The Association Between Routine Car Seat Screening and Subsequent Health Care Utilization

We studied a retrospective cohort of United States infants born at 2 health systems (1 academic and 1 community hospital per system) using electronic medical record data. Late preterm (34–36 weeks) infants of any birth weight and term (≥37 weeks) infants with a birth weight <2268 g born at 1 of the study hospitals between January 1, 2014 and December 31, 2018 and surviving to discharge were included. Newborns discharged from any location (nursery [ie, mother–baby unit], NICU, or pediatric floor) were included. Floor discharges only occurred at 1 hospital, which used these beds for NICU overflow or nursery newborns requiring prolonged hospitalization (eg, neonatal abstinence syndrome). Postdischarge data were only available from a single system (System B) but included care received at all 10 affiliated system hospitals. We previously used birth hospitalization data for System B newborns to examine predictors of screening adherence and CSTS failure.^8,14  Approval was granted by the institutional review boards at each institution.

Both systems had CSTS policies specifying that infants <37 weeks gestational age or <2268 g should be screened after 24 hours of life and before discharge (Table 1). Other conditions (eg, respiratory compromise) were noted as appropriate to screen at physician discretion. Failure criteria included hypoxia, bradycardia, or apnea using similar but not identical saturation and heart rate thresholds (Table 1). After a failed CSTS, both policies recommended repeating CSTS before discharging with a car bed. Additional diagnostic testing (eg, chest radiograph, echocardiogram) was left to physician discretion. Reason for failure and information on car bed utilization was inconsistently documented and unavailable for analysis.

Birth hospitalization outcomes included total LOS and time to discharge after the first CSTS for newborns undergoing screening (measured in hours). Postdischarge outcomes (System B only) included hospital revisits (emergency department or readmission) and brief resolved unexplained event (BRUE) diagnosis within 30 days of birth hospitalization discharge. Hospital revisits were characterized as definitely CSTS-related (eg, apnea while in the car seat), potentially CSTS-related (eg, laryngomalacia with positional work of breathing), or CSTS-unrelated (eg, neonatal fever) on the basis of a chart review by a single author (WNH). We examined both CSTS-related and all-cause revisits as seemingly unrelated visits could be influenced by second-order screening effects (eg, reduced hyperbilirubinemia readmission risk after a prolonged birth hospitalization with additional lactation support or inpatient phototherapy).

Car seat tolerance screening, categorized as “not screened”, “passed initial screen”, or “failed initial screen,” represented our primary exposure. Covariates used for adjustment included study year, calendar month, sex, race/ethnicity (White, non-Hispanic; African American, non-Hispanic; Asian American; Hispanic; other/unknown), gestational age (days), birth weight, delivery type (cesarean; vaginal) payer (Medicaid; private; other), discharging unit (nursery; NICU/floor), and health system. Our decision to include race and ethnicity as a covariate does not assert race-based biologic differences but reflects the social dimensions of race and racism in medicine given disparities in care processes and health care utilization.^16–18 

Descriptive statistics were used to examine cohort characteristics. Differences in newborn characteristics by CSTS category were compared by using χ² and ANOVA. Median quantile regression was used to evaluate the relationship between CSTS and LOS, whereas logistic regression was used for the all-cause hospital revisit outcome. The association between CSTS and outcomes was analyzed in 3 ways owing to 2 considerations: (1) imperfect adherence to routine CSTS screening, and (2) newborns who failed their initial CSTS may have received additional interventions that impacted subsequent health outcomes. First, we compared newborns undergoing screening to those meeting criteria but not screened, estimating the effect of screening itself. Second, we compared screened newborns who failed their initial screen to those who passed to estimate whether infants failing screening are at greater risk for adverse outcomes. Third, we compared all 3 groups (not screened, passed, failed), estimating risks related to both screening and CSTS result. Given the potential for effect modification, subgroup analyses stratified by discharging unit were performed. Adjustment covariates, the 3-part analytic approach, and subgroup analysis were chosen a priori. Infants failing the initial CSTS in the nursery and subsequently transferred to the NICU or floor were recategorized as having been discharged from the nursery to reduce bias because their transfer was influenced by the primary exposure. Analysis was performed by using Stata, version 15.1 (StataCorp, College Station, TX).

Of 5222 study newborns, 64% met screening criteria because of gestational age, 9% because of birth weight, and 27% because of both (Table 2). Screening adherence was 91% and differed by system (A: 88% vs B: 93%; P ≤ .001). Sixty-one percent of newborns were discharged from the nursery (including 34 newborns transferred to the NICU or floor after failing the CSTS in the nursery) in which screening adherence was 93% as compared with 86% in the NICU/floor (P ≤ .001). Screening adherence was positively associated with birth at 35 to 36 weeks’ gestation (P ≤ .001), higher mean birth weight (P = .04), private insurance (P = .02), and later study year (P ≤ .001). Screening did not differ by sex, race/ethnicity, delivery type, or birth month. Among screened newborns, 8% failed their initial CSTS (nursery: 10% vs NICU/floor: 5%). Of the 379 newborns who failed their initial screen, 269 (71%) passed on their second attempt whereas 341 (90%) passed a repeat CSTS at any point before discharge. Among screened newborns, <1% (38/4728) had no passing CSTS recorded before discharge. Performance on repeat CSTS was similar for newborns tested in the nursery as compared with those in the NICU/floor with second attempt pass rates of 72% and 68%, respectively (89% and 93% before discharge). Supplemental Table 5 describes the 2833 newborns for whom postdischarge data were available and is similar to the overall study population.

Unadjusted median total LOS was 88 hours (IQR 61–196) but varied substantially by discharging unit (nursery: 68 vs NICU/floor: 224) (Table 3). Newborns undergoing screening appeared to have a significantly shorter total LOS by 15.3 hours (95% confidence interval [CI]: 22.6–8.1). After adjustment for discharging unit and other covariates, this was no longer significant (6.1 hours longer; 95% CI: -1.2–13.3) (Table 4). Newborns who failed screening did experience significantly longer adjusted total LOS by 16.1 (5.6–26.7) and 10.5 (95% CI: 3.0–18.0) hours as compared with newborns who were not screened or who passed the initial CSTS, respectively. Newborns who passed the CSTS did not experience significantly longer birth hospitalization LOS as compared with unscreened newborns (5.1 hours; 95% CI: -2.2–12.3).

Initial screening occurred a median of 21 hours before discharge, regardless of unit (Table 3). However, newborns who passed the initial CSTS were screened a median of 20.4 hours before discharge, whereas newborns who initially failed screening were discharged 37.3 hours after their initial CSTS, a difference of 17.0 hours (95% CI: 14.4–19.5) (Fig 1). This differed by unit (nursery: 13.9 hours; 95% CI: 10.9–16.9 vs NICU/inpatient service: 49.2 hours; 95% CI: 42.5–55.9).

Among 2883 infants from System B with available follow-up data, 152 (5.3%) experienced an all-cause hospital revisit within 30 days of discharge (Table 3). No difference in subsequent hospital utilization was noted between screened and unscreened newborns (adjusted odds ratio: 0.78; 95% CI: 0.44–1.40) or between newborns who failed the CSTS and those who passed (adjusted odds ratio: 0.81; 95% CI 0.43–1.56) (Table 4). Analyses stratified by discharging unit also showed no differences. No infants experienced a hospital revisit for a definite CSTS-related reason. Ten or fewer infants experienced a possibly CSTS-related hospital revisit or a BRUE diagnosis within 30 days of discharge (Table 3). All BRUE diagnoses occurred in infants who passed the CSTS, and none were associated with a car seat. Owing to the small numbers, no statistical analysis was performed for these outcomes.

Screening adherence was high, and CSTS failure was relatively common in this cohort of LPT and term low birth weight newborns. Neither routine screening nor CSTS failure were associated with 30 day hospital revisit or BRUE diagnosis although event rates were low. However, newborns failing screening experienced significantly longer birth hospitalization LOS. These findings highlight the necessity of additional investigation into the value of routine CSTS in regard to neonatal care experiences and health outcomes.

Using a limited subset of this data, we previously reported that higher gestational age (but younger chronologic age), higher birth weight, and later birth year were positively associated with CSTS failure, whereas similar factors plus insurance status were associated with screening adherence.^8,14  The current study expands on our previous work by including data from an additional health system and on postdischarge utilization. This updated analysis improves the generalizability of this work while offering insight into the important but understudied question as to the impact of routine CSTS on newborn outcomes.

Our results differed for newborns screened in the nursery or the NICU/floor. Although newborns failing the CSTS experienced longer total LOS in both settings, this difference was much longer for newborns screened outside of the nursery and appears to be explained by the delay between initial screening and discharge which, was 13.9 hours in the nursery (approximating the recommended 12 hours until repeating the CSTS) as compared with 49.2 hours in the NICU/floor. We observed this discrepancy despite similar second attempt pass rates in both settings. It is possible that clinicians perceive infants cared for outside of the nursery as higher risk despite otherwise having been deemed ready for discharge and may take a more conservative approach to the management of a failed screen (eg, resetting the apnea/bradycardia clock or pursuing additional diagnostic tests). However, delaying discharge beyond the time necessary to repeat screening may not be warranted given the poor reproducibility and reliability of the CSTS.^19,20 

Infants failing the CSTS are most commonly retested, although practice varies.^6,7,21  This is consistent with our study hospitals’ CSTS policies, which recommend repeat testing but defer additional decisions, including transfer to a higher level of care, to clinician discretion. Ultimately, 11% of study newborns who failed in the nursery were transferred to the NICU or floor before discharge, which is lower than what has been previously reported.^12,15,22  In 1 study, 21% transferred to the NICU in which all 3 infants were found to have significant pathology requiring intervention.¹²  In another, 55% transferred to the NICU in which two-thirds received additional interventions, including chest radiography, echocardiography, and sepsis work-ups, all of which were normal or failed to explain the CSTS result.¹⁵  Twenty-eight percent of those transferred did, however, receive supplemental oxygen for desaturations noted while being monitored in the NICU. In the final study, every newborn failing screening was transferred to the NICU in which 39% were diagnosed with apnea.²²  This suggests that CSTS plus clinical suspicion can uncover meaningful pathology in some infants. Additional study is needed to investigate whether standardizing CSTS processes could improve the ability to identify which infants failing the initial screen would most likely benefit from additional evaluation.

Our findings add to existing evidence that newborns who fail the CSTS experience similar postdischarge outcomes to those who pass.^10,11  Using parent surveys, Sharma and Davis found that newborns who failed screening were more likely to be diagnosed with asthma or obstructive sleep apnea in the first years of life but experienced similar numbers of emergency department visits and hospital readmissions.¹⁰  Jensen et al performed a larger study among newborns discharged from the NICU, finding no significant difference in the primary outcome (a composite of 30 day mortality or all-cause readmission) between newborns who passed and those who failed CSTS.¹¹  Failing was associated with significantly lower odds of readmission alone. The authors speculated this may have been due to interventions in response to the failed screen, including delaying discharge by 3 days, similar to the 49.2-hour difference we observed. Our study adds to these in 2 important ways. First, in contrast to Jensen et al, we included newborns cared for in the nursery, a population with unique risks, care processes, and potential to benefit from screening. In addition, a majority of CSTS take place in the nursery. Second, we included newborns who qualified for routine screening yet failed to undergo CSTS. Because this group encompasses both newborns who would have passed and who would have failed had the CSTS been administered, it serves as a control group to assess the population-level impact of routine screening because comparisons limited to screened newborns may be confounded by subsequent interventions after CSTS failure.

Routine CSTS became the standard of care after small studies revealed altered physiologic parameters.^1,2  Three decades later, high-quality evidence revealing improved patient outcomes is still lacking. For this reason, the Canadian Pediatric Society withdrew its support for routine CSTS in 2016, instead recommending families be counseled on appropriate car seat use emphasizing fit and limiting use to travel.²³  In addition, clinician surveys reveal ambivalence regarding CSTS utility, with just greater than a third of nursery providers feeling that it is a good screening test to assess discharge readiness.⁷  In contrast, two-thirds of neonatologists believe the CSTS medically necessary, yet 75% retained equipoise to support a randomized controlled trial of more versus less restrictive criteria, whereas nearly half endorsed a trial of CSTS versus no screening.²¹  This equipoise justifies and supports prospective studies of routine CSTS focused on patient outcomes, including neurodevelopment and mortality, as well as subsequent health care utilization.²¹  Studies should also seek to understand that families’ experiences as caregivers may cause them to feel a range of negative emotions after their newborn fails the CSTS, as has been described with other routine newborn screens.^24–26  Screening may also impact caregiver behavior in ways that are protective should a failed screen prompt closer attention to their newborn’s breathing in the car seat, or in ways that are problematic if reassurance after a negative screen result leads to inappropriate use (ie, prolonged, unattended stretches outside of travel).

Our results suggesting no definitive benefit of routine CSTS in LPT and term low birth weight newborns must be interpreted considering our study’s limitations and should be viewed as exploratory. We attempted to minimize selection bias by studying a cohort of infants recommended for universal screening; however, it is possible that screening adherence was nonrandom, and important unobserved confounders affected our results, highlighting the necessity of future randomized controlled trials. Variable missingness or misclassification is another limitation, given our retrospective design. An example is our inability to measure car bed use since this or other postscreening interventions may have been protective for infants failing screening. We were likely underpowered for the postdischarge outcomes because this information was only available for approximately half of our total cohort. Additionally, we observed readmission rates lower than those typically reported for LPT newborns.²⁷  This may be explained by System B serving as a referral center for high-risk pregnancies, so postdischarge utilization closer to home would not be captured. However, we did measure utilization across the system’s statewide network of hospitals, and it is unlikely that newborns from the 3 screening groups would have differentially presented for care outside of the system (ie, distance between birth hospital and home is not associated with risk of CSTS failure). Finally, nearly all revisits were for CSTS-unrelated reasons. This highlights a challenge likely to be encountered in designing prospective trials, further supporting the inclusion of more proximal measures, such as costs and caregiver experience. It also supports pursuing large, population-based epidemiologic studies (ie, Canadian versus United States experiences) for rare outcomes like mortality. The final limitation is that our findings should not be applied to newborns otherwise identified as higher-risk (eg, trisomy 21, airway anomalies, etc) because we aimed to assess the utility of routine CSTS rather than targeted testing.

In this large, retrospective study, routine CSTS was not associated with health care utilization within 30 days of discharge among LPT and term low birth weight newborns, whereas a failed screen was associated with prolonged birth hospitalization LOS, particularly for newborns screened outside the nursery. Prospective trials are needed to evaluate this routine practice and its utility for otherwise low-risk infants.

We would like to acknowledge Emily Hannon, MD, for her help in collecting the data.