Article Text

other Versions

Download PDFPDF
Efficacy and safety of caudal dexmedetomidine in pediatric infra-umbilical surgery: a meta-analysis and trial-sequential analysis of randomized controlled trials
  1. Ushma Jitendra Shah1,2,
  2. Derek Nguyen3,
  3. Niveditha Karuppiaah2,4,
  4. Janet Martin1,4 and
  5. Herman Sehmbi1
  1. 1Anesthesia & Perioperative Medicine, Western University, London, Ontario, Canada
  2. 2Anesthesia & Perioperative Medicine, London Health Sciences Centre Children's Hospital, London, Ontario, Canada
  3. 3Bachelor of Medical Sciences, Western University Schulich School of Medicine and Dentistry, London, Ontario, Canada
  4. 4Centre for Medical Evidence, Decision Integrity & Clinical Impact (MEDICI), Western University, London, Ontario, Canada
  1. Correspondence to Dr Herman Sehmbi, Anesthesia & Perioperative Medicine, Western University, London, Ontario, Canada; Herman.Sehmbi{at}


Background Dexmedetomidine is used as a local-anesthetics adjuvant in caudal block to prolong analgesia in pediatric infra-umbilical surgery.

Objective We evaluated the analgesic efficacy and safety of the addition of caudal dexmedetomidine to local anesthetics (vs local anesthetics alone) in pediatric infra-umbilical surgery.

Evidence review We searched 10 databases for randomized controlled trials (RCTs) of pediatric patients undergoing infra-umbilical surgery, comparing caudal block with and without dexmedetomidine as local anesthetic adjuvant. We performed a frequentist random-effects meta-analysis (R statistical package). We analyzed continuous outcomes as a ratio of means (ROM) and dichotomous data as relative risk (RR), along with 95% CI. We included 19 RCTs (n=1190 pediatric patients) in the meta-analysis. The primary outcome was duration of analgesia (defined as ‘the time from caudal injection to the time at which the study-specific pain score was greater than a cut-off threshold’).

Findings Data from 19 included RCTs (n=1190) suggested that compared with control (mean duration 346 min), the addition of caudal dexmedetomidine significantly prolonged the duration of analgesia (ratio of means 2.14, 95% CI 1.83 to 2.49, p<0.001; ‘moderate’ evidence). Trial-sequential analysis showed adequate ‘information size’ for the primary outcome. Caudal dexmedetomidine also reduced the number of analgesic administrations (‘low’ evidence), total acetaminophen dose (‘moderate’ evidence) and the risk of emergence delirium (‘moderate’ evidence). There were no significant differences in adverse effects such as hypotension, bradycardia, post-operative nausea and vomiting, urinary retention and respiratory depression.

Conclusions Our results suggest that the addition of dexmedetomidine to local anesthetic in caudal block significantly improves the duration of analgesia and reduces the analgesic requirements, while maintaining a similar risk-profile compared with local anesthetic alone. Further data on neurological safety are needed.

  • pediatrics
  • anesthesia
  • local
  • pain
  • postoperative
  • acute pain
  • regional anesthesia

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Caudal block is a time-tested and safe regional anesthesia technique that is commonly used for analgesia in pediatric infra-umbilical surgery.1 2 However, a limited duration of analgesia (4–12 hours) with a standard caudal block (using local anesthetics (LA) only) caps its usefulness as an analgesic technique.3 Using a higher dose to increase this duration may result in toxicity. Unsurprisingly, the Pediatric Regional Anesthesia Network database study identified a 25% incidence of the use of excessive doses for extending analgesia.1 Several adjuvants, in combination with LA, have been used to prolong the duration of analgesia to overcome this reliance on such higher doses. Thus, appropriate use of adjuvants to prevent this excessive dosing may be paramount for patient safety.

The European Society of Regional Anesthesia and Pain Therapy and the American Society of Regional Anesthesia and Pain Medicine joint committee practice advisory on pediatric regional anesthesia has recommended α2-agonists (such as clonidine and dexmedetomidine) for caudal blocks.3 Evidence from neuraxial4 and non-neuraxial studies5 suggest that dexmedetomidine may be more efficacious than clonidine. Thus, current practice will benefit from evidence regarding the pooled estimates of this analgesic effect, and its potential for adverse effects. Given this use is an ‘off-label’ indication, with inherent potential for neurological toxicity, its essential to investigate the existing evidence gap concerning the dose-response relationship of caudal dexmedetomidine. This may allow us to ascertain the lowest possible dose of dexmedetomidine that may provide adequate analgesia. While previous meta-analysis and narrative reviews6–8 have indicated the analgesic benefits of caudal dexmedetomidine, an assessment of adequacy of this ‘information size’ to guide clinical recommendations and a dose-response relationship has not been performed.

In this review, we sought to evaluate the analgesic efficacy and safety of caudal dexmedetomidine, when added to LA (vs LA alone) for analgesia in pediatric infra-umbilical surgery. Our primary outcome was the ‘duration of analgesia’, and we hypothesized that caudal dexmedetomidine significantly prolongs this. In particular, we aimed to perform a comprehensive search, ascertain the adequacy of information size (trial-sequential analysis (TSA)) and investigate dose-response using subgroup analysis.


We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)9 reporting guidelines to prepare this review and registered our protocol (PROSPERO registration number: CRD42018106027). We used the R-statistical package (R Studio V.3.6.0) for statistical analysis. We have summarized deviations from our original protocol in the online supplemental file (sections 3).

Supplemental material

Eligibility criteria

We sought full-text published manuscripts using the following characteristics:

Participants: pediatric patients (aged <18 years) undergoing infra-umbilical surgery (abdominal, urogenital or lower limbs) with or without sedation or general anesthetic.

Interventions: caudal epidural block with dexmedetomidine, regardless of the guidance method, using any concentration, dose or volume of long-acting LA (bupivacaine, levobupivacaine or ropivacaine), with or without epinephrine, and any dose of dexmedetomidine used.

Comparator: caudal block without dexmedetomidine (or other adjuvants).

Outcomes: analgesia—duration of analgesia, the total number of analgesics administrations in 24 hours, total amount of analgesics used in 24 hours; block related—duration of the motor block; adverse effects—emergence time, the incidence of hypotension, bradycardia, respiratory depression, postoperative nausea and vomiting (PONV), urinary retention, emergence delirium and neurological complications.

Studies: published, full text, indexed or non-indexed, randomized controlled trials (RCTs) of interventions.

Restrictions: human subjects; English language.

Exclusions: adult patients (aged ≥18 years), volunteer studies, supra-umbilical surgeries, use of only short-acting (eg, chloroprocaine or lignocaine) or intermediate-acting (mepivacaine) LA.

Information sources and search strategy

We searched 10 databases including the US National Library of Medicine database (Medline and Medline In-Process), PubMed, Excerpta Medica Database (Embase), Cochrane Central Controlled Trials Database Register, Cochrane Systematic Reviews Database, Web of Science, Biosys Previews, Scopus and Cumulative Index to Nursing and Allied Health Literature from inception to July 2018. We used medical subject headings, text words and controlled vocabulary terms relating to ‘caudal’, ‘dexmedetomidine’, ‘pediatric’, ‘infra-umbilical’ and ‘randomized controlled trial’. We retrieved non-indexed articles using Google Scholar. We excluded unpublished studies, conference proceedings, theses and abstracts. We ran an updated search in June 2020 to identify any recently published articles. The search strategy employed is summarized in online supplemental file (section 2).

Study selection

Two authors (UJS and NK) screened retrieved studies independently using a two-level screening process (titles, followed by abstract and full texts). We employed a web-based systematic review software (DistillerSR, Evidence Partners, Ottawa, Canada) for this, and applied the stated criteria (as above) to exclude studies from the analysis. We sought the opinion of a third author (HS) in the case of a disagreement.

Data collection process

We developed and piloted a data extraction template using DistillerSR, and two authors (NK and DN) extracted relevant study characteristics and outcome data independently. The third author (UJS) audited the data. We did not contact original study authors due to multiple studies and resource limitations.

Data items

We extracted and collected data on the following item:

  • Study characteristics: author name, year of publication, study title, journal name, indexing status (indexed or non-indexed), study country, American society of Anesthesiology (ASA) class, age, type of surgery, type of intraoperative anesthesia used and details of the groups with the number of patients analyzed in each group.

  • Block technique and analgesic regimens: LA type, dose (mg/kg), concentration (%), dose of dexmedetomidine (μg/kg), block localization technique (ultrasound, peripheral nerve stimulation or landmark guided), procedural sedation, perioperative analgesic regimen (preoperative, intraoperative and postoperative) and rescue analgesia.

  • Outcomes: analgesic outcome (duration of analgesia, the total number of analgesic administrations within 24 hours, the total dose of analgesic consumption within 24 hours), block-related outcomes (duration of the motor block) and adverse effects (emergence times, the incidence of hypotension, bradycardia, PONV, urinary retention, respiratory depression, emergence delirium, and neurological complications).

We chose the duration of analgesia (defined as ‘the time from caudal injection to the time at which the study-specific pain score was greater than a cut-off threshold’) as our primary outcome. If a cut-off was not specified, but the duration of analgesia was provided, we extracted such published outcome data for analysis. We designated all other outcomes as secondary. We extracted study-specific definitions of the primary outcome from each paper to assess homogeneity.

Risk of bias of individual studies

Two authors (NK and UJS) independently assessed the quality of included RCTs using the Cochrane Collaboration risk of bias tool (V.2, 2016) for RCTs.10 This tool assesses risk of bias across the following five domains: bias arising from the randomization process (including random sequence generation and allocation concealment); due to deviations from intended interventions; due to missing outcome data; due to measurement of the outcome and from selective reporting of outcomes. The authors assigned a score (low, some concern or high risk of bias) to each type of bias category, with the majority bias rating representing the overall bias rating. We consulted a third author (HS) to resolve disagreements. We included all studies regardless of the risk of bias status.

Evidence appraisal

We used the guidance from the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) working group to assess the certainty of the evidence for outcomes reported in this review.11 We assessed the certainty of the evidence for each outcome using the risk of bias, imprecision, inconsistency, indirectness and publication bias. When we assessed one of the above domains as a risk, we downgraded the evidence by two levels (very serious risk) or one level (serious risk). We presented these results in a summary of findings table for efficacy and safety outcomes.

Summary measures

We extracted continuous data as mean and SD and presented it as the ratio of means (ROM) with 95% CIs. We employed transformation methods described by Hozo et al12 and Wan et al13 when these measures were not readily available as means and SD. We analyzed the data from figures using Web Plot Digitizer14 and imputed missing SD when missing.15 We presented dichotomous data as a relative risk (RR) with 95% CI. In the case of multiarm trials involving multiple adjuvants, we extracted data from relevant arms.

Synthesis of results

Two authors (HS and UJS) performed the statistical analysis, which was checked by the third author (JM) for errors. We recognized a priori that inclusion of non-indexed articles (defined as published articles not indexed in any major database) is comprehensive but may lead to a biased estimate due to low quality. We compared such eligible articles against lists of predatory journals and publishers before including them in the final analysis. Moreover, we planned to pool the indexed and non-indexed articles only if there were no differences between these subgroups. We conducted a pairwise frequentist meta-analysis using the restricted maximal likelihood estimator; employed inverse variance method was used for weighting of studies with continuous outcomes and Mantel-Haenszel method for binary outcomes. Finally, we used Hartung-Knapp-Sidik-Jonkman modification16 in meta-regression analysis to avoid bias underestimation in smaller studies. We considered differences as statistically significant when p<0.05 (two-sided), and when 1 (null value) was not included in the 95% CI for outcomes. We used the I2 statistical method to identify statistical heterogeneity. To avoid overstating the results and inflating type I error, we used secondary outcome results for hypothesis generation. We defined a ‘minimal clinically important difference’ for the primary outcome as an effect size of 0.2, such that the 95% CI of the ratio of means excludes the interval of 0.8 to 1.2. We employed a similar test to judge other outcomes. We have described our detailed statistical methods in online supplemental file (section 4).

Dose-response relationship

To examine of there is a dose-response relationship for caudal dexmedetomidine, we conducted subgroup analysis of important outcomes for which such data were available. This included:

  1. Analgesic outcomes (duration of analgesia and the number of analgesic administrations).

  2. Adverse effects such as emergence time, duration of motor block, hypotension and PONV.

Publication bias

We assessed publication bias using funnel plots (if the number of studies was ≥10), as well as statistical tests such as Egger’s regression test.17 We generated a contour funnel plot to reduce the risk of false positives in funnel plot asymmetry. If we suspected publication bias, then we imputed the impact of missing studies using the ‘Copas selection model’.18

Exploration of heterogeneity

We hypothesized a priori that small differences in methodology, interventions, definitions and measurements of outcomes would result in high heterogeneity. We explored this using predefined subgroups analyses (risk of bias, LA type, dexmedetomidine dose, guidance technique) and meta-regression (LA dose and concentration; dexmedetomidine dose) methods. We explored the impact of outlier studies using Baujat plots,19 influence diagnostics and the leave-one-out test.20

Trial-sequential analysis

Finally, we performed a TSA to inform the risk of type I error in our estimates, as described by the Copenhagen Trial Unit.21 TSA does this by constructing threshold bounds (Z-boundaries) for efficacy, as well as futility, using an alpha-spending function (such as O’Brien-Fleming method). These bounds are used to test the statistical significance of the observed effect, as each RCT is added to the meta-analysis cumulatively. Thus, the overall alpha (rate of type I error) is maintained at 0.05. Also, the TSA estimates the required information size (RIS) that informs us of the sample size required in case the effect curve does not cross any bounds (efficacy or futility).

We performed TSA for primary outcome using a random-effects model employing the empirical versions of mean difference, variance estimator and heterogeneity. We tested other assumptions by varying the variance estimator (‘low risk’), mean difference and studies based on the risk of bias (‘low risk’ vs all studies). For other outcomes which were expected to have fewer trials or event rate, we chose to downgrade evidence for imprecision (using GRADE handbook guidelines). This approach is comparable to the TSA.22


Study selection

We search identified 827 records (after de-duplicates) and excluded 763 records on initial screening. Next, we assessed 64 full-text manuscripts for eligibility and excluded 31 studies. We compared another 14 non-indexed articles against lists of predatory journals, judged them as ‘suspect’ or ‘potentially suspect’, and excluded them. Thus, 19 RCTs met the inclusion criteria and included in this review by consensus.4 23–40 Figure 1 summarizes the PRISMA flow chart.9

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of study inclusion and exclusion.

Study characteristics

We included 19 RCTs with 1190 patients in this review, with all studies categorized as ‘indexed’.4 23–40 Selected RCTs were conducted from 2009 to 2018 and recruited pediatric patients of age 0–12 years and ASA class I–II, undergoing various infra-umbilical procedures. All studies employed a general anesthesia (GA), and most performed a landmark-guided caudal block (except ultrasound-guidance in one study25). Ten studies employed bupivacaine,23 28–31 33–35 38 40 six employed ropivacaine,4 24 26 27 32 37 while three used levobupivacaine.25 36 39 Fourteen studies used dexmedetomidine in a dose of 1 μg/kg4 23 25–28 30 33–35 37–40 and five used a dose of 2 μg/kg.24 29 31 32 36 The included studies employed different regimens and routes of premedication, intraoperative rescue analgesics, postoperative analgesics and pain assessment scales. We have summarized the study characteristics in table 1 and online supplemental file 1, section 5–8.

Table 1

Characteristics of studies meeting inclusion criteria

Risk of bias within studies

We categorized 8 RCTs as ‘low’ overall risk of bias,23 24 27 29 35 38–40 and 11 as ‘some risk’.4 25 26 28 30–34 36 37 We found an inadequate description of allocation concealment in 11 RCTs23–25 27–29 32 35 38–40 and potential measurement bias (blinded assessor) in 7 RCTs.4 26 30 33 34 36 37 The reviewers’ consensus assessment of the risk of bias is detailed in figure 2 and online supplemental file, section 9.

Figure 2

Risk of bias summary: judgments about each risk of bias item for each included study, across five domains (D1–D5). Green and yellow circles indicate ‘low’ and ‘some’ risk of bias, respectively.

Primary outcome: duration of analgesia

We extracted the duration of analgesia from 990 patients in 15 studies.4 23–25 27–30 32 34–37 39 40 Compared with control (mean duration of analgesia 346 min), the addition of caudal dexmedetomidine prolonged the duration of analgesia (ROM 2.14 (95% CI 1.83 to 2.49; p<0.001, I2=100%); figure 3). The 95% predictive interval for the same outcome (ROM 1.14–4.17) suggests that caudal dexmedetomidine will likely prolong analgesia in 95% of future studies. We explored associated high heterogeneity using prespecified subgroup analysis and meta-regression. We found no subgroup differences based on risk of bias and the type of LA; however, we found significant subgroup differences for dexmedetomidine dose (1 μg/kg—ROM 1.96 (95% CI 1.66 to 2.32)4 23 25 27 28 30 34 35 37 39 40 vs 2 μg/kg 2.69 (95% CI 2.09 to 3.36)24 29 32 36). Univariable meta-regression did not reveal differences based on LA dose (mL/kg) or LA concentration (%). Multivariable meta-regression modeling using LA dose, LA concentration and dexmedetomidine dose showed the greatest impact on reducing unaccounted heterogeneity (R2=0.52, p=0.05; permutation test p=0.04). We found consistent results despite the exclusion of two outliers (Baujat plot).24 28 Influence diagnostic plots or leave-one-out analysis did not reveal any outlier studies. We found no evidence of publication bias on visual inspection of funnel plots and Egger’s test (p=0.64), or selection bias using Copas selection models. We have summarized these results in online supplemental file (section 10).

Figure 3

Forest plot for the duration of analgesia. The vertical dotted line represents that mean of the pooled effect for the primary outcome. Dexmed, dexmedetomidine; ROM, ratio of means. Red bar indicates the 95% prediction interval.

We performed TSA for the primary outcome using five different models. The base model (model 1) assumed empirical mean differences and variance estimators. The result demonstrated that the estimates of primary outcome crossed the Z-boundary favoring intervention group (figure 4). TSA using other assumptions (model 2–5), showed that the estimate crossed the Z-boundary in each instance, suggesting ‘firm evidence’ of effect favoring dexmedetomidine. Finally, we judged the certainty of evidence for the primary outcome as ‘moderate’ (table 2 and online supplemental file, section 16). We evaluated the dose-response relationship of 1 vs 2 μg/kg caudal dexmedetomidine on several outcomes via subgroup analyses. Our results demonstrate that compared with control, a higher dose (2 μg/kg—ROM 2.66 (95% CI 2.35 to 3.01); 4 RCTs24 29 32 36) significantly improves the duration of analgesia than a lower dose (1 μg/kg—ROM 1.96 (95% CI 1.71 to 2.24); 11 RCTs4 23 25 27 28 30 34 35 37 39 40). This is presented in table 3 and online supplemental file 1 (section 15).

Figure 4

Trial sequential analysis of duration of analgesia (primary outcome). The depicted figure refers to model 1 and assumes empirical mean differences and variance estimators. The x-axis depicts accumulating number of studies (and thus participants) and is called the ‘information size’. The z-values are shown on the y‐axis and represent the statistical summary of the sequentially accrued data from studies. The horizontal lines ‘conventional threshold’ represent threshold for statistical significance at a constant Z-value of 1.96 (local anesthetic (LA) alone is better) and −1.96 (LA and dexmedetomidine (Dexmed) combination is better) and corresponds to a p value of 0.05 in conventional analysis. The curved lines at the top and bottom left‐hand corners (trial sequential boundaries) represent the sequential-analysis thresholds for statistical significance, and a diamond representing incremental addition of data from a trial. The diagonal lines inside the conventional threshold lines represent the futility boundaries. The cumulative Z-curve at the bottom represents the accumulating amount of information as data from trials are added, with each square denoting an individual trial. In this case, the z-values from accrued trial data (ie, the Z-curve) crosses the ‘efficacy boundary for analgesic benefit’ and approaches the required information size.

Table 2

Summary of findings—caudal dexmedetomidine+LA compared with LA alone for postoperative analgesia and adverse effects in pediatric infra-umbilical surgery

Table 3

Subgroup analysis of important outcomes to assess dose response of 1 vs 2 μg/kg caudal dexmedetomidine (Dexmed)

Secondary (continuous) outcomes

Number of analgesic administrations

We extracted data on the ‘number of analgesic administrations’ from five studies (n=341).23 30 32 34 35 Subgroup analysis using risk of bias, dexmedetomidine dose or the type of LA found no differences. Compared with control group (mean requirement 2.28 doses), caudal dexmedetomidine reduced the number of analgesic administrations required (ROM 0.34, 95% CI 0.16 to 0.76; p=0.008, I2=97%). The visual inspection of the funnel plot for this outcome did not suggest publication bias, but we did not do Egger’s test due to a few studies. Our dose-response analysis showed that caudal dexmedetomidine reduced the number of analgesic administrations, regardless of the dose used. While this was estimated to be as a ROM of 0.33, (95% CI 0.12 to 0.91, four RCTs) for 1 μg/kg, it was a ROM of 0.41 (95% CI 0.36 to 0.47, one RCT) for 2 μg/kg (table 3 and online supplemental file, section 15).

Total analgesic (acetaminophen) dose

We extracted data on total analgesic (acetaminophen) dose from two studies (n=102).25 35 Compared with control group (mean requirement 346 mg acetaminophen), caudal dexmedetomidine reduced the total acetaminophen dose (mg) required (ROM 0.50, 95% CI 0.39 to 0.64; p≤0.001, I2=98%). Few studies precluded any subgroup analysis or meta-regression, or assessment of publication bias.

Emergence time

We analyzed emergence time from nine studies (n=520).23 24 26 27 29 32 34 38 39 Subgroup analysis using risk of bias, dexmedetomidine dose or the type of LA revealed no differences. Expressed as ROM (95% CI), caudal dexmedetomidine prolonged the emergence time by an estimate of 1.41 (0.99 to 2.01; p=0.05, I2=96%) compared with control group (mean 10 min). We did not detect publication bias (Egger’s test p=0.69).

Duration of motor block

We assessed the duration of the motor block from three studies (n=159).26 32 34 Caudal dexmedetomidine did not prolong motor block (ROM 1.11, 95% CI 1.02 to 1.22; p=0.02, I2=91%) compared with control group (mean block duration 118 min). Fewer studies precluded any subgroup analysis or meta-regression, or assessment of publication bias.

We have summarized the results from these continuous secondary outcomes in figure 5, table 2 and online supplemental file (section 12).

Figure 5

Forest plot of the impact of caudal Dexmed on other continuous outcomes (duration of motor block, emergence time, number of analgesic administrations and total acetaminophen dose). Each diamond represents a separate outcome. Dexmed, dexmedetomidine; ROM, ratio of means.

Other secondary outcomes

Many studies reported ‘no events’ for many outcomes. We included such studies our analysis and have provided a detailed description in the online supplemental file 1. Expressed as RR (95% CI), caudal dexmedetomidine did not increase the risk of hypotension (1.31 (0.31 to 5.82), p=0.70, I2=0%; 15 RCTs (n=890)4 23–26 28 29 31 32 34 36–40; ‘low’ certainty) compared with control group (7 episodes per 1000). Caudal dexmedetomidine significantly increased the risk of bradycardia (4.83 (0.54 to 41.22; p=0.15, I2=0%); 15 RCTs (n=890)4 23–26 28 29 31 32 34 36–40; ‘very low’ certainty) compared with control group (0 episodes per 1000). We found caudal dexmedetomidine did not increase the risk of PONV (1.01 (0.63 to 1.64; p=0.95, I2=0%); 16 RCTs (n=1006)4 23–32 34 36 38–40; ‘low’ certainty) compared with control group (56 episodes per 1000). The was no evidence of publication bias (Egger’s test p=0.43). Caudal dexmedetomidine did not increase the risk of urinary retention (0.52 (0.09 to 2.91; p=0.45, I2=0%); 14 RCTs (n=829)4 23–26 29 31 32 34–36 38–40; ‘moderate’ certainty) compared with control group (7 episodes per 1000). Caudal dexmedetomidine did not increase the risk of respiratory depression (5.18 (0.26 to 103.17; p=0.28, I2=not applicable); 15 RCTs (n=951)4 23–25 28–37 40; ‘low’ certainty) compared with control group (0 episodes per 1000). Caudal dexmedetomidine reduced the risk of emergence delirium (0.25 (0.11 to 0.54; p<0.01, I2=0%); 5 RCTs (n=295)26 27 34 39 40; ‘moderate’ certainty) compared with control group (216 episodes per 1000). None of the included studies evaluated neurological complications comprehensively. The results from these secondary outcomes are summarized in figure 6, table 2 and online supplemental file (section 13). Finally, we found no dose-response for other outcomes (emergence times, duration of motor block, hypotension and PONV—table 3 and online supplemental file, section 15).

Figure 6

Forest plot of the impact of caudal Dexmedetomidine (Dexmed) on adverse effects (bradycardia, emergence delirium, hypotension, PONV, respiratory depression and urinary retention). Each diamond represents a separate outcome. PONV, postoperative nausea and vomiting; RR, risk ratio.


Principal findings

Our results indicate that the addition of dexmedetomidine to caudal LA prolongs the duration of analgesia by 1.8–2.5 times (‘moderate’ evidence). TSA suggests that we have the RIS to determine this outcome reliably. We also observed a >50% reduction in the number of analgesic administrations and total acetaminophen dose (‘low’ evidence for both). Caudal dexmedetomidine may marginally prolong emergence time (‘moderate’ evidence) and duration of motor block duration (‘low’ evidence). While our results indicate that the risk for hypotension, bradycardia, PONV, urinary retention or respiratory depression is likely unchanged, we believe more data are needed to confidently assert this, particularly given the ‘very low’ to ‘low’ ratings of evidence in some of these outcomes. Finally, caudal dexmedetomidine reduced the risk of emergence delirium by 46%–89% (‘moderate’ evidence). We found a dose-response relationship for the duration of analgesia. In absolute terms, this implies a mean duration of analgesia of 15.9 hours with a higher dexmedetomidine dose (2 μg/kg) compared with 12.1 hours using a lower dose (1 μg/kg) and 5.9 hours with control. While we explored dose-response relationships for other outcomes, these were underpowered (especially for adverse events). Due to the lack of data, we are unable to obtain conclusive results on neurological complications.

Caudal dexmedetomidine—a dose-response relationship

To date, three RCTs have compared different doses of caudal dexmedetomidine. Bharti et al26 studied the effects of 0.5, 1.0 and 1.5 μg/kg dexmedetomidine in 78 patients undergoing infra-umbilical surgery, with control. While they found a lack of dose-response effect for analgesic outcomes, emergence agitation or hemodynamic effects, they reported a higher risk for prolonged sedation and urinary retention. Al-Zaben et al26 evaluated 1 vs 2 μg/kg dexmedetomidine, with control, in 91 patients. They report a higher risk of adverse effects, such as postoperative sedation, hypotension, bradycardia and urinary retention, with a higher dose. Our post hoc testing of their results shows a dose-response effect in their ‘time to first analgesia’. Similarly, Karuppiah et al40 reported a dose-response effect for the need for rescue analgesia and a higher incidence of hypotension and bradycardia with 2 μg/kg. Our post hoc testing of their results shows borderline evidence of dose-response effect in analgesic duration (online supplemental file, section 17).

Thus, direct external evidence points to a dose-response effect with analgesic outcomes, although with a higher risk of complications. As opposed to this, our subgroup analysis (stratified by dose) suggests a dose-related improvement in only the duration of analgesia, without an increased risk of adverse effects. This observation is likely due to fewer studies and few overall events. We advocate readers to interpret this lack of an increased risk for adverse effects with escalating doses in our review with caution in light of the low power. Until such evidence is available, it may be best to reserve the higher dose for select patients based on clinical needs (such as a more extensive surgery).

Caudal versus intravenous dexmedetomidine

Perineurally administered adjuvants are redistributed to the central compartment through systemic uptake41 and may produce systemically mediated analgesia or adverse effects.5 The similarity of the effect size of bradycardia (RR ~4) and emergence delirium (RR ~0.4) observed with intravenous dexmedetomidine42 to our results (caudal route), suggests a systemic action in part. It is unknown, which of the two actions, local or systemic, plays a predominant role in peripheral and neuraxial blocks using dexmedetomidine. Animal43 and human volunteer44 studies indicate that dexmedetomidine exerts its antinociceptive effects locally, through perineural mechanisms. Previous meta-analyses of intrathecal dexmedetomidine45 assert that it prolongs the time to the first analgesic request by 2.23 times compared with control, while intravenous dexmedetomidine46 prolongs it by 1.6 times. Both routes exhibit a similar risk of bradycardia. These results indicate a substantial perineural mechanism for dexmedetomidine, which may be an advantage over the intravenous route. In a study of 75 patients, Al-Zaben et al47 evaluated both these routes for time to rescue analgesia and found a ROM of 2.18 and 1.39 for caudal and intravenous dexmedetomidine, compared with control. Contrary to this, Yao et al39 assessed these routes in a three-arm study with 90 patients and found no difference. Therefore, until further robust evidence is available, intravenous dexmedetomidine cannot not be recommended as a substitute for the caudal route.

Caudal dexmedetomidine versus other adjuvants

Caudal adjuvants such as clonidine (20 RCTs, 4 hours),48 ketamine (13 RCTs, 5.6 hours),49 dexamethasone (9 RCTs, 5.4 hours)50 and magnesium (4 RCTs, 3.2 hours)51 also prolong the duration of analgesia over control. None of these reviews has evaluated a dose-response effect. Moreover, these reviews are currently outdated and have not used potential indirect evidence (network meta-analysis) for estimating the best adjuvant. Indeed, preliminary evidence from our network meta-analysis of caudal adjuvants suggests that dexmedetomidine may represent the best overall adjuvant for analgesia.52 53 Despite this, the choice of a given adjuvant should be informed by the trade-off between their efficacy, adverse effects, safety, local availability and cost.3

Neurological safety of caudal dexmedetomidine

As such, the use of all adjuvants for neuraxial blocks (except epinephrine) remains an off-label indication. None of the included studies in our review evaluated the neurological safety of caudal dexmedetomidine. Such effects are best ascertained by examination or a delayed (2 weeks) follow-up questionnaire to evaluate deficits. Unfortunately, a pediatric population hinders a reliable neurological assessment. Reassuringly, animal54–56 and human57 outcome data indicate the safety of dexmedetomidine. Underlying mechanisms point to the local anti-inflammatory action of dexmedetomidine as a protective factor.58 59 However, drawing firm conclusions will require robust data on neurological safety, and current evidence does not provide this. It is unlikely that a large-sized RCT would be carried out to assess this, and in its absence, we will have to rely on animal data or observational evidence. We could not identify any case report of a neurological insult following the use of caudal dexmedetomidine, which could be attributed to the drug.

Limitations and strengths

Our review has several limitations. Available RCTs involved diverse demographics and methods, including variations in age, gender and the type of infra-inguinal surgery (including abdominal wall and genitourinary surgery). We observed variations in the type, dose and concentration of LA and dexmedetomidine; and in analgesia. Finally, we noted variations in the definitions and assessment of outcomes that may contribute to heterogeneity. We mitigated this by employing a priori subgroups and meta-regression to explore heterogeneity. Since we were unable to resolve this sufficiently, we downgraded the evidence where appropriate. Second, the included studies were small, and we cannot exclude a potential for publication and selection bias, and available statistical tests for these biases themselves remain underpowered. Third, we were unable to ascertain opioid consumption (most studies used non-opioids) or neurological complications (mostly unreported). Given that peri-neural dexmedetomidine use is currently considered off-label, a lack of data on neurological outcomes is concerning. Fourth, we analyzed TSA data for only the primary outcome. For other outcomes (eg, adverse effects), we chose to assess imprecision and downgrade evidence appropriately, as this method is comparable to TSA.22 We did so due to the lack of reliable data on baseline consumption of analgesics, or risk for such outcomes. Finally, the lack of individual-level patient data precludes the analysis of pain and analgesic consumption as a composite outcome, as recommended.60

Despite these limitations, our manuscript has several strengths. We employed a pre-specified and registered protocol, and conducted a comprehensive search of the literature. We employed ROM rather than mean differences, to adjust for the variation in the type and doses of LA used and improve the translational application of findings. Unlike previous work, we deployed a restricted maximal likelihood estimator with Hartung-Knapp-Sidik-Jonkman modification, rather than Der-Simonian Laird model, to avoid bias underestimation in smaller studies. We leveraged TSA methodology to assess the adequacy of information size for our primary outcome, besides appraising the evidence quality using GRADE methods. Finally, our review is the first to quantify the clinical benefits of a higher dexmedetomidine dose (2 μg/kg). Thus, it represents a significant advance compared with previous reviews that have included a small number of studies, unable to establish a dose-response effect or appraise the adequacy and strength of evidence.6–8


Our meta-analysis provides moderate-quality evidence that the addition of dexmedetomidine to LA in caudal block significantly improves the duration of analgesia and reduces the analgesic requirements. Furthermore, there is moderate-quality evidence that it reduces the risk of emergence delirium, and slightly prolongs emergence. While we estimated a similar risk-profile with caudal dexmedetomidine compared with LA alone, the existing data on adverse effects are generally of low quality and further data are needed for a definitive conclusion. Compared with a lower dose (1 μg/kg), a higher dose of dexmedetomidine (2 μg/kg) increases the duration of analgesia, but higher-quality data on its safety are lacking. Until such evidence is available, it may be best to reserve the higher dose for select patients based on clinical needs (such as a more extensive surgery). Future studies should focus on assessing neurological safety of caudal dexmedetomidine and its comparison with intravenous dexmedetomidine.


The authors would like to thank Rachel Sandieson (MLIS), Western University, for her invaluable help with constructing and conducting the literature search for this review.



  • Contributors UJS, JM and HS conceptualized the study and wrote the protocol; UJS and NK performed the initial study screening; DK and NK performed the full-text review and data extraction; UJS, NK and HS performed the risk of bias assessment; HS and JM did the statistical analysis. All authors contributed to the final version of the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available on reasonable request. Extracted data are available on reasonable request. This can also be provided as dataset files and R codes (or R Markdown files) for scrutiny or re-analysis for future studies. Detailed methods and R codes are already provided in the supplementary file.