Use of Prophylactic Indomethacin in Preterm Infants: A Systematic Review and Meta-analysis

Background Prophylactic indomethacin has been widely used as an effective intervention for reducing mortalities and morbidities in preterm infants including the cardiopulmonary and neurodevelopmental morbidities as intraventricular hemorrhage (IVH), but many studies have reported contraindicated outcomes of its signicance. Therefore, we aim to systematically review and meta-analyze the data of prophylactic indomethacin on preterm infants. Methods Our to include assessed the use of prophylactic indomethacin in 2020. The nal list of our included studies is comprised of 22 randomized trials and cohort studies. Our analysis of observational data showed that intubation in the delivery room/rst day (74%), bronchopulmonary dysplasia (BPD) (33.2%), and patent ductus arteriosus (PDA) (32.2%) were the most prevalent outcomes in infants that received prophylactic indomethacin. Among all the studies outcomes, the only signicant favorable outcome was lowering the rate of PDA (P< 0.001) while no signicance was recorded with BPD, pulmonary hemorrhage, neurodevelopmental delays (IVH), mortality, length of hospital stays, and time spent on ventilators outcomes (P = 0.106, 0.123, 0.460, 0.340, 0.625, and 0.732, respectively). Moreover, necrotizing enterocolitis was signicantly increased when applying prophylactic indomethacin in these infants (P< 0.001). The use in should be generally discouraged due to its neutral effect on most of the mortality and morbidity outcomes and the signicant occurrence of its adverse events despite the positive effect on ductal closure. The rates of mortalities, bronchopulmonary dysplasia (BPD), or long-term neurodevelopmental decays reportedly seem to have been not affected. A previously published large RCT advised against using indomethacin as a prophylactic agent (15). Although the study showed favorable outcomes in terms of reducing incidence rates of PDA, PDA ligation, IVH, and pulmonary hemorrhage, no improvement regarding the incidence of death and neurodevelopmental disorders rates has been found. Therefore, in this systematic review, we aim to analyze the data of previously published investigations on the use of prophylactic indomethacin in preterm infants. the dosages injection, denition between


Abstract
Background Prophylactic indomethacin has been widely used as an effective intervention for reducing mortalities and morbidities in preterm infants including the cardiopulmonary and neurodevelopmental morbidities as intraventricular hemorrhage (IVH), but many studies have reported contraindicated outcomes of its signi cance. Therefore, we aim to systematically review and meta-analyze the data of prophylactic indomethacin on preterm infants.

Methods
Our systematic search included the following databases: Pubmed, Google Scholar, Scopus, Web of Science, The New York Academy of Medicine (NYAM), Virtual health library (VHL), and the System for Information on Grey Literature in Europe (SIGLE) to include studies that assessed the use of prophylactic indomethacin in preterm infants until August 12, 2020.

Results
The nal list of our included studies is comprised of 22 randomized trials and cohort studies. Our analysis of observational data showed that intubation in the delivery room/ rst day (74%), bronchopulmonary dysplasia (BPD) (33.2%), and patent ductus arteriosus (PDA) (32.2%) were the most prevalent outcomes in infants that received prophylactic indomethacin. Among all the studies outcomes, the only signi cant favorable outcome was lowering the rate of PDA (P< 0.001) while no signi cance was recorded with BPD, pulmonary hemorrhage, neurodevelopmental delays (IVH), mortality, length of hospital stays, and time spent on ventilators outcomes (P = 0.106, 0.123, 0.460, 0.340, 0.625, and 0.732, respectively). Moreover, necrotizing enterocolitis was signi cantly increased when applying prophylactic indomethacin in these infants (P< 0.001).

Conclusion
The use of prophylactic indomethacin in preterm infants should be generally discouraged due to its neutral effect on most of the mortality and morbidity outcomes and the signi cant occurrence of its adverse events despite the positive effect on ductal closure.

Structured Summary
Background Many authors believe neurologic disabilities have been associated with preterm labor presumed due to either over circulation or under circulation secondary to patent ductus arteriosus.

Objectives
Systemically revied all published RCT and meta-analysis to nd if prophylactic indomethacin can decrease CNS complications.

Data sources
Following the Preferred Reporting Items for Systematic Review and Meta-analyses statement (PRISMA) recommendations, we performed this systematic review and meta-analysis (32). A systematic electronic database search was conducted for relevant studies published, from inception to 12th August 2020, in seven databases: Pubmed, Google Scholar, Scopus, Web of Science, The New York Academy of Medicine (NYAM), Virtual health library (VHL), and the System for Information on Grey Literature in Europe (SIGLE).

Study eligibility criteria, participants, and interventions
Participants were any preterm infants, the intervention was the prophylactic indomethacin, the comparison was placebo or no treatment groups, and all possible outcomes were included. The systematic search was followed by a manual search in references of the included papers to include missed papers. We included all original studies that assessed the use of prophylactic indomethacin in preterm infants.
Papers were excluded if there were one of the following exclusion criteria: I. non-original studies II. in vitro or animal studies; III. data duplication, overlapping or unreliably extracted or incomplete data; IV. abstract only articles, reviews, thesis, books, conference papers, or articles without available full texts (conferences, editorials, author response, letters, and comments.

Risk of bias
The revised Cochrane quality assessment tool (RoB 2) was used to determine the quality of randomized studies and the risk of bias in non-randomized studies -of interventions (ROBINS-I) tool for non-randomized studies.

Data analysis
All data were analyzed using R software version 4.0.2. Using the "meta" package, odds ratios (OR) and prevalence rates of different outcomes were calculated. The corresponding 95% con dence intervals (CI) of pooled effect size were calculated using a xed-effects or random-effects, according to heterogeneity level.
Heterogeneity was assessed with Q statistics and I2 test considering it signi cant with I2 value > 50% or P-value < 0.05.
The publication bias was assessed using Egger's regression test and represented graphically by Begg's funnel plot when there were ten or more studies/effect sizes. Egger's regression test P-value < 0.10 was considered signi cant. Whenever publication bias was found, the trim and ll method of Duvall and Tweedie was applied to add studies that appeared to be missing to enhance the symmetry.

Limitations
Limitations to our study include the signi cant heterogeneity in the analysis of some outcomes due to the different study designs that were included in this study. However, we estimated the risk of bias in most cases when signi cant heterogeneity was estimated, and no signi cant risk was found in these events.

Conclusions
Prophylactic indomethacin in VLBW infants has proven e cient in preventing short-term events as PDA. My analysis, however, showed no signi cance on IVH despite the many investigations that were published before which showed that indomethacin can reduce the incidence of IVH and other neurodevelopmental abnormalities.
Implications of key ndings I tried to lose the debate in this matter, so the clinician can work on other interventions or quality projects to help in decreasing CNS complications in preterm infants.

Systematic review registration number
We could not obtain because there are a big waiting list and priority for UK researchers.

Background
Many cardiopulmonary and neurologic disabilities have been associated with preterm labor including patent ductus arteriosus (PDA), pulmonary hemorrhage, intracranial hemorrhage, and mental retardation (1)(2)(3)(4). Although advances in modern medicine have improved the survival rates of very low birth weight (VLBW) infants, many neurodevelopmental complications are still present due to preterm birth such as blindness, deafness, and cerebral palsy. VLBW infants are liable to intraventricular hemorrhage (IVH) which is usually associated with the neurodevelopmental decays when related to the brain parenchyma. IVH grade 3-4 is a major risk factor for the occurrence of these complications in preterm infants (5)(6)(7)(8). Although the incidence rate of IVH has been markedly reduced since the 1980s (9, 10), no or minimal reductions have been recorded recently (11,12).
Many pre-and postnatal interventions have been reported to effectively treat IVH and reduce its incidence in preterm infants (13). One of these is indomethacin prophylaxis which is better administered within the rst six hours after birth (14)(15)(16)(17). Besides, it helps in the closure of ductus arteriosus and therefore, can prevent the complications of PDA as pulmonary hypertension (14,15,18). Its mechanisms of action include prostaglandin synthesis inhibition by inhibiting the cyclooxygenase pathways, reduction of hyperemic responses resulting from cerebrovascular hypoxia and hypercapnia, increasing the bloodbrain barrier permeability, and prevention of cerebral perfusion-induced ischemia (19)(20)(21)(22)(23). Moreover, it enhances microvascular development in the germinal matrix (24). Perfusion-related factors as hypoxia, hypercapnia, and hypotension usually develop after birth in VLBW infants (25). Most cases of preterm infants develop IVH within 6-8 hours after birth regardless of the gestational age (26). It happens probably due to the increased levels of angiopoietin 2 and vascular endothelial growth factor in the germinal matrix that normally decreases within hours after birth (13).
The results of previously published randomized controlled trials (RCTs) have shown that early administration of indomethacin after birth lowers the incidence of symptomatic PDA and severe IVH as a prophylactic measurement (16,(27)(28)(29). Although indomethacin administration showed favorable outcomes in reducing IVH incidence, many concerns have aroused concerning its effect on cerebral perfusion (30,31). The rates of mortalities, bronchopulmonary dysplasia (BPD), or long-term neurodevelopmental decays reportedly seem to have been not affected. A previously published large RCT advised against using indomethacin as a prophylactic agent (15). Although the study showed favorable outcomes in terms of reducing incidence rates of PDA, PDA ligation, IVH, and pulmonary hemorrhage, no improvement regarding the incidence of death and neurodevelopmental disorders rates has been found. Therefore, in this systematic review, we aim to analyze the data of previously published investigations on the use of prophylactic indomethacin in preterm infants.

Search strategy and study selection
In accordance with the Preferred Reporting Items for Systematic Review and Meta-analyses statement (PRISMA) recommendations, we performed this systematic review and meta-analysis (32). A systematic electronic database search was conducted for relevant studies published, from inception to 12 th August 2020, in seven databases: Pubmed, Google Scholar, Scopus, Web of Science, The New York Academy of Medicine (NYAM), Virtual health library (VHL), and the System for Information on Grey Literature in Europe (SIGLE). The search process conducted using keywords, medical subject (MeSH) terms, and publication types based on the PICO framework (participants, comparison, intervention, and outcomes). Participants were any preterm infants, the intervention was the prophylactic indomethacin, the comparison was placebo or no treatment groups, and all possible outcomes were included. The systematic search was followed by a manual search in references of the included papers to include missed papers (33).
We included all original studies that assessed the use of prophylactic indomethacin in preterm infants. Papers were excluded if there were one of the following exclusion criteria: i) non-original studies ii) in vitro or animal studies; iii) data duplication, overlapping or unreliably extracted or incomplete data; iv) abstract only articles, reviews, thesis, books, conference papers or articles without available full texts (conferences, editorials, author response, letters, and comments.
The title and abstract screening were performed by four independent reviewers. Furthermore, three independent reviewers performed full-text screening to ensure the inclusion of relevant papers in our systematic review. Any disagreement was done by discussion and consulting the senior member when necessary.

Data extraction
Two authors made the pilot extraction of a few papers for building the data extraction sheet. The data extraction sheet included: patient's characteristics, outcomes, and risk of bias tool. Two authors extracted the data and reviewed by a third reviewer when necessary. If a disagreement occurs, a senior author was consulted.

Risk of bias
Three independent authors assessed the risk of bias among different included studies using the revised Cochrane quality assessment tool (RoB 2) was used to determine the quality of randomized studies (34) and the risk of bias in non-randomized studies -of interventions (ROBINS-I) tool for non-randomized studies (35). Any discrepancy between the reviewers was solved by discussion or in conjugation with the senior author.

Statistical analysis
All data were analyzed using R software version 4.0.2 (36). Using the "meta" package, odds ratios (OR) and prevalence rates of different outcomes were calculated (37). The corresponding 95% con dence intervals (CI) of pooled effect size were calculated using a xed-effects or random-effects, according to heterogeneity level. Heterogeneity was assessed with Q statistics and I 2 test considering it signi cant with I 2 value > 50% or P-value < 0.05 (38).
The publication bias was assessed using Egger's regression test (39,40) and represented graphically by Begg's funnel plot (41) when there were ten or more studies/effect sizes. Egger's regression test P-value < 0.10 was considered signi cant. Whenever publication bias was found, the trim and ll method of Duvall and Tweedie was applied to add studies that appeared to be missing (42) to enhance the symmetry.

Search results
We identi ed 3,801 records after excluding of 506 duplicates by using Endnote software version X9. Title and abstract screening resulted in 36 records for further full-text screening. The later yielded 20 eligible papers for inclusion in our study. Two papers were added after performing manual search trials. Finally, we included 22 studies for this systematic review and meta-analysis ( Figure 1).

Study characteristics and risk of bias
Out of the 22 included studies; seven were randomized controlled trials and the remaining 15 were cohort in design. The sample size of the included studies was highly variable ranging from 19 and as high as 34,602 pre-term infants. The average mean age in all reported treatment groups was 28 weeks (ranging from 25 to 50 weeks), while it was 27 weeks in control ones (ranging from 26 to 29 weeks). Table 1 shows the main characteristics of the included studies.
The overall risk of bias in the included RCTs was low in more than 50% of all items. The domains with major problems were deviations from intended interventions, randomization processes, and missing outcome data ( Figure 2). For non-randomized trials, the risk of bias was higher with more than 50% of all items showing serious of critical risks of bias. Domains with the biggest issues were deviations from intended interventions, selective reporting, and cofounding of the results (Figure 3).

Cardiopulmonary Outcomes
Eight studies, with a total of 20,708 preterm infants, were included in the analysis of patent ductus arteriosus (PDA) rates. Infants with prophylactic doses of indomethacin showed signi cantly lower rates of PDA compared to those who did not (OR= 0.33; 95% CI= 0.27-0.39; P-value< 0.001). However, there was signi cant heterogeneity among the included studies (I 2 = 72%; P-value< 0.001) (Figure 4).

Necrotizing enterocolitis/Intestinal perforation, and death
Nine studies, with 35,759 preterm infants, were included in the analysis of necrotizing enter colitis (NEC)/Intestinal perforation (IP) rates. There was a signi cant increase in NEC/IP rates in prophylactic indomethacin group when compared to the placebo/no treatment group (OR= 1.36; 95% CI= 1.24-1.49; P-value< 0.001). Moreover, there was no signi cant heterogeneity among the included studies (I 2 = 44%; P-value= 0.088) ( Figure 6).
In the same context, death rates were compared in 47,265 patients of 13 included studies. There were no signi cant differences in death rates of prophylactic indomethacin group when compared to the placebo/no treatment group (OR= 1.06; 95% CI= 0.94-1.19; P-value= 0.340). However, there was signi cant heterogeneity among the included studies (I 2 = 69%; P-value< 0.001). There was no signi cant risk of bias when tested using Egger's regression test (P-value= 0.428) (Figure 7).

Hospitalization outcomes
Two studies of 340 patients were included in the analyses of hospitalization days and days spent of ventilation. On comparing these outcomes among the prophylactic indomethacin and placebo/no treatment groups, there was no statistically signi cant difference for hospitalization days (MD= 3.67; 95%CI= -11.03-18.37; P-value= 0.625) and days spent of ventilation (MD= -1.00; 95%CI= -6.71-4.71; P-value= 0.732). There was a signi cant heterogeneity in the analysis of hospitalization days (I 2 = 56%; P-value= 0.132), while it was not present in the analysis of days spent of ventilation (I 2 = 0%; P-value= 1.000) ( Figure  8).

Discussion
In this study, we have included 22 studies from the systematic and manual search to be analyzed to study indomethacin as a prophylactic measurement in pre-term infants from many aspects including the cardiopulmonary, neurodevelopmental delays, necrotizing enterocolitis/Intestinal perforation and death, days of hospitalization and the time spent on mechanical ventilation together with the prevalence of these outcomes from eligible study designs. The pooled results of our analysis showed that intubation on the rst day at the delivery room was the most prevalent outcome, while pulmonary hemorrhage was the least. In general, these rates are like previously published reports. Laptook et al. (43) reported a prevalence rate of 29.2% for neurodevelopmental delays VLBW infants. PDA was prevalent in 65.57% in neonates at birth and 41% of the patient who received indomethacin did not have favorable outcomes in Koch et al. study (44). The quality of the included studies was good in general which is suggestive of valid data included in the analysis. However, analyzed data showed various heterogeneity in some outcomes which is probably due to the difference in study designs, the different dosages of indomethacin injection, and outcome de nition between studies.
As for the cardiopulmonary outcomes, our analysis showed that prophylactic indomethacin administration in infants signi cantly lowers the rates of PDA formation (P-value< 0.001). Although, signi cant heterogeneity was estimated (I 2 = 72%; P-value< 0.001) due to the different study designs that were included in the analysis, our results were similar to previously published studies (45,46). On the other hand, the effect of indomethacin on BPD and pulmonary hemorrhage rates was statistically non-signi cant (P-value =106 and 0.132, respectively). This is consistent with previously published meta-analysis studies of RCTs which found signi cance in terms of lowering PDA rates, but no difference in terms of BPD (45). Jensen et al. (46) in their analysis of observational data found that prophylactic indomethacin did not increase or decrease the risk of developing BPD. Moreover, the authors compared these results with another analysis of RCTs, however, the analysis indicated the same information that prophylactic indomethacin had no bene cial effects of BPD.
In terms of neurodevelopmental outcomes, our analysis showed that indomethacin prophylaxis did not signi cantly reduce the risk of IVH and CP/neurodevelopmental outcomes in preterm infants when compared to the placebo group (P-value =0.460 and 0.997, respectively). Our results are consistent with the results of Cooke et al. (47) which found no signi cance on the effect of prophylactic indomethacin on IVH and other neurodevelopmental abnormalities as retinopathy of prematurity. On the other hand, Fowlie et al. found a signi cant reduction in IVH incidence in infants that were prophylactically injected with indomethacin. However, signi cant heterogeneity in this study was estimated due to the inconsistency of treatment e cacy among their included studies (45). None of the studies, however, measured the long-term outcomes, they have only focused on the short ones. Schmidt et al. (15) in their big trial on 18-month infants reported statistical insigni cance on long term neurodevelopmental outcomes although IVH grade 3 and 4 were signi cantly reduced.
Therefore, concerns should be made to assess the overall quality of the effect of indomethacin on the long-term neurodevelopmental outcomes and the rate of adverse events incidence due to the vasoconstrictive nature of the drug which may alter the cerebral blood ow.
Other adverse events of prophylactic indomethacin application include increase incidence rates of necrotizing enter colitis (NEC)/Intestinal perforation (IP). In our study, a signi cant increase in the rate of these outcomes was estimated (P-value< 0.001). Our study results are inconsistent with the results of other previously published systematic reviews which found no signi cance on the effect of indomethacin on increasing the incidence of NEC (14,45,47).
Additionally, other events were discussed by these studies as the effect of indomethacin on renal function. Fowlie et al. (14) found signi cance in the increased rate of oliguria but not creatinine levels.
Furthermore, we found no signi cance between the use of prophylactic indomethacin on infants in reducing the time of hospital stay and time spent on mechanical ventilation. Fowlie et al. (14) result favored the control groups in terms of time spent in the hospital with no signi cance (P =0.087), however, the opposite was reported with the time spent on ventilation with reported statistical signi cance (P =0.33). Besides, the data of 13 included studies showed no signi cance in reducing the rates of mortality in infants that received prophylactic indomethacin (P =0.34). Despite the heterogeneity in our analysis, the results are consistent with these shown by Fowlie et al. (14,45). On the other hand, Jensen et al. (46) subgroup analysis showed a low rate of mortality in the infants that received prophylactic indomethacin with no intrauterine growth restriction and with no requirement for subsequent PDA treatment. However, the authors included observational data only.
Limitations to our study include the signi cant heterogeneity in the analysis of some outcomes due to the different study designs that were included in this study. However, we estimated the risk of bias in most cases when signi cant heterogeneity was estimated, and no signi cant risk was found in these events.

Conclusion
Prophylactic indomethacin in VLBW infants has proven e cient in preventing short-term events as PDA. Our analysis, however, showed no signi cance on IVH despite the many investigations that were published before which showed that indomethacin can reduce the incidence of IVH and other neurodevelopmental abnormalities. Consequently, and based on the ndings that higher rates of NEC/IP and the neutral effect on infant mortality, the use of prophylactic indomethacin should be generally discouraged until a proper assessment of the case has been provided.