© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Acta Psychiatr Scand 2014: 130: 238–239 All rights reserved
ACTA PSYCHIATRICA SCANDINAVICA
Debate section Letter to the editor
General indicators confounding Kraepelin: predictive indicators of misdiagnosing borderline as bipolar DOI: 10.1111/acps.12302 In their article titled ‘Bipolar or borderline: a clinical overview’, Ghaemi and colleagues found the two illnesses to be similar with regard to the nosological validators of mood lability and impulsivity (1). The authors conclusion that such a similarity often leads to misdiagnosis despite the two conditions being seen as different clinical entities is consistent with previous research (2). We recently reviewed data obtained from a clinical assessment tool used by the Division of Psychosomatic Medicine at Robert Wood Johnson University Hospital in New Brunswick, New Jersey. The tool (GICK) consists of pooled questions from the routine History, Physical and Mental Status Examination included in medical students’ orientation packets at the commencement of their third-year clerkship in psychiatry. We looked at 83 consecutive new patient consults requested over a 1-month period and found that 12 patients had a preadmission diagnosis of bipolar affective disorder (BAD). We chose to exclude patients with a co-occurring substance use disorder (n = 3). Of the nine patients identified, 3 were reevaluated as having a primary diagnosis of borderline personality disorder. While all three patients were also diagnosed with post-traumatic stress disorder (PTSD), none ruled-in for bipolar affective disorder or schizoaffective disorder. Items in the patient’s history that demonstrated predictive value of misdiagnosis included (i) a history of physical or sexual trauma, (ii) female gender, (iii) a positive screen to the question, ‘Do you consider yourself a nervous person?,’ and (iv) a pattern of antipsychotic use characterized by initial efficacy, followed by a loss of effect. Patients with such a pattern may also have related aspects in their history including extrapyramidal side-effects (EPS) and use of benztropine or similar acting anticholinergic medication. While we did find that past sexual abuse was correlated, our study did not find the genetic validator of a bipolar family history being predictive
of misdiagnosis. This may have been due to our extremely small sample size. While ours was a clinical performance improvement project that reviewed data routinely obtained by medical students during a core clerkship, we did find that 3.6% of patients with a pre-admission diagnosis of BAD were misdiagnosed and instead afflicted with borderline personality disorder and PTSD. We plan on applying the identified predictive variables to a larger database of patients with diagnoses of schizophrenia, schizoaffective disorder, and bipolar affective disorder to test their statistical and clinical significance. Such a study would further clarify the role specific indicators in patients’ histories play in the accurate diagnosis and treatment planning of individuals who present with mood lability and impulsivity (3, 4). A. Tobia, D. Mikkilineni, V. Bisen, A. Zimmerman and A. Trenton Department of Psychiatry, Rutgers-RWJMS, Piscataway, NJ, USA E-mail: [email protected]
References 1. Ghaemi SN, Dalley S, Catania C, Barroilhet S. Bipolar or borderline: a clinical overview. Acta Psychiatr Scand 2014; 130:99–108. 2. Coentre R, Power P. A diagnostic dilemma between psychosis and post-traumatic stress disorder: a case report and review of the literature. J Med Case Rep 2011;5:97. 3. Davidson JT. Recognition and treatment of posttraumatic stress disorder. JAMA 2001;286:584–588. 4. Leibschutz J, Saitz R, Brower V et al. PTSD in urban primary care: high prevalence and low physician recognition. J Gen Intern Med 2007;22:719–726.
Misdiagnosis: Predictive value versus sensitivity DOI: 10.1111/acps.12303 Reply In this analysis, 3 of 9 patients previously diagnosed with bipolar illness were felt to be misdiagnosed as having borderline personality disorder (BPD), using DSM criteria (1). I will note that the DSM criteria for BPD are very broad, exclude sexual
238
trauma as a criterion, and list self-harm as one of nine criteria, of which five are sufficient to make that diagnosis. Thus, if my view is correct that sexual trauma and self-harm are central to the borderline personality concept, then the DSM definition of BPD, as used in most research, may not be scientifically valid. With that proviso, these data reflect predictive values, and not sensitivity, of the bipolar diagnosis. In this case, the positive predictive value (PPV) of bipolar illness was 67% (6/9).
Letter to the editor This is actually much higher than other studies, where the PPV tends to be about 40% (2). PPV is sensitive to the baseline prevalence rates of diagnoses; as each specific psychiatric diagnosis has ‘low’ prevalence, even in general psychiatric clinical populations, it is very difficult to achieve PPV about 50–60%, no matter how narrowly and conservatively we define our diagnoses (2). In fact, reliability of many psychiatric conditions, including ‘simple’ ones like major depressive disorder (MDD, kappa value of only 0.25), was rather low in DSM-5 field trials (3). These low kappa values would predict, if predictive values had been assessed in the DSM-5 field trials, correspondingly low PPVs, even with less controversial diagnoses such as MDD. Hence, it is important to emphasize that the identification of 3 of 9 subjects as possibly having BPD rather than bipolar illness is NOT evidence of ‘overdiagnosis’, but rather reflective of limitations in reliability of psychiatric diagnoses in general, being rather similar for all conditions (2). ‘Overdiagnosis’ would reflect sensitivity, that is, of all those individuals, in the original sample in this analysis of 83 new patients, who truly do not have bipolar illness (were identified by the researchers as not meeting DSM criteria for bipolar illness), how many were previously mistakenly diagnosed as having bipolar illness? Other studies consistently find that the frequency of this misdiagnosis is at least two-fold less than the opposite case, that is, persons who truly have bipolar illness (based on
researchers’ diagnoses) and are previously mistakenly diagnosed as NOT having bipolar illness. In addition to their proposed analyses, the researchers would help clarify mistaken claims of ‘overdiagnosis’ of bipolar illness if they assess sensitivity, not just predictive value. N. Ghaemi Mood Disorders Program, Tufts Medical Center, Boston, MA, USA E-mail: [email protected]
References 1. Tobia A, Mikkilineni D, Bisen V, Zimmerman A, Trenton A. General indicators confounding Kraepelin: predictive indicators of misdiagnosing borderline as bipolar. Acta Psychiatr Scand 2014;130:238. 2. Phelps J, Ghaemi SN. The mistaken claim of bipolar ‘overdiagnosis’: solving the false positives problem for DSM-5/ ICD-11. Acta Psychiatr Scand 2012;126:395–401. 3. Regier DA, Narrow WE, Clarke DE et al. DSM-5 field trials in the United States and Canada, Part II: test-retest reliability of selected categorical diagnoses. Am J Psychiatry 2012;170:59–70.
239
Acta Psychiatr Scand 2014: 130: 238–239 All rights reserved
ACTA PSYCHIATRICA SCANDINAVICA
Debate section Letter to the editor
General indicators confounding Kraepelin: predictive indicators of misdiagnosing borderline as bipolar DOI: 10.1111/acps.12302 In their article titled ‘Bipolar or borderline: a clinical overview’, Ghaemi and colleagues found the two illnesses to be similar with regard to the nosological validators of mood lability and impulsivity (1). The authors conclusion that such a similarity often leads to misdiagnosis despite the two conditions being seen as different clinical entities is consistent with previous research (2). We recently reviewed data obtained from a clinical assessment tool used by the Division of Psychosomatic Medicine at Robert Wood Johnson University Hospital in New Brunswick, New Jersey. The tool (GICK) consists of pooled questions from the routine History, Physical and Mental Status Examination included in medical students’ orientation packets at the commencement of their third-year clerkship in psychiatry. We looked at 83 consecutive new patient consults requested over a 1-month period and found that 12 patients had a preadmission diagnosis of bipolar affective disorder (BAD). We chose to exclude patients with a co-occurring substance use disorder (n = 3). Of the nine patients identified, 3 were reevaluated as having a primary diagnosis of borderline personality disorder. While all three patients were also diagnosed with post-traumatic stress disorder (PTSD), none ruled-in for bipolar affective disorder or schizoaffective disorder. Items in the patient’s history that demonstrated predictive value of misdiagnosis included (i) a history of physical or sexual trauma, (ii) female gender, (iii) a positive screen to the question, ‘Do you consider yourself a nervous person?,’ and (iv) a pattern of antipsychotic use characterized by initial efficacy, followed by a loss of effect. Patients with such a pattern may also have related aspects in their history including extrapyramidal side-effects (EPS) and use of benztropine or similar acting anticholinergic medication. While we did find that past sexual abuse was correlated, our study did not find the genetic validator of a bipolar family history being predictive
of misdiagnosis. This may have been due to our extremely small sample size. While ours was a clinical performance improvement project that reviewed data routinely obtained by medical students during a core clerkship, we did find that 3.6% of patients with a pre-admission diagnosis of BAD were misdiagnosed and instead afflicted with borderline personality disorder and PTSD. We plan on applying the identified predictive variables to a larger database of patients with diagnoses of schizophrenia, schizoaffective disorder, and bipolar affective disorder to test their statistical and clinical significance. Such a study would further clarify the role specific indicators in patients’ histories play in the accurate diagnosis and treatment planning of individuals who present with mood lability and impulsivity (3, 4). A. Tobia, D. Mikkilineni, V. Bisen, A. Zimmerman and A. Trenton Department of Psychiatry, Rutgers-RWJMS, Piscataway, NJ, USA E-mail: [email protected]
References 1. Ghaemi SN, Dalley S, Catania C, Barroilhet S. Bipolar or borderline: a clinical overview. Acta Psychiatr Scand 2014; 130:99–108. 2. Coentre R, Power P. A diagnostic dilemma between psychosis and post-traumatic stress disorder: a case report and review of the literature. J Med Case Rep 2011;5:97. 3. Davidson JT. Recognition and treatment of posttraumatic stress disorder. JAMA 2001;286:584–588. 4. Leibschutz J, Saitz R, Brower V et al. PTSD in urban primary care: high prevalence and low physician recognition. J Gen Intern Med 2007;22:719–726.
Misdiagnosis: Predictive value versus sensitivity DOI: 10.1111/acps.12303 Reply In this analysis, 3 of 9 patients previously diagnosed with bipolar illness were felt to be misdiagnosed as having borderline personality disorder (BPD), using DSM criteria (1). I will note that the DSM criteria for BPD are very broad, exclude sexual
238
trauma as a criterion, and list self-harm as one of nine criteria, of which five are sufficient to make that diagnosis. Thus, if my view is correct that sexual trauma and self-harm are central to the borderline personality concept, then the DSM definition of BPD, as used in most research, may not be scientifically valid. With that proviso, these data reflect predictive values, and not sensitivity, of the bipolar diagnosis. In this case, the positive predictive value (PPV) of bipolar illness was 67% (6/9).
Letter to the editor This is actually much higher than other studies, where the PPV tends to be about 40% (2). PPV is sensitive to the baseline prevalence rates of diagnoses; as each specific psychiatric diagnosis has ‘low’ prevalence, even in general psychiatric clinical populations, it is very difficult to achieve PPV about 50–60%, no matter how narrowly and conservatively we define our diagnoses (2). In fact, reliability of many psychiatric conditions, including ‘simple’ ones like major depressive disorder (MDD, kappa value of only 0.25), was rather low in DSM-5 field trials (3). These low kappa values would predict, if predictive values had been assessed in the DSM-5 field trials, correspondingly low PPVs, even with less controversial diagnoses such as MDD. Hence, it is important to emphasize that the identification of 3 of 9 subjects as possibly having BPD rather than bipolar illness is NOT evidence of ‘overdiagnosis’, but rather reflective of limitations in reliability of psychiatric diagnoses in general, being rather similar for all conditions (2). ‘Overdiagnosis’ would reflect sensitivity, that is, of all those individuals, in the original sample in this analysis of 83 new patients, who truly do not have bipolar illness (were identified by the researchers as not meeting DSM criteria for bipolar illness), how many were previously mistakenly diagnosed as having bipolar illness? Other studies consistently find that the frequency of this misdiagnosis is at least two-fold less than the opposite case, that is, persons who truly have bipolar illness (based on
researchers’ diagnoses) and are previously mistakenly diagnosed as NOT having bipolar illness. In addition to their proposed analyses, the researchers would help clarify mistaken claims of ‘overdiagnosis’ of bipolar illness if they assess sensitivity, not just predictive value. N. Ghaemi Mood Disorders Program, Tufts Medical Center, Boston, MA, USA E-mail: [email protected]
References 1. Tobia A, Mikkilineni D, Bisen V, Zimmerman A, Trenton A. General indicators confounding Kraepelin: predictive indicators of misdiagnosing borderline as bipolar. Acta Psychiatr Scand 2014;130:238. 2. Phelps J, Ghaemi SN. The mistaken claim of bipolar ‘overdiagnosis’: solving the false positives problem for DSM-5/ ICD-11. Acta Psychiatr Scand 2012;126:395–401. 3. Regier DA, Narrow WE, Clarke DE et al. DSM-5 field trials in the United States and Canada, Part II: test-retest reliability of selected categorical diagnoses. Am J Psychiatry 2012;170:59–70.
239
