advertisement

---

Glaucoma Dialogue IGR 10-4

Response

Comment by Boel Bengtsson on:

22358 Disc hemorrhages and treatment in the early manifest glaucoma trial, Bengtsson B; Leske MC; Yang Z et al., Ophthalmology, 2008; 115: 2044-2048

See also comment(s) by Don Budenz • Kenji Kashiwagi • Chris Leung • James Morgan • Jody Piltz-Seymour • Hidenobu Tanihara •

Find related abstracts

---

We appreciate the great interest and several comments on our recently published paper on presence and frequency of disc hemorrhages in the Early Manifest Glaucoma Trial. We are glad for the opportunity to address the most important questions raised in those comments. We were not able to prove any difference in the presence or frequency of disc hemorrhages between treated and untreated glaucoma patients included in the Early Manifest Glaucoma Trial. There were somewhat (but not significantly) fewer patients, 46.4% vs. 51.2% with disc hemorrhages in the control group than in the treated group. As has been shown in numerous earlier studies, disc hemorrhages were significantly associated with progression.
The main objection in some of the comments was that IOP reduction in EMGT was not large enough to see any effect on disc hemorrhages. In the results section of the paper we actually presented two sub analyses among the treated patients. In one of these we compared IOP reduction in patients with and without disc hemorrhages. IOP reduction was similar in both groups. In the second analysis, where no censoring for progression was used, there was no association between presence of disc hemorrhages and the magnitude of the IOP reduction. This finding is a clear indication that the magnitude of IOP reduction is non-important. It seems unlikely that any possible IOP reducing effect on disc hemorrhages should diminish at some arbitrary threshold level of IOP reduction.
As so often, when the null hypothesis not is rejected, i.e., no statistically significant differences were found, the question of statistical power was raised. We included 129 treated and 126 untreated glaucoma patients. Our paper presents analyses that compare presence and frequency of disc hemorrhages in an already completed clinical trial; therefore we chose not to present any power calculation. The statistical power was sufficient (≥ 80%) to detect a difference of 20% or more between the two groups, but not enough to detect a difference of 10%. What is a relevant difference? � that can of course be discussed. The EMGT material is pretty large with slightly more than 7500 observations of 255 patients, and we felt that the results were both interesting and thought-provoking and certainly worth reporting, even the power certainly is not 95% to detect a 5% difference!
The censoring of data after progression can, of course, increase the risk of bias, since control patients progressed earlier than treated patients, and therefore treated patients had more visits and thereby increased chance for detection of disc hemorrhages. Therefore, we also analyzed, and pointed out in the discussion of the paper, that the frequency of disc hemorrhages was similar in both groups � 12.4% of photographs in the treated group and 11.2% in the control group.
Gender was clearly associated with disc hemorrhages detected on both photographs and noted in clinical forms, while refractive error and baseline/follow-up IOP was significantly associated with disc hemorrhages noted in clinical forms only.

Gender was clearly associated with disc hemorrhages

These were our results, and they are difficult to explain. The discrepancies can possibly be due to stepwise procedures or other model selection, but, as we mentioned in the discussion section, in clinical forms the odds ratios for both refraction and IOP were much closer to 1 than gender, with an odds ratio of 0.48.
Our results suggest that the increased risk for progression caused by disc hemorrhages is independent of IOP reduction. As pointed out in one of the comments, extra IOP reduction may well be motivated in the management of patients with disc hemorrhages � markers for increased risk of progression � but we should not expect the increased therapy to prevent future hemorrhages. That hemorrhages seem to be more frequent at lower IOP levels, however, certainly raises at least some little concern.

---

• ← Previous
• Next →

---