Prevalence of sleep disturbance and associated risk factors in UK Biobank participants with alcohol use disorders and major depression

Introduction

Major depressive disorders (MDD) and alcohol use disorders (AUD) are among the most common psychiatric illnesses, with lifetime prevalence rates of 20.6% and 29% respectively (1, 2). These disorders are frequently comorbid; those with AUD have a 1.2-fold increased likelihood of having MDD in the previous year (2). The co-occurrence of these conditions is associated with increased severity of symptoms and overall worse prognosis (3, 4). Sleep disturbance (SD) is common in both conditions and when present, impacts associated morbidity, treatment response, and relapse risk (5-8). There is also evidence to suggest that SD is causally linked to alcohol dependence and is associated with certain circadian clock genes in those with AUD (9-12). However, the relative prevalence of SD in patients with AUD and/or MDD and its underlying etiological and associated risk factors are not well understood.

Prior studies of patients with AUD have shown that SD is associated with younger age, depressive symptoms, increasing severity and frequency of alcohol use, and psychosocial stress (7). In patients with MDD, the incidence of insomnia symptoms appears to increase with age (13, 14). SD occurring in the context of comorbid AUD and MDD has not been previously explored. The rates of SD in subjects with AUD, MDD, and both AUD+MDD in a single large population-based sample have not been compared before and research examining whether clinical and demographic factors associated with SD differ in each of these groups is lacking. Knowledge regarding the prevalence and factors associated with SD in patients with AUD and/or MDD could potentially help guide future screening, prevention, and treatment strategies for these conditions (15).

In this study, we aimed to utilize data from the UK Biobank (UKB) to examine the rates of SD and mean sleep duration in patients with AUD, MDD, and MDD+AUD. Further, we examined the association between SD and various clinical and demographic risk factors in each group. Finally, we investigated whether the prevalence of SD and self-reported sleep duration in these groups differed from controls after adjusting for associated demographic and clinical risk factors.

Methods

Sleep measurements and case definitions

Figure 1 summarizes the case and control definitions of SD, AUD, and MDD, as well as exclusion criteria.

• Download figure
• Open in new tab

Figure 1:

AUD/MDD Diagnosis group definitions derived from the UK Biobank

Abbreviations: AUD = alcohol use disorders; AUDIT = Alcohol Use Disorders Identification Test; GP = general practitioner; ICD = International Classification of Disorders - 10^th Revision; MDD = major depressive disorders; MPD = multiple personality disorder; UKB = U.K. Biobank

SD was defined based on the response to the question: “Do you have trouble falling asleep at night or do you wake up in the middle of the night?” Those who responded “usually” were considered to have SD. Those who responded “never/rarely” or “sometimes” were considered to not have SD. Sleep duration was determined by subject self-report based on response to the question “About how many hours of sleep do you get in every 24 hours?”

Responses to the Alcohol Use Disorders Identification Test (AUDIT) were used to identify patients with AUD. An AUDIT score of ≥ 8 was considered to indicate AUD (17). About 100,000 participants completed the AUDIT, which was included an optional Mental Health Questionnaire (MHQ) (Figure 1). We defined participants as having “probable MDD” using criteria from previous studies, which also relied on completion of the MHQ (18)(Figure 1). Briefly, participants were classified as cases if they self-reported anhedonia or depression lasting longer than two weeks and had seen a general practitioner or a psychiatrist for anxiety/depression OR had a depression diagnosis (ICD-10 codes of F32, F33, F34, F38, F39, F40) in the medical record. The sample that completed the MHQ had similar prevalence rates of mental disorders when compared to a contemporaneous epidemiological study (19). To remove potential confounders of sleep with AUD and MDD, we excluded participants with a diagnosis of schizophrenia, bipolar disorder, personality disorder, and those currently taking antipsychotic medication(s). Finally, to ensure independence among the observations, we removed related individuals (second cousin or more related) by keeping only one participant from each family. Using these criteria, the cohort was divided into those with MDD alone, AUD alone, both (MDD+AUD), and controls with neither disorder (Figure 1).

Results

Requisite data were available for 47,825 subjects from the UKB, allowing them to be categorized into those with AUD, MDD, and controls. Based on our predetermined case definition criteria, the cohort was divided into those with MDD alone (n = 5,991), AUD alone (n =12,952), both (MDD+AUD) (n = 3,219), and controls (n = 25,663). Sample sizes varied slightly between analyses based on the amounts of missing data for clinical covariates.

Discussion

This is the first examination of sleep-related complaints in a single population-based sample comprising of subjects with AUD, MDD and MDD+AUD to our knowledge. The main findings were that the clinical and demographic factors associated with SD and sleep duration were similar in controls and across all diagnostic categories. After adjusting for diagnostic category, older age, female sex, white ethnicity, and greater BMI, neuroticism score and alcohol consumption were associated with SD. Patients with both AUD and MDD had the greatest sleep-related morbidity, with unadjusted and adjusted prevalence of SD significantly higher in this group compared to any other group. While the prevalence of SD in the AUD only category did not differ from controls in the unadjusted models, after adjusting for demographic and clinical factors, the prevalence was higher in the AUD group. Finally, after adjusting for both demographic and clinical factors, the prevalence of SD did not differ between the AUD only and MDD only groups.

In the general population, middle to older age has been shown to be associated with greater risk of insomnia symptoms, especially sleep maintenance difficulties (22), and women have been reported to have a 1.4-fold increased risk of SD compared to men (23). Lower socioeconomic status and psychological distress have also been found to be associated with SD (24),(25). On the other hand, in subjects with AUD, SD has been shown to be associated with a younger age, greater severity of AUD and alcohol use, and psychosocial stress (7). In our study, SD was associated with similar demographic and clinical factors across the different diagnostic categories, suggesting some degree of shared etiological and/or risk factors for SD in each of these conditions.

Further, when we examined whether the predictors of SD differed in terms of the magnitude of their influence in each of the groups, age showed a smaller effect in the MDD only group while sex showed a smaller effect in controls versus the other groups. The prevalence of insomnia symptoms was shown to increase with age in subjects with MDD in a single study where the baseline prevalence of insomnia was high at 77% in the 16-to-24 year age group and 90% in the 55-to-64 year age group (13). In our study, the prevalence of SD in the MDD group was 39%, higher than in controls and the AUD only group. The UKB included subjects were older, between the ages of 40-69 years at enrollment, which may have reduced the likelihood of detecting the effects of age on SD. Multiple previous studies have shown that the rates of SD are similar in men and women in early alcohol recovery (5, 6, 26, 27). The differential impact of sex on the prevalence of SD in AUD ± MDD compared to controls that we observed needs to be explored further in future studies.

Previous population-based estimates from the US and UK indicate that about 36% of adults endorse at least one insomnia symptom (13, 22). Data from our sample which utilized a single question demonstrated rates that were slightly lower (27.8%) in the controls, which may be partly due to the exclusion of those with MDD or AUD from this group. SD is highly prevalent in AUD in and the rates vary depending on whether patients are treatment seeking and the temporal relationship with last exposure to alcohol i.e. current active use, acute withdrawal, early recovery, or sustained recovery. The rates of SD in clinical samples are estimated to range between 36-91%, depending on the stage of AUD (5-7). SD is also extremely common in clinical samples with MDD, with a prevalence as high as 70-80% (14). In a population-based survey from the UK, 83.2% of subjects with MDD endorsed SD and 52% noted moderate severity of sleep-related symptoms (13). The overall prevalence of SD was lower in our sample compared to most of these previous studies, likely reflecting the different populations studied, especially in terms of age restrictions in the UKB, and different questions used to ascertain SD. To our knowledge, no previous studies have examined the prevalence of SD in subjects with comorbid AUD and MDD. Patients with MDD+AUD had greater rates of SD compared to controls and those with either condition alone in our study; this is in keeping with prior literature demonstrating that other morbidity is worse in patients with both conditions versus either condition (3, 4).

Neuroticism is a stable personality trait that has been shown to be associated with increased risk of both MDD and substance use disorders (SUD) and appears to predict MDD more strongly than SUD (28). Neuroticism has been linked with poor sleep quality and sleep habits (29-31). In our study, the neuroticism score was much higher in patients with MDD alone or MDD+AUD and was strongly predictive of SD after accounting for diagnostic category. Further, as expected, those with AUD reported much higher alcohol consumption than those with MDD alone and controls. Greater levels of alcohol consumption have also been shown to be associated with SD and short sleep duration (32, 33). The differences in the rates of SD between the diagnostic groups were attenuated as expected in our study after accounting for clinical factors, including neuroticism and alcohol consumption.

Our study has some limitations. The UKB sleep data relied on self-report and used a single question inquiring about sleep disturbance and duration. Our case definition for MDD utilized general practitioner diagnoses of ICD codes as well as self-reported depression/anhedonia, which could have led to misclassification of cases. Furthermore, our analysis of associated demographic and clinical risks was limited to a select set of measures that are known to affect sleep. After restricting our sample to patients who had the requisite data from the MHQ, we were left with 47,825 out of the total > 400,000 independent participants from the UKB, which may have biased the results. Recent investigations have revealed that those who completed the MHQ were better educated and of higher socioeconomic status; however, the rates of self-reported diagnoses have been found to be similar to population-based estimates (19). Even in light of these limitations, ours is the largest investigation of sleep disturbance and duration in a sample with AUD and MDD and first to show greater sleep morbidity in the AUD+MDD group.

Conclusion

Clinical and demographic factors associated with sleep disturbance are similar across AUD, MDD, and MDD+AUD groups, suggesting potentially similar underlying risks and etiology. Those with MDD+AUD had worse sleep disturbance, further demonstrating that when these conditions co-occur they are associated with significantly greater morbidity. The lower prevalence of sleep disturbance in the AUD alone group prior to adjusting for demographic factors was likely influenced by overrepresentation of younger men in this category. Future studies should examine a more expansive range of demographic, clinical, and genetic factors to further understand the underpinnings of sleep disturbance in these disorders and identify potential etiological differences.