Introduction

Sleeping problems across populations may be related to changes in lifestyle, increasing use of technology, increased work, social demands and tertiary stress (changes from secondary school to university such as reduced adult surpervision, new social opportunities and commitments) transition among university students [1–3]. Developing country populations are undergoing rapid demographic, epidemiologic and health transition [4]. Increasing urbanisation results in increased noise and stress providing a poorer physical and mental environment for sleep [5]. Poverty and mental health are negatively associated in developing countries [6]. Societal changes contributing to sleep loss may particularly impact on adolescents and young adults. Technology such as the internet and mobile phones may contribute globally to the problem of sleep loss and mental health [7]. Sleep problems might be an unrecognised public health issue in low- and middle-income and emerging economy countries, especially among young adults and university student populations.

The prevalence of sleep disorders, poor sleep quality or sleeping problems or insomnia among university students has mainly been studied in high-income countries such as in the USA (9.5 [8], 15 [9], 22.5 [10], 27 [11], 50.9 [12] and 60 % [13]), Italy (26.7 % [14]), Spain (60 % [15]), Australia (42 % [3]), Korea (36.2–60 % [16, 17]), Hongkong (68.6 % [18]), Taiwan (54.7 % [19]) and Japan (25.6 % [20]) and a few studies from low- and middle-income countries such as Brazil (28.1 % [21]), Ethiopia (55.8 % [22]), India (17.3 % [23]) and Palestine (9.5 % [24]). Some research seems to show an increase in the prevalence of sleep problems among university students over time, e.g. in the USA over two decades from 1978 to 2000, an increase from 24 to 71 % of self-reported dissatisfaction with their sleep was found [25].

Various risk factors for sleeping problems among university students have been identified as follows: \ sociodemographic factors: being a woman [9, 19, 26–28], older age [12, 14], and living away from parents [29]; (2) stress [3, 10, 13, 18, 22, 28] and childhood adversity [10]; (3) poor mental health, including anxiety, depression or minor psychiatric disturbance [3, 10, 14, 18, 23, 26–33]; (4) poor health status [14, 16, 20]; (5) health risk (behaviour): short sleep [34], smoking [20], drinking more alcohol [13], use of stimulants [35], lack of exercise [12, 36], problem gambling [37], tendency toward Internet addition [19], media use [38], skipping breakfast [19], obesity [36, 39], and motor vehicle accidents [2]; (6) lack of social support [19, 26]; and (7) poor academic performance [11, 32, 40].

The aim of this study was to investigate sleeping problems and its associated factors among university students in mainly low- and middle-income countries.

Methods

Sample and procedure

This cross-sectional study was carried out with a network of collaborators in participating countries (see Acknowledgements). The anonymous, self-administered questionnaire used for data collection was developed in English, then translated and back-translated into languages (Arabic, Bahasa, Chinese, French, Lao, Russian, Spanish, Thai and Turkish) of the participating countries. The study was initiated through personal academic contacts of the principal investigators. These collaborators arranged for data to be collected from 400 male and 400 female undergraduate university students aged 16–30 years by trained research assistants in 2013 in one or two universities in their respective countries. The universities involved were located in the capital cities or other major cities in the participating countries. Research assistants working in the participating universities asked classes of undergraduate students to complete the questionnaire at the end of a teaching class. Classes were recruited according to timetable scheduling in a quasi-random fashion. The students who completed the survey varied in the number of years for which they had attended the university. A variety of majors were involved, including education, humanities and arts, social sciences, business and law, science, engineering, manufacturing and construction, agriculture, health and welfare and services. Informed consent was obtained from participating students, and the study was conducted in 2013. Participation rates were in most countries over 90 %. Ethics approvals were obtained from all participating institutions.

Measures

Sleeping problems

The prevalence of nocturnal sleeping problems was estimated based on the question: ‘Overall in the last 30 days, how much of a problem did you have with sleeping, such as falling asleep, waking up frequently during the night, or waking up too early in the morning?’ Response options ranged from 1 (none) to 5 (extreme/cannot do). Sleeping problems were defined by the response to this question with ‘severe’ or ‘extreme/cannot do’ [41].

Sociodemographic questions

Sociodemographic questions included age, gender, population group, and residence, and socioeconomic background was assessed by rating their family background as wealthy (within the highest 25 % in ‘country’, in terms of wealth), quite well off (within the 50 to 75 % range for their country), not very well off (within the 25 to 50 % range from ‘country’) or quite poor (within the lowest 25 % in their country, in terms of wealth) [42].

Stress and health status

History of child physical and sexual abuse was assessed with two questions, and subjective health status with one question.

Centres for Epidemiologic Studies Depression Scale

We assessed depressive symptoms using the ten-item version of the Center for Epidemiologic Studies Depression Scale [43]. Scoring is classified from 0 to 9 as having a mild level of depressive symptoms, 10 to 14 as moderate depressive symptoms and 15 representing severe depressive symptoms [44]. The Cronbach alpha reliability coefficient of this ten-item scale was 0.74 in this study.

Post-traumatic stress disorder

A seven-item screener was used to identify post-traumatic stress disorder (PTSD) symptoms in the past month [45]. Items asked whether the respondent had experienced difficulties related to a traumatic experience (e.g. ‘Did you begin to feel more isolated and distant from other people?’, ‘Did you become jumpy or get easily startled by ordinary noises or movements?’). Consistent with epidemiological evidence, participants who answered affirmatively to at least four of the questions were considered to have a positive screen for PTSD [45]. The Cronbach alpha reliability coefficient of this seven-item scale was 0.75 in this study.

Health risk (behaviour)

Substance use and gambling

Tobacco use was assessed with the question: Do you currently use one or more of the following tobacco products (cigarettes, snuff, chewing tobacco, cigars etc.)? Response options were ‘yes’ or ‘no’ [46]. Heavy alcohol consumption was measured by asking participants ‘how often do you have (for men) five or more and (for women) four or more drinks on one occasion?’ The South Oaks Gambling Screen (SOGS), a standardised measure of pathological gambling and gambling behaviours in their lifetime [47] was used to assess nine different gambling behaviours, e.g. ‘Played cards for money’. Response options ranged from 1 = not at all to 3 = Once a week or more. Students who scored positive (in terms of more than once a week) on any of the nine gambling behaviours were classified as engaged in gambling. Cronbach alpha for this nine-item scale was 0.84 in this sample. Heavy Internet use was assessed with the question how many hours they normally spend on the internet per day. A cutoff of 6 h or more for heavy Internent use was chosen, in line with some previous studies, e.g. [48, 49] Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ) short version, self-administered last 7 days (IPAQ-S7S). We used the instructions given in the IPAQ manual for reliability and validity, which is detailed elsewhere [50]. To sum up the single indicators to an overall indicator of PA-related energy expenditure (EE; metabolic equivalent (MET) min−1) is a major goal of the IPAQ instruments. We used the recommended, following MET estimates of IPAQ: Vigorous PA = 8 METs, moderate PA = 4 METs and walking on average = 3.3 METs. For calculating the overall METs PA, each category was multiplied with its special MET estimate value. We also used the recommended categorical score, three levels of PA (low, moderate and high) as proposed in IPAQ Scoring Protocol (short form). Low activity represented individuals who do not meet the criteria for moderate and vigorous intensity categories (<599 MET-min/week). Moderate activity represented moderate—or vigorous—intensity activities achieving a minimum of at least 600 Met-min/week. High activity represented achieving a minimum of at least 3,000 Met-min/week [50].

Skipping breakfast was assessed with the question, ‘How often do you eat breakfast?’ Response options ranged from 1 = almost every day to 3 = rarely or never [42].

Anthropometric measurements

Height (without footwear) using a stadiometer and weight (without footwear and any heavy accessories) using a calibrated weighing scale were measured.

Body mass index (BMI) was calculated as weight in kilogrammes divided by height in metre squared. There was a low response rate of anthropometric measurements for Grenada and Cameroon and for the China Hongkong sub-sample and Indonesia body weight and height was collected by self-report. BMI was used as an indicator of obesity (≥27.5 kg/m2) in the East and South Asian participants [51], and for the other countries, obesity was defined as BMI = 30 kg/m2 [52].

Injury

Participants were asked, ‘During the past 12 months, how many times were you seriously injured?’ (serious injury was defined as ‘when it makes you miss at least one full day of usual activities (such as university, sports or a job) or requires treatment by a doctor or nurse’). Eight options were provided, ranging from 1 = 0 times to 8 = 12 or more times. A response of ‘0’ was described as not having sustained a serious injury, while a response of one or more times was classified as having experienced a serious injury [53].

Social support

Three items were drawn from the Social Support Questionnaire to assess perceived social support [54]. The items were selected to reflect perceived tangible and emotional support: If I were sick and needed someone to take me to a doctor, I would have trouble finding someone; I feel that there is no one I can share my most private concerns and fears; and I feel a strong emotional bond with at least one other person. These items were responded to on 4-point scales, 1 = completely true, to 4 = completely false, and summed to a score with a range of 3–12. Cronbach’s alpha for this sample was 0.69.

Academic performance

Academic performance was assessed with the question, ‘how would you rate your academic performance?’ Response options ranged from 1 = excellent to 5 = not satisfactory.

Data analysis

The data were analysed using IBM SPSS (version 20.0). Stratified analysis was conducted for male and female students. The prevalence of sociodemographic factors, stress and health status, health risk (behaviour), social support, academic performance and nocturnal sleeping problem was calculated as a percentage. Associations between key outcomes of nocturnal sleeping problem and sociodemographic, stress and health status, health risk (behaviour), social support and academic performance variables were evaluated calculating odds ratio (OR). Unconditional multivariable logistic regression was used with STATA for evaluation of the impact of explanatory variables for key outcome of nocturnal sleeping problem (binary-dependent variable). All variables statistically significant at the p < 0.05 level in bivariate analyses were included in the multivariable models. Missing cases were excluded from the analysis. The country was entered as the primary sampling unit for survey analysis in STATA in order to achieve accurate CIs, given the clustered nature of the data [42].

Results

Sample characteristics

The total sample included 20,222 undergraduate university students (mean age, 20.8; SD = 2. 8) from 26 countries. By population group, 40.1 % of the sample was Asian, 26.9 % African, 16.7 % mixed or coloured, 6.6 % White or Caucasian and 9.5 % other. Overall, 10.4 % reported severe or extreme nocturnal sleeping problems (male, 10.2 %; female, 10.5 %). Noctural sleeping problems differed by country, from 32.9 % in Indonesia to 3.0 % in Thailand among Asian countries, from 13.7 % in Mauritius to 7.5 % in South Africa and from 11.8 % in Jamaica to 6.1 % in Columbia in the Americas (see Table 1).

Table 1 General demographics
Full size table

More than half of the participants (53.6 %) come from a wealthy family background. Regarding stress and health status, 8.5 % reported poor subjective health status, 5 % child physical abuse, 2.6 % child sexual abuse, 36.8 % screened positive for moderate or severe depression symptoms and 20.9 % screened positive for PTSD symptoms. In terms of health risk (behaviour), 12.8 % of the students were currently using tobacco, 11.8 % binge drinking, 20.3 heavy Internet use, 8.2 % had been gambling more than once a week and 47.5 % engaged in inadequate physical activity. Almost half (46.2 %) skipped breakfast, 5.5 % were obese and 23.9 % had sustained at least one injury in the past year. From the university students, 6.2 % rated their academic performance as dissatisfactory or poor (see Table 2).

Table 2 Sample characteristics by study variables
Full size table

Predictors of nocturnal sleeping problem

In multivariate logistic regression analysis, coming from a poor family background, staying off campus (on their own or with their parents or guradians), stress (history of child sexual abuse), poor mental health (depression and PTSD symptoms), health risk behaviour (tobacco use, heavy Internet use, gambling, skipping breakfast and having sustained an injury), lack of social support and poor academic performance were associated with nocturnal sleeping problems (see Table 3).

Table 3 Bivariate and multivariate logistic regression with nocturnal sleeping problem
Full size table

In multivariate logistic regression analysis by male and female gender separately, child sexual abuse and heavy internet use was only for female gender associated with nocturnal sleeping problem, while coming from a poor family background, frequent gambling, lack of social support and poor academic performance were only for male gender associated with nocturnal sleeping problem (see Table 4).

Table 4 Multivariate logistic regression with nocturnal sleeping problem by gender
Full size table

Discussion

The overall prevalence of severe or extreme noctural sleeping problems in this sample of 20,222 undergraduate university students from 26 countries was 10.4 %. Using different types of classifications for sleeping problems, previous studies seemed to have found generally higher prevalences of sleeping problems among university students [3, 9–23], while findings from two studies found similar rates [8, 24]. However, the study found large country variations in terms of prevalence of sleeping problems, ranging from 32.9 % in Indonesia to 3.0 % in Thailand. Previous studies seem to have found higher rates of sleeping problems among university students, e.g. in Thailand (48.1 %) [35] compared with 3.0 % in this study, and in Hongkong (68.6 %) [18] compared with 7.1 % in China (mainly in students from Hongkong) in this study. There was a within China country variation in terms of 4.6 and 8.1 % of students having sleep problems in Chengdu and Hongkong, respectively, in this study. This difference may be explained by the higher industrialised and stressful Hongkong compared with Chengdu in mainland China. Within the country differences found, one could group the prevalence of self-reported sleeping problems in this study into three groups, namely low (<10 %: Thailand, Singapore, Russia, Colombia, China, Egypt, Laos, South Africa, India, Venezuela, Nigeria and Pakistan), medium (10–15 %: Cameroon, Philippines, Madagascar, Barbados, Kyrgyzstan, Ivory Coast, Bangladesh, Grenada, Tunesia, Jamaica and Turkey) and high (>15 %: Mauritius, Namibia and Indonesia). Some of these differences in the prevalence of sleeping problems are difficult to explain, since there is little evidence on sleeping problems in these low and middle income and transitional countries [41]. One possible explanation for these differences may be a cultural pattern to the ways in which students from different countries report their sleeping problems, with students may be downplaying in some countries and in other countries students may more likely complain [55]. One other explanation could be that in countries with a high rate of sleeping problems among students such as in Mauritius, Namibia and Indonesia in this study, this could be a reflection of increased social stress in these countries, which needs further investigation.

In terms of sociodemographic factors, this study only found that poorer economic background and not gender, age and living status was associated with sleeping problems. Low socioeconomic status has been found of importance among the sociodemographic determinants of insomnia [56]. Contrary to a previous study [36], this study found that residing off campus (on their own or with parents or guardians) was associated with sleeping problems. This finding may be related to increased stress of commuting from the residence of the students to the campus for studying. In accordance with other studies [3, 10, 14, 18, 22, 29–34], this study found that a history of child sexual abuse and poor mental health (depression and PTSD symptoms) were associated with sleeping problems. Insomnia is now considered not only a symptom of but also a possible predictor of depression [5, 57]. The consistency of the findings of this study with those of the published research stresses the public health importance and implications of more thoroughly investigating links between sleep problems with mental health [22], indicating intervention programmes among young adults.

Regarding health risk behaviour (tobacco use, heavy internet use, gambling, skipping breakfast and having sustained an injury) were found in this study to be associated with sleeping problems. Other studies [2, 19, 20, 37, 38] also found that smoking, problem gambling, tendency toward internet addition, skipping breakfast and accidents (injury) were associated with sleeping problems. It is possible that heavy Internet users stay often online late into the night and may get up too late and skip breakfast before classes or work, which may lead to excessive fatigue and insomnia [19, 58]. Regarding injury, it is possible that as a result of functional impairment in sleep disorder, university students are placed at higher risk for unintentional injury [59]. Pickett et al. [60] found that gradients in risk for youth injury increased in association with numbers of risk behaviours reported. This clustering of unhealthy problem behaviours may suggest that university students who are likely to have sleeping problems are also likely to be involved in other problem behaviours. Contrary to some previous studies [12, 13, 36, 39], this study did not find any association between heavy drinking, lack of exercise, obesity and sleeping problem. It is possible that in our study, the level of heavy drinking was low so that there was no relationship between drinking and sleep problems. Regarding sleep problems and obesity, a recent review [36] concludes that predominantly cross-sectional studies make it difficult to ascertain clearly whether it is sleep disorder contributing to obesity or obesity contributing to the sleep disorder calling for more research. They hypothesised a combination of both, each contributing to a downward spiral of worsening sleep habits and body adiposity [36]. Furthermore, in this study, physical activity was assessed by self-report, and it is possible that students overestimated their physical activity, and thus the positive influence of exercise in sleep problems was not identified.

This study also found that lack of social support was related to sleeping problems. The association between social support and sleep quality has also been found in previous studies [19, 24] and therefore the finding in this study reasonable. Furthermore, poor academic performance was found to be associated with sleeping problems in this university student population, as also found in some previous studies [11, 33, 40]. The finding that sleep quality is an important component of academic success is important, as a significant proportion of university students do not have good quality sleep [40]. ‘If poor sleep appears to be problematic or contributory to presenting concerns and/or academic functioning, clients should be provided with patient education about the importance of sleep, be given information on sleep hygiene, and be encouraged and helped to improve their sleep habits’ [40].

Study limitations

This study had several limitations. The study was cross-sectional, so causal conclusions cannot be drawn. The investigation was carried out with students from one or two universities in each country, and inclusion of other centres could have resulted in different results. University students are not representative of young adults in general, and the prevalence of sleeping problems and its risk factors may be different in other sectors of the population [42].

The measures in this study were simple self-report items, e.g. on sleep problems and child abuse, and more refined assessments with objective verification would have been desirable [22, 42].

To improve on the specificity of the sleeping problem measure, only those students reporting severe or extreme sleeping problems were included, excluding those with mild or moderate sleeping problems [41]. Another limitation was the retrospective reporting and the possible negative bias that may have pervaded measures across domains.

Conclusions

There was a significant prevalence of past-month nocturnal sleeping problems in the sample of undergraduate university students from 26 low- and middle-income and emerging economy countries. Potential factors associated with the risk for reporting sleeping complaints were related to poorer economic status, stress, poor mental health, health risk behaviour, lack of social support and poor academic performance. These findings may assist in prevention strategies to promote a better quality of sleep and subsequent quality of life for this population.