2002 International International Biennial Scientific Meeting

DETERMINING CLINICALLY SIGNIFICANT CHANGE ON THE EPDS IN POSTNATAL DEPRESSION TREATMENT OUTCOME STUDIES

Stephen Matthey

Paediatric Mental Health Service, South West Sydney Area Health Service

Stephen.matthey@swsahs.nsw.gov.au

Objectives: To discuss the usefulness of traditional statistical techniques when evaluating change on the Edinburgh Postnatal Depression Scale (EPDS); ii) to discuss the method proposed by Jacobson and Truax (1991) to determine clinical significance in treatment studies, and apply this to the EPDS.

Method: The reporting of statistical significance, using p values and null hypothesis significance testing, tells us little about the size of the change in EPDS scores between treatment groups. Adding the Effect Size statistic improves this, but still tells us nothing about how many participants showed a meaningful change on the measure. To report this clinical significance requires the Reliable Change (RC) Index to be calculated, and a decision as to what post treatment score is considered sufficient for a person to be considered either 'recovered' or 'improved'. These will be discussed, and data from 22 women referred to our Parent-Baby day unit for perinatal mood disorders will be examined.

Using this information, I shall show how conclusions based upon the p statistic are quite different from those based upon the clinical significance calculations.

Conclusions: To make the results of research studies more clinically useful, we need to report information about how many participants improve following treatment. This information cannot be gained from the averaging statistics of p or effect size. We must also include the clinical significance calculation, which requires the RC Index to be calculated for whatever measure is being used, as well as determining what post treatment score is considered in the recovered or improved ranges. These will be discussed in relation to the EPDS.