HEPATITIS B virus (HBV), hepatitis C virus (HCV), and human cytomegalovirus (HCMV) are viruses of the hepadnavirus, Flavivirus, and herpesvirus groups, respectively.
Although each virus has unique aspects in terms of mode of transmission and risk factors for transmission, all may be transmitted by percutaneous or permucosal exposure to infectious body fluids or by perinatal transmission. 1–3 Epidemiologic evidence worldwide suggests that overall, the sexual transmission of any of these viruses is not the predominant mode of transmission. However, in adolescents and young adults in the United States, multiple studies have identified sexual activity and behaviors as significant risks for acquiring both HCMV and HBV. 1–3 In addition, seroepidemiologic studies demonstrate marked increases in seroprevalence of both HCMV and HBV throughout adolescence and early adulthood. 4–10 11,12
The clinical course of HCMV and HBV are potentially altered in persons infected with HIV, with consequent implications for both the clinical management and the further transmission of these viruses.
There are no reports on the prevalence or risk factors for infection of these viruses in an adolescent population with sexually acquired HIV. The purpose of this study was to examine the seroprevalence and risk factors for HBV, HCV, and HCMV infections in a population of male and female adolescents with sexually acquired HIV infection, and to compare these variables with age- and risk behavior-matched HIV-uninfected adolescents. 13–16
This study reports on data collected from March 1996 to August 1998 as part of an ongoing cohort study of HIV infection in adolescents-the REACH (Reaching for Excellence in Adolescent Care and Health) Project of the Adolescent Medicine HIV/AIDS Research Network (AMHARN). The description of the cohort, recruitment, follow-up, and consent procedures have been described in detail previously.
Briefly, adolescent women and men 13 to 18 years who were known to be HIV-infected and in health care were recruited into the REACH study through 16 different clinical sites in 13 US cities. Regions of the US include (1) northeastern US (New York, NY, Brooklyn, NY, Bronx, NY, Newark, NJ); (2) mid-Atlantic (Baltimore, MD, Washington, DC, Philadelphia, PA); (3) midwestern (Chicago, IL); (4) south-southeastern (Birmingham, AL, New Orleans, LA, Atlanta, GA, Memphis, TN, Miami, FL, Ft. Lauderdale, FL); and western (Los Angeles, CA). All subjects consented to participate in this study according to the guidelines of their respective institutions. 17 The enrollment criterion for this study targets adolescents who by history have acquired HIV through sexual activity or injection drug use. Only two subjects included in the analysis admitted to injection drug use. The study also recruited HIV-uninfected controls of comparable gender, age, and HIV-risk behaviors in a 2:1 ratio of HIV-infected: HIV-uninfected participants. 18
Data Collection and Laboratory Analysis
Demographic information was obtained by a face-to-face interview. Information on sensitive sexual and substance-use behaviors was collected using a confidential, interactive computer interview designed specifically for the REACH study.
All subjects had sera drawn at baseline for HCMV, HBV, and HCV antibodies, and for CD4 T-cell counts and HIV viral load. 17
Quantitative immunophenotyping of T cells for the CD4 marker was performed at the individual clinical sites in certified laboratories using standardized AIDS Clinical Trials Group (ACTG) protocols. CD4 T-cell levels were categorized as follows: < 200, 200 to 499, ≥ 500. HIV plasma viral loads were tested at a centralized laboratory (Children's National Medical Center, Washington, DC) using nucleic acid sequence base amplification (Organon Technika, Durham, NC).
Blood collected at the baseline visit was sent for testing at the research site's local laboratory using standard methods for HCMV antibody, HBV surface antibody, HBV surface antigen, and HBV core antibody. The presence of HCV antibody was detected using an enzyme immunoassay (EIA). All positive tests were confirmed using a second EIA, a recombinant immunoblot assay, or a polymerase chain reaction assay. For the purpose of this analysis, a positive HBV core antibody was used as the outcome variable for HBV seropositivity; too few subjects with HBV surface antigen were available for separate analysis (n = 10). Subjects who had evidence of vaccination (HBV surface antibody positive or core antibody negative) were excluded from the analysis.
Gynecologic examination for female subjects at time initial study visit included urine and endocervical sampling for
Chlamydia trachomatis and Neisseria gonorrhoeae testing. Male examinations included urine testing for C trachomatis and N gonorrhoeae. Samples for C trachomatis and N gonorrhoeae testing were placed into transport media and refrigerated or placed on ice before freezing at −70 C until processed. Samples were sent to a centralized laboratory for processing using the amplification-based ligase chain reaction technique (LCX STD system, Abbott Laboratories, Abbott Park, IL).
Three outcomes were defined for this analysis: HBV infection, HCV infection, and HCMV infection. Because the infection rate for HCV is very low, only HBV and HCMV were further statistically analyzed. Because male and female subjects have different sexual behaviors and different possible risk factors for infection with each virus, we performed the analysis separately by gender.
Potential risk factors that may have been associated with HBV or HCMV infection were first analyzed by univariate methods. These risk factors included (1) sexual activity ever and during the past 3 months (vaginal, oral, receptive anal, insertive anal); (2) unprotected sexual activity during the past 3 months; (3) age of first sexual activity (vaginal for females, vaginal or anal for males); (4) number of lifetime sexual partners; (5) evidence of other sexually transmitted diseases (STDs) (i.e., chlamydia or gonorrhea); (6) infection with HBV as a risk for HCMV infection; (7) infection with HCMV as a risk for HBV infection; (8) self-reported sexual orientation for males (heterosexual versus homosexual, bisexual, or “not sure”); (8) race; (9) region of the country; (10) mother's country of birth; (11) tattooing or body piercing; (12) homelessness; and (13) CD4 count (HIV-infected subjects) and HIV viral load (HIV-infected subjects).
Continuous risk factors were dichotomized or trichotomized. Three different tests were applied to those risk factors according to their characteristics. As a standard method, Pearson chi-square tests were used.
When cell counts were small, Fisher exact tests were applied. 19 In addition, for variables with three or more ordinal levels, Cochran-Armitage trend tests were applied (results not shown). 20 Variables showing at least marginal association with the outcome variables ( 21 P < 0.1) were considered candidates for multivariate analysis.
Logistic regression was used to examine the association between the risk factors and the outcomes simultaneously.
We chose to dichotomize continuous variables to avoid extreme values and linear assumptions. Tree-based modeling was applied to choose the best cutoff points for continuous variables. 22 23
At the time of the analysis, 372 subjects were enrolled in the REACH study. To be eligible for analysis, the subject had to have serology for HBV, HCV, and HCMV completed at the time of first visit. Five subjects were excluded because of missing data; thus, 367 subjects were eligible for the analysis. For analysis of HBV infection, an additional 73 subjects were excluded because of hepatitis B surface antibody (HBsAb) positivity and Hepatitis B core antibody (HBcAb) negativity, which indicated that they had been vaccinated against HBV. Characteristics of the eligible cohort are shown in
Table 1. The percentages of subjects who had positive serology for HBV, HCV, and HCMV, separated by HIV status and gender, are shown in Table 2. TABLE 1:
Demographics of REACH Subjects*
Seroprevalence of HBV, HCV, and HCMV Infections for HIV-Infected and HIV-Uninfected Male and Female REACH Subjects
Hepatitis C Virus
Only 6 of 367 subjects (1.6%) were HCV antibody positive. There was no difference in HCV infection rates between HIV-infected and HIV-uninfected subjects. Four of the six HCV-infected subjects were among the HIV-infected group. Three of these four subjects were female heterosexuals, and the fourth male subject was homosexual. Only one of the four HIV-infected subjects admitted intravenous drug use. The two HIV-uninfected subjects who were HCV infected were both female heterosexuals with no reported intravenous drug use. There was no statistically significant difference in number of lifetime sexual partners between HCV- infected subjects and the remainder of the cohort (
P = 0.41, two-tailed Fisher exact test). Piercing and tattooing were examined as risk factors for HCV infection because 12% of the males and 25% of the females had tattoos and 39% of the males and 48% of the females had piercing; however, these factors were not found to be significant. However, the power to detect the association was small because only six HCV-infected subjects were studied. Because of the small number of HCV-infected subjects, no further analysis of risk factors for HCV infection was performed.
Hepatitis B Virus
Although there was higher seroprevalance of HBcAb positivity in HIV-infected females, this was not significantly different from the seroprevalance among HIV-uninfected females. All male HBcAb-positive subjects were in the HIV-infected group; the seroprevalence in this group was significantly different from that of HIV-uninfected males. Of the 48 HIV-infected subjects who were HBcAb positive, 10 (21%) were positive for hepatitis B surface antigen (HBsAg), which is indicative of an active replicating HBV infection. In comparison, none of the HIV-uninfected subjects who were HBcAb positive were also HBsAg positive (
P = 0.33), and all were hepatitis B surface antibody (HBsAb) positive. The rates for active infection were 15% (5 of 34 subjects) in the female HIV-infected group and 36% (5 of 14 subjects) in the male HIV-infected group. There was also a statistically significant association between CD4 values and current HBV infection in HIV-infected subjects. ( P = 0.035, two-tailed Fisher exact test). However, this finding must be interpreted with caution because we cannot show evidence of persistence in this cross-sectional study, and the number of subjects is small.
Significantly more HIV-infected females were HCMV antibody positive than HIV-uninfected females. Likewise, significantly more HIV-infected males were HCMV antibody positive than HIV-uninfected males.
To determine whether the differences in infection rates for HBV and HCMV could be explained by factors other than HIV status, the possible risk factors (see Methods) were examined in separate univariate analyses for males and females. Injection drug use could not be examined because only one subject who was eligible for this analysis reported ever using injection drugs.
The risk factors that were significant or nearly significant in univariate analyses differ between females and males and between HBV and HCMV. These factors are shown separately by gender and virus in
Tables 3 and 4, and were then considered as possible covariates in the regression models. Because the risk factors were different, four separate models were created: (1) female HBV; (2) male HBV; (3) female HCMV; and (4) male HCMV. Of particular importance is that in our sample there were no HBV infections in HIV-uninfected males or in heterosexual males. Therefore, the regression model for male HBV infection is intended to identify risk factors for HBV infection in HIV-infected, homosexual or bisexual males. TABLE 3:
A Univariate Analysis of Risk Factors for HBV Infection
A Univariate Analysis of the Risk Factors for HCMV Infection
Besides HIV status and sexual preference, age of first sex was also associated with male HBV infection. HBV infection rates were higher for males who had had anal or vaginal sex before 13 years of age (
P = 0.049; odds ratio [OR], 5.4; 95% CI, 1.1-26.9). For females infected with HBV, only number of lifetime partners was significantly associated with HBV infection; those who had more than 10 partners had a higher risk of infection ( P = 0.02; OR, 2.4; 95% CI, 1.2-5.1).
For males infected with HCMV, sexual preference and HIV status were associated significantly with HCMV infection. Homosexual or bisexual males had a higher infection rate (OR, 5.32; 95% CI, 1.2-9.5), as did subjects who were HIV infected (OR, 3.10; 95% CI, 1.1-8.8). For females, only HIV status was associated with HCMV infection (
P = 0.003). Subjects who were HIV infected were more likely to be HCMV positive (OR, 2.4; 95% CI, 1.3-4.3).
We report the results of a seroprevalence and risk-factor analysis for HBV, HCV, and HCMV infections in a unique cohort of HIV-infected and high-risk, HIV-negative adolescents. Aside from the high seroprevalence rates for these viral infections, we have substantiated previous findings of the association of sexual-activity behaviors and increased risk for HCMV and HBV acquisition.
These data support the need for continuing primary and secondary prevention programs in adolescents to reduce the transmission of these viruses, and the need for clinical screening and treatment for HBV and HCMV in adolescents. 4–10
The seroprevalence rates of HBcAb in this cohort are among the highest reported rates for adolescents in areas of low HBV endemicity,
including subpopulations of similar racial composition. These rates approximate the reported rates for high-risk ethnic or immigrant groups within the United States; however, this was not a significant risk factor in our analysis because only 13 mothers of the subjects were born in known endemic areas. In the female cohort in the multivariate analysis, the only significant risk factor among those analyzed that was associated with HBV seropositivity was having more than 10 lifetime sexual partners. Previous studies have identified the number of partners, both lifetime and recent, as risk factors associated with HBV seropositivity among heterosexuals. 5 4
In our sample of males, all of the HBV infections were found in HIV-positive, homosexual or bisexual males. The HBV seropositive rate in the HIV-positive homosexual male cohort was 35% (14 of 40 subjects). Previous studies have found HBV seroprevalence to be as high as 61.5% among older male homosexuals.
In that study, risk factors that were associated with HBV seropositivity were duration of regular homosexual activity, number of nonsteady partners during the past 4 months, anal-genital or oral-anal intercourse, and rectal douching. Our finding of an association of earlier sexual debut with HBV seropositivity is consistent with the primary mode of transmission being related to sexual behavior. The high HBV chronic-carrier rates among HIV infected individuals and the high-risk sexual practices make these male homosexual youth who practice unsafe sex with high-risk partners particularly vulnerable to these coinfections. 6
In the cohort described, the overall rate of HIV infected subjects with HBV infections who had evidence of active replicating HBV is 21% (15% among the females and 36% among the male cohort). This is the first description of these rates in an HIV-infected adolescents who appear to have acquired both HIV and HBV through sexual activity. Given the cross-sectional design of this study, we cannot define any of our subjects with active HBV infections as chronic carriers; however, based on other studies,
we would project that the subjects reported here with active infections are likely to be chronic carriers. There is an extensive literature on the chronic-carrier state of HBV in older male homosexuals infected with HIV. These studies demonstrated increased persistence of HBV 24,25 and more active HBV replication in HIV-infected persons compared with HIV-uninfected persons. 24,26 However, there is no evidence that hepatic necrosis or disease due to HBV progresses more rapidly in HIV-infected persons, and the evidence regarding the effect of active HBV infections on HIV disease progression is not consistent. 25,27 Overall, our data have implications in the area of public health because these youth with active infections are more likely to transmit HBV than they are to be individually affected by HBV. 28,29
The seroprevalence of HCMV infection in the female cohort reported here (69.8%) is similar to that seen in a racially similar adolescent population in an STD clinic (68%),
in a lower-socioeconomic stratified group (67%), 9 and in a lower-income adolescent group in an STD clinic (79%). 8 In our cohort, there was a higher seroprevalence of HCMV in the HIV-infected female group compared with the HIV-uninfected group (78% versus 61%, 7 P = 0.003). The lack of an association of sexual-activity variables other than HIV infection with HCMV seroprevalence in this study might be accounted for either by the high background seroprevalence rates in the total cohort or by the high-risk sexual behaviors of the cohort as a whole. This is one of the first studies that found no association between higher HCMV seroprevalence and black racial background, which may reflect more similarity in the overall socioeconomic status of different racial groups in our cohort than that of previous studies.
The HCMV seroprevalence in the male cohort reported here (69.7%) is considerably higher than that reported in previous seroepidemiologic studies,
but is lower than the rates seen in older male homosexuals in STD clinics (94%). 11 In univariate analyses, multiple sexual-practice risk factors associated with HCMV seropositivity include receptive and insertive anal-genital intercourse, younger age at sexual debut, and higher number of lifetime partners. However, in a multivariate analysis model, only a homosexual or bisexual sexual orientation and HIV seropositivity remained significantly associated with HCMV infection, suggesting that the high-risk sexual activity variables are associated with these two significant risk factors in our cohort. This finding is also similar to the HBV analysis, and suggests that these young homosexual males are being exposed to high-risk partners. 30
The overall prevalence of HCV infection in this adolescent cohort (1.6%) is higher than that reported for adolescents as a whole (0.4%), but similar to the rate reported for non-Hispanic black adolescents (1.2%).
Given the low rate of admitted intravenous drug use in this cohort, which was substantiated by prospective urine drug screening (data not shown), the HCV infections most likely represent sexual transmission. Sexual transmission of HCV most likely occurs, but this transmission is inefficient compared with HCMV, HBV, and HIV. It is estimated that 20% of acute HCV is associated with high-risk sexual behavior, whereas 60% is associated with high-risk drug-taking behavior (10% cannot be accounted for). 29 31
Interpretation of specific relationships between the risk factors examined and the risk of acquiring HCV, HBV, or HCMV reported in this study should be tempered by the cross-sectional design and the retrospective data collection of several risk-factor variables. However, the consistent finding of the association of various sexual risk behaviors and seropositivity is consistent with other studies that suggest the sexual transmission of these viruses. The unique nature of this cohort, including the selection of HIV-negative subjects for their histories of high-risk behaviors and the limited number of adolescents with drug-injecting behaviors, may limit the overall ability to generalize from these data. However, the risk of acquisition of any of these viruses through risky sexual behaviors is clear.
In conclusion, we have demonstrated the high seroprevalence of HBV, HCMV, and HCV in this unique cohort of HIV-infected and HIV-uninfected high-risk adolescents. These data support the risks of acquisition of these infections through sexual behaviors in both male and female adolescents, but particularly in males with same-sex sexual orientation. In addition, the high rates of evidence of active HBV infection in both male and female HIV-infected cohorts indicate a higher risk for secondary transmission of HBV. These data reinforce the need for continuing development and application of both primary and secondary prevention programs directed at adolescents to reduce the acquisition and transmission of these viruses, and the need for screening and treatment of HBV and HCMV.
1. Hollinger FB. Hepatitis B virus. In: Fields BN, Knipe DM, Howley PM, eds. Fields Virology. Philadelphia: Lippincott-Raven, 1995; 2738–2807.
2. Houghton M. Hepatitis C virus. In: Fields BN, Knipe DM, Howley PM, eds. Fields Virology. Philadelphia: Lippincott-Raven, 1995; 1035–1058.
3. Britt WJ, Alford CA. Cytomegalovirus. In: Fields BN, Knipe DM, Howley PM, eds. Fields Virology. Philadelphia: Lippincott-Raven. 1995; 2493–2524.
4. Alter MJ, Alexander PJ, Kramer E, et al. Importance of heterosexual activity in the transmission of hepatitis B and non-A, non-B hepatitis. JAMA 1989; 262:1201–1205.
5. Margolis MJ, Handler SC. Hepatitis B: evolving epidemiology and implications for control. Semin Liver Dis 1991; 11:81–92.
6. Schreeder SE, Hadler SC, Berquist KR, et al. Hepatitis B in homosexual men: prevalence of infection and factors related to transmission. J Infect Dis 1982; 146:7–15.
7. Chandler SH, Wentworth BB, Gutman LT, Wiesner PJ, Alexander ER, Handsfield HH. The epidemiology of cytomegaloviral infection in women attending a sexually transmitted disease clinic. J Infect Dis 1985; 152:597–605.
8. Davis JA, Garvin S. Cytomegalovirus infection: a seroepidemiologic comparison of nuns and women from a venereal disease clinic. Am J Epidemol 1975; 102:327–330.
9. Sohn MK, Balcarok KB, Cloud GA, Pass RF. Cytomegalovirus infection in sexually active adolescents. J Infect Dis 1991; 163:460–463.
10. Stango G, Pass RF, Britt WJ, Alford CA. Factors associated with primary cytomegalovirus infection during pregnancy. J Med Virol 1984; 13:347–353.
11. Wentworth ER. Seroepidemiology of infections due to members of the herpesvirus group. Am J Epidemol 1971; 94:496–507.
12. McQuillan G, Coleman PJ, Kruszon-Moran D, Moyer LA, Lambert SB, Margolis HS. Prevalence of hepatitis B virus infection in the United States: the National Health and Nutrition Examination Surveys, 1976–1994. Am J Public Health 1999; 89:14–18.
13. Spector SA, Wong R, Hsia M, Pilcher M, Stempien J. Plasma cytomegalovirus (CMV) DNA load predicts CMV disease and survival in AIDS patients. J Clin Invest 1998; 101:497–502.
14. Spector SA, Hsia K, Crager M, Pilcher M, Cabral S, Stempien M. Cytomegalovirus (CMV) DNA load is an independent predictor of CMV disease and survival in advanced AIDS. J Virol 1999; 73:7027–7030.
15. Sever JL, Coinfection with herpesviruses in young children of HIV-infected women. Pediatr AIDS HIV Infec 1995; 2:75–82.
16. Kovacs A, Schluchter M, Easley K, et al. Cytomegalovirus infection and HIV-1 disease progression in infants born to HIV-1-infected women. N Engl J Med 1999; 341:77–84.
17. Rogers AS, Futterman D, Moscicki AB, Wilson C, Ellenberg J, Vermund SH. The REACH project of the adolescent medicine HIV/AIDS research network: design, methods, and selected characteristics of participants. J Adolesc Health 1998; 22:300–311.
18. Rogers AS, Schwarz DF, Weissman, G, English A. A case study in adolescent participation in clinical research: eleven clinical sites, one common protocol, and eleven IRBs. Rev Hum Subjects Res 1999; 21:6–10.
19. Fienberg SE. The Analysis of Cross-Classified Data. Cambridge: MIT Press, 1977.
20. Mehta CR, Patel NR. A network algorithm for performing Fisher's exact test in contingency tables. J Am Stat Assoc 1987; 78:427–434.
21. Agresti A. Categorical Data Analysis. New York: John Wiley & Sons, 1990.
22. Hosmer DW, Lemesshow S. Applied Logistic Regression. New York: John Wiley & Sons, 1989.
23. Clark LA, Pregibon D. Tree-based models. In: Chambers JM, Hastie TJ, eds. Statistical Models. Pacific Grove: Wadsworth & Brooks, 1992:377–417.
24. Hadler SC, O'Malley PM, Altman NL, et al. Outcome of hepatitis B virus infection in homosexual men and its relation to prior human immunodeficiency virus infection. J Infect Dis 1991; 163:454–459.
25. Bodsworth ND, Nightingale BN. The effect of concurrent human immunodeficiency virus infection on chronic hepatitis B: a study of 150 homosexual men. J Infect Dis 1989; 160:577–582.
26. Bodsworth NJ, Cooper DA, Donovan B. The influence of human immunodeficiency virus type 1 on the development of the hepatitis B carrier state. J Infect Dis 1991; 163:1138–1140.
27. Krogsgaard KL, Nielsen JO, Andersson P, et al. The influence of HTLV-III infection on the natural history of hepatitis B virus infection in male homosexual HBsAg carriers. Hepatology 1987; 7:37–41.
28. Gilson RJ, Beecham MR, Ross E, et al. Interactions between HIV and hepatitis B virus in homosexual men: effects on the natural history of infection. AIDS 1997; 11:597–606.
29. Alter MJ, Kriszon-Moran D, Nainan OV, McQuillan GM, Gao F, Moyer LA, Kaslow RA, Margolia HS. The prevalence of hepatitis C virus infection in the United States, 1988–1994. N Engl J Med 1999; 341:556–562.
30. Drew WL, Miner RC, Sands M, Ketterer B. Prevalence of cytomegalovirus infection in homosexual men. J Infect Dis 1981; 143:188–192.
31. Alter MJ. Epidemiology of hepatitis C. Hepatology 1997; 26:62S-65S.
The following investigators, listed in order of the numbers of subjects enrolled, participated in this study: University of Miami: L. Friedman, L. Pall, D. Maturo, A. Pasquale; Montefiore Medical Center: D. Futterman, D. Monte, M. Alovera-DeBellis, N. Hoffman, S. Jackson; University of Pennsylvania and the Children's Hospital of Philadelphia: B. Rudy, D. Schwarz; Children's Hospital of Philadelphia: M. Tanney, A. Feldman; Children's Hospital of Los Angeles: M. Belzer, D. Tucker, C. Kallal, D. Fuchs; Tulane Medical Center: S.E. Abdalian, L. Green, M. Ales, L. Wenthold; Children's National Medical Center: L. J. D'Angelo, C. Trexler, C. Townsend-Akpan, R. Hagler, J.A. Morrissy; University of Maryland: L. Peralta, C. Ryder, S. Miller; Cook County Hospital/University of Chicago: L. Henry-Reid, R. Camacho, D. Johnson; Children's Hospital, Birmingham: M. Sturdevant, A. Howell, J.E. Johnson; Children=s Diagnostic and Treatment Center: A. Puga, D. Cruz, P. McLendon; Emory University: M. Sawyer, J. Tigner; St. Jude Children's Research Hospital: P. Flynn, K. Lett, J. Doss; Mt. Sinai Medical Center: L. Levin, M. Geiger; University of Medicine and Dentistry of New Jersey: P. Stanford, F. Briggs; SUNY Health Science Center at Brooklyn: J. Birnbaum, M. Ramnarine. The following investigators have been responsible for the basic science agenda: C. Holland, Center for Virology, Immunology, and Infectious Disease, Children's Research Institute, Children's National Medical Center; A.B. Moscicki, University of California at San Francisco; D.A. Murphy, University of California at Los Angeles; S.H. Vermund, University of Alabama at Birmingham; P. Crowley-Nowick, The Fearing Laboratory, Brigham and Women's Hospital, Harvard Medical School, Boston, MA; S. D. Douglas, University of Pennsylvania and the Children's Hospital of Philadelphia. Network operations and analytic support are provided by C.M. Wilson, C. Partlow at University of Alabama; B. Hobbs, J. H. Ellenberg, L. Paolinelli, S. J. Durako, L. Muenz, R. Mitchell, K. Clingan, P. Ohan, V. Junankar, O. Leytush, A. Bennett, M. Rakheja, C. Xue, Y. Ma, J. Houser at Westat, Inc.