Of 5261 participants enrolled in the NHS, 2671 had an HIV SC window of ≤3 years. Of those, 2352 (88%) had HBV testing results available to determine baseline HB status and were included in these analyses. Compared to the participants in our analyses, the 12% excluded due to not having a categorizable HB status had an earlier year of HIV diagnosis, were more likely to self-report black race/ethnicity, and were less likely to have had prior HB vaccination. Among those included in the analyses, baseline HB status was classified as HB negative (n = 1732; 73.6%), resolved HB (n = 474; 20.2%), isolated HBcAb (n = 82; 3.5%), or chronic HB (n = 64; 2.7%). Characteristics at the time of HIV diagnosis are shown in . Age (overall median, 26.8 years; IQR, 23.2–332.5) differed significantly among the 4 baseline groups, as did the proportion male (n = 2229; 94.8% overall), the year HIV infection was diagnosed (overall median, 1996; IQR, 1991–2002), and the history of HBV vaccination (n = 527; 22.4%) or sexually transmitted infections (n = 849; 36.1%). Among those with a baseline CD4 cell count available, there were no differences between the groups (median, 507 cells/μL; IQR, 374–663). HIV RNA levels, which were generally not available before 1996, were available for 1411 participants (60%) and were similar among the groups (median, 4.4 log10 copies/mL; IQR, 3.7–4.9).
Baseline Characteristics by Hepatitis B Virus Infection Status at Diagnosis of HIV Infection
The Kaplan-Meier curve () suggests that those with chronic HB were at increased risk of an AIDS or death event (P
< .001). During 16
946 person years of follow-up (range, June 1986 to January 2010), there were 469 AIDS-defining or death events (AIDS in 305, death in 164). presents the number and unadjusted rate of AIDS or death events (per 100 person-years of follow-up) for the 4 baseline groups. Compared with those with no HB, the unadjusted rates were significantly higher for those with resolved HB (rate ratio [RR], 1.35; 95% CI, 1.09–1.66), those with isolated HBcAb (RR, 1.54; 95% CI, 1.02–2.34), and those with chronic HB (RR, 2.23; 95% CI, 1.51–3.28).
AIDS or death events by hepatitis B virus infection (HB) status at seroconversion. The number of participants at risk of an AIDS or death event are provided for each HB group. Abbreviation: HBcAb, total antibody to hepatitis B core antigen.
Follow-up, Number of AIDS or Death Events, and Rates (per 100 Person-Years) by Hepatitis B Virus Infection Status at Time of HIV Infection
Results from proportional hazards models are shown in . Without any adjustment for possible confounders (model 1), those with chronic HB (HR, 1.73; 95% CI, 1.17–2.55) or resolved HB (HR, 1.24; 95% CI, 1.01–1.53) had increased risk of an AIDS or death event compared with those who were HB negative, whereas those with isolated HBcAb had increased risk that was not statistically significant (HR, 1.39; 95% CI, .91–2.10). With both of the final MV models (models 2 and 3), the HRs for all HB categories are similar to the model 1 results, but only those with chronic HB were at a significantly increased risk of an AIDS or death event. Comparing those with chronic HB with those who were HB negative, the HRs were similar for the 2 MV model (HR, 1.65 [95% CI, 1.11–2.45] in model 2, which included baseline CD4 cell count categories; HR, 1.80; 95% CI, 1.20–2.69 in model 3, with time-updated categories for nadir CD4 cell count and indicators for use of HAART and non-HAART ART).
Proportional Hazardsa Models for the Risk of AIDS or Death Event
Other covariates that were associated with a significantly increased risk of an AIDS or death event in the MV models included lower CD4 cell count category (models 2 and 3) and HCV status (positive and unknown, models 2 and 3). Factors associated with a decreased risk of an AIDS or death event included black race/ethnicity (compared with white, model 3), later year of HIV diagnosis (models 2 and 3), and use of ART or HAART (model 3). Among those who were HB negative and those who had resolved HB, isolated HBcAb, or chronic HB, respectively, 31%, 25%, 21%, and 27% did not start ART before the event or censoring time; 86%, 81%, 68%, and 66% of those with any ART experience were receiving an HB-active drug (lamivudine, tenofovir, or emtricitabine) sometime during follow-up; and 71%, 64%, 62%, and 55% of those with any ART experience were receiving an HB-active drug at the time of the event or at the time of censoring. The median year of the event or censoring for the 4 groups was 2006, 2004, 2000 and 1999, respectively.
The results from the sensitivity analyses are similar to those from model 3, although the confidence intervals for models 5 and 6 are wider owing to the smaller number of events considered (). With the sensitivity analysis models 4 and 6, those with chronic HB had a consistent and significantly increased risk of an AIDS or death event compared with those with no HB (HR, 1.81 [95% CI, 1.20–2.73] for model 4; and HR, 2.16 [95% CI, 1.26–3.70] for model 6). In Model 5, which censored everyone by 1 January 1996, the results were consistent but not significant (HR, 1.57; 95% CI, .85–2.90). Finally, in an additional model (not shown) similar to model 3 but also including baseline categories for HIV RNA levels (missing, <1000 copies/mL, and ≥1000 copies/mL), the results were consistent with all other models (HR, 1.17 [95% CI, .94–1.46] for resolved HB; HR, 1.12 [95% CI, .73–1.71] for isolated HBcAb; HR, 1.78 [95% CI, 1.19–2.66) for chronic HB).
Risk of AIDS or death events by HB infection status at HIV seroconversion.
The proportional hazards models considered only hepatitis B status at baseline. Of note, 23 (1.3%) of the 1732 who were HB negative at baseline subsequently developed chronic HB, and 9 of those experienced an AIDS or death event after chronic HB. Similarly, 104 (6.0%) of those who were HB negative at baseline subsequently were determined to be HB positive (isolated HBcAb or resolved HB), and 18 of them experienced an AIDS or death event.