Effects of Home Visiting and Maternal Mental Health on Use of the Emergency Department among Late Preterm Infants

Abstract

Objective:

To describe use of the emergency department (ED) among late preterm versus term infants enrolled in a home visiting program and to determine whether home visiting frequency was associated with outcome differences.

Design:

Retrospective, cohort study.

Setting:

Regional home visiting program in southwest Ohio from 2007–2010.

Participants:

Late preterm and term infants born to mothers enrolled in home visiting. Program eligibility requires one of four characteristics: unmarried, low income, < 18 years, or suboptimal prenatal care.

Methods:

Data were derived from vital statistics, hospital discharges, and home visiting records. Negative binomial regression was used to determine association of ED visits in the first year with late preterm birth and home visit frequency, adjusting for maternal and infant characteristics.

Results:

Of 1,804 infants, 9.2% were born during the late preterm period. Thirty-eight percent of all infants had at least one ED visit, 15.6% had three or more. No significant difference was found between the number of ED visits for late preterm and term infants (39.4% vs. 37.8% with at least one ED visit, p = .69). In multivariable analysis, late preterm birth combined with a maternal mental health diagnosis was associated with an ED incident rate ratio (IRR) of 1.26, p = .03; high frequency of home visits was not significant (IRR = .92, p = .42).

Conclusions:

Frequency of home visiting service over the first year of life is not significantly associated with reduced ED visits for infants with at-risk attributes and born during the late preterm period. Research on how home visiting can address ED use, particularly for those with prematurity and maternal mental health conditions, may strengthen program impact and cost benefits.

The elevated risk of mortality and morbidity for late preterm infants (LPIs) born at 34 weeks 0 days to 36 weeks 6 days gestation, who represent more than 70% of all preterm infants, has been increasingly well described (Bird et al., 2010; Engle, Tomashek, & Wallman, 2007; Martin, Kirmeyer, Osterman, & Shepherd, 2009; Raju, Higgins, Stark, & Leveno, 2006). Compared with infants born full term (≥ 37 weeks), LPIs have higher rates of hospitalization and emergency department (ED) use in the neonatal period and through the first year of life (Escobar et al., 2005; Jain & Cheng, 2006; McLaurin, Hall, Jackson, Owens, & Mahadevia, 2009). Importantly, for certain conditions like neonatal jaundice, risk of hospitalization for LPIs is higher compared with full-term infants as well as infants born at earlier gestational ages (Ray & Lorch, 2013), suggesting an interplay of immature physiology and current systems of care for this population. In contrast to very preterm infants, LPIs are often discharged home from the hospital without a prolonged period of observation (Goyal, Fager, & Lorch, 2011), and many are not seen by any health care professional during the first week home (Hwang et al., 2013). Moreover, the majority of these infants are not enrolled in systematic, high-risk infant follow-up programs, which generally focus on very early preterm infants (Walker, Holland, Halliday, & Badawi, 2012). For LPIs, therefore, further research is needed to develop models of follow up care that can improve outcomes (National Perinatal Association, 2012; Premji, Young, Rogers, & Reilly, 2012).

One potential strategy to address these concerns is home visiting, a voluntary service delivered in a family’s home to provide care coordination, parenting education, and social support for at-risk child-bearing women and their children (American Academy of Pediatrics Council on Child and Adolescent Health, 1998; Kitzman et al., 1997; Sweet & Appelbaum, 2004). Several national models of home visiting, including Nurse Family Partnership and Healthy Families America, have developed specific program curricula and protocols; qualifications of home visitors range from nurses to social workers to paraprofessionals (U.S. Department of Health and Human Services, 2013). Currently, an estimated 400 publicly and privately funded home visiting programs serve at least 500,000 families in the United States, and an additional $1.5 billion was allocated through the Patient Protection and Affordable Care Act to expand these services (Astuto & Allen, 2009; Health Resources and Services Administration, 2010). Despite significant public investment in this intervention, to date, a paucity of literature on outcomes such as ED use for preterm infants enrolled in such programs (Goyal, Teeters, & Ammerman, 2013).

Given the late preterm birth rate among at-risk infants, practices and policies related to their care have the potential for a large public health impact.

The study objectives were to characterize ED use over the first year of life among late preterm and full-term infants enrolled in home visiting and to determine whether increased frequency of home visiting participation is associated with improvement in this outcome. Our logic model for this study was based on the social-ecological model of child health that underpins the role of home visiting for at-risk families. A strong body of literature has linked social and environmental risk factors with adverse child health outcomes, including avoidable hospitalizations and ED visits, that may be mitigated through early detection, parental education, and care coordination (McLaren & Hawe, 2005; Paul, Phillips, Widome, & Hollenbeak, 2004; Shanley, Mittal, & Flores, 2013; Shonkoff & Garner, 2012). Given the known contribution of LPIs to pediatric morbidity and health care costs and the fact that preterm birth is likely to disproportionately affect at-risk mothers eligible for home visiting, a more detailed understanding of program effectiveness for LPIs may be critical to addressing gaps in care for this important population.

Methods

Setting and Participants

In this retrospective, cohort study we examined ED use among late preterm and term infants born to at-risk, first-time mothers enrolled in a well-established, regional home visiting program serving southwest Ohio. This community-based home visiting program, which has to date served more than 19,000 families, comprises 11 local home visiting agencies which adhere to program, training, and evaluation standards established by a central office at Cincinnati Children’s Hospital Medical Center (CCHMC). To track and document process and outcome measures within and across agencies, the program uses rigorous continuous quality improvement procedures under the supervision of CCHMC quality improvement staff and is facilitated by a web-based data entry system (Ammerman et al., 2007).

In addition to being first- time mothers, women eligible for this program must have at least one of four risk characteristics: unmarried, low income (up to 300% of poverty level, receipt of Medicaid, or reported concerns about finances), < 18 years of age, or suboptimal prenatal care. Participants may be enrolled during pregnancy or postdelivery, before their child reaches age 3 months. Referrals to the program may be self-initiated or come from clinics, hospitals, and other community sources. Home visits are provided by social workers, child development specialists, or other professionals who employ a core program curriculum that is based on the Healthy Families America model of home visiting. The overall goals of the program are to (a) provide nutrition education and substance use reduction during pregnancy; (b) support parents in providing children with a safe, nurturing, and stimulating home environment; (c) optimize child health and development; (d) link families to health care and other services; and (e) promote economic self-sufficiency. To achieve these goals as outlined within the curriculum, the home provider offers printed materials for families but primarily focuses on interactive sessions with parents that may address curriculum content as well other issues or concerns specific to the family. Screening inventories for home safety, parenting stress, substance use, and other items are also performed at scheduled intervals to identify and address risks and to generate appropriate service referrals. Expected visit frequency consistent with the curriculum is weekly through the first 3 months of infancy, tapering to biweekly through the remainder of the first year.

For this analysis, infants born prior to 34 weeks gestation were excluded, resulting in 1,852 late preterm and term infants born during the years 2007 to –2010 whose mothers enrolled in home visiting either prenatally or within 3 months after delivery. Of these infants, 43 additional infants were excluded from analysis due to major congenital anomalies, as their patterns of health care use were expected to vary significantly from otherwise healthy infants. For similar reasons, six infants who died of any cause before their first birthdays were also excluded from analysis.

Data Sources

Home visiting data were abstracted from the program’s web-based data entry system described above. This system contains detailed information on each participant including enrollment timing, home visit history, and maternal demographic and psychosocial screening information. Enrolled participants consented to data being used for the purpose of quality improvement benchmarking and research. These data were linked to Ohio vital statistics, available from the Ohio Department of Health, and birth-related hospital discharge records of mother and infant, available from the Ohio Hospital Association. Because no common unique identifier across data sources was available, record linkage was accomplished using LINKS (University of Manitoba), a SAS-based probabilistic and deterministic matching program. Selected variables used for linking included maternal and infant dates of birth, hospital of birth, delivery method, sex, and maternal address. Further details of linkage for these data sources has been previously described (Hall et al., 2014). The resulting data set contained information regarding maternal/child health including demographics, social factors, pregnancy-related conditions, and infant characteristics. Lastly, this data core was linked to electronic health record data at CCHMC for outcome measures of hospital service use. The Ohio Department of Health and CCHMC Institutional Review Boards approved this study.

Covariates and Key Predictors

As described previously, data for maternal covariates were obtained through a combination of linked vital statistics, hospital discharge records, and home visiting data (Hall et al., 2014). These variables included race, ethnicity, payer source, maternal age, education level, marital status, substance use, household membership, and paternal involvement. Indicator variables for relevant clinical factors were constructed using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes and vital statistics data (Centers for Disease Control and Prevention, 2014). The ICD-9-CM codes used to derive a composite maternal mental health diagnosis were obtained from the maternal birth hospitalization record.

Late preterm birth was defined as infant birth from 34 weeks 0 days to 36 weeks 6 days gestation; gestational age measures were obtained from vital statistics and represented the best clinical estimates. Additionally, as a sensitivity analysis we repeated analyses using a combined gestational age estimate from vital statistics rather than the clinical gestational age estimate, as prior studies have demonstrated discordance between these measures (Wingate, Alexander, Buekens, & Vahratian, 2007). To measure home visiting service intensity, we adapted a prior approach from Duggan et al. (2004), counting the number of home visits conducted over the first year of life and then dividing this by the number of expected home visits over the infant’s first year per the program curriculum to calculate a percentage of expected visits. Mother/infant pairs were classified as receiving a high dose of service if they received ≥75% of expected visits, a commonly used cutoff for service evaluation in home visiting programs (Healthy Families New York, 2014). Timing of program enrollment was dichotomized as enrollment prenatally or after birth of the infant.

Analysis

Bivariate analyses using chi-squared or t tests were used to identify covariates associated with any ED use and number of unique ED visits over the first year after birth. Factors deemed to be empirically or statistically important (p values less than 0.25) were considered and tested using step-wise multivariable modeling to derive parsimonious models. To account for overdispersion of the ED visit data due to excess zeros, we used a random-effects negative binomial regression model as an alternative to standard Poisson regression, adjusting for clustering by individual home visiting agency. Models were tested for goodness of fit using Akaike Information Criterion values and link tests for model specification. Multicollinearity was also assessed, with variance inflation factors for all retained variables < 10 (O’brien, 2007).

The final multivariable model included the following variables: infant sex, maternal race, ethnicity, insurance status, maternal age, paternal living arrangement, smoking status, and mental health diagnosis. To prevent individual patients from biasing rates and to reduce measurement error, a small number of infants with >8 ED visits in the first year (1% of the sample) were deemed to be outliers based on visual assessment of the data and were omitted from the final analysis. Interaction terms between late preterm status with number of home visits, maternal age, smoking status, and mental health diagnosis were added and tested for statistical significance using likelihood ratio tests and only an interaction term with mental health diagnosis was retained. All statistical tests were two sided, and type I error was controlled at 0.05. Analyses were performed using STATA 11.0.

Results

Of the 1,804 infants meeting study inclusion criteria, 9.2% were born late preterm. No significant differences in maternal characteristics were observed among late preterm versus full term infants (56% vs. 52% with maternal age <20 years, p = .34), 96% versus 95% with single marital status (p = .56), and 81% versus 80% insured by Medicaid (p = .96). Forty-eight percent of the sample enrolled in the home visiting program prenatally. Approximately 17.5% of infants were classified as receiving a high dose of home visiting, with no significant difference between late preterm compared with term infants (18% vs. 17%, p = .82).

The number of ED visits in the first year of life for all infants ranged from 0 to 17, with 38% of infants having at least one ED visit and more than 15% of infants having three or more ED visits. As shown in Figure 1, the distributions of primary diagnoses accounting for more than 80% of ED visits were similar, although late preterm compared with term infants had a higher incidence of visits for feeding difficulty (3.3% vs. 0.7%) and asthma/wheezing not otherwise specified (2.2% vs. 0.8%).

Figure 1.

Primary diagnoses accounting for >80% of emergency department visits in the first year among term and late preterm infants.

Unadjusted Analysis

As shown in Table 1, bivariate comparisons demonstrated no significant difference in any ED use by gestational age category (term vs. late preterm birth). In univariable analysis, late preterm birth was also not associated with a significantly increased number of ED visits (incident rate ratio (IRR) 1.06, 95% confidence interval (CI) [0.91, 1.24]). High intensity of home visits over the first year of life was associated with a reduced incidence of any ED visit (31.7% vs. 39.3% among the low-intensity home visiting group); however, as shown in Table 2 this association was not observed for multiple ED visits, with an IRR = .88, 95% CI [0.72, 1.08]. The association of prenatal versus postnatal enrollment was significant in univariable analysis, both when modelled as any ED use (Table 1) and as a count of ED visits, IRR = 1.18, 95% CI [1.02, 1.36].

Table 1:

Bivariate Comparisons of Predictors with any Emergency Department (ED) use in the First Year of Life

	No ED use	Any ED use	p value
Gestational age, % (n)
Late preterm	60.6 (100)	39.4 (65)	0.70
Full term	62.2 (1019)	37.8 (619)
Home visiting service, % (n)
<75% expected visits	60.7 (903)	39.3 (585)	0.01
≥75% expected visits	68.4 (216)	31.7 (100)
Timing of enrollment, % (n)
Prenatal	59.2 (514)	40.8 (354)	0.02
Postnatal	64.6 (605)	35.4 (331)
Infant gender, % (n)
Female	64.1 (579)	35.9 (324)	0.07
Male	59.9 (540)	40.1 (361)
Race, % (n)
White	65.8 (379)	34.2 (197)	0.60
Black	59.5 (672)	40.5 (458)
Asian/Pacific Islander	66.7 (10)	33.3 (5)
Multirace	67.3 (37)	32.7 (18)
Other	74.1 (20)	25.9 (7)
Ethnicity, % (n)
Hispanic	82.8 (135)	17.2 (28)	<.001
Non-Hispanic	60.0 (984)	40.0 (657)
Insurance, % (n)
Medicaid	60.0 (861)	40.0 (573)	0.002
Private	67.5 (187)	32.5 (90)
Self-pay	76.3 (58)	23.7 (18)
Other	81.8 (9)	18.2 (2)
Maternal age, % (n)
<20 years	61.4 (578)	38.6 (364)	0.02
20–30 years	62.5 (530)	37.5 (318)
>30 years	78.6 (11)	21.4 (3)
Mental health diagnosis, % (n)
Yes	53.4 (95)	46.6 (83)	0.01
No	63.0 (1024)	37.0 (602)
Smoking status, % (n)
Yes	56.5 (309)	43.5 (238)	0.001
No	64.4 (810)	35.6 (447)
Lives with infant’s father, % (n)
Yes	68.3 (198)	31.7 (92)	.05
No	60.4 (836)	39.6 (549)

Table 2:

Bivariate Comparisons of Predictors with ≥ 3 Emergency Department (ED) Visits in the First Year of Life

	< 3 ED visits	≥ 3 ED visits	p-value
Gestational age, % (n)
Late preterm	84.2 (139)	15.8 (26)	0.88
Full term	84.7 (1387)	15.3 (251)
Home visiting service, % (n)
<75% expected visits	84.5 (1257)	15.5 (231)	0.77
≥ 75% expected visits	85.1 (269)	14.9 (47)
Timing of enrollment, % (n)
Prenatal	83.1 (721)	16.9 (147)	0.08
Postnatal	86.0 (805)	14.0 (131)
Infant gender, % (n)
Female	86.5 (781)	13.5 (122)	0.03
Male	82.7 (745)	17.3 (156)
Race, % (n)
White	89.2 (514)	10.8 (62)	0.001
Black	81.9 (926)	18.1 (204)
Asian/Pacific Islander	100.0 (15)	0.0 (0)
Multi-Race	83.6 (46)	16.4 (9)
Other	88.9 (24)	11.1 (3)
Ethnicity, % (n)
Hispanic	92.6 (151)	7.4 (12)	0.003
Non-Hispanic	83.8 (1375)	16.2 (266)
Insurance, % (n)
Medicaid	83.7 (1200)	16.3 (234)	0.09
Private	87.0 (241)	13.0 (36)
Self-pay	90.8 (69)	9.2 (7)
Other	100.0 (11)	0.0 (0)
Maternal age, % (n)
<20 years	85.7 (806)	14.3 (135)	0.02
20–30 years	82.7 (672)	17.3 (141)
>30 years	95.9 (47)	4.1 (2)
Mental health diagnosis, % (n)
Yes	78.1 (139)	21.9 (39)	0.01
No	85.3 (1387)	14.7 (239)
Smoking status, % (n)
Yes	81.7 (447)	18.3 (100)	0.03
No	85.8 (1079)	14.2 (178)
Lives with infant’s father, % (n)
Yes	87.9 (247)	12.1 (34)	0.80
No	83.8 (1160)	16.3 (225)

Multivariable Analysis

Table 3 depicts results of negative binomial regression analysis adjusting for clustering by home visiting agency as well as covariates. Although the incidence of mental health diagnosis in this sample was not significantly different between late preterm and term infants (7% vs. 10%, p = .15), we observed a significant modifying effect of mental health diagnosis on the association between late preterm birth and ED visits. Compared with term infants of mothers without a diagnosis, LPIs of mothers with a mental health diagnosis had a 2.26 IRR for ED visits in the first year, 95% CI [1.10, 4.67]. Maternal mental health was also a statistically significant predictor for term infants; however, the effect size was smaller (IRR = 1.27, 95% CI [1.01, 1.60]). In this multivariable model, neither timing of enrollment nor frequency of home visiting service during the first year was significantly associated with a reduced rate of ED use. Other covariates in the model that were statistically significant included Hispanic ethnicity (adjusted odds ratio [AOR] = .52, 95% CI [0.34, 0.80]), maternal smoking (AOR = 1.26, 95% CI [1.07, 1.49]), Black race (AOR = 1.23, 95% CI [1.01, 1.49]), and maternal age >30 years (AOR = .41, 95% CI [0.20, 0.81]). Results of a sensitivity analysis using multivariable logistic regression were similar and therefore not depicted.

Table 3:

Random Effects Negative Binomial Regression Analysis^a of Emergency Department Visits in the First Year of Life

	Incident rate ratio	95% confidence interval
Late preterm	0.97	0.74, 1.27
Maternal mental health diagnosis	1.27	1.01, 1.60
Late preterm mental health diagnosis	2.26	1.10, 4.67
> 75% expected visits	0.92	0.74, 1.13
Prenatal enrollment	1.05	0.91, 1.22
Female	0.79	0.69, 0.92
Race
White	reference
Black	1.23	1.01, 1.49
Asian/Pacific Islander	1.42	0.57, 3.52
Multi-Race	1.15	0.71, 1.85
Other	0.71	0.29, 1.71
Hispanic	0.52	0.34, 0.80
Insurance
Medicaid	reference
Private	0.85	0.68, 1.06
Self-pay	0.63	0.39, 1.02
Other	0.72	0.18, 2.81
Maternal age, % (n)
<20 years	1.00	0.86, 1.16
20–30 years	reference
>30 years	0.41	0.20, 0.81
Smoking	1.26	1.07, 1.49
Lives with infant’s father	0.84	0.68, 1.05

Note.

^aAdjusts for clustering by home visiting agency with an exchangeable correlation structure; 11 outlier observations (>8 ED visits in the first year) were omitted from analysis.

Discussion

The vulnerability of LPIs in terms of mortality, morbidity, and increased health care use in the neonatal period and later in infancy and early childhood, has been established in previous literature (Bird et al., 2010; Engle et al., 2007; Martin et al., 2009; Medoff-Cooper et al., 2012; Raju et al., 2006). However, a gap remains in the evidence to support models of postdischarge care that may improve outcomes for this population (Premji et al., 2012). The majority of LPIs are not enrolled in high-risk infant follow-up programs that generally focus on very preterm infants (Walker et al., 2012). Additionally, recent evidence suggests that many families of LPIs do not access timely primary care follow-up (Hwang et al., 2013). Moreover, given known social and environmental risks associated with preterm birth, LPIs compared with term infants may be more likely to be affected by poverty, social isolation, and other factors that place them at further risk for adverse health and developmental outcomes (Eunice Kennedy Shriver National Institute of Child Health and Human Development, 2014). We evaluated ED visit outcomes in a cohort of late preterm and term infants enrolled in a home visiting program, an intervention aimed at addressing social determinants of health for disadvantaged families through parent education, social support, and care coordination. Prior literature on at least one national model of home visiting (Nurse Family Partnership) suggests that infants receiving home visits may incur fewer hospitalizations for injuries and ingestions compared with those not receiving home visits (Kitzman et al., 1997). Our results demonstrate an association between intensity and timing of home visits and ED use in the first year; however, this relationship does not persist after adjusting for other clinical and social factors. We did not find evidence to support that LPIs benefit differently from this intervention than do other at-risk infants born full term.

We did not detect an independent association between late preterm birth and ED use, perhaps due to relatively small sample size as well as the overall high level of use across the cohort regardless of gestational age. Importantly, we did find that within this at-risk cohort, maternal mental health conditions were a significant modifier on the association between late preterm birth and ED use in infancy. That is, LPIs whose mothers had diagnosed mental health conditions had more than double the rate of ED visits in the first year of life than full-term infants of mothers without diagnosed mental health conditions. Prior work on maternal mental health has demonstrated a link between maternal depression and anxiety with increased infant acute and emergency visits (Chung, McCollum, Elo, Lee, & Culhane, 2004; Mandl, Tronick, Brennan, Alpert, & Homer, 1999; Minkovitz et al., 2005; Sills, Shetterly, Xu, Magid, & Kempe, 2007). In a recent study of term infants, the authors noted that the degree of association with infant ED visits differed by timing of maternal depression and anxiety with even higher use for those infants whose mothers’ mental health conditions began during the postpartum period (Farr et al., 2013). To our knowledge, the differential effect of maternal mental health on ED visits for late preterm versus term infants has not been previously evaluated. However, prior researchers have described the relationship between late preterm birth and maternal anxiety, particularly relating to feeding difficulty (DeMauro, Patel, Medoff-Cooper, Posencheg, & Abbasi, 2011; McDonald et al., 2013). Combined with the fact that LPIs have rates and patterns of feeding dysfunction in the first year of life that are similar to those of very preterm infants, this result likely contributes to our finding of higher incidence of ED visits for feeding problems compared with term infants.

Finally, our current findings with this cohort enrolled in a home visiting program are generally consistent with previously reported data on increased risk of ED use among late preterm versus term infants (Jain & Cheng, 2006). However, overall rates of ED visits in this sample are high compared with national data on ED use (published estimates for young children age 0–4 years are 17% with at least one ED visit, and 11% with two or more visits). This finding likely reflects the higher risk level of families eligible for and receiving home visiting services, as many sociodemographic factors including poverty, single-parent status, Black race, and Medicaid coverage are associated with more frequent ED use (Bloom, Cohen, & Freeman, 2011; Halfon, Newacheck, Wood, & St Peter, 1996). Maternal smoking status was also a significant factor in our analysis, consistent with prior literature demonstrating its association with respiratory and gastrointestinal disorders in infancy (Carroll et al., 2007; Shenassa & Brown, 2004). Given that more than 30% of mothers in this sample were classified as smokers, this behavior may be a critical point of targeted intervention within home visiting prenatally and during infancy.

Further efforts to develop and refine community-based programs serving late preterm infants and their families may need to address maternal mental health conditions.

Limitations

Several limitations related to use of administrative data in this retrospective analysis are acknowledged. Complications and comorbidities identified using vital statistics and hospital discharge data may result in a misclassification bias (Hsia, Krushat, Fagan, Tebbutt, & Kusserow, 1988; Iezzoni et al., 1992; Romano & Mark, 1994). In particular, the mental health diagnosis variable, which was based on the list of discharge diagnoses from the birth hospital, has high specificity but lower sensitivity in that there are likely women in the sample with uncoded or undiagnosed conditions, which could drive risk estimates towards the null (Yasmeen, Romano, Schembri, Keyzer, & Gilbert, 2006). Moreover, this measure would not include new mental health conditions that emerged after the birth hospitalization period. However, to further assess the validity of this variable, we compared it to maternal scores on the Interpersonal Support Evaluation List (ISEL), a validated screening tool collected by home visitors at enrollment that measures perceived social support and correlates with stress and negative effect (Merz et al., 2014); as expected, the mental health variable was significantly associated with lower (worse) ISEL scores in our bivariate assessments, p = .001. Another potential limitation is that outcome measures were reliant on data captured through the CCHMC system, the region’s only pediatric ED services provider. This system sees nearly 90% of pediatric admissions within its eight- county primary catchment zone, with that percentage increasing for the youngest patients. These data may underestimate rates of ED visits and introduce selection bias, though our findings are robust to adjustment for geographic clustering by zip code. Another limitation may be generalizability due to the regional population represented in the study. Finally, though home visitors place a strong emphasis on identification of a medical home and adherence to well child care visits for program participants, we were unable to assess individual data on primary care use as a predictor of ED visits. These limitations are offset by strengths of the study, which include our ability to evaluate ED use for a regional, at-risk population and link this outcome to predictors not otherwise captured in hospital administrative data (i.e., gestational age, social risk factors). Outcomes were obtained through electronic heath system records and therefore not subject to parental reporting bias.

Practice Implications

Specific implications for practice based on these findings may include enhanced screening and support systems for maternal mental health issues prior to hospital discharge, at postpartum follow-up, and in pediatric settings. Within home visiting programs, targeted counseling and guidance for mothers to reduce maternal stress and improve coping after delivery of a preterm infant may serve as a useful curriculum enhancement. Additional components of an enhanced home visiting curriculum targeting LPIs may include emphasis on elevated health risks associated with late preterm birth, particularly with regards to respiratory conditions and feeding difficulty. Finally, for all infants enrolled in home visiting programs, further program refinements focused on modifiable risk factors for ED use, such as improved connection to primary care and a more systematic approach to maternal smoking cessation, may increase the cost benefits of this intensive, preventive service.

Conclusions

Late preterm infants have been previously shown to be at higher risk for ED use and rehospitalization compared with full-term infants in the first year of life. This study is one of the first to focus on the utility of a community-based program to mitigate outcomes in this population. We observed that within a socially at-risk population of infants enrolled in home visiting, late preterm birth in combination with maternal mental health conditions is associated with more than twofold higher rate of ED visits compared with infants lacking these risk factors. Utilization outcomes in this population did not appear to improve with early program engagement or high intensity of service after adjustment for clinical and social factors. Further research may focus on development and refinement of approaches to address needs of LPIs and their families in a home-based setting.