Independent and joint association of physical activity and sedentary behavior on all-cause mortality

Introduction

The lack of physical activity (PA) is a primary risk factor for mortality.^[1] Previous studies indicated that over 5 million deaths worldwide each year are due to inadequate PA.^[2] High amounts of sedentary behavior (SB) are associated with increased risks for chronic diseases and mortality.^[3–5] Furthermore, self-reported SB in several domains such as sitting,^[6] driving a car,^[7] and TV viewing,^[8,9] is positively associated with mortality. A study on the dose-response association between sitting time and both all-cause and cardiovascular disease mortality revealed that people who reported sitting almost the whole time had a 54% higher risk than those who reported sitting almost no time.^[10] SB was nearly responsible for 3.8% of all-cause mortality based on a meta-analysis involving 54 countries.^[11] Unfortunately, SB is very widespread. According to objective monitoring, adults in western countries spend an average of 9 to 11 h for daily sedentary.^[12,13] Meanwhile, several studies emphasized that a large proportion of the global population still present with high levels of SB and low levels of PA.^[14–16] A multi-center study conducted in ten countries including China found that the mean sedentary time of adults aged 18 to 66 years was 8.7 h/day.^[17] The largest study involving 1,005,791 participants to date examined the joint associations of SB and PA with mortality, and the result indicated that the relative risks (RRs) associated with SB were higher among the population who were physically inactive, and high levels of PA might attenuate the harmful effects associated with prolonged SB.^[18]

To our knowledge, the associations of PA and SB with mortality have been examined in quite a few studies from western countries. However, the availability of similar data is rare among Chinese population. Therefore, in the present study, we aimed to prospectively examine the independent and joint associations of PA and SB with all-cause mortality in Jiangxi Province of China.

Methods

Results

The present study included 11,744 eligible participants. The average age of all participants was 58.9 ± 13.3 years; 4813 were men (41.0%). Mean sedentary time was 3.8 ± 2.6 h. There were 1781 (15.2%), 3171 (27.0%), and 6792 (57.8%) participants engaged in low, moderate, and high PA, respectively. The characteristics of participants are presented in Table 1. SB (<6 h, ≥6 h) and PA (low to moderate, high) were crosswise combined into four groups of high PA and SB < 6 h, high PA and SB ≥ 6 h, low to moderate PA and SB < 6 h, and low to moderate PA and SB ≥ 6 h. Compared to the other three groups, participants with low to moderate PA and SB ≥ 6 h had the highest age, VAI, SB, SBP, DBP, the largest proportion of male, current smoking, hypertension, stroke, education ≥10 years, urban residents, unemployed, the lowest BMI, and the lowest proportion of married participants. Characteristics of participants stratified by SB (<6 h, ≥6 h) and PA (low to moderate, high) were also presented in Supplementary Table 1.

During an average of 5.4 years of follow-up (median: 5.6 years; interquartile range: 5.3–5.7 years), there were totally 796 deaths, including 133 stroke deaths, 240 cardiovascular deaths, 68 cancer deaths, 109 respiratory failure deaths, and 246 other deaths. The risk of all-cause mortality was lower among participants with high PA than those with low to moderate level (5.2% vs. 8.9%) and higher (7.8% vs. 6.0%) among participants with SB ≥6 h compared to those with <6 h.

Table 2 presented the independent association of PA and SB with all-cause mortality by the Cox proportional hazards regression analysis. There was a statistically positive linear association between SB (continuous) and all-cause mortality (HR: 1.04, 95% CI: 1.02–1.06). When SB <4 h group was assessed as the reference group, the HR (95% CI) values in groups of 4 h ≤ SB < 6 h, 6 h ≤ SB < 8 h, and SB ≥ 8 h were 0.97 (0.78–1.15), 1.28 (1.03–1.59), and 1.46 (1.17–1.83), respectively. There was no significant difference on HR values between SB < 4 h group and 4 h ≤ SB < 6 h group. Therefore, the SB was finally combined into <6 h group and ≥6 h group. High SB (≥6 h) was associated with a greater risk of all-cause mortality compared with low SB (<6 h); the HR (95% CI) was 1.37 (1.17–1.61) in the all-adjusted model (Model II). Compared with low PA group, the HR for all-cause mortality was not significant in the moderate PA group. Thus, SB was divided into low to moderate PA group and high PA group. After adjustment for pertinent covariates, participants with high PA had a lower risk of all-cause mortality than participants with low to moderate PA (HR: 0.75, 95% CI: 0.61–0.87). Sensitivity analysis showed that the independent association of PA and SB with all-cause mortality remained statistically significant in participants excluding those who died in the first 2 years of follow-up [Supplementary Table 2].

When PA and SB were assessed together, the Kaplan–Meier curves [Figure 1] showed that the cumulative hazards of all-cause mortality significantly differed among the four groups (log-rank P < 0.001). Based on RERI of −0.090 (95% CI: −0.175 to −0.005), AP of −0.054 (95% CI: −0.076 to −0.032), and S of 0.881 (95% CI: 0.874–0.889), the analyses revealed an interaction on the additive scale between SB and PA on all-cause mortality. There was a significant interaction between PA (low to moderate, high) and SB (<6 h, ≥6 h) on all-cause mortality. A stronger positive association between SB and all-cause mortality was found in low to moderate PA group (HR: 1.74, 95% CI: 1.18–2.28) compared with in high PA group (HR: 1.28 [95% CI: 1.04–1.57], P_interaction = 0.023) [Supplementary Figure 1].

Table 3 showed that the highest risk of all-cause mortality was in the group with prolonged SB (≥6 h) and inadequate PA (low to moderate), with an HR of 1.67 (95% CI: 1.35–2.06). Among participants with high PA, compared with SB < 6 h, prolonged SB (≥6 h) was still associated with a higher all-cause mortality (7.0% vs. 4.9%; HR: 1.33, 95% CI: 1.12–1.56). Consistent results were found among participants excluding those who died in the first 2 years of follow-up [Supplementary Table 3].

Discussion

In recent years, the association of PA and SB with health outcomes is an object of intense concern. Our study lends further evidence to the increasing literature pointing to PA and SB as independent predictors of all-cause mortality. We found that Chinese participants with adequate PA (high) had lower mortality risk, while those with prolonged SB (≥6 h) had higher mortality, and participants with prolonged SB (≥6 h) combining with inadequate PA (low to moderate) had the highest risk of all-cause mortality.

Previous studies demonstrated the inverse association of self-reported PA with all-cause mortality. Samitz et al^[30] pooled data of 1,338,143 participants and found that RR for mortality was 0.65 (95% CI: 0.60–0.71) comparing highest with lowest levels of PA. Investigating different intensity categories of PA with all-cause mortality, Lollgen et al^[31] reported highly active men and women had a 22% (RR: 0.78, 95% CI: 0.72–0.84) and 31% (RR: 0.69, 95% CI: 0.53–0.90) lower risk of all-cause mortality, respectively. An increasing number of papers comprising the literature documented the positive association between self-reported SB and all-cause mortality. Biswas et al^[32] found a significant HR association of prolonged SB with all-cause mortality (HR: 1.24, 95% CI: 1.09–1.41). Matthews et al^[33] revealed that black adults with SB > 12 h/day had a 20% to 25% increased risk of all-cause mortality than those with SB < 5.8 h/day. In recent decades for China, there has been a substantial transformation from a labor-intensive lifestyle to more sedentary and physically inactive lifestyle, which is close to the Western lifestyle. However, the status of PA and SB and the associations with mortality risk were unclear.^[34] Our study made efforts in this regard. In agreement with the above results mainly in Western countries, consistent conclusions were found in our study among Chinese population.

Similar results were also reported in prior studies relying on objective measurements. A study on 7-day waist-worn accelerometry data of 1906 participants aging ≥50 years from the U.S. nationally representative National Health and Nutrition Examination Survey 2003 to 2004 with 2.8 years of follow-up, indicated that participants in the fourth of SB quartile had a 5.94 times higher risk of mortality than those in the lowest quartile.^[35] Lee et al^[36] reported that among 16,741 women with PA measured by wearing a triaxial accelerometer on the hip for 7 days, during a follow-up of 2.3 years, there was a 60.0% to 70.0% of risk reduction for the fourth quartile vs. the first quartile of PA. A subset of the European Prospective Investigation Into Cancer and Nutrition-Norfolk study prospectively investigated 5249 adults 40 to 79 years of age wearing an accelerometer on the right hip for 1 week; strongly inverse and positive linear associations were, respectively, observed for PA and SB with mortality.^[37] Comparing to accelerometry measurements, self-report measurements have not quantified PA and SB, but they had longer follow-up durations to reflect long-term patterns and could be accessed easily. The two assessment approaches overlapped, but each would provide unique information.^[38] Further studies should effectively integrate them to explore a more comprehensive estimation of PA, SB, and their associations with all-cause mortality.

Although recommendations on minimizing SB have begun to appear in public health guidelines,^[5] few quantitative guidelines exist for SB. Previous cut-off duration of daily SB required to minimize mortality was inconsistent. A meta-analysis including 13 studies (all self-reported measures) indicated that ≥4 h/day in SB resulted in an increased risk of all-cause mortality.^[18] Based on six studies (five self-reported vs. one device-based), Chau et al^[4] revealed that SB of >7 h/day was associated with increased mortality risk. Basing on studies with device-based measures, Ku et al^[39] reported that the appropriate daily SB cut-off duration in adults was around 9 h. In our study, we found the significant cut-off duration of self-reported SB was 6 h/day. After considering the report indicating that the subjective assessment might lead to an underestimation of daily SB in the range of 2 to 3.5 h,^[40] it is our inference that the specific duration of SB for recommendations in China needs further study.

Some mechanisms may explain the association of SB and PA with health. SB has been significantly associated with metabolic risk factors including fasting glucose, triglycerides, and high-density lipoprotein cholesterol,^[41,42] which may be a part of the explanation for the higher risk of mortality. Moreover, SB would lead to mitochondrial dysfunction, dysregulation of cellular redox status, and increased inflammation, and also may alter mitochondrial DNA deletions or mutations and increase reactive oxygen species-mediated toxicity. Finally, this chain of factors results in cellular senescence and cell death.^[43] Nevertheless, more study is necessary for the mechanisms underlying the harmful influence of SB on health. Possible mechanisms underlying the benefit of PA on mortality involve the favorable alterations in glucose tolerance, lipid levels, blood pressure, and BMI.^[44,45] Recent studies have demonstrated the direct vascular deconditioning and conditioning effect of PA, which may contribute to decreased health risk.^[46] Furthermore, PA has other potential benefits on health covering attenuation of plaque progression, improvement of endothelial cell function, reduction of myocardial oxygen and thrombosis, stabilization of vulnerable plaques, and strengthened collateralization.^[47,48] This may explain how adequate PA attenuates the harmful influence of SB.

The strengths of our study include a Chinese population-based design, a large random sample size, and strict adjustment to minimize residual confounders. Several limitations should be noted. First, we did not objectively quantify PA and SB, so the association of PA and SB with mortality might be underestimated in questionnaire- based studies, possibly due to recall bias when relying on imprecise PA and SB self-reports.^[20,49,50] Second, although a range of confounding covariates we adjusted, the influence of potential confounding effects could not be excluded entirely. Third, all participants included in our study were from southern China, so it would be difficult to generalize to other populations.

Conclusions

A significant and independent association of prolonged SB on all-cause mortality appeared among participants with high PA. High amounts of PA effectively attenuate the risk of SB on mortality. In conclusion, PA and SB have been among the leading modifiable risk factors for all-cause mortality not only in Western countries but also in China. Reduction of SB will be an effective strategy, ancillary to increasing PA, for preventing all-cause mortality in physically inactive or sedentary populations.

Acknowledgements

The authors thank to all the investigators and subjects who participated in the Chinese Hypertesion Survey.

Funding

This work was supported by grants from the National Natural Science Foundation of China (No. 81760049), the Jiangxi Science and Technology Innovation Platform Project (No. 20165BCD41005), the National Key R&D Program of China (No. 2018YFC1312902), the Science and Technology Plan of Health Commission of Jiangxi Province (No. 20185215), and the Key Project of Education Department of Jiangxi Province (No. GJJ170013).

Conflicts of interest

None.

Supplementary Material