Background: Neuroleptic malignant syndrome requires prompt recognition for effective management, but there are no established diagnostic criteria. This is the first validation study of recently published international expert consensus (IEC) diagnostic criteria, which include priority points assigned on the basis of the importance of each criterion for making a diagnosis of neuroleptic malignant syndrome.
Methods: Data were extracted from 221 archived telephone contact reports of clinician-initiated calls to a national telephone consultation service from 1997 to 2009; each case was given a total priority point score on the basis of the IEC criteria. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision, (DSM-IV-TR) research criteria, in original form and modified to accept less than "severe" rigidity, served as the primary diagnostic reference standard. Consultants' diagnostic impressions were used as a secondary reference standard. Receiver operating characteristic curve analysis was used to optimize the priority point cutoff score with respect to the reference standards.
Results: Area under the receiver operating characteristic curve ranged from 0.715 (95% confidence interval, 0.645-0.785; P = 1.62 × 10) for consultant diagnoses to 0.857 (95% confidence interval, 0.808-0.907; P < 5 × 10) for modified DSM-IV-TR criteria. The latter was associated with 69.6% sensitivity and 90.7% specificity.
Conclusions: Agreement was best between IEC criteria with a cutoff score of 74 and modified DSM-IV-TR criteria (sensitivity, 69.6%; specificity, 90.7%); this cutoff score demonstrated the highest agreement in all comparisons. Consultant diagnoses showed much better agreement with modified, compared with original, DSM-IV-TR criteria, suggesting that the DSM-IV-TR criterion of "severe" rigidity may be more restrictive than what most knowledgeable clinicians use in practice.