Bedtime negative affect, sleep quality and subjective health in rural China

Negative affect (NeA), one of the integral components of affective subjective wellbeing (SWB) [1], is associated with a wide range of adverse physical and mental health outcomes. For instance, stress and depression can increase the risk of premature mortality, coronary heart disease, disability, and other chronic disorders [2]. Fear and anxiety are related to phobias and other anxiety disorders and may also compromise immune functioning and create susceptibilities to stress-related physical disorders when combined with acute and chronic stress [3, 4]. Daily worry has been shown to be associated with heart rate volatility (e.g., high heart rate), which is a physiological risk factor for cardiovascular disease [5]. Anger and its poor management are associated with the etiology of heart disease [6] and some cancers [7]. Severe sadness and grief can also result in immunosuppression [4], and even suicide [8].

Prior research examining the NeA-health relationship has focused exclusively on an overall level of NeA. In these studies, overall NeA was commonly measured by asking a single retrospective question, like “During the past 12 months, would you say that you experienced a lot of stress, a moderate amount of stress, relatively little stress, or almost no stress at all?” [9], or using rating scales, such as the Positive and Negative Affect Schedule, where people rate the intensity of NeA over a specific period of time [10]. However, it should be noted that first, the lengthy reporting period used for assessing affective feelings is likely to produce data that are more analogous to global affective SWB than to more momentary affective states (which we used in this study), and the reported global affective states might be biased by several factors, such as the person’s current mood, personality traits, the sequence of questions, and limited recalled memories [11]. Second, the global retrospective measurements of NeA also shed no light on the relationship between time information over the day and the intensity of affective states, so the effect of the diurnal timing of NeA on health has been overlooked. This is a potentially serious neglect, largely because, as Kahneman et al. [12] mentioned, the quantitative information about time throughout the day and the intensity of affective states is potentially useful for assessing the burden of various diseases and the health consequences of negative emotions; investigating some social and environmental stressors; and evaluating policies and interventions. Therefore, research on the effect of the diurnal timing of NeA on health could also be meaningful and conducive to creating a finer-grained profile of the NeA-health relationship than only aggregating the NeA score into an overall level over a long period of time. However, to our best knowledge, no reported evidence directly looked at the effect of the diurnal timing of NeA on health.

Regarding the effect of the diurnal timing of NeA on health, NeA occurring at bedtime is likely to play a more critical role in shaping health than non-bedtime NeA, and sleep might mediate the NeA-health relationship. For example, research by Lancee et al. [13] among 64 participants from Netherlands compared the effects of daytime and nighttime sleep-related worry (e.g., I worried about the amount of sleep I am going to get) on sleep impairment, and found that nighttime sleep-related worry was significantly related to impaired sleep (e.g., difficulty falling asleep), while daytime sleep-related worry was not associated with sleep impairments. This indicates that nighttime NeA might play a more important role in predicting sleep-related problems and disrupting recovery process compared to daytime NeA. It is acknowledged that sleep plays a vital role in brain function and systemic physiology including appetite regulation, metabolism, and the functioning of immune, hormonal, and cardiovascular systems [14]. Impaired sleep and disrupted circadian rhythm by bedtime NeA could then cause various damaging consequences to health. For instance, poor sleep quality has been shown to be associated with impaired immune and metabolic function, obesity, mood disorders, cardiovascular disease and other chronic diseases, and increased risk of mortality [15, 16]. However, no published evidence has simultaneously examined in one study the effect of the diurnal timing of NeA on health by analyzing whether bedtime NeA is a more important determinant of poorer health than non-bedtime NeA, as well as the mediating role of sleep quality in the NeA-health relationship.

Much of the research that attempts to demonstrate the NeA-health relationship has been cross-sectional rather than longitudinal [9, 10]. The cross-sectional models might restrict the ability to identify the potential causal direction underlying the NeA-health association due to unobserved between-person heterogeneity (e.g., personality traits) [17]. Ambrona and López-Pérez [18] conducted a longitudinal analysis of the relationship between NeA and health wherein physical health was subsequently measured one month and one year later after NeA. However, Delongis et al. [17] argued that in such research designs, the time between the initial measurement of NeA and the subsequent measurement of health status was widely spaced and not monitored, it is therefore difficult and perhaps impossible to disentangle what has been going on not only psychobiologically, but also environmentally. Therefore, Delongis et al. [17] suggested that it might be better and useful to use a within-person (intraindividual) model of analysis rather than an across-person (interindividual) model with multiple measures of NeA levels and health obtained over time when examining the NeA-health relationship. In a longitudinal, within-person analysis, it is possible to observe whether fluctuations in NeA levels covary with changes in health with the effects of between-person differences being eliminated. Based upon above, the fixed effect model (which was used in this study) may be an appropriate approach to achieve the goal of examining the longitudinal, within-person NeA-health relationship. One of the superiorities of fixed effect models over other statistical approaches (e.g., random effect models) is that using each individual as his or her own control, it controls for easily measured variables such as gender, race, ethnicity, and region of birth, as well as for other variables that are difficult to measure such as intelligence, personality trait, and genetic makeup; thus, the bias in the estimate of the NeA-health relationship may be reduced [19].

For relevant SWB studies from China, most of them either exclusively used cross-sectional data and focused on reported evaluative SWB (e.g., life satisfaction and global happiness) [20, 21], overlooking the impact of affective SWB (e.g., negative emotions) on health, or only used health as one of the control variables in statistical models when examining the association between SWB and other aspects such as income [22]. To our knowledge, no prior studies have investigated the longitudinal, within-person NeA-health relationship, as well as the mediating effect of sleep quality, in China. The present study targeted rural Chinese for several reasons. First, according to the latest census, rural Chinese comprised over one-third (36.11%) of the entire Chinese population in 2021 [23], but the current SWB literature provides limited evidence regarding this subgroup. An analysis of Chinese SWB that excludes the rural residents is therefore far from comprehensive. In addition, it is acknowledged that China has a remarkable rural–urban disparity, in terms of the socioeconomic status and the entitlement to social welfare [24], which has been shown to have long term consequences in terms of mental wellbeing [25, 26]. Compared with their urban counterparts, Chinese rural dwellers are living in greater socioeconomic disadvantage. Conducting such research among rural Chinese can provide some evidence to shift policy priorities to improve rural living standards, health status and quality of life.

In brief, this study set out to examine the effects of the diurnal timing of NeA (both bedtime NeA and non-bedtime NeA) on subjective health, and the mediating role of sleep quality in the NeA-health relationship, using a longitudinal, within-person design in rural China. We hypothesize that in rural China, bedtime NeA is a stronger determinant of poorer health than non-bedtime NeA, and the deleterious effects of NeA on health would be mediated by sleep quality.

This study is amongst the first to examine the impact of the diurnal timing of NeA including bedtime NeA and non-bedtime NeA on subjective health, and the mediating role of sleep quality in the NeA-health relationship, using a longitudinal, within-person design in rural China. The core research questions addressed by this study were as follows: in rural China, (a) is bedtime NeA a stronger determinant of poorer health than non-bedtime NeA? (b) To what extent the effect of NeA on health can be mediated through sleep quality?

For the first research question, when only controlling for time-variant covariates (from model 1 to model 3), within-person analysis revealed that non-bedtime NeA was unrelated to health, while bedtime NeA was found to be predictive of health, suggesting that bedtime NeA is a stronger predictor of poorer health than non-bedtime NeA. No reported studies have directly analyzed the impact of the diurnal timing of NeA on health; however, some studies examining the relationship between the timing of cortisol (a biochemical marker of chronic stress) and health could provide some enlightenment. For example, a prospective study by Kumari et al. [36] in a sample from Whitehall II study demonstrated that bedtime cortisol was predictive of subsequent cardiovascular-related mortality, while waking cortisol levels were unrelated to subsequent mortality. Tene et al. [37] reported consistent results among 182 cognitively intact ischemic stroke patients that higher bedtime cortisol levels were associated with larger neurological deficits, brain atrophy, and worse cognitive results, while post-awakening cortisol levels were not associated with any neuroimaging findings or cognitive scores. The present study consistently mirrored the cortisol-health patterns found in previous studies [36, 37].

The second research question was to what extent the effect of NeA on health could be mediated through sleep quality. After further controlling for sleep quality, the coefficient for bedtime NeA was no longer statistically significant, suggesting that the deleterious effect of bedtime NeA on health was mediated by impaired sleep quality. However, the potential mechanism underlying this phenomenon warrants further empirical investigation. Studies have shown that NeA, such as stress, is associated with increased heart rate reactivity, increased systolic blood pressure, and reduced vagal tone [38, 39]. Reduced vagal tone has been shown to be associated with increased sleep disruptions and poor sleep quality [40, 41]. Bedtime NeA which occurred right before sleep could result in these cardiac autonomic changes but might leave people with no time to recover from these physiological changes before going to sleep. Therefore, compared with non-bedtime NeA, bedtime NeA was more likely to directly undermine sleep quality (as shown in Table 3) and recovery processes, such as by leading to heightened pre-sleep arousal. Sleep plays a critical role in brain function and systemic physiology, including metabolism, appetite regulation, and the functioning of immune, hormonal, and cardiovascular systems [14]. Disruption of circadian rhythm and sleep deprivation have been shown to accelerate the risk of ill health, and even mortality [42]. In such a case, the deleterious impact of bedtime NeA on health might be explained by impaired sleep, such as poor sleep quality in this study. However, this study sheds no light on the potential physio-biological mechanisms linking bedtime NeA, poor sleep quality, and undermined health; thus, the above explanations should be treated with caution.

This study suggests that the standard method of calculating the NeA score through a (diurnal) time-weighted aggregated activity negative affect score may not be as important for future health as the bedtime negative affect score. The standard time-weighted aggregated measure may under-estimate the effect of bedtime NeA on future health, which suggests researchers, at the very least, should give consideration to bedtime values of NeA, rather than solely relying on overall time-weighted measures of NeA. Therefore, another significance of this study is about the importance of measuring bedtime NeA separately from the standard time-weighted approach.

Our findings could inform future interventions to focus more on reducing bedtime NeA so as to protect and promote sleep quality, thereby improving health levels and quality of life. For example, research by Sun et al. [11] demonstrated that some social environmental factors, such as the type of activity (e.g., shopping, watching TV) and quality of social interaction, represented by the friendliness felt towards the interacting partners, have a significant effect on the diurnal rhythms of people’s affective experiences in daily life. It is possible that the affective feelings could change in response to stimulation from the environment. Therefore, improving some social environmental factors, such as organizing and participating in pleasant activities and creating a friendly interactive environment in daily life, could effectively contribute to the decline in NeA, especially when NeA is high at bedtime, thereby preventing people from impaired sleep quality and then improving health levels.

Several limitations should be considered. First, the data were collected from four counties of three provinces in China; thus, the findings might not be generalized to all rural Chinese or other populations. Second, the present study only took into account a limited number of variables to examine the NeA-health relationship. Some other factors, such as healthy behaviors, utilization of health service, and some physiological markers, might also need to be considered in order to gain a comprehensive understanding about the NeA-health relationship. Third, the present study only focused on the global self-rated health. Different dimensions of health status, such as the health-related somatic symptoms, might also need to be evaluated in order to depict a panorama of the NeA-health relationship. Fourth, the present study employed DRM to measure people’s NeA. However, as the DRM still to some extent relies on retrospective self-reports, it could possibly produce some general methodological problems. For example, individuals with higher levels of cognitive ability may be better able to reconstruct their day, whereas participants of lower ability may forget what they were doing or how they felt about the activity [43]. Fifth, there is a potential overlap between the measures of NeA and subjective health, while they tap into different dimensions: self-reported health—a measure of general health and well-being, and NeA—a measure of negative affect. For example, an individual experiencing a deterioration in mental health might report lower scores on both measures. This overlapping nature might introduce a challenge in discerning the distinct contribution of NeA to the observed health outcomes. Furthermore, while the longitudinal, within-person analysis employed in this study is suitable for discerning the potential causal direction underlying the NeA-health association, it is not without limitations. For example, although this approach can effectively handle time-invariant covariates (and confounders), it falls short in handling unobserved time-variant factors. Finally, the calculation of bedtime NeA was based on participants’ negative feelings they experienced about the activity right before their last activity “went to sleep for the night”. Given that participants may be involved in different bedtime activities with varying durations, there is no established standard to determine the timeframe considered as bedtime NeA.

The present study uniquely sheds light on the effect of bedtime NeA and non-bedtime NeA on subjective health, and the mediating role of sleep quality in the NeA-health relationship using a longitudinal, within-person design in rural China. We found that bedtime NeA is a stronger predictor of poorer health than non-bedtime NeA, and the deleterious influence of bedtime NeA on health seems to operate through poor sleep quality. Therefore, interventions to reduce bedtime NeA could potentially improve subsequent sleep quality, thereby protecting people to some extent from impaired health status.