Intravenous immunoglobulin for suspected or subsequently proven infection in neonates

Congenital and nosocomial infections are important causes of neonatal morbidity and mortality. Maternal transport of immunoglobulins to the fetus mainly occurs after 32 weeks gestation and endogenous synthesis does not begin until several months after birth. Administration of intravenous immunoglobulin provides IgG that can bind to cell surface receptors, provide opsonic activity, activate complement, promote antibody dependent cytotoxicity, and improve neutrophilic chemoluminescence. Theoretically infectious morbidity and morbidity could be reduced by the administration of intravenous immunoglobulin.

To assess the effectiveness of intravenous immunoglobulin (IVIG) to reduce mortality/morbidity caused by suspected infection in newborn infants. In secondary analyses to assess the effectiveness of IVIG to reduce mortality/morbidity in those neonates who entered into the studies with suspected infection and who later were confirmed as being infected.

Medline, Embase, and Reference Update Databases were searched in November 1997, and the Cochrane Library in July 1998 using the following keywords: immunoglobulin and infant-newborn, and random allocation, or controlled trial, or randomized controlled trial (RCT). The reference lists of identified RCTs and meta-analyses, personal files and Science Citation Index were also searched. No language restrictions were applied. Unpublished data were requested from authors and information has been obtained from one author to date.

The criteria used to select studies for inclusion were: 1) DESIGN: RCT (including quasi-randomized trials) 2) Newborn infants (< 28 days old) 3) INTERVENTION: IVIG for treatment of suspected (and in some infants subsequently proved) bacterial/fungal infection compared to placebo or no intervention. Suspected infection was defined as clinical symptoms and signs consistent with infection without isolation of causative organism. Proved infection was defined as: clinical symptoms and signs consistent with infection in association with isolation of causative organism from either blood culture, cerebrospinal fluid culture, urine culture (urine obtained by suprapubic tap) or a normally sterile site (e.g., liver, spleen, meninges, lung) at autopsy. 4) At least one of the following outcomes was reported: mortality during initial hospital stay; length of hospital stay; side effects; psychomotor development/growth at follow up.

Two reviewers independently abstracted information for the outcomes of interest and one researcher (AO) checked for any discrepancies and pooled the results. Relative risk (RR) and Risk Difference (RD) with 95% confidence intervals (CI) using the fixed effects model are reported for dichotomous outcomes and weighted mean difference (WMD) for continuous data. NNT was calculated for outcomes that showed a statistically significant reduction in RD. Data from quasi-randomized trials were excluded in sensitivity analyses.

Study quality was generally poor. Four of 7 identified studies (n = 208), reported on the outcomes of all randomized patients with clinically suspected infection. Mortality was reduced [RR 0.52 (95% CI; 0.28, 0.98), RD -0.102 (95% CI; -0.005, -0.199, NNT 10 (95% CI; 5, 200]. When, in a sensitivity analysis (n = 126), the results from a quasi-randomized trial (n = 82) were excluded, RR and RD remained similar, [RR 0.53 (95% CI; 0.25, 1.15), RD -0.106 (-0.232, 0.021)] but statistical significance was lost. Treatment with IVIG (six trials, n = 234) in cases of subsequently proved infection did not result in a statistically significant reduction in mortality [RR 0.62 (95% CI; 0.34, 1.12, RD -0.074 (95% CI; -0.163, 0. 014)]. Excluding in a sensitivity analysis (n = 199) a quasi-randomized trial (n = 35) changed the results slightly [RR 0. 68 (95% CI; 0.36, 1.29), RD -0.059 (-0.152, 0.035)]. There was no statistically significant between-study