Journal of Perinatology (2016), 1–5 © 2016 Nature America, Inc. All rights reserved 0743-8346/16 www.nature.com/jp
ORIGINAL ARTICLE
Enteral feeding during indomethacin treatment for patent ductus arteriosus: association with gastrointestinal outcomes D Louis1, R Torgalkar1, J Shah1, PS Shah2 and A Jain1 OBJECTIVE: Enteral feeds are often discontinued or reduced during indomethacin treatment for patent ductus arteriosus (PDA) in preterm neonates, but the clinical impact of this practice is unknown. The objective of this study was to study the associations between enteral feed volume at the time of indomethacin therapy in preterm neonates with PDA and subsequent gastrointestinal outcomes. STUDY DESIGN: Retrospective cohort study. Single-center level III Neonatal Intensive Care Unit. RESULTS: All consecutive preterm neonates who had received treatment with indomethacin for PDA over a 5-year period were included and categorized based on enteral feed volume exposure during treatment (Group A: nil per oral (NPO, N = 229); Group B: ⩽ 60 ml kg − 1 day − 1 (N = 142); Group C:460 ml kg − 1 day − 1 (N = 44)). Baseline characteristics and clinical outcomes were compared between the three groups. The primary outcome was necrotizing enterocolitis (NEC) ⩾ stage IIa, while secondary outcomes included other gastrointestinal complications and common prematurity-related morbidities. Group C had a higher gestational age (mean ± s.d.; A: 26.3 ± 1.8; B: 26.1 ± 1.8; C: 27.0 ± 2.0 weeks), birth weight (A: 864 ± 239; B: 847 ± 202; C: 932 ± 234 g) and postnatal age at the time of indomethacin treatment (A: 5.3 ± 2.9; B: 7.2 ± 4.9; C: 15.4 ± 6.6 days). All groups had similar rates of the primary outcome NEC (A: 6.1%, B: 7.8% and C: 4.6%, respectively). They also had similar rates of the secondary outcomes with the exception of days to reach enteral feeds of 120 ml kg − 1 day − 1 (A: 22.8 ± 8.5; B: 20.5 ± 8.6; C: 16.8 ± 7.7; P o 0.05 for all inter-group comparisons). Secondary analysis including only those neonates who were not already NPO before indomethacin treatment (N = 261) and categorized based on preemptive management (made NPO; enteral feed volume reduced; enteral feed volume unchanged/increased) also showed similar results. CONCLUSIONS: This large retrospective study did not identify any association between enteral feed volumes during indomethacin treatment or preemptive reduction in enteral feeds and subsequent incidence of adverse gastrointestinal outcomes in preterm neonates. Preemptive reduction in enteral feed volume was associated with longer time to reach full enteral feeds. Journal of Perinatology advance online publication, 25 February 2016; doi:10.1038/jp.2016.11
INTRODUCTION A hemodynamically significant patent ductus arteriosus (PDA), by virtue of ‘stealing’ blood from the descending aorta to pulmonary circulation, is thought to result in a chronic state of relative gut hypoperfusion. In addition, indomethacin, a cyclo-oxygenase inhibitor commonly used for treatment of PDA in preterm neonates, also decreases blood flow to the gastrointestinal tract, presumably by its vasoconstrictive action on mesenteric arteries.1–4 Several epidemiological studies have reported an association between both PDA as well as indomethacin treatment and necrotizing enterocolitis (NEC) in preterm neonates, although this has not been conclusively proven in randomized controlled trials.5 Management of enteral feeding for preterm neonates during indomethacin therapy has remained an unresolved area in neonatal medicine, resulting in a wide variation in practice.6 Although some clinicians preemptively discontinue or reduce enteral feed volume at the time of initiation of indomethacin treatment for PDA, presumably in an effort to reduce the probability of adverse gastrointestinal complications, in particular NEC, others tend to continue enteral feeds unchanged. Currently, there is limited evidence to support either approach. A recently
concluded randomized controlled trial demonstrated that continuation of trophic feeds during treatment of PDA with nonsteroidal anti-inflammatory drugs reduced the time to reach full enteral feeds in comparison to neonates who were kept nil per oral (NPO).7 The effect of higher volumes of enteral feeds during indomethacin treatment on subsequent adverse gastrointestinal outcomes is not known. In this study, our primary objective was to investigate the association between enteral feeding volume (EFV) intake during indomethacin treatment for PDA in preterm neonates and the subsequent incidence of NEC as well as other relevant gastrointestinal and prematurity-related morbidities. Our secondary objective was to evaluate the association between preemptive enteral feed related clinical management prior to initiating indomethacin treatment for PDA in preterm neonates. We hypothesized that there is no association between EFV during indomethacin therapy and subsequent adverse gastrointestinal outcomes. We further hypothesized that preemptive discontinuation or reduction in enteral feeds prior to initiating indomethacin treatment is not associated with better outcomes.
1 Department of Pediatrics, Mount Sinai Hospital, Toronto, ON, Canada and 2Department of Pediatrics and Health Policy, Management, and Evaluation, Mount Sinai Hospital, University of Toronto, Toronto, ON, Canada. Correspondence: Dr A Jain, Department of Pediatrics, Mount Sinai Hospital, University of Toronto, 600 University Avenue, Suite 19-231, Toronto, ON M5G 1X5, Canada. E-mail: [email protected] Received 31 July 2015; revised 3 December 2016; accepted 12 January 2016
Indomethacin and feeding in preterm neonates D Louis et al
2
PATIENTS AND METHODS This retrospective cohort study included consecutive preterm neonates who received at least one treatment course of indomethacin for PDA over a 5-year period from July 2008 to June 2013 in the level III Neonatal Intensive Care Unit (NICU) of Mount Sinai Hospital in Toronto, Ontario, Canada. Potential eligible neonates were identified from our Institute’s electronic pharmacy record. Neonates with a history of NEC prior to indomethacin therapy or those with any known major congenital or chromosomal abnormalities were excluded. During the study period, the decision to treat PDA with indomethacin was made by the clinical team based on either clinical signs or echocardiographic confirmation of PDA or both. Indomethacin was usually administered at a dose of 0.2 mg kg − 1 every 12 h infused over 15 to 20 min for a total of three doses. Management of enteral feeds during indomethacin therapy was at the discretion of the attending neonatologist. Practices varied from discontinuing or not initiating feeds (that is, NPO) to preemptive reduction in EFV or maintaining EFV the same as before initiating treatment (that is, not increasing) or increase EFV as per the unit’s feeding protocol. This resulted in a range of EFV exposure during indomethacin treatment. The hospital records of all identified neonates were screened to define eligibility for enrollment. For all included neonates, data regarding baseline demographics, preexisting morbidities prior to indomethacin treatment, age at PDA treatment, dose of indomethacin and enteral feeds 24 h prior to initiating indomethacin and during indomethacin were collected. Small for gestational age was defined as a birth weight less than the 10th percentile for the gestational age using Fenton’s gender-specific growth charts.8 Donor human milk was not available in our unit during the study period. All neonates were preferentially given their mother’s expressed breast milk. Preterm formula was only used if expressed breast milk was not available or was contraindicated. For this study, we recorded details of the type of feeds received during indomethacin treatment. Each neonate was only enrolled once. For neonates who received multiple courses of indomethacin, EFV at the time of first course was included in the analysis. However, data regarding use of prophylactic indomethacin as well as all subsequent courses of indomethacin was recorded to allow calculation of total indomethacin exposure during NICU stay.
Exposure Neonates were categorized into three groups based on the actual EFV received at the time of initiation of indomethacin therapy: 1. Group A: NPO, 2. Group B: EFV ⩽ 60 ml kg − 1 day − 1 (that is, ⩽ 50% full enteral feeds) and 3. Group C: EFV460 ml kg − 1 day − 1 (that is, 450% of full enteral feeds). EFV was calculated by dividing the daily total enteral feed volume received by the infant’s current weight. We chose these categories because intravenous fluid therapy is usually discontinued in our unit when neonates reach an EFV of 120 ml kg − 1 day − 1. We also performed secondary analysis based on the preemptive clinical decisions made regarding enteral feed management at the time of initiating indomethacin treatment. For this, we only included neonates who were on at least some enteral feeds during the 24-h period prior to indomethacin treatment with no evidence of feed intolerance. This subcohort was again divided into three groups: 1. Group 1: electively made NPO, 2. Group 2: feed volume reduced by at least 25% from pretreatment volume but not made NPO and 3. Group 3: feed volume remained unchanged or increased during treatment.
Outcomes The primary outcome for this study was NEC ⩾ stage IIA based on the modified Bell’s staging criteria9 that occurred at any time after initiating treatment with indomethacin during NICU stay. The final diagnosis and staging were recorded from health charts as documented by the attending clinical team. Neonates diagnosed with posttreatment spontaneous intestinal perforation (n = 2) were excluded from the analysis. Secondary outcomes included NEC occurring within 7 days of Journal of Perinatology (2016), 1 – 5
indomethacin treatment, surgical procedure for NEC (pen rose drain or laparotomy), feeding intolerance during indomethacin treatment (defined as at least two consecutive naso-gastric/oro-gastric aspirates of 450% of previous feed volume or any aspirate containing bile or blood), time to reach EFV of 120 and 160 ml kg − 1 day − 1 (calculated for each neonate in days from birth to reach respective EFV for the first time during NICU stay) and duration of total parenteral nutrition (TPN). Other secondary outcomes included duration of central venous catheter use, incidence of culturepositive sepsis (blood, urine or cerebrospinal fluid positive for pathogenic organisms), chronic lung disease,10 retinopathy of prematurity ⩾ stage III,11 duration of hospital stay and mortality during NICU stay. Using data from the entire cohort, we also explored the relationship between EFV taken as a continuous variable and the primary outcome.
Ethics Our institutional research ethics board approved this study.
Statistical consideration Baseline characteristics were described using descriptive statistics. The incidence of primary and secondary outcomes between groups was compared using Chi-square test or Fischer’s exact test for categorical variables and Student's t-test or Wilcoxon rank-sum test for continuous variables depending on their distribution. The data obtained were used to identify whether EFV had an independent association with primary outcome after adjusting for gestational age at birth, need for surfactant therapy, postnatal age and need for invasive ventilation at the time of treatment using multivariate logistic and linear regression analysis. A P-value of o0.05 was considered significant.
RESULTS A total of 415 preterm neonates received indomethacin treatment for PDA during the study period and were included in this study. Of these, 229 infants were NPO (Group A), 142 received EFV ⩽ 60 ml kg − 1 day − 1 (Group B) and 44 received EFV460 ml kg − 1 day − 1 (Group C) during indomethacin treatment. In Group A, 155 neonates (68%) were already NPO before indomethacin treatment, while 74 neonates (32%) were receiving some enteral feeds but were made NPO for the duration of indomethacin treatment. Comparison of baseline characteristics and demographic parameters at the time of initiation of indomethacin treatment revealed some inter-group differences. Neonates in Group C had a slightly higher gestational age and birth weight and were more likely to be born to hypertensive and diabetic mothers but less likely to have received surfactant treatment or umbilical arterial catheter insertion (Table 1). Prophylactic indomethacin use did not differ between the three groups. As expected, neonates in Group C received indomethacin at a later postnatal age and were less likely to have required invasive ventilation at that time (Table 2). None of the study neonates had a previous history of spontaneous intestinal perforation. The range of EFV during indomethacin treatment for Group B was 1 to 60 ml kg − 1 day − 1 and for Group C was 62 to 160 ml kg − 1 day − 1. First-course total indomethacin dose as well as cumulative indomethacin exposure during entire NICU stay were similar across the three groups. Comparison of primary outcome revealed no significant difference between groups (Table 3). Neonates in Group A experienced lower incidence of feeding intolerance compared with the other two groups. There was an inverse relationship between volume of feeds during indomethacin treatment and time to reach full enteral feeds, duration of TPN and duration of exposure to central venous catheter, although not all comparisons reached statistical significance. Multivariate logistic regression analysis using NPO as the reference category and adjusted for the effects of gestational age at birth, surfactant use, postnatal age and invasive ventilation at the time of indomethacin treatment did not show any association between NPO over other categories for © 2016 Nature America, Inc.
Indomethacin and feeding in preterm neonates D Louis et al
3 Table 1.
Baseline demographics of the study population
Gestational age (weeks) Birth weight (g) Male SGA Multiple gestation Maternal PIH Maternal diabetes Received any antenatal steroid PPROM Cesarean section Apgar score at 5 min Surfactant use 41 surfactant doses Grade 3 or 4 IVH UAC use Treatment with inotrope in the first 48 h after birth Prophylactic indomethacin use
P-value
Group A
Group B
Group C
Nil per oral
1–60 ml kg − 1 day − 1
460 ml kg − 1 day − 1
(N = 229)
(N = 142)
(N = 44)
26.3 864 130 3 117 40 16 183 62 105 7 220 40 33 164 21 13
(1.8) (239) (56.7) (1.3) (51.8) (17.5) (7.0) (82.4) (27.3) (46.3) (5, 9) (96.1) (18.0) (14.4) (71.6) (13.0) (5.7)
26.1 847 85 1 77 12 6 121 43 62 7 138 25 14 113 8 8
(1.8) (202) (59.9) (0.7) (54.6) (8.5) (4.3) (88.3) (30.9) (44.0) (5, 8) (97.2) (18.1) (9.9) (79.6) (8.1) (5.6)
27.0 932 24 1 20 10 7 40 14 19 8 36 4 4 27 2 4
(2.0) (234) (54.6) (2.3) (45.5) (22.7) (15.9) (93.0) (32.6) (43.2) (6, 9) (81.8) (11.1) (9.1) (61.4) (5.6) (9.1)
B vs A
C vs A
C vs B
0.31 0.48 0.56 0.99 0.60 0.02 0.28 0.13 0.46 0.67 0.19 0.57 0.98 0.20 0.09 0.22 0.97
0.02 0.08 0.79 0.51 0.44 0.41 0.05 0.08 0.48 0.71 0.05 o 0.01 0.31 0.47 0.17 0.26 0.40
o0.01 0.02 0.53 0.42 0.29 0.01 0.01 0.38 0.84 0.93 0.01 o0.01 0.45 0.99 0.01 0.99 0.48
Abbreviations: IVH, intraventricular hemorrhage; PIH, pregnancy-induced hypertension; PPROM, preterm premature rupture of membranes; SGA, small for gestational age; UAC, umbilical artery catheter. Values are mean (s.d.), n (%) or median (interquartile range).
Table 2.
Demographic parameters at the time of first-course indomethacin therapy Group A
Group B
P-value
Group C
Nil per oral 1–60 ml kg − 1 day − 1 460 ml kg − 1 day − 1 B vs A (N = 229) Postnatal age (days) Postnatal weight (g) Invasive ventilation FiO2 (%) Volume of enteral feeds prior to indomethacin therapy (ml kg − 1 day − 1) Volume of feeds during indomethacin therapy (ml kg − 1 day − 1) Number exposed to formula feed during indomethacin therapy Total dose of first-course indomethacin (mg kg − 1) Cumulative indomethacin exposure during NICU stay (mg kg − 1)
(N = 142)
C vs A
C vs B
o0.01 o0.01 o0.01 0.30 o0.01
o0.01 o0.01 o0.01 0.38 o0.01
(N = 44)
5.3 (2.9) 803 (236) 152 (66.4) 25.0 (5.9) 7.0 (20.7)
7.2 794 104 25.2 19.5
(4.9) (242) (73.8) (6.4) (29.5)
15.4 974 19 26.5 110.1
(6.6) (229) (43.2) (9.3) (42.3)
o0.01 0.71 0.14 0.77 o0.01
0 (0) — 0.6 (0.4) 0.8 (0.5)
16.7 5 0.6 0.8
(17.3) (3.6) (0.1) (0.3)
98.6 0 0.7 0.8
(31.4) (0) (0.8) (0.8)
o0.01 o0.01 o0.01 — — 0.34 0.67 0.48 0.42 0.76 0.95 0.86
Abbreviation: NICU, Neonatal Intensive Care Unit. Values are mean (s.d.) or n (%).
the primary outcome (odds ratio (OR) and 95% confidence interval (CI): 0.78 (0.33, 1.87) for ⩽ 60 ml kg − 1 day − 1 and 0.75 (0.10, 5.51) for 460 ml kg − 1 day − 1). Linear regression analysis including data from entire cohort, using EFV as a continuous variable and NEC as the outcome adjusted for gestational age and postnatal age at treatment, also did not show a significant association (adjusted OR 0.99 (95% CI: 0.98, 1.01) for every ml kg − 1 day − 1 of enteral feeds). Secondary analysis based on preemptive clinical decisions made prior to initiating treatment with indomethacin included 261 neonates. Of these, 74 neonates were made NPO, 40 had enteral feeds continued, but volume was reduced compared with preindomethacin level and 147 had EFV either kept the same or increased. The rates of NEC in these three groups were 4.1, 12.5 and 5.4%, respectively (P40.05 for all inter-group comparisons). Comparison for secondary outcomes did not show any difference between groups (data not shown), except for total days from birth to reach EFV of 120 ml kg − 1 day − 1 for the first time, which was © 2016 Nature America, Inc.
significantly longer in babies who were made NPO for treatment (mean (s.d.): 21.9 (7.9) days) compared with those where the feed volume was reduced (18.4 (8.5) days; P = 0.04). DISCUSSION In this retrospective cohort study including a relatively large number of preterm neonates, we found no association between variation in actual intake of EFV or preemptive changes in feeds during indomethacin treatment for PDA and subsequent incidence of NEC or other gastrointestinal and common prematurity-related morbidities. Lower EFV during indomethacin treatment was associated with a slight delay in reaching full enteral feeds as well as a reciprocal increase in duration of exposure to TPN and central venous catheter. Management of enteral feeds during pharmacological treatment for PDA in preterm neonates is a common clinical Journal of Perinatology (2016), 1 – 5
Indomethacin and feeding in preterm neonates D Louis et al
4 Table 3.
Outcomes in the study population
NEC NEC within 7 days of indomethacin treatment Surgical NEC Feeding intolerance Days to reach enteral feeds of 120 ml kg − 1 day − 1 Days to reach enteral feeds of 160 ml kg − 1 day − 1 Duration of TPN (days) Duration of central line (days) PDA ligation CLD ROP ⩾ stage III Total length of hospital stay (days) Death before discharge
Group A
Group B
Group C
Nil per oral
1–60 ml kg − 1 day − 1
460 ml kg − 1 day − 1
(N = 229)
(N = 142)
(N = 44)
14 5 6 11 22.8 29.9 25.1 20.8 31 171 25 75.9 27
(6.1) (2.1) (2.6) (4.9) (8.5) (17.2) (13.1) (13.2) (13.5) (74.7) (13.0) (43.1) (11.7)
11 2 2 15 20.5 29.4 23.8 19.1 24 114 25 77.4 19
(7.8) (1.4) (1.4) (10.6) (8.6) (19.3) (11.1) (11.0) (16.9) (80.9) (19.2) (42.8) (13.4)
2 2 0 4 16.8 23.5 21.4 16.4 5 38 3 71.3 2
(4.6) (4.6) (0) (9.3) (7.7) (12.3) (10.7) (11.9) (11.4) (86.4) (7.3) (38.4) (4.6)
P-value B vs A
C vs A
C vs B
0.54 0.99 0.72 0.04 0.02 0.80 0.32 0.19 0.38 0.17 0.14 0.74 0.78
0.69 0.18 0.59 0.28 o0.01 0.01 0.08 0.05 0.70 0.09 0.20 0.52 0.19
0.74 0.24 0.99 0.99 0.01 0.03 0.20 0.18 0.38 0.40 0.09 0.40 0.17
Abbreviations: CLD, chronic lung disease; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosus; ROP, retinopathy of prematurity; TPN, total parenteral nutrition. Values are mean (s.d.) or n (%).
scenario that is encountered on a day-to-day basis in NICUs. The literature to inform clinical practice is scant resulting in a wide variability of practice. Only two studies, both recently published, have examined the relationship between enteral feeding during indomethacin treatment and gastrointestinal complications in preterm neonates.7,12 The ductus arteriosus feed or fast with indomethacin or ibuprofen (DAFFII) trial was a controlled trial that randomized 177 neonates o 30 weeks gestational age to a NPO group vs trophic feeds (15 ml kg − 1 day − 1) when their PDAs were treated with indomethacin or ibuprofen.7 The primary outcome was the time to reach full enteral feeds (120 ml kg − 1 day − 1). This study reported a shorter duration of time to reach full enteral feeds in the trophic arm by a mean of 2.8 days without any significant difference in the rates of NEC (13% vs 10%) or other neonatal morbidities. Recently, Kelleher et al.12 performed a retrospective database study including 15 751 extremely low birth weight infants and investigated the association between trophic feeds during prophylactic indomethacin treatment given for the first 3 days of life and occurrence of spontaneous intestinal perforation (SIP). This study also found no significant association. The incidence of SIP in neonates who received feeds and prophylactic indomethacin was 3.3% compared with 4.9% in neonates who received indomethacin while being NPO.12 The results of these studies are keeping with our observation of a lack of association between EFV during indomethacin treatment and adverse gastrointestinal sequelae. To the best of our knowledge, ours is the first report providing outcome data for neonates who received larger feed volumes in comparison to those who were kept NPO (controls) during indomethacin treatment. This is important as a significant number of preterm neonates are expected to be receiving a considerable volume of enteral feeds at the time of initiation of treatment, especially if treated after the first week of age. In our study, we found a wide variability in practice among clinicians regarding management of enteral feeds during indomethacin treatment. Although some preferred to completely stop enteral feeds, others preemptively reduced EFV by an arbitrary chosen percentage. For the remaining neonates, EFV was held at the pretreatment level throughout the duration of the indomethacin course. This variability in practice at our institution is similar to a recently published survey of 281 US and 134 non-US neonatologists. The survey reported that 71% of US Journal of Perinatology (2016), 1 – 5
neonatologists and 22% of non-US neonatologists preferred stopping enteral feeding during the duration of indomethacin therapy.6 Although this variation in practice stems from a lack of research in the field, it allowed us to test the associations between EFV exposure during indomethacin treatment and relevant outcomes. The hypothetical concern of exposing the preterm gut to the additional metabolic ‘stress’ of enteral feeds during a potential period of gut underperfusion from indomethacin, especially in the setting of a preexisting chronic relative hypoperfusion secondary to ‘ductal steal’, prompts clinicians to electively discontinue or reduce enteral feeding during indomethacin treatment. These concerns are backed by small physiological studies that have demonstrated reduction in gut perfusion secondary to presence of PDA as well as treatment with indomethacin. However, the clinical correlation of these observations and their relationship with the ability of the preterm gut to handle enteral feeds is not known. Further, it is plausible that enteral feeds by virtue of inducing an increase in mesenteric blood flow (postprandial increase) may be protective against the vasoconstrictive action of indomethacin. A Doppler velocimetry nested study conducted in a subgroup of 34 neonates in the DAFFII trial reported changes in blood flow velocities in superior mesenteric artery after a 4 ml kg − 1 test feed in babies randomized to NPO or trophic feeds 18 to 24 h after completing indomethacin treatment.13 At 10 min postfeeding, infants with trophic feeds exhibited an increase in the peak systolic velocity and a marginal increase in mean velocity, while the NPO group did not show similar changes; whereas at 30 min after feeding, the increase in systolic, mean and diastolic velocities were similar in the two groups. Further, while evaluating the effect of prophylactic indomethacin and trophic feeds on SIP in a secondary analysis, Kelleher et al.12 reported a significantly lower incidence of SIP in neonates who received both indomethacin and feeds during the first 3 days of life in comparison to those who received neither. The practice of electively discontinuing enteral feeds is contrary to the important clinical goal of reaching full enteral feeds and minimizing exposure to parenteral nutrition. This was demonstrated by our study where we observed an inverse relationship between EFV during indomethacin treatment and time to reach full enteral feeds, duration of need for TPN therapy and central catheters. Other morbidities were similar across all groups. We © 2016 Nature America, Inc.
Indomethacin and feeding in preterm neonates D Louis et al
found the incidence of feeding intolerance to be lower in the NPO group than other groups, which was not surprising given the definition of feeding intolerance used in our study. All infants classified as having feeding intolerance in the NPO group had aspirates recorded as containing bile and/or blood, while in the other two groups neonates with two consecutive aspirates of 450% previous feed volume were also classified as having feeding intolerance. Our study has some limitations. First, this was a retrospective evaluation. Although the variability in EFV decisions prior to initiating indomethacin treatment may represent physician preferences, it may also be secondary to their perception of risks for individual patients. In an attempt to tease out the factors that may have governed such risk assessment and differential clinical management of enteral feeds, we compared neonates who were preemptively made NPO vs those for whom EFV was reduced or remained unchanged (that is, secondary analysis; data not provided) and found no difference in the cardio-respiratory indices of clinical stability, demographic features at the time of treatment, incidence of preexisting feeding intolerance in the 24 h prior to initiating treatment or the presence of any other gastrointestinal complication, suggesting that the differences in preemptive management of feeds observed is likely a reflection of practice variability among clinicians. However, because of a lack of specific documentation, we are unable to confirm this. Second, we cannot rule out the possibility of our study being underpowered to detect a significant difference in primary or secondary outcomes. Nevertheless, this is the largest study available so far on this topic and included all patients treated with indomethacin in our unit during the study period. Third, we are unable to assess the effect of nonsteroidal anti-inflammatory drugs other than indomethacin as these were seldom used in our unit. In many units, ibuprofen is now used as a primary agent for the medical treatment of PDA. Although the results of this study may not be generalizable to other nonsteroidal anti-inflammatory drugs, the use of ibuprofen is unlikely to have altered the outcome of this study as ibuprofen has been shown to cause even less vasoconstriction in splanchnic circulation in comparison to indomethacin.14 Fourth, our method of dividing neonates on enteral feeds during treatment into two separate groups was somewhat arbitrary. However, in an attempt to find a significant cutoff we did evaluate the entire cohort taking EFV as a continuous variable but identified no relationship between EFV and measured outcomes. Fifth, although we found no baseline difference between Group A and Group B, Group C has several important differences. Some of these differences such as preindomethacin feed volume and need for invasive ventilation at the time of treatment were expected given the late postnatal age at treatment in group C. Because of relatively small number of neonates who developed the primary outcome of NEC, we were limited in the number of variables that could be adjusted for in the analyses. We adjusted for what we perceived as key modifying variables (gestational age, postnatal age at treatment, need for surfactant and invasive ventilation at treatment) and demonstrated lack of inter-group differences by comparing as many factors and prematurity-related outcomes as feasible. Although lack of differences in these parameters suggest against major inter-group baseline differences, we are unable to confirm this assumption. Sixth, due to a lack of availability of routine pretreatment and posttreatment echocardiography, we are unable to investigate the effect of relevant PDA characteristics such as PDA size or flow reversal in abdominal aorta on the interaction between EFV during indomethacin treatment and NEC. For the same reason, we are unable to rule out interaction between EFV during treatment and success rate of indomethacin. However, the fact that total indomethacin exposure during NICU stay and rate of PDA ligation were similar between the groups suggest lack of
© 2016 Nature America, Inc.
significant impact of EFV and success rate of first course of idomethacin. Finally, we only collected data on type of enteral feeds (that is, formula vs expressed breast milk) during indomethacin and not throughout NICU stay. Although we are unable to provide a firm clinical recommendation, the results of this study questions the common clinical practice of reducing or withholding enteral feeds during indomethacin treatment in preterm neonates and provides the background and justification for future pragmatic randomized control trials. CONCLUSION There was no association between EFV during indomethacin treatment and adverse gastrointestinal outcomes in preterm neonates. Results of this study do not support the common clinical practice of preemptive discontinuation/reduction in enteral feed volume in preterm neonates receiving treatment with indomethacin for PDA. Further large sample size multicenter randomized controlled trials would be necessary to provide definite evidence. CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS We thank Junmin Yang, statistician in Mother and Infant Care Program of Mount Sinai Hospital, Toronto, ON, Canada for performing the statistical analysis for this study and Josie Chundamala, scientific editor at Mother and Infant Care Program of Mount Sinai Hospital, Toronto, ON, Canada for proofreading the manuscript.
REFERENCES 1 McCurnin D, Clyman RI. Effects of a patent ductus arteriosus on postprandial mesenteric perfusion in premature baboons. Pediatrics 2008; 122: e1262–e1267. 2 Coombs RC, Morgan ME, Durbin GM, Booth IW, McNeish AS. Gut blood flow velocities in the newborn: effects of patent ductus arteriosus and parenteral indomethacin. Arch Dis Child 1990; 65: 1067–1071. 3 Laudignon N, Chemtob S, Bard H, Aranda JV. Effect of indomethacin on cerebral blood flow velocity of premature newborns. Biol Neonate 1988; 54: 254–262. 4 Christmann V, Liem KD, Semmekrot BA, van de Bor M. Changes in cerebral, renal and mesenteric blood flow velocity during continuous and bolus infusion of indomethacin. Acta Paediatr 2002; 91: 440–446. 5 Fowlie PW, Davis PG, McGuire W. Prophylactic intravenous indomethacin for preventing mortality and morbidity in preterm infants. Cochrane Database of Syst Rev 2010, Issue 7. Art. No.: CD000174. doi:10.1002/14651858.CD000174.pub2. 6 Jhaveri N, Soll RF, Clyman RI. Feeding practices and patent ductus arteriosus ligation preferences-are they related? Am J Perinatol 2010; 27: 667–674. 7 Clyman R, Wickremasinghe A, Jhaveri N, Hassinger DC, Attridge JT, Sanocka U et al. Enteral feeding during indomethacin and ibuprofen treatment of a patent ductus arteriosus. J Pediatr 2013; 163(2): 406–411.e4. 8 Fenton TR, Kim JH. A systematic review and meta-analysis to revise the Fenton growth chart for preterm infants. BMC Pediatr 2013; 13: 59. 9 Kliegman RM. Neonatal necrotizing enterocolitis: implications for an infectious disease. Pediatr Clin North Am 1979; 26: 327–344. 10 Jobe AH, Bancalari E. Bronchopulmonary dysplasia. Am J Respir Crit Care Med 2001; 163: 1723–1729. 11 International Committee for the Classification of Retinopathy of Prematurity. The International Classification of Retinopathy of Prematurity revisited. Arch Ophthalmol 2005; 123: 991–999. 12 Kelleher J, Salas AA, Bhat R, Ambalavanan N, Saha S, Stoll BJ et al. Prophylactic indomethacin and intestinal perforation in extremely low birth weight infants. Pediatrics 2014; 134: e1369–e1377. 13 Yanowitz TD, Reese J, Gillam-Krakauer M, Cochran CM, Jegatheesan P, Lau J et al. Superior mesenteric artery blood flow velocities following medical treatment of a patent ductus arteriosus. J Pediatr 2014; 164: 661–663. 14 Pezzati M, Vangi V, Biagiotti R, Bertini G, Cianciulli D, Rubaltelli FF. Effects of indomethacin and ibuprofen on mesenteric and renal blood flow in preterm infants with patent ductus arteriosus. J Pediatr 1999; 135: 733–738.
Journal of Perinatology (2016), 1 – 5
5
ORIGINAL ARTICLE
Enteral feeding during indomethacin treatment for patent ductus arteriosus: association with gastrointestinal outcomes D Louis1, R Torgalkar1, J Shah1, PS Shah2 and A Jain1 OBJECTIVE: Enteral feeds are often discontinued or reduced during indomethacin treatment for patent ductus arteriosus (PDA) in preterm neonates, but the clinical impact of this practice is unknown. The objective of this study was to study the associations between enteral feed volume at the time of indomethacin therapy in preterm neonates with PDA and subsequent gastrointestinal outcomes. STUDY DESIGN: Retrospective cohort study. Single-center level III Neonatal Intensive Care Unit. RESULTS: All consecutive preterm neonates who had received treatment with indomethacin for PDA over a 5-year period were included and categorized based on enteral feed volume exposure during treatment (Group A: nil per oral (NPO, N = 229); Group B: ⩽ 60 ml kg − 1 day − 1 (N = 142); Group C:460 ml kg − 1 day − 1 (N = 44)). Baseline characteristics and clinical outcomes were compared between the three groups. The primary outcome was necrotizing enterocolitis (NEC) ⩾ stage IIa, while secondary outcomes included other gastrointestinal complications and common prematurity-related morbidities. Group C had a higher gestational age (mean ± s.d.; A: 26.3 ± 1.8; B: 26.1 ± 1.8; C: 27.0 ± 2.0 weeks), birth weight (A: 864 ± 239; B: 847 ± 202; C: 932 ± 234 g) and postnatal age at the time of indomethacin treatment (A: 5.3 ± 2.9; B: 7.2 ± 4.9; C: 15.4 ± 6.6 days). All groups had similar rates of the primary outcome NEC (A: 6.1%, B: 7.8% and C: 4.6%, respectively). They also had similar rates of the secondary outcomes with the exception of days to reach enteral feeds of 120 ml kg − 1 day − 1 (A: 22.8 ± 8.5; B: 20.5 ± 8.6; C: 16.8 ± 7.7; P o 0.05 for all inter-group comparisons). Secondary analysis including only those neonates who were not already NPO before indomethacin treatment (N = 261) and categorized based on preemptive management (made NPO; enteral feed volume reduced; enteral feed volume unchanged/increased) also showed similar results. CONCLUSIONS: This large retrospective study did not identify any association between enteral feed volumes during indomethacin treatment or preemptive reduction in enteral feeds and subsequent incidence of adverse gastrointestinal outcomes in preterm neonates. Preemptive reduction in enteral feed volume was associated with longer time to reach full enteral feeds. Journal of Perinatology advance online publication, 25 February 2016; doi:10.1038/jp.2016.11
INTRODUCTION A hemodynamically significant patent ductus arteriosus (PDA), by virtue of ‘stealing’ blood from the descending aorta to pulmonary circulation, is thought to result in a chronic state of relative gut hypoperfusion. In addition, indomethacin, a cyclo-oxygenase inhibitor commonly used for treatment of PDA in preterm neonates, also decreases blood flow to the gastrointestinal tract, presumably by its vasoconstrictive action on mesenteric arteries.1–4 Several epidemiological studies have reported an association between both PDA as well as indomethacin treatment and necrotizing enterocolitis (NEC) in preterm neonates, although this has not been conclusively proven in randomized controlled trials.5 Management of enteral feeding for preterm neonates during indomethacin therapy has remained an unresolved area in neonatal medicine, resulting in a wide variation in practice.6 Although some clinicians preemptively discontinue or reduce enteral feed volume at the time of initiation of indomethacin treatment for PDA, presumably in an effort to reduce the probability of adverse gastrointestinal complications, in particular NEC, others tend to continue enteral feeds unchanged. Currently, there is limited evidence to support either approach. A recently
concluded randomized controlled trial demonstrated that continuation of trophic feeds during treatment of PDA with nonsteroidal anti-inflammatory drugs reduced the time to reach full enteral feeds in comparison to neonates who were kept nil per oral (NPO).7 The effect of higher volumes of enteral feeds during indomethacin treatment on subsequent adverse gastrointestinal outcomes is not known. In this study, our primary objective was to investigate the association between enteral feeding volume (EFV) intake during indomethacin treatment for PDA in preterm neonates and the subsequent incidence of NEC as well as other relevant gastrointestinal and prematurity-related morbidities. Our secondary objective was to evaluate the association between preemptive enteral feed related clinical management prior to initiating indomethacin treatment for PDA in preterm neonates. We hypothesized that there is no association between EFV during indomethacin therapy and subsequent adverse gastrointestinal outcomes. We further hypothesized that preemptive discontinuation or reduction in enteral feeds prior to initiating indomethacin treatment is not associated with better outcomes.
1 Department of Pediatrics, Mount Sinai Hospital, Toronto, ON, Canada and 2Department of Pediatrics and Health Policy, Management, and Evaluation, Mount Sinai Hospital, University of Toronto, Toronto, ON, Canada. Correspondence: Dr A Jain, Department of Pediatrics, Mount Sinai Hospital, University of Toronto, 600 University Avenue, Suite 19-231, Toronto, ON M5G 1X5, Canada. E-mail: [email protected] Received 31 July 2015; revised 3 December 2016; accepted 12 January 2016
Indomethacin and feeding in preterm neonates D Louis et al
2
PATIENTS AND METHODS This retrospective cohort study included consecutive preterm neonates who received at least one treatment course of indomethacin for PDA over a 5-year period from July 2008 to June 2013 in the level III Neonatal Intensive Care Unit (NICU) of Mount Sinai Hospital in Toronto, Ontario, Canada. Potential eligible neonates were identified from our Institute’s electronic pharmacy record. Neonates with a history of NEC prior to indomethacin therapy or those with any known major congenital or chromosomal abnormalities were excluded. During the study period, the decision to treat PDA with indomethacin was made by the clinical team based on either clinical signs or echocardiographic confirmation of PDA or both. Indomethacin was usually administered at a dose of 0.2 mg kg − 1 every 12 h infused over 15 to 20 min for a total of three doses. Management of enteral feeds during indomethacin therapy was at the discretion of the attending neonatologist. Practices varied from discontinuing or not initiating feeds (that is, NPO) to preemptive reduction in EFV or maintaining EFV the same as before initiating treatment (that is, not increasing) or increase EFV as per the unit’s feeding protocol. This resulted in a range of EFV exposure during indomethacin treatment. The hospital records of all identified neonates were screened to define eligibility for enrollment. For all included neonates, data regarding baseline demographics, preexisting morbidities prior to indomethacin treatment, age at PDA treatment, dose of indomethacin and enteral feeds 24 h prior to initiating indomethacin and during indomethacin were collected. Small for gestational age was defined as a birth weight less than the 10th percentile for the gestational age using Fenton’s gender-specific growth charts.8 Donor human milk was not available in our unit during the study period. All neonates were preferentially given their mother’s expressed breast milk. Preterm formula was only used if expressed breast milk was not available or was contraindicated. For this study, we recorded details of the type of feeds received during indomethacin treatment. Each neonate was only enrolled once. For neonates who received multiple courses of indomethacin, EFV at the time of first course was included in the analysis. However, data regarding use of prophylactic indomethacin as well as all subsequent courses of indomethacin was recorded to allow calculation of total indomethacin exposure during NICU stay.
Exposure Neonates were categorized into three groups based on the actual EFV received at the time of initiation of indomethacin therapy: 1. Group A: NPO, 2. Group B: EFV ⩽ 60 ml kg − 1 day − 1 (that is, ⩽ 50% full enteral feeds) and 3. Group C: EFV460 ml kg − 1 day − 1 (that is, 450% of full enteral feeds). EFV was calculated by dividing the daily total enteral feed volume received by the infant’s current weight. We chose these categories because intravenous fluid therapy is usually discontinued in our unit when neonates reach an EFV of 120 ml kg − 1 day − 1. We also performed secondary analysis based on the preemptive clinical decisions made regarding enteral feed management at the time of initiating indomethacin treatment. For this, we only included neonates who were on at least some enteral feeds during the 24-h period prior to indomethacin treatment with no evidence of feed intolerance. This subcohort was again divided into three groups: 1. Group 1: electively made NPO, 2. Group 2: feed volume reduced by at least 25% from pretreatment volume but not made NPO and 3. Group 3: feed volume remained unchanged or increased during treatment.
Outcomes The primary outcome for this study was NEC ⩾ stage IIA based on the modified Bell’s staging criteria9 that occurred at any time after initiating treatment with indomethacin during NICU stay. The final diagnosis and staging were recorded from health charts as documented by the attending clinical team. Neonates diagnosed with posttreatment spontaneous intestinal perforation (n = 2) were excluded from the analysis. Secondary outcomes included NEC occurring within 7 days of Journal of Perinatology (2016), 1 – 5
indomethacin treatment, surgical procedure for NEC (pen rose drain or laparotomy), feeding intolerance during indomethacin treatment (defined as at least two consecutive naso-gastric/oro-gastric aspirates of 450% of previous feed volume or any aspirate containing bile or blood), time to reach EFV of 120 and 160 ml kg − 1 day − 1 (calculated for each neonate in days from birth to reach respective EFV for the first time during NICU stay) and duration of total parenteral nutrition (TPN). Other secondary outcomes included duration of central venous catheter use, incidence of culturepositive sepsis (blood, urine or cerebrospinal fluid positive for pathogenic organisms), chronic lung disease,10 retinopathy of prematurity ⩾ stage III,11 duration of hospital stay and mortality during NICU stay. Using data from the entire cohort, we also explored the relationship between EFV taken as a continuous variable and the primary outcome.
Ethics Our institutional research ethics board approved this study.
Statistical consideration Baseline characteristics were described using descriptive statistics. The incidence of primary and secondary outcomes between groups was compared using Chi-square test or Fischer’s exact test for categorical variables and Student's t-test or Wilcoxon rank-sum test for continuous variables depending on their distribution. The data obtained were used to identify whether EFV had an independent association with primary outcome after adjusting for gestational age at birth, need for surfactant therapy, postnatal age and need for invasive ventilation at the time of treatment using multivariate logistic and linear regression analysis. A P-value of o0.05 was considered significant.
RESULTS A total of 415 preterm neonates received indomethacin treatment for PDA during the study period and were included in this study. Of these, 229 infants were NPO (Group A), 142 received EFV ⩽ 60 ml kg − 1 day − 1 (Group B) and 44 received EFV460 ml kg − 1 day − 1 (Group C) during indomethacin treatment. In Group A, 155 neonates (68%) were already NPO before indomethacin treatment, while 74 neonates (32%) were receiving some enteral feeds but were made NPO for the duration of indomethacin treatment. Comparison of baseline characteristics and demographic parameters at the time of initiation of indomethacin treatment revealed some inter-group differences. Neonates in Group C had a slightly higher gestational age and birth weight and were more likely to be born to hypertensive and diabetic mothers but less likely to have received surfactant treatment or umbilical arterial catheter insertion (Table 1). Prophylactic indomethacin use did not differ between the three groups. As expected, neonates in Group C received indomethacin at a later postnatal age and were less likely to have required invasive ventilation at that time (Table 2). None of the study neonates had a previous history of spontaneous intestinal perforation. The range of EFV during indomethacin treatment for Group B was 1 to 60 ml kg − 1 day − 1 and for Group C was 62 to 160 ml kg − 1 day − 1. First-course total indomethacin dose as well as cumulative indomethacin exposure during entire NICU stay were similar across the three groups. Comparison of primary outcome revealed no significant difference between groups (Table 3). Neonates in Group A experienced lower incidence of feeding intolerance compared with the other two groups. There was an inverse relationship between volume of feeds during indomethacin treatment and time to reach full enteral feeds, duration of TPN and duration of exposure to central venous catheter, although not all comparisons reached statistical significance. Multivariate logistic regression analysis using NPO as the reference category and adjusted for the effects of gestational age at birth, surfactant use, postnatal age and invasive ventilation at the time of indomethacin treatment did not show any association between NPO over other categories for © 2016 Nature America, Inc.
Indomethacin and feeding in preterm neonates D Louis et al
3 Table 1.
Baseline demographics of the study population
Gestational age (weeks) Birth weight (g) Male SGA Multiple gestation Maternal PIH Maternal diabetes Received any antenatal steroid PPROM Cesarean section Apgar score at 5 min Surfactant use 41 surfactant doses Grade 3 or 4 IVH UAC use Treatment with inotrope in the first 48 h after birth Prophylactic indomethacin use
P-value
Group A
Group B
Group C
Nil per oral
1–60 ml kg − 1 day − 1
460 ml kg − 1 day − 1
(N = 229)
(N = 142)
(N = 44)
26.3 864 130 3 117 40 16 183 62 105 7 220 40 33 164 21 13
(1.8) (239) (56.7) (1.3) (51.8) (17.5) (7.0) (82.4) (27.3) (46.3) (5, 9) (96.1) (18.0) (14.4) (71.6) (13.0) (5.7)
26.1 847 85 1 77 12 6 121 43 62 7 138 25 14 113 8 8
(1.8) (202) (59.9) (0.7) (54.6) (8.5) (4.3) (88.3) (30.9) (44.0) (5, 8) (97.2) (18.1) (9.9) (79.6) (8.1) (5.6)
27.0 932 24 1 20 10 7 40 14 19 8 36 4 4 27 2 4
(2.0) (234) (54.6) (2.3) (45.5) (22.7) (15.9) (93.0) (32.6) (43.2) (6, 9) (81.8) (11.1) (9.1) (61.4) (5.6) (9.1)
B vs A
C vs A
C vs B
0.31 0.48 0.56 0.99 0.60 0.02 0.28 0.13 0.46 0.67 0.19 0.57 0.98 0.20 0.09 0.22 0.97
0.02 0.08 0.79 0.51 0.44 0.41 0.05 0.08 0.48 0.71 0.05 o 0.01 0.31 0.47 0.17 0.26 0.40
o0.01 0.02 0.53 0.42 0.29 0.01 0.01 0.38 0.84 0.93 0.01 o0.01 0.45 0.99 0.01 0.99 0.48
Abbreviations: IVH, intraventricular hemorrhage; PIH, pregnancy-induced hypertension; PPROM, preterm premature rupture of membranes; SGA, small for gestational age; UAC, umbilical artery catheter. Values are mean (s.d.), n (%) or median (interquartile range).
Table 2.
Demographic parameters at the time of first-course indomethacin therapy Group A
Group B
P-value
Group C
Nil per oral 1–60 ml kg − 1 day − 1 460 ml kg − 1 day − 1 B vs A (N = 229) Postnatal age (days) Postnatal weight (g) Invasive ventilation FiO2 (%) Volume of enteral feeds prior to indomethacin therapy (ml kg − 1 day − 1) Volume of feeds during indomethacin therapy (ml kg − 1 day − 1) Number exposed to formula feed during indomethacin therapy Total dose of first-course indomethacin (mg kg − 1) Cumulative indomethacin exposure during NICU stay (mg kg − 1)
(N = 142)
C vs A
C vs B
o0.01 o0.01 o0.01 0.30 o0.01
o0.01 o0.01 o0.01 0.38 o0.01
(N = 44)
5.3 (2.9) 803 (236) 152 (66.4) 25.0 (5.9) 7.0 (20.7)
7.2 794 104 25.2 19.5
(4.9) (242) (73.8) (6.4) (29.5)
15.4 974 19 26.5 110.1
(6.6) (229) (43.2) (9.3) (42.3)
o0.01 0.71 0.14 0.77 o0.01
0 (0) — 0.6 (0.4) 0.8 (0.5)
16.7 5 0.6 0.8
(17.3) (3.6) (0.1) (0.3)
98.6 0 0.7 0.8
(31.4) (0) (0.8) (0.8)
o0.01 o0.01 o0.01 — — 0.34 0.67 0.48 0.42 0.76 0.95 0.86
Abbreviation: NICU, Neonatal Intensive Care Unit. Values are mean (s.d.) or n (%).
the primary outcome (odds ratio (OR) and 95% confidence interval (CI): 0.78 (0.33, 1.87) for ⩽ 60 ml kg − 1 day − 1 and 0.75 (0.10, 5.51) for 460 ml kg − 1 day − 1). Linear regression analysis including data from entire cohort, using EFV as a continuous variable and NEC as the outcome adjusted for gestational age and postnatal age at treatment, also did not show a significant association (adjusted OR 0.99 (95% CI: 0.98, 1.01) for every ml kg − 1 day − 1 of enteral feeds). Secondary analysis based on preemptive clinical decisions made prior to initiating treatment with indomethacin included 261 neonates. Of these, 74 neonates were made NPO, 40 had enteral feeds continued, but volume was reduced compared with preindomethacin level and 147 had EFV either kept the same or increased. The rates of NEC in these three groups were 4.1, 12.5 and 5.4%, respectively (P40.05 for all inter-group comparisons). Comparison for secondary outcomes did not show any difference between groups (data not shown), except for total days from birth to reach EFV of 120 ml kg − 1 day − 1 for the first time, which was © 2016 Nature America, Inc.
significantly longer in babies who were made NPO for treatment (mean (s.d.): 21.9 (7.9) days) compared with those where the feed volume was reduced (18.4 (8.5) days; P = 0.04). DISCUSSION In this retrospective cohort study including a relatively large number of preterm neonates, we found no association between variation in actual intake of EFV or preemptive changes in feeds during indomethacin treatment for PDA and subsequent incidence of NEC or other gastrointestinal and common prematurity-related morbidities. Lower EFV during indomethacin treatment was associated with a slight delay in reaching full enteral feeds as well as a reciprocal increase in duration of exposure to TPN and central venous catheter. Management of enteral feeds during pharmacological treatment for PDA in preterm neonates is a common clinical Journal of Perinatology (2016), 1 – 5
Indomethacin and feeding in preterm neonates D Louis et al
4 Table 3.
Outcomes in the study population
NEC NEC within 7 days of indomethacin treatment Surgical NEC Feeding intolerance Days to reach enteral feeds of 120 ml kg − 1 day − 1 Days to reach enteral feeds of 160 ml kg − 1 day − 1 Duration of TPN (days) Duration of central line (days) PDA ligation CLD ROP ⩾ stage III Total length of hospital stay (days) Death before discharge
Group A
Group B
Group C
Nil per oral
1–60 ml kg − 1 day − 1
460 ml kg − 1 day − 1
(N = 229)
(N = 142)
(N = 44)
14 5 6 11 22.8 29.9 25.1 20.8 31 171 25 75.9 27
(6.1) (2.1) (2.6) (4.9) (8.5) (17.2) (13.1) (13.2) (13.5) (74.7) (13.0) (43.1) (11.7)
11 2 2 15 20.5 29.4 23.8 19.1 24 114 25 77.4 19
(7.8) (1.4) (1.4) (10.6) (8.6) (19.3) (11.1) (11.0) (16.9) (80.9) (19.2) (42.8) (13.4)
2 2 0 4 16.8 23.5 21.4 16.4 5 38 3 71.3 2
(4.6) (4.6) (0) (9.3) (7.7) (12.3) (10.7) (11.9) (11.4) (86.4) (7.3) (38.4) (4.6)
P-value B vs A
C vs A
C vs B
0.54 0.99 0.72 0.04 0.02 0.80 0.32 0.19 0.38 0.17 0.14 0.74 0.78
0.69 0.18 0.59 0.28 o0.01 0.01 0.08 0.05 0.70 0.09 0.20 0.52 0.19
0.74 0.24 0.99 0.99 0.01 0.03 0.20 0.18 0.38 0.40 0.09 0.40 0.17
Abbreviations: CLD, chronic lung disease; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosus; ROP, retinopathy of prematurity; TPN, total parenteral nutrition. Values are mean (s.d.) or n (%).
scenario that is encountered on a day-to-day basis in NICUs. The literature to inform clinical practice is scant resulting in a wide variability of practice. Only two studies, both recently published, have examined the relationship between enteral feeding during indomethacin treatment and gastrointestinal complications in preterm neonates.7,12 The ductus arteriosus feed or fast with indomethacin or ibuprofen (DAFFII) trial was a controlled trial that randomized 177 neonates o 30 weeks gestational age to a NPO group vs trophic feeds (15 ml kg − 1 day − 1) when their PDAs were treated with indomethacin or ibuprofen.7 The primary outcome was the time to reach full enteral feeds (120 ml kg − 1 day − 1). This study reported a shorter duration of time to reach full enteral feeds in the trophic arm by a mean of 2.8 days without any significant difference in the rates of NEC (13% vs 10%) or other neonatal morbidities. Recently, Kelleher et al.12 performed a retrospective database study including 15 751 extremely low birth weight infants and investigated the association between trophic feeds during prophylactic indomethacin treatment given for the first 3 days of life and occurrence of spontaneous intestinal perforation (SIP). This study also found no significant association. The incidence of SIP in neonates who received feeds and prophylactic indomethacin was 3.3% compared with 4.9% in neonates who received indomethacin while being NPO.12 The results of these studies are keeping with our observation of a lack of association between EFV during indomethacin treatment and adverse gastrointestinal sequelae. To the best of our knowledge, ours is the first report providing outcome data for neonates who received larger feed volumes in comparison to those who were kept NPO (controls) during indomethacin treatment. This is important as a significant number of preterm neonates are expected to be receiving a considerable volume of enteral feeds at the time of initiation of treatment, especially if treated after the first week of age. In our study, we found a wide variability in practice among clinicians regarding management of enteral feeds during indomethacin treatment. Although some preferred to completely stop enteral feeds, others preemptively reduced EFV by an arbitrary chosen percentage. For the remaining neonates, EFV was held at the pretreatment level throughout the duration of the indomethacin course. This variability in practice at our institution is similar to a recently published survey of 281 US and 134 non-US neonatologists. The survey reported that 71% of US Journal of Perinatology (2016), 1 – 5
neonatologists and 22% of non-US neonatologists preferred stopping enteral feeding during the duration of indomethacin therapy.6 Although this variation in practice stems from a lack of research in the field, it allowed us to test the associations between EFV exposure during indomethacin treatment and relevant outcomes. The hypothetical concern of exposing the preterm gut to the additional metabolic ‘stress’ of enteral feeds during a potential period of gut underperfusion from indomethacin, especially in the setting of a preexisting chronic relative hypoperfusion secondary to ‘ductal steal’, prompts clinicians to electively discontinue or reduce enteral feeding during indomethacin treatment. These concerns are backed by small physiological studies that have demonstrated reduction in gut perfusion secondary to presence of PDA as well as treatment with indomethacin. However, the clinical correlation of these observations and their relationship with the ability of the preterm gut to handle enteral feeds is not known. Further, it is plausible that enteral feeds by virtue of inducing an increase in mesenteric blood flow (postprandial increase) may be protective against the vasoconstrictive action of indomethacin. A Doppler velocimetry nested study conducted in a subgroup of 34 neonates in the DAFFII trial reported changes in blood flow velocities in superior mesenteric artery after a 4 ml kg − 1 test feed in babies randomized to NPO or trophic feeds 18 to 24 h after completing indomethacin treatment.13 At 10 min postfeeding, infants with trophic feeds exhibited an increase in the peak systolic velocity and a marginal increase in mean velocity, while the NPO group did not show similar changes; whereas at 30 min after feeding, the increase in systolic, mean and diastolic velocities were similar in the two groups. Further, while evaluating the effect of prophylactic indomethacin and trophic feeds on SIP in a secondary analysis, Kelleher et al.12 reported a significantly lower incidence of SIP in neonates who received both indomethacin and feeds during the first 3 days of life in comparison to those who received neither. The practice of electively discontinuing enteral feeds is contrary to the important clinical goal of reaching full enteral feeds and minimizing exposure to parenteral nutrition. This was demonstrated by our study where we observed an inverse relationship between EFV during indomethacin treatment and time to reach full enteral feeds, duration of need for TPN therapy and central catheters. Other morbidities were similar across all groups. We © 2016 Nature America, Inc.
Indomethacin and feeding in preterm neonates D Louis et al
found the incidence of feeding intolerance to be lower in the NPO group than other groups, which was not surprising given the definition of feeding intolerance used in our study. All infants classified as having feeding intolerance in the NPO group had aspirates recorded as containing bile and/or blood, while in the other two groups neonates with two consecutive aspirates of 450% previous feed volume were also classified as having feeding intolerance. Our study has some limitations. First, this was a retrospective evaluation. Although the variability in EFV decisions prior to initiating indomethacin treatment may represent physician preferences, it may also be secondary to their perception of risks for individual patients. In an attempt to tease out the factors that may have governed such risk assessment and differential clinical management of enteral feeds, we compared neonates who were preemptively made NPO vs those for whom EFV was reduced or remained unchanged (that is, secondary analysis; data not provided) and found no difference in the cardio-respiratory indices of clinical stability, demographic features at the time of treatment, incidence of preexisting feeding intolerance in the 24 h prior to initiating treatment or the presence of any other gastrointestinal complication, suggesting that the differences in preemptive management of feeds observed is likely a reflection of practice variability among clinicians. However, because of a lack of specific documentation, we are unable to confirm this. Second, we cannot rule out the possibility of our study being underpowered to detect a significant difference in primary or secondary outcomes. Nevertheless, this is the largest study available so far on this topic and included all patients treated with indomethacin in our unit during the study period. Third, we are unable to assess the effect of nonsteroidal anti-inflammatory drugs other than indomethacin as these were seldom used in our unit. In many units, ibuprofen is now used as a primary agent for the medical treatment of PDA. Although the results of this study may not be generalizable to other nonsteroidal anti-inflammatory drugs, the use of ibuprofen is unlikely to have altered the outcome of this study as ibuprofen has been shown to cause even less vasoconstriction in splanchnic circulation in comparison to indomethacin.14 Fourth, our method of dividing neonates on enteral feeds during treatment into two separate groups was somewhat arbitrary. However, in an attempt to find a significant cutoff we did evaluate the entire cohort taking EFV as a continuous variable but identified no relationship between EFV and measured outcomes. Fifth, although we found no baseline difference between Group A and Group B, Group C has several important differences. Some of these differences such as preindomethacin feed volume and need for invasive ventilation at the time of treatment were expected given the late postnatal age at treatment in group C. Because of relatively small number of neonates who developed the primary outcome of NEC, we were limited in the number of variables that could be adjusted for in the analyses. We adjusted for what we perceived as key modifying variables (gestational age, postnatal age at treatment, need for surfactant and invasive ventilation at treatment) and demonstrated lack of inter-group differences by comparing as many factors and prematurity-related outcomes as feasible. Although lack of differences in these parameters suggest against major inter-group baseline differences, we are unable to confirm this assumption. Sixth, due to a lack of availability of routine pretreatment and posttreatment echocardiography, we are unable to investigate the effect of relevant PDA characteristics such as PDA size or flow reversal in abdominal aorta on the interaction between EFV during indomethacin treatment and NEC. For the same reason, we are unable to rule out interaction between EFV during treatment and success rate of indomethacin. However, the fact that total indomethacin exposure during NICU stay and rate of PDA ligation were similar between the groups suggest lack of
© 2016 Nature America, Inc.
significant impact of EFV and success rate of first course of idomethacin. Finally, we only collected data on type of enteral feeds (that is, formula vs expressed breast milk) during indomethacin and not throughout NICU stay. Although we are unable to provide a firm clinical recommendation, the results of this study questions the common clinical practice of reducing or withholding enteral feeds during indomethacin treatment in preterm neonates and provides the background and justification for future pragmatic randomized control trials. CONCLUSION There was no association between EFV during indomethacin treatment and adverse gastrointestinal outcomes in preterm neonates. Results of this study do not support the common clinical practice of preemptive discontinuation/reduction in enteral feed volume in preterm neonates receiving treatment with indomethacin for PDA. Further large sample size multicenter randomized controlled trials would be necessary to provide definite evidence. CONFLICT OF INTEREST The authors declare no conflict of interest.
ACKNOWLEDGEMENTS We thank Junmin Yang, statistician in Mother and Infant Care Program of Mount Sinai Hospital, Toronto, ON, Canada for performing the statistical analysis for this study and Josie Chundamala, scientific editor at Mother and Infant Care Program of Mount Sinai Hospital, Toronto, ON, Canada for proofreading the manuscript.
REFERENCES 1 McCurnin D, Clyman RI. Effects of a patent ductus arteriosus on postprandial mesenteric perfusion in premature baboons. Pediatrics 2008; 122: e1262–e1267. 2 Coombs RC, Morgan ME, Durbin GM, Booth IW, McNeish AS. Gut blood flow velocities in the newborn: effects of patent ductus arteriosus and parenteral indomethacin. Arch Dis Child 1990; 65: 1067–1071. 3 Laudignon N, Chemtob S, Bard H, Aranda JV. Effect of indomethacin on cerebral blood flow velocity of premature newborns. Biol Neonate 1988; 54: 254–262. 4 Christmann V, Liem KD, Semmekrot BA, van de Bor M. Changes in cerebral, renal and mesenteric blood flow velocity during continuous and bolus infusion of indomethacin. Acta Paediatr 2002; 91: 440–446. 5 Fowlie PW, Davis PG, McGuire W. Prophylactic intravenous indomethacin for preventing mortality and morbidity in preterm infants. Cochrane Database of Syst Rev 2010, Issue 7. Art. No.: CD000174. doi:10.1002/14651858.CD000174.pub2. 6 Jhaveri N, Soll RF, Clyman RI. Feeding practices and patent ductus arteriosus ligation preferences-are they related? Am J Perinatol 2010; 27: 667–674. 7 Clyman R, Wickremasinghe A, Jhaveri N, Hassinger DC, Attridge JT, Sanocka U et al. Enteral feeding during indomethacin and ibuprofen treatment of a patent ductus arteriosus. J Pediatr 2013; 163(2): 406–411.e4. 8 Fenton TR, Kim JH. A systematic review and meta-analysis to revise the Fenton growth chart for preterm infants. BMC Pediatr 2013; 13: 59. 9 Kliegman RM. Neonatal necrotizing enterocolitis: implications for an infectious disease. Pediatr Clin North Am 1979; 26: 327–344. 10 Jobe AH, Bancalari E. Bronchopulmonary dysplasia. Am J Respir Crit Care Med 2001; 163: 1723–1729. 11 International Committee for the Classification of Retinopathy of Prematurity. The International Classification of Retinopathy of Prematurity revisited. Arch Ophthalmol 2005; 123: 991–999. 12 Kelleher J, Salas AA, Bhat R, Ambalavanan N, Saha S, Stoll BJ et al. Prophylactic indomethacin and intestinal perforation in extremely low birth weight infants. Pediatrics 2014; 134: e1369–e1377. 13 Yanowitz TD, Reese J, Gillam-Krakauer M, Cochran CM, Jegatheesan P, Lau J et al. Superior mesenteric artery blood flow velocities following medical treatment of a patent ductus arteriosus. J Pediatr 2014; 164: 661–663. 14 Pezzati M, Vangi V, Biagiotti R, Bertini G, Cianciulli D, Rubaltelli FF. Effects of indomethacin and ibuprofen on mesenteric and renal blood flow in preterm infants with patent ductus arteriosus. J Pediatr 1999; 135: 733–738.
Journal of Perinatology (2016), 1 – 5
5
