ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /ea r l humdev

A controlled trial of skin-to-skin contact in extremelypreterm infants

Rachel Miles a, Frances Cowan a, Vivette Glover a,Jim Stevenson b, Neena Modi a,*

a Faculty of Medicine, Imperial College London, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, UKb Department of Psychology, University of Southampton, UK

Accepted 24 November 2005

0378-3782/$ - see front matter D 200doi:10.1016/j.earlhumdev.2005.11.008

* Corresponding author.E-mail address: n.modi@imperial.a

KEYWORDSPreterm infant;Outcome;Skin-to-skin;Cortisol;Behaviour;Maternal mentalstate

Abstract

Background: Extremely preterm birth, even in the absence of significant neurologicalimpairment, is associated with altered pain responses and impaired memory and behaviour.Preterm birth increases the risk of maternal depression and may impede the development of themother—infant relationship, factors that in turn are also associated with impaired infantoutcome. Mother—infant skin-to-skin contact has been recommended as a simple means ofameliorating these effects.Methods: We conducted a pragmatic, prospective, controlled, intention-to-treat trial in twoneonatal intensive care units. Infants born below 32 weeks gestation were recruited within thefirst week after birth and assigned to a control group receiving standard care, or an interventiongroup in whichmothers were encouraged to provide a session of skin-to-skin contact once daily for4 weeks. We assessed infant behaviour at time of discharge from hospital, responses to immu-nisation at 4 and 12 months of age, and memory, behaviour and development at 1 year corrected(postmenstrual) age. Indices of maternal depression, stress, anxiety, lactation performance andinfant interaction were assessed at time of infant discharge, 4 months and 1 year.Results: No significant difference was identified in any infant or maternal measure at any timepoint.Conclusions: Mother—infant skin-to-skin contact after extremely preterm birth results in neitherbenefit nor adverse consequences. Although there is no reason to dissuade mothers who wish toprovide STS contact, we are unable to recommend resource allocation for the implementation ofSTS programmes for extremely preterm infants in a neonatal intensive care unit setting.D 2005 Elsevier Ireland Ltd. All rights reserved.

5 Elsevier Ireland Ltd. All rights re

c.uk (N. Modi).

1. Introduction

Infants born below 32 weeks gestation have significantmaturational delay in brain development at term [1] and areat risk of altered pain responses [2] and impaired cognitive

Early Human Development (2006) 82, 447—455

served.

R. Miles et al.448

[3—5] and behavioural development [6,7] in the absence ofovert brain injury. There are several possible mediators ofthis altered trajectory of brain development.

Primate and other animal research demonstrate that theactivity of the hypothalamic—pituitary—adrenal (HPA) axiscan be permanently up-regulated by excessive glucocorti-coid (GC) exposure in the perinatal period [8]. Glucocorti-coids are neurotoxic and chronic endogenous andtherapeutic GC exposure is a causal factor for adversememory and anxiety related outcomes in many humandiseases [9—11]. There is mounting evidence that exposureto therapeutic GC during early development adverselyaffects the immature brain [12]. Neonates in intensive careare exposed to a dysregulated environmental milieu,repeated invasive procedures and prolonged illness andmay be at risk of chronic endogenous cortisol exposure [13].

HPA axis reactivity can be attenuated by maternal care[14] and sensitive maternal behaviour buffers infant cortisolresponses [15]. However preterm birth often results inmaternal psychological distress including postnatal depres-sion, with important implications for infant outcome.Postnatal depression is associated with less optimal infantbehaviour and lower cognitive ability [16,17] possiblythrough a reduction in maternal sensitivity [18] andincreased risk of insecure/disorganised attachment, factorsthat are predictive of impaired infant social and emotionaldevelopment [19—21].

Mother—infant skin-to-skin (STS) contact is held to bringabout enhanced maternal psychological well-being andmother—child relationships [22—24] and improved lactation[25—27], which in turn is associated with improved devel-opmental outcomes [28]. We have shown that a singlesession of STS significantly reduces infant salivary cortisol[29] suggesting that repeated sessions might bring about asustained reduction. We therefore aimed to address thehypothesis that mother—infant STS contact improves pre-term infant behavioural and developmental outcomes and toexplore potential mediating pathways, namely maternalpsychological state, lactation and infant endogenous cortisolproduction.

2. Methods

This was a pragmatic, controlled trial. Participant alloca-tion was by a crossover, cluster recruitment designbetween two tertiary referral neonatal intensive careunits, serving a multicultural urban population (Hammer-smith Hospital and Queen Charlotte’s and Chelsea Hospi-tal). The study was approved by the Research EthicsCommittee of Imperial College Faculty of Medicine andHammersmith Hospitals Trust. Written informed, parentalconsent was obtained.

Mothers were approached as soon as practicable afterdelivery in order to explain the study. Infants were eligible ifborn below 32 weeks gestation, less than 7 days of age andwithout major immediate life-threatening congenital ab-normalities. Allocation was to either standard nursing careor to the intervention group. Initial selection of hospital toprovide intervention or control was by randomisation. Eachhospital remained in this group for 4 months and thencrossed over following a wash-out phase, during which no

recruitment was undertaken. The wash-out phase ensuredthat infants from the previous period of recruitment hadbeen discharged before crossing over to the next phase.Medical and nursing protocols for the two neonatal units areidentical.

2.1. Skin-to-skin intervention

If the infants were not being ventilated and considered bythe staff to be medically stable (not having repeatedepisodes of apnoea or bradycardia, not on inotropic supportand with a stable core-peripheral temperature gap of lessthan 2 8C), mothers were invited to provide STS contact.Nasal CPAP (continuous positive airway pressure), or lowflow oxygen delivery via nasal cannula were not consideredcontra-indications. There was no time of day criteria for thedelivery of the intervention in order to accommodate thedifferent times at which mothers visited.

Mothers were seated on a chair at a 608 angle, wearing afront opening blouse and no strong perfume. The infant wasplaced naked except for a nappy and a woollen hat, in avertical position directly on to the skin between themother’s breasts and covered with a light blanket. The backand buttocks were supported by the mother’s hands, thelower limbs were tucked up in flexion and head and neckpositioned so as not to obstruct the airway. Continuous pulseoximetry and heart and respiratory rate monitoring contin-ued during STS contact. As this was a pragmatic trial,designed to test the intervention in a standard neonatalintensive care setting, responsibility for the care of theinfant and supervision of the mother remained with thenurse providing clinical care. The intention was to provide20 min of STS once daily for 4 weeks. Unit policy did notrequire nursing staff to provide care for the same infant on alongitudinal basis and this would have served to minimiseany possibility of bias towards either STS or controlmanagement.

Mothers in both intervention and control groups wereprovided with comparable levels of informal support. Duringthe intervention phase this was provided by the researchnurse, following discharge during monthly telephone callsmade by the research co-ordinator and at follow upassessments, by the research psychologist.

Data collection was undertaken at recruitment, inhospital, at discharge, at 4 and 12 months post birth andat 1 year from term. All questionnaire data were coded andchecked in batches by two assessors.

2.2. Infant measures

Data collected at trial entry were gestational age, birthweight, gender and prenatal and postnatal steroid expo-sure. Infant level of care, coded according to a UKstandard [30] was documented daily throughout the in-patient stay. Infant plasma cortisol concentrations duringweeks 1—4 post-birth were measured in once weekly 0.5ml blood samples obtained during venepuncture forclinical indications and residual plasma after clinicalsampling.

Infant immunisations were performed at 4 months and1 year in accordance with the national UK schedule.

Skin-to-skin contact 449

Behavioural (Modified Behavioural Pain Scale [31] andmeasures of vocal distress latency and duration) andsalivary cortisol responses [32,33] to single intramuscularinjection of DTP-Hib vaccine were assessed. Maternal andinfant responses were video-recorded and analysed retro-spectively by an assessor blind to infant group. Immunisa-tions were administered at standard times (between 12pm and 4 pm at 4 months; between 10 am and 12 pm at12 months) in order to control for possible diurnalvariation. Saliva samples were obtained immediatelybefore and 20 min after injection. Fruit juices, sweetdrinks and milk feeds were avoided before immunisation.The infant behavioural reactivity had an overall inter-rater reliability level of 0.95. Test—retest reliability,assessed by re-rating video data after 12 months was 0.95.

Eligible infants admitted to N

Randomised n =

Allocated to control n = 32

Assessed at discharge n = 26

A4

Pre-discharge Died = 5

Withdrew = 1

Available for assessment at 4 months n = 25 Assessed at 4 months n = 25

As

Assessed at 12 months from birth n = 24

m

m

Pre-4 month assessment

Defaulted = 1

Pre-12 month assessment

Defaulted = 1

Assessed at 12 months from term

n = 22

Pre-12 month from term assessment

Defaulted = 2

Figure 1 Cons

At 1 year corrected age infants underwent astandardised neurological examination (the HammersmithInfant Neurological Examination [34]) a test of visualperception and recognition memory (the Fagan Test ofInfant Intelligence [35]), a developmental assessment(the Revised Griffiths Mental Development Scales [36])and two maternal report measures, the Infant ToddlerSocial and Emotional Assessment (ITSEA) [37] measuringacquisition of mental age appropriate social—emotionalskills and the child domain of the Parenting StressIndex (PSI) [38]. The standardised neurological exami-nation and Griffiths developmental assessment at 1 yearwas carried out by a consultant/senior lecturer inperinatal neurology (FC), who was unaware of studyallocation.

NUs n = 145

Excluded Refused = 13 Early deaths = 10 Early transfer = 3 Researcher not available = 30 Other (Mother too unwell; poor English language) = 11

78

Allocated to intervention n = 46

ssessed at discharge n = 3

Pre-discharge Died = 2

Withdrew = 1

Available for assessment at 4 months n = 42

sessed at 4 months n = 37

Available for assessment at 12 onths from birth n

= 40. Assessed at 12

onths from birth n = 37

Pre-4 month assessment

Defaulted = 5 Withdrew = 1

Pre-12 month from birth assessment Defaulted = 3 Withdrew = 2

Assessed at 12 months from term

n = 32

Pre-12 month from term assessment Defaulted = 5

ort diagram.

R. Miles et al.450

2.3. Maternal measures

Baseline variables were past obstetric history and demo-graphic and social support indices. Psychological well beingwas assessed by history of previous anxiety or depressionand by identification at recruitment of depressive symptomson the Edinburgh Postnatal Depression Scale (EPDS) [39] andanxiety levels on the State-Trait Anxiety Inventory for Adults(STAI) [40].

Measures administered at infant discharge were theParental Stressor Scale: Neonatal Intensive Care [41] whichmeasures maternal stress relating to infant NICU admission,the perception of own care giving confidence using theMother and Baby Scale (MABS) [42], the EPDS and the STAI;at 4 months, the EPDS, STAI and a score of maternal soothingof infant immunisation distress based upon a modified

Table 1 Mother and infant baseline characteristics

Gestational age at birth (nearest week)Birth weight (g)Gender (number female)Infant intensive care daysInfant high dependency care daysInfant special care daysInfant surfactant administration (number of doses)Infant postnatal steroid administration (number of doses)Maternal age (years)Maternal full time education (years)Number with rupture of membranes N24 h before deliveryMaternal antenatal steroid administration (doses)Number of previous live birthsNumber of previous abortions, stillbirths and neonatal deathsPrevious psychological morbidity (%)Index of Perceived Social Support (%)1=never234=sometimes567=always

Cohabitation status (%)Living aloneLiving with partnerLiving with friends or familyOther

Relationship length (%)b1 year1—2 years3—5 yearsN5 years

Relationship quality (%)CloseFew tensions and disagreementsSome frictionConstant friction/breakdown

Maternal EPDS at trial entryMaternal STAI at trial entry

version of Lewis and Ramsay’s index [43]; and at 1 yearcorrected age, maternal soothing of immunisation distress,mental health as measured by the General Health Question-naire (GHQ-28) [44], the parent domain of the PSI (whichidentifies dysfunctional parental behaviour and stressors inthe parenting relationship) and maternal attachment totheir infant as assessed by the Parent to Infant Attachmentquestionnaire [45].

2.4. Lactation and breast milk intake

We derived a lactation performance index based on thenumber of inpatient days the infant received maternalbreast-milk. Following discharge we recorded ongoingbreast-feeding at monthly intervals up to 1 year. Breast-milk intake was rated on a 5-point ordinal scale as, 1=fully

Control (mean (S.D.)) STS (mean (S.D.)) P

28 (2.3) 28 (2.1) ns1133 (367) 1086 (402) ns

11 19 ns14.8 (16.5) 21.3 (22.8) ns5.9 (9.4) 4.2 (5.1) ns

14.1 (19.4) 14.0 (15.2) ns1.2 (0.6) 1.4 (0.7) ns3.8 (10.3) 2.3 (6.4) ns

30.6 (6.6) 30.3 (6.2) ns12.8 (3.5) 12.7 (2.3) ns

9 19 ns2.6 (2.6) 2.3 (1.5) ns0.7 (0.9) 0.8 (1.0) ns0.8 (1.0) 1.1 (1.4) ns63 50 ns

ns0 20 50 5

12 1612 919 1458 50

ns17 776 783 133 2

ns0 2

20 1224 2656 59

ns46 6242 314 04 7

11.6 (6.9) 12.4 (4.9) ns40.7 (18.2) 42.0 (12.5) ns

Table 2 Infant and maternal scores

Time of assessment Composition of aggregatedscores

Description of scale STS(mean (S.D.))

Control(mean (S.D.))

p

(a) Infant scores, discharge to 4 monthsInfant discharge home Mother and Baby Scale [42] Neonatal Behavioural

AssessmentAlert-responsive 27.68 (5.80) 32.55 (8.10) nsUnsettled-irregular 34.88 (13.82) 27.64 (12.46) nsEasiness 18.76 (3.62) 20.00 (2.05) nsAlert during feeds 13.04 (5.05) 14.55 (4.32) nsIrritable during feeds 12.64 (7.36) 12.45 (7.29) ns

4 months post-birth HPA axis response toimmunisation stress [32,33]

Salivary cortisol response(Dnmol/l)

3.25 (6.36) 5.22 (3.20) ns

4 months post-birth The Modified BehaviouralPain Scale [31]

Motor, vocal and facialresponse to immunisation

5.40 (1.16) 5.68 (1.20) ns

4 months post-birth Vocal Distress LatencyResponse to immunisation

Time delay in seconds fromskin puncture to response

1.33 (1.24) 2.16 (1.50) ns

4 months post-birth Infant Distress DurationResponse to Skin Puncture

Time duration in secondsfrom onset of distressto achieving calmbehavioural state

100.60 (64.68) 138.70 (113.10) ns

(b) Infant scores, 12 months post-birth and 1-year post-term12 months post-birth HPA axis response to

immunisation stress [32,33]Salivary cortisol response(Dnmol/l)

�0.28 (5.78) �0.03 (5.43) ns

12 months post-birth The Modified BehaviouralPain Scale [31]

Motor, vocal and facialresponse to immunisation

5.13 (1.61) 5.67 (1.01) ns

12 months post-birth Vocal Distress LatencyResponse to immunisation

Time delay in seconds fromskin puncture to response

3.03 (1.56) 2.94 (1.61) ns

12 months post-birth Infant Distress DurationResponse to Skin Puncture

Time duration in secondsfrom onset of distressto achieving calmbehavioural state

70.50 (58.49) 78.44 (66.99) ns

1 year post-term The Fagan Test of InfantIntelligence [35]

Visual Recognition Memory 56.86 (8.01) 55.81 (5.32) ns

1 year post-term The Griffiths MentalDevelopment Scale [36]

Locomotor 91.54 (21.47) 100.26 (16.33) ns

Personal—Social 96.46 (15.54) 87.84 (17.83)Hearing and Speech 99.32 (13.39) 96.42 (16.52)Eye—hand coordination 103.11 (20.03) 93.47 (17.38)Performance 98.36 (14.51) 99.47 (19.99)

1 year post-term Infant Toddler Social andEmotional Scale [37]

Externalising 0.39 (0.20) 0.43 (0.41) ns

Internalising 0.30 (0.11) 0.50 (0.56)Deregulation 0.48 (0.18) 0.57 (0.26)Competence 1.00 (0.26) 0.99 (0.32)

1 year post-term Parenting Stress Index [38] Child domain score ofbehavioural development

89.70 (13.17) 93.31 (16.68) ns

(c) Maternal scores, discharge to 4 monthsInfant discharge home Edinburgh Postnatal

Depression Scale [39]Presence of symptoms ofpostnatal depression

7.69 (5.31) 9.26 (5.29) ns

Infant discharge home State-Trait AnxietyInventory [40]

State anxiety level 34.09 (10.78) 35.26 (12.09) ns

Infant discharge home Parental Stressor Scale Level of maternal stressrelated to infant admissionto NICU

Neonatal IntensiveCare [41]

Sights and Sounds 2.85 (1.04) 2.54 (1.06) nsInfant appearance 3.35 (0.91) 3.53 (1.58) nsParental role alteration 3.32 (1.07) 3.40 (1.00) ns

(continued on next page)

Skin-to-skin contact 451

Table 2 (continued)

Time of assessment Composition of aggregatedscores

Description of scale STS(mean (S.D.))

Control(mean (S.D.))

p

(c) Maternal scores, discharge to 4 monthsInfant discharge home Mother and Baby

Scale [42]Maternal perception ofconfidence in care giving

14.29 (2.82) 15.00 (2.14) ns

4 months post birth Edinburgh PostnatalDepression Scale [39]

Presence of symptoms ofpostnatal depression

6.57 (4.71) 6.00 (5.09) ns

4 months post birth State-Trait AnxietyInventory [40]

State anxiety level 31.49 (10.52) 30.79 (11.94) ns

4 months post birth Maternal Soothing OfInfant Distress toImmunisation [43]

Quantity/quality ofmaternal soothingbehaviour

2.23 (0.46) 2.21 (0.79) ns

(d) Maternal scores, 1 year post-term1 year post-term General Health

Questionnaire [44]Maternal mental health

Somatic symptoms 4.63 (3.26) 6.65 (4.37) nsAnxiety 3.50 (2.62) 4.59 (4.08) nsSocial dysfunction 6.33 (3.35) 6.29 (1.53) nsDepression 0.40 (1.30) 0.82 (1.24) ns

1 year post-term Parenting StressIndex [41]

Identification ofdysfunctional parentalbehaviour/stressorsin the parentingrelationship

24.48 (5.95) 25.23 (5.59) ns

1 year post-term Parent to InfantAttachment [45]

Competence Qualityof maternal attachmentto infant

21.51 (2.30) 21.85 (2.60) ns

R. Miles et al.452

breast-fed, 2=mainly breast-fed, 3=both breast milk andformula-fed, 4=mainly formula fed, 5=fully formula-fed.

2.5. Cortisol assays

Blood samples were collected into heparinised vacutainersand spun at 3000 rpm for 15 min. Plasma was removed andfrozen at �80 8C in polypropylene storage vials. Salivasamples were collected by allowing the infant to suck on acotton dental roll. No oral stimulant was used. Saliva wasextracted by immediate centrifugation and stored at�20 8C.Plasma and saliva sample were assayed in batches using astandard radioimmunossay.

2.6. Sample size and data analysis

We calculated that a total sample size of 39 subjects in eachgroup would allow the detection of a 0.75 S.D. (90% power;5% significance level) or 0.65 S.D. (80% power; 5% signifi-cance level) difference in mean outcome measures. Theplanned primary outcomes related to infant measures, aslisted above and the planned secondary outcomes to infantcortisol levels and maternal measures.

Data were analysed by RM and JS using SPSS for Windows(Release 10.1; SPSS, Inc., Chicago, IL). Kolmogrov—Smirnovtests were used to test the normal distribution of alloutcome data. Data were log transformed if necessary. Aninitial parametric or non-parametric analysis was conducted

on individual outcome measures. Effect sizes were calcu-lated in addition to tests of statistical significance. Statis-tical analysis was conducted separately for each individualoutcome measure. Aggregate scores were also computed toovercome the effects of missing data. These were createdfor each time point from the mean Z scores for thosemeasures for which a score was available at that time point.If necessary scores were reverse coded so high aggregatescores indicate optimal profiles. The pain parameters wereaggregated so that a disordered pain response was indicatedby a low behavioural response and short response latencyand a high cortisol response and prolonged distress duration.Mean plasma cortisol during weeks 1 to 4 was calculated foreach infant and these data were log transformed.

3. Results

Eligibility, recruitment and loss to follow-up rates are shownin the Fig. 1. We recruited 46 infants into the STS group and32 into the control group. The difference in the number ofinfants recruited to each group arose because of variationsin the wash-out phases and the need to complete recruit-ment within a finite time period. There were no significantdifferences between the groups in maternal or infantcharacteristics at trial entry (Table 1). No infant had anadverse clinical response to STS contact.

Table 2a—d shows the individual outcome measures formeasures comprising the aggregated scores. There wereno significant differences between the groups in any

Table 3 Mean infant plasma cortisol concentrations (nmol/l),weeks 1—4 post birth and change in salivary cortisol (D) beforeand after immunisation at 4 and 12 months

Plasmacortisol

Control(mean (S.D.))

STS(mean (S.D.))

Effectsize

p

Weeks 1—4 418 (379) 333 (128) �0.22 nsD 4 months 5.22 (7.48) 3.25 (6.36) �0.26 nsD 12 months �0.03 (5.43) �0.28 (5.78) �0.06 ns

Skin-to-skin contact 453

individual outcome measure nor in any of the aggregatedmeasures. The mean infant plasma cortisol levels for eachchild in weeks 1—4 ranged from 17 to 2041 nmol/l. Table 3shows the mean plasma cortisol concentrations weeks 1—4and the salivary cortisol response to 4 and 12-monthimmunisation.

There was considerable variation in the duration of STScontact provided by the 46 mothers in the intervention group(mean=507 min, S.D.=414 (range 0—3350 min), number ofsessions mean=11.26, S.D.=8.17 (range 0—27)). The influ-ence of variation in the amount of STS contact on outcomeswas investigated in two ways. Within the STS group, thosegiving less than the target 20 min STS care were comparedwith those giving more, in relation to the cortisol, child andmaternal aggregate measures. In addition correlations werecalculated for cases providing any STS (i.e. 1 min/day ormore) between the amount of care and outcomes. OnceBonferroni corrections were applied to these comparisons,there were no significant relationships between variation inthe amount of STS provided and cortisol, child and maternalaggregate measures.

4. Discussion

This pragmatic, prospective, controlled, intention to treattrial evaluated the impact of mother—infant STS contact inextremely preterm infants in a neonatal intensive caresetting using quantitative measures of infant and maternaloutcome. The purpose was to investigate hypotheses basedupon animal and human studies and on our pilot data, thatSTS contact would significantly reduce the neonatal endog-enous cortisol response, reduce maternal postnatal depres-sion, and improve infant and maternal outcomes.

Our study failed to demonstrate any effect of STS contactupon later infant responses to the standardised stress ofimmunisation, or upon cognitive and behavioural develop-ment. The small effect sizes indicate that the lack ofsignificant difference in infant outcome aggregates isunlikely to be due to low power. STS contact neitherlowered nor raised plasma cortisol; the latter is animportant point given the vulnerability of these extremelypreterm infants. We have previously shown that STS contactreduces saliva cortisol acutely [29]. The current studysuggests that this acute response is not translated into asustained effect.

We also predicted that STS contact would improvelactation, increase maternal care-giving confidence andenhance mental health and parenting skills. These hypoth-eses were based upon a very limited but highly influentialliterature reporting greater parenting confidence, self

esteem, psychological adaptation to premature birth [22—24] and improved lactation [25—27,47], in relation to STScontact. However we were unable to replicate thesefindings. In previous studies, details of intervention havenot always been reported clearly and have largely beenuncontrolled, with qualitative assessments derived frominterviews conducted by persons responsible for the deliveryof the intervention. This considerably increases the possi-bility of response bias. A combination of these factors mayhave lead to an inflation of the beneficial effects of STScontact and the disparity with our findings. In this emotivearea, publication bias may also have played a role.

A limitation of our study is that there was wide variationin the amount of STS contact actually administered. It wasour impression that many mothers did not appear to enjoySTS contact. This was unexpected and we regret that ourstudy was not designed to explore such a possibility. Afurther limitation was the lack of a measure of intention tobreast-feed, but we judged it inappropriate to try toestablish intentions in mothers who were coming to termswith the birth of a preterm infant. Unit policy was topromote breast-feeding strongly and it is also possible thatthis may have masked the impact of STS on lactation.

The STS intervention has also been termed Kangaroocare. The original Kangaroo Mother Care programme wasdesigned as an alternative to conventional care to addressthe high low birth weight infant mortality in Colombia [46].The core feature was the constant positioning of the nakedinfant on the mother’s breasts enclosed in her clothing.Although these studies differ substantially from thatreported here and cannot be directly compared, aCochrane review [47] concluded that there is insufficientevidence to recommend its routine use in low birth weightinfants.

Many recent studies of STS contact in preterm babieshave demonstrated beneficial short-term outcomes. Theseinclude improved analgesia during heel stick blood sam-pling [48] and accelerated behavioural and autonomicmaturation [49]. However if STS contact is to be activelypromoted in neonatal intensive care units, evidence oflong-term benefit is desirable. Feldman et al. [50] con-ducted a non-randomised study utilising matched controlsand evaluated Kangaroo care in a neonatal care settingcomparable to our own. Kangaroo care was associated withimproved parenting and higher infant perceptual—cognitiveand motor development scores at 6 months of age. Thisstudy differed from ours in several important aspects. Theinfants were more mature with a mean gestational age atbirth of 31—33 weeks, Kangaroo care was commenced at arange of 3—40 days from birth and only mothers whoagreed to administer Kangaroo care for at least 1 h dailyfor 14 days were considered eligible. In contrast the infantsin our study were considerably more immature, with amean gestational age of 28 weeks, Kangaroo care wascommenced within 1 week of birth, and was administeredfor a shorter period. Thus, possibilities that requireconsideration are that Kangaroo care is only be beneficialif administered during an appropriate developmentalwindow, that there is a bdose—responseQ relationship, andthirdly, that any developmental gains might not besustained. We hope that these observations will help informthe design of future trials.

R. Miles et al.454

It is important to note that we did not detect any harmfuleffects of STS contact and we would not seek to dissuademothers who wished to provide this from doing so. Howevergiven the lack of objective evidence of benefit for motherand baby we are unable to recommend the allocation ofscarce health care resources into STS programmes forextremely preterm neonates in neonatal intensive caresettings.

Acknowledgements

This study would not have been possible without the helpwith data collection by Diana Adams. We acknowledgeSimon Matta’s assistance with cortisol assays. This workwas undertaken with funding from the NHS LondonRegional Office, Research and Development Programme,the Sir Halley Stewart Trust and the Touch ResearchInstitute. The views expressed in the paper are those ofthe authors and not necessarily those of the NHS or theDepartment of Health. The authors have no competinginterests to declare.

References

[1] Huppi PS, Schuknecht B, Boesch C, Bossi E, Felblinger J, FuschC, Herschkowitz N. Structural and neurobehavioural delay inpreterm infants. Pediatr Res 1996;39:895–901.

[2] Whitfield MF, Grunau RE. Behavior, pain perception and theextremely low birth weight survivor. Clin Perinatol 2000;27:363–79.

[3] Luciana M, Lindeke L, Georgieff M, Mills M, Nelson CA.Neurobehavioral evidence for working memory deficits inschool-aged children with histories of prematurity. Dev MedChild Neurol 1999;41:521–33.

[4] Isaacs EB, Lucas A, Chong WK, Wood SJ, Johnson CL, MarshallC, Vargha-Khadem F, Gadian DG. Hippocampal volume andeveryday memory in children of very low birth weight. PediatrRes 2000;47:713O–72O.

[5] Taylor HG, Klein N, Minich NM, Hack M. Verbal memory deficitsin children with less than 750 g birth weight. Child Neuropsy-chol 2000;6:49–63.

[6] Horwood LJ, Mogridge N, Darlow BA. Cognitive, educational,and behavioural outcomes at 7 to 8 years in a national very lowbirthweight cohort. Arch Dis Child 1998;79:F12–20.

[7] Hille ETM, den Ouden AL, Saigal S, et al. Behavioural problemsin children who weigh 1000 g or less at birth in four countries.Lancet 2001;357:1641–3.

[8] Anisman H, Zaharia MD, Meaney MJ, Merali Z. Do early lifeevents permanently alter behavioural and hormonal responsesto stressors. Int J Dev Neurosci 1998;16:149–64.

[9] Bremner JD, Vermetten E. Stress and development: behav-ioral and biological consequences. Dev Psychopathol 2001;13:473–89.

[10] Heffelfinger AK, Newcomer JW. Glucocorticoid effects onmemory function over the human life span. Dev Psychopathol2001;13:491–513.

[11] Mathews SG. Antenatal glucocorticoids and the programmingof the developing CNS. Pediatr Res 2000;47:291–300.

[12] Barrington KJ. The adverse neuro-developmental effects ofpostnatal steroids in the preterm infant: a systematic review ofRCTs. BMC Pediatr 2001;1:1 [http://www.biomedcentral.com/1471-2431/1/1].

[13] Ng PC, Wong GWK, Lam CWK, et al. Pituitary—adrenal responsein preterm very low birth weight infants after treatment with

antenatal corticosteroids. J Clin Endocrinol Metab 1997;82:3548–52.

[14] Liu D, Diorio J, Day JC, Francis DD, Meaney MJ. Maternal care,hippocampal synaptogenesis and cognitive development inrats. Nat Neurosci 2000;3:799–806.

[15] Gunnar M. Quality of early care and buffering of neuroendo-crine stress reactions: potential effects on the developinghuman brain. Prev Med 1998;27:208–11.

[16] Murray L, Sinclair D, Cooper P, Ducournau P, Turner P. Thesocioemotional development of 5-year-old children of postnatallydepressed mothers. J Child Psychol Psychiatr 1999;40:1259–71.

[17] Hay DF, Pawlby S, Sharp D, Asten P, Mills A, Kumar R.Intellectual problems shown by 11-year old children whosemothers had postnatal depression. J Child Psychol Psychiatr2001;42:871–89.

[18] Appelbaum M, Batten DA, Belsky J, et al. Chronicity ofmaternal depressive symptoms, maternal sensitivity, and childfunctioning at 36 months. Dev Psychol 1999;35:1297–310.

[19] Atkinson L, Paglia A, Coolbear J, Niccols A, Parker KCH, GugerS. Attachment security: a meta-analysis of maternal mentalhealth correlates. Clin Psychol Rev 2000;20:1019–40.

[20] Martins C, Gaffan EA. Effects of early maternal depression onpatterns of infant—mother attachment: a meta-analytic inves-tigation. J Child Psychol Psychiatr 2000;41:737–46.

[21] VanIjzendoorn MH, Schuengel C, Bakermans-Kranenburg MJ.Disorganized attachment in early childhood: meta-analysis ofprecursors, concomitants, and sequelae. Dev Psychopathol1999;11:225–49.

[22] Affonso D, Wahlberg V, Persson B. Exploration of mothers’reactions to the Kangaroo method of premature care. NeonatalNetw 1989;7:43–51.

[23] Affonso D, Bosque E, Wahlberg V, Brady J. Reconciliation andhealing through skin-to-skin contact provided in an Americantertiary level intensive care nursery. Neonatal Netw 1993;12:25–32.

[24] Gale G, Frank L, Lund C. Skin—skin holding (Kangaroo) of theintubated premature infant. Neonatal Netw 1993;12:49–57.

[25] Whitelaw A, Heisterkamp G, Sleath K, Acolet D, Richards M.Skin to skin contact for very low-birth-weight infants and theirmothers. Arch Dis Child 1988;63:1377–81.

[26] Wahlberg V, Affonso D, Persson B. A reterospective compara-tive study using the Kangaroo method as a complement to thestandard nursing care. Eur J Public Health 1992;2:24–37.

[27] Bier JA, Ferguson AE, Morales Y, Liebling JA, Archer D, Oh W,Vohr BR. Comparison of skin-to-skin contact with standardcontact in low-birth-weight infants who are breast-fed. ArchPediatr Adolesc Med 1996;150:1265–9.

[28] Lucas A, Morley R, Cole TJ, Lister G, Leeson-Payne C. Breastmilk and subsequent intelligence quotient in children bornpreterm. Lancet 1992;339:261–4.

[29] Gitau R, Modi N, Giannakoulopoulos X, Bond C, Glover V,Stevenson J. Acute effects of maternal skin-to-skin contact andmassage on saliva cortisol in preterm babies. J Reprod InfantPsychol 2002;20:83–8.

[30] Cooke R. Report of working group of the British Association ofPerinatal Medicine and Neonatal Nurses Association on Cate-gories of babies requiring neonatal care. Arch Dis Child 1992;67:868–9.

[31] Taddio A, Nulman I, Sheva Koren B, Stevens B, Koren G. Arevised measure of acute pain in infants. J Pain SymptomManage 1995;10:456–63.

[32] Ramsay DS, Lewis M. Developmental change in infant cortisoland behavioral response to inoculation. Child Dev 1994;65:1491–502.

[33] Lewis M, Thomas D. Cortisol release in infants in response toinoculation. Child Dev 1990;61:50–9.

[34] Haataja L, Mercuri E, Regev R, Cowan F, Rutherford M,Dubowitz V, Dubowitz L. Optimality score for the neurologic

Skin-to-skin contact 455

examination of the infant at 12 and 18 months of age. J Pediatr1999;135:153–61.

[35] Fagan J. The Fagan Test of Infant Intelligence version 4.1.Cleveland7 Infantest Corporation; 1991.

[36] Griffiths R. The Griffiths Mental Development Scales from birthto 2 years. Oxford7 The Test Agency; 1996.

[37] Briggs-Gowan MJ, Carter AS. Preliminary acceptability andpsychometrics of the Infant Toddler Social and EmotionalAssessment (ITSEA): a new adult-report questionnaire. InfantMent Health J 1998;19:422–45.

[38] Abidin PR. Parenting Stress Index, 3rd ed. Odessa (FL)7 Psy-chological Assessment Resources Inc; 1995.

[39] Cox JL, Holden JM, Sagovsky R. Detection of postnataldepression: development of the Edinburgh Postnatal Depres-sion Scale. Br J Psychiatry 1987;150:782–6.

[40] Spielberger CD, Gorsuch R, Lushene R, Vagg PR, Jacobs GA.Manual for the State-Trait Anxiety Inventory. California7Consulting Psychologists Press; 1983.

[41] Miles MS, Funk SG, Carlson J. Parental Stressor Scale: NeonatalIntensive Care unit. Nurs Res 1983;42:148–52.

[42] Wolke D, St. James-Roberts I. Mother and baby scale. In:Brazelton TB, Nugent JK, editors. Neonatal behaviouralassessment scale. Cambridge, England7 Cambridge UniversityPress; 1987. p. 116–49.

[43] Lewis M, Ramsay D. Effect of maternal soothing on infant stressresponse. Child Dev 1999;70:11–20.

[44] Goldberg D. Manual of the General Health Questionnaire.Windsor, England7 NFER-NELSON; 1978.

[45] Condon JT, Corkindale CJ. The assessment of parent-to-infantattachment: development of a self-report questionnaire in-strument. J Reprod Infant Psychol 1998;16:57–76.

[46] Charpak N, Ruiz-Pelaez JG, Charpak Y, Rey-Martinez. KangarooMother Program: an alternative way of caring for low birthweight infants? One year mortality in a two cohort study.Pediatrics 1994;94:804–10.

[47] Conde-Agudela A, Diaz-Rossello JL, Belizan JM. Kangaroomother care to reduce morbidity and mortality in low birth-weight infants (Cochrane Review). The cochrane library, issue,vol. 2. Oxford7 Update Software; 2003.

[48] Ludington-Hoe SM, Hosseini R, Torowicz DL. Skin-to-skincontact (Kangaroo care) analgesia for preterm infant heelstick. AACN Clin Issues 2005;16:373–87.

[49] Feldman R, Eidelman AI. Skin-to-skin contact (Kangaroo care)accelerates autonomic and neurobehavioural maturation inpreterm infants. Dev Med Child Neurol 2003;45:274–81.

[50] Feldman R, Eidelman AI, Sirota L, Weller A. Comparison of skin-to-skin (Kangaroo) and traditional care: parenting outcomesand preterm infant development. Pediatrics 2002;110:16–26.