PAIN 141 (2009) 79–87

w w w . e l se v i e r . c o m / l o c a t e / p a i n

Long-term impact of neonatal intensive care and surgery on somatosensoryperception in children born extremely preterm

Suellen M. Walker a,b,*, Linda S. Franck a, Maria Fitzgerald b, Jonathan Myles c, Janet Stocks a, Neil Marlow d

a Portex Unit, Pain Research and Respiratory Physiology, UCL Institute of Child Health, 6th Floor Cardiac Wing, 30 Guilford St, London WC1N 1EH, UKb Neuroscience Physiology and Pharmacology, UCL, Medawar Building, Malet Place, London WC1E 6BT, UKc Barts and the London School of Medicine and Dentistry, Queen Mary University of London, Turner St, London E1 2AD, UKd School of Human Development, University of Nottingham, Queen’s Medical Centre, Nottingham, NG7 2UH, UK

a r t i c l e i n f o

Article history:Received 3 July 2008Received in revised form 18 September2008Accepted 20 October 2008

Keywords:NeonateIntensive careChildrenSomatosensoryPainDevelopmentQuantitative sensory testing (QST)

0304-3959/$34.00 � 2008 International Association fdoi:10.1016/j.pain.2008.10.012

* Corresponding author. Address: Portex AnaestheRespiratory Physiology, UCL Institute of Child HealthGuilford St, London WC1N 1EH, United Kingdom. Te+44(0)20 78298634.

E-mail address: suellen.walker@ich.ucl.ac.uk (S.M.

a b s t r a c t

Alterations in neural activity due to pain and injury in early development may produce long-term effectson sensory processing and future responses to pain. To investigate persistent alterations in sensory per-ception, we performed quantitative sensory testing (QST) in extremely preterm (EP) children (n = 43)recruited from the UK EPICure cohort (born less than 26 weeks gestation in 1995) and in age and sexmatched term-born controls (TC; n = 44). EP children had a generalized decreased sensitivity to all ther-mal modalities, but no difference in mechanical sensitivity at the thenar eminence. EP children who alsorequired neonatal surgery had more marked thermal hypoalgesia, but did not differ from non-surgical EPchildren in the measures of neonatal brain injury or current cognitive ability. Adjacent to neonatal tho-racotomy scars there was a localized decrease in both thermal and mechanical sensitivity that differedfrom EP children with scars relating to less invasive procedural interventions or from those without scars.No relationship was found between sensory perception thresholds and current pain experience or paincoping styles in EP or TC children. Neonatal care and surgery in EP children are associated with persistentmodality-specific changes in sensory processing. Decreases in mechanical and thermal sensitivity adja-cent to scars may be related to localized tissue injury, whereas generalized decreases in thermal sensitiv-ity but not in mechanical sensitivity suggest centrally mediated alterations in the modulation of C-fibrenociceptor pathways, which may impact on responses to future pain or surgery.

� 2008 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

The normal development of somatosensory and pain processingis dependent upon sensory activity in early life [13,16]. As ourunderstanding of the plasticity of developing pain mechanismshas increased, an important question has arisen as to whether pro-longed or repeated noxious inputs in newborn infants have long-term effects upon the central nervous system. The rate of pretermbirth is rising, and improvements in neonatal intensive care meanthat babies are surviving from much younger postconceptionalages [11,44]. Repeated procedural interventions, which are essen-tial for monitoring and intensive care management [49], producesignificant cortical pain responses in even the youngest infant[51], which may not always be reflected by behavioral changes[50]. In addition, major and/or repeated surgery may be requiredto treat complications of prematurity or to correct congenital

or the Study of Pain. Published by

sia Unit, Pain Research and, 6th Floor Cardiac Wing, 30l.: +44(0)20 79052382; fax:

Walker).

anomalies. The impact of altered neural activity due to pain and in-jury at these early developmental stages on subsequent sensoryfunction requires specific evaluation.

Current evidence for long-term effects following early injury iscompelling but incomplete. In clinical studies, both long-term in-creases and decreases in pain sensitivity have been reported fol-lowing surgical, procedural and intensive care interventions inthe neonatal period, as studies vary in methodology (e.g. baselinechange vs response to subsequent painful stimulus), outcome(e.g. parental or self-report questionnaires, observer scores ofbehavioral responses), and age at evaluation [18,38,52]. However,the cognitive and behavioral impairments reported in large cohortsof children born preterm [3,33] may confound the interpretation ofalterations in pain behavior. More specific evaluation of changes insomatosensory processing may be achieved with quantitative sen-sory testing (QST), which provides objective, reliable measures ofsensory thresholds in children [22,34]. Generalized reductions insensory threshold have been demonstrated in children after sur-gery and/or intensive care in the neonatal period [21,26,47], butthe extent to which sensory changes reflect the degree of tissue in-jury or are influenced by cognitive impairment requires furtherevaluation.

Elsevier B.V. All rights reserved.

80 S.M. Walker et al. / PAIN 141 (2009) 79–87

We performed quantitative sensory testing in extremely pre-term (EP) and term-born control (TC) children with the primaryaims of quantifying generalized changes in mechanical detectionand thermal perception thresholds at a reference point on the the-nar eminence, and comparing modality-specific generalizedchanges in sensory thresholds with localized changes adjacent toscars. Secondarily, we evaluated associations between (i) the de-gree of sensory change in EP children and the degree of proceduralor surgical intervention during the neonatal period and (ii) neona-tal experience and report of current pain, pain interference andpain coping strategies.

2. Methods

2.1. Participants

A cohort study was conducted to compare sensory thresholds inextremely preterm (EP) children (n = 43) and matched healthy full-term children (term controls, TC; n = 44). EP children were re-cruited from the EPICure cohort comprising children born < 25weeks 6 days gestation in the United Kingdom and Republic of Ire-land from March to December 1995 [7]. Of the 307 EP survivors,219 (71%) participated in a range of follow-up evaluations at 11years of age (see Table 1). An additional 6 EP children born in early1996, who were not included in the initial cohort but consented toall subsequent follow-up, also participated. Initial studies of respi-ratory function were performed at school, and spirometry readingswere obtained in 186 EP children. Of these children, a conveniencesample of 49 children who lived within 2 h travelling time fromLondon attended UCL Institute of Child Health for further analysisof cardiorespiratory function [31]. Forty-three children were avail-able for sensory testing. Based on the closest previous study [21],our initial aim was to recruit at least 27 EP and TC children for80% power to see similar effects. Ongoing recruitment increasedthe proportion who had undergone neonatal surgery. Term-bornclassmate controls were identified by EPICure assessors andteachers, as previously described [33], and recruitment packages

Table 1Sample recruitment.

EPICure cohort - children born <26 weeks PMA admitted to intensive care between March 1st 1995 and Dec 31st 1995- 276 maternity units in UK and Republic of Ireland: 314 survived to discharge

307 SURVIVORS at 11 years

11 abroad, 20 refused participation, 56 non-responders

219 born in 1995 + 6 born in 1996* ASSESSED AT SCHOOL AT AGE 11 - include tests of cognitive function, neurosensory disability, spiromtery

13 unable to perform spirometry, 12 poor quality recordings, 11 no consent for respiratory tests, 3 technical problems

186 COMPLETE SPIROMETRY TESTING

132 subjects more than 2 hours travelling time from London, 4 did not consent (too far to travel), 1 lost contact

49 ATTEND UCL INSTITUTE OF CHILD HEALTH- cardiorespiratory and sensory testing

3 (1F, 2M) inadequate time for sensory testing (single day attendance)3 (2F, 1M) investigator unavailable

43 EP SUBJECTS§ FOR SENSORY TESTING

PMA, postmenstrual age.*Children born in early 1996 who were recruited into the EPICure study, but notincluded in the initial cohort analysis (ceased December 31st 1995), but havecompleted subsequent follow-up.§Includes two children born in 1996 who had mild scars only (i.e. no proceduralscars or neonatal surgery).

containing parent and child information sheets, an invitation to at-tend, and a screening questionnaire were distributed. As fewercontrol than EP children agreed to attend for comparison, addi-tional control group participants were recruited from primaryschools local to UCL Institute of Child Health, London to completematching for sex, age and ethnic group.

The study was approved by the Great Ormond Street Hospitalfor Children NHS Trust and UCL Institute of Child Health ResearchEthics Committee. Written parental consent and child assent wereobtained prior to commencement of testing, and children and par-ents were informed that they could withdraw from the study atany time without consequences.

2.2. Sensory testing

Sensory testing was performed in a quiet, temperature-con-trolled room. The same investigator (SW) performed all the testingusing standardized verbal instructions and the same equipment forall subjects. As it was necessary to test scar sites related to neona-tal interventions, it was not possible for the investigator to beblinded to the treatment group. Prior to data acquisition, childrenwere shown the instruments and given a demonstration of the testprocedure to allay any anticipatory anxiety. A parent or guardianwas present during testing, and all efforts were made to minimisedistraction of the child. Children were not given auditory cues toindicate the start of the stimulus, and could not see the computerscreen used for thermal testing. No child withdrew from the proto-col or reported pain or distress during testing.

Thermal and mechanical testings were performed on the thenareminence of the non-dominant hand, and on skin of normalappearance adjacent to scars (at a distance of 5–10 mm). As thelargest standardized group of scars related to procedural or surgi-cal interventions were located on the thorax, additional thoracictesting (T6-8 dermatome) on the posterior axillary line was per-formed in a randomly selected convenience sample of EP childrenwithout thoracic scars (n = 7) and TC children (n = 16).

Skin temperature was measured in the webspace of the non-dominant hand to ensure skin temperature was >28 �C prior totesting. Thermal stimulation was performed with a handheld Sens-elab MSA Thermal Stimulator (Somedic, Sweden) with an18 � 18 mm (3.24 cm2) contact thermode to provide uniform con-tact on the rounded thenar eminence of children [23,24]. The ther-mode was set at a baseline temperature of 32 �C, with a rate ofchange of 1 �C/s and lower and upper limits of 10 and 50 �C,respectively. Using the method of limits, we sequentially deter-mined cool, warm, cold, and hot perception thresholds. Childrenwere familiarized with the baseline temperature of 32 �C duringthe demonstration phase, and during testing were asked to pressthe button when they first felt the temperature change to cool orwarm, and subsequently when the temperature felt cold (like anice-block) or hot (like a cup containing tea or chocolate that wastoo hot to drink). This allowed assessment of the response to differ-ent intensity stimuli, while avoiding the requirement for noxiousstimuli to assess pain thresholds. Participants were asked to pressa response button as soon as the specified sensation was perceived.Each stimulus was repeated five times, and the mean of the lastthree measures was calculated to determine the threshold. At theend of each test sequence, children were asked about paradoxicalresponses, which were classified as (i) cold felt as hot, (ii) hot feltas cold, (iii) rapid transition to hot, (iv) rapid transition to cold,or (v) others.

The mechanical detection threshold was determined usinghandheld von Frey hairs applied at slightly different sites withina small area to avoid habituation. Children kept their eyes closedand reported ‘‘yes” when they felt the filament touch the skin.Using an up–down method, the appearance and disappearance

S.M. Walker et al. / PAIN 141 (2009) 79–87 81

thresholds were sequentially determined, and the threshold wascalculated as the geometric mean of ten values. Threshold datawere analysed using von Frey hair number units which correspondto a logarithmic transformation [47]. Brush allodynia was tested byapplying strokes with a calibrated Senselab brush (200–300 mN).The sensation was scored between 0 and 10 (0 = neutral,10 = unpleasant/painful).

2.3. Pain experience and coping style

All children completed a brief questionnaire about their painexperience as previously described [47]. This included history ofrecurrent pain (frequency, site, treatment); the degree to whichpain interfered with activity (rated on a 0–10 numeric ratingscale); most remembered pain and its intensity (rated on a 0–10numeric rating scale); previous operations; questions regardingscars (aetiology, age, associated sensory changes, pain, allodynia)and a body map to mark the site of any scars and/or sensorychanges. Children also completed the Pain Coping Questionnaire(PCQ), which classifies typical behaviors and cognitions whenexperiencing pain in eight domains (information-seeking, problemsolving, seeking social support, positive self-statements, behavioraldistraction, cognitive distraction, externalizing, and catastrophiz-ing), which are then summarised into one of three higher ordercoping styles: approach, distraction and emotion-focused avoid-ance. The PCQ is validated in children over 8 years of age [40].

2.4. Cognitive ability

As EP children were participants in the EPICure follow-up stud-ies, data from the main EPICure database were accessed. Of 307survivors from the 1995 cohort, 219 were assessed in schools bya paediatrician and a psychologist. Compared with those assessed,children not seen at 11 years were more likely to be boys and tohave had severe disability at 2.5 years [12]. The Mental ProcessingComposite from the Kaufman Assessment Battery for Children (K-ABC) [29] was used as a measure of cognitive function. Disabilitywas categorised in motor, hearing, visual, communication and cog-nitive domains and was graded as previously described [33].Briefly, disability was considered to be severe if it was likely tomake the child highly dependent on caregivers, moderate if it al-lowed a reasonable degree of independence to be reached, andmild if there were neurological signs with minimal functional con-sequence or other impairments, such as squint or refractive error.

2.5. Statistical analysis

Continuous and ordinal data were analysed with unpaired two-tailed Student’s t-test or one way analysis of variance (ANOVA)with Tukey’s post hoc tests, as appropriate for the number ofgroups. Categorical data were analysed using non-parametric Pear-

Table 2Sample characteristics.

EP (n = 43) TC (n = 44)

Gestational age at birth, M (SD) 24.6 (0.67)Age (years), M ± SD 11.2 ± 0.4 11.1 ± 0.4Height (cm), M ± SD** 142.3 ± 6.2 146.3 ± 6.8Height (SDS)** �0.37 ± 0.9 0.34 ± 1.0Weight (kg), M ± SD* 36.4 ± 8.4 40.2 ± 9.4Weight (SDS)** �0.20 ± 1.2 0.40 ± 1.2Boys, n:Girls, n 15:28 16:28

EP, extremely preterm; TC, term control; M, mean; SD, standard deviation; SDS,standard deviation score.

* p < 0.05.** p < 0.01.

son’s chi-square or Fisher’s exact tests. Univariate comparisonsevaluated factors associated with changes in sensory thresholds.Levels of significance were reported at p < 0.05 and p < 0.01 (SPSS14). Multivariable analyses were performed to test for an interac-tion of sex and EP status using STATA 9.

3. Results

3.1. Demographic data

Demographic data are shown in Table 2. The 43 EP subjects forsensory testing did not differ in terms of sex, gestational age atbirth, current height or weight from either the 49 EP children pre-senting in London or the larger group assessed in school. Therewere no significant differences in the age or sex distribution be-tween the EP and TC sensory testing groups. EP children were sig-nificantly shorter and weighed less than the TC group.

3.2. Scars and sensory descriptors

All EP children had differing degrees of visible scarring relatedto neonatal treatment: 20 had mild scarring only on the dorsumof the hands related to intravenous cannulae; 11 had additionalmoderate scars related to procedural interventions (e.g. chest draininsertion, marked scarring at vascular access sites); and 12 hadscars related to neonatal surgery (Table 3). One EP child had visiblescars proximal to the thenar eminence of the dominant hand,presumably related to surgical insertion of an arterial line, but nochildren had significant scarring proximal to the thenar eminenceused for thermal testing. No TC children had surgery during theneonatal period. A higher proportion of EP children than TC chil-dren required surgery at different times during childhood, and inone EP case this was repeat surgery in the region of neonatalsurgery (closure of abdominal stoma). The variable sites and agesof the childhood scars precluded any comparison of sensorythresholds.

Although many EP children reported an altered response tobrush in the region of a scar, only one child (with abdominal scarsrelated to treatment of necrotizing enterocolitis) reported the sen-sation as unpleasant (VAS 6/10). One TC child reported pain andredness in the region of a scar related to recent inguinal surgery.A higher proportion of EP children than TC children reported para-doxical or unusual thermal responses. A rapid transition to ‘‘hot”rather than a graded change was reported on the thenar eminenceby 5 of 41 EP and 1 of 44 TC children, and adjacent to scars in 5 EPchildren and 1 TC child. Cold was paradoxically felt as hot on thethenar eminence in three cases (2 EP and 1 TC).

3.3. Thenar eminence sensory thresholds

In all children the pre-test skin temperature was at a level(28 �C and above) previously shown not to influence thresholdmeasurements [23]. All 44 TC children reliably completed sensorytesting. Of the 43 EP children, 2 did not complete thermal testingand were excluded from the analysis of thermal thresholds. Onechild with moderate cognitive impairment had variable responseswith no clear differentiation between cool/cold and warm/hot; andthe other child reported an inability to detect any thermal changesduring the test but not during the pre-test sequence. In a further 2EP children consistent measures for thermal testing but not formechanical testing were obtained (i.e. able to perceive von Freyhair stimulus during the demonstration phase but variable re-sponses during testing). In all included individuals, ‘‘hot” was de-tected at a higher temperature than ‘‘warm”, and ‘‘cold” at lowertemperatures than ‘‘cool”. In addition to evaluating the responseto different intensity stimuli, this provided an internal control

Table 3Scars and sensory descriptors.

Extremely preterm (n = 43) Term control (n = 44)

Scars Neonatal 43/43 20: minor 11: moderate procedural 12: surgery Childhood 8/44 5: laceration 1: burn/scald2: surgery

Neonatal surgery 12/43 6: thoracotomy (PDA) 1: abdominal (NEC)2: abdo+inguinal(BIH+pyloromyotomy) 3: inguinal (BIH)

0/44

Surgery afterneonatal period

14/43 1: ENT 1: minor orthopedic 1: suture laceration5: inguinal surgery 2: scar revision 1: VP shunt1: squint 2: abdominal stoma

8/44 2: ENT 2: minor orthopedic3: suture laceration 1: inguinalsurgery

Scar dysesthesia 11 (7 neonatal) 10: itching 1: tickling 4 2: itching 1: numbness (burn scar)1: tightness

Altered skinsensation

12/43 10: postural pins and needles 1: tingling1: ‘‘vibration” in chest

11/44 9: postural pins and needles1: tingling 1: itching (eczema)

Paradoxicalsensation (thenar)

6/41 1: cold as hot 5: rapid heat 2/44 1: cold as hot 1: rapid heat

Paradoxicalsensation (majorscar)

6/12 1: cold as hot 5: rapid heat 1/8 1: rapid heat

Brush allodynia 1 VAS 6/10 (abdominal scar) 0

PDA, patent ductus arteriosus; NEC, necrotizing enterocolitis; BIH, bilateral inguinal hernia; ENT, ear, nose and throat; VP shunt, ventriculoperitoneal shunt.

82 S.M. Walker et al. / PAIN 141 (2009) 79–87

demonstrating that children understood the verbal instructionsand could perceive the difference between modalities.

Fig. 1 shows thermal perception thresholds measured on thethenar eminence. EP children were significantly less sensitive thanTC children to all thermal modalities. The degree of difference wasgreater for the more intense stimuli of cold (�2.8 �C; 95%CI �4.8 to�0.8 �C; p = 0.005; SE = 01.0 �C) and hot (+2.5 �C; 95%CI +0.8 to+4.2 �C, p = 0.005, SE = 0.9 �C). In TC children, males were moresensitive than females to cold (28.9 ± 0.3 �C vs 26.2 ± 0.7 �C; 95%CI of difference in means 0.8 to 4.8, p = 0.008) and hot(35.6 ± 0.4 �C vs 38.2 ± 0.7 �C; 95%CI of difference in means �4.7to �0.5, p = 0.017), but thresholds for cool and warm did not differsignificantly. Thermal thresholds were not influenced by sex in theEP group (male and female values for cold: 24.5 ± 1.5 �C vs24.3 ± 1.0 �C and hot: 40.4 ± 1.4 �C vs 39.5 ± 0.8 �C). In the multi-variable analyses the interaction of sex and EP status was neversignificant.

Fig. 1. Thenar eminence thermal perception thresholds. The temperatures (�C) atwhich the sensations of cool, cold, warm and hot were perceived on the thenareminence of the non-dominant hand are plotted for full-term control (TC; n = 44)and extremely preterm (EP; n = 41) children. Extremely preterm children were lesssensitive to all thermal modalities. **p < 0.01 unpaired two-tailed Student’s t-test.

There was no significant difference in mechanical detectionthresholds on the thenar eminence in EP and TC groups (Fig. 3A),and no sex differences in mechanical thresholds in either group.

3.4. Generalized thermal sensitivity and neonatal surgery

Changes in thermal perception for cold and hot were most ex-treme in EP children who underwent surgery during the neonatalperiod (n = 12) (Fig. 2). Differences between EP children with sur-gery and TC children for cold (�4.8 �C) and hot (+4.1 �C) were sta-tistically significant after adjustment for the multiple comparisons(p < 0.01 for both). The subgroup of EP children with neonatal sur-gery did not differ from the remainder of the EP group in the de-gree or incidence of cognitive disability at 11 years of age (Table4). Surgery during later childhood did not have an effect on ther-mal thresholds in EP or TC children, although numbers were smalland the type of surgery was more variable.

3.5. Sensory thresholds adjacent to thoracic scars

In EP children the most standardized scars in a similar bodyarea were on the thorax. Localized decreases in sensitivity werefound adjacent to lateral thoracotomy scars, with differences fromTC for mechanical detection thresholds (Fig. 3B), and significantdifferences in cold and hot thresholds from TC and EP childrenwithout scars (Fig. 4). Due to the small size of these subgroups,thermal differences between TC and EP children without scarswere not detectable on the chest. Sensory thresholds adjacent tothoracic scars related to less invasive procedures (i.e. percutaneouschest drain insertion) did not differ from TC or EP children withoutscars.

3.6. Pain experience, interference and coping

There were few differences in recent pain experience or paincoping styles between the EP and TC children (Table 5). A similarproportion of EP and TC children reported pain-related interfer-ence with daily activity, but EP children reported a higher degreeof interference and more frequently used social support strategiesfor coping with pain than TC children. Overall, more girls reportedheadaches than boys (43% vs 19%, p < 0.05), but there were no dif-ferences in any other aspect of pain experience or in pain copingstyles related to sex. EP children who had surgery (p < 0.01) or amajor procedure (p < 0.05) in the neonatal intensive care unit

Fig. 2. Effect of neonatal surgery on thenar eminence thermal thresholds. Thetemperatures at which the sensations of cold and hot (�C) were perceived on thethenar eminence of the non-dominant hand are plotted for full-term control (TC;n = 44), extremely preterm children requiring neonatal intensive care but not majorsurgery (EP no surgery; n = 29), and extremely preterm children who alsounderwent surgical operations in the neonatal period (EP + surgery; n = 12).**p < 0.01 one way ANOVA with Tukey’s post hoc comparison.

S.M. Walker et al. / PAIN 141 (2009) 79–87 83

(NICU) reported more frequent use of information-seeking strate-gies compared to EP children who had not. EP children who hadsurgery during childhood more often used externalizing copingstrategies (p < 0.05) than EP children who did not have surgery inchildhood. TC children who had surgery during childhood moreoften used information-seeking (p < 0.05) and social support strat-egies (p < 0.01) than TC children who did not have childhood sur-gery. However, none of these factors had an influence on sensorythresholds.

4. Discussion

We have demonstrated persistent modality-specific changes insensory processing in children born extremely preterm. These chil-dren have a generalized decrease in thermal sensitivity that is

Fig. 3. Mechanical detection thresholds. (A) The mechanical detection threshold on tsignificantly in full-term control (TC; n = 44) and extremely preterm (EP; n = 39) groups.hair number, vFh no.) was significantly increased compared to that in control childrenprocedural scars (chest drain; n = 6) and in EP children without scars (EP; n = 7) did not d[vFh no. = log scale of increasing force: vFh no. 4 = 0.08 g; 6 = 0.22 g; 8 = 0.6 g; 10 = 1.6 g

more marked following neonatal surgery, suggesting that thedegree of tissue injury has an additional impact. Decreases inmechanical and thermal sensitivity adjacent to surgical scarsmay be related to localized tissue injury, whereas generalizeddecreases in thermal sensitivity but not in mechanical sensitivitysuggest centrally mediated alterations in the modulation ofC-fibre nociceptor pathways, which may influence future painresponses.

Thermal sensitivity on the thenar eminence was reduced in EPchildren compared with TC children, in whom values were consis-tent with the previous normative data [24,34,47]. Similarly, chil-dren born preterm (<31 weeks) or fullterm who required NICUcare showed generalized decreases in thermal sensitivity [21],and a trend to reduced sensitivity to warm and cool was noted ina pilot study of children following neonatal thoracotomy [47]. Itwas recently shown that heat pain thresholds in both ex-NICUand control children were higher when a parent was present (asin our study), possibly due to the child being more distracted orless anxious, but the difference in sensory threshold betweengroups was unaffected [26]. Noxious stimuli and pain thresholdswere avoided in the current study to comply with local ethics com-mittee requirements.

Generalized changes in thermal sensitivity but not in mechani-cal sensitivity were found in EP children. QST allows assessment ofsmall fibre function [8], with heat detection traditionally ascribedto unmyelinated C-fibres and cold detection to A-delta fibres[4,56]. This division may not be accurate given the complexity oftransient receptor potential (TRP) channels involved in thermosen-sation [5]. TRPA1 and TRPM8 contribute to cold sensation and arepresent on both A-delta and C-fibres [35]. The expression of TRPchannels changes during postnatal development [25], but the im-pact of early injury on receptor distribution and discrimination ofgraded stimuli has not been investigated. Assessment of mechani-cal perception thresholds with von Frey hairs tests A-beta function[2,8]. The absence of generalized changes in mechanical detectionthresholds in EP children, both currently and elsewhere [21], mayreflect modality-specific changes in modulation of A-delta and C-fi-bres but not of A-beta sensory input [15,41]. This is not inconsis-tent with the long-term decreases in thermal and mechanicalsensitivity following neonatal inflammation in rats [41], as hind-limb withdrawal from a noxious mechanical stimulus more likelyreflects A-delta and C fibre function rather than A-beta fibrefunction.

he thenar eminence (expressed as von Frey hair number, vFh no.) did not differ(B) The mechanical detection threshold on the lateral thorax (expressed as von Freyadjacent to scars from neonatal thoracotomy (PDA; n = 6). Thresholds adjacent toiffer from those in TC. **p < 0.01 one way ANOVA with Tukey’s post hoc comparison.; 12 = 4.5 g; 14 = 12.6 g].

Table 4Neonatal characteristics and outcomes of EP children involved in QST study and full EPICure cohort.

EP no surgery(QST subjects; n = 31)

EP + surgery (QSTsubjects; n = 12)

EPICure EP (cohort testedat 11 years; n = 219)

Neonatal measuresGestational age at birth, M ± SD 25.1 ± 0.61 24.6 ± 0.74 24.9 ± 0.73CRIB score, M ± SD 7.3 ± 2.5 9.5 ± 3.2 8.0 ± 3.5Neonatal surgery 0 Total: 12/43 (28%) Total: 77/200* (38%)

PDA 6/43 (14%) PDA 28/199* (14%)NEC 1/43 (2.3%) NEC 5/198* (2.5%)

Cerebral Ultrasound§

� No abnormality 17 (59%) 8 (67%) 145 (67%)� IVH with out ventriculomegaly 8 (28%) 4 (33%) 37 (17%)� Ventriculomegaly without parenchymal changes 0 (0%) 0 (0%) 5 (2%)� Parenchymal cysts (unilateral + /� ventriculomegaly) 3 (10%) 0 (0%) 22 (10%)� Parenchymal cysts (bilateral + /� ventriculomegaly) 1 (3%) 0 (0%) 9 (4%)

Outcomes at 11 year follow-upMPC (standardized), M ± SD 89 ± 13 88 ± 10 84 ± 18

Cognitive disability (%)� None 18 (58%) 8 (67%) 36 (16%)� Mild 12 (39%) 3 (25%) 84 (39%)� Moderate 1 (3%) 1 (8%) 63 (29%)� Severe 0 0 36 (16%)

M, mean; SD, standard deviation; CRIB, clinical risk index for babies; PDA, surgical ligation of patent ductus arteriosus; NEC, necrotizing enterocolitis requiring abdominalsurgery; §, last cerebral ultrasound scan performed before discharge from intensive care; *details of neonatal surgery were missing for 19–21 patients in the full cohort; IVH,intraventricular hemorrhage; MPC, Mental Processing Composite (from the Kaufman-Assessment Battery for Children).

84 S.M. Walker et al. / PAIN 141 (2009) 79–87

The current findings suggest that the degree of injury has an im-pact on the magnitude of sensory change, although confirmation inlarger subgroups is required. Generalized sensory changes weremore marked in EP children who also required neonatal surgery,and localized changes were found adjacent to surgical scars butnot to procedural scars. By contrast, no differences were reportedbetween preterm and term-born children despite a longer NICUstay and more procedural interventions in the preterm group,but very few children in either group had surgery [21]. Surgery

Fig. 4. Lateral thorax thermal perception thresholds. The temperatures (�C) atwhich the sensations of cold and hot were perceived on the lateral thorax areplotted for full-term control (TC; n = 16), extremely preterm children withoutthoracic scars (EP: no scar; n = 7), adjacent to chest drain scars (EP: chest drain;n = 6) and adjacent to lateral thoracotomy scars (EP: PDA; n = 6). Thermalsensitivity was significantly decreased adjacent to neonatal thoracotomy scars forpatent ductus arteriosus (PDA) ligation. *p < 0.05 one way ANOVA with Tukey’s posthoc comparison.

during initial hospitalisation in preterm neonates has been associ-ated with an increased risk of visual and auditory sensorineuraldeficits [10,53], and neurodevelopmental impairment is increasedby surgical rather than medical treatment of necrotizing enteroco-litis [39,48] or patent ductus arteriosus [28]. As our EP childrenwere recruited from mainstream schools, the incidence and sever-ity of cognitive deficit were lower than the full cohort and did notdiffer between those who underwent surgery and the remainder ofthe EP group. Therefore, the increased change in surgical EP chil-dren is independent of cognitive function, and is unlikely to reflectnon-specific differences due to poor understanding of the instruc-tions, reduced concentration, or delayed motor responses. Data onanalgesia were not available for the whole cohort, as children weretreated at centres across the UK. However, a survey of UK paediat-ric anaesthesiologists at that time reported routine perioperativeadministration of analgesia and/or local anaesthesia in newborns,although systemic opioids were used less in preterm neonates thanin term neonates [9].

While repeated heel prick in preterm infants produces local-ized hyperalgesia [14], which may persist throughout the firstyear [1] and be associated with generalized alterations in re-sponse to subsequent noxious stimuli [19], it is possible that thisinitial increased sensitivity resolves or is progressively masked atolder ages. In rodents, the adult pattern of generalized hypoalge-sia only emerges later in development [41]. The major clinicalimportance of baseline hypoalgesia may lie in a link with alteredresponses to future painful injuries. Neonatal inflammation inthe rat produces a dual pattern of generalized hypoalgesia, butan enhanced hyperalgesic response when repeat inflammation[41] or surgical incision [6] is performed in the previously in-jured paw. A similar pattern of baseline thermal hypoalgesia,but enhanced sensitization to a prolonged stimulus, was shownusing QST in ex-NICU children [21]. Increased perioperative anal-gesic requirements have been reported following repeat surgerymonths or years after neonatal surgery, particularly when per-formed in the same dermatome [37]. The first postnatal weekin the rat has been identified as a critical period for persistentsensory alterations following inflammation [41,43] and skinwounding [42]. Defining critical periods clinically is difficult,

Table 5Pain experience.

EP (n = 43) TC (n = 44)

How often do you have pain?� Never 13 (30%) 21 (48%)� Monthly or less 18 (42%) 13 (29%)� Daily or weekly 12 (28%) 10 (23%)

What kinds of pain do you get?� Headache, n (%) 19 (44%) 11 (25%)� Abdominal pain, n (%) 6 (14%) 7 (16%)� Limb pain, n (%) 4 (9%) 4 (9%)� Other, n (%) 1 (2%) 1 (2%)

Do you have pain that interferes with the things you would like to do? Yes; n (%) 10 (23%) 13 (27%)If yes, how much does it interfere (0 = not at all; 10 = worst possible)? Median (IQR)* 6.5 (5–8) 4 (3–5)M (SD)* 6.1 (2.1) 4.0 (1.8)Think of the pain you most remember in your life; How painful

was it? (0 = not at all; 10 = worst possible) Median (IQR)5 (0–8) 6 (0–7)

M (SD) 4.4 (4.2) 4.3 (3.9)

Pain coping, M(SD)� Approach 3.07 (0.78) 2.88 (0.56)

s Information-seeking 2.59 (1.01 2.42 (0.84)s Problem solving 3.18 (1.05) 3.09 (0.76)s Positive self-statements 3.03 (1.05) 3.04 (0.84)s Seeking social support* 3.37 (0.96) 2.89 (1.00)� Distraction 3.26 (0.97) 3.31 (0.82)

s Behavioral distraction 3.23 (1.06) 3.06 (0.99)s Cognitive distraction 3.28 (1.15) 3.56 (0.91)� Avoidance 2.00 (0.61) 1.84 (0.61)

s Catastrophizing 2.44 (0.76) 2.19 (0.69)s Externalizing 1.56 (0.77) 1.50 (0.57)

IQR, interquartile range; M, mean; SD, standard deviation.* p < 0.05.

S.M. Walker et al. / PAIN 141 (2009) 79–87 85

but important [15]. As similar changes were found in childrenfollowing NICU care and preterm (<31 weeks) or full-term birth[21] and in the current extremely preterm cohort, it appears thatthe ‘‘critical period” for initiating prolonged thermal changesextends at least throughout the preterm and early neonatalperiods.

Fewer boys were recruited as mortality and morbidity areincreased in extremely preterm boys [7,32]. No sex differenceswere found in sensory thresholds in the EP children, but sensitiv-ity to cold and hot was greater in boys than in girls in the TCgroup. The study had low power for tests of interaction, so thelack of a significant interaction between sex and EP status leavesone unable to conclude whether the effect of sex differs in extre-mely preterm and term children. Sex differences have not beenreported previously for cold and warm [23,54] or heat painthresholds [58]; but not all studies separate results for boys andgirls [21,34].

The thermal hyposensitivity in EP children was not accompa-nied by marked differences in everyday pain report. Similarly,the frequency or level of current pain did not account for the en-hanced sensitization to a prolonged stimulus in children followingNICU care [21,26]. Epidemiological surveys have shown that recur-rent pains are common in children [36,55]. We found a similarprevalence, with no difference between EP and TC children,although the study was not powered for this outcome. Interferencein daily activities due to pain was reported by similar proportionsof EP and TC children, and is consistent with the reported fre-quency of functional impairment due to pain in other cohorts ofpreterm-born adolescents and adults [17,45,46]. Subtle differencesin everyday pain sensations and discomfort at neonatal thoracot-omy scars have been reported [47]. We found minor dysesthesiasin approximately one-third of the neonatal scars, one child re-ported brush allodynia in relation to abdominal scars, and onechild reported chronic chest pain in the region of a neonatal PDArepair. Further investigation of the persistence of pain or abnormalsensations after injury is needed, and larger studies will be re-

quired to evaluate the impact of neonatal experience on chronicpain in later life.

Although pain coping style has been shown to influence exper-imental pain tolerance and intensity in children [30], we found norelationship between sensory perception thresholds and paincoping styles in EP or TC children. EP children reported a higherdegree of pain interference and were more likely to seek socialsupport (i.e., verbalising pain to others). Seeking social supportis associated with greater pain catastrophizing and elicits moreprotective behaviors from parents in both healthy children andthose with chronic pain [57], and increased catastrophizing hasrecently been reported in ex-NICU children [26]. Parenting styleand more solicitous responses to pain by parents of pretermchildren [26], altered child behavior (e.g. increased risk-avoidanceand lower resilience) [20], and other social factors throughoutchildhood may influence behavioral pain responses, and shouldbe differentiated from alterations in sensory processing. Specificevaluation of the sensory responses of preterm children to clinicalpain or acute injury (e.g. pain intensity, degree of hyperalgesia) iswarranted to evaluate the possible link between baseline hypoal-gesia and the enhanced response to future injury found in animalstudies.

With ongoing reductions in mortality in preterm neonates, thefocus is shifting to long-term morbidity [44] and how this can bemodified by improved neonatal care. Higher prevalence, lowerseverity neurological dysfunction is increasingly recognised, par-ticularly after school age [11,27]. Previous studies have focusedon neurodevelopment and motor function [32], but our current re-sults suggest that changes in sensory function can also be detectedmany years following extreme preterm birth. Implications of thesedata are twofold: quantification of alterations in sensory process-ing may alert health care providers to patients with altered re-sponses to future injury; and identification of factors associatedwith long-term changes in sensory function may highlight inter-ventions during neonatal intensive care and surgery that will mod-ulate persistent effects.

86 S.M. Walker et al. / PAIN 141 (2009) 79–87

Conflicts of interest

The authors declare that they have no conflicts of interest,including specific financial interests, relationships or affiliationsrelevant to the manuscript.

Acknowledgements

The study was undertaken on behalf of the EPICure StudyGroup (http://www.epicure.ac.uk) and the authors gratefullyacknowledge assistance from Enid Hennessy for statistical advice,and Samantha Johnson for data from the full EPICure database.We wish to thank Sooky Lum, Jane Kirkby and Liam Welsh fromthe Portex Unit at UCL Institute of Child Health for their invaluableassistance with recruitment. We are also grateful for the time andenthusiasm of our subjects and their families, which made thisstudy possible. The study was funded by the UK Medical ResearchCouncil (MRC), Grant No. 72524.

References

[1] Abdulkader HM, Freer Y, Garry EM, Fleetwood-Walker SM, McIntosh N.Prematurity and neonatal noxious events exert lasting effects on infant painbehaviour. Early Hum Dev 2008;84:351–5.

[2] Bell-Krotoski J, Tomancik E. The repeatability of testing with Semmes-Weinstein monofilaments. J Hand Surg [Am] 1987;12:155–61.

[3] Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive andbehavioral outcomes of school-aged children who were born preterm: a meta-analysis. JAMA 2002;288:728–37.

[4] Campero M, Serra J, Ochoa JL. C-polymodal nociceptors activated by noxiouslow temperature in human skin. J Physiol 1996;497:565–72.

[5] Caterina MJ. Transient receptor potential ion channels as participants inthermosensation and thermoregulation. Am J Physiol Regul Integr CompPhysiol 2007;292:R64–76.

[6] Chu YC, Chan KH, Tsou MY, Lin SM, Hsieh YC, Tao YX. Mechanical painhypersensitivity after incisional surgery is enhanced in rats subjected toneonatal peripheral inflammation: effects of N-methyl-D-aspartate receptorantagonists. Anesthesiology 2007;106:1204–12.

[7] Costeloe K, Hennessy E, Gibson AT, Marlow N, Wilkinson AR. The EPICurestudy: outcomes to discharge from hospital for infants born at the threshold ofviability. Pediatrics 2000;106:659–71.

[8] Cruccu G, Anand P, Attal N, Garcia-Larrea L, Haanpaa M, Jorum E, et al.EFNS guidelines on neuropathic pain assessment. Eur J Neurol 2004;11:153–62.

[9] De Lima J, Lloyd-Thomas AR, Howard RF, Sumner E, Quinn TM. Infant andneonatal pain: anaesthetists’ perceptions and prescribing patterns. Br Med J1996;313:787.

[10] Doyle LW. Outcome at 5 years of age of children 23 to 27 weeks’ gestation:refining the prognosis. Pediatrics 2001;108:134–41.

[11] Fawke J. Neurological outcomes following preterm birth. Semin Fetal NeonatalMed 2007;12:374–82.

[12] Fawke J, Johnson S, Rowell V, Trikic R, Smith R, Palmer H, et al. The EPICurestudy: neurosensory disability at 11 years. Arch Dis Child 2008;93:A33.

[13] Fitzgerald M. The development of nociceptive circuits. Nat Rev Neurosci2005;6:507–20.

[14] Fitzgerald M, Millard C, MacIntosh N. Hyperalgesia in premature infants.Lancet 1988;1:292.

[15] Fitzgerald M, Walker S. Infant pain traces. Pain 2006;125:204–5.[16] Fitzgerald M, Walker S. The role of activity in developing pain pathways. In:

Dostrovsky J, Carr D, Koltzenburg M, editors. Proceedings of the 10th WorldCongress on Pain. Progress in Pain Research and Management, Vol.24. Seattle: IASP Press; 2003. p. 185–96.

[17] Gray R, Petrou S, Hockley C, Gardner F. Self-reported health status and health-related quality of life of teenagers who were born before 29 weeks’ gestationalage. Pediatrics 2007;120:e86–93.

[18] Grunau RE, Holsti L, Peters JW. Long-term consequences of pain in humanneonates. Semin Fetal Neonatal Med 2006;11:268–75.

[19] Grunau RE, Oberlander T, Whitfield MF, Fitzgerald C, Morison SJ, Saul JP. Painreactivity in former extremely low birth weight infants at corrected age 8months compared with term born controls. Infant Behav Dev 2001;24:41–55.

[20] Hack M, Cartar L, Schluchter M, Klein N, Forrest CB. Self-perceived health,functioning and well-being of very low birth weight infants at age 20 years. JPediatr 2007;151:635–41. 41 e1-2.

[21] Hermann C, Hohmeister J, Demirakca S, Zohsel K, Flor H. Long-term alterationof pain sensitivity in school-aged children with early pain experiences. Pain2006;125:278–85.

[22] Hilz MJ, Axelrod FB, Hermann K, Haertl U, Duetsch M, Neundorfer B. Normativevalues of vibratory perception in 530 children, juveniles and adults aged 3–79years. J Neurol Sci 1998;159:219–25.

[23] Hilz MJ, Glorius SE, Schweibold G, Neuner I, Stemper B, Axelrod FB.Quantitative thermal perception testing in preschool children. Muscle Nerve1996;19:381–3.

[24] Hilz MJ, Stemper B, Schweibold G, Neuner I, Grahmann F, Kolodny EH.Quantitative thermal perception testing in 225 children and juveniles. J ClinNeurophysiol 1998;15:529–34.

[25] Hjerling-Leffler J, Alqatari M, Ernfors P, Koltzenburg M. Emergence offunctional sensory subtypes as defined by transient receptor potentialchannel expression. J Neurosci 2007;27:2435–43.

[26] Hohmeister J, Demirakca S, Zohsel K, Flor H, Hermann C. Responses topain in school-aged children with experience in a neonatal intensive careunit: Cognitive aspects and maternal influences. Eur J Pain 2009;13:94–101.

[27] Johnson S. Cognitive and behavioral outcomes following very preterm birth.Semin Fetal Neonatal Med 2007;12:363–73.

[28] Kabra NS, Schmidt B, Roberts RS, Doyle LW, Papile L, Fanaroff A. Neurosensoryimpairment after surgical closure of patent ductus arteriosus in extremely lowbirth weight infants: results from the Trial of Indomethacin Prophylaxis inPreterms. J Pediatr 2007;150:229–34. 34 e1.

[29] Kaufman A, Kaufman N. Kaufman Assessment Battery for Children (K-ABC). MI: American Guidance Service: Circle Pines; 1983.

[30] Lu Q, Tsao JC, Myers CD, Kim SC, Zeltzer LK. Coping predictors of children’slaboratory-induced pain tolerance, intensity, and unpleasantness. J Pain2007;8:708–17.

[31] Lum S, Kirkby J, Welsh L, Marlow N, Stocks J. The EPICure study: extremelypremature birth and implications for respiratory health at age 11. Am J RespirCrit Care Med 2008;177:A558.

[32] Marlow N, Hennessy EM, Bracewell MA, Wolke D. Motor and executivefunction at 6 years of age after extremely preterm birth. Pediatrics2007;120:793–804.

[33] Marlow N, Wolke D, Bracewell MA, Samara M. Neurologic and developmentaldisability at six years of age after extremely preterm birth. N Engl J Med2005;352:9–19.

[34] Meier PM, Berde CB, DiCanzio J, Zurakowski D, Sethna NF. Quantitativeassessment of cutaneous thermal and vibration sensation and thermal paindetection thresholds in healthy children and adolescents. Muscle Nerve2001;24:1339–45.

[35] Munns C, AlQatari M, Koltzenburg M. Many cold sensitive peripheralneurons of the mouse do not express TRPM8 or TRPA1. Cell Calcium 2007;41:331–42.

[36] Perquin CW, Hazebroek-Kampschreur AA, Hunfeld JA, Bohnen AM, vanSuijlekom-Smit LW, Passchier J, et al. Pain in children and adolescents: acommon experience. Pain 2000;87:51–8.

[37] Peters JW, Schouw R, Anand KJ, van Dijk M, Duivenvoorden HJ, Tibboel D. Doesneonatal surgery lead to increased pain sensitivity in later childhood? Pain2005;114:444–54.

[38] Porter FL, Grunau RE, Anand KJ. Long-term effects of pain in infants. J DevBehav Pediatr 1999;20:253–61.

[39] Rees CM, Pierro A, Eaton S. Neurodevelopmental outcomes of neonates withmedically and surgically treated necrotizing enterocolitis. Arch Dis Child FetalNeonatal Ed 2007;92:F193–8.

[40] Reid GJ, Gilbert CA, McGrath PJ. The Pain Coping Questionnaire: preliminaryvalidation. Pain 1998;76:83–96.

[41] Ren K, Anseloni V, Zou SP, Wade EB, Novikova SI, Ennis M, et al.Characterization of basal and re-inflammation-associated long-termalteration in pain responsivity following short-lasting neonatal localinflammatory insult. Pain 2004;110:588–96.

[42] Reynolds ML, Fitzgerald M. Long-term sensory hyperinnervation followingneonatal skin wounds. J Comp Neurol 1995;358:487–98.

[43] Ruda MA, Ling QD, Hohmann AG, Peng YB, Tachibana T. Altered nociceptiveneuronal circuits after neonatal peripheral inflammation. Science2000;289:628–31.

[44] Saigal S, Doyle LW. An overview of mortality and sequelae of preterm birthfrom infancy to adulthood. Lancet 2008;371:261–9.

[45] Saigal S, Rosenbaum PL, Feeny D, Burrows E, Furlong W, Stoskopf BL, et al.Parental perspectives of the health status and health-related quality of life ofteen-aged children who were extremely low birth weight and term controls.Pediatrics 2000;105:569–74.

[46] Saigal S, Stoskopf B, Pinelli J, Streiner D, Hoult L, Paneth N, et al. Self-perceivedhealth-related quality of life of former extremely low birth weight infants atyoung adulthood. Pediatrics 2006;118:1140–8.

[47] Schmelzle-Lubiecki BM, Campbell KA, Howard RH, Franck L, Fitzgerald M.Long-term consequences of early infant injury and trauma uponsomatosensory processing. Eur J Pain 2007;11:799–809.

[48] Schulzke SM, Deshpande GC, Patole SK. Neurodevelopmental outcomes of verylow-birth-weight infants with necrotizing enterocolitis: a systematic review ofobservational studies. Arch Pediatr Adolesc Med 2007;161:583–90.

[49] Simons SH, van Dijk M, Anand KS, Roofthooft D, van Lingen RA, Tibboel D. Dowe still hurt newborn babies? A prospective study of procedural pain andanalgesia in neonates. Arch Pediatr Adolesc Med 2003;157:1058–64.

[50] Slater R, Cantarella A, Franck L, Meek J, Fitzgerald M. How well do clinical painassessment tools reflect pain in infants? PLoS Med 2008;5:e129.

[51] Slater R, Cantarella A, Gallella S, Worley A, Boyd S, Meek J, et al. Cortical painresponses in human infants. J Neurosci 2006;26:3662–6.

[52] Taddio A, Katz J, Ilersich AL, Koren G. Effect of neonatal circumcision on painresponse during subsequent routine vaccination. Lancet 1997;349:599–603.

S.M. Walker et al. / PAIN 141 (2009) 79–87 87

[53] The Victorian Infant Collaborative Study. Surgery and the tiny baby:sensorineural outcome at 5 years of age. J Paediatr Child Health1996;32:167–72.

[54] Thibault A, Forget R, Lambert J. Evaluation of cutaneous and proprioceptivesensation in children: a reliability study. Dev Med Child Neurol1994;36:796–812.

[55] van Dijk A, McGrath PA, Pickett W, VanDenKerkhof EG. Pain prevalencein nine- to 13-year-old schoolchildren. Pain Res Manag 2006;11:234–40.

[56] Verdugo R, Ochoa JL. Quantitative somatosensory thermotest. A key methodfor functional evaluation of small calibre afferent channels. Brain1992;115:893–913.

[57] Vervoort T, Craig KD, Goubert L, Dehoorne J, Joos R, Matthys D, et al. Expressivedimensions of pain catastrophizing: a comparative analysis of school childrenand children with clinical pain. Pain 2008;134:59–68.

[58] Zohsel K, Hohmeister J, Oelkers-Ax R, Flor H, Hermann C. Quantitative sensorytesting in children with migraine: preliminary evidence for enhancedsensitivity to painful stimuli especially in girls. Pain 2006;123:10–8.