We aimed to study whether national and local antibiotic stewardship projects have reduced the antibiotic use in newborns and to monitor potential changes in adverse outcomes.
In a nationwide, population-based study from Norway, we included all hospital live births from 34 weeks' gestation (n = 282 046) during 2015 to 2019. The primary outcome was the proportion of newborns treated with antibiotics from 0 to 28 days after birth. The secondary outcomes were the overall duration of antibiotic treatment and by categories: culture-positive sepsis, clinical sepsis, and no sepsis.
A total of 7365 (2.6%) newborns received intravenous antibiotics during the period, with a reduction from 3.1% in 2015 to 2.2% in 2019 (30% decrease; P < .001). Hospitals with antibiotic stewardship projects experienced the largest reduction (48% vs 23%; P < .001). We found a small decrease in the median duration of antibiotic treatment in newborns without sepsis from 2.93 to 2.66 days (P = .011), and geographical variation was reduced during the study period. The overall number of days with antibiotic treatments was reduced by 37% from 2015 to 2019 (119.1 of 1000 vs 75.6 of 1000; P < .001). Sepsis was confirmed by blood culture in 206 newborns (incidence rate: 0.73 cases per 1000 live births). We found no increase in sepsis with treatment onset >72 hours of life, and sepsis-attributable deaths remained at a low level.
During the study period, a substantial decrease in the proportion of newborns treated with antibiotics was observed together with a decline in treatment duration for newborns without culture-positive sepsis.
Antibiotic treatment in newborns may be associated with long-term adverse outcomes and an increase in antibiotic resistance, but adequate and timely antibiotic treatment is needed to prevent sepsis-related morbidity and mortality.
Our findings suggest that in a high-resource setting with low newborn mortality, a stricter policy than previously practiced regarding antibiotics is safe without increased risk for adverse outcomes. Hospitals with antibiotic stewardship projects saw the largest reduction in antibiotic use.
Newborns are at risk for sepsis, and suspected sepsis is a major reason for admission to the NICU.1 Clinical signs of sepsis in the newborn are nonspecific and biomarkers are insufficient to make a certain diagnosis.2 Intravenous antibiotics early in life affect the developing microbiota3 and may be associated with long-term adverse outcomes,4–6 increasing antimicrobial resistance,3,7 and impaired growth up to school age.8
In recent years, there has been a push internationally to decrease the use of antibiotics.9 The Norwegian Ministry of Health published a national strategy in 2015 against antibiotic overuse, aiming to reduce the use by 30% from 2012 to the end of 2020.10
There is a large variation between countries in the use of antibiotics in neonatal care.11–13 During 2009–2014 there was also a geographical variation within Norway, with a twofold difference between hospitals in the proportion of newborns treated with antibiotics.14 This is an unwanted variation not explained by differences in incidence of sepsis.
In this study, we describe changes in the use of intravenous antibiotics in all term and near-term newborns in Norway between 2015 and 2019 and geographical variations. Moreover, we present epidemiological data on culture-positive and clinical sepsis. Our hypothesis was that during this 5-year period national and local initiatives had led to a decrease in antibiotic consumption without increased incidence of readmissions, sepsis-attributable mortality, or delayed diagnosis of sepsis in the newborn period.
Methods
Study Population, Data Sources, and Setting
In an observational study we included all liveborn newborns with a gestational age (GA) ≥34 weeks born in Norway during the 5-year period between January 1, 2015, to December 31, 2019 (Fig 1). The Regional Ethical Committee for Medical and Health Research Ethics approved the study.
Data on the total number of live births were obtained from the Medical Birth Registry of Norway.15 Prospectively collected clinical data were obtained from the Norwegian Neonatal Network (NNN), a Web-based nationwide registry governed by the Norwegian Institute of Public Health.16 In the NNN, data on investigations, treatments, and diagnoses are entered on a daily basis by the attending physician on all infants admitted to each participating neonatal unit. There are 21 neonatal units across Norway, all of which included all admissions during the study period. Virtually all neonates receiving intravenous antibiotics are admitted to one of these neonatal units. Lastly, we analyzed data from the Norwegian Patient Registry (NPR) on diagnosis of sepsis in all newborns up to 4 weeks of age during the study period. The NPR provides data on patients treated at all hospitals, including diagnoses labeled with International Classification of Diseases, 10th Revision, (ICD-10) codes.17 Reimbursement for hospital stays is linked to the NPR, providing high completeness of discharge data. At some of the hospitals in Norway, newborns may be admitted to regular pediatric wards for complications arising after discharge from the maternity ward. To capture all newborns with sepsis after discharge from the maternity ward, the NPR was used as a supplementary data source to identify all cases admitted to any department >72 hours of life with a discharge code of sepsis.
Norway has 5.4 million inhabitants and consists of 11 counties, which are grouped into 4 regional health care trusts (South-East, West, Central, and North). The neonatal death rate in 2019 was 1.3 per 1000 live births,18 and health services during pregnancy and childhood are universal and free of charge. Most of the deliveries are conducted at hospitals with specialized obstetric departments. Because of large geographical distances there are also some smaller delivery units for low-risk deliveries with a limited number of annual births. Newborns from such facilities who need antibiotic treatment are transferred to the nearest neonatal unit. Less than 0.3% of the deliveries are conducted as planned home deliveries. We excluded home deliveries because of missing information in regard to which region the children were born.
During the study period, the following antibiotic stewardship strategies were implemented:
Standardized criteria for neonatal sepsis diagnosis published in October 2015.19
A reminder message given to the physician during daily registration in the NNN suggesting to reconsider antibiotics after 3 days therapy, introduced in 2015.
The national Choosing Wisely campaign launched in 2018 included a recommendation of early discontinuation of antibiotics.20
In 2017, neonatal units in 3 large hospitals performed a quality improvement (QI) project, including an automatic 48-hour stop order on antibiotic use and implementing procalcitonin as a biomarker to reduce antibiotic treatment duration.21
During 2017–2018, another large hospital implemented serial physical examinations for suspected sepsis as a QI project.22
Study Definitions
Diagnoses registered in NNN were defined according to the ICD-10 codes. Bacterial sepsis in the newborn (P36.0 to P36.8) is defined as growth of bacteria in blood cultures together with clinical signs and symptoms compatible with infection. Growth of coagulase negative staphylococci (CoNS) in blood culture in this age group was considered a contamination.1 Unspecified bacterial sepsis (P36.9) or “clinical sepsis” is applied when there are clinical and biochemical signs of sepsis, without growth of bacteria in blood cultures or when blood cultures were not obtained. In 2006 (reviewed 2014–2015), neonatologists within the Norwegian Pediatric Association suggested the following 4 criteria for the use of P36.9: (1) clinical signs of infection, (2) maximum C-reactive protein level (CRP) of >30 mg/L, (3) minimum duration of 5 days antibiotic treatment, and (4) other explanations for the clinical picture excluded.19 The CRP cutoff value was chosen to find a balance between sensitivity and specificity for sepsis, and observational studies show that most healthy newborn have CRP values well <30 mg/L.23,24
We classified children treated with antibiotics in 3 groups: culture-positive sepsis (P36.0 to P36.8), clinical sepsis (P36.9), and no sepsis (antibiotics but no P36.0 to P36.9 code at discharge).
Main Outcome
The main outcome of the study was any exposure to systemic intravenous antibiotics during first 28 days of life. The secondary outcome was the duration of antibiotics in days, counted as calendar days from the first to the last dose. The secondary outcome was analyzed for all newborns who were commenced on antibiotics, and thereafter separately for those with culture-positive sepsis, clinical sepsis, and no sepsis.
Additional Variables
We included GA, type of blood culture pathogen, and diagnosis in the analysis. Our data do not include information on maternal risk factors for neonatal sepsis such as fever during delivery or clinical signs of chorioamnionitis.
Statistical Analysis
Data were analyzed by using SPSS version 25 statistical software (IBM, SPSS Statistics, IBM Corporation, Armonk, NY) and Stata version 16.0 (Stata Corp, College Station, TX). Results are expressed as a percentage with 95% confidence interval or as means with SD, as appropriate. To test the significance of our findings, we used χ2 test for categorical analyses, analysis of variance with logarithmic transformation for continuous data, and a P value of <.05 as the significance level.
Results
Between 2015 and 2019, a total of 288 623 children were born in Norway. After excluding children with missing data on GA, missing health region of birth, and a GA <34 weeks, we were left with a cohort of 282 046 children (Fig 1), of whom 7365 (2.6%) were treated with antibiotics. Near-term infants (GA: 34.0–36.6 weeks) contributed with 12 917 (4.5%) of the cohort, and proportions did not vary by birth year or region (data not shown).
During the study period, we found a 30% reduction in the proportion of children started on antibiotics at a national level (from 3.1% in 2015 to 2.2% in 2019 during days 0 to 28 after birth; P < .001). Data on antibiotic exposure only during days 0 to 7 after birth are presented in Table 1.
. | Year and Live Births With GA ≥34 Wk . | |||||
---|---|---|---|---|---|---|
2015 (n = 58 644) . | 2016 (n = 58 744) . | 2017 (n = 56 297) . | 2018 (n = 54 916) . | 2019 (n = 54 167) . | P . | |
Antibiotics started on day 0–3 of life, n (%) | 1693 (2.9) | 1479 (2.5) | 1282 (2.3) | 1129 (2.1) | 1104 (2.0) | <.001 |
Antibiotics started on day 4–7 of life, n (%) | 55 (0.10) | 63 (0.11) | 58 (0.11) | 54 (0.10) | 39 (0.07) | .35 |
Region, n (%) | ||||||
South-East | 1052 (3.3) | 920 (2.8) | 798 (2.6) | 724 (2.4) | 714 (2.3) | <.001 |
West | 379 (2.8) | 302 (2.3) | 252 (2.0) | 189 (1.6) | 170 (1.4) | <.001 |
Central | 185 (2.3) | 186 (2.3) | 180 (2.4) | 175 (2.4) | 170 (2.3) | .99 |
North | 132 (2.9) | 134 (3.0) | 111 (2.6) | 95 (2.2) | 89 (2.2) | .06 |
By GA, n (%) | ||||||
GA ≥ 34 wk | 1748 (3.0) | 1542 (2.6) | 1341 (2.4) | 1183 (2.2) | 1143 (2.1) | <.001 |
GA ≥ 37 wk | 1485 (2.7) | 1321 (2.4) | 1152 (2.1) | 1029 (2.0) | 1008 (1.9) | <.001 |
GA 34–36 wk | 263 (9.7) | 221 (8.1) | 189 (7.6) | 154 (6.4) | 135 (5.5) | <.001 |
Local QI project,a n (%) | ||||||
Yes | 413 (2.9) | 380 (2.7) | 262 (1.9) | 238 (1.8) | 200 (1.5) | <.001 |
No | 1335 (3.0) | 1162 (2.6) | 1079 (2.5) | 945 (2.3) | 943 (2.3) | <.001 |
. | Year and Live Births With GA ≥34 Wk . | |||||
---|---|---|---|---|---|---|
2015 (n = 58 644) . | 2016 (n = 58 744) . | 2017 (n = 56 297) . | 2018 (n = 54 916) . | 2019 (n = 54 167) . | P . | |
Antibiotics started on day 0–3 of life, n (%) | 1693 (2.9) | 1479 (2.5) | 1282 (2.3) | 1129 (2.1) | 1104 (2.0) | <.001 |
Antibiotics started on day 4–7 of life, n (%) | 55 (0.10) | 63 (0.11) | 58 (0.11) | 54 (0.10) | 39 (0.07) | .35 |
Region, n (%) | ||||||
South-East | 1052 (3.3) | 920 (2.8) | 798 (2.6) | 724 (2.4) | 714 (2.3) | <.001 |
West | 379 (2.8) | 302 (2.3) | 252 (2.0) | 189 (1.6) | 170 (1.4) | <.001 |
Central | 185 (2.3) | 186 (2.3) | 180 (2.4) | 175 (2.4) | 170 (2.3) | .99 |
North | 132 (2.9) | 134 (3.0) | 111 (2.6) | 95 (2.2) | 89 (2.2) | .06 |
By GA, n (%) | ||||||
GA ≥ 34 wk | 1748 (3.0) | 1542 (2.6) | 1341 (2.4) | 1183 (2.2) | 1143 (2.1) | <.001 |
GA ≥ 37 wk | 1485 (2.7) | 1321 (2.4) | 1152 (2.1) | 1029 (2.0) | 1008 (1.9) | <.001 |
GA 34–36 wk | 263 (9.7) | 221 (8.1) | 189 (7.6) | 154 (6.4) | 135 (5.5) | <.001 |
Local QI project,a n (%) | ||||||
Yes | 413 (2.9) | 380 (2.7) | 262 (1.9) | 238 (1.8) | 200 (1.5) | <.001 |
No | 1335 (3.0) | 1162 (2.6) | 1079 (2.5) | 945 (2.3) | 943 (2.3) | <.001 |
Four large neonatal units conducted antibiotic stewardship QI projects during the study period as explained in the Methods section. The n (%) of newborns started on antibiotics in these units is compared with hospitals without such local initiatives.
The difference between the region with the highest versus lowest proportion of infants commenced on antibiotics did not change appreciably over time (1.0% in 2015 vs 0.9% in 2019; Table 1). Four hospitals covering approximately one-fourth of all deliveries in Norway had local antibiotic stewardship QI-projects during the study period.21,22 Collectively, there was a 48% reduction in children started on antibiotics in these 4 hospitals, compared with a 23% reduction in all the other hospitals who did not have such projects, with a similar baseline (Table 1; P < .001).
Antibiotic treatments were mainly started during the first week of life (6706 of 7365; 91%), and antibiotic initiation during this first week decreased from 2.9% to 2.1% (Table 1).
We found an overall reduction in duration of antibiotic treatment, from a mean of 3.9 days in 2015 to 3.4 days in 2019 (Table 2; P < .001). This decrease was observed in 3 of 4 health regions. In the West region, the duration was unchanged but had been markedly lower compared with that of other health regions at the start of our study in 2015. There was no significant change in treatment duration for culture-positive sepsis but a significant decrease for clinical sepsis and for those with no sepsis diagnosis (Table 2). Differences in duration of treatment between regions were reduced over time (Table 2). The annual number of days with antibiotics decreased by 37%, from 119.1 per 1000 newborns in 2015 to 75.6 per 1000 newborns in 2019.
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Pa,b . |
---|---|---|---|---|---|---|
Overall, mean (SD), d | 3.9 (3.0) | 3.9 (3.1) | 3.5 (2.6) | 3.6 (2.9) | 3.4 (2.6) | <.001 |
Duration by region, mean (SD), d | ||||||
South-East | 4.0 (2.9) | 4.0 (3.1) | 3.6 (2.6) | 3.8 (3.0) | 3.5 (2.3) | <.001 |
West | 3.1 (2.4) | 3.5 (3.3) | 3.0 (2.3) | 3.2 (2.8) | 3.4 (3.5) | .54 |
Central | 4.3 (3.6) | 4.0 (3.2) | 3.8 (3.1) | 3.6 (2.8) | 3.4 (2.6) | .09 |
North | 4.3 (4.0) | 3.3 (2.4) | 3.3 (2.8) | 3.1 (1.9) | 3.2 (2.2) | .006 |
Duration by diagnosis, mean (SD), d | ||||||
Culture-positive sepsis | 8.9 (5.9) | 8.8 (5.5) | 8.9 (5.2) | 9.5 (5.8) | 9.1 (5.0) | .98 |
Clinical sepsis | 5.3 (2.9) | 5.4 (2.9) | 5.0 (2.6) | 5.0 (2.5) | 5.0 (2.3) | .01 |
No sepsis | 2.9 (2.3) | 2.8 (2.3) | 2.6 (1.9) | 2.8 (2.3) | 2.7 (1.9) | .02 |
Numbers with diagnosis, n (%) | ||||||
Culture-positive sepsis | 47 (0.08) | 52 (0.09) | 34 (0.06) | 38 (0.07) | 34 (0.06) | .34 |
Clinical sepsis | 609 (1.0) | 523 (0.9) | 413 (0.7) | 364 (0.7) | 325 (0.6) | <.001 |
No sepsis | 1159 (1.9) | 1044 (1.7) | 957 (1.7) | 885 (1.6) | 859 (1.6) | <.001 |
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Pa,b . |
---|---|---|---|---|---|---|
Overall, mean (SD), d | 3.9 (3.0) | 3.9 (3.1) | 3.5 (2.6) | 3.6 (2.9) | 3.4 (2.6) | <.001 |
Duration by region, mean (SD), d | ||||||
South-East | 4.0 (2.9) | 4.0 (3.1) | 3.6 (2.6) | 3.8 (3.0) | 3.5 (2.3) | <.001 |
West | 3.1 (2.4) | 3.5 (3.3) | 3.0 (2.3) | 3.2 (2.8) | 3.4 (3.5) | .54 |
Central | 4.3 (3.6) | 4.0 (3.2) | 3.8 (3.1) | 3.6 (2.8) | 3.4 (2.6) | .09 |
North | 4.3 (4.0) | 3.3 (2.4) | 3.3 (2.8) | 3.1 (1.9) | 3.2 (2.2) | .006 |
Duration by diagnosis, mean (SD), d | ||||||
Culture-positive sepsis | 8.9 (5.9) | 8.8 (5.5) | 8.9 (5.2) | 9.5 (5.8) | 9.1 (5.0) | .98 |
Clinical sepsis | 5.3 (2.9) | 5.4 (2.9) | 5.0 (2.6) | 5.0 (2.5) | 5.0 (2.3) | .01 |
No sepsis | 2.9 (2.3) | 2.8 (2.3) | 2.6 (1.9) | 2.8 (2.3) | 2.7 (1.9) | .02 |
Numbers with diagnosis, n (%) | ||||||
Culture-positive sepsis | 47 (0.08) | 52 (0.09) | 34 (0.06) | 38 (0.07) | 34 (0.06) | .34 |
Clinical sepsis | 609 (1.0) | 523 (0.9) | 413 (0.7) | 364 (0.7) | 325 (0.6) | <.001 |
No sepsis | 1159 (1.9) | 1044 (1.7) | 957 (1.7) | 885 (1.6) | 859 (1.6) | <.001 |
One-way analysis of variance with logarithmic transformation for trend for duration during the period of 2015–2019.
χ2 test for change in numbers with diagnosis during the period of 2015–2019.
A blood culture was obtained in 6758 (91.8%) of the children receiving antibiotics. Sepsis was confirmed by blood culture in 206 newborns (0.73 per 1000 live births at GA ≥34 weeks); 181 of these were born at term (incidence 181 of 269 851; 0.67 per 1000 live births). Group B Streptococcus (GBS) was the predominant pathogen in culture-positive sepsis (n = 73 of 213; 34%), with an incidence of 0.26 per 1000 live births in near-term and term infants. The other commonly encountered pathogens were Escherichia coli and Staphylococcus aureus (Table 3). Overall, during the 5-year period, 2.6% of newborns in Norway with a GA ≥34 weeks were treated with antibiotics. Only 2.9% of newborns treated with antibiotics had a culture-positive sepsis; the number needed to treat (NNT) was 36 for each culture-positive sepsis episode.
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Total . |
---|---|---|---|---|---|---|
GBS | 15 | 20 | 14 | 13 | 11 | 73 |
Other streptococci | 4 | 7 | 2 | 5 | 4 | 22 |
S aureus | 7 | 8 | 5 | 6 | 5 | 31 |
E coli | 9 | 10 | 9 | 7 | 7 | 42 |
Other specified bacteriaa | 11 | 8 | 5 | 5 | 7 | 36 |
Unspecified growth | 2 | 2 | 0 | 2 | 3 | 9 |
Total (excluding CoNS,b fungi, and viruses) | 48 | 55 | 35 | 38 | 37 | 213 |
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Total . |
---|---|---|---|---|---|---|
GBS | 15 | 20 | 14 | 13 | 11 | 73 |
Other streptococci | 4 | 7 | 2 | 5 | 4 | 22 |
S aureus | 7 | 8 | 5 | 6 | 5 | 31 |
E coli | 9 | 10 | 9 | 7 | 7 | 42 |
Other specified bacteriaa | 11 | 8 | 5 | 5 | 7 | 36 |
Unspecified growth | 2 | 2 | 0 | 2 | 3 | 9 |
Total (excluding CoNS,b fungi, and viruses) | 48 | 55 | 35 | 38 | 37 | 213 |
If the culture had growth of 2 virulent organisms both were included (n = 2). Five children had 2 separate blood cultures with growth of different pathogens, thus the total number of infants with positive blood cultures is 206.
Other specified bacteria (descending frequency): Enterococcus, other Gram-positive cocci, other Gram-positive rods, Bacillus spp, Bacillus cereus, Enterobacter cloaca, Klebsiella oxytoca, Acinetobacter spp, other Gram-negative cocci, and 1 anaerobe.
A total of 133 blood cultures revealed growth of CoNS, and 83 of 133 (62%) received ≥5 d of antibiotics.
We found a nonsignificant decrease in the incidence of sepsis with the onset at 7 to 28 days from 62 of 59 932 (0.10%) in 2015 to 45 of 55 246 (0.08%) in 2019 (Table 4). The number of repeated courses of antibiotics was stable over time (Table 4). Newborn deaths remained low, at ∼0.5 per 1000 with no significant change over time (P = .54; Table 4). The mortality rate did not change appreciably among those treated with antibiotics or those not treated with antibiotics (Table 4). Death due to sepsis was uncommon, and a significant decrease in both culture-positive and clinical sepsis was observed during the study period. During the 5-year period, 14 children who died in a neonatal unit had a sepsis-related ICD-10 diagnosis. Ten of these 14 had other conditions as the primary cause of death (severe congenital anomalies, fulminant viral infections, or inborn errors of metabolism). Four children were classified as sepsis-attributable deaths (details in Table 4).
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Pc . |
---|---|---|---|---|---|---|
Admission from day 4–28 (from NPRa), n | ||||||
Age at admission: 4–6 d | 20 | 11 | 18 | <5 | 10 | .008 |
Age at admission: 7–28 d | 62 | 62 | 58 | 42 | 45 | .35 |
Repeated treatments with antibiotics (from NNNb), n | ||||||
1 course | 1781 | 1587 | 1377 | 1246 | 1182 | .56 |
≥2 courses | 51 | 33 | 37 | 37 | 35 | — |
Deaths (from NNNb) | ||||||
Total, n (%) | 33 (0.06) | 33 (0.05) | 41 (0.07) | 27 (0.05) | 29 (0.05) | .54 |
With any antibiotic treatment, n (%) | 25 (0.04) | 23 (0.04) | 26 (0.05) | 19 (0.03) | 14 (0.03) | .46 |
With diagnosis P36.9,d n | 5 | 1 | 1 | 0 | 0 | .03 |
With culture-positive sepsis,e n | 0 | 4 | 3 | 0 | 0 | .02 |
. | 2015 . | 2016 . | 2017 . | 2018 . | 2019 . | Pc . |
---|---|---|---|---|---|---|
Admission from day 4–28 (from NPRa), n | ||||||
Age at admission: 4–6 d | 20 | 11 | 18 | <5 | 10 | .008 |
Age at admission: 7–28 d | 62 | 62 | 58 | 42 | 45 | .35 |
Repeated treatments with antibiotics (from NNNb), n | ||||||
1 course | 1781 | 1587 | 1377 | 1246 | 1182 | .56 |
≥2 courses | 51 | 33 | 37 | 37 | 35 | — |
Deaths (from NNNb) | ||||||
Total, n (%) | 33 (0.06) | 33 (0.05) | 41 (0.07) | 27 (0.05) | 29 (0.05) | .54 |
With any antibiotic treatment, n (%) | 25 (0.04) | 23 (0.04) | 26 (0.05) | 19 (0.03) | 14 (0.03) | .46 |
With diagnosis P36.9,d n | 5 | 1 | 1 | 0 | 0 | .03 |
With culture-positive sepsis,e n | 0 | 4 | 3 | 0 | 0 | .02 |
—, not applicable.
Recorded in the NPR: data are complementary to NNN because all newborns are captured regardless of admitting unit (neonatal or other).
Reported to the NNN. P value calculated for repeated courses relative to the total number of treatments.
χ2 test or Fisher’s exact test (lower 2 rows).
One without severe other morbidity; started antibiotics the first day of life but died the same day. Autopsy concluded with sepsis as the probable cause of death.
Three without severe other morbidity: 2 infants with GBS sepsis (one started antibiotics on day 10 and died on day 14. The other started antibiotics on day 12 and died on day 34). One infant with E coli sepsis (started antibiotics on day 1 and died on day 4).
Discussion
In this nationwide, population-based study we found a 30% reduction in the proportion of near-term and term newborns commenced on antibiotics over the 5-year study period. Concomitantly a shorter duration of treatment in newborns without culture-positive sepsis reduced the number of days with systemic antibiotics by 37%. National initiatives and local antibiotic stewardship QI-projects were implemented during this period. The occurrence of sepsis after first week of life and sepsis-attributable mortality remained low.
The population-based design including virtually all children born in Norway at ≥34 weeks' gestation is a strength and avoids any selection bias. Another major strength is the sample size allowing for robust estimates. Daily recording in the national registry ensures almost complete data sets including blood cultures; however, the exact start and end of treatment in hours was not registered.
The main limitation of the study is the reliance on a large number of physicians for the registration into the Web-based system. Inaccuracies in the data sets are inevitable in such a registry-based study, although this is unlikely to change systematically over time or vary by geographic region. National initiatives within the Norwegian Pediatric Association for uniform use of the ICD-10 codes on discharge were implemented before the start of the study. The diagnosis of P36.9 (clinical sepsis) is controversial, and it is likely that the steep fall in the number of children discharged with this diagnostic code was at least partly driven by a stricter use of the diagnostic criteria during the study period.19,25
Blood culture growth of CoNS was considered to be a likely contaminant, in line with previous studies on this subject.1,16 However, there is a possibility that some of the blood cultures with growth of such low-grade pathogens represented true sepsis. We did not have information on maternal antibiotics during labor, and therefore we cannot estimate the impact of such antibiotics on incidence of early-onset sepsis in our population.
A previous study from the NNN during 2009–2011 showed that 2.3% of term infants were treated with antibiotics the first week of life.16 At the end of the current study, the corresponding proportion in term infants was 1.9%. Several interventions took place during the study period, including an electronic reminder in the NNN daily registration platform to reconsider antibiotic prescription after 3 days instituted in 2015. Limited by the observational design, we cannot draw conclusions regarding the effect of the specific interventions. Interestingly, institutions that had antibiotic stewardship QI-projects during the study period saw a substantially larger reduction in antibiotic use compared with that of other hospitals.21,22 This suggests that local projects may strengthen the effect of national initiatives. The geographic variation in the proportion of newborns started on antibiotic treatment persisted, but the duration of antibiotics was more uniform at the end of our study period.
A national neonatal sepsis guideline was not available during the study period. However, by tradition most Norwegian neonatal units do not treat asymptomatic infants just on the basis of risk factors. It is well known that screening guidelines based on maternal risk factors for sepsis in newborns may result in extensive use of systemic antibiotics. By using the Centers for Disease Control and Prevention 2010 guidelines, 7% of infants born at ≥35 weeks' gestation received empirical antibiotics for suspected early-onset sepsis.26 The National Institute for Health and Care Excellence guidelines,27 also using a risk-based approach, led to empirical treatment of far more than 10% of near-term and term infants in a UK study.28 Implementation of an electronic neonatal sepsis calculator, which takes into account clinical observations in addition to maternal risk factors, may safely reduce antibiotic use during the first 3 days of life by ∼50%, compared with that of traditional risk-based management.28–31
The decision to start antibiotics could, in principle, be made by risk factors or when the newborn presents with symptoms. Most decisions tools use a combination of these 2 in addition to laboratory markers, as is the current clinical practice in Norway. There will always be a trade-off between treating infants without infections and delaying treatment in those who do have infection. The NNT to cover 1 culture-positive sepsis in the current study was 36, which is lower than in 2009–2011 (NNT: 44).16 Delayed treatment should be avoided as much as possible, although the clearly adverse events (permanent morbidity or death) are rare. The sensitivity of any guideline or screening algorithm will never reach 100%, and the balance between optimal sensitivity and specificity to detect sepsis will continue to challenge neonatologists.31,32
Algorithms and guidelines need to consider the setting, including the incidence of sepsis in the population, recommendations for GBS screening and intrapartum antibiotics. Similar to the UK guidelines,33 the Norwegian national policy does not recommend a systematic screening for GBS during pregnancy in healthy female patients. Intrapartum antibiotics are recommended if a previous child has had GBS sepsis, if a urinary tract infection with GBS has been diagnosed in pregnancy, and in the case of maternal fever or prolonged rupture of membranes >18 hours.34 The usual standard of care in Norway is observation of newborns 48 hours postdelivery, which allows for recognition of signs occurring before discharge. Early signs of infection are subtle, and serial monitoring of vital signs in newborns with risk factors should capture symptoms as soon as they appear and lead to therapy.22 Serial monitoring is currently also one of the possible strategies to identify infants with suspected sepsis, suggested by the American Academy of Pediatrics.35 Early hospital discharge may increase the risk of missing symptoms and signs, and careful selection of low-risk newborns should be implemented before any change in discharge policy.
In the current study, the decreased duration of antibiotic treatment in noninfected newborns was minor, especially with regard to the reduction in the proportion of newborns started on antibiotics. However, because of the large reduction in newborns commenced on antibiotics, the newborns treated in 2019 had a higher likelihood of true sepsis, as indicated by a lower NNT. They are therefore not directly comparable to those started on antibiotics in 2015. In the Surveillance and Correction of Unnecessary Antibiotic Therapy (SCOUT) study, the decision to start treatment was deemed inappropriate in only 4% of the cases, but continuation was deemed inappropriate in 39% of cases.13 The ultimate goal to discontinue antibiotics in nonseptic infants within 36–48 hours remains, and future initiatives should be aimed toward early discontinuation. In 1 of the 2 antibiotic stewardship projects conducted in Norway, an automatic stop order was part of the interventions, as also described from other institutions.21,36,37 Similar to our study, the SCOUT study saw a 27% reduction in overall antibiotic use in newborns in an antibiotic stewardship program, without changes in any safety outcomes.13
Our findings suggest that in a high-resource setting with low newborn mortality, a stricter policy than previous practice regarding antibiotics is safe without increased risk for adverse outcomes. The number of sepsis-attributable deaths should, however, be interpreted with caution, given the low frequency of this event. With a limited accuracy of decision tools and biomarkers to identify sepsis, clinical vigilance is paramount, and a sufficiently low threshold to start antibiotics is necessary to avoid unnecessary deaths due to missed sepsis diagnoses.
Conclusions
In this nationwide study, we found a reduction in the proportion of term and near-term newborns who were treated with antibiotics over a 5-year period. We found no increased incidence of readmissions, sepsis-attributable mortality, or delayed diagnosis of sepsis during the same period. The duration of antibiotic therapy in nonseptic newborns was reduced and better aligned across the country but needs further efforts to reduce unnecessary prolonged treatment.
Acknowledgments
We thank the neonatal units from the following hospital in Norway for contributing with data to the NNN and thus making this study possible: Oslo University Hospital-Rikshospitalet, Oslo; Oslo University Hospital-Ullevål, Oslo; Akershus University Hospital, Lørenskog; Drammen Hospital, Drammen; Østfold Hospital Trust,Fredrikstad; Innlandet Hospital Trust, Lillehammer; Innlandet Hospital Trust, Elverum; Telemark Health Trust, Skien; Hospital of Southern Norway, Kristiansand; Hospital of Southern Norway, Arendal; Stavanger University Hospital, Stavanger; Haukeland University Hospital, Bergen; Fonna Health Trust, Haugesund; Health Sunnmøre Trust, Ålesund; Førde Health Trust, Førde; St Olav University Hospital, Trondheim; Nord-Trøndelag Health Trust, Levanger; Nordland Central Hospital, Bodø; University Hospital of North Norway, Tromsø; and Finnmark Health Trust, Hammerfest.
Dr Mundal conducted the analyses, drafted the initial manuscript and revised the manuscript; Dr Stensvold was responsible for data retrieval and processing, contributed to data analyses and reviewed the manuscript; Dr Rønnestad helped conceptualize the study and ensured complete data collection and reviewed the manuscript; Dr Klingenberg conceptualized and designed the study, contributed during data analysis and reviewed the manuscript; Dr Størdal conceptualized and designed the study and contributed considerably to data analyses and in the writing process; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: No external funding.
References
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.
Comments