Physical activity reduces the risk of numerous diseases, like ischemic heart disease,[1] stroke,[2] diabetes mellitus[3], and cognitive disorders,[4] as well as total mortality [5]. Health-related quality of life (HRQoL) is a global indicator of health resulting from the individual's perception of the impact that diseases exert on different spheres of life (physical, mental and social). Most of the evidence on the relation between leisure-time physical activity (LTPA) and HRQoL has been obtained in cross-sectional studies in middle-age adults [6, 7]. However, little evidence exists in the case of the elderly. This evidence is based on clinical trials of the short-term effect of exercise programs in patients with chronic diseases, who are often institutionalized,[8, 9] and in cross-sectional studies, which have limited capacity to establish causal relations because HRQoL itself may influence the ability to do physical activity [10–16]. To our knowledge, only one study in elderly women has shown a longitudinal association between higher LTPA and better HRQoL;[17] however, it was limited to examining the effect of LTPA on the mental components of HRQoL, and did not include the physical components.

Finally, there is growing consensus that sedentary behavior has a harmful effect on health independently of the total volume of physical activity performed [18]. Specifically, it is known that a greater number of hours spent sitting down or watching television is associated with a higher risk of cardiovascular disease,[19, 20] diabetes,[21] and general mortality [22]. However, we know of no study that has yet examined the influence of the number of sitting hours on HRQoL in older adults. Accordingly, we conducted a longitudinal study of the association of LTPA and number of sitting hours with HRQoL in older adults. This association is important because of several reasons: the elderly are the population segment with the largest increase over the last decades, their HRQoL worsens with age, and LTPA and LTSB are modifiable behaviors.

The most frequent activities in 2003 were walking (87.7% of subjects), gardening (28.7%), and swimming (19.1%). With regard to total volume of LTPA, the median was 21.3 MET-hr/week, and the 25^th and 75^th percentiles were 4.6 and 26.3 MET-hr/week, respectively. Men performed more physical activity than women (median MET-hr/week was 23.8 in men and 13.8 in women; p < 0.001).

Study participants may suffer from long-term disability or chronic diseases that limit physical function, so that they have both low LPTA and poor SF-36 outcomes. Thus, for these people, poor SF-36 may have led to lower LPTA. To rule out this mechanism of reverse causation, we rerun the analyses with the 285 individuals who, at baseline, were free of coronary disease, stroke, cancer at any site, chronic obstructive pulmonary disease, diabetes mellitus and arterial hypertension). Results went in the same direction to those in table 2, but the statistical power was necessarily much lower (data not shown). Also to exclude reverse causation, we conducted analyses using as reference the individuals in the second quartile of LTPA (i.e. those showing some minimal capacity to do physical activities at baseline). Results were generally consistent with those in table 2 though, as expected, the association tended to be weaker (data not shown).

Our results show that greater LTPA and less LTSB are independently associated with better long-term HRQoL in older adults. This association affects both the physical and mental scales of HRQoL. Specifically, doing more LTPA showed a positive linear trend with physical functioning, physical role, bodily pain, vitality, social functioning, emotional role and mental health. Moreover, meeting the ACSM/AHA recommendations on physical activity was associated with better physical functioning, social functioning and emotional role. Finally, the number of sitting hours showed a gradual and inverse relation with the score on the scales of physical functioning, physical role, bodily pain, vitality, social functioning and mental health.

The literature is heterogeneous regarding physical activity and HRQoL in older adults, partly because of the different study designs used and different ways of measuring physical activity and HRQoL. In a randomized clinical trial with deconditioned institutionalized elderly persons, Dechamps et al showed that a tai chi program and a cognition-action program lasting 6 months slowed down the decline in HRQoL, evaluated as limitations in the activities of daily living at 12 months follow-up [8]. In contrast, a non-randomized intervention study with a small sample size in elderly nursing home residents with minor disabilities found no differences in HRQoL associated with a 12-month program of supervised exercise [9]. In the latter study, HRQoL was assessed by limitations in instrumental activities of daily living, tests of cognitive functioning, health tests, and a scale measuring the fear of falling.

Some cross-sectional studies in older adults have shown a positive association between LTPA and HRQoL. Lobo et al showed that physical activity measured for 7 days with an accelerometer in older institutionalized adults was associated with better physical functioning, physical role, vitality and less bodily pain on the SF-36 [11]. Likewise, in a study of 112 healthy volunteers aged 60 and over, more physical activity was associated with better score on various scales of the SF-36 [13]. In addition, in the Behavioral Risk Factor Surveillance System survey in the United States, the proportion of persons age 65 and over who reported 14 or more unhealthy days (physical or mental) was lower in those who did moderate or vigorous physical activity than in those who did not [14]. Finally, in a small Japanese study, in persons aged 65-85 years with physical activity measured by accelerometer for one year, HRQoL was poorer among those in the lowest quartiles of both step count and duration of activity > 3 METs [16].

In one longitudinal study in older adults, women who maintained or increased their physical activity improved their scores on various mental health scales of the SF-36 with respect to those who were always sedentary [17]. Our study extends knowledge on the longitudinal relation between LTPA and HRQoL because it includes standard measures of all the dimensions of HRQoL. Furthermore, our study is unique in showing an inverse association between number of sitting hours and HRQoL in the elderly, which is independent of the total volume of LTPA.

Because the total amount of leisure time is finite, a reduction in LTSB requires the increase in the time devoted to several types of physical activities, which can be heterogeneous in terms of intensity (light, moderate, vigorous) but also in terms of the relative expense and sacrifice needed to engage in them. As pointed out by Mekary et al,[32] the advantage of the isotemporal model is that it allows comparing substitution of a fixed time of an activity type for the same time engaged in sedentary behavior; it, thus, helps to answer the relevant public health question of how to spend the leisure time for optimal HRQoL. Our results suggest that many dimensions of HRQoL can be effectively improved by the isotemporal replacement of LTSB by physical activity of just light intensity (e.g., walking, dancing, gardening); moreover, in our population, replacement of LTSB by activities of moderate-to-vigorous intensity led to better physical functioning but not to clinically relevant improvements in other dimensions of HRQoL.

The enhanced HRQoL associated with increased LTPA may be a consequence of several mechanisms, such as reduced cardiovascular risk factors, prevention and management of chronic diseases,[37, 38] lower risk of falls,[39] prevention of functional limitation[40] and lower risk of mental disorders like depression and anxiety[41] and cognitive decline [42]. Another possible mechanism is satisfaction arising from self-efficacy for physical activity. McAuley et al suggest that physical activity directly influences self-efficacy and, through it, acts on HRQoL, especially on the components of mental health [15, 43, 44]. Our results are consistent with these studies, since improvements were observed on the mental scales from the first quartile of LTPA, whereas for the scales of physical functioning and physical role clinically relevant effects are not seen until the third quartile (table 2). Longer sitting time has been associated with overweight and obesity, independently of physical activity [45]. Obesity, diabetes and hypertension are possible mediating mechanisms that may explain the association between sedentary behavior and HRQoL [46]. Furthermore, most sedentary activities, such as watching television, reading or sitting at the computer, decrease communication with the family, reduce the social network[47] and increase the risk of depression, anxiety and stress,[48] which would explain the poorer quality of life associated with sedentary behavior.

Our study has several strengths and limitations. An important strength is that the we used validated instrument to measure LTPA and LTSB [28, 49] as well as HRQoL [31, 50]. Another strength is that our analyses were adjusted simultaneously for LTPA and LTSB, and for a considerable number of confounding factors.

Losses to follow-up were the main study limitation. Measuring HRQoL in older adults over the long term is complex, since the SF-36 does not allow for replies by third parties, and it is inevitable that a certain proportion of elderly persons experience mental and/or physical decline that impede completing the questionnaire over time. Losses to follow-up could affect the representativeness of our cohort; nonetheless, the association between LTPA and HRQoL had already been observed in a cross-sectional analysis of this cohort in 2001 [23]. Another study limitation is that the question on sitting time does not allow to know the specific activity done while seated (watching television, reading, driving, etc.); thus this study cannot make specific recommendations to reduce sedentary behavior. Lastly, studies on the effect of physical activity on health are potentially affected by a reverse causation bias, because individuals with better health status and less disability are more able to do physical activity. However, the analyses conducted with participants free of disease at baseline, and those using the second quartile of physical activity as reference, produced results in the same direction to those in the whole study sample. Although it does not entirely rule out reverse causation, it suggests that its contribution to our results is likely to be small.

Greater LTPA and less LTSB were independently associated with better long-term HRQoL in older adults. These findings have practical importance because they illustrate that LTPA can reduce the age-associated decline in HRQoL. Moreover, most of the LTPA in our cohort consisted of walking, which is the safest activity in older adults,[51] and there is evidence of the short-term efficacy of brief counseling and the use of pedometers to increase the time spent walking [52]. Finally, this study suggests that HRQoL might be improved by replacing time spent seated by time performing light physical activity, and that doing it for just one hour/day may have clinically relevant benefits.