Study design and sample

The survey “The Elderly of Rio Grande do Sul”, supported by the State Council on Aging [20] focused on adults aged 60 years old and over who lived in private households in the state of Rio Grande do Sul, Brazil. Briefly, this cross-sectional study of noninstitutionalized persons was based on a multistage stratified random sample. The first step was to stratify the 333 municipalities in the state into five categories according to basic economic activity and number of inhabitants. The proportion of each category in each homogeneous zone was calculated, and the number of subjects in each stratum needed to obtain a representative proportion of elderly community residents was determined. Second, the municipalities were randomly selected proportionally in each stratum. The third step was to obtain a random sample of urban census areas for each municipality as supplied by the Brazil Geography and Statistics Institute (Instituto Brasileiro de Geografia e Estatística). Fourth, to get a sample of private households from each of these census areas, a block was randomly selected, and every eighth house was systematically visited in person by the interviewer. If a household had more than one age-eligible person, the respondent was randomly selected. Houses with no eligible person were replaced by the next neighbor.

Trained lay interviewers carried out face-to-face computer-assisted structured interviews on 7040 individuals in eight of the nine regions into which the state was divided (information from one region had to be discarded because of data entry problems). The response rate was 99%, yielding a sample of 6,963 subjects, 6,924 (99.4%) of whom provided information on physical inactivity. All subjects gave oral consent. Recruitment and consent procedures were approved by the ethics committee of the Federal University of São Paulo which specifically approved this study. Details of the survey methodology have been published previously [21].

Measures

Physical inactivity. The specific question that determined physical inactivity was: “In the last three months did you engage in any kind of regular physical activity?” All participants were asked to name the type of physical activity (for example walking, swimming, cycling, workout, etc.), and to indicate how often they practiced the chosen activities, providing less room for ambiguity. Absence of participation indicated physical inactivity. As there was no statistically significant difference in gender or age between those reporting ‘No’ and ‘Don’t know’ both were noted as physically inactive (coded 1), physical activity once or more a week was coded 0 (i.e., not physically inactive).

Demographic Variables. Interviewers asked participants basic questions about their sociodemographic background, including age, gender, education (<4 years, ≥4 years; the cut-point reflects the number of years of basic education in Brazil for people of this age), income (<$US200/month [low income] vs. ≥$US200/month [higher income]), race/ethnicity (White, Afro-Brazilian, other).

Social ties and health behaviors. Social ties and health behaviors included marital status (married, never married, no longer married [divorced, separated, widowed]); report of any current participation in social activities (participation in any formal association on a list that included cultural, sports, recreational, religious, charitable/aid providing, community, union, political, and ‘other’ associations), was coded ‘Yes’ (0), no participation (report of ‘No’ or ‘Don’t know’) was coded ‘No’ (1); and presence of children (No [1], Yes [0]). A response of ‘Yes’ to living alone, use tobacco, and inappropriate alcohol use (the latter defined as a positive answer to any of the following questions: 1. Has your family, your friends, your physician or your priest ever commented or suggested that you were drinking too much? 2. Have you ever tried to stop drinking but been unable to do so? 3. Have you ever had trouble at work or school because of alcohol, such as drinking or missing work? 4. Have you ever been involved in fights or arrested for being drunk? 5. Has it ever seemed to you that you were drinking too much?), was coded 1, and ‘No’ was coded 0.

Physical health. Self-rated health was assessed by asking participants “How would you rate your health?” Response was on a 5-point scale, dichotomized as impaired (fair, bad, very bad) vs. non-impaired (good, very good).

Dependence in activities of daily living (ADL) was assessed by response to each of the following five items: Do you need help (a) performing household activities (cleaning, maintenance, meal preparation), (b) taking medications, (c) with personal hygiene (bathing, combing hair, dressing, cutting nails), (d) feeding yourself, and (e) with mobility (sitting down, getting up, lying down, walking, going up stairs)? Response categories of “don’t know” and “no response” were rare and were recoded conservatively to “no” (need no help), and coded as 0; response of “yes” (need help) was coded as 1. Responses were summed (possible range 0–5), and the summed score dichotomized as yes (1) (≥1 activity for which help was needed) vs. no (0) (no help needed for any activity).

Enquiry into health conditions for which treatment had been sought in the previous six months included hypertension, heart problems, stroke, varicosities, diabetes, arthritis, back problems, osteoporosis, bronchitis, pneumonia, urinary infection, renal problems, dermatologic problems, headache (in the previous week), gastro-intestinal problems and cancer.

Sensory impairments. Vision was assessed by asking participants “How would you rate your visual ability?” Response was on a 6-point scale, dichotomized as impaired (blind, very bad, bad) vs. non-impaired (fair, good, very good).

Hearing was assessed by asking participants “How would you rate your hearing ability?” Response was on a 5-point scale, dichotomized as impaired (deaf, very hard to hear, hard to hear) vs. non-impaired (minimal or no difficulty hearing).

Depression. The presence of depression within the previous 30 days was determined by the six-item Short Psychiatric Evaluation Schedule (Short-SPES), validated for the older Brazilian population [22]. Each interviewer-administered question required a yes/no answer from the respondent. The total score (potential scoring range 0–6) was categorized as depression (score ≥4) vs non-depression (scores 0–3). A study of inter-rater reliability showed complete scoring agreement between examiners. Performance on the Short-SPES is not affected by sex, age, marital status, income, education, or minority status [22].

Service use. Use of health care was defined as self-report of any outpatient visit in the previous 6 months (yes/no), and any hospitalization within the previous 12 months (yes/no). If hospitalized, we asked whether there had been one or more than one hospitalization.

Statistical analysis

Descriptive statistics (frequencies, χ² tests) were used to characterize the sample and to test the bivariate relationships between physical inactivity and the variables of interest. The distribution of some variables was skewed, hence we decided to keep them dichotomized in order to get a balanced distribution. Adjusted logistic regression was used to study possible correlates of physical inactivity using three models: Model 1 included sociodemographic characteristics, Model 2 added social ties and health behaviors, and Model 3 further controlled for self-rated health, ADL, health conditions and depression. Separate logistic regression analyses were run to examine the association between physical inactivity and outpatient visits, hospitalization, and number of hospitalizations, first with an unadjusted model, and then with a model that adjusted for age, gender, ethnicity and socioeconomic status. These variables were selected because previous work suggested that they are associated both with the likelihood of physical inactivity as well as with health service use. Multivariate significance tests were performed using Wald χ² tests. Statistical significance was evaluated using 2-tailed design-based tests. To control for Type I error inflation we adopted a more conservative p-value for significance: P<0.01. Analyses were carried out using SPSS version 20.0.

Adjusted association between physical inactivity and sociodemographic variables (Table 2, model 1)

Physical inactivity was significantly related to all sociodemographic variables. In particular, women, older persons, those with less education, lower income, and of Afro-Brazilian ethnicity had increased odds of being inactive than their reference groups (respectively by 48%, 37%-70% depending on age category, 67%, 37%, and 49%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Logistic regression.

https://doi.org/10.1371/journal.pone.0117060.t002

Adjusted association between physical inactivity and sociodemographic variables, social ties and health behaviors (Table 2, model 2)

All sociodemographic associations were slightly modified but remained significant. Absence of involvement in social activities was the only new variable significantly associated with physical inactivity (OR 1.77).

Adjusted association between physical inactivity and sociodemographic variables, social ties, health behaviors and health conditions (Table 2, model 3)

Odds ratios remained significant for all variables examined previously. With respect to measures of health and specific health conditions, physical inactivity was significantly associated only with impaired self-rated health (OR 1.37), dependence in ADL (OR 1.24), visual impairment (OR 1.28), and depression (OR 1.28). Association with other health conditions did not meet criteria for significance. Alternate analyses (not shown), including only self-rated health and ADL (i.e., excluding health conditions), and including specific health conditions but not self-rated health and ADL, modified little the associations already found.

Physical inactivity and use of health services (Table 3)

In unadjusted analyses, physical inactivity was associated with increased odds of hospitalization (but not of multiple hospitalizations) in the previous 12 months (OR 1.47). Statistical significance for reduced odds of an outpatient visit in the previous 6 months was just missed. After adjusting for demographic characteristics, the odds of an outpatient visit in the previous 6 months was reduced significantly by 19%. Association with any hospitalization in the previous 12 months remained positive, with increased odds of 41%, but there was no association with multiple hospitalizations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Unadjusted and adjusted odds ratios of the association between physical inactivity and outpatient visits, hospitalization and number of hospitalizations.

https://doi.org/10.1371/journal.pone.0117060.t003

Health services utilization

Our findings on use of health services were somewhat unexpected. In an adjusted analysis, physically inactive older participants were 19% less likely than those who were physically active to report any outpatient visit in the past 6 months. In contrast, they were 41% more likely to use be hospitalized in the previous 12 months (but not more likely to have multiple hospitalizations). This may reflect personal neglect by the physically inactive, or greater obstacles accessing regular outpatient care, for instance, common outpatient medications are paid for out-of-pocket (which may be difficult when income is low), but medications when hospitalized are covered [40]. Outpatient care neglect may result in an increased odds of hospitalization. Canadian findings indicate that physical inactivity was strongly associated with statistically significant increases in hospitalization, length of stay and healthcare visits, together with higher concomitant costs to the healthcare system [41].

Physical inactivity involves multiple, complex behavior processes [19–23], such that over time individuals may adopt a lethargic attitude to health care, reinforced by obstacles to health service use [40,41]. Despite investigatory efforts, little is known as to whether these factors affect receipt of health services among those with the disorders reported here, and under the circumstances under which they live.

Strengths and Limitations

The strength and novelty of the current study is that we could take into consideration a wide range of confounders that could influence associations with physical inactivity. These included sociodemographic characteristics, health behaviors and social ties, alternative measures of health, a wide array of health conditions, and also depression. Other strengths of our study include the focus on older people (for whom information is needed), the representativeness of the sample, the relatively large number of participants, and the high response rate.

This study has certain limitations. Owing to the cross-sectional design, causality cannot be determined. In addition, we sampled only community residents, excluding persons in hospitals or nursing homes, where prevalence of physical inactivity is likely to be higher, however, the number in these settings is likely to be small. Due to socioeconomic and cultural disparities across geographic areas in Brazil the findings may not be generalizable beyond the area under investigation. Standard physical activity questionnaires, such as the International (or Global) Physical Activity Questionnaire [42], were not yet available at the time. The information on physical inactivity is based on self-report to a single item, and as is common in epidemiological surveys, did not use objective methods. It reflects absence of planned physical activity, but does not indicate intensity of physical activity, or physical activity in other areas. Since planned exercise is considered desirable, our approach may have resulted in over-reporting of physical activity because of social desirability, and so underestimate the prevalence of physical inactivity [43]. Our analyses may therefore be unduly conservative. Health conditions were self-reported, but self report has been found to provide valid assessments of health conditions [44]. While the diagnosis of depression was not based on the “gold standard” of a clinical interview, the Short-SPES is a valid and reliable tool for identification of depression in Brazilian community resident elderly [22]. The data were not available to discuss the major cultural, social, physical, and economic barriers that need to be addressed if behavior change is to be promoted in favor of increased physical activity or to discourage inactivity. While the data were gathered long ago, and the emphasis, through different medias mostly towards adolescents and adults, to increase physical activity have spread out, more recent information have shown similar figures than ours suggesting that the numbers found in this study can represent the current situation [15].

In summary, this study has shown that physical inactivity among older adults in the southern part of Brazil is highly prevalent, particularly among Afro-Brazilians. It is associated with low income and education; lack of social interaction; and poor self-rated health, ADL, vision, and depression; although not with other health conditions. Self-care may be neglected, resulting in hospitalization. Special attempts need to be made to target this population in order to encourage physical activity. Additional resources may be needed to provide the means whereby physical activity will be more accessible to and feasible for older persons with these kinds of problems. Recent studies and reviews indicate that several physical activity interventions have been identified as promising for future research and implementation [45]. Efforts to address active leisure will require taking safety, economic, personal, psychological, cultural, and social barriers into account [46]. Emphasizing physical activity among the elderly requires creativity in developing acceptable, relevant, culturally meaningful activities (e.g., dancing, typical local sports, tai chi, etc), and studies to ascertain outcomes and cost-effectiveness. Findings from the present study may help the public, health care providers, and those who determine public policy better address the issues of active lifestyle among the older population with the aim of improving health and well-being, and reducing health care costs.