From Hospital to Home: Continuity between Skin-to-Skin Care and Later Verbal Engagement in Infants Born Preterm

Abstract

This descriptive cohort study documented continuity between family-delivered skin-to-skin care rates for preterm infants in the NICU and amounts of child-directed speech at child age 9 months. Involvement in skin-to-skin care may be an early marker of caregiver-infant engagement and a target for early interventions that support positive caregiver-infant interactions.

Children born very preterm (VPT; <32 weeks gestational age [GA]) are at elevated risk for neurodevelopmental delays.^1,2 Family-administered skin-to-skin (STS) care, in which caregivers hold their infant against their bare chest, is a developmental care practice in the Neonatal Intensive Care Unit (NICU) associated with favorable neurodevelopmental outcomes at 12 and 18 months (all ages adjusted for the degree of prematurity).^3,4 These associations may reflect direct, long-term benefits of STS care on neurodevelopment. Alternatively, STS care may be linked to later caregiver-infant engagement that directly or cumulatively leads to positive neurodevelopmental outcomes. Exploring these possibilities requires data on how STS care in the hospital relates to caregiver-infant engagement at home.

The current descriptive cohort study of infants born VPT investigated relations between rates of family-administered STS care in the NICU and caregiver-infant engagement in the home environment during infancy. One index of caregiver-infant engagement is the quantity of child-directed caregiver speech, which has been linked to neurodevelopmental outcomes in full term and preterm infants.^5,6 Assessing continuity in caregiver-infant engagement from hospital to home may help explain how early experiences contribute to neurodevelopmental outcomes. Rates of STS care may also serve as an early marker of caregiver-infant engagement, indicating families who may benefit from interventions that support positive caregiver-infant interactions.

Methods

Participants

We recruited 74 infants from Lucille Packard Children’s Hospital (LPCH) and California’s High-Risk Infant Follow-Up (HRIF) Clinic between August 2020 and September 2024. Infants were eligible if they were born < 32 weeks GA, had no congenital/genetic anomalies, no hearing/vision loss, and caregivers who primarily spoke English or Spanish. Of these, 58 met our eligibility criterion for developmental care – born at or admitted to LPCH within seven days of birth and hospitalized for over 7 days. Verbal engagement data were available for 32 families, who comprised the final sample. Included and excluded families did not significantly differ on demographic or clinical variables (all p > 0.10). Our sample is representative of English- and Spanish-speaking families in the San Francisco Bay Area. The study was approved by the Stanford School of Medicine Institutional Review Board (IRB; Protocol #53904, #44480) and informed consent was obtained from a primary caregiver.

Measures

Clinical and Demographic Information

Demographic and clinical characteristics of infants, including GA (weeks), length of hospital stay (days), and child sex (male or female), were extracted from the electronic medical record (EMR). Caregivers reported on their race, ethnicity, and language spoken at home. Socioeconomic status (SES) was measured using the Hollingshead Four Factor Index⁷, which combines both primary caregivers’ education and occupation into a score ranging from 8 to 66, with higher scores indicating higher SES. Clinical data regarding four common comorbidities of premature birth were also obtained from the EMR. Bronchopulmonary dysplasia (BPD) was defined as grade > 2 severity; sepsis as a positive blood culture or > 7 days of antibiotic treatment; intraventricular hemorrhage (IVH) as any grade; and necrotizing enterocolitis as requiring medical or surgical treatment. A binary health acuity variable then classified infants as having none (health acuity = 0) or > 1 (health acuity = 1) of these conditions.

Family-Involved Visitation and Developmental Care Practices

Visitation and developmental care data were extracted from the EMR. Nurses at the LPCH NICU routinely document all instances of family visits to bedside. To account for varying lengths of hospitalization, visitation frequency was calculated as the total number of visitation instances, divided by the length of hospital stay (days). Furthermore, nurses record all instances of STS care, including the session duration (minutes) and the individual(s) involved. Rates of family-delivered STS care were computed as the total minutes of STS, divided by the length of hospital stay (days). Duration was calculated as the total minutes of STS, divided by the number of STS instances.

Caregiver Verbal Engagement

Caregiver verbal engagement was assessed at child age ~ 9 months using day-long audio recordings via Language Environment Analyses (LENA)⁸, a high-quality digital recording device worn by infants in specialized clothing. Families were instructed to record a typical day at home for up to 16 hours. The LENA speech-recognition software classifies all adult speech “near and clear” to the child and generates adult word counts (AWC) for each 5-minute segment of the recording. Recordings were subsequently cleaned using an automated classifier to identify segments in which the child was likely to be awake and that contained primarily child-directed speech (CDS).⁹ Child-directed adult word count/hour (CD-AWC/hour) was computed as the sum of AWC during non-sleep, CDS segments, divided by the length of the cleaned recording.

Statistical Analysis

All analyses were conducted in R version 4.4.3. Both STS care rates and CD-AWC/hour were inspected for normality using histograms and Shapiro-Wilks tests. STS care rates were transformed using a base 10 log due to skewness. Hierarchical linear regression models were used to examine relations between STS care rates and later CD-AWC/hour. SES and health acuity were selected a priori as covariates, given their anticipated relevance to STS care rates and verbal engagement measures. Family visitation frequency was included as a covariate to assess the specific effects of engaging in STS care beyond hospital visitation alone.

Results

This sample (56.2% female) had a mean GA at birth of 28.8 weeks (SD = 2.5). The average length of hospital stay was 71.6 days (SD = 43.9); 53.1% (n = 17) of infants were classified as having health acuity = 1. The mean SES score was 40.9 (SD = 17.8). Table I outlines NICU visitation frequency, STS care rates and durations, day-long recording lengths, and CD-AWC/hour. On average, families visited around once a day and infants experienced STS for about 25 mins/day, with instances lasting about 70 mins, on average. However, there was substantial variation, such that one infant received more than 100 mins/day, whereas two infants received none. STS rates were not significantly correlated with SES (r = .29, p = .10). At child age 9 months, adjusted for degree of prematurity, portions of the day-long recordings on which CD-AWC/hour were computed averaged about 8 hours. Caregiver verbal engagement varied across families, ranging from 426 to more than 3500 CD-AWC/hour.

Table 1.

Descriptive statistics for NICU measures and day-long home recordings (n = 32).

	M (SD)	Min	Max
Visitation frequency (visits/day)a	1.4 (0.5)	0.2	2.6
Skin-to-skin care rateb	24.8 (26.5)	0.0	107.8
Skin-to-skin care durationc	70.8 (18.2)	40.0	124.4
Child age at day-long recordingd	9.7 (0.8)	8.6	11.9
Duration of total day-long recordingd	15.4 (2.2)	4.6	16.0
Duration of CDS in day-long recordingd	7.9 (2.5)	2.4	14.6
CD-AWC/houre	1782.3 (772.3)	425.8	3565.0

Note:

^aNumber of family visits to bedside/day of NICU hospitalization;

^bMinutes of family-involved skin-to-skin care/day of NICU hospitalization;

^cMinutes of family-involved skin-to-skin care/number of STS instances;

^dChild age (adjusted for degree of prematurity), total recording length, and CDS recording length of day-long recordings;

^eNumber of child-directed adult words/hour during day-long recording.

We next examined whether infants who received more STS care in the NICU were exposed to greater verbal engagement at 9 months. Table II shows that SES, health acuity, and visitation frequency (Model 1) accounted for approximately 11% of the variance in CD-AWC/hour. Model 2 shows that STS care rate significantly increased the overall prediction of CD-AWC/hour (F(1, 27) = 7.09, p = 0.01), accounting for approximately 18% additional variance. Figure 1 illustrates that caregivers who provided more STS care (mins/day) provided more CD-AWC/hour when the child was 9 months. For example, increasing STS care rate from 25 to 50 minutes/day (a 0.29 increase) is associated with an increase of 184 CD-AWC/hour.

Table 2.

Multiple regression models (unstandardized coefficients (SE)) predicting CD-AWC/hour at 9-months (adjusted for degree of prematurity), after consideration of covariates (n = 32).

Predictors	Model 1	Model 2
SESa	12.87 (7.76)	6.63 (7.42)
Health acuityb	141.91 (271.11)	169.10 (245.91)
Visit frequencyc	247.96 (258.38)	105.05 (240.22)
Skin-to skin care rated	-	631.30 (237.03)*
R2	11.44%	29.87%*
Δr 2	-	18.43%*

Note:

^*p < .05

^aHollingshead Four Factor Index of Socioeconomic Status (HI; Hollingshead, 1975; range = 8–66);

^bPresence/absence of ≥ 1 common comorbidities of premature birth: bronchopulmonary dysplasia, intraventricular hemorrhage, sepsis, or necrotizing enterocolitis;

^cNumber of family visits to bedside/days of NICU hospitalization;

^dMinutes of log-transformed skin-to-skin care/days of NICU hospitalization.

Figure 1.

Model predictions showing relations between skin-to-skin (STS) rate in the NICU and child-directed adult word count (CD-AWC/hour) during day-long recording at 9 months (adjusted for degree of prematurity), controlling for covariates.

Note: Black line represents model-predicted values of CD-AWC/hour based on skin-to-skin care rate. Shaded area indicates 95% confidence interval of the model prediction. Dots represent individual observations.

Discussion

To investigate continuity in caregiver-infant engagement for infants born VPT, this study explored relations between STS care rates in the NICU and later caregiver verbal engagement. Our findings extend prior work^3,4 by demonstrating that caregivers who engage in more STS care in the NICU are also more verbally engaged at home. Models controlled for visitation, indicating that effects were due to engagement in STS care, rather than visitation more generally. Notably, this relation emerged despite families engaging in STS care on average about 25 minutes/day, rates well below the WHO recommendations of 8 or more hours/day.¹⁰ However, the durations observed were comparable with those found in infants in highly resourced communities.¹¹ Taken together, these findings underscore that variation in STS rates is meaningful even at these levels.

There are many possible interpretations of the link between STS care rates and verbal engagement. STS care in early infancy increases oxytocin levels and strengthens parent-infant bonding, which may, in turn, enhance caregiver confidence and promote behaviors that support language-rich home environments.^12,13 Engaging in STS care may indirectly affect caregiver behaviors by reducing infant stress and promoting infant development, resulting in a relatively responsive child, which may lead to more responsive caregiving behaviors.^14,15 Alternatively, engagement in STS care in the NICU may signal an underlying pattern of caregiver-infant engagement that persists across time and caregiving contexts. In any case, clinical attention to the rates of STS care in the NICU may help to identify caregivers who might benefit from education and support in establishing favorable home environments for infants born VPT.

Our study had both strengths and limitations. Although our sample was diverse, it was relatively small, which may limit generalizability. Our measures of STS care relied on the accuracy of nurse charting; however, this practice is well-established at our center and integrated into standard clinical training.¹⁶ Our study was limited to a single site and could not explore how hospital- or population-level factors may have influenced STS rates. Still, our findings offer valuable insights into caregiver engagement and should guide future research on systemic factors. Day-long audio recordings provide more objective, extensive, and less obtrusive estimates of caregiver verbal engagement than other available methods. However, they do not capture non-verbal behaviors that may also reflect positive caregiver-infant engagement. In addition, although caregivers were instructed to record a typical day at home, we could not verify who contributed to our measure of CD-AWC/hour. Finally, our design is correlational and therefore, we cannot draw causal conclusions.

In conclusion, this study highlights continuity from hospital to home in the caregiving environments of infants at-risk for neurodevelopmental delays and underscores links between patterns of early developmental care and children’s later language environments. Future work should link developmental care practices in the NICU, caregiver verbal engagement in the home, and subsequent neurodevelopmental outcomes. Routine charting of STS care offers a practical way for clinicians to monitor caregiver engagement in the NICU and to identify families who may benefit from additional post-discharge support, such as connections to community resources or access to early interventions. Promoting STS care as part of standard NICU practice may help establish enriched caregiving environments that support long-term developmental trajectories.