sign in sign up
<<
Home , Emil Kraepelin

head to head

Is underdiagnosis Daniel J Smith clinical senior lecturer in , of the definition of to include Cardiff University School of Medicine, Cardiff in 1994 and marketing of new drugs [email protected] the main pitfall Nassir Ghaemi professor of psychiatry, Tufts University School have begun to address a century of relative of Medicine, Boston, Massachusetts, USA neglect; yet, perhaps predictably, objections about overdiagnosis have been raised. when diagnosing Bipolar disorder is a complex condition, and patients can Defining misdiagnosis bipolar disorder? present with the entire range Let’s start by agreeing on some definitions. of psychiatric symptoms.1 Its underdiagnosis Overdiagnosis implies that a disorder is Daniel Smith and has always been, and continues to be, a major frequently diagnosed when absent, as problem. well as when present. Underdiagnosis, Nassir Ghaemi believe that Ever since Emil Kraepelin defined manic conversely, implies that a disorder is often not many people with bipolar depressive insanity as recurrent mood diagnosed when present and also infrequently episodes (either or , but not diagnosed when absent. Underdiagnosis and disorder remain undiagnosed, necessarily requiring mania) in 1898,2 the overdiagnosis are claims of validity (based on a condition has been underdiagnosed. For almost definitive diagnosis from pathological findings but Mark Zimmerman a century it was commonly misdiagnosed or, in psychiatry, a formal research diagnostic argues that overdiagnosis is as .3 In 1980, the American interview5) not reliability (whether clinicians Diagnostic and Statistical Manual (DSM-III) agree on what they have diagnosed). We should the bigger problem narrowed the definition of schizophrenia and therefore start with certain knowledge about divided manic depressive insanity into two diagnosis (from a research interview, not groups: broadly defined major depressive clinicians’ diagnoses) and assess the accuracy disorder and narrowly defined bipolar of earlier diagnoses, not the other way around. disorder.4 Reasons for underdiagnosis since Stated simply, poor reliability means only that then include lack of insight into mania by clinicians could not agree on the diagnosis. It patients, lack of systematic assessment of says little about true rates of misdiagnosis. mania by clinicians, stigma, and the aggressive In Zimmerman and colleagues’ widely marketing of antidepressants. The broadening cited study,6 43% (63/145) of patients who

Mark Zimmerman director, outpatient psychiatry, Rhode phenomenon in bipolar disorder—clinician Island Hospital, Bayside Medical Building, 235 Plain Street, overdiagnosis in patients without a history of a Providence, RI 02905, USA [email protected] manic or hypomanic episode.

Overdiagnosis is the bigger problem Bipolar disorder is a serious We therefore conducted a study to determine illness resulting in significant how often bipolar disorder might be psychosocial morbidity and overdiagosed and underdiagnosed.15 We excess mortality. Research reports, reviews, interviewed 700 psychiatric outpatients with and commentaries have suggested that the structured clinical interview for DSM-IV bipolar disorder is under-recognised, and (SCID), a widely used, validated, diagnostic that many depressed patients have, in fact, interview. About a fifth of the sample reported bipolar disorder.1‑4 The treatment and clinical having had bipolar disorder diagnosed (145, implications of the failure to recognise 21%), significantly higher than the number we bipolar disorder in depressed patients diagnosed with the structured interview (90, include the underprescription of mood (13%), P<0.001). Only 63 (43%) of the 145

sp l stabilising drugs, an increased risk of rapid patients who reported a previous diagnosis of / cycling, and increased costs of care.5‑8 bipolar disorder had the condition diagnosed brown During the past decade, my hospital by the structured interview. Bipolar disorder

l l pau has introduced semistructured diagnostic was also underdiagnosed in some patients, interviews into outpatient clinical practice. but three times as many patients had been Use of these interviews initially documented overdiagnosed as had been underdiagnosed underdetection of psychiatric comorbidity,9‑11 (82 v 27). and these findings were replicated in other Supporting the validity of our diagnoses of settings.12‑14 However, in recent years, we bipolar disorder, we found that the patients observed the emergence of an opposite who were previously wrongly diagnosed All references are in the version on bmj.com with bipolar disorder were significantly less “Overdiagnosing bipolar disorder likely to have a first degree relative with WHERE DO YOU STAND ON THE ISSUE? can unnecessarily expose patients Tell us on bmj.com bipolar disorder than patients who were to serious drug side effects” accurately diagnosed with bipolar disorder.

686 BMJ | 27 march 2010 | Volume 340 head to head

answered “yes” to a questionnaire item (“Have reliability in this study was reported as evidence “Studies that have reported you been diagnosed with bipolar or manic of overdiagnosis when in fact the (validity) data overdiagnosis actually describe 6 depressive disorder by a healthcare pro­ suggest underdiagnosis. only low reliability” fessional?”) subsequently­ had bipolar disorder We have found no study reporting diagnosed by formal research interview. overdiagnosis that started with definitive underdiagnosis of bipolar disorder. This suggests only that the earlier diagnoses diagnoses and compared these with earlier Low diagnostic reliability is not unique were unreliable (which is unsurprising given clinical diagnoses. Studies that have reported to bipolar disorder because all psychiatric that these diagnoses were made by different overdiagnosis in substance abuse7 8 or diagnoses have, at best, only moderate health professionals using different diagnostic adolescents9 actually describe only low reliability14 and complex psychiatric criteria over long periods of time). This kind of reliability because, as with the Zimmerman disorders may be difficult to diagnose disagreement among clinicians and researchers study, they start with a relatively unreliable without necessarily implying overdiagnosis. does not prove overdiagnosis because it does measure of diagnosis (routine clinical practice) Diagnostic disagreement is common not begin with findings on validity. and then compare this with the formal research in other disorders that are known to be assessment. underdiagnosed—for example, asthma,15 Validity and reliability In contrast, four studies that used the ,16 chronic cholecystitis,17 and We carried out the required validity analysis validity approach all identified underdiagnosis coeliac .18 These are all, like bipolar based on the research interview data from (40-67% of patients with bipolar disorder were disorder, unreliably diagnosed. that same study. According to the formal previously misdiagnosed).10‑13 These include In summary, although clinicians’ research interview, 90 patients had bipolar detailed case series studies of adults with diagnoses of bipolar disorder can be disorder and 610 did not.6 Thirty per cent (27) mood disorders10 11 and substance abuse12 and unreliable, this does not contradict the of patients with true bipolar disorder had not in children.13 Although sample sizes in these reality that if patients truly have bipolar been previously diagnosed whereas only 13% studies were smaller (pooled n=271 versus disorder they are more likely to be missed (82) of those without bipolar disorder had n=700 in the Zimmerman study), they were not than correctly diagnosed. Underdiagnosis, previously had the condition diagnosed. In less rigorous in design because all misdiagnosis not overdiagnosis, remains the major effect, bipolar disorder had been missed more studies must retrospectively assess previous problem for bipolar disorder within than twice as frequently as it was mistakenly diagnoses. Taken together, these studies (as contemporary clinical practice. diagnosed (relative risk 2.23, 95% confidence well as the Zimmerman study when fully and Competing interests: None declared. interval 1.53 to 3.25). Unfortunately, low correctly analysed) are strong evidence of Cite this as: BMJ 2010;340:c854

The proportion with affected relatives in bipolar disorder. Consistent with other this is followed by a discussion of methods the overdiagnosed patients was the same as studies, we found that nearly one third of the clinicians can use to improve the detection of in patients who never had bipolar disorder patients we diagnosed with bipolar disorder the disorder. These discussions of diagnostic diagnosed. were previously undiagnosed. practice are usually not balanced by a Although other studies have reported However, most discussions of the summary of studies of showing overdiagnosis overdiagnosis of bipolar disorder, we are misdiagnosis of bipolar disorder have and the risks associated with overdiagnosis. aware of only one other study with data on focused on the personal and societal costs Because clinicians are probably inclined both overdiagnosis and underdiagnosis. of underdiagnosis. Only occasionally to diagnose disorders that they feel more Hirschfeld and colleagues16 interviewed have authors discussed the possible comfortable treating, when confronted with 180 depressed primary care outpatients problems associated with overdiagnosis. patients with mood instability who do not receiving antidepressant drugs with the Unnecessary side effects are a potentially meet criteria for a hypomanic episode, doctors structured clinical interview for DSM-IV. Forty serious consequence of overdiagnosis. may nonetheless diagnose a potentially drug three patients reported a prior diagnosis of Mood stabilisers are the treatment of choice responsive disorder such as bipolar disorder bipolar disorder, and this diagnosis was not for bipolar disorder and, depending on the rather than a disorder such as borderline confirmed in 14 (33%). The overdiagnosis drug, can affect renal, endocrine, hepatic, that is less responsive to rate of 33% was higher than the 22% immunological, or metabolic function. drugs. underdiagnosis rate in the 137 patients who Thus, overdiagnosing bipolar disorder can Use of screening scales, such as the mood had not had bipolar disorder previously unnecessarily expose patients to serious drug disorders questionnaire, for bipolar disorder diagnosed. side effects. can also result in overdiagnosis. Screening Thus, the only two studies examining both questionnaires prioritise sensitivity, at a cost underdiagnosis and overdiagnosis have both Causes of overdiagnosis of false positive results, because it is presumed found evidence that overdiagnosis is a greater Why might the phenomenon of false positive that they are followed by expert clinical problem than underdiagnosis. bipolar disorder diagnoses be arising at this evaluation. Insufficient diagnostic rigour time? One important answer is likely to be after the use of screening scales can result in Accuracy is crucial the increased availability of drugs to treat greater rates of overdiagnosis. Routine use Whether bipolar disorder is more frequently bipolar disorder and the accompanying of bipolar disorder screening scales is not overdiagnosed or underdiagnosed is not marketing efforts. Many continuing medical supported by the data and does not make really important. Rather, it is critical that programmes on bipolar disorder conceptual sense.17 it is accurately diagnosed. Use of thorough begin with a summary of research suggesting Competing interests: None declared. diagnostic evaluations is important to detect bipolar disorder is underdiagnosed, and Cite this as: BMJ 2010;340:c855

BMJ | 27 march 2010 | Volume 340 687