Oversight or Trompe l'oeil?

Article Outline

• Reference
• Copyright

To the Editors:

Vercellini et al recently reported that the postoperative use of a low-dose oral contraceptive pill (OCP) reduces the risk of endometrioma recurrence.¹ They retrospectively followed up on 277 patients who had undergone laparoscopic excision of ovarian endometriomas and who were offered OCP, and examined the recurrence rate in 3 groups: (1) patients who used OCP for the entire follow-up period (always users); (2) those who used OCP discontinuously (ever users); and (3) those who declined treatment (never users). By showing that group 3 has a significantly higher rate of recurrence than 1 group and 1 that, within group 2, patients used OCP for ≥ 12 months have a significantly lower recurrence rate than those who used less than 12 months, they concluded that the OCP used postoperatively is beneficial. Recognizing the possibility of a self-selection bias in patients, they argued that “women taking OCP could have been considered at higher risk of endometrioma recurrence because the suggestion to use postoperative medical treatment would be expected to be more probable after difficult or nonradical surgery than after routine procedures,”¹ and hence their results could be conservative.

However, a closer look at their presented data suggests that the selection bias is in the direction opposite to what the authors articulated, which might undermine their conclusions. This can be seen from their Table 1, which shows that ever and always users were significantly younger than never users, suggesting the possibility that the former might be slightly, but decisively, healthier, and thus less prone to recurrence. In fact, Table 1 shows clearly that always users have a significantly lower proportion of bilateral cases (22% vs 39%, P = .03 by Fisher's exact test) and lower proportion of stage IV patients (22% vs 35%, P = .16). In addition, even though always users were younger than never users, they actually had a higher proportion of previous pregnancies (26% vs 17%). All these appear to indicate that always users were younger, healthier, less sick, likely to be sexually more active (hence less likely to have dyspareunia and more motivated to take OCP), and had a higher proportion of previous pregnancies than never users. These observations strongly suggest the existence of a selection bias in favor of group 1, which is difficult, if not impossible, to eliminate post hoc by multivariate regression techniques as used by the authors.

Another rather remarkable aspect of the analysis is the comparison (or lack of) between groups 2 and 3. Although ever users were younger than never users, they appeared to resemble to the latter in disease severity. Naturally, one would compare the recurrence rate between these 2 groups, which could also serve the purpose of demonstrating an effect of length of OCP use, if any. In doing so the authors would have solidified their conclusion if the results were favorable. Unfortunately, the authors chose not to carry out such an analysis. Instead, they split ever users into 2 groups and showed a length-dependent effect only within this group. This is rather curious.

Although postoperative use of OCP might very well reduce recurrence risk in ovarian endometriomas, the results reported by Vercellini et al should be viewed with caution because of its retrospective nature and, most importantly, the apparent selection bias as alluded to previously. Until some well-powered randomized clinical trials demonstrate the benefit of OCP use more conclusively, the current retrospective studies should acknowledge their deficiencies openly, objectively, and candidly so that readers can have a better perspective on the matter.

Back to Article Outline

Reference 

1. Vercellini P, Somigliana E, Daguati R, Vigano P, Meroni F, Crosignani PG. Postoperative oral contraceptive exposure and risk of endometrioma recurrence. Am J Obstet Gynecol. 2008;198:e1–e5
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (99 KB)
   • CrossRef