The substantial declines in HIV-related mortality, as a consequence of the introduction of combination antiretroviral therapy (cART), have seen liver-related death (LRD) assume increasing relative importance among HIV-positive individuals [1,2]. Although progression of liver disease is common with hepatitis C virus (HCV) infection and known to be accelerated in the presence of HIV , LRD is often associated with older age as complications of HCV-related liver disease usually take decades to develop . During this period, HCV-co-infected individuals have many competing risks of death, such as mortality associated with injecting drug use, AIDS, cardiovascular disease, malignancies, bacterial infections, violent death and renal disease .
While the uptake of treatment for HCV has so far been low among HIV/HCV-co-infected individuals in Europe, in particular, among injecting drug users (IDUs), this is partially explained by the costs of treatment, potential contraindication to interferon-based treatments, anticipated poor treatment adherence and low treatment efficacy of pegylated interferon and ribavirin . However, with the recent approval of less toxic, oral, directly acting antivirals (DAAs) for HCV, fewer co-infected individuals will have contraindications to HCV treatment, while treatment outcomes will be significantly improved and approach the HCV cure rate of above 90% seen for HCV-monoinfected individuals [6,7].
The potential benefits of curative treatment with new DAAs are numerous, including reductions in fibrosis progression, LRD and extra-hepatic manifestations, along with limiting ongoing transmission . However, regardless of the advances in HCV treatment, the approximate costs of €50 000–90 000 per treatment for the currently approved DAAs and the expectation that new oral combination regimens will be more expensive still  will, in many countries, necessitate the prioritization of individuals for treatment. Therefore, a better understanding of the spectrum of causes of death among HIV/HCV-co-infected individuals, factors that affect progression to LRD and potential competing risks is essential in determining who to prioritize for treatment of HCV infection.
The aims of this study were to describe causes of death among HIV/HCV-co-infected individuals naive to interferon-based treatment, to identify factors associated with LRD and to provide guidance on who should be prioritized for treatment of HCV infection with DAAs.
EuroSIDA is a large prospective observational cohort of 18 786 HIV-positive individuals in 107 centres across Europe, Israel and Argentina. The study has been described in detail previously . At recruitment, in addition to demographic and clinical information, a complete antiretroviral treatment history is obtained together with the most recent CD4+ cell count and HIV-RNA measurements. At each follow-up, visit details on all CD4+ cell count and HIV-RNA values measured since last follow-up are collected, as are the dates of starting and stopping each antiretroviral drug used. Since 2010, qualitative data on alcohol abuse have also been collected at each follow-up visit. Alcohol abuse is defined as above 25 units per week for men and above 20 units for women (see forms at http://www.cphiv.dk).
Information on the collection of HCV-related parameters has been published elsewhere . In brief, data on HCV genotype, HCV antibody (HCVAb) and HCV viremia levels are collected at each follow-up visit, along with hepatitis B surface antigen (HBsAg) status. A number of measures are available to calculate liver fibrosis levels. Clinical centres report all liver biopsy and Fibroscan test results to the coordinating centre. Data are also available for a subset of individuals on the biomarker hyaluronic acid, while aspartate transaminase (AST) and platelet counts are routinely collected allowing calculation of the AST-to-platelet ratio (APRI) .
Causes of death were classified using CoDe methodology and EuroSIDA algorithms [13,14]. Deaths were classified into the following categories: LRD, AIDS-related death, non-LRD/non-AIDS-related death and unknown causes. Non-LRD/non-AIDS-related death includes causes such as bacterial infection, cardiovascular events, drug and accidental death and cancer, although, hepatocellular carcinoma (HCC) is included with LRD along with liver failure, cirrhosis or complications as a result of HCV or HBV infection.
Throughout this study, baseline is defined as 1 January 2000 or entry in the EuroSIDA study, if this was after 1 January 2000, or the first fibrosis measurement available after HCVAb-positivity if this occurred after entry into EuroSIDA. Follow-up prior to 2000 was excluded to avoid the inclusion of individuals receiving sub-optimal ART. Follow-up was censored at the date of starting interferon-based treatment.
Crude death rates (CDRs) were calculated per 1000 person-years of follow-up (PYFU). CDRs were stratified by the time-updated variables: HIV transmission route, age, region of EuroSIDA, HCV viremia status (positive, negative, unknown), HBsAg status (positive, negative, unknown), CD4+ cell count, HIV-RNA, reported alcohol abuse and stage of liver fibrosis (Metavir F0/F1, F2/F3 and F4 ).
Competing-risks Cox proportional-hazards models adopting the Fine and Gray methodology were used to describe factors associated with the sub-distribution hazards of LRD . The competing risks accounted for were death from causes other than LRD and loss to follow-up, defined as no contact for 1 year. Multiple imputation, with four sets of imputations, was used to impute missing data on HCV-RNA, HCV genotype, fibrosis staging and hepatitis B virus (HBV) status. Only those HCV-RNA positive were then included in modelling. The following baseline explanatory variables were included: age, sex, HIV transmission route, region of EuroSIDA, cardiovascular event (stroke, myocardial infarction, angina, endarterectomy, high blood pressure), diabetes diagnosis, HCV genotype, calendar year, CD4+ cell count, CD4+ cell count nadir, HIV-RNA, HBsAg status, minimum duration of HCV infection (calculated using the first available HCVAb test) and staging of liver fibrosis. Non-parametric cumulative incidence functions were calculated to estimate the 5-year probability of LRD according to liver fibrosis staging and CD4+ cell count.
A combined definition of liver fibrosis staging was used throughout according to the Metavir scoring system . F0/F1 fibrosis was defined from validated liver biopsy, Fibroscan measurements less than 7.6 Kpa, an APRI score below 1.5 and hyaluronic acid measurements below 100 ng/ml. F4 fibrosis was defined from validated liver biopsy, Fibroscan measurements greater than 12.5 Kpa, an APRI score above 2 and hyaluronic acid measurements above 250 ng/ml, in line with previous EuroSIDA work and other studies of HCV infection [17–19]. When more than one fibrosis measurement is available, they are prioritized in the order given above. The majority of fibrosis data came from the APRI score (77.6%), followed by hyaluronic acid (20.0%), Fibroscan (1.4%) and liver biopsy (0.9%).
All statistical analyses were performed using SAS (version 9.4; SAS Institute, Cary, North Carolina, USA). Fine and Gray models were fitted using the %PSHREG macro and PHREG procedure, whereas cumulative incidence functions were calculated using the %CIF (cumulative incidence function) macro [20,21].
There were 670 deaths recorded in the study population of 3941 HIV/HCV-co-infected individuals, contributing a total of 16 091 PYFU [median 3.5 years, inter-quartile range (IQR) 1.3–6.4 per individual] to January 2013, giving an overall incidence of all-cause mortality of 41.6 [95% confidence interval (CI) 38.6–44.7] per 1000 PYFU. There were 145 (21.6%) deaths classified as liver-related, equating to an overall incidence of LRD of 9.0 (95% CI 7.6–10.5) per 1000 PYFU.
Baseline characteristics, overall and stratified by cause of death, are shown in Table 1. The study population was mostly white (93.6%) men (67.9%), with a median age of 37 years. The majority of the study population resided in either eastern (31.0%) or southern (25.5%) Europe, although all European regions were well represented. The most common HIV transmission routes were injecting drug use (70.0%), followed by heterosexual exposure (15.3%). Eastern Europe accounted for 49.4% of AIDS-related deaths, but only 15.9% of LRDs. Duration anti-HCV positive prior to baseline was highest in those that progressed to LRD [4.1 years (IQR 2.2–7.0)], and varied significantly by region of EuroSIDA (P < 0.0001).
Crude death rates
The most common causes of death were AIDS (24.2%) and LRD (21.6%), followed by unknown causes (19.4%), drug/violent death (9.6%), bacterial infection (7.9%), cardiovascular disease (6.9%) and cancer (4.5%). Interestingly, although LRD was one of the leading causes of death, in univariable analysis, the rate of LRD has declined 14.8% (95% CI 11.0–18.5; P < 0.0001) per year over the study period (see Supplementary figure, http://links.lww.com/QAD/A684). CDRs for LRD, AIDS-related, non-LRD/non-AIDS and unknown causes of death stratified by demographics and HCV-related factors are shown in Table 2.
Whereas all-cause mortality and non-LRD rates peaked in those aged 55 and over, LRD rates were highest in the years 35–45 and 45–55 before tapering off after 55. LRD rates were two-fold higher among those positive for HCV-RNA compared to those who were negative (CDR 10.1, 95% CI 8.1–12.2; and CDR 5.6, 95% CI 2.5–8.6, respectively), and 2.5-fold higher in those positive for HBsAg compared with those who were HBsAg-negative (CDR 21.3, 95% CI 12.9–29.7; and CDR 8.3, 95% CI 6.8–9.8, respectively). LRD rates were 13-fold higher among current alcohol abusers compared with those not reporting alcohol abuse (CDR 36.8, 95% CI 18.5–55.0; and CDR 2.9, 95% CI 1.0–4.8, respectively).
Most importantly, LRD rates were 35-fold higher among those with F4 fibrosis compared with those having F0/F1 fibrosis (CDR 42.4, 95% CI 31.0–53.7; and CDR 1.2, 95% CI 0.5–1.9, respectively), and 8-fold higher than those with F2/F3 fibrosis (CDR 10.0, 95% CI 2.6–17.4). Further, whereas rates of AIDS-related death and non-LRD/non-AIDS-related death were similar for those with F0/F1 (CDR 5.6, 95% CI 4.1–7.1; and CDR 7.9, 95% CI 6.1–9.7) and F2/F3 fibrosis (CDR 5.7, 95% CI 0.1–11.3; and CDR 8.6, 95% CI 1.7–15.4, respectively), AIDS-related death rates were 3-fold higher (CDR 14.1, 95% CI 7.5–20.8) and non-LRD/non-AIDS-related death rates 3.75-fold higher (CDR 30.7, 95% CI 21.0–40.5) among those with F4 fibrosis.
Factors associated with sub-distribution hazards of liver-related death
Multivariable sub-distribution hazard ratio (sHR) estimates of factors associated with the cumulative incidence of LRD from Cox proportional-hazards modelling, using the Fine and Gray methodology for competing risks, are shown in Fig. 1. The strongest association with LRD was seen for F4 and F2/F3 fibrosis, which was associated with 6-fold (sHR 6.25, 95% CI 4.08–9.58, P < 0.0001) and 2.5-fold (sHR 2.52, 95% CI 1.53–4.15, P < 0.0001) increased sub-distribution hazards compared with F0/F1, respectively. Other factors associated with LRD were HBsAg-positivity (sHR 2.15, 95% CI 1.31–3.51, P = 0.0024), minimum duration of HCVAb-positivity above 10 years (sHR 1.95, 95% CI 1.03–3.71, P = 0.041) compared with that below 2 years, CD4+ cell count (sHR 0.83, 95% CI 0.73–0.95, P = 0.0052) per doubling, and age between 35 and 45 years (sHR 1.61, 95% CI 1.01–2.57, P = 0.045) compared with less than 35 years. Whereas baseline CD4+ cell count was included in the model, CD4+ nadir was not associated with LRD. However, when omitting baseline CD4+ cell count, CD4+ nadir became a significant predictor of LRD (sHR 0.90, 95% CI 0.83–0.97, P = 0.0045).
Interaction terms between liver fibrosis staging, CD4+ cell count and age were tested and found to be non-significant (P > 0.3 and P > 0.7, respectively), meaning there is no evidence to suggest that the effect of liver fibrosis stage on LRD is modified by CD4+ cell count or age.
Cumulative incidence of liver-related death
The association between liver fibrosis staging, CD4+ cell count and LRD can be further illustrated by the cumulative incidence functions of LRD stratified by these groups. Cumulative incidence functions for time to LRD, stratified by liver fibrosis staging, are shown in Fig. 2. The 5-year probability of LRD was low in those with F0/F1 fibrosis (2.2%, 95% CI 1.7–2.9), but substantial in those with F2/F3 (10.3%, 7.6–13.5) and F4 fibrosis (14.0%, 10.3–18.3, P < 0.0001) for Gray's test of separation between the three cumulative incidence functions.
Cumulative incidence functions for time to LRD, stratified by CD4+ cell count and liver fibrosis staging, are shown in Fig. 3; here fibrosis stages F2–F4 are considered together due to low numbers in some categories. The 5-year probability of LRD was low in those with F0/F1 fibrosis and CD4+ cell count at least 300 cells/μl (1.7%, 95% CI 1.1–2.5), although higher in those with less than 300 cells/μl (3.3%, 95% CI 2.2–4.8). The 5-year probability of LRD was substantial in those with at least F2 fibrosis and CD4+ cell count at least 300 cells/μl (8.6%, 95% CI 5.9–11.9) and higher still in those with less than 300 cells/μl (15.3%, 95% CI 11.7–19.3, P<0.0001) for Gray's test of separation between the four cumulative incidence functions.
Liver-related death, along with AIDS, was the most common cause of death in this population, highlighting the importance of HCV disease management in order to reduce the burden of LRD among HIV/HCV-co-infected individuals. LRD occurred almost exclusively among those with at least F2 fibrosis at baseline, with F4 and F2/F3 fibrosis associated with 35 and 8-fold increased risk, compared with F0/F1 fibrosis, respectively. Further, the 5-year probability of LRD was low among those with F0/F1 fibrosis, whereas it was substantial at 10.3% in those with F2/F3 and as high as 14.0% in those with F4 fibrosis. However, non-LRD death rates were substantially higher among those with F4 fibrosis, suggesting the risk of death from causes not related to HCV should be considered during HCV treatment prioritization.
These findings are in agreement with a recent study which also found that low levels of fibrosis (Metavir F0/F1), as measured by the FIB-4 index, were associated with a minimal 5-year risk of end-stage liver disease (ESLD), but that this risk increased to 17% for those with F3/F4 fibrosis . Therefore, liver fibrosis staging should be considered the most important risk factor for LRD, and treatment with DAAs should be prioritized for those with at least F2 fibrosis. Current European guidelines recommend deferral of HCV treatment for those with F0/F1 fibrosis, whereas treatment is encouraged for those with significant liver fibrosis – a stance that is strongly supported by our findings . These recommendations are further supported by studies showing a reduction in LRD rates and regression of liver fibrosis staging following SVR to HCV therapy , including among those with cirrhosis .
In multivariable analysis, in addition to significant fibrosis, concurrent HBV co-infection, lower CD4+ cell count and minimum duration of HCVAb-positivity were all associated with the cumulative incidence of LRD. Further, when omitting baseline CD4+ from the model, CD4+ nadir was strongly associated with LRD. These findings suggest that efforts to reduce late presentation of co-infected individuals are essential so that current treatment guidelines can be employed, such that cART should be initiated early during the course of HIV progression, and in the case of concurrent HBV co-infection should contain potent HBV-active treatment, so that time with low CD4+ cell count and uncontrolled HBV co-infection can be minimalized .
Further, although the interaction between CD4+ cell count and liver fibrosis staging did not reach significance, we saw some evidence from the stratified cumulative incidence of LRD to suggest that among those with at least F2 fibrosis, the risk of LRD was higher when baseline CD4+ cell counts were below 300 cells/μl. Low CD4+ cell count, high HIV viral load and circulating HBV DNA have all been associated with rapid progression of liver fibrosis [26–28], highlighting the public health problem associated with late presentation of HIV/HCV co-infection. However, each of these risk factors could be controlled by initiating appropriate cART early in the co-infected individuals, potentially reducing the need for expensive treatment with DAAs.
Individuals reporting current alcohol abuse were found to have a 13-fold higher risk of LRD from analysis of CDRs. However, data on alcohol abuse were only recently added to EuroSIDA, and 40% of the study population have no data available. Further, estimation of Fine and Gray competing-risk models does not allow time-updating variables, which are linked to the outcome of interest. Consequently, the effect of alcohol abuse on LRD could not be assessed in this model. Alcohol consumption is a known contributor to the rapid progression of liver fibrosis and acts synergistically with HCV infection [29,30]. HCV-infected individuals should be warned of the potential for accelerated liver damage caused by drinking excessively and aim to reduce alcohol intake to prevent the rapid development of fibrosis. However, the reality is that many co-infected individuals will belong to difficult-to-treat populations engaging in active injecting drug use and alcohol abuse, requiring careful clinical management.
Although 31.0% of individuals in this study resided in Eastern Europe, where co-infection was most prevalent , they accounted for just 15.9% of the LRDs recorded in this study. In comparison, 49.4% of all AIDS-related deaths occurred in this region, illustrating the competing risks that these individuals face . It also indicates that co-infected individuals in Eastern Europe may initially see more benefit from a comprehensive cART program than the introduction of DAAs.
Age was found to be associated with LRD with borderline statistical significance in this study, suggesting that those aged between 35 and 45 years are at the highest risk of LRD. However, the estimated effect sizes were similar for those aged 45–55 and above 55 years, which may suggest that the risk of LRD is increased but stable after the age of 35. Similarly, minimum duration HCVAb-positivity was associated with LRD, with those diagnosed for at least 10 years at increased risk. As expected, age was strongly associated with non-LRD from causes such as cancer and bacterial infections, consequently lowering the risk of LRD in later life as death from other causes becomes more prominent.
Unlike other viral infections, clearance of HCV with or without treatment infers only partial protection against re-infection . Many studies have documented a high incidence of re-infection with HCV after viral clearance among IDUs , including a recent EuroSIDA study in which 18% of all co-infected individuals who achieved spontaneous clearance became re-infected within a few years, 94% of those re-infected having acquired HIV via injecting drug use . However, re-infection is not restricted to IDUs with studies documenting re-infection rates as high as 15% among MSM . Consequently, in the setting of treatment prioritization, risk behaviour associated with re-infection must be considered before initiating HCV treatment regimens.
Data are beginning to emerge which suggests that HCV eradication may lead to a reduction in all-cause mortality . A recent study of a large number of HCV-seropositive and seronegative individuals found that those who were HCV-seropositive were at an increased risk of death from hepatic and extra-hepatic diseases, potentially due to complications associated with fibrosis and cirrhosis, or an inflammatory effect of circulating HCV-RNA . Our study, although not designed to determine the effect of HCV-RNA eradication on mortality, also found consistently higher CDRs among those who were HCV-RNA-positive compared with those who were HCV-RNA-negative.
This evidence, taken together with observed higher SVR rates among those with low levels of fibrosis and potential for reductions in extra-hepatic manifestations and transmission risk [6,8], may mean that eventually it will be beneficial to treat all co-infected individuals regardless of fibrosis staging. However, while the pathways by which active HCV infection may be linked with non-LRD remain unclear and treatment remains prohibitively expensive, prioritization of those at greatest risk of LRD is still necessary. Further, the pathogenesis of liver disease varies among individuals, and those who are not initially prioritized for treatment should still be monitored closely for progression of liver fibrosis.
The main limitation of this study is that due to limited power, we were unable to compare the individual stages of fibrosis and had grouped F0/F1 and F2/F3. Moreover, as the fibrosis data were drawn from a combination of clinical procedures and biomarkers, consensus on the definitions of the stages of fibrosis was difficult to attain. However, cut-off points for low levels of fibrosis and cirrhosis at the other end of the scale were relatively well defined . In particular, the APRI score, where the majority of the fibrosis data are taken from in this study, has been validated in a number of studies [12,40]. Furthermore, misclassification of liver fibrosis levels in this study would make differentiation between the fibrosis staging levels less precise. This would lead to an under-estimation of the differences between liver fibrosis levels with respect to progression to LRD.
We found a significant relationship between lower CD4+ cell count and progression to LRD; however, due to few LRDs among those with higher CD4+ cell counts, we were unable to determine the optimum time to start cART in order to prevent progression to LRD. A further limitation of this study is that data were not complete at baseline, with multiple imputations required to impute missing data, while alcohol abuse data were only available from 2010 onwards, and the qualitative nature of the data means we do not know how much each individual is drinking, just that they are considered to be an alcohol abuser. Also, minimum duration of HCV infection was found to be associated with LRD, calculated using the first positive HCVAb test result. However, this is likely to underestimate the true duration of HCV infection in this population as seroconversion likely occurred around the same time as HIV infection.
In conclusion, new DAAs for treatment of HCV infection may offer impressive outcomes and less toxicity; however, the costs of treatment will necessitate the prioritization of those at the highest risk of LRD for therapy. In this regard, after considering the risk of re-infection, those with significant liver fibrosis (≥F2) should be prioritized for treatment with DAAs. Moreover, as part of a comprehensive strategy to prevent LRD in HIV/HCV co-infected individuals, it is essential to identify co-infected individuals as soon as possible so that they may start cART early in the course of HIV infection, which in the case of concurrent HBV infection includes potent HBV active treatment, to prevent rapid progression of liver fibrosis and reduce the need for expensive treatments.
Statement of funding: Primary support for EuroSIDA is provided by the European Commission BIOMED 1 (CT94–1637), BIOMED 2 (CT97-2713), the 5th Framework (QLK2-2000-00773), the 6th Framework (LSHP-CT-2006-018632) and the 7th Framework (FP7/2007-2013, EuroCoord n° 260694) programmes. Current support also includes unrestricted grants by Janssen R&D, Merck and Co. Inc., Pfizer Inc., GlaxoSmithKline LLC. The participation of centres from Switzerland was supported by The Swiss National Science Foundation (Grant 108787).
Author contributions: D.G. led the study with supporting contributions from L.P., J.D.L. and A.M. D.G. performed the statistical analyses. D.G., J.D.L., L.P., O.K. and A.M. designed the study and D.G. drafted the manuscript. All co-authors contributed to redrafting and refinement of the manuscript. A.M. supervised the study.
EuroSIDA study group: The multicentre study group, EuroSIDA (national coordinators in parenthesis). Argentina: (M. Losso), M. Kundro, Hospital JM Ramos Mejia, Buenos Aires. Austria: (N. Vetter), Pulmologisches Zentrum der Stadt Wien, Vienna; R. Zangerle, Medical University Innsbruck, Innsbruck. Belarus: (I. Karpov), A. Vassilenko, Belarus State Medical University, Minsk, V.M. Mitsura, Gomel State Medical University, Gomel; D. Paduto, Regional AIDS Centre, Svetlogorsk. Belgium: (N. Clumeck), S. De Wit, M. Delforge, Saint-Pierre Hospital, Brussels; E. Florence, Institute of Tropical Medicine, Antwerp; L. Vandekerckhove, University Ziekenhuis Gent, Gent. Bosnia-Herzegovina: (V. Hadziosmanovic), Klinicki Centar Univerziteta Sarajevo, Sarajevo. Bulgaria: (K. Kostov), Infectious Diseases Hospital, Sofia. Croatia: (J. Begovac), University Hospital of Infectious Diseases, Zagreb. Czech Republic: (L. Machala), D. Jilich, Faculty Hospital Bulovka, Prague; D. Sedlacek, Charles University Hospital, Plzen. Denmark: (J. Nielsen), G. Kronborg, T. Benfield, M. Larsen, Hvidovre Hospital, Copenhagen; J. Gerstoft, T. Katzenstein, A.-B.E. Hansen, P. Skinhøj, Rigshospitalet, Copenhagen; C. Pedersen, N.F. Møller, Odense University Hospital, Odense; L. Ostergaard, Skejby Hospital, Aarhus, U.B. Dragsted, Roskilde Hospital, Roskilde; L.N. Nielsen, Hillerod Hospital, Hillerod. Estonia: (K. Zilmer), West-Tallinn Central Hospital, Tallinn; Jelena Smidt, Nakkusosakond Siseklinik, Kohtla-Järve. Finland: (M. Ristola), Helsinki University Central Hospital, Helsinki. France: (C. Katlama), Hôpital de la Pitié-Salpétière, Paris; J.-P. Viard, Hôtel-Dieu, Paris; P.-M. Girard, Hospital Saint-Antoine, Paris; P. Vanhems, University Claude Bernard, Lyon; C. Pradier, Hôpital de l’Archet, Nice; F. Dabis, D. Neau, Unité INSERM, Bordeaux, C. Duvivier, Hôpital Necker-Enfants Malades, Paris. Germany: (J. Rockstroh), Universitäts Klinik Bonn; R. Schmidt, Medizinische Hochschule Hannover; J. van Lunzen, O. Degen, University Medical Center Hamburg-Eppendorf, Infectious Diseases Unit, Hamburg; H.J. Stellbrink, IPM Study Center, Hamburg; C. Stefan, J.W. Goethe University Hospital, Frankfurt; J. Bogner, Medizinische Poliklinik, Munich; G. Fätkenheuer, Universität Köln, Cologne. Georgia: (N. Chkhartishvili) Infectious Diseases, AIDS & Clinical Immunology Research Center, Tbilisi. Greece: (J. Kosmidis), P. Gargalianos, G. Xylomenos, J. Perdios, Athens General Hospital; H. Sambatakou, Ippokration General Hospital, Athens. Hungary: (D Banhegyi), Szent Lásló Hospital, Budapest. Iceland: (M. Gottfredsson), Landspitali University Hospital, Reykjavik. Ireland: (F. Mulcahy), St. James’ Hospital, Dublin. Israel: (I. Yust), D. Turner, M. Burke, Ichilov Hospital, Tel Aviv; E. Shahar, G. Hassoun, Rambam Medical Center, Haifa; H. Elinav, M. Haouzi, Hadassah University Hospital, Jerusalem; Z.M. Sthoeger, AIDS Center (Neve Or), Jerusalem. Italy: (A. D’Arminio Monforte), Istituto Di Clinica Malattie Infettive e Tropicale, Milan; R. Esposito, I. Mazeu, C. Mussini, Università Modena, Modena; R. Pristera, Ospedale Generale Regionale, Bolzano; F. Mazzotta, A. Gabbuti, Ospedale S Maria Annunziata, Firenze; V. Vullo, M Lichtner, University di Roma la Sapienza, Rome; M. Zaccarelli, A. Antinori, R. Acinapura, G. D’Offizi, Istituto Nazionale Malattie Infettive Lazzaro Spallanzani, Rome; A. Lazzarin, A. Castagna, N. Gianotti, Ospedale San Raffaele, Milan; M. Galli, A. Ridolfo, Osp. L. Sacco, Milan. Latvia: (B. Rozentale), Infectology Centre of Latvia, Riga. Lithuania: V Uzdaviniene, Lithuanian AIDS Centre, Vilnius. Luxembourg: (T. Staub), R. Hemmer, Centre Hospitalier, Luxembourg. Netherlands: (P. Reiss), Academisch Medisch Centrum bij de Universiteit van Amsterdam, Amsterdam. Norway: (V. Ormaasen), A. Maeland, J. Bruun, Ullevål Hospital, Oslo. Poland: (B. Knysz), J. Gasiorowski, M. Inglot, Medical University, Wroclaw; A. Horban, E. Bakowska, Centrum Diagnostyki i Terapii AIDS, Warsaw; A. Grzeszczuk, R. Flisiak, Medical University, Bialystok; M. Parczewski, M. Pynka, K. Maciejewska, Medical Univesity, Szczecin; M. Beniowski, E. Mularska, Osrodek Diagnostyki i Terapii AIDS, Chorzow; T. Smiatacz, Medical University, Gdansk; E. Jablonowska, E. Malolepsza, K. Wojcik, Wojewodzki Szpital Specjalistyczny, Lodz; I. Mozer-Lisewska, Poznan University of Medical Sciences, Poznan. Portugal: (M. Doroana), L. Caldeira, Hospital Santa Maria, Lisbon; K. Mansinho, Hospital de Egas Moniz, Lisbon; F. Maltez, Hospital Curry Cabral, Lisbon. Romania: (R. Radoi), C. Oprea, Spitalul de Boli Infectioase si Tropicale: Dr Victor Babes, Bucarest. Russia: (A. Rakhmanova), Medical Academy Botkin Hospital, St Petersburg; A. Rakhmanova, St Petersburg AIDS Centre, St Peterburg; T. Trofimora, Novgorod Centre for AIDS, Novgorod, I. Khromova, Centre for HIV/AIDS & and Infectious Diseases, Kaliningrad; E. Kuzovatova, Nizhny Novgorod Scientific and Research Institute, Nizhny Novogrod. Serbia: (D. Jevtovic), The Institute for Infectious and Tropical Diseases, Belgrade. Slovakia: A. Shunnar, D Staneková, Dérer Hospital, Bratislava. Slovenia: (J. Tomazic), University Clinical Centre Ljubljana, Ljubljana. Spain: S. Moreno, J. M. Rodriguez, Hospital Ramon y Cajal, Madrid; B. Clotet, A. Jou, R. Paredes, C. Tural, J. Puig, I. Bravo, Hospital Germans Trias i Pujol, Badalona; J.M. Gatell, J.M. Miró, Hospital Clinic Universitari de Barcelona, Barcelona; P. Domingo, M. Gutierrez, G. Mateo, M.A. Sambeat, Hospital Sant Pau, Barcelona; J.M. Laporte, Hospital Universitario de Alava, Vitoria-Gasteiz. Sweden: (A. Blaxhult), Venhaelsan-Sodersjukhuset, Stockholm; L. Flamholc, Malmö University Hospital, Malmö, A. Thalme, A. Sonnerborg, Karolinska University Hospital, Stockholm. Switzerland: (B. Ledergerber), R. Weber, University Hospital, Zürich; M. Cavassini, Centre Hospitalier Universitaire Vaudois, Lausanne; A. Calmy, Hospital Cantonal Universitaire de Geneve, Geneve; H. Furrer, Inselspital Bern, Bern; M. Battegay, L. Elzi, University Hospital Basel; P. Schmid, Kantonsspital, St. Gallen. Ukraine: (E. Kravchenko), N. Chentsova, Kiev Centre for AIDS, Kiev; V. Frolov, G. Kutsyna, I. Baskakov, Luhansk State Medical University, Luhansk; S. Servitskiy, Odessa Region AIDS Center, Odessa; A. Kuznetsova, Kharkov State Medical University, Kharkov; G. Kyselyova, Crimean Republican AIDS Centre, Simferopol. United Kingdom: (B. Gazzard), St. Stephen's Clinic, Chelsea and Westminster Hospital, London; A.M. Johnson, E. Simons, S. Edwards, Mortimer Market Centre, London; A. Phillips, M.A. Johnson, A. Mocroft, Royal Free and University College Medical School, London (Royal Free Campus); C. Orkin, Royal London Hospital, London; J. Weber, G. Scullard, Imperial College School of Medicine at St. Mary's, London; M. Fisher, Royal Sussex County Hospital, Brighton; C. Leen, Western General Hospital, Edinburgh.
The following centres have previously contributed data to EuroSIDA: Bernhard Nocht Institut für Tropenmedizin, Hamburg, Germany; 1st I.K.A Hospital of Athens, Athens, Greece; Ospedale Riuniti, Divisione Malattie Infettive, Bergamo, Italy; Ospedale Cotugno, III Divisione Malattie Infettive, Napoli, Italy; Hospital Carlos III, Departamento de Enfermedades Infecciosas, Madrid, Spain.
EuroSIDA Steering Committee: J. Gatell, B. Gazzard, A. Horban, I. Karpov, B. Ledergerber, M. Losso, A. d’Arminio Monforte, C. Pedersen, A. Rakhmanova, M. Ristola, A. Phillips, P. Reiss, J. Lundgren, J. Rockstroh, S. De Wit
Coordinating Centre Staff: D. Podlekareva, L. Peters, J.E. Nielsen, C. Matthews, A.H. Fischer, A. Bojesen, D. Raben, D. Kristensen, K. Grønborg Laut, J.F. Larsen. Statistical Centre Staff: A. Mocroft, A. Phillips, A. Cozzi-Lepri, D. Grint, L. Shepherd, A. Schultze.
Conflicts of interest
O.K. has received honorarium, consultancy and/or lecture fees from Abbott, Gilead, GSK, Janssen, Merck, Tibotec and Viiv. A.M. has received honorarium, consultancy or guest speaker fees from Pfizer, Merck, Gilead, BI and BMS. J.R. has received consultancy or lecture fees from Bionor, BMS, BI, GSK, ViiV, Abbott, Gilead, Pfizer, Merck, Tibotec and Janssen. All other authors have no conflicts of interest to report for the present study.