The advent of HAART in 1996 changed HIV disease in developed countries from a condition with a poor short-term prognosis to one with prolonged survival needing long-term management. Major improvements in survival were observed in the first years following its introduction, with over 87% of individuals expected to be alive 10 years following infection in 1999–2001 . However, an unacceptably high proportion of infected persons still present at advanced stages of HIV. Indeed, one-third of the estimated 63 500 individuals aged between 15 and 59 years living with HIV in the UK in 2005 were unaware of their infection  and are therefore at risk of missing the benefits of initiating HAART before becoming immunologically compromised [3,4] or progressing to symptomatic disease.
Advances in patient management over the last decade, as well as the availability of more potent drugs, have led to continuing improvement in virological response to therapy [5,6]. However, whether this will translate into further survival benefits, given the emergence of drug resistance and the increasing risk of cardiovascular disease, is less clear.
A small proportion of those presenting for care do so close to the time when they first became infected with HIV. If a recent negative HIV antibody test result is available, this allows a reliable estimate to be made of their date of HIV seroconversion. Although such individuals are likely to differ from the majority of the clinical population, given their perception of risk which led them to seek repeated HIV tests, their known duration of infection provides a unique opportunity to estimate survival expectations following HIV infection and therefore evaluate the population effectiveness of therapy. These data are important not only to patients and their clinicians, but also to healthcare planners attempting to predict the HIV disease burden on healthcare systems. Since seroconverters are by definition diagnosed early in their HIV infection, this allows maximum opportunity for intervention. Derived survival estimates from such a cohort in a country with universal access to effective therapy, which is free at the point of care, can therefore be interpreted as the optimum which may be expected with current management practice and therapy.
We provide current survival estimates and temporal changes in the risk of death, along with corresponding changes in therapy use, from a cohort of patients followed up from HIV seroconversion in the UK.
We used data from the UK Register of HIV Seroconverters, a longitudinal cohort established in October 1994 with the initial aim of studying the natural history of HIV infection and factors related to the length of the incubation period. Currently its aims are to assess response to therapy and associated factors, evaluate the population effectiveness of HAART, and estimate the prevalence of transmitted drug resistance and its consequences. The study has been described elsewhere . Briefly, eligible patients are aged 16 years or older with a positive HIV test result within 3 years of a negative HIV antibody test result, or with laboratory evidence of seroconversion. Since November 2004 the eligibility criteria only allowed for persons to be enrolled if the HIV test window was 12 months or less. Data are collected at enrolment and annually thereafter, mainly through Clinical Report Forms at clinical centres, and include demographic information, laboratory measurements (including CD4 cell count and HIV viral load), details of all antiretroviral therapy, clinical events and vital status. The records of persons who became lost to follow-up at their clinical centre were cross-checked with national death registries, through the Office of National Statistics (England and Wales) and the General Registrar Office (Scotland). These records were available electronically for deaths that occurred only from 1 January 1993 and were restricted to those aged 55 years or less at the time of death. For the current analyses, cross-checks were made with deaths that occurred by 31 December 2005.
We analysed the time from estimated seroconversion (midpoint between negative and positive HIV test dates, or date of laboratory evidence of seroconversion where available) to death or the date last known alive using time to event methods. Since the data had been cross-checked with national mortality data, individuals known to be alive after January 1993 and not known to have died could be assumed to be alive to at least the earlier of 31 December 2005 or their 55th birthday, and this was incorporated into the censoring date.
Cox proportional hazards models allowing for late entry into the cohort  were used to assess changes in the risk of death over time by fitting calendar period as a time-updated covariate, categorized to capture the era before widespread availability of HAART (pre-1996) and regular intervals subsequently (1996–1997, 1998–1999, 2000–2001, 2002–2003, 2004–2006). We adjusted for sex, age at seroconversion, HIV exposure category, and short HIV test interval (≤ 30 days between last negative and first positive HIV tests, or laboratory evidence of seroconversion) as a proxy for seroconversion illness . We investigated the effects of these potential prognostic factors on survival and considered whether these effects had changed over calendar time by including interactions with calendar period at risk. P-values were obtained using likelihood ratio tests. Kaplan–Meier methods were used to determine the expected survival in each calendar period. Time was always measured from seroconversion, but in each calendar period persons were only considered at risk from the latest of: the date of seroconversion; the date of enrolment into the cohort (late entry); or the first day of the calendar period. Follow-up was censored on the last day of the period of interest. Similar Kaplan–Meier methods were used to assess temporal trends in the time from seroconversion to initiation of HAART (defined as three or more drugs, either containing two or more classes, or containing tenofovir or abacavir).
By August 2006, 2275 seroconverters were enrolled from 126 UK centres and included in the analysis. Seven large treatment centres in south-east England, including London, and Scotland accounted for 57% of patients. Twelve patients with unknown HIV exposure category were excluded.
The overall median age at seroconversion was 30 years (interquartile range, 25–36 years), and appeared to increase over calendar time (Table 1). The median year of seroconversion was 1994 (range, 1982–2006). The majority of patients (1836/2275, 81%) were infected through sex between men, with those infected through sex between men and women increasing as a proportion of enrolled seroconverters in later calendar periods (8% pre-1996 compared with 19% in 2004–2006).
Changes over calendar time in mortality and survival estimates
During 18 695 person years of follow-up, 444 patients (20%) died. After adjusting for sex, age at seroconversion, HIV exposure category and short HIV test interval, the relative risk of death, compared with pre-1996, was 0.63 [95% confidence interval (CI), 0.48–0.81] in 1996–1997, and continued to decrease in each calendar period, to 0.03 (95% CI, 0.02–0.06) in 2004–2006 (Fig. 1).
Older age at seroconversion was associated with an increased risk of death [hazard ratio (HR), 1.49 per 10-year increase; 95% CI, 1.34–1.66; P < 0.001]. HIV exposure category was an important prognostic factor (P = 0.001): compared with men infected through sex between men, injecting drug users (IDU) were at greater risk (HR, 1.53; 95% CI, 1.17–2.00), as were those with needlestick injury/occupational exposure (HR, 3.54; 95% CI, 1.28–9.82), though there were only eight patients in this category (< 1%). Those infected through sex between men and women had similar risk to men infected through sex between men (HR, 0.98; 95% CI, 0.62–1.53). There was no evidence of an effect of sex (P = 0.16) or short HIV test interval (P = 0.65) on the risk of death.
To investigate whether the prognostic effects of these covariates had changed over time, we included interactions between calendar time and each of the covariates in the Cox model. As such interactions can cause model instability when there are many categories, we collapsed the calendar periods to pre-1996, 1996–1999 and 2000–2006, and excluded the eight patients who reported needlestick injury/occupational as the HIV exposure category. There was no evidence of a change over time in the effect of sex or age (P = 0.46 and 0.34, respectively; Table 2). However, there was strong evidence of a change in the effect of HIV exposure category on survival over time (P < 0.0001): while IDU were at similar risk to those infected through sex between men prior to 1996 (HR, 1.13; 95% CI, 0.80–1.61), their risk of death did not fall as much as that of other groups in the HAART era, leaving them at increasingly elevated risk in later periods (1996–1999: HR, 1.55; 95% CI, 0.90–2.69; 2000–2006: HR, 7.45; 95% CI, 4.26–13.05). However, those infected through sex between men and women appeared to be at similar risk to those infected through sex between men in all periods (pre-1996: HR, 0.78; 95% CI, 0.38–1.59; 1996–1999: HR, 1.02; 95% CI, 0.47–2.18; 2000–2006: HR, 1.25; 95% CI, 0.45–3.45).
Prior to 1996, the proportion of individuals expected to survive 5, 10 and 15 years following seroconversion was 87%, 50% and < 28%, respectively. The respective proportions increased to 91%, 75% and 60% in 1996–1999, and to 99%, 94% and 89% in 2000–2006. The improved survival estimates were seen across all age groups, but were most pronounced among those aged 45 years and older, for whom the expected proportion surviving 10 years increased from 12% pre-1996 (95% CI, 1–39) to 83% (95% CI, 65–92) in 2000–2006 (Fig. 2). IDU had lower estimated 10-year survival compared to other groups in 2000–2006 (74%, compared with 95% for men infected through sex with men and 98% for those infected through sex between men and women).
As expected, the proportion of AIDS-related deaths fell over time: of the 219 deaths pre-1996, 67 (31%) were attributed to AIDS, though cause of death was unknown in 78 (36%) patients. By 2000–2006, 13 (of 86, 15%) deaths were AIDS related, and cause of death was unknown in 11 (13%).
Changes in HAART use
The uptake of HAART in the cohort increased over time, from < 2% of person years pre-1996 to 8%, 33%, 46%, 52% and 58% in 1996–1997, 1998–1999, 2000–2001, 2002–2003 and 2004–2006, respectively. Considering the 940 patients seroconverting in 1996 or later and with follow-up therapy data, the median time from seroconversion to HAART initiation was 3.9 years (interquartile range, 1.3–9.2 years). To assess temporal trends in the time to HAART initiation, we considered the 25th percentile (i.e., the time by which 25% of individuals had started HAART), since the median had not been reached in all calendar periods. This changed from 2.1 years in 1996–1997 to 1.5, 0.8, 1.2 and 1.3 years in 1998–1999, 2000–2001, 2002–2003 and 2004–2006, respectively (P-trend = 0.0001). The strong statistical significance can mainly be explained by the difference between 1996–1997, when HAART was less widely available, and subsequent years (excluding 1996–1997 year group, P-trend = 0.32). In addition, the results could be driven by the proportion of individuals starting treatment in primary infection. Excluding persons treated within 6 months of seroconversion, the median time from seroconversion to HAART initiation was 5.0 years (range, 2.1 to > 10 years), and the 25th percentile of time to starting HAART was 2.5 years in 1996–1997, falling to 2.0, 2.0, 2.0 and 1.4 years in 1998–1999, 2000–2001, 2002–2003 and 2004–2006, respectively (P-trend = 0.001 overall and 0.23 excluding 1996–1997).
We examined changes in the class of HAART regimen prescribed over time. Among patients initiating HAART in 1996–1997, 62% were prescribed a regimen based on at least one unboosted protease inhibitor (PI). From 1998 onwards, the majority of first HAART regimens were non-nucleoside reverse transcriptase inhibitor (NNRTI)-based (56% of patients initiating HAART in 1998–1999, and 76%, 73% and 74% in 2000–2001, 2002–2003 and 2004–2006, respectively), although the use of regimens based on boosted PI increased in later periods (7%, 15% and 20% in 2000–2001, 2002–2003 and 2004–2006, respectively). Of the 133 patients initiating HAART in 2004–2006, 37 (28%) started with lamivudine, zidovudine and efavirenz, and 20 (15%) with emtricitabine, tenofovir and efavirenz, though in total 29 different combinations were used.
We found that survival following HIV seroconversion has continued to improve over calendar time with a 97% reduction in the risk of death in 2004–2006 compared with pre-1996 in our cohort. This is extremely encouraging and it remains to be seen how much further reduction in the risk of death is possible for infected individuals. Of note, recent studies have indicated that, even in the HAART era, mortality rates among those who are HIV-infected are still at least four times that of the general population [10,11], indicating room for further improvement.
In contrast with our findings, a recent large study following patients from HAART initiation found no improvement in survival to match the observed improvements in virological outcome over the first decade of HAART among 20 000 patients in Europe and North America . However, this study only considered follow-up to 1 year from HAART initiation, so would not pick up the improvements in medium- to long-term survival. Two further studies, one using seroconverter data from the GEMES cohort of Spain, observed a similar continuing reduction in the risk of death to that reported here from this UK-based study [12,13]. Healthcare planners will need to account for these ongoing improvements in terms of the number of people under care.
Concurrent with the observed survival improvements, we saw an increase in the proportion of person-time spent on HAART, and changes in the make-up and timing of HAART. The first choice treatment regimen has changed from being predominantly based on an unboosted PI, to being either NNRTI- or boosted PI-based. Improved simplicity and potency of more recent drug regimens, as well as better drug reaction profiles, are likely to have contributed to the continuing improvements in outcome, and hopefully will continue to do so.
We observed a trend towards initiating treatment earlier following seroconversion in later calendar periods, even after excluding treatment in primary infection. This was mainly driven by the differences between 1996–1997 and later, most likely due to the availability of HAART; nevertheless there remained a suggestion of shorter time to HAART initiation in the latest calendar period 2004–2006. This may be accounted for by the shorter average interval between the last negative and first positive HIV tests in later periods, since individuals were being diagnosed closer to their date of seroconversion. However, we also observed a similar pattern in the time from first positive HIV test to HAART initiation (results not shown). This suggests a possible move towards a choice of earlier treatment, despite no change in the UK guidelines to recommend this [14,15]. One possible explanation for initiating HAART sooner after seroconversion may be the decline over calendar time in initial CD4 cell counts after seroconversion, as observed in a recent study based on 22 seroconverter cohorts in Europe, Australia and Canada . However, if there is a real trend towards initiating HAART at higher CD4 cell counts, then the long-term consequences of such strategies will need to be continually monitored and evaluated.
Survival has improved across all age groups and, despite the natural ageing of the cohort, we found no evidence to suggest that the relative risk of death between age groups has changed over time. However, we found that patients infected through injecting drug use continue to have poorer outcomes than other HIV exposure groups, in agreement with a number of other studies [5,12,17]. Although survival among IDU has improved considerably, these improvements have not matched those of other risk groups, with the hazard of death from any cause in 2000–2006 being over seven times that of those infected through sex between men. One possible reason may be limited access to treatment; indeed, only 45% of IDU had initiated HAART 5 years from seroconversion, compared to 57% of non-IDU. However, IDU spent a similar proportion of person-time as other groups reportedly on HAART. Furthermore, the percentage of time with viral load < 1000 copies/ml while reportedly on HAART (a surrogate for adherence to prescribed therapy which was not explicitly captured in these data) was similar amongst IDU and non-IDU (74% and 78%, respectively). It is likely that high levels of coinfection with hepatitis B and C [18–20] compromise survival improvements in IDU. In the era of HAART and improved survival, these coinfections are likely to contribute increasingly to the impaired survival observed in this subgroup, for example through liver diseases associated with hepatitis C [21,22]. Indeed, in 2000–2006, 12% of deaths among IDU were attributed to hepatitis C compared to just 2% among non-IDU.
In 2000–2006, we found that 94% of patients who are newly infected with HIV and identified early are expected to survive at least 10 years from infection. This is likely to be the best that can be achieved with current treatments, given the optimal circumstances of patients in this study. The importance of identifying infection early in terms of future health is clear, yet in the UK in 2005 over one-third of adults newly diagnosed with HIV already had CD4 cell counts below the minimum threshold for initiating therapy as recommended by national guidelines (< 200 cells/μl), and 11% presented with AIDS . In countries with good access to therapy, our study shows the promising potential survival benefits that HAART can now offer. However, further promotion of HIV testing should be encouraged so that diagnosis can be made as early as possible and therapy initiated when the full benefits can be realised.
The authors thank all the UK Register participants for allowing their routine clinical data to be included. The tireless work of all colleagues at the clinical centres is gratefully acknowledged, and the following are worthy of a mention: Pat Byrne, Royal Free NHS Trust, London; Laura Ellis, The Lothian University Hospitals NHS Trust, Edinburgh; Lisa Heald, Brighton & Sussex University Hospitals NHS Trust, Brighton; Chris Higgs, Chelsea & Westminster NHS Trust, London; Ruth Johnstone, Mortimer Market Centre, London; Sheila Morris, The Lothian University Hospitals NHS Trust, Edinburgh; Scott Mullaney, St. Mary's Hospital, London.
Ethical approval: The UK Register has research ethics approval 04/Q2707/155.
Funding: The UK Register is funded by the Medical Research Council.
1. CASCADE collaboration. Determinants of survival following HIV-1 seroconversion after the introduction of HAART. Lancet
2. The UK Collaborative Group for HIV and STI Surveillance. A complex picture. HIV and other sexually transmitted infections in the United Kingdom: 2006
. London: Health Protection Agency, Centre for Infections; 2006.
3. Chadborn TR, Delpech VC, Sabin CA, Sinka K, Evans BG. The late diagnosis and consequent short-term mortality of HIV-infected heterosexuals (England and Wales, 2000–2004). AIDS 2006; 20:2371–2379.
4. Sabin CA, Smith CJ, Youle M, Lampe FC, Bell DR, Puradiredja D, et al
. Deaths in the era of HAART: contribution of late presentation, treatment exposure, resistance and abnormal laboratory markers. AIDS 2006; 20:67–71.
5. May MT, Sterne JA, Costagliola D, Sabin CA, Phillips AN, Justice AC, et al
, Antiretroviral Therapy (ART) Cohort Collaboration. HIV treatment response and prognosis in Europe and North America in the first decade of highly active antiretroviral therapy: a collaborative analysis. Lancet 2006; 368:451–458.
6. Lampe FC, Gatell JM, Staszewski S, Johnson MA, Pradier C, Gill MJ, et al
. Changes over time in risk of initial virological failure of combination antiretroviral therapy: a multicohort analysis, 1996 to 2002. Arch Intern Med 2006; 166:521–528.
7. UK Register of HIV Seroconverters Steering Committee. The UK register of HIV seroconverters: methods and analytical issues. Epidemiol Infect
8. Clayton C, Hills M. Statistical models in epidemiology. Oxford: Oxford University Press; 1993.
9. Tyrer F, Walker AS, Gillett J, Porter K, UK Register of Seroconverters. The relationship between HIV seroconversion illness, HIV test interval and time to AIDS in a seroconverter cohort. Epidemiol Infect 2003; 131:1117–1123.
10. Ormaasen V, Sandvik L, Dudman SG, Bruun JN. HIV related and non-HIV related mortality before and after the introduction of highly active antiretroviral therapy (HAART) in Norway compared to the general population. Scand J Infect Dis 2007; 39:51–57.
11. Lohse H, Hansen AB, Pedersen G, Kronborg G, Gerstoft J, Sorensen HT, et al
. Survival of persons with and without HIV infection in Denmark, 1995–2005. Ann Intern Med 2007; 146:87–95.
12. Perez-Hoyos S, Ferreros I, del Amo J, Muga R, del Romero J, de Olalla PG, et al
. Survival and progression to AIDS in a seroconverter cohort in the post-highly active antiretroviral therapy era: the effectiveness goes on. AIDS 2006; 20:289–291.
13. Lima VD, Hogg RS, Harrigan PR, Moore D, Yip B, Wood E, et al
. Continued improvement in survival among HIV-infected individuals with newer forms of highly active antiretroviral therapy. AIDS 2007; 21:685–692.
14. British HIV Association. BHIVA treatment guidelines. HIV Med
15. British HIV Association. BHIVA treatment guidelines. HIV Med
16. Dorrucci M, Rezza G, Porter K, Phillips A, CASCADE collaboration. Temporal trends in postseroconversion CD4 cell count and HIV load: the Concerted Action on Seroconversion to AIDS and Death in Europe Collaboration, 1985–2002. J Infect Dis 2007; 195:525–534.
17. Van Sighem A, Danner S, Ghani AC, Gras L, Anderson RM, de Wolf F, ATHENA National Observational Cohort Study. Mortality in patients with successful initial response to highly active antiretroviral therapy is still higher than in non-HIV-infected individuals. J Acquir Immune Defic Syndr 2005; 40:212–218.
18. Jimenez Fabrega X, Carballo Almeida A, Batalla Martinez C, Comin Bertran E, Cuenca Oliva AM, Ezpeleta Garcia A, et al
. The prevalence of infection by the hepatitis B, C and human immunodeficiency viruses in drug users. Aten Primaria 1999; 24:368–371.
19. Rodriguez-Mendez ML, Gonzalez-Quintela A, Aguilera A, Carballo E, Barrio E. Association of HCV and HBV markers in Spanish HIV-seropositive patients in relation to risk practices. Hepatogastroenterology 2003; 50:2093–2097.
20. Mohsen AH, Murad S, Easterbrook PJ. Prevalence of hepatitis C in an ethnically diverse HIV-1-infected cohort in south London. HIV Med 2005; 6:206–215.
21. Rockstroh JK, Mocroft A, Soriano V, Tural C, Losso MH, Horban A, et al
. Influence of hepatitis C virus infection on HIV-1 disease progression and response to highly active antiretroviral therapy. J Infect Dis 2005; 192:992–1002.
Analysis and writing committee
F. M. Ewings, K. Bhaskaran, K. McLean, D. Hawkins, M. Fisher, S. Fidler, R. Gilson, D. Nock, R. Brettle, M. Johnson, A. Phillips, A. Johnson, K. Porter for the UK Register of HIV Seroconverters.
Members of the UK Register of Seroconverters steering committee
A. Phillips (Chair), Royal Free & University College Medical School (RFUCMS), London; A. Babiker, MRC Clinical Trials Unit, London; R. Brettle, The Lothian University Hospitals NHS Trust, Edinburgh; J. Darbyshire, MRC Clinical Trials Unit, London; V. Delpech, Health Protection Agency, London; P. Easterbrook, King's College Hospital, London; S. Fidler, St. Mary's Hospital, London; M. Fisher, Brighton & Sussex University Hospitals NHS Trust, Brighton; R. Gilson, UCL Centre for Sexual Health & HIV Research and the Mortimer Market Centre, Camden Primary Care Trust, London; D. Goldberg, Health Protection Scotland, Glasgow; D Hawkins, Chelsea & Westminster NHS Trust, London; H. Jaffe, University of Oxford, Oxford; A. Johnson, RFUCMS, London; M. Johnson, Royal Free NHS Trust and RFUCMS, London; K. McLean, The West London Centre for Sexual Health, London; D. Pillay, RFUCMS, London.
22. Weis N, Lindhardt BO, Kronborg G, Hansen AB, Laursen AL, Christensen PB, et al
. Impact of hepatitis C virus coinfection on response to highly active antiretroviral therapy and outcome in HIV-infected individuals: a nationwide cohort study. Clin Infect Dis 2006; 42:1481–1487.