In this study clinical insomnia was reported by one tenth of the adult population. This study also showed that in increasing age, higher education, umemployment, tobacco use and generalized anxiety disorder increased the odds for having clinical insomnia. Among these, older age is an established risk factor for clinical insomnia³². However, other four findings (tobacco use, generalized anxiety disorder, unemployment and higher education) are novel as they have not been investigated general population, to best of our knowledge. Even more interesting was the fact that in multivariate analysis, socio-demographic factors like socio-economic status, marital status, family type, crowding, major depressive disorder, and alcohol dependence were excluded from the model, many of which are established risk factor for insomnia.

In previous epidemiological studies, prevalence if insomnia has been reported to vary between 10.2-19.7%^2,9,23,33. This difference is related to difference in methodology of the study. For example, studies using sleep diary²³ reported higher prevalence compared to those using structured interview². Other factors e.g., criteria followed to diagnose insomnia, characteristics of population included viz., age, race, gender, use of addictive substances and comorbidities also influenced the results of previous studies³². In present study, ISI was used to diagnose insomnia, which contained items related to initial, middle and terminal insomnia as well as effect of insomnia on daytime functions^27,28. It assessed symptoms within a span of past seven days, which was contrary to the present definition of insomnia that focuses on frequency as well as duration of symptoms³².

A recent meta-analysis reported that ISI could be used as a reliable tool to diagnose insomnia with 88% sensitivity and 81% specificity, substantiating the results of present study³⁴. Another issue is- whether objective methods would have influenced the results of the study? In one of the earlier studies, objective assessment had shown higher prevalence of insomnia, compared to subjective report, which is contrary to the standard definition of insomnia, which focuses on subjective report^32,33. Hence, we opine that results of present study are reliable.

Contrary to the established fact that females are predisposed to insomnia, we did not find effect of gender in present study on prevalence of insomnia³². In earlier studies, female preponderance for insomnia has been ascribed to multiple factors e.g., genetic factors, predisposition towards anxiety, depression and coping styles to name a few²³. However, this data was from western population and a meta-analysis has also reported male predisposition for insomnia in East Asian population³⁵. Though the reasons are not entirely clear, it is possible that other factors interacted with gender to nullify the gender predisposition in present study. For example, three factors- increasing years of education, urban population and middle socio-economic group had higher prevalence of insomnia in this study.

It is possible that these factors exposed male subjects to emotional stresses, most of who were bread-earners for their family compared to female subjects who were largely home-makers. Gender difference in coping-skills is known with women more frequently using socialization while men focus on the problem and may get indulged in rumination³⁶. Rumination related to daytime issues has been reported in this population when they were not able to fall asleep, and can perpetuate insomnia by increasing stress, as described in stress-diathesis theory of insomnia^32,37. Other possibility could be related to use of addictive substances like tobacco and alcohol, which are culturally more acceptable among males. This could have increased the rates of insomnia among males leading to disappearance of gender effect. Lastly, in the Indian society, males are mainly responsible for providing financial support to the family, and high rates of unemployment, as discussed above could have led to higher prevalence of insomnia among males.

Higher education and unemployment were associated with insomnia in present study. However, previous studies have shown higher prevalence of insomnia among subjects with lower education and among unemployed subjects^38,39. Difference in the results of the present study could be related to high proportion of unemployment seen among subjects with graduation or higher education. This could have led to stress and consequent insomnia. Consistent with the results of present study, role of socio-economic variables is unclear with one study showing increased prevalence of insomnia among socially disadvantaged persons, while other showing inconsistent relationship^38,39. Interestingly, type of family and crowding were not associated with insomnia in this study.

Previous data have shown that single women parents have higher risk of having insomnia, whereas two-parents living in a nuclear family have lesser chances to have insomnia^40,41. Joint families may improve the sleep by sharing the household responsibilities among members, however, this effect may be culture specific. In traditional Indian setting, younger women in the joint family carry the burden of household work, that of child-care as well as care of elders. These responsibilities increase the stress and could be one reason why we did not find any difference between family types. Though theoretically, crowding in the bedroom may impair sleep quality, yet we did not find any effect of overcrowding in this study. In Indian settings, bed-room is usually shared among parents and young children. Most of people grow with this habit and hence, it may not impair their sleep. This is in concordance with results of previous study⁴². However, these findings need to be examined in future.

Insomnia has been found to be associated with clinically significant depression as well as anxiety in previous studies^19,23. However, it must be remembered that methodology was different across studies. For diagnosis of depression different methods have been used e.g., Beck’s depression inventory (BDI)²³ or patient health questionnaire-9 (PHQ-9)¹⁹. Among these, BDI contains items related to somatic as well as cognitive components, both of which are seen in subjects with insomnia. This can lead to spurious diagnosis of depression in these patients³². On the other hand, MINI focuses on cardinal (i.e., emotional) symptoms of depression. This could have led to exclusion of cases with prominent somatic and cognitive symptoms that are usually seen as daytime manifestations of insomnia. This could be one reason, why we did not find association between major depressive disorder and insomnia in present study in all models (Table 2).

This finding is reiterated by studies which reported improvement in depressive symptoms after adequate management of insomnia⁴³. Secondly, most of the studies have focused on depression and did not assess anxiety concurrently, thus comparative effect of anxiety and depression could not be assessed^2,19. van Mill et al.⁴⁴ assessed both major depressive disorders as well as generalized anxiety disorder and reported that both disorders increased odds for insomnia. Among all anxiety disorders, only subjects with generalized anxiety disorder have shown the polysomnographic evidence of disturbed sleep^45,46. Still, there is dearth of literature that has assessed effect of comorbid depression and anxiety disorders on insomnia, and this study was an attempt to fill that gap⁴⁶.

Alcohol dependence did not influence insomnia in this study while tobacco use increased the odds for insomnia. Though the insomnia was more prevalent in subjects with alcohol dependence in univariate analysis, as seen in a previous studies^47,48, this effect disappeared during multivariate analysis. In Indian clinical settings, it has been observed that most of the patients use nicotine in any form (chewing or smoking) after taking alcohol. It is possible that insomnia in these subjects is actually related to nicotine rather than alcohol, as seen in present as well as earlier study⁴⁸. In addition, psychiatric disorders have been found to influence occurrence of insomnia among subjects with alcohol dependence in various studies, similar to the findings of the present study^48,49.

It must be noted that previous studies assessing association between insomnia and alcohol had different methodology compared to present study. Their cohort was subjects having alcohol related disorders, on the contrary, present study was population based; hence, head to head comparison of these studies was not possible. Another interesting finding was lesser prevalence of insomnia among cannabis users. This could be related to pharmacological properties of cannabis and requires further investigation. Nabinol and cannabinol, both of which are cannabinoids are known to reduce sleep onset latency and relieve obstructive sleep apnea; contrarily, cannabis withdrawal is associated with sleep disruption^50,51. However, effect of cannabis disappeared when socio-economic variables were entered (model 3, Table 2) suggesting an interaction between cannabis and socio-demographic factors, which requires further investigation.

Like any other scientific investigation, this study also had some methodological limitations. First, diagnosis of insomnia as well as psychiatric disorders was made using structured interview rather than clinical interview. However, measures that were used have optimal psychometric properties making them suitable for epidemiological study, as mentioned in methodology. Second, because of the study design we could not assess differential effect of amount and frequency of addictive substances on prevalence of insomnia. This is an area worth investigating and will be assessed in future. Third, we did not rule out effect of other disorders that may interfere with sleep e.g., circadian rhythm sleep disorder, sleep apnea and restless legs syndrome. Fourth, other medical disorders e.g., congestive heart failure and chronic pain to name a few, are known to be associated with sleep disturbances. They were not addressed in the present study.

In conclusion, this study showed that insomnia affects around 10% of the population. It was associated with unemployment, higher education, generalized anxiety disorder and tobacco use in the community based sample.