The natural history of HIV in hemophiliacs before the introduction of highly active antiretroviral therapy (HAART) has been well described in cohorts from Western countries.1-6 HAART has reduced HIV progression and AIDS-associated mortality in hemophilic men, which has resulted in an increase in the cumulative incidence of non-AIDS-related mortality.7-10 HIV-infected hemophiliacs are also at risk for death from other non-AIDS causes, such as liver disease and hemorrhagic complications, that were common in this population before the HIV epidemic and may have re-emerged in the HAART era.11-18 To what extent the increases in non-AIDS-related mortality in the HAART era are attributable to the decreases in AIDS mortality requires analyses that take into account the competing risks among the different causes of death.
Another important factor to take into account in these types of studies is the effect of age on HIV progression. Because most hemophiliacs became infected by HIV in the late 1970s and early 1980s, the impact of HAART is inevitably confounded by the survival bias introduced by the patients who have lived long enough as to receive treatment and have thus had a greater chance to develop hepatitis C virus (HCV)-related disease. Because age is one of the strongest determinants of HIV progression,3,5,6,8,9,19 those infected at a younger age may have benefited more from the introduction of HAART.
The Spanish Multicenter Study Group of Seroconverters (GEMES) has identified all hemophiliacs who became infected by HIV in the late 1970s and early 1980s in 3 of the largest hemophilia units in Spain. The aim of these analyses is to study the trends in progression to AIDS, all-cause mortality, and cause-specific mortality (AIDS-related, liver disease, and hemorrhagic complications) over calendar periods with different exposure to HAART in a cohort of hemophiliacs in Spain, taking into account the competing risks of the causes of death.
PATIENTS AND METHODS
The GEMES identifies HIV seroconverters from all transmission categories retrospectively or prospectively from the 1980s to the current date and follows them up over time. All subjects who fulfill the criteria of a "seroconverter," an individual in whom the date of HIV seroconversion can be reliably ascertained, are included in the study, which has been described in detail elsewhere.20 All hemophiliacs with a diagnosis of HIV infection ever seen in any of the 3 hemophilia units in the hospitals La Paz in Madrid, Vall d'Hebron in Barcelona, and Virgen del Rocío in Sevilla were included in this retrospective cohort.
Ascertainment of all HIV-1-positive hemophiliacs ever seen in these hospitals was complete; records were well kept for medical and legal reasons, because the Spanish National Health System compensated these persons by 1993. Previous HIV-negative test results, dated in 1979, were only available for some cases from Vall d'Hebron Hospital in Barcelona.21 All subjects born before 1979 were assumed to have become infected with HIV between 1979 and 1985, the date by which all plasma concentrates were, by law, heat-treated in Spain. For those born after 1979, date of birth was taken as the date of start exposure to HIV. Date of HIV seroconversion was estimated through the probability of cumulative distribution of infections using mathematic techniques for interval-censored data from 1979 to 1985.22
The medical records of these patients were reviewed by the clinicians in charge of their medical care to collect information on sociodemographic characteristics, (age and sex), clinical data regarding the type of hemophilia (hemophilia A and B) and severity (severe if factor level <1%, moderate if factor level 1%-5%, and mild if factor level >5%), clinical data regarding HIV infection, AIDS-defining conditions during follow-up, antiretroviral treatments and duration of those, CD4 counts and viral load measurements, and cause of death. The cause of death was that identified by the senior clinician after inspecting the medical records. When the cause of death could not be found, it was recorded as unknown, even if it occurred in a patient with an AIDS diagnosis. The groups used in the current analyses were created after inspecting the distribution of individual causes of death that could be grouped in categories with clinical entities and sufficient numbers of events to perform analyses.
The entry date for the analyses was staggered to the date of the first HIV-positive test result in each of the participating centers or 3 years after estimated seroconversion, whichever occurred later, to correct for the potential survivor bias introduced in the analysis when including patients who might have been identified because they survived long enough, thus excluding rapid progressors. Patients were censored on the last time they were seen alive before December 2003. Updated information from clinic notes was terminated by March 2004, and although completeness of follow-up was high, to minimize potential losses, cross-checks with the Spanish National AIDS Register were performed by September 2004. AIDS was defined according to the 1993 European definition.23
Follow-up time was divided into the following calendar periods, which reflect availability of the different antiretroviral treatments: before 1992 (no therapy or zidovudine monotherapy), 1993 through 1996 (combination therapy), 1997 through 2000 (early HAART), and 2001 through 2003 (late HAART). The incidence rates of AIDS (only initial events), death from all causes, and death from specific causes were calculated for each of these periods using Poisson regression, allowing for late entry to the first HIV-positive visit to the recruiting center. Analyses were adjusted for age at seroconversion, type of clotting disorder (hemophilia A or B), severity of the clotting disorder (mild, moderate, or severe), and the previously described calendar periods (fitted as time-dependent covariates). To explore the modifying effect of these variables on the effectiveness of HAART, interactions were looked for in the regression models.
For cause-specific mortality, causes of death were divided into AIDS-related deaths, deaths from liver disease, deaths from hemorrhagic complications, deaths from other causes, and deaths from unknown causes. To study the evolution of the causes of death over time, 2 approaches were taken to allow for the potential interdependence of the different causes. To calculate cause-specific rates, a standard censoring strategy was used: subjects dying from causes different from the cause of interest were censored at the date of death. Subjects lost to follow-up were censored on their last known contact, and the rest were censored by December 2003. This approach assumes that causes of death are independent and that censoring is noninformative. When multiple causes of death are present, censoring is informative because it is dependent on the probability of survival. Ignoring this would produce biased results that overestimate the risk of death from the cause of interest. In this context, there is a need to use a competing risk approach. Under the competing risks assumptions, when studying the risk of death from cause A, subjects dying from cause B were not censored at their date of death; instead, they were censored at the end of the maximum follow-up period, and thus contribute to the denominator.24,25
Analyses were performed in Stata 8.0 (StataCorp, College Station, TX) using robust methods to estimate confidence intervals.
Sociodemographic and Clinical Characteristics
A total of 585 subjects were included. Sociodemographic and clinical characteristics are described in Table 1. Two hundred fifty-eight (44%) were younger than 15 years of age at the time of HIV seroconversion, 505 (86%) had type A hemophilia, and 478 (82%) had severe hemophilia. The median date of estimated seroconversion was October 1982, the median date of entry into the cohort was November 1985, the median date of AIDS diagnosis was October 1990, and the median date of death was December 1992.
TRENDS IN AIDS RATES OVER TIME
AIDS rates were strongly influenced by age and calendar period (Fig. 1). Subjects older than 35 years of age had higher AIDS rates in all calendar periods except for the late-HAART period, a period during which AIDS rates fell dramatically in all age groups compared with the mid-1990s, and differences among age groups were not statistically significant.
The strong calendar period effect observed in univariate analyses is maintained in multivariate analyses. Compared with 1993 through 1996, decreases of 75% (relative risk [RR] = 0.25, 95% confidence interval [CI]: 0.14 to 0.43) and 72% (RR = 0.26, 95% CI: 0.12 to 0.63) in the risk of AIDS were observed in the early- and late-HAART periods, respectively. Age remains a strong determinant of progression to AIDS (Table 2). Having severe hemophilia was associated with a 94% higher risk of progression to AIDS compared with mild or moderate hemophilia (see Table 2). As shown in Figure 2, a gradual increase in the uptake of HAART was seen from 1996 onward for all age groups.
TRENDS IN MORTALITY RATES OVER TIME
Three hundred fifty-four (60%) persons died, and all-cause mortality was strongly influenced by age at HIV infection and calendar period (Fig. 3). For all age groups, mortality rates peaked by 1993 through 1996. Mortality was higher for subjects older than 35 years of age, although for this group, mortality did not decrease in the early-HAART period but dropped in the late-HAART period. For hemophiliacs younger than 35 years of age, mortality dropped in the early-HAART years and remained low thereafter (see Fig. 3). The strong calendar period association observed in crude analyses is maintained in multivariate analyses (Table 3). Compared with 1993 through 1996, there was a 72% decrease in all-cause mortality during 1997 through 2000 (RR = 0.28, 95% CI: 0.18 to 0.42) and an 83% (RR = 0.17, 95% CI: 0.09 to 0.33) decrease for 2001 through 2003.
Of the 354 deaths, 208 were AIDS related, 34 were related to liver disease, 11 were related to hemorrhagic complications, and 21 were related to other causes. A significant variation in the proportional mortality attributable to different causes was observed in different calendar periods. During the period 1993 through 1996, among those with known causes of death, 86% of deaths were attributable to AIDS and 6% were attributable to liver disease. By 2001 through 2003, these proportions were 9% and 81%, respectively (Table 4).
Cause-specific mortality rates are shown in Figure 4. Whereas a marked decline is seen for AIDS-related deaths in the early-HAART and late-HAART periods, an upward trend is observed for liver-related deaths, which had become the most common cause of death by the period 2001 through 2003. Differences between AIDS- and liver-related death rates for period 2001 through 2003 are not statistically significant, however. No change is observed for deaths from hemorrhagic complications or for those from other causes. For deaths from unknown causes, the pattern resembles that of AIDS-related deaths.
Multivariate analyses for specific causes of death are described in Table 3. For AIDS-related deaths, determinants and trends were similar to those for deaths from all causes. For liver-related deaths, a steady increase is observed in the early-HAART period, and by the late-HAART period, deaths from liver disease were 3 times more common than in 1993 through 1996 (RR = 2.80, 95% CI: 0.94 to 8.36; see Table 3). Age at HIV infection and the severity of hemophilia also had strong associations with liver-related deaths (see Table 3).
No statistically significant changes were seen for trends over time for deaths from hemorrhagic complications. Being older than 35 years of age carried a 7-fold increase in the risk of death (RR = 6.92, 95% CI: 1.20 to 39.92). Having severe hemophilia was associated with a 75% reduction in the risk of death (RR = 0.25, 95% CI: 0.07 to 0.97), however, compared with mild or moderate hemophilia. For death from other causes, nonstatistically significant declines are seen for the early- and late-HAART periods. For death from unknown causes, the observed pattern resembles that of AIDS-related death (see Table 3).
There were differences between the cause-specific model approach and the competing risk approach. Allowing for competing risks (Table 5), the increases observed for liver-related deaths in period 2001 through 2003, although still apparent, were smaller (RR = 1.70, 95% CI: 0.57 to 5.04) than those in the cause-specific model (RR = 2.80, 95% CI: 0.94 to 8.36).
Remarkable reductions in the risks of AIDS and all-cause mortality were seen in hemophiliacs in Spain in the early- and late-HAART periods compared with 1993 through 1996. This increased survival has been largely attributed to reductions in AIDS-related deaths and has been accompanied by an increase in the number of persons at risk for death from liver-related conditions from 1997 onward. By 2001 through 2003, liver disease was the most common cause of death in HIV-infected hemophiliacs. Although an increase in the cumulative incidence and cause-specific rate is observed for liver-related deaths in period 2001 through 2003, analyses that do not take into account the competing risks overestimate this rise. Taking the competing risks into consideration suggests that these increases are largely secondary to the decline in AIDS mortality rather than to primary increases in liver-related deaths. This finding is coincident with the natural history of both infections, because the incubation period for HCV is longer than that for AIDS and AIDS was an important competing cause of death in the pre-HAART era. No changes over time in the death rate attributable to hemorrhagic diseases or other causes are observed. Because the changes coincide with the widespread use of HAART, which was prescribed to more than three quarters of cohort members, we can infer that HAART has had a large impact on the hemophilia population in our setting.
Other authors have previously described the reductions in AIDS rates and all-cause mortality experienced by hemophiliacs and have attributed it to HAART.7,8,9,10,26 It is worth pointing out that most of these patients were not naive to antiretrovirals when started on HAART. The increasing importance of liver disease as a rising cause of death in hemophiliacs has been reported by numerous authors,8-10,13-16,18 although some claim that HAART may also have a beneficial effect on liver disease mortality.26 In the United Kingdom Haemophilia cohort, liver disease accounted for 25% of deaths from 1997 through 1999.10 We report a similar proportion (27%) in 1997 through 2000, which rose to 76% by 2001 through 2003. The leading role of liver disease as a major cause of death in HIV-infected patients is also being reported in other groups with high rates of HIV-HCV coinfection, such as intravenous drug users,25,27-30 although its relative importance is determined by the competing risks operating in those populations.
Liver-related deaths in our population are likely to be associated with long-standing HCV infection, although other causes, such as hepatitis B virus (HBV) infection and drug hepatotoxicity, may also play a role.31,32 The fact that most of the excess of liver-related deaths is attributable to longer survival and the higher number of persons at risk goes against a synergistic effect of HAART and hepatitis virus in the evolution to end-stage liver disease. A low number of patients were treated for HCV infection, because HCV treatments are difficult to follow and have multiple side effects, particularly in this population.33 Weighted analyses need to be conducted regarding treatment, given the rising importance of liver-related deaths in hemophiliacs.
This work also highlights the well-described strong effect of age at seroconversion on HIV progression to AIDS, all-cause mortality, and cause-specific mortality.3,5,6,8,19,20 Reductions in AIDS and mortality rates were seen in subjects of all ages compared with 1996 through 1997, although for subjects infected with HIV who were older than 35 years of age, decreases in mortality are only apparent in the late-HAART period.
Subjects with severe hemophilia had higher rates of AIDS and all-cause mortality compared with those with mild or moderate hemophilia. Whereas higher rates were observed for AIDS and liver disease, severe hemophilia carried a lower risk of death from hemorrhagic complications among HIV-infected hemophiliacs. This is indeed an unusual finding and may be related to closer surveillance of hemorrhagic complications in these patients. Nevertheless, no changes are seen over time. The higher death rate from liver disease in subjects with hemophilia B has not been described in the literature and may be a chance finding.
This is the largest study conducted in Spain in HIV-infected hemophiliacs. Dates of HIV infection have been estimated homogeneously and allow for adequate classification of individuals by the maturity of their HIV infection.22 Individuals were "left-censored" to their first HIV-positive visit to the hospital to minimize the bias introduced by missing early deaths,34 and using competing risks models allows one to correct for the informative censoring that takes place when multiple causes of death are present.24,25 Loss of follow-up in this cohort was low, but to minimize it, cross-checks were performed with the National AIDS Register. Cause of death was missing in 22% of the reported deaths; however, because they follow a similar pattern to AIDS-related deaths, they are likely to be AIDS-related. We have not been able to detect increases of causes of death other than liver disease, and this may be attributable to lack of power for infrequent causes. Larger collaborations are needed to explore these issues in more depth.
In summary, major reductions in AIDS and death rates are observed from 1997 to 2003 in hemophiliacs in Spain. These survival gains are largely attributable to HAART and have been accompanied by an increase in the death rate from liver disease, which is overestimated if competing risks are not taken into account. Treatment of HCV infection and end-stage liver disease has become the next challenge in this population.
Julia del Amo, Santiago Pérez-Hoyos, Patricia García de Olaya, and Ildefonso Hernández initiated this project. These authors, together with Manolo Quintana, Isabel Ruiz, José Miguel Cisneros, Charo Pérez, and Cristina González, were responsible for study design and data collection. Santiago Pérez-Hoyos, Alicia Moreno, and Inmaculada Ferreros were responsible for statistical analyses. Julia del Amo wrote the first draft of the paper. All authors were involved in the study, commented on interim drafts, and reviewed the final manuscript.
1. Lee CA, Phillips AN, Elford J, et al. Progression of HIV disease in a haemophilic cohort followed for 11 years and the effect of treatment. BMJ
2. Rosenberg PS, Goedert JJ, Biggar RJ. Effect of age at seroconversion on the natural AIDS incubation distribution. Multicenter Hemophilia
Cohort Study and the International Registry of Seroconverters. AIDS
3. Eyster ME, Gail MH, Ballard JO, et al. Natural history of human immunodeficiency virus infections in haemophiliacs: effects of T-cell subsets, platelet counts, and age. Ann Intern Med
4. Darby SC, Ewart DW, Giangrande PL, et al. Mortality before and after HIV infection in the complete UK population of haemophiliacs. UK Haemophilia Centre Directors Organisation. Nature
. 1995; 377:79-82.
5. Touloumi G, Karafoulidou A, Gialeraki A, et al. Determinants of progression of HIV infection in a Greek hemophilia
cohort followed for up to 16 years after seroconversion. J Acquir Immune Defic Syndr Hum Retrovirol
6. Collaborative Group on AIDS Incubation and HIV survival including the CASCADE EU Concerted Action. Time from HIV-1 seroconversion to AIDS and death before widespread use of highly-active antiretroviral therapy: a collaborative re-analysis. Collaborative Group on AIDS Incubation and HIV Survival including the CASCADE EU Concerted Action. Lancet
7. Sabin CA, Yee TT, Devereux H, et al. Two decades of HIV infection in a cohort of haemophilic individuals: clinical outcomes and response to highly active antiretroviral therapy
. 2000; 14:1001-1007.
8. Quintana M, Del Amo J, Barrasa A, et al. Progression of HIV infection and mortality by hepatitis C infection in patients with haemophilia over 20 years. Haemophilia
9. Chorba TL, Holman RC, Clarke MJ, et al. Effects of HIV infection on age and cause of death for persons with haemophilia A in the United States. Am J Hematol
10. UK Haemophilia Centre Doctors Organisation. The impact of HIV on mortality rates in the complete UK haemophilia population. AIDS
11. Rosendaal FR, Varekamp I, Smit C, et al. Mortality and causes of death in Dutch haemophiliacs, 1973-86. Br J Haematol
. 1989;71: 71-76.
12. Prins M, Sabin CA, Lee CA, et al. Pre-AIDS mortality and its association with HIV disease progression in haemophilic men, injecting drug users and homosexual men. AIDS
. 2000;14: 1829-1837.
13. Diamondstone LS, Aledort LM, Goedert JJ. Multicentre Hemophilia
Cohort Study. Factors predictive of death among HIV-uninfected persons with haemophilia and other congenital coagulation disorders. Haemophilia
14. Ragni M, Belle S. Impact of human immunodeficiency virus infection on progression to end-stage liver disease in individuals with hemophilia
and hepatitis C virus
infection. J Infect Dis
15. Darby SC, Ewart DW, Giangrande PL, et al. Mortality from liver cancer and liver disease in haemophilic men and boys in UK given blood products contaminated with hepatitis C. UK Haemophilia Centre Directors Organisation. Lancet
16. Goedert JJ, Eyster ME, Lederman MM, et al. End-stage liver disease in persons with hemophilia
and transfusion-associated infections. Blood
17. Walker IR, Julian JA, Association of Haemophilia Clinic Directors of Canada. Causes of death in Canadians with haemophilia 1980-1995. Haemophilia
18. Eyster ME, Diamondstone LS, Lien JM, et al. Natural history of hepatitis C virus
infection in multitransfused hemophiliacs: effect of coinfection with human immunodeficiency virus. The Multicenter Hemophilia
Cohort Study. J Acquir Immune Defic Syndr
. 1993; 6: 602-610.
19. Darby SC, Ewart DW, Giangrande PLF, et al for the UK Haemophilia Centre Directors Organisation. Importance of age at infection with HIV-1 for survival and development of AIDS in UK hemophilia
20. GEMES (Grupo Español Multicéntrico para el Estudio de Seroconvertores). AIDS incubation period in Spain before highly active antiretroviral therapy
. Med Clin (Barc)
21. Altisent Roca C, Ayestaran Altuna A, Tusell Puigbert JM, et al. Progression of HIV infection in a cohort of hemophiliacs over an 11-year period. Med Clin (Barc)
22. Pérez-Hoyos S, Ferreros I, del Amo J, et al. Imputation of the date of HIV seroconversion in cohorts of haemophiliacs. Gac Sanit
. 2003;17: 474-482.
23. Ancelle-Park R. Expanded European AIDS case definition. Lancet
24. Clayton D, Hills M. Statistical Methods in Epidemiology
. Oxford: Oxford University Press, 1998.
25. CASCADE Collaboration. Changes over calendar time in the risk of specific first AIDS-defining events following HIV seroconversion, adjusting for competing risks
. Int J Epidemiol
26. Qurishi N, Kreuzberg C, Luchters G, et al. Effect of antiretroviral therapy on liver-related mortality in patients with HIV and hepatitis C virus
27. Klein MB, Lalonde RG, Suissa S. The impact of hepatitis C virus
coinfection on HIV progression
before and after highly active antiretroviral therapy
. J Acquir Immune Defic Syndr
. 2003;33: 365-372.
28. Salmon-Ceron D, Lewden C, Morlat P, et al. Liver disease as a major cause of death among HIV infected patients: role of hepatitis C and B viruses and alcohol. J Hepatol
29. Soriano V, Garcia-Samaniego J, Valencia E, et al. Impact of chronic liver disease due to hepatitis viruses as cause of hospital admission and death in HIV-infected drug users. Eur J Epidemiol
30. Klein MB, Lalonde RG, Suissa S. The impact of hepatitis C virus
coinfection on HIV progression
before and after highly active antiretroviral therapy
. J Acquir Immune Defic Syndr
31. Lee C, Dusheiko G. The natural history and antiviral treatment of hepatitis C in haemophilia. Haemophilia
32. Rumi MG, De Filippi F, Santagostino E, et al. Hepatitis C in haemophilia: lights and shadows. Haemophilia
. 2004;10(Suppl 4): 211-215.
33. Romero M, Perez-Olmeda M, Garcia-Samaniego J, et al. Management of chronic hepatitis C in patients co-infected with HIV: focus on safety considerations. Drug Saf
34. Porter K, Johnson AM, Phillips AN, et al. The practical significance of potential biases in estimates of the AIDS incubation periods distribution in the UK Register of HIV Seroconverteres. AIDS
Coordinating Center: Department of Public Health, Miguel Hernández University, Alicante: Ildefonso Hernández Aguado and Julia del Amo
Center of Data Analysis: Valencian School for Health Studies: Santiago Pérez-Hoyos, Alicia Moreno, and Inmaculada Ferreros
Participating Centers: Centers for AIDS Attention and Prevention of Barcelona: Joan Cayla, Patricia Garcia de Olalla, and Teresa Brugal; Valencian Centers for AIDS Information and Prevention: Ildefonso Hernández Aguado, Manuela Garcia de la Hera, Isabel Hurtado, Josefina Belda, Elisa Fernández, Ignacio Alastrúe, and C. Santos; Hospital Germans Trias I Pujol, Badalona: Roberto Muga, Arantza Sanvicens, Bonaventura Clotet, and Jordi Tor; Sandoval-Madrid: Jorge del Romero, Carmen Rodríguez, Soledad Garcia, Montserrat Raposo, Carmen de Mendoza, Vicente Soriano, and Grupo de Seroconvertores de la Comunidad de Madrid; Catalonia Prisions: Rafael Guerrero and Andrés Marco; Hospital la Paz: Manolo Quintana, Cristina González, and Julia del Amo; Hospital Vall d'Hebron: Isabel Ruiz and Joan Tussell; Hospital Virgen del Rocío de Sevilla: Rosario Pérez and José Miguel Cisneros; and National AIDS Register: Isabel Noguer and Paz Sobrino