(027) Increased risk of retinal vascular occlusion in patients receiving cataract surgery: causation or biased association?

Publication Number: 365

Background: A recent study from the American Academy of Ophthalmology IRIS Registry reported showed a small but significantly higher risk of retinal vascular occlusion (RVO) after cataract surgery. However, potential concerns due to the lacked consideration of preoperative examination and possible detection bias should be noted.

Objectives: We replicate the previous IRIS study but address these critical concerns in this present study.

Methods: This retrospective cohort study analyzed Taiwan's National Health Insurance Research Database. We included patients aged over 50 years old with at least two diagnosis records of cataracts from 2003 to 2018. Patients receiving cataract surgery were defined as the treatment group, and the index date was defined as the date of surgery. The control group consisted of patients not receiving cataract surgery, and the index date was randomly selected after the diagnosis of cataracts in the control group. We excluded patients with RVO, aphakia, pseudophakia, epiretinal membrane (ERM), and macular hole (MH) before the index date. To generate more homogeneous comparisons, we conducted the 1:1 propensity score matching approach by considering the comprehensive baseline characteristics. We first replicated the IRIS Registry study to evaluate the risk of RVO after the index date until the end of the study period or death. Considering that the lacked consideration of preoperative assessment may introduce a protopathic bias, we conducted the additional analysis by only including patients with preoperative assessment within six months before the index date. Furthermore, we conducted the control analysis to evaluate the detection bias by using the ERM and MH as false outcomes.

Results: We included a total of 576,736 patient pairs with comparable baseline characteristics in this study. The incidence rate of RVO was 13.25 and 10.89 per 10,000 person-years in the cataract surgery and control groups, respectively, yielding an HR of 1.25 (95% CI 1.19-1.31) which was comparable with the IRIS report (HR: 1.26; 95% CI 1.16-1.38). In the additional analysis, no increased risk of RVO (HR 1.02; 95% CI 0.94-1.10) was found after accounting for the effect of protopathic bias. In the control analysis, the increased risk of the false outcome (HR 2.09; 95% CI 2.03-2.15) was possibly attributable to detection bias, since we also found that more outpatient visits for ophthalmologists in the cataract surgery group compared to the controls.

Conclusions: Contrary to the IRIS registry reports, our findings suggested cataract surgery was not associated with the increased risk of RVO after the consideration of preoperative examination and possible detection bias.