Abstract
OBJECTIVE:
Identify perinatal risk factors associated with SIP
STUDY DESIGN:
This was a retrospective case-control study of SIP in infants born at ≤28 weeks of gestation and admitted between 1995 and 2016 at a tertiary care NICU. Infants with NEC or other GI abnormalities were excluded. Cases of SIP were matched with gestational age-matched controls with the closest birth date. Maternal, infant and birth related factors were evaluated using univariate analyses and significant factors were evaluated using multiple logistic regression.
RESULT:
25 cases of SIP were matched with 25 controls. No maternal factors reached statistical significance. Being one of twins increased the odds of SIP 29-fold. Birth-order or weight-discrepancy in twin had no association of SIP within twin pairs.
CONCLUSION:
Twins are at significantly higher risk for SIP. The association of SIP and twin gestation was independent of previously reported risk factors of perinatal indomethacin or magnesium sulfate and merits further study.
Abbreviations
Necrotizing enterocolitis
Extremely low birth weight
Very low birth weight
Neonatal intensive care unit
Spontaneous intestinal perforation
Introduction
Spontaneous intestinal perforation (SIP) of the newborn is defined as the sudden occurrence of perforation of the intestine during the first 14 days of life involving less than five centimeters of bowel and not associated with any concomitant signs of NEC or other intestinal pathology [1–3]. SIP incidence is reported to be two to three percent in very low birth weight infants (VLBW) and five percent in extremely low birth weight (ELBW) infants [2, 4–5]. The incidence is likely under-reported because in some cases it is misdiagnosed or classified as NEC [3–5]. Low birth weight and prematurity are the well-characterized risk factors for SIP. The average gestational age at birth of infants that develop SIP ranges from 25–27 weeks, and the average birth weight is between 670 and 970 grams [6]. SIP is more commonly seen in males [7]. Early administration of post-natal steroids, such as dexamethasone [8–10], and the use of early postnatal indomethacin [6, 10–12] are well established risk factors for SIP. The combination of postnatal steroids and postnatal indomethacin was found to be additive in predisposing to SIP [13, 14]. The contribution of prenatal obstetric factors to SIP such as chorioamnionitis, placental abruption, pregnancy-induced hypertension, and medications such as antenatal steroids, magnesium sulfate, and indomethacin have been reported but evidence has been inconsistent [6, 12–14]. The aim of the current study is to evaluate the association of perinatal factors with the development of SIP in premature infants born at ≤28 weeks’ gestation.
Methods
This was a retrospective case-control study done at a single tertiary care NICU in Farmington, CT. Infants born at ≤28 weeks gestational age and admitted to the NICU between January 1, 1995 and December 31, 2016 were studied. Infants who developed perforations during their NICU stay were identified using a prospectively collected neonatal patient database (Neonatal Information Systems Version 5, Medical Data Systems, Philadelphia, PA).
Definition of SIP
SIP was defined as presence of intraperitoneal air in non-anomalous bowel without clinical or laboratory signs of NEC such as pneumatosis intestinalis, portal venous air or hematological changes, such as thrombocytopenia, anemia and increased or decreased neutrophil counts. Patients with perforation associated with NEC, congenital intestinal anomalies or other genetic conditions were excluded. Each case of SIP was matched with a control of the same gestational age at birth that was born within six months of the case. The clinical database (NIS-5) included items that were prospectively collected from daily input of data elements. The fidelity of these data elements was verified at time of discharge by a dedicated information management team to conform to pre-determined definitions of terms mostly derived from the Vermont-Oxford Handbook of definitions [15]. Medical records of all cases and controls were reviewed for demographic factors as well as information regarding putative perinatal risk factors for SIP.
Antenatal data collection
Antenatal data included dosage and use of antenatal medications (steroids, magnesium sulfate, and indomethacin), maternal conditions (presence of preterm labor, premature prolonged rupture of membranes, pregnancy-induced hypertension, and placental abruption), maternal substance use and group B streptococcal (GBS) infection status at delivery. Intrapartum data included mode of delivery, Apgar scores, and resuscitative procedures in the delivery room (chest compressions, epinephrine and volume use).
Postnatal risk data collection
Postnatal risk data included gestational age, birth weight, growth classification (small for gestational age - SGA, appropriate for gestational age - AGA, or large for gestational age - LGA), multiple gestation, infant sex, umbilical arterial line use and dose and timing of indomethacin and dexamethasone administration. Additional postnatal risk factors investigated were postnatal age at diagnosis of bowel perforation, need for surgery with details of type and timing of surgical intervention and the outcomes at discharge. Approval for the study was obtained from the Institutional Review Board of Connecticut Children’s Medical Center.
Statistical analysis
All analyses were performed using IBM SPSS version 25. Demographic and clinical characteristics of the study group were described using median and interquartile ranges as appropriate. Univariate comparisons of risk factors and outcomes were performed using Chi square test, Student’s t-test or Mann-Whitney-U test as appropriate. Multivariate logistic regression modeling strategies were used to identify factors significantly associated with SIP after accounting for confounding variables.
Results
During the 21-year study period there were 9349 neonates admitted to the Connecticut Children’s Medical Center NICU in Farmington, CT. Of these, 1387 infants were born at ≤28 weeks of gestation. Among these patients, 62 (4.5%) were diagnosed with intestinal perforations. After excluding other causes of intestinal perforations (NEC, anomalies, etc.), 25 infants met the criteria for diagnosis of a spontaneous intestinal perforation (SIP), and they were matched with 25 controls based on gestational age, birth weight, and birth within six months of the corresponding case.
Maternal and obstetric characteristics
Table 1 shows the maternal and obstetric characteristics of the cases and controls. The two groups did not differ significantly in maternal age, gravidity or parity status. There were no differences in GBS colonization or use of tobacco, alcohol or illicit drugs. The occurrence of preterm labor, placental abruption, chorioamnionitis, pregnancy-induced hypertension or the mode of delivery were not statistically different between SIP cases and controls. Maternal medications such as steroids, indomethacin and magnesium sulfate used prior to birth of the infant were also not statistically different in the two groups.
Maternal Factors in SIP
Maternal Factors in SIP
Statistical analysis based on Fisher’s Exact Test. aMissing data 3, bMissing data 3.
Table 2 shows the infant factors associated with SIP. There were no differences in GA and birthweight in the two groups as they were matched for these variables. Twenty-five cases of SIP were noted, of which 11 cases were of singleton gestation and 14 were one of twin gestations. Twin gestations were significantly associated with SIP cases compared to controls (56%vs 4%; p < 0.0001). However, twin birth order or birthweight discrepancy between twins were not significant risk factors. The mean time of onset of SIP was 6.35 days with a range of 2–16 days. Of the infants with SIP, three died and 22 survived to discharge from the NICU. The location of perforation was in the small intestine in all patients studied. For three infants, perforation was noted in the terminal ileum and four infants had perforation in the jejunum. Twelve infants were managed by placement of peritoneal drain and seven required laparotomy. Of the seven infants requiring laparotomy, bowel was resected in six infants. Two infants with SIP developed NEC at a later date in the NICU.
Infants Factors in SIP
Infants Factors in SIP
All analyses based on T-test or Fischer’s Exact Test; Values expressed as N (%) or Mean (SD). *Statistically significant.
Table 2 shows post-natal risk factors in cases and controls. There was no difference between the two groups for race and sex. Need for resuscitation at delivery, including need for chest compressions, epinephrine, and volume bolus with normal saline were not significantly different. Apgar scores at 1 and 5 minutes and use of postnatal indomethacin or ibuprofen use were statistically similar between both groups. Interestingly, postnatal dexamethasone was not noted in any of the SIP cases but 9%in the control group had received postnatal dexamethasone within 16 days postnatal age. Postnatal usage of hydrocortisone in the first 14 days was studied but not found to be statistically significant.
Multivariate analyses
Table 3 shows the Wald statistics and odds ratios from a multiple logistic regression analysis done using SPSS version 25.0. After meeting all assumptions for this analysis, the logistic regression model was found to be statistically significant, χ2 (8) = 23.15 and p < 0.001. The model explained 51.6%(Nagelkerke R2) of the variance in SIP. Twins were 29.6 times more likely to develop SIP than singletons. None of the other putative factors including race, antenatal steroid use, postnatal steroid use, epinephrine use in delivery room, or use of antenatal magnesium sulfate had any significant effect in the model.
Multiple Logistic Regression Analysis of Variables implicated in SIP
Multiple Logistic Regression Analysis of Variables implicated in SIP
a. The reference category is: Non-perforation; b. Coefficient showing positive or negative association.
14 of the 25 cases were from a set of twins allowing us to perform within pair comparisons as shown in Table 4. All twin pairs were di-chorionic and di-amniotic and did not share a common vascular supply in the placenta. The average weight difference among twin pairs was 91.8 gm (range 0–330 gm) and the average percent difference in birth weight was 9.67%(range 0%–30.6%). Of the 12 twins, 8 with higher birth weight developed SIP and 4 with lower birth weight developed SIP. The occurrence of SIP in the larger twin was statistically significant (p < 0.005). In one set of twins, both had the same birth weight. In two sets of twins, both developed SIP and in three sets of twins, the non-SIP twin was stillborn. In total mortality was 8%in the SIP group.
Twin-Pair Comparisons
Twin-Pair Comparisons
P < 0.005. BW- Birth Weight in grams; GA –gestational age in completed weeks; Rel. BW- Relative difference in birth weight; %Diff.–Percent Difference in birth weight, UK-unknown. *In this twin set, both developed SIP; **In this twin set one twin was delivered deceased.
This retrospective case control study suggests that infants of twin gestation are at a significantly higher risk of developing SIP. Within twin pairs, the one with higher birth weight was more at risk but sex, and birth order were not significant factors.
The influence of gestational age and birth weight on SIP is well reported in the literature and these factors were controlled for in this report to focus on other lesser-recognized risk factors [16–19]. No racial or ethnic differences have been noted in previous studies on SIP [20, 21]. Our study also found no significant association.
Previous studies have not shown a consistent association of male sex with SIP [17, 22]. In our study, there were more males with SIP but there was no significant association either on univariate or on multivariate analysis. A larger sample size is needed for a more definitive analysis.
The use of antenatal indomethacin as a tocolytic prior to delivery was noted to be a risk factor in previous reports [23, 24]. We were not able to detect a statistically significant association in this case-control study. An association of SIP with antenatal steroid use or magnesium sulfate use previously suspected in SIP was not detected in our study [14].
Postnatal indomethacin use was found to be an important risk factor for SIP in a study of hydrocortisone use in the first week after birth for the prevention of BPD [21, 26]. Further, use of prophylactic indomethacin was associated with increased odds of SIP independent of early enteral feeding shown in a study by Stavel et al. [11]. We do not use hydrocortisone in the first week of life and prophylactic indomethacin is not a routine practice at our center. Therefore, this specific association could not be studied. In general, we found no association between postnatal indomethacin use and SIP.
Some studies have shown an association of SIP with use of inotropes in premature infants [24, 27]. In addition, there is indirect evidence from a report from Singh et al who reported a higher incidence of SIP in infants who were acidotic at birth [28]. It is possible that circumstances related to perinatal distress and acidosis may have an effect on gut perfusion and predispose premature infants to SIP. However, both the use of epinephrine in the delivery room and the use of dopamine infusion in the first postnatal week were not associated with SIP in our study.
The mortality rate of SIP in our study was 8%, which is relatively low compared to the mortality in infants who develop perforation due to NEC. A similar finding was recently reported by Clyman et al. [29]. The finding of a higher risk of SIP in one of a pair of twins is intriguing. In a review of previous reports evaluating risk factors for SIP, there has always been a trend for a higher rate of occurrence in twins [16, 29]. Case reports of spontaneous ileal perforation in a pair of twins have been published [30]. In another case series, Hay et al. described SIP in three pairs of identical male twins [31]. SIP has been reported in the donor twin after twin-twin transfusion syndrome [32]. In one report by Suply et al., the adjusted odds ratio was 2.9 times higher risk of SIP in multiple pregnancies, although twin pregnancies were not specified [19]. Also, in a large cohort studied by Attridge et al., there is a slightly higher but non-significant increased incidence of SIP in multiple gestations [8, 24]. From data published by Singh et al. it is notable that twins had twice the rate of SIP compared to singletons but this finding was not elaborated in that paper [28]. In a retrospective analysis of putative risk factors for intestinal perforation in premature infants, twin births was noted to be one of the risk factors by Holland et al. [22]. Despite accumulating evidence from previous case reports and retrospective review of risk factors, we believe that this may be one of the first case-control studies describing the association of SIP in twins.
We also studied factors within the twin pairs that could increase the risk for SIP. Interestingly, when only one of the twins developed SIP, it was usually the larger twin that was at increased risk. No previous reports of this observation have been published. There were a few cases in our study in which both twins developed SIP and a similar phenomenon was noted in a published case report [33].
Two infants in our study who had developed SIP early in life later developed ≥stage 2 NEC. Similar to our experience, Drewett et al. described NEC and other gastrointestinal pathology in nine infants who developed SIP initially [34]. This highlights the fact that infants who developed SIP are at higher risk of developing further intestinal pathology and thus should be monitored closely. It is reasonable to speculate that one of the reasons for higher risk of SIP in twins may be a compromise of intestinal blood supply due to placental or other factors. The fact that both twins had developed SIP in some cases and the association with a stillborn twin is some other cases raises the possibility that there may be a role of the in-utero environment in the etiology of this condition. In relation to the relative blood and nutrient flow to the uterus and placenta, it is important to note that at any given week of gestation, the total body weight of twins exceeds that of the singleton infant. Therefore, unlike in a singleton, where only the ipsilateral uterine artery (on the placental side) undergoes changes in pulsatility, in twin gestation both uterine arteries need to undergo similar changes. This makes it more likely that the uterine and placental blood flow may be compromised in twin gestation. Volume flow rates in the uterine arteries in twin pregnancies need to exceed those in singleton pregnancies. Differences in pulsatility index have been noted in uterine arteries in twin versus singleton gestations and this factor may play a role in the higher risk of gut compromise and SIP in twins [35].
Given the limitations of a single center, retrospective, case-control study, our findings need to be interpreted with caution. Multiple gestation pregnancies are prone to preterm labor, so the potential risk of SIP is not insignificant. There is an urgent need for larger studies and corroboration of our findings from other centers to help confirm our findings of increased risk of spontaneous intestinal perforation among multiple gestation births.
Footnotes
Acknowledgments
The authors would like to thank the database management team at Connecticut Children’s Medical Center in Farmington, especially Vicki Pehmoeller, Lisa Dion and Shari Galvin, for their invaluable help in authenticating the data and helping in its collection and retrieval. We are grateful to Dr. Aaftab Husain for his help with collecting data from the charts.
Conflict of interest
The authors declare no conflict of interest.
Author contribution statement
NH was involved in development of the protocol. AM and UP were involved in data collection and statistical analyses. NH, AM and UP were involved in writing the manuscript. NH was the principal investigator.
Funding information
No funding was received for this project.