Relationship of 30-Year Changes in Obesity to Sleep-Disordered Breathing in the Western Collaborative Group Study

Transcription

1 Relationship of 30-Year Changes in Obesity to Sleep-Disordered Breathing in the Western Collaborative Group Study Dorit Carmelli,* Gary E. Swan,* and Donald L. Bliwise Abstract CARMELLI, DORIT, GARY E. SWAN, AND DONALD L. BLIWISE. Relationship of 30-year changes in obesity to sleep-disordered breathing in the Western Collaborative Group Study. Obes Res. 2000;8: Objective: Obesity is an important etiologic factor in sleepdisordered breathing (SDB), but the extent to which changes in obesity across adult life contribute independently to SDB in old age has not been studied. In this study, we examined the association between changes in obesity from midlife to late adulthood and overnight recording of respiration during sleep. Research Methods and Procedures: Subjects in this study are from the Western Collaborative Group Study, a longitudinal cardiovascular epidemiological study that began in 1960 through Overnight sleep recordings were obtained from 281 male participants in the 1995 through 1996 follow-up of the Western Collaborative Group Study. Subjects were 75 to 91 years old when assessed for SDB as indexed by the respiratory disturbance index and an oxygen desaturation index (O 2 DI). Long-term changes in anthropometrics were evaluated and examined in relation to SDB severity. Results: Over the 30 years of follow-up, body mass index and waist circumference increased significantly for this sample and were associated with SDB severity as indexed by respiratory disturbance index and O 2 DI. Waist circumference at baseline and gain in waist circumference over the 30 years of follow-up (both p 0.01) were significantly and independently associated with SDB severity as assessed by O 2 DI. However, percentage of variance as accounted for by waist circumference was modest. Submitted for publication January 24, Accepted for publication in final form July 5, *Center for Health Sciences, SRI International (formerly Stanford Research Institute), Menlo Park, California; and Sleep Disorders Center, Emory University School of Medicine, Atlanta, Georgia. Address correspondence to Dorit Carmelli, Center for Health Sciences, SRI International, 333 Ravenswood Avenue, Menlo Park, CA doritc@unix.sri.com Copyright 2000 NAASO Discussion: This study supports the hypothesis that gain in waist circumference over adult life is significantly associated with SDB severity in older men. Key words: aging, sleep-disordered breathing, central obesity Introduction Epidemiological studies in recent years have identified sleep-disordered breathing (SDB) as a prevalent condition among the elderly linked to cardiovascular disease (CVD), daytime sleepiness, cognitive dysfunction, and motor vehicle accidents, but crucial knowledge of causality is still lacking. The strongest risk factors of SDB are age, gender, obesity, and obesity-by-gender interaction (1). Because obesity has the greatest impact on SDB in older men (2,3), the basis for this marked increase in prevalence could be changes in body fat distribution that favor an increase in central fat deposition (e.g., waist and neck). CVD and associated comorbidities (e.g., diabetes and hypertension) are also more prevalent in the elderly and may enhance the association of obesity with SDB (4). Understanding the impact of midlife obesity on SDB in old age is extremely important because it may provide the key to targeted intervention and reduction of associated morbidity. Previous reports in the literature have suggested multiple mechanisms by which obesity increases the risk for SDB. Independent associations with respiratory indices have been found with various indices of body habitus including body mass index (BMI), waist-to-hip ratio, and neck circumference (5 7). Weight loss has been shown to improve SDB in morbidly obese patients, suggesting a causal role of obesity in SDB (8). A recent study that examined 1-year weight histories in newly diagnosed patients with SDB found a significant weight gain in subjects with SDB compared with those without (9). The extent, however, that weight gain across adult life is associated with SDB in the elderly is not known. 632 OBESITY RESEARCH Vol. 8 No. 9 December 2000

2 A large amount of research has been devoted to the description of changes in obesity and fat distribution in late adulthood and their contribution to an increased risk of cardiovascular morbidity and mortality (10 12). Little is known, however, about the role these changes have on SDB occurrence and severity in old age. The objective of this study was to address this question using available information on changes in obesity across adult life for a cohort of elderly men. Given the strong relationship between obesity and SDB, we hypothesize that midlife obesity and changes in obesity over 30 years of adult life are associated with SDB occurrence and severity in old age. Research Methods and Procedures Subjects In the present analysis, we used data from the Western Collaborative Group Study (WCGS), a prospective cardiovascular epidemiological study that began in 1960 through 1961 with 3152 healthy white males drawn from 10 large California corporations (12). Age of participants at intake ranged from 39 to 59 years. All participants were determined to be healthy and free of CVD. Seven examinations occurred between 1960 and 1970 that carefully monitored CVD incidence and changes in CVD risk factors (13). A comprehensive battery of anthropometric measurements was obtained during the 1964 through 1965 follow-up examination (10), and these were used as baseline measures in the present analysis. Subsequent follow-ups to ascertain the vital status of this cohort occurred in 1982 through 1983 and again in 1986 through 1988 (14). Two additional rounds of physical examinations of survivors of this cohort were conducted in 1991 through 1992 and 1995 through Overnight sleep recordings were conducted on a subset of 281 male participants in the 1995 through 1996 follow-up of the WCGS. These subjects were selected because of geographic proximity to the examination site. A comparison of exam participants with and without sleep-recordings with respect to age, education, and obesity at baseline or follow-up did not find significant differences. Measures Anthropometric measures taken in the 1960s and taken again during the 1995 through 1996 follow-up included the following: height to the nearest millimeter; weight to the nearest 0.25 pound, using a research grade scale (Health- O-Meter, Bridgeview, IL); and circumferences obtained with a flexible metal tape measure, maintaining close contact with the skin without compression of the underlying tissues. Body circumferences included relaxed arm midway between the elbow and the acromial process of the shoulder, waist at the umbilicus or at the maximum girth for larger subjects, and calf at the widest part of the calf. Measurements were read to the nearest 0.1 cm. Minimal neck circumference was measured to the nearest 0.1 cm, with the tape perpendicular to the long axis of the neck, and was taken at the most recent follow-up but not in the 1960s. All measurements were taken twice, and the average was used in the final analyses. Respiratory and Cardiac Activity during Sleep Subjects were monitored in their own homes with ambulatory technology, which was applied by a technologist the evening of the recording. The Edentrace II Recording System (Mallinckrodt, Hazelwood, MO), an ambulatory microprocessor, which recorded breathing sounds, airflow, respiratory effort, oxygen saturation, body position, and heart rate, was used to derive measures of SDB severity. Standard disposable Edentrace sensors for airflow and pulse oximetry were used. Heart rate and respiratory effort were recorded with silver chloride transthoracic impedance sensors (Vermed, Bellows Falls, VT). Oxygen saturation was measured using standard pulse oximetry hardware and software contained within the Edentrace II. The Edentrace II has been used widely in numerous epidemiological studies recording respiration during sleep in the home setting (15). Validation studies have been reported by Redline et al. (16) and Emsellem et al. (17), who noted high ( 90%) sensitivity/ specificity and high correlations ( 95%) between data collected with this system and data obtained using standard polysomnographic techniques. After morning downloading, the computerized scoring of all breathing events for the entire night was edited manually on an event-by-event basis by a registered polysomnographic technologist blind to all subjects identifying information. Recording parameters on the Edentrace II were set as follows: 1) Apnea, an airflow pause or absence of respiratory effort of at least 10 seconds from previous baseline, where baseline average excluded the lowest 7 of the last 10 breaths; 2) Hypopnea, a fall in airflow amplitude of 50% below average amplitude for at least 10 seconds, accompanied by at least a 4% drop in oxygen saturation; and 3) Desaturation, any fall in oxygen saturation to 90% or a drop of 4% below baseline saturation. Based on these definitions, we computed for each subject a respiratory disturbance index (RDI) equal to the total number of apneas and hypopneas divided by total hours of recording time. In addition, we calculated an oxygen desaturation index (O 2 DI), defined as the number of desaturations per hour of total recording time. Statistical Analysis Data were analyzed for the group as a whole and after subdivision of participants into obese (BMI 28) and non-obese (BMI 28) as of their last exam. Participants characteristics at baseline and when monitored for SDB are presented as mean SD unless otherwise specified. Using Spearman s correlation coefficient, we first evaluated the OBESITY RESEARCH Vol. 8 No. 9 December

3 relationship of anthropometric variables at both baseline and follow-up with respiratory indices. Multivariate analysis, using multiple regression, was then performed to select independent predictors of respiratory measurements. Variables were entered into the regression models after a screen for colinearity was conducted so that dependent variables exhibiting significant correlations were included in separate regression models. The model with the best fit (largest R 2 value) was selected as the final model. To investigate differences in the prediction of SDB in obese and non-obese subjects, separate regression models were developed. All statistical analyses were performed with the SAS statistical package (SAS Inc., Cary, NC). Results Sample Characteristics Subjects in the present study are 281 community-dwelling older men, mean age years (range 75 to 91) when monitored for SDB. Mean BMI and mean waist circumference at baseline were kg/m 2 and cm, respectively; these increased significantly to kg/m 2 and cm, respectively, over the 30 years of follow-up. The increase in waist circumference from midlife to old age was strongly and positively associated with the increase in BMI (r 0.80). Mean neck circumference at follow-up was cm (range 35 to 51); no measurement of neck circumference was taken at baseline on these subjects. Mean RDI was 4.4 per hour (median was 2.6; interquartile range was 1 to 6 apneic/hypopneic episodes per hour of sleep). By the conventional criteria of RDI 5, 29.5% of monitored subjects could be classified as having mild SDB. Mean RDI SD of subjects who met SDB criteria was Similarly, mean O 2 DI in this sample was 13 desaturations per hour of sleep (median was 9; interquartile range was 4.2 to 18.2 desaturation episodes per hour of recording time). For those who met SDB criteria, mean O 2 DI SD was compared with a mean SD of in those without SDB. In univariate analyses, obesity measurements at baseline and follow-up were significantly correlated with the RDI and O 2 DI (Table 1). Overall, the correlations with O 2 DI were stronger than those with RDI. In addition, gains in obesity were significantly associated with O 2 DI but not with RDI; similarly, neck circumference at follow-up was significantly associated with O 2 DI (r 0.16) but not with RDI (r 0.08). Age showed a significant association with RDI but not with O 2 DI. Subgroup Analyses Classified by BMI at follow-up, 187 subjects were in the normal weight group (BMI 28) and 94 were in the obese group (BMI 28). Compared with non-obese, the obese Table 1. Correlations between anthropometric measurements at baseline and follow-up and sleeprecorded respiratory variables Variable RDI O 2 DI Age (yr) 0.13* s variables BMI (kg/m 2 ) Waist circumference (cm) variables BMI (kg/m 2 ) Waist circumference (cm) Neck circumference (cm) Change variables BMI (kg/m 2 ) * Waist circumference (cm) * p p p were younger, had a larger mean BMI and mean waist circumference at baseline, and experienced a greater increase with age in both BMI (mean SD, in obese vs in non-obese; p 0.001) and waist circumference (mean SD, in obese vs in non-obese; p 0.001). As expected and seen in Table 2, the prevalence of SDB measured by RDI (36% vs. 26%; p 0.06) and O 2 DI (mean SD, vs ; p 0.001) was higher among the obese than nonobese participants. Independent Correlates of SDB There was significant colinearity between the obesity measures at each examination cycle and between changes in BMI and changes in waist circumference (Table 3). Because of colinearity, baseline measurements and change in both BMI and waist circumference were not entered in the same regression model. Instead we adopted the strategy of testing different models, using various combinations of obesity measurements, and selecting the model with the largest R 2 value as the final model. Table 4 summarizes the best predictive model for the obese and non-obese subgroups. Independent predictors of RDI in the non-obese subgroup were older age ( coefficient ; p 0.04) and gain in waist circumference ( coefficient ; p 0.03), accounting together for 6% of the observed variability. The contribution of age to prediction of RDI was modest, compared with that of the change in waist circumference. In the obese subgroup, only the gain in waist circumference was an independent predic- 634 OBESITY RESEARCH Vol. 8 No. 9 December 2000

4 Table 2. Anthropometric and respiratory measurements in subjects by obesity in old age Variable Non-obese (BMI < 28) (n 187) Obese (BMI > 28) (n 94) p Value Age in BMI in 1960s BMI in BMI Waist circumference in 1960s Waist circumference in Waist circumference Neck circumference RDI in O 2 DI in RDI 5 in % 36% 0.06 tor of RDI. Furthermore, SDB severity as assessed by O 2 DI in both obese and non-obese subjects was significantly and independently associated with both waist circumference at baseline ( coefficient ; p 0.01) and gain in waist circumference over the 30 years of follow-up ( coefficient ; p 0.01). However, the percentage of variance as accounted for by waist circumference was modest (12% in the obese and 7% in the non-obese subgroups). Substituting either BMI or neck circumference for waist circumference resulted in poorer fit in both the obese and non-obese subgroups. Discussion For this sample of surviving men from the WCGS, we found that both midlife waist circumference and the increase in waist circumference over 30 years of adult life were independently associated with SDB in old age. Substituting changes in BMI for changes in waist circumference did not improve the association with SDB occurrence and severity. Previous epidemiological studies have found strong cross-sectional associations between central obesity and SDB (1). Others have reported that external neck circumference can explain most or all of the observed association between obesity and SDB (15 16). Davies and Stradling (18) studied 66 patients referred for investigation of sleep apnea and found that neck circumference explained 42% of the observed variability in severity of sleep apnea. Moreover, the authors of this study found that the relationship between overall obesity and SDB was fully explained by variation in neck size. Katz et al. (19) studied 123 patients and found that neck circumference explained 29% of the variability in SDB, compared with only 4% for BMI. Neither study reported any body circumference measurements apart from neck circumference. Hoffstein and Mateika (5) Table 3. Intercorrelations between adult anthropometric measurements BMI 1960 Waist 1960 BMI 1996 Waist 1996 Neck 1996 BMI BMI Waist BMI Waist Neck BMI Waist All the bivariate correlations except the correlation between BMI in the 1960s and change in waist circumference are significantly different from zero. OBESITY RESEARCH Vol. 8 No. 9 December

5 Table 4. Independent predictors of RDI and O 2 DI index in subjects by obesity in old age Model Independent variable Partial correlation p Value Non-obese (n 187) RDI Age Waist Model R O 2 DI Waist 1960s Waist Model R Obese (n 94) RDI Waist 1960s Model R O 2 DI Age Waist 1960s Waist Model R measured BMI and neck and abdominal circumferences in 670 clinic patients and reported that only neck circumference and BMI were independently related to SDB. In addition, they observed that in the non-obese patient subgroup, neck circumference alone and not abdominal circumference predicted sleep apnea. In this sample of healthy elderly survivors from the WCGS, we found that central obesity in midlife was associated with SDB in old age. In addition, the cross-sectional correlation between neck circumference and respiratory measures in the WCGS was lower than that reported in other studies. However, an important factor that complicates analysis of the relationship between anthropometric variables and SDB is colinearity. Close correlation between explanatory variables can invalidate multiple regression analysis. For example, changes in BMI are strongly associated with changes in waist circumference and cannot be entered into the same regression model. Similarly, colinearity is further emphasized by the correlation between neck circumference and waist circumference (r 0.70). However, if the link between central obesity and SDB is neck fat deposition, then it would be expected that neck circumference rather than waist circumference is the better correlate of SDB. This was not the case in the present sample, suggesting that in elderly men, neck circumference may be a weaker correlate of SDB than in younger patients. Why is abdominal mass in older men an important correlate of SDB? It is known that abdominal obesity will reduce lung volume, particularly in the supine position, which in turn, can influence upper-airway dimensions (20,21). Impaired respiratory muscle force has also been noted in centrally obese patients (22). The central obesity SDB link also may be related to abnormal airway muscle function rather than structure. A reduction in muscle fibers in the middle pharyngeal constrictor muscle has been demonstrated in non-obese habitual snorers (23), and studies of sleep apnea patients before and after weight loss have shown changes in upper-airway function rather than structure (24). In older men, age-related changes in fat distribution favor an increase in central obesity more so than in younger patients. In addition, the influence of central obesity in this selected sample of survivors from the WCGS may also have been underestimated because of loss to follow-up and excess mortality among nonparticipants. The present study is clearly limited by the nature of the design and population studied. The WCGS is largely white, upper-middle class, well-educated, and probably healthier than most typical elderly populations. The lengthy period of follow-up (30 years) no doubt resulted in a biased group of survivors. The mean RDI for this sample was also lower than that seen in some, but not all, studies of elderly populations. For example, Ancoli-Israel et al. (25) reported a mean RDI of 32.2 (SD 45) among 427 elderly community volunteers. However, Bliwise (26) reported a mean RDI of 4.5 (SD 9.9) in 256 elderly volunteers, Phillips et al. (27) reported a mean RDI of 2.7 in 92 elderly volunteers, and Dickel and Mosko (28) reported a mean RDI of 5.5 (SD 6.7) in their sample of 100 community volunteers. Because the latter three studies employed full polysomnography and therefore adjusted the total number of recorded breathing events by the hours of polysomnographically recorded sleep, this may have resulted in a reduction in the RDI compared with the data of Ancoli-Israel et al., who relied on a system that did not allow direct measurement of time asleep. Nonetheless, the most likely explanation of the relatively low RDI encountered in this sample is that these individuals represent survivors of a cohort that was more likely to be healthier. Acknowledgments This study was supported by Grant AG from the National Institute on Aging and Grant HL from the National Heart, Lung, and Blood Institute. References 1. Grunstein RR, Wilcox I. Sleep-disordered breathing and obesity. Baillieres Clin Endocrinol Metab. 1994;8: Young T, Palta M, Dempsey J, et al. The occurrence of sleep-disordered breathing in middle-age adults. N Engl J Med. 1993;328: Redline S. Age-related differences in sleep apnea: generalizability of findings in older populations. In: Kuna S, Suratt P, Remmers J, eds. Sleep and Respiration in Aging Adults. New York: Elsevier; 1991, pp OBESITY RESEARCH Vol. 8 No. 9 December 2000

6 4. Grunstein RR. Endocrine and metabolism disturbances in obstructive sleep apnea. In: Saunders NA, Sullivan CE, eds. Sleep and Breathing. 2nd ed. New York: Marcel Dekker; 1993, pp Hoffstein V, Mateika S. Differences in abdominal and neck circumferences in patients with and without obstructive sleep apnoea. Eur Respir J. 1992;5: Shinohara E, Kihara S, Yamashita S, et al. Visceral fat accumulation as an important risk factor for obstructive sleep apnoea syndrome in obese subjects. J Int Med Res. 1997;241: Vgontzas AN, Tan TL, Bixler EO, et al. Sleep apnoea and sleep disruption in obese patients. Arch Intern Med. 1994;154: Pillar G, Peled R, Lavie P. Recurrence of sleep apnoea without concomitant weight increase after weight reduction surgery. Chest. 1994;106: Phillips BG, Hisel TM, Kato M, et al. Recent weight gain in patients with newly diagnosed obstructive sleep apnea. J Hypertens. 1999;17: Chumlea WC, Baumgartner RN. Status of anthropometry and body composition data in elderly subjects. Am J Clin Nutr. 1989;50: Carmelli D, McElroy M, Rosenman RA. Longitudinal changes in fat distribution in the Western Collaborative Group Study. Int J Obes. 1991;15: Carmelli D, Zhang H, Swan GE. Obesity and 33-year follow-up for coronary heart disease and cancer mortality. Epidemiology. 1997;8: Rosenman RH, Brand RJ, Jenkins D, et al. Coronary heart disease in the Western Collaborative Group Study: final follow-up experience of years. JAMA. 1975;233: Ragland DR, Brand RJ. Coronary heart disease mortality in the WCGS: follow-up experience of 22 years. Am J Epidemiol. 1988;127: Redline S, Tishler PV, Tosteson TD, et al. The familial aggregation of obstructive sleep apnea. Am J Respir Crit Care Med. 1995;151: Redline S, Tosteson T, Boucher MA, Millman RP. Measurement of sleep-related breathing disturbances in epidemiologic studies: assessment of the validity and reproducibility of a portable monitoring device. Chest. 1991;100: Emsellem HA, Corson WA, Rappaport BA, Hackett S, Smith LG, Hausfeld JN. Verification of sleep apnea using a portable sleep apnea screening device. South Med J. 1990;83: Davies RJO, Stradling JR. The relationship between neck circumference, radiographic pharyngeal anatomy, and obstructive sleep apnea. Eur Respir J. 1990;3: Katz I, Stradling J Slutsky AS, et al. Do patients with obstructive sleep apnea have thick necks? Am Rev Respir Dis. 1990;141: Begle RL, Badr S, Skatrud JB, Dempsey JA. Effect of lung inflation on pulmonary resistance during NREM sleep. Am Rev Respir Dis. 1990;141: Hoffstein V, Zamel N, Phillipson EA. Lung volume dependence of pharyngeal cross-sectional area in patients with obstructive sleep apnea. Am Rev Respir Dis. 1984;130: Lopata M, Onal E. Mass loading, sleep apnea, and the pathogenesis of obesity hypoventilation. Am Rev Respir Dis. 1982;126: Smirne S, Iannaccone S, Ferini-Strambi L, et al. Muscle fiber type and habitual snoring. Lancet. 1991;337: Rubinstein I, Colaptino N, Rotstein LE, et al. Improvement in upper airway function after weight loss in patients with obstructive sleep apnea. Am Rev Respir Dis. 1988;138: Ancoli-Israel S, Kripke DF, Klauber MR, Mason WJ, Fell R, Kaplan O. Sleep-disordered breathing in communitydwelling elderly. Sleep. 1991;14: Bliwise DL. Sleep and aging. In: Pressman MR, Orr WC, eds. Understanding Sleep: The Evaluation and Treatment of Sleep Disorders. Washington, DC: American Psychological Association; 1997, pp Phillips BA, Berry DTR, Schmitt FA, Magan LK, Gerhardsteim DC, Cook YR. Sleep-disordered breathing in the healthy elderly. Chest. 1992;101: Dickel MJ, Mosko SS. Morbidity cut-offs for sleep apnea and periodic leg movements in predicting subjective complaints in seniors. Sleep. 1990;13: OBESITY RESEARCH Vol. 8 No. 9 December