In the present study, for the first time in a nationally representative adult sample, we observed that blunted rest-activity rhythm, indexed by decreased levels of relative amplitude, was associated with all-cause mortality as well as CVD- and cancer-specific mortality. Those associations were independent of age, sociodemographic factors, health status and objective measurements of sleep behaviors. Furthermore, after excluding participants with diabetes, CVD, cancer or stroke at the baseline visit, our finding stays the same for the all-cause mortality (Supplementary Table 2), indicating that impaired robustness of rest-activity circadian rhythm was associated with increased risk of all-cause mortality even in the relatively healthy population. We further compared the prediction performance of the rest-activity circadian rhythm with sleep parameters and other traditional risk factors, and found relative amplitude to be predictive of mortality above and beyond traditional predictors with the exception of age. Our results indicate overall robustness of circadian rhythm is a strong predictor of mortality in the general population.
Our results are in consistent with the findings from the UK Biobank study [14] and the Rotterdam Study[25] for the association of rest-activity rhythm with all-cause mortality as well as the findings from the Study of Osteoporotic Fractures (SOF) in elderly women (mean age 84 years)[12] and the MrOS Sleep Study in elderly men (mean age 76 years)[11] for the association of rest-activity rhythm with all-cause mortality and CVD mortality. In the SOF study, adding total sleep time or sleep efficiency to the models did not change the associations between rest-activity rhythm parameters and mortality. Zuurbier et al. [25] specifically compared the prediction performance of rest-activity circadian rhythm parameters with both subjectively and objectively measured sleep parameters in the Rotterdam Study and observed that sleep measures were not related to mortality after adjustment for health parameters. In line with their results, we also observed that relative amplitude, the index of the overall robustness of the rest-activity circadian rhythm, was independently and more strongly associated with mortality than sleep parameters. This suggests that although sleep/wake cycles are intertwined with the circadian system, the circadian rhythm has its independent role in the risk of mortality.
The mechanism underlying the association of rest-activity circadian rhythms with mortality cannot be interpreted based on these observational data. However, animal and human laboratory studies have shown that the number of neurons in the suprachiasmatic nucleus (SCN), where the endogenous clock locates, decreased along with age [26–28]. This may lead to a decreased amplitude of circadian rhythm, and a decreased ability of the SCN to drive synchronization, causing a complex deviation from homeostasis, and driving risks for mortality with the aging process. In addition to the normal aging process, disrupted rest-activity circadian rhythm has also been suggested to play a role in multiple disorders such as diabetes [29], depression [30], dementia [31–33] and cancer [34] as well as pro-inflammatory status [6, 9] and pro-diabetes [10], indicating disrupted rest-activity rhythm may also serve as a biomarker for accelerating biological aging. On the other hand, rest-activity rhythm parameters also reflect a combination of the interaction between the endogenous circadian rhythm and rhythmic environmental and behavioral factors. It has been suggested that the rest-activity circadian rhythm could be modified through environmental and lifestyle changes [35, 36]. Future studies are warranted to examine whether environmental and lifestyle interventions or pharmacological treatments may correct the disrupted circadian rhythm and thus reduce the risk of mortality.
Our study has several strengths. NHANES population is from a nationally representative sample, which increases the generalizability of our findings. The linked mortality data is through death certificate record from the National Death Index, with minimal missing on the link. The large number of events for the outcomes studied and well-characterized data on multiple risk factors and confounders, provide us the opportunity to comprehensively compare the prediction effect of rest activity rhythm parameters with the traditional risk factors on mortality risk. Nevertheless, there are limitations. First of all, changes in rest-activity circadian parameters over time were not accounted for in the analyses, which might have underestimated the magnitude of the associations (regression dilution bias). Second, there may be other residual confounding factors that were not included, although we have adjusted for multiple confounding factors at baseline. Third, we did not observe consistent associations between rest-activity circadian parameters and other cause specific (except CVD and cancer) mortality, and this may due to the limited sample size of the cause-specific death events.
In summary, we observed in a nationally representative adult sample, participants with disrupted rest-activity circadian rhythms had higher risk of all-cause, CVD and cancer mortality. Future studies will be needed to test whether interventions (e.g. the arrangements of physical activity during the day/night, the sleep/wake cycle, and bright light exposure) that regulate rest-activity circadian activity rhythms will improve health outcomes.