Dr. Torres is with the Department of Neurology and Psychiatry, Botucatu Medical School, Universidade Estadual Paulista (UNESP), Distrito de Rubião Jr., Botucatu (São Paulo), Brazil, 18618-970 (e-mail: torresar@fmb.unesp.br). Dr. Prince, Dr. Bhugra, Dr. Farrell, and Dr. Jenkins are with the Institute of Psychiatry, King's College, London. Dr. Bebbington is with the Department of Mental Health Sciences, University College, London. Dr. Brugha and Dr. Meltzer are with the Department of Health Sciences, University of Leicester. Dr. Lewis is with the Division of Psychiatry, University of Bristol. Ms. Singleton is with the Home Office, London.

Many people with obsessive-compulsive disorder (OCD) avoid or delay seeking help (1), perhaps because of embarrassment about their symptoms (2). The literature about help seeking among persons with OCD is scant (3), particularly findings from epidemiological studies. Women and persons with obsessions usually predominate in community samples (4), whereas most clinical samples have an equal gender distribution and a predominance of persons who have both obsessions and compulsions (5), indicating a possible differential pattern of treatment seeking.

Having a comorbid psychiatric disorder usually increases the likelihood of treatment among individuals with mental disorders, including OCD (1,6). However, the literature lacks information about the adequacy of management for those who have obtained services. Recent good-quality clinical trials and meta-analyses have informed evidence-based recommendations for the management of OCD. Serotonergic antidepressants, such as clomipramine, and selective serotonin reuptake inhibitors (SSRIs) are the drugs consistently shown to be the most effective in the treatment of OCD (7,8). There is also consistent evidence of the effectiveness of behavioral and cognitive approaches, currently considered the treatments of choice for OCD (9,10).

Our aim was to analyze the use of services by people with OCD from a nationally representative sample and compare it with use by those with other neuroses. We also wanted to determine possible differences in services used between genders, individuals with different types of the disorder (only obsessions, only compulsions, or both), and those with or without a comorbid anxiety or depressive disorder. We were also interested in evaluating the appropriateness of therapeutic approaches for individuals who were receiving treatment. We hypothesized that men, those with both obsessions and compulsions, and those with at least one co-occurring neurotic disorder would be more likely to receive treatment.

The major design features of the British Survey of Psychiatric Morbidity of 2000 are described here, and further methodological details are available elsewhere (11,12). Ethical approval for the survey was obtained from the London Multi-Centre Research Ethics Committee (a National Health System body), which considered all aspects of the survey, including the consent procedures and documents for participants. Because this was a general household survey, verbal consent only was considered sufficient. However, the interviewers and all documentation for respondents stressed that participation was voluntary and that the respondent could refuse to answer any questions or terminate the interview at any time.

Sample

The survey covered persons aged 16 to 74 years living in private households in England, Wales, and Scotland. Primary sampling units were 438 post-code sectors selected from the Postcode Address File and stratified by region and socioeconomic group. Thirty-six addresses were systematically selected per sector. Only one eligible person was selected at random per household by using the Kish grid method (13). The individuals could freely refuse to participate in the survey, and the response rate was 69.5% (8,886 cooperating adults from 12,792 eligible persons, and 8,580 complete interviews).

Instruments

The instrument used to assess neurotic disorders was the revised version of the Clinical Interview Schedule (CIS-R) (14,15), which is made up of subsections covering 14 different symptom clusters. Initial filter questions in each section establish the existence of a particular symptom in the past month, leading to a more detailed assessment focusing on the past week. Each symptom with a score of 2 or higher was considered of relevant severity (range for each symptom, 0 to 4 or 5). The overall threshold score for significant psychiatric morbidity is 12 or above (summed scores from all 14-symptom sections range between 0 and 57). Additional questions enable application of ICD-10 diagnostic criteria (16) by use of computerized algorithms. Six diagnostic categories were obtained: OCD, generalized anxiety disorder, depressive episode, phobias, panic disorder, and mixed anxiety and depressive disorder (11).

CIS-R has eight questions about obsessions and compulsions. The obsession score reflects the respondent's experience in the past week of repetitive unpleasant thoughts or ideas, and the compulsion score reflects the experience of doing something over and over again. One point was assigned for each of four responses: the symptoms were present on four or more days, the respondent tried to stop thinking any of these thoughts or repeating the behaviors, the symptoms made the respondent upset or annoyed, and the episode of obsession lasted at least 15 minutes or a behavior was repeated at least three times.

The algorithm for generating an ICD-10 diagnosis of OCD required a duration of two weeks or longer, at least one act or thought resisted, social impairment, and overall scores of four for obsessions or compulsions or at least six for obsessions and compulsions (17). The validity of the ICD-10 diagnosis generated by the CIS-R has never been formally established, because the number of cases of OCD in general validations of the measure has been inadequate to draw robust conclusions.

Data collection

Trained nonclinical interviewers carried out the initial structured interviews, which were in-person and computer assisted. Interviews were completed with 8,580 individuals. The full assessment of OCD and other neurotic disorders was made in the first stage, as was collection of all data about service use.

Statistical analyses

Statistical analyses were conducted using Stata version 8 software (18). Given the multistage stratified sampling design, the data were weighted, first to take account of differing selection probabilities and second to compensate for nonresponse by using poststratification. To take account of the complex sample design and weighting procedures used in this survey, all estimates of prevalence and association were made using the appropriate Stata survey commands, generating robust standard errors. Use of health services by persons with OCD and those with other neuroses was compared by using the chi square test. A conventional p value of .05 was considered as the level of statistical significance.

Comparisons between persons with different types of OCD (with only obsessions, only compulsions, or both) and between individuals with OCD with and without a comorbid disorder were also made by using the chi square test or Fisher's exact test, when necessary. A multiple logistic regression analysis was conducted to adjust the association between receipt of treatment and having OCD (compared with other neuroses) and having comorbid OCD (compared with noncomorbid OCD) for age, gender, level of income, and marital and employment status, as well as some clinical factors.

A total of 114 survey respondents with OCD (74 women) were identified, which yielded a weighted one-month prevalence of 1.1%. Among the 114 respondents with OCD, 61 (55%) had purely obsessive symptoms, 14 (11%) had purely compulsive symptoms, and 39 (34%) had both obsessions and compulsions. The respondents with OCD did not differ significantly in ethnicity from those with other neuroses. Among those with OCD 103 (88%) were Caucasian, four (3.6%) were of African descent, and five (8.9%) were Indian or Pakistani. Among those with other anxiety or depressive disorders, 1,292 (93%) were Caucasian, 27 (1.8%) were of African descent, 29 (2.5%) were Indian or Pakistani, and 35 (2.6%) had other ethnicities.

Seventy-six individuals with OCD (62%) had at least one comorbid neurosis: 47 (37%) had a depressive episode, 40 (31%) had generalized anxiety disorder, 25 (22%) had agoraphobia or panic disorder, 19 (17%) had social phobia, and 20 (15%) had specific phobia (19). Some participants had more than one comorbid disorder.

Service use by respondents with OCD

As shown in Table 1, few respondents with OCD maintained contact with specialist psychiatric or psychological services. Eight percent had made an outpatient visit to a mental health center in the past three months, and in the past year 9% had seen a psychiatrist and 5% had seen a psychologist. Lifetime contacts were more frequent; 20% reported an admission to a psychiatric inpatient service at some time in their lives. Recent contacts with community services were also more common; 62% had spoken to their general practitioner about an emotional problem in the past year, and 27% had used some kind of community service (that is, day treatment, a self-help or support group, or home care).

Treatment of OCD

As shown in Table 1, 40% of those with OCD were receiving some kind of treatment. Twenty percent were receiving medication only, 5% were receiving counseling or psychotherapy only, and 15% were receiving both types of treatment. However, there was scant use of evidence-based therapies. Only 2% of those with OCD were using SSRI antidepressants, compared with 10% taking anxiolytics or hypnotics and 7% taking antipsychotics. Of the 20% receiving counseling or psychotherapy, only 5% were receiving cognitive-behavioral therapy.

Comparison of service use

Persons with OCD were approximately twice as likely as those with other neuroses to be receiving some kind of intervention for a mental health condition (40% compared with 23%, p<.001) (Table 1). This estimate of association (crude odds ratio [OR]=2.2, 95% confidence interval [CI]=1.4—3.5) was almost unchanged after the analysis adjusted for age, gender, level of income, and marital and employment status (adjusted OR=2.0, CI=1.2—3.3). Those with OCD were more likely than those with other neuroses to be receiving psychotropic medication (35% compared with 19%, p<.001) and counseling or psychotherapy (20% compared with 8%, p<.001) and also much more likely to have been hospitalized for mental health treatment at some time in their life (20% compared with 6%, p<.001). Those with OCD were also more likely to be in contact with mental health professionals of all types, except social workers.

Factors predicting service contact and receipt of treatment

Persons with OCD and a comorbid disorder were significantly more likely to be receiving treatment than those with OCD alone (56% compared with 14%, p<.001) (Table 1). They were more likely to have visited a mental health center as outpatients in the three months before the interview, more likely to be taking medication and to be receiving counseling or psychotherapy. Self-help or support groups were used more by persons with OCD without comorbidity. In our sample, 36% of those with only obsessions (N=61), 44% with only compulsions (N=14), and 44% with obsessions and compulsions (N=39) were receiving treatment—a difference that was not significant.

There was only a trend (a difference that did not reach statistical significance) for older people with OCD to be more likely to be receiving treatment, but other sociodemographic factors were not associated with treatment receipt. No significant gender differences were found, except that men with OCD were more likely than women to be receiving cognitive-behavioral therapy (four men, compared with only one woman, p=.046). In a univariate analysis, comorbidity with other neurotic disorders, overall severity of psychological morbidity (as indexed by a CIS-R total score greater than 18), and ever having had suicidal thoughts were all strongly associated with being in treatment.

In a multivariate analysis the association of treatment receipt with comorbidity (crude OR=7.8, CI=2.9—22.1) was largely independent of severity of psychological morbidity and also remained unaltered after the analysis adjusted for age, gender, level of income, and marital and employment status (adjusted OR=6.2, CI=2.0—19.8). In fact, having a comorbid disorder was the only clinical factor that remained associated with receipt of treatment after the logistic regression. Severity of psychological morbidity (CIS-R total score greater than 18) and lifetime history of suicidal thoughts were no longer associated with receipt of treatment after adjustment for other clinical factors, including presence of comorbid neuroses.

This analysis used data from a carefully designed and conducted survey of a representative household sample of adults from all areas in England, Scotland, and Wales. A comprehensive structured diagnostic assessment was coupled with a careful investigation of contact with a wide variety of services, including primary and secondary health and mental health care services, together with receipt of pharmacological and nonpharmacological interventions. Considering that ICD-10 and DSM-IV criteria for neuroses in general are quite similar, we believe that our results can be compared to those from U.S. studies.

The study has some limitations that must be highlighted. First, the validity of the ICD-10 diagnosis as generated by CIS-R has never been formally established. Second, the structured diagnostic assessments were carried out by lay interviewers, the CIS-R OCD questions could lead to overdiagnosis, and the diagnosis was not validated with a systematic clinical reappraisal (20). Third, the relatively small sample of OCD patients limited the statistical power for some subsample analyses. Fourth, the specific contents of obsessive-compulsive symptoms, which may play a role in treatment seeking (1,2), were not assessed. Finally, the information provided by respondents was not compared with records from the health services and it is unclear how generalizable our findings are to other countries and cultures.

Previous studies have reported low rates of treatment seeking by persons with neuroses in general (17,21) and anxiety disorders in particular (22). Consistent with other studies (23) our findings indicate that those with OCD are relatively high-intensity users of mental health services, receiving significantly more psychological and pharmacological treatments than participants with other neuroses.

Despite this finding, most people with OCD in the community are not being treated (3). In our survey, fewer than 10% had been seen by a mental health professional in the year before the interview. The limited receipt of specific evidence-based treatments for OCD, particularly cognitive-behavioral therapy (5% of respondents with OCD) and SSRI antidepressants (2%) suggests considerable deficiencies at one or more stages in the process from help seeking to treatment. Those with OCD were three times as likely as those with other neuroses to have ever been admitted to an inpatient mental health facility. Again, the fact that one in five of those with OCD had been hospitalized reflects in part the severity of the disorder but also suggests that clinicians fail to detect, diagnose, and manage the disorder effectively.

There are certainly important questions regarding the readiness of persons with OCD to seek treatment (24). They tend to keep their symptoms hidden from others, because the symptoms are usually egodystonic and generate considerable embarrassment. They may also be unaware of the pathological nature of the problem or of the availability of effective treatments (25,26). Previous surveys (1,27,28) reported that only about a third of persons with OCD had ever disclosed their problem to a doctor, despite the distress and impairment associated with the disorder. Reluctance to talk about the symptoms may also be due to fear of being considered "insane," fear of hospitalization or even arrest, and fear that talking about the problems may make them happen. Avoidance because of fears of contamination or other obsessions may also prevent some persons from seeking services (29). Some may consult doctors only for secondary complaints, such as skin problems from washing compulsions (2).

An important aspect to be taken into account is that OCD severity can influence treatment-seeking behaviors and that community studies are more likely to include mild cases (or even false-positive ones), which may partly explain the fact that a majority of persons with OCD were not in treatment. The availability of treatments also has to be considered. Cognitive-behavioral therapy is available in many of the National Health Service mental health trusts and several of the primary health care services in Great Britain; it is conducted either by a psychologist or psychology nurses with training in this approach. However, there are concerns about unmet needs for cognitive-behavioral therapy (30).

We have no direct data on the extent of underrecognition of OCD. However, despite the difficulties referred to above, 62% of those with OCD in our survey had talked to a general practitioner about emotional problems in the past year. This finding implies a missed opportunity for engagement in many cases. To seize this opportunity, general practitioners would need to go beyond reliance on spontaneous disclosure of obsessive-compulsive symptoms. The average gap between the onset of OCD and initiation of appropriate treatment was found in one study to be as long as 17 years (31), partly because of misdiagnosis related to the presence of co-occurring conditions. Of course, even direct disclosure of the problem to a health professional does not guarantee that it will be recognized accurately or that the patient will be referred for appropriate treatment (32).

An encouraging finding in the study reported here is the apparent increase in receipt of treatment by persons with OCD compared with the last British National Psychiatric Morbidity Survey—40% in 2000 compared with 19% in 1993 (33). Consistent with much other research (1,3,24,34), in our analysis the major factors predicting receipt of treatment for those with OCD were the overall severity of psychological morbidity and the presence of comorbid mental disorders. It should be noted that these findings are not unique to OCD but apply to common mental disorders in general (21,34). However, although correlated, these were strong and independent predictors, arguing against the previous suggestion that the association between comorbidity and treatment is merely confounded by OCD severity (1). The more parsimonious explanation is that the comorbid condition may provide a more acceptable indication for seeking help. Of course, it may be that those with a comorbid disorder then disclose the symptoms of that condition rather than their obsessions and compulsions (35) and that despite the higher levels of service contact, the diagnosis of OCD is still not being made. This inference is supported by the relatively low rates of evidence-based therapies received by this group. However, the treatment gap was much more striking for persons with OCD and no comorbid mental disorders, none of whom had visited a mental health service in the three months before the interview, whereas only 1% had seen a psychiatrist in the previous year and none was receiving cognitive-behavioral therapy or SSRI antidepressants.

Previous suggestions that among those with OCD, women, the unemployed, and Caucasians were more likely to be receiving treatment (3) were not borne out in our analysis. Mayerovitch and colleagues (1) reported that persons who had only compulsions had especially low rates of treatment. In our sample, persons with the three different clinical types of OCD (only obsessions, only compulsions, or both) did not differ significantly in rates of contact with any category of service or receipt of any type of treatment. Our sample may have been too small to detect significant differences between these three groups. Nevertheless, fewer individuals with only obsessions were in treatment. The direction of this finding could help to explain the observation that persons with obsessions alone predominate in most community samples whereas clinical populations have more mixed symptoms (4).

Our results confirm the existence of a considerable number of untreated individuals with OCD in the community in Great Britain. A majority of individuals who had sought help were being seen by nonpsychiatric physicians, which is consistent with findings of other studies (3,33,35). Although persons with OCD had higher rates of treatment than their counterparts in the previous British survey (17) and than individuals with other neurotic disorders, most persons with OCD are still not receiving treatment, especially by mental health professionals. Enhancing public awareness about the condition and available effective treatments is warranted.

Comorbidity with other neuroses, which occurs in a majority of cases of OCD, substantially increases treatment-seeking behaviors (24) but may also obscure diagnosis of OCD (31,35). Only a minority of persons with the disorder were receiving psychological or pharmacological approaches currently considered more effective, such as cognitive-behavioral therapy or SSRIs. The condition may be going undetected, because patients find it more acceptable to describe other co-occurring emotional symptoms. They may be concealing their OCD symptoms, resulting in inappropriate treatment strategies and prolonging suffering and impairment. Therefore, primary care professionals should be aware of the high rates of comorbidity associated with OCD and should directly inquire about the occurrence of obsessions and compulsions when a patient presents with depressive or anxiety symptoms.

Identification of OCD is a crucial step to initiate appropriate treatment, which can lead to considerable relief from suffering and impairment.

The survey was carried out by the Social Survey Division of the Office for National Statistics on behalf of the Department of Health, the Scottish Executive, and the National Assembly for Wales. Dr. Torres received a postdoctoral scholarship from CAPES Foundation (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) of the Ministry of Education of Brazil to develop this study at the Institute of Psychiatry, King's College.

The authors report no competing interests.