512 instance, the optional fourth number used to assign a severity rating for disability codes), such assessment is made within the impairment, disability, or handicap, and not between them. Secondly, Veen et al provide little detail about the identification of criteria for inclusion in the impairment, disability, and handicap categories for each of the eight outcome assessments. General boundaries for category inclusion are presented, but not what specifically marked the difference, for instance, between a hearing impairment and a hearing disability. Thirdly, all impairments are treated as being equally important. However, is it reasonable to assign equal weight to a hearing impairment and to a motor impairment in respect of overall outcome? Finally, Veen et al equate the ICIDH definition of handicap with social disadvantage. Irrespective of the fact that this excludes the other six dimensions of experience defmed by the ICIDH as part of handicap,’ is handicap an appropriate outcome for measurement of the consequences of extreme prematurity or low birthweight? Handicap is the expression of an impairment and/or disability interacting with the environment in which the individual exists. Thus, although disadvantages may exist as a result of impairment or disability, it is only in the interaction with the environment that a handicap is manifested. These deficiencies in the definition of handicap adopted by Veen et al make the interpretation of handicap as measured in this study difficult. For many parents of very premature or low-birthweight babies the critical questions are "Will my baby walk?" or "Will my baby go to normal school?" Appropriate understanding and definition of the differences between impairment, disability, and handicap as health outcomes are clearly needed. If they are to be usable, however, both in respect of epidemiological studies and in planning long-term care they need a more precise definition than that used by Veen et al. Department of Public Health, Darlington Health Authority, Darlington DL3 6HX, UK PHIL C. MACKIE 1. Wood PHN. International classification of impairment, disabilities and handicaps. Geneva: World Health Organisation, 1980. 2. Pharoah POH. Impairments, disabilities and handicaps. Arch Dis Child 1990; 65: 819. Bilirubin in newborn infants SIR,-Dr Benaron and Dr Bowen (July 13, p 78) suggest that bilirubin is consumed as an antioxidant in newborn infants. We recently assayed nine serum proteins, together with indirect and total bilirubin, in both cord and three-day serum samples from 44 newborn infants ranging from 28 to 41 weeks’ gestation. Cord alpha1-antichymotripsin (AACT) activities, a measure of inflammation,l did not show a statistically significant correlation with the rise in bilirubin concentrations in the group as a whole. However, there was a bimodal distribution of AACT activities, with 2 infants having AACT above 75% of pooled normal adult serum (which contained about 45 mg of AACT per dl) and the remainder having less than 35 %. The difference in bilirubin rise between the two groups was significant despite the small numbers (mean rise in total bilirubin 12-4 mg/dl in the low AACT group vs 3-8 mg/dl in the high group; p < 0-001). By contrast, the 2 infants with respiratory distress syndrome and low activities of all three proteinase inhibitors assayed (alphal-antitrypsin, AACT, and alpha2-macroglobulin2,3) had bilirubin rises comparable with those in normal infants. This finding supports the results of Benaron and Bowen. Interestingly, indirect bilirubin concentrations in both cord and three-day serum correlated positively with ceruloplasmin and antitrypsin activities, suggesting an effect of oestrogen on conjugation even at birth. Since quantities of these two proteins are increased as a result of either inflammation or oestrogen/,4 they were not useful indicators of inflammation alone. Unlike the study by Benaron and Bowen, our data suggest that maturational effects of stress on the newborn liver may have had an important role in limiting the rise in indirect and total bilirubin concentrations. Firstly, increased AACT activities in cord serum suggest pre-existing inflammation, since induction of AACT synthesis requires several hours.1 Secondly, the two infants with high AACT activities had low cord concentrations of alpha1- fetoprotein, suggesting relatively mature hepatic function. Thirdly, cord indirect bilirubins were higher, but three-day indirect bilirubins were lower, than in the non-inflamed infants. These findings do not exclude the possibility of bilirubin destruction by free radicals. University of North Carolina School of Medicine at Chapel Hill, Moses Cone Memorial Hospital, Greensboro, North Carolina 27401-1020, USA A. MYRON JOHNSON 1. Ganrot K. Plasma protein pattern in acute infectious diseases Scand J Clin Lab Invest 1974; 34: 75-81. 2. Evans HE, Levi M, Mandl I. Serum enzyme inhibitor concentrations in the respiratory distress syndrome. Am Rev Resp Dis 1970; 101: 359-63. 3. El-Bardeesy MW, Johnson AM. Serum proteinase inhibitors in infants with hyaline membrane disease. J Pediatr 1972; 81: 579-87. 4. Ganrot PO. Variation of the concentrations of some plasma proteins in normal adults, in pregnant women and in newborns. Scand J Clin Lab Invest 1972; 29 (suppl 124); 83-88. Sensitivity of HTLV-I antibody assays for HTLV-II S:R,&mdash;Concern about infection with the human T-lymphotropic viruses (HTLV) types I and II has prompted blood transfusion services to screen donors. Most assays available for screening use disrupted whole HTLV-1 antigen.1 These assays also detect some people infected with HTLV-11, and reactive sera are referred to as HTLV-1/11 reactive. Kline et al2 evaluated seven assays, finding over 90% sensitivity and specificity on sera collected in Haiti. However, studies that discriminate between HTLV I and II on the basis of genomic differences indicate that Haiti, like other Caribbean areas, is endemic for HTLV-I, not HTLV-11 (W. A. B., unpublished). In the USA, about half of HTLV-I/II reactive blood donors are infected with HTLV-II, probably because of the high prevalence of this virus in intravenous drug users (IVDU).3-S We present evidence that some HTLV-1 assays are relatively insensitive in detecting HTLV-II. We screened sera from 9226 pregnant women from Jamaica, an HTLV-I endemic area, and 223 IVDUs from New Orleans in whom over 80% of seropositive individuals are infected with HTLV-11.’’ We used a commercially available whole-virus enzyme immunoassay (EIA) (Dupont) and an investigational EIA which has as antigen a recombinant envelope peptide from a conserved region of HTLV I and II (Cambridge Bioscience). Sera positive in either test and all IVDU sera were confirmed by HTLV-I immunoblotting supplemented with a recombinant HTLV-I envelope protein rgp21e (Biotech).6 gag (p24) and env (rgp2le and/or gp46) reactivity were required for positivity. Although the immunoblot uses HTLV-I it detects HTLV I and II antibody with similar sensitivity (S. Z. W., unpublished). Samples positive on either EIA but not confirmed on immunoblotting were excluded SENSITIVITY OF HTLV-I EIAs IN DETECTING ANTIBODIES IN THREE POPULATIONS WITH DIFFERRENT PROPORTIONS OF HTLV-I AND HTLV-II Group A= pregnant Jamaican women, 100% HTLV-I on basis of polymerase chain reaction testing of other Jamaican samples, group B=New Orleans IVDU, 82% HTLV-II, on basis of polymerase chain reaction testing of 49 seropositive samples, and group C= New Orleans outpatients, 64% HTLV-II on basis of serological testing with type-specific synthetic peptides (Olympus). WV (D) and WV (A)=whole-virus Dupont or Abbott, Rec=recombmant (Cambridge) Pos=positive, Neg=negative, Sens=sens!t!V!ty, Spec=speaficity, PPV and NPV= posit!ve and negatme predictive values