Methodological pitfalls in early detection studies – the NAPE Lecture 2002


  • This paper is part of the TIPS project with the following research group: Thomas McGlashan, M.D. (PI), Per Vaglum M.D. (PI), Svein Friis, M.D., Ulrik Haahr, M.D., Jan Olav Johannessen, M.D., Tor K. Larsen, M.D., Ingrid Melle, M.D., Stein Opjordsmoen, M.D., Bjørn Rishovd Rund, Ph.D., Erik Simonsen, M.D.
    An earlier version of this paper was presented at the NAPE (Nordic Association for Psychiatric Epidemiology) Lecture at the 2002 Annual Meeting, Oslo, Norway, November 16, 2002. This issue of Acta Psychiatrica Scandinavica also includes a short biography of Svein Friis.

Svein Friis, Division of Psychiatry, Ullevål University Hospital, N-0407, Oslo, Norway E-mail:


Objective: To identify and discuss methodological pitfalls that may help explain why many questions around early detection (ED) and duration of untreated psychosis (DUP) are still unsolved.

Method: This paper concentrates on pitfalls in the following areas: sampling, measurement and data analyses.

Results: The main problems seem to be: Sampling: Referral bias, exclusion of patients, patient refusal, and patients lost to follow-up. Measurement: Reliability, which is particularly cogent for multisite investigations, and validity, which includes: Start of illness, start of psychosis, diagnoses, start of treatment, the relationship between ED and DUP and choice of outcome measures.

Data analyses:  Overlooking threshold effects of DUP, improper control for baseline scores, and lack of control for confounders.

Conclusion: Methodological pitfalls may bias ED studies. Several pitfalls are unavoidable, but proper design and quality assurance can reduce their impact. Researchers ought to identify the pitfalls, and to estimate and discuss their influence.