Hematoma block or procedural sedation and analgesia, which is the most effective method of anesthesia in reduction of displaced distal radius fracture?

The inclusion criteria were the following: (1) published randomized controlled trials (RCTs) comparing the efficacy of HB and PSA during reduction of distal radius fracture on pain control, (2) articles including patients with the diagnosis of distal radius fracture, and (3) articles that were clinical trials including adult or pediatric populations. No limit was set for the length of study follow-up. The exclusion criteria were (1) articles that were not controlled intervention studies, (2) animal studies, and (3) articles comparing the efficacy of HB and PSA after reduction or operation.

Two authors (YP Chen and PT Tseng) searched PubMed, ScienceDirect, Cochrane Library, and ClinicalTrials.gov using the keywords of “(radius fracture) AND (hematoma block OR local anesthesia)” from inception to July 28, 2017 without any limitations applied. In addition, the reference lists of review articles relevant to this topic were manually searched to identify potentially eligible papers [9, 14].

After completing the searches, duplicate studies were removed, and the above two authors independently screened the titles and abstracts for eligibility. A list of potentially eligible studies was developed by consensus, and the full texts were assessed by the two authors. Both authors then applied the eligibility criteria and developed a final list of studies to be included. A third reviewer was available for mediation in the event of any inconsistencies.

The methodological quality of the included studies was determined by two authors using the Jadad scale [15]. In brief, the Jadad score was calculated for each study and included three categories of study quality: randomization, blindness, and withdrawals and dropouts. The Jadad score ranged from zero (poor quality) to five (high quality).

The primary outcomes were the differences in pain obtained by HB and PSA during reduction procedures. The outcomes of interest were recorded using validated scales with specific scoring systems, e.g., visual analogue scale (VAS) [16], numeric pain rating scale (NRS) [17], Wong-Baker FACES Pain Rating Scale [18], or Procedure Behavior Checklist (PBCL) [19].

The secondary outcomes of interest included differences in pain severity after reduction procedure by HB or PSA, differences in reduction failure, and adverse effects (AEs) between HB or PSA during reduction.

The two authors extracted data from the eligible studies into a database. The variables of interest included mean age (years), gender, pain severity, ethnicity (Caucasian, African-American, Hispanic, Asian, or Native American), and Jadad score. For articles that did not provide data, we contacted the corresponding authors twice over a month-long period in an attempt to acquire the variables of interest.

Under the presumed heterogeneity of the sample populations among all of the recruited studies, the data were analyzed using Comprehensive Meta-Analysis software, version 3 (Biostat, Englewood, NJ). The analytical models were random-effects meta-analysis models rather than fixed effect models [20]. Since not all studies used the same metrics to estimate the same target (pain severity), we converted and merged all the effect sizes (ESs) of different scales regarding pain severity with the Hedges’ g and 95% confidence intervals (95% CIs) according to the protocol in the Comprehensive Meta-Analysis manuals and guidance on the Comprehensive Meta-Analysis website (https://​www.​meta-analysis.​com/​downloads/​Meta-analysis%20​Converting%20​among%20​effect%20​sizes.​pdf). A two-tailed p value less than 0.05 was considered statistically significant.

Heterogeneity was assessed using the Q statistic to evaluate the dispersion of the true effect among the recruited studies [21]. The I ² statistic was used to indicate the proportion of heterogeneity among study estimates [22]. Publication bias was evaluated by visually inspecting funnel plots [23] and by performing Egger’s regression tests [24]. In cases of publication bias, the Duval and Tweedie trim and fill test was performed to impute potential missing trials and re-calculate the effect size [25]. Sensitivity was tested by performing one-study removal to examine whether the results of our meta-analysis were influenced by any outliers within the recruited studies [26].

Finally, meta-regression and subgroup meta-analyses were performed to investigate the effects of clinical variables as possible sources of heterogeneity. Meta-regression analyses were performed using the unrestricted maximum likelihood method only when data were available across more than five studies. Subgroup meta-analysis was performed when at least two datasets were available.

After screening the titles/abstracts, nineteen full-text articles were assessed for eligibility. Of these, fourteen were excluded (Fig. 1 and Additional file 2: Table S2). Therefore, this meta-analysis included five articles [10‐12, 27, 28]. Of these, four were randomized controlled trials [10, 12, 27, 28], and one was a non-randomized controlled trial [11] in which subjects were subdivided according to patients and family’s own will. Pain severity was evaluated by PBCL in one study [27], by Wong-Baker FACES in one study [11], by NRS pain score in one study [10], and by VAS in two studies [12, 28]. Furthermore, the subjects were adolescents in two studies (children with HB = 73, mean age = 10.3, mean female proportion = 32.7; children with PSA = 81, mean age = 10.3, mean female proportion = 34.7) [11, 27], and the subjects were adults in the other three studies (patients with HB = 152, mean age = 43.9, mean female proportion = 48.0; patients with PSA = 153, mean age = 43.4, mean female proportion = 47.7) [10, 12, 28]. To avoid potential bias, these two distinct groups were not pooled to perform meta-analysis. To be specific, subgroup meta-analysis was performed according to age difference.

Across the five studies, the average Jadad score was 2.80 with a standard deviation (SD) of 1.48 (Additional file 3: Table S3).

The meta-analysis results further revealed that, in pediatric patients, HB also obtained a larger reduction in pain severity compared to PSA (Hedges’ g − 0.402, 95% CI − 0.718 to − 0.085, p = 0.013) (Fig. 2(C)) without significant heterogeneity (Q value = 0.300, df = 1, p = 0.584; I ²  < 0.001%, tau < 0.001). Publication bias was not evaluated because less than three datasets were used.

Further meta-analyses of secondary outcome, including differences of reduction failure after reduction procedure as well as differences in complications and AEs, were not compared between HB or PSA because less than two datasets were available.

Additional file 5: Table S4 summarizes the secondary outcomes. In the adult patient group, Myderrizi et al. and Singh et al. reported no difference in the rate of reduction failure at the early or late follow-up respectively [12, 28]. In the pediatric patient group, Bear et al. also reported no difference in the rate of further reduction at the final follow-up [11]. Furthermore, nausea, vomiting and respiratory distress were the most common AEs developing after HB or PSA. In the adult patients analyzed by Fathi et al., 11% had early AEs in the PSA group, but no patients had AEs in the HB group [10]. In the pediatric group, Bear et al. reported 7.7% self-limited paresthesia after HB [11]. However, the rates of other AEs did not significantly differ between HB and PSA in Bear et al. and Luhmann et al. [11, 27].

This meta-analysis has several limitations. First, although only RCTs and well-controlled intervention studies were considered for inclusion, study quality varied among the recruited studies. Potential sources of bias in these trials included inadequate methods to conceal random allocation as well as lack of blinding. Second, the intent was to compare the benefits and effectiveness between HB and PSA. However, the regimens of intervention varied among these enrolled studies [10‐12, 27, 28]. For example, options of local anesthesia for HB in our enrolled studies included lidocaine and xylocaine of varying concentrations. In addition, the PSA regimens also widely varied among adult and pediatric patients, which may also contribute to variable AEs and result in heterogeneity and potential bias. However, in clinical practice, it was unavoidable to introduce additional sedation procedure to avoid irritation or agitation during reduction of distal radius fracture when HB was performed in pediatric patients. This additional sedation procedure with oral midazolam or inhalant nitrous oxide had been seen in the current study and may also increase events of AEs [11, 27]. To avoid this potential bias, adult and pediatric groups were separated to perform subgroup meta-analysis according to adult/pediatric age difference. However, since subgroup analysis dilutes the number of datasets and cases, it would have decreased the power of the conclusion achieved by our meta-analysis. Third, the technique used to perform HB in the adult population also varied. For example, in Fathi et al., HB was guided by ultrasound to augment the efficacy of pain relief owing to better and more accurate local injection of anesthetics into the fracture site so as to potentially enhance effect of pain relief [10]. Last but not least, quality of fracture reduction plays an important role in successful treatment of displaced distal radius fracture and may decide the risk of reduction loss. Besides, cost-effectiveness is also another important point of view for comparing the outcome between HB and PSA during reduction of displaced distal radius fracture. However, the pity of it is that all of the enrolled studies in our meta-analysis focus neither on the quality of fracture reduction nor on the cost of the different methods of anesthesia. More well-organized randomized control trials, including considerations for outcomes of reduction quality and cost-effectiveness under HB or PSA, may be necessary for future research in this field.