Our institutional database was searched for all NICU admissions the time period from January 1, 2002 through December 31, 2011 to identify infants with a gestational age at birth £34 weeks who developed stage 2a or greater NEC using modified Bell's Criteria [6]. Infants with known chromosomal anomalies, congenital heart disease, or a diagnosis of an isolated spontaneous intestinal perforation were excluded. Eligible matched controls were identified by matching 1 to 1 with each case patient for gestational age at birth (±10 days), admission date (±4 weeks) and birth weight (±150g). The electronic medical records for each of these cases and controls were then reviewed and data elements collected for further analysis.

Maternal and infant characteristics historically associated with an increased risk of developing NEC were recorded. Maternal factors included the following: pregnancy induced hypertension, chorioamnionitis, administration of antenatal steroids, preterm premature rupture of membranes (PPROM), and prolonged rupture of membranes (PROM). Infant data elements collected included; gestational age at birth, birth weight, gender, mode of delivery, Apgar scores at 1 and 5 minutes, pre-transfusion hematocrit, corrected gestational age at onset of NEC, relationship of NEC onset to a transfusion in the preceding 48 hours, and severity of NEC along with any therapy provided. Additional infant data elements included the presence of a hemodynamically significant patent ductus arteriosus (PDA) and treatment, if any, hypotension and therapy provided, type of feeding (breast milk or formula feedings), administration of postnatal steroids for prevention or treatment of chronic lung disease, and enteral feeding status at the time of any transfusion. Data on outcome variables included: diagnosis of bronchopulmonary dysplasia (BPD) as defined by oxygen requirement at 36 weeks corrected gestational age, periventricular leucomalacia (PVL) as defined by ultrasonography scan at 36 weeks corrected gestational age, retinopathy of prematurity (ROP), length of stay, and death prior to hospital discharge. The decision to provide a PRBC transfusion was at the discretion of attending physician. Our hospital policy throughout the study period was that infants who required a PRBC transfusion were transfused with PRBC that had been stored for less than five days. We did not divide an adult unit of blood into aliquots to limit donor exposure. All transfused PRBC units were cytomegalovirus (CMV) negative, irradiated and leuko-reduced. Donor blood was collected and stored in citrate phosphate dextrose (CPD) solution.

Statistical methods: Statistical analyses were performed with SPSS 20. Chi-square tests and Fisher's exact tests were done to measure degree of association between categorical variables. Analysis of variance (ANOVA) was used for comparisons involving continuous variables between NEC and non-NEC. A simple logistic regression was performed to calculate odds of having major outcomes variables like mortality and PVL. Statistical significance was set at a P <0.05.

Results

Over the ten year period from January 1, 2002 to December 31, 2011, there were 9022 admissions to the NICU, and 131 neonates were identified as having had NEC. Thirty-one infants did not meet the predetermined inclusion criteria leaving a total of 100 patients with NEC for evaluation (Fig. 1).

Fig. 1 Patient and control selection flow chart with inclusion and exclusion criteria.

The overall incidence of NEC during the study period was 1.45% of all admissions to the NICU, and 3.67% in infants £34 weeks gestation. The incidence of NEC did not vary significantly at our center during the 10 year study period. The NEC and matched control groups were similar in gestational age at birth and most other demographic variables but transfusion-associated NEC infants were significantly smaller in birth weight compared to other infants with NEC and non NEC controls (P<0.05) (Table I). There were more female and Hispanic subjects in the transfusion-associated NEC group, and a larger proportion of them had hemodynamically significant PDA which required surgical ligation (P<0.05). A significantly greater number of mothers of transfusion-associated NEC infants were diagnosed with pregnancy-induced hypertension as compared to controls (Table I).

TABLE I Demographic Characteristics of NEC Infants and Controls

Infants with transfusion-associated NEC had a significantly lower hematocrit at birth, and also a lower hematocrit just before onset of NEC (Table II). Infants with transfusion-associated NEC were of lower gestational age at birth, had lower birth weight, and were more likely to have had a 5 minute Apgar score of less than 6. Infants in the transfusion-associated group were more likely to be of blood type B+ and less likely to be blood type A+. The rates of premature rupture of membranes, maternal PIH and absent end diastolic umbilical flow were similar between the two groups of NEC.

TABLE II Clinical Status at Onset of NEC and Outcomes in Two Groups

Infants with transfusion-associated NEC were older at the time when NEC developed, and had a higher rate of surgical (stage 3a+3b) NEC. They were also more likely to have had their feedings held around the time of blood transfusion.

Our study demonstrated an association between PRBC transfusion and onset of NEC in about a quarter of our infants over the 10-year study period. These infants were smaller in their birth weight, were born at lower gestational age, and were more likely to have had a hemodynamically significant PDA. Infants in our study did not demonstrate a protective effect of withholding feedings before blood transfusion.

This main limitations of the study are: a retrospective observational design; predominantly African-American population, and a small sample size. Absence of a standard feeding protocol or a policy on withholding feedings at the time of transfusion during the study period precluded demonstration of a strong relationship between PRBC transfusion and NEC.

Some recent studies have identified a similar association between transfusion and transfusion-associated NEC [5, 7-15]. A meta-analysis of several of these studies concluded that there is a strong association between PRBC transfusion and odds of developing NEC [16]. A more recent meta-analysis has been more cautious in reaching the same conclusion [17]. On the other hand, two recent reports did not find PRBC transfusion as a risk factor for NEC [18,19]. Some investigators have reported a significant decrease in the overall NEC rate following the institution of a policy of withholding enteral feedings during and after PRBC transfusion [20]. However, in our study, 84.6% of transfusion-associated NEC cases had their enteral feedings withheld during the time of transfusion as well as for several hours afterwards, suggesting that withholding enteral feedings around the time of a transfusion may not prevent NEC.

In conclusion, our study suggests a temporal association of a PRBC transfusion with the development of NEC that has also been reported by others. Our data also suggest that infants with transfusion-associated NEC represent a subset that is more likely to be seen in smaller VLBW infants. Withholding enteral feedings around the time of a PRBC transfusion also does not seem to be an effective strategy to decrease the occurrence of transfusion-associated NEC.