Mortality risk factors in the Catalan long-term care system

ALBERT PRADES-COLOMÉ

Documento de Trabajo 2025/10

Agosto de 2025

fedea

Las opiniones recogidas en este documento son las de sus autores y no coinciden necesariamente con las de Fedea.

Albert Prades-Colomé * 1 2 3

Ministry of Social Rights and Inclusion, Generalitat de Catalunya, Barcelona, Spain

Department of Business, Tecnocampus, Universitat Pompeu Fabra, Mataró, Spain

Centre for Research in Economics and Health (CRES), Universitat Pompeu Fabra, Barcelona, Spain

*Corresponding author. E-mail address: albert.pradesi@upf.edu. Mailing address: Edifici Mercè Rodoreda (campus de la Ciutadella), Ramon Trias Fargas, 25-27, 08005 Barcelona, Spain. ORCID ID: 0009-0004-6139-1495.

Abstract:

As populations age, understanding the health impact of long-term care systems is critical for shaping effective policy. This study investigates the association between long-term care benefits and mortality risk among older adults in Catalonia, Spain, using comprehensive administrative data from July 2015 to December 2024. The analysis focuses on individuals aged 50+ who were assessed for long-term care needs, categorizing them by severity (Grades I–III) and type of benefit received: home care, residential care, a combination of both or no benefit. Applying survival analysis techniques—including Kaplan-Meier estimators and Cox proportional hazards models—it finds that individuals with long-term care needs receiving benefits have significantly lower mortality hazards. Notably, individuals transitioning from home to residential care exhibit the most favourable hazard ratios, suggesting that responsive care pathways are associated with better survival outcomes, potentially due to a most accurate matching of care to needs. Residential care alone is associated to higher mortality risk than home care in the population with the highest grades of long-term care needs. Individuals with recognized long-term care needs who do not receive any benefits face significantly higher risks, a pattern that may reflect the consequences of unmet care needs. Mortality risk varies by sex, age, and clinical profile, with higher hazards observed among men, older individuals, and those with haematological, neoplastic, or respiratory conditions. These findings underscore the association between formal long-term care systems and lower mortality risk and emphasize the importance of timely, adaptive care pathways in mitigating health decline among aging populations.

Keywords: Mortality, Long-Term Care, Proportional Hazard Models, Catalonia, Ageing.

Statements and Declarations:

- Conflict of interest: The author declares no conflicts of interest.

- Ethical approval: This article does not contain any studies with human participants performed by any of the authors.

- 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 https://creativecommons.org/licenses/by/4.0/

JEL Codes: I18, I38, J14

Acknowledgments: The author is thankful to Ministry of Social Rights and Inclusion at Catalan Government for providing the database, the UPF Scientific Computing Core Facility for help in the data processing, the participants in the XLIV Spanish Health Economics Association annual meeting (Jornadas AES), and Roberto Martínez-Lacoba and his colleagues at the Grupo de Investigación en Economía, Alimentación y Sociedad (GEAS) at the University of Castilla-la-Mancha for their comments and suggestions.

Funding: This work was supported by the Doctoral Industrial Plan of the Ministry of Research and Universities at Generalitat de Catalunya (Government of Catalonia).

1. Intro

Population ageing is transforming the demographic landscape of many high-income countries, raising urgent questions about how to design and deliver health and long-term care systems that can meet the needs of a growing older population with complex conditions. In Spain, demographic projections are striking: by 2050, 30% of the population will be over 65 years-old, up from 20% today, while the share of those aged 80 and older is expected to nearly double, from 6.1% in 2021 to 11.2% by mid-century (OECD (2023); Instituto Nacional de Estadística (2022)). These shifts are placing increasing pressure on public health and long-term care systems (Lorenz et al. (2020); Eggink et al. (2017)), as ageing is associated with worsening health, multimorbidity, and greater vulnerability (WHO, 2015), which are key predictors of mortality (Carreras et al. (2018); Palladino et al. (2016); Howdon & Rice (2018)).

The health-related challenges of ageing are clearly reflected in self-reported health measures, another predictor of mortality (Reinwarth et al. (2023); Palladino et al. (2016)). In Spain, only 44% of people 65 years or older report their health as good or very good, and a substantial share face limitations in Activities of Daily Living (ADLs) . Around 37% of older Spaniards report some degree of functional limitation and 12% report severe limitations, which can significantly restrict their autonomy (OECD, 2023). Such impairments often necessitate formal long-term care (LTC), defined as a spectrum of services that support individuals with sustained physical or mental impairments in managing their daily lives (European Commission, 2024). Historically, LTC was largely delivered by family members—predominantly women—within the household setting (WHO Centre for Health Development et al., 2021). However, sociocultural and economic changes, including declining family size, geographical dispersion, and rising female labour force participation (Esping-Andersen (2016); Charmes (2019)), have eroded the capacity of families to provide care, placing pressure on public institutions to step in.

Recognizing the growing importance of LTC, Spain introduced a formal Long-Term Care System (LTCS) in 2007 with the passage of the Dependency Act (DA) (Law 39/2006). While the legislation was approved at the national level, the implementation and administration of the system was delegated to regions, leading to significant heterogeneity in its functioning (Oliva et al., 2023). Catalonia, Spain's second most populous region, represents approximately of the Spanish population and of LTCS beneficiaries (IMSERSO (2024); Departament de Drets Socials i Inclusió (2024)). Despite the region's demographic weight and the institutional complexity of its LTCS, the health impacts of LTC benefits in Catalonia remain understudied. Most research conducted to date has been limited to data from the period before the full implementation of the DA in July 2015 (Peña-Longobardo et al., 2016), leaving a gap in understanding the effects of LTC (Serrano-Alarcón et al. (2022), Hernández-Pizarro (2016)).

Therefore, this study aims to examine the association between receipt of LTC benefits on mortality risk among older adults with recognized LTC needs in Catalonia between 2015 and 2024. The overarching aim of the LTCS is to improve the wellbeing and quality of life of its beneficiaries. However, Spain lacks systematic indicators of wellbeing for LTCS beneficiaries -such as those provided by ASCOT (Malley et al., 2012)- which limits the possibilities for direct evaluation. In response, several studies have used health-related outcomes as proxies for wellbeing. Costa-Font et al. (2018), Serrano-Alarcón et al. (2022), and Hernández-Pizarro et al. (2024) have examined healthcare use, including hospitalizations, hospital admissions, length of hospital stay, unscheduled visits to hospitals or primary care, and admissions for nursing home-avoidable conditions. Another relevant indicator is mortality, given its strong association with both physical and mental wellbeing (Chida & Steptoe (2008); Steptoe et al. (2015); Tang et al. (2008)).

Hernández-Pizarro (2016) previously examined the relationship between LTC benefits and mortality in Catalonia for the period 2008–2014, finding that access to LTC benefits may contribute to increased life expectancy among beneficiaries. However, no comparable analysis using individual-level data has been conducted since the full implementation of the DA in 2015. Moreover, recent trends in deinstitutionalization in Spain —enabling individuals to remain in their own homes for as long as possible, rather than being placed in residential care –or institutional- settings (Bermejo-Patón et al., 2023)- and the disproportionate impact of the COVID-19 pandemic on those with recognized LTC needs (see Figure 5 in Appendix 1) highlight key challenges. These effects were particularly severe among people with LTC needs who were in residential care homes (Zunzunegui et al. (2022); Costa-Font et al. (2021)), underscoring the need to revisit and update this evidence base with more granular and contemporary data.

The activities of daily living (ADLs) is a term used to collectively describe fundamental skills required to independently care for oneself, such as eating, bathing, and mobility (Edemekong et al., 2023).

To address this gap, this paper draws on individual-level administrative data covering more than 320,000 individuals in Catalonia aged 50 and over who have been recognized as having LTC needs between July 2015 and December 2024. Individuals are categorized according to their officially assessed level of LTC needs, with controls for pre-existing medical conditions and demographic characteristics (age, sex, nationality and civil status) included in the analysis to reduce selection bias. We use survival analysis techniques to estimate the association between types of LTC benefits and mortality risk —home-based care, institutional care, a combination of both, or no benefit—on mortality risk. This classification allows us to explore how different care settings and transitions between them affect longevity.

The results reveal substantial variation in survival probabilities depending on both the Grade of LTC needs and the type of benefit received. Individuals with the highest level of needs (Grade III) experience significantly higher mortality. Those who receive home care at first and then transition to residential care exhibit the highest survival probabilities, suggesting that transitions between care modalities are associated with higher survival probabilities, which may reflect underlying differences in care needs and access. While this survival advantage is most pronounced early on, the univariate analysis indicates that it narrows over time, possibly reflecting limits to the ability of institutional care to compensate for an underlying health decline. On the other hand, individuals in residential care show higher mortality risk relative to home care when compared alone. By contrast, individuals who qualify for—but do not receive—LTC benefits face the lowest survival probabilities, especially in the early years, a finding consistent with the risks posed by unmet care needs and systemic delays.

Building on these findings, this study contributes to the growing evidence base on the health impacts of LTC interventions. It offers empirical insights into the comparative mortality hazard ratios of different care modalities, at a time when Spain and other OECD countries are reevaluating the balance between institutional and home-based care. By leveraging a comprehensive administrative dataset covering nearly a decade after the full implementation of Spain’s DA, this paper revisits these debates in a contemporary setting, with implications for ongoing deinstitutionalization policies and the future design of equitable, need-based LTC systems.

More broadly, this paper contributes to international research on survival models analysing the relationship between LTCS, care settings, and health outcomes, complementing studies from Castilla-la-Mancha (Pardo-Garcia et al., 2024), the Netherlands (Wouterse et al., 2023), Belgium (Van den Bosch et al., 2013), Germany (Koller et al., 2014), and Japan (Lin et al., 2017). As countries across the OECD deal with rapidly ageing populations, understanding how LTCS affect mortality risk is essential not only for improving health outcomes but also for designing equitable and financially sustainable care infrastructures.

The remainder of this paper is organized as follows. The section titled 'Institutional Background' offers an overview of the Spanish LTCS. The 'Methodology' section describes the dataset and approach used in the analysis. The 'Results' section presents the findings, while the 'Conclusion and policy discussion' section discusses the results, highlights the key conclusions, policy implications, and limitations of the study.

2. Institutional Background

Spain's LTCS was established through the enactment of the DA (Law 39/2006), approved in December 2006 and implemented in January 2007. This legislation introduced a nationwide entitlement to LTC benefits, marking a significant shift toward a formalized, needs-based care model in Spain. Importantly, while the framework was developed at the national level, the system was designed to be decentralized: the country's 17 regions, called Autonomous Communities (CCAA), are responsible for administering and managing the services, operating under shared national guidelines. This has resulted in substantial regional variation in implementation, funding, service delivery, and waiting times, all of which shape the experience of care recipients across the country (Oliva et al., 2023).

The Spanish LTCS is universal and designed to provide benefits based on individuals' assessed levels of LTC needs. In this sense, the system aligns with the principles of equity and needs-based allocation. Eligibility is determined through a standardized functional assessment, using the Baremo de Valoración de la Dependencia (BVD), a validated scale that ranges from 0 to 100 points (Decree 504/2007; Royal Decree 174/2011). Individuals with scores above 25 are entitled to receive benefits, categorized into three levels of need: Grade I (moderate needs, 25–49), Grade II (severe needs, 50–74), and Grade III (major needs, 75–100). Examiners apply the official scale through direct observation and must link the assessed loss of autonomy to a medical diagnosis within the National Health System (Serrano-Alarcón et al., 2022). These categories determine the type and intensity of care to which individuals are entitled and form the foundation of benefit assignment within the system. Figure 1 illustrates the assessment process and corresponding benefit pathways across grades.

In Catalonia, these assessments are carried out by independent multidisciplinary teams—called SEVAD (Servei de Valoració de la Dependència or Dependency Assessment Services)—comprising healthcare and social workers, who act as evaluators-. Each region has a designated SEVAD, and individuals are assigned to a specific team based on their area of residence. Evaluators also collect sociodemographic and health-related information (e.g., age, sex, nationality, marital status, previous medical diagnoses and personal income levels), but this information does not influence the final BVD score (Vidiella-Martin et al., 2024). Rather, the system aims to offer an objective, comparable measure of LTC needs levels across individuals and regions. Nonetheless, recent research has identified patterns suggesting upward score rounding, particularly when a slight increase results in eligibility for more generous benefits (Hernández-Pizarro et al., 2020). Additionally, because evaluators are randomly assigned, differences in their leniencies have also been shown to affect the grading outcome (Serrano-Alarcón et al., 2022). The official scale has shown a moderate correlation with the actual time spent on caregiving (Rodríguez-González & Rodríguez-Míguez, 2021). In Catalonia, the assessment process takes an average of six months (Departament de Drets Socials i Inclusió, 2024). Lastly, individuals can ask for a revaluation if their level of LTC needs increases.

Following the needs assessment, eligible individuals and their families—or their legal representatives—select their preferred LTC benefits from a menu of available options in consultation with the administration. This selection is formalized through an Individual Attention Plan (PIA), which defines both the type and intensity of the care package. While the menu of benefits is the same for all Grades (except for residential care, which is restricted to Grades II and III), the intensity of the benefits varies depending on the assessed need. Higher grades typically correspond to greater hours of formal care or higher-value monetary transfers. At the same time, the income level of the beneficiary determines the amount of co-payments.

Available benefits include both in-kind services—such as telecare, home care, day-care centres, and residential facilities—and cash allowances, notably a provision for informal care. The latter involves unconditional transfers designed to compensate family members or other caregivers for their time and effort. These payments have been instrumental in recognizing and supporting informal care, although they have also sparked debates about their adequacy and potential to reinforce traditional gender roles in caregiving (WHO Centre for Health Development et al., 2021). In terms of delivery, in-kind services may be provided directly by public entities or outsourced to certified private providers that meet specific quality standards. Indeed, due to the limited availability of publicly managed facilities, especially in residential and day-care centres, the system relies heavily on private providers, who may offer both publicly subsidized and market-rate services under distinct financing models (Vidiella-Martin et al., 2024).

Users may combine different services depending on their preferences and the available supply within their region. For example, a person with moderate LTC needs may choose to combine telecare with day-care centre attendance. However, supply-side constraints—such as waiting lists, limited geographic coverage, and administrative bottlenecks—can restrict the availability of preferred options. In December 2023, the average waiting time for the PIA agreement was 112 days (Departament de Drets Socials i Inclusió, 2024), but further delays may occur before the benefit is actually delivered, influenced by factors such as service capacity, budget constraints—as there is a co-payment in function of the income level of beneficiaries-, and the timing of the individual's application. Notably, individuals often defer applying for LTC benefits until their situation becomes critical or other informal care arrangements fail, further complicating service planning and responsiveness (Vidiella-Martin et al., 2024).

The rollout of the system was staggered by design, initially prioritizing individuals with the most severe needs. Grade III benefits were incorporated in 2007, followed by Grade II benefits between 2008 and 2011. The extension to Grade I individuals was initially planned for gradual implementation between 2013 and 2015. However, the financial strain imposed by the Great Recession led to significant delays. In particular, Royal Decree-Law 8/2010 introduced austerity measures that curtailed retroactive payment rights, while Royal Decree-Law 20/2012 further reduced the generosity of benefits (Peña-Longobardo et al., 2016). As a result, the inclusion of Grade I recipients was postponed until July 2015, which serves as the baseline for the observation period used in this study.

Despite the system's universal ambition, Spain remains an outlier among high-income countries in terms of public expenditure on LTC. While approximately of individuals aged 65 and over receive formal LTC benefits—a proportion higher than in France or the Netherlands —public LTC spending remains relatively low, at just of GDP. This figure falls well below the OECD average of and is significantly lower than in countries such as the Netherlands or France ((INE, 2024); (OECD, 2023)). This discrepancy has long raised concerns regarding the adequacy, equity, and sustainability of care provision in Spain (Ramírez-Navarro et al., 2024).

The COVID-19 pandemic further exposed structural vulnerabilities in the Spanish LTCS. Residential care homes, in particular, were unprepared to handle the public health emergency, resulting in disproportionately high mortality rates among institutionalized individuals. Chronic underfunding, staff shortages, and limited crisis response capabilities have been identified as key factors behind this outcome (Zunzunegui et al. (2022); Costa-Font et al. (2021)). In response, the Spanish government incorporated substantial reforms into its Recovery and Resilience Plan Gobierno de España (2021), which reorients policy priorities toward deinstitutionalization (Bermejo-Patón et al., 2023). In line with this policy shift, the present study classifies LTC benefit recipients into three categories—home care, institutional care, and individuals who transition from home to institutional care—allowing for an in-depth analysis of how mortality risk varies by care modality.

3. Methodology

3.1. Population

This study draws on administrative data covering all individuals who applied for recognition of LTC needs in Catalonia between July 1, 2015, and December 31, 2024. The dataset, comprising 523,515 individuals, was provided by the Ministry of Social Rights and Inclusion of the Catalan Government—the competent authority responsible for managing the LTCS in the region. All data were anonymized before analysis and remain confidential, in accordance with data protection agreements.

The database includes precise records on the timing of each application for LTC needs assessment and the date of death (if applicable) for each individual. It also contains detailed information on the type and timing of the LTC benefits awarded, along with the severity of need as measured by the BVD score. For the purposes of this study, LTC benefits were classified into four mutually exclusive categories based on the care pathway: (1) Home Care, in which individuals remain in their homes while receiving support; (2) Residential Care , where individuals move into institutional care settings; (3) Home and Residential Care, capturing transitions for people who first receive a home-care benefit and eventually switches to residential care; and (4) Without Benefit, referred to individuals who were eligible for care but never received any services.. A detailed classification of benefit types within each category is presented in Table 3 in Appendix 2.

The dataset contains a set of covariates detailing individual characteristics. These include socioeconomic factors such as age at the time of application, sex (male or female), nationality (Spanish or foreign), civil status (married –including having a partner-, single, widowed, or unknown), and income level. Additionally, the dataset captures clinical diagnoses recorded at the time of the LTC application. These were grouped according to the International Classification of Diseases, 10th Revision (ICD-10) (World Health Agency, 2019), and aggregated into 15 Diagnostic Groups (DGs). More details of the grouping procedure are provided in Table 4 in Appendix 3. Additionally, fixed effects were included for the individual’s Health Area of residence—an administrative division used by the regional health system to organize service delivery- and reference SEVAD - the centre responsible for assessing LTC-. The geographical and population distribution of these Health Areas and SEVADs is shown in Appendix 4.

The target population was defined by applying two main inclusion criteria: individuals had to be at least 50 years old at the time of application and have met the eligibility threshold for LTC benefits under the Spanish LTCS, as indicated by a BVD score equal to or greater than 25.

The data cleaning process is summarized in Figure 2. Of the initial 523,515 individuals, 158,551 (30.3%) were excluded for having a BVD score below 25—either due to ineligibility or dropout prior to assessment, sometimes as a result of death during the evaluation period. An additional 29,347 individuals (5.6%) were removed for being under 50 years of age. This yielded a target population of 335,617 individuals.

Further exclusions were made to refine the analytical sample. A total of 8,915 individuals (2.7%) were removed due to receiving atypical or special-category benefits not considered in the study—most of which are reserved for complex clinical cases and represent a minor share of the data-. Another 4,846 individuals were discarded due to missing values in key variables: 2,678 had missing information on DGs, 1,867 were missing Health Area data, and 301 lacked a recorded date of application. Finally, 42 individuals were excluded due to discrepancies in the sequencing of events, such as death being recorded previously to application.

In Catalonia, residential care homes primarily provide non-medical care, staffed by low-skill workers offering personal and social services. Although some may offer limited nursing care—such as managing medical treatments and providing skin care for a few hours each day—they do not have continuous medical staff. This sets them apart from medical nursing homes, which deliver round-the-clock healthcare and are funded by the National Health System (National Institute on Aging (2025), Costa-Font et al. (2023)).

After applying these exclusion criteria, the final number of subjects included in the database –Full Sample-consists of 321,814 individuals, which corresponds to 61.5% of the original dataset and 95.9% of the target population.

For further analysis, we stratify the final sample into three mutually exclusive subsamples based on the initial grade of LTC needs at entry into the system. These subsamples are individuals classified as Grade I (n = 193,750), Grade II (n = 88,558), and Grade III (n = 39,506) of LTC needs. This stratification enables a nuanced examination of whether the relationship between benefit type and mortality risk varies by the severity of initial LTC needs. In particular, it allows us to explore potential heterogeneity in outcomes across the spectrum of LTC needs, which is essential for assessing the allocative effectiveness of the system under a needs-based framework.

3.2. Empirical Strategy.

The empirical approach in this study uses survival analysis, a statistical framework used to estimate the time until the occurrence of a specific event—in this case, death. The variable of interest, , represents the number of days from the date of application for LTC needs assessment under the Spanish LTCS to either the recorded date of death or the end of the observation period. The follow-up spans from July 1, 2015, coinciding with the full implementation of benefits for individuals with Grade I, to December 31, 2024.

The central object of analysis is the survival function, , which captures the probability that an individual i remains alive beyond time t. Kaplan–Meier (KM) estimators are initially used to provide a nonparametric depiction of survival probabilities over time (Rich et al. (2010); Goel et al. (2010)). The KM estimate at time t, denoted , is calculated in (1):

\[\hat {S} _ {K M} (t) = \prod_ {s \leq t} \frac {\bar {Y} (s) - d (s)}{\bar {Y} (s)}\tag{1}\]

where denotes the number of individuals at risk immediately prior to time s, and is the number of deaths occurring at that time. These curves serve as a descriptive, visual tool for comparing survival patterns across different groups, such as benefit types or grades of LTC needs.

However, the KM estimator is inherently univariate (Dudley et al., 2016). To address this limitation, a multivariate semi-parametric Cox proportional hazards model is applied (Cox (1972); T. M. Therneau (2000)). The Cox model estimates the hazard rate , or the risk of death for individual i at time t, as (2):

\[\lambda (t _ {i}) = \lambda_ {0} (t _ {i}) e x p (\beta_ {1} X _ {1} + \beta_ {2} X _ {2} + \dots \beta_ {k} X _ {k})\tag{2}\]

where is the unspecified baseline hazard function, and represent covariates included in the model. These covariates capture: (1) LTCS variables, which are our variables of interest -the initial Grade of LTC needs and the type of benefit received-; (2) individual-level characteristics used as control variables (sex, age, nationality, civil status and pre-existing medical diagnoses); and (3) fixed effects for Health Area and SEVAD.

The empirical analysis is conducted in two stages. First, univariate models are estimated to explore raw associations between survival outcomes and each explanatory variable, with corresponding KM survival curves presented. Second, multivariate Cox models are used to assess adjusted hazard ratios. Analyses are conducted on the full sample of 321,814 individuals, and for the three subsamples, stratified by initial Grade of LTC needs: Grade I (n = 193,750), Grade II (n = 88,558), and Grade III (n = 39,506).

Although individual-level income data are available, it is not included as a covariate in the analysis. This is because income is only recorded at the time of benefit allocation; therefore, individuals who die before receiving a benefit or who never receive one are excluded due to missing income data, introducing a risk of selection bias. Moreover, the population under study tends to have relatively homogeneous and low-income levels, typically consisting of non-varying pension payments. Further details on the distribution and characteristics of income data can be found in Appendix 5, and a model including the variable has been included in the robustness checks, as it is described on section 4.4.

To validate the assumptions underlying the Cox model, particularly the proportionality of hazards over time, the Schoenfeld residuals test is applied. When this assumption is violated for the variables of interest, time-varying coefficient models are used to account for potential linear changes in covariate effects over time (Zhang et al. (2018); Tian et al. (2003); Fox & Weisberg (2023)). All analyses have been performed using R's survival package (Therneau, 2024), with visualizations generated using the survminer package (Kassamabra et al., 2024).

Finally, several robustness checks are conducted to assess the sensitivity of our results, including re-estimating models using the continuous BVD score instead of the categorical Grade of LTC needs, adding personal income as a covariate, and modelling age as a categorical variable rather than a continuous one.

4. Results

The results presented in this section are particularly significant as they represent the first analysis of the individual-level administrative database in Catalonia following the full implementation and consolidation of the Spanish LTCS. Previous research using individual-level data has focused on data up until 2014 (Serrano-Alarcón et al. (2022); Vidiella-Martin et al. (2024)), making this study a valuable addition to the literature. The section is structured as follows: first, a descriptive analysis is provided for both the full sample and the three subsamples based on initial Grades of LTC need. Next, Kaplan-Meier survival curves for the univariate models are presented. This is followed by a discussion of the findings from the multivariate Cox regression model. Lastly, the robustness of the results is examined.

4.1. Descriptive statistics

Table 1 presents the descriptive statistics for the Full Sample and the three Subsamples categorized by the Grade of LTC needs, highlighting significant differences across the groups. Of the total sample, 42% have deceased, with this percentage increasing to 68% in the Grade III subsample. In terms of Type of Benefit received, Home Care is the most common benefit, accounting for 52% of the main cohort. However, as the grade of LTC needs increases, the proportion receiving Home Care decreases, while Residential Care increases accordingly. The category labelled "Home & Residential Care" represents individuals who initially received Home Care and subsequently transitioned to Residential Care. This benefit type accounts for 8% to 14% of the sample across the different grades, with the higher share in the Grade II subsample. Notably, the proportion of individuals without any benefits is 27% in the main cohort and in the Grade I subsample, but decreases to 24% in the Grade II subsample, and rises to 33% in the Grade III subsample.

Regarding individual characteristics, the sample is predominantly female, with 62% of the total population being women. This proportion slightly decreases in the Grade II and Grade III subsamples, where women constitute 59% of each group. Age increases with Grade of LTC needs; the average age of individuals in the Grade I subsample is 81.49 years, while the Grade III subsample has an average age of 83.85 years. A vast majority of the sample (99%) are of Spanish nationality, although the share of individuals with foreign nationality is slightly higher in the Grade III subsample (2%), suggesting that non-Spanish individuals may apply for LTC benefits in more severe cases. The proportion of widowed individuals with Grade of LTC needs, particularly in the Grade III subsample, while the share of single individuals decreases across Grades, representing around 10% of the sample. Notably, the civil status information for 9% of the sample is missing.

Each individual may have up to five diagnoses in the DG variable. Osteoarticular diagnoses are the most common, affecting 66% of the sample, though their prevalence decreases as the Grade of LTC needs increases. Neurological diagnoses, which affect 61% of the cohort, show the opposite trend, increasing with assessed LTC grade. Circulatory diagnoses, the third most prevalent (50%), also decrease with grade, though to a lesser extent. Other diagnostic groups have lower prevalence, with notable representations in Mental, Endocrine-metabolic, and Genitourinary categories. The remaining diagnostic groups appear in fewer than 21% of the cases.

When examining the distribution of fixed-effect control variables (see Appendix 4), approximately 25% of individuals in all subsamples are residents of Barcelona city. This proportion increases to over 60% when including the broader Barcelona metropolitan area. Girona is the only other Health Area accounting for more than 10% of the population, whereas all remaining regions each comprise less than 10%. A similar pattern is observed for SEVADs, with the majority of the population concentrated within the metropolitan area of Barcelona.

Table 1 Descriptive Statistics for the Full Sample and the different Subsamples

This table presents descriptive statistics collected at the time of LTC needs assessment. The first column shows data for the full sample, while the following columns correspond to the subsamples based on the initial Grade of LTC needs. For numerical variables (age), the mean is reported with standard deviations in parentheses. Depending on the nature of each variable, either an , Anova or t-Student test has been conducted to assess differences across groups.

Individuals initially classified as Grade I are not eligible for Residential Care. However, some may not receive benefits immediately. If their condition worsens and they are reassessed, they may be assigned a higher grade and become eligible for Residential Care. Therefore, the presence of individuals with Residential Care in the Grade I subsample is possible and reflects these cases.

4.2. Kaplan-Meier estimates.

Figure 3a shows the Kaplan–Meier estimates of the survival function for the Full Sample and figure 3b for each of the subsamples. Survival probabilities decline steadily over time across all groups, with significant differences in mortality patterns based on Grade of LTC needs. In the first year, 13.2% of individuals with Grade I LTC needs deceases, compared to 28.7% in Grade III, underscoring the increased vulnerability of individuals requiring more intensive LTC. Mortality rates peak in the second year, reaching 22.2% for Grade I and 24.5% for Grade II. From the third year onward, annual mortality rates decrease relative to the initial population but continue to rise when compared to the remaining cohort. Mortality remains consistently higher among individuals with greater LTC needs, reinforcing the strong association with survival outcomes.

Figure 4 presents four panels, one for the Full Sample and one for each Subsample, displaying Kaplan–Meier estimates of the survival function by type of benefit received, as classified in Section 4.1. Across all panels, a clear and consistent pattern emerges, indicating lower hazards for people receiving LTC benefits. Individuals without benefits exhibit the lowest survival probabilities over time, with the greatest disparities occurring in the initial years. This suggests that individuals without benefits may either die before receiving any formal support or face higher mortality risks due to unmet care needs. In particular, longer waiting lists for certain services, such as residential care, could increase these risks. Notably, this pattern becomes more pronounced as the Grade of LTC needs increases (further details are provided in Table 6 in Appendix 6).

Conversely, individuals who have received a combination of home and residential care exhibit the highest survival probabilities throughout most of the follow-up period. This pattern may indicate that access to both types of care provides comprehensive and flexible support, ensuring adaptability to new needs and addressing a broader range of LTC needs. In this group, individuals spent the majority of their time either waiting for benefits or receiving home care, transitioning to residential care at a later stage (see Table 7 Appendix 7 for more information). This transition may reflect a response to health deterioration, where residential care could potentially help mitigate further declines and extend life expectancy. However, with the exception of individuals in the Grade III subsample, this advantage diminishes in later years, with survival probabilities in the combined-care group declining more rapidly from year 6 in the Grade I subsample and from year 7 in the Grade II subsample. This reversal may reflect the fact that transitioning to residential care often signals a deterioration in health status, which might be temporarily mitigated by the change in benefits but ultimately fails to offset underlying declines in health.

Finally, individuals receiving only home care generally exhibit higher survival probabilities than those in residential care. This gap widens over time, highlighting the limited capacity of institutional care to offset ongoing health decline. However, as LTC needs become more severe, this difference narrows—suggesting that the comparative advantage of home-based care diminishes with increasing care intensity and health deterioration.

Fig. 3 Kaplan–Meier Mortality Plots for the Full Sample (3a, Left Panel) and the different Subsamples (3b, Right Panel)

Note: Kaplan-Meier survival curves for all-cause mortality for the Full Sample and the three Subsamples. The log-rank test confirms statistically significant differences among groups (p < 0.0001).

Fig. 4 Kaplan–Meier Mortality Plots for in function of the Type of Benefit received for the Full Sample and the different Subsamples

Kaplan-Meier survival curves for all-cause mortality, depending on the Type of Benefit Received for the Full Sample (Top-Left Panel), Grade I of LTC needs subsample (Top-Right Panel), Grade II of LTC needs subsample (Bottom-Left Panel), Grade III of LTC needs subsample (Bottom-Right Panel). The log-rank test confirms statistically significant differences among type-of-benefit groups (p < 0.0001).

4.3. Multivariate Cox regression model

Table 2 presents the results of the multivariate Cox proportional hazards model for the full sample and the stratified subsamples based on initial grades of LTC needs (I, II, III). In the full-sample model, initial LTC grade is included as a covariate. Time-varying coefficients (Zhang et al., 2018) are included for the interest LTCS variables when the proportional hazards assumption was violated —a common issue in models with extended follow-up periods (Thomas & Reyes, 2014). Fixed effects to control for unobserved heterogeneity across heath and LTCS administrative units were included. All models demonstrated strong fit, with concordance statistics ranging from 0.73 to 0.77 and all global tests (likelihood ratio, Wald, and log-rank) showing high statistical significance (p < 0.001).

The initial grade of LTC need emerged as the most influential predictor of mortality. Individuals assessed at Grade III had a hazard ratio (HR) of 4.30, implying a 330% higher risk of death compared to those at Grade I. Receiving LTC benefits is significantly associated with improved survival outcomes. Relative to individuals without any benefit, all types of care—home care, residential care, and their combination—exhibited much lower HRs.

For individuals in Grade II and III subsamples, receiving home care is associated with a lower HR compared to residential care, suggesting a potentially more favourable impact of home-based services on survival outcomes. However, this pattern does not extend to the Grade I subsample, where residential care appears to be linked to a lower HR than home care. This counterintuitive result likely reflects a selection mechanism: individuals initially classified as Grade I are not eligible for residential care and must undergo a reassessment to be reclassified as Grade II or III before accessing institutional services. Consequently, those who eventually enter residential care from Grade I experience a delay, introducing a distortion in the observed hazard ratios.

A similar pattern is observed among individuals who receive a combination of home and residential care—those who initially benefit from home-based services and later transition to institutional care. As shown in Appendix 7, this group spends the majority of their care trajectory in home settings, with the shift to residential care typically occurring at a later stage, likely in response to a deterioration in health status or the completion of waiting periods for institutional placement. However, this subgroup is particularly noteworthy as it suggests that, in cases of advanced LTC needs, moving to residential care may temporarily mitigate the progression of health decline compared to home care, thereby potentially extending survival. The consistently low hazard ratios associated with this group (HR = 0.04 across all samples and subsamples, and 0.03 in Grade I) underscore the strength of this association and highlight the potential value of flexible, adaptive care pathways that respond to evolving needs.

Time interactions reveal small but significant trends. The hazard associated with Grade III decreased modestly over time, while the hazards of all LTC benefit types slightly augmented. This likely reflects progressive health deterioration that attenuates the initial benefits of receiving care.

Among individual-level covariates, women exhibited lower mortality risks than men across all subsamples, though the effect diminishes as care needs increased. Age was positively associated with hazard and showed a consistent effect across models. Spanish nationality was associated with a higher hazard relative to other nationalities, which may reflect demographic composition rather than intrinsic risk, as a 99% of the sample is Spanish. Regarding civil status, only being married was statistically significant, with slightly higher hazard ratios; other statuses showed no meaningful effect.

The effects of DGs on mortality risk demonstrated considerable variability. Using neurological conditions as the reference category, individuals diagnosed with osteoarticular, mental health, and sensory (eye and ear) conditions exhibited lower mortality risks. In contrast, all other diagnostic categories were associated with increased mortality hazards. Notably, individuals with haematological diseases (HR = 1.49), neoplasms (HR = 1.41), and respiratory conditions (HR = 1.30) experienced significantly elevated risks. A convergence effect was observed with increasing levels of LTC need, wherein hazard ratios tended to approach to one across most DGs. However, the magnitude of this convergence effect varied depending on the specific DGs.

Table 2 Multivariate Cox Regression models for the Full Sample and the different Subsamples

HR: Hazards Ratio. CI: Confidence Interval. Ref: Reference. DG: Diagnostic Group. *** p<0.001. ** p<0.01. * p<0.05. Fixed-Effect variables include Health Area and SEVAD covariates. For concordance, standard deviation is 0.00 in the four models.

4.4. Robustness checks

To assess the robustness of the results, three additional analyses were performed, as shown in Table 8 of Appendix 8. First, the numeric BVD score was used in place of the categorical Grade to measure LTC needs. Second, income level was added as an additional covariate. Third, age was modelled as a categorical variable in intervals rather than as a continuous variable. The results from these analyses closely mirror those presented in Section 4.3, reinforcing the robustness of the findings.

5. Conclusion and policy discussion

This study provides a comprehensive analysis of the relationship between LTC system and its benefits with mortality risk in Catalonia. Using individual-level administrative data on over 320,000 individuals aged 50 and above from 2015 to 2024, survival analysis techniques were applied—Kaplan-Meier curves and Cox proportional hazard models—to assess how mortality outcomes vary by Grade of care needs and the type of benefit received. The analysis accounts for demographic, clinical, and geographic covariates, enabling robust estimation of adjusted hazard ratios. The findings represent the first analysis of this administrative database since the full implementation of the Dependency Act in July 2015 (Peña-Longobardo et al., 2016), offering a contemporary perspective on the Catalan LTCS.

The results suggest a strong association between LTC benefits and lower mortality risk. Across all levels of LTC needs, individuals receiving any form of LTC support faced significantly lower hazards of death compared to those without benefits, although the magnitude and timing of this survival advantage varied. These findings align with and extend previous evidence suggesting that formal care provision can improve health outcomes among frail older adults (Costa-Font et al. (2018) Hernández-Pizarro et al. (2018); Serrano-Alarcón et al. (2022)).

Notably, individuals who received home-based care and subsequently transitioned to residential care exhibited the highest survival probabilities, which may reflect patterns of care use in response to changing health needs. The elevated survival rates in this group likely reflect a dual mechanism: the short-term stabilizing effect of home care, followed by the capacity of institutional care to meet escalating needs. However, the survival benefit associated with combined care pathways diminishes over longer follow-up periods, particularly among individuals with lower initial LTC needs. This attenuation suggests that while formal LTC may delay mortality by addressing immediate care needs, it may not fully arrest the progression of underlying health deterioration in ageing populations.

Results also indicate that residential care, when considered independently, is associated with a slightly increased mortality risk compared to receiving care at home, which shows lower HRs. However, the evidence suggests that residential care may be associated with better outcomes when appropriately matched to care needs and implemented at the right time, underscoring the importance of smooth transitions between care settings. Ensuring that institutionalization is well-timed and appropriately matched to care needs is crucial for maximizing outcomes.

In contrast, individuals not receiving any formal LTC benefits face higher mortality risks, likely due to both unmet care needs and systemic delays in access—effects that are especially pronounced among those with severe care requirements (Grade III), for whom baseline mortality hazard is already elevated. The paper also reinforces well-established patterns: mortality risk increases with age, is higher for men, and is significantly influenced by specific comorbidities (like haematological and respiratory diseases).

This study has some limitations. First, while the Cox models adjust for a wide range of covariates, they cannot fully eliminate selection bias. Individuals who receive benefits may differ in unobserved ways from those who do not—particularly regarding family support or health literacy. Secondly, benefit allocation is not randomized, compromising the causality of the results. Third, although robustness checks consolidate the findings, certain subsample exclusions (e.g., those who die before receiving a Grade of LTC needs) may still introduce bias. Additionally, while mortality is an important and objective outcome, it does not capture the quality of life, wellbeing, or satisfaction of beneficiaries—dimensions increasingly emphasized in international frameworks such as ASCOT (Malley et al., 2012). This is relevant specially in cases of people with advanced LTC needs, as it would be crucial to determine whether low HRs could alternatively reflect prolonged decline rather than improved life quality. Finally, our findings are based on the Catalan implementation of the Spanish LTCS and may not be fully generalizable to other regional or international contexts.

Future work could extend this analysis by incorporating quality-of-life outcomes alongside mortality, exploring care transitions in greater temporal detail, or comparing results across regions to assess institutional variation.

Linking LTC data with healthcare utilization or cost records would also support evaluations of system efficiency and cost-effectiveness.

To conclude, this investigation demonstrates that formal LTC benefits—especially when flexible and timely—are associated with lower mortality risk among older adults in Catalonia. The results support the value of Spain’s LTCS framework, while highlighting areas for urgent improvement, specifically the timeliness of benefit allocation and the ability to ensure continuity of care across settings. These findings lend support to ongoing deinstitutionalization efforts, but also caution that such shifts must be accompanied by robust, well-resourced home care systems to avoid unintended negative health consequences, and reinforce the role of residential care in specific situations. As ageing societies seek to balance equity, efficiency, and sustainability in care provision, this evidence reinforces the importance of flexible, integrated, and responsive care systems that can adapt to the evolving needs of older populations.

Funding: This work was supported by the Doctoral Industrial Plan of the Ministry of Research and Universities at Generalitat de Catalunya (Government of Catalonia).

Data availability: The dataset used in this paper is not publicly available due to privacy reasons. It is possible to apply for data access at the Department of Social Rights and Inclusion at the Catalan Government (https://dretssocials.gencat.cat/ca/contacte_dasc/index.html).

6. Bibliography

References

1. Bermejo-Patón, F., del Pozo-Rubio, R., Amo-Saus, M. E., & Moya-Martínez, P. (2023). The recovery and resilience plan on the long-term care system. Towards a deinstitutionalization? Frontiers in Public Health, 11, 1130132. https://doi.org/10.3389/FPUBH.2023.1130132/BIBTEX

References

1. Carreras, M., Ibern, P., & Inoriza, J. M. (2018). Ageing and healthcare expenditures: Exploring the role of individual health status. Health Economics, 27(5), 865–876. https://doi.org/10.1002/HEC.3635

References

1. Charmes, Jacques. (2019). The unpaid care work and the labour market: an analysis of time use data based on the latest world compilation of time-use surveys. ILO.

References

1. Chida, Y., & Steptoe, A. (2008). Positive psychological well-being and mortality: A quantitative review of prospective observational studies. Psychosomatic Medicine, 70(7), 741–756. https://doi.org/10.1097/PSY.0B013E31818105BA

References

1. Costa-Font, J., Jimenez-Martin, S., & 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

References

1. Costa-Font, J., Jimenez-Martin, S., Vilaplana-Prieto, C., Viola, A., Jimenez-Martin Universitat Pompeu Fabra Ramon Trias Fargas, S., & Gse, B. (2023). Universalizing the Access to Long-term Care: Evidence from Spain. https://doi.org/10.3386/W31825

References

1. Costa-Font, J., Jiménez Martin, S., & Viola, A. (2021). Fatal Underfunding? Explaining COVID-19 Mortality in Spanish Nursing Homes. Journal of Aging and Health, 33(7–8), 607–617. https://doi.org/10.1177/08982643211003794/ASSET/IMAGES/LARGE/10.1177_08982643211003794-FIG10.JPEG

References

1. Cox, D. R. (1972). Regression Models and Life-Tables. Journal of the Royal Statistical Society: Series B (Methodological), 34(2), 187–202. https://doi.org/10.1111/J.2517-6161.1972.TB00899.X

References

1. Decreto 504/2007, de 20 de Abril, Por El Que Se Aprueba El Baremo de Valoración de La Situación de Dependencia Establecido Por La Ley 39/2006, de 14 de Diciembre, de Promoción de La Autonomía Personal y Atención a Las Personas En Situación de Dependencia., Gobierno de España 17646 (2007). https://www.boe.es/buscar/doc.php?id=BOE-A-2007-8350

References

1. Departament de Drets Socials i Inclusió. (2024). Informe de la situació de la dependència (desembre 2023).

References

1. Dudley, W. N., Wickham, R., & Coombs, N. (2016). An Introduction to Survival Statistics: Kaplan-Meier Analysis. Journal of the Advanced Practitioner in Oncology, 7(1). https://doi.org/10.6004/JADPRO.2016.7.1.8

References

1. Edemekong, P. F., Bomgaars, D. L., Sukumaran, S., & Schoo, C. (2023). Activities of Daily Living. Ryan's Occupational Therapy Assistant: Principles, Practice Issues, and Techniques, Fifth Edition, 513–532. https://doi.org/10.4324/9781003526346-41

References

1. Eggink, E., Ras, M., & Woittiez, I. (2017). Dutch long-term care use in an ageing population. The Journal of the Economics of Ageing, 9, 63–70. https://doi.org/10.1016/J.JEOA.2016.08.001

References

1. Esping-Andersen, G. (2016). Families in the 21st Century. In Population and Development Review (Issue 4). SNS förlag. https://doi.org/10.1111/PADR.12019

References

1. European Commission. (2024). Long-term care - Employment, Social Affairs & Inclusion - European Commission. https://ec.europa.eu/social/main.jsp?langId=en&catId=792

References

1. Fox, J., & Weisberg, S. (2023). Cox Proportional-Hazards Regression for Survival Data in R. http://tinyurl.com/carbook

References

1. Gobierno de España. (2021). Plan de Recuperación, Transformación y Resiliencia. COMPONENTE 22 Plan de choque para el Economía de los Cuidados y Refuerzo de las políticas de igualdad e inclusión.

References

1. Goel, M. K., Khanna, P., & Kishore, J. (2010). Understanding survival analysis: Kaplan-Meier estimate. International Journal of Ayurveda Research, 1(4), 274. https://doi.org/10.4103/0974-7788.76794

References

1. Hernández-Pizarro, H. M. (2016). The Effect of Long-Term Care Benefits on Mortality (University Pompeu Fabra Working Paper Series).

Hernández-Pizarro, H. M., Nicodemo, C., & Casasnovas, G. L. (2020). Discontinuous system of allowances: The response of prosocial health-care professionals. Journal of Public Economics, 190, 104248. https://doi.org/10.1016/J.JPUBECO.2020.104248

References

1. Hernández-Pizarro, H. M., Serrano-Alarcon, M., López-Casasnovas, G., Nicodemo, C., & Vall, J. (2024). Should I stay or Should I go: the effects of Long-Term care provision setting on hospitalizations.

References

1. Howdon, D., & Rice, N. (2018). Health care expenditures, age, proximity to death and morbidity: Implications for an ageing population. Journal of Health Economics, 57, 60–74. https://doi.org/10.1016/J.JHEALECO.2017.11.001

References

1. IMSERSO. (2024). Información estadística destacada del SAAD. https://imserso.es/ca/el-imserso/documentacion/estadisticas/sistema-autonomia-atencion-dependencia-saad

References

1. INE. (2024). Nota de Prensa: Contabilidad Nacional Trimestral de España. Cuarto trimestre 2023. Avance. https://www.ine.es/dyngs/Prensa/avCNTR4T23.htm

References

1. Instituto Nacional de Estadística. (2022). Nota de Prensa: Proyecciones de Población 2022-2072. Notas de Prensa.

References

1. Kassamabra, A., Kosinski, M., & Biecek, P. (2024). survminer: Drawing Survival Curves using “ggplot2.” In R package vignette, version 0.5.0. https://cran.r-project.org/web/packages/survminer/survminer.pdf

References

1. Koller, D., Schön, G., Schäfer, I., Glaeske, G., Van Den Bussche, H., & Hansen, H. (2014). Multimorbidity and long-term care dependency - A five-year follow-up. BMC Geriatrics, 14(1), 1–9. https://doi.org/10.1186/1471-2318-14-70/TABLES/3

References

1. Ley 39/2006, de 14 de Diciembre, de Promoción de La Autonomía Personal y Atención a Las Personas En Situación de Dependencia. (2006). https://www.boe.es/eli/es/l/2006/12/14/39/con

References

1. Lin, H. R., Otsubo, T., & Imanaka, Y. (2017). Survival analysis of increases in care needs associated with dementia and living alone among older long-term care service users in Japan. BMC Geriatrics, 17(1), 1–10. https://doi.org/10.1186/S12877-017-0555-8/TABLES/5

References

1. Lorenz, N., Ihle, P., & Breyer, F. (2020). Aging and Health Care Expenditures: A Non-Parametric Approach (8216; CESifo Working Paper,). Munich: Center for Economic Studies and ifo Institute (CESifo). https://www.econstor.eu/handle/10419/216612

References

1. Malley, J. N., Towers, A. M., Netten, A. P., Brazier, J. E., Forder, J. E., & Flynn, T. (2012). An assessment of the construct validity of the ASCOT measure of social care-related quality of life with older people. Health and Quality of Life Outcomes, 10(1), 1–14. https://doi.org/10.1186/1477-7525-10-21/TABLES/6

References

1. National Institute on Aging. (2025). Long-Term Care Facilities: Assisted Living, Nursing Homes, and Other Residential Care. https://www.nia.nih.gov/health/assisted-living-and-nursing-homes/long-term-care-facilities-assisted-living-nursing-homes

References

1. OECD. (2023). Health at a Glance 2023: OECD Indicators (OECD Publishing, Ed.). https://doi.org/10.1787/7a7afb35-en

References

1. Oliva, J., Sancho Castiello, M., & del Pozo-Rubio, R. (2023). Sistema para la Autonomía y Atención a la Dependencia: primeros 15 años y una mirada hacia el futuro. Gaceta Sanitaria, 37, 102253. https://doi.org/10.1016/J.GACETA.2022.102253

References

1. Palladino, R., Lee, J. T., Ashworth, M., Triassi, M., & Millett, C. (2016). Associations between multimorbidity, healthcare utilisation and health status: evidence from 16 European countries. Age and Ageing, 45(3), 431–435. https://doi.org/10.1093/AGEING/AFW044

References

1. Pardo-Garcia, I., Martinez-Lacoba, R., Moya-Martínez, P., Amo-Saus, E., del Pozo-Rubio, R., & Escribano-Sotos, F. (2024). Mortality risk factors in the dependent population of Castilla-La Mancha (Spain) before and during the first COVID-19 wave. Heliyon, 10(7). https://doi.org/10.1016/J. HELIYON.2024.E28931

References

1. Peña-Longobardo, L. M., Oliva-Moreno, J., García-Armesto, S., & Hernández-Quevedo, C. (2016). The Spanish long-term care system in transition: Ten years since the 2006 Dependency Act. Health Policy, 120(10), 1177–1182. https://doi.org/10.1016/J.HEALTHPOL.2016.08.012

References

1. Ramírez-Navarro, J., Revilla Castro, A., Fuentes Jiménez, M., Sanz Yagüez, D., Nartñibez i Llopis, M., García Aonso, E., & Cavero Cano, G. (2024). Informe XVIII Aniversario de la Ley de Dependencia. https://directoressociales.com/wp-content/uploads/2024/12/XVIII-Aniversario-INFORME-FINAL-v2.pdf

References

1. Real Decreto 174/2011, de 11 de Febrero, Por El Que Se Aprueba El Baremo de Valoración de La Situación de Dependencia Establecido Por La Ley 39/2006, de 14 de Diciembre, de Promoción de La Autonomía Personal y Atención a Las Personas En Situación de Dependencia., 18567 (2011). https://www.boe.es/eli/es/rd/2011/02/11/174

References

1. Reinwarth, A. C., Wicke, F. S., Hettich, N., Ernst, M., Otten, D., Brähler, E., Wild, P. S., Münzel, T., König, J., Lackner, K. J., Pfeiffer, N., & Beutel, M. E. (2023). Self-rated physical health predicts mortality in aging persons beyond objective health risks. Scientific Reports 2023 13:1, 13(1), 1–9. https://doi.org/10.1038/s41598-023-46882-7

References

1. Rich, J. T., Neely, J. G., Paniello, R. C., Voelker, C. C. J., Nussenbaum, B., & Wang, E. W. (2010). A practical guide to understanding Kaplan-Meier curves. Otolaryngology—Head and Neck Surgery, 143(3), 331–336. https://doi.org/10.1016/J.OTOHNS.2010.05.007

References

1. Rodríguez-González, A. M., & Rodríguez-Míguez, E. (2021). La medición de la dependencia a partir del tiempo de cuidado y del baremo de valoración de dependencia. Revista Española de Geriatría y Gerontología, 56.1, 35–40. https://doi.org/10.1016/j.regg.2020.06.006

References

1. Serrano-Alarcón, M., Hernández-Pizarro, H., López-Casasnovas, G., & Nicodemo, C. (2022). Effects of long-term care benefits on healthcare utilization in Catalonia. Journal of Health Economics, 84, 102645. https://doi.org/10.1016/J.JHEALECO.2022.102645

References

1. Steptoe, A., Deaton, A., & Stone, A. A. (2015). Subjective wellbeing, health, and ageing. The Lancet, 385(9968), 640–648. https://doi.org/10.1016/S0140-6736(13)61489-0/ATTACHMENT/E951A360-9053-415F-B773-FCC9A4F2B3DC/MMC1.PDF

References

1. Tang, K. K., Chin, J. T. C., & Rao, D. S. P. (2008). Avoidable mortality risks and measurement of wellbeing and inequality. Journal of Health Economics, 27(3), 624–641. https://doi.org/10.1016/J.JHEALECO.2007.08.003

References

1. Therneau, T. (2024). A package for survival analysis in R. R package vignette, version 3.5-8. https://cran.r-project.org/web/packages/survival/vignettes/survival.pdf

References

1. Therneau, T. M. (2000). Modeling Survival Data: Extending the Cox Model (Springer Science+, Ed.). Business Media New York.

Thomas, L., & Reyes, E. M. (2014). Tutorial: Survival Estimation for Cox Regression Models with Time-Varying Coefficients Using SAS and R. Journal of Statistical Software, 61(Code Snippet 1), 1–23. https://doi.org/10.18637/JSS.V061.C01

References

1. Tian, L., Zucker, D., & Wei, L. J. (2003). On the Cox Model with Time-Varying Regression Coefficients. http://biostats.bepress.com/harvardbiostat/paper4

Van den Bosch, K., Geerts, J., & Willemé, P. (2013). Long-term care use and socio-economic status in Belgium: a survival analysis using health care insurance data. Archives of Public Health, 71(1), 1–9. https://doi.org/10.1186/0778-7367-71-1/TABLES/4

Vidiella-Martin, J., Hernández-Pizarro, H. M., García-Gómez, P., & López-Casasnovas, G. (2024). Unravelling hidden inequities in a universal public long-term care system. The Journal of the Economics of Ageing, 29, 100527. https://doi.org/10.1016/J.JEOA.2024.100527

WHO. (2015). World report on Ageing And Health (World Health Organization, Ed.). www.who.int

References

1. WHO Centre for Health Development, Organisation for Economic Co-operation and Development, Barber, S. L., Van Gool, K., Wise, S., Woods, M., Or, Z., Penneau, A., Milstein, R., Ikegami, N., Kwon, S., Bakx, P., Schut, E., Woutersee, B., Flores, M., & Lorenzoni, L. (2021). Pricing long-term care for older persons. World Health Organization. https://iris.who.int/handle/10665/344505

World Health Agency. (2019). ICD-10 Version:2019. https://icd.who.int/browse10/2019/en

References

1. Wouterse, B., Bakx, P., & Wong, A. (2023). Measuring Nursing Home Performance Using Administrative Data. Medical Care Research and Review, 80(2), 187–204. https://doi.org/10.1177/10775587221108247/ASSET/IMAGES/LARGE/10.1177_10775587221108247-FIG9.JPEG

Zhang, Z., Reinikainen, J., Adeleke, K. A., Pieterse, M. E., & Groothuis-Oudshoorn, C. G. M. (2018). Time-varying covariates and coefficients in Cox regression models. Annals of Translational Medicine, 6(7), 121–121. https://doi.org/10.21037/ATM.2018.02.12

Zunzunegui, M. V., Rico, M., Béland, F., & García-López, F. J. (2022). The Impact of Long-Term Care Home Ownership and Administration Type on All-Cause Mortality from March to April 2020 in Madrid, Spain. Epidemiologia 2022, Vol. 3, Pages 323-336, 3(3), 323–336. https://doi.org/10.3390/EPIDEMIOLOGIA3030025

Appendix

Appendix 1.

Fig. 5 Weekly crude mortality rate (MR) per 1000 people with recognized levels of long-term care (LTC) needs

The graph shows the weekly mortality rate (per 1000 individuals) from January 2016 to the January of 2024. There is a clear seasonal pattern with periodic peaks during the winter and a significant spike around 2020, corresponding to the first wave of the COVID-19 pandemic. The trend re-stabilizes afterwards.

Appendix 2. Benefits

Table 3 Types of LTC benefits included in each categorical group

Appendix 3.

Table 4 List of Diagnostic Groups (DG) and ICD-10 classification of the diseases included in each DG

Appendix 4.

Table 5 Distribution of the population by Health Area and SEVAD

Appendix 5.

Fig. 8 Histogram of the Distribution of Personal Income

The figure displays the distribution of personal income among individuals in our sample, measured at the time of benefit allocation. For better readability, it includes only those with income below €100,000, who represent 99.6% of the sample. The median is located at 12,727€, with the first quartile at 9,483€ and the third at 17,912€.

Appendix 6.

Table 6 Distribution of Cases Without Benefits by Time Since Application and termination reason

This table presents a breakdown of individuals in the “Without Benefit” category of the “Type of Benefit” variable, focusing on whether their time in the system fell below the average waiting periods for receiving LTC benefits. Two reference thresholds are used: 177 days, which corresponds to the average time between the application for LTC recognition and the completion of the needs assessment, and 289 days, which includes both the assessment period and the average time required to assign a specific benefit. The table reports how many individuals died or reached the end of the observation period (December 31, 2024) before exceeding these thresholds. These two outcomes help explain why a considerable proportion of individuals in this category did not receive any benefit. In the table, n refers to the total number of individuals falling under each condition, and % indicates the proportion of those individuals relative to the total number of people without benefit in the corresponding sample. A third potential waiting period—related to the allocation of specific services such as residential care—is not included, as it depends on various factors including geographical location, timing, service type, and centre-specific availability. Additionally, for voucher or cash benefits, the administrative data reflect the date from which individuals are entitled to receive the benefit, rather than the actual start of benefit reception, as these are retroactively granted six months after application. It is important to emphasize that these thresholds are based on averages and may not represent the experience of every individual. However, since all individuals included in this table have had their needs assessment completed, it is likely that the absence of benefits is primarily due to systemic waiting times rather than individual choice.

Appendix 7.

Table 7 Total Time Interval and Proportion in Institutional Care for the “Home & Residential Care” Group This table presents the average and median interval durations (in days) from the application for LTCS to death for the Full Sample and each of the Subsamples used in the analysis. Additionally, it shows the proportion of time individuals spend in residential care, relative to the time spent awaiting benefits or receiving home care. The results indicate that time in residential care begins at relatively low levels and increases with Grade, although it never exceeds 45% of the total duration. These findings suggest that the transition to residential care may reflect a response to health deterioration, with residential care potentially serving to mitigate further declines in health.

Appendix 8.

Table 8 Robustness checks

HR: Hazards Ratio. Ref: Reference. DG: Diagnostic Group. Signif.: Significance. *** p<0.001. ** p<0.01. * p<0.05. Fixed-Effect variables include Health Area and SEVAD covariates. For the concordance, standard deviation is 0.00 in the four models. Model A is the one used in the study, in this case without adding the time-varying coefficients. Model B includes the BVD punctuation as a numerical value instead of categorizing by Grades of LTC Needs. Model C includes income level of the individuals as a control variable. For this model, the category “Without Benefit” in “Type of Benefit” variable is excluded as it loses an 85% of the observations. In Model D, age has been included as a categorical variable with three groups (50-65, 65-80, 80+).