Source: Journal of General Internal Medicine Vol. 22, #6, pp. 818-821. Date: June 2007 URL: http://www.springerlink.com/content/j602073621w413x6/ Feeling Bad in More Ways than One: Comorbidity Patterns of Medically Unexplained and Psychiatric Conditions -------------------------------------------------------------------- Ellen A. Schur, MD(1,5,*), Niloofar Afari, PhD(2), Helena Furberg, PhD(3), Megan Olarte(3), Jack Goldberg, PhD(4), Patrick F. Sullivan, MD, FRANZCP(3), and Dedra Buchwald, MD(1) 1 Department of Medicine, University of Washington School of Medicine, Seattle, WA, USA; 2 Department of Psychiatry, University of California, San Diego & VA San Diego Healthcare System, San Diego, CA, USA; 3 Department of Genetics, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; 4 Department of Epidemiology, University of Washington & Vietnam Era Twin Registry, Seattle, WA, USA; 5 Harborview Medical Center, Box 359780, 325 Ninth Avenue, Seattle, WA 98104, USA. * Corresponding Author: Ellen A. Schur, MD; Harborview Medical Center, Box 359780, 325 Ninth Avenue, Seattle, WA 98104, USA (e-mail: ellschur@u.washington.edu). Received September 5, 2006 Accepted January 17, 2007 Published online February 21, 2007 BACKGROUND Considerable overlap in symptoms and disease comorbidity has been noted among medically unexplained and psychiatric conditions seen in the primary care setting, such as chronic fatigue syndrome, low back pain, irritable bowel syndrome, chronic tension headache, fibromyalgia, temporomandibular joint disorder, major depression, panic attacks, and post-traumatic stress disorder. OBJECTIVE To examine interrelationships among these 9 conditions. DESIGN Using data from a cross-sectional survey, we described associations and used latent class analysis to investigate complex interrelationships. PARTICIPANTS 3,982 twins from the University of Washington Twin Registry. MEASUREMENTS Twins self-reported a doctor's diagnosis of the conditions. RESULTS Comorbidity among these 9 conditions far exceeded chance expectations; 31 of 36 associations were significant. Latent class analysis yielded a 4-class solution. Class I (2% prevalence) had high frequencies of each of the 9 conditions. Class II (8% prevalence) had high proportions of multiple psychiatric diagnoses. Class III (17% prevalence) participants reported high proportions of depression, low back pain, and headache. Participants in class IV (73% prevalence) were generally healthy. Class I participants had the poorest markers of health status. CONCLUSIONS These results support theories suggesting that medically unexplained conditions share a common etiology. Understanding patterns of comorbidity can help clinicians care for challenging patients. KEY WORDS primary care; fibromyalgia; chronic fatigue syndrome; back pain; depression. INTRODUCTION Clinicians have long observed that some patients are burdened with a large number of medical complaints, and the challenge of caring for these patients has been acknowledged.1 A subset of these patients has "medically unexplained" or "functional" conditions like chronic fatigue syndrome, irritable bowel syndrome, and fibromyalgia. Common psychiatric disorders such as major depression and anxiety are also frequently comorbid with these unexplained conditions but do not appear to fully explain them.2 It has been proposed that the spectrum of unexplained conditions may reflect one underlying syndrome.3 We therefore examined the associations among 9 conditions usually considered medically unexplained or psychiatric - chronic fatigue syndrome, low back pain, irritable bowel syndrome, chronic tension headache, fibromyalgia, temporomandibular joint disorder, major depression, panic attacks, and posttraumatic stress disorder - using a community-based twin registry. The goals of this study were to (a) describe the associations among these conditions, (b) use latent class analysis (LCA) to investigate complex interrelationships between these conditions, and (c) describe the demographic and clinical characteristics of individuals falling within each class identified by LCA. METHODS Sample All twins were participants in the University of Washington Twin Registry (UWTR), a community-based registry of twin pairs derived from applications for drivers' licenses in Washington State. The University of Washington receives lists of applicants who are twins, and each member of the pair is invited to join the UWTR and complete a health survey. Based on age and sex data for nonrespondents, registry participants are slightly younger and more likely to be female than the pool of potential twins. All UWTR procedures and the data collection involved in this study were approved by the University of Washington Institutional Review Board. Informed consent was obtained from all twins. Primary Measures Nine conditions usually considered medically unexplained or psychiatric were chosen a priori for analyses. The presence of each condition was coded by the response to the question "Has your doctor ever told you that you have (specific condition)?" We chose this self-report method because diagnostic criteria for the conditions were not agreed upon or validated measures were unavailable or too lengthy for survey administration. Demographics and Zygosity Assignment Demographic characteristics were self-reported. We dichotomized race, marital status, education, and general health status and calculated body mass index (BMI; weight/height2). Twins were asked about childhood similarity. Such questions classify zygosity with an accuracy of approximately 95% of that achieved with biological indicators.4 Statistical Analyses Initial analyses were conducted with SAS v9.1 (SAS Institute Inc., 2004). Odds ratios (OR) and 95% confidence intervals were calculated as measures of association among all of the conditions. For all analyses, data from each individual twin were considered separately. We used generalized estimating equations to account for the nonindependence of twins and clustering of twins within a pair. For these initial analyses, age and sex were chosen as a priori covariates in the adjusted model. Pairwise comparisons were performed only on adjusted ORs. Latent Class Analysis (LCA). Like cluster analyses, LCA5 attempts to find groupings of individuals defined by responses to a number of items. We used LCA because it has yielded convergent results in prior studies of complex human traits6 and is readily compatible with discrete data. We included all 9 conditions along with sex in the LCA. To perform the LCA, we used FORTRAN with an efficient estimation-maximization algorithm7 for maximum likelihood estimation. To determine the number of latent classes that gave rise to the observed data, we fit up to 10 latent class models to the data (50 separate runs with randomized starting values run for each to avoid the known problem of local minima). Number of classes was ascertained by using the Schwarz Bayesian criterion,8 an index of parsimony that penalizes the goodness-of-fit statistic by the number of model parameters times the natural logarithm of the sample size. The Schwarz Bayesian criterion helps to determine the number of classes with a balance of goodness-of-fit and model complexity. Demographic and Clinical Characteristics of Classes. We analyzed variables not entered into the LCA including age; marital status; education; medical, psychiatric, or alternative practitioner visits; total number of visits to a health professional; general health status; BMI; and exercise levels. We used Chi-square tests and linear regression to assess group differences among classes. RESULTS Participants The sample included 1,042 monozygotic pairs, 828 dizygotic pairs, and 121 pairs of undetermined zygosity for a total of 3,982 individual twins. Of these, 3,937 had nonmissing data for all conditions and were included in the LCA. The mean age was 32.4 years (SD=14.7); 86% were white and 61% were female. Associations Among the 9 Conditions Table 1 presents unadjusted and adjusted ORs for all possible combinations of the 9 conditions. Strikingly, 31 of 36 comparisons performed were significant. At an alpha level of 0.05, we would expect approximately 2 associations to be significant by chance. Only irritable bowel syndrome had consistently low ORs. Latent Class Analysis The Schwarz Bayesian criterion declined for 1 to 4 class solutions, had minimal change for 5 classes, and then increased thereafter. Review of the solutions suggested that a 4-class solution best fit the data (see Table 2) and was more interpretable than the 5-class solution, as there was a clear minimum of the Schwarz Bayesian criterion. Class I (2% of sample) had markedly high proportions of individuals who reported all of the 9 syndromes. Class II (8% of sample) participants had high proportions of multiple mental health conditions including major depression, panic attacks, and posttraumatic stress disorder, and low back pain. Class III (17% of sample) participants reported high proportions of depression, low back pain, and headache. Class IV (73% of sample) had the lowest proportion of females and low probabilities for all 9 conditions. There was poor agreement between class assignments for twin pairs (Cohen's kappa=0.1). Demographic and Clinical Characteristics of Classes Demographic and clinical characteristics of each class are presented in Table 2. The results suggest differences across groups, most notably between class I and class IV. Class I individuals most often reported fair or poor general health status and had the highest mean BMI, the highest total number of health care visits, and the lowest exercise frequencies. DISCUSSION We found consistent patterns of comorbidity between medically unexplained conditions in a community sample of twins. Two percent of our sample had high prevalences of all 9 conditions. Although depression and anxiety also commonly co-occurred with medically unexplained conditions, twins with predominantly mental health conditions appeared to be in distinct classes. Twins with a high burden of physical and mental illness were distinguished from healthy participants by their increased health care use, worse self-reported health status, higher BMI, and lower levels of physical exertion. Our findings support proposals that a common pathway9 or a single disease process3 may underlie these clinically defined syndromes. The recent identification of candidate genes associated with chronic fatigue syndrome10 could eventually reveal not only the physiological underpinnings of chronic fatigue, but common pathways for multiple currently unexplained syndromes. Frequencies of medically unexplained syndromes were not markedly increased in the twins with the greatest psychiatric comorbidity and utilization of psychiatric care (class II). In contrast, the primary physical complaint was localized: low back pain. Again, localized symptoms of low back pain and headache predominated over medically unexplained syndromes in sufferers of depression alone (class III). These results concur with literature on both psychiatric comorbidity11 and the association of mental health diagnoses with back pain12 and chronic tension headache.13 The comorbidity of mental illness and localized symptoms may reflect a limited ability to cope with common symptoms like headache or back pain in individuals with depression and anxiety. In sum, we found that twins with predominantly mental illness were distinct from those with the highest burden of medically unexplained syndromes, echoing prior findings.14 For clinicians, these data reiterate the importance of screening for psychiatric conditions when painful or medically unexplained conditions are present. This study was cross-sectional, thus limiting any conclusions about causality for the observed associations. In addition, the clinical conditions were assessed by single-item self-report of a doctor's diagnosis, which can be subject to response bias. However, our overall prevalences are comparable to available published general population rates for chronic fatigue syndrome,15 low back pain,16 fibromyalgia,17 major depression,11 panic, 18 and posttraumatic stress disorder.19 Our use of self-report data may explain the lower-than-expected overall prevalence of irritable bowel syndrome, as research suggests that a large undiagnosed symptomatic population may exist.20 In summary, this study is unique in its use of a community-based sample of twins and diverse statistical approaches to examine patterns of comorbidity for 9 conditions seen in primary care. We defined a group with multiple medically unexplained syndromes who appeared to be distinct from individuals with primarily mental health diagnoses. Clinicians may find our results helpful for several reasons. First, they suggest individuals should be screened for comorbid disorders such as posttraumatic stress disorder when depression and anxiety are present and for major depression when low back pain and chronic headache are reported. Second, our findings also help clinicians understand that distressed patients are not merely accumulating diagnoses or somatic complaints, but have a constellation of conditions that frequently coexist, analogous to metabolic syndrome. Multiple medications for symptom management alone can be minimized, focusing instead on proven behavioral strategies such as graduated exercise or cognitive behavioral therapy. Finally, research seeking unified etiologies and treatment strategies is needed. Such ground-breaking work may eventually allow clinicians to feel confidence instead of consternation when treating patients with multiple unexplained, comorbid conditions. Acknowledgements This research was supported by National Institutes of Health awards 5 U19 AI038429 (Buchwald) and R55AR051524 (Afari). Dr. Schur is funded by a National Institutes of Health Career Development Award K23 DK070826. Conflict of Interest None disclosed. TABLES Table 1. [No electronic version available] Table 2. Latent Class Analysis Results for Sex and 9 Clinical Conditions with Demographic and Clinical Characteristics by Class ----------------------------------------------------------------------------------------------------------------------- Item Overall Class I Class II Class III Class IV P sample (2%) (8%) (17%) (73%) value* (N=3,937) multiplex depression depression unaffected and anxiety ----------------------------------------------------------------------------------------------------------------------- Variables in latent class analysis Female sex % 61 89 77 86 53 - Low back pain % 27 91 47 47 19 - Major depression % 20 85 97 53 2 - Panic attacks % 12 55 92 9 3 - Chronic tension headache % 8 49 20 30 1 - Irritable bowel syndrome % 6 63 16 18 0 - TMJ syndrome % 5 45 10 18 1 - Posttraumatic stress disorder % 4 45 31 0 0 - Chronic fatigue syndrome % 2 61 4 6 0 - Fibromyalgia % 2 61 0 4 0 - Demographic and clinical characteristics Married or living with partner % 40 51 42 50 37 <0.001 College education or more % 57 63 56 62 56 <0.010 Health fair or poor % 6 40 17 11 3 <0.001 Age in years, mean (SD) 32.3 (14.7) 35.7 (14.9) 42.4 (12.8) 34.9 (14.6) 31.1 (14.6) <0.001 Psychiatric visits in past 3 months, 0.30 (1.9) 0.81 (2.1) 1.4 (3.8) 0.51 (2.2) 0.11 (1.4) <0.001 mean (SD) and median 0 0 0 0 0 Medical visits in past 3 months, 1.3 (3.0) 2.9 (4.9) 2.6 (6.9) 1.7 (2.6) 1.0 (2.2) <0.001 mean (SD) and median 1 1 1 1 0 Alternative practitioner visits in past 0.68 (3.7) 2.3 (5.5) 1.6 (7.6) 0.81 (2.7) 0.5 (3.1) <0.001 3 months, mean (SD) and median 0 0 0 0 0 Total health visits in past 3 months, 2.4 (5.5) 6.0 (8.2) 5.6 (11.2) 3.1 (4.6) 1.6 (4.5) <0.001 mean (SD) and median 1 3 2 1 1 Body mass index, mean kg/m2 (SD) 24.7 (5.1) 27.5 (6.9) 25.6 (6.4) 25.4 (5.7) 24.4 (4.7) <0.001 Episodes of vigorous exercise per 1.4 (2.7) 0.77 (1.6) 1.1 (2.0) 1.0 (2.0) 1.6 (2.9) 0.115 week, mean (SD) Episodes of moderate exercise per 2.6 (4.1) 2.2 (3.8) 2.9 (7.1) 2.2 (2.6) 2.7 (3.9) <0.001 week, mean (SD) ----------------------------------------------------------------------------------------------------------------------- To aid the reader, arrows indicate probabilities that deviate by 15% from the overall prevalence * P values are for differences across all 4 classes TMJ = temporomandibular joint REFERENCES 1. Short D. Difficult patients. Br J Hosp Med. 1994;51:128-30. 2. Aaron LA, Herrell R, Ashton S, et al. Comorbid clinical conditions in chronic fatigue: a co-twin control study. J Gen Intern Med. 2001;16: 24-31. 3. Wessely S, Nimnuan C, Sharpe M. Functional somatic syndromes: one or many? Lancet. 1999;354:936-9. 4. Torgersen S. The determination of twin zygosity by means of a mailed questionnaire. Acta Genet Med Gemellol (Roma). 1979;28:225-36. 5. McCutcheon AL. Latent Class Analysis. Beverly Hills, CA: Sage Publications; 1987. 6. Sullivan PF, Kessler RC, Kendler KS. Latent class analysis of lifetime depressive symptoms in the national comorbidity survey. Am J Psychi- atry. 1998;155:1398-406. 7. Dempster AP, Laird NM, Rubin DB. Maximum likelihood from incom- plete data via the EM algorithm. J R Stat Soc. 1977;39:1-38. 8. Williams LJ, Holahan PJ. Parsimony-based fit indices for multiple- indicator models: do they work? Struct Equation Model. 1994;1:161-89. 9. Clauw DJ, Schmidt M, Radulovic D, Singer A, Katz P, Bresette J. The relationship between fibromyalgia and interstitial cystitis. J Psychiatr Res. 1997;31:125-31. 10. Goertzel BN, Pennachin C, de Souza Coelho L, Gurbaxani B, Maloney EM, Jones JF. Combinations of single nucleotide polymorphisms in neuroendocrine effector and receptor genes predict chronic fatigue syndrome. Pharmacogenomics. 2006;7:475-83. 11. Kessler RC, Berglund P, Demler O, et al. The epidemiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R). JAMA. 2003;289:3095-105. 12. Linton SJ. A review of psychological risk factors in back and neck pain. Spine. 2000;25:1148­56. 13. Puca F, Genco S, Prudenzano MP, et al. Psychiatric comorbidity and psychosocial stress in patients with tension-type headache from head- ache centers in Italy. The Italian Collaborative Group for the Study of Psychopathological Factors in Primary Headaches. Cephalalgia. 1999;19:159-64. 14. Henningsen P, Zimmermann T, Sattel H. Medically unexplained physical symptoms, anxiety, and depression: a meta-analytic review. Psychosom Med. 2003;65:528-33. 15. Afari N, Buchwald D. Chronic fatigue syndrome: a review. Am J Psychiatry. 2003;160:221-36. 16. Andersson GB. Epidemiology of low back pain. Acta Orthop Scand Suppl. 1998;281:28-31. 17. Wolfe F, Ross K, Anderson J, Russell IJ, Hebert L. The prevalence and characteristics of fibromyalgia in the general population. Arthritis Rheum. 1995;38:19-28. 18. Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry. 1994;51:8-19. 19. Kessler RC, Sonnega A, Bromet E, Hughes M, Nelson CB. Posttrau- matic stress disorder in the National Comorbidity Survey. Arch Gen Psychiatry. 1995;52:1048-60. 20. Hahn BA, Saunders WB, Maier WC. Differences between individuals with self-reported irritable bowel syndrome (IBS) and IBS-like symptoms. Dig Dis Sci. 1997;42:2585-90. -------- (c) 2007 Spinger (c) 2007 Society of General Internal Medicine