Cognitive Behavior Therapy for Hypochondriasis a Randomized Controlled Trial

ORIGINAL CONTRIBUTION

Cognitive Behavior Therapy for Hypochondriasis A Randomized Controlled Trial

Arthur J. Barsky, MD Context Hypochondriasis is a chronic, distressing, and disabling condition that is preva- David K. Ahern, PhD lent in ambulatory medical practice. Until recently, no specific treatment has been clearly demonstrated to be effective. YPOCHONDRIASIS IS DEFINED Objective To assess the efficacy of a cognitive behavior therapy (CBT) for hypo- as a persistent fear or belief chondriasis. that one has a serious, un- Design A randomized, usual care control group design, conducted between Sep- diagnosed medical illness. tember 1997 and November 2001. The individual primary care physician was the HOccurring in as many as 5% of medi- unit of randomization, and all patients clustered within each physician’s practice cal outpatients, it is a prevalent, dis- were assigned to the experimental treatment (individual CBT and a consultation let- abling, and chronic condition that has ter to the primary care physician) or to the control condition. Subjects were generally been refractory to psycho- assessed immediately before and 6 and 12 months after the completion of treat- logical and pharmacological treatment. ment and costly for the health care sys- Setting and Participants Participants were 80 patients from primary care prac- tem.1-6 Until recently, there was no tices and 107 volunteers responding to public announcements, all of whom exceeded empirically validated treatment for the a predetermined cutoff score on a hypochondriasis self-report questionnaire on 2 suc- hypochondriacal patient’s somatic cessive occasions. symptoms, belief in the presence of an Intervention A scripted, 6-session, individual CBT intervention was compared with undiagnosed disease, health-related medical care as usual. The CBT was accompanied by a consultation letter sent to the anxiety, and bodily preoccupation. Al- patient’s primary care physician. though limited by small samples, lack Main Outcome Measures Hypochondriacal beliefs, fears, attitudes, and somatic of randomization and control groups, symptoms; role function and impairment. the absence of long-term follow-up, lim- Results A total of 102 individuals were assigned to CBT and 85 were assigned to ited generalizability, and the absence of medical care as usual. The sociodemographic and clinical characteristics of the 2 groups validated outcome measures,7-12 the ex- were similar at baseline. Using an intent-to-treat analytic strategy, a consistent pat- isting treatment literature suggests a tern of statistically and clinically significant treatment effects was found at both 6- and role for a variety of psychosocial inter- 12-month follow-up, adjusting for baseline covariates that included educational level, ventions, although none has been con- generalized psychiatric distress, and participant status (patient vs volunteer). At 12- month follow-up, CBT patients had significantly lower levels of hypochondriacal symp- clusively validated. In the only large- toms, beliefs, and attitudes (PϽ.001) and health-related anxiety (P=.009). They also scale, rigorous, randomized controlled had significantly less impairment of social role functioning (P=.05) and intermediate trial to date, cognitive therapy and be- activities of daily living (PϽ.001). Hypochondriacal somatic symptoms were not im- havioral stress management were both proved significantly by treatment. more effective than a waiting list con- Conclusion This brief, individual CBT intervention, developed specifically to alter hy- 13 trol condition. pochondriacal thinking and restructure hypochondriacal beliefs, appears to have sig- We have proposed that hypochon- nificant beneficial long-term effects on the symptoms of hypochondriasis. driasis can be understood as a self- JAMA. 2004;291:1464-1470 www.jama.com perpetuating and self-validating disorder of cognition and bodily per- 14-16 Author Affiliations: Department of Psychiatry, Brigham Corresponding Author: Arthur J. Barsky, MD, Depart- ception. In this formulation, per- and Women’s Hospital and Harvard Medical School, ment of Psychiatry, Brigham and Women’s Hospital, sonally threatening life events prompt Boston, Mass. 75 Francis St, Boston, MA 02115 ([email protected]).

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

predisposed individuals to suspect that The precise cutoff score of 150 (score cantly in the next 12 months; somato- they have become ill. This suspicion range, 52-258) was determined from form pain disorder; psychosis or sui- leads them to selectively attend to be- previous studies3,17-19 and was chosen cide risk; and ongoing, symptom- nign bodily sensations and health in- because we believed that many sub- contingent, disability determinations, formation that confirm their suspi- threshold individuals were sufficiently workers’ compensation proceedings, or cion and to ignore disconfirmatory hypochondriacal to be in need of treat- litigation. evidence. Benign bodily sensations are ment and would be able to benefit from thereby amplified and misattributed to it. Signed informed consent was ob- Treatment Conditions the putative disease, which further sub- tained in accordance with the Brigham and Therapists stantiates their disease convictions. We and Women’s Hospital Human Re- Cognitive behavior therapy was admin- have developed a cognitive behavior search Committee requirements. Sub- istered individually in six 90-minute ses- therapy (CBT) that specifically targets jects meeting initial eligibility criteria and sions at weekly intervals. Each session the cognitive and behavioral amplifi- exceeding the screening cutoff com- was tightly scripted (manual available ers of benign bodily symptoms that pleted the baseline research battery, from the authors) and devoted to 1 of 5 propel this hypochondriacal cycle of which included a second administra- factors that cause patients to amplify so- disease conviction and symptom am- tion of the screening questionnaire and matic symptoms and misattribute them plification. Patients are helped to cor- a determination of their medical mor- to serious disease: attention and bodily rect faulty symptom attributions, re- bidity. hypervigilance, beliefs about symptom structure beliefs and expectations about Physicians were randomized imme- etiology, circumstances and context, ill- health and disease, correct misunder- diately following the baseline re- ness and sick role behaviors, and mood. standings about the medical care pro- search interview by a staff member not Each session consisted of educational in- cess, modify maladaptive illness behav- connected to the research, using a ran- formation about the symptom amplifi- iors, and learn techniques of selective dom numbers table. Randomization was ers, an illustrative exercise, and a dis- attention and distraction. stratified by physician seniority (house cussion to personalize the material We hypothesized that the CBT in- staff or attending staff) and by prac- presented. The 3 study therapists had tervention would alleviate the symp- tice volume (part-time or full-time for master’s or doctoral degrees and prior toms of hypochondriasis more effec- attending staff, primary care program CBT experience. Treatment sessions tively than medical care as usual. or traditional internal medicine pro- were held in offices in the department gram for house staff). All research data of psychiatry. METHODS were collected by research assistants Because it was considered essential Study Design who were blind to the patient’s treat- that the patients’ ongoing medical man- A randomized, usual care control group ment status, and the study therapists agement be coordinated with the CBT design was used. To avoid a “contami- and investigators had no foreknowl- to alter their understanding of symp- nation” effect, the individual primary edge of treatment assignment. The toms, disease, and medical care, a stan- care physician was the unit of random- therapists had no role in data collec- dardized consultation letter was sent to ization, and all patients clustered within tion. Subjects were compensated for each patient’s primary care physician. each physician’s practice were as- each research interview, but not for un- This letter (available from the authors) signed to the experimental treatment dergoing treatment. contained the following 5 practical sug- (individual CBT and a consultation let- gestions for medical management that ter to the primary care physician) or to Subjects and Settings were designed to augment the indi- the control condition. Subjects were Subjects were accrued from 2 sources: vidual therapy: (1) make improved cop- assessed immediately before and 6 (1) successive patients attending ing with somatic symptoms rather than and 12 months after the completion of primary care practices in 2 large, aca- symptom elimination the goal of medi- treatment. demic medical centers, and (2) volun- cal management; (2) uncouple access to Subjects were recruited by screening teers responding to public announce- the physician from symptom status by with a self-report hypochondriasis ques- ments of a treatment study for “health scheduling regular appointments; (3) tionnaire (composed of the Whiteley anxiety and hypochondriasis.” provide only limited reassurance; (4) ex- Index and Somatic Symptom Inventory), The inclusion criteria were age older plain the patient’s symptoms using the using a predetermined cutoff score than 18 years, English fluency and lit- model of cognitive and perceptual symp- selected to identify a sample, half of eracy, having seen a primary care phy- tom amplification; and (5) be conser- whom met Diagnostic and Statistical sician in the last 12 months, and ex- vative in medical diagnosis and treat- Manual of Mental Disorders, Fourth Edi- ceeding the hypochondriasis screening ment, within the bounds of appropriate tion (DSM-IV) criteria for hypochon- cutoff score on both occasions. Exclu- medical practice. driasis and half of whom had subthresh- sion criteria included major medical Treatment fidelity was assessed by au- old (subdefinitional) hypochondriasis. morbidity expected to worsen signifi- diting audiotapes of randomly se-

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

lected therapy sessions from all 3 thera- ing (eg, doing errands, working both individual and sets of outcome pists; adherence to the CBT manual was around the house) and social activi- measures as a function of treatment ef- excellent. Receipt of the consultation ties (eg, seeing friends, participating fects (between subjects), time of as- letter was acknowledged by 96.8% of in community activities). sessment (within subjects), and covar- the primary care physicians. Covariates. Psychiatric comorbid- iates (educational level, generalized ity was assessed with the Hopkins psychiatric distress, and participant sta- Assessment Symptom Checklist-90.36,37 This widely tus [patient vs volunteer]). Hence, the Outcome Variables. The primary out- used, 90-item instrument has excel- interaction effect of treatmentϫtime come measure was the Whitely Index, lent psychometric properties and pro- was of most interest. In cases where the a widely used self-report question- vides an overall measure of general- overall MANOVA was significant, or a naire of hypochondriacal attitudes and ized psychiatric distress.37 Participant priori contrasts were planned, only the beliefs, whose validity, reliability, and status indicated whether the subject was ANCOVA results are presented, using sensitivity to change have been dem- a patient accrued in the clinic or a vol- the Greenhouse-Geisser adjustment. In- onstrated.20,21 Secondary outcomes unteer recruited from the community. dividual, planned contrasts were cal- included health-related anxiety, as- Aggregate medical morbidity was as- culated for each outcome measure com- sessed with the Health Anxiety Inven- sessed in 2 ways: the Duke Severity of paring baseline and 6-month, and tory, a 14-item, self-report question- Illness Scale38,39 was used for patients baseline and 12-month assessments, re- naire that is minimally influenced by accrued from primary care practices. spectively. Effect sizes and threshold the presence of major medical illness This structured audit of the medical rec- values for clinical significance were also and has good validity, internal consis- ord was conducted by a research phy- derived for each analysis. All tests of sta- tency, and reliability.22 The frequency sician who was blind to treatment statistical significance were interpreted of hypochondriacal thoughts was as- tus. For study volunteers, their primary with a criterion of PϽ.05. Statistical sessed with the Hypochondriacal Cog- care physicians provided 2 ordinal rat- analyses were performed using SPSS, re- nitions Questionnaire, requiring the pa- ings of aggregate medical morbidity. lease 12.0.1 for Windows (SPSS Inc, tient to rate how often each of 18 Chicago, Ill). disease-related thoughts occurs. Hy- Data Analysis pochondriacal somatic symptoms were An intent-to-treat approach was used RESULTS assessed with the Somatic Symptom In- in all analyses. The last-observation- Subject Accrual ventory.17,23 The reliability, internal con- carried-forward approach was used to The study was conducted between Sep- sistency, and convergent validity of this impute scores for missing data values. tember 1997 and November 2001. A 13-item questionnaire have been dem- Although the physician was the unit of total of 6307 individuals completed the onstrated previously.17,18,23-26 A clini- randomization, only 11 physicians had screening questionnaire, of whom 776 cal diagnosis of DSM-IV hypochondria- more than 1 subject. Thus, any ap- (12.3%) exceeded the cutoff score sis was made by trained research proach to estimating clustering as an ad- (FIGURE). Two hundred nineteen in- assistants using the Structured Diag- justment for physician effects would dividuals declined to participate, 156 nostic Interview for Hypochondriasis, likely lead to unstable estimates of the proved to be ineligible, and 214 could a highly structured interview shown in within-cluster correlation. Therefore, to not be reached subsequently, result- previous work to have concurrent, con- address the potential effect of clustering in a total of 187 subjects (30.2% of vergent, and discriminant validity.27-29 ing on the results, we selected 1 sub- those eligible) who participated. A ran- The tendency to amplify benign ject at random from each of these 11 dom sample (n=191) of the 589 non- bodily sensation and experience it as physicians for analysis, resulting in the participants was compared with those noxious, unpleasant, and alarming was exclusion of 24 subjects (n=163). These who participated. The nonpartici- assessed with the Somatosensory Am- results did not differ in any respect from pants were older (46.4 vs 42.3 years, plification Scale. This questionnaire has those obtained with the full sample t376=2.96, P = .003), had less educa- good reliability and validity and is sen- (N=187), which was therefore used for tion (19.1% completed college vs ␹2 Ͻ sitive to change as a result of attention all subsequent analyses. The 2 groups 29.4%, 6=42.6, P .001), and were training.14,30-33 were compared 6 and 12 months after more likely to be male (38.9% vs 23.5%, ␹2 Ͻ Role impairment and functional treatment on the primary and second- 1=10.5, P .001). status were determined with the ary outcome variables. General linear Of the 102 patients in the treatment Functional Status Questionnaire.34 modeling was used as the primary ana- arm, 63 (61.8%) attended all 6 ses- This valid and reliable self-report lytic approach, in which a series of uni- sions, 13 (12.7%) attended 4 or 5 ses- questionnaire was developed for use variate and multivariate repeated mea- sions, 12 (11.8%) attended 1 to 3 in ambulatory medical popula- sures analysis of covariance (ANCOVA sessions, and 14 (13.7%) attended none. tions.34,35 It includes subscales assess- and MANOVA) models were used. This Six-month follow-up was obtained for ing intermediate activities of daily liv- approach permitted the modeling of 85 (83.3%) of the 102 treatment pa-

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

tients and for 76 (89.4%) of the 85 con- (F1,181=5.04, P =.03), but not at 6 ment effect size for the Whiteley Index trol patients (87% in person and 13% months (F1,181=2.37, P=.13). Conse- was r=0.31 at 6 months and r=0.27 at by telephone). Twelve-month fol- quently, participant status (patient or 12 months. low-up was obtained for 92 (90.2%) of volunteer) also was included as a co- The secondary outcome measures of treatment patients and 78 (91.8%) of variate in the models. hypochondriacal symptoms were ana- control patients (84.7% in person and lyzed in the same manner as the White- 15.3% by telephone). These attrition Treatment Outcomes ley Index. For hypochondriacal thought rates do not differ significantly be- The results for the Whiteley Index, the frequency, health anxiety, and somato- tween groups. Six-month follow-up was primary outcome measure, are shown sensory amplification, the interac- obtained for 92% of treatment compl- in TABLE 2, using repeated measures tions between group and assessment eters (those attending Ն4 sessions) and ANCOVA. There was a statistically sig- were statistically significant, indicat- 57.6% of treatment dropouts. Twelve- nificant interaction effect for group ing a significant treatment effect on month follow-up was obtained for (treatment vs control) by assessment these measures (see Table 2). Post- 94.7% of treatment completers and time (baseline, 6-month, and 12- hoc tests of within-subjects contrasts for 76.9% of treatment dropouts. month follow-up). Post-hoc tests of hypochondriacal thought frequency re- within-subjects contrasts disclosed a vealed a significantly greater improve- Treatment Groups at Baseline statistically significant improvement in ment in the treatment vs control group The sociodemographic characteristics the treatment vs control group at both at 6-month (F1,182=6.97, P=.009), and Ͻ of the 2 treatment groups did not dif- 6-month (F1,182=20.1, P .001) and 12- at 12-month follow-up (F1,182=5.64, fer significantly (TABLE 1). They were month follow-up, compared with base- P=.02). For health anxiety, post-hoc Ͻ predominantly women, middle-aged, line (F1,182=14.2, P .001). The treat- tests of within-subjects contrasts re- and reported a history of hypochondriasis for approximately 11 years. Edu- cational level and generalized psychi- Figure. Profile of Randomized Controlled Trial atric distress did not differ significantly in the 2 groups at baseline but were 6307 Participants Screened for Eligibility used as covariates in the analyses because of their established relation- 776 Participants Had Hypochondriasis Screening Score >150 (Cutoff) ships to the outcome variables of interest. The treatment and control 589 Participants Excluded groups did not differ significantly in ag- 156 Did Not Meet Inclusion Criteria gregate medical morbidity at baseline. 219 Declined Participation Eighty patients were recruited from 214 Unreachable or Other primary care practices and 107 were 187 Participants Enrolled volunteers. At baseline, volunteers were 80 Patients significantly more symptomatic and 107 Volunteers more disabled than patients on the

major outcome measures of hypochon- 163 Primary Care Physicians Randomized driacal symptoms (PϽ.001), health- related anxiety (PϽ.001), and interme- 87 Primary Care Physicians Randomized to 76 Primary Care Physicians Randomized to diate activities of daily living (PϽ.001). Experimental Treatment (102 Participants Usual Medical Care (85 Participants Repeated measures of ANCOVA were [43 Patients; 59 Volunteers]) [37 Patients; 48 Volunteers]) 1 Physician with 7 Participants 2 Physicians with 4 Participants performed on the Whiteley Index mod- 1 Physician with 4 Participants 1 Physician with 3 Participants eled as a function of participant status 1 Physician with 3 Participants 1 Physician with 2 Participants (patients vs volunteers), treatment 4 Physicians with 2 Participants 72 Physicians with 1 Participant 80 Physician with 1 Participant (CBT vs usual care) and assessment point (baseline, 6 months, and 12 85 Participants Completed 6-mo Outcome 76 Participants Completed 6-mo Outcome months), and all 2- and 3-way interac- Assessment Assessment tion terms. The 3-way interaction was

significant (F2,362=3.38, P=.04), indi- 92 Participants Completed 12-mo Outcome 78 Participants Completed 12-mo Outcome cating that volunteers had higher Assessment Assessment Whiteley Index scores at baseline and experienced a larger treatment effect 102 Participants Included in Primary Analysis 85 Participants Included in Primary Analysis 43 Patients 37 Patients (difference between CBT and usual 59 Volunteers 48 Volunteers care) at 12 months than the patients

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

vealed a significantly greater improve- plification, individual tests of within- (F1,182=7.70, P=.006). Hypochondria- ment in the treatment vs control group subjects contrasts revealed a signifi- cal somatic symptoms were not signifi- at both 6 months (F1,182=6.73, P=.01) cantly greater improvement in the cantly improved by treatment. and 12 months compared with treatment vs control group at 6 months The 2 treatment groups were com- baseline (F1,182=7.01, P=.009). For am- (F1,182=8.47, P=.004) and at 12 months pared on measures of functional status using MANCOVA as the omnibus test, with intermediate and social ac- Table 1. Sociodemographic Characteristics tivities combined as dependent mea- Treatment Group, No. (%) Control Group, No. (%) (n = 102 [59 Volunteers, (n = 85 [48 Volunteers, P sures, again including baseline educa- 43 Patients]) 37 Patients]) Value tional level, psychiatric comorbidity, Age, mean (SD), y 40.66 (12.60) 44.29 (13.75) .06 and participant status as covariates. Women 76 (74.5) 67 (78.8) .49 TABLE 3 indicates a statistically signifi- Ethnicity cant interaction effect for group by as- White 72 (70.6) 63 (74.1) sessment time for intermediate activi- Black 17 (16.7) 11 (12.9) .23 ties (F =12.32, PϽ.001), but no Hispanic 2 (2.0) 7 (8.2) 2,364 statistical significance for social activi- Other 11 (10.7) 4 (4.8) ties (F =2.29, P=.10). Post-hoc tests Education 2,364 Graduate/professional 29 (28.4) 17 (20.0) of within-subjects contrasts for inter- College graduate 34 (33.3) 21 (24.7) mediate activities of daily living re- Some college 21 (20.7) 23 (27.0) .12 vealed that there was a statistically sig- High school graduate 11 (10.8) 19 (22.4) nificant improvement in the treatment Grades 7-11 7 (6.8) 5 (5.9) vs control group at both 6 months Ͻ Employed 70 (68.6) 53 (62.4) .37 (F1,182=18.44, P .001) and at 12 Ͻ DSM-IV diagnosis 67 (65.7) 48 (56.6) .20 months (F1,182=18.30, P .001). The of hypochondriasis treatment effect size for intermediate ac- Age of onset, mean (SD), y 30.54 (14.00) 32.65 (15.79) .34 tivities at 12 months was r=0.30. For Abbreviation: DSM-IV indicates Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. social activities, post-hoc tests of within- subjects contrasts revealed that there Table 2. Hypochondriacal Symptoms (N = 187) was a statistically significant improve- Treatment Group, Control Group, P ment in the treatment vs control group Mean (SE [95% CI])* Mean (SE [95% CI])* Value† at 12 months (F1,182=3.74, P=.05), but Whiteley Index (score range, 1-5)‡ no significance at 6 months (F1,182=3.06, Baseline 3.58 (0.054 [3.47-3.68]) 3.51 (0.060 [3.38-3.62]) P=.07). The treatment effect size for so- 6-mo follow-up 2.82 (0.075 [2.68-2.97]) 3.21 (0.083 [3.05-3.38]) Ͻ.001 cial activities at 12 months was r=0.23. 12-mo follow-up 2.65 (0.084 [2.48-2.81]) 30.02 (0.093 [2.85-3.21]) No therapist effect was found, ie, Hypochondriacal thought frequency (score range, 1-5)‡ when the patients treated by each of Baseline 2.09 (0.066 [1.98-2.21]) 2.29 (0.064 [2.16-2.41]) the therapists were compared, no 6-mo follow-up 1.75 (0.057 [1.63-1.86]) 2.14 (0.063 [2.01-2.26]) .008 significant differences were found for 12-mo follow-up 1.62 (0.056 [1.51-1.73]) 2.02 (0.062 [1.90-2.14]) any of the outcome measures. Health anxiety (score range, 1-4)‡ Although the intervention did not Baseline 2.68 (0.047 [2.58-2.77]) 2.71 (0.051 [2.61-2.81]) include treatment for comorbid psy- 6-mo follow-up 2.29 (0.047 [2.20-2.38]) 2.51 (0.051 [2.41-2.61]) .004 chiatric disorder, subjects were free 12-mo follow-up 2.20 (0.051 [2.10-2.30]) 2.44 (0.056 [2.33-2.55]) to obtain such treatment as they or Somatosensory amplification (score range, 1-5)‡ their physicians saw fit. At 6-month Baseline 3.25 (0.067 [3.12-3.38]) 3.04 (0.073 [2.90-3.19]) follow-up, 20 CBT patients (19.6%) 6-mo follow-up 2.92 (0.068 [2.79-3.06]) 2.96 (0.074 [2.81-3.11]) .003 and 19 control patients (22.4%) had 12-mo follow-up 2.82 (0.070 [2.68-2.96]) 2.87 (0.077 [2.72-3.03]) initiated a psychotropic medication Hypochondriacal somatic or care with a mental health profes- symptoms (score range, 1-5)‡ Baseline 2.73 (0.048 [2.63-2.82]) 2.81 (0.053 [2.70-2.91]) sional since inception. At 12 months, 6-mo follow-up 2.16 (0.056 [2.05-2.27]) 2.42 (0.061 [2.30-2.55]) .08 6 CBT patients (5.9%) and 5 control 12-mo follow-up 2.00 (0.056 [1.88-2.11]) 2.24 (0.061 [2.18-2.36]) patients (5.9%) had initiated a psy- Abbreviation: CI, confidence interval. chotropic medication or mental *Adjusted for baseline educational level, psychiatric comorbidity, and participant status. †P value based on Greenhouse-Geisser adjustment for the interaction between group and assessment from analysis health care since their 6-month of variance (ANOVA) model covarying for educational level, psychiatric comorbidity, and participant status. follow-up. These rates do not differ ‡Higher scores indicate more severe symptoms. significantly between groups.

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

COMMENT Table 3. Functional Status and Role Impairment (N = 187) This 6-session CBT, specifically target- Treatment Group, Control Group, P ing the cognitive and perceptual mecha- Mean (SE [95% CI])* Mean (SE [95% CI])* Value† nisms thought to underlie hypochon- Intermediate activities of driasis, appears to significantly improve daily living (0-100)‡ a range of hypochondriacal symp- Baseline 74.92 (2.01 [70.89-78.96]) 74.99 (2.25 [70.57-79.43]) Ͻ toms, beliefs, and attitudes. These ef- 6-mo follow-up 82.54 (2.11 [78.39-86.71]) 70.61 (2.31 [66.05-75.18]) .001 12-mo follow-up 83.42 (2.03 [79.43-87.42]) 72.23 (2.22 [67.85-76.62]) fects are evident at 6-month follow-up Social activities (0-100)‡ and persist at 12 months. The effects Baseline 81.41 (2.35 [76.78-86.05]) 81.61 (2.58 [76.52-86.69]) on role functioning are not consis- 6-mo follow-up 86.02 (2.20 [81.69-90.35]) 80.55 (2.41 [75.80-85.30]) .10 tently significant at 6 months but 12-mo follow-up 86.11 (2.34 [81.50-90.72]) 79.26 (2.56 [74.21-84.32]) emerge at 12 months, the primary end *Adjusted for baseline educational level, psychiatric comorbidity, and participant status. †P value based on Greenhouse-Geisser adjustment for the interaction between group and assessment from analysis point. These treatment effects are seen of variance (ANOVA) model covarying for educational level, psychiatric comorbidity, and participant status. using an intent-to-treat analysis, after ‡0-100 (100 = maximal function). adjusting for psychiatric comorbidity, sociodemographic characteristics, and is not unusual for such studies how- Hypochondriacal attitudes and con- participant status (patient vs volun- ever; Kashner et al,40 for example, found cerns improved more than somatic teer) at baseline. The findings are com- that 56% of somatization disorder pa- symptoms did. This finding, although patible with the only other major trial tients randomized to group therapy it might seem counterintuitive, was ac- reported to date and expand on it by failed to attend a single session. tually expected: the treatment was in- having a control group available for Second, we lacked an “attention” con- tended to improve coping with symp- comparison at long-term follow-up.13 trol, ie, a generic psychosocial interven- toms rather than curing them outright Though the magnitude of the treat- tion providing nonspecific attention, sup- (“care rather than cure”). This had both ment effect is modest, it is important port, concern, and positive expectation. an empirical and a conceptual basis. to remember that hypochondriasis gen- This limits our ability to attribute the Empirically, clinical experience and in- erally has been considered a refrac- treatment effect to the specific cogni- tervention trials for a variety of func- tory and chronic disorder (the mean du- tive and behavioral strategies of the in- tional somatic syndromes suggest that ration of illness was 11 years in this tervention. However, the fact that the the patients who do best are those who study) for which there has been no em- cognitive processes thought to underlie learn to compensate for, rather than at- pirically validated treatment. In addi- the disorder (eg, hypochondriacal cog- tempting to eliminate, their somatic dis- tion, this CBT was brief (only 6 ses- nitions, health beliefs, amplification) im- tress. Conceptually, hypochondriacal sions) and included no follow-up proved with treatment suggests that the somatic symptoms cannot simply be “booster” sessions. Finally, patients treatment had a specific effect. stripped away with symptomatic treat- were not treated in the study for co- Third, considerable improvement oc- ment because they exist for underly- morbid psychiatric disorder, and the curred in the control group. This was ing psychological and interpersonal rea- continued presence of these disorders likely due to the inadvertent inclusion sons. This suggests that a realistic goal likely moderated the treatment effect. of patients with transient hypochon- in treating hypochondriasis is amelio- The study has several limitations. driasis, probably because the 2 screen- ration of distressing fears and beliefs and First, many eligible patients did not paring measures were too close to each improved coping, rather than the elimi- ticipate, limiting the generalizability of other in time (approximately 3 weeks). nation of somatic symptoms per se. the findings. Those who did consent to Additionally, regression to the mean We are unable to partial out the vari- participate might have been more re- and the supportive effect of being en- ance in treatment effect between the ceptive to a psychosocial approach and rolled in a longitudinal study contrib- CBT and the physician consultation let- hence benefited more from it than those uted to the high rate of spontaneous im- ter. That the latter may have been ben- who did not consent. On the other hand, provement. A Hawthorne effect may eficial is suggested by 2 studies with so- the study participants might be less se- also have occurred whereby control matization disorder patients in which verely hypochondriacal and more re- physicians, having learned of the study, a psychiatric consultation letter alone sponsive to their medical physician’s made a greater effort to help their hy- resulted in lower health care costs along ministrations and/or more likely to im- pochondriacal patients. with either improved or stable physi- prove spontaneously, which would tend Finally, study subjects came from 2 cal functioning.41-43 This points to the to reduce the treatment effect. In addi- different sources. Participant status, how- critical importance of seamlessly inte- tion, 25% of the subjects attended less ever, was included as a covariate in all grating the psychosocial care of these than 4 treatment sessions, suggesting analyses, and the fact that these 2 groups patients with their medical care. that future efforts must be directed to- differed at baseline confers some mea- The treatment offered in this study ward reducing treatment dropout. This sure of generalizability on the findings. was not attractive to many hypochon-

Downloaded From: https://jamanetwork.com/ on 09/25/2021 COGNITIVE BEHAVIOR THERAPY FOR HYPOCHONDRIASIS

driacal patients, and only 30% of those treatment must be made more attrac- Analysis and interpretation of data: Barsky, Ahern. Drafting of the manuscript: Barsky, Ahern. eligible entered the trial. Hypochon- tive in the future by seamlessly inte- Critical revision of the manuscript for important in- driacal individuals are by definition grating it into the primary care pro- tellectual content: Barsky, Ahern. Statistical expertise: Ahern. convinced of the medical nature of their cess and conducting it in the medical Obtained funding: Barsky, Ahern. condition and therefore psychosocial setting (as our treatment was not). The Administrative, technical, or material support: Barsky. treatment seems nonsensical to them. treatment effect could also be strength- Supervision: Barsky. Funding/Support: This investigation was supported Although a major problem, this should ened by increasing the number of by research grant MH40487 from the National Insti- not detract from the fact that those pa- sessions to 8 and by adding booster tute of Mental Health. Role of the Sponsor: The sponsor played no role in tients who did undergo treatment sessions. the data collection and analysis, manuscript prepara- benefited from it. And since hypochon- tion, or authorization for publication. Author Contributions: Dr Barsky, as principal inves- Acknowledgment: We wish to acknowledge the enor- driasis is a prevalent problem in am- tigator, had full access to all of the data in this study mous contributions made by Duff Bailey, MD, Heli M. bulatory medical practice,6,44 this frac- and takes responsibility for the integrity of the data Peekna, PhD, and Daniel T. Beck, MSW; and by the and accuracy of the data analysis. research assistants Susan L. Clancy, Beth A. Dela- tion of hypochondriacal patients still Study concept and design: Barsky, Ahern. mater, Jeanne M. Fama, Shera D. Gruen, Elizabeth Liu, represents a sizeable population. The Acquisition of data: Barsky, Ahern. and Lindsay B. Renner.

REFERENCES 1. Kellner R. Somatization and Hypochondriasis. New 16. Barsky AJ. A 37-year-old man with multiple so- 31. Gramling SE, Clawson EP, McDonald MK. Per- York, NY: Praeger; 1986. matic complaints. JAMA. 1997;278:673-679. ceptual and cognitive abnormality model of hypo- 2. Noyes R, Kathol RG, Fisher MM, Phillips BM, Su- 17. Barsky AJ, Wyshak G, Klerman GL. Transient hy- chondriasis: amplification and physiological reactiv- elzer M, Woodman CL. One-year follow-up of medi- pochondriasis. Arch Gen Psychiatry. 1990;47:746- ity in women. Psychosom Med. 1996;58:423-431. cal outpatients with hypochondriasis. Psychosomat- 752. 32. Wise TN, Mann LS. The relationship between so- ics. 1994;35:533-545. 18. Barsky AJ, Wyshak G, Klerman GL. Hypochondria- matosensory amplification, alexithymia, and neuroti- 3. Barsky AJ, Fama JM, Bailey ED, Ahern DK. A pro- sis: an evaluation of the DSM-III criteria in medical out- cism. J Psychosom Res. 1994;38:515-521. spective 4-5 year study of DSM-III-R hypochondria- patients. Arch Gen Psychiatry. 1986;43:493-500. 33. Barsky AJ, Wyshak G, Latham KS, Klerman GL. sis. Arch Gen Psychiatry. 1998;55:737-744. 19. Barsky AJ, Cleary PD, Sarnie MK, Klerman GL. The The Somatosensory Amplification Scale and its rela- 4. Lipsitt DR, Starcevic V. Hypochondriasis: Modern course of transient hypochondriasis. Am J Psychia- tionship to hypochondriasis. J Psychiatr Res. 1990; Perspectives on an Ancient Malady. New York, NY: try. 1993;150:484-488. 24:323-334. Oxford University Press; 2001. 20. Pilowsky I. Dimensions of hypochondriasis. Br J 34. Jette AM, Davies AR, Cleary PD, et al. The Func- 5. Kellner R, Abbott P, Pathak D, Winslow WW, Um- Psychiatry. 1967;113:89-93. tional Status Questionnaire: reliability and validity when land BE. Hypochondriacal beliefs and attitudes in fam- 21. Pilowsky I. A general classification of abnormal used in primary care. J Gen Intern Med. 1986;1:143- ily practice and psychiatric patients. Int J Psychiatry illness behaviours. Br J Med Psychol. 1978;51:131- 149. Med. 1983;13:127-139. 137. 35. Cleary PD, Jette AJ. Reliability and validity of the 6. Barsky AJ, Wyshak G, Klerman GL, Latham KS. The 22. Salkovskis PM, Rimes KA, Warwick HMC, Clark Functional Status Questionnaire. Qual Life Res. 2000; prevalence of hypochondriasis in medical outpa- DM. The Health Anxiety Inventory: development and 9:747-753. tients. Soc Psychiatry Psychiatr Epidemiol. 1990;25: validation of scales for the measurement of health anxi- 36. Derogatis LR. The SCL-90-R: Administration, Scor- 89-94. ety and hypochondriasis. Psychol Med. 2002;32:843- ing, and Procedures Manual II. Towson, Md: Clinical 7. Warwick HM, Clark DM, Cobb AM, Salkovskis PM. 853. Psychometric Research; 1983. A controlled trial of cognitive behavioural treatment 23. Weinstein MC, Berwick DM, Goldman PA, Mur- 37. Derogatis LR. SCL-90-R: Administration, Scor- of hypochondriasis. Br J Psychiatry. 1996;169:189- phy JM, Barsky AJ. A comparison of three psychiatric ing, and Procedures Manual. 3rd ed. Minneapolis, 195. screening tests using receiver operating characteris- Minn: National Computer Systems Inc; 1994. 8. Avia MD, Ruiz A, Olivares ME, et al. The meaning tic (ROC) analysis. Med Care. 1989;27:593-607. 38. Parkerson GR Jr, Michener JL, Wu LR, et al. As- of psychological symptoms: effectiveness of a group 24. Barsky AJ, Wyshak G, Latham KS, Klerman GL. sociations among family support, family stress, and per- intervention with hypochondriacal patients. Behav Res Hypochondriacal patients, their physicians, and their sonal functional health status. J Clin Epidemiol. 1989; Ther. 1996;34:23-31. medical care. J Gen Intern Med. 1991;6:413-419. 42:217-229. 9. Warwick HMC, Marks IM. Behavioural treatment 25. Kroenke K, Spitzer RL, deGruy FV, Swindle R. A 39. Parkerson GP, Broadhead WE, Chin-Kit JT. Qual- of illness phobia and hypochondriasis: a pilot study of symptom checklist to screen for somatoform disor- ity of life and functional health of primary care pa- 17 cases. Br J Psychiatry. 1988;152:239-241. ders in primary care. Psychosomatics. 1998;39:263- tients. J Clin Epidemiol. 1992;45:1303-1313. 10. Stern R, Fernandez M. Group cognitive and be- 272. 40. Kashner TM, Rost K, Cohen B, Anderson M, Smith havioral treatment for hypochondriasis. BMJ. 1991; 26. Barsky AJ, Wyshak G, Klerman GL. Medical and GR. Enhancing the health of somatization disorder pa- 303:1229-1231. psychiatric determinants of outpatient medical utili- tients: effectiveness of short-term group therapy. Psy- 11. Papageorgiou C, Wells A. Effects of attention train- zation. Med Care. 1986;24:548-560. chosomatics. 1995;36:462-470. ing in hypochondriasis: a brief case series. Psychol Med. 27. Noyes R, Happel RL, Yagla SJ. Correlates of hy- 41. Rost K, Kashner TM, Smith GR Jr. Effectiveness 1998;28:193-200. pochondriasis in a nonclinical population. Psychoso- of psychiatric intervention with somatization disor- 12. Bouman TK, Visser S. Cognitive and behavioural matics. 1999;40:461-469. der patients: improved outcomes at reduced costs. Gen treatment of hypochondriasis. Psychother Psycho- 28. Noyes R, Kathol R, Fisher M, Phillip B, Suelzer M, Hosp Psychiatry. 1994;16:381-387. som. 1998;67:214-221. Holt C. The validity of DSM-III-R hypochondriasis. Arch 42. Smith GR, Rost K, Kashner TM. A trial of the effect 13. Clark DM, Salkovskis PM, Hackmann A, et al. Two Gen Psychiatry. 1993;50:961-970. of a standardized psychiatric consultation on health psychological treatments for hypochondriasis. BrJPsy- 29. Barsky AJ, Cleary PD, Spitzer RL, Williams JBW, outcomes and costs in somatizing patients. Arch Gen chiatry. 1998;173:218-225. Wyshak G, Klerman GL. A structured diagnostic in- Psychiatry. 1995;52:238-243. 14. Barsky AJ, Wyshak G. Hypochondriasis and so- terview for hypochondriasis: a proposed criterion stan- 43. Smith GR Jr, Monson RA, Ray DC. Psychiatric matosensory amplification. Br J Psychiatry. 1990;157: dard. J Nerv Ment Dis. 1992;180:20-27. consultation in somatization disorder: a randomized 404-409. 30. Speckens AEM, Spinhoven P, Sloekers PPA, Bolk controlled study. N Engl J Med. 1986;314:1407- 15. Barsky AJ. Somatosensory amplification and hy- JH, van Hemert AM. A validation study of the Whitely 1413. pochondriasis. In: Lipsitt DR, Starcevic V, ed. Hypo- Index, the Illness Attitude Scales, and the Somatosen- 44. Escobar JI, Gara M, Waitzkin H, Cohen Silver R, chondriasis: Modern Perspectives on an Ancient sory Amplification Scale in general medical and gen- Holman A, Compton W. DSM-IV hypochondriasis in Malady. New York, NY: Oxford University Press; 2001: eral practice patients. J Psychosom Res. 1996;40:95- primary care. Gen Hosp Psychiatry. 1998;20:155- 223-248. 104. 159.

Downloaded From: https://jamanetwork.com/ on 09/25/2021