Survey Mode and Administration
The current study is based on a survey conducted by the Social and Economic Survey Research Institute at Qatar University in April of 2018 using computer-assisted telephone interviewing. The interviewers received training and were monitored to ensure questions were asked appropriately and answers were recorded accurately.
The current study was approved by the Qatar University institutional review board (IRB) ethics approval reference number is QU-IRB 882-E/18 and participants provided informed consent.
Study Sample and Characteristics
Qatar has a diverse population with three distinct social classes or population groups and shares this social structure with most other countries in the Arabian Gulf. The first group is Qatari nationals (QNs) or natives; most are of tribal origins with shared ancestry and traditions. The second group is higher income white-collar expatriates (WCEs) who are highly skilled, educated, and come from all over the world. The third group is blue-collar workers (BCEs) who are mostly male laborers from South Asia and South East Asia with little or no formal education. The target population was sampled from three diverse social groups (QNs, WCEs, and BCEs) within Qatar’s resident population.
Given the majority of adults in Qatar own at least one cellphone, our sample was selected from a frame of cellphone numbers using list-based dialing techniques [22]. A total of 10,579 phone survey interviews were attempted of which 5,872 were eligible and a total of 2,523 surveys were completed giving a response rate of 43%. After removing cases where sleep duration was missing (n = 19) and excluding out of bound cases greater than 18 hours or less than 1 hour of sleep (n = 5), a final sample of 2,499 was retained. The final study sample consisted of 832 QNs, 935 WCEs, and 732 BCEs.
Measures
Sleep Duration
The average sleep duration was based on the question "On an average night, how many hours do you sleep?” To maintain power and precision, responses were collapsed into three categories: less than 7 hours, 7–8 hours and more than 8 hours based on previous literature of the U-shaped association and health related outcomes [23, 24].
Diagnosed Chronic Illness
Chronic disease status was defined as having any of the following conditions in response to the following survey question and response options: "Have you been diagnosed or told by your doctor that you have any of the following conditions" Which included: “Hypertension or high blood pressure?”, “Cardiovascular or heart disease?”, “Diabetes?”, “Asthma?”, “Gastrointestinal disorder?”, “Depression?”, “Other mental or Psychological Problems such as anxiety, or sleep problems?”, “Cancer”, “Disability (physical, mental, visual, hearing, etc.)“, “Thyroid disorder?”, “Any other condition not mentioned?”.
Depression
Depressive symptoms were assessed using the ultra-brief patient health questionnaire (PHQ–2) [25]. The performance of the PHQ–2 was reported to be acceptable compared to longer depression screening instruments. A score that ranges from 0 to 6 was computed, and a cutoff point ≥3 was used as proxy-measure of clinically significant depression [26].
Body Mass Index & Life style Variables
Body Mass Index (BMI) was calculated from self-reported weight and height measurements. Accordingly, participants were classified as underweight if BMI ≤ 18.4, normal weight if BMI≥18.5 and BMI ≥ 24.9, and overweight or obese if BMI ≥ 25.Sedentary behavior was measured using the question “Physical activity or exercise includes activities such as walking briskly, jogging, team sports, or any other activity in which you breathe harder or feel warmer, do you currently engage in physical activity on a regular basis?” Smoking status was also assessed as current versus non-current smoker (never and former smoker status), perceived health status was measured on a scale from 0–100 and categorized into three quantiles of poor, good and very good.
Statistical Analysis
Descriptive analysis explored characteristics of the sample. Weighted proportions with corresponding 95% confidence intervals (CI) were used to account for the complex survey design and nonresponse. To correct for survey design effects on the variances of reported proportions, the F-transformed version of the Pearson Chi-square statistic was used.
Multinomial logistic regression models with sampling weights that account for complex survey design were used to examine the association between socioeconomics, lifestyle variables, and health characteristics that could potentially determine sleep duration (dependent variable). All models were adjusted for age, gender and social group/class. Statistical significance was determined using an alpha value of 0.05. All analyses were conducted in STATA 14.