We included eight randomised and two quasi-randomised controlled trials in the review, encompassing 505 participants. Seven studies were from Europe (Germany, Sweden and the UK), two from East Asia (Hong Kong and Japan) and one from Australia. Nine of the studies examined treatments for episodes of recurrent corneal erosions and one study considered prophylaxis to prevent development of recurrent corneal erosions after injury. Two of the nine treatment studies also enrolled participants in a study of prophylaxis to prevent further episodes of recurrent corneal erosions. The studies were poorly reported; we judged only one study low risk of bias on all domains.
Two studies compared therapeutic contact lens with topical lubrication but one of these studies was published over 30 years ago and used a therapeutic contact lens that is no longer in common use. The more recent study was a two-centre UK study with 29 participants. It provided low-certainty evidence on resolution of symptoms after treatment with similar number of participants in both groups experiencing resolution of symptoms at four months (risk ratio (RR) 0.97, 95% confidence interval (CI) 0.62 to 1.53). There was very low-certainty evidence on recurrence after partial or total resolution at seven months’ follow-up (RR 1.07, 95% CI 0.07 to 15.54). There was no evidence of an important difference in pain score (score of 3 in the contact lens group and score of 2 in the topical lubrication group, low-certainty evidence) and no adverse effects were reported. The older study, using a contact lens no longer in common use, found an increased risk of pain and complications with the contact lens compared with hypromellose drops and paraffin ointment at night.
A single-centre, Australian study, with 33 participants, provided low-certainty evidence of an increased risk of recurrence with phototherapeutic keratectomy compared with alcohol delamination but with wide confidence intervals, compatible with increased or decreased risk (RR 1.27, 95% CI 0.48 to 3.37). Time to recurrence was similar in both groups (6.5 and 6 months, low-certainty evidence). On average people receiving phototherapeutic keratectomy reported less pain but confidence intervals included no difference or greater pain (mean difference (MD) -0.70, 95% CI -2.23 to 0.83, low-certainty evidence). No adverse effects were reported.
A 48-participant study in Hong Kong found recurrences were less common in people given diamond burr superficial keratectomy after epithelial debridement compared with sham diamond burr treatment after epithelial debridement (RR 0.07, 95% CI 0.01 to 0.50, moderate-certainty evidence). The study did not report pain scores but adverse effects such as corneal haze (RR 0.92, 95% CI 0.06 to 13.87, low-certainty evidence) and astigmatism (0.88 versus 0.44 dioptres, moderate-certainty evidence) were similar between the groups.
A study comparing transepithelial versus subepithelial excimer laser ablation in 100 people found low-certainty evidence of a small increased risk of recurrence of corneal erosion at one-year follow-up in people given the transepithelial compared with subepithelial technique, however, the confidence intervals were wide and compatible with increased or decreased risk (RR 1.20, 95% CI 0.58 to 2.48, low-certainty evidence). Other outcomes were not reported.
Other treatment comparisons included in this review were only addressed by studies published two decades or more ago. The results of these studies were inconclusive: excimer laser ablation (after epithelial debridement) versus no excimer laser ablation (after epithelial debridement), epithelial debridement versus anterior stromal puncture, anterior stromal puncture versus therapeutic contact lens, oral oxytetracycline and topical prednisolone (in addition to ‘standard therapy‘) versus oral oxytetracycline (in addition to ‘standard therapy‘) versus ‘standard therapy‘.