Sir,

We thank Dr Ng et al. for their interest and comments on our report. In our study, we use a broader definition as: codes of endophthalmitis in subsequent outpatient visits or admissions, or at the index cataract surgery were considered as the occurrence of postoperative endophthalmitis. Concerning the use of this definition, for we believe that postoperative endophthalmitis is not necessarily treated in admission, and this condition can also be treated in outpatient visits. Unlike EPSWA study and other previous studies using hospital-based data or define admission for endophthalmitis as the occurrence of postoperative endophthalmitis.1, 2, 3 If we use the definition as EPSWA study for endophthalmitis as the index of occurrence, we would exclude patients that were only treated in outpatient clinics but not in hospitals. Therefore, the incidence of postoperative endophthalmitis will be underestimated. The 2-year incidence of postoperative endophthalmitis in Taiwan will be 0.26%. It is similar to Dr Ng's report.

Different from what Dr Ng pointed out, we did consider the length of stay as an important factor and control this variable in Cox regression analysis in our study, as presented in Table 4 of our previous article.4 In our study population, there were 88.1% outpatient cataract surgery and 11.9% inpatient cataract surgery with 1.53±0.80 days of the mean length of stay. As the percentage of inpatient cataract surgery was relatively low and the length of stay was short, we used ‘site of operation’ as the variable and divided patients into the outpatient and inpatient cataract surgery categories. Even after we adjusted site of operation and other factors, we found that hospital volume and surgeon volume were still significantly associated with endophthalmitis risks. Unlike EPSWA report,5 we found that inpatient cataract surgery posed a higher risk for postoperative endophthalmitis than outpatient cataract surgery in hospital volume model (HR=1.33, P=0.014), but not in surgeon volume model (HR=1.25, P=0.065).