11.07.2015 Views

Clinical Trials

Clinical Trials

Clinical Trials

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Clinical</strong> <strong>Trials</strong>: A Practical Guide ■❚❙❘Table 4. Hormone replacement therapy (HRT) trial example: comparison of proportion of mental functionimprovement in each socioeconomic group.Socioeconomic Improvement in mental function Total number Proportion of womenstatus Yes No of women showing improvementHigh 120 180 300 40%Low 40 460 500 8%Odds ratio = 0.13 (95% CI 0.08, 0.20; P < 0.0001).Step 2. Assess whether the potential confounder is a predictorof the outcomeThe second condition for a variable to be a confounder requires that the potentialconfounder must also be related to the outcome being measured. The relationshipbetween socioeconomic status and improvement in mental function is examinedin Table 4. The results show that the odds of having improved mental function inthe low socioeconomic status group are only about 13% of that in the highsocioeconomic status group. The low P-value suggests that there is strongevidence of a difference in improvement in mental function between the twosocioeconomic groups in favor of the high socioeconomic status group. Based onthese results, socioeconomic status is an important predictor of an improvementin mental function among postmenopausal women.Step 3. Assess that the potential confounder is not a consequenceof treatmentIn the HRT example, it is not possible for the treatment allocation to influencethe socioeconomic class of a woman on admission, since this is determined beforetreatment randomization. Hence we can conclude that socioeconomic statuswould not lie on the causal path between HRT treatment and the primaryendpoint (improvement in mental function).Hence, the socioeconomic status variable satisfies all the criteria for being aconfounding factor.Evaluating the degree of confoundingIf a prognostic factor satisfies the three conditions for being a confounder,the next step is to evaluate the degree of confounding. This should be done bycomparing the unadjusted (also known as crude) estimates of treatment affect –ie, the estimates that are unadjusted for the potential confounding factor –with the adjusted estimates. There is no specific test to determine whethera factor is a confounder in respect of any given treatment effect, but, if we adjust299

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!