The preventive effect of dexmedetomidine on anesthesia complications in strabismus surgery: a systematic review and meta-analysis

The report of this article was followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [13]. The protocol of this systematic review was registered in the International prospective register of systematic reviews (PROSPERO), National Institute for Health research (ID: CRD42023438847). In adherence to the principles of Cochrane, three researchers independently searched PubMed, Medline, Embase, Ovid, Cochrane Library, ISI Web of Science, Clinical Trials, and Chinese databases, including China National Knowledge Infrastructure (CNKI), Wanfang Database, and Chinese Biomedical Literature Database (SinoMed), to retrieve eligible studies published up until May 2022 since the inception of these databases. Keywords used in the search included "dexmedetomidine," "α2 agonist," "children," "strabismus," "eye surgery," and "ophthalmic surgery." After generating a list of potentially eligible studies, the researchers conducted manual review and verification to determine the final selection for inclusion in the analysis.

This study only included Randomized Controlled Trials (RCTs) related to strabismus surgery that compared the use of dexmedetomidine with a placebo or other anesthetic agents, without restrictions on the route of administration or dosage of the drug. Studies that were of types such as reviews, cohort studies, or case reports, and other non-RCT studies were excluded. We only included the studies written in English or Chinese. The three independent researchers then screened the retrieved literature, removed duplicates that arose during database integration, and read the titles and abstracts of the articles for initial selection. The full texts of the preliminarily selected articles were obtained and thoroughly read to determine if they met the inclusion criteria. Any discrepancies during the process were resolved through consultation and discussion.

The primary outcomes of interest in this study were the incidence of postoperative delirium, postoperative nausea and vomiting (PONV), and postoperative pain as well as the number of patients requiring additional analgesics postoperatively. Additionally, the study also assessed the incidence of the oculocardiac reflex (OCR).

Two independent researchers extracted the necessary information from the included studies, including author, publication year, patient recruitment period, inclusion criteria, sample size, baseline data, comparison groups, intervention duration, primary anesthetic agents, and data on primary and secondary outcomes. A third researcher was involved in information verification and data cross-checking to ensure the accuracy and reliability of the information and data.

Two researchers assessed the bias and risk of the eligible studies using the Cochrane Risk of Bias tool [14]. This tool includes the following sections: selection bias due to random sequence generation or allocation concealment; performance bias due to participants or personnel being aware of the assigned interventions; detection bias due to blinding deficiencies when obtaining outcomes; reporting bias due to selective reporting of outcomes, and other biases. Each risk of bias level is categorized as low, high, or unclear, and the results are represented with different colored blocks and corresponding risk of bias graphs. In case of discrepancies in the evaluation, the assessment of a third researcher is adopted for the final evaluation.

For dichotomous variables, this study employs the risk ratio (RR) along with the corresponding 95% confidence intervals (CI) for outcome analysis. The heterogeneity of the studies is assessed using the I² statistic and the chi-square test, and if substantial heterogeneity is observed, subgroup analyses are conducted to explore the reasons. I² value is more than 50%, which means the high heterogeneity and the random model should be used. I² value is less than 50%, which means the low heterogeneity and fixed model should be used. Egger's test are employed to assess publication bias. Forest plots and funnel plots are created using Review Manager software (Version 5.4, The Nordic Cochrane Centre, Copenhagen, Denmark). Statistical analyses are performed using STATA (Version 14.0, Stata Corp, College Station, TX, USA). A difference is considered to be statistically significant when P < 0.05.

Figure 1 depicts the flow of study selection and screening. In this research, a comprehensive and exhaustive search was carried out across common databases, yielding a total of 478 search results. After merging the databases, 125 duplicate records were removed. An initial screening was conducted by reviewing titles and abstracts, resulting in the exclusion of 276 articles, along with the removal of 16 articles for which the full text was not accessible. This left a total of 61 articles. All 61 articles underwent detailed full-text review and examination. Subsequently, 32 articles that did not meet the inclusion criteria, 8 non-randomized controlled trial (RCT) articles, and 8 articles where key data needed for the study were not obtainable were excluded. Ultimately, 13 studies that fulfilled the research criteria were included in the analysis for further quantitative and qualitative assessments [11, 12, 15‐25]. Of these, 4 studies utilized intranasal administration [11, 12, 18, 23], while the remaining 9 employed intravenous administration [15‐17, 19‐22, 24, 25]. The characteristics of the included studies were shown in the Table 1.

Figure 2 displays the results of the bias risk assessment. The majority of studies had a low risk of bias, with two instances containing high-risk bias scores: one study did not employ blinding during random allocation [16], and another study exhibited high risk during outcome evaluation [21]. Additionally, three studies were considered to have unclear other risks because the sample size was not calculated during the trial design [18, 20, 21].

This study analyzed the incidence of postoperative delirium and severe delirium with dexmedetomidine compared to a placebo (saline solution). A total of 13 articles reported the number of cases of postoperative delirium, and no significant heterogeneity was found among the studies (P = 0.88, I2 = 0%). A fixed-effects model was utilized. The pooled RR value was 0.73, with a 95% CI of (0.60, 0.88), and the combined effect size Z = 3.30 (P = 0.001). The results showed that compared to the placebo group, the dexmedetomidine group had a significantly lower incidence of postoperative delirium, as seen in Fig. 3. Egger's tests (P = 0.0085) indicated that there was no publication bias.

Furthermore, eight articles reported the number of cases with severe postoperative delirium. Again, no significant heterogeneity was found among the studies (P = 0.50, I2 = 0%). A fixed-effects model was employed. The pooled RR value was 0.45, with a 95% CI of (0.26, 0.79), and the combined effect size Z = 2.81 (P = 0.005). The results showed that compared to the placebo group, the dexmedetomidine group had a significantly lower incidence of severe delirium, as seen in Fig. 4. Egger's tests (P = 0.0063) indicated that there was no publication bias.

This study also analyzed the incidence of postoperative nausea and vomiting (PONV) with dexmedetomidine compared to a placebo (saline solution). A total of 8 articles reported the number of cases of postoperative nausea and vomiting, and no significant heterogeneity was found among the studies (P = 0.63, I2 = 0%). A fixed-effects model was used. The pooled RR value was 0.48, with a 95% CI of (0.33, 0.69), and the combined effect size Z = 3.98 (P < 0.0001). The results showed that compared to the placebo group, the dexmedetomidine group had a significantly lower incidence of PONV, as seen in Fig. 5. Egger's tests (P = 0.0072) indicated that there was no publication bias.

This study analyzed the incidence of the need for supplemental analgesia for postoperative pain when using dexmedetomidine compared to a placebo (saline solution). A total of 7 articles reported the number of cases requiring supplemental analgesia for postoperative pain, and no significant heterogeneity was found among the studies (P = 0.39, I2 = 5%). A fixed-effects model was used. The pooled RR value was 0.60, with a 95% CI of (0.43, 0.85), and the combined effect size Z = 2.89 (P = 0.004). The results showed that compared to the placebo group, the dexmedetomidine group had a significantly lower incidence of the need for supplemental analgesia for postoperative pain, as seen in Fig. 6. Egger's tests (P = 0.013) indicated that there was no publication bias.

This study analyzed the incidence of the oculocardiac reflex (OCR) when using dexmedetomidine compared to a placebo (saline solution). A total of 9 articles reported the number of cases with oculocardiac reflex postoperatively, and significant heterogeneity was found among the studies (P = 0.02, I2 = 59%). A random-effects model was employed. The pooled RR value was 0.82, with a 95% CI of (0.55, 1.21), and the combined effect size Z = 1.01 (P = 0.31). The results showed that there was no significant change in the incidence of OCR in the dexmedetomidine group compared to the placebo group, as shown in Fig. 7. Egger's tests (P = 0.432) indicated that there was publication bias.

Due to the significant heterogeneity and bias observed in the OCR analysis, a subgroup analysis was conducted. The studies were categorized into intravenous administration and intranasal administration groups for subgroup analysis. A total of 5 articles reported the number of cases with OCR postoperatively following intravenous administration of dexmedetomidine, and no significant heterogeneity was found among these studies (P = 0.42, I2 = 0%). A fixed-effects model was used. The pooled RR value was 0.50, with a 95% CI of (0.33, 0.77), and the combined effect size Z = 3.18 (P = 0.001). The results showed that the incidence of OCR was significantly reduced in the group receiving dexmedetomidine intravenously compared to the placebo group, as seen in Fig. 8. Egger’s tests (P = 0.0074) indicated that there was no publication bias. A total of 4 articles reported the number of cases with OCR postoperatively following intranasal administration of dexmedetomidine, and no significant heterogeneity was found among these studies (P = 0.59, I2 = 0%). A fixed-effects model was used. The pooled RR value was 1.22, with a 95% CI of (0.93, 1.58), and the combined effect size Z = 1.46 (P = 0.15). The results showed that there was no significant change in the incidence of OCR in the group receiving dexmedetomidine intranasally compared to the placebo group, as seen in Fig. 8. Egger’s tests (P = 0.0113) indicated that there was no publication bias. There was high heterogeneity between the two subgroups, with P = 0.0005, I2 = 91.7%.