Health Related Quality of Life among the Elderly: a Population-based Study Using Sf-36 Survey

As life expectancy continues to rise, one of the greatest challenges of public health is to improve the quality of later years of life. The aim of this present study was to analyze the quality of life profile of the elderly across different demographic and socioeconomic factors. A cross-sectional study was carried out in two stages, involving 1,958 individuals aged 60 years or more. Health related quality of life (HRQOL) was assessed using the SF-36 questionnaire. The lowest scores were found among measures for vitality, mental health and general health and the highest among factors including social functioning and role limitations due to emotional and physical factors. HRQOL was found to be worse among women, in individuals at advanced ages, those who practiced evangelical religions and those with lower levels of income and schooling. The greatest differences in SF-36 scores between the categories were observed in functional capacity and physical factors. The results suggest that healthcare programs for the elderly should take into account the multi-dimensionality of health and social inequalities so that interventions can target the most affected elements of HRQOL as well as the most vulnerable subgroups of the population.

Health related quality of life among the elderly: a population-based study using SF-36 survey Qualidade de vida relacionada à saúde em idosos, avaliada com o uso do SF-36 em estudo de base populacional

Introduction
The progressive rise in life expectancy contributes to an increase in the prevalence of chronic illnesses in the elderly population 1 .Despite suffering from chronic conditions, elderly individuals can have a good level of health and remain capable of administering basic survival activities, their social lives and finances 2 .Therefore, one of the greatest public health challenges is to increase the number of years of a healthy and quality life.
The concept of quality of life encompasses satisfaction and wellbeing, containing subjective and multi-dimensional characteristics 3,4 .Quality of life can be addressed as general quality of life or health-related quality of life (HRQOL).The former is a broad-based term that includes the sense of wellbeing and happiness regardless of illnesses and dysfunctions.In HRQOL, a multidimensional approach is employed that takes into account physical, mental and social aspects that are more clearly related to symptoms, disabilities and limitations caused by disease 5,6 .Self-assessed health and health-related quality of life instruments generate a set of important health indicators for individuals and populations and are significant predictors of mortality, especially in the elderly.In a broad-based literature review, Idler & Benyamini 7 detected a greater risk of death in individuals who assessed their health status as regular or bad compared Cad.Saúde Pública, Rio de Janeiro, 25 (10):2159-2167, out, 2009 to those with a more favorable self-assessment of health.
However, HRQOL measurements are not generated by the Brazilian national health information system 8 .Subjective health indicators can be obtained through health surveys that counterbalance the lack of traditional information systems and are valuable when it comes to the formulation and assessment of public health policies.
One of the most widely used instruments to assess health-related quality of life is the SF-36 (Medical Outcomes Study 36-item Short-Form Health Survey) that is drawn from the Medical Outcomes Study (MOS) questionnaire published in English in 1990.The literature on this instrument is documented by the International Quality of Life Assessment Project (IQOLA) 9 .The SF-36 contains 36 items combined in eight scales, which can also be grouped into two components: physical and mental.SF-36 has been translated and validated in several languages and cultures.There are surveys applying the SF-36 in more than 40 countries 9 .The instrument allows the measurement of various health dimensions and can assess the impact of disease as well as the benefits of treatment.It is also a good predictor of mortality.In a cohort study with elderly individuals, Tsay et al. 10 found a greater risk of mortality among those who scored low on the SF-36 measures.
In Brazil, the instrument was translated and validated by Ciconelli et al. 11 in a study involving individuals with rheumatoid arthritis.It was considered suitable for administration under the socioeconomic and cultural conditions of the Brazilian population.
Studies developed in other countries demonstrate that some SF-36 domains, such as vitality and general health, are more compromised than others, such as mental health and social functioning 12,13 .A number of studies have assessed the extent to which demographic and socioeconomic conditions are associated with HRQOL using the SF-36 13,14 and have found significant differences between subpopulations, which points out the need for a differentiated approach to public health planning in order to improve equity.
However, there have been no previously published Brazilian population-based studies using the SF-36 for comparisons with international data.
The aim of the present study was to provide a profile of SF-36 scales and analyze the influence of demographic and socioeconomic factors on health-related quality of life in an elderly Brazilian population.A two-stage stratified cluster sample was obtained.Census tracts were grouped into three strata according to the percentage of heads of household with college education: less than 5%, 5% to 25% and over 25%.Ten census tracts were selected from each stratum totaling 120 sectors in the four areas.After the fieldwork to update maps, the selection of households was performed.In order to obtain satisfactory subpopulation sample sizes the following gender and age domains were defined: < 1 year, 1 to 11 years, 12 to 19 yearold-men, 12 to 19 year-old-women, 20 to 59 yearold-men, 20 to 59 year-old-women, men aged 60 and over and women aged 60 and over.For each domain in each study area a minimum sample size of 200 was estimated, based on a prevalence of 0.5, an error of 0.07, an alpha error of 0.05 and a design effect of 2. Considering a possible loss of 20%, 250 individuals were selected for each domain 16 .For the present study, only two domains were included -those with people aged 60 years or more.Data were collected by trained interviewers directly to the selected individual using a pre-codified questionnaire.The questionnaire was mostly made up of closed questions organized into 19 theme blocks.

Material and methods
The variables analyzed in this study were obtained from three thematic sets of questions: health related quality of life, constituted using the SF-36 and sets of socioeconomic and demographic characteristics.
The dependent variables were the scores of the SF-36 scales: physical functioning, role limitations due to physical health problems (referred to here as role-physical), bodily pain, general health (general health perceptions), vitality, social functioning, role limitations due to emotional health problems (referred to here as role-emotional) and mental health.
The scores were attributed to each item according to the proposed methodology 11 .The total scores from each of the eight domains were then converted to a scale ranging from 0 to 100, with higher scores representing better health 11 .
The independent variables of this study were the demographic and socioeconomic characteristics: gender; age (60 to 69, 70 to 79 and 80 years or more); skin color/ethnicity (white and black/ mixed); marital status (with and without spouse); religion (Catholic, Evangelic, and others or no religion); monthly per capita family income (less than 1 minimum wage; 1 to 4 times the minimum wage; and more than 4 times the minimum salary); and schooling (0 to 3; 4 to 8; and 9 or more years of study).
Estimates of means, standard error and confidence intervals were performed for each of the SF-36 scales.Differences in means according to demographic and socioeconomic variables were tested using simple linear regression analysis.Multiple regression models were used to control the effect of gender, age and per capita monthly income and schooling.All data analysis took into account the sample design considering the weights and the intra-cluster correlations.Analyses were performed with

Results
A total of 1,958 elderly individuals were interviewed: 929 men and 1,029 women, with a mean age of 69.6 years.Most of the interviewees were in the 60 to 69 age group (55.8%), lived with a spouse (58.9%), were Catholic (75.5%) and referred to themselves as being white (80.2%).About 75% had a per capita monthly income less than four times the minimum salary and 42.6% had less than four years of schooling (Table 1).
Women obtained lower scores than men in all domains except for role-physical (Table 3).The greatest difference between genders was found in the physical functioning scale, with a difference of 9.2 points between mean scores.
Unadjusted analysis of the difference in scores according to skin color/ethnicity revealed that white individuals obtained significantly higher mean scores in the general health scale.However, this difference failed to remain significant in multiple linear regression analysis (Table 3).Regarding the mean scores by marital status, differences between elderly individuals with and without spouse were no longer significant after adjusting for gender, age, schooling and per capita income (Table 3).
Considering the age groups (Table 4), mean scores diminish progressively with the advance in age, with statistically significant differences in all the scales except for mental health and bodily pain, comparing the age groups "80 or more" with those aged 60 to 69.Individuals of the Catholic faith obtained better scores than those from Evangelical religion for role-physical and vitality indicators, even after adjusting for gender, age, per capita monthly income and schooling (Table 4).
Scores were higher in the strata with higher income.The greatest differences in mean scores between the lowest and highest income strata were found in the following scales: role-physical (14.1), social functioning (10.4) and physical functioning (9.7).Differences between income strata were non-significant in the role-emotional, mental health and bodily pain scales (Table 5).
Comparing years of education, better healthrelated quality of life was observed among those with more years of schooling.Differences were significant in all scales, except role-emotional and social functioning, between the segment with 9 or more years of schooling and that with less than 4 years.The highest differences were found in bodily pain (10.6), physical functioning (10.0 points) and role-physical (8.3).Differences were non-significant between the stratum with 4 to 8 years of schooling and that with less than 4 years in the following scales: general health, vitality, social functioning, role-emotional and mental health (Table 5).

Discussion
The SF-36 is an instrument that enables the investigation of health-related quality of life, addressing multiple dimensions: role-physical, physical functioning, bodily pain, general health, vitality, role-emotional, social functioning and mental health 11,17 .Based on the reviewed literature, this is the first Brazilian paper that analyzes health-related quality of life in elderly using the SF-36 in a population-based study.Among the eight dimensions assessed by the SF-36, the population studied in the present survey obtained the worst scores in the scales of: vitality, mental health and general health.Other studies showed similar results.Lam et al. 18 in a study carried out in China in individuals aged 14 years or older, also found the lowest scores in Dif ( 2)-( 1) p ( 2)-( 1) these three domains.Leplège et al. 19 , in research developed in France, found the worst mean scores in the general health, role-emotional and vitality domains.In a sample of 3,802 individuals aged 15 years or more, Wyss et al. 13 observed in Tanzania, in individuals aged 65 and over, the lowest scores in general health and vitality.
Analyzing health-related quality of life according to gender, this study showed that women were in a worse situation than men in all SF-36 scales except role-physical.Similar results were found in other studies.In a sample of 1,688 individuals aged 18 years or older in China, Li et al. 14 found lower scores among women in the following dimensions: physical functioning, bodily pain, general health and vitality.Wyss et al. 13 also observed that women obtained lower scores than men in all SF-36 scales.In Brazil, studies published on self-rated health using a general question found a worse self-assessment of health among women 20,21,22,23 .The fact that women exhibit a worse self-assessed level of health may be attributed to the greater perception and knowledge that they have regarding diseases and symptoms 1 .The role as a family health caregiver makes women dedicate more attention to the signs of diseases.Studies generally demonstrate a greater prevalence of reported illness and use of healthcare services among women in comparison to men 1,24 .
The influence of skin color/ethnicity on the health situation has been studied by some authors 23,25,26 .In relation to this variable, the present study found no significant associations.The difference encountered in unadjusted analysis can be attributed to socioeconomic inequality and not to the condition of skin color per se.Dachs 25 found no significant differences in selfassessed health according to skin color when the analyses were adjusted for schooling and income.A study on the prevalence of 12 chronic diseases in a Brazilian population (PNAD-2003), showed slight differences between black and white individuals, with a lower prevalence, for seven of the 12 diseases, among individuals with mixed skin color in comparison to those with white skin, after adjusting for age, gender and schooling 1 .
Considering marital status, elderly individuals with spouses reported a better health status than those with no spouse in two dimensions.However, the differences were no longer significant in the multiple linear regression, as elderly individuals without spouses are generally older and female.Thus no influence from marital status on HRQOL was detected in the present study.This finding differs from the study of Wyss et al. 13 , in which single individuals obtained higher scores than widow/widowers, even after adjusting for age and gender.
The age factor has considerable influence in HRQOL.As expected, older individuals have poorer health status than younger obes.No significant differences by age were detected in the bodily pain and mental health scales, revealing that these two dimensions are not greatly compromised by the advance in age.Populationbased studies carried out in other countries using the SF-36 also found lower scores with an increase in age, especially in the physical component, along with a weak or lack of a decline in the mental component, similar to the results of this Brazilian study 12,13,14 .The influence of age on self-assessed health is also documented by the Brazilian literature 20,21,22,23 .
According to religion, elderly individuals pertaining to Evangelical faiths obtained lower scores than those of the Catholic religion in rolephysical and vitality domains, even after adjusting for age, gender, per capita income and schooling.One of the limitations of cross-sectional studies, however, is that they do not allow the identification of cause and effect.It is possible that individuals in a poorer state of health migrate from one religion to another in search of greater spiritual support.A number of authors have studied the relationship between religious affiliation and health events, finding no association with preventive practices for women's cancers 27 or the prevalence of hypertension 28 .In a systematic literature review, Moreira-Almeida et al. 29 found that greater religious involvement is associated with better mental health.Two studies derived from the Multi-Center Intervention Study on Suicide Behavior (SUPRE-MISS) in Brazil 30,31 found associations between religious affiliation and suicidal behavior as well as between religious affiliation and the prevalence of alcohol abuse.The former observed a greater proportion of suicidal ideation among those of the Spiritualist doctrine when compared to those of the Evangelical, whereas the latter found a greater prevalence of alcohol abuse among Spiritualists and Catholics when compared to those of the Evangelical faiths.
In the present study, there was a positive association between socioeconomic levels and HRQOL.The worst scores in all the SF-36 scales were found in the lowest strata of income and schooling.Studies from other countries using the SF-36 also found that individuals from lower socioeconomic strata obtained lower average scores in all eight dimensions 19,26 .Other studies carried out in Brazil have found differences in self-rated health status according to the level of schooling 20,21,32 .Lima-Costa et al. 32 found that even slight differences in family income exert an influence in self-rated health status among the elderly.
The present study detected significant social inequality in HRQOL of the elderly, especially with regard to physical functioning and rolephysical, which were more compromised in relation to the analyzed variables.Health-related quality of life were shown to be worse among: elderly women, individuals with more advanced ages, those with lower incomes, with lower levels of schooling and those who practice evangelical religions in comparison to the catholic faith.According to bibliographic review this is the first paper providing a Brazilian elderly profile of SF-36 scores by demographic and social factors.These data can be used for future comparison and to monitor Brazilian elderly HRQOL.
The rapid demographic changes occurring in the country, with a growing number of elderly individuals and those with chronic illnesses, stressed the need to assess and to monitor different health dimensions in order to guide specific interventions 33 .Measures of HRQOL are especially required from the perspective of promoting active ageing that foresees the inclusion of the elderly in social contexts, with autonomy and independence in their activities, as well as actively contributing in the community 34 .When working with healthcare programs targeting the elderly, it is also necessary to take into account significant social inequalities and to provide conditions to protect the more vulnerable segments of this population.
Stata 8.0 (Stata Corp., College Station, USA) application software.The ISA-SP project was approved by the Ethics Committees of the School of Public Health at the University of São Paulo (USP), the School of Medical Science at the State University of Campinas (UNICAMP) and the School of Medicine at the State University of São Paulo-Botucatu (UNESP).All subjects signed a consent form and the confidentiality of data was assured.The present study was approved by the Research Ethics Committee of the School of Medical Science (UNICAMP) under protocol number 369/2000.
This is a cross-sectional population-based study, developed with data obtained from the Multi-Center Health Survey in the State of São Paulo (ISA-SP) carried out in 2001 and 2002 in four areas of the State of São Paulo, Brazil 15

Table 1
Sample characteristics according to demographic and socioeconomic variables.Multi-CenterHealth Survey in the State of São Paulo (ISA-SP),2001-2002.

Table 3
Mean scores, mean differences and confi dence intervals (95%) of SF-36 scales according to gender, skin color and conjugal situation.Multi-Center Health Survey in the State of São Paulo (ISA-SP),2001-2002.
* Differences adjusted by gender, age, per capita income and schooling using multiple linear regression model.

Table 4
Mean scores, mean differences and confi dence intervals (95%) of SF-36 scales according to age and religion.Multi-Center Health Survey in the State of São Paulo (ISA-SP),2001-2002.
* Differences adjusted by gender, age, per capita income and schooling using multiple linear regression model.

Table 5
Mean scores, mean differences and confi dence intervals (95%) of SF-36 scales according to per capita monthly income and schooling.Multi-Center Health Survey in the State of São Paulo (ISA-SP),2001-2002. 1) * Differences adjusted by gender, age, per capita income and schooling using multiple linear regression model.