How the amount and quality of naps reduce tiredness and sleepiness in nurses during long night shifts

Our investigation used a prospective observational design and was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (see Additional file 1). To examine the effect of napping during the night shift, our target population was nurses who were provided with sufficient rest breaks, such as at least 2 h, as described in the guidelines of the Japanese Nursing Association [29]. However, because nurses frequently experience missed, interrupted, or delayed rest breaks due to job demands and resources [30], a survey of Japanese hospitals showed an average nap break of 1.3 h and an average nap duration of 0.9 h in two-shift work schedules in Japan [31]. Further, different environmental factors in each ward affect nurses’ total nap duration during night shifts, and TND affects fatigue [22]. Hence, to adjust for these biases and achieve our aims, our investigation was conducted over multiple night shifts in one hospital ward, which was one of the few workplaces where sufficient nap break durations were available. Further, to exclude the impact of the COVID-19 pandemic on hospitals, our research was conducted in November 2021, when the Japanese government lifted the state of emergency.

Participants were nurses employed in a general hospital with approximately 300 beds in northern Japan. The exclusion criteria were a present or past history of sleep-related problems and current pregnancy. Sample size calculations were not performed because all nurses who fulfilled the criteria were included in the sample.

The study setting was a five-department mixed ward with approximately 50 beds, where nurses engage in a two-shift system with night shifts of over 16-h—the most common shift pattern in Japan [31]. Night shifts lasted from 16:30 to 9:15 and always included four nurses, one of whom had a leadership role in addition to caring for the same number of patients as the others.

Nurses were allowed a 30-min supper break and at least a 2-h nap break, the duration of which varied depending on the situation. The nurses tried to take the longest possible nap breaks without compromising patient safety, using one of two rooms in the ward where they could control the lighting and air conditioning themselves. The first room, located in front of the nurses’ station, was a break room with a convertible sofa that could be used as a bed. The second one, located near the nurses’ station, was an informed consent room with a folding bed.

Data were presented using means (standard deviations) or frequencies (percentages). We excluded from later analyses the night shifts in which participants did not intend to nap. Pearson’s correlation analysis was performed to examine the association between continuous variables.

We examined the associations between the napping parameters to consider napping quantity and quality together. Regarding napping quantity, the TND used in previous studies of napping during night shifts [22, 23] is the typical parameter used to measure napping quantity. For napping quality, according to the National Sleep Foundation, SE is one of the appropriate indicators of objective sleep quality [28]. However, as these two indicators are very strongly correlated (r = 0.86, p < 0.001), we thought it improper to use them together. On the other hand, TIB, often used to measure sleep quantity in the past, was related to fatigue and sleepiness as well [42, 43]. Additionally, TND is the product of TIB and SE, and the correlation between TIB and SE was moderate (r = 0.44, p < 0.001). Hence, we used TIB for the napping quantity and SE for the napping quality as the target value instead of TND. Focusing on lying-down duration first, rather than sleep duration, may make it easier for nurses to practice napping.

Since most nurses (81.2%) worked four night shifts per month, we used a restricted maximum likelihood-based mixed-effects model for repeated measures (MMRM) to examine the associations between napping quantity and quality and fatigue or sleepiness. An unstructured covariance matrix was used to model the correlations among the repeated measures on different night shifts. Four time point outcome variables were set for both fatigue and sleepiness: (1) after nap breaks; (2) end of the night shift; (3) Δbefore to after nap breaks (after nap breaks minus before nap breaks); and (4) Δstart to end of the night shift (end of the night shift minus start of the night shift). As the nap timing during the night shift affects both fatigue and sleepiness [44–46], the midpoint between the first sleep-onset time and last awakening time during the TIB was used as the nap timing, with reference to the Munich Chronotype Questionnaire [47]. This nap timing was included as a covariate in all models. All continuous variables were centered at the grand mean to avoid multicollinearity.

For each outcome variable, we first conducted Model 1 using the continuous variables TIB and SE, and their interactions as fixed effects to examine the effects of both napping quantity and quality during night shifts. Second, to help interpret the quantity and quality of the napping and to confirm the details of the associations, we conducted an exploratory analysis of Model 2 that transformed TIB and SE into category variables. Grouping sleep parameters into categories is a strategy often used to associate sleep-related parameters and outcomes [26, 48–50]. Regarding TIB, we selected the cutoff points 120 and 180 min. This is because the investigated hospital’s regulation for nap breaks was at least two hours. The other reason is the guideline of the Japanese Nursing Association also recommends nap breaks of at least two hours to achieve one non-rapid eye movement (NREM)/rapid eye movement (REM) cycle based on sleep physiology [29]. Previous studies also divided the napping quantity by 120 or 180 min [23, 51–54]. According to these reasons, TIB was divided into three groups: (< 120 min, 120–180 min, and > 180 min). Regarding SE, ≥ 85% is generally considered indicative of good sleep quality [28]. However, the shorter the TIB, the more difficult it is to achieve a high SE. A calculated factor that reduces SE is the awakening duration, which includes SL, WASO, and BOL. In particular, SL does not shorten with a shorter TIB, and the relative rate of one min in the numerator increases when the TIB of the denominator becomes shorter. Hence, we considered that SE ≥ 85% could not be applied to napping during the night shift. Previous studies have also shown that the mean range of SE in napping during 16-h night shifts—the same shift pattern as in this study—is approximately 60–77% [22, 55] and no more than 85%. Hence, based on the assumption that taking a nap with a duration of one NREM/REM sleep cycle is effective [22, 29], the TND to be achieved was 85 min, which corresponds to a median of 70–100 min for the first sleep cycle duration [56]. As the SE to achieve this TND of 85 min in a TIB of 120 min was approximately 70%, we classified SE into two groups: ≥ 70% and < 70%. We combined these three groups of TIB and two groups of SE to classify the night shift into six groups, which were used as fixed effects in Model 2. The least squares means and 95% confidence intervals (CIs) for each time point of fatigue and sleepiness in each group were estimated by the Model 2. If a significant main effect of the group was found, post-hoc t-tests were conducted to estimate the mean difference and their 95% CIs between groups. Third, to perform a sensitivity analysis of the effects of sleep duration, we used Model 3 with the continuous variable of TND as a fixed effect.

Work demands may affect night-shift napping, fatigue, and sleepiness [30, 34, 57]. The correlation between the number of total steps, which reflects one of the work demands [34], and night shift napping was examined. The effects of work demands on fatigue and sleepiness were also examined by using the number of total steps as fixed effects alone in the above-mentioned models.

Finally, univariate analyses were conducted to explore nurse characteristics associated with napping that achieved our primary aim. We counted the number of naps that met each napping parameter identified in the above analysis over one month. Variables regarding night shifts were summarized as one-month averages for each nurse and used as night shift-related factors. The outcome variables were the rates of achieving each napping parameter per month. Nurse-related and night shift-related factors were the explanatory variables. A generalized linear model from a binomial family and a logit link function were used to estimate the odds ratios and their 95% CIs for each explanatory variable. In these explanatory analyses, pairwise deletion was used for missing data.

Significance was set at α = 0.05 since this study was exploratory. All analyses were conducted using JMP® Pro software version 17.2 (SAS Institute Inc., Cary, NC, USA) and R version 4.3.1 (R Core Team, 2023).

Napping during the night shift was classified by TIB and SE, which were combined into six groups. TIB > 180 min and SE ≥ 70% was appeared in 23 night shifts (21.9%), TIB > 180 min and SE < 70% was present in 12 night shifts (11.4%), TIB 120–180 min and SE ≥ 70% was present in 24 night shifts (22.9%), TIB 120–180 min and SE < 70% appeared in 31 night shifts (29.5%), TIB < 120 min and SE ≥ 70% was present in four night shifts (3.8%), TIB < 120 min and SE <70% occurred in 11 night shifts (10.5%). SE ≥ 70% was 26.7% (4/15 night shifts) when TIB < 120 min, whereas SE ≥ 70% was 43.6% (24/55 night shifts) when TIB was 120–180 min, and 65.7% (23/35 night shifts) when TIB was > 180 min.

Note that the means (standard deviations) of the TND for each group are as follows: TIB < 120 min and SE ≥ 70% was 87.0 (16.2) min, TIB < 120 min and SE < 70% was 39.8 (17.8) min, TIB 120–180 min and SE ≥ 70% was 129.5 (16.5) min, TIB 120–180 min and SE < 70% was 90.8 (19.4) min, TIB > 180 min and SE ≥ 70% was 160.6 (16.4) min, and TIB > 180 min and SE < 70% was 104.7 (16.2) min.

There were significant main effects of TND on fatigue at the end of the night shift and Δfatigue from the start to the end of the night shift (Table 3, Model 3). However, TND did not have any significant effect on any of the sleepiness outcomes. The partial regression coefficient (standard error) of TND on fatigue at the end of the night shift was −0.09 (0.02) (p < 0.001) and onΔfatigue from the start to the end of the night shift was −0.06 (0.03) (p = 0.046).

There was no correlation between total steps and TND (r = −0.09, p = 0.339) or SE (r = 0.06, p = 0.569), while the correlation between total steps and TIB was weak (r = −0.23, p = 0.018). However, the number of total steps were no main effects on fatigue or sleepiness at any time point in the Model (Table 3, Model 4).

It may be noteworthy to focus on achieving a duration of at least 120 min. Previous studies reported that nurses who napped for ≥ 120 min during night shifts had lower cumulative fatigue and recovery from fatigue than those who napped for < 120 min [23, 52, 53]. The Japanese Nursing Association’s guidelines also recommend nap breaks of at least two hours [29]. Our results align with these previous studies. Additionally, a longer TIB is needed to achieve napping with higher SE. There was a moderate correlation between TIB and SE, and the exploratory analysis also showed that achieving SE ≥ 70% was easier when TIB was longer. However, a longer TIB did not significantly reduce fatigue and sleepiness. Similar findings have been reported in studies conducted in Brazil, where long naps during night shifts are routine. Those studies showed that naps > 3 h were not as effective for recovery after work as were 2–3-h naps [52, 53].

In contrast, napping with higher SE, indicating better napping quality, mitigated fatigue and sleepiness at the end of the night shift. This result does not conflict with the fact that good sleep quality has positive effects [27]. Additionally, if nurses achieve TIB ≥ 120 min and further achieve SE ≥ 70%, they may obtain at least one NREM/REM sleep cycle including slow wave sleep—the deepest NREM sleep—which plays an important role in energy recovery [56, 59]. Thus, we conclude that sufficient quantity and high-quality napping have benefits for nurses to mitigate fatigue and sleepiness.

Napping benefits not only nurses but also patients. Regarding the assumptions, the participating nurses were engaged in long night shifts (from 16:30 to 9:15), which were associated with fatigue and sleepiness [9, 10]. Fatigue and sleepiness, which peak at the end of the night shift without countermeasures [10, 60–62], impair nurses’ cognitive function and performance [63, 64]. In addition, nurses become busier in the early morning because they must provide care for patients who wake up. This is supported by our results that the number of total steps, indicating work demands [34], increased considerably from 6:00 a.m. to the end of the night shift. Thus, nurses face a higher risk of errors, such as medication administration errors and needlestick injuries, toward the end of a night shift. Previous studies showed that fatigue and other performance deficits in night-shift workers in the early morning were linked with major disasters [21], with more incidents during night than day shifts [65]. Our results showed that SE ≥ 70% reduced the KSS score by more than one point at the end of the night shift compared to SE < 70% when TIB was 120–180 min. A systematic review of sleepiness while driving reported a 1.4–1.9 times increased risk and a 5.4 times increased odds of lane deviation per one-point increase in KSS score [66]. If this is true for nurses, it may lead to errors in patient care. Napping with TIB ≥ 120 min and SE ≥ 70% during night shifts may prevent these errors and accidents, contributing to safer nursing.

Additionally, the difference in sleepiness at the end of the night shift can also reduce the risk of nurses getting into traffic accidents while driving home. Night-shift nurses are at a greater risk of dozing off while driving, driving off the road, or being involved in a car accident compared to nurses who work other shifts [15, 67]. Reducing these risks can enhance public safety.

TND was associated with fatigue but not sleepiness. A previous study showed that only SE in main sleep predicted sleepiness among sleep parameters, but not total sleep time [26]. Thus, SE may be the ideal napping parameter, as it is strongly associated with sleepiness. We cannot conclude why only SE was associated with sleepiness rather than TND and TIB, despite the strong correlation between SE and TND. As SE is a complex index of TND, TIB, SL, WASO, and BOL, further research is necessary to determine why SE is associated with sleepiness.

The number of total steps, one of the indicators of work demands [34], was not associated with either fatigue or sleepiness in this study. This supports that fatigue and sleepiness at the end of a night shift were mitigated by napping with sufficient quantity and quality.

Therefore, we propose using TIB and SE as target values instead of TND. As a first step, we recommend that nurses first aim to have a TIB of at least 120 min.

Exploring the factors associated with TIB ≥ 120 min and SE ≥ 70% revealed three factors that were commonly associated. The first one is sleep reactivity, which characterizes the degree to which stress exposure disrupts sleep, resulting in difficulty falling and staying asleep [68]. A trend similar to our results has been reported by Kalmbach et al. [68], who found that people with high sleep reactivity have low SE during their main sleep. Further, workers with high sleep reactivity had over five times the odds of developing shift work disorder compared to those with low sleep reactivity after transitioning to rotating shifts [69]. Therefore, this factor could be an important individual attribute for nurses working night shifts.

The second factor is the time spent using electronic devices before napping. The use of electronic devices during limited nap breaks not only reduces the duration to nap but also prolongs the time needed to fall asleep and increases alertness [70, 71]. A previous study found that the time spent on electronic devices was associated with TND [22]. The timing of use must be carefully considered.

The third factor is a prophylactic nap taken before the night shift, which has been reported as effective in lowering sleepiness during night shifts [54]. Our results showed that nurses who did not tend to take a prophylactic nap found it easier to achieve TIB ≥ 120 min. This may be because not taking a prophylactic nap before a night shift increases sleep pressure based on the two-process model [72, 73]. On the contrary, it is difficult to explain why nurses who took a prophylactic nap found it easier to achieve high-quality napping (SE ≥ 70%). Since our results used the one-month averages for each nurse, further studies should examine a prophylactic nap of each night shift in more detail, including individual circadian rhythm phases [74].

We recommend that night-shift naps involve a TIB of at least 120 min and an SE of at least 70%. Assessing sleep reactivity, pre-nap time on electronic devices, and prophylactic naps taken before the night shift may be useful in achieving effective napping quantity and quality. Additionally, nurse managers should implement hospital regulations to support nurses’ napping habits, and nurses should fully utilize the full nap break.

As previously reported [25, 75], it is important to note that nurses also must be aware that long naps during night shifts do not effectively reduce fatigue or sleepiness after nap breaks. Our results also showed that neither the quantity nor the quality of night-shift naps impacted fatigue and sleepiness after nap breaks, likely because of sleep inertia—defined as a brief period of decreased cognitive function or performance immediately after waking, which can temporarily obscure the recuperative effects of sleep [21]. Therefore, nurses should ensure that they have as much time as possible to recover from sleep inertia before restarting work.

One strength of this study is that it investigated all night shifts in a hospital ward for one month. Further, there were no deficits in napping parameters, fatigue, or sleepiness. However, there are some limitations. First, we showed that napping during night shifts had the best effect when TIB was at least 120 min and SE was at least 70%, but the effects of napping during night shifts on fatigue and sleepiness could not be fully explained by using TND, TIB, and SE. Second, selection bias may have been present. Our results were limited to one ward, and nurses who did not nap were excluded from the analysis. Third, napping was measured using wearable devices only, and sleep depth was not considered. Fourth, the effects of napping during night shifts were limited to its effects on fatigue and sleepiness, while its effects on other outcomes, such as performance and accidents, remain unclear. Hence, it is impossible to generalize our results to all shift workers. To establish the evidence for proper napping, validation involving other wards and hospitals, shift patterns, and different outcomes is required for large samples.