Does duration of untreated psychosis bias study samples of first-episode psychosis?

While findings are contradictory, many studies report that long Duration of Untreated Psychosis (DUP) correlates with poorer outcome in first episode psychosis. In an outcome study of first-episode psychosis, we compared the patients who refused to participate in a follow-along with those who consented to estimate the importance of this factor in sample recruitment bias. Our questions were: (i) What is the percentage of refusers? (ii) Are there systematic differences between refusers and consenters on DUP and/or other admission variables? (iii) What is the risk of refusal for different values of DUP?

In an unselected group of consecutively admitted patients we compared follow-along refusers and consenters on the following admission variables: sex, age, diagnostic group, substance abuse, being in-patient, coming from an early detection site and DUP. We conducted a logistic regression analysis with refusal as the outcome variable.

Ninety-three of 397 patients (23%) were refusers. In univariate analyses the only significant difference was found for DUP. The median DUP for consenters was 10 weeks and for refusers 32 weeks. DUP remained significant when all independent variables were controlled for.

DUP was significantly related to risk for refusal. For tests of the impact of DUP on outcome, this may introduce a type II error bias.