Email updates

Keep up to date with the latest news and content from BMC Neurology and BioMed Central.

Open Access Highly Accessed Research article

Validity of Simpson-Angus Scale (SAS) in a naturalistic schizophrenia population

Sven Janno1*, Matti M Holi2, Katinka Tuisku3 and Kristian Wahlbeck45

Author Affiliations

1 Department of Psychiatry, University of Tartu, Raja 31, 50417, Tartu, Estonia

2 Department of Psychiatry, Helsinki University, Helsinki, Finland

3 Finnish Institute of Occupational Health, Helsinki, Finland

4 STAKES National Research and Development Centre for Welfare and Health, Helsinki, Finland

5 Vaasa Central Hospital, Vaasa, Finland

For all author emails, please log on.

BMC Neurology 2005, 5:5  doi:10.1186/1471-2377-5-5

Published: 17 March 2005

Abstract

Background

Simpson-Angus Scale (SAS) is an established instrument for neuroleptic-induced parkinsonism (NIP), but its statistical properties have been studied insufficiently. Some shortcomings concerning its content have been suggested as well. According to a recent report, the widely used SAS mean score cut-off value 0.3 of for NIP detection may be too low. Our aim was to evaluate SAS against DSM-IV diagnostic criteria for NIP and objective motor assessment (actometry).

Methods

Ninety-nine chronic institutionalised schizophrenia patients were evaluated during the same interview by standardised actometric recording and SAS. The diagnosis of NIP was based on DSM-IV criteria. Internal consistency measured by Cronbach's α, convergence to actometry and the capacity for NIP case detection were assessed.

Results

Cronbach's α for the scale was 0.79. SAS discriminated between DSM-IV NIP and non-NIP patients. The actometric findings did not correlate with SAS. ROC-analysis yielded a good case detection power for SAS mean score. The optimal threshold value of SAS mean score was between 0.65 and 0.95, i.e. clearly higher than previously suggested threshold value.

Conclusion

We conclude that SAS seems a reliable and valid instrument. The previously commonly used cut-off mean score of 0.3 has been too low resulting in low specificity, and we suggest a new cut-off value of 0.65, whereby specificity could be doubled without loosing sensitivity.