Dependency Change with Aging and Associated Factors in Uruguay: A Cohort Study

Abstract

Objectives: To assess the heterogeneity of transitions toward dependency in older adults and to explore the robustness of results to different operationalizations of dependency. Method: Using data from people aged 60 years and older from a national representative study in Uruguay (Encuesta Longitudinal de Protección Social, N = 5071), we fitted multinomial regressions adjusted by sociodemographic and health characteristics to model transitions into dependency and death. We used a harder operationalization with basic activities of daily living (Katz-dependency) and Comprehensive-dependency with basic, instrumental, and advanced activities. Results: Increasing age (RRR = 1.08, CI = [1.05; 1.12], p < .001) and having comorbidities (RRR = 2.16, CI = [1.31; 3.57], p = .003) increased the risk of transition from nondependent to dependent using Katz-dependency. Women with at least two chronic conditions have increased risk of Comprehensive-dependency (RRR = 1.79, CI = [1.15; 2.80], p = .010). Discussion: Inconsistencies in findings emerged when evaluating transitions into dependency with the different measures, which may have social care implications.

Keywords

dependency activities of daily living older adults Uruguay

Introduction

Older populations exhibit higher prevalence of chronic diseases, disability, and dependency, and as a consequence of these higher prevalences, national social security and healthcare systems are usually placed under stress (At et al., 2015; Prina et al., 2020; Sousa et al., 2010). With increasing age, the risk of dependency tends to increase as functional or mental impairments result from declines in health. Yet, the need for daily help or care of older adults living in the community has been shown to be heterogeneous and depend on sociodemographic characteristics such as age, sex, or education (At et al., 2015; Sousa et al., 2009, 2010).

Demographic changes experienced in the past decades in low- and middle-income countries have increased the older segments of the population, stressing the social security, healthcare system, and also could be leading to a deterioration in the quality of life of the older population. For instance, in Latin American countries, the share of older people has been increasing since 1990 and is expected to double between 2019 and 2050 (United Nations, 2020). Uruguay, a country with a population of 3.4 million, exhibits the oldest population in Latin America, with 19% of this population aged 60 years or older in 2015 (Rofman et al., 2016). Furthermore, it has been reported that Uruguay began its demographic transition earlier than other Latin American countries (Rofman et al., 2016). Yet, despite being the country with the oldest population in Latin America, the need for help, care, or support of older Uruguayan adults, and its changes over time, has not been studied widely.

Developed countries have already gone through these demographic changes and have therefore planned, implemented, and reformed their policies to deal with problems associated with dependency of the older population (Matus-López, 2015). However, only some Latin American countries have implemented measures to address these problems, in particular, the ones associated with increased dependency to care or support (Galiani et al., 2017; Matus-López, 2015; Matus-López & Rodríguez-Modroño, 2014). Uruguay stands out among Latin American countries due to the implementation of a new Long-Term Care (LTC) program within the framework of the National Care System (NCS) initiated in 2015. Among its objectives, the NCS aims to promote the development of autonomy of dependent older adults, providing them care through services and assistance (Matus-Lopez & Cid Pedraza, 2016).

The World Health Organization (WHO) defined disability as an umbrella term for impairments, activity limitations, and participation restrictions which denotes the negative aspects of the interaction between an individual with a health condition and that individual’s environmental and personal factors (WHO, 2001). Dependency, a multidimensional construct that could refer to functional, psychosocial, and supportive dimensions and that could also include an economic dimension (Edjolo et al., 2016; Rely et al., 2020), is defined as the need for human help or care beyond that habitually required by a healthy adult (Harwood et al., 2004).

Although the definitions of disability and dependency are clear, there is no consensus on how to operationalize dependency. For instance, in large national representative studies with a multidisciplinary scope such as the Cognitive Function and Ageing Study (CFAS) from England and the Survey of Health, Ageing and Retirement in Europe (SHARE), different operationalizations were used. Additionally, these differences in the operationalization were evidenced in previous analysis for Latin American countries (Monteverde et al., 2016). Specifically, while in the British study, dependency was operationalized using the interval measure, which is based on the lapsed time between periods when participants might require help (Kingston et al., 2017), in SHARE and for Latin American countries, it was operationalized counting the limitations in daily activities (Monteverde et al., 2016; Rodríguez-Sampayo et al., 2011). Moreover, it has been suggested that to adequately operationalize dependency, in addition to the actual limitations that individuals may experience, it is also necessary to consider information about the daily need for help (Albarrán Lozano et al., 2009; Edjolo et al., 2016; Harwood et al., 2004; Prina et al., 2020). Hence, it could be argued that it is possible to operationalize dependency using information about the need for assistance in performing activities of daily living (ADLs). So, in sum, to assess dependency, information about the need for human help will be used, in addition to information on the limitations to perform daily activities.

Much of the literature on disability and functional status of older adults have used the basic activities of daily living (BADLs) that includes activities such as bathing, dressing, toileting, transferring, continence, and feeding (Katz et al., 1963). Later, these activities were extended to include instrumental activities of daily living (IADLs) such as the ability to use the telephone, shopping, and housekeeping, among others (Lawton & Brody, 1969). Recently, advanced activities of daily living (AADLs), that includes social and community participation, communicating, and making decisions, are also considered (WHO, 2011).

Including different types of activities and its various combinations can inform about different aspects of dependency experienced by older adults (Millán-Calenti et al., 2010). For instance, a harder operationalization of dependency would consider only BADLs whilst more advanced and comprehensive assessment of the construct could also consider IADLs and AADLs.

Previous analyses have reported that trajectories toward dependency are characterized by a first need for help in IADLs and then an overlapping need for help in BADLs and IADLs (Edjolo et al., 2016).

The “male–female health-survival paradox” that postulates that women have longer life expectancies but they live in poorer health is a well-documented phenomenon (Austad, 2006; Bora & Saikia, 2015; Cameron et al., 2010). For instance, a national representative study for England concludes that the proportion of independent older people will increase in the next 20 years but men will be successively more independent than women (Kingston et al., 2017). Using data from several Latin American countries, it has been reported that women spend a longer time with difficulties than men (Prina et al., 2020). The patterns of dependency by sex could also be affected by men and women’s different roles in household activities, and a comparison between different cultures and contexts could provide new insights into patterns of heterogeneity (Millán-Calenti et al., 2010). Sex differences may be relevant to inform the actions to be taken to ensure quality of life of older people despite their difficulties.

In turn, educational level, socioeconomic conditions, physical or functional impairments, and the existence of comorbidities are risk factors for disability and dependency (Mayston et al., 2014; Prina et al., 2020), although more evidence is needed to understand the dynamics of disability and dependency (Prina et al., 2020).

The different assessment of dependency may provide complementary perspectives of the problem that informs policy design. For instance, to ensure that care or assistance meets needs of dependent older people, it is relevant to analyze differences in the type of activities that individuals will need help with. Previous research has not examined transitions to dependency and death or the robustness of results to different operationalizations of dependency. A better understanding of dependency, its changes with increasing age including death, and its associated factors is critical to inform care policies designed to improve quality of life of older people living in the community and their families.

In this first study of dependency of the population of older Uruguayans, we aim to generate knowledge about their dependency with increasing age, using data from a nationally representative cohort study and modeling transitions including death. We address the following questions:

Do transitions toward dependency vary by age, sex, and education after controlling for comorbidities and impairment in the population of older adults in Uruguay?

Are these transitions sensitive to the measures of dependency considered? Is there heterogeneity by age, sex, education, comorbidities, and impairments?

Methods

Data

We used data from the Encuesta Longitudinal de Protección Social (ELPS) for Uruguay. Baseline data were collected between 2012 and 2013 and the follow-up assessments between 2015 and 2016.

This survey is self-reported and representative of the national population of individuals aged 14 years or older. Participants were asked basic questions about their sociodemographic characteristics and health, in addition to questions about whether they received government benefits, their employment history, social security, heritage, and household composition. Specifically, participants were asked their age, sex, and the highest level of education completed. They were asked whether they had hearing or sight difficulties and were asked if they had been diagnosed with any of the following conditions: asthma, emphysema, arthrosis or arthritis, tendinitis, rheumatism, hypertension, diabetes, osteoporosis, kidney failure, cardiac problems, spine problems, anemia, cancer, and HIV.

For this study, we considered individuals aged 60 years and older at baseline. Of the 6197 individuals in this age-group surveyed at baseline, 1716 (28%) were not included in the follow-up wave (11% died, 1% moved to a nursing home, and 16% were untraceable). Hence, the analytical sample included 4462 individuals with complete data in both waves and 609 individuals with complete data in the first wave who died before the second wave, totalizing 5071 cases. For further information about the design of the baseline and follow-up data collection waves of the ELPS, see BPS (2015).

For the purposes of our analyses, we derived a series of dummy variables to account for differences in education (1 = highest educational level completed is primary school or less and 0 = otherwise) and sex (1 = female and 0 = male). Dummy variables were also created to indicate the presence of physical health conditions or difficulties. Specifically, a series of independent dummy variables were derived indicating whether an individual had difficulties with their sight (1 = have difficulties and 0 = otherwise), had hearing difficulties (1 = have difficulties and 0 = otherwise), and had two or more chronic diseases (1 = have two or more chronic conditions and 0 = have less than two chronic conditions). Age was measured in years.

Dependency measures

We measured dependency in two different ways. First, we defined an indicator variable that took the value of 1 if an individual needed help on a regular basis with at least one of the following activities: eating, dressing, personal care activities (toothbrush, to comb, or face washing), displacing inside home, or using the restroom (Katz et al., 1963). If an individual needed help with at least one of these activities on a regular basis (i.e., declared the need for help “Sometimes”, “Many times,” or “Always”), we classified the individual as dependent (Katz-dependency from now on). For the analysis of the Katz-dependency, we also considered the number of activities in which the person needs help from these five.

Second, we defined dependency using data on the frequency of need for help in any of the activities available in the ELPS study which includes basic, instrumental, and advanced activities (Comprehensive-dependency from now on). This approach adds to the five previous activities included in the definition of the Katz-dependency measure the following activities: avoiding health risk (e.g., taking medications), changing/maintaining position, displacing outside home, performing housework, participating in social life, and communicating. We also employ the number of activities in which the person needs help from these 11 for the analysis of Comprehensive-dependency and its associated factors.

The use of these two operationalizations of dependency allows us to assess dependency with a harder operationalization (Katz-dependency) that considers only BADLs, and with a more comprehensive assessment (Comprehensive-dependency) that considers also the IADLs and AADLs available in the survey, and therefore to perform a sensitivity analysis of results to these two different measures.

Statistical Analysis

To analyze transitions into dependency, we used multinomial logistic regressions to investigate the association between sociodemographic characteristics, comorbidities, and other limitations with transitions into dependency states and death.

In the first wave, there are two possible states: no dependency (ND) and dependency (D), while in the second wave, individuals may have died, and therefore, the additional deceased state (Dth) was considered (see Figure 1 for a pictorial representation of the model). In our multinomial model, staying nondependent (ND–ND) is the reference category, interpreting the coefficients relative to this transition.

Figure 1.

Three-state model of transitions into dependency states and death.

We were interested in the association of the different factors with the odds of experiencing any transition. Henceforth, we look for the relative risk ratio (RRR) of a given transition, relative to the risk of being nondependent in both periods (ND–ND). As an example, suppose we were interested in the differences between men and women in the risk of transition ND–D, given a set of control variables $X$ . The RRR representing the risk ratio women have relative to men of transition ND–D can be written as

RRR = \frac{P (transition = ND - D | X, woman = 1) / P (transition = ND - ND | X, woman = 1)}{P (transition = ND - D | X, woman = 0) / P (transition = ND - ND | X, woman = 0)}

We further investigated factors associated with both measures of dependency at baseline considering a multivariate perspective. We ran ordinary least squares (OLS) to evaluate the association of factors with the number of activities requiring help at baseline and logistic regressions to inquire on factors associated with the probability of being dependent.

The multivariate models were adjusted simultaneously by age, sex, education, presence of two or more chronic conditions, and visual and hearing difficulties. In addition, we included interactions of sex with age and with the presence of two or more chronic conditions in order to reflect the well-known “male-female health-survival paradox” documented in previous literature.

Individuals who died before the second wave were included in the regressions. Besides, there was no information that allowed us to determine the state at follow-up of those untraceable in the second wave (995 cases); hence, to analyze if they were different from those with complete data, we checked for differences in the mean of variables between lost individuals and those present in both waves.

All the statistical analyses were done first with the hardest measure of dependency (Katz-dependency) and then considering the more comprehensive measure of dependency (Comprehensive-dependency), and results were compared. Finally, we ran all analyses using Stata 13.

Results

Descriptive Statistics

Table 1 presents descriptive baseline statistics for the sample analyzed. The average age of individuals in the sample was 72 (SD = 8.2) years, and 61% of the sample were women. Almost 60% of the individuals reported that primary school was the highest educational level completed. Sight difficulties (7%) were more prevalent than hearing difficulties (2%), and 42% of individuals had at least two chronic diseases.

Table 1.

Descriptive statistics of analytical sample at baseline (N=5071).

Age	Mean (SD)	Sex	N (%)
Age (years)	71.9 (8.2)	Female	3074 (60.6)
Age ranges	N (%)	Male	1997 (39.4)
60–64	1104 (21.8)	Chronic conditions (CC)	N (%)
65–69	1153 (22.7)	Fewer than two CC	2917 (57.5)
70–74	984 (19.4)	Two or more CC	2154 (42.5)
75–79	797 (15.7)	Sight difficulties	N (%)
80–84	634 (12.5)	Sight difficulties (yes)	365 (7.2)
85 or more	399 (7.9)	Sight difficulties (no)	4706 (92.8)
Education	N (%)	Hearing difficulties	N (%)
Primary school or less (yes)	3011 (59.4)	Hearing difficulties (yes)	105 (2.1)
More than primary school (yes)	2060 (40.6)	Hearing difficulties (no)	4966 (97.9)

Table 2 presents the descriptive statistics for the dependency measures at baseline and the distribution of the sample for transitions among dependency states and death. Individuals classified as Katz-dependent at baseline are 4.4%, and this percentage almost doubles when using the Comprehensive-dependency measure (8.1%). On average, individuals required assistance in .09 (SD = .5) of the five Katz activities, while this number rises to .25 (SD = 1.1) when considering the 11 activities of the comprehensive measure.

Table 2.

Descriptive statistics of dependency measures of analytical sample at baseline (N=5071).

Number and percentage of people requiring help in each activity
Katz activities	N (%)	Other activities	N (%)
Eating	32 (.6)	Avoiding health risks	52 (1.0)
Dressing	116 (2.3)	Changing/maintaining position	68 (1.3)
Personal care	114 (2.2)	Displacing outside house	269 (5.3)
Displacing inside house	121 (2.3)	Performing housework	231 (4.6)
Using the restroom	92 (1.8)	Participating in social life	108 (2.1)
		Communicating	61 (1.2)
Katz-dependency	N (%)	Comprehensive-dependency	N (%)
Katz-dependent (yes)	224 (4.4)	Comprehensive-dependent (yes)	409 (8.1)
Katz-dependent (no)	4847 (95.6)	Comprehensive-dependent (no)	4662 (91.9)
Number of activities requiring help from	Mean (SD)	Number of activities requiring help from	Mean (SD)
5 activities of Katz	0.09 (.5)	11 activities of Comprehensive	0.25 (1.1)
Transitions
Katz-dependency	N (%)	Comprehensive-dependency	N (%)
ND–ND	4061 (80.1)	ND–ND	3650 (72.0)
ND–D	244 (4.8)	ND–D	527 (10.4)
ND–Dth	542 (10.7)	ND–Dth	485 (9.6)
D–ND	61 (1.2)	D–ND	90 (1.8)
D–D	78 (1.5)	D–D	177 (3.5)
D–Dth	85 (1.7)	D–Dth	142 (2.8)

Notes: ND–D: nondependent to dependent; ND–Dth: nondependent to death; D–ND: dependent to nondependent; D–D; dependent to dependent; D–Dth: dependent to death.

The analytical sample did not differ from individuals not followed in the second wave in terms of age (t-test, p = .18), sex (t-test, p = .68), education (t-test, p = .13), hearing impairments (t-test, p = .30), number of chronic conditions (t-test, p = .47), Katz-dependency status (t-test, p = .89), or Comprehensive-dependency status (t-test, p = .21). However, there is difference in terms of visual impairments (t-test, p = .05). These results suggest that there is no selection in the missing data.

Risk factors of dependency using the Katz-dependency measure

The results show (see Figure 2) increasing age is associated with a higher number of Katz activities requiring help on average (β = .004, SE = .001, p = .006). Also, the presence of at least two chronic conditions (β = .080, SE = .025, p = .001) and having sight difficulties (β = 0.156, SE = 0.028, p < .001) increase this number. Women have a lower mean number of activities that require help at baseline (β = −.509, SE = .129, p < .001), but, since the interaction of age and sex is positive and significant (β = .007, SE = .002, p < .001), increasing age increases the mean number of activities requiring help even more for women than for men. Although having hearing difficulties (β = .090, SE = .050, p = .074) or low educational level (β = .027, SE = .015, p = .072) may have a positive association with the activities requiring help, the coefficients do not reach traditional statistical significance. Finally, results did not show significant association of the interaction between having two or more chronic conditions and sex (β = −.028, SE = .031, p = .352).

Figure 2.

Associated factors to dependency at baseline. Panel (a): OLS for number of activities requiring help. Katz-D counts help in five activities and Comprehensive-D in eleven activities. Panel (b): logistic regression for dependent state at baseline. Katz-D defines dependent if required help in at least one activity of Katz-dependency measure and Comprehensive-D defines dependent if required help in at least one activity of the Comprehensive-dependency measure. Abbreviations: CC, chronic conditions. CI, confidence interval. OR, odds ratio.

Turning to the risk of being Katz-dependent at baseline, results show increasing age (OR = 1.07, CI = [1.04; 1.11], p < 0.001), having two or more chronic conditions (OR = 3.06, CI = [1.79; 5.24], p < 0.001), having sight difficulties (OR = 3.06, CI = [1.79; 5.24], p < .001), and a lower educational level (OR = 1.61, CI = [1.16; 2.24], p = .005) are associated with a higher risk of being Katz-dependent. However, there are no differences in the risk of being Katz-dependent between women and men (OR = .55, CI = [.03; 10.15], p=.690) either the interaction of age with sex (OR = 1.02, CI = [.98; 1.05], p = .431) nor having two or more chronic conditions with sex (OR = .66, CI = [.35; 1.26], p = .209) present statistically differences. These results are shown in Figure 2 and in the Supplementary Material (see Supplementary Table S1).

Table 3 panel (a) shows the results of the multinomial logistic regression using the Katz-dependency measure to define a dependent state of the individual. As individuals age, the results show that the relative risk of each transition is greater than that of the transition of being nondependent in both waves with the exception of remaining in a dependent state, where the estimate was not significant. Thus, we could expect that per additional year of older age individuals is more likely to move from a nondependent toward dependent state, relative to remain nondependent (RRR = 1.08, CI = [1.05; 1.12], p < 0.001). Having at least two chronic conditions increases the relative risk of moving from nondependent state toward dependent (RRR = 2.16, CI = [1.31; 3.57], p = .003) and from dependent to death (RRR = 5.35, CI = [2.30; 12.44], p < 0.001). Women also have a lower relative risk of transition to death from nondependent state (RRR = .05, CI = [.01; .31], p = .001). Having a lower educational level increases the risk of transition to death from a nondependent state (RRR = 1.43, CI = [1.16; 1.76], p = .001) and from dependent state (RRR = 2.00, CI = [1.11; 3.58], p = 0.020), relative to the risk of remaining nondependent in both waves. Finally, having sight difficulties increases the risk of transition from dependent to death (RRR = 3.76, CI = [2.22; 6.34], p < .001), recover from a state of dependency (RRR = 4.24, CI = [2.34; 7.69], p < .001), and remain dependent (RRR = 2.55, CI = [1.39; 4.65], p = .002), relative to stay nondependent in both waves.

Table 3.

Relative risk ratios for transitions to dependency and death. Base transition ND–ND.

	(a) Katz-dependency					(b) Comprehensive-dependency
	ND–D	ND–Dth	D–ND	D–D	D–Dth	ND–D	ND–Dth	D–ND	D–D	D–Dth
Age	1.08***	1.08***	1.11**	1.04	1.14***	1.09***	1.09***	1.05*	1.08***	1.15***
Age	(1.05; 1.12)	(1.06; 1.10)	(1.02; 1.20)	(0.98; 1.10)	(1.09; 1.20)	(1.07; 1.12)	(1.07; 1.11)	(.99; 1.11)	(1.03; 1.12)	(1.10; 1.19)
≥ 2 CC	2.16***	1.22	3.26*	2.03	5.35***	1.30	1.16	2.21*	2.43**	3.90***
≥ 2 CC	(1.31; 3.57)	(.92; 1.61)	(.91; 11.72)	(.86; 4.75)	(2.30; 12.44)	(.89; 1.91)	(.87; 1.55)	(.90; 5.39)	(1.23; 4.80)	(2.05; 7.40)
Female	2.87	.05***	211.61	0.01**	0.05	3.12	.06***	4.31	.37	.01**
Female	(.18; 45.04)	(.01; 0.31)	(.28; 159516)	(.00; .71)	(.00; 7.84)	(.41; 23.54)	(.01; 0.43)	(.04; 517.30)	(.01; 14.07)	(.00; .62)
Female*age	.99	1.03**	.94	1.07**	1.04	.99	1.03**	.98	1.02	1.06**
Female*age	(.95; 1.02)	(1.01; 1.05)	(.87; 1.03)	(1.01; 1.14)	(.98; 1.11)	(.96; 1.02)	(1.00; 1.05)	(.92; 1.05)	(.97; 1.07)	(1.01; 1.12)
Female*≥2 CC	1.31	1.09	.96	1.10	.36**	1.79**	1.15	2.24	1.66	.83
Female*≥2 CC	(.72; 2.41)	(.74; 1.60)	(.23; 4.05)	(.39; 3.07)	(.13; .99)	(1.15; 2.80)	(.76; 1.73)	(.76; 6.60)	(.75; 3.66)	(.37; 1.84)
Primary	1.20	1.43***	1.57	1.61*	2.00**	1.33***	1.43***	1.18	1.32	2.02***
Primary	(0.90; 1.59)	(1.16; 1.76)	(0.88; 2.70)	(0.95; 2.73)	(1.11; 3.58)	(1.08; 1.63)	(1.15; 1.78)	(0.76; 1.85)	(0.94; 1.85)	(1.28; 3.19)
Sight difficulties	1.43	1.35*	4.24***	2.55***	3.76***	1.86***	1.42*	4.80***	3.86***	4.32***
Sight difficulties	(.92; 2.20)	(.97; 1.88)	(2.34; 7.69)	(1.39; 4.65)	(2.22; 6.34)	(1.34; 2.58)	(.98; 2.08)	(.98; 2.08)	(2.57; 5.82)	(2.77; 6.74)
Hearing difficulties	.89	1.57*	.94	1.41	1.29	1.52	1.63	2.96**	1.42	2.28**
Hearing difficulties	(.37; 2.14)	(.94; 2.63)	(.22; 4.10)	(.48; 4.16)	(.51; 3.29)	(.83; 2.78)	(.90; 2.97)	(.90; 2.97)	(.60; 3.36)	(1.09; 4.74)
N	5071	5071	5071	5071	5071	5071	5071	5071	5071	5071

Notes: 95% confidence intervals in parentheses; ***p<0.01, **p<0.05, *p<0.1. ND–ND: nondependent to nondependent; ND–D: nondependent to dependent; ND–Dth: nondependent to death; D–ND: dependent to nondependent; D–D; dependent to dependent; D–Dth: dependent to death; CC: chronic conditions.

In sum, increasing age and having two or more chronic conditions were found to be the factors associated with a higher risk of transition from a nondependent to a dependent state, relative to remain nondependent with the Katz-dependency measure.

Risk factors of dependency using the Comprehensive-dependency measure

The frequency of each transition was sensitive to the dependency measure (see Table 1). While 4.8% of the sample transitioned from a nondependent to a dependent state when dependency was operationalized using Katz measure, this percentage has a twofold increase (10.4%) when considering the Comprehensive-dependency measure. However, for both measures, the permanence in a nondependent state is the most frequent.

In terms of dependency at baseline, the results for the Comprehensive-dependency measure are similar to the results obtained for Katz-dependency measure. The differences are that for the total of activities, the effect sizes are greater than for Katz’s activities and that hearing difficulties increased the risk of being Comprehensive-dependent, while it was not associated with being Katz-dependent (see Figure 2 and Supplementary Table S1).

In addition, when we use the Comprehensive-dependency measure rather than the Katz-dependency measure, the results show inconsistencies with respect to the factors associated with risk of transitions into dependent and nondependent states and death (see Table 2). While the effect of age does not change in direction and magnitude when the different measures are used, having two or more chronic conditions was not found to be significantly associated with the risk of becoming Comprehensive-dependent relative to remain nondependent, contrary to the previous significant result when the Katz-dependency measure was used. In addition, having a low educational level (RRR = 1.33, CI = [1.08; 1.63], p = .010) and sight difficulties (RRR = 1.86, CI = [1.34; 2.58], p = .000) were associated with a greater risk of transiting to a Comprehensive-dependent state, relatively to remain nondependent. More importantly, results show that the interaction between being female and having at least two chronic conditions increases the risk of becoming Comprehensive-dependent (RRR = 1.79, CI = [1.15; 2.80], p = .010). For instance, we could expect that a woman with two or more chronic conditions has a risk 1.79 times higher of moving from nondependent to Comprehensive-dependent relative to stay nondependent.

Sensitivity analyses and additional analyses

We performed two sensitivity analyses to study the robustness of our results. First, we changed the Katz-dependency measure, replacing the activity of “Displacing inside home” with “Changing/maintaining position.” Second, we analyzed the subsample of individuals who responded the survey by themselves, which implies a loss of 3% of the analytical sample (the subsample corresponds to 4468 cases). The results are in the Supplementary Material (see Supplementary Table S2–S5).

Further, we performed additional analyses which are also included in the Supplementary Material. We have fitted a series of multinomial logistic models adding the covariates one at a time and using interaction terms to capture the “male–female health–survival paradox.” Also, we have added a dummy variable to indicate whether the person lives alone or not (“Lives alone” = 1, the person lives alone; “Lives alone” = 0, otherwise). For the sake of brevity, we have reported in the Supplementary Material (Supplementary Table S6–S7) only the relative risk ratios for the transition ND–D, for both dependency operationalizations (Katz- and Comprehensive-dependency). Finally, we have considered the net household income of the last month (in logs) and included it as a covariate in the model. However, as 20% of the data on income is missing, the number of observations for models including income as a covariate is 4162. Hence, we added an additional estimation of the multinomial logistic regression with this subsample, and the results are reported in Table in the Supplementary Material (see Supplementary Table S8).

Discussion

Our work reports the first investigation about dependency and its dynamics for older adults in Uruguay using data from a national representative study. The main aims of the current work were to assess the heterogeneity of transitions toward dependency by sociodemographic characteristics in the population of older adults and to explore the sensitivity to different measures of dependency. In our analyses, heterogeneity of transitions toward dependency emerged, and results showed inconsistencies when evaluating these transitions with different measures.

First, we showed that it is more likely to transition to dependency when a comprehensive measure of the construct is used. Although this could be due to the greater number of activities in the measure, it also may be related to the different type of activities included. As previously reported in the literature, IADLs are the first in which people become dependent, and older people tend to have, on average, less difficulty with BADLs than with IADLs (Carmona-Torres et al., 2019; Edjolo et al., 2016; Graciani et al., 2004; Millán-Calenti et al., 2010; Rodríguez-Sampayo et al., 2011). The Katz-dependency measure includes only BADLs, while the Comprehensive-dependency includes also IADLs and AADLs and this may be increasing transitions toward dependency with the comprehensive measure. Consequently, operationalizing dependency with a comprehensive measure, including BADLs and IADLs, expands the target population of dependency care policies, even more than when the operationalization uses only BADLs. Moreover, in Uruguay’s current NCS, the official measure that determines whether a person can be a beneficiary of the program or not is a comprehensive operationalization of dependency (MIDES, 2018b). Although the methodology of the official scale is not publicly available, it is known that all activities asked in ELPS are included in the official scale construction (MIDES, 2018a, 2018b). Therefore, our results suggested that the comprehensive operationalization of dependency used in the NCS of Uruguay, by expanding the program’s target population, may increase program’s costs more than when the operationalization of dependency uses only BADLs. While it is desirable to consider a comprehensive measure of dependency (WHO, 2011), it may be also to consider a more restrictive operationalization to take into account the sustainability of the policy, especially in the long term.

In addition, our results show that increasing age, poorer health, and sight impairments are factors associated with a greater number of activities requiring help and higher probability of being dependent at baseline, in keeping with previous studies (Rodríguez-Sampayo et al., 2011; Sousa et al., 2009, 2010). These results are consistent either when we consider the Katz-dependency or the Comprehensive-dependency measures. Besides, hearing difficulties also emerged as a risk factor when we use the Comprehensive-dependency measure, which may be due to the activities included in this measure. In particular, hearing impairments could be affecting Comprehensive-dependency due to “participating in social life” and “communicating” included only in that measure.

Furthermore, in agreement with previous research, the analyses of transitions toward dependency including death show heterogeneity by age, sex, education, health, and physical impairments for older people in Uruguay (At et al., 2015; Kingston et al., 2017; Prina et al., 2020; Sousa et al., 2010). Moreover, we showed that results are sensitive to the dependency measure. For instance, moving from nondependent to dependent state with the comprehensive measure is associated with sex rather than when the harder operationalization of dependency is used (the Katz-dependency measure). For the latter case, only age and the presence of comorbidity are factors associated with higher risk of transition toward a dependency state.

Consistently with previous literature, our results suggest that women age with poorer health than men (At et al., 2015; Kingston et al., 2017; Prina et al., 2020; Sousa et al., 2010). Older women with chronic conditions are more likely to enter into dependency state when considering a comprehensive measure of dependency than men. So, the “male–female health–survival paradox” is combined with a higher risk of dependency for older woman indicating that care policies should take this into account to improve their quality of life (Alexandre et al., 2014; Alvarado et al., 2008; Tareque et al., 2017). It is possible that the role of women in the household combined with their poorer health results in women’s higher risk of transition into a Comprehensive-dependent state (Millán-Calenti et al., 2010).

Finally, our findings suggest that how dependency is measured is also relevant to understand dependency dynamics over time. Older people with chronic conditions have unmet care needs (Abdi et al., 2019), so it is important to develop care models based around the needs of older people. The differences identified in the risk factors of transiting to a dependent state when using different measures of the construct provide new insights about how to design and implement policies that provide care to older dependent population living at home. As mentioned above, in the NCS of Uruguay, the official scale operationalizes dependency in a comprehensive manner, and our results suggest that in the comprehensive operationalization, women with worse health condition are at greater risk of transitioning to dependency. Therefore, not considering sex to determine whether or not a person is a beneficiary of the program could generate gender inequities between the older population, to the detriment of women. In addition, this could be worsened if restrictions are applied to program access due to budgetary issues or long-term sustainability of the policy. This could affect the design of interventions to improve quality of life of dependent older people as our results suggest the need to design sex-specific interventions. Further work could deepen knowledge about the type of care required to improve the design of care policies as our results suggest that the transition to dependency is associated with different characteristics when using a hard operationalization of the construct rather than when a comprehensive operationalization is used.

A limitation of the current work is that the ELPS study only has two waves. Even though attrition is not a major problem, the analysis of transitions into dependency is limited when data have only two observations maximum per individual. Another drawback is that international comparisons may be hampered by the absence of databases with comparable information in nationally representative samples.

Despite full harmonization of databases is challenging, our analysis extends previous reports from other Latin American countries such as those produced by the 10/66 Dementia Group (Prince, 2009), adding evidence about the population of older adults in Uruguay, and thus allowing a better understanding of dependency in the context of the aging process. Future research on dependency in Uruguay should guarantee the harmonization, that may provide new insights regarding international comparisons, the heterogeneity in the evolution of dependency as individuals age, and the associated need for help.

Supplemental Material

sj-pdf-1-jah-10.1177_08982643211017726 – Supplemental Material for Dependency Change with Aging and Associated Factors in Uruguay: A Cohort Study

Supplemental Material, sj-pdf-1-jah-10.1177_08982643211017726 for Dependency Change with Aging and Associated Factors in Uruguay: A Cohort Study by Alejandra Marroig, Maximiliano Machado and Graciela Muniz-Terrera in Journal of Aging and Health

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Alejandra Marroig

Supplementary Material

Supplemental material for this article is available online.

References

Abdi

Spann

Borilovic

de Witte

Hawley

. (2019). Understanding the care and support needs of older people: a scoping review and categorisation using the WHO international classification of functioning, disability and health framework (ICF), BMC geriatrics, 19, 195. doi: 10.1186/s12877-019-1189-9

Albarrán Lozano

Alonso González

Bolancé Losillas

. (2009). Comparación de los baremos español, francés y alemán para medir la dependencia de las personas con discapacidad y sus prestaciones. Revista Española de Salud Pública, 83(3), 379-392. doi: 10.1590/S1135-57272009000300004

Alexandre

T. d. S.

Corona

L. P.

Nunes

D. P.

Santos

J. L. F.

Duarte

Y. A. D. O.

Lebrão

M. L

. (2014). Disability in instrumental activities of daily living among older adults: Gender differences. Revista de Saúde Pública, 48, 379-389. doi: 10.1590/S0034-8910.2014048004754

Alvarado

B. E.

Zunzunegui

M.-V.

Béland

Bamvita

J.-M

. (2008). Life course social and health conditions linked to frailty in Latin American older men and women. The Journals of Gerontology: Series A, 63(12), 1399-1406. doi: 10.1093/gerona/63.12.1399

Bryce

Prina

Acosta

Ferri

C. P.

Guerra

Huang

Rodriguez

J. J. L.

Salas

Sosa

A. L.

Williams

J. D.

Dewey

M. E.

Acosta

Liu

Beard

Prince

. (2015). Frailty and the prediction of dependence and mortality in low- and middle-income countries: A 10/66 population-based cohort study. BMC Medicine, 13, 138. doi: 10.1186/s12916-015-0378-4

Austad

S. N.

. (2006). Why women live longer than men: Sex differences in longevity. Gender Medicine, 3(2), 79-92. doi: 10.1016/s1550-8579(06)80198-1

Bora

J. K.

Saikia

. (2015). Gender differentials in self-rated health and self-reported disability among adults in India. Plos One, 10, e0141953. doi: 10.1371/journal.pone.0141953

BPS . (2015). Encuesta Longitudinal de Protección Social. Presentación y primeros resultados tabulados 2012-2013. https://www.elps.org.uy/elps/file/1086/1/principales-resultados-junio-2015.pdf. https://www.elps.org.uy/

Cameron

K. A.

Song

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. doi: 10.1089/jwh.2009.1701

10.

Carmona-Torres

J. M.

Rodríguez-Borrego

M. A.

Laredo-Aguilera

J. A.

López-Soto

P. J.

Santacruz-Salas

Cobo-Cuenca

A. I

. (2019). Disability for basic and instrumental activities of daily living in older individuals. Plos One, 14(7), e0220157. doi: 10.1371/journal.pone.0220157

11.

Edjolo

Proust-Lima

Delva

Dartigues

J.-F.

Pérès

. (2016). Natural history of dependency in the elderly: A 24-year population-based study using a longitudinal item response theory model. American Journal of Epidemiology, 183(4), 277-285. doi: 10.1093/aje/kwv223

12.

Galiani

Ibarrarán

Caruso Bloeck

. (2017). Long-term care in Latin America and the Caribbean? Inter-american development bank. doi: 10.18235/0000813

13.

Graciani

Banegas

J. R.

López-García

Rodríguez-Artalejo

. (2004). Prevalence of disability and associated social and health-related factors among the elderly in Spain: A population-based study. Maturitas, 48(4), 381-392. DOI: 10.1016/j.maturitas.2003.10.004.

14.

Harwood

R. H.

Sayer

A. A.

Hirschfeld

. (2004). Current and future worldwide prevalence of dependency, its relationship to total population, and dependency ratios. Bulletin of the World Health Organization. 82, 251-258.

15.

Katz

Ford

A. B.

Moskowitz

R. W.

Jackson

B. A.

Jaffe

M. W

. (1963). Studies of Illness in the Aged. JAMA, 185(12), 914-919. doi: 10.1001/jama.1963.03060120024016

16.

Kingston

Wohland

Wittenberg

Robinson

Brayne

Matthews

F. E.

Jagger

, Cognitive Function and Ageing Studies Collaboration. (2017). Is late-life dependency increasing or not? A comparison of the cognitive function and ageing studies (CFAS). Lancet, 390(10103), 1676-1684. doi: 10.1016/S0140-6736(17)31575-1

17.

Lawton

M. P.

Brody

E. M

. (1969). Assessment of older people: Self-maintaining and instrumental activities of daily living. The Gerontologist, 9(3), 179-186. doi: 10.1093/geront/9.3_Part_1.179

18.

Matus-Lopez

Cid Pedraza

. (2016). New long-term care policies in Latin America: The national system of care in Uruguay. Journal of the American Medical Directors Association, 17(7), 663-665. doi: 10.1016/j.jamda.2016.04.001

19.

Matus-López

. (2015). Pensando en políticas de cuidados de larga duración para América Latina. Salud Colectiva, 11, 485-496. doi: 10.18294/sc.2015.785

20.

Matus-López

Rodríguez-Modroño

. (2014). Presiones de oferta y demanda sobre políticas formales de cuidados en Latinoamérica. Revista Del CLAD Reforma y Democracia, 60, 103-130.

21.

Mayston

Guerra

Huang

Sosa

A. L.

Uwakwe

Acosta

Ezeah

Gallardo

de Oca

V. M.

Wang

Guerchet

Liu

Sanchez

Lloyd-Sherlock

Prince

M. J

. (2014). Exploring the economic and social effects of care dependence in later life: Protocol for the 10/66 research group INDEP study. SpringerPlus, 3(1), 379. doi: 10.1186/2193-1801-3-379

22.

MIDES. (2018a). Construcción de baremos para valoración de dependencia. Ministerio de Desarrollo Social, Secretaría Nacional de Cuidados. https://www.gub.uy/sistema-cuidados/sites/sistema-cuidados/files/documentos/publicaciones/construccion-de-baremos-para-valoracion-de-dependencia.pdf

23.

MIDES . (2018b). Aproximación a la población dependiente mediante encuestas. Ministerio de Desarrollo Social, Secretaría Nacional de Cuidados, Dirección Nacional de Evaluación y Monitoreo. https://www.gub.uy/sistema-cuidados/comunicacion/publicaciones/aproximacion-poblacion-dependiente-mediante-encuestas

24.

Millán-Calenti

J. C.

Tubío

Pita-Fernández

González-Abraldes

Lorenzo

Fernández-Arruty

Maseda

. (2010). Prevalence of functional disability in activities of daily living (ADL), instrumental activities of daily living (IADL) and associated factors, as predictors of morbidity and mortality. Archives of Gerontology and Geriatrics, 50(3), 306-310. doi: 10.1016/j.archger.2009.04.017

25.

Monteverde

Tomas

Acosta

L. D.

Garay

. (2016). Envejecimiento poblacional y magnitud de la dependencia en Argentina y México: Perspectiva comparada con España. Revista Latino-Americana de Población, 10(8), 135-154.

26.

Prina

A. M.

Y.-T.

Kralj

Acosta

Guerra

Huang

Jotheeswaran

A. T.

Jimenez-Velazquez

I. Z.

Liu

Rodriguez

J. J.

Salas

Sosa

A. L.

Prince

. (2020). Dependence- and disability-free life expectancy across eight low- and middle-income countries: A 10/66 study. Journal of Aging and Health, 32(5–6), 401-409. doi: 10.1177/0898264319825767

27.

Prince

M. J

. (2009). The 10/66 dementia research group—10 years on. Indian J Psychiatry, 51(Suppl1): S8-S15.

28.

Rely

Vargas-Chanes

García-Peña

Salinas-Escudero

Gutiérrez-Robledo

L.-M.

Wong

. (2020). Multidimensional dependency subgroups in community-dwelling older adults: A latent class analysis. Salud UIS, 52, 101-109. doi: 10.18273/revsal.v52n2-2020004

29.

Rodríguez-Sampayo

Rodríguez-Míguez

Álvarez-García

. (2011). Distribución territorial de la dependencia en España y Europa. Papeles de economía española, 129, 27-47.

30.

Rofman

Amarante

Apella

. (2016). Demographic change in Uruguay: Economic opportunities and challenges. Directions in Development. World Bank. doi: 10.1596/978-1-4648-0844-9

31.

Sousa

R. M.

Ferri

C. P.

Acosta

Albanese

Guerra

Huang

Jacob

Jotheeswaran

Rodriguez

J. J. L.

Pichardo

G. R.

Rodriguez

M. C.

Salas

Sosa

A. L.

Williams

Zuniga

Prince

. (2009). Contribution of chronic diseases to disability in elderly people in countries with low and middle incomes: A 10/66 dementia research group population-based survey. Lancet, 374(9704), 1821-1830. doi: 10.1016/S0140-6736(09)61829-8

32.

Sousa

R. M.

Ferri

C. P.

Acosta

Guerra

Huang

Jacob

Jotheeswaran

Hernandez

M. A. G.

Liu

Pichardo

G. R.

Rodriguez

J. J. L.

Salas

Sosa

A. L.

Williams

Zuniga

Prince

. (2010). The contribution of chronic diseases to the prevalence of dependence among older people in Latin America, China and India: A 10/66 Dementia Research Group population-based survey. BMC Geriatrics, 10(1), 53. doi: 10.1186/1471-2318-10-53

33.

Tareque

Tiedt

A. D.

Islam

T. M.

Begum

Saito

. (2017). Gender differences in functional disability and self-care among seniors in Bangladesh. BMC Geriatrics, 17(1), 177. doi: 10.1186/s12877-017-0577-2.

34.

United Nations . (2020). World population ageing 2019. Department of economic and social affairs, Population Division.

35.

WHO (2001). International classification of functioning, disability and health: ICF. World Health Organization.

36.

WHO (2011). World report on disability. Retrieved from https://www.who.int/disabilities/world_report/2011/report.pdf?ua=1.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.35 MB