Glycerin Suppositories and Enemas in Premature Infants: A Meta-analysis

A systematic search was conducted using MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials for RCTs of premature infants treated with glycerin suppositories or enemas. A medical librarian queried each database to identify studies that included the concepts premature infants and glycerin enemas or glycerin suppositories. Additional similar terms were used to avoid missing studies because of variation in nomenclature. The search strategy and results from each database can be seen in Supplemental Tables 4–6. We also searched the available clinical trial registries^12–15  and included all citations up through January 2022. Our study protocol is registered online through the International Prospective Register of Systematic Reviews.¹⁶ 

Abstracts were screened independently and in duplicate by the first two authors (P.R.B. and R.L.). Our inclusion criteria were (1) premature infants <32 weeks gestation and/or birth weight <1500 g, (2) interventions with glycerin enemas or suppositories used prophylactically or as rescue therapy, and (3) studies that were RCTs. The subgroups glycerin suppositories and glycerin enemas were used for the meta-analysis.

Studies that met the inclusion criteria were reviewed in full with data extraction performed by the first two authors. To ensure consistency between reviewers, data extraction templates were created and used. Participant data included number of participants, years of recruitment, gestational age, sex, birth weight, major congenital or gastrointestinal anomalies, and type of feeds. Intervention data included study location, years of recruitment, type of active treatment (glycerin enemas or glycerin suppositories), type of control (no intervention or placebo procedure), and treatment dosing, frequency, and duration.

The majority of trials on the use of glycerin suppositories or enemas were powered to detect a difference in time to full enteral feeds, and this was selected as the primary outcome measure. Additional outcomes of interest were mortality, NEC, rectal perforation, rectal bleeding, meconium evacuation, and jaundice. These outcomes were selected because of their importance and potential to be modified by the study treatment. We hypothesized that the routine use of glycerin suppositories or enemas in premature infants may expedite meconium evacuation but would have no consistent effect on enteral feeding or other outcomes.

Outcome data were pooled using Review Manager version 5.4. We used a random-effects model because of anticipated heterogeneity between studies given the variation in terms of interventions and participants. Trials were stratified by focus on glycerin suppositories or glycerin enemas. Dichotomous outcomes were reported using risk ratio (RR), whereas continuous outcomes were reported using mean difference. All values were reported with 95% confidence intervals (CIs) and P values.

Heterogeneity was reported using the I² statistic, where 0% to 40% might not be important, 30% to 60% may represent moderate heterogeneity, 50% to 90% may represent substantial heterogeneity, and 75% to 100% is considerable heterogeneity.¹⁷  SD was estimated when not explicitly reported through the following imputations: range / 4 (when n = 15–70) and range / 6 (when n >70).¹⁸ 

Risk-of-bias assessment was performed independently and in duplicate by the first two authors using the Cochrane risk-of-bias tool.¹⁷  This measure classifies risk of bias across 6 domains as high, low, or unclear. Discrepancies between reviewers were discussed until consensus was reached.

The quality of evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation system with GRADEpro.¹⁹  Using this approach, we assessed each outcome across multiple studies and graded the quality of evidence as very low (1 of 4 possible point), low (2 points), moderate (3 points), or high (4 points). Results from RCTs are presumed to be high quality, whereas those from observational studies are considered low quality. These default ratings are then upgraded or downgraded by 1 or 2 points each for risk of bias, inconsistency, indirectness, imprecision, and effect size.

The systematic search of MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials resulted in 108 articles and abstracts. Articles and abstracts were reviewed independently and in duplicate. Of the RCTs, we excluded 5 that focused on term infants^20–24  and 1 that lacked a proper control arm,²⁵  resulting in the inclusion of 6 RCTs.^26–31  The intervention arms of 4 of these studies included glycerin suppositories,^27,28,30,31  and those of two studies included glycerin enemas.^26,29  A flow diagram of the results from the systematic search is shown in Fig 1.

The 6 RCTs comprised a total of 389 premature infants. Four included patients <32 weeks gestation and/or birth weight <1500 g,^26–28,31  one included premature infants 30 to 35 weeks gestation who required phototherapy for jaundice,³⁰  and 1 did not use gestational age as an inclusion criterion and only included premature infants with a birth weight of 500 to 1250 g.²⁹ 

All 6 trials excluded neonates with congenital or gastrointestinal malformations. Three specifically excluded premature infants with hemodynamic instability or shock.^28,29,31 

Active study treatments consisted of scheduled administration of glycerin suppositories or glycerin enemas. Control treatments included some form of placebo procedure (ie, opening and closing the diaper or leaving the glycerin suppository in the diaper, 3 trials) or no treatment at all (3 trials). Each trial differed slightly in terms of study treatment start time, frequency, and duration (Table 1).

The definition of full enteral feeding varied across studies as follows: 100 mL/kg/day,²⁹  150 mL/kg/day,^26,31  180 mL/kg/day for 24 hours,²⁸  or tolerance of enteral feeds (without prespecified volume) and discontinuation of parental nutrition for >48 hours without holding or reducing feeds.²⁷  One study did not report any outcomes related to enteral feeding.³⁰ 

The type of enteral feeds was reported in 4 of the 6 trials. Premature infants in Haiden et al²⁶  were fed expressed breast milk or 5% hydrolysate of ultrafiltrated whey-dominated milk protein until breast milk was available. Shinde et al,²⁸  Khadr et al,²⁷  and Mena et al²⁹  reported that premature infants received expressed breast milk and/or formula.

The timing of enteral feeds was similar between treatment groups within each study. Between studies, however, the timing of initiating feeds varied considerably, including starting feeds on median day 1 (range, 1–1.8 days),²⁷  median day 2 (range, 0–9 days),²⁶  median day 3 (range, 2–5 days),²⁹  and mean day 5 (SD range, 2.2–2.8 days).²⁸ 

Transition to full enteral feeding was defined as the difference in days between the start of enteral feeds and full enteral feeding; however, this was defined within each study. Across all 5 trials, there was no statistical difference in time to full feeds between treatment groups (mean difference, −0.47 days; 95% CI, −1.76 to 0.82 days; I² = 1%; P = .48) (Fig 2).

Mortality rates ranged from 0% to 17%^27–31  (N. Haiden, personal communication, 2015). There was no statistically significant difference in mortality between infants who were administered glycerin suppositories or glycerin enemas compared with the control treatment (RR, 1.06; 95% CI, 0.55 to 2.06; I² = 0%; P = .86). There were also no differences in mortality for the predefined subgroups of glycerin suppositories (RR, 1.06; 95% CI, 0.36 to 3.19; I² = 0%; P = .91) or glycerin enemas (RR, 1.08; 95% CI, 0.44 to 2.66; I² = 14%; P = .88) (Fig 3).

There were no statistically significant differences in the incidence of NEC between treatment groups in any of the trials.^9,26–31  The meta-analysis demonstrated a nonsignificant trend toward a higher incidence of NEC with glycerin suppositories (RR, 2.75; 95% CI, 0.68 to 11.18; I² = 0%; P = .16); however, there were no apparent trends for glycerin enemas (RR, 0.83; 95% CI, 0.12 to 5.57; I² = 57%; P = .13) or overall (RR, 1.30; 95% CI, 0.46 to 3.63; I² = 0%; P = .62) (Fig 4).

Investigators of 4 trials explicitly reported no cases of rectal perforation.^27,30,31  Additionally, the incidence of rectal bleeding was reported in 4 trials and ranged from 0% to 13%.^30,31  These data were obtained from Haiden et al²⁶  and from the corresponding author (N.H., personal communication, 2015). There were no significant differences between the intervention and control groups.

Initiation of meconium evacuation was reported in 2 trials. In 1 trial, treatment with glycerin suppositories was associated with earlier initial meconium evacuation (mean, 2 vs 4 days; P = .016).²⁷  This difference was not observed with glycerin enemas (median, 1 day; P = .73).²⁶  In the meta-analysis, there was no difference in days to initial meconium evacuation between treatment groups (mean difference, −1.02 days; 95% CI, −2.98 to 0.94 days; I² = 90%; P = .31) (Fig 5).

Three trials reported completion of meconium evacuation.^26,29,31  Haiden et al²⁶  reported a nonsignificant trend toward earlier completion of meconium evacuation with glycerin enemas (median, 6.5 vs 9 days; P = .11). The meta-analysis revealed significantly earlier completion of meconium evacuation overall (mean difference, −1.52 day; 95% CI, −3.03 to −0.01 days; I² = 5%; P = .05) (Fig 6).

Two studies reported outcomes related to jaundice. In the trial of infants 30 to 35 weeks gestational age requiring phototherapy, there was no significant difference in administration of glycerin suppositories versus control treatment in peak total serum bilirubin (11.3 vs 12.0v mg/dL; P = .15), total phototherapy duration (72 vs 61 hours; P = .92), or rate of bilirubin decline (0.4 vs 0.4 mg/dL/hour; P = .74).³⁰  Mena et al²⁹  also found no differences in peak serum bilirubin among randomly assigned infants receiving glycerin enemas compared with placebo (mean 7.9 vs 7.7 mg/dL; P = .20).

Additional outcomes are shown in Table 2. None of the studies reported statistically significant differences between the intervention and control groups in terms of intraventricular hemorrhage, retinopathy of prematurity, patent ductus arteriosus, sepsis, oxygen requirements, weight gain, or length of stay.

Four of the 6 trials had a high risk of bias in ≥2 of the 6 domains of the Cochrane risk-of-bias tool (Fig 7).^26–28,30  Mena et al²⁹  had the only trial that did not have high risk of bias in any domain.

Random sequence generation was created with random number software in 4 trials^26,28,30,31  (K. Nandkishor, personal communication, 2015; N. Haiden, personal communication, 2015; C. D’Angio, personal communication, 2021). In the 2 other trials, random sequence generation was accomplished through the shuffling of sealed envelopes.^27,29 

Study group allocation was concealed in 4 studies using opaque envelopes^27–29  or bags.³⁰  The 2 other trials were open label and unblinded.^26,27  Three studies relied on nurses who were not involved in direct patient care to administer study interventions, which maintained blinding of the health care team and study staff.^28,29,31  Butler-O’Hara et al³⁰  had bedside nurses administer or withhold glycerin suppositories according to random assignment, whereas the remainder of the patient care team and study staff stayed blinded.

Incomplete outcome data were assessed in all 6 trials. In Shinde et al,²⁸  3 patients in each group were transferred to another hospital before complete outcome assessment. We performed a sensitivity analysis using best- and worst-case scenarios for mortality and NEC and found no differences in mortality across subgroups and overall. The worst-case scenario for NEC (ie, 3 of 3 neonates in the intervention group developing NEC vs 0 of 3 in the control group) would have made our subgroup analysis of glycerin suppositories statistically significant (RR, 3.93; 95% CI, 1.03 to 14.98; P = .05). We therefore concluded that this study was at high risk of bias because of incomplete outcomes.

The study protocols for 3 trials were registered in clinical trials repositories (ISRCT Register identifier: ISRCTN47065764²⁷ ; ClinicalTrials.gov identifiers: NCT02153606³⁰  and NCT01746511³¹ ). The other 3 trials were not registered, and the risk of bias as a result of selective reporting was unclear^26,28,29  (K. Nandkishor, personal communication, 2015; N. Haiden, personal communication, 2015).

Four trials were at high risk of bias because of inadequate power.^26–28,30  Haiden et al²⁶  designed their study to detect a 30% difference in days to complete meconium evacuation. Shinde et al²⁷  and Khadr et al²⁸  designed their studies to detect reductions of 3 and 3.63 days to full enteral feeds, respectively. These effect sizes were larger than any possible differences between treatment groups. Adequate power to determine the significance of the observed changes would have required a much larger sample size. The trial by Butler-O’Hara et al³⁰  was designed to detect a noninferiority margin of a difference in phototherapy duration of 18 to 23 hours with an SD of 30 to 38 hours. The observed SD was much larger than anticipated, and the sample size was therefore too small to detect a true effect if one actually existed. The study by Livingston et al³¹  was a pilot study and not powered to detect differences in clinical outcomes, if these existed.

Three trials were also at high risk of bias because of protocol violations.^26,30,31  In the trial by Livingston et al,³¹  50% of patients experienced at least 1 protocol violation as a result of study treatments concluding too early or continuing too long. In Haiden et al,²⁶  38% (15 of 42) infants in the intervention group missed 1 of their daily enemas, whereas 21% of infants (8 of 39) in the control group received at least 1 enema despite being assigned to no treatment. In the trial by Butler-O’Hara et al,³⁰  7 of 39 patients (18%) in the control group and 2 of 40 (5%) randomly assigned to the active treatment received at least 1 nonstudy glycerin suppository.

Funnel plots were used to assess for publication bias for transition to enteral feeds (Fig 8) and completion of meconium evacuation (Fig 9). The shape of the funnel plot for transition to enteral feeds was concerning for possible publication bias, particularly due to the small trial by Livingston et al.³¹  If this study were removed, then the summary statistic for transition to enteral feeding would remain clustered around no difference between treatment groups. The funnel plot for completion of meconium evacuation seemed to be symmetrically distributed around 1.5 days faster with active treatment. Only 3 studies were included in this second funnel plot, but there were no apparent outliers or signs of publication bias.

We conducted a post hoc sensitivity analysis with a meta-analysis of data from the 2 trials with the lowest risk of bias.^29,31  There were no changes in the results of the meta-analysis for mortality (RR, 0.73; 95% CI, 0.25 to 2.14; I² = not applicable; P = .57), NEC (RR, 0.64; 95% CI, 0.11 to 3.71; I² = 31%; P = .23), or transition to enteral feeds (mean difference, −1.89 days; 95% CI, −5.35 to 1.56 days; I² = 15%; P = .28). The summary statistic for completion of meconium evacuation was not significant (mean difference, −1.07 days; 95% CI, −3.10 to 0.96 days; I² = 20%; P = .26).

The quality of evidence across the 6 trials was downgraded for all outcomes because of serious risk of bias (Table 3). Ratings were further downgraded because of imprecision when 95% CIs for relative risk crossed 1.0 or the mean difference crossed 0. The very small numbers for NEC, rectal bleeding, and rectal perforation resulted in additional downgrades for very serious imprecision. Initiation of meconium evacuation was downgraded because of inconsistency, as the CIs of the 2 trials did not overlap. The quality of evidence for rectal bleeding was upgraded by 1 point because of strong association. In light of these issues, the quality of evidence was very low (1 out of 4 possible points) for rectal perforation and initiation of meconium evacuation, low (2 points) for other outcomes, and moderate (3 points) for completion of meconium evacuation.

Our systematic review of clinical trial registries revealed 1 ongoing RCT in Saudi Arabia where the researchers are investigating the use of prophylactic glycerin suppositories for feeding intolerance in very low birth weight premature infants.³²  Recruitment for this study started in 2013 with an estimated enrollment of 220 patients. The inclusion criteria are premature infants with a birth weight of <1250 g, and the primary outcome is days to full enteral feeding, defined as 140 mL/kg/day. Secondary outcomes include incidence of feeding intolerance and NEC. The online trial registration indicates that this study is expected to complete enrollment by June 2022.

In this review, we have identified 6 single-center RCTs that explored the use of glycerin suppositories or glycerin enemas in premature infants.^26–31  The quality of evidence for the administration of these medications is predominantly low to very low largely because of underpowered studies and risk of bias due to methodology, inadequate blinding, frequent protocol violations, incomplete outcome data, possible selective reporting, and possible publication bias.

On the basis of our meta-analysis, we suggest that the use of glycerin suppositories and enemas does not affect mortality or the transition to enteral feeding. There does not seem to be any definitive association with NEC, although this may change as additional data become available. Earlier initiation of meconium evacuation was observed in 1 study of premature infants randomly assigned to glycerin suppositories, but there was no difference seen in another trial with glycerin enemas. Completion of meconium evacuation was faster among premature infants randomly assigned to glycerin suppositories or enemas. The clinical significance of this finding is uncertain. Among the 4 trials reporting rectal bleeding, there were 3 occurrences in the glycerin suppository and enema groups compared with 1 occurrence in the control groups.^26,27,30,31 

Given the heterogeneity and methodologic limitations of published RCTs, the evidence for the use of glycerin suppositories and enemas in premature infants remains inconclusive. Our analysis was complicated by the fact that glycerin suppositories and enemas represent similar, but different treatment modalities. We attempted to account for this issue by stratifying trials by the type of treatment, but it remains uncertain whether systematic differences exist in outcomes between these 2 medications.

The use glycerin suppositories and enemas was associated with earlier completion of meconium evacuation without any differences in transition to full enteral feeds or mortality. There does not seem to be any associated trends in harm other than rectal bleeding, although rectal bleeding occurs in a very low number of infants and does not seem to be associated with rectal perforation or NEC. Evaluation of these rare outcomes completely would require a much larger sample size and is likely not feasible in the context of an RCT.

Our group performed a similar meta-analysis in 2015 and identified a trend toward increased risk of NEC with the administration of glycerin suppositories and enemas. This earlier review included data from 3 of the 6 studies in the current review, and we reported a relative risk of 2.72 (P = .13) with active versus control treatment. With the additional data from 3 more recent studies, we see a reduction in that trend, with a relative risk of 1.30 (P = .62). This effect is likely due to the larger trial on glycerin enemas, where there were 3 instances of NEC in the intervention arm compared with 8 in the control arm.²⁹ 

A Cochrane review was previously published in 2015 to evaluate the use of glycerin laxatives for prophylaxis of feeding intolerance in very low birth weight infants.³³  These findings were similar to our previous meta-analysis where we found no significant difference in time to full enteral feeds, duration of hospital stay, mortality, and weight at discharge. The study authors did find a higher instance of meconium passage over the first 48 hours.

Two of the trials included in the current review assessed the effects of glycerin suppositories or enemas on jaundice in preterm infants.^29,30  Butler-O’Hara et al³⁰  did not find a significant difference in peak serum bilirubin, phototherapy duration, or rate of bilirubin decline between groups. Similarly, Mena et al²⁹  found no difference in peak serum bilirubin between glycerin enema and placebo treatments. These results are similar to a systematic review published in 2011 that focused on 3 RCTs in healthy term neonates, where no identifiable difference was noted with the use of glycerin suppositories on neonatal hyperbilirubinemia.^20,24,34  In contrast, a trial of term neonates combined phototherapy with glycerin suppositories and found an earlier reduction in total plasma bilirubin and shorter hospital stays after the intervention.²²  Based on the limited data, the combination of phototherapy with glycerin suppositories should continue to be an area of investigation.

Our search strategy identified 1 trial on the use of prophylactic glycerin suppositories to improve feeding intolerance in very low birth weight preterm infants in Saudi Arabia.³¹  This study began in 2013 with an estimated enrollment of 220 participants, which would make it the largest trial on the use of glycerin suppositories in premature infants to date. Currently, the study is still in active recruitment, with an estimated completion date of June 2022. Conclusions may change on the basis of the results of this study.

The quality of evidence for the use of glycerin suppositories or glycerin enemas in premature infants is predominantly low to very low. Most RCTs were underpowered and have frequent protocol violations. As a result, the evidence for the administration of glycerin suppositories and enemas is largely inconclusive. Our meta-analysis of 6 published RCTs reveals earlier completion of meconium evacuation with the use of glycerin suppositories and enemas without differences in enteral feeding or mortality. Despite the statistical difference in completion of meconium evacuation, the clinical significance of this finding is uncertain. Our previous meta-analysis in 2015 demonstrated a trend toward increased risk of NEC with active treatment, but this updated analysis no longer reflects this concern. The few trials that have assessed the effects on jaundice also find no difference with active treatment compared with control. The largest RCT on this topic is estimated to complete enrollment in June 2022 and may provide additional insight on this common, yet still unproven treatment strategy.

CI

confidence interval

NEC

necrotizing enterocolitis

RCT

randomized controlled trial

RR

risk ratio