Disability-free life expectancy and life expectancy in good self-rated health in Chile: Gender differences and compression of morbidity between 2009 and 2016

Background Chile has one of the highest life expectancies at 60 years in South America. This study was aimed to determine healthy life expectancies among Chilean older people, according to self-rated health and disability, and to explore gender differences. Methods Data from the National Survey of Health (2009 and 2016) were used to estimate prevalence of less than good self-rated health and disability among people aged 60 years and above. Health expectancies were calculated with the Sullivan method. Results In both years, women expected to live a lower proportion of their life expectancy in good self-rated health (54.5% [95% CI 50.0–58.8] for men and 37.6% [95% CI 34.3–40.8] for women in 2009; 46.1% [95% CI 42.6–49.7] for men and 38.5% [95% CI 35.6–41.4] for women in 2016). Life expectancy in less than good self-rated health increased for men (9.4 years [95% CI 8.4–10.3] in 2009; 11.5 years [95% CI 10.7–12.2]). Women expected to live a lower proportion of their remaining life without disabilities (65.3% [95% CI 61.2–69.4] for men and 44.9% [95% CI 41.9–47.9] for women in 2009; 71.9% [95% CI 68.7–75.0] for men and 61.1% [95% CI 58.5–63.8] for women in 2016). In 2016, disability-free life expectancy increased among women, but they still had a higher life expectancy with mild disability (2.8 years [95% CI 2.3–3.4] for men and 6.0 years [95% CI 5.4–6.7] for women). Conclusions Women expected to spend more years in less than good self-rated health and disabled. There was an expansion of life expectancy in less than good SRH among men and a compression of disability in both sexes. The high proportion of years expected to be lived in less than good self-rated health and gender differences in disability-free life expectancy of older adults should be addressed by public health policies in Chile.

a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 than good self-rated health and gender differences in disability-free life expectancy of older adults should be addressed by public health policies in Chile.

Background
During the last century, a remarkable increase in life expectancy (LE) was observed, which resulted in a higher proportion of people reaching old age [1]. Aging of the population in Latin America has occurred at a higher speed, compared to many countries in Europe and North America, which had more time to adapt to changing social and economic demands [2]. In Chile, LE at birth in 2019 was 80 years [3]. Chile has one of the highest life expectancies at 60 years in South America. In 2017, women at 60 years had a LE of 24.6 years, and men of the same age expected to live 21.0 more years [4]. The global increase in LE has raised the question whether the additional years are lived in good health [5]. The theory of expansion of morbidity posits that an extension of LE has as a consequence the expansion of diseases and disabilities [6]. The compression of morbidity approach assumes that the burden of disease can be prevented and postponed and as LE extends, diseases and disabilities are reducing [7]. Compression or expansion of morbidity can be absolute or relative. Absolute expansion or compression of morbidity are expressed in more years lived with or without diseases or disabilities, respectively [8]. A decrease in the proportion of LE in good health combined with an increase in the number of years in good health is considered a relative expansion of morbidity, and a decrease of the proportion of LE in poor health, in combination with an increase in the number of years in poor health is considered a relative compression of morbidity [8]. Unlike the expansion of morbidity theory, the dynamic-equilibrium hypothesis states that severe morbidity does not increase as LE extends [9].
Health expectancies are measures that combine information on LE and health indicators, in order to estimate the remaining years to be lived in good health from a specific age [10]. Previous research on health expectancies has shown that there is an advantage for women in survival, but that they spend more years in worse health status, compared to men [11][12][13]. Specifically, older women are more likely to spend more years with negative self-rated health, functional impairment in instrumental activities of daily living, mobility limitations, depressive symptoms and cognitive impairment [14][15][16][17][18]. Several dimensions are included in possible explanations for the gender health-survival paradox. Biological differences could result in a higher resilience and better immune response among women [19,20]. Cultural patterns could also have an impact on lifestyles, with a higher tobacco and alcohol consumption [19], more frequent risk-taking behavior and a lower tendency to seek health care and to comply with treatments among men [20].
Health indicators to calculate health expectancies include self-rated health (SRH) and disability [21]. Although SRH is a subjective measure of health status, an important number of studies have shown its association with mortality among the older population [22][23][24][25], along with negative outcomes in terms of morbidity, falls and hospitalization [21]. According to results from the Chilean National Study of Dependency in Older People, 44.1% men and 54.2% women aged 60 years or more rated their health as less than good in 2009 [26]. Poor SRH among Chilean older women was associated with a higher risk of dying [27]. It has also been observed that Chilean older people with less chronic diseases were more likely to report good SRH [28]. Two Brazilian studies have reported LE in good SRH [29,30], but no information is available with respect to the Chilean population. Disability-free life expectancy (DFLE) has been used to monitor health expectancies of the population in the United States [31,32] and Europe [33][34][35] over the past several decades. Several studies on DFLE including Latin American populations exist [36][37][38][39][40][41][42][43][44][45], but only two studies have reported DFLE among Chilean older people, and they have focused on the population living in Santiago [46,47]. Estimations of LE in good SRH and DFLE at national level and comparisons over time are needed.
This study was aimed to estimate LE in good SRH and DFLE among Chilean men and women aged 60 years in 2009 and 2016, and to explore gender differences.

Data
We analyzed data from the National Survey of Health (NSH), a Chilean cross-sectional study with three waves. Data from the two last waves (2009 and 2016) were employed. Stratified cluster sampling was used in both waves, based on the master sample frame of the Chilean National Institute of Statistics and the Population and Housing Census of Chile [48,49]. The samples for the two versions included people aged 15 years or more (n = 5293 in 2009; n = 6233 in 2016) and were nationally representative. Taking into account the complex study designs, weights for both datasets were calculated [48,50].
The protocols of the NSH 2009 and 2016 were approved by the Ethics Committee of the Pontificia Universidad Católica de Chile (Pontifical Catholic University of Chile). Participants signed informed consent to take part in the NSH. The databases were fully anonymized by the Chilean Ministry of Health, before they were made publicly available for research purposes.
Data collection was carried out via face to face household interviews. A detailed set of sociodemographic and health variables were gathered. The question employed to assess SRH was: "In general, would you say your health is?" The possible answers were: Very good, good, fair, poor, very poor. The answers were collapsed into good (very good and good) and less than good (fair, poor and very poor). Disability was determined according to the definition proposed by Albala et al. [51] and Fuentes-Garcia et al. [52], based on information about activities of daily living (ADL), instrumental activities of daily living (IADL) and cognitive status, ascertained by a modified version of the Mini-Mental State Examination (MMSE) and the Pfeffer Functional Activities Questionnaire (PFAQ), which have been validated as a screening test for cognitive impairment among the Chilean population [53,54]. In 2009, ADLs included were: bathing, dressing and grooming. IADLs included: housekeeping activities and transportation. In 2016, ADLs were: dressing, eating, toileting and transferring from a bed. IADLs were: housekeeping activities, transportation and shopping. Participants were classified as without disability or with disability (mild or severe), according to the following criteria: -Without disability: no difficulties or difficulties in one IADL. No difficulties in ADL.
-Severe disability: difficulties in one or more ADLs or MMSE<13 and PFAQ>11.

Health expectancies calculation
Abridged life tables for the Chilean population in 2009 and 2016 were used. Life tables were retrieved from the Global Health Observatory Data Repository of the World Health Organization. Weighted sex-specific prevalence of less than good SRH and disability was estimated for each five year age group. The Sullivan method [55] was employed to calculate years of life expected to be lived in good SRH and free of disability, based on LE at specific ages and estimated prevalence of less than good SRH and disability. Based on the proposal by Jagger et al. [56], we followed these steps: 1. Calculation of person years to be lived in good SRH/disability-free, by multiplying the person years lived in a specific age group by the proportion of people with good SRH/disability-free (1-weighted prevalence). 2. Addition of the person years with good SRH/disability-free of the successive age intervals, to calculate the total number of years in good SRH/disability-free. 3. Calculation of LE in good SRH/disability-free, by dividing the total years lived in good SRH/disability-free by the number surviving to each age interval. Standard errors were calculated to obtain 95% confidence intervals. This process considered: 1.Calculation of the variance of the prevalence (product of the prevalence by its complement, divided by the number of participants in an age interval). 2. Sum of the product of the square of the person years lived in an age interval and the variance of the prevalence rate, from a specific age interval to the final age interval. 3. To obtain the variance, the previous result for each age interval was divided by the square of the number of people surviving to the beginning of the age interval. 4. The standard error was the square root of the variance of the health expectancy. Stata 15 was employed for descriptive statistics and prevalence estimation. Life expectancies and health expectancies were calculated with R 3.6.1.

Results
The NSH 2009 included a total of 1390 participants aged 60 years or more, and the NSH 2016 had a sample of 2031 people of the same age (Table 1).
In 2016, a higher proportion of women had no education or had completed less than primary level of school (less than 8 years) and a higher proportion of men had attended at least one year of higher education (13 or more years) ( Table 2). More than four fifths of men and women lived in urban areas. A lower proportion of women had private health insurance in 2016.
As observed in Table 3, prevalence of less than good SRH was higher among women. Mild disability decreased across time among women, but the prevalence was higher compared to men in both years.
In 2016, LE at age 60 was 21.3 years for men and 25.3 years for women, showing an increase of 0.8 years among men and 0.4 years among women. Sex differences in LE in good SRH were observed (Table 4). Men and women expected to live a similar number of years in good SRH, but women expected to live more years with less than good SRH (6 more years in 2009; 4.1 more years in 2016) and had a higher proportion of years to be lived in less than good SRH in 2009 (16.2 percent points) and 2016 (7.6 percent points), compared to men. LE in less than good SRH increased 2.1 years for men across time.

PLOS ONE
Disability-free life expectancy and life expectancy in good self-rated health in Chile As shown in Table 5, women had a higher disabled life expectancy (DLE) (6.4 more years in 2009; 3.8 more years in 2016), compared to men. Fig 1 shows that the proportion of years to be lived with disabilities was higher for women (19.7 more percent points in 2009 and 10.8 more percent points in 2016). In 2016, disability-free life expectancy increased among men and women, but women still had a higher life expectancy with mild disability, compared to men.

Discussion
According to our results, Chilean women aged 60 years expected to live a higher proportion of years in less than good SRH in 2009 and 2016, compared to men. The proportion of DLE was higher among women. In particular, they expected to live a higher proportion of their lives with mild disability, compared to men. There was an absolute expansion of LE in less than good SRH for men, and an absolute compression of disability for both sexes.

PLOS ONE
Disability-free life expectancy and life expectancy in good self-rated health in Chile A recent study by our group analyzed data from the NSH from 2003 to 2016, to estimate life expectancy free of cognitive impairment [57]. That report found no sex difference in the proportion of years to be lived free of cognitive impairment between Chilean men and women aged 60 years. A higher LE among women resulted in more years to be lived free of CI, compared to men. Hence, gender differences in health expectancies among the Chilean older population vary depending on the health indicator used.
Sex differences with respect to LE in less than good SRH have been previously reported. Unlike our results, other studies in England [58] and Brazil [29,30] found a higher LE in good SRH among older women. In England, older women also had a longer LE in less than good SRH, compared to men [59].
Brazilian studies that focused on LE in good SRH [29,30] reported considerably higher LE in good SRH for Brazilian men and women aged 60, compared to our results. LE in good SRH at 60 years was 16.8 years for Brazilian men and 20.8 years for Brazilian women in 2008 [29]. In 2013, LE in good SRH at the same age was 17.3 years for Brazilian men and 20.2 years for Brazilian women [30]. According to Jagger et al. [58], English older people in 2011 had higher LE in good SRH at 65 years (12.6 years for men; 14.3 years for women), compared to Chilean older men and women aged 60 in 2009 and 2016. Chilean men and women expected to live more than half of their LE at 60 years in less than good SRH, which is much higher than what other studies have reported [29,30,58].
Although SRH is a subjective measure, a substantial number of studies have found that a negative SRH is associated with mortality among older adults [22][23][24][25]. Accordingly, a study found that poor SRH increased the risk of mortality among Chilean older women [27]. Some

PLOS ONE
Disability-free life expectancy and life expectancy in good self-rated health in Chile studies have shown that the validity of SRH as a predictor of mortality varies depending on educational attainment [59,60]. Literacy among the population 15 years or older in Chile is 95.7% [61]. The NSH did not measure literacy, but no education can be considered a proxy for illiteracy, since during the first year of schooling reading and writing are taught in Chile. We observed a slightly higher proportion of people with no education in 2009, which persisted in 2016 among women (Table 1). A previous study from 2009 reported a higher level of illiteracy among rural, population aged 60 years or more in Chile, compared to urban population of the same age [26]. The proportion of older people living in rural areas in Chile in 2016 was 15.8% for men and 13.1% for women. The accuracy of SRH as a health status indicator among Chilean rural older population and those groups with lower education should be further examined. Our findings showed a high proportion of years expected to be lived in less than good SRH among Chilean older people, which did not decrease between 2009 and 2016. Therefore, more research is needed to disentangle the factors associated with the high prevalence of less than good SRH among the Chilean older population. Chilean public policies should address the high proportion of LE in less than good SRH observed among older men and women.
Chilean older women expected to live a lower proportion of their lives without disability and a higher proportion with disabilities, compared to men. Similar findings have been reported in other countries by previous studies [38,40,58]. In 2009, LE with mild or severe disability was higher for women. A previous longitudinal study found that up to 2010 women aged 60 years expected to live a higher proportion of their lives with any level of disability, compared to men of the same age in Santiago, Chile [47]. Data from Brazil in 2008 [29] and England in 2011 [58] showed the same sex difference. However, in 2016 only LE with moderate disability was higher for Chilean women.
An analysis of trends in DFLE in Brazil between 2000 and 2010 found an expansion of severe disability [40]. For the period 1991-2011 in England, Jagger et al. [58] also reported an increase of DLE, with an expansion of the proportion of life with mild disability only. According to our results in Chile, between 2009 and 2016, LE with any level of disability remained stable for men and LE with severe disability decreased for women.

PLOS ONE
Disability-free life expectancy and life expectancy in good self-rated health in Chile This is the first study to report LE in good SRH and DFLE among Chilean older people at a national level. To date, there were no analyses of LE in good SRH or trends in DFLE in Chile. It is important to consider some limitations. Only cross-sectional data were available to make comparisons between different points in time. However, this approach is widely used in health expectancies studies and it was used in most of the studies discussed above [29,30,58]. Second, European studies have shown that people with less education, lower socioeconomic position or with poor health have a lower participation rate in surveys [62,63]. If this is the case in the NSH, the health status of the Chilean older population could be overestimated. Thirdly, there were differences in the number of items measuring ADL and IADL in the NSH between 2009 and 2016, with two more items considered the last year. A lower number of questions about these difficulties could have decreased the probability of reporting IADL or ADL limitations in 2009. The observed decrease in disability prevalence could have been underestimated. Lastly, previous research has found social inequalities in health status and disability trajectories in Chile [52,64]. Furthermore, a recent study reported differences of 8.9 years for men and 17.7 years for women in LE at birth, between the most deprived and the least deprived areas of Santiago, the capital city of Chile [65]. The data considered in this study did not allow to calculate life tables by socioeconomic position or years of education. Hence, it was not possible to examine the impact of socioeconomic factors in health expectancies.

Conclusions
This is the first study to analyze LE in good SRH among Chilean older adults. Also, this is the first report that compares DFLE at different points in time in Chile. Longer female LE was accompanied by more years in less than good self-rated health and disability for women. During the period considered, less than good SRH expanded among men. There has been a compression of severe disability among women, but mild disability has remained stable between 2009 and 2016. The high proportion of years expected to be lived in less than good self-rated health and gender differences in disability-free life expectancies of older adults should be addressed by public health policies in Chile.