advertisement
Editors Selection IGR 7-2
Clinical examination methods: Progression and IOP fluctuation
Comment by Kouros Nouri-Mahdavi on:
12483 Diurnal IOP fluctuation: not an independent risk factor for glaucomatous visual field loss in high-risk ocular hypertension, Bengtsson B; Heijl A, Graefe's Archive for Clinical and Experimental Ophthalmology, 2005; 243: 513-518
Bengtsson and Heijl (562) recently reported results of a retrospective analysis of 90 patients (90 eyes) from the Malmö Ocular Hypertension Study. Their study is notable for the very long follow-up available for the recruited patients. All patients had at least
10 years of follow-up and for some eyes, retrospective data were available up to 17 years after initial randomization. The authors found that in univariate analyses with Cox's proportional hazard model, variables related to both IOP level (largest hazard ratio # 1.23, p = 0.001) and IOP fluctuation (largest hazard ratio # 1.38 for mean of daily IOP range, p = 0.06) were predictive of future progression. However, on multivariate analyses, only the IOP level (mean of all IOP measurements) was significantly associated with future glaucoma progression. The authors are to be commended for their effort to clarify the independent roles of IOP level vs. IOP fluctuation in predicting future field loss. The Malmö Ocular Hypertension Study database is an especially valuable one in this regard given the very long follow-up of recruited patients. There are, however, methodological and statistical issues that need to be considered before embracing the results. The authors have provided readers with little detail regarding the statistical methodology used. Cox's proportional hazard model, like other models, relies heavily on assumptions that are not necessarily always met. It is not clear why the authors chose to
On multivariate analysis only IOP level but not IOP fluctuation was significantly associated with progressionput only IOP-related variables into the multivariate model. Epidemiologists generally agree that in order to evaluate the true influence of any variable in relation to a given outcome, all the known and potentially influential variables must be entered into the model. Additionally, potential interactions need to be explored. In a more recent analysis of the AGIS database with the Cox's proportional hazard model, our group at Jules Stein Eye Institute has observed significant interactions among the predictive variables that needed to be explored with alternative statistical methods.
It is quite understandable that such long-term investigations may lack enough power to address the question(s) the authors attempted to address. The 95% confidence interval for the hazard ratio for the surrogate variable used to evaluate IOP fluctuation (mean of daily range) ranged from 0.8 to 1.6 in the multivariate analysis. This means that potentially an increased risk of up to 60% related to IOP fluctuation could exist and might have been missed given the sample size and variability of IOP fluctuation across the study sample. Therefore, although the study demonstrates the significance of IOP level with regard to future field loss, it is inconclusive regarding the association, if any, of IOP fluctuation with progression of glaucomatous damage. Further evidence is needed before this important issue is settled, knowing that the mutual roles of IOP level and IOP fluctuation and their interplay could vary in different types of glaucoma or at different stages of the disease.
Comments
The comment section on the IGR website is restricted to WGA#One members only. Please log-in through your WGA#One account to continue.Log-in through WGA#One
