Developmental Trajectories in High-Risk NICU Graduates During the First Year of Life

Abstract

Objective:

We examined whether early medical factors predicted variability in developmental level and trajectories in high-risk neonatal intensive care unit (NICU) graduates during the first year of life.

Method:

Infants (n=53) who met criteria for the High-Risk Infant Follow-up Program were enrolled. Simple linear models predicted 12-month developmental abilities and linear mixed models predicted 6- to 12-month trajectories based on length of NICU stay and birthweight.

Results:

Length of NICU stay was more clearly predictive of 12-month developmental level, while birthweight better explained variability in trajectories. Communication and daily living skills varied based on early medical factors, whereas social skills did not. Motor skills varied by length of stay but not birthweight.

Conclusion:

Results support the need for close developmental monitoring of NICU graduates, particularly those with lengthier stays and lower birthweight. Developmental support needs should be based not just on current delays, but on the rate at which infants learn new skills.

Introduction

Infants who experience extended neonatal intensive care unit (NICU) hospitalizations are at increased risk for neurodevelopmental challenges, including developmental delays, autism, and behavior problems^1–5. There is wide variability in neurodevelopmental outcomes, with some children meeting or exceeding developmental expectations and others falling far behind their peers^6–8. There is a critical need to better understand infant development in the context of early medical risk to determine who is most in need of intervention and to inform programs that may serve to ameliorate adverse neurodevelopmental outcomes. To this end, the current study examined developmental trajectories in relation to early medical factors during the first year of life in a cohort of high-risk NICU graduates.

There have been a number of studies that have identified characteristics of newborn behavior, family factors, and medical concerns during the NICU stay that predict neurodevelopmental outcomes within specific early medical risk groups. In a multi-site study of children born at under 30 weeks gestation, high behavioral risk (based on NICU Network Neurobehavioral Scales profile) and higher medical risk (defined by the number of comorbidities) predicted lower cognitive and motor abilities on the Bayley Scales of Infant Development and higher problem behaviors on the Child Behavior Checklist at age two⁹. A novel study using machine learning methods in a large cohort of very preterm infants in Sweden further found that cognitive delay at age two could be reasonably predicted by factors such as length of NICU stay, birthweight, and family language¹⁰. While these large studies provide insight into important predictors of neurodevelopmental outcomes, they often lack a more detailed analysis of early developmental trajectories that would increase our understanding of how these differences unfold over time.

The current study reports on a cohort of infants recruited from a High-Risk Infant Follow-up Clinic whose development is being followed closely over the first two years of life. We sought to describe and explain variability in developmental abilities from 6 to 12 months of age in these infants with complex medical histories. As this study was conducted during the COVID-19 pandemic, we utilized the parent interview version of the Vineland Adaptive Behavior Scales, 3^rd Edition (Vineland-3)¹¹, which can be collected remotely, to measure the acquisition of important developmental skills. We asked two primary research questions related to the role of key medical factors (length of NICU stay, birthweight) in the early development of high-risk NICU graduates. 1) In line with previous work, we first examined the degree to which length of NICU stay and birthweight predicted the level of developmental abilities at 12 months of age. 2) In a unique extension of the existing research base, we then analyzed how these early medical factors predicted the rate of developmental gains from 6 to 12 months. We hypothesized that infants with longer NICU hospitalizations and lower birthweight would demonstrate lower developmental abilities at age one and slower growth in these abilities over the first year of life.

Method

Participants

Participants were enrolled in a longitudinal study of brain and behavioral development in high-risk NICU graduates, the Neurodevelopment and Early Social-Emotional Trajectories in NICU Graduates (NESTING) study. All infants (n=53) met criteria for the High-Risk Infant Follow-up (HRIF) Program as defined by California Children’s Services. Infants who were born at < 32 weeks and/or ≤ 1500 g automatically qualified for HRIF. Additional qualifying characteristics included neurological injuries, cardiac issues, and other medical concerns and interventions that indicated increased medical risk. Additional eligibility criteria for the NESTING study included English proficiency in the family and no major sensory impairment in the infant.

Infants were recruited directly from a university-based HRIF Clinic. A research associate approached all potentially eligible families during their initial visit and verbally described the study and handed them a brochure. Interested families provided their contact information and a team member followed up by phone or email to undergo screening and schedule their first visit. All participants were enrolled by 6 months of age (corrected age used for scheduling if gestational age [GA] < 37 weeks). Two sets of twins (one monozygotic, one dizygotic), in which both infants met HRIF criteria, were included. For a third set of enrolled twins, only the twin who met HRIF criteria was included in the current sample. Eligibility for the present study required at least one time point with Vineland-3 data, which all study participants met. Participants were roughly even with regard to gender, with diverse representation related to ethnicity, race, and other socioeconomic characteristics (see Table 1). Medical experiences also varied substantially across the sample (see Table 2).

Table 1.

Demographic Characteristics

Variable	n (%)
Gender
Female	25 (47.2%)
Male	28 (52.8%)
Ethnicity
Hispanic	25 (47.2%)
Not Hispanic	28 (52.8%)
Race
Asian/Pacific Islander	7 (13.5%)
Black/African-American	10 (19.2%)
White	30 (57.7%)
Multi-racial	5 (9.6%)
Family Status
Single	7 (13.2%)
Living Together	7 (13.2%)
Married	36 (67.9%)
Other/Separated	3 (5.7%)
Maternal Education
High School	8 (15.1%)
Some College	16 (30.2%)
College Degree	17 (32.1%)
Graduate/Professional Degree	12 (22.6%)
Paternal Education
High School	19 (37.2%)
Some College	8 (15.1%)
College Degree	12 (22.6%)
Graduate/Professional Degree	12 (22.6%)
Family Income
Under $25,000	9 (17.6%)
$25,000-$49,999	9 (17.6%)
$50,000-$74,999	9 (17.6%)
$75,000-$124,999	6 (11.8%)
$125,000+	18 (35.3%)
Insurance Type
Private	26 (49.1%)
Public	27 (50.9%)

Note. Race not reported for one Hispanic participant. “Maternal” education includes one father from a male same-sex parent family and “paternal” education includes one mother from a female same-sex parent family. Paternal education unavailable for two single-parent families. Family income not reported for two participants.

Table 2.

Medical Characteristics

Variable	M (SD)	Range
Gestational age (weeks)	33.01 (5.26)	23.29–41.00
Birthweight (grams)	1872.91 (962.56)	555–3843
z-Score for birthweight	−1.268 (1.95)	−5.817–2.311
Length of NICU stay (days)	58.17 (44.57)	5–176
Apgar Score
1 minute	5.38 (2.63)	1–9
5 minutes	7.27 (2.10)	2–9
Variable	n (%)
Neurological issue	22 (41.5%)
Preterm status
Extremely preterm	10 (18.9%)
Very preterm	14 (26.4%)
Moderately preterm	4 (7.5%)
Late preterm	9 (17.0%)
Full term	16 (30.2%)
HRIF Primary Qualifying Condition
Preterm (<32 weeks)	24 (45.3%)
Low Birthweight (≤1500 g, ≥32 weeks)	8 (15.1%)
Major Neurological Issue (>1500 g, ≥32 weeks)	10 (18.9%)
Cardiac (No Major Neuro, >1500 g, ≥32 weeks)	6 (11.3%)
Other (No Major Neuro/Cardiac, >1500 g, ≥32 weeks)	5 (9.4%)

Note. Apgar scores missing for one participant. Neurological issues included intraventricular hemorrhage (n=8), hypoxic-ischemic encephalopathy (n=8), seizures (n=5), germinal matrix hemorrhage (n=3), meningitis (n=2), and agenesis of the corpus collosum (n=1). Infants were placed into only one HRIF group based on priority order above (e.g., 31-week infant who weighed 1300 g placed in preterm group). “Other” group comprised of infants who had severe respiratory issues, pulmonary hypertension, hernia, necrotizing enterocolitis, sepsis, and severe hypoglycemia. One infant in preterm group and three in cardiac group had a genetic diagnosis of Trisomy 21.

Procedure

This study used a hybrid data collection design to minimize family burden and health risks during the COVID-19 pandemic. Data from the 6-, 9-, and 12-month visits were included. Data collection for the current study spanned from August 2021 to October 2023. The 6- and 9-month visits were fully remote and conducted via Zoom by a research associate. The 12-month visit had remote and in-person components. The Vineland-3 was collected by a trained clinician as part of the Zoom visit. On a few occasions (due to family preference or internet connection issues), the interview was completed over the phone or at the in-person visit. Family demographic and medical information was collected via parent interview at the 6-month visit. Additional information on the NICU stay was collected via medical chart review. Informed consent for study involvement and a HIPAA consent form to access medical records were obtained from parents for all participants. The NESTING study has current approval through the UCLA Medical Institutional Review Board 3. The study was performed in accordance with the ethical principles detailed in the Declaration of Helsinki.

Measures

Vineland Adaptive Behavior Scales, 3^rd Edition (Vineland-3).

The Vineland-3¹¹ is a norm-referenced measure of adaptive skills across the lifespan. The parent interview version was utilized to increase the reliability of information gained via parent report. This measure yields standard scores (M=100, SD=15) in Communication, Daily Living Skills, Socialization, and Motor Skills, as well as an Adaptive Behavior Composite (ABC) based on the first three subdomains. For preterm infants (GA < 37 weeks), scores were calculated using expected due date rather than birthdate.

Medical and demographic data.

At the 6-month visit, information related to family demographic and infant medical characteristics were collected via parent interview. Additional information and confirmation of parent-reported data were obtained via each infant’s NICU discharge summary obtained through their medical record. Medical variables of particular interest included length of NICU hospitalization (days), birthweight (grams), and gestational age at birth (weeks). Given the very high correlation between gestational age and birthweight, r=.90, we did not include both variables in analyses. Based on prior literature^8,10,12, and in an attempt to reduce overlap with length of NICU stay (r=−.401 with birthweight vs. r=−.610 with gestational age), our analysis focused on length of NICU stay and birthweight.

For descriptive purposes, children were also grouped according to their primary HRIF qualifying criteria (listed in priority order): preterm birth (< 32 weeks), low birthweight (≥ 32 weeks, ≤ 1500g), major neurological issue (e.g., hypoxic ischemic encephalopathy; ≥ 32 weeks, > 1500g), major cardiac issue (e.g., congenital heart disease; ≥ 32 weeks, > 1500g, no major neurological diagnosis), or other medical issue (e.g., severe hypoglycemia, necrotizing enterocolitis; ≥ 32 weeks, > 1500g, no major neurological or cardiac diagnosis; see Table 2).

Analytic Plan

First, descriptive statistics and graphical summaries were conducted to assess for outliers and violations of model assumptions. Then, we assessed for differences in Vineland-3 scores based on demographic variables to determine inclusion of potential covariates. Preliminary analyses based on select medical variables were also conducted to better describe the sample.

Our main analyses aimed to understand the impact of two primary predictors, the duration of infants’ NICU stay (i.e., Days in NICU) and Birthweight, on the Vineland-3 Adaptive Behavior Composite Standard Score (ABC SS) and Motor Skills Standard Score (Motor SS). To facilitate interpretability within our modeling framework, both predictors were mean-centered in the reported models. Our primary outcome of interest, ABC SS, is comprised of three subscales: Communication Standard Score (Comm SS), Daily Living Skills Standard Score (DLS SS), and Socialization Standard Score (Social SS). Our secondary outcome variable, Motor SS, is not a part of the ABC SS.

Correlations of the two primary predictors with outcomes at 12 months were assessed using simple linear regression models via the lm function in R. Subsequently, effects of the two predictors on longitudinal trajectories of the outcomes were assessed using linear mixed models (LMMs), including the main predictor of infant age (at 6 months, 9 months, and 12 months), predictor-by-age interactions, and subject-level random intercepts. This analysis was implemented using the lmer function in R. The LMM approach accounts for correlations between repeated measures within subjects via the subject-level random intercepts. Furthermore, LMMs can automatically handle missing data, thereby ensuring unbiased estimates under the assumption that observations are missing at random. Consequently, all available observations from each subject are utilized in modeling via the LMM. Data from all 53 children were included in these models, with Vineland-3 data available at all three time points for 47 of 53 children (88.7%). Models were conducted with and without both members of the two twin pairs included. Results did not change substantially, so all participants were included in reported analyses.

Results

Preliminary Analyses

ABC SS and Motor SS did not differ significantly across ages measured based on any demographic variable, including gender (ps=0.222–0.933), ethnicity (ps=0.171–0.691), race (ps=0.201–0.951), maternal education (ps=0.206–0.916), paternal education (ps=0.319–0.650), family income (ps=0.491–0.878), family status (ps=0.335–0.894), or type of insurance (ps=0.204–0.868). In terms of medical variables, Vineland-3 scores did not differ significantly based on the presence/absence of a diagnosed neurological issue during the NICU stay (ps=0.117–0.944). Differences in ABC SS (p=0.015) and Motor SS (p=0.031) based on preterm status were observed at 6 months, with extremely preterm infants scoring lowest in motor skills and very preterm infants scoring highest in overall developmental abilities at this age. To better characterize differences in the Days in NICU variable, we compared infants’ length of stay across HRIF primary qualifying condition groups. There was an overall difference (p<.001), with infants in the preterm group staying the longest on average (M=82.75, SD=40.67, 34–163), followed by other (M=81.40, SD=69.11, 17–176) and cardiac (M=47.00, SD=26.78, 11–78 days), with the low birthweight (M=22.75, SD=6.41, 14–30 days) and major neurological injury (M=22.60, SD=13.52, 5–40 days) groups having the lowest mean stays. Days in NICU did not differ by the presence/absence of a neurological diagnosis during the NICU stay (p=.457).

Predictors of 12-Month Developmental Abilities

Days in NICU.

See Supplemental Table 1 and Supplemental Figure 1 for detailed results. Our analysis revealed a significant negative correlation between Days in NICU and 12-month ABC SS, r(45)=−0.35, p=0.018, such that infants who experienced longer NICU stays exhibited lower overall developmental abilities at 12 months old. This correlation predominantly stemmed from the negative associations observed between Days in NICU and two of the three ABC SS subscales: Comm SS, r(45)=−0.31, p=0.032, and DLS SS, r(45)=−0.47, p<0.001. No correlation was identified between Days in NICU and Social SS, r(45)=−0.08, p=0.606.

Additionally, a significant negative correlation was observed between Days in NICU and the secondary outcome variable, 12-month Motor SS, r(45)=−0.48, p<0.001, indicating that infants who experienced longer NICU stays had lower motor skills at 12 months. To better understand this finding, an exploratory follow-up analysis of Motor subscales (Gross Motor and Fine Motor v-scale scores) was conducted, indicating that this finding was driven by moderate negative correlations with both gross motor, r(47)=−0.47, p<0.001, and fine motor skills, r(47)=−0.42, p<0.001 at 12 months.

Birthweight.

A different pattern of results was found for infant birthweight. Birthweight was not significantly correlated with any of the Vineland-3 scores at 12 months, including ABC SS, r(45)=0.24, p=0.105, and Motor SS, r(45)=0.12, p=0.431. However, a trend-level positive correlation was observed for Comm SS, such that infants with higher birthweights tended to have higher communication skills at 12 months, r(45)=0.29, p=0.052. Results for the other subscales did not approach significance (DLS SS: r(45)=0.20, p=0.178; Social SS: r(45)=0.07, p=0.661). See Supplemental Table 2 and Supplemental Figure 2 for detailed results.

Predictors of 6- to 12-Month Developmental Trajectories

Individual and mean-level developmental trajectories are displayed in Figure 1 and detailed in Supplemental Table 3. Overall, Vineland-3 scores stayed relatively stable over time, with notable variability observed among individual trajectories. Mean scores for the current sample all fell below the standard score mean of 100 (based on norming sample), but mostly fell within one standard deviation of this mean.

Figure 1. Raw individual and sample-level Vineland-3 trajectories.

Dashed lines indicate standard score mean and area within one standard deviation of the mean based on the Vineland-3 norming sample. Middle graph depicts mean Vineland-3 composite scores by age and domain, with error bars representing standard deviation.

Days in NICU.

See Table 3 for detailed results of LMMs. Analysis of longitudinal trajectories revealed a marginally significant interaction effect of Days in NICU and Age on ABC SS. Infants who required shorter NICU stays tended to show greater gains in developmental abilities over time. For instance, infants with a NICU stay of 28 days (1^st quantile of Days in NICU) demonstrated an estimated mean ABC SS of 91.143 and 93.267 at 6 months and 12 months, respectively, while those with a stay of 85 days (3^rd quantile of Days in NICU) showed estimated mean scores of 88.684 and 86.713 at the same time points. Follow-up analyses of ABC SS subscales identified a significant interaction effect of Days in NICU and Age in Comm SS and DLS SS. Infants with shorter NICU stays exhibited faster acquisition of communication and daily living skills as they aged compared to infants with longer stays. For instance, infants with a NICU stay of 28 days had an estimated mean Comm SS of 93.610 and 101.097 at 6 and 12 months, respectively, while those with a stay of 85 days showed estimated mean scores of 92.450 and 94.167 at the same time points. Similarly, infants with a NICU stay of 28 days showed an estimated mean DLS SS of 92.282 and 92.920 at 6 and 12 months, respectively, while those with a stay of 85 days demonstrated estimated mean scores of 87.688 and 82.825 at the same time points. No main or interaction effects were found for Days in NICU on Social SS.

Table 3.

Results from linear mixed models (LMMs) using Days in NICU as the primary predictor of Vineland-3 longitudinal outcomes from 6, 9 and 12 months

Main outcome	ABC SS
	Mean					SD					F-statistic & p-value
(Intercept)	89.885					2.797					F1,141=1032.915, p<0.001
Age	−0.007					0.271					F1,97=0.001, p=0.979
Days in NICU	1.278					2.794					F1,141=0.209, p=0.648
Days in NICU * Age	−0.534					0.270					F1,97=3.916, p=0.051
Subscales of ABC SS	Comm SS					DLS SS					Social SS
	Mean	SD		F-statistic & p-value		Mean	SD		F-statistic & p-value		Mean	SD	F-statistic & p-value
(Intercept)	88.563	3.448		F1,135=659.771 p<0.001		92.124	3.181		F1,138=838.855 p<0.001		96.634	2.820	F1,128=1150.024 p<0.001
Age	0.739	0.346		F1,98=4.559 p=0.035		−0.379	0.315		F1,97=1.452 p=0.231		−0.541	0.290	F1,96=3.484 p=0.065
Days in NICU	3.606	3.442		F1,135=1.098 p=0.297		0.709	3.176		F1,138=0.050 p=0.824		−1.131	2.814	F1,128=0.162 p=0.688
Days in NICU * Age	−0.752	0.344		F1,98=4.785 p=0.031		−0.717	0.313		F1,97=5.265 p=0.024		−0.021	0.288	F1,96=0.005 p=0.943
Secondary outcome	Motor SS
	Mean					SD					F-statistic & p-value
(Intercept)	90.206					3.389					F1,145=708.624, p<0.001
Age	−0.377					0.311					F1,95=1.472, p=0.228
Days in NICU	0.033					3.388					F1,145=0.0001, p=0.992
Days in NICU * Age	−0.787					0.309					F1,95=6.500, p=0.012
Motor SS Subscales	Gross Motor							Fine Motor (VS subscale score)
	Mean		SD		F-statistic & p-value			Mean		SD		F-statistic & p-value
(Intercept)	14.014		0.769		F1,143=332.005, p<0.001			12.544		0.659		F1,135=1150.363, p<0.001
Age	−0.180		0.073		F1,95=6.012, p=0.016			0.063		0.066		F1,97=0.912, p=0.342
Days in NICU	0.781		0.768		F1,143=1.034, p=0.311			−0.317		0.658		F1,136=0.233, p=0.630
Days in NICU * Age	−0.251		0.073		F1,95=11.844, p<0.001			−0.073		0.065		F1,97=1.229, p=0.270

Note. Outcome variables measured via Vineland-3 parent interview. LMMs used all available data to obtain estimates. ABC SS = Adaptive Behavior Composite Standard Score. Comm SS = Communication Standard Score. DLS SS = Daily Living Skills Standard Score. Social SS = Socialization Standard Score. Motor SS = Motor Skills Standard Score. SD = Standard Deviation.

There was a significant interaction effect of Days in NICU and Age on Motor SS, indicating that infants with shorter NICU stays learned new motor skills more quickly as they grew older, compared to infants with longer stays. For example, infants with 28 days in the NICU had an estimated mean Motor SS of 91.116 and 92.051 at 6 and 12 months, respectively, while those with 85 days in the NICU showed scores of 85.118 and 80.014 at the same time points. To follow up on this significant result, we reproduced these models using the Gross Motor and Fine Motor subscale scores, finding that differential changes in gross motor skills over time drove the above interaction. Specifically, we observed a significant interaction effect of Days in NICU and Age on the Gross Motor subscale, while no significant interaction effect was found on the Fine Motor subscale. According to the LMM analysis, infants with shorter NICU stays demonstrated quicker development of gross motor skills as they grew older, compared to infants with longer stays. For instance, infants with 28 days in the NICU had an estimated mean Gross Motor v-scale score of 13.431 and 14.104 at 6 and 12 months, respectively, while those with 85 days in the NICU had scores of 12.468 and 12.584 at the same time points.

Birthweight.

See Table 4 for detailed results. Results of the LMM indicated a negative main effect of Birthweight on ABC SS and a significant interaction effect between Birthweight and Age. Infants with lower birthweights had slightly higher developmental skills at 6 months, but demonstrated slower improvements in developmental skills over time than higher birthweight infants. For example, infants with a birthweight of 2814 grams (3^rd quantile of birthweight) had estimated mean ABC SS of 88.746 and 93.062 at 6 and 12 months, respectively, while those with a birthweight of 1200 grams (1^st quantile of birthweight) exhibited scores of 90.609 and 87.438 at the same time points.

Table 4.

Results from linear mixed models (LMMs) using Birthweight as the primary predictor of Vineland-3 longitudinal outcomes from 6, 9 and 12 months

Main outcome	ABC SS
	Mean			SD			F-statistic & p-value
(Intercept)	89.882			2.796			F1,143=1033.430, p<0.001
Age	−0.008			0.267			F1,96=0.001, p=0.975
Birthweight	−5.576			2.794			F1,143=3.982, p=0.048
Birthweight * Age	0.744			0.265			F1,96=7.874, p=0.006
Subscales of ABC SS	Comm SS			DLS SS			Social SS
	Mean	SD	F-statistic & p-value	Mean	SD	F-statistic & p-value	Mean	SD	F-statistic & p-value
(Intercept)	88.538	3.440	F1,136=662.290 p<0.001	92.150	3.277	F1,142=790.959 p<0.001	92.637	2.820	F1,128=1150.363 p<0.001
Age	0.740	0.343	F1,98=4.646 p=0.034	−0.383	0.316	F1,97=1.470 p=0.228	−0.542	0.289	F1,96=3.506 p=0.064
Birthweight	−6.214	3.436	F1,137=3.271 p=0.073	−5.330	3.274	F1,142=2.650 p=0.106	−1.512	2.815	F1,129=0.289 p=0.592
Birthweight * Age	0.894	0.341	F1,97=6.859 p=0.010	0.712	0.314	F1,96=5.157 p=0.025	0.222	0.288	F1,96=0.593 p=0.443
Secondary outcome	Motor SS
	Mean			SD			F-statistic & p-value
(Intercept)	90.191			3.564			F1,144=640.292, p<0.001
Age	−0.376			0.318			F1,95=1.399, p=0.240
Birthweight	−5.058			3.566			F1,144=2.012, p=0.158
Birthweight * Age	0.512			0.316			F1,95=2.632, p=0.108

Note. Outcome variables measured via Vineland-3 parent interview. LMMs used all available data to obtain estimates. ABC SS = Adaptive Behavior Composite Standard Score. Comm SS = Communication Standard Score. DLS SS = Daily Living Skills Standard Score. Social SS = Socialization Standard Score. Motor SS = Motor Skills Standard Score. SD = Standard Deviation.

Follow-up LMMs suggested that this interaction effect was driven by two of the ABC subscales, Comm SS and DLS SS. Infants with higher birthweights exhibited greater improvements in their communication and daily living skills compared to infants with lower birthweights as they grew older. For instance, infants with a birthweight of 2814 grams exhibited an estimated mean Comm SS of 92.141 and 101.819 at 6 and 12 months, respectively, while those with a birthweight of 1200 grams demonstrated estimated mean scores of 93.571 and 94.260 at the same time points. Similarly, infants with a birthweight of 2814 grams showed an estimated mean DLS SS of 88.821 and 90.704 at 6 and 12 months, respectively, while those with a birthweight of 1200 grams had estimated mean scores of 90.594 and 85.311 at the same time points. No main or interaction effects were found for birthweight on Social SS.

An additional LMM analyzed the role of birthweight in relation to growth in motor skills over time. No significant main or interaction effects were found for Birthweight on Motor SS.

Discussion

The current study examined the role of key medical risk factors—length of NICU hospitalization and infant birthweight—in predicting the level and change in developmental trajectories in high-risk NICU graduates over the first year of life. Interestingly, we found a somewhat different pattern of results for each predictor. The length of the NICU stay was more clearly predictive of the overall level of developmental abilities at 12 months, while birthweight held more significance in explaining variability in developmental trajectories from 6 to 12 months. When looking more in depth at individual developmental abilities, we found that communication and daily living skills appeared to be dependent on early medical factors, whereas social skills seemed largely independent of these medical indices. Variability in the overall level and trajectory of motor skills was explained by length of NICU stay but not birthweight.

Infants who experience extended NICU hospitalizations have a very different early life experience. The NICU environment often limits the close physical contact that typically characterizes the first months of life. NICU patients also experience a very different sensory environment in the hospital versus being at home or in the womb^13,14. Experimental studies show promise in modifying the NICU environment to enhance infants’ social and sensory experiences to optimize development^15–17, although barriers exist to fully implementing these programs^18–20, and they are unlikely to fully remediate the impact of these risk factors.

Consistent with prior work^10,21, we found that infants who spent more days in the NICU had lower overall parent-reported developmental and motor skills at 12 months. We also detected a trend-level interaction, in which infants who required longer NICU stays had slower growth in developmental abilities from 6 to 12 months, which was largely driven by slower acquisition of communication and daily living skills. This interaction effect was clearest for motor skills—a finding that was driven by slowed gross motor skill development—suggesting that longer NICU stays, during which infant movement is necessarily limited, may have a particularly notable effect on infants’ motor development. These motor skill limitations may then have downstream effects on other areas, such as social-communication and cognitive skills.

While it is likely that the variability in developmental outcomes associated with the duration of NICU hospitalization is at least in part explained by differences in early life experiences, a longer stay also suggests increased medical complexity, which can have a direct or indirect impact on brain development and neurodevelopmental outcomes⁹. In our sample, infants had a wide range in the duration of their NICU stay (5–176 days). Infants with longer hospitalizations tended to be infants with lower gestational ages and lower birthweights who were observed to have several co-occurring medical complications, though they were not more likely to have neurological diagnoses in the NICU. Survival and physical health must remain the priority within the NICU, and these needs will continue to require lengthy hospitalizations for many neonates. However, a recent study analyzing contributors to length of NICU stay found that while medical indicators (birthweight and number of days with ventilator support) accounted for most of the variance in the duration of NICU stays, infants who had more visits per day had shorter stays, suggesting the potential for decreasing hospitalization length through increased family support and outreach²². Our results further support the need for enhanced developmental care in the NICU, along with close monitoring and very early intervention supports once discharged, particularly for infants with longer stays.

Low birthweight is a well-documented risk factor for neurodevelopmental conditions, including autism and developmental delay^3,12. The potential predictive power of infant birthweight on early developmental trajectories is less well studied. In the current study, infant birthweight did not significantly predict variability in the overall level of developmental abilities at 12 months of age in our high-risk NICU graduates, but it did predict the rate at which infants gained new skills. Lower birthweight infants started at a similar level at 6 months (after adjusting for preterm birth) but demonstrated slower growth in parent-reported developmental skills from 6 to 12 months. As above, this effect was largely driven by gains in communication and daily living skills, but not social skills. Likewise, motor development was independent of infant birthweight in this sample. These results are consistent with a previous study using latent class analysis, finding that a subset of children born at extremely low birthweight showed decreases in intelligence test scores across the school-age years (from about age 5 to 11 years old)²³. Low birthweight, fetal growth restriction, and preterm birth are associated with differences in brain development, which may underlie the altered developmental trajectories observed in the current study. Reduced thalamocortical connectivity has been identified in infants with early medical risk, with evidence that these differences in brain function are proceeded by learning and emerging neurodevelopmental concerns (e.g., lower cognitive scores, autism features)^24–26. Together, these results suggest that subtle differences in brain development may impact how efficiently infants with lower birthweights process and acquire new information. Clinically, these data suggest the importance of monitoring not just the overall level of developmental skills, but also the rate at which skills are gained over time, particularly for infants born at lower birthweights.

Limitations and Future Directions

This study longitudinally examined the development of a heterogeneous and socioeconomically diverse group of high-risk NICU graduates over the first year of life. This study is primarily limited by small sample size, which disallowed a more in-depth analysis of medical and demographic factors and early intervention exposure that may have further elucidated variability in the developmental trajectories of these infants. To maintain sufficient power for our analysis, predictor variables were analyzed separately, so differential patterns of results should be interpreted with some caution. This study was also impacted by the COVID-19 pandemic, necessitating reliance on parent report to measure developmental abilities, which could more consistently be collected while maintaining the health and safety of this vulnerable population. We utilized a clinician-rated parent interview measure to reduce potential bias from parent report, but future studies should confirm findings via direct assessment. In particular, it will be important to follow-up on our lack of findings related to the acquisition of social skills, for which the Vineland-3 may not have been well suited at these early ages to measure more subtle differences across infants. Also, as this study was conducted during the pandemic, social experiences were quite limited for many of the infants upon their return home, which could have overshadowed lingering effects of the NICU stay and medical complications.

Conclusions

The current study investigated the role of medical risk factors in the early development of high-risk NICU graduates recruited from a university-based High-Risk Infant Follow-up Program in California. We found that length of NICU stay predicted the level of developmental abilities at age one, while birthweight predicted how quickly infants gained new skills over the first year. Motor skills varied by the length of NICU stay, but not birthweight. Results support the need for close developmental monitoring of NICU graduates, particularly those with lengthier stays and lower birthweight. It is important to identify development support needs based not just on the presence of a delay, but on the rate at which infants are learning new skills. If skill acquisition begins to slow, providers should consider a referral for early intervention services, even if the overall level of skills has not yet been identified as a delay. We will continue to follow this sample through at least age two.

Highlights.

• We examined the role of key medical factors in predicting the development of NICU graduates over the first year of life.

• Length of NICU stay predicted developmental abilities at 12 months, while birthweight better explained trajectories.

• Communication and daily living skills, but not social skills, were dependent on early medical factors.

• Variability in the overall level and trajectory of motor skills was explained by length of NICU stay but not birthweight.

• It is important to identify support needs based not just on delays, but on the rate at which infants learn new skills