Results from this large prospective population-based study indicate that high amounts of daily sitting time increase the risk of MI incidence and all-cause mortality. Compared to sitting less than 6 hours per day, sitting for 10 hours or more per day was associated with a 38% higher risk of MI and a 31% higher risk of all-cause mortality. For CHD incidence, the picture was less clear. High amounts of daily sitting time combined with physical inactivity in leisure time were associated with increased risk of MI, CHD and all-cause mortality. Being physically active in leisure time attenuated the association between sitting time and all-cause mortality.
To our knowledge, this is the first study on the association between daily total sitting time and CHD incidence in a large cohort of both men and women. Our study reinforces previous work by demonstrating that individuals who sit for 10 hours or more daily were at increased risk of all-cause mortality [8, 10–14]. The finding of a significant association between sitting time and MI but not CHD is somewhat ambiguous and in contrast to previous studies showing a positive association between sitting time and cardiovascular mortality [11, 12] and between sedentary behavior and CHD incidence [16, 18–20]. Comparison with these studies is complicated by differences in the measurements. Sitting while watching television or using a computer are different behavioral settings to other reasons for sitting, and thus may show different risk profiles. Most previous studies of CHD incidence measured sitting time as television viewing time or screen time. Only two studies have examined total sitting time in relation to CHD incidence and these only included women [17, 18]. In a large prospective cohort study of 73,743 postmenopausal women, Manson et al. found that those who were sitting for 16 or more hours per day had a 68% higher risk for incident CVD compared with less than 4 hours of daily sitting . However, in a recent study of 1654 mid-aged Australian women, Herber-Gast et al. found no association between sitting time and cardiovascular disease incidence . Variations in categorization of sitting time may explain some of the differences in the findings as sitting for 16 hours or more is more extreme than a mean of 8.4 hours/day which was considered the highest category in the study by Herber-Gast et al.
This study found that sitting for more than 10 hours per day combined with being physically inactive in leisure time increased the risk of MI and CHD incidence and all-cause mortality. For all-cause mortality, the significant interaction between sitting time and leisure-time physical activity indicated that physical activity in leisure time counterbalanced the risk associated with a higher amount of total daily sitting time. This is in contrast with previous studies documenting the negative effect of sedentary behavior on cardiovascular morbidity and mortality independent of physical activity level [11–13]. However, for MI and CHD, regular leisure time physical activity seemed to attenuate but not eliminate or counterbalance the health consequences of sitting which partly support the independent effect of sitting.
Although, the difference in the cardio-metabolic biomarkers may have little clinical relevance, the indication of a dose response relationship is relevant in understanding potential physiological mechanisms. In both men and women, we found significant associations between increased sitting time and increased levels of the included cardio-metabolic biomarkers. Our finding is in agreement with previous observational studies showing a positive association between sitting time and cardiovascular risk factors including levels of blood lipids [2, 27–30].
Strengths and limitations
The strengths of this study include the prospective population-based design, the large sample of men and women, and the detailed evaluation of participants at baseline, which was conducted by questionnaire in the complete sample and also by clinical health examination in a subsample. Also, an important strength of the study is the use of nation-wide registers to collect data on hospitalizations both prior to baseline and during follow-up. The registers are considered to be of high quality and nearly complete (less than 1% was lost to follow-up) [31, 32]. Validation studies have found the national registers to be a valid and powerful tool for estimating the population incidence of MI (predictive values of 82-94% and sensitivity of 78%) whereas, the diagnosis of overall CHD are considered less accurate (predictive value 66%) [33, 34]. Presumably, any misclassification of the outcomes was non-differential and will therefore tend to bias the association towards the null because it is not related to the exposure variable (sitting time).
Preclinical or diagnosed disease might influence sitting time and level of physical activity, either biologically or as a stimulant to improve health. If so, this could bias the results. However, individuals with preexisting coronary heart disease were excluded prior to baseline. In addition, also events that occurred during the first 2 years of follow-up and diagnosis of cancers and cardiovascular disease were excluded. Although, the associations were attenuated and no longer significant in these sensitivity analyses, they showed the same tendencies.
Although self-reported sitting time is a reasonable well-validated measure, results should be interpreted given the limitations of self-reporting . However, objective measurement instruments were not feasible in a large-scale population study. An advantage of the present study is that it covered the sum of daily sitting during work and in leisure time in both weekdays and weekends. However, the IPAQ sitting question did not include sitting during transportation which also contribute to total sitting time. Therefore, time spent sitting while travelling in a motor vehicle (as likely sitting time) was added, even though the question does not specifically asks for sitting time during transportation. However, we expect the misclassification of sitting time during transportation to be low as standing in a bus or a train often is of short durations, and most travel time is spent sitting.
Physical activity in leisure time was assessed in four predefined categories and further dichotomized in the analyses of the combined effects, which can have resulted in loss of relevant information. Also, the level of physical activity obtained through a questionnaire could be subject to misclassification and likely to have underestimated the associations. Although not validated, the 4-category question on leisure-time physical activity was used for main analyses as it has shown the ability to discriminate between inactive persons and their more active counterparts with respect to maximal oxygen uptake  and has shown the ability to predict mortality . The present study focused on leisure-time physical activity only in combination with total sitting time. Physical activity in other domains such as occupation, transport and household contributes to total physical activity but, studies have indicated domain-specific variations in the health benefits of physical activity. In contrast to leisure-time physical activity, the association between occupational physical activity and CHD morbidity and mortality is less clear. To support the findings, results were reanalyzed by applying the IPAQ measure of total and leisure-time physical activity. The IPAQ-questionnaire has been found to be valid measure of total physical activity , but as it included a large proportion of missing values (approximately 25%) and no measure of light intensity physical activity, this was used a supplementary measuring instrument. Results were similar although attenuated, which indicate that the applied 4-category question on leisure time physical activity ranks participants according to physical activity similar to the more comprehensive and time consuming IPAQ.
Some factors, such as BMI, hypertension and diabetes, may be regarded as intermediate factors rather than confounding factors and adjusting for these may have conservatively biased our results. Conversely, factors such as triglycerides or total cholesterol were not included as confounding factors as these were regarded as intermediate variables. Confounding from risk factors such as dietary habits, family history of cardiovascular disease and stress were not controlled for which may have influenced results, However, the magnitude of this potential bias is unknown.
Only 14% of the invited individuals participated and thus results may not be representative for the Danish population . To reduce non-response bias, data were weighted for non-response. but given the cohort design of the study, selection should not affect the internal validity. However, relative to other populations, the Danish population is known to be fairly active both in leisure time  and in using active transportation, especially bicycling . Thus, it is possible that the results may not apply to a more inactive population.
The relatively short period of follow-up can be viewed both as a strength and a limitation. On one hand, the number of cases is small and provides less statistical power for the analyses. On the other hand, the short period of follow-up is less likely to be influenced by changes in health behaviors over time which may complicate risk estimation.