Survival with HIV-1 varies with age at infection, with an overall median survival of 11 years in developed countries . Recent data indicate that the survival in Africa may be similar . The other human immunodeficiency virus, HIV-2, is less well studied, but the available evidence indicates a longer survival for patients with HIV-2 infection [3,4].
Two community-based studies and one occupational cohort study from Guinea-Bissau found that the mortality rate of HIV-2-infected patients was between 2.3 and 6.6 times higher than that of seronegative patients [5–7]. This compares with a mortality rate ratio of 11.9 in men and 13.9 in women, comparing HIV-1 to seronegative individuals in a community-based study in rural Uganda .
In West Africa, where both viruses circulate, it is not uncommon for one patient to be infected with both . The clinical course and the associated mortality rate of dual infection with HIV-1 and HIV-2 (HIV-D) has not been described so far, apart from a small study in Côte d'Ivoire in tuberculosis patients , which reported no difference in mortality rates between HIV-1 and HIV-D.
We compared the mortality rates and survival of patients with HIV-D infection with those who had HIV-1 or HIV-2 single infections, in a hospital-based cohort study. This represents a continuation of an earlier study of the same cohort .
Patients and methods
The first case of HIV in The Gambia was identified in 1986. In 1995, the prevalence of HIV-1 among pregnant women was 0.5% [95% confidence interval (CI) 0.4–0.6%], of HIV-2 1.1% (95% CI 1.0–1.2%), and of HIV-D 0.05% (95% CI 0.03–0.09%) . At the clinic and hospital of the Medical Research Council (MRC) Laboratories in Fajara, The Gambia (West Africa), individuals are tested for HIV antibodies for a variety of reasons. All blood donors and self-identified female commercial sex workers are routinely tested. Most patients with tuberculosis or a suspected sexually transmitted disease, and partners of HIV and sexually transmitted disease cases are also tested. Finally, patients with a clinical presentation suggestive of HIV disease are tested. The study population therefore consists of both healthy and symptomatic patients. The clinic is situated in an urban area near the capital, but also attracts patients from rural areas. It is the national referral centre for HIV patients in The Gambia. Patients aged 15 years or older attending the clinic who were HIV-positive upon screening, and who gave informed consent, were included in the study. All patients were counselled before and after HIV testing.
The date of seroconversion was unknown for most patients. The date of the first positive HIV test was taken as the date of enrolment. The clinic started recruiting patients in May 1986, and enrolment for this study closed on 30 September 1997.
Patients who were tested in outside laboratories and not re-tested at the MRC (as a result of early death or loss to follow-up) were excluded. None of the patients is known to have been on antiretroviral therapy during the study period. Patients with tuberculosis were treated with multidrug therapy according to the national guidelines. Prophylaxis against opportunistic infections or tuberculosis was not provided during the period of the study.
Serological diagnosis of HIV infections
Serum was screened by the Wellcozyme HIV 1+2 (Murex Diagnostics Ltd., Dartford, UK) until August 1996, and after that date by the ICEHIV-1.O.2 (Murex). If reactive, samples were re-tested by type-specific enzyme-linked immunosorbent assays (ELISA). For HIV-1 this was the Wellcozyme HIV recombinant-1 (Murex), and for HIV-2 the Wellcozyme HIV-2 (Murex) from the start until April 1996, and after that the ICE*-HIV-2 test (Murex). Samples that were clearly reactive in only one type-specific ELISA were assigned a serological diagnosis accordingly. Samples positive in both ELISA were further tested by a synthetic peptide-based strip method, Pepti-Lav 1-2 (Sanofi Diagnostics Pasteur, Marne la Coquette, France). We interpreted the appearance of a clear band (++) or a very clear band (+++) as evidence of infection with the relevant HIV type; samples with clear or very clear lines for both virus types were considered to be dually infected. A second, confirmatory serum sample, usually taken 2–8 weeks later, was tested in the same way. Samples with at least one positive test result, but inconclusive ELISA and Pepti-Lav results, or with insufficient serum to do all the required tests, and patients with two samples having incompatible results were classified as indeterminate. A serological diagnosis was assigned by investigators unaware of the clinical condition of the patient.
Polymerase chain reaction confirmation of dual infection
Peripheral blood mononuclear cells (PBMC) stored at −70°C from patients who were serologically dually reactive, were tested by qualitative polymerase chain reaction (PCR), using methods that have been described elsewhere [12,13]. If both the HIV-1-specific PCR and the HIV-2-specific PCR signals were positive, the patient was considered to have dual infection. Cases from whom PBMC were not available but in whom the serological pattern unequivocally suggested dual infection, were also considered to be HIV-D. Finally, some cases positive by PCR for HIV-1 but not for HIV-2, with unequivocal serological evidence of dual infection, were considered to be HIV-D, as in HIV-D patients with advanced immunodeficiency the PCR signal for HIV-2 may disappear [14,15]. The HIV status at recruitment was used to classify patients as HIV-1, HIV-2, or HIV-D.
At the first visit after a positive test result a research clinician took a history and conducted a full physical examination. All patients were given a score on the Karnofsky performance scale, ranging from 10 (moribund) to 100 (asymptomatic and well) . Patients were staged according to the Centers for Disease Control (CDC) 1993 system . From 1 January 1993 patients were also categorized according to the World Health Organization (WHO) clinical staging system, ranging from 1 (asymptomatic infection, persistent generalized lymphadenopathy or acute retroviral syndrome), to 4 (AIDS) . As a result of limited options for investigations, most diagnoses and staging were clinical rather than laboratory-confirmed.
CD4 cell count measurement
The CD4 cell count was estimated by FACScan (Becton-Dickinson, Oxford, UK). Lymphocyte subset measurements were started routinely in November 1988. These were performed as soon as possible after patient enrolment. In some patients, a CD4 cell count was not available for the first visit, but for a subsequent visit. These were used as the baseline CD4 cell count if the measurement was performed within 3 months of recruitment.
Patients were invited to attend the clinic at least every 3 months, regardless of symptoms. Those who failed to do so were visited at home by a fieldworker to ascertain the vital status. A patient was considered to be lost to follow-up if the study team had no information on his or her vital status at the close of the study. The observation time of patients who were lost to follow-up was censored at the last date they were known with certainty to be alive. The observation time of patients who refused further contact with the study team was censored at the date of their refusal. The observation period closed at 31 December 1997, and we aimed to establish the vital status of all originally recruited patients at that date.
The cause of death was unknown in the majority of cases, because most patients died at home. For patients who died in the MRC hospital the date of death was extracted from the hospital records. For patients dying at home a field worker obtained a date of death by interviewing relatives.
Data management and statistical analysis
Data were entered initially in dBASE III PLUS (Ashton-Tate, CA, USA), and later in FoxPro 2.6a for DOS (Microsoft Corporation, WA, USA). Statistical analysis was performed using the SAS System (SAS Institute Inc., NC, USA) and Stata version 6.0 (Stata Corporation, College Station, TX, USA).
Continuous data were compared by Wilcoxon test, or by the t-test if approximately normally distributed. Proportions were compared with the χ2 test or Fisher's exact test, as appropriate. Differences in survival were examined using Kaplan–Meier survival curves, the log-rank test, and multivariate analysis by Cox regression. Significance was assessed using the likelihood ratio test.
The study was approved by the Gambian Government/MRC Joint Ethics Committee.
A total of 1534 adult HIV-infected patients were recruited; from 642 patients (42%) of these only one sample was available for serology. A total of 746 patients were HIV-1 infected, 666 were HIV-2 infected and 107 patients had dual infection (HIV-D); 15 patients with indeterminate HIV infection were excluded from the analysis. PBMC were available for PCR from 82 of the 107 patients diagnosed as dually infected by serology, and in 73 of these, dual infection was confirmed by PCR. The other nine, who had advanced disease with a median CD4 cell count of 40 cells/μl [interquartile range (IQR) 30–250], tested positive by PCR for only one of the two HIV types.
The characteristics of the patients are shown in Table 1. The annual number of recruited HIV-1 cases increased over the years, but the number of HIV-2 cases stabilized after 1993. HIV-1-infected patients were significantly younger than HIV-2 and HIV-D-infected patients on first presentation (P < 0.0001 and P = 0.04, respectively). A higher proportion of HIV-2-infected patients than HIV-1 patients were female (P < 0.001). On average, women were 6 years (95% CI 5–7) younger than men at first presentation.
There were significant differences between the three types of infection at presentation according to the distribution of the clinical stage by Karnofsky score, by CDC and WHO classifications (P = 0.0007, P = 0.03, and P < 0.0001, respectively). HIV-1 and HIV-D-infected patients tended to present at a similar, more advanced stage of disease than HIV-2-infected patients. CD4 cell counts were measured in 1126 patients (74%), and were available within 3 months of their first presentation from 894 patients (59%). HIV-2-infected patients presented with a significantly higher median CD4 cell count than HIV-1-infected patients (P < 0.001); there was no significant difference in CD4 cell count between HIV-D and HIV-1-infected patients (P = 0.3, Wilcoxon test). The median CD4 count at presentation was 330 (IQR 160–570) for women and 180 (IQR 60–370) for men (P < 0.0001). The median Karnofsky score at presentation was 80 (IQR 70–90) for women and 70 (IQR 50–80) for men (P < 0.0001).
Overall, 161 patients (11%) were lost to follow-up either because they had moved or emigrated, or had refused further contact with the clinic (Table 2). There were no significant differences in the proportions lost to follow-up between the three types of infection. Those lost to follow-up were younger, more often female, more often sex workers, and had a higher Karnofsky score, better WHO and CDC stages, and a higher CD4 cell count (data not shown). These differences were highly significant (P < 0.001 in each case).
The median follow-up was 12 months (range 0–128). A total of 423 HIV-1-infected (57%), 342 HIV-2-infected (51%) and 63 HIV-D-infected patients (59%) died (Table 2). The crude mortality rate in HIV-1 was higher than in HIV-2, but was similar to that in HIV-D-infected patients. The survival of HIV-2 patients was significantly longer than that of HIV-1 (log rank test, P = 0.006). There was no significant difference in survival of HIV-D compared with HIV-1 (P = 0.8) or HIV-2 (P = 0.13).
Mortality rates stratified by age and sex
When individuals were divided into four age groups, the mortality rate increased with increasing age (Table 3). This effect remained significant after adjusting for sex, CD4 cell count category, and HIV type: the hazards ratio (HR) for patients aged 25–34 years (95% CI) was 1.44 (1.01–2.07;P = 0.04), for patients aged 35–44 years 1.47 (1.00–2.15;P = 0.05), and for patients aged 45 years and above 1.93 (1.30–2.88;P = 0.0012) compared with those aged 15–24 years .
Men had a higher mortality rate than women. This effect was significant in HIV-1 and HIV-2 (P < 0.0001 for both;P = 0.06 in HIV-D). After adjusting for age, CD4 cell count category, and HIV type, the HR (95% CI) for men was 1.63 (1.3–2.0;P < 0.0001) compared with women.
Mortality rates by CD4 cell count category
Absolute CD4 cell counts were available within 3 months of the first visit for 894 patients. There were no significant differences between this subgroup and the total patient group with regard to age, sex and mortality rates (data not shown). The mortality rate was inversely related to the CD4 cell count for all three infection types. Fig. 1 shows the Kaplan–Meier survival curves for patients in the high (≥ 500 cells/μl), intermediate (200–499 cells/μl), and low (< 200 cells/μl) CD4 cell count categories, comparing the three types of infection. The median survival of HIV-1 patients in the highest CD4 cell count category was 4.9 years (95% CI 3.7–8.7); for HIV-2 and HIV-D the median survival could not be calculated because of insufficient data. The median survival for patients in the lowest CD4 cell count category was 6 months for HIV-1, 8 months for HIV-2, and 6 months for HIV-D.
When adjusting for age and sex in a Cox proportional hazards regression, there was little difference in mortality rates between HIV-1 and HIV-2 in the low CD4 cell count (< 200 cells/μl) and the intermediate (200–499 cells/μl) categories (see Table 4). In the high CD4 cell count category (≥ 500 cells/μl) the HIV-2 mortality rate was significantly lower than that in HIV-1 (HR 0.50, 95% CI 0.28–0.88;P = 0.02). The mortality rate in HIV-D was not significantly different from that in HIV-1 in any of the categories, but was significantly higher than that of HIV-2 in the high CD4 cell count category (P = 0.04).
Mortality rates by Karnofsky score
Karnofsky scores were divided into three categories (≤ 60, 70–80, and 90–100), according to tertiles. Mortality rates were inversely related to the Karnofsky category in all three infection types (Table 4). They were lower for HIV-2 than for HIV-1 in all three Karnofsky categories; the overall HR adjusted for age, sex, and Karnofsky score was 0.75 (95% CI 0.64–0.87;P < 0.0001). The HIV-D mortality rate was similar to that of HIV-1 in all strata, and the overall HR of HIV-D relative to HIV-1 was 1.09 (95% CI 0.83–1.42;P = 0.5).
Mortality rates by Centers for Disease Control and Prevention clinical stage
The CDC clinical stage was associated significantly with mortality rates in all three infection types (Table 4). The overall HR, adjusted for sex, age, and CDC stage, of HIV-2 compared with HIV-1 was 0.76 (95% CI 0.65–0.88;P = 0.0003). HIV-D mortality rates were similar to those of HIV-1 in each stratum; the overall HR adjusted for sex, age, and CDC stage of HIV-D compared with HIV-1 was 1.19 (95% CI 0.91–1.55;P = 0.2).
Mortality rates by World Health Organization clinical stage
The WHO clinical stage at baseline was available for only 930 patients (61%), as this information has only been recorded since 1993. The stage was significantly associated with mortality rates in HIV-1 and HIV-2; for HIV-D not enough data were available (Table 4). The overall HR of the mortality rate of HIV-2 compared with HIV-1 was 0.72 (95% CI 0.61–0.84;P < 0.0001); and for HIV-D compared with HIV-1: 1.07 (95% CI 0.87–1.40).
Restricting analysis to polymerase chain reaction confirmed HIV-1/HIV-2 dual infection
We repeated the analyses restricting the group of HIV-D to those with PCR-confirmed dual infection status (n = 73); the results were similar to those in the main analysis (data not shown).
This is the largest comparative survival study of HIV-1, HIV-2 and HIV-1/HIV-2 dual infection. We found that among patients with CD4 cell counts of 500 cells/μl or greater, those infected with HIV-2 had a significantly lower mortality rate than HIV-1 or HIV-D-infected patients. In the low CD4 cell count range (< 200 cells/μl) there appeared to be no difference in mortality rates between HIV-1, HIV-2, or HIV-D. Mortality rates appeared to be similar between HIV-1 and HIV-D in all CD4 cell count categories. Mortality rates were significantly higher among older patients (after adjusting for sex, CD4 cell count, and HIV type), and among men (after adjusting for age, CD4 cell count, and HIV type). The Karnofsky performance status was a very good predictor of mortality.
To overcome the problem of an unknown time since infection, we decided to stratify HIV patients by CD4 cell count. This is not ideal, as some patients are known to progress to low CD4 cell counts rapidly and others appear not to progress for more than 10 years, both in HIV-1  and in HIV-2 [20,21]. One cannot assume that the time since infection was similar for HIV-1 and HIV-2 patients. The resulting bias will lead to an underestimation of the difference in mortality rates between the two infections . In the same way, a real difference between HIV-1 and HIV-D could exist, although we were unable to detect it. In seroprevalent cohort studies [23,24], subjects can be categorized into CD4 cell count strata or clinical stages. Although the CD4 cell count does not indicate the time since infection, it indicates disease progression, and is a more reliable predictor of AIDS and death in HIV-1 than the time since infection . The high CD4 cell category consists of long-term non-progressors and recently infected individuals in an unknown case mix.
We used the HIV serology at enrolment to assign patients to the HIV-1, HIV-2, or HIV-D group. In some patients, seroconversion from single infection to HIV-D may have occurred. As HIV-2 has been prevalent for longer than the more recently introduced HIV-1 , we assumed that most seroconversions will have been from HIV-2 to HIV-D. We decided to ignore these seroconversions. This may have led to a dilution of the differences between HIV-2 and the other two types of infection, and therefore the differences that we found may be underestimates.
Routine CD4 cell counts were started in November 1988, 2.5 years after the beginning of the study. The reasons for the missing CD4 cell counts of patients recruited after that date were rapid clinical deterioration and death, early loss to follow-up, non-attendance, or the FACScan being temporarily out of service. CD4 cell counts are lacking from a larger proportion of HIV-2 than HIV-1 patients (P = 0.00005). Those missing data have reduced the power of the study and led to a different distribution of patients among the HIV infection types. However, as CD4 cell count was the key stratifying factor, it is unlikely to have caused bias.
The losses to follow-up were limited to 11%. Those lost to follow-up were less ill and had a higher CD4 cell count than those who remained in the study. As the main predictor of death was CD4 cell count and the analysis was stratified by CD4 cell count, the loss to follow-up reduced the power of the study marginally in the high CD4 cell count category, but is unlikely to have introduced much bias.
There is broad agreement that in some patients HIV-2 can be pathogenic and can cause immunodeficiency and AIDS. However, the excess mortality risk of HIV-2-infected patients compared with HIV-negative patients appears to be limited [5–7]. Because the proportion of HIV-2 patients in the high CD4 cell count category is much larger in the population at large than in our study population, the contrast between HIV-1 and HIV-2 is most appropriately characterized by the differences found in the high CD4 cell count category. The significantly lower mortality rate in HIV-2 in that category (HR 0.5, P = 0.02) is compatible with findings of limited excess mortality among HIV-2-infected individuals [5–7]. Our finding that individuals in the advanced stage of infection (CD4 cell count < 200 cells/μl) have the same high mortality rate, irrespective of HIV type, suggests that HIV-1 and HIV-2 run the same course once the immune system is severely affected. The finding that the prognosis for patients with CD4 cell counts of 500 cells/μl or greater is better for HIV-2, coupled with the identification in community-based studies of many old and healthy HIV-2-infected individuals with normal CD4 cell counts [5,27], suggests that a substantial proportion of HIV-2-infected individuals is not harmed by the infection; others may experience a disease course that is indistinguishable from that of HIV-1.
HIV-D-infected patients had a mortality rate similar to HIV-1 in all CD4 cell count categories, although a non-significant trend towards worse survival rates was seen in the category of CD4 cell counts of 500 cells/μl or greater. The power to detect significant differences in survival between HIV-1 and HIV-D was limited; given the number of deaths, the study had 90% power to detect as statistically significant a HR (comparing HIV-D with HIV-1) of 1.56 or greater. Therefore, smaller differences cannot be excluded. The poor survival of HIV-D suggests that preceding HIV-2 does not act as a ‘vaccine’ mitigating the disease course of subsequent HIV-1 infection.
The CASCADE study  showed that in HIV-1 the age at infection is strongly predictive of mortality. In our seroprevalent study, the age at enrolment was also significantly associated with mortality. The CASCADE study  did not find a significant difference in survival between men and women, nor did several other cohort studies of HIV-1 in Africa [8,28,29], but in our study men had a significantly higher mortality rate than women in both HIV types. Even after adjustment for CD4 cell count category, HIV type, and age, the mortality HR for men compared with women was 1.63 (95% CI 1.3–2.0). In an occupational cohort study in Tanzania , men with HIV-1 had a mortality rate almost twice that of women. In our study, men were at a much more advanced stage of infection than women, and perhaps adjusting for the CD4 cell count did not capture that difference fully. Other potential reasons for this higher mortality rate among men, such as higher viral loads, should be further examined.
The WHO clinical classification does not require sophisticated laboratory support; it is a scoring system taking into account a wide range of clinical conditions. It has been used with success in research settings [23,24], but it is unclear how widespread its use is in general hospitals or health centres. The Karnofsky index is a simple clinical assessment that is made without the need for a laboratory, and which was originally created to assess the prognosis of patients with cancer . It correlated remarkably well with mortality rates, and may be a simple and useful tool for routine clinic settings and home care to classify HIV-positive patients.
This study confirms that HIV-2 infection is associated with a lower mortality rate than HIV-1, but shows for the first time that this difference is limited to patients with CD4 cell counts above 500 cells/μl. Earlier studies demonstrated lower heterosexual [31,32] and lower perinatal [11,33] transmission rates. Asymptomatic HIV-2 patients should be counselled that they have a better prognosis, that they will not necessarily get AIDS, and that the risk of transmitting the infection to an infant is much smaller than in HIV-1. Infection with both HIV-1 and HIV-2 carries the same prognosis as single HIV-1 infection.
The authors would like to thank all the patients in the study for their participation. They would also like to thank Tom Blanchard, Pa Tamba Ngom, Ramu Sarge-Njie, Bakary Sanneh, Mamady Njie, and Andrew Norris for skillful laboratory work; Ken Joof, Ramu Jaigne, Alieu Jatta, Musukebba Sanyang, and Babucarr Jawneh for counselling in the clinic and follow-up in the field; Amie Sarjo, Awa Kendah and Fatou Njie for data entry; and Roel Coutinho for comments on drafts and helpful discussions. They also thank two anonymous referees for helpful comments and suggestions, and appreciate the continued support of Saihou Ceesay, director of the Gambian National AIDS Secretariat. This study built on earlier work by Drs Andrew Wilkins, Jacques Pepin, Arinze Egboga, and Sarah Hawkes.
1. Collaborative Group on AIDS Incubation and HIV Survival
including the CASCADE EU Concerted Action. Time from HIV-1 seroconversion to AIDS and death before widespread use of highly-active antiretroviral therapy
. Lancet 2000, 355: 1131–1137.
2. Morgan D, Mahe C, Mayanja B, Okongo JM, Lubega R, Whitworth J. HIV-1
infection in rural Africa
: is there a difference in median time to AIDS and survival
compared with that in industrialized countries? AIDS 2002, 16: 597–603.
3. Matheron S, Mendoza-Sassi G, Simon F, Olivares R, Coulaud JP, Brun-Vezinet F. HIV-1
AIDS in African patients living in Paris [Letter]. AIDS 1997, 11: 934–936.
4. Whittle H, Morris J, Todd J. et al
-infected patients survive longer than HIV-1
-infected patients. AIDS 1994, 8: 1617–1620.
5. Ricard D, Wilkins A, N'Gum PT. et al
. The effects of HIV-2
infection in a rural area of Guinea-Bissau. AIDS 1994, 8: 977–982.
6. Poulsen AG, Aaby P, Larsen O. et al
. 9-year HIV-2
-associated mortality in an urban community in Bissau, west Africa
. Lancet 1997, 349: 911–914.
7. Norrgren H, Da Silva ZJ, Andersson S. et al
. Clinical features, immunological changes and mortality in a cohort of HIV-2-infected individuals in Bissau, Guinea-Bissau
. Scand J Infect Dis 1998, 30: 323–329.
8. Nunn AJ, Mulder DW, Kamali A, Ruberantwari A, Kengeya-Kayondo JF, Whitworth J. Mortality associated with HIV-1
infection over five years in a rural Ugandan population: cohort study
. BMJ 1997, 315: 767–771.
9. Nkengasong JN, Kestends L, Ghys PD. et al
. Dual infection with human immunodeficiency virus type 1 and type 2: impact on HIV type 1 viral load and immune activation markjers in HIV-seropositive female sex workers in Abidjan, Ivory Coast. AIDS Res Hum Retroviruses 2000, 16: 1371–1378.
10. Kassim S, Ackah A, Abouya L, et al.Prospective cohort study of HIV-1/HIV-2 dual infection among HIV-infected tuberculosis patients in Abidjan, Cote d'Ivoire
. In:13th World AIDS Conference
. Durban, 2000 [Abstract TuPeC3355].
11. O'Donovan D, Ariyoshi K, Milligan P. et al
. Maternal plasma viral RNA levels determine marked differences in mother-to-child transmission rates of HIV-1 and HIV-2 in The Gambia
. AIDS 2000, 14: 441–448.
12. Berry N, Ariyoshi K, Jobe O. et al
. HIV type 2 proviral load measured by quantitative polymerase chain reaction correlates with CD4+ lymphopenia in HIV type 2-infected individuals. AIDS Res Hum Retroviruses 1994, 10: 1031–1037.
13. Berry N, Ariyoshi K, Jaffar S. et al
. Low peripheral blood viral HIV-2
RNA in individuals with high CD4 percentage differentiates HIV-2
infection. J Hum Virol 1998, 1: 457–468.
14. Dieng Sarr A, Hamel DJ, Thior I. et al
dual infection: lack of HIV-2
provirus correlates with low CD4+ lymphocyte counts. AIDS 1998, 12: 131–137.
15. Ariyoshi K, Cham F, Harding E. et al
. Diagnosis of HIV-1
/2 dual infection using dilutional analysis of type specific antibody [Letter]. AIDS 1998, 12: 2504–2505.
16. Karnofsky DA, Abelmann WH, Craver LF, Burchenal JH. The use of the nitrogen mustards in the palliative treatment of carcinoma. Cancer 1948, 2: 634–656.
17. Centers for Disease Control and Prevention. Revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults
. MMWR 1992, 41/RR-17: 1–19.
18. World Health Organization. Acquired immune deficiency syndrome (AIDS): interim proposal for a WHO staging system for HIV infection and disease
. Wkly Epidemiol Rec 1990, 65: 221–228.
19. Keet IPM, Krijnen P, Koot M. et al
. Predictors of rapid progression to AIDS in HIV-1 seroconverters
. AIDS 1993, 7: 51–57.
20. van der Ende ME, Schutten M, Ly TD, Gruters RA, Osterhaus AD. HIV-2
infection in 12 European residents: virus characteristics and disease progression. AIDS 1996, 10: 1649–1655.
21. Ancelle R, Bletry O, Baglin AC, Brun-Vezinet F, Rey MA, Godeau P. Long incubation period for HIV-2
infection [Letter]. Lancet 1987, 1: 688–689.
22. Brookmeyer R, Gail MH. Biases in prevalent cohorts
. Biometrics 1987, 43: 739–749.
23. Morgan D, Maude GH, Malamba SS. et al
. HIV-1 disease progression and AIDS-defining disorders in rural Uganda
. Lancet 1997, 350: 245–250.
24. Kassa E, Rinke De Wit TF, Hailu E. et al
. Evaluation of the World Health Organization system for HIV infection and disease in Ethiopia: association between clinical stages and laboratory markers. AIDS 1999, 13: 381–389.
25. Philips AN, Pezzotti P, Lepri AC, Rezza G, and the Italian Seroconversion Study. CD4 lymphocyte count as a determinant of the time from HIV seroconversion to AIDS and death from AIDS: evidence from the Italian Serocconversion Study. AIDS 1994, 8: 1299–1305.
26. Wilkins A, Oelman B, Pepin J. et al
. Trends in HIV-1
infection in The Gambia [Letter]. AIDS 1991, 5: 1529–1530.
27. Larsen O, da Silva Z, Sandstrom A. et al
. Declining HIV-2
prevalence and incidence among men in a community study from Guinea-Bissau. AIDS 1998, 12: 1707–1714.
28. Todd J, Balra R, Grosskurth H. et al
. HIV-associated mortality in a rural Tanzanian population. AIDS 1997, 11: 801–807.
29. Sewankambo NK, Wawer MJ, Gray RH. et al
. Demographic impact of HIV infection in rural Rakai District, Uganda: results of a population-based cohort study
. AIDS 1994, 8: 1707–1713.
30. Senkoro KP, Boerma JT, Klokke AH. et al
. HIV incidence and HIV-associated mortality in a cohort of factory workers and their spouses in Tanzania, 1991 through 1996. J Acquired Immune Defic Syndr 2000, 23: 194–202.
31. Kanki PJ, Travers KU, Mboup S. et al
. Slower heterosexual spread of HIV-2
. Lancet 1994, 343: 943–946.
32. Donnelly C, Leisenring W, Kanki P, Awerbuch T, Sandberg S. Comparison of transmission rates of HIV-1
in a cohort of prostitutes in Senegal. Bull Math Biol 1993, 55: 731–743.
33. Adjorlolo-Johnson G, De Cock KM, Ekpini E. et al
. Prospective comparison of mother-to-child transmission of HIV-1
in Abidjan, Ivory Coast. JAMA 1994, 272: 462–466.