Meta-analysis of visual pretreatment for the prevention of emergence delirium in children undergoing ophthalmic surgery

Eligible studies published up to October 2021 were independently identified by two reviewers using four databases: PubMed, Embase, the Cochrane Library and China National Knowledgement Infrastructure. The terms ‘visual pretreatment’ and ‘ED’ and their analogs were used to develop the basic search query. References of included studies were also independently checked by two reviewers to identify additional studies. The detailed search strategy used in the individual databases is described in Supplementary Table 1.

The authors included studies based on the following characteristics: study design (randomized controlled trials fully published in English and Chinese), participants (preschool children with American Society of Anesthesiologists physical status 1 and 2 assigned to receive elective ophthalmic surgery), intervention (visual pretreatment) and outcomes (incidence of ED, modified Yale Preoperative Anxiety Scale [m-YPAS] score, incidence of propofol rescue and adverse events). The exclusion criteria were as follows: ineligible study design (e.g., case reports and comments), essential data missed and not provided by corresponding author and repeated reports of the same group.

Two reviewers independently extracted the relevant data from the eligible studies. Data included author names, publication year, country of origin, study duration, type of surgery, total number of participants, baseline patient characteristics (mean age, sex), details of intervention and control regimens, outcomes and details of methodological quality. For data that were reported as median and interquartile range, the authors estimated mean and standard deviation according to recognized formulas [23]. Disagreements were resolved through consulting a third reviewer. When necessary, the authors contacted the corresponding author to obtain additional information.

The incidence of postoperative ED was regarded as the primary outcome in this meta-analysis and was defined as a peak score of ≥10 on the Pediatric Anesthesia ED scale [24], a score of ≥3 on the Watcha scale [25] or a score of ≥2 on the Nursing Delirium Screening Scale [26]. The level of preoperative anxiety and the incidence of propofol rescue and adverse events (recorded by individual study) were recognized as secondary outcomes. The level of preoperative anxiety was evaluated using m-YPAS [27]. The incidence of propofol rescue was defined as the proportion of patients requiring an additional dose of propofol for the treatment of severe agitation [17].

Two independent reviewers used the Cochrane risk of bias tool [28] to assess the methodological quality of individual eligible studies based on seven items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other bias. According to the criteria of the Cochrane tool, the bias risk of each item was labeled low, unclear or high.

Pooled estimates of categorical and continuous variables were expressed as risk ratio and mean difference, respectively. A 95% CI was used for each effect size. Cochran's Q statistic [29] was used to qualitatively assess statistical heterogeneity, and the I² statistic [30] was used to quantify the level of statistical heterogeneity across studies. Nevertheless, the authors used a random effects model [31] to perform all statistical analyses regardless of the level of heterogeneity because variations between studies could not be ignored in real settings. The authors did not draw a funnel plot to inspect publication bias because an insufficient number of eligible studies were identified [32]. A p < 0.05 was considered to indicate a statistically significant difference. The authors used RevMan 5.3 software (Cochrane Collaboration, London, UK) to perform statistical analysis [33].

After initially searching four electronic databases, a total of 14 records were identified. The authors subsequently used literature management software (EndNote, Clarivate Analytics) to remove three duplicate records. Next, the authors excluded seven ineligible studies after carefully checking the titles and abstracts of the remaining 11 studies. At the final stage of the eligibility evaluation, no ineligible study was identified based on full texts. Finally, four studies involving 679 preschool children were included in the meta-analysis [17,19–21]. The process of study retrieval and selection is displayed in Figure 1.

Of the four included studies, one was published in 2018 [17] and three were published in 2021 [19–21]. All included studies [17,19–21] were performed in China. The four included studies involved different types of surgeries. Three studies [17,19,21] were designed with two arms and one study [20] was designed with three arms. The sample size of the individual studies ranged from 95 to 279, with a total number of 679. The basic characteristics of the included studies are summarized in Table 1.

The Cochrane risk of bias tool was applied to assess the methodological quality of the four included articles, and the risk of bias summary is shown in Figure 2. Results suggested that all included studies [17,19–21] correctly generated a random sequence; however, only one study [17] reported the details of allocation concealment with the appropriate method. The authors labeled all included studies [17,19–21] with low risk of bias in performance bias because behaviors of participants and personnel could not bias outcomes. With regard to the remaining four items, the four included studies were deemed to have a low risk of bias. Overall, the methodological quality of the included studies was moderate to high.

All included studies [17,19–21] reported the incidence of postoperative ED. Based on results of the statistical heterogeneity examination, it was observed that there was no obvious statistical heterogeneity in the incidence of postoperative ED between the visual pretreatment and standard control groups (I² = 0%; p = 0.96). The results revealed that the estimate of the incidence of postoperative ED in the visual pretreatment and standard control groups was 0.39 (95% CI: 0.31–0.49). In summary, the difference in the incidence of postoperative ED between both groups was statistically significant (21.43 vs 55.79%; Z = 8.09; p < 0.001) (Figure 3). In addition, one study investigated the comparative efficacy of visual pretreatment versus dexmedetomidine, and the meta-analysis suggested no statistical difference in terms of incidence of postoperative ED (risk ratio: 0.84; 95% CI: 0.24–2.91; p = 0.78).

The authors analyzed the differences in m-YPAS score between the visual pretreatment and standard control groups (Figure 4). Based on results of the heterogeneity examination, obvious statistical heterogeneity was detected in m-YPAS score between the two groups (I² = 81%; p = 0.005). The results revealed that the pooled effect of the m-YPAS score in both groups was -3.66 (95% CI: -9.96 to 2.65). In summary, the m-YPAS score in the visual pretreatment group was numerically lower than that observed in the standard control group, although the difference was not statistically significant (Z = 1.14; p = 0.26). The meta-analysis also suggested no statistical difference in terms of m-YPAS score between the visual pretreatment and dexmedetomidine groups (mean difference: 3.00; 95% CI: -1.98 to 7.98; p = 0.24).

The authors compared and analyzed differences in the incidence of propofol rescue between the visual pretreatment and standard control groups (Figure 5). Statistical heterogeneity examination revealed no significant heterogeneity (I² = 0%; p = 0.99), and results of the meta-analysis suggested a significantly lower incidence of propofol rescue in the visual pretreatment group compared with the standard control group (5.22 vs 17.99%; risk ratio: 0.29; 95% CI: 0.13–0.65; Z = 3.01; p = 0.003).

Among included studies, three studies [17,20,21] reported the incidence of PONV, and two studies [17,21] reported the incidence of breath holding and SpO₂ <95%. Statistical heterogeneity examinations did not detect obvious heterogeneity in terms of the incidence of PONV (heterogeneity not applicable), breath holding (I² = 0%; p = 0.98) or SpO₂ <95% (heterogeneity not applicable). Results of the meta-analyses revealed a comparable incidence between the visual pretreatment and standard control groups in terms of PONV (1.14 vs 2.20%), breath holding (2.24 vs 2.16%) and SpO₂ (0.75 vs 0.72%) (Figure 6).

ED is commonly observed in children undergoing pediatric anesthesia, especially preschool children maintained via the inhalation of sevoflurane [11]. Nonpharmacological methods have been suggested to reduce the incidence of postoperative ED because anesthetics and sedatives have been found to be associated with an increased risk of many undesirable side effects [16]. In the present meta-analysis of four eligible studies, the authors demonstrated that, as a nonpharmacological method, visual pretreatment is effective in decreasing the incidence of postoperative ED and propofol rescue without significant adverse effects in children undergoing ophthalmic surgery.

Application of eye ointment and an eye patch after ophthalmic surgery may play a critically important role in the occurrence of postoperative ED [17]. Owing to immature neural development, visual disturbance has been considered to be the main reason for the high incidence of postoperative ED [34], and therefore visual pretreatment using an eye patch on the eyes the day before surgery has been speculated to have a positive impact on the occurrence of postoperative ED in children undergoing ophthalmic surgery [17,18]. All studies [17,19–21] included in the current meta-analysis consistently revealed a beneficial effect of visual pretreatment with regard to reducing the incidence of postoperative ED. The authors' meta-analysis further demonstrated the beneficial role of visual pretreatment in the prevention of postoperative ED through increasing statistical power after combining the results of individual studies [33]. It should be noted that the use of visual pretreatment is also comparable to dexmedetomidine in the prevention of postoperative ED [20]. Moreover, the authors' meta-analysis suggested that visual pretreatment significantly reduces the proportion of patients requiring propofol rescue, which further confirms the preventive efficacy of visual pretreatment for postoperative ED in children undergoing ophthalmic surgery.

Preoperative anxiety has been speculated to be associated with the occurrence of postoperative ED in children undergoing ophthalmic surgery [35]. To date, several studies have investigated the relationship between preoperative anxiety and the incidence of postoperative ED [11]; however, conflicting results were reported [36–38]. Of the three included studies [17,19,20] that reported preoperative anxiety, only one [19] reported a significant reduction in m-YPAS score in the visual pretreatment group. In this meta-analysis, the authors detected a greater reduction in m-YPAS score in the visual pretreatment group, although no statistical difference was observed. Moreover, Lin et al. suggested that preoperative anxiety was still an independent risk factor for postoperative ED [17], although preoperative anxiety scores were comparable between both groups studied. Thus, it is necessary to further clarify whether preoperative anxiety is related to the occurrence of postoperative ED in children undergoing ophthalmic surgery.

As mentioned previously, because pharmacological interventions are associated with a higher risk of adverse effects, several nonpharmacological methods have been developed and recommended for the alleviation of ED [16]. In the current meta-analysis, the authors detected a comparable incidence of PONV, breath holding and SpO₂ <95% between the visual pretreatment and standard control groups, which further demonstrates the safety profile of visual pretreatment in practice.

Certainly, the authors' meta-analysis has some limitations that should be further explained. First, as is the case with some studies, most included studies had a relatively insufficient sample size, which may introduce choice, biasing the authors' results. Second, only four eligible studies were included in this meta-analysis, which might weaken the robustness and credibility of the authors' findings. Third, although all included studies used visual pretreatment regimens as an intervention, a variety of components were used in the design of the visual pretreatment protocols (Table 2). Because of the insufficient number of eligible studies, the authors could not conduct subgroup analysis to eliminate the possible impact of components on pooled results. Fourth, various ophthalmic surgeries were utilized in the included studies, which may also introduce bias, negatively affecting the authors' results. Fifth, different scales were used to identify ED in all included studies, and it is impossible to eliminate the impact of scales on the pooled results. Sixth, all studies included in this meta-analysis were conducted in China, and therefore the authors' findings should be cautiously interpreted in different clinical settings. Seventh, the formal protocol of the present meta-analysis was not registered on a public platform; however, the authors performed this meta-analysis in strict accordance with the Cochrane handbook. In spite of these limitations, the authors' study has two major advantages: it is the first meta-analysis to systematically evaluate the efficacy and safety of visual pretreatment for the prevention of postoperative ED in children undergoing ophthalmic surgery, and a systematic literature search was utilized to significantly reduce the risk of reporting bias.