Pain relief for neonatal circumcision

Cover sheet

Title

Reviewers

Dates

Contact reviewer

Contribution of reviewers

Internal sources of support

External sources of support

What's new

The review has been updated to correct the numbers reported for total patients randomized for the included trials. Results of meta-analyses were not effected except in one case, comparison 03.02 where the effect size changed from [WMD = 15.8%, 95% CI -20.8 to -6.8] to [WMD = 15.22%; 95% CI 21.08 to 9.36].

Abstract was changed to note that data from three (not four) trials of DPNB versus EMLA including 133 patients were subjected to meta-analysis.

No new trials were identified as a result of the most recent search.

Dates

Text of review

Synopsis

Abstract

Background

Circumcision is a painful procedure that many newborn males undergo in the first few days after birth. Interventions are available to reduce pain at circumcision; however, many newborns are circumcised without pain management.

Objectives

The objective of this review was to assess the effectiveness and safety of interventions for reducing pain at neonatal circumcision.

Search strategy

We searched Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 2, 2004), MEDLINE (1966 - April 2004), EMBASE (1988 - 2004 week 19), CINAHL (1982 - May week 1 2004), Dissertation Abstracts (1986 - May 2004), Proceedings of the World Congress on Pain (1993 - 1999), and reference lists of articles. Language restrictions were not imposed.

Selection criteria

Randomised controlled trials comparing pain interventions with placebo or no treatment or comparing two active pain interventions in male term or preterm infants undergoing circumcision.

Data collection & analysis

Two independent reviewers assessed trial quality and extracted data. Ten authors were contacted for additional information. Adverse effects information was obtained from the trial reports. For meta-analysis, data on a continuous scale were reported as weighted mean difference (WMD) or, when the units were not compatible, as standardized mean difference.

Main results

Thirty-five trials involving 1,997 newborns were included. Thirty-three trials enrolled healthy, full term neonates, and two enrolled infants born preterm.

Fourteen trials involving 592 newborns compared dorsal penile nerve block (DPNB) with placebo or no treatment. Compared to placebo/no treatment, DPNB demonstrated significantly lower heart rate [WMD -35 bpm, 95% CI -41 to -30], decreased time crying [WMD -54 %, 95% CI -64 to -44], and increased oxygen saturation [WMD 3.7 %, 95% CI 2.7 to 3.7]. Six trials involving 200 newborns compared eutectic mixture of analgesics (EMLA) with placebo. EMLA demonstrated significantly lower facial action scores [WMD -46.5, 95% CI -80.4 to -12.6], decreased time crying [WMD - 15.2 %, 95% CI -21 to -9.3] and lower heart rate [WMD -15 bpm, 95% CI -19 to -10]. DPNB, compared with EMLA in three trials involving 139 newborns (133 of whom were included in the analysis), demonstrated significantly lower heart rate [WMD -17 bpm, 95% CI -23 to -11] and pain scores. When compared with sucrose in two trials involving 127 newborns, DPNB demonstrated less time crying [MD -166 s, 95% CI -211 to -121], and lower heart rate [WMD -27 bpm, 95% CI -33 to -20]. Results obtained for trials comparing oral sucrose and oral analgesics to placebo, and trials of environmental modification were either inconsistent or were not significantly different.

Adverse effects included gagging, choking, and emesis in placebo/untreated groups. Minor bleeding, swelling and hematoma were reported with DPNB. Erythema and mild skin pallor were observed with the use of EMLA. Methaemoglobin levels were evaluated in two trials of EMLA, and results were within normal limits.

Reviewers' conclusions

DPNB was the most frequently studied intervention and was the most effective for circumcision pain. Compared to placebo, EMLA was also effective, but was not as effective as DPNB. Both interventions appear to be safe for use in newborns. None of the studied interventions completely eliminated the pain response to circumcision.

Background

Neither the American Academy of Pediatrics nor the Canadian Paediatric Society recommends routine or elective circumcision of the male newborn. Nevertheless, elective circumcision of male newborns is commonly performed in the first few days after birth. Approximately 1.2 million newborn males are circumcised in the United States annually at a cost of 150 to 270 million dollars (AAP 1999). Precise Canadian data are not available because the procedure has been delisted in many provinces, but it is estimated that 48% of male neonates born in Canada are circumcised (CPS 1996). The practice of male neonatal circumcision is not limited to North America; it is performed worldwide for religious and cultural reasons.

As an invasive, painful procedure, unanaesthetized circumcision elicits systemic stress responses in the vulnerable newborn which negatively affect major body systems. Documented physiological and behavioral responses include increased output of adrenal corticoids (Gunnar 1981; Talbert 1976), increased heart rate and respiratory rate, decreased arterial oxygen (Rawlings 1980), skin flushing, vomiting and cyanosis (Poma 1980), changes in sleep/wake state, increased crying (Anders 1974; Gunnar 1981), and diminished responsiveness to parents (Dixon 1984). Unanaesthetized circumcision has also been linked with complications such as apnea and choking (Lander 1997), gastric rupture (Connelly 1992), and recurrence of pneumothorax (Auerbach 1978). Infants circumcised without anaesthesia exhibit stronger pain responses to routine immunizations during the first six months of life than infants who were not circumcised (Taddio 1997b), suggesting that circumcision pain may exert long term effects on infant behavior.

INTERVENTIONS FOR CIRCUMCISION PAIN

Numerous interventions to prevent or reduce circumcision pain have been examined. These include penile blocks, topical anaesthetics, oral analgesia and sucrose administration, non-nutritive sucking, music and other environmental interventions.

The technique of dorsal penile nerve block (DPNB) for newborn circumcision was first described in 1978 (Kirya 1978), and it has since been extensively evaluated. More recently, subpubic (Dalens 1989) and penile ring block techniques (Hardwick Smith 1998; Lander 1997) have been examined. Adverse effects of penile blocks appear to be limited to bruising and slight bleeding at the injection site (Snellman 1995). Of note, the rapidity of onset of the anaesthetic used for penile blocks (generally 1% lidocaine without epinephrine) is intermediate and a "wait time" of 5 minutes is recommended to achieve anaesthesia (Taddio 2001). Wait time is a concern for clinicians because it increases the total time required for the circumcision surgery; however, inadequate "wait time" influences anaesthetic effect (Kharasch 2003).

Several types of topical anaesthetics have been used for neonatal circumcision, including eutectic mixture of local anaesthetics (EMLA) and 4 to 30% lidocaine creams. EMLA is a water-based cream that contains 2.5% lidocaine and 2.5% prilocaine. Compared with placebo, EMLA attenuates the pain responses of increased heart rate, facial activity and crying, and decreased oxygen saturation (Lander 1997; Taddio 1997). A meta-analysis of three studies examining this intervention indicated that the use of EMLA results in a significantly lower increase in heart rate (from baseline) and less crying during the various phases of circumcision surgery compared to placebo. In two of the included studies, lower facial action scores suggested less pain in the EMLA treated groups compared to placebo (Taddio 2002).

Potential difficulties with drug administration and the presurgical wait time may limit the feasibility of topical anaesthesia as a pain intervention for circumcision in many settings (Lander 1997). Considerable technical skill is required to apply the drug, and to secure the occlusive dressing needed to keep it in place. For adequate absorption, EMLA must be applied for at least 60 minutes prior to surgery (Taddio 1998), and must be reapplied if the infant voids during the wait time.

Methaemoglobinaemia (MetHb), caused by oxidation of haemoglobin by the metabolites of prilocaine, is a serious but relatively rare risk associated with EMLA use in infants less than 12 months of age. A recent systematic review of the use of EMLA for acute pain in infants demonstrated that the risk of significant MetHb is low with single dose applications of 0.5 to 2g applied for 10 - 180 minutes for full term neonates, and 0.5 to 1.25g applied for 3 to 180 minutes for preterm neonates (Taddio 1998). Local skin reactions, such as blanching, erythema, and edema of the skin have been reported with the use of EMLA , but these are usually transient and not considered serious.

Sucrose or other sugar solutions alone or in combination with non-nutritive sucking have recently been recommended as interventions for procedural pain management (Mitchell 2000). Oral sucrose is thought to activate central endogenous pathways, and may stimulate release of endorphins from the hypothalamus. Non nutritive sucking (NNS) is also thought to have an analgesic-like effect through stimulation of orotactile and mechanoreceptor mechanisms (Gibbons 2001; Mitchell 2003). The sensations created by NNS may deflect attention away from pain and facilitate self regulation because the infant is in control of the sucking. Sucrose and NNS appear to operate synergistically when offered in combination, and may provide more effective pain relief (Carbajal 1999; Gibbons 2001; Gibbons 2002). The analgesic effect of sucrose is activated within two minutes, and lasts for three to five minutes (Haouari 1999; Mitchell 2003). Although sucrose in a wide variety of dosages (concentrations from 12 to 24%, and volumes from 0.05 to 2.0 ml) has generally been found to decrease acute, procedural pain responses in neonates (Mitchell 2000; Stevens 1997), the optimal dose has not yet been identified. Meta-analyses results indicate that a 0.24g dose is effective to reduce pain responses in term infants, and higher doses do not appear to increase effectiveness (Stevens 1997). In comparison, relatively small doses (0.01 to 0.02g) appear to be effective for preterm infants (Johnston 1997). Interest in sucrose or other sugar solutions as a single or adjunctive intervention for circumcision pain is reflected in the design of recent research (e.g. Kass 2001; Kaufman 2002).

Acetaminophen is the most frequently prescribed non-opioid oral analgesic used to treat mild to moderate pain in pediatric populations (Berde 2002; McGrath 1990). Acetaminophen is safe and effective for neonates and can be administered orally or rectally (Stevens 1999). Acetaminophen has been used as an intervention for circumcision pain (Howard 1994).

A variety of non-pharmacological interventions have been evaluated for treatment of acute procedural pain in neonates. In theory, these interventions provide nonpainful stimuli that compete with painful stimuli for the neonate's attention, and thus may blunt the perception of pain (Bellieni 2002). Interventions such as rocking, massage, facilitated tucking, and cuddling reduce pain responses during invasive procedures (Campos 1994; Corff 1995; Gray 2000). Music and other sounds (intrauterine, heartbeat) provide an auditory stimulus which may modulate pain perception and these have been evaluated as interventions for circumcision pain (Marchette 1989; Marchette 1991).

NEONATAL PAIN RESPONSES

It is difficult to evaluate the effectiveness of interventions for circumcision pain because newborns are non-verbal and display stereotypic responses to a variety of painful and non-painful stimuli. To maximize the validity of pain assessment in newborn populations three classes of pain indicators or outcomes, biochemical, physiological, and behavioural, are generally employed for research. Salivary and serum cortisol, the most frequently measured biochemical indicators, serve as markers of the stress response to pain because hormones of the hypothalamic-pituitary-adrenal axis are assayed. Physiological indicators include heart rate, respiratory rate, blood pressure, transcutaneous oxygen saturation (Tc pO2), transcutaneous carbon dioxide (Tc pCO2), oxygen saturation (SaO2), palmar sweat, intracranial pressure (ICP) and vagal tone. In newborn populations, heart rate is the most frequently studied physiological indicator (Sweet 1998). Behavioral indicators include facial expression, cry, gross motor movement, and changes in behavioral state. Facial expression (Grunau 1987) is the most comprehensively studied behavioral indicator for neonatal pain.

Multidimensional measurement tools that employ more than one parameter usually contain physiological and behavioral indicators, and occasionally add contextual information to obtain an overall pain score. The Neonatal Infant Pain Scale (NIPS) (Lawrence 1993) and the Premature Infant Pain Profile (PIPP) (Stevens 1996) are multidimensional tools frequently utilized as outcome measures for investigation of acute procedural pain in term and preterm neonates. Although a number of pain measures are available for use with neonatal populations, no single measure has proven to be the best for all situations. Accordingly, all outcomes evaluated in the included studies as measures of neonatal pain were included in this review.

SUMMARY

The substantial amount of research conducted to date suggests a willingness to address the problem of circumcision pain. However, the majority of neonates are circumcised without interventions for pain (Myron 1991; Ryan 1994; Snellman 1995; Wellington 1993). This situation persists despite growing awareness that newborns may perceive pain more intensely than older children or adults (Anand 2001; Fitzgerald 1993) and can be significantly compromised by it.

It has been suggested that training to manage circumcision pain is inadequate to promote consistent use of available interventions (Howard 1998). Recent surveys indicate that significant numbers of obstetricians (75%), family practitioners (44%), and pediatricians (29%) do not use analgesia/anaesthesia for circumcision because of concerns about adverse drug effects or because they believe that the procedure does not require pain management (Maxwell 1999; Stang 1991; Stang 1998).

Although a wide variety of interventions for circumcision pain have been examined, the individual and relative effectiveness of each has not been systematically assessed. Thus, the apparent reluctance of practitioners to adopt the regular use of pain interventions for circumcision may reflect beliefs that the findings of research conducted to date are collectively un-interpretable. At the same time, negative perceptions of the technical and practical difficulties associated with pain interventions may diminish clinician motivation to implement their regular use.

A systematic review of the research in this area was needed to summarize and identify implications arising from the existing evidence, and to provide an informed basis for practice and to identify gaps in knowledge which require further investigation. This review adds to knowledge gained from a previous systematic review which examined the efficacy of a single intervention for circumcision pain (Taddio 2002) by evaluating the efficacy and safety of all interventions for reducing pain at neonatal circumcision.

Objectives

To determine the safety and effectiveness of interventions to relieve pain associated with neonatal circumcision. Subgroup analyses were prespecified according to wait time (after anaesthetic administration and prior to start of surgery) for penile blocks, and for dose delivered for sucrose interventions.

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs). Studies reported only as abstracts were included if relevant.

Types of participants

Male term or preterm neonates undergoing circumcision during the neonatal period (with postnatal age maximum of 28 days after reaching 40 weeks corrected gestational age).

Types of interventions

Any intervention intended to relieve pain during the circumcision procedure, for example, penile blocks, topical anaesthetics, oral sucrose administration, oral analgesics, surgical devices or techniques, or environmental manipulation such as music therapy or special restraints. This review included trials of interventions for circumcision pain in which any intervention was compared with placebo, no treatment, or with another active intervention.

Types of outcome measures

The primary outcome was pain as assessed by:

1. Physiological variables, such as heart rate (HR), respiratory rate (RR), oxygen saturation, or blood pressure (whether reported as change in, mean or absolute values)
2. Biochemical variables, such as salivary or serum cortisol levels (whether reported as pre- and post- measures or as change from baseline values)
3. Cry variables, for example, latency and duration of first cry, total cry duration, and/or percentage of time crying during the circumcision procedure
4. Validated pain measures, for example:
- Neonatal Infant Pain Score (Lawrence 1993);
- Neonatal Facial Action Coding System (Grunau 1987);
- Premature Infant Pain Profile (Stevens 1996);
- Other pain measures.

Secondary outcomes:

Complications of pain interventions were assessed as secondary outcomes. The outcomes included but were not limited to:
1) occurrence/incidence of methaemoglobinaemia (topical anaesthesia)
2) blanching and local skin irritations (topical anaesthesia)
3) bleeding, bruising and hematoma formation (penile blocks)
4) behavioral responses such a choking, spitting up, etc. during circumcision (all interventions)

Difficulties encountered in implementation of pain interventions, as reported by researchers, were noted.

Search strategy for identification of studies

Standard methods as per the guidelines of the Cochrane Neonatal Review Group (CNRG) were utilized.
1. CIRCUMCISION/exp
2. circumcision surgery.mp
3. newborn circumcision.mp
4. 1 OR 2 OR 3
5. local anaesthes*
6. penile block.mp/exp
7. dorsal penile nerve block.mp/exp
8. ring block.mp/exp
9. 5 OR 6 OR 7 OR 8
10. eutectic mixture of local anaesthetics.mp/exp
11. EMLA.mp/exp
12. LIDOCAINE.mp/exp
13. 10 OR 11 OR 12
14. acetaminophen.mp/ OR paracetamol.mp/exp
15. sucrose.mp
16. pacifiers.mp
17. music therapy.mp
18. Gomco clamp.mp
19. Mogen clamp.mp
20. 9 OR 13 OR 14 OR 15 OR 16 OR 17 OR 18 OR 19
21. 4 AND 20
22. HUMAN
23. MALE
24. 22 and 23
25. infant, newborn
26. neonat*
27. 25 OR 26
28. 24 AND 27
29. 21 AND 28
30. clinical trial
31. 29 AND 30

Databases searched included: Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Issue 2, 2004; MEDLINE 1966 - April 2004; EMBASE 1988 - 2004 week 19; CINAHL 1982 - May week 1 2004; PubMed 1966 - May 2004; Web of Science 1975 - May 2004; Dissertation Abstracts 1986 - May 2004. Keywords and (MeSH) terms included infant/newborn, male, circumcision, penile blocks, sucrose, lidocaine, EMLA, acetaminophen. Abstracts of the World Congress on Pain were searched for the years 1993 - 1999 inclusive. Reference lists of all articles were screened to identify any additional studies. Language restrictions were not imposed.

Methods of the review

Study Selection

The titles and abstracts of all reports identified through the electronic and other searches were scanned independently by two reviewers and full study reports were obtained for those that appeared to meet the inclusion criteria. Study reports were then evaluated independently by two reviewers for possible inclusion in the review. Disagreements were resolved by consensus. Studies rejected at this stage were included in the Table of Excluded Studies.

Quality Assessment

Assessment of the quality of all included studies was undertaken independently by two reviewers as a component of the data extraction process. Standard methods of the CNRG were used to assess: 1) the randomisation procedure, 2) concealment of allocation/blinding of randomisation, 3) blinding of intervention, 4) subject attrition and follow-up, and 5) blinding of outcome measurement. As per the CNRG guidelines, an overall quality score was not assigned. Reviewers were not blind to trial authors or institutions during the study selection or quality assessment processes.

Data Extraction

Data were extracted from included studies by two independent reviewers using a data extraction form designed specifically for this review. The data extraction form was developed in a draft format and piloted on several studies and modified as required before use. The reviewers abstracted data independently, compared results and resolved differences.

Sixteen trials included in this review either did not report outcome data, or did not report data in a format that could be analysed in this review (Arnett 1990; Benini 1993; Blass 1991 A; Holliday 1999; Holve 1983, Joyce 2001; Kass 2001; Marchette 1989; Marchette 1991; Maxwell 1987; Mohan 1998; Mudge 1989; Spencer 1992; Williamson 1997; Zahorodny 1998; Zahorodny 1999). Additional information was sought from ten authors and means and standard deviations were subsequently obtained for three trials (Benini 1993; Joyce 2001; Kass 2001). Where means and standard deviations were not available, data were imputed or derived from graphs contained in the reports (Arnett 1990; Benini 1993; Blass 1991 A; Holliday 1999; Maxwell 1987; Mohan 1998; Mudge 1989; Taddio 1997). Missing standard deviations were either calculated from other summary statistics or imputed using singular or mean standard deviations from similar trials.

Several authors reported total sample size only and information about the number of subjects per group was obtained from the authors (Benini 1993; Joyce 2001). When additional information about sample size could not be obtained from the authors, we assumed equal distribution to study groups in our data analyses (Blass 1991 A; Zahorodny 1998; Zahorodny 1999).

Data Analysis

The outcomes presented in this review were reported as results obtained during the whole circumcision procedure. Usually, the circumcision surgery was described as commencing with application of forceps to the dorsal foreskin of the penis (referred to as dorsal or lateral clamping) and ending with removal of the surgical clamp (the Gomco, Mogen, or Plastibell surgical device, also referred to as a clamp). Some authors reported a single numerical outcome result for the entire circumcision procedure (e.g. Butler O'Hara 1998; Howard 1999; Maxwell 1987; Taddio 1997). Others reported numerical results by procedure phase or step (e.g. dorsal foreskin grasped with forceps, adhesion lysis, dorsal incision, surgical clamp application, foreskin amputation, surgical clamp removal) (Benini 1993; Lander 1997; Woodman 1999). For the latter studies, depending on the outcome, we calculated either the arithmetic mean (e.g. heart rate) or total (e.g. time crying) across the phases or steps of the circumcision (as defined by the authors), and did not include the baseline or recovery phase data. Variance formulae for these arithmetic means and these totals were derived according to the general formula for linear combinations of variance (i.e. Var (X+Y) = Var(x) + Var(Y) + 2Cov(X,Y)). We assumed a correlation of 0.5 as proposed by Follmann 1992. Additional Tables 1 - 7 provide specific details on summary estimate extractions from the included studies.

Data were analysed using the statistical package (RevMan 4.2) provided by the Cochrane Collaboration. When two or more studies were identified that examined the same comparison and clinically similar outcomes, data were pooled using fixed effects. Random effects accounting for inter-study heterogeneity were considered in sensitivity analyses. Studies that compared an active intervention with placebo were analysed separately from those that compared the same active intervention with no treatment.

Continuous data summaries are reported as weighted mean differences (WMD) when the units provided were compatible. When the units were not compatible, the standardized mean difference (SMD) is reported. The SMD describes the difference between the treatments in terms of units of standard deviations (SDs). To improve interpretability, we also report estimates of WMDs derived from the estimated SMDs. To derive the WMDs from the SMDs, we selected either the unit used in the majority of the trials or the most clinically relevant unit under a particular comparison and pooled the available SDs from the trials that used that unit. We then multiplied this pooled SD by the SMD to obtain an estimate of the WMD. The WMDs thus derived are reported along side the SMDs in the results. An example of how a SMD was converted to a WMD is provided in Figure 1.

Individual study outcomes were reported as both final values (FVs) and change from baseline values (CVs). It is appropriate to combine FVs and CVs when combining mean differences to calculate a WMD. However, it is not appropriate, generally, to combine FVs and CVs when combining SMD to calculate an overall SMD. CVs often have smaller standard errors (SEs) than FVs since some of the intra-patient variation is removed from their SEs. Thus the individual study CV SMD tend to be in smaller SD units than the individual study FV SMD. However, in this systematic review, many of the SEs for CVs were either within the range of the FV SEs or they were larger, which is counterintuitive to the argument presented here. Hence, some of the SMD calculated in this review do combine CVs and FVs (Metagraphs 01.03; 01.08; 01.09; 03.02). Additional Tables 1 - 7 provide specific details on summary estimate extractions from the included studies.

Heterogeneity was assessed quantitatively with the I-squared statistic (Higgins 2003). The I-squared statistic indicates the percent variability due to between study (or inter-study) variability as opposed to within study (or intra-study) variability. An I-squared greater than 50% may be considered large. Only non-null heterogeneity statistics are presented here. Too few studies under a single comparison did not allow for any assessment of publication bias nor extensive sub-group or sensitivity analyses. However, post-hoc, we selected heart rate (the most frequently reported outcome) for between-study subgroup analyses using a chi-square method proposed by Deeks 2001. We selected the following subgroup analyses a priori: for penile block interventions, "wait time" from anaesthesia administration to start of the circumcision procedure were considered by the following three categories: no wait time reported, wait time </= 5 minutes, wait time >/= 5 minutes; for sucrose administration interventions, dose of sucrose administered was to be considered but could not be due to the lack of information provided in the reports. Surgical clamp type, use of pacifiers as a co-intervention, and choice of control group were selected for consideration post-hoc.

Description of studies

210 unique references were identified through search of the electronic databases. The full text of forty-two potentially relevant articles were obtained and reviewed for possible inclusion in this review. Six studies were excluded (see Table - Characteristics of Excluded Studies). In two excluded studies, subjects were not randomised and the intervention was chosen by the attending physician (Malnory 2003; Olson 1998). Two of the excluded studies had no comparison group (Mintz 1989, Russell 1996). One study was a cohort design and the outcome data for the control group was obtained from a previously conducted trial (Taddio 2000), and one (Taeusch 2002) was a head to head comparison of surgical clamps used for the circumcision procedure rather than a direct comparison of interventions for pain relief.

Thirty-five studies (thirty-six reports) were included in this systematic review. Details of each are given in the Table - Characteristics of Included Studies. Two reports outlined different outcome data from the same trial (Dixon 1984, Holve 1983). Two trials were reported as abstracts only (Zahorodny 1998, Zahorodny 1999) and we were unable to obtain additional information from the authors. One unpublished report of Master's thesis research was included (Zolnoski 1993).

Thirty-three of the thirty-five included studies enrolled healthy, full term neonates. One trial included infants born preterm (and less than 28 days age after reaching 40 weeks corrected gestational age) who were ready for discharge from the neonatal intensive care unit (NICU) at the time of circumcision (Butler O'Hara 1998), and one trial enrolled infants born preterm and weighing 1600 - 2500g at the time of circumcision (Holliday 1999).

Nineteen trials examined the effectiveness of penile blocks (Arnett 1990; Butler O'Hara 1998; Hardwick Smith 1998; Herschel 1998; Holliday 1999; Holve 1983; Howard 1999; Kass 2001; Kurtis 1999 A; Lander 1997; Masciello 1990; Maxwell 1987; Newton 1999; Spencer 1992; Stang 1988 A; Stang 1997; Williamson 1983; Williamson 1986; Williamson 1997). Twelve trials assessed topical anaesthetics (EMLA, lidocaine creams) (Benini 1993; Butler O'Hara 1998; Holliday 1999; Howard 1999; Joyce 2001; Lander 1997; Mohan 1998; Mudge 1989; Taddio 1997; Weatherstone 1993; Woodman 1999; Zahorodny 1998), and nine evaluated oral sucrose in a variety of concentrations and doses (Blass 1991 A; Herschel 1998; Kass 2001; Kaufman 2002; Mohan 1998; Stang 1997; Zahorodny 1998; Zahorodny 1999; Zolnoski 1993). In two trials, subjects received an oral analgesic (acetaminophen) (Howard 1994; Macke 2001). Three trials evaluated forms of environmental manipulation (e.g. music, intrauterine sounds) (Joyce 2001; Marchette 1989; Marchette 1991). For trial details see Table - Characteristics of Included Studies.

In the trials, the interventions were compared with placebo/sham treatments (e.g. saline penile block or inactive topical cream), no treatment, or with other active interventions. In several trials, all subjects received an active baseline intervention (e.g. EMLA cream or DPNB) prior to administration of the study intervention (Butler O'Hara 1998; Kaufman 2002; Stang 1997).

Methodological quality of included studies

All of the studies included in this systematic review were described as RCTs. However, fifteen of the study reports provided insufficient information or described inadequate procedures for assurance of blinding of randomisation (see Table - Characteristics of Included Studies). Nine were double-blind for delivery of all interventions (Howard 1994; Howard 1999; Joyce 2001; Macke 2001; Mudge 1989; Taddio 1997; Weatherstone 1993; Woodman 1999; Zolnoski 1993). Some studies compared interventions which could not be masked, for example, block techniques (Masciello 1990; Newton 1999; Spencer 1992). Partial blinding was achieved in several trials through inclusion of a sham or placebo group (Arnett 1990; Blass 1991 A; Holve 1983; Kass 2001; Kaufman 2002; Lander 1997; Stang 1988 A; Stang 1997). Blinding was occasionally achieved on a temporary basis during baseline assessments (Butler O'Hara 1998; Holliday 1999).

There was considerable methodologic diversity between the included studies. For example, there was variation between all of the trials as to what constituted the circumcision "procedure". In one trial (Williamson 1983), outcome data were reported for an undefined three minute "dissection period". In other trials, data were reported for each of multiple steps of a standardized procedure (Benini 1993; Hardwick Smith 1998; Herschel 1998; Lander 1997; Woodman 1999), or reported as a single summary statistic for the entire procedure (Taddio 1997). Other authors did not describe any details of the circumcision procedure followed for the trial (Blass 1991 A; Holliday 1999; Maxwell 1987; Stang 1988 A; Weatherstone 1993). In general, not enough information was provided by authors to be certain that outcome results were directly comparable across studies, as the events that constituted the procedure may not have been equivalent.

There were differences within the group of trials of DPNB (the most frequently studied intervention) in length of time fasting prior to surgery, anaesthetic dose, wait time after anaesthetic administration, and in type of surgical clamp used. In some cases, a single operator performed all circumcisions (Butler O'Hara 1998; Hardwick Smith 1998; Howard 1994), in others, the circumcisions were performed by a number of different operators (Howard 1999; Macke 2001; Stang 1997). Differences in operator technique or in the circumcision procedure could have effected outcome results. For most of the trials, subjects were required to fast prior to the surgery, however, the fasting period varied between trials from 30 - 90 minutes (Arnett 1990; Blass 1991 A; Herschel 1998; Kurtis 1999 A; Maxwell 1987) to 2 - 4 hours (Butler O'Hara 1998; Howard 1994; Kaufman 2002; Masciello 1990). Hunger could have influenced outcomes such as duration of infant crying or other behavioral responses. In a number of studies, subjects were offered pacifiers (Holliday 1999; Howard 1994; Howard 1999; Kurtis 1999 A; Mohan 1998; Spencer 1992; Stang 1997) although pacifiers were not the study intervention. In one trial, all subjects were offered sugar pacifiers (Butler O'Hara 1998). The potential effect of NNS on the outcomes measured in the trials providing pacifiers was not addressed in the reports.

Results

ACTIVE VERSUS PLACEBO OR NO TREATMENT COMPARISONS

• Penile block interventions

Fourteen trials compared dorsal penile nerve block (DPNB) to no treatment or placebo (sham injection). A total of 592 infants were included. Three trials employed pain scores (Metagraph 01.01) as an outcome measure (Arnett 1990; Holliday 1999; Kass 2001). These trials were not combined for meta-analysis of effect on pain because the scores used are not similar in conceptual development or measurement technique. However, outcomes significantly favoured DPNB using all four scores reported: infant irritability score [MD -1.8, 95% CI -2.4 to -1.2], modified behavioral pain scale (MBPS) [MD -3.2 , 95% CI -4.5 to -1.9], Holliday's behaviour score [MD -8.8, 95% CI -11.1 to -6.5], and the crying component of the same behavioral score [MD -9.8, 95% CI -13 to -6.6]. Another behavioral measure, time crying, also significantly favoured the DPNB group [WMD -54 %, 95% CI -64 to -44; SMD -1.74, 95% CI -2.1 to -1.4; Metagraph 01.02; SMD displayed, WMD derived from SMD; data not shown].

Among the physiological measures, heart rate significantly favoured DPNB [WMD - 35 bpm, 95% CI -41 to -30; SMD -1.6, 95% CI -1.8 to -1.3; I² = 73%; Metagraph 01.03; ; SMD displayed, WMD derived from SMD; data not shown]. Oxygen saturation results also significantly favoured DPNB [WMD 3.2 %, 95% CI 2.7 to 3.7; I² = 97%; Metagraph 01.06]. Results were heterogeneous, and one author reported the loss of large amounts of data (Herschel 1998). A single trial (Williamson 1983) reported results for transcutaneous oxygen saturation that also significantly favoured DPNB [MD 9.3 torr, 95% CI 1.8 to 16.9; Metagraph 01.07].

Respiratory rate (Metagraph 01.08) and serum B-endorphin (Metagraph 01.12) were not significantly different. Systolic blood pressure was reported in two studies. The combined result was significant and favoured DPNB [WMD -9 mmHg, 95% CI -16 to - 2; SMD -0.66, 95% CI -1.18 to -0.13; I²=92%; Metagraph 01.09; SMD displayed, WMD derived from SMD; data not shown] but the effect was not significant in the random effects model or when Maxwell 1987 was removed. The populations for these two trials were different. Maxwell 1987 recruited healthy newborns in the first few days of life, while Holliday 1999 enrolled low birthweight preterm infants. The preterm infants were cared for NICU and experience with other invasive procedures prior to circumcision may have affected their pain responses.

Serum cortisol (Metagraph 01.10) outcomes were reported in mg/dL, ug/dL and nmol/dL. Results were converted to nmol/dL using standard conversion factors and the outcomes expressed in these units were combined but were not significantly different. A single study (Kurtis 1999 A) reported salivary cortisol results and these were not significantly different.

In two studies (comparing DPNB to no treatment), authors did not report means and SDs. Williamson 1997 found significantly lower oxygen saturation and higher heart rates in the no treatment group during adhesion lysis and application of the surgical clamp (p< 0.001). There was a significant difference in duration of crying between the groups (p <0.001) with the DPNB group crying less. The second study found that the mean increase in heart rate and percent of time crying during circumcision was 50% less for infants in the DPNB group (p< 0.01) (Holve 1983). The DPNB group infants were more attentive to stimuli following circumcision, and were better able to quiet themselves when disturbed (Dixon 1984).

Ring block

Two trials compared ring block to no treatment and included 65 subjects (Hardwick Smith 1998; Lander 1997). Percent time crying was significantly reduced in the ring block group [WMD -26.3%, 95% CI -38 to -15, SMD -1.25, 95% CI -1.82 to -0.69; I² = 68%; Metagraph 02.01; SMD displayed, WMD derived from SMD; data not shown]. Only single studies reported other measures. In one (Lander 1997) heart rate significantly favoured the ring block group [MD -29 bpm, 95% CI -52 to - 7; Metagraph 02.02]. Oxygen saturation (Metagraph 02.03) and respiratory rate (Metagraph 02.04) were reported by Hardwick Smith 1998 and were not significantly different.

• Topical anaesthetics

Six studies compared EMLA to placebo for a total 200 patients (Benini 1993; Joyce 2001; Lander 1997; Taddio 1997; Woodman 1999; Zahorodny 1998). Two studies measured infant behavioral responses using the same pain score, the Neonatal Facial Coding System (Grunau 1987). The trials used the same measure, but the researchers scored a different set of facial actions (see Additional Table 01), and calculated the summary pain score differently. In both summation techniques, a lower score indicated less facial action and less pain. When combined, the results significantly favoured EMLA [WMD -46.5, 95% CI -80.4 to -12.6; SMD -0.6, 95% CI -1.0 to -0.2; Metagraph 03.01; SMD displayed, WMD derived from SMD; data not shown].

Cry time was also significantly decreased with EMLA treatment [WMD - 15.2 %, 95% CI -21 to - 9.3; SMD -0.78, 95% CI -1.08 to - 0.48; Metagraph 03.02; SMD displayed, WMD derived from SMD; data not shown]. One study (Joyce 2001) did not favour the EMLA treatment, but for this study cry time was measured from the start of circumcision until crying stopped or until 30 minutes elapsed. The other studies measured cry time by phases of the procedure or gave a summary value for the procedure and thus only time spent crying during circumcision surgery could be calculated.

Heart rate was significantly decreased for infants treated with EMLA [WMD -15 bpm, 95% CI -19 to -10; Metagraph 03.03]. The effect on oxygen saturation was not significant [WMD 0.9%, 95% CI -0.2 to 2.0; Metagraph 03.04], and heterogeneity was large (I²= 86%). Respiratory rate, systolic and diastolic blood pressure (Metagraphs 03.05, 03.06, 03.07) were not significantly different.

Lidocaine cream

Three trials compared topical lidocaine to placebo and included 115 patients (Mudge 1989; Weatherstone 1993; Woodman 1999). One study measured percentage of time spent in Brazelton behavioral state 6 (full cry) as a proxy for pain (Weatherstone 1993) and the results were insignificant [MD -8, 95% CI -23 to 7; Metagraph 04.01]. Cry time was significantly reduced [WMD -60 s, 95% CI -99 to -20; Metagraph 04.02] and favoured lidocaine. Heart rate was also significantly reduced [WMD -9 bpm, 95% CI -14 to - 4; I²=12%; Metagraph 04.03]. A single study examined B-endorphin levels, and these were significantly reduced for the group treated with lidocaine [MD -49 pg/mL, 95% CI -89 to -9; Metagraph 04.06]. One study (Mudge 1989) did not report standard deviations for oxygen saturation (Metagraph 04.04) and respiratory rate (Metagraph 04.05) and these could not be calculated from the information available. However, the direction of results favoured treatment with lidocaine. Oxygen saturation results for another study (Woodman 1999) were not significantly different.

• Oral sucrose/dextrose

Cry time results were not significantly different overall [WMD - 1.3 %, 95% CI -5.8 to -8.3; SMD 0.07, 95% CI -0.31 to 0.44; I² = 78%; Metagraph 05.02; SMD displayed, WMD derived from SMD; data not shown]. Individual results from five trials were inconsistent in direction. Zahorodny 1998 reported means only and SDs were substituted from another study using the same intervention and same outcome measure. One study (Kaufman 2002) reported a different measure of cry time. They averaged time spent crying in 10 second intervals, then took the cumulative average of that for each group. In this study, cry time was statistically significant and favoured the sucrose group (56 vs 86 s; P=0.0001). Zahorodny 1999 did not report means or standard deviations, but did report that both the sucrose and the water group cried much less than the no treatment group (p<0.001), and that subjects receiving the sucrose pacifier cried the least (p<.03). The sucrose and water groups in this trial also had smaller increases in heart rate compared to those receiving no intervention (p<.017). These authors did not comment on any differences between the sucrose group and the water group.

The effect on heart rate was not significant [WMD -4 bpm, 95% CI -9 to 2; Metagraph 05.03] overall in three trials. Heterogeneity was large (I²=55%) with two trials favouring the water treatment and one trial favouring the sucrose treatment. In two trials (Herschel 1998; Kass 2001) oxygen saturation was significantly greater in the sucrose group [WMD 1.8%, 95% CI 0.5 to 3.1; Metagraph 05.04] although heterogeneity was again large (I²=88%) and the random effects estimate was not significant [WMD 1.3%, 95% CI -2.7 to 5.2]. Serum cortisol (Metagraph 05.05) was measured in a single study (Stang 1997) and results were not significant.

The inconsistent results in these trials may be related to the volume and concentration of sucrose provided and to the sucrose delivery method. For example, in two studies the treatment group received a dose of 10 ml of 50% sucrose as the treatment intervention (Herschel 1998; Zahorodny 1999), while in two other studies, the treatment group received 2 ml of 50% sucrose (Kass 2001; Zahorodny 1998). The treatment groups in the other trials received 1.5 ml (Blass 1991 A), 2.3 ml (Zolnoski 1993) or an unspecified volume of 24% sucrose (Kaufman 2002; Stang 1997) respectively. The delivery method for the sugar solution also varied between studies. Five administered the sugar/water solution via a nipple/pacifier (Blass 1991 A; Herschel 1998; Kaufman 2002; Stang 1997; Zahorodny 1999) thus providing the opportunity for non-nutritive sucking. In one trial (Herschel 1998), the sucrose group had a nipple (and the opportunity for non-nutritive sucking throughout the circumcision procedure), while the no treatment control group did not receive a pacifier at all. In two studies, the sugar solution was delivered using oral syringes (Kass 2001; Zolnoski 1993). In one trial, the method of delivery was not specified (Zahorodny 1998).

• Oral analgesics

Two trials compared acetaminophen to placebo with a total of 104 patients (Howard 1994; Macke 2001). The studies employed two different pain scales, and the results were not combined because the measures are not similar in conceptual development or measurement technique. Howard 1994 used a comfort score that measures 10 behaviours (sleep, facial expression, motor activity, tone, etc.) to arrive at a composite score of 0 to 20. The lower the score, the more uncomfortable the infant. Macke 2001 used the Nursing Child Assessment Feeding Scale (NCAFS) which measures mother-infant feeding interactions using 76 behavioral items based on the concepts of synchronism and adaptation. Lower scores on the NCAFS indicate less positive responses on the part of the infant. Results using the post-operative comfort score were not significant, but the total infant scores on the NCAFS were significant and favoured acetaminophen [MD 4.0, 95% CI 1.0 to 7.1; Metagraph 06.01]. All other outcomes (cry time, heart rate, respiratory rate Metagraphs 06.02, 06.03, 06.04) were not statistically significant.

ACTIVE VERSUS ACTIVE TREATMENT COMPARISONS

• DPNB versus EMLA

Cry time was measured in a single study and was not significantly different [MD -10%, 95% CI -30 to 10; Metagraph 07.02]. Heart rate was significantly reduced for the DPNB group [WMD -17 bpm, 95% CI -23 to -11; Metagraph 07.03] but heterogeneity was large (I²=93%). The random effects estimate was not statistically significant. Butler O'Hara 1998 had a large mean difference [MD -40 bpm, 95% CI -51 to -29 ]; when this study was removed, heterogeneity was absent and the overall fixed effects WMD was no longer significant [WMD -7 bpm, 95% CI -14 to 0.5]. The large heterogeneity may be related to differences in the characteristics of the study subjects. Infants enrolled in the Butler O'Hara 1998 trial were born prematurely and hospitalised in the neonatal intensive care unit (NICU). Postnatal age was 3 - 105 days by the time of circumcision (but less than 28 days after reaching 40 weeks corrected gestational age). Exposure to invasive treatments during their NICU stay may have caused the infants to become sensitized and thus respond differently than infants in the other two trials who were healthy newborns in the first few days of life. Respiratory rate (Metagraph 07.05), measured by a single study, was not significantly different.

• DPNB versus sucrose

• DPNB versus ring block

• DPNB versus local block

• Ring block versus EMLA

• Buffered lidocaine DPNB versus lidocaine DPNB

• EMLA versus topical lidocaine

• EMLA versus sucrose

• EMLA versus music

ENVIRONMENTAL COMPARISONS

• Music versus no treatment

COMPLICATIONS/ADVERSE EFFECTS

Ten studies reported adverse effects (see Additional Tables - Table 08). Adverse effects including gagging, choking, and emesis were reported in untreated groups, while DPNB groups exhibited minor bleeding, swelling and hematoma at the block injection site post-circumcision. EMLA use was associated with erythema and minor skin pallor. In one study (Holliday 1999), two subjects who received EMLA had redness and blistering of the foreskin, leading to closure of the EMLA arm of the study. Methaemoglobin levels were measured in two trials of EMLA and found to be within normal limits (Lander 1997; Taddio 1997). All adverse effects of pain interventions were reported to be transient in nature and were not considered serious. Several authors reported on the questionable clinical utility of topical anaesthetic interventions (Herschel 1998; Howard 1999; Lander 1997) given the dexterity required to apply the creams properly and the lengthy application time.

SUBGROUP AND SENSITIVITY ANALYSES

We examined two subgroups: length of wait time after penile block interventions (a priori) and choice of clamp for all procedures (post-hoc) on the most frequently reported outcome heart rate. One study compared different lengths of wait time and two anaesthetics (Spencer 1992) but did not report means and SDs. The authors did report that DPNB groups that received either anaesthetic exhibited decreased pain responses compared to the control group. We made indirect (or between study) comparisons. Six trials comparing DPNB to no treatment prescribed and reported wait times. The trials with the longer wait time (>5 minutes) did not perform significantly better than short wait time trials (</=5 minutes) [Metagraph 01.04]. In fact, the probability under the null hypothesis was close to significant (P=0.09 vs P=0.65) when Maxwell 1987 was removed and favoured shorter wait times. A similar and statistically significant result was calculated when comparing wait times in DPNB vs EMLA (P=0.04; Metagraph 07.04). Using an indirect comparison, the Mogen clamp trial performed significantly better on reducing heart rate compared to the Gomco clamp trials when Maxwell 1987 was again removed (P=0.05 vs P=0.07) under the DPNB versus no treatment comparison (Metagraph 01.05). Sucrose dose (a priori) was not analysed because there were too few studies under the same comparison and not enough information was provided.

Post-hoc, we considered two other potential treatment effect size modifiers: control intervention and choice of pacifiers. For ethical considerations, use of saline DPNB in pain research was generally abandoned since the early 1990's. Among the included studies for this review, three used both saline DPNB treatment (sham) and no treatment control arms (Arnett 1990, Holve 1983, Stang 1988 A). In one study saline DPNB was used to blind comparison of lidocaine DPNB with another active intervention (Howard 1999). The researchers wanted to control for the effects of the injection and fluid volume compression on penile sensation. None of the studies found statistical differences between these control arms. In our review, the two control arms were displayed separately in the metagraphs when the data were reported separately in the referenced study (Stang 1988 A). Visually, we also see no difference. Other concerns for blinding involve placebo creams. One study (Mohan 1998) did not use a placebo cream and one study (Benini 1993) reported using petroleum jelly as a placebo for EMLA cream.

In nine trials pacifiers were made available to all patients (Butler O'Hara 1998; Holliday 1999; Howard 1994; Howard 1999; Kurtis 1999 A; Mohan 1998; Spencer 1992; Stang 1988 A; Stang 1997). In one (Butler O'Hara 1998) all infants were provided with sugar pacifiers although sucrose was not the intervention under study and its use may have affected results obtained on outcome measures. In another (Herschel 1998) only one out of the three study groups received a pacifier because it was used to deliver a sucrose intervention. At least two studies (Kass 2001; Zolnoski 1993) strictly prohibited the use of pacifiers and used oral syringes to deliver the sucrose intervention. The remaining studies did not report pacifier use. There were too few studies to compare within outcomes, and we could not identify obvious deviations with use or non-use of pacifiers. Of mention, Blass 1991 A and Zahorodny 1999 both found that in a water via pacifier group, cry time was significantly reduced compared with the no treatment control group (P<0.001; P<0.001).

Discussion

This systematic review incorporates data from 35 trials enrolling 1997 neonates to examine a variety of interventions for circumcision pain relief. Although the results are generally applicable to current practice, the review identified a number of important limitations of the primary studies included in the review and thus the results should be interpreted with some caution. Sample size in the majority of the trials was small (total sample size was < or = to 80 in 32 out of 35 trials), and there were some differences in the characteristics of the study subjects. Butler O'Hara 1998 enrolled neonates from an NICU that were between 3 and 105 days postnatal age at the time of circumcision (although still less than 28 days after reaching 40 weeks corrected gestational age). Holliday 1999 enrolled low birthweight neonates aged 25 - 27 days at circumcision. Each of these groups of subjects could have experienced numerous painful or invasive treatments during their stay in NICU prior to circumcision. Accordingly, their responses during circumcision, regardless of the intervention, could have been different from those of the healthy newborns that were recruited for the remainder of the trials.

All of the studies included in this review were described as randomised, but 15 of the trial reports did not include sufficient information or describe adequate procedures for assurance of blinding of randomisation. Nine of the trials were double-blind for all interventions, but some interventions such as block technique could not be blinded. In six trials of DPNB, a standardized approach to the circumcision procedure was not described in the reports, making it impossible to tell whether every infant underwent exactly the same surgical process. The impact of this could be intensified within individual trials where more than one operator performed the circumcisions (e.g. Howard 1999, Macke 2001). Other differences that may have affected outcome results between trials of the same comparison include the variability in wait time after block administration, length of fasting prior to circumcision, provision of pacifiers or other non-study interventions, and use of different surgical clamps. Finally, differences between trials in the structure of pain interventions were evident, especially in trials of oral sucrose where the dose and method of delivery varied substantially.

The studies included in this review reported on measurement of a variety of pain outcomes (physiological, behavioral, biochemical). Techniques and methods for measurement of outcomes were more dissimilar than similar across the trials, even within a single outcome variable (e.g. heart rate), and this presented significant challenges to combining outcome results. In particular, the dissimilarity in outcome measures severely limited the feasibility of combining pain scores across the included studies. None of the reports included in this review offered a definition of pain, and in general, the reports did not differentiate between the painful versus the distressing/stressful aspects of the circumcision procedure (e.g. removal of foreskin versus application of restraints). Reasons for selection of pain scores as outcome measures were not articulated in most cases, and among the included studies, a variety were used that differed in conceptual development and in measurement technique. Some pain scores were author-devised measures with no reported psychometric testing (Arnett 1990; Holliday 1999; Weatherstone 1993), while others measured behavioral indicators that were not conceptually linked to the neonate's experience of pain (Dixon 1984; Macke 2001; Newton 1999; Stang 1988 A). Others were subjective in their measurement technique (Howard 1999; Stang 1997). In six trials, researchers employed validated pain scales developed specifically to measure neonatal pain (Benini 1993; Butler O'Hara 1998; Howard 1994; Joyce 2001; Kass 2001; Taddio 1997).

Sixteen trials included in this review either did not report outcome data, or did not report data in a format that could be analysed in this review. One of the strengths of this review was that we were able to obtain additional information for three trials. Where means and standard deviations were not available, data were imputed or derived from graphs contained in the reports, and missing standard deviations were either calculated from other summary statistics or substituted with singular or mean standard deviations from similar trials allowing us to maximize the data included under each comparison.

DPNB was identified as the most effective intervention and demonstrated decreases in time crying and heart rate that were statistically and clinically significant (time crying 54% less, heart rate 30 beats per minute less) when compared with placebo or no treatment. EMLA also reduced pain responses when compared with placebo but the differences in time crying (15% less) and heart rate (15 beats per minutes less) were not as large as those observed with DPNB. Topical lidocaine demonstrated statistically significant decreases in time crying (60 seconds less), heart rate (9 beats per minute less) and serum B-endorphin levels (49 pg/ml less) compared to placebo. The issue of the statistical versus clinical significance of the outcome results was not discussed in any of the study reports, and no author identified a threshold for clinically significant (as opposed to statistically significant) intervention effects. It should be emphasized that none of the interventions examined in these trials completely eliminated pain responses to circumcision.

Ease of administration of the pain interventions will influence the applicability of the results of this review to current clinical practice. The relative ease of establishing the different penile blocks was not systematically evaluated, but it was suggested that the ring block technique is easier and safer because it eliminates the risk of injection of lidocaine into the dorsal vessels (Hardwick Smith 1998). A single study reported on use of local penile block (Masciello 1990) which appears to be similar in technique to ring block. Few adverse effects were reported with use of any of the penile blocks. EMLA and lidocaine topical anaesthetics are effective for circumcision pain when compared with placebo or no treatment, but their use may be precluded because of difficulties in application and the time required for maximum anaesthetic effect. Adverse effects such as transient skin reactions were reported but not considered serious, and methaemoglobin levels, when measured, were within normal range.

Reviewers' conclusions

Implications for practice

Circumcision is a painful procedure and routine or elective newborn circumcision is not recommended by either the American Academy of Pediatrics or the Canadian Paediatric Society. However, if circumcision is performed, the results of this review show that DPNB, RB and the topical anaesthetics EMLA and lidocaine cream can be recommended over no treatment for attenuation of circumcision pain. DPNB demonstrated the most consistent results, has been the most comprehensively studied, and was the most effective in terms of clinically significant reductions in pain responses. RB is also effective to reduce circumcision pain compared with placebo. The RB technique may be easier and safer to use because it eliminates the risk of injection of lidocaine into the dorsal vessels.

EMLA and lidocaine topical anaesthetics are effective for circumcision pain when compared with placebo or no treatment, but their use may be precluded because of difficulties in application and the time required for maximum anaesthetic effect. Adverse effects with EMLA use such as transient skin reactions were reported but not considered serious, and methaemoglobin levels, when measured, were within normal range. Topical anaesthetics are a less effective alternative to no treatment when expertise with penile blocks is not readily available.

Results for oral sucrose, oral analgesics and environmental modification interventions were either inconsistent or did not produce significantly different outcome results. These therapies cannot be recommended as treatments for circumcision pain.

None of the studied interventions completely eliminated the pain response to circumcision.

Circumcisions performed using the Mogen clamp take less time than is required using Gomco clamps. Shorter procedure time may reduce the total amount of pain experienced during circumcision, and may be important in terms of practitioner time to do the surgery.

Implications for research

Future studies should compare two or more active interventions for pain relief - a placebo or no-treatment control group is no longer acceptable. The impact of different "wait times" on the effectiveness of penile blocks and the relative acceptability and ease of administration of DPNB versus RB for practitioners should be systematically investigated. Use of the Mogen clamp in combination DPNB and RB should be investigated further to identify an optimal target time for circumcision surgery and to maximize anaesthetic effect. Although sucrose cannot be recommended as an intervention for circumcision pain at this time, research to determine the optimal dose and delivery method and the effect of combining oral sucrose with other interventions and comfort measures (e.g. nonnutritive sucking) should be pursued.

Lidocaine block and topical anaesthetic interventions could be useful in other situations where neonates undergo acute procedural pain. The pain associated with chest tube insertion, lumbar puncture, insertion of percutaneous central lines and other procedures commonly performed on high risk neonates may be significantly reduced with use of an appropriately adapted lidocaine block technique or topical anaesthetics. Further research should be pursued to identify situations where this potential can be examined.

Acknowledgements

We are grateful to Dr. J. R Holman, Dr. C Johnston, and Dr. J. F. Keck for their willingness to provide us with additional data from their studies.

Potential conflict of interest

Two of the reviewers, Brady-Fryer and Lander, were authors of a trial, Lander 1997, included in this review.

Characteristics of included studies

Characteristics of excluded studies

References to studies

References to included studies

Arnett RM, Jones JS, Horger EO. Effectiveness of 1% lidocaine dorsal penile nerve block in infant circumcision. Americal Journal of Obstetrics and Gynecology 1990;163:1074-80.

Benini 1993 {published data only}

Benini F, Johnston CC, Faucher D, Aranda J V. Topical anesthesia during circumcision in newborn infants. JAMA 1993;270:850-3.

Blass 1991 A {published data only}

Blass EM, Hoffmeyer LB. Sucrose as an analgesic for newborn infants. Pediatrics 1991;87:215-18.

Blass 1991 B {published data only}

Blass EM, Hoffmeyer LB. Sucrose as an analgesic for newborn infants. Pediatrics 1991;87:215-18.

Butler O'Hara 1998 {published data only}

Butler-O'Hara M, LeMoine C, Guillet R. Analgesia for neonatal circumcision: A randomized controlled trial of EMLA cream versus dorsal penile nerve block. Pediatrics 1998;101:e5.

Dixon 1984 {published data only}

Dixon S, Snyder J, Holve R, Bromberger P. Behavioral effects of circumcision with and without anesthesia. Journal of Developmental and Behavioral Pediatrics 1984;5:246-50.

Hardwick Smith 1998 {published data only}

Hardwick-Smith S, Mastrobattista JM, Wallace PA, Ritchey ML. Ring block for neonatal circumcision. Obstetrics and Gynecology 1998;91:930-4.

Herschel 1998 {published data only}

Herschel M, Khoshnood B, Ellman C, Maydew N, Mittendorf R. Randomized trial of a sucrose pacifier for pain control. Archives of Pediatrics & Adolescent Medicine 1998;152:279-84.

Holliday 1999 {published data only}

Holliday MA, Pinckert TL, Kiernan SC, Kunos I, Angelus P, Keszler M. Dorsal penile nerve block vs topical placebo for circumcision in low-birth-weight neonates. Archives of Pediatrics & Adolescent Medicine 1999;153:476-80.

Holve 1983 {published data only}

Holve RL, Bromberger PJ, Groveman HD, Klauber MR, Dixon SD, Snyder JM. Regional anesthesia during newborn circumcision. Clinical Pediatrics 1983;22:813-8.

Howard 1994 {published data only}

Howard CR, Howard FM, Weitzman ML. Acetaminophen analgesia in neonatal circumcision: the effect on pain. Pediatrics 1994;93:641-6.

Howard 1999 {published data only}

Howard CR, Howard FM, Fortune K, Generelli P, Zolnoun D, tenHoopen C et al. A randomized, controlled trial of a eutectic mixture of local anesthetic cream (lidocaine and prilocaine) versus penile block for pain relief during circumcision. American Journal of Obstetrics and Gynecology 1999;181:1506-11.

Joyce 2001 {published data only}

Joyce BA, Keck JF, Gerkensmeyer J. Evaluation of pain management interventions for neonatal circumcision pain. Journal of Pediatric Health Care 2001;15:105-14.

Kass 2001 {published data only}

Kass FC, Holman JR. Oral glucose solution for analgesia in infant circumcision. Journal of Family Practice 2001;50:785-8.

Kaufman 2002 {published data only}

Kaufman GE, Cimo S, Miller LW, Blass EM. An evaluation of the effects of sucrose on neonatal pain with 2 commonly used circumcision methods. American Journal of Obstetrics and Gynecology 2002;186:564-8.

Kurtis 1999 A {published data only}

Kurtis PS, DeSilva HN, Bernstein BA, Malakh L, Schechter NL. A comparison of the Mogen and Gomco clamps in combination with dorsal penile nerve block in minimizing the pain of neonatal circumcision. Pediatrics 1999;103:e23.

Kurtis 1999 B {published data only}

Kurtis PS, DeSilva HN, Bernstein BA, Malakh L, Schechter NL. A comparison of the Mogen and Gomco clamps in combination with dorsal penile nerve block in minimizing the pain of neonatal circumcision. Pediatrics 1999;103:e23.

Lander 1997 {published data only}

Lander J, Brady-Fryer B, Metcalfe J, Nazarali S, Muttitt S. A comparison of ring block, dorsal penile nerve block, and topical anesthesia for neonatal circumcision. JAMA 1997;278:2157-62.

Macke 2001 {published data only}

Macke JK. Analgesia for circumcision: Effects on newborn behavior and mother/infant interaction. Journal of Obstetric, Gynecologic, and Neonatal Nursing 2001;30:507-14.

Marchette 1989 {published data only}

Marchette L, Main R, Redick E. Pain reduction during neonatal circumcision. Pediatric Nursing 1989;15:207-10.

Marchette 1991 {published data only}

Marchette L, Main R, Redick E, Bagg A, Leatherland J. Pain reduction interventions during neonatal circumcision. Nursing Research 1991;40:241-4.

Masciello 1990 {published data only}

Masciello A L. Anesthesia for neonatal circumcision: Local anesthesia is better than dorsal penile block. Obstetrics and Gynecology 1990;75:834-8.

Maxwell 1987 {published data only}

Maxwell LG, Yaster M, Wetzel RC, Niebyl JR. Penile nerve block for newborn circumcision. Obstetrics and Gynecology 1987;70:415-19.

Mohan 1998 {published data only}

Mohan CG, Risucci DA, Casimir M, Gulrajani-LaCorte, M. Comparison of analgesics in ameliorating the pain of circumcision. Journal of Perinatology 1998;18:13-9.

Mudge 1989 {published data only}

Mudge D, Younger J B. The effects of topical lidocaine on infant response to circumcision. Journal of Nurse-Midwifery 1989;34:335-40.

Newton 1999 {published data only}

Newton CW, Mulnix N, Baer L, Bovee T. Plain and buffered lidocaine for neonatal circumcision. Obstetrics and Gynecology 1999;93:350-2.

Spencer 1992 {published data only}

Spencer DM, Miller K A, O'Quin M, Tomsovic J P, Anderson B, Wong D et al. Dorsal penile nerve block in neonatal circumcision: Chloroprocaine versus lidocaine. American Journal of Perinatology 1992;9:214-8.

Stang 1988 A {published data only}

Stang H J, Gunnar M R, Snellman L, Condon L M, Kestenbaum R. Local anesthesia for neonatal circumcision - Effects on distress and cortisol response. JAMA 1988;259:1507-11.

Stang 1988 B {published data only}

Stang HJ, Gunnar MR, Snellman L, Condon LM, Kestenbaum R. Local anesthesia for neonatal circumcision - Effects on distress and cortisol response. JAMA 1988;259:1507-11.

Stang 1997 {published data only}

Stang HJ, Snellman LW, Condon LM, Conroy MM, Liebo R, Brodersen L et al. Beyond dorsal penile nerve block: A more humane circumcision. Pediatrics 1997;100:e3.

Taddio 1997 {published data only}

Taddio A, Stevens B, Craig K, Rastogi P, Ben-David S, Shennan A et al. Efficacy and safety of lidocaine-prilocaine cream for pain during circumcision. New England Journal of Medicine 1997;336:1197-1201.

Weatherstone 1993 {published data only}

Weatherstone KB, Rasmussen LB, Erenberg A, Jackson EM, Claflin KS, Leff RD. Safety and efficacy of a topical anesthetic for neonatal circumcision. Pediatrics 1993;92:710-4.

Williamson 1983 {published data only}

Williamson PS, Williamson ML. Physiologic stress reduction by a local anesthetic during newborn circumcision. Pediatrics 1983;71:36-40.

Williamson 1986 {published data only}

Williamson PS, Evans ND. Neonatal cortisol response to circumcision with anesthesia. Clinical Pediatrics 1986;25:412-5.

Williamson 1997 {published data only}

Williamson ML. Circumcision anesthesia: A study of nursing implications for dorsal penile nerve block. Pediatric Nursing 1997;23:59-63.

Woodman 1999 {published data only}

Woodman PJ. Topical lidocaine-prilocaine versus lidocaine for neonatal circumcision: A randomized controlled trial. Obstetrics and Gynecology 1999;93:775-9.

Zahorodny 1998 {published data only}

Zahorodny W, Surez Y, Marshall R, Regan M, Holland B, Brendel D. Efficacy of EMLA and sucrose for crying associated with circumcision. In: Pediatrric Research. Vol. 43 Supplement 2. 1998:204.

Zahorodny 1999 {published data only}

Zahorodny W, David ES, Estrada P, Co J, Marshall R. Efficacy of a sucrose pacifier for newborn pain. In: Pediatric Research. Vol. 45 PART 2 of 2. 1999:17A.

Zolnoski 1993 {unpublished data only}

Zolnoski BAP. Effect of sucrose ingestion on pain during newborn circumcision [Masters thesis]. Univ. of Florida, 1993.

References to excluded studies

Malnory M, Johnson TS, Kirby RS. Newborn behavioral and physiological responses to circumcision. MCN. The American Journal of Maternal Child Nursing 2003;28:313-7.

Mintz 1989 {published data only}

Mintz MR, Grillo R. Dorsal penile nerve block for circumcision. Clinical Pediatrics 1989;28:590-1.

Olson 1998 {published data only}

Olson T, Wessman Downey V. Infant physiological responses to noxious stimuli of circumcision with anesthesia and analgesia. Pediatric Nursing 1998;24:385-9.

Russell 1996 {published data only}

Russell CT, Chaseling J. Topical anesthesia in neonatal circumcision: a study of 208 consecutive cases. Australian Family Physician 1996;25:530-4.

Taddio 2000 {published data only}

Taddio A, Pollock N, Gilbert-MacLeod C, Ohlsson K, Koren G. Combined analgesia and local anesthesia to minimize pain during circumcision. Archives of Pediatrics & Adolescent Medicine 2000;154:620-3.

Taeusch 2002 {published data only}

Taeusch HW, Martinez AM, Partridge JC, Sniderman S, Armstrong-Wells J, Fuentes-Afflick E. Pain during Mogen or Plastibell circumcision. Journal of Perinatology 2002;22:214-8.

* indicates the primary reference for the study

Other references

Additional references

American Academy of Pediatrics Task Force on Circumcision. Circumcision Policy Statement. Pediatrics 1999;103:686-93.

Abu-Saad 1998

Abu-Saad H, Bours G, Stevens B, Hamers J. Assessment of pain in the neonate. Seminars in Perinatology 1998;22:402-16.

Anand 2001

Anand KJS. Consensus statement for the prevention and management of pain in the newborn. Archives of Pediatrics & Adolescent Medicine 2001;155:173-80.

Anders 1974

Anders TF, Chalemian RJ. The effects of circumcision on sleep-wake states in human neonates. Psychosomatic Medicine 1974;36:174-9.

Auerbach 1978

Auerbach MR, Scanlon JW. Recurrence of pneumothroax as a possible complication of circumcision. American Journal of Obstetrics and Gynecology 1978;132:583.

Bellieni 2002

Bellieni CV et al. Effect of multisensory stimulation on analgesia in term neonates: A randomized controlled trial. Pediatric Research 2002;51:460-3.

Berde 2002

Berde CB, Sethna NF. Analgesics for the treatment of pain in children. New England Journal of Mecicine 2002;347:1094-1103.

Brazelton 1973

Brazelton TB. Neonatal behavioral assessment scale. In: Clinics in Developmental Medicine. Vol. 50. Philadelphia, PA: JB Lippincott, 1973:1-66.

Brazelton 1984

Brazelton TB. Neonatal behavioral assessment scale. Philadelphia: J B Lippincott, 1984.

Campos 1994

Campos RG. Rocking and pacifiers: two comforting interventions for heelstick pain. Research in Nursing & Health 1994;17:321-31.

Carbajal 1999

Carbajal R, Chauvet X, Couderc S, Olivier-Martin M. Randomized trial of analgesic effects of sucrose, glucose, and pacifiers in term infants. BMJ 1999;319:1393-7.

Connelly 1992

Connelly KP, Shropshire LC, Salzberg A. Gastric rupture associated with prolonged crying in a newborn undergoing circumcision. Clinical Pediatrics 1992;31:560-1.

Corff 1995

Corff KE, Seideman R, Venkataraman PS, Lutes L, Yates B. Facilitated tucking: a nonpharmacologic comfort measure for pain in preterm neonates. Journal of Obstetric, Gynecologic, and Neonatal Nursing 1995;24:143-7.

CPS 1996

Canadian Pediatric Society Fetus and Newborn Committee. Neonatal circumcision revisited. CMAJ 1996;154:769-80.

Dalens 1989

Dalens B, Vanneuvile G, Dechelotte P. Penile block via the subpubic space in 100 children. Anesthesia and Analgesia 1989;69:41-5.

Deeks 2001

Deeks JJ, Altman DG, and Bradburn MJ. Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Egger M, Davey Smith G, and Altman DG. Systematic reviews in health care. 2nd edition. London, UK: BMJ Books, 2001:300.

Fitzgerald 1993

Fitzgerald M, Anand KJS. Developmental neuroanatomy and neurophysiology of pain. In: Schecter N L, Berde C B, Yaster M, editor(s). Pain in infants, children and adolescents. Baltimore: Williams & Wilkins, 1993:11-31.

Follmann 1992

Follmann D, Elliot P, Suh I, Cutler J. Variance imputation for overviews of clinical trials with continuous response. Journal of Clinical Epidemiology 1992;45:769-73.

Frank 1997

Frank LS, Miaskowski C. Measurement of neonatal responses to painful stimuli: A research review. Journal of Pain and Symptom Management 1997;14:343-78.

Gibbons 2001

Gibbons S, Stevens B. Mechanisms of sucrose and non-nutritive sucking in procedural pain management in infants. Pain Research & Management 2001;6:21-8.

Gibbons 2002

Gibbonns S, Stevens B, Hodnett E, Pinelli J, Ohlsson A, Darlington G. Efficiacy and safety of sucrose for procedural pain relief in preterm and term neonates. Nursing Research 2002;51:375-82.

Gray 2000

Gray L, Watt L, Blass EM. Skin-to-skin contact is analgesic in healthy newborns. Pediatrics 2000;105:e14.

Grunau 1987

Grunau RVE, Craig KD. Pain expression in neonates: facial action and cry. Pain 1987;28:395-410.

Gunnar 1981

Gunnar MR, Fisch RO, Korsvik S, Donhowe JM. The effects of circumcision on serum cortisol and behavior. Psychoneuroendocrinology 1981;6:269-75.

Haouari 1999

Haouri N, Wood C, Griffiths G, Levene M. The analgesic effect of sucrose in full term infants: a randomized controlled trial. BMJ 1999;310:1498-1500.

Higgins 2003

Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-60.

Howard 1998

Howard CR, Howard FM, Garfunkel LC, de Blieck EA, Weitzman M. Neonatal circumcision and pain relief: Current training practices. Pediatrics 1998;101:423-8.

Johnston 1997

Johnston C, Stremler R, Stevens B, Horton L. Effectiveness of oral sucrose and simulated rocking on pain response in preterm neonates. Pain 1997;72:193-9.

Kharasch 2003

Kharasch S, Saxe G, Zuckerman B. Pain treatment: opportunities and challenges. Archives of Pediatrics & Adolescent Medicine 2003;157:1054-6.

Kirya 1978

Kirya C, Werthmann MW. Neonatal circumcision and penile dorsal nerve block - a painless procedure. Journal of Pediatrics 1978;92:998-1000.

Lawrence 1993

Lawrence J, Alcock D, McGrath P, Kay J, MacMurray SB, Dulberg C. The development of a tool to assess neonatal pain. Neonatal Network 1993;12:59-66.

Maxwell 1999

Maxwell LG, Yaster M. Analgesia for neonatal circumcision: No more studies, just do it. Archives of Pediatrics & Adolescent Medicine 1999;153:444-5.

McGrath 1990

McGrath PA. Pain in children. New York: The Guilford Press, 1990.

Mitchell 2000

Mitchell A, Brooks S, Roane D. The premature infant and painful procedures. Pain Management Nursing 2000;1:58-65.

Mitchell 2003

Mitchell A, Waltman P A. Oral sucrose and pain relief for preterm infants. Pain Management Nursing 2003;4:62-9.

Myron 1991

Myron AV, Maguire DP. Pain perception in the neonate: Implications for circumcision. Journal of Professional Nursing 1991;7:188-95.

Poma 1980

Poma PA. Painless neonatal circumcision. International Journal of Gynaecology and Obstetrics 1980;18:308-9.

Rawlings 1980

Rawlings DJ, Miller A, Engel RR. The effect of circumcision on transcutaneous pO2 in term infants. American journal of diseases of children 1980;134:676-8.

Ross 1988

Ross DM, Ross SA. Childhood pain. Baltimore: Urban & Schwarzenberg, 1988.

Ryan 1994

Ryan CA, Finer NN. Changing attitudes and practices regarding local anesthesia for newborn circumcision. Pediatrics 1994;94:230-3.

Snellman 1995

Snellman LW, Stang HJ. Prospective evaluation of complications of dorsal penile block for neonatal circumcision. Pediatrics 1995;95:705-8.

Stang 1991

Stang HJ. Response to : Pain perception in the neonate: Implications for circumcision. Journal of Professional Nursing 2001;7:194-5.

Stang 1998

Stang HJ, Snellman LW. Circumcision practice patterns in the United States. Pediatrics 1998;101:e5.

Stevens 1996

Stevens B, Johnston CC, Petryshun P, Taddio A. Premature infant pain profile: Development and initial validation. Clinical Journal of Pain 1996;12:13-22.

Stevens 1997

Stevens B, Taddio A, Ohlsson A, Einarson T. The efficacy of sucrose for relieving procedural pain in neonates - a systematic review and meta-analysis. Acta Paediatrica 1997;86:837-42.

Stevens 1999

Stevens B. Pain in infants. In: McCaffery M, Pasero C, editor(s). Pain Clinical Manual. 2nd edition. St. Louis: Mosby, 1999:626-673.

Sweet 1998

Sweet SD, McGrath PJ. Physiological measures of pain. In: Finley G A, McGrath P J, editor(s). Measurement of pain in infants and children. Seattle: IASP Press, 1998:59 - 81.

Taddio 1997b

Taddio A, Katz J, Ilersich A L, Koren G. Effect of neonatal circumcision on pain response during subsequent routine vaccination. Lancet 1997;349:599-603.

Taddio 1998

Taddio A, Ohlsson A, Einarson TR, Stevens B, Koren G. A systematic review of lidocaine-prilocaine cream (EMLA) in the treatment of acute pain in neonates. Pediatrics 1998;101:e1.

Taddio 2001

Taddio A. Pain management for neonatal circumcision. Paediatric Drugs 2001;3:101-11.

Taddio 2002

Taddio A, Ohlsson K, Ohlsson A. Lidocaine-prilocaine cream for analgesia during circumcision in newborn boys (Cochrane Review). In: The Cochrane Library, Issue 4, 2002. Oxford: Update Software.

Talbert 1976

Talbert LM, Kraybill EN, Potter HD. Adrenal cortical response to circumcision in the neonate. Obstetrics and Gynecology 1976;48:208-10.

Wellington 1993

Wellington N, Rieder MJ. Attitudes and practices regarding analgesia for neonatal circumcision. Pediatrics 1993;92:541-54.

Yarrow 1975

Yarrow L, Rubenstein JL, Pedersen FA. Infant and environment. New York: John Wiley and Sons, 1975.

Other published versions of this review

Brady-Fryer B, Wiebe N, Lander JA. Pain relief for neonatal circumcision. In: The Cochrane Database of Systematic Reviews, Issue 4, 2004.

Comparisons and data

Additional tables

01 Trials assessing pain/behavior scores

02 Trials assessing heart rate outcome variables

03 Trials assessing cry outcome variables

04 Trials assessing oxygen saturation outcome variables

05 Trials assessing respiratory rate outcome variables

06 Trials assessing biochemical outcome variables

07 Trials assessing blood pressure outcome variables

08 Trials reporting adverse effects

Additional figures

Figure 01

Notes

Published notes

Contact details for co-reviewers

Dr Janice A Lander, PhD
Professor
Faculty of Nursing
University of Alberta
5-120A Clinical Sciences Building
University of Alberta
Edmonton
Alberta CANADA
T6G-2G3
Telephone 1: 780-492-6317
Facsimile: 780-492-9954
E-mail: jlander@ualberta.ca

Ms Natasha Wiebe, M Math
Statistician
Alberta Research Centre for Child Health Evidence
University of Alberta - Department of Pediatrics
Room 9417 ABC
11402 University Avenue
Edmonton
Alberta CANADA
T6G 2J3
Telephone 1: 780-482-1273
Facsimile: 780-407-6435
E-mail: nwiebe@ualberta.ca
Secondary contact person's name: Natasha Wiebe