Related Experiment Videos
Effect of antipsychotic withdrawal on extrapyramidal symptoms: statistical methods for analyzing single-sample
S Arndt1, C S Davis, D D Miller
1Department of Preventive Medicine and Environmental Health, University of Iowa, Iowa City.
Summary
This study highlights issues with analyzing symptom severity data over time. Less common statistical methods like Friedman Analysis of Ranks and Cochran-Mantel-Haenszel (CMH) statistics revealed significant changes in extrapyramidal side effects (EPS) during antipsychotic washout, unlike traditional methods.
Area of Science:
- Psychiatry
- Clinical Psychology
- Biostatistics
Background:
- Extrapyramidal side effects (EPS) are common adverse events associated with antipsychotic medications.
- Assessing changes in symptom severity over time requires robust statistical methods, especially in psychiatric research.
- Repeated measures of symptom severity can present analytical challenges.
Purpose of the Study:
- To illustrate problems with repeated measures of symptom severity in clinical trials.
- To compare the effectiveness of different statistical methods in detecting changes in symptom severity during antipsychotic washout.
- To highlight the utility of less common statistical techniques for psychiatric data analysis.
Main Methods:
- Utilized symptom severity ratings of extrapyramidal side effects (EPS) during a 4-week antipsychotic washout period.
- Compared four statistical analysis methods: Analysis of Variance (ANOVA), Multivariate Analysis of Variance (MANOVA), Friedman Analysis of Ranks, and Cochran-Mantel-Haenszel (CMH) statistics.
- Evaluated the ability of each method to detect changes in weekly Simpson Angus scores.
Main Results:
- ANOVA and MANOVA failed to detect any mean change in weekly Simpson Angus scores, despite 43% of participants experiencing clinically significant EPS before drug discontinuation.
- Friedman Analysis of Ranks and CMH statistics identified significant changes in symptom severity over the washout period.
- CMH statistics demonstrated particular advantages due to its ability to handle incomplete data sets.
Conclusions:
- Traditional statistical methods like ANOVA and MANOVA may be insensitive to detecting changes in symptom severity in psychiatric populations.
- Less restrictive and more sensitive methods, such as Friedman Analysis of Ranks and CMH statistics, are more appropriate for analyzing repeated measures in psychiatric research.
- The CMH method is highly valuable for its robustness to missing data, enhancing the generalizability of findings in clinical studies.