Sir,

The article by Lyall et al1 entitled ‘Post-intravitreal anti-VEGF endophthalmitis in the United Kingdom: incidence, features, risk factors, and outcomes’ is timely, given current efforts to minimize the risk of endophthalmitis following intravitreal injections. We take great interest in this topic, and support the authors’ efforts to identify risk factors for endophthalmitis. However, we are concerned with biases inherent to their study design and the limitations of using data from questionnaires. Our three major issues include questionnaire validation, lack of povidone-iodine data, and inadequate control group selection.

The use of questionnaires to obtain information about clinical case histories, treatments, outcomes, and complications is subject to inaccuracies. The authors do not state whether the questionnaires used were validated. Without an attempt to validate the questionnaire, one cannot be certain about the accuracy or validity of the data.2

Although the authors acknowledge that data regarding povidone-iodine were not collected, the failure to administer povidone-iodine may have been the underlying risk factor for many of the 47 endophthalmitis cases reported. Povidone-iodine is well known to reduce the rate of endophthalmitis after intraocular surgery.3

The authors conclude that failure to administer both immediate pre and post-injection antibiotics is a risk factor for endophthalmitis. The data provided in Table 2 reported that only 8.7% (n=4) of the eyes in the study group with endophthalmitis did not receive immediate post-injection topical antibiotics vs 0% of the control group. The control group used 10 randomly selected sites and was not an appropriate control group. The control cases should have been obtained from the same sites where the study cases were obtained, in order to decrease any unknown biases.

Lyall et al’s1 conclusions are over-reaching regarding the ‘protective’ effects of administering immediate pre and post-injection antibiotics. The lack of questionnaire validation and povidone-iodine data as well as the presence of an inadequate control group should have been addressed. Furthermore, the study should not have been used to serve as an endorsement for the use of topical antibiotics.