Association between various sedentary behaviours and all-cause, cardiovascular disease and cancer mortality: the Multiethnic Cohort Study

Abstract

Background It has been proposed that time spent sitting increases all-cause mortality, but evidence to support this hypothesis, especially the relative effects of various sitting activities alone or in combination, is very limited.

Methods The association between various sedentary behaviours (time spent: sitting watching television (TV); in other leisure activities; in a car/bus; at work; and at meals) and mortality (all-cause and cause-specific) was examined in the Multiethnic Cohort Study, which included 61 395 men and 73 201 women aged 45–75 years among five racial/ethnic groups (African American, Latino, Japanese American, Native Hawaiian and White) from Hawaii and Los Angeles, USA.

Results Median follow-up was 13.7 years and 19 143 deaths were recorded. Total daily sitting was not associated with mortality in men, whereas in women the longest sitting duration (≥10 h/day vs <5 h/day) was associated with increased all-cause (11%) and cardiovascular (19%) mortality. Multivariate hazard ratios (HR) for ≥5 h/day vs <1 h/day of sitting watching TV were 1.19 in men (95% confidence interval (CI) 1.10–1.29) and 1.32 in women (95% CI 1.21–1.44) for all-cause mortality. This association was consistent across four racial/ethnic groups, but was not seen in Japanese Americans. Sitting watching TV was associated with an increased risk for cardiovascular mortality, but not for cancer mortality. Time spent sitting in a car/bus and at work was not related to mortality.

Conclusions Leisure time spent sitting, particularly watching television, may increase overall and cardiovascular mortality. Sitting at work or during transportation was not related to mortality.

Introduction

Sedentary behaviours, conventionally defined as activities that do not increase energy expenditure above the resting level of 1.0–1.5 METs (metabolic equivalents) or as reclining postures, include activities such as sitting, lying down, watching television, writing and reading.^1–3 Time spent sitting during work and leisure has increased considerably with modern technological advances in transportation, the workplace and entertainment.⁴

An effect on disease status of time spent in sedentary behaviours, such as sitting, is unlikely to be due to confounding from time spent engaged in moderate to vigorous physical activity, as large cohort studies have shown no association between leisure sitting time and physical activity.^5,6 A cross-sectional study of 11 037 adults showed that time sitting for television or computer use was not correlated with time spent in physical activity.⁷ Individuals can achieve high levels of moderate to vigorous physical activity while spending the remainder of the day being sedentary.¹

Recently, results from a meta-analysis of several cohort studies suggested that prolonged television (TV) viewing was associated with an increased risk of the incidence of cardiovascular disease,⁸ type 2 diabetes and all-cause mortality.⁹ In addition to TV watching, people usually spend additional time doing other sitting activities at leisure, such as reading, listening to music, relaxing, using a computer or playing video games.¹⁰

Although evidence suggesting deleterious effects of prolonged sitting on premature mortality has been growing, most previous studies have focused on only one type of sitting behaviour, thus failing to address the problem more broadly. Sitting behaviours have different energy expenditure levels, depending on other activities performed simultaneously while sitting.³ Moreover, the association between sitting and mortality across different racial/ethnic groups that may have different activity patterns^6,11–16 has rarely been examined, although a recent study compared Whites and African Americans.⁵

Our goal was to examine the association between total sitting, time spent in specific sitting behaviours (watching TV, doing other leisure activities, in a car or bus, at work and at meals) and mortality (all-cause and cause-specific) in a multiethnic population, adjusting for physical activity and other potential confounders. We also examined possible interactions with other factors, including age and ethnicity, as well as different types of physical activity and sedentary behaviours.

Methods

Study population

The design and characteristics of the Multiethnic Cohort Study (MEC) have been described in detail elsewhere.^17,18 In brief, the MEC was established between 1993 and 1996 in Hawaii and California, USA, to examine the association of lifestyle factors with the risk of cancer and other chronic diseases in diverse populations. More than 215 000 men and women, aged 45–75 years, comprising mainly five racial/ethnic groups (African American, Latino, Japanese American, Native Hawaiian and White) were enrolled by completing a 26-page mailed questionnaire. The study was approved by the institutional review boards of the University of Hawaii and the University of Southern California.

In the current analyses, we excluded 13 989 participants who did not self-identify as one of the five major racial/ethnic groups, and 8263 participants with implausible dietary data based on total energy intake or its components. We further excluded subjects with missing data on hours spent sleeping (n = 6308), spent in any physical activity (n = 10 461) or spent in two or more of the sitting activities (n = 3578). We also excluded subjects with missing data on height or weight (n = 2318) or smoking history (n = 5149). Additionally, men and women who reported a personal history of cancer, heart attack or stroke at baseline (n = 30 590), and participants who died within the 1st year after cohort entry (n = 530) were excluded. Finally, a total of 134 596 men and women were included in this analysis.

The excluded subjects were very similar to the analytical sample on median follow-up time, body mass index (BMI), energy intake, hours spent sleeping per day and prevalence of smoking. On the other hand, the excluded subjects were older (by 4.9 years on average), consumed somewhat more fruits and vegetables, drank an average of 2.0 fewer grams of alcohol per day, were more likely to have a history of hypertension and/or diabetes and were less likely to have a college education. These factors are adjusted for in our analysis.

Ascertainment of mortality

Deaths among MEC participants were identified through linkage to death certificate files in Hawaii and California, which were augmented through periodic linkages with the National Death Index in the USA. Causes of death were classified according to the International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Revision (ICD-10). Specific causes of death were grouped into three categories: cardiovascular disease (ICD-9 codes 390–434, 436–448; ICD-10 codes I00–I78), cancer (ICD-9 codes 140–208; ICD-10 codes C00–C97), and all other causes. Up to 31 December 2007, a total of 19 143 deaths were identified during an average 13.7 years of follow-up.

Measures of time spent in sitting and physical activity

In our baseline questionnaire, we ascertained sedentary activities (sleeping and five types of sitting) on a 24-h basis, and three categories of physical activity on a weekly basis. The full questionnaire can be viewed at (http://www.crch.org/multiethniccohort/index.htm). Sitting activities included: ‘sitting in a car or bus’, ‘sitting at work’, ‘sitting at meals’, ‘sitting watching television’ and ‘other leisure sitting activities (such as reading, playing cards, sewing)’. Each sitting activity was asked in seven categories: ‘never’, ‘<1 hour’, ‘1–2 hours’, ‘3–4 hours’, ‘5–6 hours’, ‘7–10 hours’, and ‘11 hours or more’ per day. Total daily sitting was the sum of the midpoints of the specific sitting categories, using 0 for ‘never’, 0.5 for ‘<1 hour’, 1.5 for ‘1–2 hours’, 3.5 for ‘3–4 hours’, 5.5 for ‘5–6 hours’, 8.5 for ‘7–10 hours’ and 11 for ‘11 hours or more’. The analysis was also performed using the Pareto Curve¹⁹ to assign the midpoint for the open-ended category of 11 h or more, to account for differences in patterns by sitting activities. Results were unchanged and are not shown.

Vigorous and moderate physical activities included: 'strenuous sports (such as jogging, bicycling on hills, tennis, racquetball, swimming laps, aerobics)', 'vigorous work (such as moving heavy furniture, loading or unloading trucks, shovelling, weight lifting or equivalent manual labour)' and 'moderate activity (such as housework, brisk walking, golfing, bowling, bicycling on level ground, gardening)'. Each type of physical activity was asked in eight categories: ‘never’, ‘30 minutes–1 hour’, ‘2–3 hours’, ‘4–6 hours’, ‘7–10 hours’, ‘11–20 hours’, ‘21–30 hours’ and ‘31 hours or more’ per week. Light physical activity was estimated by subtracting the total time spent in all activities (sitting, physical activity and sleeping) from 24 h. The metabolic equivalents (METs) for physical activity per week were computed using the following formula:²⁰ [(Number of hours in moderate activities × 4.0) + (Number of hours in vigorous work and in strenuous sports × 7.2)]. In a validation study, the correlation for energy expenditure based on the MEC activity questionnaire and the gold standard of doubly-labelled water was reasonable (r = 0.31)²¹ and comparable to those reported in the literature using other instruments.^22,23

Statistical analysis

Cox proportional hazard models, with age as the time metric, were used separately for men and women. Observation started at cohort entry, and exit time was defined as the date of death or at the end of follow-up (31 December 2007), whichever occurred earlier. Sitting variables were parameterized as indicator variables representing three categories of duration. Trend tests were conducted by inclusion of a continuous variable in the model assigned the median value for the appropriate category of sitting. Because smoking is a strong risk factor for early death, we carefully adjusted for this variable using a complex time-dependent model previously developed for a study of tobacco use and lung cancer incidence.²⁴

Multiple imputation was used for hours of sitting activities when only one type was missing, so that total sitting time could be computed for these individuals. The percentage missing data for sitting behaviours was: 1.6% for sitting watching TV; 1.6% for sitting in other leisure activities; 2.1% for sitting in a car or bus; 6.6% for sitting at work; and 0.3% for sitting at meals. The missing values were imputed by the Markov Chain Monte Carlo method based on education, age, ethnicity and smoking status and intensity. Five imputation data sets were created and the results were aggregated across data sets using standard multiple imputation methods.²⁵ All hazard ratios presented in this report were derived from the multiple imputation method.

We began with a minimally adjusted model consisting of race/ethnicity, age at cohort entry (5-year age groups) to further account for any cohort effects, and educational level (less than college, and college graduate or higher) as strata variables, in addition to the adjustment for smoking. In the fully adjusted model, we made additional adjustment for the following potential confounders: prevalent diabetes and/or hypertension at baseline (yes or no), energy intake divided at the median (2184 Kcal/day for men and 1747 Kcal/day for women), alcohol intake divided at the median (2.8 g/day for men and 0 g/day for women) and physical activity (METs/week by sex-specific quartiles). Fully adjusted models for duration of sitting by type also mutually adjusted for hours spent in all other sitting behaviours as trend variables. Global and pairwise comparisons from competing risk models were used to examine whether the hazard ratios associated with sitting were similar across causes of death.²⁶ Sensitivity analysis was performed by excluding men and women who died within 5 years from the initial date of follow-up. We also ran the analysis for individual sitting activities, including all cohort members with data, to determine the influence of the exclusions which produced similar results (data not shown).

We further stratified the analysis of sitting and mortality by several factors: age, racial/ethnic group, education level, diabetes and/or hypertension, smoking status, BMI, vigorous and moderate physical activity, light physical activity, fruit and vegetable intake and sleep duration. As in other similar studies,^5,9 BMI, which may be a mediator of the effect of prolonged sitting on mortality, was not included as an adjustment variable. Because detailed current employment status was not asked for in the baseline questionnaire, to minimize the effect of classification bias of employment status, we restricted the’ sitting at work’ analyses to subjects who reported at least some time spent sitting at work, which included 45 115 men and 51 180 women.

The effects on mortality for different types of sitting activities were compared in a model including total hours of sitting per day, as well as the proportion of total sitting contributed by each of five sitting activities. This model is called the differential effects model in this paper. A global Wald test was used to compare the parameters for proportion in each sitting activity, allowing us to compare different sitting activities, adjusting for total sitting time. The joint effects of different sitting activities were then examined in pairwise fashion. Interaction was assessed by a Wald test of the cross-product terms of the proportions contributed by the two sitting activities being considered, grouped into three categories. This avoided the detection of an interaction between sitting activities due solely to the increase in total sitting time.

Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC, USA).

Results

Characteristics of the study subjects

Table 1 shows the distributions of the study participants on several demographic and exposure characteristics by sex and race/ethnicity. Whites, Japanese Americans and Latinos comprised about 20–30% each of our study population, followed by African Americans (15%) and Native Hawaiians (7.5%). The mean age at cohort entry was 58.7 and 58.5 years in men and women, respectively. Among the five ethnic groups, both in men and women, African Americans were less physically active, were more likely to be current smokers and had a higher prevalence of hypertension and/or diabetes. Whites and Japanese were more likely to have a college education than other ethnic groups, among both men and women.

Table 1.

Basic characteristics for 134 596 men and women in the Multiethnic Cohort Study, 1993–1996^*

		Racial/ethnic groups
Variables	Statistics	White (N = 15 894)	African American (N = 7252)	Native Hawaiian (N = 4499)	Japanese American (N = 19 639)	Latino (N = 14 111)
Men
Age at entry, years	Mean ± SD	57.4 ± 8.7	60.0 ± 8.9	55.5 ± 8.2	59.7 ± 9.1	59.2 ± 7.7
Body mass index, kg/m2	Mean ± SD	25.9 ± 3.9	26.7 ± 4.2	28.5 ± 5.1	24.8 ± 3.3	26.9 ± 3.9
Energy intake, Kcal/day	Mean ± SD	2312.0 ± 899.0	2240.5 ± 1170.5	2786.8 ± 1308.3	2276.5 ± 835.9	2640.8 ± 1365.5
Fruit and vegetable intake, g/1000 Kcal/day	Mean ± SD	289.1 ± 143.8	286.1 ± 153.0	261.9 ± 137.2	272.0 ± 132.6	305.3 ± 143.0
Physical activity, METs/weeka	Mean ± SD	60.3 ± 64.4	43.1 ± 60.0	69.3 ± 77.8	48.6 ± 54.3	51.2 ± 68.1
Sleeping hours per day	Mean ± SD	7.2 ± 1.0	6.9 ± 1.3	6.9 ± 1.2	6.9 ± 1.1	7.1 ± 1.2
Current smokers	N (%)	2625 (16.5)	1971 (27.2)	1023 (22.7)	3128 (15.9)	2645 (18.7)
Alcohol drinkers	N (%)	11 647 (73.3)	4305 (59.4)	2750 (61.1)	11 258 (57.3)	9471 (67.1)
Hypertension and/or diabetes	N (%)	4621 (29.1)	3935 (54.3)	2203 (49.0)	8796 (44.8)	5521 (39.1)
College or graduate school	N (%)	8187 (51.5)	1830 (25.2)	927 (20.6)	6945 (35.4)	1916 (13.6)

Variables	Statistics	White (N = 18 587)	African American (N = 12 476)	Native Hawaiian (N = 5584)	Japanese American (N = 22 378)	Latino (N = 14 176)
Women
Age at entry, years	Mean ± SD	57.6 ± 8.8	59.0 ± 9.0	55.0 ± 8.2	59.9 ± 8.9	58.4 ± 7.7
Body mass index, kg/m2	Mean ± SD	25.1 ± 5.2	28.2 ± 5.7	27.8 ± 6.0	23.1 ± 3.8	27.2 ± 5.1
Energy intake, Kcal/day	Mean ± SD	1815.8 ± 698.5	1897.9 ± 981.1	2355.7 ± 1237.9	1808.2 ± 675.2	2179.0 ± 1152.6
Fruit and vegetable intake, g/1000 Kcal/day	Mean ± SD	362.2 ± 169.9	364.2 ± 181.8	339.9 ± 173.0	370.1 ± 163.7	379.8 ± 175.9
Physical activity, METs/weeka	Mean ± SD	46.6 ± 47.1	29.8 ± 39.1	46.3 ± 56.4	33.6 ± 36.1	32.4 ± 45.1
Sleeping hours per day	Mean ± SD	7.2 ± 1.1	6.9 ± 1.3	6.9 ± 1.2	6.8 ± 1.0	7.1 ± 1.2
Current smokers	N (%)	3088 (16.6)	2487 (19.9)	1314 (23.5)	2101 (9.4)	1528 (10.8)
Alcohol drinkers	N (%)	11 408 (61.4)	5025 (40.3)	2176 (39.0)	5258 (23.5)	5361 (37.8)
Hypertension and/or diabetes	N (%)	4797 (25.8)	6765 (54.2)	2355 (42.2)	8407 (37.6)	5288 (37.3)
College or graduate school	N (%)	7214 (38.8)	3226 (25.9)	999 (17.9)	6871 (30.7)	1347 (9.5)

METs, metabolic equivalents; SD, standard deviation.

^aMETs for moderate activity, vigorous work and strenuous sports.

*P-values < 0.001 in every subgroup comparison across racial/ethnic groups.

Table 2 compares men and women on the various sedentary behaviours by several subgroup strata. Time spent sitting watching TV was similar between the sexes. Women reported longer durations for sitting doing other leisure activities and at work, whereas men spent more time sitting in a car or bus. For both sexes, African Americans spent more time on average sitting watching TV, whereas Whites and Japanese spent more time sitting at work. On average, men and women who were older, had less education, had higher BMIs, were more likely to be current smokers, were less physically active and consumed fewer fruits and vegetables spent more time sitting watching television. Men and women who had higher BMIs also spent more time sitting in a car or bus and sitting at work. Younger men and those with more education spent longer durations sitting at work. Men and women with higher energy intakes sat longer at meals (data not shown).

Table 2.

Average daily sitting (hours) spent in various sedentary behaviours by various subgroup classifications among men and women in the Multiethnic Cohort Study, 1993–1996^*

	Men							Women
		Average daily sitting (hours)							Average daily sitting (hours)
Variable	No.	Total sitting	Watching television	Other leisure activitiesa	In a car or bus	At meals	At workb	No.	Total sitting	Watching television	Other leisure activitiesa	In a car or bus	At meals	At workb
All subjects	61 395	8.0	2.5	1.3	1.2	0.9	3.3	73 201	8.2	2.5	1.6	1.0	0.9	3.8
Ethnicity
White	15 894	8.6	2.3	1.4	1.1	1.0	3.6	18 587	8.6	2.3	1.7	0.9	0.9	3.7
African American	7252	8.7	2.9	1.4	1.4	0.8	3.0	12 476	9.0	2.7	1.7	1.2	0.9	3.6
Native Hawaiian	4499	8.7	2.6	1.4	1.3	0.9	3.0	5584	9.1	2.4	1.8	1.1	0.9	3.6
Japanese American	19 639	8.5	2.6	1.3	1.0	1.0	3.4	22 378	8.7	2.6	1.6	0.8	1.0	3.7
Latino	14 111	7.3	2.3	1.1	1.4	0.8	2.8	14 176	7.7	2.3	1.3	1.2	0.8	3.5
Age at enrolment, years
<65	43 073	8.6	2.4	1.2	1.3	0.9	3.4	52 068	8.8	2.3	1.5	1.0	0.9	3.9
≥65	18 322	7.6	2.8	1.5	1.0	1.0	2.5	21 133	7.9	2.8	1.8	0.9	1.0	2.6
Education level
Less than college	41 565	7.8	2.7	1.2	1.3	0.8	2.7	53 483	8.5	2.6	1.6	1.0	0.9	3.7
College or graduate school	19 830	9.3	2.2	1.4	1.1	1.0	4.2	19 718	8.8	2.1	1.7	1.0	1.0	3.6
Diabetes and/or hypertension
No	36 319	8.2	2.4	1.3	1.2	0.9	3.3	45 589	8.5	2.3	1.6	1.0	0.9	3.7
Yes	25 076	8.3	2.7	1.3	1.2	0.9	3.2	27 612	8.6	2.7	1.7	1.0	0.9	3.6
Smoking status
Nonsmokers	19 828	8.2	2.3	1.2	1.2	0.9	3.4	42 229	8.3	2.4	1.5	1.0	0.9	3.6
Former smokers	30 175	8.3	2.6	1.3	1.2	0.9	3.2	20 454	8.7	2.5	1.7	1.0	0.9	3.7
Current smokers	11 392	8.4	2.8	1.3	1.3	0.8	3.0	10 518	9.1	2.8	1.8	1.0	0.9	3.8
Body mass index, kg/m2
<24.9	26 637	8.1	2.4	1.3	1.1	0.9	3.2	38 318	8.3	2.3	1.6	0.9	1.0	3.6
25.0–29.9	26 327	8.3	2.5	1.3	1.2	0.9	3.2	21 989	8.6	2.5	1.6	1.1	0.9	3.6
≥30.0	8431	8.9	2.7	1.3	1.4	0.9	3.3	12 894	9.2	2.8	1.7	1.2	0.9	3.8
Physical activity, METs/weekc
<33.4 (men), <20.0 (women)	29 773	8.4	2.6	1.3	1.2	0.9	3.5	37 018	8.6	2.5	1.5	1.0	0.9	3.9
≥33.4 (men), ≥20.0 (women)	31 622	8.1	2.4	1.3	1.2	0.9	3.0	36 183	8.5	2.4	1.7	1.0	1.0	3.4
Fruit and vegetable intake, g/1000 Kcal/day
<260.2 (men), <339.3 (women)	30 700	8.6	2.7	1.3	1.2	0.9	3.3	36 609	8.9	2.6	1.6	1.0	0.9	3.9
≥260.2 (men), ≥339.3 (women)	30 695	8.0	2.4	1.3	1.2	0.9	3.2	36 592	8.2	2.3	1.6	1.0	0.9	3.4
Sleeping, h/day
≤6	20 388	8.4	2.4	1.3	1.2	0.9	3.4	23 958	8.6	2.4	1.6	1.0	0.9	3.7
7	20 109	8.6	2.5	1.3	1.3	0.9	3.4	25 094	8.7	2.5	1.6	1.0	0.9	3.8
≥8	20 898	7.9	2.6	1.3	1.2	0.9	2.9	24 149	8.3	2.5	1.7	1.0	0.9	3.3

METs, metabolic equivalents.

^aOther leisure activities includes behaviours such as reading, playing cards and sewing.

^bRestricted to subjects reporting any time sitting at work including 45 115 men and 51 180 women.

^cMETs for moderate activity, vigorous work and strenuous sports.

*P-value < 0.001 in every subgroup comparison.

Total daily sitting time and mortality

In the fully adjusted model, total daily sitting was not associated with all-cause mortality, cardiovascular disease mortality or cancer mortality for men, but there was a suggested increase in ‘all other causes’ mortality (Table 3). Among women, total daily sitting of 10 h per day or more was associated with an increased risk for all-cause mortality (Hazard Ratio (HR) 1.11, 95% Confidence Interval (CI) 1.04–1.19) compared with <5 h/day. There was also an increased risk of mortality from cardiovascular disease and from ‘all other causes’ among women that was markedly different from the null association for mortality from cancer [P’s for heterogeneity <0.01 (global and all pairwise)].

Table 3.

Hazard ratios for all-cause and cause-specific mortality according to total sitting time per day in the Multiethnic Cohort Study, 1993–2007

	Men				Women
Cause of death	Total daily sitting (hours)				Total daily sitting (hours)
	<5	5 − <10	≥10	P trenda	<5	5 − <10	≥10	P trenda
All-cause
No. deaths	2041	5465	3116		1476	4085	2960
Person-years	138 819	392 269	251 500		162 708	455 692	341 107
Minimally adjusted model, HR (95% CI)b	1.00	0.98 (0.93–1.03)	1.07 (1.01–1.13)	0.01	1.00	0.96 (0.90–1.02)	1.10 (1.03–1.18)	<0.01
Fully adjusted model, HR (95% CI)c	1.00	0.99 (0.94–1.04)	1.04 (0.98–1.11)	0.09	1.00	0.99 (0.93–1.05)	1.11 (1.04–1.19)	<0.01
Cardiovascular disease
No. deaths	725	1892	1104		500	1312	1002
Minimally adjusted model, HR (95% CI)b	1.00	0.98 (0.90–1.07)	1.11 (1.00–1.22)	0.02	1.00	0.92 (0.83–1.03)	1.17 (1.05–1.31)	<0.01
Fully adjusted model, HR (95% CI)c	1.00	0.98 (0.90–1.07)	1.06 (0.96–1.18)	0.14	1.00	0.96 (0.85–1.07)	1.19 (1.06–1.34)	<0.01
Cancer
No. deaths	709	1913	1053		528	1477	1018
Minimally adjusted model, HR (95% CI)b	1.00	0.96 (0.88–1.05)	0.97 (0.88–1.08)	0.71	1.00	0.96 (0.86–1.07)	0.98 (0.88–1.09)	0.89
Fully adjusted model, HR (95% CI)c	1.00	0.96 (0.88–1.05)	0.97 (0.87–1.07)	0.62	1.00	0.97 (0.87–1.07)	0.97 (0.87–1.09)	0.75
All other causes
No. deaths	608	1660	959		448	1296	940
Minimally adjusted model, HR (95% CI)b	1.00	1.01 (0.92–1.11)	1.14 (1.03–1.27)	0.01	1.00	0.99 (0.88–1.11)	1.18 (1.05–1.33)	<0.01
Fully adjusted model, HR (95% CI)c	1.00	1.02 (0.93–1.12)	1.11 (1.00–1.23)	0.04	1.00	1.04 (0.92–1.16)	1.20 (1.07–1.35)	<0.01
Pheterogeneityc,d			0.01				<0.01

HR, hazard ratio; CI, confidence interval.

^aTrend tests were conducted by inclusion of a continuous variable in the model assigned as the median value within the appropriate category of total sitting.

^bHazard ratios were calculated with age as the time metric, adjusted for age at cohort entry (5–year age groups), education, ethnicity and smoking history [including the following variables: smoking status, average number of cigarettes, average number of cigarettes squared, number of years smoked (time dependent), number of years since quitting (time dependent) and interactions between ethnicity and the smoking variables].

^cIn addition to the above model, also adjusted for history of hypertension and/or diabetes at enrolment, alcohol consumption, energy intake and physical activity (METs per week for moderate activity, vigorous work and strenuous sports).

^dThe test of heterogeneity across causes of death was performed using a Wald test from a competing risk model, where each cause was a different event. P–values for pairwise comparisons: men: P = 0.02 between cardiovascular disease and cancer, P = 0.26 between cardiovascular disease and all other causes, P = 0.02 between cancer and all other causes; women: P's < 0.01 for all comparisons.

Time spent in specific sitting categories and mortality

Table 4 shows the associations between mortality and sitting duration, by type of sitting. The associations with all-cause mortality were statistically different between sitting types based on the differential effects models (P < 0.01 for both men and women). Sitting watching TV had a stronger effect than did other sitting behaviours both in men and women (P < 0.001 for all paired comparisons).

Table 4.

Hazard ratios for all–cause and cause–specific mortality according to sedentary behaviours in the Multiethnic Cohort Study, 1993–2007

	Men								Women
	Daily exposure (hours)a								Daily exposure (hours)a
	<1		1–4		≥5	(Watching TV; work)			<1		1–4		≥5	(Watching TV; work)
	<1		1–2		≥3	(Other leisure; car/bus; meal)			<1		1–2		≥3	(Other leisure; car/bus; meal)
	No. deaths	HR	No. deaths	HR (95% CI)b	No. deaths	HR (95% CI)b	Ptrendc	Phetd	No. deaths	HR	No. deaths	HR (95% CI)b	No. deaths	HR (95% CI)b	Ptrendc	Phetd
Sitting watching television
All–cause mortality	1134	1.00	8004	1.02 (0.96–1.08)	1484	1.19 (1.10–1.29)	<0.01		943	1.00	6268	1.01 (0.94–1.08)	1310	1.32 (1.21–1.44)	<0.01
Cardiovascular disease	402	1.00	2782	0.99 (0.89–1.10)	537	1.20 (1.05–1.37)	<0.01	<0.01	297	1.00	2070	1.02 (0.90–1.15)	447	1.33 (1.14–1.55)	<0.01	<0.01
Cancer	371	1.00	2840	1.11 (0.99–1.24)	463	1.16 (1.00–1.33)	0.16		374	1.00	2265	0.96 (0.85–1.07)	384	1.07 (0.92–1.25)	0.09
All other causes	361	1.00	2382	0.95 (0.85–1.06)	484	1.21 (1.05–1.40)	<0.01		272	1.00	1933	1.06 (0.93–1.21)	479	1.62 (1.39–1.89)	<0.01
Sitting in other leisure activities
All–cause mortality	4623	1.00	4213	0.99 (0.95–1.03)	1786	1.06 (1.00–1.12)	0.05		2645	1.00	3653	0.99 (0.94–1.04)	2223	1.07 (1.01–1.14)	0.01
Cardiovascular disease	1605	1.00	1471	1.00 (0.92–1.07)	645	1.09 (0.99–1.20)	0.08	0.49	859	1.00	1193	0.97 (0.89–1.07)	762	1.10 (0.99–1.22)	0.04	0.08
Cancer	1603	1.00	1471	1.00 (0.93–1.07)	600	1.04 (0.94–1.14)	0.51		963	1.00	1311	1.00 (0.92–1.09)	749	1.05 (0.95–1.17)	0.27
All other causes	1414	1.00	1271	0.98 (0.90–1.06)	542	1.05 (0.95–1.17)	0.38		823	1.00	1149	0.99 (0.90–1.09)	712	1.07 (0.96–1.19)	0.18
Sitting in a car or bus
All–cause mortality	6248	1.00	3396	1.00 (0.95–1.04)	977	1.00 (0.94–1.08)	0.97		5542	1.00	2402	1.00 (0.95–1.05)	577	1.04 (0.95–1.13)	0.51
Cardiovascular disease	2153	1.00	1199	1.01 (0.94–1.09)	369	1.08 (0.96–1.21)	0.22	<0.01	1776	1.00	819	1.06 (0.97–1.15)	219	1.16 (1.01–1.34)	0.03	<0.01
Cancer	2149	1.00	1187	1.01 (0.94–1.09)	338	1.03 (0.91–1.16)	0.64		1972	1.00	855	0.98 (0.91–1.07)	196	1.03 (0.88–1.19)	0.91
All other causes	1947	1.00	1010	0.96 (0.89–1.04)	270	0.90 (0.79–1.02)	0.08		1794	1.00	728	0.96 (0.87–1.05)	162	0.92 (0.78–1.09)	0.24
Sitting at meals
All–cause mortality	6760	1.00	3567	0.99 (0.94–1.03)	295	1.22 (1.08–1.38)	0.08		5342	1.00	2867	1.03 (0.99–1.08)	312	1.22 (1.09–1.37)	<0.01
Cardiovascular disease	2387	1.00	1234	0.97 (0.91–1.05)	100	1.16 (0.94–1.42)	0.65	<0.01	1774	1.00	930	1.03 (0.94–1.12)	110	1.25 (1.02–1.52)	0.05	<0.01
Cancer	2340	1.00	1239	0.98 (0.91–1.05)	95	1.14 (0.93–1.41)	0.70		1935	1.00	1004	0.98 (0.90–1.06)	84	0.95 (0.76–1.19)	0.54
All other causes	2034	1.00	1093	1.01 (0.93–1.09)	100	1.40 (1.14–1.72)	0.02		1632	1.00	934	1.10 (1.01–1.20)	118	1.50 (1.24–1.82)	<0.01
Sitting at worke
All–cause mortality	1967	1.00	2963	0.96 (0.90–1.03)	1493	0.94 (0.88–1.02)	<0.01		1228	1.00	2199	1.03 (0.95–1.12)	1347	0.98 (0.90–1.08)	<0.01
Cardiovascular disease	689	1.00	1018	0.93 (0.84–1.03)	504	0.93 (0.82–1.05)	0.01	0.22	382	1.00	716	1.07 (0.92–1.24)	409	1.12 (0.95–1.31)	0.30	<0.01
Cancer	699	1.00	1058	0.97 (0.88–1.07)	534	0.92 (0.81–1.05)	0.05		485	1.00	809	0.95 (0.85–1.08)	556	0.87 (0.76–1.00)	0.07
All other causes	578	1.00	887	0.99 (0.87–1.13)	456	1.00 (0.87–1.15)	0.17		361	1.00	674	1.09 (0.94–1.25)	382	1.01 (0.85–1.21)	<0.01

HR, hazard ratio; CI, confidence interval; Ptrend, P–value for trend, Phet, P–value for heterogeneity.

^aExposure was categorized as <1, 1–4, and ≥5 (hours per day) for sitting watching television and for sitting at work. Exposure was categorized as <1, 1–2, and ≥3 (hours per day) for sitting in other leisure activities, sitting in a car or bus and sitting at meals.

^bHazard ratios were calculated with age as the time metric, adjusted for following variables: 5–year age groups at cohort entry, education, ethnicity, history of hypertension or diabetes at enrolment, alcohol consumption, energy intake, physical activity (METs per week for moderate activity, vigorous work and strenuous sports), trend of hours for other sitting behaviours, and smoking history by inclusion of the following variables: [smoking status, average number of cigarettes, average number of cigarettes squared, number of years smoked (time dependent), number of years since quitting (time dependent) and interactions between ethnicity and the smoking variables].

^cTrend tests were conducted by inclusion of a continuous variable in the model assigned as the median value within the appropriate category of total sitting.

^dThe test of heterogeneity across causes of death was performed using a Wald test from competing risk models, where each cause was a different event.

^eRestricted to subjects reporting any time sitting at work.

For time spent ‘sitting watching TV’, the HRs for all-cause mortality in the highest exposure level were 1.19 (95% CI 1.10–1.29) among men and 1.32 (95% CI 1.21–1.44) among women for the fully adjusted model (Table 4). ‘Sitting in a car or bus’ and ‘sitting at work’ were not associated with the risk of mortality, whereas ‘sitting in other leisure activities’ and ‘sitting at meals’ showed a pattern similar to that for ‘sitting watching television’. In further analysis, when men and women who reported ‘zero’ ‘hours sitting at work’ were included in the analysis in the reference group, those sitting ≥5 h/day hours at work (vs <1 h/day including zero) and younger than 65 years at baseline had reduced risks of all-cause mortality [men: HR 0.88 (95% CI 0.81–0.95); women: HR 0.86 (95% CI 0.79–0.94)], but no associations were observed among men and women aged 65 and over (P for interaction <0.001 both for men and for women). In general, the associations by cause of mortality were similar to overall mortality, although they varied somewhat in magnitude. However, the HRs for sitting watching TV, in other leisure activities and at meals were the lowest for cancer for both men and women (P for heterogeneity <0.01).

In stratified analyses, for time spent sitting watching television (Table 5), the HRs for the highest exposure category (sitting watching TV ≥ 5 h/day) were generally increased (HR > 1) for all subgroups. Nevertheless, many of the P-values for interaction were less than 0.01, reflecting differences in strengths of the associations across subgroups. The exceptions where the HRs were not elevated were the Japanese American subgroup in men and women, and the underweight (BMI < 18.5) among men, which may be a chance finding due to the small numbers of deaths. In stratified analyses for time spent sitting at work, ‘sitting at work’ was not associated with mortality in either men or women in any stratum defined by age (<65 y and ≥65 y) or by educational level (lower than college vs college or graduate school) that may reflect type of employment (data not shown).

Table 5.

Hazard ratios for all–cause mortality in relation to time spent sitting watching television by several participant characteristics in the Multiethnic Cohort Study, 1993–2007

	Men							Women
	Daily sitting watching television (hours)							Daily sitting watching television (hours)
	<1		1–4		≥5			<1		1–4		≥5
Variable	No. deaths	HR	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	Pintb	No. deaths	HR	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	Pintb
Ethnicity
White	245	1.00	1777	1.11 (0.96–1.27)	323	1.35 (1.13–1.62)	<0.01	227	1.00	1465	0.94 (0.81–1.09)	332	1.33 (1.11–1.59)	<0.01
African American	184	1.00	1435	1.08 (0.92–1.27)	386	1.37 (1.14–1.66)		234	1.00	1670	1.05 (0.90–1.21)	439	1.48 (1.25–1.75)
Native Hawaiian	95	1.00	598	0.97 (0.75–1.24)	115	1.16 (0.85–1.58)		96	1.00	494	1.05 (0.82–1.33)	94	1.40 (1.02–1.92)
Japanese American	312	1.00	2399	0.91 (0.81–1.03)	400	0.97 (0.83–1.13)		185	1.00	1494	0.88 (0.75–1.03)	282	1.06 (0.88–1.29)
Latino	299	1.00	1795	1.02 (0.90–1.16)	259	1.19 (1.00–1.42)		201	1.00	1145	1.13 (0.97–1.32)	163	1.22 (0.98–1.51)
Age at enrolment, years
<65	514	1.00	3381	1.07 (0.97–1.18)	559	1.32 (1.17–1.50)	<0.01	470	1.00	2619	1.01 (0.91–1.12)	520	1.48 (1.30–1.69)	<0.01
≥65	620	1.00	4623	0.98 (0.90–1.07)	925	1.11 (0.99–1.23)		473	1.00	3648	0.99 (0.89–1.09)	791	1.21 (1.08–1.37)
Education level
Less than college	868	1.00	6231	0.97 (0.91–1.05)	1267	1.13 (1.04–1.24)	<0.01	731	1.00	5111	0.99 (0.91–1.07)	1124	1.28 (1.16–1.41)	<0.01
College or graduate school	266	1.00	1773	1.16 (1.02–1.33)	217	1.41 (1.17–1.71)		213	1.00	1156	1.06 (0.91–1.24)	186	1.52 (1.23–1.88)
Diabetes and/or hypertension
No	549	1.00	3581	1.04 (0.95–1.14)	609	1.21 (1.08–1.37)	<0.01	444	1.00	2738	1.05 (0.95–1.16)	476	1.36 (1.19–1.56)	0.12
Yes	586	1.00	4422	0.99 (0.90–1.08)	875	1.15 (1.04–1.29)		498	1.00	3530	0.97 (0.88–1.07)	835	1.28 (1.14–1.44)
Smoking status
Nonsmokers	334	1.00	1841	1.03 (0.92–1.17)	257	1.18 (0.99–1.40)	0.62	493	1.00	3071	1.07 (0.97–1.18)	493	1.38 (1.21–1.57)	0.27
Former smokers	498	1.00	4032	1.05 (0.95–1.15)	748	1.21 (1.08–1.36)		246	1.00	1844	0.97 (0.84–1.11)	449	1.29 (1.09–1.52)
Current smokers	303	1.00	2131	0.92 (0.81–1.05)	478	1.09 (0.93–1.27)		204	1.00	1353	0.91 (0.78–1.07)	368	1.21 (1.01–1.45)
Body mass index, kg/m2
<18.5	33	1.00	123	0.54 (0.25–1.14)	28	0.55 (0.21–1.44)	0.51	64	1.00	321	0.83 (0.60–1.13)	83	1.60 (1.07–2.41)	<0.01
18.5–24.9	557	1.00	3532	0.98 (0.89–1.07)	604	1.13 (1.00–1.27)		417	1.00	2706	1.02 (0.92–1.13)	505	1.36 (1.19–1.57)
25.0–27.4	241	1.00	1948	1.14 (0.99–1.31)	360	1.31 (1.10–1.56)		152	1.00	1086	1.02 (0.85–1.22)	202	1.25 (0.99–1.56)
27.5–29.9	157	1.00	1187	0.99 (0.83–1.18)	234	1.22 (0.98–1.52)		126	1.00	825	0.89 (0.73–1.09)	163	1.07 (0.83–1.38)
30.0–34.9	117	1.00	911	1.13 (0.92–1.39)	175	1.40 (1.08–1.81)		106	1.00	817	1.07 (0.86–1.33)	183	1.21 (0.93–1.57)
≥35.0	29	1.00	303	0.95 (0.57–1.57)	83	1.10 (0.62–1.94)		77	1.00	513	0.89 (0.69–1.16)	174	1.25 (0.93–1.68)
Physical activity, METs/weekc
<33.4 (men), <20.0 (women)	684	1.00	4551	1.00 (0.92–1.09)	998	1.22 (1.11–1.36)	<0.01	560	1.00	3800	1.06 (0.97–1.17)	858	1.39 (1.24–1.55)	0.40
≥33.4 (men), ≥20.0 (women)	450	1.00	3453	1.05 (0.95–1.16)	486	1.12 (0.98–1.28)		383	1.00	2468	0.92 (0.82–1.03)	452	1.23 (1.06–1.41)
Light physical activity, h/dayd
<7.6 (men), <7.8 (women)	268	1.00	3272	1.08 (0.95–1.22)	1119	1.31 (1.13–1.51)	<0.01	221	1.00	2638	1.05 (0.91–1.22)	1015	1.38 (1.17–1.62)	0.57
≥7.6 (men), ≥7.8 (women)	866	1.00	4732	1.00 (0.93–1.07)	365	1.07 (0.94–1.22)		722	1.00	3630	1.00 (0.92–1.09)	295	1.34 (1.16–1.55)
Fruit and vegetable intake, g/1000 Kcal/day
<260.2 (men), <339.3 (women)	501	1.00	3771	0.95 (0.87–1.05)	855	1.16 (1.03–1.30)	<0.01	407	1.00	2980	0.99 (0.89–1.10)	755	1.28 (1.13–1.45)	0.77
≥260.2 (men), ≥339.3 (women)	633	1.00	4233	1.06 (0.98–1.16)	629	1.19 (1.06–1.34)		536	1.00	3288	1.01 (0.92–1.11)	555	1.33 (1.17–1.50)
Sleeping, h/day
≤6	281	1.00	2357	1.12 (0.98–1.27)	383	1.35 (1.15–1.59)	0.30	254	1.00	1860	1.03 (0.90–1.19)	325	1.30 (1.09–1.56)	0.01
7	412	1.00	2497	0.99 (0.89–1.11)	466	1.14 (0.99–1.31)		354	1.00	2243	1.03 (0.91–1.15)	446	1.26 (1.09–1.47)
≥8	441	1.00	3150	0.96 (0.87–1.06)	634	1.14 (1.00–1.30)		335	1.00	2165	0.94 (0.83–1.06)	539	1.31 (1.13–1.52)

HR, hazard ratio; CI, confidence interval; Pint, P–value for interaction.

^aHazard ratios were calculated with age as the time metric, adjusted for following variables but the relevant variable: 5–year age groups at cohort entry, education, ethnicity, history of hypertension or diabetes at enrolment, alcohol consumption, energy intake, physical activity (METs per week for moderate activity, vigorous work and strenuous sports), trend of hours for other sitting behaviours, and smoking history by inclusion of the following variables: [smoking status, average number of cigarettes, average number of cigarettes squared, number of years smoked (time dependent), number of years since quitting (time dependent), and interactions between ethnicity and the smoking variables].

^bTest for interaction was based on the Wald statistics for cross–product terms between median values of each category of sitting watching television and stratified variables.

^cMETs for moderate activity, vigorous work and strenuous sports.

^dLight physical activity estimated by subtracting the total time spent in all activities (total sitting, physical activity, and sleeping) from 24 h.

The patterns of the associations by specific cause of death were generally similar to those of the broader causes of death categories shown in Table 5, although more time spent sitting watching TV was associated with mortality from prostate cancer in men (HR 1.39 (≥5 h/day vs <1 h/day), P for trend <0.01) (supplementary data available at IJE online).

In our sensitivity analysis excluding men and women who died within 5 years of follow-up, the findings were basically unchanged, though some of the HRs comparing ≥5 h/day of sitting watching TV vs <1 h/day were somewhat weakened: all-cause, HR 1.14 (95% CI 1.04–1.25) for men and HR 1.30 (95% CI 1.18–1.43) for women; cardiovascular, HR 1.15 (95% CI 0.98–1.34) for men and HR 1.37 (95% CI 1.16–1.63) for women; ‘all other’, HR 1.12 (95% CI 0.96–1.32) for men and HR 1.46 (95% CI 1.23–1.73) for women.

Joint effect of different sedentary behaviours on mortality

Since several sedentary behaviours were associated with increased mortality, we examined whether combinations of prolonged sitting activities augment the risk of mortality. Table 6 gives the hazard ratios for all-cause mortality by duration spent sitting watching TV and in other sitting activities, with the statistical test for interaction based on the differential effects model. As expected, longer durations spent sitting watching TV coupled with sitting longer in other behaviours led to stronger associations with mortality among both men and women. The only combination that led to an increased risk above that expected was sitting watching TV and sitting at meals.

Table 6.

The joint effects of sitting while watching television and different sedentary behaviours on all–cause mortality in the Multiethnic Cohort Study, 1993–2007

	Men							Women
	Daily sitting watching television (hours)							Daily sitting watching television (hours)
	<1		1–4		≥5			<1		1–4		≥5
Additional sedentary behaviour	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	Pintb	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	No. deaths	HR (95% CI)a	Pintb
Sitting in other leisure activities, h/day
<1	683	1.00	3366	0.98 (0.90–1.06)	573	1.14 (1.02–1.28)	0.70	489	1.00	1800	0.92 (0.83–1.02)	356	1.34 (1.17–1.54)	<0.01
1–2	315	0.88 (0.76–1.00)	3326	0.98 (0.90–1.07)	573	1.14 (1.02–1.28)		297	0.82 (0.71–0.95)	2828	0.95 (0.86–1.05)	529	1.26 (1.11–1.43)
≥3	136	1.05 (0.87–1.27)	1312	1.04 (0.94–1.14)	338	1.24 (1.08–1.42)		157	1.09 (0.91–1.32)	1640	1.04 (0.94–1.16)	426	1.25 (1.09–1.43)
Sitting in a car or bus, h/day
<1	700	1.00	4655	1.01 (0.93–1.10)	894	1.17 (1.05–1.29)	0.28	643	1.00	4045	1.01 (0.93–1.10)	854	1.33 (1.19–1.48)	0.23
1–2	323	0.96 (0.84–1.09)	2612	1.01 (0.93–1.10)	461	1.18 (1.04–1.33)		241	1.03 (0.88–1.20)	1795	1.01 (0.92–1.10)	367	1.34 (1.17–1.54)
≥3	111	1.05 (0.85–1.29)	737	0.99 (0.89–1.11)	129	1.29 (1.07–1.57)		59	0.98 (0.75–1.29)	428	1.06 (0.94–1.21)	90	1.34 (1.06–1.69)
Sitting at meals, h/day
<1	855	1.00	5028	1.00 (0.93–1.07)	877	1.17 (1.06–1.29)	0.0495	686	1.00	3910	1.03 (0.95–1.12)	746	1.35 (1.21–1.50)	0.49
1–2	259	0.92 (0.79–1.06)	2758	0.99 (0.92–1.07)	550	1.16 (1.04–1.30)		236	1.11 (0.95–1.29)	2127	1.05 (0.96–1.15)	505	1.44 (1.28–1.62)
≥3	19	1.09 (0.68–1.74)	218	1.25 (1.07–1.45)	58	1.39 (1.05–1.83)		21	1.50 (0.95–2.37)	231	1.30 (1.11–1.52)	60	1.41 (1.06–1.86)
Sitting at work, h/dayc
<1	244	1.00	1480	0.93 (0.80–1.07)	242	1.14 (0.95–1.38)	0.32	185	1.00	902	0.86 (0.72–1.02)	142	1.01 (0.81–1.27)	0.33
1–4	314	0.91 (0.76–1.10)	2330	0.90 (0.79–1.04)	319	1.08 (0.90–1.30)		233	0.88 (0.70–1.10)	1674	0.88 (0.74–1.05)	292	1.25 (1.01–1.55)
≥5	199	0.80 (0.66–0.97)	1194	0.89 (0.77–1.03)	100	1.16 (0.91–1.48)		170	0.77 (0.62–0.96)	1072	0.85 (0.71–1.03)	105	1.32 (1.00–1.74)

HR, hazard ratio; CI, confidence interval; Pint, P–value for interaction.

^aHazard ratios were calculated with age as the time metric, adjusted for following variables but the relevant variable: 5–year age groups at cohort entry, education, ethnicity, history of hypertension or diabetes at enrolment, alcohol consumption, energy intake, physical activity (METs per week for moderate activity, vigorous work and strenuous sports), trends of hours for total sitting time, proportion contributed by other sitting behaviours, and smoking history by inclusion of the following variables: [smoking status, average number of cigarettes, average number of cigarettes squared, number of years smoked (time dependent), number of years since quitting (time dependent), and interactions between ethnicity and the smoking variables].

^bInteraction was assessed by a Wald test of the cross–product term of the two sitting activities being considered in models including total sitting time and the proportion of sitting time contributed by each type of sitting.

^cRestricted to subjects reporting any time sitting at work.

Discussion

In this large cohort study, sitting ≥10 h/day, compared with sitting <5 h/day, was associated with an increase in all-cause mortality of 11% for women, but no association was seen for men. However, among both men and women, all leisure-time sitting activities were associated with an increased risk, and the risk was strongest with TV watching. The observed increased risk of all-cause mortality among participants sitting watching TV ≥5 h/day was 19% in men and 32% in women, compared with watching TV <1 h/day. Sitting ≥3 h/day vs <1 h/day in other leisure sitting activities or at meals was also associated with 6–7% and 22% increased risks of all-cause mortality, respectively, among both men and women.

An association between total daily sitting and all-cause mortality was examined in three previous cohorts.^5,13,27 In two studies, risks were increased by 19% (≥9 vs <3 h/day)⁵ and 45% (≥11 vs 0– < 4 h/day)²⁷ for all-cause mortality among healthy men and women who were between 50 and 71 years old⁵ or older than 45 years,²⁷ respectively. Another study, which included men and women aged 18–90 years who had cancer or heart disease, showed an increased risk of 1.54 ('sitting almost all the time' vs 'almost none'), which may have been inflated due to prior disease.¹³ There is a lack of consistency in the literature in the categories used for analyses of sitting time, which certainly could have contributed to inter-study differences in the strengths of the observed associations. The subjects in the present study were somewhat older at the time the information on sitting duration was collected than those in other studies which have reported higher HRs than ours.^13,27

In contrast to the studies that examined total daily sitting, there are several prior prospective studies that examined time spent sitting watching TV.^{5,6,11,14–16} Results from a meta-analysis of three prospective studies, including Whites only, showed a 13% increase in the risk of all-cause mortality for every 2 h of TV watching.^9,11,14,16 In three additional cohort studies, all-cause mortality was increased between 17% (≥6 vs <2 h/day) and 61% (≥7 vs <3 h/day) among men and/or women who sat the longest watching TV^5,15 or in leisure time generally, including watching TV.⁶ Sitting while watching TV has been associated with unhealthy eating, such as higher intake of high-fat and high-calorie foods, energy-dense drinks and lower consumption of fruit and vegetables.²⁸ In contrast, although not specifically focused on watching TV, a recent report suggested that sedentary leisure time behavior was an independent cardiovascular disease risk factor, regardless of snacking habits and physical activity.²⁹ In our study, increased risk of mortality with longer durations of sitting watching TV was observed across levels of diet, smoking and physical activity. It appears unlikely that reverse causality due to undiagnosed diseases inflated the risks in the current study, since the association between mortality and sitting watching TV was observed even after excluding deaths that occurred up to 5 years after entry into the study.

In this study, hazard ratios of all-cause mortality according to daily hours of sitting were weaker in men than in women, largely due to the stronger association between sitting watching TV among women. Previous cohort studies have shown inconsistent results by sex for sitting watching TV: one showed similar results to ours,⁶ another found no difference by sex¹⁶ and a third found stronger associations in men than in women.¹⁵ It is possible that men and women behave differently while sitting watching TV in terms of eating, MET level of sedentary activity or number of breaks in sedentary activity.

Some studies showed that prolonged leisure sitting increased cardiovascular mortality after adjustment for various confounders, including physical activity,^5,6,11,16 which is consistent with our finding. A cohort study of Japanese men and women reported on mortality from specific cardiovascular disease and found that mortality from ischaemic heart diseases (51%) and from cerebrovascular diseases (21%) was increased among men who sat longer than 4 h/day watching TV (vs <2 h/day), whereas there was no association among women.¹⁵ This is in contrast to our results that found elevated risks for these causes among men and women.

The relationship between sedentary behaviour and cancer mortality has been inconsistent, with some studies showing no association,^11,13,16 and some larger studies finding increased cancer mortality associated with longer sitting duration watching TV,⁵ or time spent at leisure.⁶ Our study also found a weak elevated risk of total cancer mortality associated with sitting watching TV, but not with overall sitting. Associations for mortality from specific cancers and sitting watching TV were similar, although a stronger risk with increased sitting was observed for prostate cancer. Results from a prospective study conducted in Japan suggested that risks for some cancers were increased in persons who sat longer watching TV: liver cancer among men and women, lung cancer among men and non-Hodgkin's lymphoma among women.¹⁵

We found no association of longer sitting at work with all-cause mortality. We did not collect information on specific sitting activities at work, and it is possible that work-related sitting activities contribute to a greater energy expenditure compared with other types of sitting activities. Reported estimates range from 1.5 METs (sitting for light office work) to 2.5 (sitting for moderate work), which are higher values than for other sitting activities such as sitting watching TV (1.3), sitting riding in a car or a bus (1.3) and sitting for eating (1.5).³ Although some previous studies examined the relation between mortality and occupational physical activity based on job category,^30–32 no prior prospective studies, to our knowledge, have examined the independent role of occupational sitting on mortality in a general population sample. However, longer occupational sitting was correlated with a higher prevalence of overweight and obesity among 1579 full-time employed men and women after adjustment for age, sex, occupation category and physical activity in a cross-sectional study.³³ A challenge in studying the relation of occupational sitting and mortality is controlling for socioeconomic status (SES) and health behaviours, since people of higher SES tend to have healthier lifestyle behaviours and to show a lower risk of mortality, although they also are more likely to work in sedentary occupations.

Increased risk for all-cause mortality with time spent sitting watching TV was consistent among subgroups defined by age, education, diabetes/hypertension prevalence, smoking status, BMI, physical activity and diet. All ethnic groups except Japanese Americans also showed an increased risk for all-cause mortality with time spent sitting watching TV. Previous cohort studies from Japan showed that men who were most sedentary had an increased risk for all-cause mortality, whereas being sedentary was not related to mortality among women.^12,15 It is possible that Japanese Americans might behave differently while watching TV compared with other ethnic groups, in such as taking more breaks or doing light activities while watching TV. This speculation should be examined in further studies.

Strengths of our study included the prospective design, large number of subjects, racial/ethnic diversity and ability to control for many potential confounders, including physical activity. Because of the large sample size and number of deaths, we were able to conduct stratified analyses. Also, since we had collected information on several types of sitting behaviour, we were able to examine the effect of each one separately.

However, a limitation is our reliance on self-reported sedentary behaviours. Questionnaires used to assess physical activity and sedentary behaviours in large-scale epidemiological studies are known to contain substantial errors.³⁴ Also it is possible that the time categories provided on the questionnaire were too broad to categorize individuals’ activities well. Validation of the MEC activity questionnaire was done by comparison of its estimate of energy expenditure with that from doubly-labelled water. We are unable to validate sitting times specifically using this technique. However, as sitting activities occupy an average of 47% of the time in a typical day in our population, it seems unlikely that good agreement on total energy expenditure would be seen without agreement on total sedentary/sitting time or if the duration categories were too broad to distinguish activities. We also compared the responses to this questionnaire with those to another study conducted by direct interview and administered at about the same time to several overlapping participants. The interviewers were trained to probe about different types of activities to ensure a full day was covered, so the level of enquiry was more intense. We found reasonable agreement between responses on sitting activities between the two separate methods. Therefore, indirect evidence points to the sitting times being reasonably accurate, with total sitting time likely to be complete.

Another limitation of this study is that we did not assess details about current employment status at baseline, factors which would affect the association between sitting at work and mortality. A weak inverse relation between hours sitting at work and all-cause mortality was found among men and women younger than 65 years at baseline when those who reported 'zero' hours sitting at work were included in the reference group, suggesting a healthy worker effect where those who have work longer at sedentary jobs are healthier and at lower risk of death. Therefore, in an effort to minimize classification bias of employment status, we included only subjects who reported some sitting time at work in that analysis. Although we were able to adjust for many potential confounders, such as smoking, education, physical activity and other sitting activities, we cannot rule out the possibility that residual confounding could lead to biased results, due to unknown confounders or imprecise adjustment, such as for SES.

Previous reports suggested that resting can suppress lipoprotein lipase activity in skeletal muscle,^35–37 resulting in metabolic consequences such as dyslipidaemia, insulin resistance, hypertension and obesity.^38–42 In addition, survey results have shown that leisure sitting time increased an average of 15 min/day from 2003 to 2009 in the USA,¹⁰ but that physical activity levels did not increase over the same period.^10,43 More research is essential on not only how to increase physical activity but also how to reduce the amount of leisure sitting time,^5,6 and on whether the pattern of sitting makes a difference.⁴⁴ Intervention studies to reduce and to break up sedentary time are reported to be effective in reducing total sedentary time and increasing energy expenditure,^45,46 and in decreasing postprandial glucose and insulin levels.⁴⁷ It seems likely that future investigations which incorporate both self-reports and device-based measures will improve assessments of time, postures and energy expenditure associated with various sitting activities.^34,48,49

In conclusion, we found that longer time spent sitting, especially sitting watching TV, was associated with an increased risk of all-cause and cardiovascular disease mortality, but not of cancer mortality, after adjusting for physical activity and several other potential confounders. The association with sitting watching TV was consistent in men and women; in Whites, African Americans, Native Hawaiians, and Latinos; and in individuals at different BMI and physical activity levels. Interestingly, longer sitting at work or in a car or bus was not associated with increased mortality risk. Our results suggest that the amount of time spent in voluntary sitting should be limited.

Supplementary Data

Supplementary data are available at IJE online.

Funding

This work was supported by the National Cancer Institute [R37 CA54281]. The funders had no role in the study design or conduct of the study; collection, management, analysis or interpretation of the data; or preparation, review or approval of the manuscript.

Conflict of interest: None declared.

KEY MESSAGES.

• This is the first study examining various types of sitting simultaneously.

• This study underscores the adverse effect of prolonged sitting watching television, doing other leisure activities and at meals on all-cause mortality; in contrast, sitting at work or during transportation was not associated with all-cause mortality.

• Sitting watching TV was associated with an increased risk of mortality from cardiovascular disease but not from cancer.