The spread of HIV infection is most commonly measured in terms of prevalence rates, yet these rates can only provide general information on the HIV epidemic. The best way of monitoring the dynamics of HIV infection in the population, in terms of changes over time and space, is to calculate incidence rates. Although from a public health point of view describing the spread of HIV infection in the general population would be extremely useful, it is practically impossible to set up an incidence study in a population where the overall incidence rate is expected to be very low. Several subpopulations with very high prevalence and incidence rates have been identified, such as homosexual men and injecting drug users (IDU). However, even in these populations, incidence studies encounter great difficulties related to recruitment and retainment of individuals. For these reasons, a limited number of studies on HIV incidence have been published to date, and most of them have been performed among individuals at high-risk, such as homosexuals, IDU, and haemophiliacs [1-5]. Although these studies provide relevant information on the circulation of HIV in these groups, their results cannot be generalized to low-risk populations because the spread of HIV infection from these groups to the general population is not well understood. Patients with sexually transmitted diseases (STD) show a lower HIV prevalence compared with homosexuals and IDU  and may constitute an important link for the transmission of HIV from high-risk groups to the general population. Moreover, they represent a good sentinel population for monitoring changes in the HIV epidemic in that they have an increased probability of acquiring HIV through sexual contact. Despite evidence that STD are strongly associated with HIV infection , few data on HIV incidence among STD patients are available from the international literature [8-11].
An accurate description of the HIV epidemic, including recent HIV infections, cannot be performed through the analysis of notified AIDS cases because the long and highly variable incubation period only allows for an approximate reconstruction of the curve of infections that occurred in the past [12-14]. HIV prevalence data, which provide useful information on the current circulation of the infection, may be misleading if used to estimate the number of new HIV infections [9,15]. Data on HIV seroconversions, instead, provide information on the actual spread and on the trend of newly infected persons, allowing for more reliable estimates and predictions of the evolution of the epidemic. However, few studies have used incidence data because of the difficulties usually encountered in conducting longitudinal studies.
The aim of the present study was to provide data on the incidence of HIV infection among STD patients with a repeated HIV test in Italy.
A retrospective longitudinal study was conducted on individuals with a newly diagnosed STD who were tested for HIV at the time of the STD diagnosis and who had a previous documented HIV-negative test. The previous negative test might have been performed concurrently with another previous STD diagnosis, or for other reasons. These patients were diagnosed between January 1991 and December 1996 by a network of 47 public STD clinics located throughout Italy. The STD included and the diagnostic criteria used have been described elsewhere . All patients with a new STD were included, provided that they had not been diagnosed for the same disease during the previous 90 days. Patients with a recurrent STD or those coming for a follow-up visit were excluded, as were those attending the clinic only for HIV testing. Information on demographics, level of education, behavioural practices, previous STD, place of contagion, present disease, and method used for the diagnosis was collected for each patient at the moment of the STD diagnosis.
Individuals were allocated to different exposure categories which combined sexual orientation (heterosexual/homosexual) and reported use of injection drugs. The categories were as follows: heterosexual IDU, homosexual IDU, heterosexual non-IDU, homosexual non-IDU; those persons for whom information on use of injection drugs was missing were allocated to the category ‚undetermined exposure‚.
Genital ulcers included syphilis, chancroid, genital herpes, granuloma inguinale, and lymphogranuloma venereum. Genital warts, genital herpes, and molluscum were classified as ‚viral STD‚; all others were considered ‚bacterial STD‚, except Trichomonas vaginalis infection and pediculosis, which were defined as ‚other‚.
Definition of seroconverters and non-seroconverters
All eligible individuals had a documented written result of a previous HIV-negative test released by the laboratory which performed the test; this test might have been performed either in the same STD clinic where the current STD was diagnosed or in another facility. Every patient underwent an HIV test at the moment of the STD diagnosis: those who tested HIV-positive were defined as ‚seroconverters‚ and those who tested negative as ‚non-seroconverters‚. By analysing the results of the two HIV tests (the previous negative test and the test performed concurrently with the STD diagnosis), an estimate of the incidence of HIV infection in this population was calculated.
Logistic regression was used to assess the effect of sex, sexual orientation, use of injection drugs, number of partners in the previous 6 months, and previous STD as risk factors for seroconversion.
Confidence intervals (CI) at a 95% level are based on the Poisson distribution for rare events and were calculated using the ‚Confidence Interval Analysis‚ software .
Calculation of incidence
To take into account the variability of the contribution of each person in the estimation of the incidence rates, the denominator was calculated as the sum of the follow-up intervals for each of the patients and was expressed in person-years (PY). More specifically, the follow-up interval was computed calculating the time elapsed between the date (month/year) of the previous negative test and the date of the current test. When the interval between the two tests was very long, the estimated time of seroconversion was less precise. Thus, to reduce any potential bias, we excluded the relatively few individuals with an interval between the two tests that was longer than 50 months.
HIV seroconversion rates were calculated stratifying by sex, birth cohort group, exposure category, number of sexual partners in the previous 6 months, history of previous STD, and by selected STD.
Calculation of annual incidence
Nominator: the annual incidence rate was calculated under the hypothesis of uniform distribution of cases in the period between the last negative test and the first positive test; for example, a person who had a negative test for HIV in April 1990 and a positive test in May 1991, contributed as a case for 8/13 in 1990 and for 5/13 in 1991. Denominator: the specific follow-up periods were considered which, when totaled, constituted the total PY at risk; for example, a person who was tested for HIV in April 1990 and retested in May 1991, had 8 months at risk in 1990 and 5 months at risk in 1991.
Although the study began in 1991, the analysis of the annual incidence allowed us to observe the trend of seroconversions since 1988 because some participants had their previous HIV-negative test in 1988, 1989 or 1990.
During the study period, 1950 patients met all the inclusion criteria. Their median age was 31 years; 63.7% were males; 8.6% were non-Italians; 50.6% reported having had more than one sexual partner in the previous 6 months; 53.5% reported previous STD; 79.9% were heterosexuals; and 8.6% were IDU. The most frequently observed STD were genital warts (28.8%), non-gonococcal non-chlamydial cervico-vaginitis (21.2%), non-gonococcal non-chlamydial urethritis (10.2%), and genital herpes (8.6%). The median follow-up period between the two HIV tests was 13 months (range, 3-50 months).
Forty-seven individuals seroconverted, with a cumulative incidence of 2.4% and an incidence rate of 1.7 per 100 PY (95% CI, 1.2-2.2).
The comparison between the characteristics of the 47 seroconverters and the 1903 non-seroconverters showed no major differences in age, level of education, proportion of non-Italians or proportion of individuals reporting previous STD (data not shown). Genital warts was the most frequently reported STD in both groups; however, the proportion of patients with viral STD was higher among seroconverters (55.3% versus 39.7% for non-seroconverters; P<0.05). A genital ulcer was diagnosed in 19.1% of seroconverters, compared with 11.7% of non-seroconverters (P=0.12). HIV incidence rates and 95% CI are shown in Table 1.
Incidence rates calculated by birth cohort showed no definite trend, although higher incidence rates were observed for individuals born in the periods 1950-1954 and 1965-1969. The highest incidence was found among homosexual IDU (13.8 per 100 PY), whereas the lowest rate was observed among heterosexual non-IDU (0.4 per 100 PY).
The rate of seroconversion increased with the number of partners in the previous 6 months, with a maximum of 8.1 per 100 PY among patients with more than six partners. In a logistic-regression model (which included sex, sexual orientation, use of drugs, number of partners in the previous 6 months, and previous STD), homosexuality, use of injecting drugs, and having had more than six partners in the previous 6 months were significantly associated with HIV seroconversion, whereas being male and previous STD were not (data not shown).
A higher incidence was observed among patients diagnosed with an ulcerative STD compared with those with a non-ulcerative STD, and among those with a viral STD compared to other STD. The highest incidence rates by current STD were found in patients with latent syphilis, genital herpes, and genital warts. Incidences were similar among these three groups, although patients with latent syphilis showed a higher rate.
The anlyasis of seroconversions per year (Fig. 1) showed that the incidence rate decreased from 1.8 per 100 PY in 1989 to 0.9 per 100 PY in 1996. When the data were disaggregated by sexual orientation (data not shown), a decrease in the incidence was observed among heterosexuals (1 per 100 PY in 1990, 0.2 per 100 PY in 1995); among homosexuals, a decrease was observed in 1992 (from 7 per 100 PY in 1991 to 1.6 in 1992) followed by a peak of new infections in 1994 (6.8 per 100 PY in 1994, 2.6 in 1995). However, these data have to be considered carefully due to the small number of seroconversions in each subgroup.
This study analysed an open cohort, in which individuals with at-risk sexual behaviour were tested for the presence of HIV antibodies at different times and from which test results were used to estimate the frequency of seroconversion. The multivariate analysis showed that homosexuals, IDU, and persons with a high (more than six) number of sexual partners had a higher risk of seroconverting, confirming that sexual and parenteral transmission represent the most important routes of contagion.
The cumulative incidence of 2.4% found in our study is similar to the 2.8% cumulative incidence reported among STD patients in Paris  (Table 2). Incidence rates reported among STD patients by a European Study Group in different European countries ranged from 0.1 to 3.6 per 100 PY . An incidence rate of 1.7 per 100 PY was found in the present study, which is slightly lower than the 2.4-3.1 per 100 PY and the 3.1 per 100 PY reported by two studies performed in Miami [9,10] among individuals attending STD clinics. We observed a higher incidence among men compared with women, due mainly to the large contribution of seroconverted homosexual males.
The 4.6 per 100 PY incidence found in our study among homosexual non-IDU is much lower than the 14.8 per 100 PY reported by Meyer et al. in Paris  yet it is higher than the 0.7-2 per 100 PY reported by Holmberg  in the USA and closer to the 3.8 per 100 PY reported by van Hove in Europe . Differences in incidence between our study and the above mentioned studies may be due to a diverse composition of the study populations or to true differences in the circulation of the virus, but the information reported in these publications does not allow us to draw definitive conclusions. In a study performed in San Francisco  among homosexuals, individuals with a history of STD in the previous year had an HIV incidence of 5.5 per 100 PY.
The 13.8 per 100 PY incidence observed among homosexual IDU appears to be alarmingly high, although the wide CI should be taken into consideration. However, the incidences found in both homosexual groups (IDU and non-IDU) were significantly different from that of heterosexual non-IDU.
For IDU and heterosexuals our results were much more similar to those of studies conducted in other countries. Moreover, the incidence for IDU lies within the range of 0 to 4.5 per 100 PY found in Italy among IDU attending methadone clinics .
The number of partners in the previous 6 months was an independent predictor of seroconversion, even after adjusting by sexual orientation and drug use. By contrast, no association of seroconversion with previous STD was observed, suggesting that recent sexual promiscuity plays an important role in these cases.
The high incidence rates found in persons with ulcerative diseases, latent syphilis or viral STD should be interpreted with caution. Several studies have shown viral and ulcerative STD to be associated with HIV infection . These STD were assumed to have been acquired between the two HIV tests, but their long latent/subclinical phase does not allow us to understand whether the HIV infection preceded or followed the STD infection. Thus, these diseases can be considered only as indicators of possible HIV infection.
The trend of the annual incidence is consistent with the HIV incidence curves obtained by back-calculation analysis and dynamic models based on Italian AIDS cases notified as of December 1997. For heterosexuals, these curves showed a decrease in the incidence between 1987 and 1992, followed by a stabilization of new HIV infections in subsequent years; for homosexuals, they showed a decrease of infections in 1992 followed by a peak of new infecions after 1992 (M. Balducci, 1998, personal communication): both trends are very similar to the present data.
This study population constitutes an open cohort where patients are followed only for the time interval between the two HIV tests. An open cohort provides certain advantages as compared with a closed cohort, where the same individuals are observed over time: a closed cohort implies that individuals enrolled in the study are repeatedly counselled, which may lead to a change in their at-risk behaviour and, consequently, to a reduction in at-risk exposures. In these closed cohorts, the number of seroconversions generally tends to decline over time  as an effect of the change in behaviour, and the pool of very high-risk, highly-susceptible individuals shrinks over time. An open cohort is not affected by this bias because individuals are not submitted to periodical counselling sessions, thus better representing the general population of STD patients. On the other hand, our population of repeat testers may be affected by a selection bias, but it is unknown if this selection might result in an underestimation or an overestimation of the incidence rates; in any case, in most longitudinal studies, selection biases are unavoidable.
These results show that since 1988 new HIV infections have occurred among STD patients. Although a clear decrease in HIV incidence occurred between 1988 and 1994, the rate of seroconversion appears to be very high, especially in some high-risk groups. These findings suggest that HIV prevention and education programmes have been successful to some extent, although specific interventions should be addressed to individuals attending STD clinics. Moreover, STD clinics offer the possibility of reaching a population at high risk who may contribute strongly to the spread of HIV infection. The present results show that longitudinal retrospective studies performed on repeat testers who document their previous HIV-negative test can represent an easy and low-cost strategy for obtaining incidence data, especially among individuals who possibly would not comply to a prospective follow-up study.
Monitoring new HIV infections among STD patients may provide one of the sources of information on the spread of HIV that can contribute to a better understanding of the HIV epidemic at the national or regional level. Incidence rates obtained from high-risk settings, such as STD patients, can be fundamental in providing indications on the epidemiology of HIV infection, on the needs of public health, and on future research perspectives.
The authors thank M. Kanieff for linguistic revision, E. Costabile for bibliographic research, and C. Meduri for secretarial support.
1. Kingsley LA, Zhou SY, Bacellar H, et al
.: Temporal trends in human immunodeficiency virus type 1 seroconversion 1984-1989. A report from the Multicenter AIDS Cohort Study (MACS). Am J Epidemiol
2. Samuel MC, Hessol N, Shiboski S, Engel RR, Speed TP, Winkelstein Jr W: Factors associated with human immunodeficiency virus seroconversion in homosexual men in three San Francisco cohort studies. 1984-1989. J Acquir Immune Defic Syndr
3. Nelson KE, Vlahov D, Solomon L, Cohn S, Munoz A: Temporal trends of incident human immunodeficiency virus infection in a cohort of injecting drug users in Baltimore, Md. Arch Intern Med
4. Moss AR, Vranizan K, Gorter R, Bacchetti P, Watters J, Osmond D: HIV seroconversion in intravenous drug users in San Francisco, 1985-1990. AIDS
5. Kroner BL, Rosenberg PS, Aledort LM, Alvord WG, Goedert JJ: HIV-1 infection incidence among persons with hemophilia in the United States and western Europe, 1978-1990. Multicenter Hemophilia Cohort Study. J Acquir Immune Defic Syndr
6. European Centre for the Epidemiological Monitoring of AIDS: European HIV Prevalence Database Updated Version July 1995.
Saint-Maurice, France: European Centre for the Epidemiological Monitoring of AIDS; 1995.
7. Wasserheit JN: Epidemiological synergy. Sex Transm Dis
8. Meyer L, Couturier E, Brossard Y, et al
.: Trends in HIV infection among sexually transmitted disease patients in Paris. AIDS
9. Peterman TA, Zaidi AA, Wroten J: Decreasing prevalence hides a high HIV incidence: Miami. AIDS
10. Otten Jr MW, Zaidi AA, Peterman TA, Rolfs RT, Witte JJ: High rate of HIV seroconversion among patients attending urban sexually transmitted disease clinics. AIDS
11. Orskov Lindhardt B, Ulrich K, Hein Sindrup J, Avnstrop C, Fogh H, Lange Wantzin G: Seroconversion to human immunodeficiency virus (HIV) in persons attending and STD clinic in Copenhagen. Acta Derm Venereol
12. Mariotti S, Cascioli R: Sources of uncertainty in estimating HIV infection rates by back-calculation: an application to Italian data. Stat Med
13. Bacchetti P: The impact of lengthening AIDS reporting delays and uncertainty about underreporting on incidence trends and projections. J Acquir Immune Defic Syndr
14. Rosenberg PS, Gail MH: Uncertainty in estimates of HIV prevalence derived by backcalculation. Ann Epidemiol
15. Batter V, Matela B, Nsuami M, et al
.: High HIV-1 incidence in young women masked by stable overall seroprevalence among childbearing women in Kinshasa, Zaire: estimating incidence from serial seroprevalence data. AIDS
16. Suligoi B, Giuliani M, Binkin N and the STD Surveillance Working Group: The National STD Surveillance System in Italy: Results of the first year of activity. Int J STD AIDS
17. Gardner MJ, Altman DG: Statistics with Confidence. Confidence Intervals and Statistical Guidelines.
London: British Medical Journal Publishing; 1989.
18. Van Hove D, Declercq E, Stroobant A and the European Study Group: Incidence rates of HIV infection among STD patients in 17 European networks. XI International Conference on AIDS.
Vancouver, July 1996 [abstract We.C.222].
19. Holmberg SD: The estimated prevalence and incidence of HIV in 96 large US metropolitan areas. Am J Public Health
20. McFarland W, Kellogg TA, Dilley J, Katz MH: Estimation of human immunodeficiency virus (HIV) seroincidence among repeat anonymous testers in San Francisco. Am J Epidemiol
21. Rezza G, Nicolosi A, Zaccarelli M, et al
.: Understanding the dynamics of the HIV epidemic among Italian intravenous drug users: a cross-sectional versus a longitudinal approach. J Acquir Immune Defic Syndr
22. Ostrow DG: A case-control study of human immunodeficiency virus type 1 seroconversion and risk-related behaviors in the Chicago MACS/CSS Cohort, 1984-1992. Multicenter AIDS Cohort Study. Coping and Change Study. Am J Epidemiol
The STD Surveillance Working Group consists of: F. Amerio, L. Andreassi, G. Angelini, M. Aricò, A. Barba, P. Battarra, E. Beconcini, P. Biggio, F. Bonfigli, T. Cainelli, P. Calandra, E. Calzolari, M. Coppini, F. Cottoni, A. D‚Antuono, A. Di Carlo, P. Donofrio, G. Gaddoni, G. Galbiati, M. Gatti, S. Graifemberghi, B. Guerra, G. Landi, M.A. Latino, N. Licci, A. Locatelli, G. Marson, G. Moise, A. Mossini, M. Norat, F. Perino, L. Priano, E. Provenzano, P. Puiatti, A. Rafanelli, F. Ricciuti, G. Righini, C. Sabbatini, D. Simonetto, G. Tarantini, F. Urbani, L. Vittone, G. Zuccati.