Is the Newborn Individualized Developmental Care and Assessment Program (NIDCAP) effective for preterm infants with intrauterine growth restriction?

Abstract

Objective

This study investigates the effectiveness of the Newborn Individualized Developmental Care and Assessment Program (NIDCAP) on neurobehavioral and electrophysiological functioning of preterm infants with severe intrauterine growth restriction (IUGR).

Study Design

Thirty IUGR infants, 28 to 33 weeks gestational age, randomized to standard care (control/C = 18), or NIDCAP (experimental/E = 12), were assessed at 2 weeks corrected age (2wCA) and 9 months corrected age (9mCA) in regard to health, anthropometrics, and neurobehavior, and additionally at 2wCA in regard to electrophysiology (EEG).

Result

The two groups were comparable in health and anthropometrics at 2wCA and 9mCA. The E-group at 2wCA showed significantly better autonomic, motor, and self-regulation functioning, improved motility, intensity and response thresholds, and reduced EEG connectivity among several adjacent brain regions. At 9mCA, the E-group showed significantly better mental performance.

Conclusion

This is the first study to show NIDCAP effectiveness for IUGR preterm infants.

Introduction

Evidence indicates that the newborn intensive care unit (NICU) exerts deleterious sensory effects upon preterm infants’ immature brains and alters their subsequent development. Twenty percent of the 12.7% live births born prematurely in the United States each year¹ are intrauterine growth restricted (IUGR).² Placental insufficiency causes alterations in circulating fetal hormones, heightens such infants’ cortisol-mediated hypersensitivity, and is evident in significantly higher rates of mortality and morbidities including learning disabilities and school failure (>50%). IUGR prematurity is associated with considerable cost in terms of NICU care and often life-long medical, educational, and social resource requirements.²

The Newborn Individualized Developmental Care and Assessment Program (NIDCAP)³ constitutes a comprehensive effort to decrease secondary NICU effects on brain development. It has been successful in all but one⁴ randomized-controlled trial focused on appropriate-in-growth-for-gestational-age (AGA) preterm populations.^5–12 This study extends the investigation to a discreet population of highly vulnerable infants and tests the behavioral and electrophysiological effects of NIDCAP for severely IUGR preterm infants, 28 to 34 weeks gestational age (GA) at birth. The aim is to identify whether NIDCAP is ameliorative for infants already compromised by IUGR birth.

Methods

Study design

A controlled clinical trial was used with two-group randomization, blocked by gender and ethnicity (Caucasian/other). Immediately after obtaining consent, subjects were randomized to the experimental (E) or control (C) group. Randomization used opaque pre-numbered sealed envelopes, opened by the parents after consent.

Subjects

Thirty preterm infants with severe IUGR and their parent(s) were consecutively recruited from the NICU, Brigham and Women’s Hospital (BWH), Boston, an academic 48-bed level-III, inborn nursery with over 7000 births per year. The study protocol was approved by the institutional review boards of the BWH and the Children’s Hospital Boston.

Selection criteria included were as follows: For the mother—residence in Greater Boston; age ≥14 years; absence of major maternal medical/psychiatric illness, chronic medication treatment, and history of substance abuse; telephone accessibility; and some English language facility. For the infant—inborn at BWH; GA 28 to 33 weeks by estimated date of confinement based on second or first trimester fetal ultrasound, mother’s dates and/or Ballard GA assessment;¹³ IUGR as defined by birth weight and head circumference <5th percentile for GA (Gairdner’s and Pearson’s growth charts)¹⁴ because of placental insufficiency as documented by two or more of the following: maternal high blood pressure, oligohydramnios, absent or compromised Doppler diagnosed end diastolic flow velocity, and delivery initiated because of fetal growth restriction; singleton or twin with AGA sibling; 5-min Apgar ≥5; absence of major chromosomal or congenital anomalies, congenital infections, or diagnosed prenatal brain lesions; and deemed viable by the attending neonatologist.

Recruitment extended from January 1996 to May 2000. Seventy-five infants/families met the study criteria. Of these, 45 were approached (see Supplementary Figure 1) and 30 infants were enrolled; randomization yielded 18 C- and 12 E-group infants. Not participating eligible infants/families were comparable in all background criteria to those participating.

Control and experimental group experience

C-group infants received standard care practiced at the BWH NICU at time of study. This included effort at primary care nursing, staff-dependent parent inclusion, shielded incubators, side and foot rolls, pacifiers, and encouragement of skin-to-skin holding and breast feeding. Spill-over effects from E- to C-groups were not prevented systematically (for example a nurse involved in the care of an E-subject might also care for a C-subject). As such, any E-effects are conservative, as they are in excess of possible C-group contamination.

The E-group received NIDCAP intervention,³ which emphasizes each infant’s behavioral individuality. It takes into account infants’ current thresholds from behavioral organization to stress, and supports each infant’s strengths. The E-care team consisted of a developmental psychologist, a pediatrician, and certified and experienced NIDCAP Professionals.³ They provided weekly infant observations and daily contact with the NICU caregivers and parents to support understanding of the infants’ stress and comfort signals, to adjust care accordingly, and to ensure continuity and consistency of individualized care for each E-infant. Formal observations,^15,16 with neurobehavioral reports and support to individualized caregiving began with initial stabilization and ended at 2 weeks corrected age (2wCA). Families, valued as the infants’ most consistent and important nurturers, were supported to hold, and care for their infant, also during stressful and difficult procedures.¹⁷ Fidelity of intervention implementation was measured indirectly by the outcome effects obtained.

Outcome assessment

All outcome assessments and data abstraction from record review were performed by study personnel intentionally blinded to the infant’s group status.

Medical, anthropometric, and demographic background and outcome measures

Medical and anthropometric information from the infants’ charts was abstracted and coded by a study nurse into an a priori designed data base. A parent interviewer obtained demographic information. It was hypothesized that C- and E-groups would be comparable in medical, anthropometric, and demographic background. On the basis of published studies with AGA preterm infants of comparable GA,^10,18 intervention effects on medical and anthropometric outcomes were not expected; significant effects favoring the E-infants were hypothesized in neurobehavioral and electrophysiological functioning.

Neurobehavioral outcome measures at 2wCA

The assessment of preterm infants’ behavior (APIB)^16,19 (with documented sensitivity to GA and risk status)^18,20 and the Prechtl Neurological Examination of the Fullterm Newborn Infant (Prechtl)²¹ were administered at 2wCA by the certified APIB Professional and Prechtl evaluator. The APIB yields six system scores (autonomic, motor, state, attention, self-regulation, and examiner facilitation), used for analysis. They range from 1 (well modulated) to 9 (disorganized). The Prechtl was reduced to 12 summary variables,¹⁰ which document syndromes of reactivity and thresholds of functioning. In addition, eight APIB/Prechtl factor scores were generated for the study cohort using rules established on an independent sample of 312 preterm and fullterm infants.²⁰

Neurophysiologic outcome measures at 2wCA

At 2wCA, three C-infants were still hospitalized, leaving 15 C- and 12 E-infants for neuroelectrophysiological assessment. A registered electrophysiology (EEG) technologist obtained EEG data from 20 scalp electrodes with linked-ear reference. Analyses were limited to artifact-free segments of at minimum 240 s quiet sleep, identified by the senior electroencephalographer. Applying Laplacian reference techniques, spectral coherences between each pair of EEG electrodes (measures of cortical coupling between brain areas underlying the electrodes²²) were used. Forty coherence factors, derived from an independent normative sample (N = 312) studied at 2wCA,²⁰ were generated on the study cohort’s data and used for analysis.

Health, anthropometric, and neurobehavioral outcome measures at 9 months corrected age

At 9 months corrected age (9mCA), all study infants’ health status (C = 18; E = 12) was assessed based on parent interview, medical record, and anthropometrics (weight, length, head circumference, and corresponding percentiles for corrected age).¹⁴ A trained examiner assessed all infants with the Bayley Scales of Infant Development, second edition (Bayley-II),²³ which yield a mental (MDI) and psychomotor developmental index (PDI) (mean = 100; s.d. = 15), and a behavior rating scale (BRS) score (percentile).

Assessments were performed in the morning, in a quiet examination room, at 2wCA 1 h before an infant’s next feeding, at 9mCA at a time considered optimal by the parents. Parents were present for all examinations.

Data analysis

BMDP-2007 software²⁴ was used for intention to treat analysis. These included a priori multivariate analysis of variance (MANOVA) for medical and demographic background, and medical, anthropometric, behavioral and electrophysiological outcome variables, and post hoc univariate ANOVA for individual variable contribution. To account for unequal variances, the Browne–Forsythe test (F*) was used to compare means; Fisher’s exact test for categorical variables and 2 × 2-table proportions; Pearson’s χ² for all others.²⁴ A priori, a two-tailed probability level, P<0.05, was selected. A sample size of 18 C- and 12 E-group subjects provides 80% power at 9mCA for detecting mean differences of 12 points on Bayley-II MDI scores based on a two-group Satterthwaite t-test with unequal variances and two-sided α-level of 0.05 (nQuery Advisor 7.0, Statistical Solutions, Saugus, MA, USA). At 2wCA, Wilks’ λ²⁵ with jack-knifed classification^26,27 was used to establish two-group classification success based on medical outcome variables, behavioral factors, and spectral coherence factors. Canonical correlation analysis²⁸ was performed for relationship exploration of behavioral and electrophysiological domains.

Results

Medical and demographic background and outcome at 2wCA

The two groups were comparable on all medical, anthropometric, and demographic background variables (Supplementary Table 1). There was also no difference in C- and E-distribution of Doppler diagnosis (χ² = 1.67, normal; df = 3, P = 0.64; 5 reversed; 15 absent; 5 5 unavailable). An independent radiologist reported no abnormalities on any initial neurosonograms. Medical outcome also showed no significant group differences (Supplementary Table 2). Overall anthropometrics (Supplementary Table 3) favored the C-group and was accounted for by a trend toward smaller head circumference percentiles for the E-group.

Behavioral outcome at 2wCA

Corrected age at assessment was comparable for both groups [C: 21 days (±9 days); E: 19 days (±7 days); F = 0.59; df = 1,27; P = 0.45]. E-infants showed significantly better neurobehavioral performance than C-infants on three of six scores APIB systems, namely the autonomic, motor, and self-regulation systems¹⁹ (Supplementary Table 4). The Prechtl²⁹ overall did not differ between groups. Post hoc descriptive analysis of individual variables (Supplementary Table 5) revealed significantly improved E-group in motility as well as in intensity and in threshold of response, with a trend for the Prechtl Summary Score. The eight APIB/Prechtl factor scores (Supplementary Table 6) showed a trend favoring the E-group. Post hoc individual factor analysis significantly favored the E-group over the C-group in factor 3, an intensity and hypersensitivity factor.

Neurophysiological outcome at 2wCA

Discriminant function analysis using the 40 EEG spectral coherence factors identified four factors, which significantly differentiated the C-group from the E-group (Table 1). Jack-knifed classification success using the four factors showed 85.2% correct subject classification.^26,27 Misclassified were two C- and two E-subjects, for whom decision rules based on their spectral coherence factor values failed to correspond to their actual group assignment.

Table 1.

Discriminant function analysis of qEEG coherence measures, 2 weeks corrected age (2wCA)

Standard/jack-knifed classification matrix coherence factors 18, 35, 36, 38	Correct classification (%)	Control (C) (n = 15)	Experimental (E) (n = 12)
Control (C) group	86.7	13	2
Experimental (E) group	83.3	2	10
Total	85.2	15	12

Wilks’ λ = 0.4436; df = 4,22; F = 6.90; P = 0.001.

The four successfully discriminating factors depicted in Figure 1 indicate a pattern of decreased coherence in the slow θ frequency band in the frontal to pre-frontal regions (factor 38a), the central to right-parietal regions (factor 36), the occipital region to the left-anterior quadrant (factor 35), and in the broad-spectrum-β frequency band from the left-posterior-temporal to the right-occipital areas (factor 18). There were some short- and some long-distance connections, all with lower coherences in the E-group. Only one factor (38b) showed increased E-group coherence (a ‘long-distance’, across-midline right-frontal to left-parietal connection). Thus, the NICU intervention for this population primarily appears to have reduced connectivity between multiple brain regions.

Figure 1.

Electrophysiology (EEG) spectral coherence factors at 2 weeks corrected age (2wCA), control (C) (n = 15), experimental (E) (n = 12). Head shown in vertex view, nose above, left ear to left. EEG frequency and coherence electrodes shown above head. Arrow color illustrates experimental group coherence; green = decreased, red = increased.

Relationship of behavioral and spectral coherence measures at 2wCA

Canonical correlation between the eight APIB/Prechtl-behavioral factors and the 40 spectral coherence factors was performed in two steps because of sample size (first 1 to 20, then 21 to 40). Each step yielded highly significant results [(χ² = 649.41, df = 160, P<0.00001) and (χ² = 703.99, df = 160; P = 0.00001)], indicating a strong association between behavior and EEG coherence. In each step, two canonical variates were sufficient to express the relationships.

In step 1, coherence factor 9 (decreased coherence between the occipital and the parietal midline region at 10 Hz) was associated with behavior factors 2 and 6 (better motor system regulation). Coherence factor 2 (decreased coherence between the right central and the left central-temporal-parietal region at 10 Hz) and coherence factor 6 (increased coherence between the right-frontal and left-posterior temporal-occipital region at 8 to 12 Hz) were associated with better organization of the motor system.

In step 2, coherence factor 32 (increased coherence, between the midline occipital and the left anterior temporal, frontal, and frontopolar regions at 18 to 32 Hz) and coherence factor 35 (decreased coherence between the midline parietal and the left central regions at 6 Hz) were associated with behavior factor 7 (attention cues). Coherence factor 26 (increased coherence between the left posterior-temporal and the left-parietal regions at 8 to 24 Hz) and coherence factor 25 (decreased coherence between the midline central and the left midtemporal regions at 6 to 24 Hz) were associated with behavior factor 8 (motor system self-regulation cues). Figure 1 aids in interpretation of these findings [also Figure 1 in Duffy (2003)²⁰].

In summary, better motor system organization and attention expressivity were associated with increased connectivity between frontal and occipital regions, that is increased coherences among geographically distant brain regions (more mature function) and with reduced connectivity between parietal and occipital regions, that is coherences among geographically more closely adjacent brain regions (more immature function). Better attention expressivity was also associated with reduced connectivity between closely adjacent right midtemporal regions.

Health, anthropometric, and neurobehavioral outcome at 9mCA

At 9mCA, age (in months) at assessment was comparable for the two groups [C: 9.48 (0.47); E: 9.23 (0.35); F = 2.83; df = 1,28; P = 0.10]. There were no differences in medical well-being or community intervention services between the two groups (Supplementary Table 7). The two groups also were now comparable in all anthropometrics (Supplementary Table 8). They were well above birth percentiles (<5%), yet still below the 25% for age in weight and length, and the 50% for head circumference. Neurobehaviorally, the E-group showed significantly better Bayley-II performance than the C-group (Table 2). The E-group was on average 12 points above the C-group on MDI. Eighty-nine percent of C-infants, but only 42% of E-infants scored below the mean on MDI. The PDI scores were not significantly different between C- and E-groups, yet there was a trend toward more C- than E-children scoring below the PDI mean (78 versus 42%). The E-group scored significantly better than the C-group on the BRS total score; 58% of E-children, but only 17% of C-children scored ‘within normal limits’. Overall, results indicated significantly better neurobehavioral functioning for the E- than the C-groups.

Table 2.

Bayley Scales of Infant Development, second edition, 9 months corrected age^a (9mCA)

Variable	Control (C) (n = 18)	Experimental (E) (n = 12)	P
Mental developmental index
Mean (s.d.)	90.06 (11.33)	102.83 (10.99)	0.005
Number of children below/at  or above mean	16/2	5/7	0.01 b
Psychomotor developmental index
Mean (s.d.)	82.48 (20.56)	92.25 (20.33)	0.21
Number of children below/at or  above mean	14/4	5/7	0.06b
Behavior rating scale total score
Mean (s.d.)	21.26 (19.95)	34.08 (25.63)	0.17
Number of children within normal  limits/questionable/non-optimal	3/10/5	7/1/4	0.02 c

^aSummary analysis for mental developmental index, psychomotor developmental index, and behavior rating scale, total score.

Multivariate analysis of variance, F = 3.11, df = 3,26, P = 0.04.

Mental and psychomotor developmental index, mean = 100; s.d. = 15. Statistical analysis used is Brown–Forsythe univariate analysis of variance: F*, unless otherwise noted; two-tailed;

^bFisher’s exact test;

^cPearson’s chi-square test: χ². P = probability. All probabilities are two-tailed. Note: P-values in bold indicate significant differences <0.05.

Discussion

This is the first investigation of the effectiveness of NIDCAP on neurobehavioral and electrophysiological functioning of preterm infants with severe IUGR. The results consistently favored E- over C-infants and suggest that individualized experience before term is beneficial for brain function of infants with compromised intrauterine growth, even in the face of transiently somewhat lower head circumference percentiles. Although the advantages of NIDCAP for the IUGR E-group are not of the same magnitude as those published for AGA preterm infants born at comparable GA,^10,18 they nevertheless validate previous beneficial NIDCAP effects and broaden the generalizability of NIDCAP effectiveness to preterm infants with prenatal compromise. IUGR infants receiving NIDCAP functioned better at 2wCA in terms of autonomic and motor system modulation as well as self-regulation, all presumably adaptive and advantageous in engagement with parents, professional caregivers, and the environment. EEG coherence results indicated a pattern of decreased connectivity in multiple adjacent brain regions. Only one EEG factor (a long-distance and across-midline right-frontal to left-parietal factor) showed increased connectivity in the E-group. Without the experimental intervention, IUGR infants appear to show excessive cortical short-distance connectivities, possibly induced by the multiple pathological influences generated under IUGR conditions. The intervention appears to ameliorate such excessive coupling. The demonstrated aberrant IUGR preterm connectivity pattern differs from the pattern of increased anterior to posterior connectivity reported in AGA preterm infants who experienced NIDCAP.¹⁸ EEG-behavior correlations at 2wCA showed that better motor system organization and attention expressivity were associated with increased frontal-occipital and reduced parietal-occipital connectivity. Similar to the AGA sample,¹⁸ IUGR results also showed that better attention expressivity was associated with reduced connectivity between adjacent right midtemporal regions. The AGA sample¹⁸ showed a pattern of association of mature behavioral functions (motor system and attention expressivity) with increased ‘long-distance’ and decreased ‘short-distance’ connections, whereas the IUGR sample showed connectivities that appeared to encompass much more limited areas of brain.¹⁸ It may be hypothesized that the IUGR infants mature more slowly and possibly with less differentiation than the AGA infants. One exception to the IUGR pattern was a specific short-distance and unexpectedly increased connection in the left posterior region, indicative of heightened posterior secondary association cortical connectivity, which correlated with improved attention expressivity for the E-group infants despite their IUGR status.

E-group functioning at 9mCA with significantly higher Bayley Scale mental (MDI) and behavioral (BRS) and a trend to better psychomotor (PDI) performance may indicate that NIDCAP contributes to more long-term developmental improvement in IUGR preterm infants.

Although the specific mechanisms are unclear, NIDCAP may involve prevention or reduction in the resetting of the N-methyl-d-aspartate axis associated with inappropriate cell death and increased neurocitotoxic damage, lowered sensory and pain thresholds, and increased hyper-reactivity and hypersensitivity may be the result of iatrogenic damage associated with the setting of traditional NICU care.³⁰ NIDCAP furthermore may prevent the abrupt blood flow velocity changes that have been shown with standard NICU care procedures and implicated in the diffuse brain damage sustained by preterm NICU patients.³¹

Small sample size, resulting from strict selection criteria, significant parent stress at a preterm-severe-IUGR birth, and lack of staff because of funding shortage, is a notable study limitation. Further limitations include the short-term outcome to 9mCA and the restriction of outcome to behavioral and electrophysiological brain function measures. Although MRI studies have shown substantial differences between the brain structures of IUGR and AGA preterm infants who have reached term,^32–34 this methodology was not available for this study. The impact of care practice changes, including enforcement of ‘back to sleep’ recommendations,³⁵ increasing life stresses on young families (such as dual parent long work hours), and lack of affordable child care, warrant further investigations of their impact on IUGR preterm development. Replication with a larger sample, addition of brain structural measures, and follow-up to later ages is required for result validation and generalizability. The optimal starting time for the proposed experimental intervention cannot be determined from this study, which was initiated within 72 h after birth. The most vulnerable period for a high-risk IUGR preterm infant likely is the time between birth and 48 h, when blood flow velocity changes are at their most extreme. Further research must explore NIDCAP effectiveness in the delivery room and immediately after birth. Lastly, given the comparatively reduced magnitude of NIDCAP effectiveness for IUGR preterms compared with AGA preterms,^10,18 it will be important to extend the NIDCAP intervention for IUGR infants beyond the NICU and the 2wCA point into preschool and school age. An approach to support IUGR preterms after discharge, such as the Infant Behavioral Assessment and Intervention Program,^36–38 might further strengthen the effectiveness of the intervention for preterm populations already compromised in the womb.

Supplementary Material

01

Supplementary Tables Table 1. Medical, Anthropometric and Demographic Background Variables^a

Table 2. Medical Variables, 2 Weeks Corrected Age (2wCA)

Table 3. Anthropometric Variables (Absolute and Percentiles), 2 Weeks Corrected Age (2wCA)

Table 4. Assessment of Preterm Infants’ Behavior (APIB) System Scores, 2 Weeks Corrected Age (2wCA)

Table 5. Prechtl Neurological Examination Summary and Total Scores, 2 weeks Corrected Age (2wCA)

Table 6. Assessment of Preterm Infants’ Behavior (APIB) and Prechtl Factor Scores, 2 Weeks Corrected Age (2wCA)

Table 7. Medical Variables, 9 Months Corrected Age^a (9mCA)

Table 8. Anthropometric Variables, 9 Months Corrected Age^a (9mCA)

Supplementary Figure Figure 1. Flow Chart of subject enrollment and follow-up