Since the advent of AIDS, equality and nondiscrimination in access to prevention and treatment have been major issues because so high a percentage of patients belong to stigmatized or disadvantaged groups, such as drug users, ethnic minorities, immigrants, or women. Disadvantaged groups have been shown to have less access to zidovudine treatment1 and pneumocystis carinii pneumonia (PCP) prophylaxis2; they also start highly active antiretroviral therapy (HAART) later.3 Aside from the barriers to health care due to patients' financial situation and educational level, physicians' anticipation that patient compliance will be poor4,5 has been shown to explain delayed initiation of HAART. Moreover, several US studies, conducted mainly before the advent of HAART, identified physician experience as a determinant of progression to AIDS and survival.6-10
Because a very high level of adherence is required to fully benefit from HAART, a multidisciplinary approach to HIV care is needed to support patients and help them deal with the practical, social, and emotional dimensions of living with HIV.11,12 Hospitals and health-care centers should provide comprehensive HIV care with a range of medical and social services to achieve the best possible outcomes. Rapid progress in clinical research necessitates the regular updating of treatment recommendations. Physicians should update their practice regarding treatment so that all patients can benefit from the most recent scientific advances. Such rapid changes might vary according to physician characteristics, type of setting, and size of HIV caseload.
The purpose of this article is to describe the heterogeneity of the hospital HIV treatment provided in France and its effect on treatment outcome.
MATERIALS AND METHODS
This analysis used data from the 2003 ANRS-EN12-VESPA study, a large nationally representative 2-stage cross-sectional survey, aimed at studying the social situation of HIV-infected persons in France. The study design has been detailed elsewhere.13 Eighty-five hospitals providing HIV care were randomly selected among the 143 French hospitals with a HIV caseload greater than 60 patients, stratifying on geographic location and size of HIV caseload. As some large hospitals had several departments serving patients with HIV, the 85 hospitals were composed of 102 departments providing HIV care. Doctors in each of these departments randomly recruited a sample of outpatients (representing about 5% of their caseload) among those with the following eligibility criteria: having been diagnosed with HIV-1 infection for at least 6 months, aged 18 or older, and, if not a French citizen, having lived in France for at least 6 months. Patients who understood French very poorly were excluded. The study design and informed consent procedures complied with the ethical and statutory requirements of the French Data Protection Authority (Commission Nationale Informatique et Libertés).
Information was collected for each hospital department and for each individual patient.
In each department, a clinical research assistant completed an information sheet on the current HIV patient caseload, department capacity (number of physicians, inpatient beds, day hospital beds, and weekly outpatient appointments), participation in research activities (cohort studies and clinical trials), onsite provision of a range of specialized consultations and care (nutrition, hepatology, dermatology, drug treatment liaison team, psychiatry, etc), promotion of treatment adherence (group or individual education, leaflets, pillbox provision etc.), and the onsite presence of people living with HIV/AIDS (PLWHA) groups.
Information was collected on the following characteristics for all eligible patients: sex, age, employment status, mode of transmission, and last known CD4 cell count and viral load. These data allowed for the assessment of participation bias.
Participating patients answered a questionnaire administered in a face-to-face interview. Detailed information was collected on sociodemographic characteristics and on a wide range of social conditions (health-care access and use, health behavior, occupational status, income, housing, social support, HIV disclosure, discriminations encountered, sexual activity, and reproductive life). Physicians recorded for each patient the mode of HIV acquisition, disease characteristics, and treatment information (date of diagnosis, date of HAART initiation, clinical stage, CD4 count and viral load at diagnosis, at HAART initiation, and last known values).
Variables of Interest
The heterogeneity of the departments delivering HIV care was studied for adherence and various measures of treatment outcome.
Adherence to HAART was assessed with a dichotomous outcome (high versus moderate or poor adherence) validated in previous cohort studies.14,15 First, participants were asked about their compliance with HAART in the prior 7 days. They had to choose among the following items, each corresponding to a level of adherence: “I have followed scrupulously my treatment” (high adherence), “I have followed my treatment despite some small lapses” (moderate adherence), “I have frequently modified doses or time schedule,” “I have almost never complied with prescription instructions,” and “I have stopped treatment' (poor adherence). Then, participants were asked whether or not they had missed at least 1 dose during the previous weekend, whether they had taken the whole daily dose in only 1 take in the previous 7 days, or whether they had failed to respect the time schedule for at least 1 dose during the same period. Participants previously ranked as highly adherents but who had answered “yes” to 1 of these 3 questions were reallocated to the “moderate adherence” category.
Treatment outcome was assessed using 4 different indicators, 3 defining failures and 1 denoting success:
- Immunological response was defined as the absence of an increase of ≥100 CD4 cells per cubic millimeter between HAART initiation and data collection. This threshold was chosen on the basis of the review by May et al,16 showing a median 100-cell increase in CD4 cell count in the first 6 months on HAART, which was here the minimum duration;
- Virological failure was defined as having detectable HIV RNA (<400 copies/mL) at data collection;
- Immunovirological failure was defined as the combination of a CD4 cell count ≤200 cells per cubic millimeter and detectable HIV RNA at data collection;
- Lastly, immunovirological success was defined as the combination of a CD4 cell count ≥500 cells per cubic millimeter and undetectable HIV RNA at data collection.17
Population of Interest
The VESPA study included 2932 participants from 102 different departments. For the present analysis, we included only the 699 participants (of 97 different departments) who had been diagnosed in 1996 or later and had received HAART for at least 6 months at the time of the study. The medical history of patients diagnosed before and after the HAART era is indeed too different to be pooled in an analysis on this topic. Furthermore, the much higher mortality before 1996 indicates much greater selection bias among the patients diagnosed before that year. A minimum of 6 months is assumed necessary before the treatment can be judged effective and thus classified as a success or failure.
For each of the 5 outcomes, the existence of a center effect was assessed by testing the department as a random variable in a binary mixed model, without any other covariate. If this random variable seemed to be significant, the next step of the analysis was to include in the models covariates measured at the individual level (age, sex, sexual orientation, immigration status, education, hepatitis C virus coinfection, time since HAART initiation, CD4 cell count at HAART initiation, addictive behaviour, current trial participation, side effects, and adherence, when relevant). If the random variable was no longer significant, then the previously found effect was only due to differences in population structure, and a given patient had the same chance of treatment outcome regardless of facility.
Because including centers with very few subjects could result in a lack of robustness, we performed a sensitivity analysis by restricting the analysis to departments with at least 5 participating patients and then with at least 10 participants. All statistical analyses were performed using SAS 9.1.
Characteristics of Departments Regarding Service Provision
Among the 102 departments, the median size of the HIV caseload at data collection was 400 [interquartile range (IQR: 230-750)]. The median number of physicians providing HIV care per department was 3 (IQR: 2-6), and the median number of outpatient appointments per week was 7 (IQR: 5-10). Half of these departments specialized in infectious diseases, a quarter in internal medicine, and the others in hematology, dermatology, pneumonology, or hepatology. Internal medicine departments tended to provide a wider range of consultant services than the others. These most often included nutrition (85%), followed by dermatology (55%), and psychiatry (53%). More rarely offered were hepatology (36%), addiction liaison service (39%), endocrinology (32%), neurology (29%), and gynaecology (28%). Individual sessions on adherence education were provided in 83% of the departments, and PLWHA organizations offered an onsite presence in 41%. Most departments participated in clinical research, such as clinical trials (83%) or cohort studies (79%).
Characteristics of Participants
The median age of these patients was 42 years (IQR: 35-51); men accounted for 75% of the sample; 39% self-identified as homosexual or bisexual men, 6% were injection drug users (IDUs) (4% men, 2% women), 29% were French-born heterosexuals with no IDU history (19% men, 10% women), and 26% were immigrant heterosexuals with no IDU history (13% men and women). Most of the heterosexual immigrants came from sub-Saharan Africa (72% of the men and 86% of the women). Late diagnosis (defined either as symptoms of clinical AIDS or CD4 cell count <200 cells per cubic millimeter during the year after diagnosis) was observed in 43% of this population diagnosed in or after 1996. At HAART initiation, 49% had a CD4 cell count <200 cells per milliliter.
Structure of Departments' Population
As shown in Table 1, these different categories of participants were not uniformly distributed among the hospital departments: the structure of the patient population seemed quite heterogeneous across centers. The median proportion of immigrants was 20%, with an IQR from 0% to 40%. Although men constituted a majority of patients, with a median proportion of 75%, this rate varied substantially (IQR: 60%-90%). The median proportion of patients younger than 40 was 43% (IQR: 20%-50%), and the median proportion of patients with late diagnosis was 40% (IQR: 29%-57%). Few departments treated patients with an IDU history: the median proportion of these patients was 0 and the upper quartile 12.5%.
For the analysis population, the mean number of participants per department was 7.2 and the median 6 (IQR: 1-10).
Adherence and Treatment Outcomes
Table 2 shows the distribution of adherence and treatment outcomes measured for each hospital department and the related tests for center effects. The mean percentage of highly adherent participants was 63.3%. The heterogeneity between departments was rather large: the IQR varied from 50% to 80%, with a median adherence rate of 62%. However, the existence of a center effect was rejected when we tested the department as a random variable (P = 0.275).
The overall rate of immunological failure was 22.6%, with a departments' median of 20% and an IQR of 0% to 40%. No center effect existed here either (P = 0.215).
The overall rate of virological failure was 12.5%, with a median rate of 7.7%. Only a quarter of the departments had rates above 20%, and 44.3% had no virological failures. Again, no center effect was detected (P = 0.098).
The mean immunovirological failure was 6.1%. Three quarters of the departments had a failure rate under 12.5%, and 61% had no participants in immunovirological failure. The existence of a center effect was rejected with a P value of 0.181.
The criteria defining immunovirological treatment success, that is, a CD4 count over 500 cells per milliliter and undetectable HIV RNA, were met by 40.2% of patients. The median department success rate was 36.3%: a quarter had rates only below 13.3% and another quarter had rates above 50%. The random variable seemed to be significant (P = 0.026) for this indicator. A center effect thus existed, at least when tested for the entire population analyzed. Nevertheless, the random term was not significant when a covariate measured on the individual level was introduced, such as the patient CD4 cell count ≤200 cells per cubic millimeter at HAART initiation (P = 0.094) or the patients' immigration status (P = 0.200).
Our sensitivity analysis consisted in rerunning these models with restricting the analysis population first to the 601 patients of the 60 departments that had at least 5 participating patients and second to the 378 patients of the 26 departments that had at least 10 participating patients.
For adherence, as for the 3 first indicators of treatment failure, the absence of a center effect was confirmed. Concerning immunovirological success, the random variable was still significant in the first restriction (P = 0.030) and almost still significant in the second restriction to departments with at least 10 patients (P = 0.053). Adjusting on covariates measured on the individual level returned as previously a nonsignificant random term.
This study used data from a large random sample of hospitals providing HIV care, both representative at the national level and very diverse in terms of size, region, localization in either metropolitan areas or medium-sized cities, activity, and specialization. We observed a significant variation in the size of the HIV caseload, the provision of consultant services offered onsite, patient education to increase adherence, and support from PLWHA organizations. Nonetheless no center effect was identified, either for adherence to treatment or in HAART outcomes: in France, regardless of where HIV patients are treated for their infection, treatment success or failure depends upon individual determinants. In a previous study of this population, we found that aside from the well-established determinants (such as age, time on HAART, history of interruption in HIV care, suboptimal adherence, hepatitis C virus coinfection, CD4 cell count and HIV RNA level, and AIDS at HAART initiation), immigrant status and continuing intravenous drug use were associated with a greater risk of treatment failure.13
Because of the cross-sectional design of the VESPA study, treatment failure might have been underestimated: deaths that occurred between diagnosis and data collection were not taken into account in the calculation of failure rate. In the HAART era, mortality remains much higher among patients diagnosed with advanced infection than among patients diagnosed before reaching the criteria for treatment initiation.18 In the former group, the physician's experience may play a critical role.4,6,8-10
Our study considered only hospitals caring for a substantial number (60 or more) of persons with HIV. Patients followed in smaller HIV-oriented services are not included although they might have different HIV outcomes. However, the hospitals included provided care for an estimated 90% of the HIV patient population followed in hospitals.
Random sampling ensured the representativeness of each hospital department caseload, but differential participation was related to individual patient characteristics, including lower CD4 cell counts. Participation bias might thus explain the absence of center effects, especially if patients with poor treatment outcomes had tended to participate less in the hospitals with the poorest outcome rates. As last known CD4 cell count and viral load were available for every eligible patient, we were able to verify that this was not the case, using a mixed binary model which explained the participation of eligible patients by their health status and the treatment outcome rate of their hospital department.
Despite a low number of participants in some departments, we believe that the absence of an observed center effect is unlikely to be due to lack of power. The assumptions of the random coefficient model we use seem appropriate here, and neither the estimates nor P values changed substantially when the analysis was restricted to the 2 subsamples with the hospital departments reaching either 5 or 10 patients with at least 6 months on HAART.
The data were collected in 2003 and could perhaps not reflect the current situation. The treatment practices and the structure of the patient population have evolved. However, because the context of high innovation and frequent improvements in treatment practices is the same today as it was from 1996 to 2003, it is plausible that the observed homogeneity still applies today.
With these limitations, adherence and response to HAART seemed not different across treatment sites despite the diversity in medical care provision. These findings contrast with studies in the United States during the 1980s and early 1990s.1-4 The very high effectiveness of the current combinations of antiretroviral drugs, when accessible to all who need them, may well be acknowledged as the principal factor for the observed results. The frequent updating of treatment guidelines (at least every 2 year) may also contribute to the improvement of treatment practices. Moreover, in France, HIV care has been coordinated at the local level since 1988 to facilitate scientific communication, availability of the range of treatment services and care coordination. Dissemination of recent scientific findings among HIV specialists may be facilitated by their participation in clinical research: 83% of the departments included participate in clinical trials and 79% in cohort studies. Such participation provides the opportunity of frequent contacts between HIV experts at national and international levels.
Two decades after the advent of AIDS, at country level, HIV care has become a lay medical specialty. The HIV medical community has remained relatively small, although becoming closely networked. The pioneers have not yet started to retire.19 These factors, combined with the experience acquired, may account for the relatively homogeneity of good care that could explain the results we observed in the VESPA study. International goals for 2010 include universal antiretroviral treatment for all who need it,20 including in countries with a scarcity of skilled human resources. Studies show that when treatment is provided at no charge, its efficacy reach the same levels in southern countries as it does in northern countries.21 Accessibility will require that antiretroviral treatment be provided through medical facilities in the community. Long-term networking and training of HIV care providers to improve and maintain high-standard skills should be planned to ensure the equal outcomes that should accompany truly equal access to effective treatment.
The authors would like to thank all the patients, physicians, and nurses who have participated in the ANRS-EN12-VESPA Study.
1. Stein MD, Piette J, Mor V, et al. Differences in access to zidovudine (AZT) among symptomatic HIV-infected persons. J Gen Intern Med
2. Easterbrook PJ, Keruly JC, Creagh-Kirk T, et al. Racial and ethnic differences in outcome in zidovudine-treated patients with advanced HIV disease. Zidovudine Epidemiology Study Group. JAMA
3. Wilson IB, Landon BE, Ding L, et al. A national study of the relationship of care site HIV specialization to early adoption of highly active antiretroviral therapy. Med Care
4. Wong MD, Cunningham WE, Shapiro MF, et al. Disparities in HIV treatment and physician attitudes about delaying protease inhibitors for nonadherent patients. J Gen Intern Med
5. Escaffre N, Morin M, Bouhnik AD, et al. Injecting drug users' adherence to HIV antiretroviral treatments: physicians' beliefs. AIDS Care
6. Kitahata MM, Koepsell TD, Deyo RA, et al. Physicians' experience with the acquired immunodeficiency syndrome as a factor in patients' survival. N Engl J Med
7. Kitahata MM, Van Rompaey SE, Dillingham PW, et al. Primary care delivery is associated with greater physician experience and improved survival among persons with AIDS. J Gen Intern Med
8. Laine C, Markson LE, McKee LJ, et al. The relationship of clinic experience with advanced HIV and survival of women with AIDS. AIDS
9. Cunningham WE, Tisnado DM, Lui HH, et al. The effect of hospital experience on mortality among patients hospitalized with acquired immunodeficiency syndrome in California. Am J Med
10. Bennett CL, Adams J, Gertler P, et al. Relation between hospital experience and in-hospital mortality for patients with AIDS-related Pneumocystis carinii pneumonia: experience from 3,126 cases in New York City in 1987. J Acquir Immune Defic Syndr
11. Sherer R, Stieglitz K, Narra J, et al. HIV multidisciplinary teams work: support services improve access to and retention in HIV primary care. AIDS Care
. 2002;14(Suppl 1):S31-S44.
12. Hammer SM, Saag MS, Schechter M, et al. Treatment for adult HIV infection
: 2006 recommendations of the International AIDS Society-USA panel. JAMA
13. Dray-Spira R, Spire B, Heard I, et al. Heterogeneous response to HAART
across a diverse population of people living with HIV: results from the ANRS-EN12-VESPA Study. AIDS
. 2007;21(Suppl 1):S5-S12.
14. Carrieri MP, Raffi F, Lewden C, et al. Impact of early versus late adherence to highly active antiretroviral therapy on immunovirological response: a 3-year follow-up study. Antivir Ther
15. Peretti-Watel P, Spire B, Schiltz MA, et al. Vulnerability, unsafe sex and non-adherence to HAART
: evidence from a large sample of French HIV/AIDS outpatients. Soc Sci Med
16. May MT, Sterne JA, Costagliola D, et al. HIV treatment response and prognosis in Europe and North America in the first decade of highly active antiretroviral therapy: a collaborative analysis. Lancet
17. Lewden C, Chene G, Morlat P, et al. HIV-infected adults with a CD4 cell count greater than 500 cells/mm3
on long-term combination antiretroviral therapy reach same mortality rates as the general population. J Acquir Immune Defic Syndr
18. Mauskopf J, Kitahata M, Kauf T, et al. HIV antiretroviral treatment: early versus later. J Acquir Immune Defic Syndr
19. Lert F, Chastang JF, Castano I. Psychological stress among hospital doctors caring for HIV patients in the late nineties. AIDS Care
20. WHO, UNAIDS, UNICEF. Towards Universal Access: Scaling up Priority HIV/AIDS Interventions in the Health Sector: Progress Report. April 2007. World Health Organization; Geneva, Switzerland.
21. Braitstein P, Brinkhof MW, Dabis F, et al. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet