Routine fecal occult blood testing does not predict necrotizing enterocolitis in very low birth weight neonates

Abstract

OBJECTIVE: To determine sensitivity, specificity, predictive value of routine fecal occult blood (FOB) testing on the identification of Bell’s Stage II or III necrotizing enterocolitis (NEC) in very low birth weight (VLBW) infants.

METHODS: Retrospective medical record review of VLBW infants from 2012– 2013 evaluating FOB results and clinical and demographic risk factors. We determined predictive values of positive FOB testing within 48 hours of definite NEC diagnosis. We performed logistic regression analyses for predictors of NEC and for predictors of having positive FOB during NICU admission.

RESULTS: The incidence of NEC in our cohort of 203 infants was 3.9% (n = 8). None had positive FOB results within 48 hours of diagnosis, and only 12.5% had any positive FOB within 7 days. Sensitivity of positive FOB for predicting definite NEC = 0%, specificity = 34.4%, and positive predictive value = 0%. A majority of VLBWs (67.0%) had > one positive FOB result during their NICU course. On logistic regression, intrauterine growth restricted (IUGR) infants had significantly higher odds of both developing NEC and of having positive FOB. Positive FOB was not a significant predictor of NEC. Those with lower birth gestational ages had higher odds of positive FOB.

CONCLUSIONS: Positive FOB testing occurred in a majority of VLBW infants, with higher odds in the more preterm and IUGR. However, the sensitivity, specificity, and predictive value of routine FOB testing for identifying NEC were all very poor. Our data demonstrates that this test offers no advantages in the early diagnosis of NEC.

Keywords

FOB testing guaiac testing fecal occult blood testing necrotizing enterocolitis NEC very low birth weight prematurity

1 Background

Despite a trend towards decreased overall morbidity and mortality in very low birth weight (VLBW, <1500 grams [g]) infants over the past decade, rates of necrotizing enterocolitis (NEC) in this population remain 5– 6% [1]. An incomplete understanding of the pathophysiology has left providers with fewstrategies for early diagnosis. Routine fecal occult blood (FOB) testing has been adopted by many NICUs based on the hypothesis that intestinal injury and inflammation associated with NEC would lead to occult blood in the feces [2]. This hypothesis has been applied in adult medicine where FOB has been used as a screening test for colorectal cancer due to the associated intestinal mucosal injury [3].

Abramo et al. [2] demonstrated poor sensitivity, specificity, and predictive values of FOB testing on NEC diagnosis in a cohort born <1800 g. However, over 25 years later, routine FOB testing is still being performed. It is unclear whether this is achieving the goal of early detection, or is leading to false-positives and unnecessary medical work-ups. More evidence is needed to help guide clinicians in the use of FOB testing for VLBW infants.

The objective of this study was to evaluate the sensitivity, specificity, and predictive value of routine FOB testing on the identification of VLBW infants with NEC. We evaluated predictors of NEC and of positive FOB in a large VLBW cohort. We hypothesized that a majority of VLBW infants would have positive FOB but this test would have poor sensitivity for predicting NEC.

2 Methods

We performed a retrospective medical record review of a convenience sample of VLBW infants born over a two-year period, January 1, 2012 to December 31, 2013, and admitted the University of Maryland Children’s Hospital NICU in Baltimore, MD. Our Institutional Review Board approved this study. Inclusion criteria included: 1) birth gestational age (GA) > = 24 weeks; 2) birth weight <1500 g. Exclusion criteria included: 1) transferred to our NICU at >one week of age; 2) died prior to onset of FOB testing; 3) no available fecal testing; 4) NEC or spontaneous intestinal perforation diagnosis prior to FOB testing.

Medical records were reviewed for primary outcome of diagnosis of definite NEC [4] as well as clinical and demographic risk factors listed in Tables 1 and 2. We defined definite NEC as Stage II and/or III based on modified Bell’s criteria [4]. Medical records were reviewed to confirm proper staging of NEC, including daily progress notes, X-ray reports, surgical consultation notes, operative reports. We noted dates of FOB testing and results. FOB testing was performed at our institution usingHemoccult Brand (Beckman Coulter, Inc., Brea, CA), whose product insert notes a 1% false-positive rate. One sample was sent from each feces collected, which was placed immediately by a trained NICU nurse onto the guaiac card. The card with the sample was taken to our inpatient laboratory where FOB testing was performed by lab technicians trained in assessing appropriateness of the sample (including quantity of feces and that the sample was appropriately dry for testing), applying developer, and interpreting FOB results. Nurses obtaining the specimens were aware of demographics and clinical concerns in the infants, lab technicians performing the test were not.

2.1 Statistical analysis

We constructed a 2 × 2 table comparing FOB results vs. diagnosis of NEC. We calculated sensitivity, specificity, and predictive value of positive FOB within 48 hours of NEC diagnosis. We compared demographic and clinical risk factors between infants with and without NEC diagnosis as well between infants with and without positive FOB test result using T-Test, Wilcoxon Rank-Sum Test, Chi-square, or Fisher-Exact testing as appropriate for continuous, binary, and categorical variables. We performed univariate and multivariable logistic regression analyses to assess for significant predictors of NEC and significant predictors of positive FOB. Effects are reported as odds ratios (OR) and 95% confidence intervals (CIs). We retained those predictors with p < = 0.05 and subsequently added back each variable to assess for confounding, retaining those variables noted to be confounders. Collinearity was assessed by evalua-ting standard error and p-values of covariables. All statistical analyses were performed using SAS 9.3 (SAS Institute, Carey, NC).

3 Results

3.1 Predictors of NEC

We identified 273 neonates who met inclusion criteria (Fig. 1). A total of 70 met exclusion criteria due to being admitted after one week of age (n = 35), death prior to FOB testing (n = 30), no FOB results available (n = 4), or being admitted with NEC or bowel perforation (n = 1). Of the 203 eligible infants, 3.9% (n = 8) were diagnosed with Stage II/ III NEC and appropriate staging based on modified Bell’s criteria was confirmed upon medical record review [4].

Demographic and clinical risk factors in those with and without diagnosis of NEC are shown in Table 1. In our unadjusted model, the only significant predictor of developing NEC was intrauterine growth restriction (IUGR, <10th percentile for GA). All other demographic and clinical factors tested were not statistically significant. Specifically, having positive FOB result was not significantly different.

Regression analysis of predictors of NEC in our cohort demonstrated that when controlling for confounder of birth GA, IUGR remained the only significant predictor of the development of NEC with OR = 14.2 [95% CI 2.5– 79.5]. Infants with IUGR had 14x the odds of developing NEC than those who were not IUGR. Positive FOB was not a significant predictor of NEC in our model.

3.2 FOB testing and NEC

No infants had a positive FOB result within 48 hours of diagnosis. Only one infant (12.5%) with NEC had a positive FOB result during the week prior to diagnosis. Of the 195 infants who did not develop NEC, 65.6% (n = 128) had at least one positive FOB result. Positive FOB within 48 hours of diagnosis was found to have a sensitivity = 0% [95% CI 0 – 36.9%] and specificity = 34.4% [95% CI 27.7 – 41.5%] for predicting definite NEC. The predictive value of a positive FOB result for identify those diagnosed with NEC = 0% [95% CI 0 – 2.8%]. To calculate these values, we included any positive FOB result for those without NEC and any positive FOB result within 48 hours of diagnosis for those with NEC. One infant diagnosed with NEC was excluded as they did not have any feces tested within 48 hours of diagnosis.

In addition to the infants diagnosed with Stage II/III NEC, there were 29 cases (14.3% of eligible infants) of suspected or Stage I NEC noted on medical record review, defined using modified Bell’s criteria as “temperature instability, apnea, bradycardia, residuals, mild abdominal distension, normal motility or mild ileus, with occult positive stools”[4]. None of these infants had pneumatosis or free air on X-ray and none met criteria for Stage II/III NEC. These infants were therefore not included in analysis of Stage II/III NEC. Twenty-four percent of the Stage I cases (n = 7) had at least one positive FOB result in the week preceding their diagnosis. Five of these infants (17.2%) were diagnosed clinically by their medical team with milk-protein allergy and restarted on hydrolyzed formula feeds while the remaining were restarted on breast-milk or cow’s-milk formula. Of note, none of these infants went on to develop strictures, definite NEC, or to require any gastrointestinal surgery during their NICU stay.

3.3 Predictors of positive FOB

Demographic and clinical predictors of having ≥ one positive FOB result are shown in Table 2. In our unadjusted model, those with positive FOB had significantly lower birth weights, birth GA, and Apgar scores. Infants with positive FOB were also significantly more likely to receive mechanical ventilation, caffeine, post-natal steroids, surfactant, umbilical lines, and to have been diagnosed with a PDA (patent ductus arteriosus). In terms of feeding course, infants with at least one positive FOB result had later initiation of feeds, were older at the time of caloric supplementation and at attainment of full feeds. Interestingly, infants with positive FOB had a higher likelihood of feeds being initiated with breast-milk and lower likelihood of initiation with formula or both breast-milk and formula than those with all negative FOB testing. There was a trend toward increased risk of positive FOB in those infants who were IUGR, but this was not statistically significant (p = 0.07).

Of the infants with at least one positive FOB result, the first positive result was a median of 11 days from the initiation of enteral feeds. The first positive result was a median of four days before fortification to 22 calories, seven days before fortification to 24 calories, and ten days before full feeds were reached.

On logistic regression evaluating risk factors for positive FOB result, when controlling for birth weight, both birth GA and IUGR status were significant predictors (Table 3). Infants who were IUGR had 3.4x the odds of positive FOB during their NICU stay. And for every additional day of birth GA, the odds of having a positive FOB result decreased by 0.07, in other words, for every day that an infant is more premature, the odds of positive FOB increases by 1.07. All other factors, including Stage II/III NEC diagnosis, were not significant predictors and did not affect the model.

4 Discussion

NEC continues to be a devastating disease process for VLBW infants in the NICU. The lack of a clear understanding of the pathophysiology contributes to the challenge in prediction, early diagnosis, or prevention of NEC cases. FOB testing is used as a screening test, but it is unclear whether this truly identifies infants at risk of NEC or leads to false-positives and unnecessary medical work-ups. In this retrospective medical record review, we evaluated a large cohort of VLBW infants over a two-year period to determine the efficacy of routine FOBtesting on early identification of Stage II or higher NEC. We found a 3.9% incidence of definite NEC in our cohort, of which only one had documented positive FOB within seven days of NEC diagnosis. None had documented positive FOB within 48 hours of NEC diagnosis. The sensitivity, specificity, and positive predictive value of a positive FOB test are very poor (0%, 34.4%, and 0% respectively) for identifying those who went on to develop NEC. We found that positive FOB results are very common in our VLBW cohort, with 67% of infants having at least one occult positive feces during their admission. When controlling for confounding variables, IUGR was a risk factor that was associated with both having positive FOB as well as with developing definite NEC.

These results are similar to those of Abramo et al. [2] who performed a prospective blinded study testing for occult hematochezia within the first six weeks of life in 95 neonates with birth weights <1800 g who were receiving at least some enteral feeds. Five of six infants (83.3%) diagnosed with NEC (as defined by clinical signs and presence of pneumatosis intestinalis on X-ray) had negative FOB testing the day before NEC developed (the 6th infant did not have any fecal testing results prior to NEC diagnosis), and none of the five had any positive results within the two days prior to diagnosis. Positive FOB within 48 hours of diagnosis had a sensitivity and positive predictive value of 0% for predicting NEC. Fifty-eight percent of their infants had one or more positive FOB results during the first six weeks of life, and the incidence of NEC was higher in those infants with negative FOB (9.8%) than those with positive FOB (3.7%). They concluded that occult hematochezia was frequent in this population and did not seem to have a relationship to NEC.

Since histopathology of intestinal tissue affected by NEC shows ischemia, inflammation, and necrosis, the detection of microscopic blood in the feces seems plausible as a tool for early detection. However, in studies looking at preterm infants <30 weeks GA, positive FOB is uncommon as part of their presentation with NEC [5, 6]. The most common location of NEC onset is proximal to the ileocecal valve, and when a subsequent ileus occurs due to inflamed and irritated bowel, blood may fail to pass through the colon [6]. After 30 weeks birth GA, bloody feces become more common as a presenting feature of NEC, indicating that the findings for more preterm infants may not be generalizable to a more mature group of neonates [5, 6]. There does appear to be a link between gross rectal bleeding and NEC. Though they did not specify the exact number, Abramo et al. did note gross rectal bleeding was seen on the day of NEC diagnosis in some infants [2]. Other studies have noted a low incidence of NEC in those premature infants with only microscopic blood detected in feces (0.47%), but a dramatic increased incidence in those who presented-with or progressed-to grossly bloody feces (63.3% and 89.6% respectively) [7]. Two of the eight infants diagnosed with NEC (25%) in our study were noted to have grossly bloody feces on the day of diagnosis.

Significant predictors for developing NEC are important to understand in an effort to identify those at risk and help focus testing efforts. Many studies have shown that prematurity is one of the major risk factors [8, 9] as is feeding with formula rather than human milk [10]. We did not find that birth GA was a significant predictor, but this is likely because we were looking at VLBW infants who were all premature. We did find, similar to previous studies [11, 12], that IUGR status was a significant predictor of NEC even when correcting for birth GA. This is likely due to in-utero oxygen deprivation. Fetal hypoxia leads to a protective circulatory redistribution towards the brain, heart, adrenals and away from the mesenteric circulation [13]. Chronic fetal hypoxia leads to poor somatic growth (IUGR) and may result in direct hypoxic injury to the intestine as well as interfering with normal development and function. Abnormal development plus hypoxic injury may further exacerbate the effects of organ immaturity in premature infants and lead to increased risk of NEC in the IUGR preterm population.

In addition to NEC risk factors, we evaluated risk factors for having positive FOB test results. Since FOB did not seem to be a predictor of NEC, we wanted to evaluate what risk factors were associated with FOB. In our multivariable model, when controlling for BW, we found the significant predictors of positive FOB were lower birth GA and IUGR status. This makes clinical sense as many of the significant differences on our univariate analysis between those infants with and without positive FOB could be explained by lower GA. The one factor that seemed to be significant for both neonates who developed NEC and those who had positive FOB testing was IUGR status. This might explain the observations between positive FOB and clinical concern for NEC. Perhaps it is not that positive FOB testing predicts NEC, but that IUGR neonates are the ones at highest risk of positive FOB and are also at the highest risk of developing NEC.

One limitation of FOB testing for identification of intestinal pathology such as NEC is that there are many unrelated factors in VLBW neonates that can cause positive FOB results. Examples include swallowed maternal blood (at the time of delivery or from breast-milk expressed from crackled nipples), swallowed neonatal blood (trauma related to care interventions such as naso-/oro-gastric feeding tubes, endotracheal intubations, suctioning), and trauma related to glycerin suppositories [14]. Due to the lack of evidence of benefit from routineFOB testing on the early diagnosis of NEC, Pinheiro et al. discontinued this practice in their NICU and did not find an increase in incidence of NEC [15]. However, they do note that further objective evidence about the outcomes of routine FOB testing isneeded.

We opted to define our outcome of NEC using modified Bell’s criteria Stage II/III [4] because Bell’s Stage I lacks specificity as a diagnosis. As pointed out by Gordon, et al., the characteristics of Bell’s Stage I can be seen with any form of systemic illness or sepsis and is not specific for NEC [6]. In fact, one of thediagnostic criteria is occult blood in the feces, presumably because this was seen as a harbinger of severe bowel pathology. However, as we have demonstrated, positive FOB results are not predictors of later stages of NEC. We did collect data on those infants who were diagnosed with Stage I NEC based on the medical record, finding that 14% (n = 29) of our cohort received this diagnosis. Twenty-four percent had at least one positive FOB result in the week prior to their diagnosis. Of these 29 infants, it is important to note that none went on to develop strictures, definite NEC, or require any gastrointestinal surgery during their NICU stay.

Strengths of this study include that it evaluated a large cohort of VLBW infants that reflects a current era of neonatology. Previous studies were performed in VLBW infants over 20 years ago, so it is important to reevaluate our testing strategies as our clinical practices and technology continue to advance. As with any retrospective medical record review, there are always limitations based on the accuracy of documentation. However, since FOB testing for all VLBW infants has been the standard of care in our institution for many years, it was extremely well documented as a laboratory test result in our electronic medical record. Only 1.5% of our eligible infants were excluded due to a lack of FOB testing data.

Future studies should continue to evaluate potential screening tests that could serve as meaningful markers of intestinal inflammation related to NEC. Although routine FOB testing does not offer any benefit to clinicians for the early detection and diagnosis of NEC, other promising point-of-care tests do exist, such as fecal calprotectin [16, 17].

5 Conclusions

Routine FOB testing has very poor sensitivity, specificity, and positive predictive value and we therefore conclude that it should not be routinely performed for prediction of NEC. We demonstrated that positive FOB results occur in a majority of VLBW infants while in the NICU. We found that IUGR neonates were more likely to have positive FOB results and were more likely to develop NEC, and therefore these infants should be very closely monitored. Feeding and management strategies should take into consideration the increased risk in the VLBW IUGR cohort.

Financial disclosures

The authors have no financial relationships relevant to this article to disclose. Aimee Pickering was awarded a stipend from the Proposed Research Initiated by Students and Mentors (PRISM) Program from the University of Maryland School of Medicine to participate in this research project. The authors have no conflicts of interest to disclose.

Human research statement

The authors affirm that research involving human infants submitted to the journal was conducted in accordance with the ethical standards of all applicable national and institutional committees and the World Medical Association’s Helsinki Declaration. We obtained IRB approval from our institution to conduct this study. No animal research was conducted.

Other disclosure or disclaimer statement(s)

Aimee Pickering was awarded a stipend from the Proposed Research Initiated by Students and Mentors (PRISM) Program from the University of Maryland School of Medicine to participate in this research project. The authors have no conflicts of interest to disclose.

Footnotes

Acknowledgments

The authors would like to thank Kelly Tracey,Jennifer Thompson, Suzanne Lee and Brandy Daniels for their assistance in the Division of Neonatology. We would like to thank the University of Maryland School of Medicine Office of Student Research and the PRISM (Proposed Research Initiated by Students and Mentors) program.

References

Horbar

, Carpenter

, Badger

, et al . Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics 2012;129:1019–26.

Abramo

, Evans

, Kokomoor

, Kantak

. Occult blood in stools and necrotizing enterocolitis. Is there a relationship?. Am J Dis Child 1988;142:451–2.

Health Quality Ontario . Fecal occult blood test for color-ectal cancer screening: An evidence-based analysis. Ont Health Technol Assess Ser 2009;9:1–40.

Kliegman

, Walsh

. Neonatal necrotizing enterocolitis: Pathogenesis, classification, and spectrum of illness. Curr Probl Pediatr 1987;17:213–88.

Sharma

, Hudak

, Premachandra

, et al . Clinical manifestations of rotavirus infection in the neonatal intensive care unit. Pediatr Infect Dis J 2002;21:1099–105.

Gordon

, Swanson

, Attridge

, Clark

. Emerging trends in acquired neonatal intestinal disease: Is it time to abandon Bell’s criteria?. J Perinatol 2007;27:661–71.

Echevarria Ybarguengoitia

, Ruelas Orozco

, Jasso Gutierrez

. Blood in the feces as an aid to the diagnosis of necrotizing enterocolitis. Bol Med Hosp Infant Mex 1981;38:771–6.

Luig

, Lui

. NSW & ACT NICUS GrouEpidemiology of necrotizing enterocolitis–Part II: Risks and susceptibility of premature infants during the surfactant era: A regional study. J Paediatr Child Health 2005;41:174–9.

Beeby

, Jeffery

. Risk factors for necrotising enterocolitis: The influence of gestational age. Arch Dis Child 1992;67:432–5.

10.

Quigley

, McGuire

. Formula versus donor breast milk for feeding preterm or low birth weight infants. Cochrane Database Syst Rev 2014;4:CD002971.

11.

Garite

, Clark

, Thorp

. Intrauterine growth restriction increases morbidity and mortality among premature neonates. Am J Obstet Gynecol 2004;191:481–7.

12.

Bernstein

, Horbar

, Badger

, Ohlsson

, Golan

. Morbidity and mortality among very-low-birth-weight neonates with intrauterine growth restriction. The Vermont Oxford Network. Am J Obstet Gynecol 2000;182:198–206.

13.

Dorling

, Kempley

, Leaf

. Feeding growth restricted preterm infants with abnormal antenatal Doppler results. Arch Dis Child Fetal Neonatal EdF 2005;90:359–63.

14.

Carter

. Nursing assessment of guaiac-positive and occult blood in preterm infant stools. Neonatal Netw 2014;33:101–105.

15.

Pinheiro

, Clark

, Benjamin

. A critical analysis of the routine testing of newborn stools for occult blood and reducing substances. Adv Neonatal Care 2003;3:133–8.

16.

Albanna

, Ahmed

, Awad

. Stool calprotectin in necrotizing enterocolitis. J Clin Neonatol 2014;3:16–9.

17.

Bin-Nun

, Booms

, Sabag

, Mevorach

, Algur

, Hammerman

. Rapid fecal calprotectin (FC) analysis: Point of care testing for diagnosing early necrotizing enterocolitis. Am J Perinatol 2015;32:337–42.