Abstract
Background
Antimicrobial resistance (AMR) is a growing threat to newborns in low- and middle-income countries (LMIC).
Methods
We performed a prospective cohort study in 3 tertiary neonatal intensive care units (NICUs) in Pune, India, to describe the epidemiology of neonatal bloodstream infections (BSIs). All neonates admitted to the NICU were enrolled. The primary outcome was BSI, defined as positive blood culture. Early-onset BSI was defined as BSI on day of life (DOL) 0–2 and late-onset BSI on DOL 3 or later.
Results
From 1 May 2017 until 30 April 2018, 4073 neonates were enrolled. Among at-risk neonates, 55 (1.6%) developed early-onset BSI and 176 (5.5%) developed late-onset BSI. The majority of BSIs were caused by gram-negative bacteria (GNB; 58%); among GNB, 61 (45%) were resistant to carbapenems. Klebsiella spp. (n = 53, 23%) were the most common cause of BSI. Compared with neonates without BSI, all-cause mortality was higher among neonates with early-onset BSI (31% vs 10%, P < .001) and late-onset BSI (24% vs 7%, P < .001). Non–low-birth-weight neonates with late-onset BSI had the greatest excess in mortality (22% vs 3%, P < .001).
Conclusions
In our cohort, neonatal BSIs were most commonly caused by GNB, with a high prevalence of AMR, and were associated with high mortality, even in term neonates. Effective interventions are urgently needed to reduce the burden of BSI and death due to AMR GNB in hospitalized neonates in LMIC.
Keywords: neonatal sepsis, antimicrobial resistance, neonatal intensive care unit, low- and middle-income countries
Among 4073 neonates admitted to the neonatal intensive care unit in Pune, India, the majority of bloodstream infections occurred during the first 2 weeks of life, were caused chiefly by resistant gram-negative bacteria, and were associated with higher mortality, including in term neonates.
Of 2.5 million babies who die within the first 4 weeks of life every year, 23% are estimated to die due to infectious causes, including sepsis and pneumonia [1]. Progress in neonatal survival has not kept pace with global reductions in child mortality, and neonatal deaths now account for 47% of deaths in children aged <5 years [1, 2]. In regions with the highest neonatal mortality, infections are responsible for up to 50% of neonatal deaths [1].
Facility delivery is promoted in many low- and middle-income countries (LMIC) to ensure access to emergency obstetric care and to decrease neonatal mortality. Low- and middle-income countries (LMIC) healthcare facilities increasingly provide care for premature and sick neonates in neonatal intensive care units (NICUs) [3–6]. Similar trends are seen in India, a lower-middle-income country where facility births have risen from 39% (2005–2006) to 79% (2015–2016) and neonatal mortality is high at 23 per 1000 live births in 2018 [1, 7, 8].
Hospitalized neonates are especially vulnerable to invasive infections, including bloodstream infections (BSIs), due to such factors as an immature immune system, poor skin integrity, need for life-sustaining invasive procedures, and prolonged hospital stays [9–11]. Data on the epidemiology of BSI in LMIC NICUs are scarce and often limited to outbreak investigations [12–18]. A 2019 review of neonatal sepsis in South Asia reported widespread antimicrobial resistance (AMR) and predominance of gram-negative infections, which are associated with significantly greater morbidity and mortality and have limited treatment options in neonates [19–22]. While AMR is a global crisis, India is particularly affected, with resistance to third-generation cephalosporins exceeding 70% for common gram-negative bacteria (GNB), as well as widespread carbapenem resistance in up to 70% of Acinetobacter baumannii and more than half of Klebsiella pneumoniae isolates [23].
There are scant data on the epidemiology of neonatal sepsis and associated prevalence of AMR in LMIC, especially in NICU populations. It is critical to better understand the local, regional, and global epidemiology of neonatal sepsis and AMR among hospitalized neonates in order to guide future prevention efforts and policies.
METHODS
Study Design and Population
We conducted a prospective cohort study from 1 May 2017 until 30 April 2018 in neonates admitted to the NICU at 3 tertiary care centers in Pune, India: Byramjee Jeejeebhoy Government Medical College (BJGMC) & Sassoon Hospital, Dr. D. Y. Patil Medical College, Hospital and Research Centre, and King Edward Memorial (KEM) Hospital. All neonates admitted to the NICU were enrolled and followed until discharge, transfer, or death. This minimal-risk study did not require consent and was approved by the Johns Hopkins Medicine Institutional Review Board, site ethics committees, and the Indian Health Ministry’s Screening Committee.
Study Sites
BJGMC is a government medical college affiliated with Sassoon Hospital that has a 60-bed NICU. Dr D. Y. Patil Medical College is a private medical college and has a 26-bed NICU. KEM Hospital is a nongovernmental facility run by a charitable trust and has a 46-bed NICU. All participant NICUs have an open ward structure and admit inborn and outborn neonates. All hospitals have an active infection control committee. Prior to study initiation, the study team assessed microbiology laboratory capacity at all sites by. Site microbiology laboratories are accredited by the Indian National Accreditation Board for Testing & Calibration Laboratories and perform routine microbiology tests, including organism identification and antimicrobial susceptibility testing (AST). Automated methods include use of VITEK for organism identification and AST; manual methods include organism identification by agar plate and biochemical workup and AST by disk diffusion methods.
Data Sources
Neonates admitted to the NICU were identified by regular review of the admission register. Study staff performed prospective review of neonates’ medical records, including baseline demographic and clinical characteristics and daily clinical and microbiology data. All laboratory studies were obtained at the discretion of the clinical teams.
Study Outcomes
The primary outcome of interest was BSI, defined as a blood culture positive for a known neonatal pathogen. Blood cultures positive for well-described contaminants were not considered BSIs. Early-onset BSI was defined as BSI occurring day of life (DOL) 0 through DOL 2, where DOL 0 is the day of birth. Late-onset BSI was defined as BSI occurring on or after DOL 3. Only first episodes of early-onset BSI and late-onset BSI were included. Secondary outcomes of interest included pathogen distribution, AMR among GNB, and all-cause mortality.
Study Cohort Definitions
All neonates admitted to the NICU were enrolled in this prospective cohort study. Neonates included in the early cohort were considered at risk for early-onset BSI by presence in the NICU during the risk period, DOL 0 through DOL 2. Neonates included in the late cohort were considered at risk for late-onset BSI by presence in the NICU during the risk period, DOL 3 or later. Neonates could be included in both cohorts. Neonates admitted to the NICU discharged home or transferred to the nursery or postpartum ward on the same day as admission were excluded from analysis.
Statistical Analyses
Descriptive analysis was performed to characterize the study population. The incidence of early-onset BSI was described as the proportion of neonates who experienced early-onset BSI among neonates in the early cohort. The incidence of late-onset BSI was described as the proportion of neonates who experienced late-onset BSI among those in the late cohort. Incidence density of late-onset BSI was expressed as the number of BSI cases per 1000 patient-days among neonates included in the late cohort; only patient-days at risk for late-onset BSI were included. All-cause mortality was expressed as a proportion of neonates deceased among admitted neonates. The Pearson χ 2 test was used to evaluate group differences in mortality. The association between variables of interest and risk of early and late-onset BSI was explored using univariate and multivariable logistic regression with multiple imputation for missing data in the adjusted model. Statistical analyses were performed using Stata version 15.0 (Stata Corp., College Station, TX).
RESULTS
During the study period, 4619 neonates were enrolled (Figure 1). Neonates discharged or transferred on the same day as NICU admission (n = 546) were excluded from analysis. An additional 94 neonates were excluded due to insufficient data. The remaining 4073 neonates served as the early and late cohort. The early cohort (n = 3341) excluded neonates admitted on DOL 3 or later (n = 732). The late cohort (n = 3178) excluded all neonates who exited the NICU before DOL 3 (n = 895). The 2 cohorts were quite similar, with a substantial proportion of inborn and low birth weight (LBW, <2500 g at birth) neonates (Table 1). Of 4073 neonates included in this study, 226 (5.5%) experienced at least 1 episode of BSI. There were 55 cases of early-onset BSI (1.6%; 95% confidence interval [CI], 1.2–2.1; Table 2). The incidence was highest in lower birth weight neonates, with 4.3% (95% CI, 1.9–8.3) of extremely LBW (ELBW, <1000 g) experiencing early-onset BSI. Late-onset BSI was more common than early-onset BSI; 176 neonates (5.7%, 95% CI, 4.9–6.5) experienced at least 1 episode of late-onset BSI (Table 1). Late-onset BSI was more common in neonates with lower birth weight, with 18.6% (95% CI, 12.6–25.9) of ELBW neonates and 10.0% (95% CI, 7.8–12.6) of very LBW (VLBW, 1000 g–1499 g) neonates developing late-onset BSI. The late-onset BSI rate per 1000 patient-days was 6.6 (95% CI, 5.6–7.6); comparable rates were seen in ELBW, VLBW, and LBW neonates. The BSI rate was highest among neonates with a birth weight of 2500 g or greater, at 8.9 per 1000 patient-days (95% CI, 6.0–12.6).
Figure 1.
Study flow diagram. Neonates included in the early cohort were considered at risk for experiencing the outcome of early-onset BSI by their presence in the NICU during the risk period, DOL 0 through DOL 2. Neonates not at risk for early-onset BSI, those admitted on or after DOL 3, were excluded from the early cohort. Neonates included in the late cohort were considered at risk for experiencing the outcome of late-onset BSI by their presence in the NICU during the risk period, DOL 3 or later. Neonates not at risk for late-onset BSI, those who exited the NICU before DOL 3, were excluded from the late cohort. *Early exit, discharge (n = 26), or transfer to ward (n = 426) on same calendar day as admission. **Insufficient data, no birth weight or admission weight. Abbreviations: BSI, bloodstream infection; DOL, day of life; NICU, neonatal intensive care unit.
Table 1.
Baseline Clinical and Demographic Characteristics of Neonates at Risk for Early- and Late-onset Bloodstream Infection
| Characteristic | Early Cohort (n = 3341) | Late Cohort (n = 3178) |
|---|---|---|
| Maternal age, median (interquartile range), y | 25 (22–28) | 25 (22–28) |
| Male, n (%) | 1791 (54) | 1703 (54) |
| Gestational age, mean (SD), wk | 34.9 (3.7) | 34.9 (3.7) |
| Preterm, <37 weeks’ gestation, n (%) | 1899 (64) | 1807 (64) |
| Birth weight, mean (SD), g | 1996 (702) | 2005 (693) |
| Low birth weight, n (%) | 2421 (72) | 2313 (73) |
| Multiple gestation, n (%) | 398 (12) | 354 (11) |
| Inborn, n (%) | 3115 (93) | 2780 (87) |
| Cesarean delivery, n (%) | 1335 (44) | 1352 (46) |
| Positive-pressure ventilation at delivery, n (%) | 691 (24) | 573 (20) |
| Mechanical ventilation on admission, n (%) | 368 (12) | 341 (11) |
| Central line on admission, n (%) | 256 (9) | 240 (8) |
| Pressors on admission, n (%) | 393 (13) | 370 (12) |
| Antibiotics on admission, n (%) | 1221 (41) | 1333 (45) |
Baseline clinical and demographic characteristics for early and late cohorts. Note that early and late cohorts overlap. A total of 2446 neonates (60% of full cohort of 4073 neonates) were included in both early and late cohorts; please see Figure 1. Select characteristics not available for all neonates in cohort; percentages calculated based on denominator of neonates with known characteristic.
Abbreviation: SD, standard deviation.
Table 2.
Bloodstream Infection Incidence by Birth Weight Strata, Early and Late Cohorts
| Birth Weight Stratum | Early Cohort BSI Cases as a Proportion of NICU Admissionsa (n = 3341) | Late Cohort BSI Cases as a Proportion of NICU Admissionsa (n = 3178) | Late Cohort BSI Rate per 1000 Patient- daysb (n = 3178) |
|---|---|---|---|
| Extremely low birth weight (<1000 g) | 8/185 = 4.3% (1.9–8.3) | 27/145 = 18.6% (12.6–25.9) | 6.0 (4.0–8.7) |
| Very low birth weight (1000 g–1499 g) | 9/635 = 1.4% (0.7–2.7) | 64/641 = 10.0% (7.8–12.6) | 5.9 (4.6–7.6) |
| Low birth weight (1500 g–2499 g) | 27/1601 = 1.7% (1.1–2.4) | 54/1527 = 3.5% (2.7–4.6) | 6.7 (5.0–8.7) |
| 2500 g+ | 11/920 = 1.2% (0.6–2.1) | 31/865 = 3.6% (2.5–5.0) | 8.9 (6.0–12.6) |
| Total | 55/3341 = 1.6% (1.2–2.1) | 176/3178 = 5.5% (4.8–6.4) | 6.6 (5.6–7.6) |
aIncidence expressed in number of cases divided by number of NICU admissions, % (95% confidence interval).
bLate cohort incidence density is expressed as the BSI rate per 1000 patient-days and takes into account patient-days during at-risk period or any days in the NICU occurring on or after day of life 3 (95% confidence interval).
Abbreviations: BSI, bloodstream infection; NICU, neonatal intensive care unit.
The range of BSI occurrence was DOL 0 through DOL 110. The highest risk period for BSI was the first 2 weeks of life; 84.4% of infections occurred in the first 2 weeks (Figure 2). Five neonates experienced both early-onset BSI and late-onset BSI. One neonate had 2 blood cultures positive for Staphylococcus aureus on DOL 1 and DOL 4; these cultures were considered a single case of early-onset BSI.
Figure 2.
Distribution of BSIs by day of life (DOL). BSI cases by DOL on which blood culture was obtained. Day of birth is DOL 0. Early-onset BSI cases (n = 55) occurred on DOL 0–2, whereas late-onset BSI cases (n = 176) occurred on DOL 3 or later. Figure is censored at DOL 20; 92% (n = 212) of BSIs occurred during the first 3 weeks of life. Abbreviations: BSI, bloodstream infection; CRO, carbapenem-resistant organism; GN, gram-negative; GP, gram-positive.
GNB accounted for 53% of early-onset BSI and 60% of late-onset BSI. Klebsiella spp. (n = 53, 23%), Acinetobacter spp. (n = 30, 13%), and Citrobacter spp. (n = 25, 11%) were the 3 most common GNB identified (Table 3). Among GNB, 61 (45%) isolates were resistant to carbapenems. Gram-positive organisms were responsible for 47% of early-onset BSI and 29% of late-onset BSI; coagulase-negative Staphylococcus (CONS) and S. aureus were the most frequently identified gram-positive organisms. Among late-onset BSI, Candida spp. were identified in 11% of cases. There were no early-onset fungal BSI cases.
Table 3.
Pathogen Distribution for Early- and Late-onset Bloodstream Infections
| Organism | Early-onset BSI n = 55 n (%) | Early-onset CROa n = 11 n (%) | Late-onset BSI n = 176 n (%) | Late-onset CROa n = 50 n (%) | Total n = 231 n (%) | Total CROa n = 61 n (%) |
|---|---|---|---|---|---|---|
| Gram-negative organisms | 29 (53) | 11 (38) | 106 (60) | 50 (51) | 135 (58) | 61 (45) |
| Klebsiella spp. | 10 (18) | 6 (67) | 43 (24) | 19 (45) | 53 (23) | 25 (48) |
| K. pneumoniae | 10 (18) | 6 (67) | 42 (24) | 18 (44) | 52 (23) | 24 (48) |
| Klebsiella spp., unspecified | 0 | ... | 1 (1) | 1 (100) | 1 (0) | 1 (100) |
| Acinetobacter spp. | 3 (5) | 0 | 27 (15) | 15 (63) | 30 (13) | 15 (56) |
| A. baumannii | 1 (2) | 0 | 11 (6) | 9 (82) | 12 (5) | 9 (75) |
| Acinetobacter spp., unspecified | 2 (4) | 0 | 16 (9) | 6 (46) | 18 (8) | 6 (40) |
| Citrobacter spp. | 10 (18) | 3 (33) | 15 (9) | 6 (43) | 25 (11) | 9 (39) |
| C. freundii | 2 (4) | 1 (50) | 7 (4) | 0 | 9 (4) | 1 (13) |
| Citrobacter spp., unspecified | 8 (15) | 2 (29) | 8 (5) | 6 (75) | 16 (7) | 8 (53) |
| Escherichia spp. | 3 (5) | 1 (33) | 5 (3) | 3 (60) | 8 (3) | 4 (50) |
| E. coli | 1 (2) | 1 (100) | 5 (3) | 3 (60) | 6 (3) | 4 (67) |
| Escherichia spp. | 2 (4) | 0 | 0 | ... | 2 (1) | 0 |
| Enterobacter spp. | 0 | ... | 5 (3) | 2 (40) | 5 (2) | 2 (40) |
| E. cloacae | 2 (1) | 1 (50) | 2 (1) | 1 (50) | ||
| Enterobacter spp., unspecified | 3 (2) | 1 (33) | 3 (1) | 1 (33) | ||
| Pseudomonas spp. | 0 | ... | 3 (2) | 1 (50) | 3 (1) | 1 (50) |
| P. aeruginosa | 2 (1) | 1 (50) | 2 (1) | 1 (50) | ||
| Pseudomonas spp., unspecified | 1 (1) | ... | 1 (0) | ... | ||
| Serratia spp. | 0 | ... | 3 (2) | 1 (33) | 3 (1) | 1 (33) |
| S. marcescens | 3 (2) | 1 (33) | 3 (1) | 1 (33) | ||
| Elizabethkingia spp. | 0 | ... | 3 (2) | 3 (100) | 3 (1) | 3 (100) |
| E. meningoseptica | 3 (2) | 3 (100) | 3 (1) | 3 (100) | ||
| Other gram-negative spp. | 3 (5) | 1 (33) | 2 (1) | 0 | 5 (2) | 1 (25) |
| Gram-negative rod, non-fermenter | 3 (5) | 1 (33) | 2 (1) | 0 | 5 (2) | 1 (25) |
| Gram-positive organisms | 26 (47) | ... | 51 (29) | ... | 77 (33) | ... |
| Staphylococcus spp. | 22 (40) | ... | 40 (22) | ... | 62 (27) | ... |
| Coagulase-negative Staphylococcus | 7 (13) | ... | 18 (10) | ... | 25 (11) | ... |
| S. aureus | 8 (15) | ... | 10 (6) | ... | 18 (8) | ... |
| Staphylococcus spp., unspecified | 7 (12) | ... | 12 (7) | ... | 19 (8) | ... |
| Enterococcus spp. | 3 (5) | ... | 8 (4) | ... | 11 (5) | ... |
| E. faecalis | 1 (2) | ... | 1 (0) | ... | 2 (1) | ... |
| E. faecium | 1 (0) | ... | 1 (0) | ... | ||
| Enterococcus spp., unspecified | 2 (4) | ... | 6 (3) | ... | 8 (3) | ... |
| Streptococcus spp. | 1 (2) | ... | 3 (2) | ... | 4 (2) | ... |
| Yeast | 0 | ... | 19 (11) | ... | 19 (8) | ... |
| Candida spp. | 0 | ... | 19 (11) | ... | 19 (8) | ... |
| C. albicans | 2 (1) | ... | 2 (1) | ... | ||
| Candida spp., unspecified | 17 (10) | ... | 17 (7) | ... |
Unique bacterial and fungal isolates from neonates included in early and late cohorts. All dashes mean “not applicable”. There are no missing data. For Gram-negative organisms, there were no early onset infections due to Enterobacter, Pseudomonas, Serratia, and Elizabethkingia (indicated by zero in the table). Therefore, the subcategories for each bacterial genus were left blank (example, E. cloacae). Additionally, the column “Early-onset CRO” included dashes for these organisms because carbapenem resistance is not applicable if there were no infections due to a particular organism. Under late-onset CRO, the dashes for Escherichia spp. and Pseudomonas spp, unspecified indicate testing was not done for carbapenem resistance. For Gram-positive organisms and yeast, the columns “Early-onset CRO” and “Late-onset CRO” are not applicable — dashes were used to indicate this.
aPercentage of carbapenem-resistant isolates is calculated using a denominator of number of isolates for which carbapenem resistance testing was performed.
Abbreviations: BSI, bloodstream infection; CRO, carbapenem-resistant organism.
In the early cohort, the median length of stay was 4 days (interquartile range [IQR], 2–11), and 343 neonates (10%) died. The median gestational age among neonates with BSI who died was 31 weeks (IQR, 29–35). Mortality was higher in early cohort neonates with BSI compared with those without BSI (31% vs 10%, P < .001). ELBW neonates had an overall mortality of 57%. Among ELBW neonates with BSI (n = 8), 75% died, whereas 56% of ELBW neonates without BSI died (P = .287; Figure 3A). Among VLBW and LBW neonates, mortality was higher among neonates with BSI (VLBW neonates: 44% vs 16%, P = .020; LBW neonates: 26% vs 6%, P < .001). Among neonates with birth weight of 2500 g or greater, all-cause mortality was 4%; there were no deaths among the 12 neonates in this birth weight stratum with early-onset BSI.
Figure 3.
A, All-cause mortality by birth weight strata and early-onset BSI, early cohort. All-cause mortality among early cohort neonates, expressed as proportion of neonates deceased by birth weight strata with 95% confidence intervals (CIs); CI for neonates with birth weight ≥2500 g and BSI is a 1-sided 97.5% CI. B, All-cause mortality by birth weight strata and late-onset BSI, late cohort. All-cause mortality among late cohort neonates, expressed as proportion of neonates deceased by birth weight strata with 95% CIs. Abbreviations: BSI, bloodstream infection; ELBW, extremely low birth weight; LBW, low birth weight; NS, not significant; VLBW, very low birth weight.
In the late cohort, the median length of stay was 6 days (IQR, 3–12) and 245 neonates (8%) died. The median gestational age among neonates with BSI who died was 32 weeks (IQR, 29–34). All-cause mortality was higher among neonates with late-onset BSI (23% vs 7%, P < .001). There was no difference in all-cause mortality among ELBW neonates with and without late-onset BSI (37% vs 40%; P = .789; Figure 3B). In the 3 other birth weight strata, there was an increase in mortality among neonates with late-onset BSI; VLBW mortality 22% vs 12% (P = .022), LBW mortality 19% vs 4% (P < .001), and mortality among neonates with birth weight of 2500 g or greater 23% vs 3% (P < .001).
Among key baseline clinical and demographic characteristics, pressors on admission and antibiotics on admission were associated with an increased risk of early-onset BSI in the adjusted model (Table 4). We performed a sensitivity analysis of neonates with complete data for variables of interest, which showed the same associations. Additionally, outborn status (adjusted odds ratio [aOR], 2.06; P = .040) and respiratory support on admission (aOR, 2.06; P = .039) were associated with early-onset BSI risk. Outborn status, any respiratory support on admission, central line on admission, and antibiotics on admission were associated with risk of late-onset BSI in the adjusted model (Table 5). A sensitivity analysis of neonates with complete data for variables of interest showed similar associations.
Table 4.
Risk of Early-onset Bloodstream Infection by Baseline Clinical and Demographic Characteristics Among Neonates Included in the Early Cohort
| Characteristic | Odds Ratio | 95% CI | P Value | Adjusted Odds Ratio | 95% CI | P Value |
|---|---|---|---|---|---|---|
| Male sex | 0.77 | .45–1.32 | .343 | 0.71 | .41–1.22 | .209 |
| Preterm, <37 weeks' gestation | 1.62 | .86–3.06 | .136 | 1.23 | .54–2.81 | .616 |
| Low birth weight | 1.53 | .79–2.97 | .210 | 1.31 | .54–3.17 | .545 |
| Multiple gestation | 0.58 | .21–1.60 | .289 | 0.57 | .20–1.63 | .296 |
| Outborn | 2.40 | 1.12–5.13 | .025 | 1.87 | .84–4.15 | .126 |
| Cesarean delivery | 0.69 | .40–1.22 | .203 | 0.81 | .45–1.44 | .468 |
| Positive-pressure ventilation at delivery | 2.28 | 1.31–3.96 | .004 | 1.60 | .87–2.95 | .130 |
| Any respiratory support on admission | 4.54 | 2.13–9.67 | <.001 | 2.20 | .92–5.26 | .074 |
| Central line on admission | 0.89 | .40–2.00 | .783 | 0.43 | .16–1.17 | .098 |
| Pressors on admission | 3.47 | 1.95–6.19 | <.001 | 2.45 | 1.29–4.64 | .006 |
| Antibiotics on admission | 3.69 | 2.02–6.74 | <.001 | 2.00 | 1.00–3.98 | .049 |
Logistic regression performed for select baseline characteristics in neonates included in the early cohort (n = 3341). Individual variables and their relationship with early-onset bloodstream infection (n = 55) were assessed using univariate logistic regression and multivariable logistic regression. Multiple imputation was used to account for missing data. P values <.05 were considered statistically significant and are in bold.
Abbreviation: CI, confidence interval.
Table 5.
Risk of Late-onset Bloodstream Infection by Baseline Clinical and Demographic Characteristics Among Neonates Included in the Late Cohort
| Odds Ratio | 95% CI | P Value | Adjusted Odds Ratio | 95% CI | P Value | |
|---|---|---|---|---|---|---|
| Male sex | 0.84 | .62–1.14 | .256 | 0.80 | .58–1.10 | .164 |
| Preterm, <37 weeks’ gestation | 2.83 | 1.84–4.35 | <.001 | 1.52 | .93–2.50 | .095 |
| Low birth weight | 1.80 | 1.21–2.67 | .004 | 1.21 | .71–2.07 | .483 |
| Multiple gestation | 1.35 | .87–2.09 | .185 | 1.07 | .67–1.72 | .773 |
| Outborn | 2.66 | 1.87–3.80 | <.001 | 2.65 | 1.80–3.90 | <.001 |
| Cesarean delivery | 0.96 | .71–1.31 | .802 | 1.02 | .73–1.40 | .925 |
| Positive-pressure ventilation at delivery | 1.41 | .98–2.01 | .064 | 0.85 | .58–1.26 | .422 |
| Any respiratory support on admission | 3.29 | 2.30–4.71 | <.001 | 1.70 | 1.11–2.60 | .015 |
| Central line on admission | 4.14 | 2.85–6.03 | <.001 | 2.72 | 1.77–4.19 | <.001 |
| Pressors on admission | 2.29 | 1.59–3.31 | <.001 | 1.15 | .76–1.76 | .505 |
| Antibiotics on admission | 4.35 | 3.04–6.23 | <.001 | 2.55 | 1.69–3.84 | <.001 |
Logistic regression performed for select baseline characteristics in neonates included in the late cohort (n = 3178). Individual variables and their relationship with late-onset bloodstream infection (n = 176) were assessed using univariate logistic regression and multivariable logistic regression. Multiple imputation was used to account for missing data. P values <.05 were considered statistically significant and are in bold.
Abbreviation: CI, confidence interval.
DISCUSSION
This prospective cohort study highlights the burden of neonatal sepsis among hospitalized neonates in India, a lower-middle-income country with widespread AMR and nearly a quarter of global neonatal deaths annually [1, 19]. In our cohort, 1.6% of neonates experienced early-onset BSI and 5.5% experienced late-onset BSI. BSIs were caused predominately by GNB, with 23% of BSIs caused by K. pneumoniae alone and an additional 24% caused by Acinetobacter spp. and Citrobacter spp. Alarmingly, AMR was pervasive, including to broad-spectrum antibiotics such carbapenems, limiting therapeutic options for clinicians caring for these neonates. Nearly one-third of early cohort neonates and one-quarter of late cohort neonates with BSIs died, highlighting the toll of infections in these vulnerable hospitalized neonates. It is imperative to note that the majority of neonates who developed BSIs and died in our cohort were moderate to late preterm neonates who had already survived the first 3 days of life, the highest risk period for neonatal death [2]. These are not neonates who are likely to develop life-threatening complications of prematurity such as severe respiratory distress syndrome, necrotizing enterocolitis, and intraventricular hemorrhage. Without BSIs, these neonates are likely to survive to hospital discharge and go home with their families. This cannot be overstated—the bulk of neonatal deaths due to infections in LMIC occur in neonates who would otherwise likely survive and thrive as children.
BSI incidence in our cohort is difficult to compare to incidence from most studies reporting infection rates among hospitalized neonates in LMIC, as rates are frequently reported as a function of live births, not NICU admissions [19, 24]. However, the 2016 Delhi Neonatal Infection Study, a prospective cohort study in 3 tertiary care NICUs in Delhi, India, did report an incidence of culture-positive sepsis of 6.2%, strikingly similar to our overall incidence, though this study included only inborn neonates [25]. Importantly, the majority of cases were early-onset, whereas the converse is true in our population.
In our cohort, gram-negative pathogens predominated, including Klebsiella spp., Citrobacter spp., and Acinetobacter spp. These data are markedly different from the data from most high-income settings, including the United States, where group B Streptococcus continues to be the most common cause of early-onset sepsis and CONS is responsible for the greatest proportion of late-onset sepsis cases in hospitalized neonates [9, 26, 27]. Escherichia coli is an important gram-negative neonatal pathogen in high-income settings and was the second most common early-onset sepsis pathogen in a large US prospective cohort study [26]. Only a small proportion of BSIs in our cohort were caused by E. coli; key GNB in our population included Klebsiella spp., Acinetobacter spp., and Citrobacter spp. While reports of NICU outbreaks due to these pathogens in high-income settings exist, the overall incidence of these infections is strikingly low compared with our study [12, 27–29].
In most LMIC studies, however, the pathogen distribution more closely mirrors our data. GNB, especially Acinetobacter spp. and Klebsiella spp., cause the overwhelming majority of neonatal sepsis in the NICU [19, 24, 25, 30–35]. However, important local, national, and regional differences exist, with gram-positive predominance reported in some LMIC NICUs [36–38]. Variations in pathogen distribution may reflect differences in the relative importance of potential reservoirs of transmission. While early-onset sepsis is classically thought to be due to vertical transmission and late-onset sepsis is primarily thought to be due to horizontal transmission, it is possible that these assumptions do not hold true in our cohort and in other LMIC settings. In our cohort, the overall similarity in pathogen distribution for both early- and late-onset sepsis may point to a single predominant reservoir of transmission, such as the hospital environment, with rapid colonization and subsequent infection of hospitalized neonates. However, it is also possible that maternal colonization in this setting may mirror environmental colonization and reflect an overall predominance of GNB. Such epidemiological differences should be carefully considered in designing interventions to prevent infections in hospitalized neonates in these settings.
Among gram-negative BSI cases in our cohort, AMR was pervasive, including to third-generation cephalosporins and carbapenems. Our data must be considered within the broader context of widespread AMR in South Asia [19, 23, 25, 31, 32, 39–41]. These data highlight the challenges of caring for hospitalized neonates with suspected sepsis in regions of high AMR prevalence. While antimicrobial stewardship principles should be embraced to limit use of broad-spectrum antibiotics and reduce risk of future resistant infections, antibiotic selection at the time of suspected sepsis must be guided by local knowledge of resistance patterns. All too frequently, so-called last resort antibiotics such as polymyxins are used for empiric therapy in hospitalized neonates, with limited safety and pharmacokinetic/pharmacodynamic data [42].
Factors associated with increased risk of early-onset BSI included pressors and the need for respiratory support, reflective of clinical illness on admission. Outborn neonates were at greater risk of late-onset BSI, as were neonates who required any respiratory support, central line, and antibiotics on admission. In high-income settings, invasive devices and preceding antibiotic exposure have been identified as risk factors for nosocomial infections in neonates [9]. Understanding the role of such risk factors for late-onset BSI risk in India and other LMIC is essential for the development of effective NICU-based infection prevention strategies.
Strengths of this study include its prospective nature, enrollment of all neonates admitted to the NICU, and large sample size. The following limitations should be considered. First, blood cultures were obtained at the discretion of the clinical team. It is possible that BSI cases were missed, due to failure to obtain a blood culture prior to antibiotic initiation or inadequate blood volume obtained, a common issue in neonates. Outborn neonates may have received antibiotics prior to admission, limiting potential yield of blood cultures obtained. Second, it is not possible to ascertain whether CONS isolates represented true BSI or potential contaminants. CONS is known to be a neonatal pathogen, especially in late-onset BSI [9]. While diagnosis of CONS BSI usually relies on confirmation with a second blood culture, this practice is not routinely followed in resource-limited settings. We therefore considered all blood cultures positive for CONS as true cases of BSI. Finally, we did not standardize the laboratory workup of blood cultures obtained per clinical practice. While our laboratory capacity assessments confirmed that high-quality, accredited microbiology laboratories staffed by experienced medical microbiologists with external quality assurance mechanisms were in place, site differences in laboratory workup and reporting of results are possible.
This study adds to the evidence base that, in contrast to high-income countries, BSIs among hospitalized neonates in LMIC are predominately caused by GNB and AMR. Mortality is significantly increased among neonates with BSI, with marked increases especially in moderate to late preterm and term neonates. This points to the crucial need for infection prevention to improve the chance of survival in neonates who otherwise are likely to survive and thrive. As LMIC acquire increasing capacity to provide advanced care to preterm and sick neonates, hospital-onset sepsis due to AMR GNB will be a growing threat to neonatal survival in these settings. We must consider this a call to action to describe local epidemiology of infection and associated AMR, identify sources of transmission of these pathogens, and design and implement locally appropriate and cost-effective interventions to reduce infections and death in hospitalized neonates in LMIC.
Notes
Acknowledgments. The authors thank Susan E. Coffin and W. Charles Huskins for their technical expertise in the conception of this study. The authors additionally thank Derek Armstrong for his leadership in microbiology laboratory assessments performed prior to study initiation.
Financial support. This work was supported by the US Centers for Disease Control and Prevention (Safe Healthcare, Epidemiology, and Prevention Research Development Program Domain 7, contract 200-2016-91781) and the National Institutes of Health (UM1AI104681 to M. L. R. and K24AI141580 to A. M. M.).
Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.
References
- 1. United Nations Inter-agency Group for Child Mortality Estimation (UN IGME). Levels & trends in child mortality: report 2019, estimates developed by the United Nations Inter-agency Group for Child Mortality Estimation. New York, NY: United Nations Children’s Fund, 2019. [Google Scholar]
- 2. Wardlaw T, You D, Hug L, Amouzou A, Newby H. UNICEF report: enormous progress in child survival but greater focus on newborns urgently needed. Reprod Health 2014; 11:82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Neogi SB, Malhotra S, Zodpey S, Mohan P. Assessment of special care newborn units in India. J Health Popul Nutr 2011; 29:500–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Neogi SB, Khanna R, Chauhan M, et al. Inpatient care of small and sick newborns in healthcare facilities. J Perinatol 2016; 36:18–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Lawn JE, Davidge R, Paul VK, et al. Born too soon: care for the preterm baby. Reprod Health 2013; 10 Suppl 1:S5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bhutta ZA, Das JK, Bahl R, et al. ; Lancet Newborn Interventions Review Group ; Lancet Every Newborn Study Group. Can available interventions end preventable deaths in mothers, newborn babies, and stillbirths, and at what cost? Lancet 2014; 384:347–70. [DOI] [PubMed] [Google Scholar]
- 7. World Bank. World Bank country and lending groups. Available at: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519. Accessed September 2, 2019.
- 8. International Institute for Population Sciences and ICF. National family health survey (NFHS-4), 2015–2016: India. Mumbai, India: International Institute for Population Sciences, 2017. [Google Scholar]
- 9. Shane AL, Sánchez PJ, Stoll BJ. Neonatal sepsis. Lancet 2017; 390:1770–80. [DOI] [PubMed] [Google Scholar]
- 10. Hooven TA, Polin RA. Healthcare-associated infections in the hospitalized neonate: a review. Early Hum Dev 2014; 90 Suppl 1:S4–6. [DOI] [PubMed] [Google Scholar]
- 11. Polin RA, Denson S, Brady MT; Committee on Fetus and Newborn ; Committee on Infectious Diseases. Epidemiology and diagnosis of health care-associated infections in the NICU. Pediatrics 2012; 129:e1104–9. [DOI] [PubMed] [Google Scholar]
- 12. Johnson J, Quach C. Outbreaks in the neonatal ICU: a review of the literature. Curr Opin Infect Dis 2017; 30:395–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Khajuria A, Praharaj AK, Kumar M, Grover N, Aggarwal A. Multidrug resistant NDM-1 metallo-beta-lactamase producing Klebsiella pneumoniae sepsis outbreak in a neonatal intensive care unit in a tertiary care center at central India. Indian J Pathol Microbiol 2014; 57:65–8. [DOI] [PubMed] [Google Scholar]
- 14. Flores-Carrero A, Labrador I, Paniz-Mondolfi A, Peaper DR, Towle D, Araque M. Nosocomial outbreak of extended-spectrum β-lactamase-producing Enterobacter ludwigii co-harbouring CTX-M-8, SHV-12 and TEM-15 in a neonatal intensive care unit in Venezuela. J Glob Antimicrob Resist 2016; 7:114–8. [DOI] [PubMed] [Google Scholar]
- 15. Rohit A, Suresh Kumar D, Dhinakaran I, et al. Whole-genome-based analysis reveals multiclone Serratia marcescens outbreaks in a non-neonatal intensive care unit setting in a tertiary care hospital in India. J Med Microbiol 2019; 68:616–21. [DOI] [PubMed] [Google Scholar]
- 16. Chen D, Hu X, Chen F, et al. Co-outbreak of multidrug resistance and a novel ST3006 Klebsiella pneumoniae in a neonatal intensive care unit: a retrospective study. Medicine (Baltimore) 2019; 98:e14285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Farzana R, Jones LS, Rahman MA, et al. Outbreak of hypervirulent multi-drug resistant Klebsiella variicola causing high mortality in neonates in Bangladesh. Clin Infect Dis 2018; 68:1225–7. [DOI] [PubMed] [Google Scholar]
- 18. Ulu-Kilic A, Gundogdu A, Cevahir F, Kilic H, Gunes T, Alp E. An outbreak of bloodstream infection due to extensively resistant Acinetobacter baumannii among neonates. Am J Infect Control 2018; 46:154–8. [DOI] [PubMed] [Google Scholar]
- 19. Chaurasia S, Sivanandan S, Agarwal R, Ellis S, Sharland M, Sankar MJ. Neonatal sepsis in South Asia: huge burden and spiralling antimicrobial resistance. BMJ 2019; 364:k5314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Tsai MH, Chu SM, Hsu JF, et al. Risk factors and outcomes for multidrug-resistant gram-negative bacteremia in the NICU. Pediatrics 2014; 133:e322–9. [DOI] [PubMed] [Google Scholar]
- 21. Hsu AJ, Tamma PD. Treatment of multidrug-resistant gram-negative infections in children. Clin Infect Dis 2014; 58:1439–48. [DOI] [PubMed] [Google Scholar]
- 22. Laxminarayan R, Matsoso P, Pant S, et al. Access to effective antimicrobials: a worldwide challenge. Lancet 2016; 387:168–75. [DOI] [PubMed] [Google Scholar]
- 23. Gandra S, Joshi J, Trett A, Lamkang AS, Laxminarayan R. Scoping report on antimicrobial resistance in India. Washington, DC: Center for Disease Dynamics, Economics & Policy, 2017. [Google Scholar]
- 24. Zaidi AK, Huskins WC, Thaver D, Bhutta ZA, Abbas Z, Goldmann DA. Hospital-acquired neonatal infections in developing countries. Lancet 2005; 365:1175–88. [DOI] [PubMed] [Google Scholar]
- 25. Investigators of the Delhi Neonatal Infection Study Collaboration. Characterisation and antimicrobial resistance of sepsis pathogens in neonates born in tertiary care centres in Delhi, India: a cohort study. Lancet Glob Health 2016; 4:e752-60. [DOI] [PubMed] [Google Scholar]
- 26. Stoll BJ, Hansen NI, Sanchez PJ, et al. Early onset neonatal sepsis: the burden of group B Streptococcal and E. coli disease continues. Pediatrics 2011; 127:817–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Greenberg RG, Kandefer S, Do BT, et al. Late-onset sepsis in extremely premature infants: 2000–2011. Pediatr Infect Dis 2017; 36:774–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Rønning TG, Aas CG, Støen R, et al. Investigation of an outbreak caused by antibiotic-susceptible Klebsiella oxytoca in a neonatal intensive care unit in Norway. Acta Paediatr 2019; 108:76–82. [DOI] [PubMed] [Google Scholar]
- 29. McGrath EJ, Chopra T, Abdel-Haq N, et al. An outbreak of carbapenem-resistant Acinetobacter baumannii infection in a neonatal intensive care unit: investigation and control. Infect Control Hosp Epidemiol 2011; 32:34–41. [DOI] [PubMed] [Google Scholar]
- 30. Viswanathan R, Singh AK, Basu S, Chatterjee S, Sardar S, Isaacs D. Multi-drug resistant gram negative bacilli causing early neonatal sepsis in India. Arch Dis Child Fetal Neonatal Ed 2012; 97:F182–7. [DOI] [PubMed] [Google Scholar]
- 31. Kabwe M, Tembo J, Chilukutu L, et al. Etiology, antibiotic resistance and risk factors for neonatal sepsis in a large referral center in Zambia. Pediatr Infect Dis J 2016; 35:e191–8. [DOI] [PubMed] [Google Scholar]
- 32. Litzow JM, Gill CJ, Mantaring JB, et al. High frequency of multidrug-resistant gram-negative rods in 2 neonatal intensive care units in the Philippines. Infect Control Hosp Epidemiol 2009; 30:543–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Muley VA, Ghadage DP, Bhore AV. Bacteriological profile of neonatal septicemia in a tertiary care hospital from western India. J Global Infect Dis 2015; 7:75– 7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Viswanathan R, Singh AK, Mukherjee S, Mukherjee R, Das P, Basu S. Aetiology and antimicrobial resistance of neonatal sepsis at a tertiary care centre in eastern India: a 3 year study. Indian J Pediatr 2011; 78:409– 12. [DOI] [PubMed] [Google Scholar]
- 35. Pokhrel B, Koirala T, Shah G, Joshi S, Baral P. Bacteriological profile and antibiotic susceptibility of neonatal sepsis in neonatal intensive care unit of a tertiary hospital in Nepal. BMC Pediatr 2018; 18:208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Sorsa A, Fruh J, Stotter L, Abdissa S. Blood culture result profile and antimicrobial resistance pattern: a report from neonatal intensive care unit (NICU), Asella teaching and referral hospital, Asella, south East Ethiopia. Antimicrob Resist Infect Control 2019; 8:42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Aku FY, Akweongo P, Nyarko K, et al. Bacteriological profile and antibiotic susceptibility pattern of common isolates of neonatal sepsis, Ho Municipality, Ghana-2016. Matern Health Neonatol Perinatol 2018; 4:2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Kohli-Kochhar R, Omuse G, Revathi G. A ten-year review of neonatal bloodstream infections in a tertiary private hospital in Kenya. J Infect Dev Ctries 2011; 5:799– 803. [DOI] [PubMed] [Google Scholar]
- 39. Thakur S, Thakur K, Sood A, Chaudhary S. Bacteriological profile and antibiotic sensitivity pattern of neonatal septicaemia in a rural tertiary care hospital in North India. Indian J Med Microbiol 2016; 34:67– 71. [DOI] [PubMed] [Google Scholar]
- 40. Jain S, Gaind R, Kothari C, et al. VEB-1 extended-spectrum beta-lactamase-producing multidrug-resistant Proteus mirabilis sepsis outbreak in a neonatal intensive care unit in India: clinical and diagnostic implications. JMM Case Rep 2016; 3:e005056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Jajoo M, Manchanda V, Chaurasia S, et al. Alarming rates of antimicrobial resistance and fungal sepsis in outborn neonates in North India. PloS One 2018; 13:e0180705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Thomas R, Velaphi S, Ellis S, et al. The use of polymyxins to treat carbapenem resistant infections in neonates and children. Expert Opin Pharmacother 2019; 20:415– 22. [DOI] [PubMed] [Google Scholar]