 SD4 days from condition5 % in total days6 cases from condition7 % of long hosp.8 av. days9 SD10 # exc. days11 share in excessive days12 Pneumonia, organism unspecified 5.46 7.61 7.17 7.36 1,024,904 6.8% 142,972 46% 11.99 8.64 424,799 5.9% Other sepsis 5.78 5.78 11.23 11.37 695,576 4.6% 61,914 64% 15.98 11.82 402,662 5.6% Heart failure 5.11 6.97 7.39 7.87 868,204 5.8% 117,441 44% 12.82 9.23 400,795 5.5% Stroke, not specified as hemorrhage or infarction 5.52 9.35 7.62 9.00 684,707 4.6% 89,896 44% 13.41 11.01 312,943 4.3% Acute myocardial infarction 5.07 6.17 8.22 9.30 431,797 2.9% 52,547 48% 14.08 10.66 225,682 3.1% Bacterial pneumonia, not elsewhere classified 5.00 8.67 7.47 7.86 472,917 3.2% 63,320 47% 12.39 9.19 219,401 3.0% Other disorders of urinary system 4.10 5.00 6.41 7.15 430,282 2.9% 67,101 50% 10.36 8.51 206,121 2.9% Other chronic obstructive pulmonary disease 4.60 6.24 6.84 9.39 384,751 2.6% 56,228 47% 11.45 12.04 182,348 2.5% Fracture of the femur 6.76 7.00 8.62 8.21 461,928 3.1% 53,559 49% 13.71 9.22 181,857 2.5% Chronic kidney disease 5.84 6.50 9.73 11.30 311,414 2.1% 32,000 54% 15.67 12.62 169,391 2.3% Angina pectoris 3.16 6.31 5.52 6.70 291,174 1.9% 52,735 46% 9.74 7.94 160,969 2.2% Bacterial infection of unspecified site 7.60 9.38 9.48 9.81 320,050 2.1% 33,760 42% 16.93 11.28 133,193 1.8% Respiratory failure, not elsewhere classified 5.10 5.10 9.37 11.14 172,712 1.2% 18,430 50% 16.05 12.57 100,645 1.4% Sequelae of cerebrovascular disease 9.78 11.00 32.63 66.18 126,058 0.8% 3,863 60% 52.07 80.16 97,169 1.3% Other diseases of digestive system 3.66 5.24 5.42 6.18 201,272 1.3% 37,167 49% 8.96 7.26 96,586 1.3% Essential (primary) hypertension 2.32 2.54 5.85 19.20 115,465 0.8% 19,723 46% 10.82 27.45 77,307 1.1% Cholelithiasis 3.07 3.07 4.34 6.17 134,577 0.9% 31,037 31% 10.23 8.52 68,385 0.9% Other bacterial diseases, not elsewhere classified 6.14 12.27 10.94 11.07 123,663 0.8% 11,303 56% 16.64 11.97 66,472 0.9% Unspecified diabetes mellitus 3.98 3.98 6.12 6.94 138,245 0.9% 22,587 55% 9.21 8.08 65,481 0.9% Erysipelas 5.02 10.04 7.41 7.24 146,680 1.0% 19,804 50% 11.45 8.36 64,149 0.9% Total 7,536,377 50% 987,387 3,656,357 51% 21 Panel B. Mexico Threshold All stays Long stays T1 av. days3 SD4 days from condition5 % in total days6 cases from condition7 % of long hosp.8 av. days9 SD10 # exc. days11 share in excessive days12 Fracture of the femur 6.34 9.66 11.46 307,947 6.6% 31,885 62% 13.42 13.15 139,905 5.4% Chronic kidney disease 3.26 5.70 10.32 228,505 4.9% 40,092 49% 9.84 13.58 128,604 4.9% Pneumonia, organism unspecified 5.66 8.13 93.93 193,211 4.2% 23,757 50% 13.23 132.29 90,487 3.5% Type 2 diabetes mellitus 3.71 5.63 8.44 158,308 3.4% 28,102 52% 9.09 10.50 79,132 3.0% Other disorders of urinary system 3.60 6.25 7.43 106,890 2.3% 17,099 63% 8.79 8.39 55,630 2.1% Other chronic obstructive pulmonary disease 3.88 5.64 7.57 120,058 2.6% 21,273 61% 8.01 8.95 53,145 2.0% Shock, not elsewhere classified 5.28 8.43 11.10 93,702 2.0% 11,118 47% 15.21 13.08 52,219 2.0% Heart failure 3.69 6.01 6.16 89,461 1.9% 14,893 61% 8.54 6.70 44,278 1.7% Other cerebrovascular diseases 4.15 6.09 9.12 92,546 2.0% 15,190 46% 10.36 12.01 43,690 1.7% Other diseases of digestive system 4.01 5.64 8.22 101,288 2.2% 17,966 47% 9.13 10.89 43,439 1.7% Cholelithiasis 3.64 4.94 8.79 88,039 1.9% 17,833 41% 9.40 12.31 42,584 1.6% Chronic ischemic heart disease 3.50 6.65 8.03 72,018 1.6% 10,829 56% 10.44 9.04 42,137 1.6% Acute myocardial infarction 3.60 6.63 21.07 71,377 1.5% 10,769 63% 9.47 26.22 39,573 1.5% Other sepsis 5.62 9.57 36.13 66,528 1.4% 6,953 52% 16.10 49.28 37,761 1.4% Unknown and unspecified causes of morbidity 2.65 7.37 8.30 54,404 1.2% 7,383 69% 9.98 8.78 37,495 1.4% Essential (primary) hypertension 3.21 4.73 7.13 60,199 1.3% 12,721 45% 8.33 9.41 29,327 1.1% Unspecified diabetes mellitus 4.71 6.28 6.72 65,272 1.4% 10,400 49% 10.49 7.55 29,215 1.1% Fracture of lower leg, including ankle 5.03 7.29 9.10 53,434 1.2% 7,327 49% 12.14 10.90 25,783 1.0% Cholecystitis 3.48 4.85 19.25 52,541 1.1% 10,832 41% 9.26 29.58 25,542 1.0% Other disorders of fluid, electrolyte and acid-base balance 3.50 6.42 19.35 44,100 1.0% 6,868 58% 9.77 24.99 24,791 1.0% Total 2,119,828 46% 323,290 1,064,735 40.8% Notes: (1) medically appropriate stay for patients without ACSC comorbidities; (2) threshold for patients with ACSC comorbidities; (3) average length of stay; (4) standard deviation of length of stay; (5) total days attributed to the condition; (6) percentage of total days attributed to the condition; (7) number of hospitalizations attributed to the condition; (8) percentage of hospitalization from condition that contain excessive days; (9) average length of stay, in sample of hospitalizations that include excessive days; (10) standard deviation of average length of stay, in sample of hospitalizations that include excessive days; (11) excessive days attributed to the condition; (12) excessive days attributed to the condition as a percentage of total excessive days. Source: Authors’ elaboration based on DATASUS hospitalization database, 2019 and Mexico's Health Sector Hospital Discharge Database, 2019. 28 after generates two data records that cannot be linked. Similarly, patients who have been transferred to another establishment or have undergone a change in procedure generate several data records from which we cannot calculate the total length of stay. These data features artificially shorten the average length of stay, thus underestimating the number of excessive days. Second, our analysis does not adjust for in-hospital mortality, a factor that truncates the length of stay of some patients. To address this issue in studies of long hospitalizations, some authors drop the observations that end with the death of the patient. Our data, however, show that the correlation between mortality and length of stay is positive in some cases and negative in others. For this reason, we make no corrections. Third, our data do not allow us to precisely identify the clinically appropriate length of stay for each hospitalization. This would only be possible by analyzing patients’ medical records and assessing, case by case, the optimal clinical length of stay. Our threshold is adjusted solely for the reason of admission and assumes that all hospitalizations due to a given reason should last the same. We plan to collect complementary information through detailed analysis of medical records for future research. Studies utilizing data from medical records or professional opinions generally focus on smaller samples drawn from a specific healthcare institution. Their results confirm that, even considering the clinically justified delays, long hospitalizations account for a large proportion of total hospitalizations. For example, in a study conducted in Italy, Bo et al. (2016) show that 31.5% of hospitalizations could be classified as long by clinical standards, while Hendy et al. (2012) estimate this figure to be at nearly 50% for a London hospital. The threshold we use to identify excessive days is arbitrary and assumes that cross-state variability provides information on the medically appropriate stay. This may be subject to some errors. For example, a state might report a short length of stay for a certain condition because of high in-hospital mortality or because complex cases are transferred to another hospital. States with a large proportion of university hospitals may report longer stays as these institutions receive complex cases and part of the stay is dedicated to training (Freitas et al., 2012; Walker et al., 2021). In our threshold, the relationship between patients’ frailty and length of stay is only adjusted for the reason of hospitalization and the existence of ACSC comorbidities. For Brazil, only approximately 5% of hospitalizations are recorded with ACSC comorbidities. For Mexico, this information is not available in our data. Due to this likely underreporting (or lack of data on) ACSC comorbidities, we may be overestimating the percentage of prolonged days overall, and the part due to hospital inefficiency. In Annex 3, we present a sensitivity analysis based on a different threshold, for each condition, set at the average length of stay in the state with the median – instead of the minimum – of this average. With this alternative threshold, 36% of hospitalization days are excessive in both Brazil and Mexico. Similar to our main analysis, almost 10% of hospital days in Brazil and 6% in Mexico are due to the lack of rehabilitation, social and long-term care services (Figure A3.1, Annex 3). Fourth, the component that measures hospital inefficiency may be affected by factors we are not controlling for. A teaching hospital, for example, may report larger average stays even with high efficiency levels. Finally, our analysis is likely to underestimate the share of hospital days that can be saved through rehabilitation, social, and long-term care services. Our index for post-discharge care needs only classifies 36 of the more than 1,400 admission conditions available in Brazil and Mexico and we only consider post-discharge care needs that are directly caused by the reason of admission. Due 29 to lack of data, we are unable to account for increased frailty resulting from long hospitalizations due to any condition, even in patients that have no previous care needs. For already frail patients, a common situation among older persons, a few days at the hospital can create significant loss of autonomy. Given the proportion of older persons with care needs in Brazil (10.5%) and Mexico (25.2%) (Aranco, Ibarrarán and Stampini, 2022), this underestimation may thus be large. 7. Conclusions and policy recommendations Our results highlight the importance of developing post-discharge care services to reduce excessive days of hospitalizations among older people. A strong care system outside hospitals allows patients who are clinically fit for discharge, but still need rehabilitation or support services, to be released in a timely manner and without compromising their wellbeing. As shown in this paper, investing in long-term care services can generate substantial savings in the healthcare system. Increasing the coverage and quality of long-term care services is still a challenge in the region. Currently, there is a limited supply of long-term care systems and services in the region, and where available, they are significantly underfunded and focused on the socioeconomically vulnerable population (Aranco et al., 2022). The two countries that are the focus of this study, Brazil and Mexico, are discussing the creation of care systems, which include the provision of long-term care (da Mota Peroni et al., 2023; López-Ortega and Aranco, 2019). The path towards a long-term care system will vary by country, but there are common steps that countries need to take (Cafagna et al., 2019; Me dellín et al., 2018). First, eligibility for services needs to be assessed through a scale that evaluates care needs (Oliveira et al., 2022). Second, the assessment needs to be translated into the definition of a care plan for every person. Third, countries need to decide how to finance the system. This may be achieved through general taxation, social insurance, co-payments, or a combination thereof. Each financing mechanism has strengths and weaknesses which need to be assessed by the countries in order for them to select a mechanism that guarantees the system’s financial, social, and political viability (Fabiani et al., 2022). Fourth, it is important to ensure quality of services. This requires the establishment and monitoring of quality standards, and training and professionalization of human resources who are essential for quality service provision (Arroyo et al., 2023; Fabiani, 2023; Villalobos Dintrans et al., 2022). Fifth, it is critical to create strong coordination mechanisms between hospitals and long-term care, social, and rehabilitation services. This is not easy, particularly in countries like Brazil, Mexico, and many others in the Latin American and Caribbean region. In these countries, healthcare and social services are delivered by different institutions, have separate funding, different regulations, and different eligibility rules. Integration requires a fundamental paradigm change, one that places persons in the center of the care delivery system, and that encompasses the adaptation of both processes and infrastructures (Lloyd-Sherlock et al. 2024; Albertson et al., 2022). The role of a care manager or coordinator, a professional who works closely with patients (and their families) to guarantee the continuity of care across all levels, has emerged as a good practice. In Latin America, there are some examples of systems that have attempted to coordinate social and health care. In Brazil, the programs Maior Cuidado in the city of Belo Horizonte, and the Programa Acompanhante de Idosos (PAI) in the municipality of São Paulo, are two promising examples of improved sociosanitary coordination that have the potential to facilitate older people’s transition from the hospital to their post-discharge destination (Lloyd-Sherlock et al., 2023; Lloyd- 30 Sherlock et al., 2024). Always in Brazil, the program Melhor em Casa of the Ministry of Health is a large-scale, national effort to reduce the number and the length of hospitalizations by providing healthcare at home (da Mota Peroni et al. 2023, Ministério da Saúde do Brasil 2024). An evaluation analysis of Belo Horizonte’s program shows that the length of hospital stays for patients that belong to the Maior Cuidado program is 0.22 days shorter compared to patients that do not belong to the program, generating savings of approximately US$100 per admission (LloydSherlock et al., 2024). The authors identify two features crucial in explaining the program's success: (i) the joint development of the program by the Department of Health and the Department of Social Assistance, with both institutions working in close collaboration; (ii) the creation of a new worker category – the family care support workers – who are fully integrated into the local health and social assistance teams (Lloyd-Sherlock et al., 2024). The benefits of implementing a long-term care system are not limited to the potential savings that can be achieved in the healthcare sector but extend to other areas (Aranco et al., 2022), Villalobos-Ditrans, 2018). Such systems can free up time from family caregivers, mostly women, promoting their labor market participation or allowing them to pursue other activities. These systems can also contribute to the professionalization of caregiving work, generating more and better jobs in a sector that is currently characterized by high feminization, low levels of education, and suboptimal working conditions (Fabiani 2023, Villalobos-Ditrans et al. 2022). 31 References Albertson, E. M., Chuang, E., O'Masta, B., Miake-Lye, I., Haley, L. A., & Pourat, N. (2022). Systematic review of care coordination interventions linking health and social services for high-utilizing patient populations. Population health management, 25(1), 73-85. Aranco, N., Ibarrarán, P., and Stampini, M. (2022). Prevalence of care dependence among older persons in 26 Latin American and the Caribbean countries. Technical Note IDB-TN-2470. Inter-American Development Bank. http://dx.doi.org/10.18235/0004250 Aranco, N., Bosch, M., Stampini, M., Azuara, O., Goyeneche, L., Ibarrarán, P., Oliveira, D., Retana Torre, M.R., Savedoff, B. and Torres, E. 2022. Aging in Latin America and the Caribbean: Social Protection and Quality of Life of Older Persons. IDB Monograph 1009. Washington, DC: Inter-American Development Bank. http://dx.doi.org/10.18235/0004287 Arroyo, E. A., Prieto, E. B., Cordero, L. F., Marín, M. L., Franco, L. C. L., Gómez, A. L., Benedetti, F., and Díaz-Veiga, P. (2023). Toward the Professionalization of Caregivers: Training and Skills Needed for Long-Term. Inter-American Development Bank, Technical Note IDB-TN 02717 https://doi.org/10.18235/0005055 Bertakis, K. D., Azari, R., Helms, L. J., Callahan, E. J., & Robbins, J. A. (2000). Gender differences in the utilization of health care services. Journal of family practice, 49(2). Bo, M., Fonte, G., Pivaro, F., Bonetto, M., Comi, C., Giorgis, V., Marchese, L., Isaia, G., Maggiani, G., Furno, E., Falcone, Y., and Isaia, G. C. (2016). Prevalence of and factors associated with prolonged length of stay in older hospitalized medical patients. Geriatrics & Gerontology International, 16(3), 314–321. https://doi.org/10.1111/ggi.12471 Cafagna, G., Aranco, N., Ibarrarán, P., Oliveri, M. L., Medellín, N., and Stampini, M. (2019). Age with Care: Long-term Care in Latin America and the Caribbean. Inter-American Development Bank, IDB-MG-74. https://doi.org/10.18235/0001972 Cameron, K. A., Song, J., Manheim, L. M., & Dunlop, D. D. (2010). Gender disparities in health and healthcare use among older adults. Journal of women's health, 19(9), 1643-1650. Carey, K. (2015). Measuring the Hospital Length of Stay/Readmission Cost Trade-Off Under a Bundled Payment Mechanism. Health Economics, 24(7), 790–802. https://doi.org/10.1002/hec.3061 Carey, M. R., Sheth, H., and Braithwaite, R. S. (2005). A prospective study of reasons for prolonged hospitalizations on a general medicine teaching service. Journal of General Internal Medicine, 20(2), 108–115. https://doi.org/10.1111/j.1525-1497.2005.40269.x Cho, Y.-M., and Kwon, S. (2022). Effects of public long-term care insurance on the medical service use by older people in South Korea. Health Economics, Policy and Law, 1–18. https://doi.org/10.1017/S174413312200024X Costa-Font, J., Jimenez-Martin, S., and Vilaplana, C. (2018). Does long-term care subsidization reduce hospital admissions and utilization? Journal of Health Economics, 58, 43–66. https://doi.org/10.1016/j.jhealeco.2018.01.002 Curry, N. (2006). Preventive Social Care: Is it cost effective? Kings Fund. https://www.kingsfund.org.uk/sites/default/files/preventive-social-care-wanless-backgroundpaper-natasha-curry2006.pdf 32 Da Mota Peroni, F., Gruchovski Veríssimo, L., Goes Shibata, L., Aranco, N. (2023). Envejecimiento y atención a la dependencia en Brasil. Inter-American Development Bank. Nota Técnica IDB-TN-02677. http://dx.doi.org/10.18235/0004792 Dintrans, P. V., Oliveira, D., and Stampini, M. (2022). Estimación de las necesidades de recursos humanos para la atención a las personas mayores con dependencia de cuidados en América Latina y el Caribe. Inter-American Development Bank. Nota Técnica IDB-TN-02556. https://doi.org/10.18235/0004487 Fabiani, B. (2023). Caring for Caregivers: The Landscape of Paid Care Work in Latin America and the Caribbean. Inter-American Development Bank, Technical Note IDB-TN 02783 https://doi.org/10.18235/0005147 Fabiani, B., Costa-Font, J., Aranco, N., Stampini, M., and Ibarrarán, P. (2022). Funding Options for Long-Term Care Services in Latin America and the Caribbean. Inter-American Development Bank, Technical Note IDB-TN 02473 https://doi.org/10.18235/0004306 Falcone, D., Bolda, E., and Leak, S. C. (n.d.). Waiting for Placement: An Exploratory Analysis of Determinants of Delayed Discharges of Elderly Hospital Patients. Freitas, A., Silva-Costa, T., Lopes, F., Garcia-Lema, I., Teixeira-Pinto, A., Brazdil, P., and CostaPereira, A. (2012). Factors influencing hospital high length of stay outliers. BMC Health Services Research, 12(1), 265. https://doi.org/10.1186/1472-6963-12-265 Gaughan, J., Gravelle, H., and Siciliani, L. (2015). Testing the Bed-Blocking Hypothesis: Does Nursing and Care Home Supply Reduce Delayed Hospital Discharges? Health Economics, 24(S1), 32–44. https://doi.org/10.1002/hec.3150 Hassan, M., Tuckman, H. P., Patrick, R. H., Kountz, D. S., and Kohn, J. L. (2010). Hospital length of stay and probability of acquiring infection. International Journal of Pharmaceutical and Healthcare Marketing, 4(4), 324–338. https://doi.org/10.1108/17506121011095182 Hauck, K., and Zhao, X. (2011). How dangerous is a day in hospital? A model of adverse events and length of stay for medical inpatients. Medical Care, 49(12), 1068–1075. https://doi.org/10.1097/MLR.0b013e31822efb09 Institute for Health Metrics and Evaluation (IHME), 2020. Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD 2019). Seattle, United States. Hendy, P., Patel, J., Kordbacheh, T., Laskar, N., and Harbord, M. (2012). In-depth analysis of delays to patient discharge: A metropolitan teaching hospital experience. Clinical Medicine, 12(4), 320–323. https://doi.org/10.7861/clinmedicine.12-4-320 Holmås, T. H., Islam, M. K., and Kjerstad, E. (2013). Interdependency between social care and hospital care: The case of hospital length of stay. European Journal of Public Health, 23(6), 927–933. https://doi.org/10.1093/eurpub/cks171 Holmås, T. H., Kamrul Islam, M., and Kjerstad, E. (2013). Between two beds: Inappropriately delayed discharges from hospitals. International Journal of Health Care Finance and Economics, 13(3–4), 201–217. https://doi.org/10.1007/s10754-013-9135-4 Hyun, K.-R., Kang, S., and Lee, S. (2014). Does long-term care insurance affect the length of stay in hospitals for the elderly in Korea?: A difference-in-difference method. BMC Health Services Research, 14(1), 630. https://doi.org/10.1186/s12913-014-0630-1 Kozma, C. M., Dickson, M., Raut, M. K., Mody, S., Fisher, A. C., Schein, J. R., and Mackowiak, J. I. (2010). Economic benefit of a 1-day reduction in hospital stay for community-acquired 33 pneumonia (CAP). Journal of Medical Economics, 13(4), 719–727. https://doi.org/10.3111/13696998.2010.536350 Krell, R. W., Girotti, M. E., and Dimick, J. B. (2014). Extended Length of Stay After Surgery: Complications, Inefficient Practice, or Sick Patients? JAMA Surgery, 149(8), 815–820. https://doi.org/10.1001/jamasurg.2014.629 Landeiro, F., Leal, J., and Gray, A. M. (2016). The impact of social isolation on delayed hospital discharges of older hip fracture patients and associated costs. Osteoporosis International: A Journal Established as Result of Cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA, 27(2), 737–745. https://doi.org/10.1007/s00198-015-3293-9 Landeiro, F., Roberts, K., Gray, A. M., and Leal, J. (2019). Delayed Hospital Discharges of Older Patients: A Systematic Review on Prevalence and Costs. The Gerontologist, 59(2), e86–e97. https://doi.org/10.1093/geront/gnx028 Lenzi, J., Mongardi, M., Rucci, P., Ruscio, E. D., Vizioli, M., Randazzo, C., Toschi, E., Carradori, T., and Fantini, M. P. (2014). Sociodemographic, clinical and organisational factors associated with delayed hospital discharges: A cross-sectional study. BMC Health Services Research, 14, 128. https://doi.org/10.1186/1472-6963-14-128 Lewis, R., and Glasby, J. (2006). Delayed discharge from mental health hospitals: Results of an English postal survey. Health and Social Care in the Community, 14(3), 225–230. https://doi.org/10.1111/j.1365-2524.2006.00614.x Lingsma, H. F., Bottle, A., Middleton, S., Kievit, J., Steyerberg, E. W., and Marang-van de Mheen, P. J. (2018). Evaluation of hospital outcomes: The relation between length-of-stay, readmission, and mortality in a large international administrative database. BMC Health Services Research, 18(1), 116. https://doi.org/10.1186/s12913-018-2916-1 Lloyd-Sherlock, P., Fialho de Carvalho, P., Giacomin, K., and Sempé, L. (2023). Addressing pressures on health services in Belo Horizonte, Brazil through community-based care for poor older people: A qualitative study. The Lancet Regional Health - Americas, 27, 100619. https://doi.org/10.1016/j.lana.2023.100619 Lloyd-Sherlock, P., Giacomin, K., Fialho de Carvalho, P., Sempé, L. (2024). Programa Maior Cuidado: An Integrated Community-Based Intervention on Care for Older People. Technical note 2889. Inter-American Development Bank. http://dx.doi.org/10.18235/0005535 López-Ortega, M., Aranco, N. (2019). Envejecimiento y atención a la dependencia en México. Inter-American Development Bank. Nota Técnica IDB-TN-1614 http://dx.doi.org/10.18235/0001826 Makowsky, M. D., and Klein, E. Y. (2018). Identifying the relationship between length of hospital stay and the probability of readmission. Applied Economics Letters, 25(6), 375–380. https://doi.org/10.1080/13504851.2017.1324605 Marfil-Garza, B. A., Belaunzarán-Zamudio, P. F., Gulias-Herrero, A., Zuñiga, A. C., Caro-Vega, Y., Kershenobich-Stalnikowitz, D., and Sifuentes-Osornio, J. (2018). Risk factors associated with prolonged hospital length-of-stay: 18-year retrospective study of hospitalizations in a tertiary healthcare center in Mexico. PLOS ONE, 13(11), e0207203. https://doi.org/10.1371/journal.pone.0207203 Matus-López, M. (2023). “Análisis del impacto y ahorro en salud de un sistema de cuidados a la dependencia en Chile”. Unpublished document. 34 Medellín, N., Ibarrarán, P., and Stampini, M. (2018). Cuatro elementos para diseñar un sistema de cuidados. Technical note 1438. Inter-American Development Bank. https://doi.org/10.18235/0001129 Micallef, A., Buttigieg, S. C., Tomaselli, G., and Garg, L. (2020). Defining Delayed Discharges of Inpatients and Their Impact in Acute Hospital Care: A Scoping Review. International Journal of Health Policy and Management, 11(2), 103–111. https://doi.org/10.34172/ijhpm.2020.94 Ministério da Saúde Brasil (n.d). Departamento de Informática do Sistema Único de Saúde, DATASUS. Available at: https://datasus.saude.gov.br/sobre-o-datasus/ Ministério da Saúde Brasil (2024). Melhor em Casa inclui equipes de reabilitação e tem novas diretrizes para gestores. Available at: https://www.gov.br/saude/ptbr/assuntos/noticias/2024/marco/melhor-em-casa-inclui-equipes-de-reabilitacao-e-temnovas-diretrizes-para-gestores Misky, G., Wald, H., and Coleman, E. (2010). Post-Hospitalization Transitions: Examining the Effects of Timing of Primary Care Provider Follow-Up. Journal of Hospital Medicine : An Official Publication of the Society of Hospital Medicine, 5, 392–397. https://doi.org/10.1002/jhm.666 Moore, L., Cisse, B., Batomen Kuimi, B. L., Stelfox, H. T., Turgeon, A. F., Lauzier, F., Clément, J., and Bourgeois, G. (2015). Impact of socio-economic status on hospital length of stay following injury: A multicenter cohort study. BMC Health Services Research, 15(1), 285. https://doi.org/10.1186/s12913-015-0949-2 National Health Service (NHS) England (n.d). “Reducing length of stay”. Available at: https://www.england.nhs.uk/urgent-emergency-care/reducing-length-of-stay/ Olivares-Tirado, D. P. (n.d.). Hospitalización Social en Adultos Mayores en el Sistema de Salud de Chile. Oliveira, D., León-Moncada, S., and Terra, F. (2022). El uso de Baremos de Valoración de la Funcionalidad y de la Dependencia de cuidados en personas mayores: Prácticas, avances y direcciones futuras. https://doi.org/10.18235/0004280 Perelman, J., and Closon, M.-C. (2011). Impact of socioeconomic factors on in-patient length of stay and their consequences in per case hospital payment systems. Journal of Health Services Research and Policy, 16(4), 197–202. https://doi.org/10.1258/jhsrp.2011.010047 Picone, G., Mark Wilson, R., and Chou, S.-Y. (2003). Analysis of hospital length of stay and discharge destination using hazard functions with unmeasured heterogeneity. Health Economics, 12(12), 1021–1034. https://doi.org/10.1002/hec.800 Rao, K. D., Vecino Ortiz, A. I., Roberton, T., Lopez Hernandez, A., & Noonan, C. (2022). Future Health Spending and Treatment Patterns in Latin America and the Caribbean: Health Expenditure Projections & Scenario Analysis . Inter-American Development Bank. Rauh, S. S., Wadsworth, E., and Weeks, W. B. (2010). The fixed-cost dilemma: What counts when counting cost-reduction efforts? A hospital’s fixed costs are a reality that can make the idea of achieving savings by reducing length of stay illusory. Healthcare Financial Management, 64(3), 60–64. Rodrigues, L. P., de Oliveira Rezende, A. T., Delpino, F. M., Mendonça, C. R., Noll, M., Nunes, B. P., de Oliviera, C., and Silveira, E. A. (2022). Association between multimorbidity and hospitalization in older adults: Systematic review and meta-analysis. Age and Ageing, 51(7), afac155. https://doi.org/10.1093/ageing/afac155 35 Rojas‐García, A., Turner, S., Pizzo, E., Hudson, E., Thomas, J., and Raine, R. (2018). Impact and experiences of delayed discharge: A mixed‐studies systematic review. Health Expectations : An International Journal of Public Participation in Health Care and Health Policy, 21(1), 41–56. https://doi.org/10.1111/hex.12619 Siddique, S. M., Tipton, K., Leas, B., Greysen, S. R., Mull, N. K., Lane-Fall, M., McShea, K., and Tsou, A. Y. (2021). Interventions to Reduce Hospital Length of Stay in High-risk Populations: A Systematic Review. JAMA Network Open, 4(9), e2125846. https://doi.org/10.1001/jamanetworkopen.2021.25846 Stone, K., Zwiggelaar, R., Jones, P., and Mac Parthaláin, N. (2022). A systematic review of the prediction of hospital length of stay: Towards a unified framework. PLOS Digital Health, 1(4), e0000017. https://doi.org/10.1371/journal.pdig.0000017 Taheri, P. A., Butz, D. A., and Greenfield, L. J. (2000). Length of stay has minimal impact on the cost of hospital admission11No competing interests declared. Journal of the American College of Surgeons, 191(2), 123–130. https://doi.org/10.1016/S1072-7515(00)00352-5 Toh, H. J., Lim, Z. Y., Yap, P., and Tang, T. (2017). Factors associated with prolonged length of stay in older patients. Singapore Medical Journal, 58(3), 134–138. https://doi.org/10.11622/smedj.2016158 Villalobos Dintrans, P. (2018). Long-term care systems as social security: The case of Chile. Health Policy and Planning, 33(9), 1018–1025. https://doi.org/10.1093/heapol/czy083 Walker, R. J., Segon, A., Good, J., Nagavally, S., Gupta, N., Levine, D., Neuner, J., and Egede, L. E. (2021). Differences in length of stay by teaching team status in an academic medical center in the Midwestern United States. Hospital Practice (1995), 49(2), 119–126. https://doi.org/10.1080/21548331.2021.1882238 Walsh, T., Onega, T., and Mackenzie, T. (2014). Variation in length of stay within and between hospitals. Journal of Hospital Administration, 3(4), 53. https://doi.org/10.5430/jha.v3n4p53 Yang, S., Zhou, M., Liao, J., Ding, X., Hu, N., and Kuang, L. (2022). Association between Primary Care Utilization and Emergency Room or Hospital Inpatient Services Utilization among the Middle-Aged and Elderly in a Self-Referral System: Evidence from the China Health and Retirement Longitudinal Study 2011–2018. International Journal of Environmental Research and Public Health, 19(19), 12979. https://doi.org/10.3390/ijerph191912979 Zhao, E. J., Yeluru, A., Manjunath, L., Zhong, L. R., Hsu, H.-T., Lee, C. K., Wong, A. C., Abramian, M., Manella, H., Svec, D., and Shieh, L. (2018). A long wait: Barriers to discharge for long length of stay patients. Postgraduate Medical Journal, 94(1116), 546–550. https://doi.org/10.1136/postgradmedj-2018-135815 36 Annex 1. Findings from interviews and focus group discussions We complement our analysis with qualitative information to get a better understanding of the causes and consequences of long hospitalizations. The data were collected through interviews (3 participants) and focus group discussions (6 participants) with health experts in the region (Mexico, Argentina, and Colombia) between June and October 2023. The data were analyzed through the method of thematic analysis. This inductive approach allowed us to identify major themes and patterns and conceptualize the information into the stages of pre-hospital, in-hospital, and post-hospital care. The conversations provided crucial insights into the different final components along the continuum-of-care referred to in our analysis. The findings from the interviews and focus group discussions emphasize the importance of appropriate primary healthcare and social services for preventing hospitalizations and strengthening a person’s health status. The participants highlighted that the inability to ensure timely access and use of such services negatively affects a person’s health conditions, ultimately increasing the risk of a higher length of stay in the hospital. “El cuidado previo a pisar urgencias/hospital es fundamental. El que el paciente acuda de manera regular a sus visitas de medicina familiar también es fundamental.” – P2, focus group “No podemos hablar de la demanda de urgencias y de la atención hospitalaria si no hablamos de las enfermedades crónicas que no se protocolizan en el manejo ambulatorio.” – P2, focus group “No tenemos plan ampliado de inmunizaciones en personas mayores, la única vacuna gratuita es la de influenza y su cobertura es muy baja; entonces si no tenemos estrategias de prevención como la vacunación contra neumococo, tosferina, herpes zoster... estamos exponiendo a la población mayor, en especial a la más frágil, a hospitalizaciones recurrentes.” – P1, interview “El hecho de que el adulto mayor no tenga soporte para buscar atención médica oportuna, lo hace que llegue al hospital en una fase más grave de la enfermedad.” – P6, focus group “La estancia hospitalaria la determina la estabilidad del paciente, es decir, su funcionalidad. Cómo está funcionando él como individuo en cuestión de condiciones motoras, en cuestión de comorbilidades.” – P4, focus group “Cualquier cosa que cambia dramáticamente la funcionalidad de una persona (…) es de riesgo para una estadía hospitalaria mayor.” – P2, interview “No tenemos rutas de detección temprana de osteoporosis y riesgo de caída y fractura; las cirugías ortopédicas generan hospitalizaciones más prolongadas en personas mayores.” – P1, interview Moreover, inadequate rehabilitation and long-term care, including information and support, and long hospital stays may increase the risk of rehospitalization. The following observations exemplify this point: “[Después de una hospitalización,] un tema importante es la orientación nutricional adaptada a la persona. Otro aspecto importante es la educación en salud, es decir, un paciente con incluso 10 años con hipertensión no entiende su enfermedad, no hemos sabido informarle al respecto.” – P2, focus group 37 “Después de los 70 años en un solo día de hospitalización, si no me muevo, puedo perder hasta el 3% de la masa muscular total lo que va a generar grandes problemas de movilidad, dependencia funcional y sobrecarga a los cuidadores familiares.” – P1, interview The discussions also showed that hospital characteristics and inefficiencies can have important repercussions for a patient’s length of stay in the hospital. According to the participants, the absence of protocols, resources, and knowledge at various levels can lead to delays in processes at the beginning, during, and at the end of a hospitalization. With regards to elderly persons, common issues seem to particularly revolve around coordination between the different levels of health care, scarcity of training in geriatrics, especially among doctors working in emergency services, insufficient resources for timely and appropriate treatment of patients, and absence of clear discharge procedures. “(…) La falta de coordinación es un asunto mayor.” – P1, focus group “Hay un retardo enorme del médico familiar para enviar a los pacientes a 2º o 3º nivel.” – P1, focus group “Si el médico familiar no está capacitado para atender a los pacientes geriátricos con enfermedades como diabetes mellitus e hipertensión, las más comunes, cuando llega al 2º nivel el médico especialista tiene que internarlo y el paciente llega con todas las patologías agravadas y complicaciones.” – P3, focus group “No existen protocolos de atención humanizada y diferencial a personas mayores [en el caso de urgencias] por la cantidad de pacientes que reciben; esto hace que los procesos de admisión hospitalaria sean largos y que (…) en los servicios de urgencias se compliquen o adquieran gérmenes oportunistas que hacen que se complique el cuadro inicial, sin contar todo lo que ocurre en personas mayores con deterioro cognitivo que generan episodios delirantes y terminan siendo inmovilizados tanto física como farmacológicamente.” – P1, interview “Los médicos de Urgencias y Hospitalización en su mayoría no han recibido capacitación en geriatría ni cuentan con médicos geriatras (…), por lo que terminan inter-consultando a varios especialistas (…), lo cual lleva a toma excesiva de laboratorios y demoras en el proceso.” – P1, interview “No tenemos geriatras en los servicios de urgencias en todos los hospitales como sería lo ideal, entonces se va retrasando la atención porque el médico general tal vez tiene miedo de abordar al paciente adulto mayor. Si se capacitara al personal ayudaría a que la atención fuera más eficiente y oportuna.” – P5, focus group “En el tema de fractura de cadera, retrasan mucho desde el diagnóstico y en las áreas hospitalarias, la cirugía. Esto es atribuible a la falta de conocimiento de que este padecimiento es una urgencia y en general los médicos esperan a que el paciente esté lo más estable posible para operarlo.” – P6, focus group “Hay hospitales de 2º nivel que no tienen recursos y tienen que esperar a que el 3º nivel les dé un espacio para el diagnóstico, el paciente puede estar hasta 10 días esperando el diagnóstico, en lugar de recibir el tratamiento.” – P1, focus group “La falta de insumos necesarios. Por ejemplo, los pacientes de fracturas de cadera se quedan mucho tiempo hospitalizados porque no hay la tuerca o el tornillo o el medicamento necesario.” – P1, focus group “Pocos hospitales tienen protocolos de "alta temprana", por lo que administrativamente existen muchas barreras y procesos (…).” – P2, interview