Sleep medication and melatonin use among Norwegian nurses – A cross‐sectional study

Abstract Aim To estimate the prevalence of sleep medication and melatonin use among nurses and to assess if factors related to work, sleep or mental health, were associated with such use. Design A cross‐sectional study. Methods A questionnaire survey including 2,798 Norwegian nurses. Associations were estimated using a modified Poisson regression model. Results In total, 7.5%, 4.6% and 2.0% of the nurses included in the present study reported prescribed sleep medication, over‐the‐counter sleep medication or melatonin use in the last year, respectively. Short sleep duration, sleep problems and psychological conditions were strongly associated with both prescribed and over‐the‐counter sleep medication use. Nurses who worked more than 60 night shifts in the last year were at increased risk of sleep medication use.


| INTRODUC TI ON
Many nurses are engaged in shift work which involves irregular or unusual working hours outside the normal daytime work schedule (Wang et al., 2011). Shift work, in particular night work, early morning shifts and less than 11 hr between shifts (quick returns), disrupts the normal sleep-wake cycle and contributes to sleep disturbances and insufficient sleep (Kecklund & Axelsson, 2016).
While many seem to tolerate shift work well, some do not. In this realm, shift work disorder (SWD) refers to symptoms of insomnia and/or sleepiness due a work schedule overlapping with the habitual sleep time and which is associated with a reduction of total sleep time (Waage et al., 2009). In a large epidemiological study among Norwegian nurses, the prevalence of SWD ranged between 6.2% and 40.7% depending on work time schedule (Flo et al., 2012). Shift workers are also at increased risk of developing insomnia (Kecklund & Axelsson, 2016). Both SWD and insomnia have detrimental effects on health. A recent meta-analysis of longitudinal studies on shift work and mental health found an increased risk of depressive symptoms in shift workers, particularly in women (Torquati et al., 2019). Moreover, a recent Mendelian randomization study found evidence of a causal effect of insomnia on depression, diabetes and cardiovascular disease (Jansen et al., 2019).
A high prevalence of sleep problems in nurses is well documented Eldevik et al., 2013;Flo et al., 2012), but less is known about how nurses cope with these problems. Treatments for sleep problems include sleep hygiene, cognitive behavioural therapy, and pharmacotherapy. In terms of the latter a range of pharmacological sleep aids are used including benzodiazepines, nonbenzodiazepines (z-hypnotics), antihistamines, antidepressants, and neuroleptics with sedative effects. In Norway, the most commonly prescribed sleep medications are z-hypnotics with zopiclone as the most frequently used type (Sakshaug et al., 2017). Beneficial effect of short-term use of z-hypnotics for insomnia has been reported (Huedo-Medina et al., 2012), but in general, long-term use is not recommended due to drug tolerance, risk of addiction and other adverse effects. In the guidelines for the diagnosis and treatment of insomnia, the European Sleep Research Society recommends cognitive behavioural therapy as the first-line choice for treatment (Riemann et al., 2017). In persons aged 55 years or older, melatonin prolongedrelease tablets are indicated in treatment of insomnia. Melatonin is used as a sleep regulating medication which does not result in drug tolerance or addiction but is not recommended for treatment of insomnia in younger individuals (Riemann et al., 2017). Melatonin has until recently (1st of January 2020) only been available on prescription in Norway. Regarding sleep problems due to shift work, a recent Cochrane review found only low-quality evidence for melatonin to improve sleep duration after shifts, while insufficient evidence was found for other sleep medications (Liira et al., 2014).

| BACKG ROU N D
In the general population, prevalence rates for chronic and current use of sleep medication are commonly estimated to be between 4% and 10% (Ohayon & Lader, 2002;Omvik et al., 2010;Sakshaug et al., 2017). Sleep medication use has been found to be associated with older age, female sex, low socioeconomic status, and symptoms of sleep disorders, anxiety and depression (Ohayon & Lader, 2002;Omvik et al., 2010). Few studies have explored the prevalence of sleep medication use in a working population, and in nurses particularly, even though such use may have important implications on nurses' health and well-being. Since sleep medication use can result in drowsiness, it may also affect nurses' performance and constitute a safety risk for themselves and their patients (Riemann et al., 2017).
Among the studies that have been conducted during the last decade, the prevalence of sleep medication use in nurses has varied between 2% and 56% Francis et al., 2019;Futenma et al., 2015;Shy et al., 2011). Studies that have explored the association between work time schedule and sleep medication use have shown conflicting results (Futenma et al., 2015;Gomez-Garcia et al., 2016;Jensen et al., 2018;Sveinsdottir, 2006). In a recent Japanese cross-sectional study (Futenma et al., 2015), the risk of sleep medication use was higher in nurses working day and night compared to nurses working day, night or evening, while a higher risk of sleep medication use in nurses working night shifts compared to nurses working day or rotating shifts was found in a Spanish cross-sectional study (Gomez-Garcia et al., 2016). A recent Danish and an Icelandic study, both cross-sectional, found no difference in sleep medication use when comparing nurses with and without night work (Jensen et al., 2018;Sveinsdottir, 2006). Only one of the above-mentioned studies have also investigated the association between use of hypnotics and demographic and health factors and found use to be increased in nurses aged ≥27 years old, and in nurses with symptoms of depression, SWD and insomnia (Futenma et al., 2015). Neither of these studies included information about over-the-counter sleep medication or melatonin use although these are commonly used sleep aids in the general population.
A better understanding of the prevalence and predictors of sleep medication use in nurses can identify risk factors, facilitate treatment of sleep problems and improve health. Against this backdrop, the aim of the present study was twofold; first to explore whether work-related factors (shift work schedule, number of night shifts and quick returns the last year) were associated with prescribed sleep medication, over-the-counter sleep medication and melatonin use (separately) when controlling for age, sex, marital status, children living at home, work experience as a nurse, hours worked per week and type of workplace. And secondly, to explore whether indications of sleep problems (short sleep duration, sleepiness, shift work disorder and insomnia) or mental health problems (anxiety and depression) were associated with sleep medication and melatonin use independent of these work-related factors. We hypothesized that nurses with a work schedule including night work, or with a higher number of night shifts or quick returns in the last year would have a higher risk of sleep medication and melatonin use than those without night work or a lower numbers of night shifts and quick returns. Further, we hypothesized that the presence of sleep problems and mental health problems would be associated with increased risk of sleep medication and melatonin use, independent of these work-related factors.

| Study design
This was a cross-sectional study.

| Setting and participants
The present study was based on data from the SUrvey of Shift work, Sleep and Health (SUSSH) of Norwegian nurses. SUSSH was initiated in 2008 when a stratified sample of 6,000 nurses registered as members of the Norwegian Nurses Organization -a national labour union for nurses encompassing about 83% of all nurses in Norway (Søbstad et al., 2020) -was invited to participate in a survey.
The sample was based on random sampling within stratas divided by time elapsed since graduation (less than 12 months, 1-3 years, >3-6 years, >6-9 years and >9-12 years). This sampling method was chosen to obtain sufficient variation in length of work experience among the nurses in the study sample. Of the 6,000 questionnaires that were sent out, 600 were returned due to wrong addresses. Of the 5,400 eligible nurses, 2059 consented to participate (response rate 38.1%). In 2009, an additional random sample of 905 newly graduated nurses (graduated June 2009) was recruited (response rate 33.0%) to increase the study population. The total sample therefore included 2,964 nurses. All included nurses responded to a postal questionnaire with questions related to demographic factors, work, health and lifestyle behaviour. We excluded 166 nurses who reported that they were not working at time of completing the questionnaire. This resulted in an analytic sample of 2,798 nurses ( Figure 1).

| Independent variables
To assess the association between work-related factors and sleep medication and melatonin use, we included shift work schedule, number of night shifts and quick returns ("0," "1-20," "21-40," "41-60" and ">60") as work-related predictors, as these variables are particularly relevant for the working conditions we wanted to study.
The categories for night shifts and quick returns were chosen to obtain approximately equally sized groups. We collected detailed information about shift work schedule, but due to small groups, "permanent day work," "only evening work" and "2-shift (day and evening)" were categorized as "schedule without night work," and "only night work," "3-shift (day, evening and night)" and "other schedule including night work" were categorized as "schedule with night work." Based on experience from clinical practice and previous research, we included measures related to sleep and mental health as well as background variables that we a priori expected to be associated with sleep medication and melatonin use (Futenma et al., 2015;Omvik et al., 2010;S. Waage et al., 2014). This included sleep duration ("<6 hr," "6-<7 hr," "7-<8 hr," "8-9 hr," ">9 hr"), excessive daytime sleepiness, shift work disorder (SWD), and insomnia as measures of sleep, and anxiety and depression as measures of mental health. Registered background variables included sex, age (included as a continuous variable), marital status ("partner," "single"), children living at home ("yes," "no"), number of years of experience working as a nurse (included as a continuous variable, less than 6 months were reported as 0 years of experience), average hours worked per week (continuous variable) and type of workplace ("acute care" (somatic and psychiatric hospital), "community-based care" (nursing home, home-based care or healthcare centre)). A total of 108 nurses had reported "other" on type of workplace -this category was coded missing.

| Instruments
The questionnaire included several validated assessment tools.
Insomnia (yes, no) was measured by the Bergen Insomnia Scale (BIS) (Pallesen et al., 2008). The BIS asks how many days per week during the last month one has experienced problems related to sleep, and daytime consequences. If the nurse reported three or more days on at least one of the first four items (nocturnal problems), and three or more days on at least one of the last two items (daytime conse- Epworth Sleepiness Scale (ESS) (Johns, 1991) is a standardized subjective measure of chronic sleepiness that has been validated for use in Norwegian (Pallesen et al., 2007). The respondents are presented with eight different situations and asked to estimate the likelihood of dozing off in each situation using a 4-point scale (0 = no risk of dozing off, 3 = high risk of dozing off). A compositive score is calculating by adding the score of each item. A composite score of 11 or higher is suggestive of excessive sleepiness. Cronbach's alpha for the ESS in the present study was 0.74. The Hospital Anxiety and Depression Scale (HADS) was used to measure symptoms of anxiety and depression (Zigmond & Snaith, 1983). A validated Norwegian version of HADS was used (Bjelland et al., 2002). HADS consists of 14 items, of which 7 measures non-vegetative symptoms of anxiety and 7 measures nonvegetative symptoms of depression. The respondents score each item from 0 to 3, and a sum score for each subscale is created by adding the score of each item. In accordance with previous studies, a cut-off of 8 or above was used for categorizing anxiety and depression, respectively (Bjelland et al., 2002). In the present study, Cronbach's alpha for the anxiety and depression subscale was 0.81 and 0.82, respectively.

| Outcome variables
The nurses were asked whether they during the last year had used any of the following: 1) prescribed, or 2) over-the-counter sleep medication, or 3) melatonin ("yes," "no").

| Analysis
We compared the prevalence of prescribed and over-the-counter sleep medication and melatonin use by categories of the predictors using Pearson's chi-square test. In this analysis, continuous variables were categorized, to make it possible to compare the prevalence of the outcomes in different groups. Age was categorized as "21-25 years old," "26-30," "31-35", "36-40" and ">40," work experience as "<1 year," "1-2 years," "3-5," "6-10" and ">10," and average hours worked per week as "≤20 hr," "20.1-30," "30.1-37.5," "37.6-40" and ">40." We used a modified Poisson model to estimate crude and adjusted relative risks (RR) with 95% confidence intervals (CI) for the association between each predictor and each of the three outcome variables (Zou, 2004). We first ran separate models to estimate the association between each of the three work-related factors continuous independent variable, we tested for departure from linearity in associations between the independent variable and each of the three outcomes by comparing a model with and without a quadratic term using a likelihood-ratio test. We also explored whether the association between SWD, insomnia, excessive sleepiness, anxiety and depression and each of the three outcome variables varied according to work time schedule (with or without night work), number of night shifts or quick returns, by including an interaction term in the model, also investigated by a likelihood-ratio test. All statistical analyses were conducted in Stata/SE version 16.0 (StataCorp, College Station, TX, USA).

| Ethics
This study was approved by the Regional Committee for Medical Research Ethics in Western Norway (No. 088.88). Informed consent was obtained from all participants.
In the adjusted regression models for the work-related predictors, we only found an association between number of night shifts last year and prescribed or over-the-counter sleep medication use ( Prescribed and over-the-counter sleep medication use were strongly associated with short sleep duration (less than 7 hr), SWD, insomnia, and symptoms of depression and anxiety when adjusting for all background variables as well as the work-related factors (Table 4). Short sleep duration of less than 6 hr, SWD, insomnia and anxiety were also strongly associated with melatonin use. We found a statistically significant interaction effect (p = .05) between SWD and work schedule for the association with prescribed sleep medication use; the association between prescribed sleep medication use and SWD was stronger for nurses with a work schedule including night work than for nurses with a schedule not including night work.

| D ISCUSS I ON
The aim of this cross-sectional study comprising 2,798 Norwegian nurses was to estimate the prevalence of sleep medication and melatonin use and to identify factors associated with such use.
The association between shift work and sleep problems in nurses is well documented, but few studies have explored the use of different pharmacological sleep aids in nurses. By including data on both prescribed and over-the-counter sleep medication as well as melatonin use, we provide a more complete overview of use than in previous studies. In total, 7.5% of the nurses reported prescribed  Contrary to what we hypothesized, we did not find any association between work schedule (schedule with or without night work) and number of quick returns and sleep medication or melatonin use.

% (N) p-value a % (N) p-value a % (N) p-value a
However, we found a higher risk of prescribed and over-the-counter sleep medication use in nurses with more than 60 night shifts last year. Shift work, especially night work, and quick returns, have been found to be associated with increased risk of SWD and insomnia (Eldevik et al., 2013;Øyane et al., 2013). The lack of an observed increased risk in sleep medication use in nurses with a work schedule with night work or a high number of quick returns in the present study could partly be explained by the "healthy worker effect" (Knutsson, 2004

TA B L E 2 (Continued)
than in 3-shift workers (working day, evening and night) (Futenma et al., 2015). However, as in the Danish study, both groups had a work schedule that included night work. In a cross-sectional study including 394 Icelandic nurses who worked either permanent day, 2shift (day and evening), or 3-shift (day, evening and night), there was no statistically significant difference in sleep medication use across groups (Sveinsdottir, 2006). In that study, the nurses were asked how often they had used sleep medication during the last 6 months.
Another cross-sectional study including 635 nurses working in seven hospitals in Spain measured sleep medication use based on an item in the Pittsburgh Sleep Quality Index (Gomez-Garcia et al., 2016).
That study reported a higher sleep medication use in nurses working only nights compared to nurses working day or rotating shifts. Based on these results and the results in the present study, the association TA B L E 3 Crude and adjusted relative risk of sleep medication (prescribed or over the counter) and melatonin use by demographic and work-related factors among Norwegian nurses (N = 2,798) between work-related factors and sleep medication use remain unclear and should be further investigated, preferably by longitudinal designs initiated before the respondents start working as nurses.
In accordance with what we hypothesized, we found short sleep duration, and symptoms of SWD, insomnia, anxiety and depression to be strongly associated with sleep medication and melatonin use. This is also in accordance with the above-mentioned Japanese study (Futenma et al., 2015). Adjustment for demographic and workrelated factors hardly affected the results. The observed associations between the sleep and mental health factors and the three outcome variables did not vary by work time schedule, number of night shifts or quick returns, except for SWD, in which the association with prescribed sleep medication use was stronger in nurses with a work schedule including night work compared to nurses without night work.
In the present sample, we found a high prevalence of SWD and insomnia (34.7% and 52.0%, respectively). A recent study with a representative sample of the Norwegian population, reported that 20.0% had symptoms of insomnia (Bjorvatn et al., 2018). In that study insomnia was defined with a time frame including the last 3 months based on the third edition of the International Classification of Sleep Disorders (ICSD-3) diagnostic criteria (Sateia, 2014), while in the present study, we used the definition based on the previous edition, the ICSD-2 (American Academy of Sleep Medicine, 2005). Still, the present study suggests that insomnia is very common among nurses.
Contrary to what we hypothesized, the difference in prevalence and risk of sleep medication and melatonin use for nurses with and without symptoms of excessive sleepiness was small and not statistically significant. Excessive sleepiness may be due to sleep problems (Boyes et al., 2017), and thereby expected to be associated with increased sleep medication use. On the other hand, prescribed sleep medication use may as well induce excessive daytime sleepiness, since drowsiness after the sleep period is a potential side effect (Liira et al., 2014). Due to the cross-sectional design of the present study, we cannot assess the direction of this association. The prevalence of excessive daytime sleepiness was high in the present sample compared to what has been reported in the general Norwegian population; 27.3% compared to 17.7% in a sample of 2,301 Norwegian adults (Pallesen et al., 2007). None of the previous studies that TA B L E 4 Crude and adjusted relative risk of sleep medication (prescribed or over the counter) and melatonin use by indication of sleep and mental health factors among Norwegian nurses (N = 2,798) An important limitation of the present study entails the small number of nurses in the groups of some of the predictors leading to low statistical power in parts of the analyses. Another important limitation was the low response rate (mid-thirties). This may have affected the internal validity and generalizability of the study. We do not know whether non-responders were more or less likely to use sleep medication. Hence, it is difficult to assess how the low response rate might have affected the results. In general, non-responders are often found to be of poorer health, but whether or not non-response leads to biased results varies by type of health outcome studied (Knudsen et al., 2010;Langhammer et al., 2012;Vercambre & Gilbert, 2012).
As a result of the low response rate, the reported overall prevalence of sleep medication and melatonin use should be interpreted with caution. In addition, the cohort is younger than the Norwegian nurse work force overall. This could lead to an underestimation of the true prevalence of sleep medication and melatonin use given that such use was found to be less prevalent among younger than older nurses in the present study. More than 90% of the nurses included in the cohort were female, this is, however, representative of the sex distribution of Norwegian nurses (Statistics Norway, 2019).
Other important limitations include the lack of information on type and amount of medication, frequency of use, and the fact that we only had self-reported data on sleep medication use. Also, as sleep medication use by many is regarded as undesirable, the results may have been influenced by social desirability bias (Krumpal, 2013).
Assessments of presence of sleep problems, mental health problems and hypnotic use were also exclusively based on self-reported information; thus, we did not have any information from clinical interviews, clinical records or health registry data. We also do not know how the respondents interpreted the term "sleep medication" (e.g. in the case of antidepressants with sedative effects), and what type of sleep medication they used. The prevalence of sleep medication and melatonin use reported in the present study is not necessarily representative of the prevalence among nurses today given that our data were surveyed more than a decade ago. In the general Norwegian population, total sale of prescribed sleep medication, measured as dispensed defined daily doses, has decreased during the last decade, while the total sale of melatonin has increased (Sakshaug et al., 2019).
Further, the association between sleep medication and melatonin use and the included predictors may have changed in the last decade, but in what direction is harder to predict. Hence, in summary, future studies should include newer and more detailed information of the specific medications used, preferably based on prescription registries.
We do not know if these data are generalizable to other countries as work hours, workload and other relevant factors differ between countries. Similar studies in other countries are therefore warranted.

| CON CLUS ION
In total, 7.5%, 4.6% and 2.0% of the nurses included in the present study reported prescribed sleep medication, over-the-counter sleep medication and melatonin use during the last year, respectively. should also include data on shifts, mental disorders and medication use from sources such as health registries.

| RELE VAN CE TO CLINI C AL PR AC TI CE
Special attention should be paid to nurses working a high number of night shifts, and to nurses with indications of sleep problems or anxiety/depression, independent of work time schedule. Longitudinal studies are needed to assess the long-term effect of night work on sleep medication and melatonin use.

ACK N OWLED G M ENTS
We are grateful to all nurses participating in the study and to the Norwegian Nurses Organization (NSF) for funding and help with recruitment.