by Michael D. Anestis, M.S.
At the beginning of each semester, I open my Abnormal Psychology course with a unit discussing how and why to integrate a systematic approach to diagnosis and a data-driven approach to treatment. Inevitably, while my students see the utility in this approach and come to appreciate it as the semester progresses, they are relatively bored by this topic early on, eagerly awaiting the portion of the course devoted to specific disorders and treatments. Admittedly, I find this portion of my class a bit dry as well, but feel that it is necessary. Well, my students for next semester are likely to roll their eyes a little more emphatically now because I just found a new study that addresses this idea quite well and, as such, my reading list just evolved. A preemptive apology to my students!
In this particular study, conducted by Bruchmuller and Meyer (2009) and published in the Journal of Affective Disorders, the authors examined whether providing irrelevant information would influence the diagnostic decision making of therapists when the client clearly meets criteria for bipolar disorder. The authors provided a strong rationale for conducting this particular study. According to prior research, approximately 73% of individuals diagnosed with bipolar disorder report having been given at least one misdiagnosis prior to being diagnosed as having bipolar. On average, they received 3.5 misdiagnoses and saw five clinicians prior to receiving the proper diagnosis. In some cases, it took as long as ten years to arrive at the proper diagnostic conclusion (Lisch et al., 1994). This is particularly problematic given the data indicating that individuals with undiagnosed bipolar disorder have elevated rates of suicide (Shi et al., 2004). Additionally, because many individuals with undiagnosed bipolar disorder are mistakenly diagnosed with depression, treatment at times involves the prescription of antidepressant medication, which has been shown to worsen the course of bipolar disorder (e.g., induced manic episodes; Altshuler et al., 1995). Given these findings, it appears pivotal for clinicians to correctly identify the diagnostic condition presented by the client in order to ensure the proper treatment and the best outcome.
There are, admittedly, a multitude of factors that can lead to confusion on the part of a clinician in diagnosing a client with bipolar disorder. First, bipolar clients quite often present during the course of a depressive episode, thus naturally leading the clinician to assume this is the primary problem. Additionally, individuals in the midst of hypomanic or manic episodes are often not motivated to seek treatment, as this portion of the illness often feels good to the client at the time and, as such, is not seen as a problem. These factors, however, are not justifiable reasons for misdiagnosis, as a proper diagnosis should be arrived at through a systematic assessment that includes probing for symptoms that are not necessarily mentioned upon arrival by the client but which correspond to disorders specified in the DSM-IV-TR (or ICD-10 outside the United States). In other words, diagnosis isn't about simply asking the client what brings them to see you, but rather evaluating the data they provide through answers to structured questions based upon the diagnostic criteria for a range of mental illnesses.
The authors point out that, in making day-to-day decisions, we rarely consider all available data. Instead, we rely upon heuristics and beliefs to influence our conclusions. Quite often, this is a helpful approach, enabling us to manage an infinite number of stimuli and to make quick decisions and appropriate responses. In diagnostics, however, this is highly problematic, as indicated by the findings mentioned above.
This particular study investigated two specific ways in which the diagnosis of bipolar disorder can be influenced. One way was through the inclusion of irrelevant information - in this case, mentioning that the client had recently fallen in love, which had left him feeling extremely positive in recent weeks. The second way was by including a symptom that is often connected with bipolar disorder, a decreased need for sleep. Participating therapists received one vignette and were asked to indicate what they believed to be the proper diagnosis for that particular client. In each vignette, the fictional client endorsed a sufficient number of symptoms to meet criteria for bipolar disorder in both the DSM-IV-TR and the ICD-10 (except one vignette, in which the fictional client met criteria for depression). Of the five vignettes, four described an individual meeting criteria for bipolar disorder. They differed from one another as follows:
- Endorsed lack of sleep and a new romantic relationship
- Endorsed lack of sleep but no new romantic relationship
- Did not endorse lack of sleep but did endorse a new romantic relationship
- Did not endorse lack of sleep or a new romantic relationship
Again, remember, all of the vignettes in these four conditions described individuals who met the diagnostic criteria for bipolar disorder. They simply differed on whether they described a new romantic connection and the lack of sleep symptom of bipolar disorder.
The authors' findings were stunning. In the bipolar vignettes, only 38% of the fictional clients were diagnosed with bipolar disorder. 53% were mistakenly diagnosed with depression. No such difficulties were found in diagnosing the fictional clients in the depression vignette, indicating that the problem is at least somewhat unique to bipolar disorder. Vignettes that included a description of lack of sleep were significantly more likely to result in a bipolar diagnosis than were vignettes that did not include such a description. Given that all of the clients actually met criteria for bipolar disorder, this difference is fairly troubling, as it seems to indicate that the therapists were making their diagnostic decision based upon the presence or absence of one particular symptom rather than considering all the potential presentations for bipolar disorder. There was no main effect for relationship (e.g., whether or not the fictional client endorsed a new romantic connection did not, on its own, influence diagnostic decision making); however, relationship status interacted with sleep information such that individuals who endorsed a lack of sleep but had a new romantic connection were significantly less likely to receive a bipolar diagnosis than were individuals who endorsed a lack of sleep but no new romantic connection. At first glance, this might seem reasonable, but the authors noted two reasons why such an interpretation would be mistaken:
- Individuals in a hypomanic or manic episode often experience high levels of sociability, hypersexuality, and need for intimacy, which might make them more likely to endorse having "fallen in love." As such, this particular new relationship is just as likely to reflect the presence of mania or hypomania as it is to justify a decreased need for sleep.
- Even if a new romantic relationship is somehow seen as a justification for dismissing a lack of sleep, all of the fictional clients met criteria for bipolar disorder regardless of whether or not the sleep criteria was included. In other words, the extra symptom, which many therapists think of as a trademark of bipolar, was actually superfluous in this case and should have therefore had no impact on diagnostic decision making
The authors reported one more important finding. The theoretical orientation of the therapist significantly predicted whether or not the correct diagnosis was given. In this case, cognitive behavioral therapists were significantly more likely to identify the fictional client as meeting criteria for bipolar disorder than were psychodynamic therapists.
As I read this article, it occurred to me that some readers may respond by questioning the utility of diagnoses altogether. A complete retort to that line of reasoning is beyond the scope of this article, but I will provide a quick and overly simplistic response in case this is on your mind as you're reading these results. First, while many individuals who feel this way indicate that they feel as though diagnoses simply represent labels that force people into a box, this is in fact far from the case. Diagnoses can help clients realize that they are not alone in their symptoms or freaks for having them, but rather one of many individuals who have encountered these symptoms, for which we have empirically supported treatments. Our July 2009 featured article will, in fact, provide data that supports the utility of proper diagnostic feedback. Second, different diagnoses respond differently to different treatments. As such, without an effective diagnosis, we are significantly less likely to engage in the most effective approach to treatment.
There are several limitations to this study and the authors properly note many of these weaknesses. The primary one I would like to mention though is the use of vignettes for diagnostic decision making. A proper diagnostic assessment involves the administration of semi-structured diagnostic interviews (see our assessment tools page for more information on such tools). As such, asking clinicians to reach a diagnostic decision after merely reading a story is a bit problematic. Unfortunately, there is not much that can be done to resolve this particular problem, as it would be extremely difficult to run a study with enough participants to effectively test these hypotheses using such a diagnostic interviews. As a result, the best we can hope for may be to see these results replicated multiple times by independent researchers, thus increasing our confidence that they represent a generalizable phenomenon.
Overall, what does this study tell us? Quite a bit, actually. First, as clinicians, we tend to selectively attend to particular information in such a way that leads us to discount relevant data in diagnostic decision making. When we do this, we're likely to misdiagnose clients. Although the symptoms as described in the DSM-IV-TR and the ICD-10 are imperfect, they provide us with our best framework within to work and, as such, fidelity to that diagnostic system is crucial. Second, it highlights the need for a standardized approach to diagnostics. This point overlaps a bit with the first, but the central idea is that, by having one unified approach that involves a systematic assessment of all available data, we can decrease the degree to which personal bias on the part of the therapist or incomplete information provided by the client upon first arriving at the clinic will influence diagnosis. The need for this was emphasized further by the finding that theoretical orientation significantly predicted misdiagnosis. How the therapist thinks about mental illness is fairly irrelevant, as the interaction between therapist and client is not about satisfying the therapist's intellectual needs. The point of this interaction is the effective identification and treatment of symptoms so as to improve the life of the client and prevent relapse. If how we think about mental illness is influencing our diagnostic decision making abilities, our thought process is flawed.
In the coming month, leading up to our July featured article, I intend to touch on other issues relevant to the diagnostic process and the utility of diagnoses in general. In the meantime, we at Psychotherapy Brown Bag would love to hear your thoughts not only on this particular study, but also on these issues in general.
If you would like to learn more about these issues, a wide variety of resources in our online store could prove useful. The store is divided into books for professionals (which is further divided by topic) and books for nonprofessionals, but this division is somewhat arbitrary and most resources in the latter are also listed in the former. Feel free to contact us with suggestions for other resources we might consider including in the store.
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University.
Subscribe in a reader
