Prevalence and associated factors of DSM-5 insomnia disorder in the general population of Qatar

As Qatar is divided into eight administrative municipalities, the households in each municipality were stratified according to geographic location and the type of household residents (non-migrants or Qataris, and migrants or resident expatriates hereafter referred to as non-Qataris). The target population included residents of Qatar aged 18 and above who were living in residential units during the survey reference period. For this study, the non-residential units such as army barracks, hospitals, dormitories, camps of blue-collar workers and prisons were excluded. A probability-based sample was then drawn from these strata. As Qataris represented the minority group in the population, they were oversampled. A validated adaptation of the Kish method [38] was utilized to randomly select one eligible adult inside each household in Qatar [43]. Constructed weights accounted for sampling disproportionality and nonresponse.

For our survey design, the following formula: n = z2(p (1 − p)/e2) deff [16, 38] was used to calculate the required sample size, which was estimated at 1160. The estimate was based on α value of 0.05 or z = 1.96; p or estimate of the proportion set at 50.0% to identify the largest sample size requirement; desired sampling error e = ±3.5%; deff or design effect = 1.48; and a survey response rate = 48.0%. Prior face-to-face surveys conducted by SESRI in the same population have determined the estimates of design effect and survey response rate.

For the interviews, the measures were available in both English and Arabic, enabling participants to select their preferred language. A team formed of four bilingual research members translated and adapted the measures in the following steps: 1) Each measure was independently translated from English to Arabic by the first author and a researcher; 2) Another two researchers who had not seen the original English version of these measures were assigned to back-translate the two Arabic versions to English; 3) Finally, all team members met and reviewed the two Arabic versions of the questionnaire and compared the original and back-translated English versions. Upon their reviews and comparisons, minor discrepancies in translation arose. These were solved by consensus among team members. Please see “Additional file 1” for details of survey questions asked in both languages.

Qatar University’s Institutional Review Board approved the study (QU-IRB 339-E/14). The survey questionnaire was programmed in BLAISE and administered using CAPI (Computer Assisted Personal Interview) during face-to-face interviews conducted by trained interviewers in the households of consenting participants who were 18 years old and above. A total of 50 interviewers visited the sampled households and introduced the study. Written informed consent was obtained from all participants after explaining that study participation is voluntary and anonymous. Out of the total sampled households, approximately 4.8% refused to participate. Other reasons for non-participation included language barrier, illness/disability, and vacant housing units. A total of 1611 completed interviews were recorded and the survey response rate was 52.1%. As per household type, the completed interviews included 803 Qataris and 808 non-Qataris. The corresponding maximum sampling error percentage was +/− 3.9 and 3.8 percentage points for Qataris and non-Qataris, respectively.

After data collection, all individual interviews were merged and saved in a single BLAISE data file. This dataset was then cleaned, coded, weighted to adjust for probability of selection and non-response, and saved in STATA version 16 for analysis.

The eight-item Sleep Condition Indicator (SCI) was used to screen for insomnia disorder in the past month based on the DSM-5 criteria [20]. Each item was scored on a 5-point scale (0–4). The scale captured sleep continuity, quality, satisfaction with sleep in addition to symptoms severity, consequences on mood, cognition and day performance and duration of sleep problems. A composite score of all items was computed ranging from 0 to 32 with higher scores indicating better sleep.

As operationalized by the SCI, we measured four main criteria for insomnia disorder as per DSM-5 criteria. The first criterion was difficulty initiating or maintaining sleep. This was based on reports of taking more than 30 min to fall asleep or if waking up during the night, staying awake for greater than 30 min in total; and a rating of sleep quality as average, poor, or very poor. The second criterion was significant distress, which was based on endorsing “somewhat”, “much” or “very much” for the extent that poor sleep in the past month has troubled the respondent in general and endorsing “somewhat”, “much” or “very much” to one of two questions about the extent has poor sleep in the past month affected respondents’ mood, energy, or relationships or affected concentration, productivity, or ability to stay awake. The third criterion was in relation to the frequency of sleep disturbances, which was based on endorsing a minimum of 3 nights per week for the frequency of encountering sleep problems. The fourth criterion was related to the duration of sleep disturbances, based on endorsing a minimum duration of 3 months for having sleep problems.

We estimated 30-day prevalence of insomnia disorder using two methods in relation to the SCI data. First, we used a previously validated cut-off score of 16 or less to denote the proportion of our sample that met minimum criteria for putative insomnia disorder [20]. Second, we used the inherent operationalization of main clinical criteria in the SCI to identify proportion of the sample that met all four main criteria for insomnia disorder.

The nine-item Physician Health Questionnaire or PHQ-9 was used to measure the frequency of nine symptoms of depression in the past 2 weeks with a 4-point response including: 0 = “not at all,” 1 = “several days,” 2 = “more than half the days,” and 3 = “nearly every day.” A composite score was calculated based on all nine items and a cut-off score of 10 or higher was used to denote moderate to severe depressive symptoms [41, 59].

The two-item Generalized Anxiety Disorder (GAD-2) scale was used to measure severity of anxiety symptoms. Although not a diagnostic instrument, the GAD-2 captures the frequency of two symptom criteria for GAD in the DSM-5 including feeling anxious, nervous or on edge and not being able to control or stop worrying over the past 2 weeks. The frequency of these symptoms was captured using the same 4-point response options as describe above for the PHQ-9 [40]. The GAD-2 was scored and a cut-off of 3 or higher was used to denote potentially clinically significant anxiety symptoms [45, 61].

A health-based rating question was used to measure self-reported health status with categories including poor, fair, good, and excellent. Respondents were classified into Qataris or non-Qataris based on self-reported nationality. Ethnicity was determined according to respondent’s country of origin and the language selected to complete the interview. We first coded ethnicity into the following main categories: Arab, South Asian, South East Asian, and other (East Asian, Asian other, African, Latin American, European, from the UK, Russia, US, Canada or Australia). To increase statistical power, we collapsed non-Arab ethnicities into one category to compare with Arab ethnicity. We collected standard socio-demographic information including age, gender, marital status, education level, and employment status.

Weighted proportions, corresponding percentages, and 95% Confidence Intervals (CI) along with raw (unweighted) frequencies (N) were generated to describe the distribution of most of the variables in our sample. Mean and standard deviation (SD) was used to describe age of respondents in years.

To our knowledge, this is the first report on prevalence of insomnia disorder and associated clinical factors in a representative sample of community dwelling adults in the state of Qatar. Our study makes an important contribution as most of the published studies from the Middle East to date have rarely used a standard set of international diagnostic criteria to estimate the burden of insomnia in the general population. Overall, our prevalence estimates ranged between 3.0% using the DSM-5 diagnostic criteria or 5.5% using the SCI cut-off score to define insomnia. Our estimates were relatively lower than those reported in developed Western countries [54]. However, comparisons between the prevalence estimates from our study and these should only be made with caution. For example, a pooled prevalence estimate of 9.8% was reported in seven European countries (as defined by meeting all of the DSM-IV criteria for insomnia disorder- including the presence of a predominant complaint of insomnia, lasting for at least 1 month, that caused significant distress or daytime consequences) [54]. The lower prevalence estimate of 5.5% in our study may stem from differences in criteria especially that the DSM-5 criteria stipulate minimum of 3 months for chronicity of insomnia symptoms as opposed to minimum of 1 month in the DSM-IV. Additionally, our time frame for assessing prevalence of insomnia was 1 month only compared to a minimum of 1 month or more in the other studies [54]. However, our estimates ranged from 9.4 to 13.5%, and were similar to that reported in the general population of Turkey, where the prevalence of meeting any DSM-IV-TR criteria was estimated at 15.3% [18].

In our study increase in age was associated with decrease in insomnia symptoms in contrast to most worldwide reports that highlight a higher prevalence of insomnia among older populations [22, 53, 56, 63]. A meta-analysis of general population studies in China found that younger participants were more likely to suffer from insomnia than older participants. This finding was attributed to higher daily stress of modern urban life and high technology media use [13]. Similar factors may explain the higher prevalence of insomnia in younger respondents in Qatar who are mostly adult migrants and Qatari nationals between the ages of 25–54 years comprising 70% of the country’s population [33]. An additional explanation of why insomnia symptoms may decrease with older age is that in Arab Muslim societies, practicing filial piety and honoring parents is a religious obligation and a social norm [66], and most of the elderly live close to at least one son or a daughter. This family support may account for lower levels of daily stress and consequently insomnia amongst older adults in Arab-Muslim societies.

Overall, findings with respect to strong associations between insomnia and depression and between insomnia and anxiety in our study population are in line with findings from around the world [3, 10, 19, 30, 50, 52, 65, 67].

These findings also reinforce our suggestion that there is a need for Public Health consideration to address insomnia symptoms early, particularly due to stigma association with mental disorders. The association between insomnia and mental illness may have important implications in the overall help-seeking and health management of individuals with symptoms of insomnia. Furthermore, considering the potential clinical importance of insomnia, it is worthwhile to include insomnia-related questions in depression and anxiety screening questionnaires.

Arab ethnicity was positively associated with insomnia in our study. Arabs reported poorer sleep continuity, severity, quality and daytime impact of poor sleep than non-Arabs. Cultural factors are likely to explain these results. Late bedtimes, interrupted sleep, and daytime napping have been reported in studies investigating sleeping patterns among Middle Eastern populations, factors that may result in suboptimal sleep durations and quality of sleep [36, 49]. Extreme hot temperatures in the daytime in the Arabian Gulf often lead to an increase in nighttime socialization and late bedtimes. In addition, religious obligations such as daybreak prayer, Sohour in the fasting month, Ramadan, require interruption of sleep and changes in eating habits. Although there is scarce scientific information on how the effects of adhering to religious teachings affect sleep and health, it is generally well established that uninterrupted sleep, particularly in the initial stages of sleep, enhances sleep quality [11]. Our results indicate that these patterns of sleep, which are usually embedded in the Arab-Muslim culture, may put them at risk of insomnia. Our findings are also consistent with available literature on the role of culture in shaping beliefs, attitudes, and practices about sleep [29] as well as cross-cultural adjustment to sleep pattern changes [28].

International guidelines for insomnia management recommend the provision of patient education of healthy sleep habits known as sleep hygiene. Sleep hygiene consists of a combination of behavioral practices and environmental conditions, which improve sleep [5]. The results of this study highlight the need for public awareness campaigns focused on sleep hygiene and optimal lifestyle habits that would help improve sleep quality and reduce insomnia symptoms in this population. A meta-analysis study assessing the efficacy of sleep hygiene education for insomnia found a significant improvement in sleep onset latency, sleep fragmentation, total sleep duration and insomnia scores indicative of improved sleep [15].

Another important finding in our study was that unemployment was positively associated with insomnia. Similar findings were reported in several studies around the world, which showed that people with higher socioeconomic status tend to have less symptoms of insomnia and better quality of sleep [17, 42]. This finding is understandable, as unemployment may be associated with stress related to losing residency status in the country and fear of financial difficulties for non-Qataris. In addition, it has been reported that unemployment can lead to symptoms of depression [48]. As such, three key determinants of insomnia in present study, such as unemployment, depression and anxiety, appear to be interconnected.

The survey was cross-sectional and based on self-report. Although survey questionnaires were thoroughly translated and administered in the mother tongue of most respondents, we cannot rule out other sources of response bias in Arab and non-Arab migrants. However, we tried to mitigate the potential risk of response bias (e.g. social desirability) by training all interviewers in and systematically monitoring their adherence to standardized survey interviewing techniques. We collected no information about lifestyle factors that may be associated with insomnia and mental health such as cigarette smoking [32, 34] and diet [8, 27, 70] or shift or night work [51, 69], which is a limitation of the current analysis. The SCI is a relatively new tool for screening for insomnia according to the DSM-5 criteria and its psychometric properties in Arabic speaking populations have not previously determined. Differential diagnosis or causes by organic medical diseases as required by the DSM-5 criteria were not verified. Additionally, clinical evaluation, the gold standard for establishing insomnia diagnosis, was not carried out in this study. It is recommended that future studies establish the clinical validity of the SCI in the context of Arabic speaking populations.