The association between proportion of night shifts and musculoskeletal pain and headaches in nurses: a cross-sectional study

Design and study population

The data is from a diary study carried out by The National Institute of Occupational Health in Norway (STAMI) through the project “Shift work, sleep and pain” from October 2014 to November 2015.

An invitation was sent by post or by e-mail to a random sample of members of the Norwegian Nurses Association (NSF) (n=22,500), to reduce selection bias. The inclusion criteria were that the participants had to be authorized nurses, working in more than 50% position, being between 18 and 63 years of age, and be working in a 2-shift (day, evening) or 3-shift (day, evening, night) schedule or night shifts only. Participants who were pregnant, breastfeeding, or had been on sick leave for more than two weeks in the last six months were excluded. A total of 5,400 participants received a further invitation to participate in the study. 1032 participants answered the questionnaire at baseline. Of these, 727 participants also answered a diary questionnaire for up to 28 days. After correcting for the inclusion criteria, 684 nurses completed both the questionnaire at baseline at the diary questionnaire for a minimum of seven days.

A priori power calculations assuming 80% power indicated that a sample size of 100 was sufficient to detect a 20% difference in pain complaints, if pain complaints were treated as a continuous variable. For the present analysis, the pain variable was dichotomized, which reduces statistical power. To compensate, the number of participants was increased 6-7 fold.

Compliance with ethical standards

All participants gave written informed consent to participate in the study, which was conducted in accordance with the Declaration of Helsinki and was approved by the Norwegian Regional Committee for Medical Research Ethics (approval number 2012/199/REK sør-øst B).

Data collection

The questionnaire at baseline contained questions about demographic conditions, lifestyle, sleep, physical and mental complaints as well as psychosocial and physical working conditions. Participants who answered the questionnaire were included in the study. After inclusion, an SMS (or e-mail if not owning a smartphone) was sent to the participants daily at 21.00 with a reminder to answer a diary questionnaire as soon as possible. Recall bias is a known limitation in questionnaire studies. To limit recall bias, the participants were asked to answer daily questions about pain, working hours, and sleep for a total of 28 consecutive days. Answers received 24 hours or more after the reminder, were not considered.

Exposure variables

Morning shift was defined as a shift that started between 05:00 and 12:00. An evening shift was defined as a shift that started between 12:01 and 18:00, while a night shift was defined as a shift that started between 18:01 and 04:59. Hence, each workday was categorized accordingly. Days without work was categorized as days off.

The number of morning, evening, and night shifts worked for each participant for 28 days was calculated by aggregating the variables. The proportion of night shifts was then calculated by dividing the number of night shifts by all shifts worked. The proportion of night shifts was categorized as an ordinal variable. The categories used were <25%, 25-49.9%, and ≥ 50%, as suggested by Härmä et al. (14).

Sleep as a potential mediator

The diary sleep questions were derived from Carney et al. (24). These questions were:

1. “What time did you get into bed?”, 2. “What time did you try to fall asleep?”, 3. “How long time (in hours and minutes) did it take for you to fall asleep?” which is described as sleep onset latency (SOL), 4. “How many times did you wake up, not counting your final awakening?”, 5. “In total, how long did these awakenings last?” which is described as wake after sleep onset (WASO), 6. “What time was your final awakening?”, 7. “What time did you get out of bed after your main sleep?”, 8. “Did you wake up earlier than planned?”, 9. “How long did you sleep in addition to your main sleep during the past 24 h?”

Daily total sleep time (TST) was calculated by subtracting SOL and WASO from the time difference between when the participants was trying to sleep and the time of the final awakening. TST for all days was then summed for each participant. Furthermore, the variable was aggregated to find the number of days responded. The total sum of the TST was finally divided by the number of days responded to find the average TST. The average TST was treated as a continuous variable.

Outcome variables

Self-reported subjective musculoskeletal pain complaints and headache were measured using a diary questionnaire in which the participants scored pain in the last 24 hours on a four-point Likert scale. The options were not troubled by pain=0, somewhat troubled by pain=1, fairly troubled by pain=2, and very troubled by pain=3. Participants rated pain in five regions: the neck, shoulder, and upper back, lower back, upper extremities, lower extremities, and headaches.

In this study we wanted to study the overall effect of night shifts on pain prevalence, and not the variation in pain intensity. Thus, for each day and pain region, the pain ratings were dichotomized and summed for the data collection period of 28 days to find the number of days with pain. The five outcome variables, musculoskeletal pain and headache, were treated as discrete count variables.

Statistical analysis

All analyzes were done in R (RStudio Team, 2022). The estimates from the analyzes were presented with a 95% confidence interval (CI). The significance level was set at ≤ 0.05. The analysis was adjusted for age and BMI.

In this study, the data collected over 28 days were treated as cross-sectional data. Negative binomial regression was used, as the model was overdispersed (25), to examine the association between the proportion of night shifts and the outcome variables. The reference category was <25% night shifts, which was compared with 25-49.9% and ≥ 50% night shifts. Furthermore, it was investigated if TST could mediate the association between the proportion of night shifts and the outcome variables. If the combined effect for path a and path b is significant, it can be concluded that there is a mediation (see figure 1) (26). The indirect effect can be found by multiplying the partial derivative of the equation for path a with respect to X by the partial derivative of the equation for path b and path c’ with respect to M (27). Linear regression was used to analyze path a, and negative binomial regression was used for path b and path c’. As described in Geldhof, Anthony (27), when linear regression has been used in path a and negative binomial regression in path b and path c’, the conditional indirect effect can be estimated by the formula:

• Download figure
• Open in new tab

Figure 1.

Mediation model. X is the exposure variable, M is the mediator, and Y is the outcome variable. The indirect effect is the combined effect between path a (X➔M) and path b (M➔Y). The direct effect is path c’ (X➔Y).

To calculate the confidence intervals for the conditional indirect effect, bootstrap was used as proposed in Geldhof, Anthony (27). 1000 replicates were calculated using bootstrap. In all analysis adjustment were made for age and body mass index (BMI) as the variables theoretically can be associated with both the exposure and outcome (28-31).

In addition, an offset variable was used for all the unadjusted an adjusted analyzes with the outcome variables to adjust for the difference in number of days responded to the diary questionnaire. Offset variables can be used in negative binomial regression to adjust for differences in time for count variables (25).

Descriptive statistics

The sample consisted of 684 nurses in shift work varying from nurses who worked a 2-or 3-shift schedule or night shift only. The diary questionnaire was answered 84-85% of the days. Of the 727 participants who responded to diary questionnaire, 43 participants were excluded due to the following reasons: they had responded less than seven days to the questionnaire (n=22), not working as a nurse (n=14), were on sick leave (n=1), or had missing data for one of the confounding variables (n=6). Table 1 presents descriptive statistics consisting of demographic variables, work related variables, pain, and sleep. Mean age was 40.9 ± 11.1 years with a range of 22-63 years. Table 2 presents descriptive statistics for total sleep time (TST) for the different exposure groups used in the analysis.

Proportion of night shifts and musculoskeletal pain and headaches

Those who worked ≥ 50 % night shifts had significantly lower occurrence of pain in the lower extremities than those who worked <25 % night shifts in the adjusted analysis (IRR 0.69 95 % CI 0.51, 0.94) (Table 3). For pain in other regions, no significant association was found with proportion of night shifts. As a sensitivity analysis, only nurses working a 3-shift schedule were included in the analysis. No significant associations were found in the sensitivity analysis (Table 4).

Mediation analysis

No indication of a mediation effect with TST as mediator between proportion of night shifts and for any of the pain regions was found (Supplementary table 1).

Strength and limitations

Strengths of the present study are a large, randomized sample of nurses and the longitudinal data collection. The data collection takes place close to exposure and outcome. It is conceivable that this can reduce reporting bias which can often be the case with questionnaires assessing complaints over weeks or months retrospectively (32).

Another strength of this study is that sensitivity analysis was carried out, analyzing only the participants who worked in a 3-shifts schedule. The sensitivity analysis produced essentially similar results, confirming no association between a higher proportion of night shifts and musculoskeletal pain and headaches.

Musculoskeletal pain and headaches were measured with a four-point Likert scale, which is a type of unidimensional NRS scale that is valid and reliable (37). The participants had a deadline of 24 hours to answer the diary questionnaire. Additionally, in this study, the cut-off value for pain was set to a minimum of “somewhat troubled”, and the variable was categorized as no pain or pain. Therefore, this study only distinguished between no pain and pain each day, making it difficult to compare the prevalence of pain to similar studies with a retrospective design. However, it is a strength that the sample was based on a randomized drawing from the Norwegian Nurses Association (NSF) and 90 % of the participant was women, which is similar to the nurse population in Norway (38), supporting the generalizability of this study.

A diary is preferable to a retrospective questionnaire, as sleep can vary greatly from night to night. Diaries have also been considered the gold standard for subjective sleep reporting (39). It is recommended to answer sleep diary questionnaire within one hour after the end of a sleep period (24). However, in the present study subjects were asked to report pain and sleep the last 24 hours. The sleep measures in this study may therefore have been prone to bias. It is also a limitation that we asked about exposure (night work) and outcome (pain) in the same questionnaire.

A limitation of this study is that the number of days data was collected from the diary questionnaire varied between participants from seven to 28 days. The data collection period had a duration of 28 days with a diary questionnaire, but the number of participants who completed the questionnaire became fewer towards the end of the study. Generally, excluding participants with missing data who may increase the risk of selection bias, and including them may increase the risk of information bias. To adjust for different number of days reported in the diary, an offset variable was used, which was the natural logarithm of the number of days answered on the diary questionnaire. This has also been done in other studies to adjust for time for follow-up (40). Thus, it is the rate of days with pain and not the number of days with pain that is analyzed (25).

Concluding remarks

In this study, a higher monthly proportion of night shifts was not associated with a higher occurrence of musculoskeletal pain and headaches. On the contrary, a lower occurrence of pain in the lower extremities was found for those who worked ≥ 50 % night shifts compared to those who worked <25 % night shifts. The results have implications for the design of future studies on the effect of shift work on musculoskeletal pain and headache.