Commentary: Too much sitting--a public health threat?

As exemplified by two articles1,2 in this issue of the journal, there has been an almost exponential interest in researching the potential detrimental health effects of sedentary behaviour during the past few years. Sedentary behaviour is a distinct behaviour, not synonymous with physical inactivity, and defined as ‘any waking behaviour characterized by an energy expenditure ≤1.5 Metabolic Equivalents while in a sitting or reclining position’.3 According to this definition, being sedentary has two prerequisites: the type of behaviour, i.e. sitting or reclining, combined with low energy expenditure. Although robust secular trend data on sedentary time are unavailable, it is likely that population levels of total sedentary time have increased during the past 50 years, possibly owing to an increase in occupational- and transport-related sedentary time. The vast majority of studies examining the health impact of sedentary time has been observational in design, and most of these addressed the associations between TV viewing and health outcomes.1 This is not surprising, as TV viewing constitutes the dominant leisure time of sedentary behaviour in many developed countries, including Australia, UK and the USA. However, TV viewing is a specific sedentary behaviour and may not be a marker of overall sedentary time. In addition, TV viewing may be associated with a generally unhealthy lifestyle difficult to statistically control for in any observational study. Ford and Caspersen1 reviewed prospective observational cohort studies that reported on the associations between total sitting time and TV viewing with fatal and non-fatal cardiovascular disease (CVD). Total sitting time was consistently associated with increased risk for fatal and non-fatal CVD diseases in all four studies included in the review. Similar observations were also recently reported by van der Ploeg et al.4 who demonstrated an increased risk of all-cause mortality with increased total sitting time, an association that appeared unaffected by sex, age, body mass index (BMI) and physical activity levels. The majority of studies (four of six) in the review by Ford and Caspersen1 also showed that higher amounts of time spent viewing TV were associated with increased risk of CVD mortality. The pooled risk ratio was increased by 17% for each additional 2 h in front of the TV, similar to another recent meta-analysis on this topic.5 In a second contribution in this issue, Stamatakis et al.2 report on the cross-sectional associations between self-reported and objectively measured sedentary time with established CVD risk factors in up to 5948 men and women using data from the Health Survey for England. Self-reported sitting and TV-viewing time was significantly associated with a number of CVD risk factors, also after controlling for self-reported and, in a subsample (N = 1150), objectively measured time spent in moderate and vigorous intensity physical activity (MVPA). However, following adjustments for BMI or waist circumference, most of the observed associations were attenuated towards the null. For objectively measured sedentary time, all associations with CVD risk factors (except for total cholesterol) were attenuated towards the null after adjustment for time spent in MVPA. In additional sensitivity analyses, time spent in MVPA was inversely associated with all CVD risk factors.2 Unfortunately, Stamatakis et al.2 did not report whether this association was independent of time spent sedentary, an observation recently observed in youth.6 The increasing number of well-conducted large-scale cohort studies, which have observed positive associations between higher amounts of sedentary time and greater risk for a wide variety in health outcomes, suggests these findings are unlikely to be spurious. However, behaviours such as physical activity, time spent sitting and TV viewing are notoriously difficult to assess by self-report, making these measures prone to reporting bias. TV-viewing and total sitting time may be underestimated, and if specific individuals differentially under-report their amount of time spent sedentary, this bias may lead to either underestimation or overestimation of the true association. Meta-analyses of observational studies may provide more robust results but do not acknowledge systematic biases.7 As pointed out by Ford and Caspersen,1 most studies in their review controlled for a number of confounding factors, including smoking status, BMI, age and sex; many also adjusted their analyses for physical activity. Despite this, we cannot exclude the possibility that the observed results could be explained by residual confounding because of poorly measured or unmeasured confounders. Sitting time in general, and TV viewing in particular, is associated with consumption of high-energy dense foods and drinks, which can be consumed in large amounts, almost unconsciously. Assessing dietary habits accurately is even more difficult than the assessment of sedentary behaviours, and few studies have included dietary habits as a potential confounder.1 TV viewing may also be associated with the exposure to advertisements for unhealthy food products and beverages, which indirectly may promote an unhealthy diet. In addition, reverse causation cannot be completely ruled out, as those who are unhealthy at the inception of the study may report higher amounts of sedentary time. However, no major differences in effect size was observed in the two studies that addressed this possibility by excluding participants during the first years of follow-up.1 A confounding or mediating effect of adiposity on the association between sedentary time and CVD risk should also be considered when interpreting results from observational studies. Wijndaele et al.8 observed that waist circumference, a marker of central adiposity, partly mediated the association between TV viewing and CVD mortality. The current study by Stamatakis et al.2 and other studies by this group9 have suggested that adjustment for BMI almost completely attenuated the cross-sectional and prospective associations between TV viewing and CVD risk factors. A mediating role of adiposity is possible given the suggested bidirectional relationship between adiposity and sedentary time.10 Self-reported and objectively measured sedentary time tend to be weakly correlated, which suggests that in practice, the two measures are different constructs. Few studies have so far examined associations between objectively measured sedentary time and CVD risk.2,11 Accelerometry may provide a non-subjective estimate of the total amount of time spent sedentary. However, accelerometry does not assess domain-specific sedentary time and specific types of activity. Methodological progress in terms of activity type recognition, monitor development and data-processing techniques is being rapidly made. Future studies, including these technologies, or reanalysing already collected data, will enhance our understanding about the independent associations between time spent sitting and CVD morbidity and mortality and other health outcomes. Taken together, although most studies suggest that total sitting and TV-viewing times are associated with CVD (and all-cause) mortality, the effect sizes are generally small. Results from studies using objective methods when examining the independent associations between sedentary time and CVD risk factors are contradictory.2,11 Therefore, it is not possible to completely rule out the possibility that the observed associations between sedentary behaviour and CVD risk are due to bias, reverse causation or confounding. Formulating public health guidelines aimed at reducing sedentary time based on the current evidence appears premature. Although it is apparent that sitting less and moving about more are associated with substantial health benefits in an individual, combating CVD and other non-communicable diseases may not be as simple as only advising people to sit less and engage in more physical activity. Still, one could argue that it is unlikely to be harmful to advise individuals to reduce sedentary time, but such advice needs to be accompanied by macro-level societal changes, including marketing legislation, structural changes in the transport system, the built infrastructure and the education system, to have any substantial impact on the public health disease burden. Conflict of interest: None declared.