Success of Clinical Care for Human Immunodeficiency Virus Infection According to Demographic Group Among Sexually Infected Patients in a Routine Clinic Population, 1999 to 2004 FREE

Since the mid 1990s, human immunodeficiency virus (HIV)–infected populations under care in industrialized countries have experienced dramatic declines in death rates, AIDS incidence,¹ and the prevalence of immunosuppression,^2- 3 reflecting the increasing uptake and highly successful use of triple-combination antiretroviral therapy (CART).⁴ However, HIV clinic populations are composed of distinct demographic subgroups. In the United Kingdom (UK), there are increasing numbers of women and heterosexually infected individuals accessing care. In 2004, heterosexually infected men and women represented the largest risk group under care (47% vs 43% for homosexually infected men).⁵ While 90% of homosexually infected men were white, 70% of heterosexually infected subjects were black African, most of whom acquired HIV in sub-Saharan Africa.⁵ The success of HIV clinical care may vary across these demographic groups because of patient- and health care–related factors. In particular, black African patients have tended to access medical care at a later stage and with more advanced disease than white patients,^6- 9 presenting a greater challenge in achieving a successful outcome of treatment. Awareness of treatment issues, social circumstances, and experience of clinical care may also differ across demographic subgroups.^10- 11 We examined the prevalence of 2 markers, 1 of adverse prognosis (low CD4 cell count) and 1 of potentially suboptimal treatment success (raised viral load [VL] while receiving ART), among patients with HIV under care at a large London clinic from July 1, 1999, to December 31, 2004. We investigated whether (1) prevalence varied across demographic subgroups and (2) changes over time in prevalence differed between demographic subgroups. Health care in the UK is free at the point of use, and so the clinic population is not selected on ability to pay.

We included all sexually infected subjects seen at the Royal Free HIV outpatient clinic from July 1, 1999, to December 31, 2004. Patients infected through intravenous drug use and other routes were excluded because they compose less than 5% of this clinic population. The HIV clinic is staffed by a regular and experienced team and is open to referrals from HIV testing services, primary care, and other hospitals, and to self-referrals. Patients are cared for according to the British HIV Association treatment guidelines,¹² and are seen approximately every 3 months or more frequently. Information on demographics, ART use, morbidity, and mortality is ascertained from a clinic database and annual medical record reviews. Viral loads were measured using commercially available assays (Amplicor and Cobas; Roche).

Subjects were categorized into 1 of 5 demographic groups based on sex, ethnicity, and risk exposure: (1) all homosexually or bisexually infected men, (2) heterosexually infected white men, (3) heterosexually infected black African/other ethnicity men, (4) heterosexually infected white women, and (5) heterosexually infected black African/other ethnicity women.

We ascertained CD4 cell count and VL during 11 six-month intervals (July 1, 1999-December 31, 1999, to July 1, 2004-December 31, 2004) for all patients seen during each period, using the single value nearest the midpoint. ART use (yes or no) was determined at measurement of CD4 cell count and VL. Three end points were chosen: (1) low CD4 cell count (≤ 200/μL, including all subjects with ≥1 CD4 cell count in an interval); (2) raised VL while receiving ART (VL >50 copies/mL, including all subjects with ≥1 VL measurement in the interval and receiving ART at the time); and (3) VL greater than 50 copies/mL among the subgroup of patients receiving ART for 24 weeks or longer (allowing sufficient time for viral suppression to be achieved). July 1, 1999, was chosen as the starting point for analysis because CART use was fully established and VLs were routinely measured to a lower limit of 50 copies/mL.

Poisson regression was used to examine the effect of demographic group and calendar period on the prevalence of low CD4 cell count and raised VL while receiving ART.¹³ We used generalized estimating equations (SAS statistical software, version 6.12¹⁴) to account for repeated observations from individual subjects in successive periods. Analyses were carried out separately for men and women. In initial models, we fitted demographic group and calendar period as categorical variables, and age at CD4 cell count (or VL measurement) as a continuous variable. We then adjusted for time since first clinic visit as a categorical variable, because laboratory measures may take time to improve for patients newly accessing care or changing clinics. In further models, we fitted calendar year as a continuous variable, with values at half-yearly intervals, to estimate average trends during the period. We assessed differences in trend between subgroups using tests for interaction.

Of 2606 subjects seen in the clinic from July 1, 1999, to December 31, 2004, 2386 sexually infected subjects were included. Of these subjects, 63.1% were homosexual men, 17.3% were black African/other ethnicity heterosexual women, 10.2% were black African/other ethnicity heterosexual men, 5.1% were white heterosexual women, and 4.3% were white heterosexual men. Of homosexual men, 89.4% were white; only 19.5% of nonwhite homosexual men were black African. Of nonwhite heterosexual subjects, 88.3% were black African. During the period, there were slight decreases in the proportions of homosexual men and white heterosexual women and slight increases in the proportions of black African/other ethnicity heterosexual men and women (Table 1). The median (range) age increased from 36.3 (19.7-69.8) years in July-December 1999 to 39.7 (17.8-76.4) years in July-December 2004. The median (range) time since first HIV diagnosis increased from 4.3 (0-17.1) years to 6.2 (0-22.3) years, and the median (range) time since first Royal Free clinic visit increased from 2.8 (0-15.3) years to 4.3 (0-20.1) years. The percentage of subjects seen whose first clinic visit was in the previous year decreased from 23.6% in July-December 1999 to 13.5% in July-December 2004.

Of all 2386 subjects, 1698 were receiving ART in at least 1 six-month period. The clinic prevalence of ART use increased from 61.9% in July-December 1999 to 75.5% in July-December 2004. The percentage of ART-treated subjects receiving a combination of 3 or more drugs increased slightly from 94.3% to 98.4%. The percentage of ART-treated subjects receiving fewer than 3 drugs did not differ significantly by demographic group (2.0%, 1.2%, 2.2%, 4.4%, and 1.4% for categories 1-5, respectively; P = .49, adjusted for calendar period). Furthermore, after excluding results within 6 months of a first clinic visit, the proportion of patients seen who were untreated with a CD4 cell count of 200/μL or less did not vary by demographic group (3.1%, 4.3%, 3.1%, 3.9%, and 3.4% for categories 1-5, respectively; P = .68, adjusted for calendar period).

Overall, 17.2% of the 2386 subjects were already receiving ART at their first Royal Free clinic visit. This proportion was lower for black African/other men and women (11.9% and 12.8%, respectively) than for the other demographic groups (16.5%-19.4%). The median (range) CD4 cell count at first visit was 345/μL (0/μL-1594/μL); 26.8% of subjects had a first visit CD4 cell count of 200/μL or less. This proportion decreased slightly over time, from 33.0% in July-December 1999 to 25.8% in July–December 2004 (1323 subjects with a first visit since July 1, 1999). The CD4 cell count at first visit varied considerably across the demographic groups, being highest among homosexual men and lowest among black African men. The medians values were 401/μL, 318/μL, 194/μL, 365/μL, and 264/μL for categories 1 to 5, respectively. The corresponding percentages with a CD4 cell count of 200/μL or less at first visit were 19.5%, 28.3%, 50.2%, 26.8%, and 39.0%, respectively. This pattern of demographic variation in first visit CD4 cell count was similar among untreated subjects at clinic entry.

There were substantial declines in the prevalence of immunosuppression and raised VL while receiving ART from 1999 to 2004. Among all 2386 subjects, the prevalence of a CD4 cell count of 200/μL or less decreased from 19.6% in July-December 1999 to 9.0% in July-December 2004. Among 1698 subjects receiving ART, the prevalence of a VL greater than 50 copies/mL decreased from 36.9% to 14.5%. Among 1533 subjects receiving ART for 24 weeks or longer, the prevalence of a VL greater than 50 copies/mL decreased from 31.2% to 10.1%.

The prevalence of a low CD4 cell count declined substantially over time within each demographic group among men (Figure 1A). However, prevalence varied by demographic group, being highest among black African/other heterosexual men, intermediate among white heterosexual men, and lowest among homosexual men, throughout the period. The prevalence of a low CD4 cell count decreased over time among black African/other ethnicity women, but little trend was apparent for white women (Figure 1B). Therefore, although prevalence was initially higher among black African compared with white women, there was no consistent difference throughout the period. Table 2 shows these associations adjusted for age and calendar period. Compared with black African/other heterosexual men, the risk of a low CD4 cell count was 41% lower for white heterosexual men and 61% lower for homosexual men (model 1, Table 2). Compared with black African/other women, white women had an average 37% lower risk of a low CD4 cell count (model 1, Table 2). The prevalence of a low CD4 cell count tended to increase with older age among men and women. The effect of calendar time was substantial: an approximate 60% relative reduction in prevalence from 1999 to 2004, among men and women. Table 2 also shows these associations after additional adjustment for new patient status (model 2, Table 2). There was increasing probability of a low CD4 cell count for patients with a first visit 6 to 12 months ago, and less than or equal to 6 months ago, compared with those with a visit greater than 12 months ago. Demographic differences in low CD4 cell count were attenuated after adjustment for new patient status. For men, this attenuation was minimal: significant differences remained between all demographic categories. Among women, demographic variation was considerably less marked after accounting for new patient status.

Figure 2 shows the prevalence of a VL greater than 50 copies/mL among all subjects receiving ART and among subjects receiving ART for 24 weeks or longer. Prevalence declined considerably within each demographic group among men and women, but demographic differences were much less marked than for low CD4 cell count. Adjusted risk ratios are shown in Table 3 (model 1). Demographic group was significantly associated with raised VL while receiving ART among men, mainly because of the lower prevalence among homosexual men compared with black African/other heterosexual men. Among women, the association was weak and nonsignificant. Among men and women receiving ART, there was an approximate 60% relative reduction in the prevalence of raised VL from 1999 to 2004. Older age was associated with a reduced prevalence of raised VL among men. A first clinic visit less than or equal to 6 months ago was strongly associated with raised VL while receiving ART among men and women (model 2, Table 3). Demographic differences in raised VL were further attenuated after adjustment for new patient status, although there remained a significant association among men. Demographic variation was similar when analysis was restricted to subjects taking ART for 24 weeks or longer (Table 3) and among the larger group of patients who had ever received ART (data not shown).

Risk ratios (95% confidence intervals) for annual trend from 1999 to 2004 (adjusted for age, new patient status, and demographic group) were 0.86 (0.83-0.88) for low CD4 cell count, 0.86 (0.84-0.88) for raised VL while receiving ART, and 0.83 (0.81-0.86) for raised VL among subjects receiving ART for 24 weeks or longer. This corresponds to relative reductions in prevalence per year (95% confidence interval) of 14% (12%-17%), 14% (12%-16%), and 17% (14%-19%), respectively. Among men, trends in low CD4 cell count were similar across the demographic groups (interaction test between calendar year and demographic group, P = .64). However, black African/other ethnicity women had a greater reduction in low CD4 cell count over time than white women (trends, 0.85 [95% confidence interval, 0.79-0.91] and 0.98 [95% confidence interval, 0.84-1.14], respectively; interaction test P = .046). Trends over time in raised VL among subjects receiving ART were similar across demographic groups among men and women (interaction tests, P>.50).

For all patients with a CD4 cell count or VL measurement within a 6-month interval, loss to follow-up was defined as having no CD4 cell count or VL measurement in the following 2 six-month intervals, unless death occurred during that time. The percentage lost to follow-up decreased slightly from 6.1% in July-December 1999 to 4.6% in July-December 2003. The median latest CD4 cell count was similar for subjects lost to follow-up and those remaining under follow-up (data not shown). There were small differences in the percentage lost to follow-up across the demographic groups: 4.4%, 5.5%, 5.8%, 4.6%, and 6.9% for categories 1 to 5, respectively. From 1999 to 2004, 80 deaths occurred among the 2386 subjects. The overall death rate (95% confidence interval) was 1.03 (0.82-1.29) per 100 person-years, with no clear trend over time.

The prevalence of immunosuppression and raised VL while receiving ART declined substantially from 1999 to 2004 among sexually infected patients in this London clinic. By 2004, only 14.5% of patients receiving ART and 10.1% of patients receiving ART for 24 weeks or longer had a VL greater than 50 copies/mL, indicating high levels of virological suppression achievable with currently available drugs in a routine clinic. The prevalence of immunological and virological markers varied according to demographic group among men, with homosexual men having the most favorable profile and black African heterosexual men having the least favorable profile. These demographic differences were much greater for low CD4 cell count than for raised VL while receiving ART. There was little evidence of consistent demographic variation in either outcome among women. Improvements over time were at least as strong among black African/other ethnicity men and women as among white patients.

We used a single demographic categorization that best captured the major subgroups in our clinic. We excluded a small number of patients with nonsexual HIV exposure, but results were similar for the entire clinic population. However, results from this clinic cannot necessarily be generalized to clinics in which intravenous drug use is a much more common risk group, or to countries with different systems of HIV care. Rates of loss to follow-up decreased slightly over time, but CD4 cell counts were similar among subjects lost to follow-up and those continuing under follow-up. Demographic differences in loss to follow-up were small. Therefore, loss to follow-up is unlikely to have substantially biased trends or demographic effects.

Demographic differences in the clinic prevalence of immunosuppression and viral suppression may be related to a series of factors throughout the pathway of care: stage of disease at presentation and starting ART, regimen started, initial response, likelihood of discontinuing or switching ART, and response to subsequent regimens. Studies in the United States^15- 18 have found evidence that HIV care differs according to race, with African Americans and Latinos being less likely to receive optimal treatment, but differences in CART uptake between white and nonwhite patients have been less evident in European studies,^19- 21 where health care is generally free at the point of use. In our clinic from 1999 to 2004, there was little evidence of demographic differences in use of optimal treatment.

The demographic variation in prevalence of low CD4 cell count in our clinic was much greater among men than women because of the low prevalence among homosexual men and because ethnic variation among heterosexual patients was greater among men. Similar demographic variation and trends in immunosuppression were apparent in treated and untreated subgroups (data not shown). These differences are likely to largely reflect CD4 cell count at presentation to the clinic, which was lowest for black African men and highest for homosexual men. Similar demographic variation in CD4 cell count at diagnosis or presentation has been reported from many UK studies.^6- 9 Such differences probably have a long-term effect on demographic variation in immunosuppression, first because of the time taken to achieve a CD4 cell count greater than 200/μL, if the pretreatment value was low, and second because any discontinuation or failure of therapy may result in decline of CD4 cell count to pretreatment levels.^22- 23 A number of issues affect stage of HIV at presentation for care. Many black Africans arrive in the UK as refugees or asylum seekers; their arrival may be the first opportunity to access HIV testing or treatment. Concerns about immigration status, stigma, and discrimination, or lack of information about services, may also delay access to care.^10- 11,24 In addition, women may be more likely to access HIV care than men because of routine antenatal HIV testing and via contact with other health services for their children.

The VL is more rapidly responsive to ART than CD4 cell count, and viral suppression to less than 50 copies/mL is potentially achievable within several months, even among patients with a high VL and a low CD4 cell count before ART.²⁵ Therefore, viral suppression while receiving ART is a marker of ART success that is less influenced by stage of disease at presentation. In this study, the prevalence of raised VL while receiving ART did not differ between demographic groups among women, and variation among men was much less than that seen for low CD4 cell count, suggesting that demographic differences in ART success were modest. Results from previous European studies^21,26- 29 differ with regard to the extent of demographic variation in virological and immunological response to CART. Adherence is a primary factor determining ART success, and ethnic variation in adherence has been reported from US studies.^30- 31 Adherence may be influenced by drug toxicity, pill burden, comorbidity, social circumstances, disclosure of HIV status, communication issues, and physician experience. The relative importance of these issues in different demographic groups may differ between countries. Our data suggest that, in this clinic, any adverse circumstances experienced by black African compared with white heterosexual patients did not greatly compromise adherence or ART success in the former group.

Declines from 1999 to 2004 in the clinic prevalence of low CD4 cell count and raised VL while receiving ART were considerable. Improvements in adherence, potency, and tolerability of ART, and clinical management, may have contributed to this trend. On the other hand, increasing prevalences of prolonged and multiple drug exposure and drug resistance might be expected to operate against it. In the UK Collaborative HIV Cohort Study,³ there were favorable trends in CD4 cell count of less than 200/μL and VL greater than 50 copies/mL from 1999 to 2002 among patients exposed to ART, including those with triple-class ART exposure and triple-class failure. Similarly, the present study gives little suggestion that drug resistance and multidrug failure are having an adverse impact at a clinic population level, with evidence of continued declines in low CD4 cell count and raised VL while receiving ART since 2003 (P<.05 for trend, from January 1-June 30, 2003, to July 1-December 31, 2004). It is particularly encouraging that black African patients seem to have benefited as much as other groups from improvements in HIV treatment and management.

In conclusion, the success of clinical care for HIV in the CART era has continued to increase over recent years in this routine clinic population. Although there was demographic variation in the prevalence of virological and immunological markers, improvements over time were just as impressive among black African patients as among white heterosexual and homosexual subgroups. A shift to earlier diagnosis and access of medical care among black Africans is probably the main factor that could further reduce demographic variation in HIV prognostic markers among patients under care in the UK, and this should be a major emphasis of public health policy.

Correspondence: Fiona C. Lampe, PhD, Department of Primary Care and Population Sciences, Royal Free and University College Medical School, Royal Free Campus, Rowland Hill Street, London NW3 2PF, England (f.lampe@pcps.ucl.ac.uk).

Accepted for Publication: December 14, 2006.

Author Contributions:Study concept and design: Lampe, Sabin, Youle, Johnson, and Phillips. Acquisition of data: Lampe, Smith, Madge, Kinloch–de Loes, Tyrer, Chaloner, Youle, Johnson, and Phillips. Analysis and interpretation of data: Lampe, Sabin, Johnson, and Phillips. Drafting of the manuscript: Lampe. Critical revision of the manuscript for important intellectual content: Lampe, Smith, Madge, Kinloch–de Loes, Tyrer, Sabin, Youle, Johnson, and Phillips. Statistical analysis: Lampe, Smith, Sabin, and Phillips. Study supervision: Madge and Phillips.

Financial Disclosure: None reported.

Additional Information: The following researchers are part of the Royal Free Centre for HIV Medicine. Clinical: S. Bhagani, R. Breen, P. Byrne, A. Carroll, Z. Cuthbertson, A. Dunleavy, A. M. Geretti, B. Heelan, M. Johnson, S. Kinloch–de Loes, M. Lipman, S. Madge, N. Marshall, D. Nair, G. Nebbia, B. Prinz, S. Shah, L. Swader, M. Tyrer, and M. Youle. Data Management: C. Chaloner, H. Grabowska, J. Holloway, J. Puradiredja, D. Ransom, and R. Tsintas. Biostatistics/Epidemiology: W. Bannister, L. Bansi, A. Cozzi-Lepri, Z. Fox, E. Harris, T. Hill, F. Lampe, R. Lodwick, A. Mocroft, A. Phillips, C. Sabin, and C. Smith. Laboratory: E. Amoah, C. Booth, G. Clewley, A. Garcia Diaz, B. Gregory, G. Janossy, W. Labbett, and M. Thomas.