The use of sleep aids among Emergency Medicine residents: a web based survey

Sleep deprivation among resident physicians is of growing concern. Sleep deprivation is a cause of fatigue, which is characterized by decreased work capacity and efficiency accompanied by feelings of weariness, sleepiness, or tiredness [1]. Research on resident sleep deprivation has primarily focused on the detrimental impact of fatigue on patient care [1‐3]. Various specialties, including Obstetrics and Gynecology [4], General Surgery [5], Pediatrics [6], and Internal Medicine [2, 3] have shown sleep deprivation among residents negatively affects performance. However, fatigue also affects resident well-being and quality of life [7, 8]. Fatigue arising from extended work shifts and shift work have been implicated as impacting both physical and psychological disorders, ranging from an increased risk of motor vehicle crashes [9, 10], depression [11], and heart disease [12], to drug and alcohol abuse [13‐15]. Recently, research has begun to focus on strategies for maintaining wakefulness and decreasing fatigue while on shift [16]. To fully counteract the negative effects of fatigue, however, there is an overwhelming need for interventions to minimize fatigue induced by sleep deprivation. In the absence of sweeping changes to the system of resident education, further decreases in work hours, and abandonment of rotating shifts, few solutions remain.

We contend that many residents have recognized this conundrum and have taken steps to reduce their personal sleep deprivation by improving their likelihood of sleep during off duty hours through the use of pharmacological sleep aids. We aimed to determine the prevalence of sleep aid use among Emergency Medicine (EM) residents. We additionally sought to preliminarily explore how their use may relate to perceptions of sleepiness, stressors and work experience.

There were 823/3791 individuals (20.7% of all EM residents) who logged onto the website. Multiple choice questions (question 1 to 8 and demographics) were completed by 602/3791 (15.9%) and detailed questions on sleep aid and stimulant use (questions 9 to 11) were completed by 458/3791 respondents (12.1%).

Among the 602 residents responding to multiple choice questions, mean age was 30 years (SD 4 years), 71.4% were male. Race was 78% white, 11% Asian, 4% black and 4.0% other, race was not given for 3%. Hispanics comprised 4% of respondents. Residency format was PGY1–3 for 71%, PGY1–4 for 14%, and PGY2–4 for 14%. There were 32% 1st year, 32% 2nd year, 28% 3rd year, and 8% 4th year residents. For all EM residents in 2004, the ACGME reported a mean age of 31.2 years, 64% males, with 33% in the first year, 32% in the second year, 31% in the third year, and 4% in the fourth year of residency [18].

The mean Epworth Sleepiness Scale (ESS) score was 10.0 (SD 4.2, Figure 1). The ESS was normal (<10) in 332/602 (55%), indicative of excessive sleepiness in 227/602 (38%), and indicative of severe sleepiness in 43/602 (7%) of residents. The time taken to fall asleep was more than 10 minutes for 22%, between 5 and 10 minutes for 40%, and less than 5 minutes for 38% of respondents.

Factors causing fatigue and difficulties initiating sleep are shown in Table 1. To differentiate usual causes of fatigue from extraordinary events, residents were asked to report factors only if present at least four times a month for two consecutive months. In addition to pre-specified factors, residents volunteered that maintaining sociality (n = 8), problems sleeping (n = 5), relationship problems (n = 4), eating habits (n = 4), financial stress (n = 2), medications (n = 2), and hormonal cycles (n = 1) were causes of fatigue. Financial stress (n = 1), stress over household maintenance (n = 2), lack of exercise (n = 1), pregnancy (n = 1), and health problems (n = 1) were identified as causes of difficulty in initiating sleep.

The purpose of this study was to estimate the prevalence of sleep aid use among emergency medicine residents. Our main finding is that among respondents, 46% used sleep aids inclusive of alcohol, while 39% used sleep aids exclusive of alcohol. While reported sleep aids in our study were primarily over-the-counter medications, many prescription medications were used, including oxycodone, codeine, and benzodiazepines. These findings should be tempered by the self-reported nature of sleep aid use. For example, no respondents in our survey reported use of illicit substances, while prior research suggests Emergency Medicine residents are more likely to use cocaine and marijuana than residents from other specialties [19].

Participants whose difficulty initiating sleep was related to changing circadian rhythms were more likely to use sleep aids. Shift work is often cited as one of the most difficult aspects of life as an Emergency Physician, and adverse effects of shift work have long been known [20‐23]. Interestingly, fatigue or difficulty initiating sleep due to work-related emotional stress and family commitments were also found to be independent predictors of sleep aid use. While interventions to minimize the impact of circadian rhythm changes on performance are achievable and may aid in improving resident well-being and reducing fatigue-related performance errors [16, 24], interventions to reduce the work-related emotional stress and the impact of familial commitments on fatigue and sleep may also be of benefit.

This is the first known study to explore the use of sleep aids among Emergency Medicine residents, or among subject groups known to be at risk of fatigue and sleepiness; there is a dearth of data against which to compare our results. Sleep aid use in EM residents responding to this survey was higher than that reported for a general population survey of adults aged 18–45 in Detroit; in that study 18% used medications, 13% used alcohol, and 5% used medications and alcohol to facilitate sleep [25]. The higher prevalence of sleep aid use in our study may be associated with the stressors of shift work and the emotional stress of the emergency department. While we can formulate hypotheses on cause and effect of sleep aid use, it should be emphasized that our findings are only associative in nature. Further work is needed to explore predictors of sleep aid use, and how sleep aid use may affect performance in resident physicians.

There are several limitations to this study. The primary limitation is the low response rate (16%). The aim of our survey was to estimate the proportion of residents who use sleep aids. The number of residents we found to use sleep aids was 278 inclusive of alcohol. The lowest estimate of the true population proportion can be determined as the ratio of respondents using sleep aids to the entire EM resident population. It might be reasonable to assume, therefore, that at least 7.3% (95 CI 6.5–8.2%) of EM residents were using asleep aids at the time of this survey.

The direction of response bias arising from the low response in this study is not possible to ascertain; in comparison to eligible respondents, we achieved a similar demographic and experience distribution. However, we do not know whether responders were more or less likely to be sleep aid users than non-responders. Factors influencing response bias in this survey include distribution of the survey through Residency Directors. Every attempt was made to contact residents directly but concerns over privacy and local privacy policies prevented us from obtaining more than a partial list of residents' e-mail addresses. It is not possible for us to confirm that Program Directors forwarded the survey URL to their residents. By electing an anonymous survey methodology, it was not possible to ascertain from which programs residents participated. Further, a number of Program Directors did not respond to our initial communication requesting resident e-mail addresses, but instead inappropriately forwarded this request directly to their residents. This likely led to some confusion among residents, and to residents at some programs distrusting the project and its performance. We aimed to maximize the response rate by making the survey anonymous, offering participation in a drawing for a $100 gift certificate, and using regular reminders to prompt response. The sensitive nature of questions and lack of willingness of Emergency Medicine residents to participate in survey research may have further added to the low response.

In addition to response bias, the self-reported nature of the data is like to have led to under reporting of sleep aid use. Prior studies have surveyed drug use in both physicians and nurses [13], and while the use of anonymous surveys has been found to have sensitivity similar to toxicology screens via random urinalysis [26, 27], underestimation of use rather than overestimation is the trend found in these studies [28]. Finally, the use of a cross-sectional survey that was designed to estimate prevalence of sleep aid use is a limitation in the analysis of predictors of sleep aid use. Due to the limitations of low response rate, the use of across-sectional survey design, and the likely under reporting of sleep aid use, our results should be interpreted cautiously.

Even with a low response rate, our data suggest sleep aid use among EM residents may be common. How this affects performance, well-being, and health remains unknown.