Time Spent Sitting as an Independent Risk Factor for Cardiovascular Disease

Abstract

Sedentary behavior is highly prevalent despite growing evidence of adverse effects on the cardiovascular and metabolic system that are independent of the level of recreational physical activity (PA). We present results for the association between sitting time and cardiovascular disease (CVD) from selected cohort and cross-sectional studies published in or after the year 2010 according to the domains where sitting time is accumulated during the day. These include TV viewing, occupational sitting, and sitting during transportation as well as overall sitting. The outcomes considered in this review are total CVD, coronary heart disease, and stroke as well as CVD risk factors—namely, hypertension, hypercholesterolemia, and type 2 diabetes and their associated biomarkers. Finally, several current issues with regard to studying the effects of sitting time on CVD are discussed, including how sedentary behavior is assessed, isotemporal substitution modeling, examination of joint associations for sitting and PA, and benefits of breaks in sitting time. Overall, the scientific evidence supports public health recommendations that encourage adults to limit their sedentary time in order to improve their cardiovascular health.

‘Sedentary activities include activities such as watching TV, listening to music, reading and writing, knitting and sewing, playing video or computer games, and riding in a car.’

Introduction

A preponderance of scientific evidence has shown that physical activity (PA) lowers risk of all-cause mortality as well as diseases such as coronary heart disease, diabetes, stroke, and some forms of cancer.^1,2 Despite the clear evidence, only 1 in 5 adults met the 2008 Physical Activity Guidelines for Americans in 2015.³ Physical inactivity is defined as not meeting the PA guidelines of 150 min/wk of moderate- to vigorous-intensity PA (MVPA). Physical inactivity is considered the fourth leading cause of death, and it is estimated that 6% of coronary heart disease, 7% of type 2 diabetes, 10% of breast cancer, and 10% of colon cancer are attributed to physical inactivity.^4,5

In addition to the many adults who do not meet the PA guidelines, there are adults who do meet the guidelines but are otherwise sedentary.⁶ Sedentary behavior is a distinct concept from physical inactivity and is defined as “any waking behavior characterized by an energy expenditure ≤1.5 metabolic equivalents [of task] (METs), while in a sitting, reclining or lying posture. (p. 9)”⁷ This definition from the Sedentary Behavior Research Network takes into account both energy expenditure and posture. Sedentary comes from the Latin word sedere and means to sit. Sedentary activities include activities such as watching TV, listening to music, reading and writing, knitting and sewing, playing video or computer games, and riding in a car.⁸

Measurement of Sedentary Behavior

General assessment techniques for sedentary behaviors and sedentary time include self-report questionnaires, accelerometers, and direct observation.⁹ The latter is rarely used in population studies because of the high cost but can serve as the criterion measure to validate other instruments.¹⁰ Historically, self-report questionnaires were used in large, epidemiological studies, whereas accelerometers were used mainly in smaller randomized controlled trials. Although large studies still commonly assess sedentary time with questionnaires, because of declining costs, accelerometers have been used in population-based cross-sectional studies such as the National Health and Nutrition Examination Survey (NHANES) and the Health Survey for England.¹¹ More recently, longitudinal studies such as the Women’s Health Study, the Women’s Health Initiative, and the REasons for Geographic and Racial Differences in Stroke (REGARDS) Study in the United States and the UK Biobank Study have begun to collect objective activity data as well.^11-14

There are benefits and limitations to both accelerometer assessment of sedentary behavior as well as self-report. Accelerometers are advantageous because they provide an objective assessment of total sedentary time, but domain-specific information is lacking. Self-report questionnaires often include information on domain but do not provide adequate measures of breaks in sitting time or estimates of light-intensity activity.¹⁵ Furthermore, there may be a great deal of measurement error for sedentary behavior when assessed by self-report. In a study by Clark et al,¹⁶ self-reported TV viewing time was modestly correlated with accelerometer-assessed total sitting time (Spearman ρ = 0.22; 95% CI = 0.20 to 0.25). This correlation based on the NHANES data from the 2003-2004 and 2005-2006 cycles was similar for men and women as well as 3 race groups (non-Hispanic white, non-Hispanic black, and Mexican American). Given the strength of the correlation being only fair, this finding emphasizes the importance of including all domains of sitting to fully capture sedentary behaviors.

In a review of reliability and validation studies, self-report measures tended to be more reliable for TV viewing and computer use compared with other sedentary behaviors such as reading, sitting while socializing, and listening to music because these activities occur more consistently and for longer time blocks.¹⁷ Additionally, questionnaires that asked participants to recall time spent sitting in a typical day were found to have higher validity compared with 7-day or 12-month recall. Finally, Clark et al¹⁷ and Bauman et al¹⁸ mention that sedentary behaviors tend to be underreported potentially because of social desirability bias. Based on NHANES and Swedish data, accelerometer-assessed sitting time was up to 20% higher compared with self-reported sitting time.¹⁸ Bauman et al, however, note that most accelerometers cannot differentiate between standing and sitting when there is no movement, which potentially explains part of the discrepancy. One possible remedy is adding an assessment of posture to the accelerometer counts that will improve classification of sedentary behavior. The activPAL (PAL Technologies, Glasgow, Scotland) has been rated as the most accurate accelerometer for measurements of sedentary time.¹⁹ The activPAL is a small device that is typically worn on the thigh that can better distinguish between sitting/lying and upright activities because of an inclinometer. In a small validation study, Kozey-Keadle et al¹⁹ compared the accuracy and precision of the activPAL and the ActiGraph GT3X triaxial accelerometer to direct observation of 20 overweight and inactive office workers. They found a very high correlation between the activPAL and direct observation for sitting time (r = 0.94), whereas the Actigraph was only moderately correlated with the criterion measure (r = 0.39). More recently, a study by Clark et al²⁰ compared the relative validity of the sitting questionnaire used in the AusDiab studies with the activPAL. They found that self-reported overall sitting time was moderately correlated with activPAL-assessed sitting time in a sample of 700 Australian adults (r = 0.46; 95% CI = 0.40 to 0.52). However, the authors also point out that even though correlations were lower for the context-specific sitting time, self-reported measures work well to rank participants according to their sitting time but not to accurately estimate sitting time.

Given all these issues, it has been recommended that population studies assess sedentary time using a combination of a self-report instrument to obtain domain-specific information and accelerometers to measure total sedentary time and patterns of sitting time throughout the day.²¹

Physiological Mechanisms Linking Sedentary Time and Cardiovascular Disease

Extended sitting time and low levels of PA have independent physiological effects.²² Hamilton and colleagues^23,24 suggest that the lack of muscle contractility, evident in sitting, induces biological consequences. Unfortunately, these results have not been replicated in humans since the work was conducted 13 years prior. In more recent studies, prolonged sitting has been associated with increased total cholesterol, triglycerides, and waist circumference as well as decreased glucose uptake.^25,26 Additionally, previous research has shown that repeated bouts of prolonged sitting result in low shear rates, leading to endothelial dysfunction, which has been linked to vascular mortality.²⁷

Methods

In this review article, we focused on sedentary behaviors in the adult general population. Similar to a previous review by Owen et al,⁶ our results are presented according to different settings where sitting typically occurs: TV watching, occupational sitting, and transportation sitting as well as total overall sitting. Outcomes included in this review are cardiovascular disease (CVD) as well as stroke, coronary heart disease (CHD), and heart failure separately. Additionally, we examined intermediate end points that are known risk factors for CVD: diabetes, hypertension, and hypercholesterolemia. To identify articles, we searched PubMed and Google Scholar for studies containing keywords related to the exposures and outcomes described above. Additionally, we reviewed reference sections of the identified articles and other recent reviews. We included studies that were published in or after the year 2010 and adjusted for PA in their statistical analysis. Preference for inclusion was given to meta-analyses and articles based on prospective cohort studies, followed by cross-sectional studies.

TV Watching

Trends

TV watching is the most common leisure-time sedentary activity. According to the American Time Use Survey 2014, of the 5.1 hours of leisure and sports time in an average day, 2 hours and 49 minutes were spent watching TV, compared with 19 minutes for reading and only 18 minutes for sports, exercise, and recreational activities.²⁸ Time spent watching TV was even higher among adults 75 years and older, who spent 4.5 hours per day watching TV.²⁹ Over the past 15 years, TV viewing has increased slightly from 2.58 h/d in 2003 to 2.8 h/d in 2015.³⁰ TV watching is not only problematic because of its sedentary nature, but also because of its association with increased caloric intake, for example, through energy-dense snacks.^31-34 In early studies, TV watching was used as an indicator of overall sedentary behavior.³⁵

Associations Between TV Watching and Cardiovascular Risk Factors

The harmful association between TV watching and diabetes is well established. In 2 large US prospective cohort studies, Hu and colleagues found a detrimental association between TV watching and the risk of type 2 diabetes in 68 497 women³⁶ and 37 918 men.³⁷ In multivariable-adjusted models, the risk ratio for diabetes comparing the highest quintile of TV watching (more than 40 h/wk) with the lowest quintile (0-1 h/wk) was 1.77 (95% CI = 1.24 to 2.52) in women³⁶ and 2.87 (95% CI = 1.46 to 5.65) in men.³⁷ Additionally, 2 meta-analyses^38,39 have reported positive associations between TV watching and type 2 diabetes, with risk ratios ranging from 1.16 to 1.37 per 2 hours of TV viewing per day³⁸ and 1.22 to 4.0 comparing the highest with the lowest categories.³⁹ The pooled risk ratios for type 2 diabetes in the 2 meta-analyses were 1.20 per 2 hours of TV viewing (95% CI = 1.14 to 1.27)³⁸ and 2.12 (95% CI = 1.61 to 2.78) comparing the highest TV viewing category with the lowest.³⁹ Whereas Grontved and Hu³⁸ included 4 prospective studies, published between 2001 and 2010, with a total of 175 938 individuals, Wilmot et al³⁹ also incorporated findings from 5 cross-sectional studies in addition to the 5 prospective cohort studies published between 2003 and 2012. Interestingly, in the Wilmot et al meta-analysis, the pooled association based on the prospective studies was attenuated compared with the pooled risk ratio of the cross-sectional studies (pooled RR = 1.93, 95% CI = 1.40 to 2.84 for prospective studies, and pooled RR = 2.36, 95% CI = 1.30 to 4.09 for cross-sectional studies). More recently, the increased risk for type 2 diabetes resulting from TV watching has been observed in other populations. Among women with a history of gestational diabetes in the Nurses’ Health Study II, Bao et al⁴⁰ found a 77% (95% CI = 28% to 145%) higher risk for type 2 diabetes in women who viewed TV more than 20 h/wk compared with less than 5 h/wk.⁴⁰ In the prospective EPIC-Potsdam study, Ford et al⁴¹ estimated a 73% (95% CI = 24% to 141%) higher risk for type 2 diabetes comparing 4 or more hours of TV viewing with less than 1 hour among middle-aged German men and women.⁴

Unlike the strong evidence for the association between TV watching and diabetes based on several prospective cohort studies, support for an association between TV viewing and hypertension is limited to cross-sectional studies. In a cross-sectional study among 5527 adults 16 to 99 years old in the Scottish Health Survey, those reporting 3 or more hours per day of TV viewing or screen time were at a 27% higher risk of developing hypertension compared with individuals reporting less than 3 h/d (odds ratio [OR] = 1.27; 95% CI = 1.13 to 1.42).⁴² In a cross-sectional study of 7445 British men and women born in 1958, Pinto Pereira et al²⁵ reported an OR of 1.11 (95% CI = 1.01 to 1.23) for hypertension per category increase in TV viewing in women only. It should be noted that this association was fully attenuated after adjustment for diet and body mass index (BMI), suggesting that BMI may be a confounder or mediator of the association between TV watching and hypertension.

Additionally, 3 studies (2 cross-sectional^25,43 and 1 cohort study⁴⁴) have reported on associations between TV viewing time and cardiovascular biomarkers. In a cross-sectional analysis among 4864 adults in the Australian AusDiab cohort, Thorp et al⁴³ found higher systolic blood pressure (SBP; β = 0.92; 95% CI = 0.36 to 1.45) and diastolic blood pressure (DBP; β = 0.59; 95% CI = 0.28 to 0.89) and lower high-density lipoprotein (HDL) cholesterol (β = −0.01; 95% CI = −0.02 to −0.001) per hour of daily TV viewing in women only. Fasting blood glucose was slightly higher with each additional hour of TV watching in both men (β = 0.01; 95% CI = 0.001 to 0.01) and women (β = 0.004; 95% CI = 0.001 to 0.01). In a prospective analysis within the same cohort, an increase of 10 h/wk of TV watching over a 5-year period was associated with higher DBP, but not SBP, in women only (β = 0.47, 95% CI = 0.02 to 0.92; men: β = 0.37, 95% CI = −0.11 to 0.86).⁴⁴ There was no association between change in TV viewing and other CVD biomarkers over the 5-year period. Finally, using data from a cohort of 7660 individuals born in 1958, Pinto Pereira et al²⁵ examined cross-sectional associations between categories of TV viewing time (0-1, 1-2, 2-3, and ≥3 h/d) and several biomarkers. Per TV viewing time category increase, they found that SBP, DBP, total cholesterol, and low-density lipoprotein (LDL) cholesterol were higher and HDL cholesterol was lower in women. However, these associations were fully mediated (SBP) or attenuated (DBP, total, LDL, HDL cholesterol) after adjustment for diet and BMI. Similarly, in men, SBP and DBP were higher and HDL cholesterol was lower per TV viewing category increase, but fully mediated (SBP, DBP) or attenuated (HDL cholesterol) after adjustment for diet and BMI.

Associations Between TV Watching and CVD

A meta-analysis of 4 prospective cohort studies that were all published in 2010 or 2011 found an increased risk of fatal and nonfatal CVD per 2 hours of daily TV watching, with a linear dose-response relationship (RR = 1.15; 95% CI = 1.06 to 1.23).³⁸ The absolute risk difference per 2 h/d of TV watching was estimated to be 38 cases of fatal CVD per 100 000 individuals per year. Overall, these 4 studies included 34 253 individuals with 1052 incident cases of fatal or nonfatal CVD. Stamatakis et al⁴⁵ examined the associations between screen time, including TV watching, and confirmed CVD events among Scottish Health Survey 2003 respondents, which included adults aged ≥35 years, with follow-up until 2007. They found a strong positive relationship between screen time and CVD events (fatal and nonfatal combined) with a hazard ratio (HR) of 2.25 (95% CI = 1.30 to 3.89) for 4 or more hours per day of screen time compared with less than 2 h/d. Interestingly, in a small subsample, the authors did a mediation analysis and found that C-reactive protein, BMI, and HDL cholesterol were mediators of the screen time-CVD association, explaining approximately 25% of the association collectively.

Wijndaele et al⁴⁶ reported that TV watching was associated with a higher risk for total CVD, CHD, and nonfatal CVD in a prospective cohort study of 12 608 middle-aged British adults in the EPIC-Norfolk study. For each hour of daily TV watching, the adjusted HRs were 1.06 (95% CI = 1.03 to 1.08) for total CVD, 1.06 (95% CI = 1.03 to 1.09) for nonfatal CVD, and 1.08 (95% CI = 1.03 to 1.13) for CHD. Additionally, effect modification by age and metabolic risk was reported for total CVD and nonfatal CVD, but not for CHD. Among older participants and those with higher metabolic risk score (standardized summary score of waist circumference, triglycerides, HDL cholesterol, SBP, DBP, and glycated hemoglobin), the associations between TV watching and CVD were weaker compared with younger participants and those with a lower metabolic risk score.

The findings have been inconsistent for the association between TV watching and cardiovascular mortality. One study using NHANES data from the 1999-2000 and 2001-2002 cycles with updated mortality status until 2006 found no significant associations between TV watching and computer use time and mortality from diseases of the circulatory system (HR = 1.14, 95% CI = 0.51 to 2.54, comparing ≥5 h/d screen time to <1 h/d).⁴⁷ Similarly, in a longitudinal study of 7744 healthy US men aged 20 to 89 years at baseline, time spent watching TV was not associated with CVD mortality.⁴⁸ However, the null association in this study may be explained by the fact that TV watching was assessed only at baseline, and follow-up time was 21 years. These findings are in contrast to those from Matthews et al⁴⁹ and Wijndaele et al,⁵⁰ both in older adults. Matthews et al⁴⁹ reported a strong positive association between TV viewing and CVD mortality in 240 819 US adults aged 50 to 71 years, with a HR of 1.85 (95% CI = 1.56 to 2.20), comparing 7 or more hours per day of TV viewing to less than 1 h/d in the prospective NIH-AARP Diet and Health Study. Joint effects with PA showed that prolonged TV viewing time was associated with higher risk for cardiovascular mortality for both physically active and inactive individuals. TV viewing for more than 7 h/d but meeting or exceeding the PA guidelines was associated with a 2- to 2.5-fold increased risk for CVD mortality. Still, the risk for CVD mortality was much higher in individuals who were inactive and were watching TV for more than 7 h/d (HR = 3.5 for <1 h/wk of MVPA and HR = 4.2 for never/rarely having MVPA, both P < .05, compared with <1 h/d of TV viewing and >7 h/wk of MVPA). In the EPIC Norfolk study among British adults aged 45 to 79 years at baseline, each additional hour of TV viewing time per day was associated with an 8% (HR = 1.08; 95% CI = 1.01 to 1.16) higher risk of CVD mortality.⁵⁰ Likewise, in a prospective study of 8800 Australian adults in the AusDiab cohort, Dunstan et al⁵¹ found a borderline significant association for CVD mortality for the highest TV watching category (≥4 h/d) compared with the lowest (<2 h/d; HR = 1.80; 95% CI = 1.00 to 3.25).⁵¹

Occupational Sedentary Time

Trends

In the past few decades, the number of low-activity occupations has greatly increased, whereas the number of high-activity and physically demanding occupations has declined.^34,52 Jobs in agriculture and manufacturing are disappearing, whereas service-providing jobs, which primarily require light or sedentary activity, are increasing.⁵³ Between the 1960s and 2008, the percentage of sedentary jobs in the US private sector increased from 15% to 25%, whereas the percentage of jobs requiring moderate-intensity PA decreased from 48% to 20%.⁵³ Individuals in sedentary occupations may accumulate large amounts of prolonged sitting time during the workday. For example, a small accelerometer-based study among Australian office workers revealed that 82% of work hours were spent sedentary.⁵⁴ Furthermore, 41% of sedentary time occurred in bouts longer than 30 minutes.⁵⁴

Associations Between Occupational Sitting Time and Cardiovascular Risk Factors

Evidence supporting a significant association between occupational sitting time and cardiovascular risk factors is limited. In a recent cross-sectional study, Garcia et al⁵⁵ examined the association between sedentary work, defined as being seated at work most of the time and walking only short distances, and cardiovascular risk factors among 47 477 Brazilian workers. Compared with nonsedentary work, they found that sedentary work was associated with a 20% higher risk of hypertension, a 41% higher risk of hypercholesterolemia, and a 25% higher risk of type 2 diabetes in men. In women, sedentary work was associated with a 16% higher risk of hypercholesterolemia only. These findings are in contrast to a recent Dutch prospective cohort study by Picavet et al,⁵⁶ which did not find significant associations between occupational sitting and cardiovascular risk factors among 1509 middle-aged men and women. Occupational sitting was assessed 4 times (every 5 years) between 1993 and 2012 using the following categories: mainly sedentary, mainly standing, manual, and involves high physical loads. This information was used to divide participants into 2 groups—stable sitter or stable nonsitter—at work over 15 years. Stable sitting was defined as having a sedentary job for at least 3 out of the 4 assessments. Similarly, stable nonsitters were defined as being in a nonsedentary occupation 3 to 4 times during the assessments. Additionally, they used a more detailed questionnaire in the last data assessment asking for the hours of sitting at work in a typical week. Compared with stable nonsitters, stable sitters did not have a higher risk of hypertension or hypercholesterolemia (longitudinal: HR_HT = 1.08, 95% CI = 0.84 to 1.39, HR_HC = 0.80, 95% CI = 0.60 to 1.07, comparing stable sitters with nonsitters at work). Furthermore, hours of occupational sitting time was also not associated with either risk factor (HR_HT = 0.97, 95% CI = 0.73 to 1.29, HR_HC = 0.92, 95% CI = 0.66 to 1.28, comparing >20 h/wk with less than 4 h/wk). The authors suggest that their null findings could be a result of the healthy worker effect or the beneficial effects of breaks in sitting time counteracting the harmful effects of occupational sedentary behavior on health.

In a cross-sectional study that assessed the separate and joint associations between leisure sitting time and occupational sitting time with cardiovascular biomarkers, Saidj et al⁵⁷ found detrimental associations of prolonged occupational sitting for HDL cholesterol (P = .0042). Examination of the joint associations indicated that, compared with low leisure/low occupational sitting time, high leisure/high occupational sitting was most harmful, followed by high leisure/low occupational sitting and low leisure/high occupational sitting (P values: HDL cholesterol, P < .001; LDL cholesterol, P = .0074; plasma glucose and total cholesterol, not significant). Similarly, in the previously mentioned study by Pinto Pereira et al²⁵ among British adults, sitting at work was associated with lower HDL cholesterol in men only. Compared with the associations for TV watching in the same study, occupational sitting was found to have weaker and fewer significant associations with cardiovascular biomarkers.

Associations Between Occupational Sitting Time and CVD

For the most part, studies examining occupational sitting time and hard CVD outcomes have found no significant associations. In a recent prospective cohort study among almost 12 000 Danish workers, sedentary work (defined as more than 25 h/wk of sitting time at work) was not significantly associated with ischemic heart disease compared with nonsedentary work (HR = 0.95; 95% CI = 0.78 to 1.16).⁵⁸ Additionally, Chau et al⁵⁹ did not find significant associations between occupational sitting and cardiometabolic mortality in the large HUNT3 Norwegian cohort consisting of 50 817 adults aged 20 years and older (P for trend across 4 categories of occupational sitting time =.185).⁵⁹ Similarly, in a pooled analysis of 7 English and Scottish cohorts, Stamatakis et al⁶⁰ found that sitting at work was not associated with higher risk for CVD mortality. In this study, Stamatakis et al recorded 177 CVD-related deaths among 11 168 men and women over a mean follow-up time of almost 13 years. Comparing individuals in standing/walking occupations with those in sitting occupations, the HR for CVD mortality was 1.53 (95% CI = 0.72 to 3.24) in women and 0.98 (95% CI = 0.66 to 1.45) in men.

In a prospective cohort study of 58 208 healthy Finnish men and women, Wang et al⁶¹ found an increased risk of heart failure for individuals in mainly sitting, office occupations. Compared with mainly sitting at work, moderate and high levels of occupational PA were associated with a 15% (HR = 0.85; 95% CI = 0.77 to 0.93) and 13% (HR = 0.87; 95% CI = 0.80 to 0.94) lower risk for heart failure, respectively (P trend <.001). Additionally, when the joint associations for occupational, commuting, and leisure-time PA were examined, moderate to high occupational PA alone (ie, no active commuting and low level of leisure PA) compared with mainly sitting at work was beneficial to lower the risk for heart failure in men (HR = 0.78, with P < .05) but not women. The combination of moderate to high occupational PA with either active commuting or moderate to high leisure-time PA or both reduced the risk for heart failure for men and women even further. The HRs for moderate to high levels in all 3 types of PA, compared with low levels, were 0.69 for men and 0.66 for women (both P values <.05).

Transportation Sedentary Time

Trends

Sitting time during transportation is highly connected with the topics of active commuting, the built-environment, and community safety.^5,6,62 According to the 2009 US National Household Transportation Survey, individuals used their personal car for daily transport 83% of the time, compared with 1.9% for public transportation, 10.4% for walking, and 4.2% for other.⁶³ The average time spent in a vehicle was 56 minutes on a typical day, a decrease compared with 62 min/d in 2001.⁶³ Residents in large urban areas report higher amounts of active commuting—walking 14.2% of the time and using public transportation 4.1% of the time—and less car use (77.3%). Compared with those commuting by car, individuals who commute by foot or public transportation are estimated to walk 19.8 and 5.0 more minutes per day, respectively.⁶⁴ Thus, utilizing other forms of transportation instead of an automobile may lower overall sedentary time.

Associations Between Transportation Sitting Time and Cardiovascular Risk Factors

The evidence between transportation sitting time and cardiovascular risk factors is currently limited to cross-sectional studies. In cross-sectional analyses of the 2007-2008 and 2009-2010 cycles of NHANES, Furie and Desai⁶⁵ found significantly lower odds of hypertension (OR = 0.69; 95% CI = 0.58 to 0.83) and diabetes (OR = 0.69; 95% CI = 0.54 to 0.88) among individuals with high levels of active transportation (walking and biking, ≥150 min/wk) compared with no active transportation (0 min/wk). Interestingly, in stratified analyses, these associations were stronger in individuals who did not meet the PA guidelines but not significant in individuals who met the guidelines. A recent cross-sectional analysis of data from 2800 participants in the 2011-2012 Australian Diabetes, Obesity and Lifestyle Study found that self-reported time spent in cars of more than 1 h/d, compared with 15 min/d or less, was associated with higher fasting blood glucose levels (β = 0.013; 95% CI = 0.000 to 0.026).⁶⁶ There were no significant associations between time spent in cars and other cardiovascular biomarkers (HDL cholesterol and SBP and DBP) in the same study.

Similarly, in a cross-sectional analysis of data from a large group of mainly younger Brazilian workers, Garcia et al⁵⁵ found that sitting during the commute to work was detrimentally associated with some CVD risk factors. Specifically, in women, car or motorcycle use increased the odds of diabetes by more than 40% compared with walking or cycling (OR = 1.48; 95% CI = 1.01 to 2.17), whereas the use of buses increased the odds of hypercholesterolemia and hypertension by more than 20% (OR_HC = 1.27, 95% CI = 1.09 to 1.50; OR_HT = 1.24, 95% CI = 1.08 to 1.42). In men, use of a car or motorcycle led to an increase in odds for hypercholesterolemia by 15% (95% CI = 1.02 to 1.29), whereas the other associations were not significant.

Associations Between Transportation Sitting Time and CVD

Recent evidence on the association between transportation-related sedentary behavior and CVD is sparse. In the Aerobics Center Longitudinal Study, which included 7744 men aged 20 to 89 years, Warren et al⁴⁸ found that more time spent riding in a car was positively associated with CVD mortality. Men who reported 10 or more hours per week riding in a car had a 50% greater risk of CVD mortality compared with men who reported less than 4 h/wk (HR = 1.50; 95% CI = 1.08 to 2.09). Subgroup analyses for effect modification revealed that this association was stronger among inactive (P for trend =.02), overweight/obese individuals (P for trend =.004) and younger (<60 years old, P for trend =.0009) individuals and not significant in physically active, normal-weight and older men. Similarly, in a large prospective Finnish cohort of healthy men and women, active commuting was inversely associated with the risk for heart failure. Compared with 0 min/d of active commuting, the HR for 1 to 29 min/d and ≥30 min/d were 0.88 (95% CI = 0.81 to 0.96) and 0.88 (95% CI = 0.80 to 0.96), respectively.⁶¹ However, these results were no longer significant once adjusted for other types of PA.

Total Sedentary Time

Trends

Matthews et al⁶⁷ were among the first groups to objectively quantify the total time spent in sedentary behaviors. Using the accelerometer data collected during the NHANES 2003-2004 cycle, they found that, on average, individuals spent more than half of their waking hours (7.7 h/d) sedentary. Additionally, they reported that older adolescents and individuals >60 years of age were most sedentary. Differences by race were also reported, with Mexican-Americans being less sedentary compared with white and African-American adults. Of the 24 hours of a day, adults spend on average 7.7 hours sedentary, 8.3 hours sleeping, 7.8 hours in light activities, and only 0.2 hours in moderate to vigorous activities.^67,68

Associations Between Total Sitting Time and Cardiovascular Risk Factors

There is growing evidence from cross-sectional studies that higher daily sitting time is associated with CVD risk factors and cardiovascular biomarkers. In their study among younger industry workers in Brazil, Garcia et al⁵⁵ found that a sedentary lifestyle was associated with higher risk of hypertension and hypercholesterolemia in men but not in women (men: OR_HT = 1.25, 95% CI = 1.13 to 1.39, and OR_HC = 1.44, 95% CI = 1.29 to 1.60; women: OR_HT = 0.87, 95% CI = 0.75 to 1.02, and OR_HC = 0.97, 95% CI = 0.83 to 1.13). Sedentary lifestyle was defined as the combination of TV viewing for more than 2 h/d, reporting a sedentary commute to work, and also being sedentary at work. This finding is consistent with results from the 45 and Up study among 63 048 middle-aged men in Australia. In this cross-sectional analysis, sitting for 8 or more hours per day was associated with an increased risk (OR = 1.06; 95% CI = 1.00 to 1.12) of high blood pressure compared with sitting for less than 4 h/d.⁶⁹

Recent evidence regarding the association between overall sitting time and type 2 diabetes is more consistent for men compared with women. In the cross-sectional study mentioned above by Garcia et al,⁵⁵ individuals reporting the combination of TV viewing of 2 or more hours per day, transportation by car/motorcycle, and predominantly sitting at work were found to have an increased risk of type 2 diabetes in men only (OR = 1.26; 95% CI = 1.02 to 1.56) compared with individuals reporting a nonsedentary lifestyle. Similarly, in a cross-sectional analysis of more than 60 000 middle-aged men in Australia, sitting for 8 or more hours per day was associated with a 21% higher risk of diabetes (95% CI = 1.09 to 1.33) compared with less than 4 h/d of sitting.⁶⁹

With regard to cardiovascular biomarkers, current evidence is strongest for an association between total sitting time and HDL cholesterol. Two cross-sectional studies in NHANES, one utilizing accelerometers⁷⁰ and the other self-report,⁷¹ found that higher sitting time was associated with lower HDL cholesterol. Furthermore, Healy et al⁷⁰ found that the association between accelerometer-assessed sedentary time and HDL cholesterol was modified by race/ethnicity (P for interaction =.004). Higher sitting time in non-Hispanic whites was found to be significantly associated with lower HDL cholesterol (P for trend =.008), but no association was found for either Mexican-Americans or non-Hispanic blacks (P for trend =.40 and .31, respectively). In a cross-sectional study of 661 Japanese adults, Honda et al⁷² found that both accelerometer-assessed total sitting time and self-reported sitting time were inversely associated with HDL cholesterol (β for each hour of accelerometer-assessed sitting = −1.312, 95% CI = −2.086 to −0.537; β for each hour of self-reported sitting = −0.434, 95% CI = −0.767 to −0.102). In the same study, self-reported daily sitting time was also positively associated with blood glucose levels (β = 0.004; 95% CI = 0.001 to 0.007). Similarly, Qi et al⁷³ also objectively assessed sitting time, using the Actical accelerometer, in 12 083 Hispanic and Latino participants in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). After adjusting for PA, sedentary time was detrimentally associated with several cardiometabolic biomarkers such as HDL cholesterol (P = .04), triglycerides, 2-hour glucose, and fasting insulin (all P < .001). These associations (with the exception of that for triglycerides) remained significant even in participants who were meeting the 2008 PA guidelines.

Cross-sectional studies by Chau et al⁷⁴ and Thorp et al⁴³ provide evidence that total sitting time is associated with SBP and DBP. Chau et al examined this association in the Norwegian HUNT study, which includes more than 40 000 adults who are 20 years of age and older. They found that individuals who reported 10 or more hours of daily sitting had a 0.98 mm Hg higher DBP (95% CI = 0.62 to 1.35) and 0.74 mm Hg higher SBP (95% CI = 0.18 to 1.29) compared with participants who sat for less than 4 h/d. Similarly, among women in the AusDiab study, each hour of daily sitting was associated with higher SBP (β = 0.39 mm Hg; 95% CI = 0.13 to 0.64) and DBP (β = 0.25 mm Hg; 95% CI = 0.11 to 0.39). In contrast, among men in the same study, the association of sitting time and DBP was not significant, whereas higher total sitting time was associated with lower SBP (β = −0.29 mm Hg; 95% CI = −0.56 to −0.03). The authors suggest that this unexpected finding might be a result of hemodynamic responses to sitting in highly sedentary men.

Associations Between Total Sitting Time and CVD

The evidence for the association between overall daily sitting time and CHD has been inconsistent. In a large Danish cohort study among more than 70 000 men and women, the risk for myocardial infarction was 38% higher in participants who reported 10 or more hours of daily sitting compared with less than 6 hours (HR = 1.38; 95% CI = 1.01 to 1.88).⁷⁵ However, in the same study, total CHD, which additionally included angina pectoris, certain current complications following acute myocardial infarction, other acute CHD, and chronic CHD, was not associated with sitting time (HR = 1.07; 95% CI = 0.91 to 1.27). Similarly, in the cross-sectional 45 and Up Study among middle-aged men in Australia, there was no association between sitting time and heart disease (OR = 0.99, 95% CI = 0.90 to 1.08, for ≥8 h/d of sitting compared with <4 h/d).⁶⁹

Prolonged sitting time was associated with incident CVD among individuals in the prospective FINRISK study.⁷⁶ In this study among 4516 Finnish adults aged 25 to 74 years, sedentary time was assessed at baseline in 2002 by self-report, and follow-up time for incident fatal and nonfatal CVD was on average 8.6 years. Each hour of daily sitting time was associated with a 6% increase in incident fatal and nonfatal CVD (95% CI = 1% to 11%).⁷⁶ In contrast, Herber-Gast et al⁷⁷ did not find significant associations between total self-reported sitting time and nonfatal and fatal CVD incidence in a prospective cohort of 6154 women from the Australian Longitudinal Study on Women’s Health (HR = 0.97; 95% CI = 0.92 to 1.03). One possible explanation for this null finding was the lower average sitting times compared with other cohorts (5.4 h/d of average sitting time overall, and 8.4 h/d of sitting in the highest quartile).

Several studies have recently used the Women’s Health Initiative Observational Study (WHI-OS) to examine the association between overall sitting time and various disease endpoints. Chomistek et al⁷⁸ found that total sitting time was associated with higher risk for CHD, stroke, and total CVD. Sitting for 10 or more hours per day, compared with 5 or fewer hours per day, was associated with HRs of 1.18 (95% CI = 1.05 to 1.32) for CHD, 1.21 (95% CI = 1.07 to 1.37) for stroke, and 1.18 (95% CI = 1.09 to 1.29) for total CVD. Additionally, subgroup analyses revealed significant effect modification by age and BMI (P for interaction =.026 and .044, respectively). Sitting time was associated with higher risk for CVD in overweight/obese (BMI ≥ 25 kg/m²) and older women (70 years and older) but not associated in normal-weight and younger women (HR_BMI≥25 = 1.26, 95% CI = 1.13 to 1.40, and HR_age≥70 = 1.22, 95% CI = 1.09 to 1.36, comparing ≥10 h/d of total sitting to ≤5 h/d). In a different study in the WHI-OS, total sitting time was associated with higher CHD mortality (HR = 1.27; 95% CI = 1.04 to 1.55) but not with CVD mortality (HR = 1.13; 95% CI = 0.99 to 1.29).⁷⁹ Finally, in the most recent study in the WHI-OS, there was no significant association between sitting time at baseline or change in sitting time during follow-up and CVD mortality.⁸⁰

Two other large prospective cohort studies also examined the association between total sitting time and cardiovascular mortality. In a study among more than 240 000 participants aged 50 to 71 years in the NIH-AARP Diet and Health Study, sitting for 9 or more hours per day was associated with 16% higher risk for CVD mortality (95% CI = 1.02 to 1.30) compared with sitting for less than 3 h/d.⁴⁹ Likewise, in the Cancer Prevention Study II Nutrition Cohort that included more than 120 000 men and women aged 50 to 74 years, self-reported sitting time of more than 6 h/d was associated with a RR of 1.33 (95% CI = 1.17 to 1.52) for women and 1.18 (95% CI = 1.08 to 1.30) for men, compared with sitting time of less than 3 h/d.⁸¹

Discussion

In recent years, the topic of sedentary behavior and CVD has received increased attention from the research community. Several reviews^9,35,82-86 and meta-analyses^38,83,87 on this topic have been published in the past 7 years. At the same time, time spent in sedentary behaviors has remained high after increases since the middle of the past century despite the evidence that sedentary time is associated with increased risk of several diseases, such as CHD, diabetes, metabolic syndrome, and obesity as well as mortality, independent of PA.^{9,22,31,35,82,88}

The purpose of the current review was to summarize the most recent evidence with regard to sitting time and CVD. We presented findings by the domains in which sedentary behaviors occur: TV viewing, occupational, transportation, and overall sitting. We considered outcomes such as total CVD, stroke, CHD, and heart failure as well as CVD risk factors, including diabetes, hypertension, hypercholesterolemia, and associated biomarkers. Based on this review, the most consistent associations were between TV watching and risk of type 2 diabetes as well as fatal and nonfatal CVD. In addition, there was evidence to support associations between transportation-related sedentary time and cardiovascular risk factors as well as CVD, and associations between total sitting time and risk of diabetes and CVD mortality. Less consistent associations were found between occupational sitting, cardiovascular risk factors, and CVD.

Throughout this article, we presented results from multivariate models that adjusted for PA. Commonly included covariates are age, gender, race, education, smoking history, alcohol consumption, and diet. Even though investigators attempted to adjust for these confounders, there is the possibility of residual confounding from measurement error or imperfect adjustment, in particular for socioeconomic status and diet. Additionally, unmeasured confounding by other covariates because they were not measured or because the data are not available remains a limitation as well. For example, mental health/depression is potentially an unmeasured confounder in most studies because it is related both to sitting and cardiovascular outcomes. On a similar note, caution should be used when interpreting results from cross-sectional studies because of the potential for reverse causation, because individuals who are obese or have type 2 diabetes or other cardiovascular risk factors may sit more as a result of their comorbidities. To minimize the potential for reverse causation in studies of sedentary behavior and cardiovascular outcomes, prospective studies should be conducted whenever possible and include additional analyses, excluding outcomes occurring in the first years of follow-up.

As pointed out in the Introduction, assessment of sedentary behavior remains difficult because objective and self-report measures both have disadvantages. Gibbs et al⁸⁹ recently recommended the expanded use of objective measures in longitudinal studies along with standardizing methods when accelerometers are used, in particular with regard to cut-points. The usual threshold to distinguish between sedentary and light activity behaviors is 100 accelerometer counts per minute (cpm) for uniaxial data, but depending on the performed activity, the use of this cut-point can lead to misclassification of these behaviors. For example, in a small validation study, time spent standing still was incorrectly classified as sedentary and riding in a car was misclassified as light activity when using the 100 cpm threshold.⁹⁰ Therefore, it was recommended that adding an assessment of posture to the accelerometer counts will improve the correct classification of sedentary behavior.

Another important issue to consider is the interdependence of time spent in activities of different intensity levels because the number of hours in a day are fixed. Recently, isotemporal substitution modeling has been used, where total time is kept fixed and one examines how increasing time spent in one activity (eg, light-intensity physical activity) while reducing time spent in another activity (eg, sedentary time) is associated with an outcome of interest.⁹¹ For example, in a subsample of 698 adults from the 2011/2012 wave of the AusDiab3 Study, Healy et al⁹² used the isotemporal modeling approach and found that replacing 2 h/d of sitting with standing and stepping was associated with improved levels of cardiometabolic biomarkers, lower BMI, and lower waist circumference. Similarly, in the Nurses’ Health Study, Mekary et al⁹¹ found that substituting TV watching with slow walking, brisk walking, or jogging/running was associated with lower body weight. Thus, using the isotemporal substitution approach can highlight the benefits of replacing sedentary time with more active behaviors, which is important for public health recommendations.

In addition to total sitting time, breaks in sitting time have also been shown to be associated with certain cardiovascular and metabolic risk factors. In a cross-sectional analysis among 168 participants of the AusDiab study, Healy et al⁹³ found that breaks in sitting were beneficially associated with waist circumference, BMI, triglycerides, and 2-hour plasma glucose, independent of total sedentary time. On top of epidemiological studies, several randomized trials, some in crossover design, have examined the effect of breaks in sitting time on glucose metabolism in adults at risk for or already diagnosed with type 2 diabetes. In these trials, when sitting was interrupted by light- or moderate-intensity walking or resistance activities, postprandial glucose and insulin levels were lower, compared with uninterrupted sitting.^94-96 Although more studies are needed, it appears that breaking up prolonged periods of sedentary behavior with short bouts of activity may counteract some of the ill effects of high amounts of sitting time. Objective assessment is preferred over self-report questionnaires to measure breaks in sitting and bouts of sitting accurately.

In this review, we have only included studies that adjusted for levels of PA. In addition to adjustment, some studies have also examined effect modification by PA as well as joint associations for the combination of PA and sitting time. In a stratified analysis of data on 7744 men in the Aerobics Center Longitudinal Study, Warren et al⁴⁸ found that the association between time spent riding in a car and CVD mortality was significant only among inactive men (P for trend =.02) but not among active men (P for trend =.13, P for interaction =.11). Likewise, in a meta-analysis of 9 prospective cohort studies, Ekelund et al⁹⁷ found that high levels of MVPA seemed to remove the increased cardiovascular mortality risk associated with high amounts of sitting time.⁹⁷ For individuals in the most active quartile (>35.5 MET-h/wk), high amounts of daily sitting time were not associated with an increased risk for cardiovascular mortality (HR = 1.07; 95% CI = 0.96 to 1.20, comparing >8 h/d of sitting with <4 h/d). Results for the association between TV viewing and cardiovascular mortality were similar. In their study on joint associations between sitting and PA in the WHI observational study, Chomistek et al⁷⁸ found that there was no association between sitting time and CVD among women in the highest category of PA (>20 MET-h/wk).⁷⁸ However, among women who reported lower levels of PA (either 1.8-8.3 MET-h/wk or 8.4-20 MET-h/wk), those who reported sitting for more than 10 h/d were still at an increased risk for CVD, although it is important to note that the interaction was not statistically significant.

In the studies described above, when the joint association of sedentary time and PA on risk of CVD was examined, individuals reporting high amounts of sitting and low amounts of PA consistently had the highest risk of CVD. In the meta-analysis by Ekelund et al,⁹⁷ individuals who reported more than 8 h/d of sitting and less than 2.5 MET-h/wk of PA had a cardiovascular mortality risk that was 74% higher than those who were most active (>35.5 MET-h/wk) and least sedentary (<4 h/d; HR = 1.74; 95% CI = 1.60-1.90). In the Chomistek et al⁷⁸ study, women who reported 1.8 MET-h/wk or less of PA and 10 h/d or more of sitting had a significantly higher rate of total CVD (HR = 1.63; 95% CI = 1.39 to 1.90) compared with women with more than 20 MET-h/wk and 5 hours or less of daily sitting.⁷⁸ Petersen et al⁹⁸ reported similar trends in a cross-sectional study among 15 235 Danish adults. Study participants who reported 10 or more hours of daily sitting time and were inactive in leisure time had much higher odds (OR = 3.29; 95% CI = 2.60 to 4.15) for metabolic syndrome compared with adults who reported leisure-time MVPA and less than 6 h/d of sitting. Thus, public health recommendations should encourage individuals to both meet the PA guidelines and avoid accumulating high amounts of sedentary time throughout the day.

Future directions of research should include conducting more prospective, longitudinal studies that assess sedentary time with an accelerometer. Accelerometers provide the best information on total accumulated sitting time throughout the day as well as time spent in activities of other intensities, which are likely correlated with sitting time. For example, using accelerometry, Healy et al⁹⁹ found that sedentary time and light-intensity activities were strongly negatively correlated (Pearson’s r = −0.96), whereas moderate-to-vigorous-intensity PA (MVPA) was only weakly correlated with sedentary time (Pearson’s r = −0.27) in a subsample of participants in the AusDiab study. Thus, to best examine the detrimental associations sedentary time may have with disease outcomes, accounting for time spent in other types of activity via isotemporal substitution modeling seems critical. Noteworthy here are findings from Maher et al.¹⁰⁰ In a series of cross-sectional models using the NHANES data from 2003/2004 and 2005/2006, Maher et al showed that the associations between sitting and several cardiometabolic biomarkers were significant when adjusting for MVPA but that these associations disappeared when adjusting for total PA instead of MVPA. This underlines the importance of carefully addressing confounding throughout the whole spectrum of PA, which includes light PA and also moderate-to-vigorous PA.

In addition to more prospective, longitudinal studies of objectively measured sedentary time, future efforts should also be more specific to the domains where sedentary behaviors occur. Many adults of working age are employed in office occupations and spend the majority of their day sedentary. More observational but also intervention studies at the workplace are needed to better understand the long-term effects of these high amounts of accumulated sitting time and to learn about successful strategies to break up these long blocks of sitting time. Generally, interventions that aim to reduce sitting time have been shown to be effective in children and adolescents,¹⁰¹ but more evidence is needed in the adult population.⁶ Prince et al⁸³ found that interventions that focus primarily on reducing sedentary behavior showed larger reductions in sitting time compared with interventions with a focus on PA or combined PA and sedentary behavior components. Owen et al⁶ describe possible strategies of behavior change for sedentary time in the home environment and at the workplace. These can include standing desks or even treadmill desks at the workplace and infrastructure (such as showers) to increase active transportation. At home, individuals can be encouraged to stand during TV commercials or while making phone calls and to iron while they are watching TV. Additionally, innovations in community infrastructure and the built environment can and should be used to study their effectiveness on reducing sitting time and, ideally, improving cardiovascular health.