Journal Logo


Intraindividual Variability in Sleep and Levels of Systemic Inflammation in Nurses

Slavish, Danica C. PhD; Taylor, Daniel J. PhD; Dietch, Jessica R. PhD; Wardle-Pinkston, Sophie MS; Messman, Brett BA; Ruggero, Camilo J. PhD; Kohut, Marian PhD; Kelly, Kimberly PhD

Author Information
doi: 10.1097/PSY.0000000000000843
  • Free



Nurses are particularly likely to experience impaired sleep, given high levels of work stress and frequently rotating work/sleep schedules (1). Approximately 50% to 89% of nurses work some form of night shift per month (2–4), 29% to 57% meet the criteria for insomnia (5,6), and 12% report short sleep duration (<6-hour sleep/24 h). Night-shift work, insomnia, and short sleep are associated with increased risk for chronic diseases such as cardiovascular disease (7), depression (8), and type 2 diabetes (9), as well as all-cause mortality (10).

Inflammation is one plausible biological mechanism linking impaired sleep with morbidity and mortality. During a night of insufficient sleep or poor quality sleep, the sympathetic nervous system releases norepinephrine and epinephrine, which upregulate production of proinflammation biomarkers interleukin (IL)-6, C-reactive protein (CRP), IL-1β, and tumor necrosis factor α (TNF-α) (11,12). Studies have shown that night-shift work, insomnia, and short sleep duration are associated with higher levels of these biomarkers (13,14). Higher levels of systemic inflammation are in turn prospectively implicated in the development of a number of common chronic diseases, including cardiovascular disease and depression (15–17).

The primary limitation of previous studies on nursessleep and health is that typically sleep is measured with one or two survey questions or with retrospective questionnaires about recent or typical sleep (18,19). These types of measures are inadequate to assess the heterogeneity of sleep across different domains and across time. Sleep is composed of multiple domains, including efficiency, duration, and timing (20), each of which may be differentially associated with inflammation and health. Furthermore, no two nights of sleep are the same. Sleep fluctuates substantially from night to night (21), particularly in those who are engaged in rotating shift work (22). These within-person fluctuations are termed intraindividual variability (IIV) and can be quantified by computing the variation in sleep across time using prospective daily measures of sleep (e.g., sleep diaries or actigraphy).

Research has shown that greater IIV in sleep is associated with disease beyond the influence of mean sleep (for a review, see Bei et al. (21)). For example, greater IIV in sleep timing (i.e., bedtime and risetime), sleep duration, sleep quality, and sleep fragmentation has been associated with physical health conditions (e.g., diabetes, metabolic syndrome, chronic pain, obesity, and heart conditions) (23–26), higher body mass index (BMI) and weight gain (23), insomnia (27), and greater depressive symptoms and negative mood (26,28,29). Greater IIV in sleep timing, sleep efficiency (i.e., total sleep time divided by time in bed), and total sleep time is also associated with increased nocturnal norepinephrine (28,30), flatter cortisol diurnal slopes (31), higher levels of systemic inflammation (31–33), and greater allostatic load (31). Similarly, experimental circadian misalignment studies have demonstrated that greater IIV in sleep timing and duration is associated with elevated inflammation (34). However, no studies have examined associations between naturally occurring IIV in sleep (measured both subjectively and objectively) as a prospective predictor of inflammation in nurses. Nurses are particularly likely to experience sleep disruptions (2,6) and are often the first line of care in most hospital settings. Therefore, it is critical to better understand associations between disruptions in their sleep and markers of health.

Although greater IIV in sleep seems to be associated with poorer physiological function and health, it is likely that this association is stronger among night-shift working nurses. Night-shift workers experience dramatic shifts in their sleep and typically are in engaged in a 3 nights on/2 days off (or other similar) work schedule. This type of work/sleep schedule may result in a stronger “dose” of IIV in sleep than IIV in sleep owing to other factors (e.g., fluctuations in mood, daily work tasks or family responsibilities, and physical symptoms). Furthermore, epidemiological research and simulated shift work schedules in laboratory studies have shown that shift work is robustly associated with increases in inflammation (35–37). Together, this work suggests that night-shift workers may have stronger positive associations between IIV in sleep and inflammation than day workers.

The primary aim of this study was to examine how both mean and IIV in multiple facets of sleep (sleep diary- and actigraphy-determined sleep efficiency and total sleep time) were associated with markers of inflammation (IL-6, CRP, IL-1β, and TNF-α) in a sample of nurses. On an exploratory basis, we examined recent night-shift work status as a moderator of these associations, given the expectation that night-shift workers would have both more IIV in sleep and higher levels of inflammation.



This study was part of a larger parent investigation on the effects of sleep on antibody response to the influenza vaccine that occurred from September 2018 to November 2018. Participants were recruited from two regional hospitals through nursing staff presentations, notification through employee e-mail systems, and flyers that directed them to an initial online consent form. Four hundred sixty-one nurses provided online consent and were asked to complete initial online Qualtrics surveys to collect demographic information as well as retrospective self-report estimates of recent health. Participants were then invited to enroll in the main portion of the study in the early fall (i.e., the start of the influenza season), which included completion of in-person informed consent approximately 1 month later. At this time, participants were given instructions on completing the sleep diaries and wearing the actigraphy device, which they completed for the subsequent 7 days. On day 7, 392 participants reported to the hospital for a serum blood draw to assess inflammation biomarkers. All study procedures were approved by the Medical City Plano Institutional Review Board.


Inclusion criteria were as follows: a) not yet received the current season’s influenza vaccine, b) between the ages of 18 and 65 years, and c) registered nurses actively working at least part time at one of two regional hospitals. Exclusion criteria were as follows: a) pregnant/nursing or planning to become pregnant or b) having an egg allergy. Table 1 reports demographic characteristics for the entire sample and split by recent night-shift work status. Generally, participants matched national demographics of nurses in the United States (38): most participants were female (92%), White (78%), and non-Hispanic (89%). A significant portion of the sample also self-reported as Asian (10%) or African American (7%). A majority (63%) of the sample was currently married, and most had children (65%). Twenty-three percent of the sample reported working at least one night shift during the 7-day daily diary period and were classified as recent night-shift workers.

Participant Characteristics


Actigraphy-Determined Sleep

For 7 days, participants were instructed to continuously wear an Actiwatch Spectrum Pro (Philips Respironics, Bend, OR), a watch-like device used to infer objective sleep/wake patterns. Participants were asked to push an “event marker” button when they intended to go to sleep and when they got out of bed each morning. Actigraphy data were independently scored in Actiware software (Version 6.0.8) by two trained individuals using an in-laboratory protocol that systematically uses a combination of event markers, sleep diary data, activity data, and light levels. Discrepancies were resolved by a third person. Data were exported using default settings (10 immobile minutes for sleep onset and offset, medium wake threshold [40 activity counts]).

Sleep Diary–Determined Sleep

An electronic version of the Consensus Sleep Diary (39) was completed by participants each morning upon awakening using REDCap (40). Diaries were used to determine total sleep time (time in bed minus the sum of sleep onset latency, wake after sleep onset, and terminal wakefulness) and sleep efficiency (i.e., total sleep time divided by time in bed multiplied by 100). These particular measures were chosen to comprehensively assess both sleep duration and sleep quality, two important facets of sleep relevant for health (20). For each sleep measure determined by both sleep diaries and actigraphy, the intraindividual mean (iM) and intraindividual standard deviation (i.e., measure of IIV) were calculated across the 7 days for each participant. (For example, if a participant reported a total sleep time of 6, 7, 8, 5, 7, 9, and 7 hours across the 7 days, their intraindividual mean would be equal to 7 hours, and their intraindividual standard deviation would be equal to 1.29 hours.)


Serum blood was drawn by trained phlebotomists. All blood draws occurred between 7 am and 12 pm to control for circadian rhythmicity of inflammation. Samples sat for 60 minutes to clot and then were centrifuged at 3000 rpms for 30 minutes and aliquoted into cryovials. Samples were temporarily frozen on dry ice and then frozen at −80°C until assaying. All inflammation samples were assayed using high-sensitivity enzyme-linked immunosorbent assays from R&D Systems, Inc. (Minneapolis, MN) within 1 year after collection. The lower limit of detection (LLD) for CRP was 0.010 ng/ml, 0.033 pg/ml for IL-1β, 0.022 pg/ml for TNF-α, and 0.039 pg/ml for IL-6. All IL-6 and TNF-α samples were within detectable limits. For IL-1β, 123 samples (31%) were below detectable limits. The standard curve revealed that these were not missing data but were instead indicative of very low values; therefore, in alignment with previous research (41,42), half the LLD was imputed (i.e., 0.0165 ng/ml). For CRP, one sample was outside of detectable limits; therefore, half the LLD was imputed (i.e., 0.005 ng/ml). Intra-assay coefficients of variation were 1.60% for CRP, 2.29% for IL-1β, 2.45% for TNF-α, and 3.26% for IL-6. Interassay coefficients of variation were 6.73% for CRP, 6.04% for IL-1β, 5.06% for TNF-α, and 7.21% for IL-6.

Recent Night-Shift Work Status

In the daily sleep diaries, participants reported whether they worked a night shift the previous day (“Did you have to be at work past 9 PM OR before 6 AM?”). Those who said “yes” to this question at least once during the 7 days were classified as a recent night-shift worker. Those who answered “no” to this question across all 7 days were classified as a day worker.

Baseline Covariates

At baseline, participants reported on their age, sex (0, female; 1, male), race/ethnicity (0, any other race; 1, White), height, weight, and health behaviors. BMI was calculated based participants’ self-reported height and weight according to the Centers for Disease Control and Prevention guidelines (43).

Analysis Plan

All data were screened for missingness, outliers, and violation of assumptions (e.g., excessive skew). As expected, inflammation biomarkers were not normally distributed (CRP skewness before half the LLD was imputed = 0.93, kurtosis = 3.22; IL-6 skewness = 3.47, kurtosis = 18.04; IL-1β skewness before half the LLD was imputed = 7.16, kurtosis = 63.43; TNF-α skewness = 12.76, kurtosis = 179.95). Therefore, as previously described, for IL-1β and CRP, values were imputed for samples below the LLD, and then all biomarkers were either square root or log-10 transformed to resolve nonnormality for each biomarker (square-root CRP skewness after half the LLD was imputed = 0.21, kurtosis = 2.15; log-10 IL-6 skewness = 0.39, kurtosis = 3.32; log-10 IL-1β skewness after half the LLD was imputed = 0.35, kurtosis = 2.86; log-10 TNF-α skewness = 0.38, kurtosis = 13.68).

Analyses were run in R, an open source-statistical program (44). Regression analyses were run using R base packages and the package apaTables (45), and all analyses covaried for age, sex, BMI, and shift work. Controlling for time of the day the blood samples were collected did not change the significance or pattern of results; therefore, this variable was not included as a covariate. Semipartial correlations squared (sr2; i.e., the unique contribution to the total variance in the dependent variable explained by each independent variable, after accounting for the variance explained by the other independent variables) were used as a measure of effect size. Cohen’s (46) heuristics were used as determination of sr2, where a small sr2 = 0.02, a medium sr2 = 0.13, and a large sr2 = 0.26. A power analysis revealed that for a multiple linear regression with eight variables in the model, a sample size of n = 392 (α = .05) yielded an 81% power to detect a small to medium effect size (f2 = 0.04).


Descriptive Results

Average sleep diary compliance was 83% (mean [standard deviation, or SD] = 5.79 [2.52] out of seven possible sleep diaries completed per participant). Compared with day workers, night-shift workers were more likely to be younger and Hispanic/Latinx (Table 1). Night-shift workers also had shorter diary- and actigraphy-determined mean total sleep time and lower sleep efficiency, as well as greater variability in diary- and actigraphy-determined total sleep time and sleep efficiency (Table 1). Approximately 24.4% of the 90 night-shift workers (n = 22) had been engaged in some type of night-shift work for <1 year; 38.9% (n = 35), for 1 to 5 years; and 36.7% (n = 33), for 5 to 30 years. Night-shift workers worked an average (SD) of 3.11 (1.44) night shifts across the 7 days (range = 1–7; median = 3). There were no significant correlations between total number of night shifts in the past 7 days and inflammation biomarkers among shift workers (IL-6: r = 0.19, p = .075; IL-1β: r = −0.002, p = .99; TNF-α: r = −0.07, p = .52; CRP: r = 0.09, p = .39). Shift workers reported working an average (SD) of 0.75 (1.24) successive night shifts before the morning of the blood draw. There were no significant correlations between successive number of night shifts worked before the blood draw and inflammation biomarkers among shift workers (IL-6: r = 0.04, p = .72; IL-1β: r = 0.01, p = .90; TNF-α: r = −0.16, p = .15; CRP: r = −0.07, p = .49). Bivariate correlations between all study variables for the entire sample are presented in Figure 1, and bivariate correlations presented separately for shift and day workers can be found in Figure 2. In general, inflammation biomarkers were more strongly associated with sleep parameters in shift workers compared with day workers (Figure 2).

Bivariate correlations for the entire sample (n = 392). Numbers represent Pearson correlation coefficients. CRP = C-reactive protein (square-root transformed); IL-6 = interleukin-6 (log-10 transformed); diary = sleep diary; SE = sleep efficiency; iSD = intraindividual standard deviation; TST = total sleep time; iM = intraindividual mean; acti = actigraphy; Shift worker = based on sleep diary/actigraphy (day worker, 0; night-shift worker, 1); Race = 0 for any other race, 1 for White race; Gender = 0 for female, 1 for male); BMI = body mass index; TNF-α = tumor necrosis factor α (log-10 transformed); IL-1β = interleukin-1β (log-10 transformed). Color image is available online only at
Bivariate correlations by shift work status. Numbers represent Pearson correlation coefficients. Day workers (n = 302; top figure), night-shift workers (n = 90; bottom figure). IL-1β = interleukin-1β (log-10 transformed); CRP = C-reactive protein (square-root transformed); IL-6 = interleukin-6 (log-10 transformed); diary = sleep diary; SE = sleep efficiency; iSD = intraindividual standard deviation; TST = total sleep time; iM = intraindividual mean; acti = actigraphy; Race = 0 for any other race, 1 for White race; Gender = 0 for female, 1 for male); BMI = body mass index; TNF-α = tumor necrosis factor α (log-10 transformed). Color images are available online only with this article at

Intraindividual Mean Sleep and Inflammation


Neither iM actigraphy-determined total sleep time nor sleep efficiency was associated with CRP, IL-6, IL-1β, or TNF-α (Table 2), controlling for age, sex, BMI, and shift work.

Unstandardized Estimates of the Associations Between Actigraphy-Determined Total Sleep Time and Sleep Efficiency With Plasma Inflammation Biomarkers CRP, IL-6, TNF-α, and IL-1β

Sleep Diary

Neither iM sleep diary-determined total sleep time nor sleep efficiency was associated with CRP, IL-6, IL-1β, or TNF-α (Table 3), controlling for age, sex, BMI, and shift work.

Unstandardized Estimates of the Associations Between Sleep Diary–Determined Total Sleep Time and Sleep Efficiency With Plasma Inflammation Biomarkers CRP, IL-6, TNF-α, and IL-1β

IIV in Sleep and Inflammation


Greater IIV in actigraphy-determined total sleep time was associated with higher IL-6 (b = 0.05, 95% confidence interval [CI] = 0.00–0.09, sr2 = 0.01; Table 2 and Figure 3) and IL-1β (b = 0.12, 95% CI = 0.03–0.21, sr2 = 0.02; Table 2 and Figure 4), but not CRP or TNF-α (Table 2), controlling for age, sex, BMI, and shift work. Controlling for the same covariates, IIV in actigraphy-determined sleep efficiency was not associated with CRP, IL-6, IL-1β, or TNF-α (Table 2).

Associations between actigraphy- and sleep diary–determined total sleep time with plasma IL-6 (in picograms per milliliter). IL-6 = interleukin-6 (log-10 transformed); TST = total sleep time. Color images are available online only with this article at
Associations between actigraphy- and sleep diary–determined total sleep time with plasma IL-1β (in picograms per milliliter). IL-1β = interleukin-1β (log-10 transformed); TST = total sleep time. Color images are available online only with this article at

Sleep Diary

Greater IIV in sleep diary-determined total sleep time was associated with higher IL-6 (b = 0.04, 95% CI = 0.00–0.08, sr2 = 0.01; Table 3 and Figure 3) and higher IL-1β (b = 0.09, 95% CI = 0.01–0.17, sr2 = 0.01; Table 3 and Figure 4), but not CRP or TNF-α (Table 3), controlling for age, sex, BMI, and shift work. Controlling for the same covariates, IIV in sleep diary–determined sleep efficiency was not associated with either CRP, IL-6, IL-1β, or TNF-α (Table 3).

Moderation by Shift Work Status

Shift work status did not moderate any of the associations between the iM or IIV in any sleep parameters with CRP, IL-6, IL-1β, or TNF-α (Tables 2, 3), controlling for age, sex, and BMI.

Exploratory Analyses

In an effort to explore other potentially important facets of sleep for inflammation, on a post hoc basis, we examined actigraphy-determined sleep fragmentation (i.e., percentage of time spent mobile during the sleep episode) as a predictor of inflammation markers. Neither mean nor IIV in sleep fragmentation was associated with any inflammation biomarkers (mean sleep fragmentation on IL-6: b = −0.001, p = .73; IL-1β: b = −0.01, p = .077; TNF-α: b = 0.0003, p = .91; CRP: b = 0.03, p = .28; and IIV in sleep fragmentation on IL-6: b = −0.001, p = .80; IL-1β: b = 0.02, p = .16; TNF-α: b = −0.0005, p = .91; CRP: b = −0.04, p = .35).

On an exploratory basis, we also examined associations of IIV in total sleep time with inflammation markers separately in day workers and night-shift workers. Among night-shift workers, IIV in actigraphy-determined total sleep time was positively associated with IL-1β (b = 0.18, standard error [SE] = 0.07, p = .018), but not with CRP (b = 0.19, SE = 0.31, p = .55), IL-6 (b = 0.02, SE = 0.04, p = .62), or TNF-α (b = −0.01, SE = 0.03, p = .81). Similarly, IIV in diary-determined total sleep time was positively associated with CRP (b = 0.62, SE = 0.29, p = .038), but not with IL-1β (b = 0.10, SE = 0.07, p = .19), IL-6 (b = 0.05, SE = 0.04, p = .17), or TNF-α (b = −0.01, SE = 0.03, p = .85).

Among day-shift workers, IIV in actigraphy-determined total sleep time was positively associated with IL-6 (b = 0.07, SE = 0.03, p = .031), but not with CRP (b = 0.13, SE = 0.22, p = .57), IL-1β (b = 0.10, SE = 0.06, p = .11), or TNF-α (b = −0.01, SE = 0.02, p = .75). IIV in diary-determined total sleep time was negatively associated with TNF-α (b = −0.04, SE = 0.02, p = .031), but not with IL-6 (b = 0.04, SE = 0.02, p = .073), IL-1β (b = 0.09, SE = 0.05, p = .063), or CRP (b = −0.05, SE = 0.17, p = .78).


This was the first study to demonstrate that greater night-to-night variability in actigraphy- and sleep diary–determined total sleep time is uniquely associated with higher levels of inflammation biomarkers IL-6 and IL-1β. Strengthening previous research on variability in sleep and inflammation (31–33), we assessed sleep for 1 week immediately before the blood draw, used both subjective and objective assessments of sleep, and assessed multiple inflammation biomarkers. Although the effects of IIV in sleep on inflammation were small, they were independent of the effects of mean sleep (which was not uniquely associated with levels of inflammation). Furthermore, findings between IIV in sleep and inflammation did not differ by night-shift work status. Together, these results highlight the importance of incorporating measures of variability in sleep to predict biomarkers of physiological functioning in future biobehavioral research.

Our findings contribute to a growing body of literature examining associations between IIV in sleep and biomarkers of inflammation. Two other studies have shown that greater IIV in facets of sleep efficiency is associated with elevated inflammation. In one study of older adults, greater IIV in wake time and time in bed was associated with higher IL-6 among good sleepers (32). In another study of midlife adults, greater IIV in sleep onset latency and wake after sleep onset was associated with a composite measure of allostatic load (which included inflammation biomarkers CRP and IL-6) (31). In contrast to these studies, we found that IIV in total sleep time, but not sleep efficiency, was associated with elevated IL-6 and IL-1β among nurses. To examine these associations, we used a rigorous naturalistic design, using both subjective and objective means to assess sleep and examining multiple facets of sleep in relation to four inflammation biomarkers particularly relevant for sleep and health. Supplementing previous research, we also assessed sleep and shift work for 1 week immediately before the blood draw, strengthening causal inference. Together with findings from other studies, our results suggest that IIV in sleep may be a unique correlate of elevated inflammation, even beyond the influence of mean sleep.

Possible Mechanisms

It is possible the associations we observed may be mediated through disruption of neuroendocrine, circadian, and/or behavioral processes. For example, greater IIV in sleep is associated with alterations in cortisol (31) and norepinephrine (28), both of which regulate production of inflammation markers. IIV in sleep also may lead to poor health outcomes via direct disruption of circadian rhythms, which regulate key aspects of cell growth and survival, including the cell cycle, cellular senescence, and metabolism (47). Behaviorally, it is also possible that IIV in sleep may lead to disturbances in health behaviors (e.g., poor diet, reduced physical activity, increased substance use), which may indirectly lead to elevated inflammation.

However, the order in which these relationships unfold is unknown: IIV in sleep may cause disruption neuroendocrine function, health behaviors, and elevated inflammation, or these factors may cause IIV in sleep. In fact, it is highly likely that this relationship is bidirectional or cyclical. It is also possible that other variables, such as stress, may simultaneously contribute to both greater IIV in sleep and elevated levels of inflammation. Future longitudinal and experimental studies are needed to untangle these relationships across time, in order to understand if IIV is a cause and/or a consequence of elevated inflammation. Given that inflammation and disturbed sleep are both prospectively linked to multiple chronic diseases (9,15–17,48), it may be critical to understand which comes first to more effectively intervene to prevent disease.

Lack of Moderation by Shift Work Status

Somewhat surprisingly, we did not find that the associations between mean or IIV in sleep and inflammation were moderated by recent shift work status among nurses. This suggests regardless of the timing of nurses’ work schedule in the past 7 days, more night-to-night variability in total sleep time across the week is associated with elevated inflammation. However, exploratory follow-up analyses conducted separately in day and night-shift workers did reveal some discrepant findings. Among night-shift workers only, greater IIV in actigraphy and diary total sleep time was associated with higher CRP. Among day-shift workers only, greater IIV in actigraphy total sleep time was associated with higher IL-6. These preliminary results suggest that variability in sleep may be associated with unique patterns of inflammation among day and night-shift workers, although results should be replicated and extended with future research.

In the current study, night-shift work also was associated with greater IL-6 independently of IIV in total sleep time, suggesting that shift work and IIV in sleep are related but partially distinct constructs. For example, even those nurses not engaged in shift work may experience high variability in their sleep duration across a week because of day-to-day fluctuations in stress, work/family obligations, caregiving responsibilities, physical symptoms, mood, health behaviors, or medication use. Supporting our findings, one study similarly showed that independent of sleep duration, more variable sleep timing (i.e., delaying bedtimes by 8.5 hours on 4 of 8 days) is associated with elevated CRP (49). Future studies should continue to examine the relative importance of variable sleep timing and variable sleep duration on markers of physiological functioning.

Null Results With CRP and TNF-α

Our analyses did not reveal any significant associations with IIV in sleep and CRP and TNF-α. Other studies have found similarly null results with IIV in sleep and these biomarkers (31,32). It is possible that IL-6 and IL-1β may be more informative or sensitive biomarkers of variability in sleep compared with CRP and TNF-α. IL-6 is particularly sensitive to sleep loss and disturbed sleep (for a review, see Rohleder et al. (50)). CRP also is more stable over short and long periods than is IL-6 and IL-1β (51–53), and therefore, it may not be as responsive to acute fluctuations in sleep duration. However, given the small effect sizes we observed for associations between IIV in total sleep time with IL-6 and IL-1β, results should be interpreted cautiously. Findings should be replicated in other samples before definitive conclusions are drawn regarding which inflammation biomarkers are most robustly associated with IIV in sleep.

Null Results With Sleep Efficiency

In our analyses, neither mean nor variability in sleep efficiency was associated with any markers of inflammation. This supports findings from another recent study, which showed that greater IIV in total sleep time, but not sleep efficiency, was associated with increased risk for a number of medical problems (i.e., pain, depression, and gastrointestinal, neurological, and breathing problems) (26). It is possible that maintaining a consistent sleep duration (i.e., total sleep time) is more important for health than is obtaining consistent sleep quality and consolidation (i.e., sleep efficiency). Future studies should examine the relative importance of IIV in different domains of sleep in relation to markers of health.

Limitations and Opportunities for Future Research

Although this study has many unique strengths (e.g., large representative sample of nurses and 7 days of self-report and actigraphy measures of sleep immediately before the blood draw), there are some limitations warranting discussion. First, we only investigated some of the potential correlates of IIV in sleep in this study (e.g., shift work, age, and sex). There are likely other factors involved in the association between IIV in sleep and inflammation, such as medication use, underlying medical conditions, perceived stress, and/or social jetlag. Without a true control group (e.g., individuals randomly assigned to shift work), we were unable to tell if our sample’s dysregulated sleep was due to the nature of their work or these other factors. Future studies should examine other potential causes and/or consequences of IIV in sleep. Intensive longitudinal studies may be useful to untangle these bidirectional associations between nightly fluctuations in sleep and inflammation levels. A few studies have examined such associations in daily life and found that short sleep duration is associated with subsequent increases in salivary IL-6 (54), and that night-shift work is associated with alterations in salivary IL-1β (37,55).

Second, although our sample demographics generally matched national demographics of nurses in the United States (38), our sample was primarily female, and therefore, we were unable to investigate potential sex or gender differences in these associations. Previous studies have shown that there are important sex differences in both IIV in sleep and inflammation, such that women may be more affected by the detrimental effects of sleep loss on inflammation (56,57). Women also tend to exhibit greater IIV in sleep than men (28).

Third, although our findings were consistent across multiple biomarkers and across sleep diaries and actigraphy, it is important to note that we did run multiple analyses, which may have increased our chances of making a type 1 error. Fourth, IL-1β levels were nondetectable in approximately 31% of our sample. Based on examination of the standard curve, we can be certain that these nondetectable samples were truly very low levels. As with many markers of inflammation, it can be difficult to obtain detectable levels in relatively healthy samples (41,58). Values below detectable limits are data missing not at random, which poses a unique challenge for data analysis. Multiple imputation techniques are not recommended for data missing not at random (59). Therefore, when missing data are due to low levels, researchers may benefit from imputing a small value for missing data (i.e., the value half the LLD). This is a common strategy when examining markers of inflammation (41,42,58) and capitalizes on using all available information. Finally, more days of sleep measurements would have provided more reliable estimates of IIV in sleep across time; however, adding measurement occasions should always be weighed against participant burden.


Nurses play a crucial role in maintaining patient health and safety. Therefore, it is essential to understand the causes and consequences of their sleep problems. We showed that greater IIV in sleep duration may be a unique correlate of elevated inflammation in nurses. Surprisingly, mean sleep was unrelated to inflammation. We urge sleep researchers to consider the unique role of IIV in sleep on health whenever possible. Given that elevated inflammation is observed in many chronic diseases, future longitudinal studies should also examine whether inflammation may be a mediator between IIV in sleep and more distal disease outcomes. Regularizing sleep duration whenever possible may be one means to promote health among nurses.

We would like to thank the hospitals, nurse principal investigators, participants, and research assistants for their help facilitating this study.

All data and code are available on Open Science Framework:

Source of Funding and Conflicts of Interest: The authors declare no conflicts of interest. This research was supported by a National Institute of Allergy and Infectious Diseases grant (1R01AI128359-01; principal investigators: D.J.T. and K.K.).


1. Lin S-H, Liao W-C, Chen M-Y, Fan J-Y. The impact of shift work on nurses’ job stress, sleep quality and self-perceived health status. J Nurs Manag 2014;22:604–12.
2. Gold DR, Rogacz S, Bock N, Tosteson TD, Baum TM, Speizer FE, Czeisler CA. Rotating shift work, sleep, and accidents related to sleepiness in hospital nurses. Am J Public Health 1992;82:1011–4.
3. Lee KA. Self-reported sleep disturbances in employed women. Sleep 1992;15:493–8.
4. Niedhammer I, Lert F, Marne MJ. Effects of shift work on sleep among French nurses. A longitudinal study. J Occup Med 1994;36:667–74.
5. Kageyama T, Nishikido N, Kobayashi T, Oga J, Kawashima M. Cross-sectional survey on risk factors for insomnia in Japanese female hospital nurses working rapidly rotating shift systems. J Hum Ergol 2001;30:149–54.
6. Øyane NM, Pallesen S, Moen BE, Akerstedt T, Bjorvatn B. Associations between night work and anxiety, depression, insomnia, sleepiness and fatigue in a sample of Norwegian nurses. PLoS One 2013;8:e70228–8.
7. Phillips B, Mannino DM. Do insomnia complaints cause hypertension or cardiovascular disease? J Clin Sleep Med 2007;3:489–94.
8. Taylor DJ, Lichstein KL, Durrence HH, Reidel BW, Bush AJ. Epidemiology of insomnia, depression, and anxiety. Sleep 2005;28:1457–64.
9. Ayas NT, White DP, Al-Delaimy WK, Manson JE, Stampfer MJ, Speizer FE, Patel S, Hu FB. A prospective study of self-reported sleep duration and incident diabetes in women. Diabetes Care 2003;26:380–4.
10. Patel SR, Ayas NT, Malhotra MR, White DP, Schernhammer ES, Speizer FE, Stampfer MJ, Hu FB. A prospective study of sleep duration and mortality risk in women. Sleep 2004;27:440–4.
11. Irwin M. Effects of sleep and sleep loss on immunity and cytokines. Brain Behav Immun 2002;16:503–12.
12. Irwin MR, Opp MR. Sleep health: reciprocal regulation of sleep and innate immunity. Neuropsychopharmacology 2016;42:129.
13. Irwin MR, Olmstead R, Carroll JE. Sleep disturbance, sleep duration, and inflammation: a systematic review and meta-analysis of cohort studies and experimental sleep deprivation. Biol Psychiatry 2016;80:40–52.
14. Sookoian S, Gemma C, Fernandez Gianotti T, Burgueno A, Alvarez A, Gonzalez C, Pirola C. Effects of rotating shift work on biomarkers of metabolic syndrome and inflammation. J Intern Med 2007;261:285–92.
15. Cesari M, Penninx BW, Newman AB, Kritchevsky SB, Nicklas BJ, Sutton-Tyrrell K, Tracy RP, Rubin SM, Harris TB, Pahor M. Inflammatory markers and cardiovascular disease (the Health, Aging and Body Composition [Health ABC] Study). Am J Cardiol 2003;92:522–8.
16. Ford DE, Erlinger TP. Depression and C-reactive protein in US adults: data from the Third National Health and Nutrition Examination Survey. Arch Intern Med 2004;164:1010–4.
17. Pradhan A, Manson J, Rifai N, Buring J, Ridker P. C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA 2001;286:327–34.
18. Ananthakrishnan AN, Khalili H, Konijeti GG, Higuchi LM, de Silva P, Fuchs CS, Richter JM, Schernhammer ES, Chan AT. Sleep duration affects risk for ulcerative colitis: a prospective cohort study. Clin Gastroenterol Hepatol 2014;12:1879–86.
19. Schernhammer ES, Laden F, Speizer FE, Willett WC, Hunter DJ, Kawachi I, Colditz GA. Rotating night shifts and risk of breast cancer in women participating in the nurses’ health study. J Natl Cancer Inst 2001;93:1563–8.
20. Buysse DJ. Sleep health: can we define it? Does it matter? Sleep 2014;37:9–17.
21. Bei B, Wiley JF, Trinder J, Manber R. Beyond the mean: a systematic review on the correlates of daily intraindividual variability of sleep/wake patterns. Sleep Med Rev 2016;28:108–24.
22. Akerstedt T. Shift work and disturbed sleep/wakefulness. Occup Med 2003;53:89–94.
23. Patel SR, Hayes AL, Blackwell T, Evans DS, Ancoli-Israel S, Wing YK, Stone KL. The association between sleep patterns and obesity in older adults. Int J Obes (Lond) 2014;38:1159–64.
24. Dautovich ND, Kay DB, Perlis ML, Dzierzewski JM, Rowe MA, McCrae CS. Day-to-day variability in nap duration predicts medical morbidity in older adults. Health Psychol 2012;31:671–6.
25. Huang T, Redline S. Cross-sectional and prospective associations of actigraphy-assessed sleep regularity with metabolic abnormalities: The Multi-Ethnic Study of Atherosclerosis. Diabetes Care 2019;42:1422–9.
26. Slavish DC, Taylor DJ, Lichstein KL. Intraindividual variability in sleep and comorbid medical problems. Sleep 2019;42:zsz052.
27. Buysse DJ, Cheng Y, Germain A, Moul DE, Franzen PL, Fletcher M, Monk TH. Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Med 2010;11:56–64.
28. Mezick EJ, Matthews KA, Hall M, Kamarck TW, Buysse DJ, Owens JF, Reis SE. Intra-individual variability in sleep duration and fragmentation: associations with stress. Psychoneuroendocrinology 2009;34:1346–54.
29. Bei B, Manber R, Allen NB, Trinder J, Wiley JF. Too long, too short, or too variable? Sleep intraindividual variability and its associations with perceived sleep quality and mood in adolescents during naturalistically unconstrained sleep. Sleep 2017;40.
30. Irwin M, Clark C, Kennedy B, Gillin JC, Ziegler M. Nocturnal catecholamines and immune function in insomniacs, depressed patients, and control subjects. Brain Behav Immun 2003;17:365–72.
31. Bei B, Seeman TE, Carroll JE, Wiley JF. Sleep and physiological dysregulation: A closer look at sleep intraindividual variability. Sleep 2017;40:zsx109.
32. Okun ML, Reynolds CF, Buysse DJ, Monk TH, Mazumdar S, Begley A, Hall M. Sleep variability, health-related practices and inflammatory markers in a community dwelling sample of older adults. Psychosom Med 2011;73:142–50.
33. Park H, Tsai KM, Dahl RE, Irwin MR, McCreath H, Seeman TE, Fuligni AJ. Sleep and inflammation during adolescence. Psychosom Med 2016;78:677–85.
34. Wright KP, Drake AL, Frey DJ, Fleshner M, Desouza CA, Gronfier C, Czeisler CA. Influence of sleep deprivation and circadian misalignment on cortisol, inflammatory markers, and cytokine balance. Brain Behav Immun 2015;47:24–34.
35. Puttonen S, Viitasalo K, Harma M. Effect of shiftwork on systemic markers of inflammation. Chronobiol Int 2011;28:528–35.
36. Morris CJ, Purvis TE, Mistretta J, Hu K, Scheer F. Circadian misalignment increases C-reactive protein and blood pressure in chronic shift workers. J Biol Rhythms 2017;32:154–64.
37. Reinhardt EL, Fernandes P, Markus RP, Fischer FM. Night work effects on salivary cytokines TNF, IL-1beta and IL-6. Chronobiol Int 2018;19:1–16.
38. Health Resources and Services Administration. The U.S. Nursing Workforce: Trends in Supply and Education. Washington, DC: Department of Health and Human Services; 2013.
39. Carney CE, Buysse DJ, Ancoli-Israel S, Edinger JD, Krystal AD, Lichstein KL, Morin CM. The consensus sleep diary: standardizing prospective sleep self-monitoring. Sleep 2012;35:287–302.
40. Harris P, Taylor R, Thielke R, Payne J, Gonzalez N, Conde J. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–81.
41. Newton TL, Fernandez-Botran R, Lyle KB, Szabo YZ, Miller JJ, Warnecke AJ. Salivary cytokine response in the aftermath of stress: an emotion regulation perspective. Emotion 2017;17:1007–20.
42. Szabo YZ, Fernandez-Botran R, Newton TL. Cumulative trauma, emotion reactivity and salivary cytokine levels following acute stress in healthy women. Anxiety Stress Coping 2018;20:1–13.
43. May AL, Freedman D, Sherry B, Blanck HM. Obesity—United States, 1988–2008. MMWR Suppl 2011;60:73–7.
44. R: A Language and Environment for Statistical Computing [program]. Vienna, Austria: R Foundation for Statistical Computing; 2013.
45. Stanley D. apaTables: Create American Psychological Association (APA) Style Tables [program]. 1.5.1 version. 2017. Available at: Accessed February 1, 2020.
46. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, NJ: Lawrence Earlbaum Associates; 1988.
47. Sahar S, Sassone-Corsi P. Metabolism and cancer: the circadian clock connection. Nat Rev Cancer 2009;9:886–96.
48. Ayas NT, White DP, Manson JE, Stampfer MJ, Speizer FE, Malhotra A, Hu FB. A prospective study of sleep duration and coronary heart disease in women. Arch Intern Med 2003;163:205–9.
49. Leproult R, Holmbäck U, Van Cauter E. Circadian misalignment augments markers of insulin resistance and inflammation, independently of sleep loss. Diabetes 2014;63:1860–9.
50. Rohleder N, Aringer M, Boentert M. Role of interleukin-6 in stress, sleep, and fatigue. Ann N Y Acad Sci 2012;1261:88–96.
51. Ferrari R, Tanni SE, Caram LM, Corrêa C, Corrêa CR, Godoy I. Three-year follow-up of interleukin 6 and C-reactive protein in chronic obstructive pulmonary disease. Respir Res 2013;14:24.
52. Izawa S, Miki K, Liu X, Ogawa N. The diurnal patterns of salivary interleukin-6 and C-reactive protein in healthy young adults. Brain Behav Immun 2013;27:38–41.
53. Shields GS, Slavich GM, Perlman G, Klein DN, Kotov R. The short-rerm reliability and long-term stability of salivary immune markers. Brain Behav Immun 2019;81:650–4.
54. Reinhardt EL, Fernandes PA, Markus RP, Fischer FM. Short sleep duration increases salivary IL-6 production. Chronobiol Int 2016;33:780–2.
55. Reinhardt ÉL, Fernandes PA, Markus RP, Fischer FM. Daily rhythm of salivary IL-1ß, cortisol and melatonin in day and night workers. Work 2012;41(Suppl 1):5788–90.
56. Suarez EC. Self-reported symptoms of sleep disturbance and inflammation, coagulation, insulin resistance and psychosocial distress: evidence for gender disparity. Brain Behav Immun 2008;22:960–8.
57. Prather AA, Epel ES, Cohen BE, Neylan TC, Whooley MA. Gender differences in the prospective associations of self-reported sleep quality with biomarkers of systemic inflammation and coagulation: findings from the Heart and Soul Study. J Psychiatr Res 2013;47:1228–35.
58. Slavish DC, Jones DR, Smyth JM, Engeland CG, Song S, McCormick NM, Graham-Engeland JE. Positive and negative affect and salivary markers of inflammation among young adults. Int J Behav Med 2020;27:282–93.
59. Sterne JA, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, Wood AM, Carpenter JR. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ 2009;338:b2393–3.

sleep; nurses; actigraphy; sleep diary; inflammation; intraindividual variability; BMI = body mass index; CRP = C-reactive protein; IL-1β = interleukin-1β; IL-6 = interleukin-6; iM = intraindividual mean; IIV = intraindividual variability; LLD = lower limit of detection; sr2 = semipartial correlations squared; TNF-α = tumor necrosis factor α

Copyright © 2020 by the American Psychosomatic Society