Secondary Logo

Journal Logo


Effectiveness of potent antiretroviral therapies on the incidence of opportunistic infections before and after AIDS diagnosis

Detels, Rogera; Tarwater, Patrickb; Phair, John P.c; Margolick, Josephb; Riddler, Sharon A.d; Muñoz, Alvaro*for the Multicenter AIDS Cohort Study

Author Information



The introduction of potent antiretroviral (combination) therapy has resulted in a dramatic decline in the overall incidence of AIDS in industrialized countries. This decline has been documented in both clinical trials and community-based observational studies in Europe and the United States [1–11]. The evaluation of the impact of potent antiretroviral therapy on the incidence of the most common opportunistic infections (OI) in individuals with less than 100 CD4 lymphocytes in the HIV Outpatient Study indicated a decline in the incidence of Pneumocystis carinii pneumonia, cytomegalovirus disease, and Mycobacterium avium complex [12]. In recent studies [13,14], the incidence of Kaposi's sarcoma has also decreased dramatically. The HIV Outpatient Study analysis, however, was limited to initial OI.

A second limitation was the inability to determine the duration of infection in that cohort. The Multicenter AIDS Cohort Study (MACS) provides an opportunity to address both the question of changes in the relative hazard (RH) of presenting OI AIDS events in men with a known duration of infection, and the risk of a subsequent complicating infection among the men in the cohort who had already progressed to AIDS [15–17]. Analysis of the RH of infections in this large cohort during the eras of nucleoside analogue monotherapy, combination therapy, and potent antiretroviral therapy allows for an evaluation of the impact of increasingly effective treatment provided by physicians in the community.

Materials and methods

Between March 1984 and April 1985, the MACS enrolled a cohort of 4954 men who had sex with men, were 18 years of age or older, and were free of clinical AIDS [15–17]. From 1987 to 1990, 668 additional men, mostly African-Americans, were recruited. The participants, 2195 of whom were seropositive and 3427 of whom were seronegative at entry, have been followed every 6 months since enrollment for sexual practices, signs of AIDS, general health, neuropsychological status, and medication use. The diagnosis of AIDS and specific AIDS-defining events were determined by self-reporting confirmed by a physician's or a hospital report. Participants were asked about the use of specific AIDS medications, including the prophylactic use of drugs for OI, and were given the opportunity to report the use of any drugs not directly queried.

Two separate analyses were conducted: (i) for the occurrence of OI as the presenting AIDS event (referred to as presenting OI AIDS event); and (ii) for the occurrence of OI as subsequent AIDS events among individuals who already had a presenting diagnosis of AIDS (referred to as subsequent OI AIDS events). The 1993 Centers for Disease Control and Prevention clinical criteria was used to define AIDS (i.e. individuals with < 200 CD4 cells/ml but free of clinical conditions were not considered to have AIDS).

Before 1 January 1990 few of the seropositive men without AIDS in the MACS were taking any drug. From 1 January 1990 to 31 December 1992 the majority of those seropositive men without AIDS on antiretroviral therapy were taking only monotherapy. From 1 January 1993 to 31 December 1995 an increasing proportion of the seropositive men were taking combination therapy, but from 1 January 1996 to 31 December 1998 the majority of the seropositive men on antiretroviral therapy were taking potent antiretroviral therapy. The pattern of antiretroviral therapy was similar among the men who had AIDS, but the proportion on treatment was higher for the men with AIDS over the entire time period, especially before 1990. For this reason we have performed the analyses using time periods as surrogates for treatment regimens: before 1990 for no treatment, 1990–1992 for monotherapy, 1993–1995 for combination therapy, and 1996–1998 for potent antiretroviral therapy.

The men in both analyses were followed up to 1 January 1999 for OI AIDS events (both presenting and subsequent). Analyses were performed for all OI and for four specific OI:P. carinii pneumonia, any cytomegalovirus disease, M. avium complex, and esophageal candidiasis. The occurrences of the other specific OI were too low to permit meaningful separate analyses. The other OI were: toxoplasmosis, crytosporidiosis, isosporiasis, histoplasmosis, progressive multifocal leukoencephalopathy, disseminated Mycobacterium tuberculosis, cryptococcal infection, cryptococcal meningitis, cryptococcal infection of the blood, cryptococcal infection of other organs, chronic mucocutaneous herpes simplex, coccidioidomycosis, salmonella, pulmonary tuberculosis, and recurrent pneumonia. The complement of the AIDS-defining conditions were considered non-OI AIDS events, and they included: Kaposi's sarcoma, non-Hodgkin's lymphoma, dementia, and wasting syndrome.

Analysis 1: presenting opportunistic infection AIDS events

To describe the incidence of OI as the initial AIDS diagnosis using the duration of infection as the time scale, the study population used was men who were seronegative at enrollment and who subsequently seroconverted before January 1999. This cut-off was used to allow for comparable periods of time to AIDS in the four calendar periods used in the analyses. Because the incidence of HIV infection consistently declined after 1986, the date of seroconversion was estimated as one-third of the time interval between the last seronegative and the first seropositive visit, instead of the midpoint of the interval, which assumes a constant incidence rate over time. Among the 543 seroconverters, 298 (55%) remained AIDS-free at the end of follow-up, 132 (24%) had presenting OI AIDS events, 71 (13%) had a presenting AIDS event that was not an OI (referred to as presenting non-OI AIDS events), and 42 (8%) were cases in whom we could not determine the type of presenting AIDS diagnosis (i.e. missing data).

For the purpose of describing the incidence of presenting OI AIDS events, we restricted our analyses to 430 seroconverters: the 298 who remained AIDS-free and at risk of developing OI as the presenting diagnosis (censored observations), and the 132 in whom we observed the occurrence of OI as the presenting AIDS diagnosis (uncensored observations). The rationale for excluding men with non-OI AIDS-defining conditions was that the primary objective of our analysis was to describe the incidence of OI as presenting and subsequent AIDS events. Obviously, those men who present with an initial non-OI AIDS event are not at risk of OI AIDS as an initial event. Also, statistically, it is inappropriate to handle these observations as censored at the time they developed a non-OI because doing so would amount to considering these individuals at risk of OI as a presenting AIDS diagnosis after they had already developed a non-OI AIDS-presenting illness. It is a common misunderstanding to think that censored observations are only included in the analysis before the times at which they are censored. They are simply not needed for the estimation of hazards because it is assumed that those who remained under observation represent those who have previously been censored. It is this representativeness that allows the estimation of the times and probabilities when a censored observation will become an event. Indeed, we have included in the Appendix an example showing how, for the Kaplan–Meier estimator, censored observations are redistributed as events past the times at which they are censored.

Although a more rigorous analysis would have been to allow for the censored (i.e. AIDS-free at last time seen) observations to have the chance of contributing to the incidence of non-OI AIDS [18,19], the impact of such an analysis would be minimal. This is because the great majority of these censored observations occurred at the date of analysis and thus past any observed AIDS event [20]. Furthermore, because the exclusion criteria were the same across all calendar periods and we are measuring RH of exposure in the calendar periods, the effect of the exclusion is further diminished.

Analysis 2: subsequent opportunistic infection AIDS events

From the set of seropositive participants (2195 seroprevalent and 543 seroconverters), 1470 men with known date and diagnosis of AIDS were followed for the occurrence of a subsequent OI AIDS event. For the analysis of subsequent OI AIDS events models, we used the time from the presenting AIDS event to the first occurrence of a subsequent OI AIDS event as our outcome measure. Although the number of subsequent OI AIDS events after the presenting AIDS event is an alternative measure of effectiveness, it was not the focus of this study. Analogous to the exclusion in the previous analysis, we excluded those men who died free of a subsequent OI AIDS event.

Two hundred and eighteen of the 717 subsequent OI AIDS events occurred before 1990, when the MACS study protocol did not require that participants be followed after their presenting AIDS events. Therefore, the analysis for subsequent OI AIDS events does not include the calendar period before 1990.

Both analyses of OI AIDS events controlled for opportunistic infection prophylaxis use. Prophylaxis use was a dichotomous variable that identified the use of specific opportunistic infection prophylaxis relative to the specific outcome under study. In particular, the prophylaxis drugs identified before the presenting AIDS diagnosis were for P. carinii pneumonia: bactrim, dapsone, pentamidine (aerosolized and intravenous), trimethoprim; for M. avium complex: clarithromycin, azithromycin, rifabutin; for cytomegalovirus disease: acyclovir; and for esophageal candidiasis: fluconozole. Prophylaxis use in the analysis of all OI identified any of the above medications and included ganciclovir. In the analysis of subsequent OI AIDS events, we also controlled for the type of presenting AIDS event, grouped into five categories: Kaposi's sarcoma, lymphoma, dementia, wasting syndrome, and OI. For the 82 individuals who presented with multiple diagnoses, a non-OI AIDS condition was considered to have occurred before an OI AIDS condition.

Statistical methods

Changes in the pattern of occurrences of specific OI can be described by incidences reported in each therapy era. However, incidence rates do not take into account an individual's progression of HIV infection and, therefore, should only be used descriptively. Inferences concerning OI should be drawn from measures obtained from techniques that correctly adjust for disease progression, such as adjustment by the duration of infection for presenting OI AIDS, and by the time with AIDS for subsequent OI AIDS. In order to control for disease progression, we used survival analysis methods, with the duration of infection and time with AIDS as the time variables for the analysis of presenting OI AIDS and subsequent OI AIDS, respectively. The comparison of hazard functions in the survival analysis setting is equivalent to the comparison of incidence rates. A proportional hazards model with staggered entries was, therefore, used to calculate RH between calendar periods. The staggered entry technique accounts for comparable HIV infection duration for analysis 1 (presenting OI) and time with AIDS for analysis 2 (subsequent OI) among at-risk individuals in each calendar period. An individual only contributed time to the calendar periods in which they were at risk of the event. Individuals who experienced only a non-OI AIDS event or died free of an OI AIDS event were considered to be a competing risk and were removed from the risk set before analysis. The era of monotherapy was used as the reference period for the estimation of RH.


Table 1 presents the descriptive statistics of the study populations used in the two analyses. The first column shows the data for the seroconverters used in the analysis of presenting OI AIDS events, and column two shows the data for the men with AIDS used in the analyses of subsequent OI AIDS events. Among the 543 men at risk of an OI as the initial AIDS-defining event, the median age at seroconversion was 33 years. There were 245 AIDS-defining events; 42 of these were removed because the diagnosis was undetermined or was determined at death, making the date of onset difficult to determine. Therefore, these have not been considered in the analyses. Of the 203 remaining presenting AIDS events, a total of 132 were OI. The remaining 71 non-OI AIDS events were not considered in the analysis. The median years of follow-up for the seroconverters was 7.19 years.

Table 1
Table 1:
Descriptive statistics of the study populations used in the analyses of presenting and subsequent opportunistic infection AIDS events in the Multicenter AIDS Cohort Study, 1984–1998.

Among the 1470 AIDS cases, the median date of AIDS was July 1990 and the median age of onset was 38.5 years. There was a total of 839 events subsequent to AIDS, of which 717 were OI. The median years of follow-up for the AIDS cases was 0.83 years.

Fig. 1 and Fig. 2 show the prevalence of therapy use among AIDS-free HIV-infected participants and participants with AIDS in the MACS, respectively. The figures also display the prevalence trends of medication use for OI (P. carinii pneumonia, M. avium complex, cytomegalovirus disease, and esophageal candidiasis). The introduction and subsequent use of different antiretroviral therapy regimens was captured by the calendar periods described for the analyses. The use of medications either as prophylaxis or treatment for the four specific OI increased steadily from 1990 until 1996, and remained constant thereafter among the AIDS-free participants but decreased markedly among participants with AIDS.

Fig. 1.
Fig. 1.:
  The prevalence of antiretroviral therapy use and opportunistic infection prophylaxis use by seropositive participants while free of AIDS in the Multicenter AIDS Cohort Study. ░ Nucleoside reverse transcriptase inhibitor monotherapy; ░ nucleoside reverse transcriptase inhibitor combination therapy; ▒ potent antiretroviral therapy; - - - -P. carinii pneumonia prophylaxis; – – – candida prophylaxis; – - –M. avium complex prophylaxis.
Fig. 2.
Fig. 2.:
  The prevalence of antiretroviral therapy use and opportunistic infection medication use by participants with AIDS in the Multicenter AIDS Cohort Study. ░ Nucleoside reverse transcriptase inhibitor monotherapy; ░ nucleoside reverse transcriptase inhibitor combination therapy; ▒ potent antiretroviral therapy; - - - -P. carinii pneumonia prophylaxis; – – – candida prophylaxis; – - –M. avium complex prophylaxis;–– cytomegalovirus prophylaxis.

Table 2 presents the descriptive statistics of the individuals who contributed time at risk to each calendar period in both analyses. The CD4 cell counts in the AIDS-free seroconverters decreased with time as HIV disease in the cohort continued to progress. Because of the overwhelming proportion of AIDS cases on potent antiretroviral therapy and the probable loss to death of those individuals who did not respond to therapy, the CD4 cell count in participants with AIDS actually showed an increase in the last calendar period. The decreased incidence rates in the last period (i.e. a 72% decrease among the AIDS-free seroconverters from the previous two periods and a 75% decrease among AIDS cases from the previous two periods) show the effect of potent antiretroviral therapy, but the use of incidence rates does not correctly account for disease progression over time, as indicated by the low incidence observed among AIDS-free participants in the era of no therapy.

Table 2
Table 2:
Descriptive statistics of the study population for analysis 1 (presenting opportunistic infections) and analysis 2 (opportunistic infections subsequent to AIDS) by calendar period.

Presenting opportunistic infection AIDS events: analysis 1

Table 3 shows the RH of presenting OI AIDS events using 1990–1993 as the reference period. A marked decline is seen in the most recent calendar period for all OI categories. The reduction in the RH was statistically significant, however, only for all OI and P. carinii pneumonia. There were also reductions in cytomegalovirus disease, M. avium complex, and esophageal candidiasis, but these were not statistically significant, probably because of the limited number of these OI that occurred. These estimates were adjusted by OI-specific prophylaxis use before the OI AIDS event. P. carinii pneumonia prophylaxis use was also significantly protective against the occurrence of P. carinii pneumonia (RH 0.44, P = 0.009), but prophylaxis use for cytomegalovirus disease was shown to have a strong deleterious effect (RH 26.1, P = 0.003).

Table 3
Table 3:
Relative hazards (P value) of opportunistic infections (or a specific opportunistic infection) as the presenting AIDS event in calendar periods relative to 1990–1992 as the reference.

Subsequent opportunistic infection AIDS events: analysis 2

Of the 1470 AIDS cases considered, 62.7% had an OI, 22.8% had Kaposi's sarcoma, 4.6% had lymphoma, 3.5% had dementia, and 6.5% had wasting syndrome as the AIDS-defining event. Among the 717 AIDS cases with subsequent OI AIDS events, 64.9% had an OI, 24.9% had Kaposi's sarcoma, 1.4% had lymphoma, 3.1% had dementia, and 5.7% had wasting syndrome as the AIDS-defining event. The difference between these two distributions is not significant (P = 0.113), thus the occurrence of presenting AIDS diagnoses in our selected sample does not require adjustment for selection bias. Similarly, the distribution of presenting AIDS diagnoses among the individuals who contributed time in each calendar period does not differ across the three periods (i.e. OI, 60, 58, and 54%; Kaposi's sarcoma, 25, 25, and 22%; lymphoma, 4, 5, and 5%; dementia, 5, 5, and 6%; and wasting syndrome, 6, 8, and 13%, respectively).

Table 4 shows the RH for subsequent OI AIDS events. A dramatic reduction in RH was seen for all OI combined and for each of the specific OI analysed. The percentage reduction in RH between the last two calendar periods was 77% for all OI, 89% for P. carinii pneumonia, 90% for cytomegalovirus disease, 72% for M. avium complex and 42% for esophageal candidiasis. The estimates were adjusted for the presenting AIDS event and OI prophylaxis use before the presenting AIDS event. The results for the other covariates showed that presenting with dementia significantly reduced an individual's chance of having a subsequent OI AIDS event (RH 0.53, P = 0.007). The use of prophylaxis was shown to have a deleterious effect on the occurrence of subsequent OI AIDS events (RH 1.43, P = 0.001), but this result was consistent with the bias to select for treatment those who most need treatment.

Table 4
Table 4:
Relative hazards (P value) of opportunistic infections (or specific opportunistic infection) subsequent to AIDS in calendar periods relative to 1990–1992 as the reference.


The results of this study indicate that potent antiretroviral therapy, as provided by the participants’ physicians, has had a dramatic impact on the incidence of OI, both before and after the onset of AIDS, among the men enrolled in the MACS. The incidence of AIDS-defining OI among men with incident infection was reduced by 72% compared with the calendar period immediately before potent antiretroviral therapy became available. The decline in OI as a secondary event among MACS participants diagnosed with clinical AIDS was 83%. Similar declines were observed in the RH. These declines in AIDS-defining and secondary events in the potent antiretroviral therapy era were accompanied by a decline in the use of prophylactic antibiotics.

The decline in incidence of specific OI as an AIDS-defining event reached significance for P. carinii pneumonia only. The relatively small number of other specific OI may have precluded observing significance. These results are in accord with the observations of Weverling et al. [21] and Furrer et al. [22], who observed a reduced incidence of P. carinii pneumonia and esophageal candidiasis in association with the introduction of effective antiretroviral therapy. This decline in OI was not a result of the enhanced adherence to prophylaxis, as the participants reported the same level of utilization of these agents (Fig. 1 and Fig. 2) during the period when potent antiretroviral therapy was available as in the previous calendar periods for those not yet having AIDS, and a declining use for those already having a diagnosis of AIDS.

The effect observed for cytomegalovirus disease prophylaxis (RH 26.1) in the analysis of presenting OI AIDS is an artifact of the strong bias to select for prophylaxis use those who indicate the most need. In the case of cytomegalovirus disease prophylaxis, the indication for treatment is extreme immunosuppression (i.e. CD4 cell count < 50). The observed effect that presenting with dementia reduced the chance of having a subsequent OI AIDS event (RH 0.53) was probably due to the fact that dementia is usually diagnosed in advanced stages of infection progression, thus shortening the time in which a subsequent OI AIDS event could occur.

We previously reported an increase in CD4 lymphocytes among MACS participants overall with the introduction of potent antiretroviral therapy [1]. Miller et al. [23], in a study of individuals with less than 50 CD4 cells/mm3 observed that individuals demonstrating an increase in these cells above 200/mm3 experienced a much-reduced rate of progression. These observations suggest that the increase in CD4 lymphocyte number is accompanied by an increase in competence of cell-mediated immune responses and possibly improved phagocytic cell function. Our results, therefore, support the concept of increased cell-mediated immune function with the use of potent antiretroviral therapy.


These data demonstrate that the risk of OI declined at the same time that prophylaxis use declined and potent antiretroviral therapy use increased. The data thus extend the original observation of a reduction in the incidence of AIDS among seroconverters in the MACS, and offer further evidence that observational data derived from cohort studies can be used to evaluate the effectiveness of a widely used intervention at a population level.


1. Detels R, Muñoz A, McFarlane G. et al. Effectiveness of potent antiretroviral therapy on time to AIDS and death in men with known HIV infection duration. JAMA 1998, 280: 1497 –1503.
2. Markowitz M, Saag M, Powderly WG. et al. A preliminary study of ritonavir, an inhibitor of HIV-1 protease, to treat HIV-1 infection. N Engl J Med 1995, 333: 1534 –1539.
3. Egger M, Hirschel B, Francioli P. et al. Impact of new antiretroviral combination therapies in HIV infected patients in Switzerland: prospective multicentre study. :Swiss HIV Cohort Study. BMJ 1997, 315: 1194 –1199.
4. Gulick RM, Mellors JW, Havilir D. et al. Treatment with indinavir, zidovudine, and lamivudine in adults with human immunodeficiency virus infection and prior antiretroviral therapy. N Engl J Med 1997, 337: 734 –739.
5. Mocroft A, Katlama C, Johnson AM. et al. AIDS across Europe, 1994–98: the EuroSIDA study. Lancet 2000, 356: 291 –296.
6. 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.
7. Hammer SC, Squires KE, Hughes MD. et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. N Engl J Med 1997, 337: 725 –733.
8. Hogg RS, O'Shaughnesy MV, Gataric N. et al. Decline in deaths from AIDS due to new antiretrovirals. Lancet 1997, 349: 1294. 1294.
9. Centers for Disease Control and Prevention. Update: trends in AIDS incidence, deaths, and prevalence – United States, 1996. MMWR 1997, 46: 155 –173.
10. Sorvillo F, Kerndt PR, Odem S, Castillon M, Carruth A, Contreras R. Use of protease inhibitors among persons with AIDS in Los Angeles County. J Acquir Immune Defic Syndr Hum Retrovirol 1997, 15: 179 –181.
11. Chiasson MA, Berenson L, Li W, Schwartz S, Mojica B, Hamburg M. Declining AIDS mortality in New York City.Program and Abstracts of the 4th Conference on Retroviruses and Opportunistic Infections. Washington, DC, January 1997 [Abstract 376].
12. Palella FJ Jr, Delaney KM, Moorman AC. et al., and HIV Outpatient Study Investigators. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med 1998, 338: 853 –860.
13. Jacobson LP, Yamashita TE, Detels R. et al. Impact of potent antiretroviral therapy on the incidence of Kaposi's sarcoma and non-Hodgkin's lymphomas among HIV-1 infected individuals. J Acquir Immune Defic Syndr Hum Retrovirol 1999, 21: 34 –41.
14. Aboulafia DM. Regression of acquired immunodeficiency syndrome-related pulmonary Kaposi's sarcoma after highly active antiretroviral therapy. Mayo Clin Proc 1998, 73: 439 –443.
15. Kaslow RA, Ostrow DG, Detels R, Phair JP, Polk BF, Rinaldo CR Jr. The Multicenter AIDS Cohort Study: rationale, organization, and selected characteristics of the participants. Am J Epidemiol 1987, 126: 310 –318.
16. Chmiel JW, Detels R, Kaslow RA, Van Raden M, Kingsley LA, Brookmeyer R. Factors associated with prevalent human immunodeficiency virus (HIV) infection in the Multicenter AIDS Cohort Study. Am J Epidemiol 1987, 126: 568 –577.
17. Detels R, Phair JP, Saah AJ. et al. Recent scientific contributions to understanding HIV/AIDS from the Multicenter AIDS Cohort Study. Am J Epidemiol 1987, 126: 568 –577.
18. Hoover DR, Peng Y, Saah AJ. Using multiple decrement models to estimate risk and morbidity from specific AIDS illnesses. Stat Med 1996, 15: 2307 –2321.
19. Hoover DR, Peng Y, Saah A, Semba R, Detels R, Rinaldo CR Jr, Phair JP. Occurrence of cytomegalovirus retinitis after human immunodeficiency virus immunosuppression. Arch Ophthalmol 1996, 114: 821 –827.
20. Roca-Antonio J, Escudero LE, Gener J, Oller B, Rodriguez N, Muñoz A. Methods for analysis of the occurrence of abscess in patients with pancreatitis. Clin Perf Qual Health Care 1997, 5: 148 –152.
21. Weverling GJ, Mocroft A, Ledergerber B. et al. Discontinuation ofPneumocystis cariniipneumonia prophylaxis after start of highly active antiretroviral therapy in HIV-1 infection. :EuroSIDA Study Group. Lancet 1999, 353: 1293 –1298.
22. Furrer H, Egger M, Opravil M. et al. Discontinuation of primary prophylaxis againstPneumocystis cariniipneumonia in HIV-1 infected adults treated with combination antiretroviral therapy. N Engl J Med 1999, 340: 1301 –1306.
23. Miller V, Mocroft A, Reiss P. et al. Relations among CD4 lymphocyte count nadir, antiretroviral therapy, and HIV-1 disease progression: results from the EurosSIDA Study. Ann Intern Med 1999, 130: 570 –577.
24. Efron B. The two sample problem with censored data. Proceedings of the fifth Annual Berkeley Symposium on Mathematical Statistics and Probability, Vol. IV : University of California Press. Berkeley, California, 1967: 831 –853.


    We offer the following example to show how censored observations are redistributed as events past the times at which they are censored.

    Assume time to event data from a study of 14 individuals (with + denoting the censored observations) are as follows: 1, 1, 1, 2+, 4, 5, 6+, 8, 10, 11+, 12 16+, 16+, 16+. The illustration below (above the table) using periods for uncensored times and open circles with arrows for censored times depicts the data and places where events occurred for uncensored times and can occur for censored times where p1, p2, . . ., p7 represent the proportions developing the event at different times.

    From these data, the Kaplan-Meier estimates of the probabilities of event are: p1 = 0.2143, p2 = 0.0786, p3 = 0.0786, p4 = 0.0898, p5 = 0.0898, p6 = 0.1122, and p7 = 0.3368. From these probabilities, the observation censored at 2 could develop the event at times 4, 5, 8, 10, 12 or >16 whose sum of the corresponding probabilities is p2+p3+. . .+p7 = 0.7857. From the self-consistency algorithm [24], the probabilities of the censored observation at 2 developing the event at 4 is p2/0.7857 = 0.0786/0.7857 = 0.100; at 5 is 0.0786/0.7857 = 0.100; at 8 is 0.0898/0.7857 = 0.114; . . . at > 16 is 0.3368/0.7857 = 0.429. Using similar procedures for the other censored observations, the following table can be derived. The table demonstrates how censored observations are ‘redistributed’ as events past the times at which they were censored. Each censored observation has their 100% of information partitioned across the times of future observed events. The values in the last row equal the above state Kaplan-Meier estimates.FIGURE


    *The Multicenter AIDS Cohort Study (MACS) includes the following:

    Baltimore: The Johns Hopkins University School of Hygiene and Public Health: Joseph B. Margolick, Principal Investigator; Haroutune Armenian, Homayoon Farzadegan, Nancy Kass, Justin McArthur, Steffanie Strathdee, Ellen Taylor.#OChicago: Howard Brown Health Center and Northwestern University Medical School: John P. Phair, Principal Investigator; Joan S. Chmiel, Bruce Cohen, Maurice O'Gorman, Daina Variakojis, Jerry Wesch, Steven M. Wolinsky.

    Los Angeles: University of California, UCLA Schools of Public Health and Medicine: Roger Detels, Principal Investigator; Beth Jamieson, Principal Investigator; Barbara R. Visscher, Co-Principal Investigator; Eric Bing, John L. Fahey, John Ferbas, Otoniel Martínez-Maza, Eric N. Miller, Hal Morgenstern, Parunag Nishanian, John Oishi, Paul Satz, Elyse Singer, Jeremy Taylor, Harry Vinters, Dorothy Wiley, Stephen Young.

    Pittsburgh: University of Pittsburgh Graduate School of Public Health: Charles R. Rinaldo, Principal Investigator; James T. Becker, Phalguni Gupta, Lawrence Kingsley, John Mellors, Sharon Riddler, Anthony Silvestre.

    Data Coordinating Center: The Johns Hopkins University School of Hygiene and Public Health: Alvaro Muñoz, Principal Investigator; Lisa P. Jacobson, Co-Principal Investigator; Linda Ahdieh, Stephen Gange, Cynthia Kleeberger, Steven Piantadosi, Sol Su, Patrick Tarwater.

    NIH: National Institute of Allergy and Infectious Diseases: Carolyn Williams, Paolo Miotti. National Cancer Institute: Sandra Melnick.


    AIDS; cohort; potent antiretroviral therapy; HAART; HIV; opportunistic infections

    © 2001 Lippincott Williams & Wilkins, Inc.