Abstract

True or classical experimental designs have often been used as the sine qua non of an effective and internally valid approach to examine causal relationships. The Solomon four-group design has been used to assess pre-test effects, but it is often difficult and expensive to implement. This study describes how a non-contact comparison group, followed through medical records, can be used in combination with a randomized experimental-control design to assess pre-test and monitoring effects. A total of 75 patients receiving chemoprophylactic tuberculosis therapy were randomly assigned to a special education intervention (SI) group or to a usual care (UC) group and followed for 12 months. Eighty-five patients who constituted a non-contact comparison (NCC) group were also followed through medical chart reviews for the same period. Life table analyses of the proportion of subjects completing prescribed therapy showed significant differences by group at 6 months of treatment, with 64% of SI subjects, 30% of UC subjects and 21% of NCC subjects still in treatment (P < 0.01). Median time in care was 52 (SI), 13 (UC) and 5 weeks (NCC). There was no difference between the UC and NCC groups in percentage of subjects completing 1 year of therapy. Implications regarding potential underestimation of true differences between experimental and control groups are discussed.

You do not currently have access to this article.

Comments

0 Comments