The usefulness of rating scales in patients with schizophrenia.

Abstract

Rating scales serve two main purposes in psychiatry: First, they are used to quantify symptoms or patient-related outcomes in clinical research, and second, they can help to monitor treatment outcomes in clinical practice. When evaluating or treating patients with schizophrenia, the former use is considerably more common than the latter, one of the reasons for this discrepancy being that most available rating scales for patients with schizophrenia are rather lengthy and time-consuming, therefore not lending themselves to being applied in everyday office or hospital practice. Ever since its introduction, the Positive and Negative Syndrome Scale (PANSS, 1) has become increasingly popular to quantify symptoms related to schizophrenia in clinical trials and the fact that regulators also favour and accept it has given this scale almost monopoly status. As outlined by ∅stergaard et al. 2 in the Journal, the original 30-item version has a number of disadvantages. Next to issues with scalability and transferability, as argued by the authors, the 30-item PANSS takes close to an hour to complete, depending on the raters’ experience and the patients’ severity of illness, which makes it impractical for use outside of research settings. In addition, the full PANSS may not always adequately reflect schizophrenia-specific symptom change, especially when calculating the commonly used response criteria of 20% or 30% improvement in the total score of the scale. This becomes evident when relating such criteria to improvements on the Clinical Global Impression Scale 3 and when taking improvements in unspecific items such as anxiety, somatic concern or uncooperativeness into account 4. We have shown that decreasing scores in these items alone can lead to a 20% improvement on the PANSS total score, which is misleading when the treatment goal is to improve the core symptoms of schizophrenia 4. Therefore, considering both issues, namely practicality and psychometric properties, the effort of ∅stergaard et al. 2 to reduce the original scale to a shorter and more scalable one is laudable. They demonstrate convincingly that a six-item PANSS not only has psychometric advantages over the original version but, given the shorter time needed for completion, it could also be useful in documenting outcomes in everyday clinical practice. As the authors have used only two phase III clinical trials to test their hypothesis, one would like to see a replication in independent samples. Ideally, attempts to replicate results should include observational and pragmatic trials, in which patients remain in the study even if they have discontinued medication or met relapse criteria. The European First Episode in Schizophrenia Trial 5 could serve as an exemplary study in this context. Lastly, the six-item PANSS would need to demonstrate its usefulness in prospective studies. Given the ever increasing complexity of clinical trials, especially those used for registration purposes, the field should welcome a rating scale modified towards the ends of being easier and more reliable to use.