Epidemiology & Social
Epidemiological aspects of major opportunistic infections of the respiratory tract in persons with AIDS: Europe, 1993–2000
Serraino, Diego; Puro, Vincenzo; Boumis, Evangeloa; Angeletti, Claudio; Girardi, Enrico; Petrosillo, Nicolaa; Ippolito, Giuseppe
From the Department of Epidemiology, and aSecond Division of Infectious Diseases, National Institute for Infectious Diseases, L. Spallanzani, IRCCS, Rome, Italy.
Correspondence to D. Serraino, Department of Epidemiology, INMI L. Spallanzani, IRCCS, Via Portuense 292, 00149 Rome, Italy.
Received: 6 September 2002; revised: 20 November 2002; accepted: 8 April 2003.
Objective(s): To describe the epidemiology of Pneumocystis carinii pneumonia (PCP), pulmonary tuberculosis (PTB) and recurrent bacterial pneumonia (RBP) as AIDS-defining illnesses (ADI) in Europe.
Design: Analysis of AIDS surveillance data collected in the World Health Organization European region by EuroHIV, Saint Maurice, France.
Methods: Adult AIDS cases notified between 1993 and 2000 were studied. Since AIDS diagnosis may be constituted by up to four concurrent illnesses, polytomous logistic regression odds ratios (OR) and 95% confidence intervals (CI) were computed. Time trends and correlates of PCP, PTB or RBP were assessed.
Results: There were 181 296 ADI among the 142 447 AIDS cases included in this study. PCP was the commonest ADI in western Europe (17.8%) and PTB (20.4%) was the commonest ADI in eastern Europe. Within western Europe, PTB was more common in the south than in the north (OR, 1.5) and increased steadily over time. RBP increased until 1998 (from 1.9% to 3.7%) and thereafter declined. Young age was associated with an excess risk for PTB and, in comparison with heterosexuals, homosexual men were at higher risk for PCP (OR, 1.3). Injecting drug users (IDU) (OR, 2.8; 95% CI, 2.6–3.1) and recipients of blood (OR, 1.7; 95% CI, 1.4–2.2) were at increased risk for RBP.
Conclusions: This analysis highlighted the continuing importance of PCP and the increasing importance of PTB as an ADI in western Europe, and it emphasized the need to investigate more thoroughly the vast epidemic of AIDS-associated PTB in eastern Europe. IDU and recipients of blood should be considered as target groups for vaccination against RBP.
Pneumocystis carinii pneumonia (PCP) still constitutes a very common AIDS-defining illness (ADI) of the respiratory tract in western countries [1,2], although a decrease in the relative frequency of PCP as an ADI was consistently documented in the era of highly active antiretroviral therapies (HAART) [3–6]. Several hypotheses have been suggested for such a continuing occurrence of PCP, including lack of HIV care, failure of prophylactic regimens and drug resistance, but specific reasons are not yet known [4,7].
Pulmonary tuberculosis (PTB) and recurrent bacterial pneumonia (RBP) were included among ADI in 1993, when the expanded European AIDS case definition was adopted . Their impact on AIDS morbidity was investigated thereafter, with particular emphasis on changes across calendar periods [9–16]. Such studies have generally produced inconsistent results. Findings from EuroSIDA, a European multicentre observational cohort of approximately 7000 patients with HIV infection, documented a non-statistically significant decrease in the incidence of PTB between 1994 and 1999 . On the contrary, a non-statistically significant increase was seen for PTB between 1990 and 1998 in the UK . Diverging information has also been accumulated as regards the occurrence of RBP. An approximately 50% decrease in the frequency of RBP episodes was recorded in HIV-infected patients treated with HAART in Italy , whereas, in the UK, the incidence of AIDS-associated RBP has remained stable over time .
From a methodological viewpoint, studies on ADI rarely took into consideration that AIDS diagnosis may be constituted by several concomitant illnesses. Few researches, therefore, appropriately accounted for competing risks of AIDS-associated morbidity [17,18].
We, thus, tried to overcome such limitation using a multivariate polytomous approach to analyse AIDS surveillance data collected in the World Health Organization (WHO) European region from 1993 through 2000. The objective of this study was to highlight correlates and time trends of the three major opportunistic infections of the respiratory tract (i.e., PCP, PTB and RBP) as ADI included in the 1993 European AIDS case definition .
The last available version of the European Non Aggregate AIDS Data Set (ENAADS) (30 June 2001 update) was analysed . ENAADS contains anonymous information for each individual AIDS case fulfilling the 1993 European case definition recorded by the national surveillance systems of 37 countries from the WHO European region. Individual data was collected by EuroHIV (the European Centre for the Epidemiological Monitoring of AIDS, Saint-Maurice, Paris, France), according to a standard core of epidemiological information that, among others, included: country of report, sex, age and year at AIDS diagnosis; HIV transmission category; and up to four ADI diagnosed within 4 months from the first one. Detailed information on ENAADS has been already published . It is worth remembering here that ENAADS contains neither information on past treatment history with antiretroviral drugs, nor on CD4 cell counts [19,20].
This analysis was based on all ADI reported among adult and adolescent AIDS cases (i.e. ≥ 13 years old), diagnosed between January 1993 and December 2000 and notified up to June 2001 in the last version of ENAADS (AIDS cases diagnosed between January and June 2001 were not included to limit the effect of incomplete notification and of reporting delay). For the aims of the present analysis, the 26 ADI included in the 1993 European definition of adult case were categorized into five diagnostic groups: (i) PCP, (ii) PTB, and (iii) RBP (i.e., the major opportunistic infections of the respiratory tract); (iv) opportunistic infections other than PCP, PTB and RBP (OI) (i.e, candidiases; coccidioidomycosis; cryptococcosis; cryptosporidiosis; cytomegalovirus diseases; herpes simplex diseases; histoplasmosis; isosporiasis; Mycobacterium avium complex; mycobacterium tuberculosis, extrapulmonary; mycobacterium, other species; progressive multifocal leukoencephalopathy; recurrent salmonella septicaemia; toxoplasmosis of brain); and (v) conditions other than opportunistic infections (non-OI; i.e., invasive cervical cancer; HIV-related encephalopathy; Kaposi's sarcoma; Burkitt's lymphoma; immunoblastic lymphoma; primary lymphoma of the brain; wasting syndrome). Candidiasis of bronchi, trachea or lungs was not included among the major opportunistic infections of the respiratory tract because of the small number of diagnoses.
The total number of diagnoses of ADI, and the relative frequency of the five diagnostic groups mentioned above, were evaluated according to year at AIDS diagnosis, sex, age, HIV transmission category, and geographic area of origin. The WHO European region was divided into four areas: three areas of western Europe: (i) north (i.e., Denmark, Finland, Germany, Iceland Ireland, Netherlands, Norway, Sweden, and the UK); (ii) centre (i.e., Austria, Belgium, France, Luxembourg, and Switzerland); and (iii) south (i.e., Greece, Israel, Italy, Portugal, Spain, and Turkey); plus (iv) eastern Europe (i.e., Albania, Azerbaijan, Belarus, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Macedonia, Moldova, Poland, Romania, Slovakia, Slovenia, Ukraine, and Yugoslavia).
To take into account the fact that, in ENAADS, AIDS diagnosis may be constituted by up to four concomitant ADI, a polytomous logistic model was used . Such a model allowed us to estimate the proportions of the five diagnostic groups of ADI over the time period examined (1993 and 1994 were grouped together, to limit potential biases due to the application of the 1993 European case definition) and to estimate their association with selected characteristics (i.e., sex, age, geographic area of origin, HIV transmission category). Maximum-likelihood polytomous logistic regression (PLR) equations were fitted , adjusted for the all above-mentioned characteristics and for the five diagnostic groups (PCP, PTB, RBP, OI and non-OI). PLR odds ratios (OR) and 95% confidence intervals (CI) were computed.
Descriptive analysis were conducted using SPSS software , while STATA software was used to fit polytomous logistic models .
Between 1993 and December 2000, 142 447 adult/adolescents [(median age, 35 years; 113 364 males (79.6%), 29 083 females (20.4%)] were diagnosed with AIDS in the WHO European region. In 78.3% of these 142 447 cases, AIDS diagnosis consisted of one illness; of the remaining, 17.1% had two concurrent illnesses, 3.5% had three and 1.1% had four concurrent illnesses (data not shown). A total of 181 296 ADI were diagnosed, and they represent the object of the present analysis. Table 1 shows the observed frequency distribution of these 181 296 ADI according to the five diagnostic groups and geographic area of origin.
PCP (17.3% of all diagnoses) constituted the single commonest OI, followed by PTB (8.7%) and RBP (3.0%). A wide geographic heterogeneity emerged between western and eastern European countries. PCP was nearly three times more commonly diagnosed in western than in eastern Europe (i.e., 17.8% versus 5.5%). The contrary was seen for PTB (the most common ADI: 20.4% in eastern Europe) and RBP that were two- to threefold more frequently diagnosed in eastern Europe (Table 1). The overall frequency of the remaining OI was similar in the two areas, whereas conditions other than OI were more common in western Europe (Table 1).
Because of the wide heterogeneity in the frequency of ADI and in the use of HAART in eastern as compared with western European countries, the analysis of time trends and of potential determinants of ADI was restricted to AIDS cases diagnosed in western Europe.
Fig. 1 illustrates time trends of 173 944 ADI diagnosed between 1993 and 2000 among 137 325 AIDS cases in western Europe. The estimate proportions of ADI presented in Fig. 1 – adjusted for the five diagnostic groups of ADI plus age, sex, HIV-transmission category and three areas of western Europe – showed that the diagnosis of PCP tended to remain substantially stable over time, from 18.0% in 1993–1994 to 18.9% in 2000. RBP diagnoses increased between 1993–1994 (1.9%) and 1998 (3.7%), and started to decline thereafter. A marked and steady upward trend was observed in the diagnoses of PTB, that increased from 6.1% of all diagnoses in 1993–1994 to 10.5% in 2000 (Fig. 1). The OI declined as of 1998 (41.4%), and thereafter started to increase: they constituted the 42.9% of all diagnoses in 2000, whereas non-OI ADI started to decline after 1998 (from 26.4% to 24.5% in 2000) (Fig. 1).
Factors potentially associated with the diagnoses of PCP, PTB and RBP as ADI in western Europe are summarized in Table 2. Females had a nearly 30% lack of PTB and a nearly 30% excess of RBP, as compared with males. Ageing was inversely associated with both PTB (PLR-OR for ≤ 24 years versus ≥ 40 years, 2.0; 95% CI, 1.8–2.1) and, though at a lesser extent, with RBP (PLR-OR for ≤ 24 years versus ≥ 40 years, 1.2). PTB and RBP were more frequently diagnosed in southern than in northern Europe, but after adjustment a 1.5-fold statistically significant excess risk persisted for PTB only (Table 2).
The mode of acquisition of HIV infection was differently associated with ADI. As compared to heterosexuals, homosexual men were at higher risk for PCP (PLR-OR, 1.3) and at reduced risk for both PTB (PLR-OR, 0.5; 95% CI, 0.4–0.5) and for RBP (PLR-OR, 0.8; upper 95% confidence limit, 0.9). An opposite picture was seen among injecting drug users (IDU), since in such groups the risk of having RBP as an ADI was approximately three times higher than in heterosexuals (PLR-OR, 2.8; lower 95% confidence limit, 2.6), and a 2.3-fold excess risk for RBP was also registered among IDU who were homosexual men. IDU showed a slight reduction in the risk for PCP (PLR-OR, 0.8) and a slight excess risk for PTB (PLR-OR, 1.1) as compared with heterosexuals. Notably, individuals who acquired HIV infection through transfusions of blood or of blood derivatives were at a 55% reduced risk for PTB and to a 1.7-fold higher risk for RBP (lower 95% confidence limit, 1.4; Table 2).
This epidemiological evaluation focused on major OI of the respiratory tract included in the 1993 European case revision, when PTB and RBP were added to the list of ADI. PCP was the most common ADI (17.3% ), followed by PTB (8.7%) and RBP (3.0%). Some of the geographic and temporal patterns, and some individuals factors associated with PCP, PTB and RBP as ADI that emerged in this analysis were not previously noted and are worth stressing.
Large variations in the proportions of ADI were noted between AIDS diagnoses made in eastern or in western European countries and, particularly for PTB, within areas of western Europe. It is known that the frequency of tuberculosis (TB) in HIV-infected individuals largely reflects background incidence rates of TB in their country of origin , and this pattern was confirmed in the present analysis. In eastern Europe, where TB incidence rates in the general population in 1995–1999 were three- to 10-fold higher than the overall rate recorded in the general population of western Europe , PTB was the most common OI of the respiratory tract, accounting for about 20% of all ADI. Moreover, within western European countries, the highest proportion of PTB as ADI was registered in southern Europe, where Portugal, Spain and Turkey are located. In 1995–1999, these three countries showed the highest rates of TB in the general population of western Europe, including the fraction of the population under 50 years of age (i.e., the population at higher risk for HIV infection and AIDS) .
As compared with ADI recorded in western Europe, PCP was nearly three times less frequent and RBP was more than twice as frequent in eastern European countries. We are not aware of previous studies investigating geographic differences of PCP and RBP as ADI in Europe. In addition to differences in access to prophylaxis and antiretroviral therapies, the delay with which the AIDS epidemic has spread in eastern Europe, in respect to the western part of Europe , might have favoured some misdiagnoses between PCP, RBP and other early manifestations of AIDS .
Within western Europe, RBP seemed to be a more common ADI in the southern than in the northern part. Such geographic heterogeneity, however, disappeared at multivariate analysis when the effect of concurrent ADI and of other potential confounders was controlled for. Most of all, the apparent increased frequency of RBP in southern Europe was likely to reflect the predominance of cases attributable to injecting drug use in such areas .
Changes across calendar years in the percent frequency of ADI (as expressed in this analysis) reflect modifications in the relative burden of a particular ADI over the totality of ADI, but do not offer information on absolute changes over time (e.g., incidence rates from longitudinal investigations of HIV-infected persons). In such a context, it is worth noting the persistent importance of PCP (nearly one out of five ADI) and the increasing importance of PTB and RBP after their inclusion among ADI, from 6.1% to 10.5% and from 1.9%, to 3.1%, respectively, between 1993 and 2000.
In this analysis, adjustment for potentially concurrent components of AIDS diagnosis and for confounding factors did not influence the steady increase in the relative frequency of PTB over time. Such a strong upward trend was not seen for other ADIs, and adds further evidence to diverging findings on time trends of AIDS-associated PTB [3,16]. We have already stressed that information on two well-known important confounders of the relationship between calendar period and occurrence of ADI, i.e., CD4 cell count and past history of antiretroviral therapy, were not available in the data set used for this analysis [19,20]. However, when the effect of CD4 cell count and of antiretroviral therapy history were accounted for in previous investigations, inconsistent results were obtained [3,16]. Moreover, the diagnosis of PTB has been documented to occur at various levels of CD4 cell count [16,26], and it is likely that other factors, that we could not identify, might have favoured such a sharp increase in the relative frequency of PTB since its inclusion among ADI. Even in the context of wide availability of combination therapies, aggressive public health programs, including rapid case finding, monitoring of adherence to treatment and therapy for latent tuberculosis infection, will continue to play a critical role in the control of HIV-associated tuberculosis epidemic.
RBP as ADI increased across time periods characterized firstly by the wider use of dual antiretroviral drug therapy (i.e., 1995–1996), and then by the early use of HAART (i.e., 1997–1998), and started to decline thereafter. This pattern agrees with longitudinal data showing that the incidence of RBP in HIV-infected patients remained stable in the HAART era  or declined at a lower rate, as compared with PCP or other OI . As seen for PTB, RBP can occur over a wide spectrum of CD4 cell count levels , and this fact may help to explain, at least in part, changes in the relative frequency of ADI. Other explanations are plausible, including a reduction in the use of preventive drugs [e.g., such as trimethoprim–sulfamethoxazole (TMP–SMX) and macrolides] active against many of the microorganisms responsible for bacterial pneumonia following the introduction of HAART .
The number of persons diagnosed with AIDS and unaware of HIV infection has increased over time [28,29]. These persons cannot benefit by HAART and by the prophylaxis against OI, and, among them, increasing proportions of PCP have been documented [12,29]. In our analysis, we observed limited temporal fluctuations in the relative frequency of PCP as an ADI in western Europe. Thus, notwithstanding the long-term availability of effective prophylaxis with TMP–SMX and the use of HAART, PCP remains the most common OI of the respiratory tract. Specific reasons for this are not yet understood: our analysis assessed that this pattern was not influenced by competing causes of AIDS morbidity, nor by confounding factors such as area of origin or HIV-transmission category.
The effect of selected characteristics on the occurrence of PCP, PTB and RBP as ADI evaluated among cases diagnosed in western Europe put in evidence that age at AIDS diagnosis and the mode of acquiring HIV infection differently affected the occurrence of ADI.
According to previous studies reporting that young age was a potential determinant of PTB in the HIV setting [14,15], also in our analysis decreasing age was associated with an increased frequency of PTB. Such observation contrasts with the overall age-related pattern in the general population of western Europe, where incidence rates steadily increase with ageing , and may be partially attributable to the fact that most of HIV-infected persons develop TB as a consequence of a recent TB infection, rather than a reactivation of a latent infection. This hypothesis is also supported by findings from molecular epidemiology studies focusing on clusters of TB isolates from HIV-infected patients [13,30].
Two distinct patterns of risk for the ADI of the respiratory tract were identified, based on modes of HIV infection acquisition different than heterosexual intercourse.
Firstly, homosexual men had a statistically significant 27% excess risk of developing PCP and a strong reduction (i.e., 54%) in the risk of PTB. Cause-specific studies of ADI have already described that homosexual men are less likely to develop PTB than other groups of population at risk for HIV infection, such as IDU [16,31]. Such observation was confirmed in this study after adjustment for other potentially concurrent components of ADI, and, among others, for area of origin—one of the most important determinant of PTB. On the other hand, the increasing proportion of homosexual men among late testing AIDS cases [28,29] may help to explain the excess risk of PCP among such populations.
Secondly, and in contrast with the picture emerging among homosexual men, IDU were at reduced risk for PCP and at increased risk for RBP. It has been well documented that IDU are less likely to be late testing AIDS cases than are homosexual men [28,29], and this fact has clear implications in their reduced risk of PCP. As in early reports [9,11], we found that injecting drug use was the main determinant of RBP: IDU had a 2.8-fold increased risk for RBP over heterosexuals, and, interestingly, the risk of RBP in homosexual men who were also IDU was 2.3-fold higher than that of heterosexuals. That the blood route acquisition of HIV infection, as opposed to the sexual route, was strongly related to RBP occurrence was further evidenced by the nearly twofold higher frequency of RBP registered among recipients of blood or blood derivatives. This latter evidence adds to the already noted epidemiological association between use of injecting drugs and risk of RBP, but epidemiological explanations of these findings are still obscure [9–13]. At this regard, an interesting biological hypothesis came from an immunological study conducted in Spain, showing that the occurrence of RBP appeared to be related, in IDU, to humoral hyper stimulation . Altogether, the findings of this study suggest that persons who acquired HIV infection through needle sharing, or through transfusion of blood or of blood derivatives, are likely target groups for pneumococcal immunization to prevent RBP [32,33].
Specific drawbacks of AIDS surveillance data have been mentioned already (e.g., lack of information on CD4 cell count and on individual history of antiretroviral therapy), while other methodological drawbacks have been thoroughly discussed elsewhere [19,34]. It is worth remembering here that these surveillance data reflect only morbidity at the time of AIDS diagnosis, and thus they do not encompass the full spectrum of HIV-related diseases. Finally, the major strengths of this multinomial statistical approach should also be highlighted, i.e., the possibility to evaluate the effect of calendar period and of several correlates on the occurrence of potentially concurrent components of AIDS diagnosis.
In conclusion, this multinomial analysis of ADI provided a thorough assessment of the main epidemiological patterns of PCP, PTB and RBP as ADI. The importance of PCP as the leading OI persisted in western Europe throughout the last decade, despite availability of prophylaxis and HAART. The increasing importance of PTB, both in western and in eastern Europe, has serious implications for the prevention and control of the disease and, in general, for its public health implications. Blood-borne acquisition of HIV infection was strongly associated with the occurrence of RBP, and these data may be useful in the identification of target groups for pneumococcal vaccination.
The European Non-Aggregate AIDS Data Set (ENAADS), release AIDS0106.DAT, was prepared by EuroHIV—European Centre for the Epidemiological Monitoring of AIDS (Saint Maurice, France). Compilation of this data file was made possible by the continuing participation of clinical doctors in mandatory and voluntary national AIDS reporting systems.
The authors thank Michela Di Pasquale for editorial assistance.
Sponsorship: Supported by the Ministero della Salute, III Progetto Nazionale AIDS, grant 20C.15, and Ricerca Corrente INMI L. Spallanzani, IRCCS, Roma, Italy.
1. European Centre for the Epidemiological Monitoring of AIDS. HIV/AIDS Surveillance in Europe. Mid-year Report, number 63. Paris: European Centre for the Epidemiological Monitoring of AIDS; 2000.
2. Centers for Disease Control and Prevention. HIV AIDS Surveil Rep1997, 9:1–43.
3. Ives NJ, Gazzard BG, Easterbrook PJ. The changing pattern of AIDS-defining illnesses with the introduction of highly active antiretroviral therapy (HAART) in a London clinic. J Infect Dis 2001, 42(Suppl 2):134–139.
4. Lundberg BE, Davidson AJ, Burman WJ. Epidemiology of Pneumocystis carinii pneumonia in an era of effective prophylaxis: the relative contribution of non-adherence and drug failure. AIDS 2000, 14:2559–2566.
5. Kovacs JA, Gill VJ, Meshnick S, Masur H. New insights into transmission, diagnosis, and drug treatment of Pneumocystis carinii pneumonia. JAMA 2001, 286:2450–2460.
6. Abgrall S, Matheron S, Le Moing V, Dupont C, Costagliola D, for the Clinical Epidemiology Group from the French Hospital Database on HIV. Pneumocystis carinii pneumonia recurrence in HIV patients on highly active antiretroviral therapy: secondary prophylaxis. J Acquir Immune Defic Syndr 2001, 26:151–158.
7. Huang L, Hecht FM. Why does Pneumocystis carinii pneumonia still occur? AIDS 2000, 14:2611–2612.
8. Ancelle-Park RA. Expanded European AIDS case definition [letter]. Lancet 1993, 341:441.
9. Caiaffa WT, Graham NMH, Vlahov D. Bacterial pneumonia in adult population with human immunodeficiency virus (HIV) infection. Am J Epidemiol 1993, 138:909–922.
10. Hirschtick RE, Glassroth J, Jordan MC, Wilcosky TC, Wallace JM, Kvale PA, et al. Bacterial pneumonia in persons infected with the human immunodeficiency virus. N Engl J Med 1995, 333:845–851.
11. Tumbarello M, Tacconelli E, de Gaetano K, Ardito F, Pirronti T, Cauda R, et al. Bacterial pneumonia in HIV-infected patients: analysis of risk fatcors and prognostic indicators. J Acquir Immune Defic Syndr Hum Retrovirol 1998, 18:39–45.
12. Tumbarello M, Tacconelli E, de Gaetano Donati K, Cauda R. HIV-associated bacterial pneumonia in the era of highly active antiretroviral therapy. J Acquir Immune Defic Syndr Hum Retrovirol 1999, 20:208–209.
13. Girardi E, Raviglione MC, Antonucci G, Godfrey-Faussett P, Ippolito G. Impact of the HIV epidemic on the spread of other diseases: the case of tuberculosis. AIDS 2000, 14(Suppl 3): S47–S56.
14. Mayoral Cortes JM, Garcia Fernandez M, Varela Santos MC, Fernandez Merino JC, Garcia Leon J, Herrera Guibert D, et al. Incidence of pulmonary tuberculosis and HIV coinfection in the province of Seville, Spain 1998. Eur J Epidemiol 2001, 17: 737–742.
15. Del Amo J, Petruckevitch A, Phillips AN, De Cock KM, Stephenson J, Desmond N, et al. Risk factors for tuberculosis with AIDS in London: a case-control study. Int J Tuberc Lung Dis 1999, 3:12–17.
16. Kirk O, Gatell JM, Mocroft A, Pedersen C, Proenca R, Brettle RP, et al. Infections with Mycobacterium tuberculosis and Mycobacterium avium among HIV-infected patients after the introduction of highly active antiretroviral therapy. Am J Respir Crit Care Med 2000, 162:865–872.
17. Pezzotti P, Serraino D, Rezza G, Dal Maso L, Vaccher E, Lepri AC, et al. The spectrum of AIDS-defining diseases: temporal trends in Italy prior to the use of highly active anti-retroviral therapies, 1982-1996. Int J Epidemiol 1999, 28:975–981.
18. Cascade Collaboration. Changes over calendar time in the risk of specific first AIDS-defining events following HIV seroconversion, adjusting for competing risks. Int J Epidemiol 2002, 31:951–958.
19. Serraino D, Dal Maso L, La Vecchia C, Franceschi S. Invasive cervical cancer as an AIDS-defining illness in Europe. AIDS 2002, 16:781–786.
20. Semaille CH, Downs AM, Hamers FF, for EuroHIV Steering Group. European HIV/AIDS case reporting: assessing past treatment history in individuals who develop AIDS in the era of highly active antiretroviral therapy. AIDS 2003, 17:270–272.
21. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: John Wiley & Sons; 1989:216–238.
22. SPSS Inc. SPSS Base 11.0 for Windows User's Guide. Chicago: SPSS Inc.;IL, 2001.
23. StataCorp. Stata Statistical Software: Release 7.0. College Station, TX: Stata Corporation; 2001.
24. Narain JP, Raviglione MC, Kochi A. HIV-associated tuberculosis in developing countries: epidemiology and strategies for prevention. Tuber Lung Dis 1992, 73:311–321.
25. EuroTB (InVS/KNCV) and the national coordinators for tuberculosis surveillance in the WHO European Region. Surveillance of Tuberculosis in Europe. Report on tuberculosis cases notified in 1999. March 2002.
26. Law MG, de Winter L, McDonald A, Cooper DA, Kaldor JM. AIDS diagnoses at higher CD4 counts in Australia following the introduction of highly active antiretroviral treatment. AIDS 1999, 13:263–269.
27. Ledergerber B, Mocroft A, Reiss P, Furrer H, Kirk O, Bickel M, et al. Discontinuation of secondary prophylaxis against Pneumocystis carinii pneumonia in patients with HIV infection who have a response to antiretroviral therapy. Eight European Study Groups. N Engl J Med 2001, 344:168–174.
28. Girardi E, Sampaolesi A, Gentile M, Nurra G, Ippolito G. Increasing proportion of late diagnosis of HIV infection among patients with AIDS in Italy following introduction of combination antiretroviral therapy. J Acquir Immune Defic Syndr 2000, 25:71–76.
29. Dore GJ, Li Y, McDonald A, Ree H, Kaldor JM, Kaldo JM. Impact of highly active antiretroviral therapy on individual AIDS-defining illness incidence and survival in Australia. J Acquir Immune Defic Syndr 2002, 29:388–395.
30. Moro ML, Salamina G, Gori A, Penati V, Sacchetti R, Mezzetti F, et al. Two-year population-based molecular epidemiological study of tuberculosis transmission in the metropolitan area of Milan, Italy. Eur J Clin Microbiol Infect Dis 2002, 21:114–122.
31. Dupon M, Texier-Maugein J, Leroy V, Sentilhes A, Pellegrin JL, Morlat P, et al. Tuberculosis and HIV infection: a cohort study of incidence and susceptibility to antituberculous drugs, Bordeaux, 1985-1993. Groupe d'Epidemiologie Clinique du SIDA en Aquitaine. AIDS 1995 9(Suppl 6):577–583.
32. Payeras A, Martinez P, Mila J, Riera M, Pareja A, Casal J et al. Risk factors in HIV-1-infected patients developing repetitive bacterial infections: toxicological, clinical, specific antibody class responses, opsonophagocytosis and Fc(gamma) RIIa polymorphism characteristics. Clin Exp Immunol 2002, 130: 271–278.
33. Pierce AB, Hoy JF. Is the recommendation for pneumococcal vaccination of HIV patients evidence based? J Clin Virol 2001, 22:255–261.
34. Dal Maso L, Serraino D, Franceschi S. Epidemiology of HIV-associated malignancies in developed and developing countries. Eur J Cancer 2001, 37:1188–1201.
AIDS; PCP; Europe; polytomous logistic regression; pulmonary tuberculosis; recurrent bacterial pneumonia
© 2003 Lippincott Williams & Wilkins, Inc.
Highlight selected keywords in the article text.