Longitudinal antibiotic susceptibilities of neonatal Escherichia coli infection

September 08, 2021


Flannery DD, Akinboyo IC, Mukhopadhyay S, Tribble AC, Song L, Chen F, Li Y, Gerber JS, Puopolo KM. Antibiotic Susceptibility to Escherichia coli Among Infants Admitted to Neonatal Intensive Care Units Across the US From 2009 to 2017. JAMA Pediatr. Published online November 09, 2020. doi:10.1001/jamapediatrics.2020.4719. PMID: 33165599.


Dr Edward Broad
Norfolk and Norwich University Hospital

Dr Paul Cawley
Evelina Neonatal Intensive Care Unit, St Thomas’ Hospital, London
The Centre for the Developing Brain, King’s College London




Among inborn infants with E. coli isolates on blood, urine or cerebrospinal fluid cultures who are admitted to participating US neonatal intensive care units from birth to 1 year of age (P), what is the annual susceptibility/ nonsusceptibility proportion of the E. coli isolates to antibiotic categories (O) and has this changed over a nine-year study period (T)?


  • Design: Retrospective Cohort Study
  • Allocation: No allocation
  • Blinding: Unblinded
  • Follow-up period: Disposition as known during the data-collection and analysis period December 1, 2018 to November 30, 2019
  • Setting: Neonatal Intensive Care Units in the United States of America, contributing microbiology data to the administrative Premier Health Database (Premier Inc) during the study period calendar years
  • Population:
    • Inborn infants from January 1, 2009, to December 31, 2017, admitted to NICUs from birth to 1 year of age with coli isolated from blood, cerebrospinal fluid, or urine cultures
    • No exclusions specified
  • Exposure:
    • Episodes of coli infection, defined as 1 or more growth of E. coli in blood cultures, CSF cultures or urine cultures during the admission
  • Outcomes:
    • Primary outcome:
      • Change in annual rate of antibiotic resistance demonstrated by coli isolates between 2009 & 2017
    • Secondary outcomes:
      1. E. coli nonsusceptibility to ampicillin
      2. E. coli nonsusceptibility to aminoglycosides
      3. E. coli Extended Spectrum Beta-Lactamase (ESBL) phenotype, defined as E. coli nonsusceptibility to at least one of cefotaxime, ceftriaxone, ceftazidime or cefepime
      4. Carbapenem-resistant Enterobacteriaceae defined as E. coli nonsusceptibility to at least one of imipenem, meropenem, doripenem or ertapenem sodium
      5. E. coli nonsusceptibility to both ampicillin and gentamicin in patients with early onset sepsis
    • Subgroup analyses for mortality, hospital size, length of stay, birth weight (<1500g and ≥1500g), sex, race/ethnicity, geographic region, teaching hospital status
  • Analysis and Sample Size:
    • 117,484 infants in 170 neonatal units in the United States with available microbiological data
    • 733 infants of these had 1 or more E. coli isolates from blood, CSF or urine culture
    • 721 of these had available antibiotic susceptibility data (from a total of 69 centres)
    • Nonsusceptibility to an antibiotic category was defined as either resistant or intermediate findings on clinical laboratory reports
    • Proportion of infants with nonsusceptible organisms calculated in the first episode of coli infection as:
      • Number of infants with E. coli nonsusceptibility to antibiotic category / Total number of infants with available susceptibility testing to antibiotic category
    • New episodes of E. coli infection in the same patient were defined as subsequent cultures positive for E. coli more than 30 days after the previous episode
    • Generalised linear models were used to estimate antibiotic category nonsusceptibility as a function of year
    • χ2 test and analysis of variance used to compare demographics and clinical characteristics between antibiotic category groups in early onset sepsis
    • Logistic regression used to compare in-hospital mortality for early onset sepsis
  • Patient follow-up: Exclusions to follow-up not applicable


721 infants had at least 1 episode of E. coli infection for which antibiotic susceptibility data were available. Of these infants, 60.2% were male (n=434), 60.6% had a birthweight of <1500g (n=437), and 69.5% of the E. coli infections were defined as late onset (>72hrs of age, n=501). 67.4% (n=486) of infants were admitted within a teaching hospital.

Median age of first episode was 14 days (interquartile range, 1-33). The majority of patients, 96.3%, only had a single episode (n=694), with 3.3% (n=24) having 2 episodes and 0.4% (n=3) having 3 episodes of E. coli infection.

The source of isolates in first episodes:

Culture Overall Early Onset Sepsis Late Onset Sepsis
Blood 49.7% (n=358) 96.4% (n=212) 29.1% (n=146)
Cerebrospinal Fluid 0.8% (n=6) 1.4% (n=3) 0.6% (n=3)
Urine 44.7% (n=322) 0.9% (n=2) 63.9% (n=320)

E. coli nonsusceptibility over the study period (incl. early and late onset sepsis):

Antibiotic Category Isolates Tested Nonsusceptibility Result
Ampicillin 99.9% (n=720) 66.7% (n=480)
Aminoglycosides 99.7% (n=719) 16.6% (n=119)
ESBL Phenotype 96.4% (n=695) 4.9% (n=34)
Carbapenems 91.8% (n=662) 0% (n=0)

Over the longitudinal study period of 9 years, no statistically significant change in annual proportions of antibiotic nonsusceptibility were observed to any antibiotic category.

Sub-group analysis demonstrated a small but statistically significant decrease in aminoglycoside nonsusceptibility in E. coli causing late onset sepsis; yearly change -1.5% (95% Confidence Interval: −2.88% to −0.18%). No other significant trends were identified.

In the subset of patients with early onset sepsis confirmed by blood or CSF culture:

    • 31.7% (n=69) displayed sensitivity to both ampicillin and gentamicin
    • 58.3% (n=127) were nonsusceptible to ampicillin, but susceptible to gentamicin
    • 0% (n=0) were susceptible to ampicillin and nonsusceptible to gentamicin
    • 10.1% (n=22) were nonsusceptible to both ampicillin and gentamicin
    • Mortality was 8.7% (n=6) in the ampicillin and gentamicin susceptible group, 9.4% (n=12) in the ampicillin non-susceptible and gentamicin susceptible group, and 18.2% (n=4) in the ampicillin and gentamicin nonsusceptible group. There was no statistically significant difference between these groups (p=0.41)


The authors concluded that:

    1. Substantial proportions of coli isolates were nonsusceptible to commonly administered antibiotics used empirically for both neonatal early onset and late onset infections
    2. No significant temporal changes in resistance patterns were observed from 2009 to 2017
    3. Longitudinal, national micro-organism resistance pattern surveillance is required in neonates to optimise empirical antibiotic therapy


Infections in the neonatal period contribute significantly to morbidity and mortality, and are responsible for 13% of neonatal deaths in the UK(1), and 36% worldwide(2). Widespread use of intrapartum antibiotics has been credited as effective in reducing early onset Group B Streptococcal (GBS) disease(3), but concurrently Escherichia coli (E. coli) infections in neonates have risen(4-5); leading to concerns that increasing pre-birth antimicrobial chemoprophylaxis may generate selective E. coli exposure to neonates, and drive antimicrobial resistance in this species(5).

Flannery et al., investigated rates of neonatal antimicrobial nonsusceptibility in E. coli isolates over a 9-year period, in US centres. Although no significant increase in resistance patterns were seen over time, the proportion of nonsusceptibility remains non-reassuring. Amongst neonates with early onset sepsis, 10% of E. coli isolates demonstrated resistance to both ampicillin and gentamicin – rendering first-line empirical antimicrobials ineffective in this group. Additionally, whilst no significant yearly change was observed, phenotypic Extended Spectrum Beta-Lactamase activity was evident in 5% of isolates.

The results appear to be concordant with previous data from the US, citing 54% ampicillin E. coli non-susceptibility from 1997-2006, increased from 20% in 1979-1992(5). Nonsusceptibility rates are similar in other high resource settings, such as UK centres(6). Global variation in antimicrobial resistance is significant and countries must be pro-active in collecting their own microbiological surveillance – including in lower resource settings, where rates of antimicrobial resistance are higher(7).

This study’s conclusions are strengthened by its large sample size – pooling from a total of 117,484 neonatal admissions. The findings are clinically relevant – derived from clinical laboratory reports on urine, blood and cerebrospinal fluid. However, the study is retrospective and consequently analysis is confined to the scope of the original database. The paper does not provide key data of microbiological interest, such as minimum inhibitory concentration, paired maternal-infant sampling for vertical transmission inquiry, use of intrapartum antimicrobials, nor genetic analysis for subtyping or known genes of virulence. Gestational ages are not specified, and data on late onset E. coli nonsusceptibility are limited.

Furthermore, the incidence and antibiotic nonsuspectibility patterns of other pathogens are not reported – choices of early neonatal empirical therapy cannot be determined by one pathogen alone, but need to incorporate sensitivities for other likely causative pathogens. Nevertheless, this paper provides a timely reminder to neonatologists – E. coli antimicrobial resistance is not insignificant. Resistance to first-line antibiotics must be considered in infants who are deteriorating despite adequate source control.

No significant temporal changes in resistance patterns were observed in this paper’s 9 year study period. Significant rises in neonatal antimicrobial resistance may not be inevitable and investment in good antimicrobial stewardship is thus warranted. The gauntlet has been thrown down – in 10 to 20 year’s time, will we have maintained E. coli antimicrobial resistance at its current proportions, and can strict stewardship reverse this impending predicament. We have a duty to safeguard effective antibiotics for future generations. This paper highlights the importance of national and international antimicrobial resistance surveillance projects, such as neonIN – and we encourage more to join(8).


  1. Oligbu G, Ahmed L, Ferraras-Antolin L, and Ladhani S. Retrospective analysis of neonatal deaths secondary to infections in England and Wales, 2013–2015. Archives of Disease in Childhood – Fetal and Neonatal Edition 2020; [online] Available at: https://fn.bmj.com/content/early/2020/11/24/archdischild-2020-319093
  2. World Health Organisation (2011). WHO | Newborn death and illness. Who.int. [online] Available at: https://www.who.int/pmnch/media/press_materials/fs/fs_newborndealth_illness/en/
  3. Phares C, Lynfield R, Farley M, Mohle-Boetani J, Harrison L, Petit S. et al. Epidemiology of invasive group B streptococcal disease in the United States, 1999-2005. JAMA 2008; 299(17):2056–65
  4. Stoll B, Hansen N, Fanaroff A, Wright L, Carlo W, Ehrenkranz R, et al. Changes in Pathogens Causing Early-Onset Sepsis in Very-Low-Birth-Weight Infants. New England Journal of Medicine 2002; 347(4), 240–247. https://doi.org/10.1056/nejmoa012657.
  5. Bizzarro M, Dembry L, Baltimore R, and Gallagher P. Changing Patterns in Neonatal Escherichia coli Sepsis and Ampicillin Resistance in the Era of Intrapartum Antibiotic Prophylaxis. Pediatrics 2008; 121(4), pp.689–696.
  6. Cailes B, Kortsalioudaki C, Buttery J, Pattnayak S, Greenough A, Matthes J, et al. Antimicrobial resistance in UK neonatal units: neonIN infection surveillance network. Archives of Disease in Childhood. Fetal and Neonatal Edition 2018; 103(5):F474–F478.
  7. Agarwal R, and Sanka M. Characterisation and antimicrobial resistance of sepsis pathogens in neonates born in tertiary care centres in Delhi, India: a cohort study. The Lancet Global Health 2016; 4(10), pp.e752–e760.
  8. neonin.org.uk. (n.d.). Home – neonIN Surveillance Network. [online] Available at: https://neonin.org.uk/#/home