Prophylactic cyclo-oxygenase inhibitor drugs for the prevention of morbidity and mortality in preterm infants: a network meta-analysis

Abstract

Background: Patent ductus arteriosus (PDA) is associated with significant morbidity and mortality in preterm infants. Cyclooxygenase inhibitors (COX-I) may prevent PDA-related complications. Controversy exists on which COX-I drug is the most effective and has the best safety profile in preterm infants. Objectives: To compare the effectiveness and safety of prophylactic COX-I drugs and 'no COXI prophylaxis' in preterm infants using a Bayesian network meta-analysis (NMA). Search methods: Searches of Cochrane CENTRAL via Wiley, OVID MEDLINE and Embase via Elsevier were conducted on 9 December 2021. We conducted independent searches of clinical trial registries and conference abstracts; and scanned the reference lists of included trials and related systematic reviews. Selection criteria: We included randomised controlled trials (RCTs) that enrolled preterm or low birth weight infants within the first 72 hours of birth without a prior clinical or echocardiographic diagnosis of PDA and compared prophylactic administration of indomethacin or ibuprofen or acetaminophen versus each other, placebo or no treatment. Data collection and analysis: We used the standard methods of Cochrane Neonatal. We used the GRADE NMA approach to assess the certainty of evidence derived from the NMA for the following outcomes: severe intraventricular haemorrhage (IVH), mortality, surgical or interventional PDA closure, necrotizing enterocolitis (NEC), gastrointestinal perforation, chronic lung disease (CLD) and cerebral palsy (CP). Main results: We included 28 RCTs (3999 preterm infants). Nineteen RCTs (n = 2877) compared prophylactic indomethacin versus placebo/no treatment, 7 RCTs (n = 914) compared prophylactic ibuprofen versus placebo/no treatment and 2 RCTs (n = 208) compared prophylactic acetaminophen versus placebo/no treatment. Nine RCTs were judged to have high risk of bias in one or more domains.We identified two ongoing trials on prophylactic acetaminophen. Bayesian random-effects NMA demonstrated that prophylactic indomethacin probably led to a small reduction in severe IVH (network RR 0.66, 95% Credible Intervals [CrI] 0.49 to 0.87; absolute risk difference [ARD] 43 fewer [95% CrI, 65 fewer to 16 fewer] per 1000; median rank 2, 95% CrI 1-3; moderate-certainty), a moderate reduction in mortality (network RR 0.85, 95% CrI 0.64 to 1.1; ARD 24 fewer [95% CrI, 58 fewer to 16 more] per 1000; median rank 2, 95% CrI 1-4; moderate-certainty) and surgical PDA closure (network RR 0.40, 95% CrI 0.14 to 0.66; ARD 52 fewer [95% CrI, 75 fewer to 30 fewer] per 1000; median rank 2, 95% CrI 1-2; moderate-certainty) compared to placebo. Prophylactic indomethacin resulted in trivial difference in NEC (network RR 0.76, 95% CrI 0.35 to 1.2; ARD 16 fewer [95% CrI, 42 fewer to 13 more] per 1000; median rank 2, 95% CrI 1-3; high-certainty), gastrointestinal perforation (network RR 0.92, 95% CrI 0.11 to 3.9; ARD 4 fewer [95% CrI, 42 fewer to 137 more] per 1000; median rank 1, 95% CrI 1-3; moderate-certainty) or CP (network RR 0.97, 95% CrI 0.44 to 2.1; ARD 3 fewer [95% CrI, 62 fewer to 121 more] per 1000; median rank 2, 95% CrI 1-3; low-certainty) and may result in a small increase in CLD (network RR 1.10, 95% CrI 0.93 to 1.3; ARD 36 more [95% CrI, 25 fewer to 108 more] per 1000; median rank 3, 95% CrI 1-3; low-certainty). Prophylactic ibuprofen probably led to a small reduction in severe IVH (network RR 0.69, 95% CrI 0.41 to 1.14; ARD 39 fewer [95% CrI, 75 fewer to 18 more] per 1000; median rank 2, 95% CrI 1-4; moderate-certainty) and moderate reduction in surgical PDA closure (network RR 0.24, 95% CrI 0.06 to 0.64; ARD 66 fewer [95% CrI, from 82 fewer to 31 fewer] per 1000; median rank 1, 95% CrI 1-2; moderate-certainty) compared to placebo. Prophylactic ibuprofen may result in moderate reduction in mortality (network RR 0.83, 95% CrI 0.57 to 1.2; ARD 27 fewer [95% CrI, from 69 fewer to 32 more] per 1000; median rank 2, 95% CrI 1-4; low-certainty) and leads to trivial difference in NEC (network RR 0.73, 95% CrI 0.31 to 1.4; ARD 18 fewer [95% CrI, from 45 fewer to 26 more] per 1000; median rank 1, 95% CrI 1-3; high-certainty), or CLD (network RR 1.00, 95% CrI 0.83 to 1.3; ARD 0 fewer [95% CrI, from 61 fewer to 108 more] per 1000; median rank 2, 95% CrI 1-3; low-certainty). The evidence is very uncertain on effect of ibuprofen on gastrointestinal perforation (network RR 2.6, 95% CrI 0.42 to 20.0; ARD 76 more [95% CrI, from 27 fewer to 897 more] per 1000; median rank 3, 95% CrI 1-3; very low-certainty). The evidence is very uncertain on the effect of prophylactic acetaminophen on severe IVH (network RR 1.17, 95% CrI 0.04 to 55.2; ARD 22 more [95% CrI, from 122 fewer to 1000 more] per 1000; median rank 4, 95% CrI 1-4; very low-certainty), mortality (network RR 0.49, 95% CrI 0.16 to 1.4; ARD 82 fewer [95% CrI, from 135 fewer to 64 more] per 1000; median rank 1, 95% CrI 1-4; very low-certainty), or CP (network RR 0.36, 95% CrI 0.01 to 6.3; ARD 70 fewer [95% CrI, from 109 fewer to 583 more] per 1000; median rank 1, 95% CrI 1-3; very low-certainty). In summary, based on ranking statistics, both indomethacin and ibuprofen were equally effective (median ranks 2 respectively) in reducing severe IVH and mortality. Ibuprofen (median rank 1) was more effective than indomethacin in reducing surgical PDA ligation (median rank 2). However, no statistically-significant differences were observed between the COX-I drugs for any of the relevant outcomes. Authors' conclusions: Prophylactic indomethacin probably results in a small reduction in severe IVH and moderate reduction in mortality and surgical PDA closure (moderate-certainty), may result in a small increase in CLD (low-certainty) and results in trivial differences in NEC (high-certainty), gastrointestinal perforation (moderate-certainty) and cerebral palsy (low-certainty). Prophylactic ibuprofen probably results in a small reduction in severe IVH and moderate reduction in surgical PDA closure (moderate-certainty), may result in a moderate reduction in mortality (low-certainty) and trivial differences in CLD (low-certainty) and NEC (high-certainty). The evidence is very uncertain about the effect of acetaminophen on any of the clinically-relevant outcomes.