advertisement
Stein et al. (1257) present a longitudinal retrospective cohort analysis of outcomes with trabeculectomy and drainage device implantation using Medicare claims data. The results of this study are certain to be controversial. In addition to patients in the cohort being representative of all individuals undergoing these procedures, random rather than systematic errors in coding are a necessary prerequisite for such analyses to be meaningful. The findings of this study contradict those of the Tube versus Trabeculectomy (TVT) Study with the latter showing that complication rates are lower with drainage device implantation relative to trabeculectomy. This discrepancy may be reflective of the obviously different methods between the two studies and the fact that long term follow up data is not as yet available from the TVT study. Unlike a randomized clinical trial such as the TVT study, this Medicare claims data based analysis may be influenced by differences in case severity between those undergoing trabeculectomy versus drainage device implantation. Drainage device implantation has historically been reserved for patients in whom trabeculectomy has already failed and are thus, all other things being equal, at a later stage in the disease process. The authors appropriately acknowledge that despite their best efforts to control for many important case mix factors using logistic regression, data limitations may have prevented them from fully accounting for such preoperative differences.
Results from large clinical trials or retrospective claims data based reviews do not necessarily reflect what is best in an individual physician's hands for a specific patientWheredoes this leave the practitioner faced with the choice of trabeculectomy versus drainage device implantation? Practitioners vary in their comfort level with various glaucoma procedures and results from large clinical trials or retrospective claims data based reviews do not necessarily reflect what is best in an individual physician's hands for a specific patient. Further studies are clearly needed to optimally guide such choices.