Cancer Institute A national cancer institute
designated cancer center

Philip W. Lavori

Publication Details

  • CLINICAL-TRIALS IN PSYCHIATRY - SHOULD PROTOCOL DEVIATION CENSOR PATIENT DATA NEUROPSYCHOPHARMACOLOGY Lavori, P. W. 1992; 6 (1): 39-48

    Abstract:

    Clinical trials methodologists recommend counting all events regardless of adherence to protocol and comparing originally randomized groups. This strict "intent-to-treat" policy implies availability and use of outcome measures taken regardless of adherence to treatment protocol. However, outcome measurement in psychiatry requires the cooperation of the patient, and usually occurs in the context of treatment management. Consequently, the patient's or clinician's decision not to adhere to the treatment protocol may be design or default cause censorship of patient data by early truncation. This disables the analysis "by intent to treat" in the strict sense. Current methods applied to such nonrandomly truncated datasets are unsatisfactory ("last value" analysis, survival analysis) or worse (imputation by last value carried forward). I review the context of clinical experimentation in psychiatry, contrast the state of design and analysis with expert recommendations on general methods, review the current statistical strategies and propose that investigators should try to obtain complete follow-up data on all patients without regard to their adherence to treatment protocol.

    View details for Web of Science ID A1992GZ63800006

    View details for PubMedID 1571068

Stanford Medicine Resources:

Footer Links: