Long-term effects of routine morphine infusion in mechanically

ventilated neonates on children’s functioning: Five-year follow-up

of a randomized controlled trial

Joke de Graaf

, Richard A. van Lingen

, Sinno H.P. Simons

, Kanwaljeet J.S. Anand

Hugo J. Duivenvoorden

, Nynke Weisglas-Kuperus

, Daniella W.E. Roofthooft

Liesbeth J.M. Groot Jebbink

, Ravian R. Veenstra

, Dick Tibboel

, Monique van Dijk

⇑

Department of Pediatric Surgery, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands

Princess Amalia Department of Pediatrics, Department of Neonatology, Isala Clinics, Zwolle, The Netherlands

Department of Pediatrics, VUMC, Amsterdam, The Netherlands

Department of Pediatrics, University of Tennessee Health Science Center, Memphis, TN, USA

Departments of Medical Psychology and Psychotherapy, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands

Department of Neonatology, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands

Department of Medical Psychology, Isala Clinics, Zwolle, The Netherlands

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

article info

Article history:

Received 4 December 2009

Received in revised form 7 February 2011

Accepted 7 February 2011

Keywords:

Follow-up randomized controlled trial

Pain

Drug therapy

Pediatrics

Child development

Neonatology and infant care

abstract

Newborns on ventilatory support often receive morphine to induce analgesia. Animal experiments sug-

gest that this may impair subsequent cognitive and behavioral development. There are sparse human

data on long-term effects of neonatal morphine. We aimed to investigate the effects of continuous mor-

phine administered in the neonatal period on the child’s functioning. We conducted a follow-up study

among 5-year-olds who, as mechanically ventilated neonates, had participated in a placebo-controlled

trial on effects of morphine administration on pain and neurologic outcome. They were now tested on

intelligence, visual motor integration, behavior, chronic pain, and health-related quality of life. Univariate

analyses showed signiﬁcantly lower overall intelligence quotient (IQ) scores for children who earlier had

received morphine, that is, mean 94 (SD 14.5) versus 100 (SD 12.9) for those who received placebo

(P = 0.049). Other between-group differences in outcomes were not found. The statistical difference dis-

appeared after correction for treatment condition, open-label morphine consumption over the ﬁrst

28 days, and a propensity score for clinically relevant co-variables in multiple regression analyses. How-

ever, scores on one IQ subtest, ‘‘visual analysis,’’ were signiﬁcantly negatively related to having received

morphine and to open-label morphine consumption the ﬁrst 28 days. The ﬁnding of a signiﬁcant effect of

morphine on the ‘‘visual analysis’’ IQ subtest calls for follow-up at a later age focusing on the higher-order

neurocognitive functions.

1. Introduction

As recently as 2 decades ago, anesthetists, for fear of respiratory

depression, were reluctant to prescribe pain-relieving periopera-

tive opioids to neonates [35]. Then in 1987, Anand et al. [3] pub-

lished a landmark article showing dramatically worse outcomes

after patent ductus arteriosus surgery for preterm neonates who

did not receive perioperative opioids compared to those who did.

In further research on human neonatal pain, it proved to be associ-

ated with changes in pain responses [18,27,47–49], behavior

[17,22], somatosensory perception [43,51], modulation of pain

[16], and stress responses [19]. Later studies brought evidence that

morphine analgesia has positive short-term effects, for example,

diminished stress response [34,37,54].

We studied the administration of continuous morphine in a

randomized, placebo-controlled trial (RCT) in 150 mechanically

ventilated preterm neonates. These neonates showed no differ-

ences in pain intensities and no beneﬁcial effect on neurologic out-

comes following continuous morphine infusions [44–46]. The

incidence of intraventricular hemorrhage (IVH) was signiﬁcantly

reduced in the morphine-treated neonates. A larger RCT, using

higher doses of morphine, showed that morphine decreased

clinical signs of pain, but no differences occurred in the rates of

doi:10.1016/j.pain.2011.02.017

⇑

Corresponding author. Address: Department of Pediatric Surgery, Erasmus MC-

Sophia, Room SK 1276, Postbus 2060, 3000 CB Rotterdam, The Netherlands. Tel.:

+31 10 7037 091; fax: +31 10 4636 288.

E-mail address: [email protected] (M. van Dijk).

PAIN

152 (2011) 1391–1397

www.elsevier.com/locate/pain

neonatal mortality, severe IVH, and periventricular leukomalacia

[2]. Both RCTs as well as a meta-analysis [6] concluded that routine

morphine use is not recommended for ventilated neonates.

Studies in animals suggest potential adverse long-term effects

of morphine. Morphine administration to neonatal rats produces

long-term changes in behavior and brain function [23] and impairs

the adult cognitive functioning [30], in particular, spatial recogni-

tion memory [28]. Basic science has shown that the opioid system

modulates neural proliferation in vivo [42]. Extrapolating from

these data suggests a harmful role of morphine treatment in dis-

turbing neurogenesis of newborn babies. Mechanistically, mor-

phine increases apoptosis of human microglial cells [26] and

increases red neuron degeneration in the rat brain, which may lead

to cerebral dysfunction [4].

Boasen et al. [9] recently showed in rodents that both neonatal

pain and morphine treatment produced long-lasting behavioral ef-

fects to a degree sufﬁcient to alter learning, while the combined

impact of the treatments did not.

In just one published human study, neonates were randomized

to receive morphine or placebo [29]. The researchers evaluated

intelligence, motor abilities, and behavior at 5–6 years. The fol-

low-up tests showed no beneﬁcial effect of morphine administra-

tion. In a cohort follow-up study, Grunau et al. [21] found that

poorer cognition and motor function were associated with higher

number of skin-breaking procedures, but independent of early ill-

ness severity, total morphine dose, or treatment with postnatal

steroids. To resolve the discrepancy between ﬁndings of animal

studies and the sparse human data of the long-term effect of mor-

phine administration, we undertook a 5-year follow-up study of

patients previously enrolled in the RCT of Simons et al. [46]. Our

aim was to investigate the effects of morphine received in the neo-

natal period on the child’s functioning at the age of 5 years in terms

of intelligence, visual motor integration, behavior, chronic pain,

and health-related quality of life.

2. Methods

2.1. The original study

The original study was performed in 2 level III neonatal inten-

sive care units (NICUs) in the Netherlands in 2000–2002 [44–46].

Neonates younger than 3 days on ventilatory support were ran-

domly allocated to receive a loading dose (100

g/kg) followed

by a continuous infusion of 10

g/kg/h of either morphine

(n = 73) or placebo (n = 77). Children judged to be in pain or

distress received open-label morphine of 50

g/kg followed by

5-10

g/kg/h.

2.2. Overview of the follow-up-study

Eighteen of the 150 participants in the original study had died

at the time we started the present study. Seven of the 132 survi-

vors were being followed elsewhere and were not invited. Two of

them had severe physical limitations; one had severe developmen-

tal delay, and 4 had both. Three of these 7 children were diagnosed

with mild IVH in the neonatal period and one with periventricular

leukomalacia. Three had been randomized to receive continuous

morphine, and 4 to receive placebo.

Parents of the 125 eligible children were invited to participate

between December 2005 and January 2008. Written informed con-

sents were obtained before follow-up testing. Parents also pro-

vided consent for the child’s teacher’s contribution to testing.

The medical ethical review boards at Erasmus MC-Sophia Chil-

dren’s Hospital (Rotterdam) and Isala Clinics (Zwolle) approved

this study.

The follow-up assessment started with measurements of height

and weight and medical evaluation by pediatricians (R.vL. and

N.W.). Subsequently, child psychologists (J.dG. and R.V.), blinded

to the neonatal treatment condition, administered an intelligence

test and a visual motor integration test. These assessments re-

quired approximately 2 hours. Two children were seen at home.

The parents and the child’s teacher completed different question-

naires about the child’s behavior, chronic pain, and quality of life.

2.3. Assessments and variables

2.3.1. Background characteristics

Baseline characteristics were available from the original study

[44–46]. Data on total open-label morphine intravenous infusion

in the ﬁrst 7 and 28 days, and other cumulative analgesic/seda-

tives/antiepileptics given in the ﬁrst 28 days were retrieved from

the medical charts of all participants and nonparticipants of the

follow-up study. Total hospitalization, including the NICU stay or

admission to other hospitals, was obtained from the referral

centers.

The Growth Analyzer, version 3.5 (Dutch Growth Foundation,

Rotterdam, The Netherlands) served to evaluate height to age

and weight to height, expressed in z scores, on the basis of Dutch

and Turkish reference values.

Parents completed a questionnaire asking about profession and

education, predominantly spoken language at home, and details

about the child’s surgeries in the past 5 years, including opioids

use before, during, or after surgery. Surgeries were classiﬁed as

minor or major based on opioids use or clinical indications.

Family socioeconomic status (SES) was derived from the highest

occupational or educational level of either one of the parents, using

a computer program, Occupation Classiﬁcation [10]. The score

could range from 1 to 9 and was reduced to 3 levels: scores 1 to

3 corresponding with low SES, 4 and 5 with middle SES, and 6 to

9 with high SES.

2.3.2. Intelligence

Intelligence was measured with the short version of the Revi-

sion Amsterdam Child Intelligence Test (RAKIT) [8]. This test pro-

duces a mean intelligence quotient (IQ) score of 100 and

standard deviation (SD) of 15, based on a Dutch norm group. This

short version consists of 6 subtests, to measure reasoning, passive

verbal learning, spatial orientation, active learning, visual analysis,

and verbal ﬂuency [40]. The full version of the RAKIT correlates

well with the Wechsler Intelligence Scale for Children – Revised

(r = 0.86) [7].

2.3.3. Visual motor integration

Visual motor integration was assessed with the Beery-

Buktenica Developmental Test of Visual-Motor Integration (Beery

VMI) [5]. This test helps to identify difﬁculties that children may

have with integrating, or coordinating, their visual-perceptual

and motor (ﬁnger and hand movement) abilities. Children are

asked to copy geometric ﬁgures in order of increased difﬁculty.

The Beery VMI produces a mean score of 100 and SD of 15.

2.3.4. Behavior

Parents completed the Dutch-language version of the Child

Behavior Checklist 1½-5 years (CBCL) [1]. Three scales are pro-

duced: the Total Problem Scale (100 items) and subscales Internal-

izing and Externalizing problems. For the 3 scales, alike T scores of

60 or higher are considered to be consistent with a problem in the

subclinical or clinical range [1]. Teachers completed the Dutch-

language version of the Teacher Report Form for ages 1½-5 years.

This tool evaluates the child’s behavior in the school setting.

Structure and scoring system are similar to those of the CBCL [1].

1392 J. de Graaf et al. / PAIN

152 (2011) 1391–1397

2.3.5. Chronic pain

Prevalence of chronic pain was assessed from the Dutch Chronic

Pain Questionnaire [33], completed by the parents. The question-

naire deﬁnes chronic pain as pain with duration longer than

3 months.

2.3.6. Health-related quality of life

Parents completed the Dutch-language version of the 15-item

Health Utility Index (HUI-15) [13] to evaluate their child’s

health-related quality of life. HUI-15 scores were transformed to

the Health Utility Index 3 (HUI3), which consists of 8 attributes

of health status: vision, hearing, speech, ambulation, dexterity,

emotion, cognition, and pain. Each attribute comprises 5 or 6 lev-

els, varying from highly impaired to normal [11,15]. Using the gi-

ven formula of the HUI3, the theoretically possible score ranges

from 0.0 (dead) to 1.0 (perfect health). Scores between 0.89 and

0.99 reﬂect mild disability, scores between 0.70 and 0.88 reﬂect

moderate disability, and scores <0.70 reﬂect severe disability [14].

2.4. Statistical analyses

Background characteristics and continuous outcome variables

were compared with Student t test for continuous and normally

distributed data, displayed as means and standard deviation. The

Mann-Whitney test was used in case of skewed continuous data

distribution; the medians and interquartile range (IQR) were given.

The Pearson

test or Fisher’s exact test were used to evaluate cat-

egorical data.

To take relevant neonatal and other background characteristics

into account, multiple linear regression analyses were performed

for the normally distributed outcome variables. A large number

of neonatal and background characteristics described in the

literature were combined into a propensity score to correct for

baseline imbalance [55]. This score was derived from logistic

regression analysis using a forced-entry model. In this analysis,

the treatment condition (morphine or placebo) was the criterion

variable; the co-variables were: center, sex, gestational age (days),

Clinical Risk Index for Babies [12], medium SES and high SES, with

low SES as the reference variable.

Birth weight was compared to national standards and expre-

ssed in z scores, depending on sex and age and corrected for

prematurity.

We performed multiple regression analyses, with and without

the propensity score to adjust for the relevant background vari-

ables and with the amount of open-label morphine in the ﬁrst

28 days after birth and treatment condition as explanatory vari-

ables. Because primary ventilation is known to be associated with

morphine treatment, we will perform analyses with and without

this explanatory variable. To determine if outliers inﬂuenced the

regression model, the Cook’s distances were calculated. Multicol-

linearity was tested with the Variance Inﬂation Factor, which

should not exceed 10 for any of the variables.

The critical level of signiﬁcance was set at 0.05 (2-sided) for all

analyses. Data were analyzed using the statistical software package

SPSS 15.0 for Windows (SPSS Inc, Chicago, IL, USA).

3. Results

3.1. Background characteristics

Parents of 90 of the 125 eligible children (72%) consented to

participate. Forty-nine children had earlier received continuous

morphine and 41 children received placebo (Fig. 1). Parents of 16

children (13%) could not be traced and parents of 19 children

Fig. 1. Flowchart of participation.

J. de Graaf et al. / PAIN

152 (2011) 1391–1397

1393

(15%) refused participation. Neonatal characteristics of the nonpar-

ticipants did not differ from those of the participating children

(Table 1).

Neonatal characteristics of the children participating in the fol-

low-up evaluation did not differ between the morphine and pla-

cebo groups (Table 2). At the age of 5 years, children who had

received morphine were signiﬁcantly shorter (height for age) than

those who had received placebo (P = 0.02). One part of the explana-

tion could be that the correlation of z-score height with small for

gestational age (SGA) at birth was 0.49, and the incidence of SGA

was 26.5% in the morphine group and 12.2% in the placebo group.

When the SGA children were excluded, the difference between the

morphine and placebo group was no longer signiﬁcant.

Six children in each group had undergone minor surgery after

the neonatal period (Table 3). One child in the morphine group

had undergone major surgery, that is, bowel resection, resulting

in short bowel syndrome.

3.2. Intelligence

IQ scores were evaluated for 89 children, as one test result

was unreliable. Ten of the 48 children (20.8%) in the morphine

group had an IQ score of 1 SD below average (IQ score <85), ver-

sus 5 of 41 (12.2%) in the placebo group (odds ratio 1.90, 95%

conﬁdence interval [CI] .59–6.08). Univariate analyses showed

that IQ signiﬁcantly differed between groups, that is, mean 94.0

(SD 14.5) for the morphine group versus mean 99.8 (SD 12.9)

for the placebo group (P = 0.049) (Table 4). Table 5 gives 3 multi-

ple regression analyses with IQ as outcome variable. Model 1 in-

cludes treatment group and open-label morphine consumption in

the ﬁrst 28 days of life as predictor variables with a statistically

signiﬁcant worse outcome for the morphine group (P = 0.046),

along with a trend towards signiﬁcance for the open label

morphine (P = 0.058). When the analysis was adjusted for the

propensity score, treatment group and open label morphine

consumption in the ﬁrst 28 days were no longer signiﬁcant. The

further addition of primary ventilation as explanatory variable

in model 3 reversed the sign of the regression coefﬁcient of open

label morphine and showed a trend for longer primary ventilation

predicting IQ (P = 0.073). The 95% CI Variance Inﬂation Factors

remained below 2.5.

Evaluation of Cook’s distances revealed no outliers among the

cases. Entering all co-variables in the model separately gave the

same results as the analyses corrected for the propensity score.

Only SES contributed signiﬁcantly to the prediction of IQ.

Regression analysis on separate subtests of the intelligence test,

again adjusted for the propensity score, revealed a signiﬁcant neg-

ative effect of the morphine condition (P = 0.021) and a signiﬁcant

negative effect of open-label morphine consumption (P = 0.029) on

the ‘‘visual analysis’’ RAKIT subtest performance.

3.3. Visual motor integration

Fifteen of the 49 children (30.6%) in the morphine group had a

Beery VMI score of <1 SD below average (VMI score <85) versus 5

of the 41 children (12.2%) in the placebo group (odds ratio 3.18,

95% CI 1.04–9.70). The mean VMI score for the morphine group

tended to be lower than that for the placebo group, that is, 93.7

(SD 12.3) versus 98.0 (SD 12.7) (P = 0.104) (Table 4).

3.4. Behavior

As 6 of the 90 parents did not complete the Child Behavior

Checklist, analysis included 84 children. Proportions of children

in the (sub)clinical range varied from 10.3% (for externalizing prob-

lems in the placebo group) to 15.6% (for internalizing problems in

the morphine group). These proportions are comparable to those

found for normal American children. The CBCL total score and

Table 1

Demographic and clinical characteristics in neonatal period from participating children versus nonparticipating children (n = 125).

Participating children (n = 90) Nonparticipating children (n = 35) P value

Center, n (%) Rotterdam 54 (60.0) 23 (65.7) 0.56

Zwolle 36 (40.0) 12 (34.3)

Sex, n (%) Male 51 (56.7) 19 (54.3) 0.81

Female 39 (43.3) 16 (45.7)

Gestational age (weeks) Median (IQR) 30.0 (27.5–31.6) 29.1 (28.1–31.2) 0.62

Birth weight (grams) Median (IQR) 1313 (975–1765) 1300 (1050–1560) 0.84

Duration primary mechanical ventilation (hours) Median (IQR) 71 (29–159) 62 (25–89) 0.23

Duration total ﬁrst hospital admission (days) Median (IQR) 58 (34-79) 61 (36–84) 0.37

CRIB Median (IQR) 3.0 (1.0-5.0) 2.0 (1.0–5.0) 0.83

IVH, n (%) No 62 (68.9) 25 (71.4)

Mild 25 (27.8) 9 (25.7) 0.96

Severe 3 (3.3) 1 (2.9)

PVL, n (%) All grades 2 (2.2) – –

Number of painful procedures ﬁrst 14 days Median (IQR) 124 (73–219) [n = 82] 117 (69–196) [n = 30] 0.26

Cumulative dose of open-label morphine (

g/kg)

administration ﬁrst 28 days

Median (IQR) 106 (15–446) 109 (0-324) 0.84

Cumulative dose other analgesics/ sedatives/ antiepileptic (

g/kg)

administration ﬁrst 28 days

Fentanyl

n (%) 5 (5.6) 4 (11.4) –

Median (IQR) 6 (5–34) 16 (11-261)

Midazolam

n (%) 3 (3.3) 3 (8.6) –

Median (IQR) 130 (–) 47 (–)

Phenobarbital

n (%) 3 (3.3) 3 (8.6) –

Median (IQR) 28347 (20,179–40,404) 28846 (20,790–45,145)

Nalaxone

n (%) 7 (7.8) – –

Median (IQR) 14 (10–20)

IQR, interquartile range; CRIB, Clinical Risk Index for Babies; IVH, intraventricular hemorrhage; PVL, periventricular leukomalacia.

Data are expressed for sample size: participant, n = 90; participants, n = 35, unless otherwise indicated between brackets.

1394 J. de Graaf et al. / PAIN

152 (2011) 1391–1397

subscale scores did not differ between the morphine and placebo

groups (Table 4).

Teachers completed the Teacher Report Form for 81 of the 90

children. Proportions of children in the (sub)clinical range varied

from 10.8% for externalizing problems in the placebo group to

22.7% for internalizing problems in the morphine group (Table 4).

There were no signiﬁcant differences between the 2 groups.

3.5. Chronic pain

Parents reported chronic pain for 6/41 (14.6%) and 5/36 (13.9%)

children in the morphine and placebo groups, respectively (Ta-

ble 4). These 11 children with chronic pain had 1–4 different pain

locations, including the abdomen (n = 5), limbs (n = 5), head

(n = 2), throat (n = 2), back (n = 1), neck (n = 1), ear (n = 2), and

chest (n = 1). Pain frequency varied between daily and once a

month. Information on number of painful procedures in the ﬁrst

14 days of life was available for 72 children. This number did not

signiﬁcantly (P = 0.57) differ between children with chronic pain

now (n = 10) and those without (n = 62), with medians (IQR) of

91 (69–181) versus 168 (79–227), respectively.

3.6. Health-related quality of life

Parents’ scores on the Health Utility Index, completed for 73

children, did not differ between the 2 groups (median 1.0, IQR

0.9-1.0 for each group) (Table 4). The 3 children for whom the

HUI score reﬂected severe disability (2 in the morphine group

and 1 in the placebo group) showed mild pain restricting daily

activities and were reported to feel irritated/angry/agitated, or un-

happy, not well understood by others, and to have difﬁculty with

problem-solving, eyesight, and memory.

4. Discussion

Follow-up at 5 years showed no signiﬁcant differences in

intelligence, visual-motor integration, behavior, chronic pain, or

Table 4

Outcome variables for the morphine group and placebo group (n = 90).

Morphine group n = 49 Placebo group n = 41 P value

IQ Mean (SD) 94.0 (14.5) [n = 48] 99.8 (12.9) [n = 41] 0.049

Visual-motor integration Mean (SD) 93.7 (12.3) 98.0 (12.7) 0.104

Chronic pain, n (%) Yes 6 (14.6) [n = 41] 5 (13.9) [n = 36] 0.93

Daily to once a week 2 (4.9) 1 (2.8)

2 to 3 times per months 3 (7.3) 2 (5.6)

Once a month or less often 1 (2.4) 2 (5.6)

Heath utility index [n = 38] [n = 35]

Median (IQR) 1.0 (.9–1.0) 1.0 (.9–1.0) 0.66

No disability 24 (63.2) 23 (65.7) 0.95

Mild disability 7 (18.4) 7 (20.0)

Moderate disability 5 (13.2) 4 (11.4)

Severe disability 2 (5.3) 1 (2.9)

Child behavior checklist [n = 45] [n = 39]

Internalizing problems

T-score Mean (SD) 47.1 (11.9) 46.5 (11.0) 0.81

(Sub)clinical range n (%) 7 (15.6) 5 (12.8) 0.72

Externalizing problems

T-score Mean (SD) 46.7 (10.2) 46.9 (9.6) 0.93

(Sub)clinical range n (%) 5 (11.1) 4 (10.3) 0.90

Total problems

T-score Mean (SD) 46.6 (11.1) 45.8 (9.9) 0.72

(Sub)clinical range n (%) 5 (11.1) 5 (12.8) 0.81

Teacher Report Form [n = 44] [n = 37]

Internalizing problems

T-score Mean (SD) 50.1 (10.3) 48.8 (10.7) 0.59

(Sub)clinical range n (%) 10 (22.7) 6 (16.2) 0.46

Externalizing problems

T-score Mean (SD) 52.6 (7.9) 49.0 (9.7) 0.07

(Sub)clinical range n (%) 7 (15.9) 4 (10.8) 0.51

Total problems

T-score Mean (SD) 52.1 (8.8) 48.7 (11.7) 0.15

(Sub)clinical range n (%) 9 (20.5) 7 (18.9) 0.86

IQ, intelligence quotient; IQR, interquartile range.

Table 5

Results of multiple regression analyses of IQ at 5 years (n = 89).

Model 1 Model 2 Adjusted for the propensity score

Model 3 Adjusted for the propensity score

B 95% CI P Value B 95% CI P Value B 95% CI P Value

Constant 101.2 96.8–105.7 <0.001 101.6 97.4–105.8 <0.001 103.2 98.2–105.7 <0.001

Treatment group 5.8 11.6–0.1 0.046 3.95 9.4–1.5. 0.155 3.73 9.1–1.7 0.174

Open-label morphine 0.003 0.01–0.001 0.058 0.002 0.006–0.001 0.141 0.001 0.004–0.005 0.77

Primary ventilation 0.028 0.058–0.003 0.073

0.083 0.039 0.069

IQ, intelligence quotient; CI, conﬁdence interval. B values are unstandardized regression coefﬁcients. R

excluding the variance explained by the propensity score in model 2

and 3.

Propensity score including: center, sex, gestational age (days), birth weight (SDS), Clinical Risk Index for Babies (CRIB), socioeconomic status.

J. de Graaf et al. / PAIN

152 (2011) 1391–1397

1395

health-related quality of life between children who as neonates

had received either morphine or placebo. However, there was a

trend towards a more negative outcome in subjects who received

more morphine during the neonatal period. We cannot rule out

that neonatal morphine consumption does clinically signiﬁcant

harm, as the 95% CI for treatment condition included a lower

boundary of 9.4 IQ points.

Also, worse performance on ‘‘visual analysis,’’ a subtest of the IQ

test, was statistically signiﬁcantly related to neonatal morphine

consumption even after adjustment for baseline characteristics

with the propensity score. The one comparable previous study

did not ﬁnd adverse effects of neonatal morphine on the intelli-

gence, motor function, and behavior of 5- to 6-year-old children

[29]. This study differed from the present study with regard to

the 5- to 10-fold higher morphine dosage, the induction of neuro-

muscular paralysis, and lack of an observational and validated pain

assessment instrument [36,37]. A limitation of the MacGregor [29]

study was that only overall IQ, overall motor ability, and overall

behavior were analyzed. The results of the present study show that

it is extremely important to investigate effects of morphine on spe-

ciﬁc sub-domains of neurodevelopment. Children born preterm

show relative vulnerability of visual processing and visual memory

[39] that, in turn, contributes to their academic difﬁculties in

mathematics [20,50]. Even when overall performance is good,

adults born very preterm activate different neural networks to pro-

cess visuospatial material [32]. If, based on the results of the pres-

ent study, morphine adversely, selectively, further adds to effects

on visual processing, this may have important long-term sequelae.

For instance, children in the morphine group of our study obtained

lower scores on ‘‘visual analysis,’’ a subtest of the intelligence test,

than the children in the placebo group. This test requires inhibitory

skills, which is one of the domains of higher-order neurocognitive

functions, the so-called executive functions. As skills like these are

still developing at the age of 5 years, follow-up at later age is called

for and may show more robust differences.

The multivariate regression analyses showed that only socio-

economic status was a signiﬁcant predictor of IQ, consistent with

our earlier ﬁndings [52]. Surprisingly, gestational age did not pre-

dict IQ at age 5 years, even though the study population consisted

of children born at gestational ages from 25 to 40 weeks. A possible

explanation could be that also, the term-born children were criti-

cally ill and needed mechanical ventilation, which may have mod-

erated the effects of preterm birth on cognitive outcomes.

We found that duration of primary ventilation was a relevant

contributing factor, although the subjects were randomized and

duration of ventilation did not differ signiﬁcantly between the 2

groups. We evaluated morphine consumption during the ﬁrst

28 days of life, while the original RCT took place in the ﬁrst 7 days

of life.

Apart from the psychological assessments, chronic pain and

health-related quality of life were other focal points of this study.

We found an incidence of chronic pain comparable to the 19.3%

incidence in healthy 4- to 7-year-old children in the Netherlands

[33]. Also, the number of painful procedures in the neonatal period

in this study did not seem to inﬂuence the development of chronic

pain later in life, consistent with previous studies [41,51]. In addi-

tion, continuous morphine therapy in the NICU did not appear to

mitigate or accentuate the risk of developing chronic pain in pre-

school-age children.

A surprising result was the difference in height between the 2

groups. One part of the explanation could be that the correlation

of z-score height with SGA at birth was 0.49, and the incidence

of SGA was 26.5% in the morphine group and 12.2% in the placebo

group. When the SGA children are excluded, the difference be-

tween the morphine and placebo group is no longer signiﬁcant.

A limitation of this study is the relatively small sample size,

which, however, inevitably resulted from death of the children in

the original sample, loss to follow-up, and parental refusal to par-

ticipate. Another limitation is the potential for selection bias due to

the exclusion of 7 children with severe mental or physical disabil-

ities. Although they were equally divided over the 2 treatment

groups, their exclusion may have inﬂuenced the outcomes. Third,

considering that pain during the neonatal period has been associ-

ated with decreased thermal sensitivity later in life [24,43,51],it

would have been interesting to investigate the relation between

pain sensitivity and chronic pain conditions. Assessment of pain

sensitivity was not feasible, however, because participants were

too young. QST is feasible from the age of 6–7 years [25,31]. Fol-

low-up at older ages should include QST as an objective method

to determine pain sensitivity. Fourth, another limitation was that

we did not assess gross motor development with, for instance,

the Movement Assessment Battery for Children. Children born pre-

term are at increased risk of motor impairment [53]

. In addition, it

would have been interesting because a short-term follow-up at

36 weeks postconceptional age of the NEOPAIN study revealed a

signiﬁcant association between morphine and poplietal angle (a

subtest of the Neurobehavioral Assessment of the Preterm Infant)

[38]. Furthermore, Grunau et al. [21] found that larger total mor-

phine dose correlated with poorer motor development at

8 months, but not at 18 months. However, because our tests took

the larger part of a day, we had to make a selection of tests because

otherwise, the 5-year-old subjects would not have been able to

complete all tests in a reliable manner.

The results of the present study suggest that morphine does not

have major long-term effects at the low doses used. However, the

differences on visual analysis of the intelligence test warrant

follow-up at later age. The long-term consequences of larger

morphine doses during the neonatal period would be highly

relevant to study as well.

Conﬂict of interest statement

The authors declare that they have no conﬂict of interest.

Acknowledgement

This work was supported by the Erasmus MC Rotterdam and

Isala Clinics Zwolle. The authors thank the parents and children

who participated in this study. Ko Hagoort is thanked for his help

in preparing the manuscript.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in

the online version, at doi:10.1016/j.pain.2011.02.017.

References

[1] Achenbach TM, Rescorla LA. Manual for the ASEBA preschool forms &

proﬁles. Burlington, VT: University of Vermont, Research Center for Children,

Youth, & Families; 2000.

[2] Anand KJ, Hall RW, Desai N, Shephard B, Bergqvist LL, Young TE, Boyle EM,

Carbajal R, Bhutani VK, Moore MB, Kronsberg SS, Barton BA. Effects of

morphine analgesia in ventilated preterm neonates: primary outcomes from

the NEOPAIN randomised trial. Lancet 2004;363:1673–82.

[3] Anand KJ, Sippell WG, Aynsley-Green A. Randomised trial of fentanyl

anaesthesia in preterm babies undergoing surgery: effects on the stress

response. Lancet 1987;1:62–6.

[4] Atici S, Cinel L, Cinel I, Doruk N, Aktekin M, Akca A, Camdeviren H, Oral U.

Opioid neurotoxicity: comparison of morphine and tramadol in an

experimental rat model. Int J Neurosci 2004;114:1001–11.

[5] Beery KE, Buktenica N, Beery VMI. Administration, scoring, and teaching

manual. ﬁfth ed. Minneapolis, MN: NCS Pearson, Inc.; 2004.

1396 J. de Graaf et al. / PAIN

152 (2011) 1391–1397

[6] Bellu R, de Waal KA, Zanini R. Opioids for neonates receiving mechanical

ventilation. Cochrane Database Syst Rev 2008:CD004212.

[7] Bleichrodt N, Drenth PJD, Zaal JN. Intelligentie-Meting bij Kinderen. Empirische

en Methodologische Verantwoording van de Gereviseerde Kinder Intelligentie

Test [Intelligence-Measurement in Children. Empirical and Methodological

Justiﬁcation of the Revised Child Intelligence Test]. Lisse: Swets & Zeitlinger

B.V; 1987.

[8] Bleichrodt N, Drenth PJD, Zaal JN, Resing WCM. Revisie Amsterdamse Kinder

Intelligentie Test [Revised Amsterdam Child Intelligence Test]. Lisse: Swets &

Zeitlinger B.V; 1988.

[9] Boasen JF, McPherson RJ, Hays SL, Juul SE, Gleason CA. Neonatal stress or

morphine treatment alters adult mouse conditioned place preference.

Neonatology 2009;95:230–9.

[10] Centraal Bureau voor de Statistiek (CBS). CBS Standaard Beroepenclassiﬁcatie

1992; verbeterde editie 2001. Den Haag, Netherlands, 1992.

[11] Cheng AK, Rubin HR, Powe NR, Mellon NK, Francis HW, Niparko JK. Cost-utility

analysis of the cochlear implant in children. JAMA 2000;284:850–6.

[12] The CRIB. (Clinical Risk Index for Babies) score: a tool for assessing initial

neonatal risk and comparing performance of neonatal intensive care units. The

International Neonatal Network. Lancet 1993;342:193–8.

[13] de Keizer NF, Bonsel GJ, Gemke RJ. Health status prediction in critically ill

children: a pilot study introducing standardized health ratios. Qual Life Res

1997;6:192–9.

[14] Feeny D, Furlong W, Saigal S, Sun J. Comparing directly measured standard

gamble scores to HUI2 and HUI3 utility scores: group- and individual-level

comparisons. Soc Sci Med 2004;58:799–809.

[15] Feeny D, Furlong W, Torrance GW, Goldsmith CH, Zhu Z, DePauw S, Denton M,

Boyle M. Multiattribute and single-attribute utility functions for the health

utilities index mark 3 system. Med Care 2002;40:113–28.

[16] Goffaux P, Lafrenaye S, Morin M, Patural H, Demers G, Marchand S. Preterm

births: can neonatal pain alter the development of endogenous gating

systems? Eur J Pain 2008;12:945–51.

[17] Grunau RE, Holsti L, Haley DW, Oberlander T, Weinberg J, Solimano A,

Whitﬁeld MF, Fitzgerald C, Yu W. Neonatal procedural pain exposure predicts

lower cortisol and behavioral reactivity in preterm infants in the NICU. Pain

2005;113:293–300.

[18] Grunau RE, Oberlander TF, Whitﬁeld MF, Fitzgerald C, Lee SK. Demographic

and therapeutic determinants of pain reactivity in very low birth weight

neonates at 32 weeks’ postconceptional age. Pediatrics 2001;107:105–12.

[19] Grunau RE, Weinberg J, Whitﬁeld MF. Neonatal procedural pain and preterm

infant cortisol response to novelty at 8 months. Pediatrics 2004;114:e77–84.

[20] Grunau RE, Whitﬁeld MF, Davis C. Pattern of learning disabilities in children

with extremely low birth weight and broadly average intelligence. Arch

Pediatr Adolesc Med 2002;156:615–20.

[21] Grunau RE, Whitﬁeld MF, Petrie-Thomas J, Synnes AR, Cepeda IL, Keidar A,

Rogers M, Mackay M, Hubber-Richard P, Johannesen D. Neonatal pain,

parenting stress and interaction, in relation to cognitive and motor

development at 8 and 18 months in preterm infants. Pain 2009;143:138–46.

[22] Grunau RV, Whitﬁeld MF, Petrie JH, Fryer EL. Early pain experience, child and

family factors, as precursors of somatization: a prospective study of extremely

premature and fullterm children. Pain 1994;56:353–9.

[23] Handelmann GE, Dow-Edwards D. Modulation of brain development by

morphine: effects on central motor systems and behavior. Peptides

1985;6:29–34.

[24] Hermann C, Hohmeister J, Demirakca S, Zohsel K, Flor H. Long-term alteration

of pain sensitivity in school-aged children with early pain experiences. Pain

2006;125:278–85.

[25] Hilz MJ, Stemper B, Schweibold G, Neuner I, Grahmann F, Kolodny EH.

Quantitative thermal perception testing in 225 children and juveniles. J Clin

Neurophysiol 1998;15:529–34.

[26] Hu S, Sheng WS, Lokensgard JR, Peterson PK. Morphine induces apoptosis of

human microglia and neurons. Neuropharmacology 2002;42:829–36.

[27] Johnston CC, Stevens BJ. Experience in a neonatal intensive care unit affects

pain response. Pediatrics 1996;98:925–30.

[28] Ma MX, Chen YM, He J, Zeng T, Wang JH. Effects of morphine and its

withdrawal on Y-maze spatial recognition memory in mice. Neuroscience

2007;147:1059–65.

[29] MacGregor R, Evans D, Sugden D, Gaussen T, Levene M. Outcome at 5–6 years

of prematurely born children who received morphine as neonates. Arch Dis

Child Fetal Neonatal Ed 1998;79:F40–3.

[30] McPherson RJ, Gleason C, Mascher-Denen M, Chan M, Kellert B, Juul SE. A new

model of neonatal stress which produces lasting neurobehavioral effects in

adult rats. Neonatology 2007;92:33–41.

[31] Meier PM, Berde CB, DiCanzio J, Zurakowski D, Sethna NF. Quantitative

assessment of cutaneous thermal and vibration sensation and thermal pain

detection thresholds in healthy children and adolescents. Muscle Nerve

2001;24:1339–45.

[32] Narberhaus A, Lawrence E, Allin MP, Walshe M, McGuire P, Rifkin L, Murray R,

Nosarti C. Neural substrates of visual paired associates in young adults with a

history of very preterm birth: alterations in fronto-parieto-occipital networks

and caudate nucleus. Neuroimage 2009;47:1884–93.

[33] Perquin CW, Hazebroek-Kampschreur AA, Hunfeld JA, Bohnen AM, van

Suijlekom-Smit LW, Passchier J, van der Wouden JC. Pain in children and

adolescents: a common experience. Pain 2000;87:51–8.

[34] Pokela ML. Pain relief can reduce hypoxemia in distressed neonates during

routine treatment procedures. Pediatrics 1994;93:379–83.

[35] Purcell-Jones G, Dormon F, Sumner E. Paediatric anaesthetists’ perceptions of

neonatal and infant pain. Pain 1988;33:181–7.

[36] Quinn MW, Otoo F, Rushforth JA, Dean HG, Puntis JW, Wild J, Levene MI. Effect

of morphine and pancuronium on the stress response in ventilated preterm

infants. Early Hum Dev 1992;30:241–8.

[37] Quinn MW, Wild J, Dean HG, Hartley R, Rushforth JA, Puntis JW, Levene MI.

Randomised double-blind controlled trial of effect of morphine on

catecholamine concentrations in ventilated pre-term babies. Lancet

1993;342:324–7.

[38] Rao R, Sampers JS, Kronsberg SS, Brown JV, Desai NS, Anand KJ. Neurobehavior

of preterm infants at 36 weeks postconception as a function of morphine

analgesia. Am J Perinatol 2007;24:511–7.

[39] Rickards AL, Kelly EA, Doyle LW, Callanan C. Cognition, academic progress,

behavior and self-concept at 14 years of very low birth weight children. J Dev

Behav Pediatr 2001;22:11–8.

[40] Rietveld MJ, van Baal GC, Dolan CV, Boomsma DI. Genetic factor analyses of

speciﬁc cognitive abilities in 5-year-old Dutch children. Behav Genet

2000;30:29–40.

[41] Saigal S, Stoskopf B, Pinelli J, Streiner D, Hoult L, Paneth N, Goddeeris J. Self-

perceived health-related quality of life of former extremely low birth weight

infants at young adulthood. Pediatrics 2006;118:1140–8.

[42] Sargeant TJ, Miller JH, Day DJ. Opioidergic regulation of astroglial/neuronal

proliferation: where are we now? J Neurochem 2008;107:883–97.

[43] Schmelzle-Lubiecki BM, Campbell KA, Howard RH, Franck L, Fitzgerald M.

Long-term consequences of early infant injury and trauma upon

somatosensory processing. Eur J Pain 2007;11:799–809.

[44] Simons SH, Roofthooft DW, van Dijk M, van Lingen RA, Duivenvoorden HJ,

van den Anker JN, Tibboel D. Morphine in ventilated neonates: its effects

on arterial blood pressure. Arch Dis Child Fetal Neonatal Ed 2006;91:

F46–51.

[45] Simons SH, van Dijk M, van Lingen RA, Roofthooft D, Boomsma F, van den

Anker JN, Tibboel D. Randomised controlled trial evaluating effects of

morphine on plasma adrenaline/noradrenaline concentrations in newborns.

Arch Dis Child Fetal Neonatal Ed 2005;90:F36–40.

[46] Simons SH, van Dijk M, van Lingen RA, Roofthooft D, Duivenvoorden HJ,

Jongeneel N, Bunkers C, Smink E, Anand KJ, van den Anker JN, Tibboel D.

Routine morphine infusion in preterm newborns who received ventilatory

support: a randomized controlled trial. JAMA 2003;290:2419–27.

[47] Taddio A, Goldbach M, Ipp M, Stevens B, Koren G. Effect of neonatal

circumcision on pain responses during vaccination in boys. Lancet

1995;345:291–2.

[48] Taddio A, Katz J, Ilersich AL, Koren G. Effect of neonatal circumcision on

pain response during subsequent routine vaccination. Lancet 1997;349:599–

603.

[49] Taylor A, Fisk NM, Glover V. Mode of delivery and subsequent stress response.

Lancet 2000;355:120.

[50] Taylor HG, Espy KA, Anderson PJ. Mathematics deﬁciencies in children with

very low birth weight or very preterm birth. Dev Disabil Res Rev

2009;15:52–9.

[51] Walker SM, Franck LS, Fitzgerald M, Myles J, Stocks J, Marlow N. Long-term

impact of neonatal intensive care and surgery on somatosensory perception in

children born extremely preterm. Pain 2009;141:79–87.

[52] Weisglas-Kuperus N, Hille EE, Duivenvoorden HH, Finken MM, Wit JM, van

Buuren S, Goudoever JV. Verloove-Vanhorick PM; Dutch POPS-19

Collaborative Study Group. Intelligence of very preterm or very low birth

weight infants in young adulthood. Arch Dis Child Fetal Neonatal Ed

2009;94:F196–200.

[53] Williams J, Lee KJ, Anderson PJ. Prevalence of motor-skill impairment in

preterm children who do not develop cerebral palsy: a systematic review. Dev

Med Child Neurol 2010;52:232–7.

[54] Wood CM, Rushforth JA, Hartley R, Dean H, Wild J, Levene MI. Randomised

double blind trial of morphine versus diamorphine for sedation of preterm

neonates. Arch Dis Child Fetal Neonatal Ed 1998;79:F34–9.

[55] Wright AA, Zhang B, Ray A, Mack JW, Trice E, Balboni T, Mitchell SL, Jackson VA,

Block SD, Maciejewski PK, Prigerson HG. Associations between end-of-life

discussions, patient mental health, medical care near death, and caregiver

bereavement adjustment. JAMA 2008;300:1665–73.

J. de Graaf et al. / PAIN

152 (2011) 1391–1397

1397