Association between positive urine cultures and necrotizing enterocolitis in a large cohort of hospitalized infants

Abstract

Objective

We used a large research database to examine the association between urinary tract infections and necrotizing enterocolitis (NEC) in premature infants.

Methods

This retrospective data analysis included infants ≤32 weeks gestational age and ≤1500 g at birth who had urine cultures obtained at one of 322 neonatal intensive care units managed by the Pediatrix Medical Group from 1997–2012. The primary outcome was a diagnosis of NEC within 7 days after urine culture. We used multivariable conditional logistic regression conditioned on postnatal age and controlling for gestational age, inotropic support on the day of culture, and mechanical ventilation on the day of culture to evaluate the association between urine culture result and NEC.

Results

We identified 25,816 infants who had 43,556 urine cultures obtained; 6586 (15.1%) of the cultures were positive. A diagnosis of NEC within 7 days after culture was made in 334 (5.1%) of the 6586 positive cultures versus 1582 (4.3%) of the 36,970 negative cultures (p<0.01). On multivariable analysis, infants with any positive urine culture had increased risk of NEC (odds ratio [OR] 1.16, 95% confidence interval [CI] 1.02–1.31); the risk was higher when limited to Gram-negative organisms (OR 1.37, 95% CI 1.17–1.59). The risk of surgical NEC was increased in infants with any positive urine culture (OR 1.46, 95% CI 1.18–1.81) and was also higher when limited to Gram-negative organisms (OR 1.99, 95% CI 1.53–2.59).

Conclusion

Positive urine cultures were associated with increased risk of NEC within 7 days of culture.

1. Introduction

Necrotizing enterocolitis (NEC) occurs in 5% to 11% of very low birth weight (VLBW; <1500 g) infants [1,2]. NEC likely results from a combination of several factors, including genetic predisposition, intestinal immaturity, excessive intestinal inflammatory response, and inappropriate microbial colonization [3,4]. NEC mortality ranges from 16% to 42% [5]. Survivors are at high risk of severe growth delay and poor neurodevelopmental outcomes [6,7].

NEC has previously been associated with preceding gastrointestinal, respiratory, and bloodstream infections [8–10]. Outbreaks of NEC have been described, suggesting a transmissible infectious factor. Despite these reports, no single pathogen has been identified. A shift in the intestinal microbial ecology is thought to contribute to NEC risk in premature infants. This shift is preceded and marked by a loss of overall diversity and an increase in the proportion of bacterial families such as the Enterobacteriaceae [11–13]. Enterobacteriaceae includes Escherichia coli, Klebsiella spp., and Enterobacter spp., which are among the most commonly isolated organisms in urinary tract infections (UTIs) affecting infants in the neonatal intensive care unit (NICU) [14,15]. We examined the association between UTIs and NEC using a large research database.

2. Methods

2.1. Study cohort

We identified a cohort of all infants discharged from 322 NICUs managed by the Pediatrix Medical Group from 1997–2012. The data were obtained from an electronic medical record that prospectively captured data from admission notes, daily progress notes, procedure notes, and discharge summaries. Information was collected regarding maternal history and demographics, medications, laboratory results, culture results, and diagnoses.

We included all infants who were ≤32 weeks gestational age and weighed ≤1500 g at birth and who had urine cultures obtained via supra-pubic tap, in-and-out catheterization, or bag in the first 120 days of life. We excluded infants whose only urine cultures were obtained after developing NEC, as well as those with congenital gastrointestinal or urogenital defects. All urine cultures obtained prior to a diagnosis of NEC or 120 days of life were included in the analysis (Figure). We included definite and probable episodes of coagulase-negative staphylococci (CoNS) infection as defined using previously reported criteria: a definite CoNS infection was defined as two positive cultures on the same day; a probable CoNS infection as two positive cultures within a 4-day period, three positive cultures within a 7-day period, or four positive cultures within a 10-day period; and a possible CoNS infection as a culture positive for CoNS that did not meet criteria for definite or probable CoNS sepsis [16–18]. The diagnosis and severity of NEC were assigned at each site by the attending neonatologist and included either medical NEC or surgical NEC. NEC severity was defined on the first day of the course of NEC regardless of change in severity thereafter.

Figure.

Patient flow chart. GA, gestational age; BW, birth weight; GI, gastrointestinal; GU, genitourinary; NEC, necrotizing enterocolitis; DOL, day of life.

2.2. Statistical analysis

The unit of observation for this study was a urine culture. The primary outcome was a diagnosis of NEC occurring within 7 days after the urine culture was obtained. Infant-level categorical variables were compared between patients with and without NEC following the urine culture using chi-square tests and Wilcoxon rank sum tests. We used multivariable conditional logistic regression conditioned on postnatal age in days at the time of urine culture to evaluate the association between urine culture and NEC. In the final model, we included the following covariates: gestational age, inotropic support on the day of culture, and mechanical ventilation on the day of culture. All analyses were performed using Stata 12 (College Station, TX) and assumed a significance limit of α = 0.05. The study was approved by the Duke University Institutional Review Board without the need for written informed consent as the data were collected without identifiers.

3. Results

We identified 25,816 infants with 43,556 urine cultures. The median gestational age and birth weight were 27 weeks (interquartile range [IQR] 25–29) and 915 g (IQR 720–1154), respectively. Infants who developed NEC had a lower median birth weight compared to those who did not develop NEC (880 g, IQR 700–1106, vs. 920 g, IQR 725–1160, respectively; p<0.001) (Table 1).

Table 1.

Demographics of infants with positive urine cultures.

		NEC N=3784	No NEC N=22,032	P
Gestational age (weeks)				<0.001
<24		5	4
24–27		55	51
≥28		40	45
Birth weight (g)				<0.001
<750		31	28
750–999		33	30
≥1000		36	42
Male		57	55	0.027
Inborn		76	80	<0.001
Cesarean section		71	72	0.345
Race/ethnicity				<0.001
White		36	41
Black		28	24
Hispanic		31	30
Other		5	5

NEC, necrotizing enterocolitis.

There were 6586/43,556 (15.1%) positive urine cultures in 5675 infants. Among those with positive urine cultures, the median postnatal age at the time of urine culture was 16 days (IQR 7–30) in those who did not develop NEC and 19 days (IQR 12–29) in those who developed NEC within 7 days. The organism most commonly cultured from the urine was Escherichia coli (1079/6586, 16.4%), followed by Klebsiella (940/6586, 14.3%) and Enterococcus (917/6586, 13.9%). Among infants with positive urine cultures who subsequently developed NEC within 7 days (N = 334), the most common organism cultured from the urine was Klebsiella (62/334, 18.6%) (Table 2). Infants with positive urine cultures who did not develop NEC (N = 6252) most commonly had urine cultures growing Escherichia coli (1028/6252, 16.4%). We identified 1916 cases of NEC within 7 days of a urine culture. NEC was diagnosed more frequently within 7 days of a positive culture than a negative culture (334/6586 [5.1%] vs. 1582/36,970 [4.3%], p<0.01). Of the 334 NEC cases following a positive urine culture, 108 (32.3%) were surgical NEC, and 226 (67.7%) were medical NEC. Using multivariable analysis, odds of NEC were increased in infants with any positive urine culture (OR 1.16, 95% CI 1.02–1.31), and the odds were higher when limited to urine cultures with Gram-negative organisms (OR 1.37, 95% CI 1.17–1.59). Odds of surgical NEC were increased in infants with any positive urine culture (OR 1.46, 95% CI 1.18–1.81), and the odds were higher when limited to urine cultures with Gram-negative organisms (OR 1.99, 95% CI 1.53–2.59). We did not observe an association between NEC and Gram-positive urinary tract infections (OR 1.00, 95% CI 0.80–1.27) (Table 3).

Table 2.

Organisms cultured from urine and risk of NEC within 7 days.

Organisms	NEC N=334	No NEC N=6252
Gram-positive cocci
CoNS	8 (2%)	244 (4%)
Enterococcus	48 (14%)	869 (14%)
Group B Streptococcus	4(1%)	133 (2%)
Staphylococcus aureus	7 (2%)	147 (2%)
Other	12 (4%)	179 (3%)
Gram-negative rods
Citrobacter	5 (2%)	128 (2%)
Escherichia coli	51 (15%)	1028 (16%)
Enterobacter	48 (14%)	717 (11%)
Klebsiella	62 (19%)	878 (14%)
Proteus	5 (2%)	128 (2%)
Pseudomonas	2 (1%)	177 (3%)
Serratia	10 (3%)	176 (3%)
Other	20 (6%)	222 (4%)
Candida	35 (10%)	808 (13%)

NEC, necrotizing enterocolitis; CoNS, coagulase-negative staphylococci.

Table 3.

Adjusted^a odds ratios (95% confidence intervals) of NEC for infants with positive urine culture within 7 days prior from conditional logistic regression conditioned on postnatal age in days.

	Medical NEC	Surgical NEC	Any NEC
Any organism	1.04 (0.90–1.21)	1.46 (1.18–1.81)	1.16 (1.02–1.31)
Gram-positive organism	0.98 (0.75–1.28)	1.10 (0.70–1.71)	1.00 (0.80–1.27)
Gram-negative organism	1.15 (0.95–1.39)	1.99 (1.53–2.59)	1.37 (1.17–1.59)

^aAdjusted for gestational age and inotrope and ventilator support on the day of culture.

NEC, necrotizing enterocolitis.

Bold indicates statistically significant.

4. Discussion

We identified UTI as a risk factor in the development of NEC. An NEC diagnosis within 7 days was made more frequently after positive than negative urine cultures, although the absolute difference was small. Compared to infants with negative urine cultures, infants with positive urine cultures had an increased risk of NEC, especially surgical NEC, and the risk was higher when limited to Gram-negative organisms.

Although not specifically studied in the premature infant, colonization and infection of the urinary tract is generally regarded as an ascending process whereby organisms originate from the intestinal microbiota and ultimately colonize the peri-urethral region [19,20]. Thus urine cultures may serve as sentinels for intestinal dysbiosis, reflecting shifts in the intestinal and peri-urethral microbiota. Non-catheterized urine collections, in particular, may reflect colonization of the peri-urethral niche and thus best reflect changes in the intestinal microbiota. We were unable to differentiate between specimens collected via bag or in-and-out catheterization.

Although there are very limited data relating UTIs to gastrointestinal pathology in hospitalized infants [21], other serious infections have been linked with NEC. There are reports of NEC following infection as well as decreased incidence of NEC with infection control measures [22]. In a 1-month outbreak of 18 NEC cases, a preceding cluster of gastroenteritis cases was observed, but no particular causative organism identified [8]. Early colonization with Clostridium was also shown in 12 VLBW infants who later developed NEC [23]. Coagulase-negative staphylococci, particularly those that produced a delta-like toxin, have also been associated with NEC [24]. Another possible risk factor for NEC in VLBW infants is Ureaplasma colonization of the respiratory tract, which may also suggest colonization elsewhere including the gastrointestinal tract [9]. Furthermore, after infection prevention and control measures were enhanced to manage a Staphylococcus aureus outbreak, there was a decrease in the incidence of NEC from 16% to 3% in 282 VLBW infants [25].

Although multiple studies have associated specific organisms with NEC, these organisms may represent markers of a shift in the intestinal microbiota to a dysbiotic state, and as such, an alteration of the pro- and anti-inflammatory constituents of the microbiota in the developing intestinal tract. Molecular analysis of the intestinal microbiota preceding NEC has revealed reduced overall diversity and a shift in the microbiota, often characterized by increased proportions of the Enterobacteriacea [11–13]. Previous studies have demonstrated that the pathogen recognition receptor, toll-like receptor 4 (TLR4), recognizing Gram-negative bacteria lipopolysaccharide, is critical for inflammation, enterocyte death, and impaired mucosal repair in animal models of NEC [26]. It has been postulated that shifts toward the Enterobacteriacea, as observed in studies of the microbiota prior to NEC, may overstimulate proinflammatory pathways [27].

Despite these reports, no single bacterial pathogen has been consistently linked with the development of NEC. However, the emergence of new sequencing techniques has allowed for detection of shifts in the premature infant intestinal microbiome, with an increase in Proteobacteria in 9 premature infants with NEC compared to 9 premature infants without NEC [12]. In our study, the organism most commonly cultured from the urine was Escherichia coli, which, along with Klebsiella pneumoniae and Enterobacter cloacae, was also one of the most common organisms implicated in a review of 17 NEC epidemics that affected both preterm and term infants [22]. In our study, Klebsiella was the most commonly isolated organism in infants who subsequently developed NEC, and Escherichia coli was the most commonly isolated organism in infants who did not develop NEC.

We found that infants with positive urine cultures for Gram-negative organisms had increased risk of NEC. In a previous study, of 478 urine cultures positive for Gram-negative organisms, 34 (7%) were shown to have concordant blood cultures [28]. Furthermore, in infants with late-onset bloodstream infections (>48 hours after birth) caused by enteric Gram-negative bacilli, 14 of 32 infants (44%) infected with Escherichia coli and 1 of 31 infants (3%) infected with Klebsiella pneumoniae also had an episode of NEC [10].

A strength of this study is its large sample size derived from 322 NICUs that represented diverse patient populations. A limitation is that we included only the covariates of gestational age, inotropic support, and mechanical ventilation; there are likely other comorbidities or interventions such as human milk versus formula feeding that are associated with NEC risk which cannot be controlled for in this study [29]. Also, the non-catheterized urine specimens included in the analysis might have contributed to false-positive urine cultures. Furthermore, true UTIs could not be confirmed in the absence of additional urinary biochemical or cellular markers to complement the culture data. It is conceivable that infants treated with antibiotics for a false-positive non-catheterized urine culture might actually have developed NEC due to antibiotic exposure, which has been shown to increase the probability of NEC by approximately 20% per day of exposure in infants without prior sepsis [30].

In summary, we found that infants with positive urine cultures had increased risk of NEC, especially surgical NEC, and the risk was higher when limited to Gram-negative organisms. Positive urine cultures with Klebsiella were more common in infants who subsequently developed NEC, whereas positive urine cultures with Escherichia coli were more common in infants who did not develop NEC.

Highlights.

• Urinary tract infection may be a risk factor for necrotizing enterocolitis (NEC).

• Infants with positive urine cultures had increased NEC, especially surgical NEC.

• Risk of NEC was higher when urine cultures were positive for Gram negative organisms.

Abbreviations

CoNS

coagulase-negative staphylococci

NEC

necrotizing enterocolitis

NICU

neonatal intensive care unit

UTIs

urinary tract infections

VLBW

very low birth weight