We used a two-source capture–recapture method to estimate the number of patients diagnosed at the time of primary HIV infection in France between 1999 and 2002. The sources were the French PRIMO cohort and the French Hospital Database on HIV. The estimated number of patients was 325 per year, which represents only 5% (approximately 6000 cases) of all new cases diagnosed each year and only 8% of all new infections (approximately 4000 cases).
aInserm U720, Université Pierre et Marie Curie-Paris 6, Paris, France
bInserm/INED U569, Université Paris XI, Le Kremlin-Bicêtre, France
cHôpital Bichat-Claude Bernard, Paris, France
dHôpital de la Conception, Marseille, France
eHôpital Saint-Louis, Paris, France.
Received 3 April, 2006
Revised 30 August, 2006
Accepted 18 September, 2006
The proportion of incident cases of HIV infection that are diagnosed during the primary phase is not known, but would be a useful indicator of the capacity of healthcare systems to identify infected patients rapidly. In addition, most studies monitoring the transmission of resistant strains of HIV are based on cases diagnosed at the time of primary infection [1–3], but it is not known whether these patients are representative of the overall infected population.
The aims of this study were to estimate the yearly number of patients diagnosed at the time of primary HIV-1 infection in France between 1999 and 2002, and to compare the characteristics of these patients with those of other patients diagnosed in the chronic stage during the same period.
The capture–recapture method consists of matching two sources of case notification in order to determine the number of common cases, and then to estimate the total number of cases (N) in the population.
The French ANRS-CO6 PRIMO cohort enrolls patients diagnosed during or soon after primary HIV-1 infection (a negative or indeterminate HIV antibody assay associated with a positive antigenemia or plasma HIV RNA, or an incomplete Western blot with an absence of antibodies to pol proteins, or an initially negative test for HIV antibody followed within 6 months by a positive HIV serology), in 77 participating hospital units . The date of infection was estimated as the date of symptom onset minus 15 days, or the date of an incomplete Western blot minus one month, or the midpoint between a negative and a positive enzyme-linked immunosorbent assay. The interval between the estimated date of infection and enrollment could not exceed 6 months.
The French Hospital Database on HIV (FHDH, ANRS-CO4) enrolls HIV-1-infected patients managed in 62 participating hospitals . For patients with available information, the date of infection is estimated as the presumed date of infection, or the date of primary infection, or the midpoint between a negative and a positive test for HIV-1.
FHDH and PRIMO both collect data anonymously. Patients are identified by an anonymous code generated from the last name, given name, and the day and month of birth, using the same algorithm; the anonymous code can thus be used as a common identifier for record linkage . For the purposes of the present study, matches were first sought on the basis of the anonymous code. In order to take into account possible errors in data entry, leading to the same patient having a different anonymous code in the two databases, matches in the remaining records were sought on the basis of sex, the month and year of birth, and the enrollment centre. The total number of patients diagnosed at the time of primary infection, N, was calculated using the Chapman estimator , and an approximate unbiased estimate of the variance of N was used . These estimates are valid if the following assumptions are satisfied: (i) all cases reported in each system are true cases; (ii) the matching procedure identifies all true matches and only true matches; and (iii) the data sources are independent.
We then compared the characteristics (sex, transmission group, and country of birth) of the patients diagnosed with chronic HIV-1 infection and enrolled in the FHDH less than one year after diagnosis, between 1999 and 2002, with those of patients diagnosed with primary infection.
Between 1999 and 2002, 276 patients were enrolled in the PRIMO cohort, and 623 of the patients from the FHDH had a documented date of HIV-1 infection and were enrolled less than 6 months after this date. A total of 107 patients were matched by the anonymous code, and a further 25 cross-matches were validated. The estimated total number of patients diagnosed at the time of primary HIV-1 infection during the period 1999–2002 was 1299 [95%confidenceinterval(CI)1158–1439], which is approximately 325 cases per year (95% CI 289–360).
Thirty-eight patients in the FHDH had the same anonymous code as a patient in the PRIMO cohort, but the date of HIV infection was not recorded for these patients in the FHDH. If we had taken these extra 38 matched cases into account, the estimated total number of patients diagnosed at the time of primary infection would have been approximately 268 cases per year (95% CI 246–289).
Between 1999 and 2002, the characteristics of the patients presenting with primary HIV infection differed from those of patients presenting with chronic infection during the same period (Table 1). Male homosexual and bisexual individuals were more likely to be diagnosed at the time of primary infection, as were young patients (under 30 years) and patients born elsewhere than in sub-Saharan Africa.
During the period 1999–2002, the estimated number of patients diagnosed in France at the time of primary HIV-1 infection was approximately 325 per year. As concerns the application of the capture–recapture method, the source of greatest bias for the estimates is deviation from the hypothesis of independence, but this condition cannot be tested when only two sources are involved . However, the fact that 38 patients in the PRIMO cohort were also enrolled in the FHDH but were not recorded as having primary infection in this latter database shows that the two sources have at worst only a small positive dependency in common centres.
Studies of the transmission of drug-resistant HIV variants use data on patients diagnosed during the primary infection. As previously underlined , a major issue when interpreting the results of such studies is the lack of exhaustiveness. In France, it is estimated that at least 4000 individuals are newly infected by HIV each year , but that only a small proportion are diagnosed during the primary infection, less than 10% according to our results. Another issue is whether patients diagnosed at the time of primary infection are representative, in terms of HIV resistance, of all patients infected during the same period. We found here that patients diagnosed at the time of primary infection were younger, were more often homo/bisexual men, and were more likely to originate from France than from sub-Saharan Africa than patients diagnosed during the chronic phase. These differences are unlikely to be only explained by a recent shift in the HIV epidemic, but rather by delayed diagnosis in some subpopulations. If transmission of resistance is indeed associated with age, the transmission group or the country of origin, this would imply that patients enrolled in resistance studies during the primary infection are not representative of all newly infected patients. This should be kept in mind when interpreting studies of the transmission of resistant viruses.
Using a capture–recapture method, we estimated that the number of patients diagnosed during or soon after primary HIV infection was approximately 325 per year in France during the period 1999–2002. This represents only 5% of all new diagnoses of HIV infection in France (approximately 6000/year) and only 8% of the estimated number of new infections occurring each year . Diagnosis during primary infection, by permitting early treatment, may improve the individual prognosis and also reduce the potential for HIV dissemination during this early highly infectious phase. Earlier diagnosis in certain subgroups is likely to be the result of different attitudes towards health, healthcare consumption, voluntary HIV testing, and better self-ascertainment of at-risk behaviours or symptoms of primary HIV infection. This implies the need for programmes promoting early diagnosis and specifically targeting certain groups.
The full list of investigators for the Primo Study Group can be found at http://u569.kb.inserm.fr/COLLAD/IDEX.htm. The full list of investigators for the Clinical Epidemiology Group of the French Hospital Database on HIV can be found on http://www.ccdc.fr.
1. Wensing AMJ, van de Vijver DA, Angarano G, Åsjö B, Balotta C, Boeri E, et al
. Prevalence of drug-resistant HIV-1 variants in untreated individuals in Europe: implications for clinical management. J Infect Dis 2005; 192:958–966.
2. Little SJ, Holte S, Routy JP, Daar ES, Markowitz M, Collier AC, et al
. Antiretroviral-drug resistance among patients recently infected with HIV. N Engl J Med 2002; 347:385–394.
3. Chaix ML, Descamps D, Harzic M, Schneider V, Deveau C, Tamalet C, et al
. Stable prevalence of genotypic drug resistance mutations but increase in non-B virus among patients with primary HIV-1 infection in France. AIDS 2003; 17:2635–2643.
4. Desquilbet L, Deveau C, Goujard C, Hubert J-B, Derouineau J, Meyer L, et al
. Increase in at-risk sexual behaviour among HIV-1-infected patients followed in the French PRIMO Cohort. AIDS 2002; 16:2329–2333.
5. Tassie JM, Gasnault J, Bentata M, Deloumeaux J, Boue F, Billaud E, et al
. Survival improvement of AIDS-related progressive multifocal leukoencephalopathy in the era of protease inhibitors. AIDS 1999; 13:1881–1887.
6. Thirion X, Sambuc R, San Marco JL. Anonymity in epidemielogical surveys: study and initiation of a new method. Rev Epidemiol Sante Publique 1988; 36:36–42.
7. Chapman DG. Some properties of the hypergeometric distribution with applications to zoological sample censuses
. Berkeley: University of California Publications in Statistics; 1951. pp. 131–160.
8. Seber GAF. The effect of trap response on tag recapture estimates. Biometrics 1970; 26:13–22.
9. Bernillon P, Lievre L, Pillonel J, Laporte A, Costagliola D, the Clinical Epidemiology Group from Centres d'Information et de Soins de l'Immunodeficience Humaine. Record-linkage between two anonymous databases for a capture–recapture estimation of underreporting of AIDS cases: France 1990–1993. Int J Epidemiol 2000; 29:168–174.
10. Desenclos JC, Costagliola D, Commenges D, Lellouch J, les membres de l'Action Coordonnée 23 de l'ANRS “Dynamique de l’épidémie”. Prevalence of HIV seropositivity in France [in French]. Bull Epidemiol Hebd 2005; 11:41–44.