Analysis of sequential treatment protocols for endometriosis-associated infertility.

The evaluation of comparative pregnancy data in clinical studies is subject to a variety of biases. One such bias, when treated patients are retrospectively compared with untreated control subjects, results from the fact that treated patients must remain infertile from time of diagnosis to time of treatment while no such requirements is maintained for untreated control subjects. This bias is also apparent when patients undergo sequential treatments and transfer from one comparison group to another. We have eliminated this problem by applying the Mantel-Byar approach for constructing and comparing modified life tables that reflect sequential changes of treatment status. To illustrate this we applied commonly used methods of analysis as well as the Mantel-Byar method to infertile endometriosis patients attempting pregnancy. One hundred thirty consecutive patients were evaluated retrospectively, with 42 undergoing expectant management only, seven experiencing a conservative surgical procedure only, and 81 having a conservative surgical procedure after a variable period of expectant management. Use of Mantel-Byar analysis revealed no significant overall increase in pregnancy rate with conservative surgical procedures versus expectant management (chi 2 = 0.225, df = 1). To further assess the potential value of conservative surgical procedures in fertility enhancement, the data were stratified by degree of endometriosis. The adjusted chi 2 was 1.621. No significant difference was noted between the two therapeutic approaches with mild disease (n = 35, chi 2 = 0.0175) or moderate endometriosis (n = 59, chi 2 = 0.424). Conservative surgical procedures appear to be therapeutic, however, in women with severe disease (n = 36, chi 2 = 5.12, p less than 0.05). These results differed substantially from those obtained with the more biased analytic methods routinely used in clinical fertility trials. This study illustrates the utility of the Mantel-Byar approach as a valuable tool in the evaluation of response time data involving transient therapeutic states and one particularly adaptable to retrospective infertility studies involving sequential treatment approaches.

Duke Scholars

Author Kerry L. Lee Biostatistics & Bioinformatics, Division of Biostatistics