Cross design synthesis : a new strategy for medical effectiveness research : report to Congressional requesters / United States General Accounting Office.

  • United States. General Accounting Office.
Date:
[1992]
Catalogue details

Licence: Public Domain Mark

Credit: Cross design synthesis : a new strategy for medical effectiveness research : report to Congressional requesters / United States General Accounting Office. Source: Wellcome Collection.

    42/128 (page 40)
    Empirical Assessments of the Representativeness of the Achieved Patient Pool Chapter 2 Methods for Assessing Randomized Studies’ Generalizability to the Patient Population (Task 1) Judgmental assessments of the patient selection process, taken by themselves, are weaker than when combined with the empirical assessments described in the following sections. Empirical assessments are designed specifically to provide information about the consequences of patient selection procedures in randomized studies; that is, the representativeness of the achieved patient pool and the impact of any nonrepresentativeness on the generalizability of the observed treatment effect. We believe that empirical assessments of representativeness should be made unless the patient pool has been randomly selected from the reference population or approximates the total patient population. The representativeness of patients participating in a randomized study refers to the extent to which the achieved patient pool reflects the full population of patients for whom the treatment is intended. In assessing representativeness, it is unrealistic to expect (or to test for) achievement of a perfect miniature of the population (Kruskal and Mosteller, 1981); instead, one must ask whether the patient pool is representative with respect to a specific criterion. Because our goal is to estimate treatment effectiveness, the appropriate criterion is whether persons participating in the study are representative of the different ways that patients are affected by the treatment in question. Thus, the key question is: Does the treatment effect observed for participating patients represent the average treatment effect that would pertain for all patients in the relevant population? Representativeness of the treatment effect cannot be measured directly; thus, existing assessments have necessarily proceeded via two-part approaches. The major two-part approach consists of (1) directly comparing the baseline characteristics (for example, patient sex and age) of participants in a randomized study to those of patients seen in medical practice, and (2) looking for evidence of whether each underrepresented and overrepresented subgroup (for example, elderly female patients) experienced a different treatment effect than other patients. Another, less well-known approach also proceeds in two parts. As explained below, levels of patient outcomes (not effects) in each randomized study are compared to levels of outcomes for patients who received the same treatment in medical practice. Then, the reviewer looks for evidence that different levels of outcomes are linked to differential effects.