ORIGINAL INVESTIGATION. The Incidence of AIDS-Defining Illnesses in 4883 Patients With Human Immunodeficiency Virus Infection

Transcription

1 ORIGINAL INVESTIGATION The Incidence of AIDS-Defining Illnesses in 4883 Patients With Human Immunodeficiency Virus Infection Amanda Mocroft, PhD; Michael Youle, MBChD; Andrew N. Phillips, PhD; Ramesh Halai; Phillipa Easterbrook, MD; Margaret A. Johnson, MD; Brian Gazzard, MD; for the Royal Free/Chelsea and Westminster Hospitals Collaborative Group Background: Acquired immunodeficiency syndrome (AIDS) defining illnesses are known to occur at different levels of immunosuppression, and the incidence of diagnoses may also vary according to the CD4 lymphocyte count strata. Information about the incidence of disease at different levels of immunosuppression would help clinicians monitoring patients and would allow prophylaxis to be targeted at the most appropriate population. Methods: Between 1982 and July 1995, 4883 patients testing positive for the human immunodeficiency virus were seen at either the Royal Free or Chelsea and Westminster Hospitals in London, England. The incidence of each diagnosis, both overall and stratified by CD4 lymphocyte count, was calculated using a person-years analysis. Patients who had no CD4 lymphocyte counts measured during follow-up were excluded from the analysis. Results: During a median follow-up period of 27.6 months, 3875 AIDS-defining illnesses were reported in 1713 patients. The incidence of AIDS-defining illnesses ranged from 6.22 per 100 person-years of follow-up for Pneumocystis carinii pneumonia (95% confidence interval, ) to 0.37 for extrapulmonary tuberculosis (95% confidence interval, ). The incidence of each AIDS-defining illness increased as the CD4 lymphocyte count declined; diagnoses such as cytomegalovirus and Mycobacterium avium intracellulare complex infection had a low incidence at CD4 lymphocyte counts above /L (50/mm 3 ), while Kaposi sarcoma, P carinii pneumonia, and esophageal candidiasis had a high incidence throughout all CD4 lymphocyte count strata. Conclusions: This study provides important information about the risk of AIDS-defining illnesses at lower CD4 lymphocyte counts, enabling disease-specific prophylaxis to be targeted at the most appropriate population. In the future, as more prophylactic therapies are developed, this study will provide historical data of the incidence of diseases before specific prophylaxis was introduced. Arch Intern Med. 1998;158: From the HIV Research Unit, Department of Primary Care and Population Sciences and Thoracic Medicine, Royal Free Hospital School of Medicine (Drs Mocroft, Phillips, and Johnson), the HIV/Genitourinary Medicine Research Unit, St Stephens Clinic (Drs Youle, Halai, and Gazzard), and the HIV Epidemiology Unit, Chelsea and Westminster Healthcare National Health Service Trust (Dr Easterbrook), London, England. A list of the members of the Royal Free/Chelsea and Westminster Hospitals Collaborative Group appears in the acknowledgment section at the end of the article. ANATURAL order of acquired immunodeficiency syndrome (AIDS) defining diseases, defined by the CD4 lymphocyte count, has been suggested in patient populations of various sizes. 1-3 Knowledge of the risk of a specific disease according to the level of CD4 lymphocyte counts allows clinicians to screen patients for the most likely diseases, given their CD4 lymphocyte count, and to target prophylaxis at the most relevant population. It is also important to establish if the risk of a particular disease varies according to basic demographic variables, such as age or sex, or if the risk of the disease continues to increase as the CD4 lymphocyte count decreases. A study by Bacellaretal 4 reportstheincidenceofopportunistic infections in patients with CD4 lymphocyte counts below /L (100/mm 3 ), but does not detail the incidence at very low CD4 lymphocyte counts ( /L). Trials of new agents for the prophylaxis of opportunistic infections are increasingly being undertaken. Disease prophylaxis can have a high cost and patients who need prophylaxis are those who require more intensive clinical monitoring. Therefore, it is of great importance to determine the optimal time to initiate prophylaxis and to estimate the cost and benefit obtained from such new interventions. 5 The Royal Free and Chelsea and Westminster Hospitals in London, England, have a large patient population and have data collected for up to 10 years. During this period, comprehensive data on more than 4500 patients infected with the human immunodeficiency virus (HIV), of whom 2500 were patients with AIDS, were collected. Information on all AIDSdefining illnesses and CD4 lymphocyte counts recorded during follow-up were 491

2 PATIENTS AND METHODS PATIENTS All patients diagnosed as having HIV and with at least 1 recorded CD4 lymphocyte count from the Chelsea and Westminster Hospital (CWH) from 1982 to July 1995 or the Royal Free Hospital (RFH) from 1986 to August 1994 were included in this study. Prospective data collection began at the CWH in 1986 and at the RFH in 1990; retrospective data for all patients with HIV ever seen at the 2 sites were added at this time. Several data exercises have since confirmed the accuracy and completeness of the data. In particular, at both sites a comprehensive search ensured that all patients were added to the database, including patients who had died or who were unavailable for follow-up before prospective data collection began. Demographic data, details of all AIDSdefining illnesses, date of death, and immunological data are maintained in a separate database at each site. A diagnosis of AIDS was made according to the definition in use at the time, and patients were not retrospectively diagnosed as having AIDS. Patient follow-up varied for each AIDS-defining illness and was measured as the time between the first CD4 lymphocyte count and the development of the AIDSdefining illness of interest. Patients who did not develop the AIDS-defining illness were eliminated from the analysis at death or at the date of their last clinic visit if they did not die during the study. In some cases, patients gave a history of an AIDSdefining illness at their first visit, and information from these illnesses was excluded from this analysis. This was because verifying the accuracy of historical diagnoses and completeness of CD4 lymphocyte counts at the time of diagnosis was mainly incomplete. For example, a patient diagnosed as having Pneumocystis carinii pneumonia as an initial AIDS-defining illness before his or her first visit to the RFH would not contribute any information to the analysis of the average CD4 lymphocyte count at diagnosis of P carinii pneumonia, or the incidence of disease in different CD4 lymphocyte count strata, even if it was diagnosed again at a later date while at the RFH. If this patient was then diagnosed as having cryptosporidiosis after 6 months of follow-up, this information would be included in the estimate of the median CD4 lymphocyte counts at which cryptosporidiosis occurs. Only the first occurrence of each AIDS-defining illness has been noted, since there are inherent difficulties in determining the end of one episode and the onset of a new, distinct episode. STATISTICAL METHODS All statistical analyses were performed using SAS software. 6 The median CD4 lymphocyte count at diagnosis was calculated for each of the AIDS-defining illnesses, and was compared in different groups using nonparametric tests, such as the Kruskal-Wallis test. Patients diagnosed as having progressive multifocal leukoencephalopathy (39 patients), histoplasmosis (6 patients), isosporiasis (9 patients), and recurrent Salmonella septicemia (10 patients) were grouped together to form a single category of diagnoses. The CD4 lymphocyte count was divided into several groups: higher than /L, /L, /L, /L, and /L or less. The incidence of each disease per 100 person-years (PY) of follow-up in each category was calculated using a PY analysis. Confidence intervals were constructed assuming a Poisson distribution of events. Follow-up in each strata begins when the patient s CD4 lymphocyte count falls in the range of the strata, and ends either with the development of an AIDS-defining illness, the death of the patient, or a decline in CD4 lymphocyte counts such that the patient moves to another strata. The incidence rate in each strata is then the number of events divided by the total PY of follow-up. Patients were not allowed to move in the reverse direction through the CD4 lymphocyte count categories; thus, the data presented are the incidence according to the minimum CD4 lymphocyte count measured. prospectively collected. To our knowledge, there has been no previous study that has determined the incidence of each AIDS-defining illness over a wide range of CD4 lymphocyte counts in such a large patient population. Therefore, the aim of this study was to determine the CD4 lymphocyte counts at which AIDS-defining illnesses are diagnosed in England, and to investigate the changing incidence of each of the AIDS-defining illnesses as the CD4 lymphocyte value declines. RESULTS In total, 4883 patients from the RFH (907 patients [18.6%]) and CWH (3976 patients [81.4%]) had at least 1 CD4 lymphocyte count measured during a median follow-up period of 27.6 months (90% range, months), measured from the date of the first hospital visit. Measurement of the CD4 lymphocyte count began in March 1983 at CWH and in February 1982 at RFH. The median number of CD4 lymphocyte count determinations during follow-up was 7 (90% range, 1-33). During follow-up, CD4 lymphocyte counts were measured more regularly at RFH; the median number of CD4 lymphocyte counts measured per year of follow-up was 7.8 at RFH and 3.4 at CWH (Wilcoxon P.001). Table 1 describes the patients included in this study. The majority of the patients were male (4513 [92.5%]) and belonged to the homosexual or bisexual exposure category (3959 [81.1%]). The median age at diagnosis of HIV was 31.9 years (90% range, years), while the median year of HIV diagnosis was The median CD4 lymphocyte count within 6 months of the initial visit was /L (90% range, /L), and this information was available for 3517 patients (72.0%). During the follow-up period, a total of 3875 incident AIDS-defining illnesses were diagnosed in 1713 patients. Figure 1 describes the AIDS-defining illnesses diagnosed in this group, together with the distribution of CD4 lymphocyte counts at diagnosis. This illustration is ordered according to the median CD4 lymphocyte count at diagnosis. The 5 most common AIDSdefining illnesses were, in decreasing order, P carinii pneumonia (649 patients [16.7%]), esophageal candidiasis (589 patients [15.2%]), Kaposi sarcoma (510 pa- 492

3 tients [13.2%]), Mycobacterium avium intracellulare complex (372 patients [9.6%]), and cytomegalovirus retinitis (367 patients [9.5%]). The median CD4 lymphocyte count at diagnosis ranged from /L in patients diagnosed as having cytomegalovirus retinitis to /L in patients diagnosed as having extrapulmonary tuberculosis. Figure 2 summarizes the incidence of each of the AIDS-defining illnesses, ordered from the most common diagnosis to the least. The incidence is expressed per 100 PY of follow-up. Patients who died were eliminated from analysis at the date of death; all other patients were eliminated from analysis at their last clinic visit. With the exception of pulmonary tuberculosis and Table 1. Description of the 4883 Patients From the Royal Free and Chelsea and Westminster Hospitals Collaborative Group Characteristics* No. (%) Site Royal Free Hospital 907 (18.6) Chelsea and Westminster Hospital 3976 (81.4) Sex M 4513 (92.5) F 364 (7.5) Age, y (14.6) (48.8) (25.9) (8.6) (2.1) Exposure category Homosexual or bisexual 3959 (81.1) Heterosexual 147 (3.0) Intravenous drug users 300 (6.1) Other 399 (8.2) Unknown 78 (1.6) Year of HIV diagnosis 1987 or earlier 978 (20.0) (20.0) (23.1) (23.8) (13.1) *HIV indicates human immunodeficiency virus. recurrent pneumonia, diagnoses added to the 1992 revision of the Centers for Disease Control and Prevention surveillance definition for AIDS, the total duration of follow-up exceeds PY of follow-up. The overall incidence ranges from 6.22 per 100 PY (95% confidence interval, ) for P carinii pneumonia to 0.37 per 100 PY for extrapulmonary tuberculosis (95% confidence interval, ). Table 2 presents the incidence of each AIDS-defining illness according to CD4 lymphocyte count. Generally speaking, the incidence of each AIDS-defining illness continued to increase as the CD4 lymphocyte count declined. Diseases such as cytomegalovirus and mycobacterial disease, which are less common as initial AIDS-defining illnesses, had a very low incidence until the CD4 lymphocyte count declined to below /L, while diseases such as Kaposi sarcoma, P carinii pneumonia, and esophageal candidiasis had a higher incidence at higher CD4 lymphocyte counts. There were some differences in the incidence of disease according to sex and exposure category, as shown in Table 3. This table has not been further divided according to CD4 lymphocyte count strata because the number of events in individual strata rapidly becomes very small. As expected, the incidence of Kaposi sarcoma was much higher in men compared with women (incidence, 4.96 and 0.91, respectively) and in homosexual men compared with all other exposure categories combined (incidence, 5.13 and 2.23, respectively). It is possible that patients included in the unknown exposure category were actually homosexual or bisexual men; this may explain the relatively high incidence of Kaposi sarcoma in the combination of other exposure groups. The incidence of esophageal candidiasis was also considerably higher among male patients compared with female patients (incidence, 5.53 and 2.64, respectively). There were also some differences in the incidence of AIDS-defining illnesses according to year of HIV diagnosis, as shown in Table 4 and Table 5. Table 4 presents the changing incidence of M avium intracellulare complex. The CD4 lymphocyte categories higher than /L have been combined due to the small number of events, as have years after There is a clear increase in the incidence of M avium intracellulare com th Percentile 75th Percentile Median 25th Percentile 5th Percentile CD4 Lymphocyte Count, 10 9 /L CMVR MAI CMVO Cryptosporidiosis Cryptococcosis TOXO HIVWS ESCAND HS PCP KS Other ADC Lymphoma Recurrent Pneumonia Diagnosis PULTB EPTB Figure 1. Acquired immunodeficiency syndrome (AIDS) defining illnesses in 1713 patients and CD4 lymphocyte counts at diagnosis. CMVR indicates cytomegalovirus retinitis; MAI, Mycobacterium avium intracellulare complex; CMVO, other cytomegalovirus infection; TOXO, toxoplasmosis; HIVWS, human immunodeficiency virus wasting syndrome; ESCAND, esophageal candidiasis; HS, recurrent herpes simplex infection; PCP, Pneumocystis carinii pneumonia; KS, Kaposi sarcoma; ADC, AIDS dementia complex; PULTB, pulmonary tuberculosis; and EPTB, extrapulmonary tuberculosis. 493

4 Incidence per 100 PY of Follow-up Upper 95% Confidence Limit Point Estimate Lower 95% Confidence Limit 0 PCP ESCAND KS MAI CMVR CMVO Recurrent Cryptosporidiosicoccosis Lymphoma HIVWS TOXO ADC Crypto- Pneumonia Diagnosis Other PULTB HS EPTB Figure 2. Incidence of each AIDS-defining illness, ordered from most common to least. PY indicates person-years. See legend for Figure 1 for an explanation of diagnosis abbreviations. Table 2. Incidence of AIDS-Defining Illnesses per 100 Person-years of Follow-up and CD4 Lymphocyte Count* CD4 Lymphocyte Count, 10 9 /L Illness Pneumocystis carinii pneumonia Esophageal candidiasis Kaposi sarcoma Mycobacterium avium intracellulare complex Cytomegalovirus retinitis Other cytomegalovirus disease Recurrent pneumonia Cyptosporidiosis Lymphoma HIV wasting syndrome Toxoplasmosis HIV encephalopathy Cryptococcosis Other AIDS-defining diagnoses Pulmonary tuberculosis Recurrent herpes simplex Extrapulmonary tuberculosis *AIDS indicates acquired immunodeficiency syndrome; HIV, human immunodeficiency virus. Patients were not allowed to move in the reverse direction through CD4 lymphocyte categories, even if their CD4 lymphocyte count increased. Therefore, these figures are the incidence according to the minimum CD4 lymphocyte count measured, expressed per 100 person-years of follow-up. AIDS-defining illnesses added to the Centers for Disease Control and Prevention surveillance definition in Includes histoplasmosis, isoporiasis, progressive multifocal leukoencephalopathy, and Salmonella septicemia. plex in patients diagnosed as having HIV after 1992 at lower CD4 lymphocyte counts, but at higher CD4 lymphocyte counts the incidence remained consistently low. Table 5 presents similar figures for P carinii pneumonia, and here the CD4 lymphocyte count categories lower than /L have been combined. After 1988, the incidence of P carinii pneumonia steadily decreased in the higher CD4 lymphocyte count categories, while at lower CD4 lymphocyte counts the incidence increased slightly. COMMENT It is clear that there are considerable differences in the CD4 lymphocyte count at which different AIDSdefining illnesses occur. Previous studies have described the distribution of CD4 lymphocyte counts according to AIDS-defining illnesses, 1-3,7 but these studies tend to be of much smaller numbers of patients, which means that the average CD4 lymphocyte count at which some of the less common diseases occurs could not be estimated. The order of disease occurrence in this study was consistent with those of other studies. This order ranges from diseases such as lymphomas and tuberculosis, which typically occur at higher CD4 lymphocyte counts, 1,7-9 to diseases such as M avium intracellulare complex 1,10-13 and cytomegalovirus disease, 7,12,14,15 which are infrequently observed in patients with CD4 lymphocyte counts higher than /L. This study reflects the incidence of AIDS-defining illnesses in a large population of HIV-positive patients. A study of patients with AIDS would find a higher overall incidence of disease than that reported here; in contrast, a study of seroconverted patients would report a 494

5 Table 3. Differences in Incidence of Disease Between Sex and Exposure Categories Category Person-years of Follow-up Cases Incidence* 95% Confidence Interval Kaposi sarcoma Sex M F Exposure category Homosexual or bisexual All others combined Esophageal candidiasis Sex M F *Incidence is expressed per 100 person-years of follow-up. Table 4. Changing Incidence of Mycobacterium avium intracellulare complex Over Time According to the Lowest CD4 Lymphocyte Count* CD4 Lymphocyte Count, 10 9 /L Year of HIV Diagnosis or earlier 0.75 ( ) 8.33 ( ) ( ) ( ) 6.08 ( ) ( ) ( ) 5.11 ( ) ( ) 1992 or later 0.53 ( ) ( ) ( ) *Incidence per 100 person-years of follow-up with 95% confidence intervals in parentheses. HIV indicates human immunodeficiency virus. Table 5. Changing Incidence of Pneumocystis carinii Pneumonia Over Time According to the Lowest CD4 Lymphocyte Count* CD4 Lymphocyte Count, 10 9 /L Year of HIV Diagnosis or earlier 1.74 ( ) 5.64 ( ) ( ) ( ) ( ) 4.23 ( ) ( ) ( ) ( ) 4.14 ( ) ( ) ( ) 1992 or later 0.66 ( ) 2.80 ( ) ( ) ( ) *Incidence per 100 person-years of follow-up with 95% confidence intervals in parentheses. HIV indicates human immunodeficiency virus. much lower overall incidence than reported here. This makes comparisons with other studies difficult, since few studies have reported the incidence of AIDSdefining diseases according to CD4 lymphocyte strata. Moore and Chaisson 3 report a higher incidence of disease than that found in this study, but this may be due to differences in study design and the duration of patient follow-up. In other studies, the incidence of disease has often been quoted as a proportion of patients who develop disease within 1 or 2 years of the CD4 lymphocyte count dropping below some arbitrary level Those studies that address the increasing incidence of disease as CD4 lymphocyte counts decrease report similar findings to this study, that is, that the incidence of disease increases as CD4 lymphocyte counts decrease, 19,20 although the categories of CD4 lymphocyte counts used in other studies are much wider than those used here. This study is unique in the number of patients included, the number of incident AIDS-defining illnesses, and its use of follow-up CD4 lymphocyte counts, which enable the occurrence of a disease to be related to the CD4 lymphocyte count at which it occurs more accurately. The incidence of all AIDS-defining illnesses was highly dependent on the CD4 lymphocyte count. This information may prove to be extremely useful in the development of prophylaxis and screening tests. For example, the incidence of cytomegalovirus retinitis in patients with a CD4 lymphocyte count higher than /L was low, hence prophylaxis may only be appropriate in patients with very low CD4 lymphocyte counts. Similarly, if a patient presented with symptoms suggestive of cytomegalovirus retinitis, measurement of the CD4 lymphocyte count may provide some indication of how likely the diagnosis is. The optimal time to initiate prophylaxis is a complex issue; prophylaxis is necessary when the disease is common, but the best improvement in outcome may be by preventing the disease at earlier CD4 lymphocyte counts. For the majority of the diseases, the incidence increased as the CD4 lym- 495

6 phocyte count declined. Late-stage diseases, such as cytomegalovirus retinitis and M avium intracellulare complex, had a very low incidence before the CD4 lymphocyte count dropped below /L. However, it should be noted that at CD4 lymphocyte counts between0.025 and /L, the incidence of, for example, cytomegalovirus was higher than the incidence of P carinii pneumonia at CD4 lymphocyte counts between 0.10 and /L. In contrast, diseases such as Kaposi sarcoma and esophageal candidiasis had a relatively high incidence throughout all CD4 lymphocyte count categories. GUIDELINES FROM the Centers for Disease Control and Prevention for the prophylaxis of opportunistic infections suggest that prophylaxis against P carinii pneumonia should begin at CD4 lymphocyte counts of /L, against toxoplasmosis at /L, and against M avium intracellulare complex at /L. 21 In this study, the incidence of P carinii pneumonia among all patients with a CD4 lymphocyte count below /L was 13.5 per 100 PY of follow-up, while the incidence of toxoplasmosis at CD4 lymphocyte counts below /L was much lower (3.33). At CD4 lymphocyte counts higher than /L, the overall incidence of both cytomegalovirus disease or M avium intracellulare complex was very low (0.49 and 0.58 per 100 PY of follow-up, respectively), which may suggest that a CD4 lymphocyte count of /L may be the most appropriate time to start prophylaxis for these infections. In this study, female patients were at a significantly reduced risk of esophageal candidiasis compared with males. This contrasts with previous reports, in which esophageal candidiasis has been reported more commonly in females The higher frequency reported in females previously has been attributed to the frequency of vaginal candidiasis observed in women infected with HIV. 24 There was also a reduced risk of Kaposi sarcoma in all exposure groups compared with homosexual men. Kaposi sarcoma has been reported to be up to 10 times more common in homosexual men, 17,26 and it has been reported in most other exposure groups to a certain extent. 27 A woman s risk of developing Kaposi sarcoma may be closely associated with her partner s exposure category. 27 Various studies now provide evidence that a newly identified human herpesvirus may be the cause of Kaposi sarcoma, and this may help to explain the high risk of Kaposi sarcoma in homosexual men. This study confirmed the increasing incidence of M avium intracellulare complex in patients diagnosed as having HIV in more recent years, but found no evidence of an increase in cytomegalovirus infections. 16,31-33 By preventing or delaying death from opportunistic infections such as P carinii pneumonia without stopping the decline in immune function, clinical manifestations more common with advanced immunosuppression become more prevalent. 4,34 In addition, improvements in diagnosing M avium intracellulare complex infection may explain its increase in recent years. It was also interesting to note that the incidence of P carinii pneumonia at higher CD4 lymphocyte counts declined steadily since 1988, consistent with the introduction of effective prophylaxis against P carinii pneumonia. 35,36 A potential bias of this study is that patients may not report their history of AIDS-defining illnesses accurately. If a patient had been diagnosed as having cryptosporidiosis but it was not reported at his or her first visit to RFH, that patient could be incorrectly included in an analysis of the average CD4 lymphocyte count for cryptosporidiosis at the first recurrence of this disease. If this bias was present, it would have the effect of lowering the CD4 lymphocyte counts at which each disease occurs, inflate the incidence of disease at lower CD4 lymphocyte counts, and decrease survival after a specific diagnosis. It is difficult to ascertain the extent to which patient self-reporting agrees with medical history. In some cases, patients bring documented evidence of their medical history, but this is not always the case. Previous studies in unrelated medical conditions have suggested that the accuracy of recall of medical histories depends on the diagnosis, the length of the recall period, and a potential reluctance to answer sensitive questions This patient group had a median survival after an initial AIDS diagnosis of 20 months, 40 which compares favorably with other reports of survival and may suggest that this potential bias is small. In addition, since both historical events and recurrences of illnesses were excluded from the analysis, the incidence rates should be regarded as minimum estimates. This study did not consider the relationship between prophylaxis against P carinii pneumonia, antiretroviral treatment, and the incidence of disease. An alternative method to investigate these relationships would be to use Cox proportional hazards models. The relationship between the risk of disease and specific factors, such as age, sex, exposure group, CD4 lymphocyte counts, and treatment, could be investigated in greater detail. In addition, information about HIV viral load may add to the results of this study and help to identify those patients at greatest risk of a specific AIDSdefining illness, but little information on this area is available to date. The clinical care and disease progression of patients with HIV and AIDS has been substantially altered by the development of protease inhibitors; patients in this study had little exposure to these drugs due to the cutoff date for the analysis. Initial evidence suggests that the incidence of AIDS-defining illnesses has been dramatically reduced by the use of protease inhibitors. 41 In the future it will be of great interest to compare the median CD4 lymphocyte counts at which AIDS-defining illnesses occur in patients who take protease inhibitors, and to confirm the declining incidence of opportunistic infections. In conclusion, this study provides important information about the incidence of AIDS-defining illnesses across a wide range of CD4 lymphocyte counts, enabling diseasespecific prophylaxis to be targeted at the most appropriate population. In the future, as more prophylactic therapies are developed, this study will provide important histori- 496

7 cal data on the incidence of diseases before specific prophylaxis was introduced. The cost of caring for patients with HIV will vary greatly according to the opportunistic infections that arise and how they are treated. This study provides important information about the incidence of disease, given a CD4 lymphocyte count, that can be used to guide decisions about the best time to begin prophylaxis, both in terms of cost and benefit. Accepted for publication July 17, Members of the Royal Free/Chelsea and Westminster Hospitals Collaborative Group include the following. At RFH: Margarita Bofill, Sanjay Bhagani, Francis Bowen, Tony Drinkwater, Angela Dykhoff, Jonathan Elford, Vince Emery, Debbie Farmer, Paul Griffiths, George Janossy, Margaret A. Johnson, Steve Jolles, Sabine Kinloch, Christine Lee, Clive Loveday, Marc Lipman, Sara Madge, Amanda Mocroft, Adeola Olaitan, Andrew Phillips, Caroline Sabin, and Mervyn Tyrer. At CWH: Simon Barton, Fiona Boag, Phillipa Easterbrook, Brian Gazzard, Ramesh Halai, David Hawkins, Adam Laurence, Julian Morcinek, Mark Nelson, and Michael Youle. Reprints: Amanda Mocroft, PhD, HIV Research Unit, Department of Primary Care and Population Sciences, Royal Free Hospital School of Medicine, Rowland Hill Street, London NW3 2PF, England. REFERENCES 1. Crowe SM, Carlin JB, Stewart KI, Lucas CR, Hoy JF. Predictive value of CD4 lymphocyte numbers for the development of opportunistic infections and malignancies in HIV-infected persons. J Acquir Immune Defic Syndr. 1991;4: Hanson DL, Chu S, Farizio KM, Ward JW, and the Adult and Adolescent Spectrum of HIV Disease Project Group. Distribution of CD4+ lymphocytes at diagnosis of acquired immunodeficiency syndrome defining and other human immunodeficiency virus related illnesses. Arch Intern Med. 1995;155: Moore RD, Chaisson RE. Natural history of opportunistic disease in an HIVinfected urban clinical cohort. Ann Intern Med. 1996;124: Bacellar H, Munoz A, Hoover DR, et al. Incidence of clinical AIDS in a cohort of homosexual men with CD4+ cell counts 100/mm 3. J Infect Dis. 1994;170: Gable CB, Tierce JC, Simison D, Ward D, Motte K. Cost of HIV+/AIDS at CD4+ counts disease stages based on treatment protocols. J Acquir Immune Defic Syndr. 1996;12: SAS Institute Inc. SAS/STAT Users Guide, Version 6. 4th ed. Cary, NC: SAS Institute Inc; 1989:2. 7. Luo K, Law M, Kaldor JM, McDonald AM, Cooper DA. The role of initial AIDSdefining illness in survival following AIDS. AIDS. 1995;9: Stoneburner R, Laroche E, Prevots R, et al. Survival in a cohort of human immunodeficiency virus infected tuberculosis patients in New York City: implications for the expansion of the AIDS case definition. Arch Intern Med. 1992;152: Korzeniewska-Kosela M, Fitzgerald JM, Vedal S, et al. Spectrum of tuberculosis in patients with HIV infection in British Columbia: report of 40 cases. Can Med Assoc J. 1992;146: Fiala M, Kermani V, Gornbein J. Role of CD8+ in late opportunistic infections of patients with AIDS. Res Immunol. 1992;143: Pertel P, Hirschtick R, Phair J, et al. Risk of developing cytomegalovirus retinitis in persons with the human immunodeficiency virus. J Acquir Immune Defic Syndr. 1992;5: Monforte AA, Mainini F, Vago L, et al. Factors associated with increased risk of acquiring cytomegalovirus disease in AIDS patients. J Infect Dis. 1993;168: Horsburgh CR Jr, Havlik JA, Ellis DA, et al. Survival of patients with acquired immune deficiency syndrome and disseminated Mycobacterium avium intracellulare complex infection with and without antimycobacterial chemotherapy. Am Rev Respir Dis. 1991;144: Baldassano V, Dunn JP, Feinberg J, Jabs DA. Cytomegalovirus retinitis and low CD4+ T-lymphocyte counts. N Engl J Med. 1995;333: Chin DP, Reingold AL, Stone EN, et al. The impact of Mycobacterium avium intracellulare complex bacteremia and its treatment on survival of AIDS patients: a prospective study. J Infect Dis. 1994;170: Gallant JE, Moore RD, Richman DD, Keruly J, Chaisson RE. Incidence and natural history of cytomegalovirus in patients with advanced human immunodeficiency virus disease treated with zidovudine. J Infect Dis. 1992;166: Gallant JE, Moore RD, Richmann DD, Keruly J, Chaisson RE. Risk factors for Kaposi s sarcoma in patients with advanced human immunodeficiency virus disease treated with zidovudine. Arch Intern Med. 1994;154: Lundgren JD, Melbye M, Pederson C, Rosenberg PS, Gerstoft J. Changing patterns of Kaposi s sarcoma in Danish acquired immunodeficiency syndrome patients with complete follow-up. Am J Epidemiol. 1995;141: Hirschtick RE, Glassroth J, Jordan MC, et al. Bacterial pneumonia in persons infected with the human immunodeficiency virus. N Engl J Med. 1995;333: Veenstra J, Krol A, van Praag RME, et al. Herpes zoster, immunological deterioration and disease progression in HIV-1 infection. AIDS. 1995;9: USPHS/IDSA guidelines for the prevention of opportunistic infections in persons infected with the human immunodeficiency virus: a summary. MMWR Morb Mortal Wkly Rep. 1995;44: Fleming PL, Ciesielski CA, Byers RH, Castro KG, Berkelman RL. Gender differences in reported AIDS-indicative diagnoses. J Infect Dis. 1993;168: Carpenter CCJ, Mayer KH, Stein MD, Leibman BD, Fisher A, Fiore TC. Human immunodeficiency virus infection in North American women: experience with 200 cases and a review of the literature. Medicine (Baltimore). 1991;70: Carpenter CJ, Mayer KH, Fisher A, Desai MB, Durand L. Natural history of acquired immunodeficiency syndrome in women in Rhode Island. Am J Med. 1989; 86: Greenberg AE, Thomas PA, Landesman SH, et al. The spectrum of HIV-1 related disease among outpatients in New York City. AIDS. 1992;6: Beral V, Peterman TA, Berkelman RL, Jaffe HW. Kaposi s sarcoma among persons with AIDS: a sexually transmitted infection? Lancet. 1990;335: Peterman TA, Jaffe HW, Beral V. Epidemiological clues to the etiology of Kaposi s sarcoma. AIDS. 1993;7: Whitby D, Howard MR, Tenant-Flowers M, et al. Detection of Kaposi s sarcoma associated herpes virus in peripheral blood of HIV-infected individuals and progression to Kaposi s sarcoma. Lancet. 1995;346: Su IJ, Hsu YS, Chang YC, Wang IW. Herpes virus like DNA sequences in Kaposi s sarcoma from AIDS and non-aids patients in Taiwan. Lancet. 1995;345: Chang Y, Cesarman E, Pessin MS, et al. Identification of new human herpesvirus like DNA sequences in AIDS-associated Kaposi s sarcoma. Science. 1994;266: Havlik JA, Horsburgh CR, Metchock B, Williams PP, Fann SA, Thompson SE. Disseminated Mycobacterium avium intracellulare complex infection: clinical identification and epidemiological trends. J Infect Dis. 1992;165: Montaner JSG, Le T, Hogg R, Ricketts M, Sutherland D, Strathdee SA. The changing spectrum of AIDS index diseases in Canada. AIDS. 1994;8: Munoz A, Carey V, Saah AJ, et al. Predictors of decline in CD4 lymphocytes in a cohort of homosexual men infected with human immunodeficiency virus. J Acquir Immune Defic Syndr. 1988;1: Hoover DR, Saah AJ, Bacellar H, et al. Clinical manifestations of AIDS in the era of Pneumocystis carinii pneumonia. N Engl J Med. 1993;329: Munoz A, Schrager LK, Bacellar H, et al. Trends in the incidence of outcomes defining acquired immunodeficiency syndrome (AIDS) in the Multicentre AIDS Cohort Study: Am J Epidemiol. 1993;137: Graham NM, Zeger SL, Park LP, et al. Effect of zidovudine and Pneumocystis carinii pneumonia prophylaxis on progression of HIV-1 infection to AIDS. Lancet. 1991;338: Boyer GS, Templin OW, Goring WP, et al. Discrepancies between patient recall and the medical record: potential impact on diagnosis and clinical assessment of chronic disease. Arch Intern Med. 1995;155: Parent ME, Ghadirian P, Lacroix A, Perret C. Accuracy of reports of familial breast cancer in a case-control series. Epidemiology. 1995;6: Githens PB, Glass CA, Sloan FA, Entmann SS. Maternity recall and medical records: an examination of events during pregnancy, childbirth and early infancy. Birth. 1993;20: Mocroft A, Youle M, Morcinek J, et al. Has survival after AIDS improved? evidence from 2625 patients with AIDS. BMJ. 1997;314: Egger M, Hirschel B, Francioli P, et al. Impact of new antiretroviral combination therapies in HIV-infected patients in Switzerland: prospective multicenter study. BMJ. 1997;315: