Nonpharmacological management of procedural pain in infants and young children : An abridged Cochrane review

1Department of Psychology, York University; 2Department of Psychiatry, The Hospital for Sick Children and University of Toronto, Toronto, Ontario; 3Department of Psychology, Dalhousie University and IWK Health Centre, Halifax, Nova Scotia; 4Centre for Nursing, The Hospital for Sick Children and Lawrence S Bloomberg Faculty of Nursing, University of Toronto, Toronto, Ontario Correspondence: Dr Rebecca Pillai Riddell, Department of Psychology, York University, 4700 Keele Street, Toronto, Ontario M3J 1P3. Telephone 416-736-2100 ext 33204, fax 416-736-5814, e-mail rpr@yorku.ca Initial misinterpretations of common infant pain outcomes, such as the lack of declarative memory for painful experiences during infancy (1), the muted responses of premature infants after a barrage of painful procedures (2), and unacceptable rates of serious adverse events due to poor knowledge of infant responses to analgesics and anesthetics during the 1950s to 1970s (3), perpetuated widespread neglect of infant pain treatment. Established research supports infants’ anatomical and functional capacity to perceive pain (4,5) and respond to tissue insult in a manner interpretable as pain (6). However, despite significant advocacy work, infant acute pain is still undermanaged or unmanaged (7). Comprehensive meta-analytic reviews of nonpharmacological pain management strategies are essential to the ethical and humane treatment of infants and young children during acutely painful procedures. While several reviews summarize certain acute pain management techniques for painful procedures in infants and young children (8-10), no comprehensive meta-analyses were attempted. Moreover, to our knowledge, there have been no meta-analyses conducted on the broad range of nonpharmacological interventions for managing acute pain and distress in young children from zero to three years of age. Given the frequency of acutely painful procedures for healthy children in early childhood (eg, immunizations) and for hospitalized infants (11), this significant gap in the literature deserves a higher priority. The full review (12) also takes a novel and in-depth look at the nonpharmacological pain management literature by controlling for age and type of pain response. Age was considered crucial due to the steep trajectory of infant development, both psychologically and physiologically. Moreover, given the different physiological and psychological mechanisms subsuming the initial reaction to a painful stimulus (more automatic and/or reflexive reactivity such as initial distress cry) and during the period of recovery from the painful insult (more under voluntary control; such as whimpering cry minutes after a review

painful stimuli), timing of pain response was also considered important to explore (see Hadistavropolis and Craig [13] for in-depth discussion regarding observational measures of reflexive and voluntary pain reactions during infancy and childhood).

MeThods search strategy
Only randomized controlled trials (RCTs) and randomized crossover trials (RCrTs) using a no-treatment control group that involved the nonpharmacological management of acute procedural pain in infants and children zero to three years of age were included.No language restrictions were used during the search.Due to the existence of metaanalyses specifically pertaining to circumcision surgery (14,15), sucrose (16), breastfeeding (17) and music (18), these types of studies were excluded from the review.
A The Cochrane Library and www.clinicaltrials.gov(2010).Appeals were made to pediatric electronic mailing lists (Pain in Child Health [PICH; pich-l@lists.dal.ca],Pediatric Pain [Pediatric-pain@lists.dal.ca], American Psychological Association Division 54 [Pediatric Psychology; div54-members@lists.apa.org]).Finally, the reference lists of recently published reviews were also consulted.

study selection
Review authors were not blinded to authors, institutions, journals or results.Using the full-text articles, authors reviewed articles for inclusion (see Acknowledgements). Figure 1 displays the summary of the literature search.

data extraction
Data extraction was conducted using a form designed for the present review.Each form also included a risk of bias/quality questionnaire (online Appendix 1).Every included article was extracted independently by at least two authors and compared.When two authors disagreed, the original article and/or the lead author was consulted to achieve consensus.
Qualitative, quantitative, and study quality data were compiled in Excel 2007 (Microsoft Corporation, USA), RevMan 5 or SPSS version 19.0 (SPSS Inc, USA), respectively.A random sampling of 25% (RevMan5) to 30% (Excel and SPSS) of all data entered across all three programs was double checked by a subteam of review authors.
Only objectively measured behaviorally-based responses to pain were analyzed (see full review (12) for detailed list of measures included).When studies had more than one unidimensional behavioral pain measure, the most specific measure available was used.For example, pain facial expression was used over cry duration.If multimodal measures were used (ie, measures that provided a total score based on behavioural and physiological measurement or multiple behavioural scales), the total score was used.

Quality and treatment integrity assessment
Every study included in the present review was scored for quality and treatment integrity by at least two review authors.The Yates Quality of Study Design and Methods Scale (19) was used to develop the study quality rating form for this review, as this scale was specifically validated for nonpharmacological treatments.
Three minor modifications were made to the Yates scale (12).The maximum score on the revised Yates scale was 23 points, with higher scores indicating higher quality (Table 3 for risk of bias scores).See online appendix for the modified measure.
While the majority of studies were deemed of satisfactory-marginal quality, five studies were classified as lower/unknown quality due to receiving a score of 13 or less (20)(21)(22)(23)(24).
Treatment integrity was also assessed because almost one-half of the analyzed studies were crossover designs.Two raters independently judged the treatment integrity on five dimensions: treatment adherence, treatment exposure, quality of delivery, participant responsiveness, and program differentiation (25) to arrive at an overall judgment.Only one study had lower/unknown treatment integrity (26).data synthesis overall strategy: Three main principles guided the meta-analysis of the data collected for the present review: 1. Certain types of methods had to be excluded to validly pool results.
Accordingly, for a treatment study to be included in the metaanalysis, it had to have at least one trial that was an RCT or RCrT that used a no-treatment control group and included a measure that included a behavioural pain outcome.Trials that studied a relevant nonpharmacological treatment that did not meet these requirements were qualitatively described within the review to further contextualize the findings.2. Studies within the 13 different treatments (Table 3

statistical analyses Primary analyses:
The results from individual studies were pooled using the generic inverse method for a random-effects model in RevMan 5. Using this statistical methodology, an index of the variability of the sample (standard error) and sample size was used to determine how influential each study would be to the final meta-analytic statistic (standardized mean difference [SMD]).A 95% CI was also reported, which incorporated the standard error of the pooled treatment effect for the treatment effect (ie, SMD).As RCTs and RCrTs were included, different procedures, obtained from a Cochrane statistical consultant, were used.Details of this procedure are presented elsewhere (12).When data were missing, study authors were contacted or recommended techniques for interpolation/estimation from P values, t scores and F scores were used (25).secondary analyses: When two or more trials were meta-analyzed in the primary analysis and heterogeneity and/or quality was an issue, secondary sensitivity analyses were conducted.The I 2 statistic ( 27) determined the impact of heterogeneity.When heterogeneity was an issue due to highly variable SMD(s), analyses were re-run without these studies to examine the impact of the pooled findings.When analyses included a study of lower/unknown quality or treatment integrity, analyses were also re-run without these studies to examine the impact on the pooled findings.Authors' conclusions were based on a synthesis of all three potential analytic steps.Qualitative information from excluded studies were only provided as additional contextual information.

ResulTs
For the final analysis, 51 trials (3396 participants) were included quantitatively (Table 1) (20)(21)(22)(23)(24)26, and 20 trials (Table 2) (73-92) were qualitatively mentioned.Thirty-five separate primary analyses were run among the 13 interventions, three age groups and two pain response types.Table 3 presents the primary meta-analytic results from this review (SMD, 95% CI and I 2 ) and, when applicable, the secondary statistics re-run with studies removed due to heterogeneity and/or study quality.
Of the 3396 participants, 1581 were in treatment conditions only, 1153 were in control conditions and 662 were in a crossover condition.Of the 51 studies, 21 used a cross-over design and 30 used a between-groups design.
The following painful procedures (determined by respective study authors rather than review authors) were included in this review: 29 studies examined treatments for heelstick, 10 studies examined needle-injection procedures, six studies assessed venipuncture, two examined NICU diaper changes, two studies investigated endotracheal suctioning and two studied a neonatal intensive care unit (NICU) weighing procedure.

disCussioN
The summary interpretation of the primary meta-analytic findings, contextualized by secondary heterogeneity and quality/treatment integrity analyses, are presented in Table 4. Based on these results, treatments were assigned a number from 1 to 4, for each age and pain response type.As will be discussed below, the ratings reflect whether, as the literature currently stands, evidence supported the specific treatment for pain management (efficacy) or did not support the specific treatment for pain management (inefficacy).Each treatment's efficacy or inefficacy was further qualified by the level of support (sufficient versus limited).
Treatment efficacy was denoted by either a 1 (sufficient evidence, ie, two or more quality trials supporting efficacy) or 2 (limited evidence, ie, either due to quality, quantity or heterogeneity of trials, supporting efficacy).Treatment inefficacy was denoted by either a 3 (limited evidence [ie, either due to quality, quantity or trial heterogeneity]) or a 4 (sufficient evidence [ie, two or more quality trials supporting inefficacy]).Blank cells indicate no applicable research for that combination of treatment, age and pain response.A discussion of each of the findings follows.

Kangaroo care (also known as skin-to-skin contact)
An infant is placed on their caregiver's bare chest before, during and after a painful procedure.Preterm infants: Sufficient evidence suggests kangaroo care is efficacious in reducing pain reactivity and improving immediate painrelated regulation.While there was substantial heterogeneity, secondary analyses confirmed this finding.
Four studies that were excluded from the statistical analyses (73-76) also indirectly support kangaroo care as efficacious in improving pain reactivity and immediate pain-related regulation in preterm infants.Neonates: Limited evidence suggests that kangaroo care is not efficacious as an intervention for pain reactivity or immediate pain-related regulation.However, heterogeneity undermines our confidence in the pooled results.Given the exposure times in the premature infant literature, future research should explore whether using a longer exposure time in kangaroo care for neonates prior to the painful procedure (ie, 10 min or longer akin to preterm techniques), could lead to a significant treatment effect.swaddling/facilitated tucking A swaddled infant is securely wrapped in a blanket to prevent excessive movement.Facilitated tucking is a hand-swaddling technique that holds the infant's extremities flexed and contained.Preterm infants: There was sufficient evidence to support the use of swaddling/tucking as an efficacious intervention for reducing painrelated distress reactivity and immediate pain-related regulation in preterm infants.Two studies (74,77), that were not included in the analysis due to use of an active control group, suggested that swaddling was as efficacious as containment but not as efficacious as kangaroo care.Neonates: Limited evidence supports the efficaciousness of swaddling/ tucking related interventions for the healthy neonate.

Non-nutritive sucking-related strategies
An object (eg, pacifier, nonlactating nipple) is placed into an infant's mouth to stimulate orotactile or sucking behaviours during a painful event.
Preterm infants: There is sufficient evidence that sucking is efficacious in reducing pain-related distress reactivity and improving immediate pain-related regulation.Pain relief may be maximized if sucking begins at least 3 min before the painful stimuli.Two studies that were not included in the analyses due to the use of an active control group (85,89) also suggest that sucking helps diminish pain reactivity.Neonates: The results show sufficient evidence for sucking to reduce pain reactivity and immediate pain-related regulation.Four studies that were not included due to the exclusion criteria (84)(85)(86)91), also lend support to the efficacy of sucking to improve immediate painrelated regulation.older infants: Limited evidence suggests that sucking may be an efficacious intervention to improve pain reactivity.

swallowing water
Water is administered for ingestion without inciting extensive sucking (eg, water administered by a dropper).Preterm infants: There was limited evidence that water is an inefficacious intervention for pain reactivity or immediate pain-related regulation for preterm infants.Neonates: There was limited evidence that water is an inefficacious intervention for pain reactivity or immediate pain-related regulation.older infants: There was limited evidence that water is an inefficacious intervention for immediate pain-related regulation.
The above studies used 'water' as a treatment arm (comparing them to a 'no-treatment' control), while most other studies in the literature used water as the 'no-treatment' control group.Given the more common use of water in the literature and the limited evidence at every age group of its inefficacy, it is not recommended that further research use water as a treatment arm for young child procedural pain studies.

Rocking and/or holding
An infant is held and/or gently moved up and down or side-to-side by a caregiver.
Neonates: In terms of pain reactivity, rocking/holding was not efficacious in reducing pain reactivity but substantial heterogeneity reduces our confidence.One study, not included due to lack of randomization (89), suggested a significant difference in pain reactivity between infants who were held and control infants.However, there was sufficient evidence to support the efficaciousness of rocking/holding interventions for immediate pain-related regulation.older infants: There was limited evidence suggesting rocking/holding is not an efficacious intervention for pain-related distress reactivity in older infants.

Artificial rocking and water
An infant is placed in a bassinet-type machine that provides a swaying motion.Water is administered via a dropper.Preterm infants: Limited evidence indicates that simulated rocking and water is not an efficacious intervention for reducing pain-related distress pain reactivity for preterm infants.
Touch/massage/therapeutic touch An infant's body (i.e.touch, massage) or energy field (therapeutic touch) is 'stroked' or rubbed to provide some type of counter-stimulation to the nociceptive input.Preterm infants: Current evidence does not support touch/massagerelated interventions as efficacious in improving the immediate painrelated regulation but caution is warranted given the presence of substantial heterogeneity.One study not included in the analysis due to exclusion criteria (79) demonstrated that massage was more efficacious at reducing preterm infant's heart rate than light pressure or no massage therapy.Neonates: Limited evidence suggests touch/massage related interventions are not efficacious to reduce pain reactivity or immediate painrelated regulation.One study that was not included in the analysis due to exclusion criteria (78) provides further support to these findings.older infants: Limited evidence suggests that therapeutic touch is not efficacious in reducing pain reactivity in older infants.

environmental modification
Interventions involved modifying the environment in which an infant experiences painful procedures (ie, low noise and lighting, clustering procedures to avoid over handling, soothing smells).Preterm infants: While the pooled result from two studies suggest that environmental modification was not efficacious for pain reactivity, this must be interpreted with caution due to substantial heterogeneity.However, there is limited evidence to suggest that environmental modification is efficacious for immediate pain-related regulation.

Toy distraction
Toy distraction is defined as the use of a toy to divert attention from the painful stimulus.older infants: Sufficient evidence suggests that toy distraction is not efficacious for reducing pain-related distress reactivity in older infants.
Limited evidence suggests that it is also not efficacious for improving immediate pain-related regulation.

video distraction
An audio-visual screen displaying two-dimensional moving images with coordinated audio is used to divert the infants attention from the painful stimulus older infants: Limited evidence suggests that video distraction is efficacious in reducing pain-related reactivity.Limited evidence also supports efficacy for improving immediate pain-related regulation.

structured parental involvement
Parents are instructed about strategies that are accepted as painreducing but are not given any materials to aid them (eg, rocking, holding, shushing, talking, rubbing, tickling, and distracting attention without toy or video).older infants: Structured parent involvement was not found to be efficacious for pain reactivity.However, caution should be applied to this finding as post-hoc heterogeneity analyses contradicted this finding when one outlying study was removed.In addition, one excluded study also suggested that structured parental involvement significantly improved time to regulate post-immunization (90).

simulated maternal voice
An infant is exposed to a reproduction of the mother's voice to help simulate the fetal environment.Preterm infants: Results from one study indicated that mother's voice was not more efficacious than a no-treatment control for reducing pain-related distress reactivity for preterm infants.

Parent presence
The parent is present during a painful procedure but not interacting extensively with the child in a manner thought to be pain reducing.older infants: Limited evidence indicates that parent presence is not efficacious for improving immediate pain-related regulation.

CoNClusioNs implications for practice
For preterm infants, there was sufficient evidence to recommend kangaroo care, sucking-related interventions, and swaddling/facilitated tucking interventions for both pain reactivity and immediate painrelated regulation.For neonates, there was sufficient evidence to • Kangaroo care for older infants' pain reactivity and immediate pain-related regulation (eg, for the two-month immunization) • Sucking-related interventions for older infants'/young children's pain reactivity • Swaddling or tucking-related interventions for older infants pain reactivity and immediate pain-related regulation (eg, two-month immunization) • Rocking/holding for older infants' immediate pain-related regulation In addition, preliminary work from other studies (excluded from our overall quantitative analyses for methodological reasons) suggests that more research is needed to explore: exposing an infant to a familiar odor (80,87,88), feeding an infant formula (89) and administering the least painful immunization first (92) as potential nonpharmacological interventions for acute pain.It is also important to note that certain treatments were grouped together based on similar mechanisms of action despite not being the same treatment (eg, swaddling and tucking, non-nutritive sucking related strategies), which was supported by our heterogeneity analyses.However, with more research, future revisions may be able to report SMDs and CIs separately for treatments encompassed within these groups.
Moreover, given the frequency of immunization during the first years of life, it was disheartening that there were no efficacious nonpharmacological treatments for the older infant/young child.Although there is a substantial evidence base for pharmacological strategies such as sucrose and topical anaesthetics (93), from an economic and pragmatic perspective, it would behoove researchers to spend more resources in investigating efficacious nonpharmacological pain management for older infants.
The lack of developmentally-informed work on parent-mediated interventions was also a cause for a concern.Over the first years of life, it has been argued (94) that the caregiver is the most important context for the infant in pain.Currently, studies that have attempted to formally structure parent behaviour have been limited and, thus, shown to be ineffective.More work on improved parent interventions, especially ones that capitalize on an infant's primary need for proximity to the parent during periods of distress (95), is needed.Teaching a parent how to soothe more efficaciously their infant or young child in acute pain is a simple, low-cost intervention that is not being used to its potential in today's acute pain context.
In conclusion, while a number of non pharmacological treatments have a sufficient body of evidence supporting their use with preterm infants and to a lesser extent, healthy neonates, more research is needed to create a sound repertoire of empirically-supported nonpharmacological treatments for procedural pain in older infants/young children.

Table 1 Quantitative studies Included in analyses
Figure 1) Literature search results.RCT Randomized controlled trial

Table 1 -CONTINUeD
Significant gaps in the existing treatment literature on non pharmacological management of acute pain in infancy have been discerned.Based on established patterns of efficacy in other age groups/pain response types, it would seem especially productive to the field of infant pain management to investigate:

Table 3 Summary of meta-analyses
*Dash indicates no research performed for that treatment, age and pain response combination