BACKGROUND AND OBJECTIVES

Neonatal opioid withdrawal syndrome (NOWS) is associated with long and costly birth hospitalization and increased readmission risk. Our objective was to examine readmissions in the first year of life for infants diagnosed with NOWS compared with infants without NOWS, adjusting for sociodemographic and clinical factors, and to describe use during readmissions in this population.

METHODS

Using data from the Pediatric Health Information System, we identified singleton term infants with NOWS and without NOWS or other major condition (by diagnosis codes and All Patient Refined Diagnosis Related Groups coding, respectively) discharged from 2016 to 2019. We predicted time to first readmission within the first year of life using Cox regression analysis. Predictors included NOWS diagnosis, sociodemographic factors, birth NICU use, and birth weight.

RESULTS

We included 155 885 birth discharges from 17 hospitals (n = 1467 NOWS) with 10 087 readmissions. Unadjusted 1-year readmission rates were 9.9% among NOWS infants versus 6.2% among those without NOWS. The adjusted hazard ratio for readmission within the first year was 1.76 (95% confidence interval: 1.40–2.22) for infants with NOWS versus those without. Readmissions for infants with NOWS were longer and costlier and more likely to require intensive care and mechanical ventilation. Readmissions among infants without NOWS were most commonly for jaundice and respiratory and other infections, whereas respiratory infections were the leading cause of readmissions among NOWS infants.

CONCLUSIONS

Infants with a NOWS diagnosis were more likely to be readmitted within the first year of life. In future work, researchers should explore potential interventions to prevent readmissions and provide resources to families affected by opioid dependence.

Incidence of neonatal opioid withdrawal syndrome (NOWS) has risen over the past 2 decades, paralleling the opioid epidemic in the United States.15  With no established treatment protocol or clinical guidelines despite calls for standardization, treatment varies widely from neonatal pharmacotherapy with opioids and/or second-line agents phenobarbital and clonidine to conservative nonpharmaceutical treatment, including maternal rooming-in, encouraging breastfeeding, and nonpharmacologic bundles.613  Previous work has revealed wide variation in NOWS treatment.7,14,15  Overall hospital use and costs for these neonates during birth hospitalization have been well-described, but readmissions have not been extensively studied.14,16,17 

Previous studies have revealed variable readmission rates among infants with NOWS, with 30-day readmissions ranging from 3% to 9% and 1 estimate of 8% within the first year.9,14,18  Most literature examining readmissions among infants with NOWS is regionally specific, with geographically diverse samples being used in few published studies. A number of studies using regional or state-level data have revealed increased odds of readmission in the neonatal period through infancy into childhood compared with infants without NOWS.1821  Readmissions among infants with NOWS tend to be for longer, higher-acuity stays and are more common for infections, respiratory diseases, and injuries.1820 

Studies in term newborns without NOWS indicate first-year readmission rates ranging from 2% to 8%, with most readmissions within the first 30-days.18,2224  In previous studies, researchers have found readmission rates associated with race, sex, and insurance, as well as clinical factors, including birth weight, gestational age, and birth length of stay (LOS).22,23,25  Male, Black, or Hispanic and infants with public insurance are at greater risk of readmission, as well as earlier term, lower birth weight infants.22,23  Common reasons for readmission are jaundice, feeding difficulties, temperature regulation issues, respiratory infections, neonatal conditions (eg, pyloric stenosis and reflux), and respiratory distress.22,23,26 

Given the limited literature examining readmissions for infants with NOWS especially in a geographically diverse cohort, we analyzed data from pediatric hospitals across the United States to examine all-cause readmissions within the first year of life for infants diagnosed with NOWS compared with infants without NOWS adjusting for sociodemographic and clinical factors, as well as describing reason for readmission, costs, and LOS.

Using administrative data from the Pediatric Health Information System (PHIS) on inpatient discharges from 50 tertiary care pediatric hospitals, we examined all-cause readmissions within the first year of life for infants diagnosed with NOWS at birth compared with infants without NOWS born during the same time period. Among infants with NOWS, we also examined differences in readmissions for those treated pharmacologically to those without pharmacotherapy.

This study included infants diagnosed with NOWS during birth hospitalization and infants discharged without a birth discharge diagnosis of NOWS or any major condition. Both groups included all infants born or transferred to a PHIS hospital on the day of birth who were born and discharged from the hospital from January 1, 2016, to December 31, 2019. We included all inpatient readmissions occurring within the first year of life during the same time period. Diagnosis of NOWS was determined by diagnosis codes (International Classification of Diseases 10th Revision code P96.1: neonatal withdrawal symptoms from maternal use of drugs of addiction). Birth discharges without a diagnosis of NOWS or other major condition were identified by All Patient Refined Diagnosis Related Groups (APR-DRGs) classification on the basis of reason for admission, severity of illness, and risk of mortality (APR-DRG 626: neonate birth weight 2000–2499 g, normal newborn or neonate with other problem; APR-DRG 640: neonate birth weight >2500 g, normal newborn or neonate with other problem).27  NOWS diagnosis results in an APR-DRG indicating a newborn requiring a higher level of care; thus, our group definitions of infants with and without NOWS are mutually exclusive.

Using flags available in PHIS, we excluded neonates with mechanical ventilation, extracorporeal membrane oxygenation, congenital or genetic abnormalities, or those who underwent any surgical procedure during birth hospitalization. Preterm or low birth weight infants were excluded by using diagnosis codes or gestational age (<37 weeks) and birth weight (≤1500 g) if available. We restricted our analysis to singleton births, excluding twins and other multiples by diagnosis codes. This approach is consistent with previous studies in which researchers used administrative data, allowing for comparisons within a relatively homogenous, healthy group of neonates by limiting comorbidities that would contribute to readmission burden.3,7,14  Similar to previous studies, we excluded hospitals caring for fewer than 15 NOWS birth discharges or fewer than 150 birth discharges without NOWS over the study period.7,14 

Variables collected during birth hospitalization included LOS, admission to the NICU, birth weight, race and/or ethnicity, sex, insurance payer, and median household income by zip code. Income was divided into 3 categories on the basis of tertiles of income in the United States.28  Pharmacologic treatment of NOWS was determined by billing codes for opioids or opioid agonists, phenobarbital, or clonidine during birth hospitalization. Information on type, timing, or duration of maternal opioid exposure during pregnancy was not available in PHIS.

LOS, admission to ICU or NICU, costs, mechanical ventilation, reason for readmission, and in-hospital mortality were collected for readmissions and linked to birth hospitalization records by encrypted medical record numbers included in PHIS. Primary reason for readmission was classified on the basis of principal diagnosis codes into major diagnoses (eg, jaundice, lower respiratory infections, and respiratory distress).

Standardized costs were provided by PHIS and are derived from charges converted to (direct and indirect) costs according to hospital-specific ratios of costs to charges, adjusted for geographic region by using Centers for Medicare and Medicaid Services wage and price index, and standardized to eliminate between and within-hospital cost variation for individual items or services.29,30  Cost standardization is based on a cost master index (CMI) derived from median-of-hospital median costs for every Clinical Transaction Classification (CTC) coded item or service.29,30  Total standardized cost for each discharge is calculated as the sum of CMI-based cost times the number of units for each CTC code for all items or services provided in a given admission with a valid CTC code and cost available in the CMI.29,30 

The primary dependent variable was time to first readmission within the first year of life, and we also examined unadjusted readmissions at discrete time points during the first year. All infants had been discharged from birth hospitalization by the end of the study period, and discharges with fewer than 1-year follow-up time were censored on the last day of the study period.

We report frequencies (percentages) for categorical variables and mean (SD) or median (interquartile range [IQR]) for continuous variables. We compared infants with NOWS with those without NOWS on birth use, birth weight, and sociodemographic factors, accounting for hospital clustering using Cochran-Mantel-Haenszel χ2 tests for categorical variables and generalized estimating equations for continuous variables. Readmissions use and costs were compared between infants with and without NOWS and by using Wilcoxon rank test for continuous variables, Poisson regression for cumulative variables (total bed days and costs), and χ2 tests for categorical variables. Differences in readmission reasons between infants with and without NOWS and were assessed by using the signed rank test.

We compared time to first readmission within the first year of life between infants with and without NOWS using Cox regression analysis. Because of differences in birth LOS between infants with NOWS versus those without, time to readmission outcome was defined as days since birth, and models accounted for delayed entry into the at-risk period (left truncation) by specifying entry at time of birth discharge. We included birth NICU use, birth weight, and sociodemographic factors as model covariates, and assessed proportionality and time-varying effect of NOWS on readmissions by testing an interaction between log(time) and NOWS diagnosis. We also tested fixed-effect interaction terms between NOWS and other predictors. All regression models accounted for hospital clustering by using a robust sandwich estimator to adjust standard errors. As an exploratory analysis, we examined association within the NOWS cohort between pharmacologic treatment and time to first readmission using Cox regression.

All analyses were performed by using SAS (version 9.4; SAS Institute, Inc, Cary, NC) and P < .05 was considered statistically significant.

A total of 155 885 birth discharges (n = 1467 NOWS) from 17 hospitals were included, with a median of 7150 discharges per hospital (range 440–22 050). A median of 7098 discharges for infants without NOWS (range 162–22 021) and 49 NOWS discharges (range 15–337) per hospital were included.

Infants diagnosed with NOWS were more likely to be white race, with public insurance, and from lower income zip codes (P < .001) (Table 1). Average birth LOS was 2.2 days (SD = 0.9) for infants without NOWS versus 17.4 days (SD = 12.5) for those with NOWS (P < .001). Nearly all infants without NOWS were discharged within 4 days, whereas the majority of infants with NOWS stayed longer than 15 days (P < .001). Only 10% of infants without a NOWS diagnosis were admitted to the NICU during birth hospitalization versus 83% of infants with NOWS (P < .001). Infants with NOWS had slightly lower birth weights compared with those without NOWS (P < .001).

TABLE 1

Sociodemographic Factors and Use During Birth Hospitalization Among Infants Diagnosed With NOWS Compared With Infants Without NOWS Across 17 PHIS Hospitals, Births 2016–2019 (N = 155 885)

Infant CharacteristicsInfants Without NOWS (n = 154c418)Infants With NOWS (n = 1467)Pa
Race and/or ethnicity, n (%)    
 White 64 605 (41.8) 1103 (75.2) <.001 
 Black 16 971 (11.0) 83 (5.7) — 
 Hispanic 31 504 (20.4) 110 (7.5) — 
 Other 31 504 (17.9) 81 (5.5) — 
 Unknown 13 658 (8.8) 90 (6.1) — 
Male sex, n (%) 78 087 (50.6) 743 (50.7) .44 
Insurance payer, n (%)    
 Commercial and/or private 83 300 (53.9) 97 (6.6) <.001 
 Public 65 866 (42.7) 1336 (91.1) — 
 Other 5252 (3.4) 34 (2.3) — 
Median household income, n (%)    
 <$40 000 60 162 (39.0) 787 (53.7) <.001 
 $40 000–$89 999 85 532 (55.4) 650 (44.3) — 
 >$90 000 6660 (4.3) 12 (0.8) — 
 Unknown 2064 (1.3) 18 (1.2) — 
Birth wt, g, mean (SD) 3378.0 (455.9) 3061.0 (484.0) <.001 
Birth wt category, n (%)    
 ≤2500 g 3752 (2.3) 170 (11.6) <.001 
 >2500 g 150 759 (97.6) 1291 (88.0) — 
 Unknown 87 (0.1) 6 (0.4) — 
LOS, d, mean (SD) 2.2 (0.9) 17.4 (12.5) <.001 
LOS category, n (%)    
 ≤4 d 151 006 (97.8) 90 (6.1) <.001 
 5–9 d 3412 (2.2) 360 (24.5) — 
 10–14 d 0 (0.0) 222 (15.1) — 
 ≥15 d 0 (0.0) 795 (54.2) — 
NICU admission, n (%) 16 041 (10.4) 1216 (82.9) <.001 
Infant CharacteristicsInfants Without NOWS (n = 154c418)Infants With NOWS (n = 1467)Pa
Race and/or ethnicity, n (%)    
 White 64 605 (41.8) 1103 (75.2) <.001 
 Black 16 971 (11.0) 83 (5.7) — 
 Hispanic 31 504 (20.4) 110 (7.5) — 
 Other 31 504 (17.9) 81 (5.5) — 
 Unknown 13 658 (8.8) 90 (6.1) — 
Male sex, n (%) 78 087 (50.6) 743 (50.7) .44 
Insurance payer, n (%)    
 Commercial and/or private 83 300 (53.9) 97 (6.6) <.001 
 Public 65 866 (42.7) 1336 (91.1) — 
 Other 5252 (3.4) 34 (2.3) — 
Median household income, n (%)    
 <$40 000 60 162 (39.0) 787 (53.7) <.001 
 $40 000–$89 999 85 532 (55.4) 650 (44.3) — 
 >$90 000 6660 (4.3) 12 (0.8) — 
 Unknown 2064 (1.3) 18 (1.2) — 
Birth wt, g, mean (SD) 3378.0 (455.9) 3061.0 (484.0) <.001 
Birth wt category, n (%)    
 ≤2500 g 3752 (2.3) 170 (11.6) <.001 
 >2500 g 150 759 (97.6) 1291 (88.0) — 
 Unknown 87 (0.1) 6 (0.4) — 
LOS, d, mean (SD) 2.2 (0.9) 17.4 (12.5) <.001 
LOS category, n (%)    
 ≤4 d 151 006 (97.8) 90 (6.1) <.001 
 5–9 d 3412 (2.2) 360 (24.5) — 
 10–14 d 0 (0.0) 222 (15.1) — 
 ≥15 d 0 (0.0) 795 (54.2) — 
NICU admission, n (%) 16 041 (10.4) 1216 (82.9) <.001 
a

P value accounting for interhospital correlations by using Cochran-Mantel-Haenszel test for categorical variables and generalized estimating equations for continuous LOS.

Among infants without NOWS, 9581 (6.2%) had any readmission within the first year of life versus 145 (9.9%) infants with NOWS (P < .001). At first readmission, the median age was 35 days (IQR = 114) among infants without NOWS versus 76 days (IQR = 119) among infants with NOWS (P < .001), whereas the time since birth discharge was 33 days (IQR = 114) vs 55 days (IQR = 122), respectively (P < .001). Of those with any readmission in the first year, 346 (4%) infants without NOWS had >1 readmission (maximum = 3), and 14 (10%) infants with NOWS had >1 readmission (maximum = 2) for a total of 10 087 readmissions.

Use, cost, and clinical characteristics for all readmissions are presented in Table 2. Median LOS per readmission was 2 days (IQR = 2) for infants without NOWS versus 2 days (IQR = 3) among infants with NOWS (P = .001). Infants with NOWS were more likely to have longer readmission stays (incidence rate ratio [IRR] =1.37; 95% confidence interval [CI]: 1.26–1.48; P < .001). Median standardized costs per readmission were higher for infants with NOWS (P = .002) and cumulatively costlier compared with those without NOWS (IRR = 1.35, 95% CI: 1.34–1.35; P < .001). Readmissions among infants with NOWS occurred later in infancy. ICU readmissions were more likely among infants with NOWS (P < .001), whereas NICU readmissions were more likely among those without NOWS (P = .002). Mechanical ventilation was also more likely during readmissions among infants with NOWS (P = .002). Mortality during readmissions was rare and did not differ significantly by NOWS diagnosis (P = .21).

TABLE 2

Use, Cost, and Clinical Characteristics During Readmissions in the First Year Among Infants Diagnosed With NOWS Compared With Infants Without NOWS Across 17 PHIS Hospitals, Births 2016–2019 (N = 155 885 infants; n = 10 087 readmissions)

Readmission UseInfants Without NOWS (n = 9928 Readmissions)Infants With NOWS (n = 159 Readmissions)Pa
LOS, median (IQR) 2 (1–3) 2 (1–4) .001 
Cumulative bed days, sum 26 766 583 <.001 
Standardized cost, $, median (IQR) 4429 (2679–7420) 5256 (3037–10426) .002 
Cumulative cost, sum, $ 79 950 667 1 719 133 <.001 
Age at readmission, d, median (IQR) 37 (7–127) 82 (40–164) <.001 
Time since birth discharge, d, median (IQR) 35 (4–25) 59 (23–148) <.001 
NICU admission, n (%) 1065 (10.7) 5 (3) .002 
ICU admission, n (%) 834 (8.4) 27 (17) <.001 
Mechanical ventilation, n (%) 244 (2.5) 11 (7) .002 
Mortality, n (%) 14 (0.1) 1 (1) 0.21 
Readmission UseInfants Without NOWS (n = 9928 Readmissions)Infants With NOWS (n = 159 Readmissions)Pa
LOS, median (IQR) 2 (1–3) 2 (1–4) .001 
Cumulative bed days, sum 26 766 583 <.001 
Standardized cost, $, median (IQR) 4429 (2679–7420) 5256 (3037–10426) .002 
Cumulative cost, sum, $ 79 950 667 1 719 133 <.001 
Age at readmission, d, median (IQR) 37 (7–127) 82 (40–164) <.001 
Time since birth discharge, d, median (IQR) 35 (4–25) 59 (23–148) <.001 
NICU admission, n (%) 1065 (10.7) 5 (3) .002 
ICU admission, n (%) 834 (8.4) 27 (17) <.001 
Mechanical ventilation, n (%) 244 (2.5) 11 (7) .002 
Mortality, n (%) 14 (0.1) 1 (1) 0.21 

IQR, interquartile range.

a

P value from Wilcoxon rank test for continuous variables, Poisson regression for cumulative bed days and cost, and χ2 test (or Fisher’s exact test, if appropriate) for categorical variables.

Among infants without NOWS, the most common reasons for readmission were neonatal jaundice, lower respiratory infections, other infections, respiratory distress, other conditions specific to neonates, fever or temperature dysregulation, and congenital and genetic disorders (Table 3). The most common reasons for readmission among infants with NOWS were lower respiratory infections, congenital and genetic disorders, other infections, injury, unintentional injury, trauma or poisoning, other conditions specific to neonates, respiratory distress, and NOWS. Ranking of reasons for readmission differed significantly between infants with and without NOWS (P < .001).

TABLE 3

Leading Reasons for First-Year Readmissions Among Infants Diagnosed With NOWS Compared With Infants Without NOWS Across 17 PHIS Hospitals, Births 2016–2019 (n = 10 087 Readmissions)

RankInfants Without NOWS (n = 9928 Readmissions)n (%)Infants With NOWS (n = 159 Readmissions)n (%)
Jaundice 1783 (18.0) Lower respiratory infectionsa 58 (36) 
Lower respiratory infectionsa 1733 (17.5) Congenital and genetic disordersb 12 (8) 
Other infections 1171 (11.8) Other infections 12 (8) 
Respiratory distress 720 (7.3) Injury, unintentional injury, trauma, or poisoning 10 (6) 
Other conditions specific to neonatesc 712 (7.2) Other conditions specific to neonatesc 10 (6) 
Fever or temperature dysregulation 629 (6.3) Respiratory distress 9 (6) 
Congenital and genetic disordersb 594 (6.0) NOWS 8 (5) 
RankInfants Without NOWS (n = 9928 Readmissions)n (%)Infants With NOWS (n = 159 Readmissions)n (%)
Jaundice 1783 (18.0) Lower respiratory infectionsa 58 (36) 
Lower respiratory infectionsa 1733 (17.5) Congenital and genetic disordersb 12 (8) 
Other infections 1171 (11.8) Other infections 12 (8) 
Respiratory distress 720 (7.3) Injury, unintentional injury, trauma, or poisoning 10 (6) 
Other conditions specific to neonatesc 712 (7.2) Other conditions specific to neonatesc 10 (6) 
Fever or temperature dysregulation 629 (6.3) Respiratory distress 9 (6) 
Congenital and genetic disordersb 594 (6.0) NOWS 8 (5) 
a

Includes pneumonia, respiratory syncytial virus, and bronchiolitis.

b

Includes congenital diagnoses such as the following: craniofacial anomalies, congenital heart disease, congenital pyloric stenosis, cystic fibrosis, sickle cell disease.

c

Includes conditions such as the following: newborn reflux, intussusception, inguinal or umbilical hernias, colic, and other conditions specific to newborns.

In Table 4, we present unadjusted readmission rates and hazard ratios (HRs) for infants with versus without NOWS for a number of discrete time points in the first year of life. The instantaneous hazard for readmission was higher among infants with NOWS through the first year. By 30 days, 2.9% of infants without NOWS had been readmitted versus 1.5% for those with NOWS (P = .02) and by 1 year, nearly 10% of infants with NOWS had been readmitted versus 6.2% for those without (P < .001). NOWS diagnosis was associated with an increased hazard of readmission at 1 year (HR = 2.40; 95% CI: 1.85–3.11; Table 4; adjusted HR =1.76; 95% CI: 1.40–2.22) (Fig 1). Male sex, public insurance, and lower birth weight were also associated with increased hazard of readmission. There were no significant interactions between NOWS diagnosis and other clinical and sociodemographic predictors. There was no evidence from the Cox model of violation of proportionality (P = .91).

FIGURE 1

Adjusted HRs for factors associated with readmission in the first year of life for infants across 17 PHIS hospitals, births 2016–2019 (N = 155 885).

FIGURE 1

Adjusted HRs for factors associated with readmission in the first year of life for infants across 17 PHIS hospitals, births 2016–2019 (N = 155 885).

Close modal
TABLE 4

Time From Birth to First Readmission Within the First Year Among Infants Diagnosed With NOWS Compared With Infants Without NOWS Across 17 PHIS Hospitals, Births 2016–2019 (n = 155 885 Infants)

Readmissions (Days Since Birth)Infants Without NOWS (n = 154c418)Infants With NOWS (n = 1467)Unadjusted HR (95% CI)b,cPb
At Risk, naReadmitted, n (%)At Risk, naReadmitted, n (%)
Within 30 d 154 418 4524 (2.9) 1293 20 (1.5) 1.66 (1.08–2.53) .02 
Within 60 d 154 418 5955 (3.9) 1453 61 (4.2) 2.42 (1.90–3.08) <.001 
Within 90 d 154 418 6718 (4.4) 1465 78 (5.3) 2.40 (1.89–3.05) <.001 
Within 180 d 154 418 7972 (5.2) 1467 113 (7.7) 2.53 (2.02–3.18) <.001 
Within 365 d 154 418 9581 (6.2) 1467 145 (9.9) 2.40 (1.85–3.11) <.001 
Readmissions (Days Since Birth)Infants Without NOWS (n = 154c418)Infants With NOWS (n = 1467)Unadjusted HR (95% CI)b,cPb
At Risk, naReadmitted, n (%)At Risk, naReadmitted, n (%)
Within 30 d 154 418 4524 (2.9) 1293 20 (1.5) 1.66 (1.08–2.53) .02 
Within 60 d 154 418 5955 (3.9) 1453 61 (4.2) 2.42 (1.90–3.08) <.001 
Within 90 d 154 418 6718 (4.4) 1465 78 (5.3) 2.40 (1.89–3.05) <.001 
Within 180 d 154 418 7972 (5.2) 1467 113 (7.7) 2.53 (2.02–3.18) <.001 
Within 365 d 154 418 9581 (6.2) 1467 145 (9.9) 2.40 (1.85–3.11) <.001 
a

Includes only infants discharged by a given time point.

b

From unadjusted Cox regression model, accounting for left truncation.

c

HR for infants with versus without NOWS diagnosis.

Among infants with NOWS, 1036 (70.6%) were treated pharmacologically; 98.1% received opiates, 8.4% phenobarbital, and 34.1% clonidine. Of those treated pharmacologically, 60.6% received opiates only, 37.5% received both opiates and a second-line treatment, and 1.9% received only phenobarbital or clonidine. Pharmacologic treatment was not associated with time to readmission (unadjusted HR = 1.38; 95% CI: 0.92–2.09; P = .12; adjusted HR = 1.19; 95% CI: 0.87–1.63; P = .27) (Supplemental Fig 2). Male sex and birth NICU use were associated with increased hazard of readmission for infants with NOWS.

In a geographically diverse cohort, we found infants with NOWS have nearly double the all-cause readmission rates at 1 year since birth and nearly double the instantaneous hazard of readmission within the first year of life. Readmissions for infants with NOWS were longer, costlier, and more likely to require ICU-level care and mechanical ventilation, whereas infants without a NOWS diagnosis were more likely to be admitted to the NICU. Reason for readmission also differed between infants with and without NOWS.

Previous studies have revealed 1-year readmissions for term newborns without NOWS ranging from 2% to 8%, with the majority occurring within the first 30 days since birth discharge.2225  Consistent with previous studies, we found an unadjusted readmission rate of 6.2% among infants without NOWS in the first year of life, with nearly half occurring in the first 30 days. Infants without a NOWS diagnosis were more likely to be readmitted to the NICU, likely explained by earlier readmission within the neonatal period. Among infants with NOWS, our unadjusted estimates of 30-day and 1-year readmission rates (1.5% and 9.9%, respectively) were similar to previously published estimates.18 

The most common reasons for readmission among infants without NOWS were jaundice, respiratory infections, temperature-regulating conditions, and neonatal conditions like pyloric stenosis and reflux, all findings which are also consistent with the literature.23  Top reasons for readmission among infants with NOWS were similar to previous studies of readmissions within the first year and into childhood.1820  Jaundice was the top reason for readmission among infants without NOWS, whereas none of the infants with NOWS were readmitted for this reason as we expected because of the substantially longer average birth LOS. In our cohort, >50% of infants with NOWS were hospitalized for longer than 15 days at birth, whereas 98% of infants without NOWS were discharged within 4 days. We, therefore, chose to examine readmission timing from birth, accounting for birth LOS directly in our Cox regression models by specifying delayed entry into the at-risk period postdischarge. Certain neonatal conditions, including jaundice, may not have manifested or been diagnosed within just a few days of birth, whereas infants with NOWS typically remain hospitalized over 2 weeks and may be more likely to have secondary diagnoses of these neonatal conditions during birth hospitalization.18  Findings were similar when examining traditional readmission timing relative to discharge rather than birth.

Although NICU admission during birth hospitalization was not associated with readmission in the primary analysis, we found NICU admission was associated with nearly double the hazard of readmission for infants with NOWS. Male sex and public insurance were associated with readmissions among infants with NOWS, as reported in previous studies.22,23  Pharmacologic treatment was not associated with readmission among infants with NOWS, similar to previous studies in which early readmissions are examined.6,14  Similar to previous studies, although the majority of infants with NOWS spent time in the NICU, there was large variation in NICU use at both the hospital and individual levels (range 20%–100% by hospital).7,14  Hospital practice regarding pharmacologic treatment and NICU use was also highly variable, as has been reported previously.7,14,15  Among those treated without pharmacotherapy, 63% were admitted to the NICU (range 0%–100% by hospital) versus 91% among those treated pharmacologically (range 23%–100%). NICU use and pharmacologic treatment were still high in this cohort despite growing evidence of the benefits of rooming-in and other conservative measures.6,810,31  As reported previously, pharmacologic treatment in the NICU was the most common, although some hospitals have begun to shift at least some pharmacotherapy outside of the NICU setting, although the impact on readmissions is unknown.14,15  We postulate that, in addition to reflecting practice variation, NICU use may be a proxy for NOWS severity, particularly having excluded other common reasons for NICU admission (eg, preterm or low birth weight).

Although less frequent, unintentional and intentional injuries were among the top reasons for readmission among infants with NOWS. This may suggest lower levels of social support and resources afforded to families and caregivers of infants with opioid exposure. Infant custody or maternal perinatal opioid exposure may play a role in these findings and are important considerations in future research and targeting interventions, although data on these factors were not available in PHIS. Our findings are similar to an Australian study which revealed children who had NOWS as infants were more likely to be readmitted for these diagnoses into later childhood and a study from Washington State in which researchers found 5-year readmissions were more common for child abuse and neglect among infants with NOWS.19,20  These findings indicate the need for additional social support into the “fourth trimester” after birth and hospital discharge for these vulnerable infants.32,33 

This study has a number of limitations. First, this analysis relied on administrative data, which may contain incomplete information on comorbidities or other clinical factors. We identified the NOWS cohort by birth discharge diagnosis codes, which may underestimate incidence because of variation in coding practices, with some hospitals coding all opioid-exposed infants as NOWS and others only capturing those requiring pharmacotherapy or additional monitoring. However, there is no other method for identification of NOWS within administrative data, and diagnosis codes have revealed high positive predictive value compared to gold standard clinical NOWS diagnosis.34  Our comparison group of infants without NOWS were identified by APR-DRG codes, which also rely on birth discharge diagnosis codes. The diagnosis codes used to identify our cohorts were mutually exclusive (ie, an infant diagnosed with NOWS could not be classified as an infant without NOWS or major condition by APR-DRG coding). It is possible for infants without a birth discharge NOWS diagnosis to be diagnosed with NOWS later during a readmission. This was extremely rare in our cohort, with only n = 7 infants discharged without NOWS at birth with a subsequent readmission for NOWS. A sensitivity analysis excluding these infants yielded virtually identical results. Second, data from tertiary care children’s hospitals available in PHIS may not be generalizable to other delivery hospitals (adult tertiary care or community hospitals). Third, we restricted the analysis to PHIS hospitals with large numbers of birth discharges and adjacent or affiliated birthing units and well-infant nurseries, which may not by representative of all PHIS hospitals. Additionally, readmissions are underestimated because only patients readmitted to the PHIS hospital are included because readmissions to non-PHIS tertiary care or community hospitals are not captured. It is difficult to comment on the potential for differential readmissions to non-PHIS hospitals by NOWS status because this may depend on the regionalization of NOWS care, urbanicity, time since birth, and other factors unavailable to us within the scope of this study. Finally, although we explored pharmacologic treatment among the NOWS cohort, there was no information available on maternal factors like perinatal medication-assisted treatment or exposure to prescribed or illicit opioids during pregnancy, which may impact birth outcomes for infants with NOWS, or nonpharmacologic factors like breastfeeding, which could impact readmission risk.

Despite these limitations, PHIS data allow for timely analysis of readmissions and robust estimates of cost. Standardized costs allow for direct comparisons by accounting for patient-, hospital-, and regional-level variation. Although only a subset of PHIS hospitals were included, this study is strengthened by the broad geographic distribution of hospitals in the United States, contributing to the existing NOWS readmissions literature in which state and regional trends are examined.1821 

Although opioid use declined in recent years, rates are rising as a result of the coronavirus disease 2019 pandemic, which has disrupted treatment of many with opioid dependence, exacerbating mental health issues and creating job, housing, and food insecurity.3537  Access to contraceptives and abortion services has been restricted, which may result in a rise in unintended pregnancies.38,39  This unfortunate confluence of factors may lead to an uptick of NOWS births. This suggests the need for ongoing surveillance of use and outcomes among these vulnerable infants. In future work, researchers should focus on readmission preventability and interventions to provide resources to families struggling with the impacts of opioid dependence.

Infants diagnosed with NOWS were more likely to be readmitted within the first year of life, compared with infants without NOWS, for longer, costlier stays. These findings underscore the importance of improving discharge support services into the postnatal period for mothers and families.

FUNDING: No external funding.

Ms Milliren conceptualized and designed the study, planned the analyses, acquired the data, conducted the initial analyses, drafted the initial manuscript, and revised the manuscript; Ms Melvin and Dr Ozonoff critically reviewed the planned analyses, reviewed initial and final results, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

1.
Bateman
BT
,
Hernandez-Diaz
S
,
Rathmell
JP
, et al
.
Patterns of opioid use in pregnancy in a large cohort of commercial insurance beneficiaries in the United States
.
Anesthesiology
.
2014
;
120
(
5
):
1216
1224
2.
Brown
JD
,
Doshi
PA
,
Pauly
NJ
,
Talbert
JC
.
Rates of Neonatal Abstinence Syndrome Amid Efforts to Combat the Opioid Abuse Epidemic
.
JAMA Pediatr
.
2016
;
170
(
11
):
1110
1112
3.
Patrick
SW
,
Davis
MM
,
Lehmann
CU
,
Cooper
WO
.
Increasing incidence and geographic distribution of neonatal abstinence syndrome: United States 2009 to 2012. [published correction appears in J Perinatol. 2015;35(8):667]
.
J Perinatol
.
2015
;
35
(
8
):
650
655
4.
Pryor
JR
,
Maalouf
FI
,
Krans
EE
,
Schumacher
RE
,
Cooper
WO
,
Patrick
SW
.
The opioid epidemic and neonatal abstinence syndrome in the USA: a review of the continuum of care
.
Arch Dis Child Fetal Neonatal Ed
.
2017
;
102
(
2
):
F183
F187
5.
Villapiano
NL
,
Winkelman
TN
,
Kozhimannil
KB
,
Davis
MM
,
Patrick
SW
.
Rural and urban differences in neonatal abstinence syndrome and maternal opioid use, 2004 to 2013
.
JAMA Pediatr
.
2017
;
171
(
2
):
194
196
6.
Wachman
EM
,
Grossman
M
,
Schiff
DM
, et al
.
Quality improvement initiative to improve inpatient outcomes for neonatal abstinence syndrome
.
J Perinatol
.
2018
;
38
(
8
):
1114
1122
7.
Patrick
SW
,
Kaplan
HC
,
Passarella
M
,
Davis
MM
,
Lorch
SA
.
Variation in treatment of neonatal abstinence syndrome in US children’s hospitals, 2004-2011
.
J Perinatol
.
2014
;
34
(
11
):
867
872
8.
Abrahams
RR
,
Kelly
SA
,
Payne
S
,
Thiessen
PN
,
Mackintosh
J
,
Janssen
PA
.
Rooming-in compared with standard care for newborns of mothers using methadone or heroin
.
Can Fam Physician
.
2007
;
53
(
10
):
1722
1730
9.
Holmes
AV
,
Atwood
EC
,
Whalen
B
, et al
.
Rooming-in to treat neonatal abstinence syndrome: improved family-centered care at lower cost
.
Pediatrics
.
2016
;
137
(
6
):
e20152929
10.
Grossman
MR
,
Berkwitt
AK
,
Osborn
RR
, et al
.
An initiative to improve the quality of care of infants with neonatal abstinence syndrome
.
Pediatrics
.
2017
;
139
(
6
):
e20163360
11.
Grossman
MR
,
Osborn
RR
,
Berkwitt
AK
.
Neonatal abstinence syndrome: time for a reappraisal
.
Hosp Pediatr
.
2017
;
7
(
2
):
115
116
12.
Sarkar
S
,
Donn
SM
.
Management of neonatal abstinence syndrome in neonatal intensive care units: a national survey
.
J Perinatol
.
2006
;
26
(
1
):
15
17
13.
Patrick
SW
,
Barfield
WD
,
Poindexter
BB
;
Committee on Fetus and Newborn, Committee on Substance Use and Prevention
.
Neonatal opioid withdrawal syndrome
.
Pediatrics
.
2020
;
146
(
5
):
e2020029074
14.
Milliren
CE
,
Gupta
M
,
Graham
DA
,
Melvin
P
,
Jorina
M
,
Ozonoff
A
.
Hospital variation in neonatal abstinence syndrome incidence, treatment modalities, resource use, and costs across pediatric hospitals in the United States, 2013 to 2016
.
Hosp Pediatr
.
2018
;
8
(
1
):
15
20
15.
Bogen
DL
,
Whalen
BL
,
Kair
LR
,
Vining
M
,
King
BA
.
Wide variation found in care of opioid-exposed newborns
.
Acad Pediatr
.
2017
;
17
(
4
):
374
380
16.
Patrick
SW
,
Schumacher
RE
,
Benneyworth
BD
,
Krans
EE
,
McAllister
JM
,
Davis
MM
.
Neonatal abstinence syndrome and associated health care expenditures: United States, 2000-2009
.
JAMA
.
2012
;
307
(
18
):
1934
1940
17.
Hall
ES
,
Wexelblatt
SL
,
Crowley
M
, et al
;
OCHNAS Consortium
.
A multicenter cohort study of treatments and hospital outcomes in neonatal abstinence syndrome
.
Pediatrics
.
2014
;
134
(
2
).
18.
Patrick
SW
,
Burke
JF
,
Biel
TJ
,
Auger
KA
,
Goyal
NK
,
Cooper
WO
.
Risk of hospital readmission among infants with neonatal abstinence syndrome
.
Hosp Pediatr
.
2015
;
5
(
10
):
513
519
19.
Uebel
H
,
Wright
IM
,
Burns
L
, et al
.
Reasons for rehospitalization in children who had neonatal abstinence syndrome
.
Pediatrics
.
2015
;
136
(
4
).
20.
Witt
CE
,
Rudd
KE
,
Bhatraju
P
,
Rivara
FP
,
Hawes
SE
,
Weiss
NS
.
Neonatal abstinence syndrome and early childhood morbidity and mortality in Washington state: a retrospective cohort study
.
J Perinatol
.
2017
;
37
(
10
):
1124
1129
21.
Hwang
SS
,
Diop
H
,
Liu
CL
, et al
.
Maternal substance use disorders and infant outcomes in the first year of life among Massachusetts singletons, 2003-2010
.
J Pediatr
.
2017
;
191
:
69
75
22.
Young
PC
,
Korgenski
K
,
Buchi
KF
.
Early readmission of newborns in a large health care system
.
Pediatrics
.
2013
;
131
(
5
).
23.
Schiltz
NK
,
Finkelstein Rosenthal
B
,
Crowley
MA
, et al
.
Rehospitalization during the first year of life by insurance status
.
Clin Pediatr (Phila)
.
2014
;
53
(
9
):
845
853
24.
McLaurin
KK
,
Hall
CB
,
Jackson
EA
,
Owens
OV
,
Mahadevia
PJ
.
Persistence of morbidity and cost differences between late-preterm and term infants during the first year of life
.
Pediatrics
.
2009
;
123
(
2
):
653
659
25.
Harron
K
,
Gilbert
R
,
Cromwell
D
,
Oddie
S
,
van der Meulen
J
.
Newborn length of stay and risk of readmission
.
Paediatr Perinat Epidemiol
.
2017
;
31
(
3
):
221
232
26.
American Academy of Pediatrics Committee on Fetus and Newborn
.
Hospital stay for healthy term newborns
.
Pediatrics
.
2004
;
113
(
5
):
1434
1436
27.
3M Health Information Systems, National Association of Children’s Hospitals and Related Institutions Inc., Medical Advisory Committee for NACHRI APR-DRG Research Project
.
All Patient Refined Diagnosis Related Groups (APR-DRGs): Methodology Overview
.
Wallingford, CT
:
3M Health Information Systems
;
2003
28.
Congressional Budget Office
.
The Distribution of Household Income, 2016
.
Washington, DC
:
Congressional Budget Office
;
2019
29.
Keren
R
,
Luan
X
,
Localio
R
, et al
;
Pediatric Research in Inpatient Settings (PRIS) Network
.
Prioritization of comparative effectiveness research topics in hospital pediatrics
.
Arch Pediatr Adolesc Med
.
2012
;
166
(
12
):
1155
1164
30.
Jonas
JA
,
Shah
SS
,
Zaniletti
I
, et al
.
Regional variation in standardized costs of care at children’s hospitals
.
J Hosp Med
.
2017
;
12
(
10
):
818
825
31.
Blount
T
,
Painter
A
,
Freeman
E
,
Grossman
M
,
Sutton
AG
.
Reduction in length of stay and morphine use for nas with the “eat, sleep, console” method
.
Hosp Pediatr
.
2019
;
9
(
8
):
615
623
32.
Ramage
M
,
Ostrach
B
,
Fagan
B
,
Coulson
CC
.
Stabilizing the mother-infant dyad for better outcomes from OB to FM: caring for patients with perinatal opioid use disorder through the 4th trimester
.
N C Med J
.
2018
;
79
(
3
):
164
165
33.
Schauberger
CW
,
Borgert
AJ
,
Bearwald
B
.
Continuation in treatment and maintenance of custody of newborns after delivery in women with opioid use disorder
.
J Addict Med
.
2020
;
14
(
2
):
119
125
34.
Maalouf
FI
,
Cooper
WO
,
Stratton
SM
, et al
.
Positive predictive value of administrative data for neonatal abstinence syndrome
.
Pediatrics
.
2019
;
143
(
1
):
e20174183
35.
Alexander
GC
,
Stoller
KB
,
Haffajee
RL
,
Saloner
B
.
An epidemic in the midst of a pandemic: opioid use disorder and COVID-19
.
Ann Intern Med
.
2020
;
173
(
1
):
57
58
36.
Slavova
S
,
Rock
P
,
Bush
HM
,
Quesinberry
D
,
Walsh
SL
.
Signal of increased opioid overdose during COVID-19 from emergency medical services data
.
Drug Alcohol Depend
.
2020
;
214
:
108176
37.
Wakeman
SE
,
Green
TC
,
Rich
J
.
An overdose surge will compound the COVID-19 pandemic if urgent action is not taken
.
Nat Med
.
2020
;
26
(
6
):
819
820
38.
Ferreira-Filho
ES
,
de Melo
NR
,
Sorpreso
ICE
, et al
.
Contraception and reproductive planning during the COVID-19 pandemic
.
Expert Rev Clin Pharmacol
.
2020
;
13
(
6
):
615
622
39.
Phumaphi
J
,
Mason
E
,
Alipui
NK
, et al
.
A crisis of accountability for women’s, children’s, and adolescents’ health
.
Lancet
.
2020
;
396
(
10246
):
222
224

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.