lable at ScienceDirect

Behaviour Research and Therapy 58 (2014) 65e74

Contents lists avai

Behaviour Research and Therapy

journal homepage: www.elsevier .com/locate/brat

Cognitive-behavioral therapy for hypochondriasis/health anxiety:A meta-analysis of treatment outcome and moderators

Bunmi O. Olatunji a, *, Brooke Y. Kauffman b, Sari Meltzer a, Michelle L. Davis b,Jasper A.J. Smits b, Mark B. Powers b

a Department of Psychology, Vanderbilt University, United Statesb Department of Psychology, University of Texas-Austin, United States

a r t i c l e i n f o

Article history:Received 18 March 2014Received in revised form12 May 2014Accepted 16 May 2014Available online 24 May 2014

Keywords:HypochondriasisHealth anxietyCognitive-behavioral therapyModeration

* Corresponding author. Department of PsychologWilson Hall, 111 21st Avenue South, Nashville, TN 372

E-mail address: olubunmi.o.olatunji@vanderbilt.ed

http://dx.doi.org/10.1016/j.brat.2014.05.0020005-7967/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

The present investigation employed meta-analysis to examine the efficacy of cognitiveebehavioraltherapy (CBT) for hypochondriasis/health anxiety as well as potential moderators that may be associatedwith outcome. A literature search revealed 15 comparisons among 13 randomized-controlled trials(RCTs) with a total sample size of 1081 participants that met inclusion criteria. Results indicated that CBToutperformed control conditions on primary outcome measures at post-treatment (Hedges's g ¼ 0.95)and at follow-up (Hedges's g ¼ 0.34). CBT also outperformed control conditions on measures ofdepression at post-treatment (Hedges's g ¼ 0.64) and at follow-up (Hedges's g ¼ 0.35). Moderator an-alyses revealed that higher pre-treatment severity of hypochondriasis/health anxiety was associatedwith greater effect sizes at follow-up visits and depression symptom severity was significantly associatedwith a lower in effect sizes at post-treatment. Although effect size did not vary as a function of blindassessment, smaller effect sizes were observed for CBT vs. treatment as usual control conditions than forCBT vs. waitlist control. A dose response relationship was also observed, such that a greater number ofCBT sessions was associated with larger effect sizes at post-treatment. This review indicates that CBT isefficacious in the treatment of hypochondriasis/health anxiety and identifies potential moderators thatare associated with outcome. The implications of these findings for further delineating prognostic andprescriptive indicators of CBT for hypochondriasis/health anxiety are discussed.

© 2014 Elsevier Ltd. All rights reserved.

The central feature of hypochondriasis is a preoccupation withthe inaccurate belief that one has, or is in danger of developing, aserious medical condition based on misinterpretations of benign(or minor) bodily sensations (American Psychiatric Association[APA], 2000). Research has shown that up to 9% of patients ingeneral medical practice clinics and up to 5% of the general popu-lation meets diagnostic criteria for hypochondriasis (Creed &Barsky, 2004; Gureje, Ustun, & Simon, 1997). The heightenedprevalence in medical practice may reflect the fact that hypo-chondriasis is also characterized by a strong “disease conviction”that persists despite appropriate medical evaluation. This preoc-cupation with medical illness often focuses on specific signs orsymptoms (e.g., sore throat), diseases (e.g., cancer), organs (e.g.,heart), or vaguely defined somatic phenomena (e.g., “my achingveins”). Typically, patients with hypochondriasis attribute

y, Vanderbilt University, 30103, United States.u (B.O. Olatunji).

unwanted bodily sensations to possible disease (e.g., “this headachemeans I have a brain tumor”) and are highly concerned with theirauthenticity. Perhaps the most readily observable sign of hypo-chondriasis is the persistent attempt to seek reassurance about thefeared symptoms or illness. Individuals with this condition mayrepeatedly contact doctors, seek additional tests, scour Internetsites and medical texts, and seek reassurance from significantothers about bodily sensations which have been appropriatelyevaluated and judged to be benign. This preoccupationwith diseasecan be disruptive to social, occupational, and family functioning,and is associated with substantial economic costs (Katon&Walker,1998).

Although hypochondriasis has historically been viewed as asomatoform disorder (APA, 2000), the validity of this categorizationhas not been without debate. It has been argued that hypochon-driasis may be best conceptualized as an anxiety disorder (Olatunji,Deacon, & Abramowitz, 2009). This view suggests that healthanxiety represents a continuous dimension (ranging fromno healthanxiety to severe health anxiety) with ‘hypochondriasis’ at its

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e7466

clinical endpoint (Marcus, Gurley, Marchi, & Bauer, 2007). Thisargument is based largely on empirical observations that symp-toms of hypochondriasis overlap with certain anxiety disorders:namely, panic disorder (PD) and obsessive-compulsive disorder(OCD). Like those with hypochondriasis, patients with PD arehypervigilant to benign, arousal-related body sensations and oftenerroneously attribute them to organic causes such as heart attacks,strokes, and other serious medical conditions (Abramowitz, 2005;Barsky, Barnett, & Cleary, 1995). Similarities have also beenobserved between hypochondriasis and OCD in terms of preoccu-pation with health and disease, and the repetitive and pervasivenature of such preoccupation (Abramowitz, 2005; Fallon, Javitch,Hollander, & Liebowitz, 1992). Much like PD and OCD, cognitive-behavioral models (Abramowitz, Schwartz, & Whiteside, 2002;Warwick & Salkovskis, 1990) posit that hypochondriasis is anextreme form of health anxiety that emerges from the misinter-pretation of benign and normally occurring experiences that lead toanxiety and the use of safety behaviors which paradoxicallymaintains the anxiety (Abramowitz & Moore, 2007; Abramowitz,Olatunji, & Deacon, 2007).

The cognitive-behavioral model of hypochondriasis is depictedin Fig. 1 (Abramowitz, Deacon, & Valentiner, 2007). This modelsuggests that dysfunctional beliefs about bodily symptoms andillness play a significant role in the development of hypochon-driasis. Such beliefs often increase the likelihood of having cata-strophic cognitions when exposed to benign bodily symptoms orhealth-related information. Once concerned about the possibilityof acquiring an illness, the person becomes vigilant for any signs ofbeing ill and is motivated to reduce their worry by gaining certainty

Fig. 1. Cognitive-behavioral model of hypochondriasi

about their health status. It is important to note that this process inhypochondriasis is driven largely by mistaken beliefs about ill-nesses. According to the cognitive-behavioral model, mistakenbeliefs in hypochondriasis are maintained (despite contradictoryinformation and repeated reassurance of good health frommedicalprofessionals) by maladaptive strategies used to cope with health-related anxiety. These strategies include attempts to prevent thefeared illness, avoidance, and attempts to attain certainty abouthealth status. These safety behaviors prevent individuals with hy-pochondriasis from acquiring information that would disconfirmtheir mistaken beliefs about illnesses. This cognitive-behavioralmodel of hypochondriasis has been empirically supported andproven to be clinically useful (Taylor & Asmundson, 2004). Based inpart on these observations, the newly published DSM-5 hasreplaced the diagnosis of hypochondriasis with “illness anxietydisorder” (APA, 2013).

Based on the findings of early, predominantly psychodynamicinterventions, hypochondriasis has historically been regarded asresistant to psychological treatment (Taylor, Asmundson, & Coons,2005). In fact, treatments were initially considered to be of limitedvalue for hypochondriasis (Ladee, 1966). Extreme interventionswere considered and some clinicians even employed prefrontallobotomies for the treatment of hypochondriasis (Bernstein,Callahan, & Jaranson, 1975). Fortunately, the emergence of acognitive-behavioral model of hypochondriasis has paved the wayfor the application of a more appropriate approach to treatment.The cognitive-behavioral approach is derived largely from theobservation that symptoms of hypochondriasis d at both a topo-graphical and functional level d overlap remarkably with certain

s outlined by Abramowitz, Deacon, et al. (2007).

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e74 67

anxiety disorders (Olatunji et al., 2009). As with other anxietydisorders, this approach emphasizes the role of dysfunctional be-liefs in maintaining hypochondriasis. This approach also suggeststhat avoidance and health-related safety behaviors preventdysfunctional beliefs from being disconfirmed, and therebyperpetuate excessive health anxiety and somatic preoccupation(Warwick & Salkovskis, 2001). This functional analysis of hypo-chondriasis would then suggest that cognitive behavioral therapy(CBT) may provide much needed symptom relief. CBT here refers tothe class of interventions that are based on the basic premise thatemotional disorders are maintained by cognitive and behavioralfactors, and that psychological treatment leads to changes in thesefactors through cognitive and behavioral techniques (Beck &Emery, 2005).

Research on the effects of CBT for hypochondriasis has producedencouraging results. In one study, CBT was found to be superior tono treatment in reducing health anxiety, the need for reassurance,and the frequency of checking behavior (Warwick, Clark, Cobb, &Salkovskis, 1996). A subsequent study found that patientsreceiving cognitive therapy showed significantly more reduction insymptoms of hypochondriasis at mid-treatment and at post-treatment compared to those receiving a stress managementintervention (Clark et al., 1998). Given that research has shown thatpatients with excessive health anxiety have a preference for psy-chological treatments over drug treatments (Walker, Vincent,Furere, Cox, & Kjernisted, 1999), a better understanding of the ef-ficacy of CBTmay provide a solid empirical foundation for making itthe gold standard treatment for hypochondriasis. Meta-analysis isthe primary means by which researchers have synthesized theresults from multiple treatment trials examining the efficacy ofvarious treatments. Although the use of meta-analytic data is notwithout limitations, this approach has offered very preliminaryinsight into delineating the general effectiveness of CBT in thetreatment of hypochondriasis. An initial meta-analytic review ofthe literature concluded that CBT is themost effective treatment forhypochondriasis (Taylor et al., 2005). However, this investigationexamined only 5 studies utilizing CBT and included uncontrolledcomparisons in the analysis. Thomson and Page's (2007) morerecent Cochrane review, which included only six studies, alsoconcluded that CBT was the most effective psychological treatmentfor hypochondriasis. However, several limitations of the existingtrial data at the time of the publication were noted. For example,the reliance on wait-list control conditions in many trials makes itdifficult to rule out the possibility that improvements after CBTwere due to nonspecific factors.

Although much remains unknown about the incremental ef-ficacy of CBT for hypochondriasis, even less is known about reli-able predictors of treatment outcome. For example, some researchsuggests that comorbid psychiatric disorders are poor prognosticindicators (Barsky, Fama, Bailey, & Ahern, 1998; Noyes et al., 1994).However, other studies report that comorbid anxiety anddepression is either related to good outcome (House, 1989) orunrelated to outcome (Greeven et al., 2009). Research has alsoshown that younger age is associated with good outcome (Barsky,1996; House, 1989) while other studies have failed to observe thisassociation (Speckens, Spinhoven, van Hemert, Bolk, & Hawton,1997). Taylor et al. (2005) suggest that the inconsistent findingswith regards to predictors of treatment outcome in hypochon-driasis may be because some variables predict outcome for sometreatments but not others. It may also be the case that theinconsistent findings reflect the fact that the variables are weakpredictors of outcome. Yet another possibility is that individualstudies lack the power necessary to detect predictors of treatmentoutcome. To the extent that this is the case, meta-analysis of manytrials may prove useful in clarifying some of these discrepancies.

Clarifying the substantive predictors of treatment outcome mayhave important implications for treatment planning in CBT. Forexample, if it is the case that comorbidity predicts worse treat-ment outcome, this may be justification for the use of adjunctivetreatments to CBT that more directly target the comorbid condi-tions. This may also have implications for the time course of CBTfor hypochondriasis. Among those with more severe comorbidconditions, implementation of CBT may be difficult, therebyinterfering with effective treatment. Treatment of comorbid dis-orders before initiating CBT for hypochondriasis in such cases maythen maximize outcome.

In addition to clarifying predictors of treatment outcome,examining the extent to which sample-specific characteristics (e.g.,gender) and study-specific characteristics (e.g., number of treat-ment sessions) predict outcome in CBT for hypochondriasis mayalso prove to be informative. Indeed, it has been observed that thequestion towards which all outcome research should ultimately bedirected is the following: “What treatment, by whom, is mosteffective for this individual with that specific problem, and underwhich set of circumstances?” (Paul, 1967, p. 111). Although no onestudy can address such a complicated question (Beutler, 1991),examination of moderators in the context of meta-analysis oftreatment outcome studies examining the efficacy of CBT for hy-pochondriasis may be an important preliminary step in beginningto address this question. Accordingly, the present investigationemploys a meta-analytic approach to examine the efficacy of CBTfor hypochondriasis. This investigation is especially timely giventhe limitations of the prior two meta-analyses CBT for hypochon-driasis, one of which consisted of only 5 CBT studies and includeduncontrolled comparisons (Taylor et al., 2005) and the other whichwas limited to only 6 studies (Thomson & Page, 2007). In addition,this investigation builds on prior work by examining moderators oftreatment outcome. Four questions derived from the existingliterature were examined.

1. Do CBT treatments outperform control conditions on hypo-chondriasis/health anxiety outcome measures at post-treatment and at follow-up?

2. Do CBT treatments outperform control conditions on depressionseverity measures at post-treatment and follow-up?

3. Does higher pre-treatment hypochondriasis/health anxietyseverity and higher pre-treatment depression scores predictlower between-group effect sizes at post-treatment and follow-up?

4. Does comparator treatment/condition, participant age, per-centage of female participants, percentage of comorbidity,number of sessions, inclusion of treatment integrity checks and/or blind assessors (i.e., clinical trial quality) moderate treatmenteffect sizes?

Method

Study selection

Randomized controlled trials (RCTs) of CBT (see inclusioncriteria below) for hypochondriasis/health anxiety were selectedusing a comprehensive search strategy. A search was conducted inPsycINFO and MEDLINE (1966 to March 2014). The searchesincluded the following terms: “cognitive behavioral” or “cognitivebehavioral therapy,” and “clinical trial” or “trial” alone and incombination with “hypochondriasis” or “health anxiety.” Thesewords were searched as key words, title, abstract, and MedicalSubject Headings. In addition, we contacted authors of CBT trials foremerging publications.

Fig. 2. Study selection and reasons for exclusions.

1 Primary and secondary outcome measures were defined in accordance withhow the measures were described in the articles with the exception of two articles(Barsky, Ahern, Bauer, Nolido, & Orav, 2013 andWarwick, 1996). Insufficient data onthe primary outcomes were provided for Barsky et al. and primary outcomes werenot clearly stated in the Warwick (1996) article, therefore we defined the primaryand secondary outcome consistent with the population of articles. Primary mea-sures were exclusively measures of hypochondriasis/health anxiety. Secondarymeasures generally included measures of hypochondriasis/health anxiety, generalanxiety, and somatization, and depression. Subsequent analyses were also con-ducted to examine the effects of CBT on depression exclusively.

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e7468

As depicted in Fig. 2, the initial search strategies produced 2988potential articles. Examinationof the abstracts identified25 relevantand non-repeating studies. Inclusion criteria for the meta-analysiswere as follows: (a) participants who met full DSM-III-R, DSM-IV,or DSM-IV-TR criteria for hypochondriasis or had clinical levels ofhealth anxiety; (b) adequate control condition (psychological pla-cebo,wait-list control, orpill placebo); and (c)more thanone sessionof CBT. Treatments were classified as CBT if they included cognitivetechniques (e.g. cognitive restructuring, behavioral experiments,etc.), behavioral techniques (e.g. in-vivo exposure, imaginal expo-sure, etc.), or a combination of these strategies. Acceptance-basedtherapies (e.g., ACT), which emphasize engaging in valued behav-iors despite anxiety, were included if they also utilized cognitive orbehavioral techniques. Exclusion criteria for the meta-analysiswere: (a) single case studies; (b) treatment conditions aimed ataugmentating a psychological treatment; (c) studies with insuffi-cient data, unless study authors were able to provide such data; and(d) studieswith redundant data. Of the 25 identified studies,14wereexcluded based on these criteria. Two additional studies were thenidentified after reviewing the references of the remaining studies.Table 1 shows the 15 comparisons within the 13 studies with a totalsample size of 1081 participants that met the final inclusion criteriaand were included in the meta-analyses.

Software

All analyses were completed with Comprehensive Meta-Analysis (Bornstein & Rothstein, 1999).

Procedure

Data on the following variables were collected when reported ineach of the 13 included studies: treatment conditions (13 studies),control type (13 studies), treatment dose (number of sessions; 13

studies), number of participants (13 studies), mean age (11 studies),and percentage of female participants (12 studies). Treatmentintegrity variables, including inclusion or exclusion of treatmentintegrity checks and blind assessors, were collected categorically andassumed to be absent if not reported in the manuscript. Dependentvariables were labeled as either primary or secondary outcomemeasures as defined by the authors in the original publication.1

Control conditions were classified into three categories: pla-cebo, treatment as usual, or wait-list. Treatments categorized asplacebo were treatments that did not include the cognitive andbehavioral components of CBT, including: attention control, relax-ation, problem-solving, and pill placebo.Wait-list (WL) was definedas a control condition in which participants did not receive anytreatment for hypochondriasis/health anxiety symptoms for aspecified amount of time. Five of the 13 studies utilized treatmentas usual as the control condition, four of the 13 studies had wait-listas the control condition, three of the 13 studies had a psychologicalplacebo as the control condition, and one study employed a pillplacebo as the control condition.

Effect size calculation

Between-group effect sizes for each study were computed usingHedges's g (Rosenthal, 1991). Studies with multiple outcomes were

Table 1Studies included in the meta-analysis.

Study Conditions N Mean age # of sessions Primary outcome measure Secondary outcome measure

Sorensen et al. (2011) CBT vs. WL 56 Not provided 16 HAMA, HAI BDI, BAI, HAMDSeivewright et al. (2008) CBT vs. TAU 49 Not provided 4.3 HAI HADS-DHedman et al. (2011) CBT vs. Psych PL 81 39.05 12 HAI IAS, WI, BAI, ASI, MADRS-S, QOLIGreeven et al. (2007) CBT vs. Pill PL 75 40.32 7.3 WI IAS-HA, IAS-Illness Behavior, SCL-90,

MADRS, BASBarsky and Ahern (2004) CBT vs. TAU 187 43.21 6 WI HCQ, HAI, SSI, SASVisser and Bouman (2001) Exposure vs. WL 42 37.7 12 IAS-HA, IAS-Illness Behavior,

IAS-Body, SAS, SCL-90(Som)BDI, MOCI, SCL-Total

Visser and Bouman (2001) CT vs. WL 40 36.7 12 IAS-HA, IAS-Illness Behavior,IAS-Body, SAS, SCL-90(Som)

BDI, MOCI, SCL-Total

Bourgault-Fagnou andHadjistavropoulos (2013)

CBT vs. WL 39 68.94 6 WI SHAI, SSI, STAI-S, STAI-T, GDS

Bourgault-Fagnou andHadjistavropoulos (2013)

ECBT vs. WL 36 68.53 6 WI SHAI, SSI, STAI-S, STAI-T, GDS

Warwick (1996) CBT vs. WL 36 37 16 HAI BDI, BAISumathipala et al. (2008) CBT vs. TAU 150 35 3 GHQ-30 BSI,# of visits,# of complaintsSpeckens et al. (1995) CBT vs. TAU 79 37.1 16-Jan Intensity-Mean, Intensity-Max IAS-HA, IAS-Illness Behavior, WI,

HADS-A, HADS-DBarsky et al. (2013) CBT vs. Psych PL 89 51.84 9 HAI SSI, MCQ, Beliefs, SCL-90, FSQMcManus, Surawy, Muse,

Vazques-Montes, andWilliams (2012)

CBT vs. TAU 74 42.6 8 HAC BAI, BDI, FFMQ

Buwalda, Bouman, andVan Duijn (2006)

CBT vs. Psych PL 48 41 6 GIAS BDI

Note. CBT ¼ Cognitive Behavior Therapy, WL ¼ Waitlist, TAU ¼ Treatment As Usual, PL ¼ Placebo, NP ¼ Not Provided, HAMA: Hamilton Anxiety Rating Scale, HAI: HealthAnxiety Inventory, BDI: Beck Depression Inventory, BAI: Beck Anxiety Inventory, HAMD: Hamilton Depression Rating Scale, HADS-D: Hospital Anxiety and Depression Scale-Depression, IAS: Illness Attitude Scale, WI: Whitely inventory, ASI: Anxiety Sensitivity Inventory, MADRS: Montgomery Asberg Depression Rating Scale, QOLI: Quality of LifeInventory, IAS-HA: Illness Attitude Scale-Health Anxiety, IAS-Illness Behavior: Illness Attitude Scale-Illness Behavior, SCL-SOM: Symptom Check List-Somatization scale, BAS:Brief Anxiety Scale, HCQ: Hypochondriacal Cognitions Questionnaire, SSI: Somatic Symptoms Inventory, SAS: Somatosensory Amplification Scale, IAS-Body: Illness AttitudeScale-Body, MOCI: Maudsley Obsessive Compulsive Inventory, SCL-Total: Symptom Check List-Total, SHAI: Short Health Anxiety Inventory, STAI-S: State Trait Anxiety In-ventory State-Scale, STAI-T: State Trait Anxiety Inventory-Trait Scale, GDS: Geriatric Depression Scale, GHQ-30: General Health Questionnaire-30, BSI: Bradford SomaticInventory, HADS-A: Hospital Anxiety and Depression Scale-Anxiety, MCQ: Metacognitions Questionnaire, SCL-90: Symptom Checklist-90, FSQ: Functional Status Question-naire, HAC: Health Anxiety Composite, FFMQ: Five Facet Mindfulness Questionnaire, GIAS: Groningen Illness Attitude Scale.

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e74 69

categorized as above (see Table 1) and then combined within eachdomain. When the necessary data were available, Hedges's g wascalculated directly using the following formula: g ¼ XT � XC=SPwhere XT is the mean of the treatment group, XC is the mean of thecomparison group, and SP is the pooled standard deviation. If thesedata were not provided, Hedges's gwas estimated using conversionequations for significance tests (e.g., t, F) (Rosenthal,1991). All effectsizes were corrected for small sample sizes according to Hedgesand Olkin (1985). Therefore, a smaller sample size reduces theestimated effect size helping control for different sample sizesacross studies. These controlled effect sizes may then be inter-preted conservatively with Cohen's convention of small (0.2), me-dium (0.5), and large (0.8) effects (Cohen, 1988). Hedges's g alsomay be computed directly from Cohen's d with the following for-mula: g ¼ d (1 � 3/4 (n1 þ n2) � 9). When there were multipleoutcomes per domain theywere combined according to Borenstein,Hedges, and Rothstein (2007). The overall mean effect size for all ofthe studies combined was computed using the following formula:g ¼ P

wjgj=P

wj wherewj is the weight for each study and gj is theeffect size for each study. Effect sizes were calculated with randomeffects models. The random effects analysis estimates the overalleffect size assuming the studies included are only a sample of theentire population of studies and/or that the studies areheterogeneous.

Results

Heterogeneity

To test the assumption that the effect sizes of this study werefrom a homogeneous sample, a heterogeneity analysis was con-ducted. A total of 13 studies and 15 comparisons across pooled time

points (post and follow-up) on primary outcome measures wereincluded in this test of heterogeneity. The test was significant,Q(14) ¼ 58.03, p < .001, suggesting that random effects analysesweremost appropriate for this study. This significant heterogeneityalso suggests it may be appropriate to employ moderator analysesto identify potential sources of study variability.

Question 1: does CBT outperform the control conditions on primaryand secondary outcome measures at post-treatment and follow-up?

Fig. 3 depicts the relative advantage of CBT compared to controlconditions on primary outcomes at post-treatment in an analysis of9 studies and 11 comparisons (Hedges's g ¼ 0.95 [SE ¼ 0.15, 95% CI:0.66e1.22, p < .001]). The analysis involving outcomes at follow-upincluded 7 studies and 7 comparisons and indicated that CBT out-performs control conditions on primary outcomes at follow-up,although the effect size was significantly smaller than thatobserved at post-treatment (Hedges's g ¼ 0.34 [SE ¼ 0.14, 95% CI:0.06e0.62, p ¼ .016]).

A post-treatment analysis of secondary outcome measureswhere CBT outperformed control conditions included 9 studies and11 comparisons (Hedges's g ¼ 0.56 [SE ¼ 0.11, 95% CI: 0.35e0.78,p < .001]). A follow-up analysis of 7 studies and 7 comparisonsrevealed that CBT also outperformed control conditions on sec-ondary outcomes at follow-up (Hedges's g¼ 0.20 [SE¼ 0.10, 95% CI:0.02e0.38, p ¼ .028]).1

Question 2: does CBT outperform the control conditions on measuresof depression at post-treatment and follow-up?

An analysis of measures of depression at post-treatmentincluded 9 studies and 11 comparisons and showed that CBT

Fig. 3. Effect size estimates (Hedges' g) for the efficacy of CBT compared to controlconditions on primary symptom reduction.

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e7470

outperformed control conditions (Hedges's g ¼ 0.64 [SE ¼ 0.12, 95%CI: 0.41e0.86, p < .001]). The pooled follow-up analysis included atotal of 4 studies and 4 comparisons, where CBT also outperformedcontrol conditions on depression measures (Hedges's g ¼ 0.35[SE ¼ 0.13, 95% CI: 0.01e0.61, p ¼ .008]).

Question 3: does higher pre-treatment hypochondriasis/healthanxiety severity and higher pre-treatment depression scores predictlower effect sizes?

The analysis for pre-treatment severity included 4 studies and 4comparisons with 218 participants and revealed that higher pre-treatment severity was not significantly associated with highereffect size at post treatment (b ¼ 0.00, p ¼ .400). The analysis ofpre-treatment severity at follow up visits included 3 studies and 3comparisons with 325 participants and revealed that higher pre-treatment severity was significantly associated with higher effectsize at follow-up (b ¼ 0.03, p < .050). The analysis for pretreatmentdepression included 5 studies and 6 comparisons with 292 par-ticipants where higher pre-treatment severity was significantlyassociated with lower effect size at post treatment (b ¼ �0.09,p < .050).2 Insufficient data was provided to analyze pre-treatmentdepression severity at follow-up visits.

Question 4: does effect size vary as a function of control type, blindassessment, number of sessions, percentage of females, mean age,and comorbidity?

The following analyses were completed using fully random ef-fects categorical moderator analyses. The 4 studies and 6 compar-isons employing wait-list control conditions (Hedges's g ¼ 1.10[SE ¼ 0.14, 95% CI: 0.83e1.37, p < .001]) showed larger effect sizesthan the 5 treatment as usual control conditions on primary out-comes across pooled time points (post and follow-up) (Hedges'sg ¼ 0.46 [SE ¼ 0.13, 95% CI: 0.20e0.71, p < .001]). Comparison be-tween other control types and placebo-controlled studies was not

2 Studies that used the Health Anxiety Inventory were used to examine therelationship between higher pre-treatment hypochondriasis/health anxiety and theeffect size for CBT. Studies that used the Beck Depression Inventory were used toexamine the relationship between higher pre-treatment depression and the effectsize for CBT. These were the most common measures used.

possible as few studies employed these control conditions (1 pillplacebo and 3 psychological placebos).

There was no significant difference in effect size for post-treatment outcomes for 7 comparisons and 6 studies with blindassessors (Hedges's g¼ 1.11 [SE¼ 0.19, 95% CI: 0.73e1.49, p� .001])and 4 comparisons and 3 studies which did not report having blindassessors (Hedges's g¼ 0.66 [SE¼ 0.18, 95% CI: 0.31e1.01, p < .001]).This comparison (4 comparisons and 4 studies) was not significantfor follow-up outcomes either for the studies with blind assessors(Hedges's g ¼ 0.72 [SE ¼ 0.19, 95% CI: �0.10 to 0.65, p ¼ .155]) andthe 3 comparisons and 3 studies without blind assessors (Hedges'sg ¼ 0.46 [SE ¼ 0.25, 95% CI: �0.03 to 0.95, p ¼ .068]). Treatmentintegrity was reported in 12 studies, but a comparison was notpossible because only 1 study failed to report treatment integrity.

The following analyses were completed using unrestrictedmaximum likelihood meta regressions. There was a significantrelation between effect size and number of sessions (11 compari-sons and 9 studies; b ¼ 0.07, p ¼ .038), with more sessions asso-ciatedwith larger effect sizes at post-treatment (Fig. 4). The relationbetween effect size and number of sessions at follow-up was notsignificant (7 comparisons and 7 studies; b ¼ �0.01, p ¼ .858).There was no significant relation between effect size and femalepercentage at post-treatment (10 comparisons and 8 studies;b ¼ �0.00, p ¼ .822) nor at the follow-up analysis (7 comparisonsand 7 studies; b ¼ �0.01, p ¼ .373). Additionally, there was not asignificant relation between effect size and mean age at post-treatment (9 comparisons and 7 studies; b ¼ 0.00, p ¼ .767) or atfollow up (6 comparisons and 6 studies; b ¼ �0.02, p ¼ .465). Aninsufficient amount of studies provided information on comor-bidity percentage; therefore, that analysis was not possible.

Publication bias: “the file drawer problem”

The omission of file drawer studies is a problem because it canlead to biased meta-analytic estimates if their psychometric prop-erties (i.e., reliability and validity) differ from psychometric prop-erties of published studies (Rosenthal, 1991). Rosenthal's fail-safe Nwas an ingenious response to this file drawer problem in theintegration of research (Orwin, 1983; Rosenthal, 1979). Thisformula allowed for a direct assessment of the threat posed bysampling bias in the literature search (Orwin, 1983). Rosenthalsuggested the following equation to compute a fail-safeN:X ¼ KðKZ2 � 2:706Þ=2:706 where K is the number of studies inthe meta-analysis and Z is the mean Z obtained from the K studies(Rosenthal, 1991). Rosenthal also suggested that findings may beconsidered robust if the required number of studies (X) to reducethe overall effect size to a non-significant level exceeded 5K þ 10,which in this study would be 85 (Rosenthal, 1991). The analysesconducted on the 15 comparisons amongst the 13 studies revealedthat it would require more than 166 current or future unpublishedstudies with an effect size of 0 in order to bring the overall effectsize of the primary analyses within the non-significant range. Thissuggests that the findings in this meta-analysis are robust as thefail-safe-N (166) is much larger than the convention for robustcutoff or 5K þ 10 (85).

Discussion

Traditionally, hypochondriasis was considered resistant to psy-chotherapy, possibly because of the shortcomings of psychody-namic and psychoanalytic conceptualizations, which historicallydominated the treatment of this condition. These models wereremoved from behavior, cognition, and biology, and the treatmentsderived from them produced little enduring benefit. Within the lasttwo decades, however, a model of hypochondriasis as “health

Regression of Number of Sessions on Hedges's g

Number of Sessions

Hed

ges'

s g

3.13 4.53 5.94 7.34 8.75 10.15 11.55 12.96 14.36 15.77 17.17

2.00

1.80

1.60

1.40

1.20

1.00

0.80

0.60

0.40

0.20

0.00

Fig. 4. The relationship between CBT for hypochondriasis/health anxiety effect size and number of sessions.

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e74 71

anxiety” has been advanced that draws from the cognitive (i.e.,dysfunctional beliefs, body vigilance, anxiety sensitivity, intoler-ance of uncertainty) and behavioral (i.e., avoidance, safety-seeking)processes implicated in the development of other anxiety disorders(Olatunji et al., 2009). This model, depicted in Fig. 1, has informedthe application of CBT to the treatment of hypochondriasis byhelping patients recognize and modify faulty beliefs about illnessand eliminate behavioral responses that prevent the self-correctionof faulty beliefs (Taylor & Asmundson, 2004). The present investi-gation employed meta-analysis to examine the efficacy of thisapproach to the treatment of hypochondriasis/health anxiety. Thefindings showed that CBT outperformed control conditions onprimary symptom outcome measures at post-treatment showing alarge effect size. This finding is consistent with prior meta-analysesdemonstrating that CBT is effective in reducing symptoms of hy-pochondriasis (Taylor et al., 2005; Thomson & Page, 2007). How-ever, the present study expands this prior research by including anumber of RCTs that have been published since these previousmeta-analyses, and thus adds to the evidence base of CBT for hy-pochondriasis/health anxiety.

The present investigation also found that CBT outperformedcontrol conditions on primary hypochondriasis/health anxietysymptom outcome measures at follow-up showing a small effectsize. Although the sustained efficacy (the ability to produce lastingsymptomatic changes) of CBT-based approaches has been ques-tioned (e.g., Eddy, Dutra, Bradley, & Westen, 2004), these meta-analytic findings suggest that treatment gains attributed to CBT(relative to controls) are observed after treatment is completed.While these findings are encouraging, additional research is neededto adequately determine the extent to which CBT produces longerlasting symptom changes for patients with hypochondriasis/healthanxiety. This will require future studies to include substantiallylonger follow-up intervals so more definitive inferences can bemade regarding the durability of CBT for hypochondriasis/healthanxiety. It is important to note that the overall effect size for CBTwas significantly larger at post-treatment (g ¼ .95) compared tofollow-up (g ¼ 0.34). This suggests that the therapeutic effects ofCBT for hypochondriasis/health anxiety may diminish after theacute treatment phase. Future research aimed at identifying stra-tegies that may be employed during and after CBT (i.e., boostersessions) in order to better sustain treatment gains for patientswith hypochondriasis/health anxiety may prove valuable. Suchapproachesmay be vital in reducing the health care costs associatedwith excessive doctor visits and requests for unnecessary medicaltests that is often observed among those with hypochondriasis/health anxiety.

Although the present findings suggest that the therapeutic ef-fects of CBT for hypochondriasis/health anxiety may diminish afterthe acute treatment phase, they also suggest that the acute thera-peutic effects of CBT for patients with hypochondriasis/healthanxiety may generalize to other symptoms. More specifically, CBToutperformed control conditions on secondary outcome measuresat post-treatment. Although secondary outcome measures didinclude assessment of symptoms of hypochondriasis/health anxi-ety, they also included assessment of psychiatric symptoms broadlydefined, as well as specific symptoms of general anxiety, somati-zation, and depression. Prior research has shown that hypochon-driasis/health anxiety does demonstrate significant associationswith general psychopathology (Gropalis, Bleichhardt, Witthoft, &Hiller, 2012; Weck, Neng, Richtberg, & Stangier, 2012). However,the present study suggests that CBT that is specifically targetedtowards the reduction of hypochondriasis/health anxiety may alsosignificantly reduce other symptoms of distress.

The present study also examined if CBT treatments out-performed control conditions on reducing depressive symptomsspecifically at post-treatment. The experience of depression is quitecommon among those with hypochondriasis/health anxiety (Noyeset al., 1994), so much so that some have speculated that hypo-chondriasis is a “masked” form of depression (Lesse, 1980). Thepresent findings revealed that CBT for patients with hypochon-driasis/health anxiety also resulted in significant reductions insymptoms of depression. These findings are consistent with priorwork showing that CBT that specifically targets anxiety-relateddisorders also outperform control conditions in reducing symp-toms of depression (Hofmann & Smits, 2008; Olatunji, Davis,Powers, & Smits, 2013). Research has shown that symptoms ofhypochondriasis/health anxiety may emerge before those ofdepression (Noyes et al., 1994; Simon Gureje, & Fullerton, 2001),suggesting that depressive symptoms may represent a response tothe distress and functional impairment associated with hypo-chondriasis/health anxiety (Abramowitz, 2004). Given thisfunctional relationship, directly targeting symptoms of hypochon-driasis/health anxiety may be expected to also lead to reductions insymptoms of depression.

The present investigation also examined the extent to whichinitial pre-treatment hypochondriasis/health anxiety severity andpre-treatment depression scores predicted CBT effect sizes. Priorlongitudinal research suggests that severity of hypochondriasismay be of prognostic significance (Buwalda & Bouman, 2008;Hartman et al., 2009). Although it is intuitive to predict thathigher pre-treatment hypochondriasis/health anxiety severity willpredict lower effect sizes, the opposite was true in the present

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e7472

study. That is, higher pre-treatment hypochondriasis/health anxi-ety was found to be associated with an increase in CBT effect size atfollow-up visits. This finding does replicate those of a recent studythat found that health anxiety at baseline was positively associatedwith symptom improvement after internet-based CBT for severehealth anxiety (Hedman, Andersson, et al., 2013; Hedman,Lindefors, et al., 2013). The present findings also complimentthose of Nakao, Shinozaki, Ahern, and Barsky (2011) who found thatmore baseline anxiety predicted larger improvements associatedwith a CBT intervention for hypochondriasis. Contrary to conven-tional wisdom, the finding that psychological treatment might bemore efficacious for high-severity than for low-severity patientshas also been observed in the treatment of other disorders(Driessen, Cuijpers, Hollon, & Dekker, 2010; Smits et al., 2013;Smits, Minhajudin, Thase, & Jarrett, 2012).

The question remains as to why CBT is more effective (relative tocontrols) for patients with more severe hypochondriasis/healthanxiety. Is this merely regression toward the mean? According toDriessen et al. (2010), evidence of moderation is most likely to befound when efficacious treatments are compared with stringentcontrols and the sample contains both more and less severepatients. In such cases, nonspecific treatments may be sufficient forlow-severity patients, but high-severity patients will require atreatment that has specific effects beyond the simple provision oftreatment as usual in order to fully benefit. Should CBT be proven tobe more efficacious relative to control conditions for patients withmore severe symptoms of hypochondriasis/health anxiety than forthose with less severe symptoms, this does not mean that treat-ment guidelines should be revised to recommend CBT as a mono-treatment for patients with severe symptoms. Indeed, the moder-ated finding of symptom severity should be interpreted withcaution given that the effect was small (b ¼ 0.03) and the clinicalsignificance of such an effect is not sufficiently apparent. Thequestion of the incremental efficacy of CBT for this group of patientsrequires more head-to-head comparisons between CBT and other“bona fide” psychological treatments before more definitive in-ferences can be made.

Unlike pre-treatment hypochondriasis/health anxiety severity,pre-treatment depression symptom severity was significantlyassociated with a decrease in the advantage of CBT over controlconditions at post-treatment. This finding is consistent with recentresearch showing that depressive symptoms at baseline werenegatively related to improvement after internet-based CBT forsevere health anxiety (Hedman, Andersson, et al., 2013; Hedman,Lindefors, et al., 2013). As previously noted, depressive symptomsoften co-occur with hypochondriasis/health anxiety (Noyes et al.,1994). The presence of severe depression may then impederesponse to CBT for hypochondriasis/health anxiety through mul-tiple mechanisms. For example, depressive symptoms maydecrease motivation and compliance with difficult exposure as-signments that are often employed as homework in CBT, therebypreventing meaningful reductions in symptoms of hypochondria-sis/health anxiety. It is important to note that the finding showingthat pre-treatment depression symptom severity may predictworse outcomes is not unique to hypochondriasis/health anxiety.Indeed, prior research has shown that pretreatment levels ofdepression also predict worse outcomes for patients with OCD(Abramowitz, Franklin, Street, Kozak, & Foa, 2000; Stewart, Yen,Egan, & Jenike, 2006), a disorder that is related to hypochondria-sis/health anxiety (Abramowitz, 2005). Based on these findings,there have been efforts to develop and implement a treatmentprogram specifically for depressed OCD patients (Abramowitz,2004). Although the present findings did indicate that CBT for hy-pochondriasis/health anxiety also reduces depressive symptoms,the findings also suggest that patients with hypochondriasis/health

anxiety that present with severe depression may benefit fromsupplementary interventions that specifically target depressivesymptoms.

Examination of the effect size as a function of treatment as usualand waitlist controls revealed that CBT showed smaller effect sizeswhen compared to treatment as usual control conditions thanwhen compared to waitlist controls. One interpretation of thesefindings is that non-specific factors may influence treatmentoutcome in CBT for hypochondriasis/health anxiety to some degree.In fact, examination of the effect size for CBT compared to waitlistcontrols and that of CBT compared to treatment as usual controlconditions in the present investigation would suggest that thedifferences are far from negligible in magnitude. Examination ofthe extent to which other sample-specific and study-specificcharacteristics predicted outcome also revealed some importantfindings. For example, the CBT effect size for hypochondriasis/health anxiety did not vary as a function of blind assessment at posttreatment or follow-up.

More CBT sessions were also found to be associated with largereffect sizes at post-treatment. Although a relationship betweennumber of CBT treatment sessions and treatment efficacy has notbeen consistently observed in other disorders (Abramowitz, 1996;Rosa-Alc�azar, S�anchez-Meca, G�omez-Conesa, & Marín-Martínez,2008), this finding suggests that there may be added benefit tomore CBT sessions for patients with hypochondriasis/health anxi-ety. This finding does appear to be in linewith recent work showingthat treatment adherence (i.e. the number of completed treatmentmodules) in internet-based CBT for severe health anxiety signifi-cantly predicted outcome (Hedman, Andersson, et al., 2013;Hedman, Lindefors, et al., 2013). In addition to the number of ses-sions, future research is needed to examine if the length of CBTtreatment sessions is associated with outcomes as there may besome advantages to a more intensive approach to the treatment ofhypochondriasis/health anxiety.

Hypochondriasis is a chronic, disabling, prevalent and costlydisorder that has been generally viewed as refractory to treatment(Barsky& Ahern, 2004). Although the specific and problem focusednature of CBT has been the basis of concerns that it may notgeneralize to “real-life” patients who frequently present with co-occurring conditions (Westen, Novotny, & Thompson, 2004), thepresent meta-analytic investigation suggests that CBT is effective inreducing both disorder-specific and nonspecific (e.g., depression)symptoms for those with hypochondriasis/health anxiety. Thepresent investigation also identified some moderators of outcomethat may have prognostic and prescriptive implications. A prog-nostic variable is one that predicts outcome irrespective of thetreatment, whereas a prescriptive variable (often referred to as amoderator) predicts a different pattern of outcomes between twoor more treatment modalities (Hollon & Beck, 1986). Importantprescriptive indicators were observed in the present study thatmayguide future research. For example, the prescriptive indicatorsshowed that CBT is more effective than control conditions for pa-tients with more severe symptoms of hypochondriasis/healthanxiety but less effective for patients with more severe symptomsof depression. However, most RCTs that have been evaluated for thetreatment of hypochondriasis/health anxiety have been CBTfocused. One study did find that patients with hypochondriasiswho received CBT did significantly better on all measures relative tothe wait-list control group, and on a specific measure of healthanxiety compared with short-term psychodynamic psychotherapy(Sorensen, Birket-Smith, Watta, Buemann, & Salkovskis, 2011).Furthermore, the short-term psychodynamic psychotherapy groupdid not significantly differ from the waiting-list group on anyoutcome measures. More examination of the efficacy of otherpsychological treatments for hypochondriasis/health anxiety will

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e74 73

be valuable in identifying more substantive prescriptive variablesthat may inform individually tailored treatment.

Although the present meta-analysis represents a significantimprovement on prior work that was limited by a small number ofstudies, some of which were uncontrolled (Taylor et al., 2005;Thomson & Page, 2007), the use of only RCTs is perhaps anotherlimitation of the study as it resulted in the availability of a relativelysmall number of studies for analysis. Thus, examination of somemoderators (i.e., effects of blind assessors) in the meta-analysiswere potentially underpowered. Accordingly, some significantmoderators had relatively small beta weights and these findingsshould be interpreted with some caution. Another issue is that aninsufficient amount of studies provided information on variables(i.e., comorbidity percentage) that would allow for a morecomprehensive examination of potential moderators of outcome.This observation highlights the importance of more thoroughassessment and reporting of data in RCTs. Although the use of RCTsdoes allow for greater confidence in the present findings, moreRCTs of CBT for hypochondriasis/health will also allow for theidentification of more reliable moderators and mediators of treat-ment outcome. A recent study examining the mediating role ofputative mechanisms in internet-delivered CBT for severe healthanxiety found that reduced perceived risk of disease, less attentionto bodily symptoms, and reduced intolerance of uncertaintysignificantly mediated improvement in health anxiety (Hedman,Andersson, et al., 2013; Hedman, Lindefors, et al., 2013). Themediating effects of these processes are consistent with thecognitive-behavioral model represented in Fig. 1. Future researchdelineating additional mediating mechanisms is crucial becausesuchmechanisms provide important insight into the processes thatshould be targeted during treatment.

It is important to note that the very nature of RCTs may make itdifficult to detect meaningful moderators and mediators, especiallyif stricter eligibility criteria and overly rigid delivery of CBT pro-tocols eliminate the very processes that influence how patientsrespond to treatment. The use of more effectiveness trials in com-munity settings may complement RCTs in efforts to identify mod-erators and mediators of treatment outcome. Indeed, a confluenceof efficacy and effectiveness trials may ultimately reveal whattreatment, by whom, is most effective for patents with hypochon-driasis/health anxiety, and under which set of circumstances.Although the newly published DSM-5 has replaced the diagnosis ofhypochondriasis with illness anxiety disorder (APA, 2013), many ofthe underlying processes remain the same. Patients with illnessanxiety disorder experience heightened bodily sensations, areintensely anxious about the possibility of an undiagnosed illness,are not easily reassured, and engage in excessive safety behaviors tominimize anxiety about health concerns. Given the overlap insymptoms, CBT should prove to be equally as effective for illnessanxiety disorder as it is for hypochondriasis.

References3

Abramowitz, J. S. (1996). Variants of exposure and response prevention in thetreatment of obsessiveecompulsive disorder: a meta-analysis. Behavior Therapy,27, 583e600.

Abramowitz, J. S. (2004). Treatment of obsessive compulsive disorder in patientswho have comorbid depression. Journal of Clinical Psychology, 60, 133e1141.

Abramowitz, J. S. (2005). Hypochondriasis: conceptualization, treatment, andrelationship with obsessiveecompulsive disorder. Annals of Clinical Psychiatry,17, 211e217.

Abramowitz, J. S., Deacon, B. J., & Valentiner, D. P. (2007). The Short Health AnxietyInventory: psychometric properties and construct validity in a non-clinicalsample. Cognitive Therapy and Research, 31, 871e883.

3 *Indicates the studies used in the meta-analysis.

Abramowitz, J. S., Franklin, M. E., Street, G., Kozak, M., & Foa, E. B. (2000). Effects ofcomorbid depression on response to treatment for obsessiveecompulsive dis-order. Behavior Therapy, 31, 517e528.

Abramowitz, J. S., & Moore, E. L. (2007). An experimental analysis of hypochon-driasis. Behaviour Research and Therapy, 45, 413e424.

Abramowitz, J. S., Olatunji, B. O., & Deacon, B. J. (2007). Health anxiety, hypo-chondriasis, and the anxiety disorders. Behavior Therapy, 38, 86e94.

Abramowitz, J. S., Schwartz, S. A., & Whiteside, S. P. (2002). A contemporary con-ceptual model of hypochondriasis. Mayo Clinic Proceedings, 77, 1323e1330.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mentaldisorders (4th ed. text revision). Washington, D.C: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mentaldisorders (5th ed., rev.). Washington, DC: Author.

Barsky, A. (1996). Hypochondriasis: medical management and psychiatric treat-ment. Psychosomatics, 37, 48e56.

* Barsky, A. J., & Ahern, D. K. (2004). Cognitive behavior therapy for hypochon-driasis. A randomized trial. Journal of the American Medical Association, 291,1464e1470.

* Barsky, A. J., Ahern, D., Bauer, A., Nolido, N., & Orav, J. (2013). A randomized trial oftreatments for high-utilizing somatizing patients. Journal of General InternalMedicine, 28, 1396e1404.

Barsky, A., Barnett, M. C., & Cleary, P. D. (1995). Hypochondriasis and panic disorder:boundary and overlap. Archives of General Psychiatry, 51, 918e925.

Barsky, A., Fama, J., Bailey, E., & Ahern, D. K. (1998). A prospective 4- to 5-yearstudy of DSM-III-R hypochondriasis. Archives of General Psychiatry, 55,737e744.

Beck, A. T., & Emery, G. (2005). Anxiety disorders and phobias: A cognitive perspective(15th annual ed.). New York: Basic Books.

Bernstein, I. C., Callahan, W. A., & Jaranson, J. (1975). Lobotomy in private practice:long-term follow-up. Archives of General Psychiatry, 32, 1041e1047.

Beutler, L. E. (1991). Have all won and must all have prizes? Revisiting Luborskyet al'.s verdict. Journal of Consulting and Clinical Psychology, 59, 226e232.

Borenstein, M., Hedges, L. V., & Rothstein, H. (2007). Meta-analysis d Fixed effectsversus random effects. Biostat Inc. (Tech. Rep).

Bornstein, M., & Rothstein, H. (1999). Comprehensive meta-analysis: A computerprogram for research synthesis. New Jersey: Biosta.

* Bourgault-Fagnou, M. D., & Hadjistavropoulos, H. D. (2013). A randomized trial oftwo forms of cognitive behaviour therapy for an older adult population withsubclinical health anxiety. Cognitive Behaviour Therapy, 42, 31e44.

Buwalda, F. M., & Bouman, T. K. (2008). Predicting the effect of psychoeducationalgroup treatment for hypochondriasis. Clinical Psychology and Psychotherapy, 15,396e403.

* Buwalda, F. M., Bouman, T. K., & Van Duijn, M. A. J. (2006). Psychoeducation forhypochondriasis: a comparison of a cognitiveebehavioural approach and aproblemsolving approach. Behaviour Research and Therapy, 45, 887e899.

Clark, D. M., Salkovskis, P. M., Hackmann, A., Wells, A., Fennell, M., Ludgate, J., et al.(1998). Two psychological treatments for hypochondriasis. A randomisedcontrolled trial. British Journal of Psychiatry, 173, 218e225.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).Hillsdale, NJ: Erlbaum.

Creed, F., & Barsky, A. (2004). A systematic review of the epidemiology of somati-sation disorder and hypochondriasis. Journal of Psychosomatic Research, 56,391e408.

Driessen, E., Cuijpers, P., Hollon, S. D., & Dekker, J. J. (2010). Does pretreatment severitymoderate the efficacy of psychological treatment of adult outpatient depression?A meta-analysis. Journal of Consulting and Clinical Psychology, 78, 668e680.

Eddy, K., Dutra, L., Bradley, R., & Westen, D. (2004). A multidimensional meta-analysis of pharmacotherapy for obsessiveecompulsive disorder. Clinical Psy-chology Review, 24, 1011e1030.

Fallon, B., Javitch, J., Hollander, E., & Liebowitz, M. (1992). Hypochondriasis andobsessive compulsive disorder: overlaps in diagnosis and treatment. Journal ofClinical Psychiatry, 52, 457e460.

Greeven, A., van Balkom, A. J., van der Leeden, R., Merkelbach, J. W., van denHeuvel, O. A., & Spinhoven, P. (2009). Cognitive behavioral therapy versusparoxetine in the treatment of hypochondriasis: an 18-month naturalisticfollow-up. Journal of Behavior Therapy and Experimental Psychiatry, 40, 487e496.

* Greeven, A., van Balkom, A. J., Visser, S., Merkelbach, J. W., van Rood, Y. R., vanDyck, R., et al. (2007). Cognitive behavior therapy and paroxetine in the treat-ment of hypochondriasis: a randomized controlled trial. American Journal ofPsychiatry, 164, 91e99.

Gropalis, M., Bleichhardt, G., Witth€oft, M., & Hiller, W. (2012). Hypochondriasis,somatoform disorders, and the anxiety disorders: sociodemographic variables,general psychopathology, and naturalistic treatment effects. Journal of Nervousand Mental Disease, 200, 406e412.

Gureje, O., Ustun, T. B., & Simon, G. E. (1997). The syndrome of hypochondriasis: across-national study in primary care. Psychological Medicine, 27, 1001e1010.

Hartman, T. C., Borghuis, M., Lucassen, P., van de Laar, F., Speckens, A., & van Weel, C.(2009). Medically unexplained symptoms, somatisation disorder and hypo-chondriasis: course and prognosis. A systematic review. Journal of Psychoso-matic Research, 66, 363e377.

Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. London: Aca-demic Press.

Hedman, E., Andersson, E., Andersson, G., Lindefors, N., Lekander, M., Rück, C., et al.(2013). Mediators in internet-based cognitive behavior therapy for severehealth anxiety. PLoS ONE, 8(10), e77752.

B.O. Olatunji et al. / Behaviour Research and Therapy 58 (2014) 65e7474

* Hedman, E., Andersson, E., Andersson, G., Lj�otsson, B., Rück, C., Asmundson, G.,et al. (2011). Internet-based cognitive-behavioural therapy for severe healthanxiety: randomized controlled trial. British Journal of Psychiatry, 198,230e236.

Hedman, E., Lindefors, N., Andersson, G., Andersson, E., Lekander, M., Rück, C., et al.(2013). Predictors of outcome in Internet-based cognitive behavior therapy forsevere health anxiety. Behaviour Research and Therapy, 51, 711e717.

Hofmann, S. G., & Smits, J. A. (2008). Cognitive-behavioral therapy for adult anxietydisorders: a meta-analysis of randomized placebo-controlled trials. Journal ofClinical Psychiatry, 69, 621e632.

Hollon, S. D., & Beck, A. T. (1986). Cognitive and cognitive-behavior therapies. InS. L. Gartield, & A. E. Bergin (Eds.), Handbook of psychotherapy and behaviorchange (Vol. III); (pp. 443e482). New York: Wiley.

House, A. (1989). Hypochondriasis and related disorders: assessment and man-agement of patients referred for a psychiatric opinion. General Hospital Psy-chiatry, 11, 156e165.

Katon, W. J., & Walker, E. A. (1998). Medically unexplained symptoms in primarycare. Journal of Clinical Psychiatry, 59, 15e21.

Ladee, G. A. (1966). Hypochondriacal syndromes. Amsterdam: Elsevier.Lesse, S. (1980). Masked depression-the ubiquitous but unappreciated syndrome.

Psychiatric Journal of the University of Ottawa, 5, 268e273.Marcus, D. K., Gurley, J. R., Marchi, M. M., & Bauer, C. (2007). Cognitive and

perceptual variables in hypochondriasis and health anxiety: a systematic re-view. Clinical Psychology Review, 27, 127e139.

* McManus, F., Surawy, C., Muse, K., Vazques-Montes, M., & Williams, M. G. (2012).A randomized clinical trial of mindfulness-based cognitive therapy versus un-restricted services for health anxiety (hypochondriasis). Journal of Consultingand Clinical Psychology, 80, 817e828.

Nakao, M., Shinozaki, Y., Ahern, D. K., & Barsky, A. J. (2011). Anxiety as a predictor ofimprovements in somatic symptoms and health anxiety associated with cog-nitiveebehavioral intervention in hypochondriasis. Psychotherapy and Psycho-somatics, 80, 151e158.

Noyes, R., Kathol, R. G., Fisher, M. M., Phillips, B. M., Suelzer, M. T., & Woodman, C. L.(1994). Psychiatric comorbidity among patients with hypochondriasis. GeneralHospital Psychiatry, 16, 78e87.

Olatunji, B., Davis, M., Powers, M., & Smits, J. (2013). Cognitive-behavioral therapyfor obsessiveecompulsive disorder: a meta-analysis of treatment outcome andmoderators. Journal of Psychiatric Research, 47, 33e41.

Olatunji, B. O., Deacon, B. J., & Abramowitz, J. S. (2009). Is hypochondriasis ananxiety disorder? British Journal of Psychiatry, 194, 481e482.

Orwin, R. G. (1983). A fail safe N for effect size in meta-analysis. Journal forEducational Statistics, 8, 157e159.

Paul, G. L. (1967). Strategy of outcome research in psychotherapy. Journal ofConsulting Psychology, 31, 109e118.

Rosa-Alc�azar, A. I., S�anchez-Meca, J., G�omez-Conesa, A., & Marín-Martínez, F. (2008).Psychological treatment of obsessiveecompulsive disorder: a meta-analysis.Clinical Psychology Review, 28, 1310e1325.

Rosenthal, R. (1979). The “file drawer problem” and tolerance for null results.Psychological Bulletin, 86, 638e641.

Rosenthal, R. (1991).Meta-analytic procedures for social research (rev. ed.). ThousandOaks, CA, US: Sage Publications, Inc.

* Seivewright, H., Green, J., Salkovskis, P., Barrett, B., Nur, U., & Tyrer, P. (2008).Cognitive-behavioural therapy for health anxiety in a genitourinary medicineclinic: randomised controlled trial. British Journal of Psychiatry, 193, 332e337.

Simon, G. E., Gureje, O., & Fullerton, C. (2001). Course of hypochondriasis in aninternational primary care study. General Hospital Psychiatry, 23, 51e55.

Smits, J. A. J., Hofmann, S. G., Rosenfield, D., DeBoer, L. B., Costa, P., Simon, N. M.,et al. (2013). d-cycloserine augmentation of cognitive behavioral group therapyof social anxiety disorder: prognostic and prescriptive variables. Journal ofConsulting & Clinical Psychology, 81, 1100e1112.

Smits, J. A. J., Minhajuddin, A., Thase, M. E., & Jarrett, R. J. (2012). Outcomes of acutephase cognitive therapy in outpatients with anxious versus nonanxiousdepression. Psychotherapy & Psychosomatics, 81, 153e160.

* Sorensen, P., Birket-Smith, M., Wattar, U., Buemann, I., & Salkovskis, P. (2011).A randomized clinical trial of cognitive behavioural therapy versus short-termpsychodynamic psychotherapy versus no intervention for patients with hypo-chondriasis. Psychological Medicine, 41, 431e441.

* Speckens, A. E. M., van Hemert, A. M., Spinhoven, P., Hawton, k. E., Bolk, J. H., &RooiJmans, H. (1995). Cognitive behavioural therapy for medically unexplainedphysical symptoms: a randomized controlled trial. BMJ, 311, 1328e1332.

Speckens, A. E. M., Spinhoven, P., van Hemert, A. M., Bolk, J., & Hawton, K.(1997). Cognitive behavioural therapy for unexplained physical symptoms;process and prognostic factors. Behavioural and Cognitive Psychotherapy, 25,291e294.

Stewart, S. E., Yen, C. H., Egan, S. D., & Jenike, M. (2006). Outcome predictors forsevere obsessiveecompulsive patients in intensive residential treatment. Jour-nal of Psychiatric Research, 40, 511e519.

* Sumathipala, A., Siribaddana, S., Abeysingha, M. R. N., De Silva, P., Dewey, M.,Prince, M., et al. (2008). A randomised controlled trial of cognitive behaviouraltherapy versus structured care for medically unexplained symptoms: a ran-domized controlled trial. British Journal of Psychiatry, 193, 51e59.

Taylor, S., & Asmundson, G. (2004). Treating health anxiety: A cognitive-behavioralapproach. New York: Guilford Press.

Taylor, S., Asmundson, G., & Coons, M. (2005). Current directions in the treatment ofhypochondriasis. Journal of Cognitive Psychotherapy: An International Quarterly,19, 285e304.

Thomson, A. B., & Page, L. A. (2007). Psychotherapies for hypochondriasis. CochraneDatabase of Systematic Reviews, 4, 1e43.

* Visser, S., & Bouman, T. K. (2001). The treatment of hypochondriasis: exposureplus response prevention vs. cognitive therapy. Behaviour Research and Therapy,39, 423e442.

Walker, J., Vincent, N., Furere, P., Cox, B., & Kjernisted, K. (1999). Treatment pref-erence in hypochondriasis. Journal of Behavior Therapy and Experimental Psy-chiatry, 30, 251e258.

* Warwick, H. M. C. (1996). A controlled trial of cognitive-behavioural treatment ofhypochondriasis. British Journal of Psychiatry, 169, 189e195.

Warwick, H. M. C., Clark, D. M., Cobb, A. M., & Salkovskis, P. M. A. (1996). Controlledtrial of cognitive-behavioural treatment of hypochondriasis. British Journal ofPsychiatry, 169, 189e195.

Warwick, H., & Salkovskis, P. (1990). Hypochondriasis. Behaviour Research andTherapy, 28, 105e117.

Warwick, H., & Salkovskis, P. (2001). Cognitive-behavioral treatment of hypo-chondriasis. In V. Starcevic, & D. R. Lipsitt (Eds.), Hypochondriasis: Modern per-spectives on an ancient malady (pp. 314e328). New York: Oxford UniversityPress.

Weck, F., Neng, J. M., Richtberg, S., & Stangier, U. (2012). Dysfunctional beliefs aboutsymptoms and illness in patients with hypochondriasis. Psychosomatics, 53,148e154.

Westen, D., Novotny, C. M., & Thompson-Brenner, H. (2004). The empirical status ofempirically supported psychotherapies: assumptions, findings, and reporting incontrolled clinical trials. Psychological Bulletin, 130, 631e663.