Our main results show that the impact of diabetes on disability manifests only in combination with stroke. Older adults with diabetes who had experienced a stroke were more likely to have ADL limitations, severe ADL limitations and higher mortality risks than those with neither condition. These respondents were also more likely to develop a disability or remain disabled over a six-year period. While having a stroke alone was a risk factor, the combination with diabetes intensified the risks of having and developing a disability as well as mortality risks. These findings expose the major burden that co-existing health conditions place on ability and quality of life among older adults. As discussed by Otiniano and colleagues, patients recovering from a stroke, especially those with diabetes, should receive additional attention upon discharge to the community, given their greater disability and mortality risks [11
Second, our study confirmed that older women in São Paulo face greater risks than older men of being, becoming and remaining ADL limited. These results contribute to a growing discussion about the greater disability burden and lower quality of life in later years among women in Brazil [25
]. Women constitute the majority of the elderly population in São Paulo, and their higher disability risks reduce their chances of maintaining social and economic autonomy. Our results showed no significant association between receiving ADL assistance and gender. These results contrast with previous findings that women’s needs for personal care assistance went unmet for a greater proportion of time that they lived with a disability compared to men [25
]. This discrepancy is related to the effects of additional control variables included in the current study. Socioeconomic and additional health variables were included in addition to age—the only control variable in the previous study.
These new findings highlight the ways in which women’s lower socioeconomic and health status influence the level of care and assistance they receive in later years. First, higher rates of illiteracy among older women in São Paulo (29% of women are illiterate compared to 19% of men) likely indicate lower socioeconomic status, which limits older women’s access to health care services in general and personal assistance in particular. Older women’s marital status and longer life expectancies can also explain their lower chances of receiving assistance in their later years [25
]. In São Paulo, older women are more likely than older men to be unmarried or widowed (63% and 25% for women and men, respectively), and they are also more likely to have additional chronic conditions (1.23 additional conditions among women compared to 0.96 among men, p
0.001). Thus, a common scenario for an older woman in São Paulo is to find herself living alone with scarce economic and social resources to attend to her needs for care in later life. This is a rather inequitable outcome for Brazilian women who are, as in most countries, the majority of informal and long-term caregivers within their families and communities [33
Our results confirm that the oldest among the old (75+ years), who are more likely to suffer from multiple health conditions, are at a higher risk of having ADL limitations and requiring ADL assistance. Within a six-year period, the oldest Brazilians were also more likely to develop ADL limitations, to remain ADL limited and to die. The implications of these results are particularly serious, given the low probabilities of recovery from disability observed in this study and the rapid increase of the elderly population in Brazil, particularly the oldest of the old.
The persistence of CNC diseases such as stroke and diabetes represents a major economic burden for the Brazilian state and society. Recent reports estimated that the labor force reduction tied to diabetes, heart disease and stroke would represent a loss of over USD $4 billion in economic outputs for the country between 2006 and 2015 [1
]. The annual costs associated with hospitalization of Brazilians with diabetes are estimated at USD $264 million [37
], and an additional USD $327 million are needed to provide acute treatment for incident stroke [38
]. Unpaid assistance and care services provided by families, along with high-cost health services, is yet another dimension of the economic and social burden of CNC diseases in Brazil.
Our results on the impact of education on disability are rather encouraging. Being literate reduced the risk of having and retaining ADL and severe ADL limitations and dying. These findings improve the scholarly understanding of the influence socioeconomic indicators such as education have on the health conditions of populations in less-developed countries. The association between lower socioeconomic conditions, higher mortality and lower healthy life expectancies is well established in developing nations [39
]. In Brazil, existing evidence based on cross-sectional data indicates similar patterns [40
]. Camargos and colleagues found that older adults with five years of schooling or more spent a higher proportion of their lives without disability than those with less education [28
Smoking is a major risk factor associated with CNC diseases such as stroke, and tobacco consumption in Brazil is widespread despite recent declines. Our results showed the positive health outcomes of not smoking, particularly reduced mortality risk. Recent health policy discussions have highlighted improvements in cardiovascular and chronic respiratory disease mortality rates in Brazil, which are partially attributable to declines in tobacco use and greater access to primary care [1
]. Nonetheless, the direction of smoking trends in Brazil remains ambiguous, with some evidence showing steady rates of smoking between 2006–2009 [1
The Brazilian state’s efforts to control tobacco use are part of a broader comprehensive strategy that has prioritized CNC diseases in Brazil’s public health policy, especially since 2006 [1
]. Some notable measures include the provision of health care; programs to increase awareness of healthy diets and promote physical activity; the control of alcohol consumption; and the widespread delivery of drugs to those at high risk of cardiovascular diseases [1
]. The universalization of health care in Brazil is at the core of these important policy developments. Recent studies evaluating the Brazilian public health strategy have emphasized the need to shift the current acute-care-oriented health care model to a chronic-care model to meet the rising burden of chronic conditions resulting from demographic and epidemiological transformations in Brazil [1
]. One premise of this model is the need to build an integrated and inter-sectorial health care system with the goal of providing cost-effective interventions to maximize health outcomes. In general, policy recommendations by Brazilian organizations, as well as international organizations such as the World Health Organization, suggest health education and awareness programs as effective strategies for linking health care and education in the prevention of CNC diseases. As Brazil anticipates a growing number of individuals with chronic conditions, expanding efforts to improve health care access and education levels can multiply the impact on CNC disease prevention and lead to a reduction in economic and social costs.
Self-reported data could potentially be a source of bias. However, methodological studies have shown that self-reported data on functional disability correspond to medical diagnosis [43
]. The use of self-reported health conditions is also problematic. This study is not able to take into account the increased disability and mortality of individuals with undiagnosed conditions; therefore, the estimates presented here are conservative estimates of the real burden faced by this population. However, undiagnosed rates tend to decrease with age, which reduces the bias in this study of older adults [27
]. In addition, the results provided in this study might not be representative of the older population residing in institutions, given the methodological design of the SABE survey that focused in non-institutionalized population in the baseline. Nonetheless, studies have estimated that the institutionalized population in Brazil is relatively small [44
], thereby minimizing the potential for bias in our results. Our analyses did not control for income, which can influence the probability of receiving personal assistance. However, many older women in São Paulo do not have personal income, which is why many studies on Brazil use education as a measure of socioeconomic status [28
]. Finally, our results reflect the experiences of older adults in the largest city in Brazil, which may differ from the general Brazilian population.