33 272 Infants, 7-Year Follow-up: Total Serum Bilirubin, Transfusions Reexamined

To the Editor.—The long-term effect of elevated total serum bilirubin (TSB) levels on neurologic outcome remains controversial.1 In 1993 we reported an analysis of data from the US Collaborative Perinatal Project, a cohort study of infants born 1959–1966 that included neurologic examinations at age 7 on 33 272 children.2 We found a statistically significant association between TSB level ≥20 mg/dL and “abnormal or suspicious” neurologic examination (22.4% vs 15.1%; P < .001), but no significant association between peak TSB level ≥20 mg/dL (342 μmol/L) and “abnormal” neurologic examinations (4.5% vs 3.8%; P = .66) at age 7 years.Brown and Johnson3 and Johnson and Bhutani4 have speculated that the reason we found no association between TSB ≥20 mg/dL and definite neurologic abnormalities is that abnormalities were prevented by exchange transfusions, done in 53% of the infants with TSB ≥20 mg/dL. To examine this possibility, we stratified the analysis by whether an exchange transfusion was done. We used SAS (SAS Corp, Cary, NC) to generate the table and Stata 6.0 (College Station, TX) to calculate risk ratios, exact P values, and 95% confidence intervals, and to test for homogeneity of the Mantel Haenszel relative risk.5 (This last test is to investigate the statistical significance of differences in the relative risk in children who did and did not receive exchange transfusions.)Results are shown in Table 1. Although there was a trend toward a higher relative risk in children who had not had exchange transfusions, neither this relative risk itself nor the test for a difference in relative risk by exchange transfusion history approached statistical significance. Although it is possible that exchange transfusion may have prevented neurologic abnormalities in some of the patients in whom it was done, the risk of abnormalities in infants with TSB ≥20 mg/dL who did not receive an exchange was also low, and similar to the risk in children who had lower TSB levels. We conclude that the low risk of abnormal 7-year neurologic examinations in children whose maximum bilirubin levels were ≥20 mg/dL in the CPP is not primarily attributable to use of exchange transfusions.