Published ahead of print on November 6, 2003, doi:10.1164/rccm.200302-265OC
© 2004 American Thoracic Society Prevalence of Sleep-disordered Breathing and Sleep Apnea in Middle-aged Urban Indian MenDepartment of Chest Medicine, P. D. Hinduja National Hospital and Medical Research Centre, Veer Savarkar Marg, Mahim, Mumbai, India Correspondence and requests for reprints should be addressed to Zarir F. Udwadia, M.D., F.R.C.P., F.C.C.P., Department of Chest Medicine, P. D. Hinduja National Hospital and Medical Research Centre, Veer Savarkar Marg, Mahim, Mumbai 400002, Maharashtra, India. E-mail: zfu{at}vsnl.com
No data are available on the prevalence of sleep-disordered breathing (SDB) and obstructive sleep apnea–hypopnea syndrome (OSAHS) in Indians. We conducted a two-phase cross-sectional prevalence study for the same in healthy urban Indian males (35–65 years) coming to our hospital in Bombay for a routine health check. We also investigated its risk factors and evaluated the significance of the most commonly asked questions that best correlated with the presence of OSAHS. In the first phase, 658 subjects (94%) returned completed questionnaires regarding their sleep habits and associated medical conditions. In the second phase, 250 of these underwent an overnight home sleep study. The estimated prevalence of SDB (apnea–hypopnea index of 5 or more) was 19.5%, and that of OSAHS (SDB with daytime hypersomnolence) was 7.5%. Multiple stepwise logistic regression determined body mass index, neck girth, and history of diabetes mellitus as the principal covariates of SDB. The presence of snoring, nocturnal choking, unrefreshing sleep, recurrent awakening from sleep, daytime hypersomnolence, and daytime fatigue was each statistically significant for identifying patients with OSAHS. The higher prevalence of OSAHS in urban Indian men is striking and may have major public health implications in a developing country.
Key Words: epidemiology • sleep study • India Obstructive sleep apnea–hypopnea syndrome (OSAHS) is a potentially disabling condition characterized by excessive daytime sleepiness (EDS), disruptive snoring, repeated episodes of upper airway obstruction during sleep, and nocturnal hypoxemia. Undiagnosed OSAHS represents a major public health hazard. Various global epidemiologic studies have demonstrated the prevalence of OSAHS to vary from 0.3–5.1% (1–8). These prevalence estimates, however, are based on data from predominantly white populations and may not be applicable to other racial groups. Major etiologic factors such as obesity (5, 6, 9) and craniofacial anatomic predisposition (9–11) are both genetically and environmentally influenced, and it is therefore pertinent to determine the prevalence of sleep apnea in different populations. There are no data on prevalence of sleep-disordered breathing (SDB) and OSAHS in the Indian population, and hence, we conducted a two-stage study to estimate this in healthy urban Indian males between 35–65 years coming for a routine health check to the P. D. Hinduja National Hospital and Medical Research Centre in Bombay, India. We also investigated the risk factors for SDB and OSAHS and evaluated the significance of the most clinically relevant questions used to diagnose SDB and OSAHS.
Sample This investigation was based on a random sample of healthy Indian males residing in Bombay. The males were aged 35–65 years and were coming for a health check to our hospital between December 1999 to December 2000. All of these patients were asymptomatic and came for a health check as part of their employment policy or for insurance reasons. A two-stage sampling scheme designed to optimize the study's precision by oversampling subjects more likely to have SDB was used to construct a cohort representing a wide range of SDB. The exclusion criteria were recent myocardial infarction (one subject) and recent upper-airway surgery (two subjects).
Procedure In the second stage, data of the questionnaire were analyzed, and the respondents were divided as per their snoring habits into habitual snorers and nonsnorers. Because most patients with sleep apnea snore habitually and loudly, we invited by phone 100% of the habitual snorers and 25% of randomly chosen nonsnorers for a home sleep study to yield a cohort with adequate variance in SDB. A technician trained in sleep medicine visited the respondent's house to attach the limited polysomnography (PSG) machine (Compumedics P series, 10-channel system). This machine recorded continuous polygraphic recordings for electrocardiography, nasal and oral airflow (thermisters), tracheal sounds, thoracic and abdominal effort (by inductance plethysmography), limb movement, body position, and oxyhemoglobin level (pulse oxymeter). The technician stayed until the patient seemed to have slept and noted this time. The awakening time was noted by the patient himself and reported to the technician when he returned to disconnect the leads. The polysomnographic records were manually scored by a doctor specialized in sleep medicine. An abnormal breathing event was defined as a complete cessation of airflow for 10 seconds or more (apnea) or a discernible 50% reduction in respiratory airflow accompanied by a decrease of 4% or more in oxyhemoglobin saturation (hypopnea). The apnea–hypopnea index (AHI) was defined as total number of apneas and hypopneas divided by the number of hours of sleep. SDB was defined as an AHI of 5 or more and OSAHS as SDB with daytime hypersomnolence.
Statistical Analysis
Calculation of Prevalence
A flow chart of the study population is available in Figure E1 in the online supplement. In the first stage, 658 of 700 men returned the questionnaire, giving a response rate of 94%. Of these, 171 (26%) were snorers and 487 (74%) were nonsnorers. In the second stage, all of the snorers (n = 171) and 25% of nonsnorers (n = 122) were contacted to undergo a home sleep study (total n = 293). Of these, 151 (88.3%) snorers and 103 (84.4%) nonsnorers agreed (total n = 254), giving a participation rate of 87%. Subjects who agreed to the home sleep study and those who refused the home sleep study in snorers and nonsnorers showed similar frequencies with regard to their responses to all questionnaire items on sleep characteristics, body habitus, and age. However, of these, four sleep studies were excluded (one in a snorer and three in nonsnorers), as they were unable to sleep with leads attached and hence disconnected them within 1–3 hours. Thus, 250 sleep studies (150 in snorers and 100 in nonsnorers) were included in the final analysis. The demographic profile of the snorers and nonsnorers who returned completed questionnaires is given in Table 1 . The snorers had a significantly higher weight, BMI, neck girth, waist girth, and hip girth as compared with nonsnorers. Habitual snoring was seen in 26% of the study population, nocturnal choking/witnessed apneas in 5%, and daytime hypersomnolence in 22% of the study population. The mean age of the sample was 47.84 years, and the mean BMI was 24.56. The mean Epworth score of snorers was 8.39 ± 4.67, and that of nonsnorers was 6.05 ± 3.67.
Prevalence of SDB and OSAHS A wide range of SDB ranging from apnea–hypopnea scores of 0–73 was found among the 250 subjects who underwent limited PSG. Of these, 93 subjects (87 snorers and 6 nonsnorers), 61 subjects (all snorers), and 45 subjects (44 snorers and 1 nonsnorer) had AHIs of 5 or more, 10 or more, and 15 or more, respectively. When the mean age and BMI of sample subjects who underwent PSG were compared with those who did not have PSG in the respective snoring groups, no significant difference was found. Thus, the prevalence rate of SDB in the PSG group of snorers (58%, 95% CI, 50–66%) and nonsnorers (6%, 95% CI, 1–11%) was considered representative of the corresponding questionnaire respondents group, and results were extrapolated (13). Thus, the prevalence of SDB in our population was 19.5%. Forty subjects (39 snorers and 1 nonsnorer) had EDS along with an AHI of 5 or more, making the prevalence of OSAHS in healthy Indian males between 35–65 years 7.5%. The prevalence of SDB and OSAHS at various cutoff points of AHI is depicted in Figure 1 .
The prevalence of SDB at apnea–hypopnea scores of 5 or more, 10 or more, and 15 or more was extrapolated from the cohort to the general population (Table 2) . The maximum prevalence of SDB was seen in the age group 45–54 years. However, age was not a significant risk factor, and no significant difference was seen among the three age groups.
Factors Associated with SDB and OSAHS The significant correlates of SDB in the PSG subjects with the odds ratios are shown in Table 3 . The variables, which were significantly affected, were included in a multiple logistic regression. Stepwise logistic regression selected and retained BMI, neck girth, and history of diabetes mellitus as principal covariates for SDB. Age-adjusted odds ratio for BMI was 3.47 (95% CI, 2.23–9.96), for neck girth was 3.85 (95% CI, 1.72–10.11), and for history of diabetes mellitus was 2.03 (95% CI, 1.11–2.83).
Table 4 shows the odds ratio estimating the increased risk of SDB as against specific questions regarding sleep characteristics of the patient. A history of snoring, EDS, nocturnal choking, unrefreshing sleep, recurrent awakening from sleep, and daytime fatigue was each significantly associated with SDB and OSAHS with high odds ratio.
A number of epidemiologic studies have been performed to evaluate the prevalence of OSAHS in various ethnic and racial groups. However, most of these studies are Western studies performed in predominantly white populations and may not be applicable to other racial groups. Therefore, the global importance and impact of this disease remain poorly understood. At present, data from studies in nonwhite groups are too sparse to even determine with confidence whether prevalence differs worldwide (14). An important Asian study was conducted by Ip and colleagues (15), who investigated the prevalence of OSAHS in Chinese office workers from Hong Kong. Her figures of 4% OSAHS were similar to those from Western studies despite obesity, a strong risk factor for OSAHS, being relatively uncommon in Asian countries. She postulated that craniofacial features that compromise the upper airways could exist in her population and account for the high OHAHS prevalence. When 75 skulls of Indian origin were compared with 98 of Tuscan origin, cephalometry showed differences in mandibular length and the anteroposterior dimensions of the nasopharynx-pharyngeal tubercle to posterior nasal spine in the Indian skulls, pointing to the possible presence of an osteogenic etiology of OSAHS (16). Another study by Li and colleagues (17) showed that when compared with white men, Far East Asian men were less obese but had a greater severity of OSAHS. A crucial conclusion from these studies is that predictive equations for the presence of OSAHS developed from populations of obese, white subjects and based on weight or facial measurements are unlikely to be accurate for Asian population, and the prevalence of OSAHS in different ethnic groups may vary considerably and should be studied individually. No data are available on the prevalence of this condition from the Indian subcontinent, and a Medline search confirms that this is the first epidemiologic study evaluating the prevalence of OSAHS from this country. Sleep medicine has been slow to develop in India, there being no more than 40 sleep laboratories in a country of one billion. High costs of equipment, large patient workloads in hospitals, and methodologic difficulties have probably been factors that have discouraged other epidemiology studies from India. We have attempted to redress this imbalance with this study. Our study is strengthened by the very high response rate of 94%. This study confirms the wide prevalence of OSAHS in the urban Indian population. Our prevalence rates of 7.5% for OSAHS are among the highest prevalence rates reported from epidemiologic studies across the globe and are higher than most Western and other Asian studies. The exact causes for this higher prevalence are unclear, but it is possible that Indian facial and anthropometric characteristics might be responsible. India is too vast and diverse a country and huge gulfs exist between urban and rural populations (70% of Indians live in villages, many remote and inaccessible). Our study population represents urban Indian males. It also represents males that are better educated and employed, have higher incomes, and are of better socioeconomic class and status than an impoverished Indian villager. These more affluent males are therefore more likely to be westernized and perhaps have a higher prevalence of diabetes, hypertension and ischemic heart disease than rural Indians. Also, urban Indian males are significantly more obese (BMI of 24) than their rural counterparts (BMI of 20) (18). Thus, the prevalence rates of our study would be representative of urban males of better socioeconomic status (and may be higher than for rural Indian men). Women were not included in our study because our point of inclusion was subjects having company checks for employment or insurance reasons; fewer women than men are sent for such checks in India, and they would have therefore been underrepresented. Another practical reason for excluding women was bearing in mind Indian social norms: most Indian women would have been reluctant to allow a male technician to enter their house for a night study. The results of a prevalence study are influenced by the characteristic of the study population. Because the prevalence found in the study group can only be generalized to a population with similar characteristics, the study group must be representative of the total population. Unfortunately, there are no studies stating national average BMI for Indians. Dhurandhar and Kulkarni (19) studied a population from various sections of society from Bombay and found the mean BMI of adult males to be 23.9 ± 3.8. The mean BMI of our study population was 24.56 ± 4.3 (statistically no significant difference). Because BMI was an important predictor of SDB, our population would be fairly representative of adult urban Indian males. We used home sleep studies for the diagnosis of OSAHS in our study population. Although the proposed gold standard for the diagnosis of OSAHS is in-patient overnight PSG (20), as recommended by American (21) and Australasian (22) guidelines, it is unclear whether such complex and expensive investigation is appropriate, especially in a developing country. A number of studies (23–25) have shown that domiciary use of a portable respiratory device can be a cost-effective yet accurate way to diagnose OSAHS. The diagnostic usefulness of the home sleep study is supported by its acceptable sensitivity and specificity. Our findings of habitual snoring in 26% of population, nocturnal choking in 5%, and daytime hypersomnolence in 22% are consistent with those found by the others (26, 27) and indicate that the symptoms of OSAHS are common in the general population. The definition of OSAHS is arbitrary, and it has been suggested that an AHI of 5 or more is a low cut-off value, especially for older people (28, 29), and many studies have used higher cut-off values for AHI. Young and coworkers in their Wisconsin study (5) found the prevalence of OSAHS to be 4%, 2.3%, and 1.4% using AHI cut-offs of 5 or more, 10 or more, and 15 or more with moderate to severe daytime sleepiness. Ip and colleagues (15) found the prevalence of OSAHS in Hong Kong as 4%, 3.2%, and 3.1% at an AHI of 5 or more, 10 or more, and 15 or more, respectively. Gislason and colleagues (3) found the prevalence of OSAHS in Sweden to be 1.4% and 0.9% with an AHI of 5 or more and 10 or more, respectively. Cirignotta and colleagues (4) found the prevalence of OSAHS to be 5.1% and 3.3% in Italy using an AHI cut-off of 5 or more and 10 or more, respectively, but only 0.5% of these had severe symptomatic OSAHS requiring treatment. Bearpark and colleagues (6) found the prevalence of OSAHS in Australia to be 3.1% using an AHI of 5 or more with sleepiness. Franceschi and colleagues (30) used the criteria of AHI of 10 or more and found the prevalence of OSAHS to be 1%. In a sample of male population using a definition of OSAHS as an AHI of 10 or more plus daytime hypersomnolence, hypertension, or other cardiovascular complications, Bixler and coworkers (31) found a prevalence of 3.3%. Using an AHI of 10 or more with EDS as definition of OSAHS, Duran and coworkers (26) found a prevalence of 3.2% in Spain. Stradling and Crosby (32) used the definition of an AHI of 20 or more with symptoms and found the prevalence of OSAHS in the United Kingdom to be 0.3%. The prevalence of OSAHS seen in our study was higher and was 7.5%, 6.1%, and 5.4% using AHI cut-offs of 5 or more, 10 or more, and 15 or more with EDS, respectively. The correlation between age and SDB has been studied by different investigators with dissimilar results, with suggestion of a rise in prevalence of SDB with age (5, 26, 31, 33). Age was not found to be a significant risk factor in our study, and no such trend in the prevalence of SDB with increasing age was seen. The highest prevalence was seen in the age group 45–54 years, but it was not significantly higher than the prevalence in the other two age groups. The lack of a continuous increase in the prevalence of SDB with increasing age in our study suggests that age is not a strong risk factor for SDB over the middle decades of life. Obesity is a significant risk factor for SDB in white populations (5–8). In this study, a higher BMI was a risk factor for SDB in Indian subjects as well. To assess further the impact of obesity on SDB in our population, the relative risks of having SDB in relation to measure of body habitus were calculated. On comparing the odds ratios in our subjects with those reported in the Wisconsin study (5) and the study of Ip and colleagues (15), the risks of having SDB due to an increase in any index of adiposity was much higher in our population (Table 3). An epidemic of obesity is sweeping across India (34), and it could be projected that the number of cases of OSAHS will increase over the next few decades. However, it is worth stating that 46% of our subjects with SDB had a BMI of less than 30, the Western cut-off for obesity (35), whereas 27% of subjects with SDB had a BMI of less than 27, which is the cut-off point for obesity in Asians (36). These observations suggest that a significant percentage of our subjects were not obese by Western or Asian standards but still had SDB. This leads us to postulate that other craniofacial risk factors for SDB, such as pharyngeal narrowing, retrognathia or micrognathia, and pharyngeal collapsibility, might assume greater pathogenic significance in Indian subjects and may be responsible for our higher prevalence. The importance of a larger neck girth in producing upper airway incompetence during sleep has been documented in patients with sleep apnea in a number of studies (13, 32, 37, 38), and the mechanism is presumably external compression of the pharynx by superficially located fat masses. In agreement with these findings, our study also showed neck girth to be an important predictor of SDB and found that the risk of SDB is 5.34 times higher for subjects with a neck girth of 17 inches (mean + SD) or more—the mean being 15.7 and the SD being 1.33 for 250 subjects with PSG. A significant association of SDB with diabetes mellitus was found in our study after adjusting for demographics (age) and anthropometric variables (including BMI and neck girth). A number of studies have shown that SDB may be causally associated with metabolic derangements such as glucose intolerance and insulin resistance (39–41). The association of diabetes and OSA has been evaluated in a sample of 116 age-stratified men with hypertension selected from subjects in a population-based study in Sweden. It was shown that although obesity was the main risk factor for diabetes mellitus, coexistent severe OSA may add to the risk independently (42). Also recent studies by Ip and colleagues (43) and Punjabi and colleagues (44) have shown that SDB is independently associated with glucose intolerance and insulin resistance. Collectively, the effects of elevated sympathetic activity, the alterations in glucocorticoid regulation induced by sleep loss, and recurrent intermittent hypoxemia associated with SDB may facilitate the development of glucose intolerance and insulin resistance. However, in our study, we did not measure blood sugar or insulin levels and considered diabetes to be present when the respondent gave that history in the questionnaire. In contrast to a systematic review in which no firm evidence for the contribution of OSAHS to hypertension was demonstrated (45), it has been recently shown that SDB and hypertension are clearly linked (46–48). In our study, although an association of SDB with hypertension was significant, multiple logistic regression analysis did not select hypertension as a principle covariate. Another objective of our study was to evaluate the significance of common symptoms that are known to be associated with SDB and OSAHS. We found that a positive history of snoring, EDS, nocturnal choking, recurrent awakening from sleep, unrefreshing sleep, and daytime fatigue were each significantly associated with SDB and OSAHS with high odds ratio and must be elicited while evaluating patients with possible OSAHS. The limitations of our study are that it only represents prevalence rates in urban Indian male population, and these rates cannot be generalized to the rural Indian population or to Indian women. Also, lack of EEG defined sleep may have caused an underestimation of AHI. The strengths of this study are that this is the only study to date investigating the prevalence of SDB and OSAHS in India. The study is also strengthened by the high response rate of 94%, and thus, overestimation of prevalence due to participation by subjects with self-perception of sleep disorders would be unlikely. In summary, we found that the prevalence of SDB was 19.5%, and that of OSAHS was 7.5% in healthy urban Indian males between 35–65 years of age. The findings of this study and the high prevalence rates in middle-aged urban Indian men might have important public health implications in a developing country with limited health resources. Indian men in this age group have among the highest rates of ischemic heart disease and hypertension worldwide (49). When compared with whites, blacks, Hispanics, and other Asians, coronary artery disease rates among Indians worldwide are two to four times higher at all ages (18). India also has among the largest number of individuals with diabetes (approximately 25 million) (50). The potential impact of undetected and uncontrolled OSAHS on the previously mentioned populations might be considerable (51, 52) and worthy of further study. In our study, BMI, neck girth, and a history of diabetes mellitus were significantly associated with SDB, and a history of snoring, EDS, nocturnal choking, recurrent awakening from sleep, unrefreshing sleep, and daytime fatigue should always be elicited while evaluating a patient with possible OSAHS. The application of these easily identifiable predictive factors would assist in appropriate referral and prioritization for PSG. With the high prevalence of OSAHS and limited funds, this might be useful for better utilization of resources.
The authors thank their statisticians, Ms. Nikita Agnihotri and Mr. Suresh Bowalekar, for helping with statistical analysis; PFT technicians Ms. Lalita Angne and Ms. Mangal Walkar for distributing and collecting the questionnaires from health check patients; Professor Mary Ip from Hong Kong and Professor John Stradling from Oxford for their constructive advice; and the Research and Ethical Committee of Hinduja Hospital for supporting this study.
This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: Z.F.U. has no declared conflict of interest; A.V.D. has no declared conflict of interest; S.G.L. has no declared conflict of interest; C.I.S. has no declared conflict of interest. Received in original form February 24, 2003; accepted in final form October 31, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
