Abstract
The use of prophylactic indomethacin in very preterm infants is controversial. The last randomized controlled trial (RCTs) to study this therapy enrolled infants over 20 years ago. More recently, observational studies have investigated the association between exposure to prophylactic indomethacin and neonatal morbidities and mortality. We performed a systematic review and meta-analysis of these studies for the outcomes of death and bronchopulmonary dysplasia (BPD). Two observational studies involving a total of 11,289 very preterm infants were suitable for meta-analysis. The pooled data showed that prophylactic indomethacin was not associated with higher or lower risk-adjusted odds of death or BPD (0.93, 95% CI 0.76–1.13) and of BPD among survivors (0.94, 95% CI 0.78–1.12). However, there was a weak association between indomethacin prophylaxis and decreased risk-adjusted odds of mortality (0.81, 95% CI 0.66–0.98). It is unknown whether this finding resulted from unmeasured confounding, chance, or represents a true benefit. To confirm the hypothesis that prophylactic indomethacin has a small effect on mortality in the current era, a contemporary RCT would need to enroll over 3,500 very immature infants at high risk of death.
Introduction
Since the earliest descriptions of ductal closure following administration of indomethacin, nearly 3000 very preterm infants have been enrolled in 19 randomized controlled trials (RCT) of prophylactic indomethacin.1–4 The trials have shown that administration of indomethacin soon after birth reduces the incidence of severe peri- and intraventicular hemorrhage and symptomatic patent ductus arteriosus (PDA).4 Despite these benefits, no difference was observed in the rates of mortality, bronchopulmonary dysplasia (BPD), or adverse long-term neurodevelopment.4 The lack of evidence for a treatment effect on BPD is noteworthy, because of the strong association between the presence of a PDA and the subsequent development of BPD.5–7 However, a key limitation of the RCTs is that most defined BPD as the use of supplemental oxygen at 28 days of life.4 Only the Trial of Indomethacin Prophylaxis in Preterms (TIPP) assessed supplemental oxygen use at 36 weeks post-menstrual age (PMA).4, 8
Importantly, the last participants in the published RCTs were enrolled over 20 years ago.4, 8 Although by most estimates, rates of BPD have not improved during that time span, survival among the most extremely preterm infants has increased.9, 10 Moreover, modern obstetrical and neonatal care includes greater use of antenatal corticosteroids and exogenous surfactant, and less use of early invasive mechanical ventilation.9 Whether the safety and efficacy of prophylactic indomethacin differ in the current era is unknown. Over the past 2 decades, several observational studies were conducted with varying methodologies and rigor and have produced conflicting results.11–19 In the absence of contemporary trial data, clinicians may increasingly be tempted to use data from these observational studies to assess the risks and benefits of prophylactic indomethacin and inform their practice.20 We undertook the present systematic review and meta-analysis to synthesize data from these observational studies and evaluate the association between exposure to prophylactic indomethacin and the outcomes of death and BPD. We compare these results to the existing pooled data from randomized clinical trials.
Review Methods
Two authors (EAJ, EEF) independently conducted a systematic search of PubMed according to the strategy summarized in Box 1. Studies selected for inclusion compared the risk of developing BPD, defined as supplemental use of oxygen at 36 weeks PMA, between infants who received prophylactic indomethacin (treatment on the day of birth or day after birth) and contemporary untreated controls. The meta-analysis was limited to studies reporting risk-adjusted estimates of effect to reduce the influence of unmeasured confounding. Separate meta-analyses were performed for the composite outcome of death or BPD and the individual outcomes of BPD among survivors and mortality. To pool data from multiple observational studies, we obtained adjusted log[odds ratio] and log[standard error] values from the respective corresponding authors. Each meta-analysis was conducted using a random effects models and the generic inverse variance method in RevMan version 5.3 (The Cochrane Collaborative, Copenhagen, Denmark).
To compare the observational and RCT data, we recreated the forest plots for the outcomes of BPD and mortality that were included in the most recent Cochrane review on the use of prophylactic indomethacin.4 The composite outcome of death or BPD was not reported in this review.4 In addition, we conducted an exploratory meta-analysis of all 10 trials reporting BPD as an outcome (defined as supplemental oxygen use at 28 days or 36 weeks PMA). Finally, we performed cumulative meta-analyses to assess whether the measured effect of prophylactic indomethacin on BPD and mortality changed over time. The cumulative plots were generated using Stata/SE version 13.1 (StataCorp LP, College Station, TX).
Literature Search Results
The PubMed search identified 1101 potential articles for inclusion, of which 4 fulfilled the pre-specified inclusion/exclusion criteria and 2 were suitable for inclusion in the meta-analysis (Figure 1).11, 12, 14, 15 The studies by Laughon et al. and Cordero et al. were excluded from the meta-analysis because neither reported adjusted estimates of effect for the association between treatment with prophylactic indomethacin and the risk for developing BPD.14, 15 Laughon et al. performed adjusted analyses for their comparison of infants treated with indomethacin after the first day of life and those who did not undergo treatment for a PDA.15 However, infants treated with prophylactic indomethacin were not included owing to differences in the demographic characteristics of these infants relative to the other evaluated babies.15 Cordero et al. compared 167 infants born < 1000g who were treated with prophylactic indomethacin to 167 untreated infants matched by year of birth, birth weight, gestational age, and gender.14 The unadjusted rates of BPD were similar between the groups, however the authors did not report adjustment for potential confounding variables.14
Figure 1.
Systematic review search flow diagram
The methods and results of the 2 studies included in the meta-analysis are summarized in Table 1. In total, the studies included 11,289 very preterm infants cared for in hospitals participating in the Canadian Neonatal Network (CNN) and the US National Institute of Child Health and Human Development Neonatal Research Network (NRN), respectively.11, 12 Among the study infants, 2,856 (25.3%) were treated with prophylactic indomethacin, but treatment rates varied between the two cohorts (CNN: 7.8%; NRN: 33.0%).11, 12
Table 1.
Summary of observational studies included in the meta-analysis
| Author Year |
Study Type | Population | Risk-Adjustment Covariates | Summary of Findings | |||
|---|---|---|---|---|---|---|---|
| Maternal | Infant | Death or BPD | BPD among survivors |
Death by study endpoint* |
|||
| Rolnitsky11 2015 | Retrospective cohort study of prospectively collected data |
|
|
|
PI: 180/269 (66.9%) No PI: 1753/3189 (55.0%) | PI: 122/213 (57.3%) No PI: 1181/2591 (45.6%) | PI: 63/269 (23.4%) No PI: 601/3189 (18.8%) |
| Unadjusted OR: 1.66 (95% CI 1.27, 2.16) | Unadjusted OR: 1.60 (95% CI 1.21, 2.12) | Unadjusted OR: 1.32 (95% CI 0.98, 1.77) | |||||
| Adjusted OR: 1.08 (95% CI 0.77, 1.52) | Adjusted OR: 1.09 (95% CI 0.76, 1.56) | Adjusted OR: 0.81 (95% CI 0.55, 1.19) | |||||
| Jensen12 2017 | Retrospective cohort study of prospectively collected data |
|
|
|
PI: 1404/2587 (54.3%) No PI: 2680/5244 (51.1%) | PI: 960/2143 (44.8%) No PI: 2042/4606 (44.3%) | PI: 444/2587 (17.2%) No PI: 638/5244 (12.2%) |
| Unadjusted OR: 1.14 (95% CI 1.03, 1.25) | Unadjusted OR: 1.02 (95% CI 0.92, 1.13) | Unadjusted OR: 1.50 (95% CI 1.31, 1.71) | |||||
| Adjusted OR: 0.87 (95% CI 0.71, 1.05) | Adjusted OR: 0.89 (95% CI 0.72, 1.10) | Adjusted OR: 0.80 (95% CI 0.64, 1.01) | |||||
Abbreviations: BW, birth weight; CPR, cardiopulmonary resuscitation; DR, delivery room; GA, gestational age; OR, odds ratio; PI, prophylactic indomethacin; SGA, small for gestational ages
Rolnitsky et al. defined mortality as death by hospital discharge. Jensen et al. defined mortality as death 36 weeks postmenstrual age.
Meta-analyses of Observational Studies
The observational data meta-analyses for the 3 study outcomes are shown in Figure 2. The risk of death or BPD was similar between infants who received prophylactic indomethacin and those who did not when assessed using pooled unadjusted data (odds ratio 1.34, 95% CI 0.93, 1.94) and risk-adjusted odds ratios (0.93, 95% CI 0.76, 1.13; Figure 2). There was also no difference between the groups for the outcome of BPD among survivors in the unadjusted and adjusted analyses (Figure 2). The pooled unadjusted odds of mortality by study endpoint were higher among infants exposed to prophylactic indomethacin than in untreated infants (1.46, 95% CI 1.30, 1.65; Figure 2). However, when the adjusted odds ratios were considered, the direction of effect was reversed, suggesting that the use of prophylactic indomethacin was associated with lower risk-adjusted odds of mortality (0.81, 95% CI 0.66, 0.98; Figure 2).
Figure 2.
Meta-analysis of observational studies for the composite outcome of death or BPD (A), BPD among survivors (B), and death by study endpoint (C)
Similarities and Differences between the Results of the Observational Studies and RCTs
The meta-analyses of RCTs reporting BPD and death as outcomes are shown in Figures 3 and 4, respectively. Ten trials assessed the risk of BPD in 2,021 preterm infants (BPD at 28 days: n=1,022; BPD at 36 weeks PMA: n=999).4 Consistent with the observational data, prophylactic indomethacin did not reduce the risk for BPD when diagnosed at 28 days (relative risk 1.08, 95% CI 0.92, 1.26), 36 weeks PMA (relative risk 1.06, 95% CI 0.92, 1.22), or when data from the two time points were combined (relative risk 1.07, 95% CI 0.96, 1.22). The cumulative meta-analysis confirms that no evidence for beneficial effects of prophylactic indomethacin on BPD emerged over time (Figure 3).
Figure 3.
“Cochrane style” and cumulative meta-analyses of randomized controlled trial of prophylactic indomethacin reporting BPD as an outcome. Plots created using data from Fowlie PW, et al. Cochrane Database Syst Rev. 2010:CD000174.
Figure 4.
“Cochrane style” and cumulative meta-analyses of randomized controlled trial of prophylactic indomethacin reporting death (bottom) by study endpoint as an outcome. Plots created using data from Fowlie PW, et al. Cochrane Database Syst Rev. 2010:CD000174.
In contrast to the meta-analysis of observational data indicating reduced mortality risk associated with prophylactic indomethacin (Figure 2), the trial results show no benefit (Figure 4). When mortality was assessed at the latest study follow-up, the relative risk closely approximates the point of equivalence (0.96, 95% CI 0.81, 1.12). The cumulative meta-analysis of the first 17, predominantly small trials, was consistent with a potential benefit of prophylactic indomethacin on mortality (Figure 4). This trend disappeared after the inclusion of data from TIPP, which was by far the largest of the 18 RCTs and the only trial that included deaths during post-discharge follow-up in the ascertainment of the primary outcome.8
Comment
The results of the present meta-analysis of contemporary observational studies conducted in 2 large neonatal networks suggest that prophylactic indomethacin is not associated with an increased or decreased risk of BPD. This finding is consistent with the available RCT data. For the outcome of mortality, however, the pooled data from observational studies and RCTs diverge. While the RCTs show no evidence of benefit,4 the pooled analysis of 2 observational studies suggests that the use of prophylactic indomethacin may be associated with a small reduction in mortality risk. It is unknown whether this finding represents a true benefit, or a spurious result due to chance or unmeasured residual confounding in the observational studies. One potential contributor to this result is survival bias – an infant must live long enough to receive indomethacin. Rolnitsky et al. excluded infants who received palliative care only and Jensen et al. excluded those who died in the first 12 hours of life.11, 12 While these exclusions likely omit some critically ill infants, it is possible that others lived long enough to meet study eligibility but died prior to receiving planned treatment with indomethacin. Treatment selection bias must also be considered. In particular, some clinicians who commonly use prophylactic indomethacin in extremely preterm infants may withhold the medication in babies with acute, life-threatening disease. Although both observational studies included in the meta-analysis performed risk adjustment for likely confounding variables, residual differences in illness severity may still contribute to the observed advantage of infants exposed to prophylactic indomethacin.
Because the association between prophylactic indomethacin and reduced mortality risk is at odds with prior evidence from RCTs, this meta-analysis of recent observational studies generates an intriguing hypothesis. A contemporary RCT would be required to confirm this finding before widespread use of prophylactic indomethacin could be recommended to prevent mortality. Assuming the same mortality rate among untreated infants in the pooled observational data (14.7%) occurred in placebo treated infants, a trial would need to enroll over 3,500 very preterm infants at high risk for mortality to replicate the result shown in the observational meta-analysis (216/1768 [12.2%] vs. 260/1768 [14.7%]; odds ratio 0.81, 95% CI 0.66–0.98).
Conclusion
In a meta-analysis of 2 large, contemporary observational studies, we found no evidence of an association between prophylactic indomethacin and higher or lower risk for the composite outcome of death or BPD or BPD among survivors. These results, combined with the published RCT data showing that prophylactic indomethacin does not reduce BPD risk, represent over 13,000 preterm infants studied across 4 decades. In contrast to the RCT data, the meta-analysis of observational studies found a small reduction in the risk-adjusted odds of mortality associated with indomethacin prophylaxis. Whether this finding arose by chance, resulted from unmeasured residual confounding, or represents a true benefit in the modern era is unknown. This question is not likely to be answered definitely unless a large randomized trial with adequate statistical power is performed.
Box 1. Observational study systematic review and meta-analysis search strategy.
Search terms
PubMed queried on 01/01/2018 for human studies identified using the following search phrase: (indomethacin) AND (infant OR neonate) AND (preterm or premature).
Inclusion criteria
Original, observational research studies comparing preterm infants (gestational age < 32 weeks and/or birth weight < 1500g) treated with prophylactic indomethacin (treatment initiated on the day of birth or day after birth) to concurrent, untreated controls.
Report bronchopulmonary dysplasia diagnosed at 36 weeks PMA as an outcome
Abstract available in English
For inclusion in the meta-analysis – studies must report risk-adjusted estimates of the association between prophylactic indomethacin and the study outcomes.
Exclusion criteria
Randomized trials
Studies using historical (non-concurrent) controls
Studies using only select infants as control (e.g. those with echocardiography diagnosed PDA)
Studies likely to have a high degree of bias or non-generalizability (e.g. studies evaluating primarily outborn infants)
Acknowledgments
We are grateful to Prakesh S Shah MSc, MBBS, MD, DCH, MRCP, FRCPC (Canadian Neonatal Network), Marie Gantz, PhD (RTI International), and Benjamin Carper, MS (RTI International) for providing additional observational data used in these meta-analyses.
Funding Source: EAJ was supported by a grant from the National Health, Lung, and Blood Institute (K23HL136843). EEF was supported by a grant from the Eunice Kennedy Shriver National Institutes of Child Health and Human Development (K23HD084727).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: The authors have no conflicts of interest or relevant financial interests to disclose
References
- 1.Elliott R, Starling M, Neutze J. Medical manipulation of the ductus arteriosus. Lancet. 1975;1:140–2. doi: 10.1016/s0140-6736(75)91432-4. [DOI] [PubMed] [Google Scholar]
- 2.Friedman W, Hirschklau M, Printz M, Pitlick P, Kirkpatrick S. Pharmacologic closure of patent ductus arteriosus in the premature infant. N Engl J Med. 1976;295:526–9. doi: 10.1056/NEJM197609022951003. [DOI] [PubMed] [Google Scholar]
- 3.Heymann M, Rudolph A, Silverman N. Closure of the ductus arteriosus in premature infants by inhibition of prostaglandin synthesis. N Engl J Med. 1976;295:530–3. doi: 10.1056/NEJM197609022951004. [DOI] [PubMed] [Google Scholar]
- 4.Fowlie PW, Davis PG, McGuire W. Prophylactic intravenous indomethacin for preventing mortality and morbidity in preterm infants. Cochrane Database Syst Rev. 2010 doi: 10.1002/14651858.CD000174.pub2. CD000174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Marshall DD, Kotelchuck M, Young TE, Bose CL, Kruyer L, O'Shea TM, et al. Risk factors for chronic lung disease in the surfactant era: a North Carolina population-based study of very low birth weight infants. Pediatrics. 1999;104:1345–50. doi: 10.1542/peds.104.6.1345. [DOI] [PubMed] [Google Scholar]
- 6.Oh W, Poindexter BB, Perritt R, Lemons JA, Bauer CR, Ehrenkranz RA, et al. Association between fluid intake and weight loss during the first ten days of life and risk of bronchopulmonary dysplasia in extremely low birth weight infants. J Pediatr. 2005;147:786–90. doi: 10.1016/j.jpeds.2005.06.039. [DOI] [PubMed] [Google Scholar]
- 7.Palta M, Gabbert D, Weinstein MR, Peters ME. Multivariate assessment of traditional risk factors for chronic lung disease in very low birth weight neonates. The Newborn Lung Project. J Pediatr. 1991;119:285–92. doi: 10.1016/s0022-3476(05)80746-2. [DOI] [PubMed] [Google Scholar]
- 8.Schmidt B, Davis P, Moddemann D, Ohlsson A, Roberts RS, Saigal S, et al. Long-term effects of indomethacin prophylaxis in extremely-low-birth-weight infants. N Engl J Med. 2001;344:1966–72. doi: 10.1056/NEJM200106283442602. [DOI] [PubMed] [Google Scholar]
- 9.Stoll BJ, Hansen NI, Bell EF, Walsh MC, Carlo WA, Shankaran S, et al. Trends in care practices, morbidity, and mortality of extremely preterm neonates, 1993–2012. JAMA. 2015;314:1039–51. doi: 10.1001/jama.2015.10244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Horbar JD, Edwards EM, Greenberg LT, Morrow KA, Soll RF, Buus-Frank ME, et al. Variation in performance of neonatal intensive care units in the United States. JAMA Pediatr. 2017;171:e164396. doi: 10.1001/jamapediatrics.2016.4396. [DOI] [PubMed] [Google Scholar]
- 11.Rolnitsky A, Lee S, Piedbouf B, Harrison A, Shah P, Network. CN Prophylactic interventions in neonatology: how do they fare in real life? Am J Perinatol. 2015;32:1098–104. doi: 10.1055/s-0035-1556882. [DOI] [PubMed] [Google Scholar]
- 12.Jensen EA, Dysart KC, Gantz MG, Carper B, Higgins RD, Keszler M, et al. Association between use of prophylactic indomethacin and the risk for bronchopulmonary dysplasia in extremely preterm infants. J Pediatr. 2017;186:34–40. e2. doi: 10.1016/j.jpeds.2017.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Nelin T, Pena E, Giacomazzi T, Lee S, Logan J, Moallem M, et al. Outcomes following indomethacin prophylaxis in extremely preterm infants in an all-referral NICU. J Perinatol. 2017;37:932–7. doi: 10.1038/jp.2017.71. [DOI] [PubMed] [Google Scholar]
- 14.Cordero L, Nankervis C, Delooze D, Giannone P. Indomethacin prophylaxis or expectant treatment of patent ductus arteriosus in extremely low birth weight infants? J Perinatol. 2007;27:158–63. doi: 10.1038/sj.jp.7211659. [DOI] [PubMed] [Google Scholar]
- 15.Laughon M, Bose C, Clark R. Treatment strategies to prevent or close a patent ductus arteriosus in preterm infants and outcomes. J Perinatol. 2007;27:164–70. doi: 10.1038/sj.jp.7211662. [DOI] [PubMed] [Google Scholar]
- 16.Gudmundsdottir A, Johansson S, Håkansson S, Norman M, Källen K, Bonamy A. Timing of pharmacological treatment for patent ductus arteriosus and risk of secondary surgery, death or bronchopulmonary dysplasia: a population-based cohort study of extremely preterm infants. Neonatology. 2015;107:87–92. doi: 10.1159/000367887. [DOI] [PubMed] [Google Scholar]
- 17.Mondal T, Chaudhuri D, Li B, Shivananda S, Dutta S. Prophylactic indomethacin worsens short-term respiratory outcomes in extremely low-birth-weight infants. Am J Perinatol. 2014;31:61–7. doi: 10.1055/s-0033-1334449. [DOI] [PubMed] [Google Scholar]
- 18.Liebowitz M, Clyman R. Prophylactic indomethacin compared with delayed conservative management of the patent ductus arteriosus in extremely preterm infants: effects on neonatal outcomes. J Pediatr. 2017;187:119–26. doi: 10.1016/j.jpeds.2017.03.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Madan JC, Kendrick D, Hagadorn JI, Frantz ID, National Institute of Child Health and Human Development Neonatal Research Network Patent ductus arteriosus therapy: impact on neonatal and 18-month outcome. Pediatrics. 2009;123:674–81. doi: 10.1542/peds.2007-2781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Reese J, Shelton E, Slaughter J, McNamara P. Prophylactic Indomethacin Revisited. J Pediatr. 2017;186:11–4. e1. doi: 10.1016/j.jpeds.2017.03.036. [DOI] [PMC free article] [PubMed] [Google Scholar]