Hospital Variation in Neonatal Abstinence Syndrome Incidence, Treatment Modalities, Resource Use, and Costs Across Pediatric Hospitals in the United States, 2013 to 2016

We used administrative data from the Pediatric Health Information System (PHIS) on inpatients in 48 tertiary care pediatric hospitals and compared neonates with NAS to other neonates admitted during the same time period. Among neonates with NAS, we compared those treated pharmacologically to those treated without pharmacotherapy. The study was determined not human subjects research by the Boston Children’s Hospital Institutional Review Board.

All inpatient encounters with patients discharged between January 1, 2013 and March 31, 2016, and patients aged <30 days at admission were included. On the basis of flags provided within PHIS, we excluded neonates with mechanical ventilation, extracorporeal membrane oxygenation, congenital or genetic abnormalities, or preterm or low birth weight (by diagnosis code) and patients who underwent any surgical procedure after admission. These exclusions were chosen to limit comorbidities and allow for comparisons within a relatively healthy and more homogenous group of neonates. Rates of preterm birth or low birth weight were similar between neonates with and without NAS (23% vs 22%), and we therefore opted to exclude these neonates. We determined NAS status by using International Classification of Diseases diagnosis codes (Ninth Revision code 779.5: drug withdrawal syndrome in infant of dependent mother; and 10th Revision code P96.1: neonatal withdrawal symptoms from maternal use of drugs of addiction). Hospitals caring for fewer than 20 NAS discharges over the study period were excluded given that the findings in these might not be generalizable to the other PHIS hospitals. This approach is similar to other articles examining NAS by using PHIS.⁹  We further stratified NAS by treatment group by using pharmacologic billing codes. Neonates who received opioids, opioid agonists, barbiturates, or benzodiazepines were classified as pharmacologically-treated; all others were classified as treated without pharmacologic agents.

Resource use during neonatal hospitalization included total LOS in days, admission to the NICU, NICU LOS in days, readmissions (30- and 90-day), and adjusted total cost. Cost data from the PHIS database are derived from charges that are converted to costs (direct and indirect) according to hospital-specific ratios of costs to charges; these are adjusted for geographic region by using the Centers for Medicare and Medicaid Services wage index. We also examined in-hospital mortality.

We reported frequencies (percent) for categorical variables and mean (SD) for continuous variables. We compared neonates with NAS to those without NAS on demographic, socioeconomic, clinical characteristics, and use. For analyses, we used generalized estimating equations (GEEs) for continuous variables and Cochran-Mantel-Haenszel χ² tests for categorical variables to account for clustering by hospital. We made similar comparisons among neonates with NAS by treatment modality. All analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). We considered P < .05 statistically significant.

Among 48 hospitals and 313 664 neonatal discharges in the study period, 176 902 discharges were excluded, including those from 24 hospitals caring for fewer than 20 NAS discharges and 1 hospital without cost data (Supplemental Fig 2). Among 136 762 eligible neonatal encounters from 23 pediatric hospitals, 2% (n = 3264) had a diagnosis of NAS. Among 23 included hospitals, the number of neonatal encounters with NAS per hospital ranged from 21 to 516 (mean: 141.9; median: 61), and the percentage of neonatal encounters with NAS per hospital ranged from <1% to 18% (mean: 4%; median: 2%).

Demographic and clinical characteristics are presented in Table 1. Compared with neonates without NAS, neonates with NAS were more likely to be white, have public insurance, and reside in zip codes with lower median household income. Neonates with NAS had slightly lower birth weights, marginally higher Apgar scores at birth, longer LOSs, and more frequent admissions to the NICU with longer stays. Adjusted total costs per admission for neonates with NAS were over 10 times those of other neonates. Neonates with NAS had lower readmission rates than neonates without NAS at both 30 and 90 days. There were no significant differences in hospital mortality.

A total of 2294 (70%) of the neonates with NAS were treated pharmacologically, and this varied by hospital (mean: 65%; median: 71%; range: 13%–90%; P < .001; Fig 1). Table 2 shows the demographic and clinical characteristics of neonates with NAS by treatment group. Treatment groups differed by sex, race, and insurance, but not by Apgar score or birth weight. Pharmacologically-treated neonates with NAS experienced longer LOSs and more frequent NICU admissions. Pharmacotherapy was highly concordant with NICU admission, with 93% of pharmacologically-treated neonates with NAS admitted to the NICU, although variation by hospital was observed (mean: 82%; median: 100%; range: 0%–100%; P < .001; Fig 1). Adjusted total costs for pharmacologically-treated neonates were more than double those of other neonates with NAS. Readmissions and in-hospital mortality did not differ by pharmacologic treatment. The pharmacologic agents used most commonly were morphine (90%), phenobarbital (23%), and methadone (13%), and 20% of pharmacologically-treated neonates received both morphine and phenobarbital.

We observed differences between neonates with and without NAS and by NAS treatment modality at tertiary care pediatric hospitals in the United States during the 2013–2016 period. Compared with other neonates, those with NAS incurred higher costs of care and use of services. Neonates with NAS cost $34 000 more per admission and stayed a cumulative excess of 52 000 hospital bed days compared with other infants, representing a cumulative cost burden of $111 000 000. These results agree with other studies suggesting high costs of care for these infants and increased use of health care resources.^1,2,5,7  Our findings underscore the economic burden of the opioid epidemic via increased health care use and costs related to in-utero substance exposure. Of note, the vast majority of neonates with NAS receive public insurance (89%), resulting in increased hospital use and costs covered primarily by public payors.

Despite the recommendations of the American Academy of Pediatrics and others for standardization of NAS treatment,^14,15  our findings and others suggest persistent variation that is unlikely explained by case mix alone, particularly with regards to the use of pharmacologic therapy for NAS.^9,16,19  This variation may be amenable to standardization of practices, with the added benefit that a more conservative approach to first-line treatment, potentially without the use of pharmacotherapy, could result in savings in terms of both hospital use and cost.¹⁷  Over two-thirds of neonates with NAS in these hospitals received pharmacotherapy. If overall rates of pharmacologic therapy were reduced by 25% to the first quartile rate of 53%, this would result in nearly 600 fewer infants receiving pharmacotherapy, resulting in a savings of ∼6300 hospital bed days, 3700 NICU bed days, and $13 500 000 over the 3.25-year study period. A more substantial reduction of 50% would result in 660 fewer infants receiving pharmacotherapy and a potential savings of 7400 hospital bed days, 4300 NICU bed days, and nearly $16 000 000 over the study period.

Interestingly, a number of hospitals provided pharmacotherapy to a proportion of neonates outside of the NICU setting, with 1 hospital completely shifting pharmacotherapy to outside the NICU (n = 17), providing another potential opportunity to reduce costs and use. The average cost for these 17 encounters was $38 687 (SD: $35 071), and the average LOS was 16.1 days (SD: 13.3). These indicate modest cost savings as well as small reductions in LOS relative to all pharmacologically-treated neonates with NAS, although conclusions are limited given the small sample. A shift toward providing pharmacotherapy outside of the NICU would save both NICU bed days as well as overall bed days and costs while fostering opportunities for rooming-in, breastfeeding, and other nonpharmacologic interventions that have demonstrated improved short-term outcomes.^{4,11–14,16–18}  Standardization of pharmacotherapy in terms of duration, location, and choice of treatment also offers an opportunity to reduce use along with duration of opioid exposure.^13,17 

This study is limited by the use of administrative data, which may include incomplete data on comorbidities and other clinical factors. By identifying neonates with NAS by diagnosis codes, we may have underestimated the incidence of NAS, particularly among those infants not requiring pharmacologic therapy whose admissions may not include a diagnosis code for NAS. However, other methods for identifying NAS are not well-established in the context of administrative data. Additionally, although nonpharmacologic interventions are becoming more common and show improved outcomes, this information is not coded, and we were therefore unable to determine the degree of use for these interventions.

Our analysis includes data from a subset of tertiary care pediatric hospitals within PHIS, which may reflect different NAS populations and care compared with delivery hospitals. Given the limitations of the data, we cannot generalize to NAS discharges at all delivery hospitals (ie, adult tertiary and community hospitals). The included hospitals all had birthing units or nurseries or were adjacent to or affiliated with a nearby birthing hospital. However, PHIS data allow for a more timely analysis of NAS care data and more accurate estimates of costs, and the size of the study population and its broad geographic distribution support the validity of this analysis.

Neonates with NAS use more hospital resources during lengthier, more expensive hospital stays and incur higher costs, without benefits in use or mortality during neonatal admission. Among neonates with NAS, there is significant variation in treatment choices across hospitals, reflecting opportunities for standardization of practices and improvement in care, particularly regarding the use of pharmacologic therapy, which is also a driver of increased usage and costs.