1. Intravenous immunoglobulin for preventing infection in preterm and/or low-birth-weight infants Intravenous immunoglobulin for suspected or subsequently proven infection in neonates Lacy J, Ohlsson A. Arch Dis Child 1995;72:F151-5 Ohlsson A, Lacy J. Cochrane Library, January 1998 Ohlsson A, Lacy J. Cochrane Library, February 2001 Unpublished update September, 2003

2. Background Nosocomial infections continue to be a significant cause of morbidity and mortality among preterm and/or low birth weight infants Nosocomial infections continue to be a significant cause of morbidity and mortality among preterm and/or low birth weight infants Maternal transport of immunoglobulins to the fetus occurs mainly after 32 weeks gestation Maternal transport of immunoglobulins to the fetus occurs mainly after 32 weeks gestation Endogenous synthesis does not begin until several months after birth Endogenous synthesis does not begin until several months after birth

3. Background Administration of intravenous immunoglobulin provides IgG that can: Administration of intravenous immunoglobulin provides IgG that can:  bind to cell surface receptors  provide opsonic activity  activate complement  promote antibody dependent cytotoxicity  improve neutrophilic chemoluminescence

4. IgG IgM

5. IVIG has the potential of: Preventing serious nosocomial infections Preventing serious nosocomial infections Altering the course of congenital or nosocomial infections Altering the course of congenital or nosocomial infections

6. Intravenous immunoglobulin for preventing infection in preterm and/or low-birth- weight infants

7. Objectives To assess the effectiveness/safety of IVIG administration - compared to placebo or no intervention - to preterm (< 37 weeks gestational age at birth) and/or low birth weight (< 2500 g) infants in preventing nosocomial infections To assess the effectiveness/safety of IVIG administration - compared to placebo or no intervention - to preterm (< 37 weeks gestational age at birth) and/or low birth weight (< 2500 g) infants in preventing nosocomial infections

8. Search strategy MEDLINE, EMBASE, and Cochrane Library were searched in November 1997 using keywords: MEDLINE, EMBASE, and Cochrane Library were searched in November 1997 using keywords:  [immunoglobulin] AND [infant-newborn] AND [random allocation OR controlled trial OR randomized controlled trial (RCT)]. The reference lists of identified RCTs, personal files and Science Citation Index were searched.  No language restrictions were applied  Search repeated in February 2001 [+ 4 RCTs; additional data on 298 neonates)  Search repeated in September 2003 (no new data)

9. Criteria used to select studies for inclusion in this (prevention) review were: Design: RCTs in which administration of IVIG was compared to a control group that received a placebo or no intervention Design: RCTs in which administration of IVIG was compared to a control group that received a placebo or no intervention Population: preterm (< 37 weeks gestational age) and/or LBW (<2500 g) infants Population: preterm (< 37 weeks gestational age) and/or LBW (<2500 g) infants

10. Criteria used to select studies for inclusion in this review were: Intervention: IVIG for the prevention of bacterial/fungal infection during initial hospital stay (> 7 days) Intervention: IVIG for the prevention of bacterial/fungal infection during initial hospital stay (> 7 days) Studies that were primarily designed to assess the effect of IVIG on humoral immune markers were excluded as were studies in which the follow-up period was one week or less Studies that were primarily designed to assess the effect of IVIG on humoral immune markers were excluded as were studies in which the follow-up period was one week or less

11. Criteria used to select studies for inclusion in this review were: At least one of the following outcomes was reported: At least one of the following outcomes was reported:  sepsis  any serious infection  death from all causes  death from infection  length of hospital stay  intraventricular haemorrhage  necrotizing enterocolitis  bronchopulmonary dysplasia

12. Data collection & analysis Two reviewers (AO, JL) independently abstracted information for each outcome reported in each study on pre-designed data abstraction forms Two reviewers (AO, JL) independently abstracted information for each outcome reported in each study on pre-designed data abstraction forms One reviewer (AO) checked for any discrepancies and entered the data following consensus into RevMan One reviewer (AO) checked for any discrepancies and entered the data following consensus into RevMan Accuracy of data entry was checked by the two reviewers (AO, JL) comparing RevMan printouts to data abstraction forms and discrepancies were resolved by consensus Accuracy of data entry was checked by the two reviewers (AO, JL) comparing RevMan printouts to data abstraction forms and discrepancies were resolved by consensus One reviewer (AO) pooled the data in RevMan One reviewer (AO) pooled the data in RevMan

13. Statistical analyses Relative Risk (RR) and Risk Difference (RD) with 95% confidence intervals (CIs) using the fixed effects model are reported Relative Risk (RR) and Risk Difference (RD) with 95% confidence intervals (CIs) using the fixed effects model are reported When a statistically significant RD was found the Number Needed to Treat (NNT) was also calculated with 95% CIs When a statistically significant RD was found the Number Needed to Treat (NNT) was also calculated with 95% CIs

14. Statistical analyses The results include all accepted studies in which the outcome of interest was reported The results include all accepted studies in which the outcome of interest was reported Statistically significant heterogeneity was reported Statistically significant heterogeneity was reported Attempts were made to explain statistically significant heterogeneity Attempts were made to explain statistically significant heterogeneity

15. Main results (as of September 2003) Twenty studies met inclusion criteria Twenty studies met inclusion criteria  Included approximately 5,000 preterm and/or LBW infants  Reported on at least one of the outcomes of interest for this systematic review

16. Main results When 10 studies (n = 3,975) were combined there was a statistically significant reduction in sepsis (one or more episodes) When 10 studies (n = 3,975) were combined there was a statistically significant reduction in sepsis (one or more episodes)  RR 0.85 (95% CI 0.74, 0.98)  RD –0.03 (95% CI 0.00, - 0.05)  NNT 33  Test for heterogeneity (p = 0.02)

17. Main results When 16 studies (n = 4,986) were combined there was a statistically significant reduction in any serious infection (one or more episodes) When 16 studies (n = 4,986) were combined there was a statistically significant reduction in any serious infection (one or more episodes)  RR 0.82 (95% CI 0.74, 0.92)  RD –0.04 (95% CI –0.02,-0.06)  NNT 25 (95% CI 17, 50)  Test for heterogeneity (p = 0.0006)

18. Main results There were no statistically significant differences for: There were no statistically significant differences for:  mortality from all causes  mortality from infection  NEC, BPD, IVH  length of hospital stay  no major adverse effects of IVIG were reported in any of the studies  a rise in serum IgG was noted in all studies that reported on this outcome

20. Attempts to explain across study heterogeneity in a systematic review of prophylactic administration of intravenous immunoglobulin (IVIG) in neonates Beyene J, Shah V, Ohlsson A

21. Comparison: IVIG vs placebo/no treatment Outcome: Any serious infection (RR)

22. Funnel Plot: IVIG vs placebo or no treatment. Outcome: Any serious infection (RR)

23. Comparison: IVIG vs placebo/no treatment Outcome: Any serious infection (RD)

24. IVIG vs placebo/no treatment Outcome: Any serious infection (RD)

25. Cumulative meta-analysis: IVIG vs Placebo or No Treatment RR 0.82 (95% CI; 0.74,0.92) RD 0.04% (95% CI 0.02, 0.06)

26. Conclusions Statistically significant in-between study heterogeneity was present due to differences in control group event rates Statistically significant in-between study heterogeneity was present due to differences in control group event rates IVIG administration results in a 3-4% reduction in sepsis and/or any serious infection IVIG administration results in a 3-4% reduction in sepsis and/or any serious infection Is not associated with reductions in other morbidities: NEC, IVH, length of hospital stay or mortality Is not associated with reductions in other morbidities: NEC, IVH, length of hospital stay or mortality Prophylactic use of IVIG is not associated with any short term serious side effects Prophylactic use of IVIG is not associated with any short term serious side effects

27. Conclusions The decision to use prophylactic IVIG will depend on the costs and the values assigned to the clinical outcomes The decision to use prophylactic IVIG will depend on the costs and the values assigned to the clinical outcomes There is no justification for further RCTs testing the efficacy of previously studied IVIG preparations to reduce nosocomial infections in preterm and/or LBW infants There is no justification for further RCTs testing the efficacy of previously studied IVIG preparations to reduce nosocomial infections in preterm and/or LBW infants The results of these meta-analyses should encourage basic scientists and clinicians to pursue other avenues to prevent nosocomial infections The results of these meta-analyses should encourage basic scientists and clinicians to pursue other avenues to prevent nosocomial infections

28. Intravenous immunoglobulin for suspected or subsequently proven infection in neonates Lacy J, Ohlsson A. Arch Dis Child 1995;72:F151-5 Ohlsson A, Lacy J. Cochrane Library, January 1998 Ohlsson A, Lacy J. Cochrane Library, February 2001 Unpublished update September, 2003

29. Objectives To assess the effectiveness of intravenous immunoglobulin (IVIG) to reduce mortality/morbidity caused by suspected infection in newborn infants 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. 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.

30. Selection criteria Design: RCT (including quasi-randomized trials) Design: RCT (including quasi-randomized trials) Newborn infants (< 28 days old) Newborn infants (< 28 days old)

31. Selection criteria Intervention: IVIG for treatment of suspected (and in some infants subsequently proved) bacterial/fungal infection compared to placebo or no intervention 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 Suspected infection was defined as clinical symptoms and signs consistent with infection without isolation of causative organism

32. Selection criteria 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. 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.

33. Selection criteria At least one of the following outcomes was reported: 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.

34. Search strategy, data collection and statistical analyses In general as per systematic review of IVIG for prevention of nosocomial infection In general as per systematic review of IVIG for prevention of nosocomial infection Update February 2001 Update February 2001  Two additional studies were included (total n = 110)  Additional information was obtained from 5 previously published studies  One ongoing international, multi-centre trial based on our previous reviews was identified Update September 2003 (no new data) Update September 2003 (no new data)

35. Main results Six of 9 identified studies (n = 318) reported on the outcomes of all randomized patients with clinically suspected infection Six of 9 identified studies (n = 318) reported on the outcomes of all randomized patients with clinically suspected infection  Mortality was not reduced (borderline statistical significance)  RR 0.63 (95% CI; 0.40, 1.00)  RD – 0.09 (95% CI; 0.00, - 0.17)  No statistically significant between study heterogeneity

36. Main results  Treatment with IVIG (seven trials, n = 262) in cases of subsequently proved infection did result in a statistically significant reduction in mortality n = 262) in cases of subsequently proved infection did result in a statistically significant reduction in mortality RR 0.55 (95% CI; 0.31, 0.98)RR 0.55 (95% CI; 0.31, 0.98) RD -0.09 (95% CI; - 0.01, - 0.18)RD -0.09 (95% CI; - 0.01, - 0.18) NNT 11 (95% CI; 6, 100)NNT 11 (95% CI; 6, 100) There was no statistically significant between-study heterogeneityThere was no statistically significant between-study heterogeneity

40. Methodological concerns Quasi-randomization Quasi-randomization Lack of blinding Lack of blinding Lack of a preset sample size Lack of a preset sample size Lack of “undistinguishable” placebo Lack of “undistinguishable” placebo Small sample size Small sample size

41. Methodological strengths/weaknesses Nine studies conducted in seven countries: Nine studies conducted in seven countries:  India = 2  Mexico  Saudi Arabia  Switzerland  Taiwan  Turkey  U.S = 2

42. Conclusions The addition of the results of two small studies (n = 110) changed the presence of statistical significance for the two major outcomes; The addition of the results of two small studies (n = 110) changed the presence of statistical significance for the two major outcomes;  Mortality in suspected cases of infection (previously significant to non-significant)  Mortality in subsequently proved infection (previously non-significant to significant)

43. Conclusions The reduced mortality following treatment with IVIG for suspected sepsis and the imprecise estimate of the effect size to prevent one death (NNT 11, 95% CI; 6, 100) justify further research The reduced mortality following treatment with IVIG for suspected sepsis and the imprecise estimate of the effect size to prevent one death (NNT 11, 95% CI; 6, 100) justify further research

44. Conclusions The role of IVIG preparations with high concentrations of antibodies to specific organisms should be evaluated The role of IVIG preparations with high concentrations of antibodies to specific organisms should be evaluated If such trials are to be undertaken, the design should include long-term follow-up assessment and cost-effectiveness evaluation If such trials are to be undertaken, the design should include long-term follow-up assessment and cost-effectiveness evaluation

45. Conclusions Researchers should be encouraged to undertake well-designed trials to confirm or refute the effectiveness of IVIG to reduce adverse outcomes in neonates with suspected infection Researchers should be encouraged to undertake well-designed trials to confirm or refute the effectiveness of IVIG to reduce adverse outcomes in neonates with suspected infection

46. Globalize the Evidence Localize the Decision Making