Risk of bias assessment
Three RCT studies reported the randomization process, intervention deviation, yet none of them had selective reporting and outcome data missing. The study of Qian L30 did not describe the approach of random sequence generation and potential intervention deviation. Vasavada et al.22 reported that the surgery type was revealed to the surgeon and outcome examiner in their study. The details of risk assessment for these RCTs were presented in the Figure 2.
Retrospective studies were assessed based on NOS. Five studies described the detailed process of cohort selection, and were assessed to be of low risk in cohort selection. The study by Autrata et al.31 did not report the source of study subjects, and was assessed to be of medium risk in cohort selection. High risk was found in the cohort selection in the study by Trivedi et al32, as the source and preoperative data on patients were not described. Given the age at the surgery and the age at the last visit were significant factors for the postoperative outcomes in cataract infants, these two factors were used to assess the cohort comparability. According to our inclusion and exclusion criteria, all these retrospective studies have similar age at surgery and last visit in both groups. So, all these studies were assessed to be of low risk in the cohort comparability. The study by Trivedi et al32. was assessed to be of high risk since they did not report the duration of follow up and rate of follow-up loss. Except for the above study, all the other studies gave detailed information on outcome assessment. The details of risk assessment for these retrospective studies were shown in the Figure 3.