The process for selection of the studies is outlined in . Two studies were published before 1990, all other studies were published between 2001 and 2009.
A single-centre RCT (LOE 1) published in 1989 addressed prophylactic surgical ligation of the PDA (10
). Subjects with BW <1000 g were randomly assigned to prophylactic surgical ligation of a PDA within 24 h of birth or medical care. The incidence of NEC was higher in the control group compared with the surgery group (13 of 44 versus three of 40, P=0.002), but there was no difference in the outcomes of severe intraventricular hemorrhage (IVH), bronchopulmonary dysplasia (BPD), retinopathy of prematurity or death. Mosalli et al (11
) reviewed this RCT in their Cochrane systematic review.
Other studies examining the relationship between primary surgical closure of the PDA and the development of NEC in preterm infants were observational, reporting the experience of centres using nonrandomized, concurrent controls for comparison (LOE 2). The studies varied in the first-line treatment with medical management, indomethacin treatment or indomethacin prophylaxis.
Five studies (n=918) used indomethacin as first-line treatment or after failed medical therapy (12
). Within each of these studies, there was a group of infants who underwent primary surgery (without any prior exposure to indomethacin) for a clinically and echocardiogram-identified PDA, allowing for comparison with an indomethacin treated group. Most studies also described a group treated with secondary surgery. These observational studies were within single centres (12
) or two centres (16
Retrospective observational studies comparing outcome of necrotizing enterocolitis (NEC) among groups treated for patent ductus arteriosus (PDA)
One study (13
) reported a statistically nonsignificant trend toward less NEC in the primary surgery group compared with a medical treatment group and an indomethacin treatment group. There were only three cases of NEC in this small sample (n=57), an incidence of 5.2%. Four studies (12
) reported comparison between two groups: indomethacin treatment and primary surgery. Two of the studies (14
) did not show a statistically significant difference in NEC outcome between the two groups. One study (12
) reported a higher incidence of the combined outcome of NEC and spontaneous intestinal perforation (SIP) in the indomethacin treatment group. The fourth study (16
), published in 1987, showed a statistically significant increase in NEC in the primary surgery group.
In a retrospective study that used the database of a multicentre health care provider group (n=12,581) (17
), comparison was made among five groups: prophylactic indomethacin, indomethacin treatment, primary surgery, PDA without treatment and no PDA. There was no difference in the incidence of NEC across the groups, except for the comparison between the primary surgery and no PDA groups ().
Two other studies were pertinent to our review question. A study (18
) using regression models on data from the National Institute of Child Health and Human Development (n=2383) showed no difference in the adjusted risk of NEC (OR 1.22 [98.3% CI 0.67 to 2.24]) between the indomethacin treatment group and the primary surgery group; acknowledging that 13% of the surgery group was exposed to prophylactic indomethacin (LOE 2). Compared with indomethacin treatment, infants undergoing primary surgery were at increased risk of BPD (OR 2.19 [98.3% CI 1.16 to 4.15]) and borderline increased risk of neurodevelopmental impairment at 18 to 22 months of age (OR 1.79 [98.3% CI 0.998 to 3.21]). Chorne et al (19
) reviewed infants <28 weeks’ gestational age (GA) (n=446) from a single centre, who were all exposed to prophylactic indomethacin. Using multivariable regression analysis to investigate potential predictors of adverse outcomes, controlling for GA and neonatal morbidities, none of the treatment variables, including primary PDA ligation or indomethacin treatment, were predictive for development of NEC (LOE 2). The authors suggested that primary PDA ligation was associated with increased risk of chronic lung disease (CLD).
Quality of the studies
The studies were only of fair quality, and were limited by lack of clear description of the comparison groups, clearly defined outcomes, and known confounders identified and controlled for <www.cebm.net/index.aspx?o=1157