Recommendations from the National Sleep Foundation indicate that adults aged 18+ years should receive at least 7 hours of sleep per night (Hirshkowitz et al., 2015). However, it is estimated that at least 30% of the US population sleeps less than 7 hours per night (Ford, Cunningham, & Croft, 2015). Around 33% of all adults experience symptoms of transient insomnia, 40% of whom develop more severe forms of insomnia (Pavlova & Latreille, 2019). In the general population, between 30–50% of men and 11–23% of women report moderate-to-severe sleep apnea symptoms (Heinzer et al., 2015; Pavlova & Latreille, 2019). This is disconcerting given that adequate sleep is necessary for proper mental, emotional and physical restoration, and that poor sleep is associated with the development of numerous chronic conditions (Parish, 2009).

Pharmacological (e.g., benzodiazepines and non-benzodiazepine receptor agonists) interventions are among the most common treatments for sleep disorders (Brandt & Leong, 2017). However, such treatments can be habit-forming and have been associated with a range of adverse effects including increased incidence of falls and fractures, dementia and other memory/cognition issues, infections, and mortality (Brandt & Leong, 2017). Excessive sleepiness and daytime fatigue are side effects observed in more than 10% of those taking pharmacological treatments (Proctor & Bianchi, 2012).

Non-pharmacological approaches used to treat sleep ailments include mindfulness or cognitive behavioral therapy, melatonin supplementation, ear plugs/eye masks (i.e., sensory reduction), bright light therapy, and exercise (Miller, Renn, Chu, & Torrence, 2019). Diet is an approach of particular importance. Adequate sleep duration (i.e., 7–8 hours per night) is associated with greater nutrient intake and high-fat diets are associated with sleep disorders (Grandner, Jackson, Gerstner, & Knutson, 2013; St-Onge, Mikic, & Pietrolungo, 2016). High-carbohydrate diets have been shown to decrease slow-wave sleep (SWS) but increase rapid-eye-movement (REM) (Afaghi, O’Connor, & Chow, 2007; St-Onge et al., 2016).

Diets resembling a Western pattern (e.g., high in total and saturated fats, protein, and added sugar) are more pro-inflammatory. Diets defined by high intake of fruits and vegetables, whole grains, and fish (e.g., Mediterranean) are more anti-inflammatory (Ahluwalia, Andreeva, Kesse-Guyot, & Hercberg, 2013). A sleep duration less than 7 hours or greater than 8 hours per night was associated with increased levels of pro-inflammatory cytokines in a meta-analysis of 72 studies (Irwin, Olmstead, & Carroll, 2016). Poor sleep quality or sleep disturbances and diagnosed sleep or sleep-related disorders such as insomnia or OSA also were found to be associated with inflammation (Irwin et al., 2016; Kapsimalis et al., 2008).

The Dietary Inflammatory Index (DII^®) was designed to measure the pro- or anti-inflammatory nature of one’s diet (Shivappa, Steck, Hurley, Hussey, & Hebert, 2014). The DII has been validated against inflammatory cytokines (Shivappa, Steck, Hurley, Hussey, Ma, et al., 2014; M. D. Wirth et al., 2014), and has been associated with inflammation-related outcomes such as cancer (Jayedi, Emadi, & Shab-Bidar, 2018), diabetes (Denova-Gutierrez et al., 2018), and cardiovascular disease (Shivappa et al., 2018). More pro-inflammatory DII scores were associated with increased severity of OSA, daytime sleepiness and dysfunction, increased REM latency, short sleep duration (i.e., <6 hours per night), and greater odds of reporting sleep disturbances or having “poor” sleep quality as defined by the Pittsburgh Sleep Quality Index (PSQI) (Bazyar et al., 2021; Godos et al., 2019; Kase, Liu, Wirth, Shivappa, & Hebert, 2021; Lopes et al., 2019; Masaad et al., 2021). Using data from an anti-inflammatory diet intervention, Wirth and colleagues found that individuals in the first tertile for the change in DII scores (i.e., anti-inflammatory diet changes) compared to individuals in the third tertile (i.e., pro-inflammatory diet changes) had a reduction of nearly 25 minutes of wake-after-sleep-onset (WASO) and about a 2.6% increase in sleep efficiency (M. D. Wirth et al., 2020).

Shiftwork has long been associated with circadian disruption and sleep disturbances (Moreno et al., 2019). In addition to this, night and rotating shift workers have higher levels of inflammation compared to their dayshift-working counterparts (Puttonen, Viitasalo, & Harma, 2011). In terms of diet, findings from a meta-analysis demonstrated that total energy intake does not differ between day and nightshift workers; however, diet quality is poorer among those who work nights (Bonham, Bonnell, & Huggins, 2016). Correspondingly, night and rotating shift workers tend to have more pro-inflammatory diets than primarily dayshift workers (M. D. Wirth et al., 2014; M. D. Wirth, Shivappa, Burch, Hurley, & Hebert, 2017). Police officers are frequently exposed to shiftwork, increased stress, and other environmental situations that may predispose them to experiencing poor sleep, increased inflammation, or poor access to healthy food options (Garbarino, Guglielmi, Puntoni, Bragazzi, & Magnavita, 2019; M. Wirth et al., 2013).

The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) cohort study was designed to examine biological processes through which stressors of police work influence adverse health outcomes (Violanti et al., 2006). Using data from BCOPS, this study tested the hypothesis that more pro-inflammatory diets would be associated with shorter sleep duration and poorer sleep quality, measured both objectively and subjectively, compared to those with more anti-inflammatory diets. Additionally, shiftwork experience was examined as an effect modifier.

Participants were mainly male (74%), European-American (77%), held a rank of police officer (70%) as compared to higher level positions, and were less than college educated or had an associate degree only (67%) at baseline. The average age was 41.5 ± 6.7 years, average BMI was 29.3 ± 4.8 kg/m², and the average years employed as a police officer was 15.0 ± 7.2 years (Table 1). The average E-DII was −0.67 ± 2.16 which indicates a neutral inflammatory diet. When comparing sample characteristics by E-DII category, those with more pro-inflammatory diets compared to more anti-inflammatory diets were more likely to be male (p<0.01), to be current smokers (p=0.05), have a higher waist circumference (p<0.01), and have higher systolic blood pressure (p=0.02) at baseline (Table 1).

In the stationary effect models, every one-unit increase (i.e., more pro-inflammatory) in the E-DII score was associated with an adjusted higher WASO of 1.28 minutes (SE=0.56, p=0.02). No other differences were observed for objective measures either when using the E-DII in its continuous or categorical form. Every one-unit increase in the E-DII was associated with a −0.14 (SE=0.06, p=0.01) decrease (i.e., improvement) in the PSQI global sleep score. Categorically, those in the pro-inflammatory E-DII category had a lower mean PSQI than the very anti-inflammatory category (6.61 vs. 7.35, p=0.04, Table 2).

For the analytic approach that examined the differential impact of cross-sectional and longitudinal effects within the same model, every one-unit increase in the change in E-DII score (i.e., becoming more pro-inflammatory over time), WASO increased by 1.36 minutes (SE=0.74, p=0.07) and the global PSQI sleep score improved (β_change=−0.22, SE=0.01, p<0.01). No other statistically significant results were observed among the full sample (Table 3). However, notable interactions (i.e., p<0.20) were observed between the change in E-DII and shift status for sleep efficiency (p=0.06), sleep latency (p=0.15), WASO (p=0.18), and sleep fragmentation (p=0.14) models (Table 4). Sleep latency appeared to decrease among the night shift group with increasing pro-inflammatory diets (β_change=−0.71, SE=0.21, p<0.01); whereas, no such relationship was observed among day or evening shift workers. Among day shift officers, but not evening or night shift officers, every one-unit increase in the change in E-DII was associated with a 3.33-minute (SE=1.24, p=0.01) increase in WASO. A one-unit increase in the change in E-DII score was associated with an improvement in sleep efficiency in night shift officers, but a worsening in sleep efficiency among day-shift officers (Table 4).

In this population of police officers, an occupational group exposed to numerous stressors that can affect sleep (M. Wirth et al., 2013), every one-unit increase in the change in E-DII over time (i.e., becoming more pro-inflammatory over time) increased WASO by 3.33 minutes in officers primarily working day shifts. Similarly, post-hoc analyses of a self-selection DII-based clinical trial (the Inflammation Management Intervention or IMAGINE) found similar results (M. D. Wirth et al., 2020). For that analysis, participants in the control and intervention arm were combined due to high rates of cross-over between study arms. Within IMAGINE, those with the most anti-inflammatory dietary changes over 3 months decreased WASO by 25 minutes per night, whereas no change was observed in participants with pro-inflammatory E-DII changes (p<0.01) (M. D. Wirth et al., 2020).

Among United Arab Emirates college students, mean E-DII scores were greater (0.55 vs. 0.07, respectively, p=0.01) among those with PSQI-based daytime dysfunction compared to those without daytime dysfunction (Masaad et al., 2021), and among Iranian female college students, every one-unit increase in the E-DII score was associated with greater odds (OR=1.22, 95%CI=1.03–1.44) of poor sleep according to the PSQI (Bazyar et al., 2021). In a cross-sectional study among those with OSA from Brazil, the E-DII was found to be a predictor of sleep apnea severity and daytime dysfunction (as measured by the PSQI) among older adults (Lopes et al., 2019). Also using the PSQI, Godos and colleagues found that Italian adults with the most pro-inflammatory diets, compared to anti-inflammatory diets, had lower odds or having “good” sleep quality (odds ratio [OR]=0.49, 95% confidence interval [95%CI]=0.31–0.78) (Godos et al., 2019). Using data from the National Health and Nutrition Examination Survey from the United States, Kase and colleagues found that the most pro-inflammatory E-DII group, compared to the most anti-inflammatory group, had elevated odds of ≤6 hours of sleep per night (OR=1.40, 95%CI=1.21–1.61), ≥9 hours of sleep per night (OR=1.23, 95%CI=1.03–1.46), and self-reported sleep disturbances (OR=1.14, 95%CI=1.02–1.27) (Kase et al., 2021).

Of the past studies described above, five were cross-sectional (Bazyar et al., 2021; Godos et al., 2019; Kase et al., 2021; Lopes et al., 2019; Masaad et al., 2021), four were international studies with two of those focusing on college students and the other on adults with OSA (Bazyar et al., 2021; Godos et al., 2019; Lopes et al., 2019; Masaad et al., 2021), and four included only self-report measures of sleep (Bazyar et al., 2021; Godos et al., 2019; Kase et al., 2021; Masaad et al., 2021). The Brazilian study used polysomnography (PSG), the most rigorous sleep assessment (Lopes et al., 2019). The IMAGINE study was the most comparable to the current study in terms of the longitudinal design and sleep measurement devices (M. D. Wirth et al., 2020). A recent review provides further support related to associations between dietary index scores and sleep quality; specifically, healthy diets being associated with better sleep quality (Godos et al., 2021). However, the authors of that review further stated that the evidence is limited given most studies were cross-sectional in nature and that most sleep assessments were subjective (Godos et al., 2021). The current study made use of a longitudinal design and objective markers of sleep.

Interestingly, statistically significant findings were in the opposite direction for sleep latency and sleep efficiency than hypothesized for those working primarily night shifts. However, fatigue may actually be associated with a decreased sleep latency in populations experiencing elevated fatigue (e.g., those undergoing cancer treatment) (Holliday et al., 2016). Study design phenomena may also explain these findings. In a longitudinal study, individuals who struggle with the adverse effects of night work (e.g., sleep disturbances) may leave the occupation or switch to another shift type. This may create a selection bias where those who continue to work the night shift may be healthier in certain domains compared to their day-working counterparts, or those no longer working in the profession. This is partly evidenced by the fact that over the full study from baseline to the second follow-up, day shift membership decreased by 22%, evening shift by 48% and night shift by 53%. Within those that remained in the night shift category, some officers may inherently be better able to deal with the adverse effects of shiftwork either through long years of training or genetics (Burch et al., 2009). This enhanced coping may lead to fewer adverse effects of shiftwork on sleep. At the same time, it is possible that some officers may be able to consume high-fat/high-sugar foods without it adversely impacting sleep as much as in other officers working the night shift. This may create a phenomenon where a subset of night shift officers has better sleep quality but more pro-inflammatory diets, which may explain results that were opposite of the hypothesized effect in night shift workers.

In the current study, more pro-inflammatory diets were associated with better subjective sleep quality. Kline and colleagues observed that actigraphically measured total sleep time was, on average, about 1.25 hours less than subjectively reported total sleep time. Subjective total sleep time more closely resembled TIB from actigraphy (Kline et al., 2010). It also is possible that reporting bias related to social desirability or the desire to make oneself feel better about their own health or lifestyle habits impacted self-reporting.

The average E-DII scores of the police officers in this study fell into the neutral category at baseline. This is consistent with shift workers in the general American population (M. D. Wirth et al., 2017). Biologically, it is possible that the E-DII is associated with various aspects of sleep. For example, sleep promoting cytokines include tumor necrosis factor (TNF)-α and interleukin (IL)-1β, both of which are pro-inflammatory. IL-1β can stimulate growth hormone-releasing hormone, which enhances NREM sleep (Obal & Krueger, 2003). IL-6 may be involved with sleep initiation as it peaks around the time of sleep onset. Additionally, administration of high levels of IL-6 can disrupt sleep structure (Kapsimalis et al., 2008). Chronic exposure to poor diets may, in a similar manner, disrupt the rhythm of IL-6 secretion leading to similar effects as seen with administration of high levels of IL-6.

The E-DII score includes a range of dietary components that impact bodily processes other than inflammation, some of which my impact sleep. High-fat diets can decrease sleep efficiency; whereas, high carbohydrate diets may improve sleep structure (St-Onge et al., 2016). Supporting this potential mechanism is the fact that those in the very anti-inflammatory group (at baseline) consumed 29% of total energy intake as fat verses 39% (p<0.01) in those in the pro-inflammatory group (data not shown). Carbohydrate consumption may increase tryptophan availability for later synthesis of serotonin and melatonin which may facilitate sleep (Doherty, Madigan, Warrington, & Ellis, 2019).

Compared to other research in this field, a longitudinal design and an analysis that differentiated between cross-sectional and longitudinal effects were study strengths. Objective measures of sleep were assessed. The E-DII very specifically focuses on dietary inflammation, which is important given the inflammatory underpinnings of sleep. A range of covariates, including stress, were evaluated as potential confounders.

However, limitations should be considered when interpreting the results. The population was primarily European-American males, which may limit generalizability. Dietary data were obtained through self-report using an FFQ. Additionally, there was considerable attrition across time points and this attrition may have been due to biases related to selective work adaptation abilities or a healthy worker effect. Lastly, work and off days were not separated in the actigraphy measures, and it is conceivable that sleep and diet are different on work versus off days.

Among day workers, a more pro-inflammatory diet change over time was associated with increased time spent awake after initially falling asleep. Specifically, if these individuals could make their E-DII score more anti-inflammatory by 5 points, then the results indicate they may decrease time spent awake by about 17 minutes per night (119 minutes per week). This nearly 2-hour increase in sleep per week could help to alleviate sleep debt. The concern of sleep debt is particularly important for police officers as research has indicated that sleep is associated with stress, metabolic abnormalities, poor mental health, and other adverse health outcomes among police officers (Garbarino et al., 2019; Garbarino & Magnavita, 2019). Future studies should employ more rigorous sleep assessments, such as PSG, to more thoroughly investigate mechanisms of action between dietary inflammation and sleep. This is important given there is a bidirectional relationship between sleep and inflammation and that sleep may impact dietary choices (Vidafar, Cain, & Shechter, 2020). Understanding if these results apply to other working populations such as nurses could potentially extend these findings to a larger segment of the population.