Letter by Cheung et al Regarding Article, “Mild Retinopathy Is a Risk Factor for Cardiovascular Mortality in Japanese With and Without Hypertension: The Ibaraki Prefectural Health Study”
To the Editor:
The Ibaraki Prefectural Health Study shows that mild hypertensive retinopathy signs are independent predictors of cardiovascular mortality among Japanese men and women, with or without hypertension.1 Although this is the largest study to date examining the link between retinal signs and mortality, some aspects of the study merit clarification and further discussion.
For instance, there are several noteworthy issues related to the methodology used for retinal grading. First, to grade hypertensive retinopathy, the authors used the Keith-Wagner-Barker classification scheme, which has known major limitations.2 Based on the strength of the reported associations between retinopathy signs and cardiovascular risk, a simplified classification scheme has been proposed.2 It offers a more sensible clinical approach to grade hypertensive retinopathy. Grade 1 and 2 retinopathy, for example, are grouped together as a single entity (mild hypertensive retinopathy), because they are difficult to differentiate clinically and there is no difference in cardiovascular risk between these 2 grades. Second, irrespective of the classification scheme used, the authors should provide reliability data on retinal grading. This is critical given the subjective nature of grading qualitative signs. The intragrader and intergrader κ-values should therefore be reported. Third, the unusually small number of participants with grade 3 and 4 retinopathy (n=89, 0.1%) casts some doubt as to the accuracy of retinal grading. Much higher prevalence rates (5%–10%) have been reported in previous population-based studies.2 Finally, since the graders were not masked to participants' blood pressure levels, the potential for grading bias cannot be excluded.
Furthermore, the authors should consider additional analyses that might help extract more clinically relevant information from the available data. For example, the hazard ratios of the reported associations range from 1.2 to 1.7. The clinical significance of such modest associations is unclear. Other analytical methods, such as those used by McGeechan and colleagues (eg, predictive values, C statistics),3 could be helpful in this regard. In addition, the associations might vary among different subgroups (eg, diabetes mellitus versus nondiabetes mellitus). It is known that clinical signs of early diabetic and hypertensive retinopathy may share many phenotypic similarities.4 Moreover, diabetes mellitus clearly has a significant effect on both retinopathy signs (as shown in Table 1)1 and cardiovascular outcomes. Thus, it would certainly be informative to have supplementary analysis stratified by diabetes mellitus status.
Nonetheless, the Ibaraki Prefectural Health Study provides welcoming data on the link between retinopathy signs and long-term cardiovascular mortality risk. The findings are in keeping with previous studies4,5 and support the notion that retinopathy signs are specific biomarkers for cardiovascular disease, above and beyond traditional risk factors, even in apparently normotensive people. Future studies are needed to determine whether incorporating retinal photography in routine clinical assessment could improve the prediction of cardiovascular disease.
Ning Cheung, MBBS
Centre for Eye Research Australia
Royal Victorian Eye and Ear Hospital
University of Melbourne
Tien Yin Wong, MD, PhD
Singapore Eye Research Institute
National University of Singapore
- © 2012 American Heart Association, Inc.
- Sairenchi T,
- Iso H,
- Yamagishi K,
- Irie F,
- Okubo Y,
- Gunji J,
- Muto T,
- Ota H
- Liew G,
- Wang JJ,
- Mitchell P,
- Wong TY