OBJECTIVES:

To estimate the percentage of hospital births receiving antibiotics before being discharged from the hospital and efficiency diagnosing proven bloodstream infection.

METHODS:

We conducted a cross-sectional study of 326 845 live births in 2017, with a 69% sample of all California births involving 121 California hospitals with a NICU, of which 116 routinely served inborn neonates. Exposure included intravenous or intramuscular antibiotic administered anywhere in the hospital during inpatient stay associated with maternal delivery. The main outcomes were the percent of newborns with antibiotic exposure and counts of exposed newborns per proven bloodstream infection. Units of observation and analysis were the individual hospitals. Correlation analyses included infection rates, surgical case volume, NICU inborn admission rates, and mortality rates.

RESULTS:

The percent of newborns with antibiotic exposure varied from 1.6% to 42.5% (mean 8.5%; SD 6.3%; median 7.3%). Across hospitals, 11.4 to 335.7 infants received antibiotics per proven early-onset sepsis case (mean 95.1; SD 71.1; median 69.5), and 2 to 164 infants received antibiotics per proven late-onset sepsis case (mean 19.6; SD 24.0; median 12.2). The percent of newborns with antibiotic exposure correlated neither with proven bloodstream infection nor with the percent of patient-days entailing antibiotic exposure.

CONCLUSIONS:

The percent of newborns with antibiotic exposure varies widely and is unexplained by proven bloodstream infection. Identification of sepsis, particularly early onset, often is extremely inefficient. Knowledge of the numbers of newborns receiving antibiotics complements evaluations anchored in days of exposure because these are uncorrelated measures.

What’s Known on This Subject:

Current appraisals of newborn antibiotic use derive from the percent of patient-days entailing antibiotic exposure, a foundation that insufficiently informs benchmarking efforts and appraisals of underlying diagnostic efficiency.

What This Study Adds:

Identification of proven bloodstream infection is often extremely inefficient. Knowledge of the numbers of newborns receiving antibiotics and proven bloodstream infection cases complements newborn antibiotic use appraisals anchored in days of exposure because these are uncorrelated measures.

To measure antibiotic prescribing practice for neonates, organizations including the California Department of Health Care Services,1  Centers for Disease Control,2  and Vermont Oxford Network,3  track NICU days of treatment exposure, reporting antibiotic use rate (AUR) as the proportion of NICU patient-days entailing specified antibiotic exposures.

Values for AUR vary widely, prompting reflection on ranges consistent with proven bloodstream infection rates.4  Stewardship initiatives and publications3,510  reflect a broad-based view that AURs often are unnecessarily high. However, benchmarking is fraught with difficulty because a particular proportion of NICU patient-days entailing antibiotic exposures can reflect a wide variety of early-onset sepsis (EOS) and late-onset sepsis (LOS) risk profiles and approaches to initiation and duration of treatment.

Estimating the percentage of newborns with therapeutic antibiotic exposure and hospital-level sepsis diagnostic efficiency may provide complementary evaluative insight. To date, this has only been considered theoretically.5  In 2017, the California Department of Health Care Services began categorically tracking newborns receiving antibiotics during the inpatient stay associated with maternal delivery. With this study, we describe variation in these measures across a large sample of California hospitals with NICUs.

California Children’s Services, within the Integrated Systems of Care Division of the Department of Health Care Services, confers state approval to a hospital conforming to specified practice standards and annual reporting requirements of specific data elements for all births and NICU admissions. Upon such approval, payment is augmented above baseline state levels.11  Three designated levels of NICU care (regional, community, and intermediate11 ) generally correspond to American Academy of Pediatrics levels IV, III, and II, respectively.12 

All approved NICUs submit data to the California Perinatal Quality Care Collaborative,13  which collects data prospectively using an expanded version of the Vermont Oxford data set14,15  and a supplemental form, and annually submits a data set to California Children’s Services. Of 148 NICUs in California,16  121 contributed to the calendar year 2017 study data set; 5 served no inborn admissions routinely and their study data comprised only LOS, surgical volume, and mortality. The Stanford University Institutional Review Board approved this study.

All neonates receiving care at hospitals reporting on variables used for this study were included.

EOS counts the number of infants with bacterial or fungal infection diagnosed by blood culture within 2 days of birth. EOS rate is the number of infants with EOS per 1000 live births.

LOS counts infants with ≥1 bacterial or fungal infection diagnosed by blood culture at ≥3 days after birth. LOS rate is the percentage of infants with LOS among those with high illness acuity on admission, based on a birth weight of 401 to 1500 g or gestational age 22 to 31 weeks and 6 days (inclusive), and additional criteria for infants with birth weight >1500 g, specified in the California Perinatal Quality Care Collaborative Manual of Definitions.17  Thus, LOS at some hospitals may include outborn infants not reflected in counts of newborns with antibiotic exposures.

We did not compute aggregated early- and late-onset infection rates because the respective denominators (number of live births and number of NICU admissions with high illness acuity) are irreconcilable.

AUR is the percentage of NICU patient-days that infants were administered ≥1 antibacterial or antifungal agent intravenously or intramuscularly.

Newborns with antibiotic exposure is the number of inborn newborns who received ≥1 doses of an antibacterial or antifungal agent administered intravenously or intramuscularly in any location in the hospital during the inpatient stay associated with maternal delivery. The percentage of newborns with antibiotic exposure is the percentage of all live births with such exposure.

We estimate proven bloodstream infection (sepsis) diagnostic efficiency by computing the number of newborns exposed to antibiotics per proven EOS or LOS case or per all proven bloodstream infections. Computational details are provided in the Supplemental Information.

Calculating a hospital-level measure of diagnostic efficiency requires at least 1 proven bloodstream infection in the denominator (cannot divide by 0). For this reason, we could not compute efficiency for all proven bloodstream infections for 14 birth hospitals (102 hospitals remaining); similarly, we could not compute 31 hospitals’ efficiency in diagnosing EOS or LOS (85 birth hospitals remaining, respectively). Importantly, all 116 birth hospitals were included in correlation analyses of antibiotic exposures with proven bloodstream infection rates.

Inborn admission rate is the percentage of live births admitted to the NICU. Of 13 hospitals with self-designated freestanding NICUs organizationally independent of a maternal delivery service, 8 serve a colocated obstetric facility, and from these births, we computed inborn admission rates; the remaining 5 are freestanding children’s hospitals. Surgical case volume counts patients undergoing a surgical procedure, excluding circumcision, extracorporeal membrane oxygenation cannulation or decannulation, peritoneal dialysis catheter placement or removal, chest tube placement, and central line placement. NICU mortality rate is the percentage of NICU patients admitted who died before discharge.

The rate of necrotizing enterocolitis (NEC) (used for purposes outside the current study) is the percentage of infants 401 to 1500 g birth weight or 22 to 29 weeks’ gestation who were diagnosed with NEC either at surgery, postmortem examination, or by radiographic pneumatosis intestinalis, hepatobiliary gas, and/or pneumoperitoneum, accompanied by bilious gastric aspirate or emesis, abdominal distention, and/or gross or occult blood in stool with no apparent rectal fissure.

Units of observation and analysis were the individual hospital. Primary outcomes were the percentage of newborns with antibiotic exposure and diagnostic efficiency for EOS and for LOS. We computed statewide outcome values using numerators and denominators summed across all reporting hospitals. Before the study began, we hypothesized that the AUR and percentage of newborns with antibiotic exposure would correlate positively with proven bloodstream infection rates and with the NEC rate. In our post hoc analyses, we examined relationships among those antibiotic use measures and their individual correlations with inborn admission rate, surgical case volume, and NICU mortality rate, hypothesizing positive correlation with the 2 resource use measures and none with mortality. We estimated linear correlation by the Pearson correlation coefficient (ρ) with Bonferroni-adjusted significance levels for post hoc analyses. We tested sample mean differences using the 2-sample t test with unequal variances. Hypothesis tests were 2 sided with a significance level at P ≤ .05. We used Stata 15 (Stata Corp, College Station, TX)18  for analyses and graphical displays. To communicate maximal information at a glance about distributional ranges and densities, we created violin plots.

Of 121 hospitals reporting for 2017, 5 served no inborn admissions routinely and their study data comprised only LOS, surgical volume, and mortality; 116 routinely served inborn births, which is a prerequisite for reporting newborn antibiotic exposure, and provided summary data for 326 845 live births. The subgroup at risk for LOS (rate denominator) with high illness acuity on admission comprised 15 223 infants, <5% the size of the birth sample.

Statewide, 8.4% of newborns received antibiotics before being discharged from the hospital; hospital values ranged between 1.6% and 42.5% (mean 8.5%; SD 6.3%; median 7.3%). As shown in Fig 1, whether a hospital reported any EOS and its level of NICU care affected this variation. Higher outlier values appear among hospitals reporting 0 EOS; the single regional NICU reporting 0 EOS was outside the interquartile range for the other regional facilities. Variation was narrowest among intermediate NICUs reporting EOS >0.

FIGURE 1

Percentage of births exposed to antibiotics. A, Early sepsis = 0. B, Early sepsis > 0. The shaded bar and vertical lines identify interquartile range, median (white dot), 1.5 × interquartile range; symmetric curves to either side estimate the distribution of values (relative number of units at each vertical axis value). The numbers below each care level describe the number of hospitals and live births, respectively.

FIGURE 1

Percentage of births exposed to antibiotics. A, Early sepsis = 0. B, Early sepsis > 0. The shaded bar and vertical lines identify interquartile range, median (white dot), 1.5 × interquartile range; symmetric curves to either side estimate the distribution of values (relative number of units at each vertical axis value). The numbers below each care level describe the number of hospitals and live births, respectively.

Close modal

Table 1 details antibiotic exposure, burden of infection, and diagnostic efficiency, and Fig 2 stratifies variation by level of care. Statewide, 34.3 newborns were treated per any proven bloodstream infection. Hospitals varied >100-fold, treating 7.3 to 781 infants per proven bloodstream infection (mean 66.4; SD 91.7; median 41.3). Statewide, 88.8 newborns were treated per EOS case. Hospitals treated 11.4 to 335.7 infants per EOS case (mean 95.1; SD 71.1; median 69.5). Statewide, 10.4 newborns were treated per LOS case. Hospitals treated 2 to 106 infants per LOS case (mean 19.6; SD 24.0; median 12.2). LOS rates exceed EOS rates by more than an order of magnitude (36.7 per 1000 high-acuity admissions statewide vs 0.75 per 1000 live births statewide), and variation in LOS diagnostic efficiency is comparatively narrower. The EOS rate denominator of birth volume was significantly lower at hospitals with no EOS (mean 1883.2 and SD 880.1 vs mean 3158.4 and SD 1525.5; P < .0001).

TABLE 1

Distribution of EOS and LOS Rates, Percentage of All Live Births Who Received a Newborn Antibiotic Exposure and Sepsis Diagnostic Efficiency

Hospital-Level Mean (SD)10th Percentile25th Percentile50th Percentile75th Percentile90the PercentileLowestHighestStatewide
Percentage of births exposed to antibiotics 8.53 (6.27) 3.67 4.69 7.35 9.55 14.14 1.59 42.54 8.43 
Diagnostic efficiency, EOS + LOS 66.35 (91.70) 16.54 26.06 41.25 69.50 122.00 7.25 781.00 34.26 
EOS          
 Rate (cases per 1000 live births) 0.72 (0.69) 0.53 1.17 1.70 2.89 0.75 
 Diagnostic efficiency 95.08 (71.14) 33.44 46.87 69.52 122.84 178.54 11.45 335.75 88.82 
LOS          
 Rate (% of admissions with high illness acuity) 3.18 (3.10) 2.99 4.69 7.25 18.75 3.67 
 Diagnostic efficiency 19.60 (24.02) 3.88 7.09 12.18 22.36 36.96 2.02 164.01 10.35 
Hospital-Level Mean (SD)10th Percentile25th Percentile50th Percentile75th Percentile90the PercentileLowestHighestStatewide
Percentage of births exposed to antibiotics 8.53 (6.27) 3.67 4.69 7.35 9.55 14.14 1.59 42.54 8.43 
Diagnostic efficiency, EOS + LOS 66.35 (91.70) 16.54 26.06 41.25 69.50 122.00 7.25 781.00 34.26 
EOS          
 Rate (cases per 1000 live births) 0.72 (0.69) 0.53 1.17 1.70 2.89 0.75 
 Diagnostic efficiency 95.08 (71.14) 33.44 46.87 69.52 122.84 178.54 11.45 335.75 88.82 
LOS          
 Rate (% of admissions with high illness acuity) 3.18 (3.10) 2.99 4.69 7.25 18.75 3.67 
 Diagnostic efficiency 19.60 (24.02) 3.88 7.09 12.18 22.36 36.96 2.02 164.01 10.35 
FIGURE 2

A, Newborns receiving antibiotics per EOS case (diagnostic efficiency for EOS). B, Newborns receiving antibiotics per LOS case (diagnostic efficiency for LOS). C, EOS rate. D, LOS rate. The shaded bar and vertical lines identify interquartile range, median (white dot), 1.5 × interquartile range; symmetric curves to either side estimate the relative number of units at each vertical axis value. Numbers below each plot describe the number of hospitals and live births, respectively.

FIGURE 2

A, Newborns receiving antibiotics per EOS case (diagnostic efficiency for EOS). B, Newborns receiving antibiotics per LOS case (diagnostic efficiency for LOS). C, EOS rate. D, LOS rate. The shaded bar and vertical lines identify interquartile range, median (white dot), 1.5 × interquartile range; symmetric curves to either side estimate the relative number of units at each vertical axis value. Numbers below each plot describe the number of hospitals and live births, respectively.

Close modal

Table 2 provides details of correlation analyses. Variation in EOS rate or LOS rate did not explain variation in the percentage of newborns receiving antibiotics (Fig 3). Neither percentage of newborns with antibiotic exposure nor sepsis diagnostic efficiency correlated with AUR. The percentage of newborns with antibiotic exposure correlated strongly with inborn admission rate (ρ = 0.47; P = .0002) but not with EOS diagnostic efficiency. Diagnostic efficiency for EOS correlated positively with diagnostic efficiency for LOS (ρ = 0.38; P = .05). Surgical case volume was uncorrelated with LOS rate or LOS diagnostic efficiency but strongly correlated with NICU mortality rate (ρ = 0.53; P < .0001). Neither NEC rate nor NICU mortality rate correlated with percentage of newborns with antibiotic exposure or sepsis diagnostic efficiency.

TABLE 2

Correlation Analyses

ρP
Percentage of births exposed to antibiotics in correlation with the AUR 0.21 0.91 
Percentage of births exposed to antibiotics in correlation with the EOS rate 0.14 .99 
Percentage of births exposed to antibiotics in correlation with the LOS rate 0.0008 .99 
Percentage of births exposed to antibiotics in correlation with the NEC rate 0.03 .78 
Percentage of births exposed to antibiotics in correlation with the inborn admission rate 0.47 .0002 
Percentage of births exposed to antibiotics in correlation with the NICU mortality rate −0.007 .99 
Diagnostic efficiency   
 (EOS + LOS)in correlation with the AUR 0.09 .36 
 EOS in correlation with the AUR 0.25 .99 
 LOS in correlation with the AUR 0.03 .99 
 EOS in correlation with the inborn admission rate 0.13 .99 
 EOS in correlation with the NICU mortality rate 0.03 .99 
 LOS in correlation with the NICU mortality rate −0.25 .80 
 EOS in correlation with the LOS rate 0.005 .99 
 EOS in correlation with the NEC rate −0.16 .15 
 LOS in correlation with the NEC rate −0.07 .54 
 EOS in correlation with the diagnostic efficiency, LOS 0.38 .05 
No. live births in correlation with the EOS rate 0.09 .32 
No. live births in correlation with the EOS rate (where EOS >0) −0.22 .05 
Surgical case volume in correlation with the diagnostic efficiency, LOS −0.24 .99 
Surgical case volume in correlation with the LOS rate 0.09 .99 
Surgical case volume in correlaton with the NICU mortality rate 0.53 <.0001 
ρP
Percentage of births exposed to antibiotics in correlation with the AUR 0.21 0.91 
Percentage of births exposed to antibiotics in correlation with the EOS rate 0.14 .99 
Percentage of births exposed to antibiotics in correlation with the LOS rate 0.0008 .99 
Percentage of births exposed to antibiotics in correlation with the NEC rate 0.03 .78 
Percentage of births exposed to antibiotics in correlation with the inborn admission rate 0.47 .0002 
Percentage of births exposed to antibiotics in correlation with the NICU mortality rate −0.007 .99 
Diagnostic efficiency   
 (EOS + LOS)in correlation with the AUR 0.09 .36 
 EOS in correlation with the AUR 0.25 .99 
 LOS in correlation with the AUR 0.03 .99 
 EOS in correlation with the inborn admission rate 0.13 .99 
 EOS in correlation with the NICU mortality rate 0.03 .99 
 LOS in correlation with the NICU mortality rate −0.25 .80 
 EOS in correlation with the LOS rate 0.005 .99 
 EOS in correlation with the NEC rate −0.16 .15 
 LOS in correlation with the NEC rate −0.07 .54 
 EOS in correlation with the diagnostic efficiency, LOS 0.38 .05 
No. live births in correlation with the EOS rate 0.09 .32 
No. live births in correlation with the EOS rate (where EOS >0) −0.22 .05 
Surgical case volume in correlation with the diagnostic efficiency, LOS −0.24 .99 
Surgical case volume in correlation with the LOS rate 0.09 .99 
Surgical case volume in correlaton with the NICU mortality rate 0.53 <.0001 

The analyses entail the percentage of live births receiving a newborn antibiotic exposure; diagnostic efficiency, EOS + LOS; diagnostic efficiency, EOS; diagnostic efficiency, LOS; AUR; proven bloodstream infection rates; NEC; inborn admission rate; surgical case volume; and NICU mortality rate.

FIGURE 3

Relationship between hospital-proven bloodstream infection rate and percentage of births with antibiotic exposures. A, EOS. B, LOS.

FIGURE 3

Relationship between hospital-proven bloodstream infection rate and percentage of births with antibiotic exposures. A, EOS. B, LOS.

Close modal

Except as described, stratification of analytical questions by NICU level of care did not alter the main inferences.

Antibiotic prescribing practices for neonates are remarkably heterogeneous. The plausible correlations we explored provide no rational explanation for the observed variation in percentage of newborns with antibiotic exposure.

Measuring the percentage of NICU patient-days entailing specified antibiotic exposures (AUR) identified 1 aspect of antibiotic prescribing practice variation and informed stewardship efforts associated with substantial recent AUR declines.1,4  However, as a solitary measure, it is context free, incompletely informing benchmarking and improvement efforts.4,19  Because AUR correlates with neither the percentage of newborns receiving antibiotic exposures nor sepsis diagnostic efficiency, hospitals with similar AUR values may differ substantially in treatment starts, and plausibly, durations. Thus, stewardship efforts require understanding in addition to AUR, numbers of patients receiving antibiotics, bloodstream infection rates, and bloodstream infection diagnostic efficiency.

Hospitals varied 27-fold in the percentage of newborns receiving antibiotics; at some hospitals, this proportion exceeded 33%. The highest outlier values occurred at hospitals reporting 0 EOS (Fig 1) where diagnostic efficiency is incalculable.

Diagnostic efficiency for EOS varied 29-fold; at 10% of hospitals, at least 200 and sometimes ≥300 newborns receive antibiotics per EOS case. A recent guideline reappraisal for suspected EOS management estimated such values between 18 and 38 for newborns with nonspecific clinical signs consistent with sepsis, assessing this range to indicate inefficient case identification.5  Even when considering all births rather than a more focused symptomatic subgroup, 90% of hospitals in the current study exceed that range. The same authors estimated EOS diagnostic efficiency values between 80 and 210 for newborns at ≥35 weeks’ gestation whose mothers had clinical chorioamnionitis.5  Although the range shown in Table 1 resembles the latter range, it describes all births rather than this more focused subgroup.

At most hospitals, LOS diagnostic efficiency appears better than for EOS. The hospital mean of 20 newborns receiving antibiotics per proven LOS was one-fifth the mean value for EOS, whereas the hospital mean LOS rate was 44-fold greater among the at-risk population (Table 1). Infants with suspected LOS typically have clinical signs, which may explain the similarity of these values to those mentioned above for symptomatic EOS. Interquartile range of LOS diagnostic efficiency was relatively narrow; 81-fold overall variation reflects a few extreme outliers among community NICUs (Fig 2), suggesting overtreatment.

Infants of African American race, gestational age <34 weeks, and very low birth weight have substantially higher EOS risk.20  Compared with the rest of a cohort of >5000 neonates with a birth weight <1500 g, those born by cesarean delivery to mothers with preeclampsia and without preterm labor or chorioamnionitis had a 12-fold lower incidence of EOS.21  Explicitly accounting for these factors could illuminate hospital-level case-mix differences between and within levels of care (Figs 1 and 2) and inform local stewardship efforts. However, lack of correlation between EOS rate and percentage of births exposed to antibiotics (Fig 3) would remain because the computed proportional contribution of each of those factors to each hospital’s data point is independent of the question whether the percentage of newborns exposed to antibiotics correlates with EOS rate.

Providers may remain skeptical about negative blood culture results guiding decisions for infants with nonspecific clinical signs.22  However, proper blood culture collection methods and current detection thresholds are highly reliable: “if the bacteria cannot grow in the blood culture bottle (an ideal medium at an ideal temperature, free of antibiotics, complement, or phagocytes), then why would they grow effectively in the infant’s bloodstream?”22  Because 15% of EOS isolates from neonates of birth weight <1500 g are strict anaerobic species, the use of separate culture bottles for aerobes and anaerobes has been recommended.21 

Evidence continues to accumulate that antibiotic exposure early in life (ie, when the newborn’s evolving microbiota and immune system are learning to live with 1 another) can be problematic.23  Early antibiotic exposure is associated with later adverse outcomes including asthma,24  obesity, atopy, and inflammatory bowel disease23 ; and for preterm infants, these outcomes include bronchopulmonary dysplasia, NEC, LOS, and death.2528  Additionally, antibiotic use is associated with selection of multidrug-resistant pathogens, themselves associated with increased morbidity, mortality, cost, and length of stay.29  Other pertinent factors, not yet quantified, include newborn and/or parental separation distress, delayed maternal bonding and establishment of breastfeeding, intravenous line insertion(s), lumbar puncture(s), and risk of medical error accompanying the NICU stay.30 

Why might inborn admission rate correlate strongly with the percentage of newborns exposed to antibiotics (ρ = 0.47; P = .0002) but not with EOS diagnostic efficiency? The latter correlation entails 31 fewer hospitals, only those with ≥1 EOS case. The former correlation thus appears driven by the hospitals that are mathematically the most inefficient EOS identifiers (ie, those with incalculable EOS diagnostic efficiency because of 0 EOS). Additionally, mean live births at hospitals reporting 0 EOS was significantly lower than at hospitals reporting at least 1 (mean 1883.2 and SD 880.1; mean 3158.4 and SD 1525.5; P < .0001), an unsurprising finding considering actual EOS rates. Among all sampled hospitals, the number of live births and EOS rate were uncorrelated but negatively correlated among hospitals where EOS was >0 (ρ = −0.22; P = .05). Thus, antibiotic treatment thresholds appear relatively less well calibrated at hospitals with relatively lower birth volumes and case experience.

Although diagnostic efficiency for EOS and LOS correlated positively (ρ = 0.38, P = .05), the magnitude of the relationship reveals within-hospital differences in treatment thresholds for suspected EOS and LOS.

NICU mortality rate did not correlate with the percentage of newborns exposed to antibiotics or with diagnostic efficiency for either EOS or LOS. Although NICU mortality rate correlated with surgical case volume (ρ = 0.53; P < .0001), surgical case volume did not correlate with LOS rate or LOS diagnostic efficiency. The bulk of surgical case volume occurs at regional-level NICUs, which serve populations with relatively high illness acuity and provide educational and consultative support to lower-level NICUs.11  The notably low and tight LOS diagnostic efficiency distribution among regional NICUs (Fig 2) may reflect active antibiotic stewardship efforts and signals opportunity for regional-level NICUs to guide improvement at high-outlier community NICUs referring to them.

Although counts of newborns with antibiotic exposures derive from all inborn admissions, LOS cases may include outborn admissions, to that extent, inflating counts of newborns treated for LOS and for all sepsis combined. An inborn admission subsequently transferred out could also inflate the proportinal contribution of outborn LOS cases. Antibiotic prescribing practices at unsampled hospitals may differ from our 69% sample.31  Less performance feedback and regulatory oversight likely occur at those unsampled hospitals; EOS diagnostic efficiency estimates especially may underestimate overall numbers treated per case in California. Audit confirming proper blood culture collection methods was infeasible. However, our data are consistent with other reported incidence rates.5,20,32  Our diagnostic efficiency computations (Supplemental Information) derive from reported antibiotic use surveillance33 ; our data did not discriminate antibiotic exposures for EOS from LOS. Such allocations likely varied across hospitals. Our computations assumed that each newborn exposed reflected either suspected EOS or LOS, counting an infant treated for both only once. NICUs did not report conditions for which antibiotic treatment is justified despite negative blood culture results (eg, meningitis, omphalitis, osteomyelitis, tracheitis, pneumonia, urinary tract infection). When present, such uncounted infections would result in overestimated counts of newborns treated per proven infection. Similar consideration applies in which fluconazole prophylaxis constituted an infant’s sole antibiotic exposure. Birth weight and gestational age–specific counts of newborns with antibiotic exposure, presently unavailable, could reveal whether EOS or LOS diagnostic efficiency varies by such subgroups.

In a 69% sample of California births in 2017, 27 455 of 326 845 newborns (8.4%) received antibiotics before being discharged from the hospital. The 27-fold variation among hospitals in the percentage of newborns exposed to antibiotics was unexplained by either bloodstream infection rate or NICU AUR. EOS case identification often is inefficient. LOS case identification often appears comparatively better calibrated. Even so, within the same hospital, diagnostic efficiency may differ. These measures complement newborn antibiotic use evaluations anchored in patient-days of exposure.

Further performance insight can emerge from answers to the following questions. What proportion of our newborns exposed to antibiotics entail infants of African American race; gestational age <34 weeks; birth weight <1500 g; and among the latter, born by cesarean delivery in which preeclampsia was present but not preterm labor or chorioamnionitis; exposure for suspected EOS and LOS; conditions for which antibiotic treatment is justified despite negative blood culture results; and fluconazole prophylaxis constituting sole antibiotic exposure? Does our estimate of service population risk mirror observed EOS and LOS rates and account for birth volume?

“When different physicians are recommending different things for essentially the same patients, it is impossible to claim that they are all doing the right thing.”34  Applying this conditional observation regarding practice variation to Fig 3 highlights the imperative to learn which, if any, of the data points arrayed above each horizontal axis EOS or LOS rate, most strikingly, when the rate is 0, denotes providers who are “doing the right thing.”

The findings and conclusions in this article are those of the authors and do not necessarily represent the views or opinions of the California Department of Health Care Services or the California Health and Human Services Agency.

Dr Schulman conceptualized and designed the study, performed the analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Benitz participated in the study design and interpretation of analyses and critically reviewed the manuscript; Drs Profit, Lee, and Gould supervised data collection and critically reviewed the manuscript; Ms Duenas and Dr Bennett coordinated and supervised data collection, aggregated data, provided the aggregated data sets to California Children’s Services, and critically reviewed the manuscript; Dr Jocson participated in the study design and interpretation of analyses and critically reviewed the manuscript; Dr Schutzengel participated in the interpretation of analyses and critically reviewed the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

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

FUNDING: No external funding.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2019-2576.

     
  • AUR

    antibiotic userate

  •  
  • EOS

    early-onset sepsis

  •  
  • LOS

    late-onset sepsis

  •  
  • NEC

    necrotizing enterocolitisrate

  •  
  • ρ

    Pearson correlation coefficient

1
Schulman
J
,
Dimand
RJ
,
Lee
HC
, et al
.
Neonatal intensive care unit antibiotic use
.
Pediatrics
.
2015
;
135
(
5
):
826
833
2
US Department of Health and Human Services
.
Surveillance for antimicrobial use and antimicrobial resistance options
.
2018
.
3
Vermont Oxford Network
.
Antibiotic stewardship in newborn care.
Available at: https://public.vtoxford.org/quality-education/universal-training/. Accessed September 20, 2019
4
Schulman
J
,
Profit
J
,
Lee
HC
, et al
.
Variations in neonatal antibiotic use
.
Pediatrics
.
2018
;
142
(
3
):
e20180115
5
Benitz
WE
,
Wynn
JL
,
Polin
RA
.
Reappraisal of guidelines for management of neonates with suspected early-onset sepsis
.
J Pediatr
.
2015
;
166
(
4
):
1070
1074
6
Kuzniewicz
MW
,
Puopolo
KM
,
Fischer
A
, et al
.
A quantitative, risk-based approach to the management of neonatal early-onset sepsis
.
JAMA Pediatr
.
2017
;
171
(
4
):
365
371
7
California Perinatal Quality Care Collaborative
.
Antibiotic stewardship collaborative
.
2017
.
8
Committee on Obstetric Practice
.
Committee opinion no. 712: intrapartum management of intraamniotic infection
.
Obstet Gynecol
.
2017
;
130
(
2
):
e95
e101
9
Hooven
TA
,
Polin
RA
.
Time to overhaul the “rule out sepsis” workup
.
Pediatrics
.
2017
;
140
(
1
):
e20171155
10
Higgins
RD
,
Saade
G
,
Polin
RA
, et al;
Chorioamnionitis Workshop Participants
.
Evaluation and management of women and newborns with a maternal diagnosis of chorioamnionitis: summary of a workshop
.
Obstet Gynecol
.
2016
;
127
(
3
):
426
436
11
California Department of Health Care Services
.
Provider standards
.
2019
.
12
American Academy of Pediatrics Committee on Fetus And Newborn
.
Levels of neonatal care
.
Pediatrics
.
2012
;
130
(
3
):
587
597
13
California Perinatal Quality Care Collaborative (CPQCC)
.
Who we are.
2014
. Available at: https://www.cpqcc.org/about/who-we-are. Accessed September 20, 2019
14
California Perinatal Quality Care Collaborative (CPQCC)
.
NICU data resources.
Available at: https://www.cpqcc.org/nicu/nicu-data-resources. Accessed September 20, 2019
15
Vermont Oxford Network
.
Databases and reporting.
Available at: https://public.vtoxford.org/data-and-reports/. Accessed September 20, 2019
16
Bhatt
DR
,
Adams
M
,
Clearly
J
, et al
NICUs & Neonatologists in California.
California Association of Neonatologists, District IX Section on Neonatal-Perinatal Medicine (SONPM). Available at: https://canneo.groupsite.com/main/summary. Accessed September 21, 2019
17
California Perinatal Quality Care Collaborative
.
CPQCC manual of definitions for infants born in 2017
.
2016
.
18
StataCorp
.
Stata Statistical Software [computer program]. Version 15.
College Station, TX
:
Stata Press
;
2017
19
Flannery
DD
,
Puopolo
KM
.
Neonatal antibiotic use: how much is too much?
Pediatrics
.
2018
;
142
(
3
):
e20181942
20
Schrag
SJ
,
Farley
MM
,
Petit
S
, et al
.
Epidemiology of invasive early-onset neonatal sepsis, 2005 to 2014
.
Pediatrics
.
2016
;
138
(
6
):
e20162013
21
Mukhopadhyay
S
,
Puopolo
KM
.
Clinical and microbiologic characteristics of early-onset sepsis among very low birth weight infants: opportunities for antibiotic stewardship
.
Pediatr Infect Dis J
.
2017
;
36
(
5
):
477
481
22
Cantey
JB
,
Baird
SD
.
Ending the culture of culture-negative sepsis in the neonatal ICU
.
Pediatrics
.
2017
;
140
(
4
):
e20170044
23
Zeissig
S
,
Blumberg
RS
.
Life at the beginning: perturbation of the microbiota by antibiotics in early life and its role in health and disease
.
Nat Immunol
.
2014
;
15
(
4
):
307
310
24
Alm
B
,
Erdes
L
,
Möllborg
P
, et al
.
Neonatal antibiotic treatment is a risk factor for early wheezing
.
Pediatrics
.
2008
;
121
(
4
):
697
702
25
Cotten
CM
,
Taylor
S
,
Stoll
B
, et al;
NICHD Neonatal Research Network
.
Prolonged duration of initial empirical antibiotic treatment is associated with increased rates of necrotizing enterocolitis and death for extremely low birth weight infants
.
Pediatrics
.
2009
;
123
(
1
):
58
66
26
Alexander
VN
,
Northrup
V
,
Bizzarro
MJ
.
Antibiotic exposure in the newborn intensive care unit and the risk of necrotizing enterocolitis
.
J Pediatr
.
2011
;
159
(
3
):
392
397
27
Kuppala
VS
,
Meinzen-Derr
J
,
Morrow
AL
,
Schibler
KR
.
Prolonged initial empirical antibiotic treatment is associated with adverse outcomes in premature infants
.
J Pediatr
.
2011
;
159
(
5
):
720
725
28
Ting
JY
,
Synnes
A
,
Roberts
A
, et al;
Canadian Neonatal Network Investigators
.
Association between antibiotic use and neonatal mortality and morbidities in very low-birth-weight infants without culture-proven sepsis or necrotizing enterocolitis
.
JAMA Pediatr
.
2016
;
170
(
12
):
1181
1187
29
Patel
SJ
,
Saiman
L
.
Antibiotic resistance in neonatal intensive care unit pathogens: mechanisms, clinical impact, and prevention including antibiotic stewardship
.
Clin Perinatol
.
2010
;
37
(
3
):
547
563
30
Hooven
TA
,
Randis
TM
,
Polin
RA
.
What’s the harm? Risks and benefits of evolving rule-out sepsis practices
.
J Perinatol
.
2018
;
38
(
6
):
614
622
31
San Francisco
Chronicle
. Rubenstein S. California birthrate plunges to lowest level in a century.
2018
. Available at: https://www.sfchronicle.com/bayarea/article/California-birthrate-plunges-to-lowest-level-in-a-12923521.php. Accessed September 20, 2019
32
Dong
Y
,
Speer
CP
.
Late-onset neonatal sepsis: recent developments
.
Arch Dis Child Fetal Neonatal Ed
.
2015
;
100
(
3
):
F257
F263
33
Cantey
JB
,
Wozniak
PS
,
Sánchez
PJ
.
Prospective surveillance of antibiotic use in the neonatal intensive care unit: results from the SCOUT study
.
Pediatr Infect Dis J
.
2015
;
34
(
3
):
267
272
34
Eddy
DM
.
Evidence-based medicine: a unified approach
.
Health Aff (Millwood)
.
2005
;
24
(
1
):
9
17

Competing Interests

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

Supplementary data