Introduction

The contribution of cardiovascular diseases to mortality increased by 34% during the last three decades in India [1]. In this same period, a concomitant rise in underlying risk factors, namely obesity, high blood pressure, fasting blood glucose, and dyslipidemia was noted. These metabolic risk factors along with psychosocial problems tend to become confluent among older adults. In India, almost one in ten (over 139 million) people belong to the elderly age group (60 years or above) in whom high prevalence of chronic diseases, depression, and anxiety constitutes a major public health challenge [2,3,4].

Sleep is a critical physiological process that maintains the body’s circadian rhythm, by not only providing physical rest, but also ensuring internal restorative functions [5]. Inadequate and poor-quality sleep can interrupt the health-promoting functions that occur during sleep, leading to poorer health outcomes [6]. Population-based data on prevalence of subjective poor sleep quality in India are limited, since most studies have beeen conducted in a clinical setting. The reported prevalence of poor sleep quality based on community-based studies was 27% from south India, 43% from urban east India, and 39% from rural east India [7, 8]. Subjective poor sleep quality has been associated with an increased risk of metabolic syndrome including obesity and diabetes [9, 10], higher rates of hypertension [11, 12], and dyslipidemia [10, 13].

The abnormal collapse of the pharyngeal airway during sleep leading to apneic episodes and the resulting repetitive arousals is termed obstructive sleep apnea (OSA) [14]. OSA symptoms, characterized by snoring and sleepiness, are strongly associated with obesity [10], and also with metabolic syndrome, including insulin resistance, and dyslipidemia through inflammatory pathways, as well as with hypertension [15]. OSA symptoms are associated with worsening hypertension, arrhythmias, stroke, and myocardial infarction [12].

In high-income countries, the metabolic and psychosocial problems co-occur with both poor sleep quality [9,10,11,12,13] and OSA symptoms [10, 15,16,17]. While the association of metabolic risk factors in obesity or cardiovascular diseases is well documented in India, the extent of their association with sleep problems or OSA symptoms remains less clear. Importantly, studies aiming to assess the independent contribution of metabolic risk factors must account for psychosocial problems, in particular symptoms of depression and anxiety, since they exist as common co-morbidities of metabolic dysregulation and sleep problems [8, 18]. However, there is no population-based study among this age group from India that assessed the quality of sleep and OSA symptoms, while considering psychosocial problems and metabolic risk factors together.

The population-based LoCARPoN study of persons aged 50 years and above in India provided an opportunity to describe the pattern of sleep quality and risk of OSA symptoms and their association with metabolic risk factors. Our research question was to determine if metabolic risk factors, independently of psychosocial problems, are associated with poor sleep quality and OSA symptoms. To fulfill this aim, we included common metabolic and frequently occurring psychosocial problems. Since the prevalence of metabolic risk factors is high in all age groups in India, we hypothesized that metabolic risk factors can explain additional variability of poor sleep quality and OSA symptoms independent of psychosocial problems and demographic and lifestyle factors in older Indian adults. As a secondary objective, we also assessed and presented the association of psychosocial factors with poor sleep quality and OSA symptoms, independently of metabolic factors.

Methods

Study population

This was a cross-sectional study using the baseline data of the Longitudinal Cognition and Aging Research on Population of the National Capital Region-LoCARPoN (Hindi—“dedication-to-people”) study [19]. The LoCARPoN is a population-based cohort of middle-aged and older adults in New Delhi, India, aimed at studying India-specific risk factors with the onset of stroke, dementia, and mental health.

Between year 2016 and 2019, 8858 participants aged 50 years and above consented and were enrolled from two areas in South Delhi by household visits. A total of 7505 individuals (85%) attended the study center for in-person assessments including cognitive testing and cardiovascular examinations, and provided blood specimens for estimates of blood glucose and lipid profile. We analyzed the data of these 7505 individuals for the present study. The LoCARPoN study was approved by the Institutional Ethics Committee of All India Institute of Medical Sciences, New Delhi, India (reference number: IEC/NP-53/2014 RP-12/2014). Written informed consent was obtained from individual participants.

Exposure assessment

Pre-tested structured questionnaires were used to collect information on (a) socio-demographics age, sex, education (assessed in continuous scale as “years of schooling”), and marital status (marital status was dichotomized to married or divorced/widow/single); (b) lifestyle factors smoking, alcohol, and physical activity. For the purpose of the current analysis, both smoking and alcohol were taken as ever smoker and ever alcohol. In the baseline assessment study, these variables were assessed by face-to-face interview questions; physical activity was assessed using the international physical activity questionnaire (IPAQ), and categorized as mild, moderate, and high levels of physical activity. (c) psycho-social factors (symptoms of depression, anxiety and perception of social support, and stress) and cognition; symptoms of depression, based on the scores of the 30-item geriatric depression scale, were categorized into normal (< 10), mild (10 to < 20), and moderate to severe (20 to 30) [20]. Symptoms of anxiety, based on the 11-item ACS anxiety scale, were categorized into normal (< 2), mild (2 to 5), and moderate to severe (6 to 11) [21]. The 7-item psychosocial stress scale from the INTERHEART study was used to assess levels of stress and categorized as mild normal (< 2) and moderate-to-severe (≥ 2) stress [22]. Social support was categorized into inadequate (< 5) and adequate (≥ 5). These categorizations were done to aid in the comparison with other studies and also to make clinical interpretations. (d) metabolic factors included fasting glucose, total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), glycated hemoglobin, and the cardiovascular risk factors blood pressure (BP) and body mass index (BMI). The cut-offs for BMI (in kg/m2) were set as underweight (< 18.5), normal (18.5–25), pre-obese (25–30), and obese (> 30), according to the World Health Organization [23]. Overnight fasting blood specimens were used to estimate fasting blood glucose, HDL, LDL, and total cholesterol using an Erba XL-640 Biochemistry Analyser. Glycated hemoglobin (HbA1C) was measured on the same blood specimen using the high-performance liquid chromatography (HPLC) method in a Biorad D-10-HbA1c Analyser. Dyslipidemia for each measure of lipid profile was defined as total cholesterol ≥ 200 mg/dL, LDL ≥ 130 mg/dL, and HDL ≤ 60 mg/dL. These cut-offs were based on the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)-NCEP ATP III guidelines, which also recommended that ideal HDL levels should be between 40 and 60 mg/dL [24]. Diabetes mellitus was defined as HbA1c ≥ 6.5% (as per the American Diabetic Association-ADA International Expert Committee report) [25], or currently on medication for diabetes mellitus. High blood pressure was defined as either SBP ≥ 140 mmHg or DBP ≥ 90 mmHg (as per the Seventh Report of the Joint National Committee-JNC on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure) [26] and/or currently on medication for hypertension. Using an electronic blood pressure instrument (OMRON HEM-8712), BP was measured with three readings on the right arm, 5 min apart. The mean of the three readings were included for SBP and DBP. For anthropometry, the mean of three readings was taken for body weight (in kilograms, digital weighing machine) and height (in centimeters, stadiometer with metallic metric tape mounted on a wall). In addition, medication for diabetes mellitus and high blood pressure was self-reported. The Mini Mental Status Examination (MMSE)-Hindi version, was used as a cognitive screening tool.

Outcome assessment

Sleep quality characteristics were assessed using the Pittsburgh Sleep Quality Index (PSQI). The PSQI is a self-rated questionnaire consisting of 19 items grouped into seven components: subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of sleeping medication, and any daytime dysfunction. Each component score ranges from 0 to 3 (higher is worse), and the components are summed to get the total score. Participants with a global PSQI score greater than 5 were categorized as having poor sleep quality. Difficulty in initiating sleep was defined as a sleep latency of more than 30 min. Disturbed sleep was also derived from the PSQI and defined as trouble sleeping experienced any time during the past month, with score range 0–3; the summed score was categorized (≥ 1 = disturbed sleep, 0 = no disturbed sleep).

Sleep components: sleep medication was defined as the reported use of any prescribed or “over-the-counter” medication to aid sleep at least once a week (yes/no). Sleep latency was defined as the duration it has taken to fall asleep each night, in the past one month (in minutes). Sleep duration (in hours) was self-reported, as hours of actual sleep that the person got at night. Total time in bed (hours) was calculated as the difference between bedtime and wake time in hours. Habitual sleep efficiency (HSE) (in %) was calculated as (sleep duration/total time in bed * 100). Since sleep duration and total time in bed were assessed separately, it may result in implausible high values, but probably represent high HSE. To balance this bias in estimates with loss of precision: values between 100 and 110% were recoded to 100%. Daytime dysfunction: self-reported, yes/no, if answered yes to any of the following: trouble staying awake while driving, eating meals, or engaging in social activity at least once a week, or, problem in keeping up enough enthusiasm to get things done.

Symptoms of obstructive sleep apnea (OSA) were measured with the Berlin questionnaire (BQ). The questionnaire consists of eight questions; five items assess symptoms of snoring (category 1), three items assess symptoms of sleepiness (category 2), and a third category consists of either hypertension or BMI ≥ 30 kg/m2 (category 3). Categories 1 and 2 are considered positive if the score is ≥ 2 points. An individual with a score of 2 and above on the Berlin questionnaire is considered as at high risk of OSA symptoms, while those having a score below 2 are designated as at low risk. Since our objective was to assess the association of metabolic risk factors with the risk of OSA symptoms, we decided not to include the third category in the definition of OSA symptoms. Having the third category, defined by high blood pressure or BMI ≥ 30 kg/m2, would have resulted in model misspecification if BMI and hypertension were also included as covariates in the model. A similar approach has been used by Peker Y. et al. who developed a modified high-risk (mHR) OSA score by ignoring obesity and hypertension [27].

Hence, we used only category 1 and category 2 to classify individuals at high risk of OSA (total score ≥ 2) so as to allow us to include BMI and hypertension in the model. We could not get any reference where BQ was deployed as mentioned above; hence, we cannot comment on the change in validity of the questionnaire. However, since essentially the outcome variable of OSA symptom reflected the cardinal two symptoms of snoring and sleepiness, which are actually the only two components of BQ specifically for OSA symptoms, we believe that using the BQ excluding the third category will not affect the validity of the instrument to identify those with symptoms of snoring and sleepiness. Use of the BQ was not intended to make a diagnosis of OSA but to enable identification of individuals who are at high risk of OSA symptoms of snoring and sleepiness.

Statistical analysis

We presented a descriptive analysis of participants’ characteristics as frequencies and percentages for categorical variables, or as mean (S.D.) for continuous variables. The unadjusted associations between categorical variables were assessed with Pearson’s chi-squared test, and those between continuous variables and categorical variables with t test or one-way analysis of variance (ANOVA). In model 1, we estimated the association of each covariate with the outcome variables, i.e., separately for sleep quality and OSAS, by binary logistic regression, adjusted for age and sex. We report adjusted odds ratio (AOR) with 95% CI.

Thereafter, we made 3 groups of variables, namely, group 1: demographic (age, sex, marital status, education) and lifestyle factors (smoking, alcohol use, physical activity); group 2: psychosocial problems (symptoms of depression, anxiety, perceived stress, and social support); and group-3: metabolic risk factors (diabetes, BMI, HDL, LDL, total cholesterol, hypertension). To assess the independent association of psychosocial problems and metabolic risk factors with each outcome, we fitted three successive multivariable models as follows: model 2a included two groups of variables, i.e., psychosocial problem and demographic and lifestyle factors; model 2b also had two groups of variables, i.e., metabolic risk factors and demographic and lifestyle factors; and model 3 was the full model which had variables belonging to all 3 groups. Thus, model 2a and 2b were essentially nested models of full model 3. The process was repeated for each outcome.

Independent contribution by nested models was quantified using McFadden’s pseudo R2 for logistic regression. Thus, we compared pseudo R2 between model 2a and model 3 to assess the independent contribution of metabolic risk factors, and between model 2b and model 3 to assess the independent contribution of psychosocial problems to each outcome. Any observed change in pseudo R2 was formally tested with a log-likelihood ratio test (LRT), and was considered significant at p < 0.05.

Sensitivity analysis after excluding individuals with a past history of stroke

A self-reported history of stroke was collected in this study, and a total of 109 (1.5%) of 7505 individuals reported having stroke in the past (males 77, females 32). We conducted a sensitivity analysis to explore if the associations would change, after excluding these 109 individuals. Therefore, after excluding these 109 individuals, we reanalyzed the data and implemented model 1, model 2a, model 2b, and model 3 for both the outcomes that are poor sleep quality and symptoms of obstructive sleep apnea.

For covariates with missing data, multiple imputation was performed with the “mice” package, using option “predictive-mean-matching.” Glycated hemoglobin had the most missing values (22.5%); other missing covariate values did not exceed 2%. Associations with p < 0.05 were considered statistically significant in multivariable analysis. All analyses were performed in R.

Power calculation showed that for a 5% level of significance, with a sample size of 7505, we would have a power of 90% to detect an estimated odds ratio of 1.11 for multivariable logistic regression, where R2 for covariates has been kept at a high of 0.35 considering their syndromic presence. Supplementary file figure-1 provides a graphical representation of the sample sizes as a function of different odds ratios for a given power of 90% for logistic regression. Power calculation was done using G-power software (G*Power Version 3.1.9.6 http://www.gpower.hhu.de/).

Since the incidence of the outcome of interest was common in our study population (> 10%), the adjusted odds ratio derived from the logistic regression may no longer approximate the risk ratio, since, with more frequent outcome, the odds ratio tends to overestimate the risk ratio, if OR > 1, or underestimate it if OR < 1. Hence, we used the formula RR = OR / (1 − Po) + (Po*OR), where Po is the prevalence in the non-exposed group, suggested by Zhang J. et al. [28] to convert OR derived from logistic regression to RR. We presented the adjusted RR for the fully adjusted model 3 in Table 4 and Table 5.

Results

We analyzed the data of 7505 individuals with a mean age of 64.5 (SD: 9.2) years, ranging 50–99 years; females comprised half of the cohort (50.1%). The demographic, lifestyle, metabolic, and psycho-social characteristics are presented in Table 1.

Table 1 Characteristics of study participants at baseline of the cohort
Full size table

Prevalence of poor sleep quality and OSA symptoms

The prevalence of poor sleep quality (global PSQI score > 5) was 16.9% (95% CI: 16.0, 17.7), significantly lower among males, 12.3% (95% CI: 11.2, 13.4), than females, 21.0% (95% CI: 19.7, 22.3), and increased with age from 14.0% (50–60 years) to 21.2% (≥ 80 years) (Supplementary Table S1). Table 2 describes the prevalence of each sleep characteristic by age and sex. A long sleep latency, i.e., difficulty initiating sleep (> 30 min), was present in 7.5% of participants; this was more common in females (9.9%) than in males (5.0%). The average sleep duration was 6.5 ± 1.1 h shorter in females than males, but did change with age. Around 4% of the participants slept < 5 h; almost half slept 7 to 8 h. On an average, the habitual sleep efficiency (HSE) was 82%. Overall, subjective sleep quality was 0.69 (on a scale of 0—better to 3—poor). Some form of sleep disturbance was found in 38% of the participants, most commonly due to “getting up for bathroom” (25%), and “waking up in middle of night” (25%). Sleep disturbances were more common among females (43.1%) than males (32.5%). Overall, 11.4% participants reported experiencing daytime dysfunction, which increased with age from 9.1% (50–60 years) to 15.9% (≥ 80 years) by age.

Table 2 Components of sleep quality measured by the Pittsburgh Sleep Quality Index, by sex and age group
Full size table

Symptoms of snoring were reported by 46% (95% CI: 44.6, 46.9) of participants and daytime sleepiness by 13.7% (95% CI: 12.9, 14.5) (Table 3). Almost one in ten participants reported nodding off at least 3–4 times a week while driving. Of 7505 participants, 525 (overlap category 1 and 2) reported symptoms of both snoring and daytime sleepiness resulting in prevalence of likely OSA as 7.0% (95% CI: 6.4, 7.6), higher among females (8%) than in males (6%); the prevalence of likely OSA decreased with age from 8.3% (50–60 years) to 7.6% (≥ 80 years).

Table 3 Prevalence of symptoms of obstructive sleep apnea, using the Berlin questionnaire
Full size table

Correlates of poor sleep quality

Age- and sex-adjusted OR of mild and moderate-severe symptoms of depression, anxiety, stress, perception of inadequate social support, hypertension, diabetes, and dyslipidemia were associated with poor sleep quality; higher education, cognition score, and moderate physical activity were associated with lower odds of poor sleep quality (Table 4). In the full model (model 3), demographic characteristics (age, female sex, being divorced/widow(er)/single), psycho-social problems (symptoms of depression and anxiety), and metabolic risk factors (hypertension and low HDL) were independently associated with poor sleep quality, while higher education and moderate physical activity were inversely associated with poor sleep quality. Diabetes mellitus was not significant in the full model. Table S2 shows factors associated with sleep latency and sleep disturbances.

Table 4 Factors associated with poor sleep quality measured using the Pittsburgh Sleep Quality Index (PSQI)
Full size table

Independent contribution of psychosocial problems and metabolic risk factors to poor sleep quality

Nested models comprising the group of variables pertaining to psychosocial problems and another for metabolic risk factors, each adjusted for demographic and lifestyle factors, were compared with the full model (Table 4). Between models 2a and 3, the pseudo R2 increased very modestly but significantly from 0.14 to 0.15 (LRT p < 0.01, df = 8, chi2 = 25.1) implying a small independent contribution of metabolic risk factors to poor sleep quality. Between models 2b and 3, the pseudo R2 increased from 0.10 to 0.15 (LRT p < 0.001, df = 7, chi2 = 303.4). This implied that psychosocial factors are independently associated with poor sleep quality (Table 4).

Correlates of OSA symptoms

Age- and sex-adjusted OR of mild and moderate-severe symptoms of depression, anxiety, stress, inadequate social support, overweight, and diabetes mellitus were associated with OSA symptoms; higher education and moderate physical activity were associated with lesser odds of OSA symptoms (Table 5). In the full model (model 3), psychosocial problems (symptoms of anxiety and moderate-severe depression), and metabolic risk factor (BMI ≥ 25 kg/m2), were independently associated with more OSA symptoms, while moderate physical activity was protective. Demographic characteristics were not significant in the full model. Supplementary Table S3 shows factors associated with snoring and sleepiness components.

Table 5 Factors associated with symptoms of obstructive sleep apnea assessed using the Berlin questionnaire
Full size table

Independent contribution of psychosocial problems and metabolic risk factors to OSA symptoms

We found an independent contribution of metabolic risk factors on the odds of having OSA symptoms. This is reflected in the increase of the pseudo R2 from 0.08 to 0.10 between model 2a and model 3 (LRT p < 0.001, df = 8, chi2 = 48.2). The psychosocial factors also had an independent effect on OSA symptoms as evident with pseudo R2 increased from 0.08 to 0.10 between models 2b and 3 (LRT p < 0.001, df = 7, chi2 = 70) (Table 5).

In sensitivity analysis, we found that excluding the 109 individuals who reported a history of stroke in the past did not lead to any change in the pattern of associations reported already (Supplementary Table S4 and Supplementary Table S5).

Discussion

We present data from an Indian cohort of older adults to study the frequency and risk factors for subjective poor sleep quality and likelihood of OSA symptoms. The most salient findings were the independent association of metabolic factors with sleep quality and the independent association of psychosocial problems with OSA symptoms.

Prevalence of poor sleep quality

Almost one in six individuals (16%) had poor sleep quality in our study. Comparatively higher prevalence was reported by two community-based studies, with small sample sizes, namely 28% by George et al. [8] (elderly population, n = 170, rural southern India) and 43% by Mondal et al. [7] (adults, mean age 38.5 years, urban eastern India). In the absence of representative population-based studies, hospital-based studies are often used to estimate symptom prevalence in India, thus giving a skewed distribution.We argue that the prevalence of 16.9% reported in the present study based on a large community sample of urban adults using a standardized and valid PSQI tool characterizes poor sleep quality in the Indian urban context best.

Prevalence of OSA symptoms

The probability of clinically relevant OSA symptoms was 7% in our study. Studies from India among the general population have reported similar prevalence estimates of 8.7% in rural [29] and 9.3% in urban areas [30]. We used the Berlin questionnaire but restricted the categorization to snoring and sleepiness, while excluding the third category, which is characterized by high BMI and hypertension, as obesity is a known risk factor of OSA symptoms [31]. Globally, prevalence of OSA reportedly varies widely from 6·5 to 17% in women, and between 17 and 34% in men [14, 31]. A systematic review of 10 community studies from Asia that used sleep questionnaires reported the prevalence of OSA symptoms between 4.9 and 27.3% [32]. The review, where most of the studies were from Japan, China, and Singapore, further found that male sex, older age, a higher BMI and waist-to-hip ratio, arterial hypertension, and smoking were associated with OSA. Asians are, on an average, less obese than the population in high-income western countries and may have a lower prevalence. The prevalence reported in our study confirms previous reports that the prevalence of OSA symptoms in Asians is similar to that in populations from North America, Europe, and Australia [33]. Comparable prevalence of OSA symptoms despite relatively lower prevalence of obesity may indicate the contribution of other risk factors in a population. We discuss an independent association of psychosocial factors with OSA symptoms in our study below.

Independent contribution of psychosocial and metabolic risk factors to poor sleep quality

Psychosocial problems, mainly symptoms of anxiety and depression, had a strong independent association with poor quality sleep, sleep latency, and disturbed sleep. Our findings were consistent with studies conducted in India [8, 34, 35], China [18], Korea [36], Turkey [37], USA [38], and Europe [39] as well as other populations [40].

Potvin et al. reported anxiety symptoms to be independently related to short sleep duration, daytime sleepiness, and sleep disturbances [41]. A cross-sectional survey among Korean adults (19–69 years) also found anxiety and depression to be more common among those with insomnia [42]. The relation between sleep quality and depression has been shown to be bidirectional [43]. Maladaptive emotional response is postulated as one of the paths through which inadequate sleep leads to depression [44]. Recently, it has also been shown that poor sleep quality affects the connectivity pathways involving the amygdala and the frontal cortex [45]. In the present cross-sectional design, we cannot infer the direction of the association, which needs to be explored further as the LoCARPoN-Cohort matures. Nonetheless, these findings reiterate the importance of screening for emotional disorders, irrespective of gender, in people presenting with sleep disorders in the Indian context.

Metabolic risk factors (hypertension and dyslipidemia—low HDL) were independently associated with poor sleep quality. Poor sleep quality has been associated with an increased risk of metabolic syndrome, including an adverse lipid profile [9, 10], but in some studies, this association was explained by BMI and sleep apnea [46, 47]. However, there is a limited number of studies that explored the association of different metabolic risk factors while adjusting for diverse psychosocial problems among community-dwelling individuals. Kiwan et al. [48] assessed the sleep characteristics and metabolic profile of medicated and non-medicated mentally ill patients compared with a control group. They found the effect size of the association of metabolic risk factors poor sleep quality was similar in both medicated and non-medicated mentally ill patients, perhaps indicating its independent effect.

Gender as a risk factor for poor sleep quality and OSA symptoms

Females were more likely to have poor sleep quality in our study than males. The gender difference in poor sleep quality, with more affected females, has been attributed to hormonal transitions, vasomotor symptoms, stress, depression, anxiety, maladjustment to night-shift work, an overactive bladder, and comorbid conditions, and cognitively females being more perceptive of sleep problems [49, 50]. Our study suggests that accounting for psychosocial and metabolic risk factors does not reduce the female preponderance for poor sleep quality.

While predominantly males are affected more by OSA, the prevalence of OSA symptoms was higher among females in our study [16]. This sex difference was observed although females reported less snoring. A study from south India also reported a higher prevalence among females (11.7%) than males (7.4%) [29]. The association of females with OSA symptoms in our study persisted after adjusting for psychosocial problems (model 2a) and metabolic risk factors (model 2b), but it was no longer significantly associated in the fully adjusted model. This suggests that the apparent association was confounded by an unequal distribution of psychosocial and metabolic risk factors between the two sexes.

The male-to-female ratio of the prevalence of OSA symptoms is less in population-based studies compared to that of hospital-based studies indicating a clinical under-recognition of OSA among females attending hospitals [51]. Moreover, the Berlin questionnaire used to assess OSA symptoms in the present study may overestimate OSA symptoms in females [52]. On the other hand, OSA may well be underreported and underdiagnosed in females, as they are less likely to report snoring due to social awkwardness [49]. This may also have biased the assessment in our study. However, our findings highlight that health-care professionals in India need to ask women of all ages about their sleep to try to identify occult OSA symptoms.

Independent contribution of psychosocial problems to OSA symptoms

In line with prior studies, we found that metabolic risk factors (BMI ≥ 25 kg/m2), psychosocial problems (symptoms of anxiety and moderate-severe depression), and the absence of moderate physical activity were all associated with OSA symptoms [16, 17, 53]. While a link between metabolic risk factors and OSA is established, recent systematic reviews have reported that symptoms of OSA are higher among patients with depressive disorders and anxiety [54]. The underlying temporal direction of the association between OSA symptoms and psychosocial problems is still being investigated. Two cohort studies reported a prospective association between OSA and subsequent development of depressive disorders [55, 56]. Mostly, results on the co-occurrence of different psychosocial problems and OSA symptoms were obtained in clinical populations; studies in the general population are scarce. The present study is one of the largest population-based studies that have assessed an exhaustive list of both metabolic risk factors and psychosocial problems.

Moreover, in our study, dyslipidemia (low HDL) and hypertension were also associated with the occurrence of snoring. In fact, the deleterious effect of OSA on cardiovascular symptoms has been attributed to the high LDL/HDL ratio [57]. In the SYNAPSE study of 834 elderly persons, low HDL levels were related to the severity of OSA [58].

Our finding has public health implications. First, given the concomitant rise in the prevalence of metabolic syndrome and obesity in India, there is a presumable increase in the occurrence of OSA. As a corollary, preventive efforts targeted at BMI are urgently needed to accrue benefits against OSA and related consequences. Second, our findings underscore an opportunity for cross-diagnosis of OSA symptoms and psychosocial problems, and the need to sensitize psychiatrists and sleep care providers. As the prevalence of psychosocial problems is rising fast in India, it is important to introduce screening for OSA in primary care settings, in particular in women.

Strengths and limitations

This is one of the few large-scale epidemiological studies with a clearly defined primary population base (based on voting and population register) exploring the determinants of subjective sleep quality as well as OSA symptoms. Furthermore, we considered multiple psychosocial problems lifestyle factors and metabolic risk factors. A standardized and validated assessment tool and a large sample size increased the internal validity and reliability of our study.

However, several limitations must also be discussed. We used PSQI for sleep quality assessment, rather than the gold standard polysomnography. The latter requires overnight evaluation, is expensive, and is not suitable for population-level estimations in low- or middle-income countries. The PSQI is a widely used, validated tool to assess sleep disturbances. Likewise, we used the Berlin questionnaire to assess OSA symptoms, instead of using the apnea–hypopnea index cut-off for logistical reasons. Finally, the cross-sectional nature of the analysis limits any temporal interpretation of the associations. Nonetheless, our results are valid for pointing to associations with relevance for public health prevention and control.

Conclusion

In older Indian adults, metabolic risk factors and psychosocial problems were associated independently with both subjective sleep quality and OSA symptoms. Our findings have public health implications because the proportion of people of elderly age in India is increasing rapidly, while the prevalence of metabolic risk factors is already high in this age group. This risk pattern has the potential to exacerbate the occurrence of sleep disorders and OSA, and their associated cardiovascular and neurologic sequelae, further straining the Indian health-care system.