Search and selection
The search included 10874 potentially relevant articles (1301 from Medline, 5190 from EMBASE, 2803 from Web of Science, 184 from CINAHL, 160 from Lillacs, 154 from SportsDiscus, 936 from PsychInfo, and 146 from Sedentary Behavior Research Database). Fourteen additional records were selected from the articles suggested by the SBRN members (Figure 1).
After removing duplicate records, a total of 9768 articles remained. After screening titles and abstracts, 56 full papers were read in their entirety. In addition, 2 articles were found in the reference list of these full papers (an additional 787 titles were screened). Of the 58 articles, only 23 met the inclusion criteria and were included in the review. The complete list of included and excluded articles is presented in the Additional file 3.
Methodological quality assessment
Additional file 2: Table S1 presents the quality assessment of the 23 articles included in the review. Of the 23 articles included, 16 (70%) were cross-sectional studies, [25–40] 1 (4%) was a case–control study, [41] and 6 (26%) were prospective cohort studies [42–47]. Concerning quality of the evidence, 12 (52%) were evaluated as very low, [25, 26, 29–36, 38, 42] 5 (22%) as low, [39, 41, 43, 44, 48] 4 (17%) as moderate, [28, 37, 40, 45] and 2 (9%) as high quality evidence [46, 47].
Risk of selection bias was identified in 9 articles (39%), [25, 28, 29, 31–34, 37, 42, 47] and information bias due to self-reported instruments was found in 20 articles (83%) [25–27, 31–34, 37, 38, 40–47]. Indirectness (surrogate outcomes) was used in 16 articles (70%), [25–37, 39–41] imprecise results were presented in 14 (61%) articles, [25–27, 30–32, 34–39, 41, 42] and an inconsistent [25] result among subgroups was found in 1 (4%) article. Most of the articles (n = 20 – 87%) received an additional point for the adjustment of potential confounders [25–29, 32–39, 41–47]. Eleven (48%) studies gained a point for magnitude of effect [30, 36, 39, 8–42, 44–47] and 5 (22%) for considering a dose–response relationship [36, 39, 42, 45, 46]. Further details concerning the quality assessment of each article are presented in the Additional file 4.
Sedentary behavior—health outcomes
Mortality
Four prospective cohort studies, [43–46] classified as low, [43] moderate [44] and high quality, [45, 46] investigated the relationship between sedentary behavior and mortality (all-cause, cardiovascular, colorectal cancer, other causes).
Martinez-Gomez et al.’s [44] study showed that individuals who spent less than 8 hours sitting/day had a lower risk of all-cause mortality (HR = 0.70, 95% CI: 0.60 to 0.82) when compared with their sedentary peers. In addition, individuals who were physically active and less sedentary (<8 hours/day of sitting) showed a lower risk of all cause-mortality (HR 0.44; 95% CI: 0.36 to 0.52) than those who were inactive and sedentary.
Similarly, Pavey et al. [45] found a dose–response relationship between sitting time and all-cause mortality. Individuals who spent 8–11 hours/day (HR 1.35; 95% CI 1.09 – 1.66) and more than 11 hours/day sitting (HR 1.52; 95% CI 1.17 – 1.98) presented a higher risk of all-cause mortality than those who spent less than 8 hours/day sitting. For each hour/day spent sitting, there was an increase of 3% (HR 1.03; CI 95% 1.01-1.05) in the risk of all-cause mortality. Moreover, the risk of all-cause mortality of individuals who were physically inactive (less than 150 minutes/week) and spent 8–11 or more than 11 hours/day sitting increased by 31% (HR 1.31, 95% CI 1.07 to 1.61) and 47% (HR 1.47, CI 1.15 to 1.93), respectively.
In León-Munoz et al., [46] individuals were classified as consistently sedentary (>median in 2001 and 2003), newly sedentary (<median in 2001 and > median in 2003), formerly sedentary (>median in 2001 and < median in 2003), and consistently nonsedentary (<median in 2001 and 2003). They found that when compared with the consistently sedentary group, subjects newly sedentary (HR 0.91; 95% CI 0.76 - 1.10), formerly sedentary (0.86; 95% CI 0.70 - 1.05), or consistently non-sedentary (0.75; 95% CI 0.62 - 0.90) were protective against all-cause mortality.
Examining a colorectal cancer survivor population, Campbell et al. [43] identified that more than 6 hours per day of pre-diagnosis leisure sitting time, when compared with fewer than 3 hours per day, was associated with a higher risk of all-cause mortality (RR, 1.36; 95% CI, 1.10 to 1.68) and mortality from all other causes (not cardiovascular and colorectal cancer) (RR, 1.48; 95% CI 1.05-2.08). Post-diagnosis (colon cancer) sitting time (>6 hours) was associated with a higher risk of all-cause mortality (RR, 1.27; 95% CI, 0.99 to 1.64) and colorectal cancer-specific mortality (RR, 1.62; 95% CI, 1.07-2.44).
Metabolic syndrome
Three cross-sectional studies, [25, 35, 36] classified as very low [25, 35] and moderate [36] quality, investigated the relationship between sedentary behavior and metabolic syndrome.
Gardiner et al., [25] showed that individuals who spent most of their time sitting (highest quartile, >3 hours/day) had an increased odds of having metabolic syndrome (men: OR 1.57; CI 95% 1.02 – 2.41 and women: OR 1.56; CI 95% 1.09 – 2.24) when compared with their less sedentary peers (lowest quartile, <1.14 hours/day). In the same study, women who watched more television (highest quartile) increased their risk of metabolic syndrome by 42% (OR 1.42; CI 95% 1.01 – 2.01) when compared with those who watched less television per day (lowest quartile).
In the same sense, Gao et al. [36] showed that individuals in the highest quartile (>7 hours/day) of television watching/day, when compared with those in the lowest quartile (<1 hours/day), had an increased odds (OR 2.2, 95% CI 1.1–4.2) of having metabolic syndrome. In a dose–response relationship, for each hour of television watching/day, there was an increase of 19% in the odds (95% CI 1.1–1.3; p for trend 0.002) of having metabolic syndrome.
Bankoski et al. [35] found that a greater percentage of the time spent in sedentary behavior increased the risk of having metabolic syndrome (only quartile 2 vs. quartile 1, the hours/day of each quartile was not reported; OR 1.58; 95% CI 1.03 - 2.24), whereas breaks in sedentary time throughout the day protected against metabolic syndrome (only quartile 2 vs. quartile 1; OR 1.53; 95% CI 1.05 - 2.23).
Cardiometabolic biomarkers
Six cross-sectional studies, [25, 28, 33, 34, 36, 39] classified as of very low [25, 28, 33, 34] and of moderate quality, [36, 39] investigated the relationship between sedentary behavior and independent cardiometabolic biomarkers.
Triglycerides
The likelihood of having high triglycerides was higher in men (Odds Ratio (OR) 1.61; 95% CI 1.01-2.58) and women (OR 1.66; 95% CI 1.14-2.41) who were in the highest quartile of overall sitting time [25]. However, Gao et al. [36] and Gennuso et al. [39] showed that the association between time spent in sedentary behavior and high triglycerides was not statistically significant.
HDL cholesterol
Gao et al., [36] found that greater time spent viewing television was associated with low HDL cholesterol (2.5; 95% CI 1.0-5.9; p < 0.05). In a study by Gardiner et al., [25] women in the highest quartile of television viewing and men in the highest quartile of overall sitting time presented an OR for low HDL cholesterol of 1.64 (95% CI 1.06-2.54) and 1.78 (95% CI 1.78; 95% CI 1.05-3.02), when compared with the lowest quartile, respectively. However, Gennuso et al. [39] found that the relationship between time spent in sedentary behavior and low HDL cholesterol was not statistically significant (p = 0.29).
Blood pressure
When compared with the lowest quartile of overall sitting time, the OR for high blood pressure in the third quartile was 1.50 (95% CI 1.03-2.19) [25]. In Gao et al.’s [36] study, greater time viewing television was associated with high blood pressure (2.5; 95% CI 1.0-6.0; p < 0.05). However, Gennuso et al. [39] found that the relationship between time spent in sedentary behavior and systolic blood pressure (p = 0.09) and diastolic blood pressure (p = 0.32) was not statistically significant.
Plasma Glucose/ Hb1Ac/ Glucose intolerance
Gennuso et al. [39] demonstrated that greater television viewing and sedentary time was associated with higher plasma glucose (p = 0.04). In Gardiner et al.’s [25] study, this relationship was observed only in women (1.45; 95% CI 1.01-2.09; p < 0.05). However, Gao et al. [36] and Stamatakis et al. [28] found that the relationship between television viewing and high fasting glucose and Hb1Ac was not statistically significant.
Cholesterol ratio and total
Gao et al. [36] demonstrated that greater time in television viewing was associated with a high total-to-HDL cholesterol ratio (OR 2.0; 95% CI 1.1-3.7; p < 0.05). In Stamatakis et al.’s [28] study, self-reported total leisure-time sedentary behavior (β 0.018; 95% CI 0.005-0.032), television viewing (β 0.021; 95% CI 0.002-0.040), and objectively assessed sedentary behavior (β 0.060; 95% CI 0.000-0.121) were associated with cholesterol ratio. However, Gennuso et al. [39] found that the relationship between time spent in sedentary behavior and total cholesterol was not statistically significant (p = 0.50).
Other cardiometabolic biomarkers
The association between objectively measured sedentary time and pericardial fat [33] and coronary artery calcification [34] was not observed after adjusting for moderate to vigorous physical activity. Gennuso et al. [39] found a positive association between sedentary hours and C-reactive protein (p < 0.01).
Waist circumference/waist-to-hip ratio/abdominal obesity
Six cross-sectional studies, [25, 26, 28, 30, 36, 39] classified as being of very low [25, 26, 28, 30] and of moderate [36, 39] quality, investigated the relationship between sedentary behavior and waist circumference/waist-to-hip/abdominal obesity.
Gardiner et al. [25] and Gomez-Cabello et al. [30] found that sitting time increased the risk of abdominal obesity by 80% (OR 1.8; 95% CI 1.20-2.64) in both sexes and 81% in women (OR 1.81; 95% CI 1.21-2.70).
In Stamatakis et al.’s [28] study, television time (β 0.416; 95% CI 0.275 - 0.558) and total self-reported leisure-time sedentary behavior (β 0.234; 95% CI 0.129 - 0.339) were positively related to waist circumference. Gao et al. [36] found that greater time in television viewing was associated with high waist-to-hip ratio (3.9; 95% CI 1.08 - 8.4; p < 0.01). Gennuso et al. [39] found that more time spent in objectively measured sedentary behavior was associated with a high waist circumference (p < 0.01). In a colorectal cancer survivor population, [26] sedentary time was not associated with waist circumference.
Overweight/obesity
Six cross-sectional studies, [28–31, 36, 37, 39] classified as being of very low [28–31, 37] and of moderate [36, 39] quality, investigated the relationship between sedentary behavior and overweight/obesity.
Gomez-Cabello et al. [30] demonstrated that sitting more than 4 hours/day increased the risk of overweight (OR 1.7; 95% CI 1.06-2.82) and obesity (OR 2.7; 95% CI 1.62-4.66). In a similar study, Gomez-Cabello et al. [31] showed that being seated more than 4 hours/day increased the risk of overweight/obesity (OR 1.42; 95% CI 1.06-1.89) and overfat (1.4 OR; 95% CI 1.14-1.74) in women and the risk of central obesity (OR 1.74; 95% CI 1.21 – 2.49) in men.
Gennuso et al. [39] found that more time spent in objectively measured sedentary behavior was associated with higher BMI (p < 0.01). In Stamatakis et al.’s [28] study, self-reported leisure-time sedentary behavior (β 0.088; 95% CI 0.047 - 0.130) was associated with BMI.
Inoue et al. [37] found that when compared with the reference category (high television(TV)/insufficient moderate to vigorous physical activity (MVPA)), the adjusted ORs (95% CI) of overweight/obesity were 0.93 (95% CI 0.65-1.34) for high TV/sufficient MVPA, 0.58 (95% CI 0.37-0.90) for low TV/insufficient MVPA, and 0.67 (95% CI 0.47-0.97) for low TV/sufficient MVPA. Stamatakis et al. [28] also showed that TV time (β 0.159; 95% CI 0.104-0.215) was positively associated with BMI. However, only Gao et al. [36] found that greater time of television viewing was statistically significantly association with BMI (OR 1.4; 95% 0.7-2.8).
In the only study that evaluated sedentary behavior in transport, Frank et al. [29] showed that ≥1 hour/day sitting in cars was not associated with overweight (0.86 OR. 95% CI 0.51-1.22) or obesity (0.67 OR; 95 CI% 0.41-1.06).
Mental health (Dementia, mild cognitive impairment, psychological well-being)
Three cross-sectional studies, [32, 38, 40] one case–control, [41] and two prospective cohort studies, [42, 47] classified as very low [32, 41] and low quality [38, 40, 42, 47] investigated the relationship between sedentary behavior and mental health (dementia, mild cognitive impairment, and psychological well-being).
In Verghese et al.’s [47] study, individuals who frequently played board games (HR 0.26; 95% CI 0.17-0.57) and read (HR 0.65; 95% CI 0.43-0.97) were less likely to develop dementia.
Buman et al. [32] demonstrated that sedentary time was negatively associated with psychosocial well-being (β -0.03; 95% CI −0.05 - -0.01); p < 0.001. However, Dogra et al. [38] found that 4 hours or more of sedentary behavior per day was not associated with psychologically successful aging.
With regards to mild-cognitive impairment (MCI), reading books (OR 0.67; 95% CI 0.49-0.94), playing board games (OR 0.65; 95% CI 0.47-0.90), craft activities (OR 0.66; 95% CI 0.47-0.93), computer activities (OR 0.50; 95% CI 0.36-0.71), and watching television (OR 0.48; 95% CI 0.27-0.86) were significantly associated with a decreased odds of having MCI [40]. According to Geda et al.’s [41] study, physical exercise and computer use were associated with a decreased likelihood of having MCI (OR 0.36; CI 95% 0.20-0.68).
However, Balboa-Castillo et al. [42] found that the highest quartile of sitting time was negatively associated with mental health (β-5.04; 95% CI −8.87- -1.21); p trend = 0.009.
Cancer
Only one study, with moderate quality, found no association between time watching television or videos and renal cell carcinoma [27].