Introduction
Early diagnosis of HIV infection is essential, not only to promote risk reduction behaviours among HIV-infected persons but also to allow patients to benefit from prophylactic treatment of opportunistic infections and, increasingly, from evolving antiretroviral therapeutic strategies [1].
In France, as in other industrialized countries, one goal of HIV-related public health policy since 1987 has been to encourage voluntary HIV testing through information campaigns, numerous HIV testing sites and free HIV tests. Case management and follow-up are further encouraged by immediate entitlement to free medical care for all HIV-infected people. Furthermore, in January 1993, the French government reinforced a recommendation to all physicians to offer voluntary HIV testing systematically to all women at their first antenatal visit and to couples at premarital consultations. As a result of the voluntary HIV screening policy, increasing numbers of tests have been requested by populations at low risk of HIV infection [2] and HIV testing has become routine practice for general practitioners [3]. HIV testing has also increased in high-risk populations, including homosexual men [4] and injecting drug users [5].
Despite the large number of HIV tests performed (5 million for a population of 58 million excluding blood donations in 1995) [6], national surveillance data recently showed that HIV infection had not been detected prior to AIDS diagnosis in 22% of AIDS cases [7]. Current clinical studies cannot accurately reflect access to HIV testing and to recommended treatments, since they are necessarily conducted among patients who are aware of their HIV-infection status and are receiving medical care. In order to study potential barriers to HIV testing and medical care management, we decided to conduct a retrospective study in patients at the time of AIDS diagnosis. This paper focuses on the first objective of the study, which was to describe the circumstances of the first HIV-positive test and the factors potentially associated with a late test.
Methods
Adult patients diagnosed with AIDS (1993 European AIDS case definition) [8] between July 1993 and May 1995 in all private and public hospitals of two French districts were eligible for the study. The first district was situated in a Paris suburb and had an AIDS incidence rate of 1632 per million population in 1996; the second, situated in south-western France, included the city of Bordeaux and had an incidence of 945 per million population. After informed consent, two sets of data were collected for each patient within 3 months of the AIDS diagnosis: (i) clinical, biological and therapeutic data concerning medical follow-up between the times of diagnosis of HIV infection and of AIDS were extracted from the medical file; (ii) data on socioeconomic characteristics, circumstances of first HIV-positive test, attitudes and behaviours regarding medical care and health services utilization, and social support before and after diagnosis of HIV infection were collected by a trained person, not belonging to the medical team, in a 'face-to-face' confidential interview. The time period between AIDS diagnosis and interview was on average 41 days (median, 31 days).
A minimum set of aggregated data including gender, age group, HIV transmission group, knowledge of HIV infection at AIDS diagnosis and various reasons for non-inclusion was also collected on eligible AIDS patients who could not be included.
Detailed information on circumstances of first HIV-positive test were collected: reasons for testing, previous HIV test(s) and reasons for the absence of a previous test. A specific question concerned the perception of risk at the time of the first HIV-positive test: 'At diagnosis of HIV infection, did you think you had taken a risk or risks?'. Reasons for testing were classified into five mutually exclusive categories: (a) clinical symptoms; (b) test routinely offered as part of a medical check-up; (c) HIV-risk behaviours or having an HIV-positive sex partner; (d) imprisonment; and (e) detoxification treatment.
A late test, defined as a first HIV-positive test within 6 months of AIDS diagnosis, was analysed according to the following parameters: age, gender, country of birth (France, sub-Saharan Africa, other), inclusion district, HIV-transmission group [homo-/bisexual male, injecting drug user (IDU), haemophiliac/coagulation disorder/transfusion recipient, heterosexual contact], level of education (< 12 years, ≥ 12 years), income earned or from welfare/family/illegal sources, monthly income level, and health insurance at the time of HIV diagnosis. Use of health care services before HIV diagnosis was evaluated by two questions: 'Did you have a regular general practitioner or a physician whom you consulted?' and 'Did you regularly consult a dentist?'. Behaviours regarding medical care were evaluated by the following two questions: 'When undergoing a treatment such as antibiotics, did you strictly follow the prescription (dose, length)?' and 'Did you ever consult someone practising in alternative medicine?'.
A second analysis was conducted for the same factors by comparing the group of patients who either were late testers as defined above or had a first CD4 count < 200 × 106 cells/l within 6 months of their first HIV-positive test, with patients who were not late testers for whom a first CD4 count within 6 months of the first HIV test was ≥ 200 × 106 cells/l.
Univariate associations were assessed using the χ2 test and, when appropriate, Fisher's exact test and Student's t test. Multivariate analysis of factors associated with a late test was performed by logistic regression (BMDP statistical software; University of California Berkeley, Berkeley, California, USA).
Results
Of the 629 patients who had AIDS diagnosed between July 1993 and May 1995 in the two districts, 359 (57.1%) were eligible for recruitment to this study (205 in the Bordeaux district, 154 in the Paris suburb). The reasons for non-inclusion were: clinical conditions not allowing the interview (114; 18.1%), refusal to participate (49; 7.8%), lost to follow-up within the 3 months following AIDS diagnosis (27; 4.3%), physicians neglecting to propose the study (25; 4%), not possible to conduct the interview (patient not able to speak French, handicapped or imprisoned) (10; 1.6%) and unknown reasons (45; 7.1%).
Compared with those patients not included, the study population comprised a higher proportion of men (83.6 versus 74.4%; P < 0.01) and, by transmission group, a higher proportion of homo-/bisexual men (39.6 versus 28.1%) and a lower proportion of IDU (29 versus 32.2%) and of patients infected through heterosexual contact (21.7 versus 25.2%) (P = 0.03). There was no age difference between the two groups. A higher proportion of patients was included in the Bordeaux district than in the Paris district (57.1 versus 42.9%; P = 0.06). The proportion of late testers (first HIV-positive test less than 6 months before AIDS diagnosis) did not differ between patients included and not included, either overall (19.2 versus 20.7%; P = 0.64) or within each district.
Among the 69 late testers, 53 (76.8%) had their first HIV-positive test within 1 month of AIDS diagnosis. The time period between first HIV-positive test and AIDS diagnosis ranged from zero (first HIV-positive test at AIDS diagnosis) to 11 years, with a median of 5 years. Among the late testers, the median CD4 count at HIV diagnosis was 65 × 106 cells/l (range, 0-1000 cells/l) compared with 278 × 106 cells/l (range, 0-1300 cells/l) in the other group. Among the late testers, 40.6% had Pneumocystis carinii pneumonia (PCP) at AIDS diagnosis compared with 15.9% in the other group.
Among the 69 late testers, 89.7% had an HIV test because of clinical symptoms compared with 38.9% of the other 290 patients (P < 0.001). Among these 290 patients, 32.1% had a test because of risk behaviours compared with 4.4% of the late testers, 20.5% because the test was routinely offered compared with 5.9% of the late testers, and 3.9 and 4.6% because of imprisonment and detoxification, compared with none of the late testers. At diagnosis of HIV infection, 53.6% of the late testers did not think they had been at risk compared with 39.3% of the patients from the other group (P < 0.05).
Among the late testers, 17 (24.6%) had already had a negative HIV test prior to the first HIV-positive test, compared with 48 (16.6%) of the 290 tested more than 6 months before AIDS diagnosis (P = 0.12). In patients tested previously, the median number of tests was one for the late testers and three for the other group. Three of the six reasons given for the absence of a previous HIV test (test never proposed, not feeling at risk, not feeling sick) were cited by more than 50% of the patients in both groups. Only one reason, 'prefer not to know', was significantly more frequently cited by the late testers than by the other group (42.3 versus 26.9%; P < 0.05).
The proportion of late testers was 34.6% among patients infected through heterosexual contact compared with 12.7% among homo-/bisexuals and 9.6% among IDU (P < 0.05) (Table 1). The proportion of late testers was higher in men than in women (20.7 versus 11.9%; P = 0.12), and after exclusion of homosexual, bisexual and haemophiliac men, men were significantly more likely to be late testers than women (28.7 versus 11.9%; P = 0.01). The proportion of late testers was 14.4% among patients aged less than 35 years at AIDS diagnosis compared with 23.8% among those aged 35 years or more (P = 0.02). Fifty per cent of patients born in sub-Saharan Africa were late testers, compared with 16.5% of patients born in France and 20.9% of those born in other countries (P < 0.01). The proportion of late testers did not vary with district, education, earned versus unearned income, monthly income level or health insurance (Table 1). Neither did the proportion of late testers differ regarding use of health services (general practitioner, dentist), but 28.8% of patients who had consulted a person practising alternative medicine were late testers compared with 17.4% who had not consulted such a person (P < 0.05) (Table 2).
To investigate whether factors associated with a late test differed by transmission group, an analysis was carried out separately for the three main HIV transmission groups (homo-/bisexual, IDU, heterosexual contact).
Among the 142 homo-/bisexual men, the proportion of late testers was higher among those with a high level of education (19.4%) than those with a low level of education (6.7%; P = 0.04) (Table 3). Patients who had consulted an alternative medical practitioner were more likely to be late testers (26.7%) than the other patients (8.9%). Other sociodemographic characteristics and use of health services were not associated with a late test. In a logistic regression analysis, a high level of education and having consulted an alternative medical practitioner remained independently associated with a late test; the adjusted odds ratios (aOR) were 3.1 [95% confidence interval (CI), 1.0-9.6] and 3.4 (95% CI, 1.2-10.0) respectively.
Among the 78 patients infected through heterosexual contact, the proportion of late testers was higher in men than in women (47.8 versus 15.6%; P < 0.01), higher in patients born in sub-Saharan Africa than in other patients (57.9 versus 27.1%; P = 0.01) and higher in patients with no earned income at the time of HIV diagnosis than in patients with an earned income (62.5 versus 27.4%; P < 0.01) (Table 3). Almost half of the patients who did not have a general practitioner were late testers (48.3 versus 26.5% for the other patients; P = 0.05). A similar difference was seen among patients who did not regularly consult a dentist (47.7 versus 17.6% for the other group; P < 0.01). No differences were seen according to medical care behaviour. In multivariate analysis, gender and earned income were independently associated with a late test. Men were more likely to be late testers than women (aOR, 5.6; 95% CI, 1.7-18.9) and patients with no earned income were more likely to be late testers (aOR, 5.2; 95% CI, 1.4-19.0) than those with earned income.
Of the 104 IDU, the proportion of late testers was higher in patients without health insurance than in those who had insurance (25 versus 7.6%; P = 0.09). No other demographic, socioeconomic or behavioural characteristics were associated with a late test, but the sample size was small (n = 10; data not shown).
Of the 290 patients who were not late testers, 36 had a first CD4 count < 200 × 106 cells/l within 6 months of the first HIV-positive test. When grouped with the 69 late testers and compared with the 186 patients who had a first CD4 count ≥ 200 × 106 cells/l within 6 months of the HIV test, factors associated with a late test were the same as in the previous analysis (data not shown).
Discussion
In our study population from two French districts, HIV infection was discovered at AIDS diagnosis for almost one-fifth of individuals questioned. The distribution by transmission group, the proportion of patients who first discovered they had HIV infection at AIDS diagnosis (22%) and the proportion of patients with PCP at AIDS diagnosis are comparable with those of all reported AIDS cases in France during the same period (1993-1995) [7,9]. Our population was recruited at AIDS diagnosis and thus may not be representative of all individuals infected with HIV. Indeed, late testers may be over-represented among those diagnosed with AIDS since early testers may have benefitted from drugs preventing them from developing AIDS. However, this bias should be minimal since our study was conducted before the advent of highly active antiretroviral therapy. Moreover, our results may not reflect the current behaviours of people recently infected with HIV. However, the fact that a large proportion of patients with AIDS discovered their HIV infection too late to benefit from new therapeutic strategies remains a major public health issue.
In a recent analysis of surveillance data from selected European countries, the proportion of AIDS patients diagnosed in 1996 who did not know they were infected with HIV before AIDS diagnosis (HIV diagnosis and AIDS diagnosis occurred in the same quarter) was 21.0% overall; it was 21.3% in France and ranged from 12.8% in Italy to 33.7% in the United Kingdom [10].
In our study, even in known high-risk groups (homo-/ bisexual men, IDU), there was still an unexpectedly high proportion of late testers (13 and 10% respectively), although lower than in the heterosexual group. In IDU, the low proportion of late testers could be explained by their frequent contacts with health care services. However in homo-/bisexual men, we found that a late test was associated with a high level of education and consultation of practitioners in alternative medicine, reflecting a subgroup probably well-informed about HIV risk but possibly reluctant to be tested. The potential benefits of the HIV test should perhaps become a more important part of the prevention messages aimed at homosexual men.
Perception of risk is one of the motivations for HIV testing. Overall, at the time of the first HIV-positive test, the perception of risk, although a little better in patients tested early, was quite low in both this group and late testers. A similar result was found in a cross-sectional survey in nine US states (1995-1996): among persons in known high-risk groups for HIV interviewed about reasons for delaying HIV testing, 29% gave as the reason 'unlikely to have been exposed' [11]. In our study, the only reason reported more frequently by the late testers than by the other group to explain the absence of a previous HIV test was that they 'prefer not to know'. This apparent discrepancy may imply that some subjects have the suspicion of being at risk without clearly accepting it. Better availability of clear information on the potential benefits of an early HIV test may also have a positive impact on these subjects.
The highest proportion of late testers (34.6%) was found in patients infected through heterosexual contact. Male gender was one factor strongly associated with a late test. Among AIDS cases reported between 1989 and 1992 in England and Wales, a late test (defined by an interval of 9 months between first HIV-positive test and AIDS diagnosis) was also associated with heterosexual transmission and male gender [12]. In the European study previously cited, similar results (association between late test and heterosexual transmission or male gender) were found overall and in each country studied [10]. This could be explained by the fact that women have a higher rate of utilization of health services and that HIV testing is more widely accessible to women, e.g., through the systematic offer of an HIV test at the time of pregnancy.
We found that patients born in sub-Saharan Africa were more often late testers than patients born in France. In the previously cited English study, one of the factors strongly associated with lack of awareness of HIV infection was ethnic group (non-whites) [12]. In another study among Africans living in London, 61% had a first diagnosis of HIV infection at AIDS diagnosis [13]. However, in our study the association with sub-Saharan African origin did not remain significant in multivariate analysis after taking into account socio-economic status. Association between late test and poor economic status may result from a lack of medical information or poor use of medical services, indepen-dently of geographical origin.
Overall, for almost half of the patients, diagnosis of HIV infection occurred in the context of HIV-related disease or other illnesses. Among patients having their first HIV-positive test within 6 months of AIDS diagnosis, it is understandable that HIV testing was for medical reasons. However, it is striking that clinical symptoms were also the main reason (39%) for HIV testing among the patients tested more than 6 months before AIDS diagnosis. This could be explained in part by a long interval between the HIV infection and the test, as suggested by the low median CD4 count at HIV diagnosis. Illness was also the most common reason for testing among US AIDS patients diagnosed between 1990 and 1992 (58%) in a study reported by Wortley et al [14]. Physicians should take advantage of the medical consultation to discuss risk perception and behaviours and to offer an HIV test to their patients, even in the absence of symptoms suggesting underlying HIV infection, particularly to those who do not have frequent contacts with the health care system.
Our results suggest that efforts to diagnose HIV infection as early as possible should be made both on the demand side, by increasing the perception of HIV risk and the potential benefit of medical care, and on the supply side, by decreasing the missed opportunities for testing during medical visits. In view of the new antiretroviral therapeutic strategies available, the benefits of early HIV testing may appear much clearer both to the individual and to the physician.
Acknowledgements
We thank all participants: C. Marimoutou, M. Decoin (Groupe d'Epidémiologie Clinique du SIDA en Aquitaine, Bordeaux); J.M. Ragnaud, M. Buisson, S. Sire, M. Dupon, M.C. Paty, M. Simon, J.C. Martin (Hôpital Pellegrin Tripode, Bordeaux); D. Lacoste, N. Bernard, E. Monlun, J. Constans (Hôpital Saint-André, Bordeaux); J.L. Pellegrin, C. Lasseur, P. Rispal, M.S. Doutre (Hôpital Haut-Lévèque, Pessac); P. Loste (Hôpital de Dax); H. Berthé, J.C. Boulard, T. Hanslik, C. Dupont, E. Rouveix, V. Giraud (Hôpital Ambroise Paré, Boulogne-Billancourt); C. Goujard, A. Levy, M. Môle, M.T. Ranoux (Hôpital Antoine Béclère, Clamart); M.H. Pichot (Hôpital Beaujon, Clichy); C. Chandemerle, F. Meier, M. Bloch, I. Cahitte, F. Cordonnier (Hôpital Louis Mourier, Colombes); M. Risbourg, M. Saillour, P. de Truchis, C. Perronne (Hôpital Raymond Poincaré, Garches); G. Force (Hôpital Notre-Dame du Perpétuel Secours, Levallois-Perret); B. Deborne (Hôpital Max Fourestier, Nanterre); D. Zucman (Hôpital Foch, Suresnes); V. Tassain, S. Haeghebaert (European Centre for the Epidemiological Monitoring of AIDS, Saint-Maurice) and A.M. Downs for reviewing the English manuscript.
References
1. Carpenter CCJ, Fischl MA, Hammer SM, et al.: Antiretroviral therapy for HIV infection in 1997: updated recommendations of the International AIDS Society-USA Panel. JAMA 1997, 277:1962-1969.
2. Spira A, Bajos N and the Analyse Des Comportements Sexuels En France Group: Sexual Behaviour and AIDS. Edited by A. Spira Aldershot: Avebury; 1994.
3. Moatti JP, Souville M, Obadia Y, et al.: Ethical dilemmas in care for HIV infection among French general practitioners. Health Policy 1995, 31:197-210.
4. Schiltz MA, Adam P: Le test de dépistage du VIH: diffusion parmi les homo et bisexuels français. In Agence Nationale de Recherches sur le SIDA. Le dépistage du VIH-Politiques et pratiques. Paris: Sciences Sociales et SIDA; 1996.
5. Six C, Hamers F, Ancelle-Park R, Brunet JB: L'infection à VIH chez les résidents des centres de soins spécialisés pour toxicomanes avec hébergement. Bull Epidemiol Hebdom 1996, 11:53-54.
6. Chauffert O, Laurent E, Lorente E, Goulet V, Les Biologistes du Réseau National du VIH: Activité de dépistage du VIH en France de 1989 à 1995. Bull Epidemiol Hebdom 1996, 39:172-173.
7. Cazein F, Lot F, Pillonel J, Pinget R, Laporte A: Connaissance du statut sérologique avant le diagnostic de SIDA. Bull Epidemiol Hebdom 1995, 46:202-203.
8. Ancelle-Park R: Expanded European AIDS case definition. Lancet 1993, 341:441.
9. Réseau National de Santé Publique: Surveillance du SIDA en France: situation au 31 décembre 1996. Bull Epidemiol Hebdom 1997, 11:46-49.
10. European Centre for the Epidemiological Monitoring of AIDS: Quarterly Report No. 52. Saint-Maurice: WHO-EC Collaborating Centre on AIDS; 1996.
11. Proceedings of The Consultation on the Future of HIV/AIDS Surveillance; 1997 May 21-22. Atlanta: CDC National AIDS Clearing House; 1997.
12. Porter K, Wall P, Evans B: Factors associated with lack of awareness of HIV infection before diagnosis of AIDS. BMJ 1993, 307:20-23.
13. O'Farrell N, Lau R, Yoganathan K, Bradbeer C, Griffin GE, Pozniak AL: AIDS in Africans living in London. Genitourin Med 1995, 71:358-362.
14. Wortley PM, Chu SY, Diaz T, et al.: HIV testing patterns: where, why, and when were persons with AIDS tested for HIV? AIDS 1995, 9:487-492.
© Lippincott-Raven Publishers.