Projections for the future course of the HIV epidemic are important for planning health services and to help set research priorities . UK projections have been derived using back-calculation  or Bayesian approaches  based on various data sources, including the unlinked HIV prevalence monitoring programme , the National Survey of Sexual Attitudes and Lifestyle  and HIV and AIDS case reports . Surveillance strategies based solely on HIV and AIDS reports have several important limitations. First, new reported HIV infections and data from unlinked anonymous seroprevalence studies may not be representative of all HIV-infected patients, and give no indication as to the duration of infection among the identified individuals. Second, the long incubation period between HIV and the development of AIDS means that recent data on AIDS cases do not reflect current epidemiology. Other factors that have made it difficult to interpret trends over time in reported AIDS cases include the heterogeneous nature of an AIDS diagnosis since the revision of the US Centers for Disease Control and Prevention (CDC) AIDS case definition to include conditions that occur at a less advanced stage of immunodeficiency [7,8], reporting delays , differences in diagnostic practices and case finding and the variable use of combination antiretroviral therapies and prophylaxis against opportunistic infections [10,11]. Finally, existing surveillance strategies cannot assess whether HIV-infected persons are being diagnosed at an early or late stage of disease. Additional approaches to monitoring of the HIV epidemic are needed in order to improve characterization of the changing epidemiology of HIV infection and access to therapies, and so to help to focus prevention activities.
The CD4 cell count is now measured routinely during the clinical monitoring of HIV infection, and it has been suggested that information on the CD4 cell count at initial HIV diagnosis (CD4HIV) and at the onset of clinical AIDS (CD4AIDS) could provide useful staging information to supplement HIV/AIDS case surveillance . In 1993, a single measured CD4 cell count less than 200 × 106 cells/l, irrespective of symptoms, was incorporated into the CDC AIDS surveillance case definition , although this component of the revised definition has not been adopted elsewhere. To date, most AIDS surveillance centres in Europe and the United States have not routinely collected information on the CD4 cell count at the time of HIV diagnosis. However, pilot studies of more detailed immunological monitoring at AIDS diagnosis and advanced immunodeficiency were initiated in Scotland in 1992  and more recently in England ; D. Bennet, personal communication.
The CD4HIV reflects the degree of HIV progression at the time of initial presentation and diagnosis. The stage of disease at diagnosis will be influenced by the willingness of infected persons to be tested, as well as by attitudes to case finding and contact tracing in different regions and risk groups. Knowledge of the CD4HIV could be used in several ways: to help target population subgroups presenting late in disease and in need of programmes encouraging HIV testing and earlier diagnosis, to help in the planning of HIV health care services and in the allocation of resources for prevention and therapy, to enable assessment of the impact of regional and national HIV awareness campaigns, and to refine existing estimates of the incubation distribution from HIV to AIDS through a back-calculation approach based on the CD4HIV .
In contrast, the CD4AIDS provides information on the level of immunosuppression at which clinical AIDS manifests itself. Differences according to time period, risk group, gender and geography may indicate variations in the natural history of disease and/or in access to antiretroviral and prophylactic therapy in these groups. Such information would be helpful for the interpretation of trends in the clinical spectrum of AIDS cases.
In this study, we evaluated the use of the CD4 count at HIV and AIDS diagnoses as additional potential surveillance measures for monitoring changes in the natural history of the HIV epidemic, through an examination of (i) the temporal trend over a 10-year period (1986–1996) in CD4HIV and CD4AIDS, and (ii) the influence of various patient characteristics and treatment factors on these values in a large London-based clinic population of HIV-1-seropositive individuals.
Data were collected from 5921 adult HIV-1-seropositive persons and 2835 patients with AIDS meeting the 1987 CDC AIDS case definition  identified from three HIV clinic sites in West London over a 10-year period from 1 January 1986 to 1 October 1996: Charing Cross Hospital (n = 334), Victoria Clinic (n = 165) and Chelsea and Westminster Hospital (n = 1492). This represented almost 60% of all AIDS cases in London. Routes of access to HIV testing in this region are primarily through self-referral to clinics for sexually transmitted diseases and drug-dependency units. Computerized records provided the following information on these patients: date of birth, gender, HIV transmission category, AIDS index diagnosis, serial CD4 cell counts, date of HIV and AIDS diagnoses and dates of initiation of antiretroviral therapy and/or prophylaxis against Pneumocystis carinii pneumonia (PCP) (oral trimethoprim–sulphamethoxazole or dapsone, and aerosolized pentamidine).
The main AIDS-defining conditions were PCP (n = 737; 26%), Kaposi's sarcoma (n = 480; 16.9%), oesophageal candidiasis (n = 285; 10.1%), cytomegalovirus (n = 150; 5.3%), wasting syndrome (n = 113; 4%), lymphoma (n = 88; 3.1%), atypical mycobacteria (n = 65; 2.3%) and various presumptive infections (n = 458; 16.2%). Other diagnoses included progressive multifocal leukoencephalopathy, dementia, toxoplasmosis, cryptosporidiosis, tuberculosis and cryptococcal disease (n = 219; 7.7%). Antiretroviral therapy (mainly zidovudine) was received by 1938 patients (77.4%) and 1836 (73.7%) received prophylaxis against PCP at some time during the follow-up period.
CD4 cell counts
All patients were reviewed periodically and their T lymphocyte subsets measured as clinically indicated by flow cytometry (FACScan, Becton Dickinson, Cockeysville, USA) up to June 1995, and Ortho (Cytoron) until 31 December 1996.
The definitions of CD4HIV and CD4AIDS were developed as a result of a preliminary exploration of different interval time cut-offs of CD4 cell measurements around the HIV and AIDS diagnoses. The decision for the most appropriate cut-off was based on a combination of optimizing the number of eligible patients, stability in the CD4 cell count and biological rationale. There was an approximate 10–15 × 106 cells/l difference in the CD4HIV value according to whether the interval between the HIV test and CD4 cell count was 0 or 6 months (Table 1). The CD4HIV was therefore defined as the nearest CD4 cell count within 2 months of HIV diagnosis: 1991 (23%) HIV-infected individuals were eligible, with a median CD4 cell count at HIV diagnosis of 334 × 106 cells/l (interquartile range 136–518).
There was little difference in the median CD4AIDS value according to whether a cut-off of 1, 2, 3, 4, 5 or 6 months before the AIDS diagnosis was used. A less rigorous definition that included a period of 1–6 months either before or after the AIDS diagnosis was about 6 × 106 cells/l lower (Table 1). The CD4AIDS was defined as the last CD4 cell count in the 2 months prior to the development of AIDS; 1093 (38.6%) patients were eligible (median 65 × 106 cells/l; interquartile range 28–156).
A natural logarithmic transformation for CD4 cell count was used in order to make the distribution closer to normal or Gaussian, and so stabilize the variance. Trends in patient characteristics over time were analysed using a chi-square test, while the Wilcoxon's rank, sum test, was used for comparisons of median CD4 cell counts between patient subgroups. CD4AIDS was presented for the following years grouped together (≤ 1988, 1989–1990, 1991–1992, 1993–1994 and 1995–1996) because of smaller numbers. Simple and multiple linear regression were used to examine the influence of various demographic and treatment covariates on CD4HIV and CD4AIDS, and on the changes in these values over time. A nested case-control analysis was used to identify any distinctive clinical or demographic characteristics (including HIV transmission category, gender, year and age at AIDS diagnosis, and nature of diagnosis) among persons who were unaware of their HIV infection at AIDS diagnosis.
As of 1 October 1996, there were 1920 deaths among the 2835 patients with AIDS. Survival following AIDS diagnosis was analysed using the Kaplan–Meier product-limit method for each year of diagnosis, and for those persons who were aware or unaware of their HIV infection at AIDS diagnosis. A Cox proportional hazards model was fitted to the data to study the impact of prior knowledge of HIV status at AIDS diagnosis, after adjusting for CD4 cell count at AIDS diagnosis, age and year of diagnosis.
HIV diagnoses recorded were 1831 prior to 1986, 558 in 1987, 452 in 1988, 454 in 1989, 465 in 1990, 459 in 1991, 419 in 1992, 329 in 1993, 355 in 1994, 312 in 1995 and 287 in 1996. Overall, 84.9% of patients were homosexual or bisexual, 7.8% were heterosexual, 7% injecting drug users and 0.3% were of unknown transmission group. Only 6.3% were among women.
Of the 2835 patients with an AIDS diagnosis: 211 diagnoses were prior to 1986, 197 in 1987, 229 in 1988, 256 in 1989, 274 in 1990, 286 in 1991, and 325 in 1992, 310 in 1993, 344 in 1994, 244 in 1995 and 159 in 1996. The mean age at AIDS diagnosis was 37.4 years (± 8.62), and the median time between HIV and AIDS diagnosis was 2.17 years (range 0.08–13.4). The distribution of risk groups was similar to those observed among patients with an HIV diagnosis.
Overall, only 1991 (33.6%) and 1093 (38.6%), respectively, of our HIV-infected patients had a CD4 cell count available within 2 months of their HIV-positive diagnosis and in the 2 months prior to their the initial AIDS diagnosis. This is largely attributable to the absence of routine CD4 cell monitoring prior to 1987. The number and frequency of patients with an available CD4HIV increased from 457 (25%) prior to 1987 to 83 (15%) in 1987, 112 (25%) in 1988, 167 (37%) in 1989, 246 (53%) in 1990 and 228 (79%) in 1996. Similarly, the percentage of patients with available CD4AIDS values increased from 91 (4%) prior to 1987, to 25 (13%) in 1987, 53 (23%) in 1988, 71 (28)% in 1989, 110 (40%) in 1990 and 76 (48%) in 1996.
Compared with patients with a CD4HIV or CD4AIDS value, patients with a missing CD4 value were slightly younger (median 30.5 versus 32.1 years;P < 0.001), more likely to have an AIDS index diagnosis of PCP (30.2% versus 19.3%;P < 0.001), to have a shorter time between their HIV and AIDS diagnosis (1.09 versus 3.71 years;P < 0.001) and were less likely to have received either antiretroviral therapy (28.9% versus 77.8%;P < 0.001) or PCP prophylaxis (16.6% versus 57.2%;P < 0.001) during follow-up.
Temporal trends in CD4 cell counts at diagnosis of HIV and AIDS
Figure 1 shows no evidence of a progressive increase in CD4HIV over a 10-year period: the median value was 350 × 106 cells/l cells prior to 1986, 374 × 106 cells/l cells in 1992 and 348 × 106 cells/l cells in 1996. There was however a slight decline in the proportion of patients presenting with advanced immunodeficiency at HIV diagnosis (i. e. CD4 cell count < 200 × 106 cells/l) from 44.6% in 1988 to 32.5% in 1996.
In contrast, there was a delay in the onset of clinical AIDS, with a decline in the CD4AIDS over time which reached a plateau in 1988 (P = 0.009) (Fig. 1). The median value of CD4AIDS decreased by approximately 50 × 106 cells/l: from 99 × 106 cells/l prior to 1986 to 99, 58, 63, 68 and 60 × 106 cells/l in 1988, 1990, 1992, 1994 and 1996, respectively. A diagnosis of AIDS was coincident with, or within 3 months of, their first HIV positive test (i.e. CD4AIDS = CD4HIV) in 289 patients.
Factors associated with trends in CD4 cell count at HIV diagnosis
Table 2 shows the changing demography of patients diagnosed with HIV infection and the median values of CD4HIV according to five 2-year periods of HIV diagnosis stratified by transmission group, gender, age at diagnosis and clinical site. Prior to 1988, 93% of HIV cases were attributed to homosexual contact, but this has declined steadily since then to 85.9% during 1989 and 1990, and 80.1% during 1995 and 1996. There has also been a proportionate increase in HIV diagnoses secondary to heterosexual contact and injecting drug use from less than 7% before 1988 to 14.1% during 1989 and 1990, and 18.0% during 1995 and 1996, although the overall number of such cases has remained small. The male-to-female ratio (approximately 10:1) and age at first HIV positive test has remained more or less constant over this 10-year period.
We observed no consistent differences in the CD4HIV value according to transmission group or gender, but there was a striking difference according to age (Table 2). A lower CD4HIV value was associated with a more advanced age at diagnosis (P < 0.001). Persons presenting with HIV infection, aged 40 years or over had a median CD4HIV count approximately 200–300 × 106 cells/l less than that of persons aged ≤ 25 years, and this was apparent over all time periods. We found no statistically significant differences according to site, although the number of patients at the Victoria Clinic accounted for less than 10%.
Table 3 summarizes the multivariate analysis of factors that may influence CD4HIV. Only younger age and injecting drug use were significant predictors of a higher CD4 cell count at first HIV diagnosis.
Factors associated with trends in CD4 cell count at AIDS diagnosis
Table 4 shows changes in the nature of AIDS index diagnoses over a 10-year period from 1986 to 1996. PCP, oesophageal candidiasis and Kaposi's sarcoma were the most frequent AIDS index diagnoses, and overall accounted for almost half of the diagnoses. The relative frequency of first AIDS diagnoses attributable to PCP decreased from 23.0% prior to 1988 to 16.8% during 1991 and 1992 and 15.3% during 1995 and 1996. Similarly, the frequency of Kaposi's sarcoma decreased from 19.5% prior to 1988 to 16.5% during 1991 and 1992 and 15.8% during 1995 and 1996. In contrast, the proportion of non-PCP opportunistic infection diagnoses has remained fairly stable over time. The fall in the incidence of PCP as an AIDS index diagnosis was coincident with the widespread adoption into clinical practice of antiretroviral therapy in 1988 (20.7% use of antiretrovirals prior to 1988, 52.4% during 1989 and 1990, 75.9% during 1991 and 1992 and 54.2% during 1995 and 1996) and of primary PCP prophylaxis in 1989 (8.0% prior to 1988, 32.0% during 1989 and 1990, 70.1% during 1991 and 1992 and 63.1% during 1995 and 1996), either alone or in combination with each other. Over the same time period, there were no significant changes in age at AIDS diagnosis; however, heterosexual and injecting drug users were more common in recent years (< 5% prior to 1988, 9.5% during 1991 and 1992 and 14.1% during 1995 and 1996).
Table 4 also shows the median values of CD4AIDS according to patient characteristics, including nature of AIDS index diagnosis and use of PCP prophylaxis and antiretroviral therapies prior to the diagnosis of AIDS. We observed changes in CD4AIDS for certain AIDS index diagnoses. There was a decline in the median CD4AIDS for PCP (66 × 106 cells/l prior to 1988 to 40 × 106 cells/l during 1995 and 1996), for atypical mycobacterial infection (104 × 106 cells/l prior to 1988 to 60 × 106 cells/l during 1991 and 1992 and 14 × 106 cells/l during 1995 and 1996), lymphoma (353 × 106 cells/l prior to 1988 to 161 × 106 cells/l during 1991 and 1992 and 152 × 106 cells/l during 1995 and 1996) and cytomegaloviral disease (203 × 106 cells/l prior to 1988 to 44 × 106 cells/l during 1991 and 1992 and 11 × 106 cells/l during 1995 and 1996). There was no consistent trend in the CD4 cell count over time for Kaposi's sarcoma, wasting, or oesophageal candidiasis.
There was also a 20–30 × 106 cell/l decline in the median CD4AIDS value with the introduction and increasing use of antiretroviral therapy in conjunction with PCP prophylaxis: from approximately 89 × 106 cells/l prior to 1991 to approximately 48 cells × 106 cells/l during 1991 and 1992 and 39 × 106 cells/l during 1995. At all time periods, a lower CD4 cell count at AIDS diagnosis was associated with combined use of antiretroviral therapy and PCP prophylaxis compared with values in those who received neither therapy prior to their AIDS diagnosis (38 versus 90 × 106 cells/l in 1989 and 1990, 48 versus 128 × 106 cells/l in 1991 and 1992 and 39 versus 132 × 106 cells/l in 1995 and 1996). Among those who received antiretroviral therapy only, after an initial reduction, there was a further rise in CD4AIDS between 1991 and 1992. We examined possible reasons for this rebound in CD4AIDS, including changes in the CD4 cell count level at which antiretroviral therapy is initiated. Overall, there was no significant change over time in the CD4 level at initiation of antiretroviral therapy except among those patients who received only antiretroviral therapy, who had a consistently higher CD4 count. In a multivariate analysis, the use of antiretroviral therapy and PCP prophylaxis prior to an AIDS diagnosis and the nature of the AIDS diagnosis were independent predictors of CD4AIDS. The temporal decline in CD4AIDS was no longer significant after adjustment for prior use of antiretroviral therapy (Table 5).
Characteristics of patients unaware of their HIV infection at AIDS diagnosis
Testing for HIV infection at or within 3 months of the development of AIDS occurred in 769 (27.1%) of the 2835 patients with AIDS. When compared with 2066 HIV-positive patients with an interval of at least 3 months between their HIV-positive and AIDS diagnoses, these patients were significantly more likely to have been diagnosed with AIDS prior to 1991 (54.9 versus 36.1%;P < 0.0001), to be diagnosed with PCP as their AIDS index diagnosis (39.1 versus 21.1%;P < 0.0001) and were much less likely to have received antiretroviral therapy (5.1 versus 44.5%;P < 0.0001) or PCP prophylaxis (7.4 versus 41.5%;P < 0.0001) prior to their AIDS diagnosis (Table 6). We found no differences between the two groups according to gender, HIV transmission group, age, CD4 cell count at diagnosis or the frequency of any other AIDS diagnoses.
In the 2835 patients with AIDS and a median follow-up of 16.3 months, the median survival from AIDS diagnosis was 21 [95% confidence interval (CI) 20.1–21.8] months. The median survival was 20.1 (95% CI 18.5–21.7) months for patients diagnosed with AIDS during or prior to 1988, 20.3 months for those diagnosed between 1989 and 1990 (95% CI 18.7–21.9), 20.95 months for those diagnosed between 1991 and 1992 (95% CI 19.0–22.9) and 22.07 months for those diagnosed between 1993 and 1994 (95% CI 19.9–24.2) (log rank test, P = 0.0005). Persons tested for HIV less than 3 months before developing AIDS had a shorter survival than those tested more than a year before diagnosis (median survival 20.6 months versus 24.1 months;P = 0.04). However, after adjustment for CD4AIDS (P < 0.0001), and use of antiretroviral therapy (P = 0.08) or PCP prophylaxis (P = 0.02), prior to AIDS diagnosis in a Cox proportional hazards model, neither year of AIDS diagnosis nor duration of time between HIV and AIDS diagnosis were independently predictive of survival.
This analysis shows that temporal changes in the CD4HIV and CD4AIDS among different demographic groups can provide useful insights into the changing natural history of HIV infection. We observed no significant change in CD4HIV over a 10-year period from 1986 to 1996 in the overall study population. These findings would initially suggest that there has been no substantial improvement in access to and uptake of HIV testing among those who perceive themselves to be at risk of infection. However, further analysis of trends in CD4HIV within selected subgroups found an approximate 100 × 106 cells/l increase in CD4HIV among homosexual men over 10 years, and a decline by at least 100 × 106 cells/l among injecting drug users and heterosexuals. Over the same time period, there was an approximate 10% decline in the number of new HIV diagnoses among homosexual men in our clinic and a two- to fourfold increase in reported new infections among injecting drug users and the heterosexual population. This suggests that the stability in CD4HIV observed over time mainly reflects the balance between a growing proportion of injecting drug users and heterosexuals, who tend to present with more advanced HIV disease, and a declining proportion of homosexual men, who have better access to and uptake of HIV testing and who, therefore, present earlier.
A wide variation in the CD4HIV in different demographic and transmission groups and health care settings has also been reported by other investigators [12,15–18]. Among newly identified HIV-infected patients at two clinics for sexually transmitted diseases in Baltimore, United States (which included a high proportion with heterosexually acquired HIV infection) in 1989–1991, 50% had initial CD4 cell counts < 500 × 106 cells/l . In contrast, the majority of patients who were diagnosed HIV positive at a same day testing site in London had CD4 counts in the normal range (median 500 × 106 cells/l), suggesting recent infection . This contrasts with the Scottish experience, where in a survey of 1679 adult HIV-infected patients, 67% of non-injecting drug users were already severely immunosuppressed at their initial presentation, but only 31% of injecting drug users were similarly affected . The variation in CD4HIV across exposure categories observed in these various studies may reflect a combination of differences in test-seeking behaviour between the risk groups, ready access to HIV testing among injecting drug users in certain treatment programmes, and the more recent spread of HIV infection into the injecting drug user and heterosexual communities.
The major caveat in the interpretation of trends in CD4HIV is the high proportion of patients with missing CD4HIV values prior to 1990, which was more likely in younger adults. However, our hospital was one of the first to offer and perform regular CD4 cell count tests and, therefore, has the most comprehensive CD4 data available from these early years. Generalizing from our CD4HIV data is also limited by the predominant representation of homosexual men in our cohort, who tend to have a particularly high level of access to HIV therapies and clinical trials compared with some other patient groups. However, between 1986 and 1996, the predominant group affected by HIV in the UK was homosexual men, and our patient cohort accounted for 50% of all HIV infections and 60% of AIDS cases notified in London. Finally, although there is no precise information as to how many of our patients may have had an initial HIV test elsewhere, a review of records from the clinic's administrative database, which includes information on transfers, suggests that these account for less than 10% of our patients. We also found no systematic differences across the three clinic sites in the profile of CD4HIV and CD4AIDS over time.
As has been observed in several other cohort studies conducted over a comparable time period, our analyses revealed a decline in the overall proportion of AIDS index diagnoses due to PCP, and to a lesser extent Kaposi's sarcoma [19–24]. This is mainly a consequence of the introduction in the late 1980s of effective primary prophylaxis against PCP, as the incidence of PCP in patients unaware of their HIV status has not changed . In confirming the well-established associations between level of immunodeficiency and susceptibility to particular AIDS-defining illnesses , we also observed a declining CD4 cell count for several AIDS-defining conditions, including atypical mycobacterial infections, wasting syndrome, cytomegaloviral disease and lymphoma.
The fall in CD4AIDS in our cohort, from a median of 100 × 106 cells/l prior to 1989 to 60 × 106 cells/l in 1989, compares with a decline from a median of 73 × 106 cells/l in 1986 to 47 × 106 cells/l in 1990 observed in a review of 6932 patients with AIDS in Germany . This phenomenon can be largely attributed to the introduction of antiretroviral therapy in 1987 and of primary prophylaxis against PCP in 1989. Patients who received antiretroviral drugs and/or PCP prophylaxis were more immunocompromised when diagnosed with AIDS than individuals who received neither, and the downward trend in CD4AIDS was apparent for both PCP and non-PCP opportunistic infections. The plateau in the value of CD4AIDS for PCP since 1990 suggests that the antiretroviral therapies available at the time and PCP prophylaxis may have reached their maximum epidemiological effect . However, further analysis of trends since the introduction of triple antiretroviral therapy with protease inhibitors and non-nucleoside reverse transcriptase inhibitors are required to confirm this observation.
We also observed a decreasing proportion of patients who were apparently unaware that they were HIV infected when they received their first AIDS diagnosis (177 (23%) prior to 1988, 128 (16.6%) in 1991 and 1992 and 99 (12.9%) in 1995 and 1996) and a corresponding increase in the proportion of patients with more than a year between their HIV and AIDS diagnoses. This compares with 49% who were considered unaware of their HIV diagnosis at AIDS (as defined by an interval of 9 months or less between first HIV-positive test date and an AIDS diagnosis) in a review of UK data up to 1992 by Porter  and the 20% reported from Australia by Kaldor . In our analysis, better survival from AIDS diagnosis was associated with a longer awareness of an HIV diagnosis prior to AIDS diagnosis. This was explained in a multivariate analysis by the higher level and duration of use of antiretroviral therapy and PCP prophylaxis pre-AIDS among those aware of their serostatus at AIDS diagnosis, as well as earlier diagnosis and treatment of AIDS indicator disease.
What are the implications of these various findings for prevention efforts in our clinic population? The observation that a third of homosexual men in our cohort continue to present with either advanced HIV disease or with less than 3 months between their HIV and AIDS diagnosis indicates that willingness to be tested among those at risk is less than optimal. It seems inconceivable, with the widespread availability of educational programmes in this community, that their delayed presentation is a result of a failure to perceive their risk of infection. This highlights the continuing need for educational programmes that reinforce the benefits of an early HIV diagnosis. The continued monitoring of the characteristics of these late presenters should help to target efforts to improve access to HIV/AIDS diagnosis and care.
There are several considerations pertinent to a decision to adopt CD4HIV and CD4AIDS as additional surveillance measures. Foremost is the lack of standardization in CD4 monitoring. We found that individuals without a CD4 cell count at HIV or AIDS diagnosis were likely to be younger (CD4HIV) and less likely to have received antiretroviral therapy (CD4AIDS), as was observed by Hanson . Since 1992, less than a quarter of our patients had a missing CD4 cell count within 2 months of HIV diagnosis, and only a third were without a CD4AIDS. This reduced to 10% and 15%, respectively, when a less stringent cut-off of 6 months was used; it is likely to continue to decrease in the future. In addition, some instances of apparently missing CD4HIV and CD4AIDS values are the result of transfers from other clinical centres, where this information may have been documented. This problem would be largely overcome by reporting at a national level to allow information on previous HIV tests to be linked through soundex and date of birth.
In summary, monitoring CD4HIV and CD4AIDS permits better epidemiological tracking of the HIV epidemic. However, meaningful interpretation of trends in these measures requires the availability of linked data on demographic characteristics, nature of the AIDS diagnosis and the CD4 cell count at the introduction of antiretroviral therapy and PCP prophylaxis. The experience in Scotland has shown that it is valuable to collect such information nationally [3,13]. These measures provide a crude assessment of the extent and uptake of medical care and therapy and could also be useful in comparing clinical practice across institutions. Monitoring of the CD4 counts at these key reference points would provide a low-cost system of surveillance since it utilizes a routine practice in clinical care. Such an approach also offers the flexibility to incorporate changes in therapeutic practice, such as the use of multiple antiretroviral combination therapies.
We are grateful to Juliet Kirk for her assistance in the preparation of this manuscript.
1. Report of an Expert Group (Chairman: Professor N E Day). The incidence and prevalence of AIDS and prevalence of other severe HIV disease in England and Wales for 1995 to 1999: projections using data to the end of 1994.
CDR Rev 1996, 6: R1 –R24.
2. Day NE, Gore SM, McGee MA. et al
. Predictions of the AIDS epidemic in the UK: the use of the back projection methods.
Philos Trans R Soc Lond (Biol) 1989, 325: 123 –134.
3. Raab GM, Gore SM, Goldberg DJ. et al
. Bayesian forecasting of the human immunodeficiency virus epidemic in Scotland.
J Stat Soc. A 1994, 157: 17 –30.
4. Department of Health. Prevalence of HIV in England and Wales. Annual Report of the Unlinked Anonymous Prevalence Monitory Programme
. London: Department of Health; 1997.
5. Johnson AM, Wadsworth J, Field J, Bradshaw S (eds). Sexual Attitudes and Lifestyle
. Oxford: Blackwell Scientific; 1994.
6. Waight PA, Rush AM, Miller E. Surveillance of HIV infection by voluntary testing in England.
Comm Dis Rep 1992, 2: R85 –88.
7. Centers for Disease control. Classification system for human T-lymphotropic virus type III/lymphadenopathy-associated virus infection. MMWR
8. Centers for Disease Control and Prevention. 1993 Revised classification system for HIV infection and expanded surveillance definition for AIDS among adolescents and adults. MMWR
9. Whitmore-Overton SE, Tillet HE, Evans BG, Allardice GM. Improved survival from diagnosis of AIDS in adult cases in the United Kingdom and bias due to reporting delays.
AIDS 1993, 7: 415 –420.
10. Rosenberg PS, Gail MH, Schrager LK. et al
. National AIDS incidence trends and the extent of zidovudine therapy in selected demographic groups.
J Acquir Immune Defic Syndr 1991, 4: 392 –401.
11. Graham NMH, Zeger SL, Park LP. et al
. Effect of zidovudine and Pneumocystis carinii prophylaxis on progression of HIV-1 infection to AIDS.
Lancet 1991, 338: 265 –269.
Collaborative Group. Use of monitored CD4 cell counts: predictions of the AIDS epidemic in Scotland. AIDS
Collaborative Group. CD4 surveillance in Scotland: perspectives on severe HIV related immunodeficiency. AIDS
14. Easterbrook PJ, Ly M Yu, Hawkins D, McLean K, Gazzard BG. CD4 cell counts of 200 × 106 cells/l or below in natural history studies or surveillance: is one enough or are two better?
CDR Rep 1997, 7: R179 –R185.
15. Hutchinson CM, Wilson C, Reichart CA, Marsiglia VC, Zenilman JM, Hook EW. CD4 lymphocyte concentrations in patients with newly identified HIV infection attending STD clinics.
:Potential impact on publicly funded health care resources.
J Am Med Assoc 1991, 10: 266: 253 –256.
16. Farizo KM, Buehler JW, Chamberland ME. et al
. Spectrum of disease in persons with human immunodeficiency virus infection in the United States.
J Am Med Assoc 1992, 267: 1798 –1805.
17. Bor R, Lipman M, Elford J, Murray D. et al
. HIV seroprevalence in a London same-day testing clinic.
AIDS 1994, 8: 697 –700.
18. Kaldor JM, French MAH. When do patients present with HIV infection?
Med J Aust 1993, 15 8: 37 –38.
19. Schwartlander B, Horshburgh CR, Hamouda O. et al
. Changes in the spectrum of AIDS-defining conditions and decrease in CD4+ lymphocyte counts at AIDS manifestation in Germany from 1986 to 1991.
AIDS 1992, 6: 413 –420.
20. Peters BS, Beck EJ, Colman DG. et al
. Changing disease patterns in patients with AIDS in a referral centre in the United Kingdom: the changing face of AIDS.
BMJ 1991, 302: 203 –207.
21. Montaner JSG, Le T, Hogg R. et al
. The changing spectrum of AIDS index diseases in Canada.
AIDS 1994, 8: 693 –696.
22. Brodt RH, Kamps BS, Gute P. et al
. Changing incidence of AIDS-defining illnesses in the era of antiretroviral combination therapy.
AIDS 1997, 11: 1731 –1738.
23. McDonald AM, Crofts N, Blumer CE. et al
. The pattern of diagnosed HIV infection in Australia, 1984–1992.
AIDS 1994, 8: 513 –519.
24. Munoz A, Schrager LK, Bacellar H. et al
. Trends in the incidence of outcomes defining acquired immunodeficiency syndrome (AIDS) in the multicenter AIDS cohort study: 1985–1991.
Am J Epidemiol 1993, 137: 423 –438.
25. Porter K, Wall PG, Evans BG. Factors associated with lack of awareness of HIV infection before diagnosis of AIDS.
BMJ 1993, 307: 20 –23.
26. Crowe SM, Carlin JB, Stewart KI. et al
. Predictive value of CD4 lymphocyte numbers for the development of opportunistic infections and malignancies in HIV- infected persons.
J Acquir Immune Defic Syndr 1991, 4: 770 –776.
27. Hanson DL, Horburgh CR, Fann SA, Havlick JA. Survival prognosis of HIV-infected patients.
J Acquir Immune Defi Syndr 1993, 6: 624 –629.