Search results
The method used to select the studies is shown in Fig 1. The initial search identified 273 studies (80 from MEDLINE, 48 from EMBASE, 125 from Web of Science, 12 from Cochrane, 1 from CNKI, 5 from Wanfang and 2 studies identified from reference lists). After removing duplicates, 150 citations were reviewed. A total of 142 publications were excluded for the following reasons: 112 did not belong to the comparison for IOP after corneal refractive surgery, 9 were conducted with animals, 15 were reviews, 3 were unable to extract the mean or standard deviation of the three kinds of IOP, 2 studies were in French without an English translation, and 1 without special follow-up time. Therefore, the meta-analysis was comprised of 8 full articles.
Study characteristics
The characteristics of the studies included in our analysis are presented in Table 1. Of the 8 articles, Kirwan’s, Qazi’s and Denise’s all included data of different follow-up time, so there were 13 groups of data in this study. 5 of the 8 studies included were prospective and 3 were retrospective. Three of the studies were conducted in America and two studies were conducted in China, others were each in Ireland, Korea, and Iran. The sample sizes varied from 28 to 148, and the total was 724. The average age is between the 20 and 40 years old. The longest follow-up time was 12 months, the shortest was only 1 week, and the rest were 1 month, 3 months and 6 months respectively.
Quality assessment
The QUADAS2 tool was applied to assess for bias and the quality evaluation results of the included literatures are shown in Fig 2. The reference standard in five articles was highly biased[12-14, 17, 18] and the patient selection in only one article was highly biased [17]. For flow and timing, all literatures showed lowly biased[12-19]. In general, the quality of the included literatures is relatively high.
Analysis of postoperative IOPcc and IOPGAT
Fig. 3 gave the forest plot of the correlation between postoperative IOPcc and IOPGAT. There was significant heterogeneity among the groups of data (p < 0.0001, I2 = 71%), so the random effect model was used for analysis. In the pooled analysis, the WMD between IOPcc and IOPGAT was 2.67 mmHg (95% CI: 2.20~3.14 mmHg, p < 0.0001, Fig 3). The Egger statistic (p = 0.028) revealed there was certain publication bias.
We performed subgroup analysis using the study design, surgical procedure and postoperative follow-up time as sub-group criteria respectively. In the subgroup analysis of postoperative IOPcc and IOPGAT, as shown in Table 2, the heterogeneity among the data on surgical procedure was zero, while the heterogeneity of other subgroups was still more than 50%.
The meta-regression results showed no statistical significance for the effect of three characteristics (study design, surgical procedure, post-op follow-up) on heterogeneity, namely, study design (p = 0.9747), surgical procedure (p = 0.0976), post-op follow-up (p = 0.2983 (1 month), p = 0.5096 (3 months)).
Analysis of postoperative IOPg and IOPGAT
Fig. 4 gave the forest plot of the correlation between postoperative IOPg and IOPGAT. There also was little heterogeneity among the data of groups (p = 0.0025, I2 = 60%), and random effect model was used for analysis. In the pooled analysis, the WMD between IOPg and IOPGAT was -0.27 mmHg (95% CI: -0.70~0.16 mmHg, p = 0.2174, Fig 4) and Egger statistics (p = 0.1339) showed no publication bias.
Although the Fig. 4 showed that there was some heterogeneity between IOPg and IOPGAT after operation, there was no significant difference between them in general, so the heterogeneity between IOPg and IOPGAT had not been analyzed.
Comparison of IOP pre- and post-operative surgery
In the 8 studies, only four groups of data[15-18] contain preoperative IOP and ∆IOP, where ∆IOP refers to the difference between the value IOP obtained from pre- and post-operation. The data were summarized in Table 3. Through the meta-analysis, the WMD between preoperative IOPcc and IOPGAT was 1.52 mmHg (95% CI: 0.97~2.07 mmHg, p < 0.0001), and the WMD between preoperative IOPg and IOPGAT was 1.16 mmHg (95% CI: 0.60~1.73 mmHg, P < 0.0001). And the three ∆IOP values from the largest to the lowest were shown as: mean-∆IOPg = 3.83 mmHg, mean-∆IOPGAT = 2.65 mmHg, mean-∆IOPcc = 1.43 mmHg.