Neonatal bowel perforations are a large contributor to the surgical disease burden among children and have a high fatality rate, but little information is known about its presentation in low- and middle-income countries. This is one of the first studies to describe the epidemiology of neonatal bowel perforation from a tertiary hospital in Uganda.

We found a small male predominance, though there is no statistical difference between male and female as per this cohort. This is comparable to other gastrointestinal perforation studies done elsewhere [12, 13]. Of note, a majority (80%) of our patients were term neonates, which seems to be contrary to the risk profile, as both SIP and NEC causing perforations appear to be more common in preterm neonates [1]. The higher prevalence of term neonates may be due to the higher likelihood of survival to presentation, as frequently these patients have to travel long distances in order to reach the tertiary health facility. Nevertheless, almost all of the patients were referred in from peripheral health facilities soon after birth, which could explain the early presentation at less than 7 days of life. The average age at diagnosis has been recorded to be 5 days in different studies [2]; however, the mean age at diagnosis for this cohort was 7.85 days. The mean age of presentation is skewed due to the few that arrived when they are many days old, with one even coming at 28 days of age.

Most patients had a birth weight above 2.5 kg, which was consistent with the fact that our cohort had many more term babies than preterm babies. This finding is similar to other studies that have shown significantly more babies with normal birth weight developing bowel perforations than those of low birth weight [13]. This could be because in our low resource setting, term babies are more likely to survive the neonatal period than preterm newborns. Pneumoperitoneum occurred in an average 60% of cases, which corroborated other reports suggesting 80% of cases [1, 12]. The next common finding in our cohort was dilated bowel. There were also instances of having no identifiable radiological findings on abdominal plain film, and this has been documented in other studies [13].

As with prior literature, the most common site of perforation is the colon [1]. There are also reports that document equal occurrence of perforations on ileum and colon [14], and with others showing a higher occurrence in the ileum [2]. In our setting as we report, the colon was the most common site for the perforations. However, perforation anatomic location was diverse ranging from gastric to rectal perforations, which may be explained by the mixed etiology and sites of perforation.

All mothers had access to antenatal care, and almost all mothers (95%) were able to receive folic acid. Only a minority of mothers took medications during pregnancy and 15% received treatment for malaria in the perinatal period. This finding suggests that maternal health may not be a large contributing factor to the likelihood of developing neonatal bowel perforation.

Majority of our babies with bowel perforations had blood group O+, suggesting the need for further research to study the relation between blood group and disease states in the neonatal period.

Outcome of mortality has been reported to be as high as 70% in some studies [12]. In our study, the mortality rate was still high at 55%. This is likely exacerbated by lack of resources and human capital to provide the neonatal intensive care that these patients needed. Furthermore, the lack of parenteral nutrition severely lowered the chances of survival while awaiting bowel function to recover.

Limitations of this study include the single-center nature which may reduce generalizability. Nevertheless, cases are referred from across the country to this national referral center, so a representative sample of the nation’s disease burden is still likely. The small number of cases also reduces the power to discern differences between groups and to make substantial associations. Furthermore, the retrospective observational design does not allow for prospective collection of data, which limits the type of clinical data that could be collected. Relatedly, missing data from chart review opened the potential for selection bias.