Dose-response relation of self-reported and accelerometer-measured physical activity to perceived health in middle age—the Northern Finland Birth Cohort 1966 Study

Perceived health (PH; also known as self-rated health and self-perceived health) is a widely used, inexpensive health measure. Most often a single-item question asking individuals to rate their current health status on a 4- or 5-point scale from poor to excellent is used. PH is a subjective health measure that includes physical, mental, emotional, and functional aspects of health [1]. PH has been shown to be associated with mortality, morbidity, and objective health status [2, 3] and to predict mortality after controlling for objective health status [2]. In some cases, it has been an even stronger predictor of survival than medical records [4].

Lifestyle, social, and work-related factors have been identified as determinants of PH [3, 5, 6], and it has been acknowledged that people with similar health conditions may provide different answers to a PH question [7]. Thus, PH includes components beyond objective health status, as Maddox and Douglass noted decades ago: “These ratings clearly measure something more—and something less—than objective medical rating” [8].

It has been widely established that sufficient and regular physical activity (PA) enhances health and decreases the risk of morbidity and mortality. The lack of physical activity, or physical inactivity, increases the risk of many non-communicable diseases and shortens the lifespan over three years [9, 10], and the global cost of physical inactivity was evaluated as more than 53 billion international dollars in 2013 [11].

Recently, objective wearable measuring devices such as accelerometers and pedometers have become popular in the PA research field [12]. They provide accurate information about PA due to the lack of common errors (recall, and social desirability bias) inherent in self-reports of PA [13]. However, both measuring conditions may cause reactivity, i.e. changes in habitual PA behavior during the measurement period.

Self-reported leisure-time PA (LTPA) [14, 15] and moderate to vigorous PA (MVPA) measured with an accelerometer [16–18] have both separately been reported to be associated with PH in many age groups. However, previous research on the determinants of PH has often relied on self-reported PA, and questions have been raised about the potential conceptual overlap between self-reported PA and health due to the complex nature of human perception of health and well-being, including physical, social, and mental constructs [16]. To our knowledge, few studies have compared objective and subjective PA measures in relation to subjective health measures. Anokye et al. reported a stronger relationship between objective PA and health-related quality of life (HRQoL) compared to subjectively measured PA [19].

Sedentary time increases in adulthood with age, and those participating in LTPA or occupational PA in early middle age are more likely to engage in LTPA as they age [20, 21]. Additionally, a connection between higher age and poorer PH has been recognized [22]. Therefore, reaching a sufficient level of PA in middle age, and maintaining this level while aging, is a key factor for healthy aging. Thus, the aim of the study was to determine and compare the relationship between self-reported and accelerometer-measured PA and PH in midlife, in the context of a wide population-based birth cohort study from Northern Finland. To our knowledge, this is the first study exploring the relationship between both measured and self-reported PA and PH in a single study setup. We hypothesized that both PA variables have a positive association with PH and that objectively measured PA might have a stronger association compared to self-reported PA due to its ubiquitous nature including all measured activities around the clock.

The aim of this population-based study was to investigate and compare the relationship of self-reported and accelerometer-measured physical activity to PH in midlife. The main finding was that both self-reported LTPA and accelerometer-measured MVPA were significantly and positively associated with PH. To our knowledge, this study for the first time reports the dose-response association between both self-reported and accelerometer-measured PA and PH in a large cross-sectional setup.

In the present study, the amount of accelerometer measured MVPA and self-reported LTPA was higher among those who perceived their health as at least good compared to those with fair, poor, or very poor health. When adjusted for socioeconomic, demographic, and health factors and ST, both MVPA and LTPA were positively associated with PH. The relationship was consistently stronger between LTPA and PH compared to measured MVPA, contrary to our hypothesis. Odds of good PH was over fourfold higher for those in the highest LTPA quartile compared to the lowest, as opposed to MVPA, where the odds of good PH were modestly higher in the highest quartile, around 1.7-fold, compared to the lowest quartile.

A similar framework was used in an earlier study involving objective and subjective measures of PA and health-related quality of life among middle aged participants [19]. However, the association between accelerometer-measured PA and HRQoL was stronger compared to questionnaire-based measurements of PA; the opposite was observed in our study, where LTPA had a stronger association with PH than objectively measured MVPA. This discrepancy might be due to different outcome measures (PH vs. HRQoL). Moreover, the possibility that PH and self-reported PA might have a conceptual overlap could explain the stronger relationship between them in this study. It has been suggested that especially LTPA would have stronger positive association with PH compared to other PA domains (like household or work related PA) [15, 31, 32]. LTPA is autonomous and often aims for recreation, enjoyment and social interaction and thus has potential to increase mood and psychological health in addition to physiological health benefits [31, 32]. In the study, the agreement between objectively MVPA and self-reported LTPA was low, which indicates that LTPA neglects other domains of PA (e.g. work related, commuting, and household activities) which can have moderate or higher intensity and thus are included in the objectively measured MVPA time.

Our findings are in line with an earlier study indicating a positive connection between PH and objectively measured PA and a lack of connection between PH and ST [16]. Previously, the dose-response relationship between LTPA and PH has also been acknowledged [14]. Other covariates for good PH found in this study, including higher education, marriage/cohabitation, higher income, and working status have also been widely acknowledged previously (see, e.g., [2, 3, 22]). In this study, gender was not related to PH.

Our results indicate that middle aged participants engaged in an average of 69 min of MVPA per day (2590 MET-min/wk), as measured with the activity monitor, with men participating in significantly more MVPA than women. In previous studies among middle-aged populations, the amount of accelerometer-measured PA has varied considerably. In the United States, MVPA duration was only 35 and 20 min/day in men and women respectively [33]. However, a study in Norway reported almost twice as much daily MVPA as the current study, with 117 min/day [34]. These studies relied on waist-worn monitors, and it has been found that wrist-worn monitors provide higher estimates of activity compared to waist-worn monitors due to the wear location [35]. Furthermore, using objective PA measurements may have effected on the PA behavior of the participants during the two-week measurement period in this study.

Self-reported time spent in light and brisk activity during leisure time in this study was around 3 h/wk., which is less than that reported in previous US study involving the same age group [36]. In this study, women reported significantly more LTPA than men, although based on the objective measurement, men engaged in more MVPA than women. The disparity between measured and self-estimated PA can be explained partly by the questionnaire addressing only the amount and duration of leisure time brisk and light activity. It has been established that 500–1000 MET-min/wk. of PA is enough to bring major health benefits [37]. With this criterion, the study population was sufficiently active based on both the questionnaire and the accelerometer.

Time spent sitting was on average 7.5 h/day based on self-report. Men reported an hour more ST than women per day. Most of the ST was reported to take place at work. Earlier studies involving accelerometer measurements show STs of around the same magnitude in middle-aged people: around 7.5 h/day in the United States and 8.5 h/day in Australia [21, 38].

This study has some limitations. Causality between PA and PH cannot be concluded due to the cross-sectional study design. The lack of accelerometer-measured ST is another limitation, as accelerometer measured and self-reported STs differ. Objective measurement is known to measure a greater amount of ST compared to self-reports [34]. Thus, the actual sitting time in this study might be underestimated. It has been established that measuring ST with wrist-worn monitor has limitations due to its difficulty in posture recognition and with methods used in this study it wasn’t possible to differentiate reliably sitting from other low intensity still activities, e.g. standing [39]. The amount of reported sitting time did not differ between good and poor PH, indicating that the amount of sitting is more weakly associated with perception of health compared to PA, which was clearly positively connected with PH.

The methodological limitations of objective and subjective measures of PA when investigating the relationships between PH and well-being should be acknowledged. While accelerometer measurement is a reliable and continuous measuring method, it does not capture all types of PA precisely, underestimating, for example, cycling and carrying loads. In contrast, PA questionnaires tend to overestimate the true PA level but can provide essential information about the PA type and context [36]. The strengths of the study include a representative population sample and high compliance of activity monitor wear.

In this population-based study on middle-aged people, the amount of both accelerometer measured moderate to vigorous physical activity and self-reported LTPA were positively associated with PH regardless of daily time spent sitting. A clear dose-response relationship was observed, with both physical activity measures on PH and higher PA connected to better health. This study encourages use of both objective and subjective measurement of PA and the results can be utilized in policymaking and planning interventions and programs promoting physical activity at population level.