- Email: [email protected]
Incidence and risk factors for pneumonia in HIV infected and non-infected drug users
Journal of Infection (1996) 32, 181-186
Incidence and Risk Factors for Pneumonia in HIV infected and Non-infected Drug Users G. H. C. Mientjes, I. J. B. Spijkerman*, E. J. C. van Ameijden, J. A. R. van den Hoek and R. A. Coutinho Municipal Health Service, Department of Public Health and Environment, Amsterdam, The Netherlands Accepted for publication 1 November •995
Objective: To study the incidence and risk factors for pneumonia in a cohort of HIV infected and non-infected drug users
(DU). Design: A prospective epidemiological study. Setting and patients: Injecting and non-injecting DU who attended the Municipal Health Service in Amsterdam for followup visits in the stu@. Main results: 203 HIV infected and 437 non-infected DU were followed for a total of 1749 person-years. HIV infected DU reported 111 episodes of pneumonia, which required hospitalization in 29 cases, and HIV negative DU reported 55 episodes, which required hospitalization in nine cases. The incidences among HIV positive and HIV negative DU were O. 19 and 0.05 per person-year respectively. With multivariate Poisson regression current injecting (RR 2.13), recent seroconversion (RR 3.92), asthmatic constitution (RR 2.72), CD4+ ceil count between 200-500 (RR 1.67 compared to >500), CD4+ cell count less than 200 (RR 2.23 compared to >500) and a previous history of pneumonia (RR 2.43) were independently associated with self-reported pneumonia among HIV infected DU. Among HIV negative DU heroin smoking (RR 1.87), asthmatic constitution (RR 3.62) and a previous history of pneumonia (RR 2.84) were independently associated with self-reported pneumonia. Also a higher Quetelet Index (OI) appeared to be protective (QI>_21 RR 0.42, QI 19-21 RR 0.82 compared to 0I<19) among HIV negative DU. Risk factors for reported and hospitalized cases of pneumonia were comparable among HIV positive DU. Conclusions: HIV infected DU are at increased risk for pneumonia and the rate increases with lower CD4 cell counts. Also behavioural characteristics, such as injecting drug use and smoking heroin, and clinical history variables, such as a history of pneumonia or an asthmatic constitution, are risk factors for pneumonia among DU. Pneumococcal vaccination should not only be focused on HIV positive DU but also on the identified risk groups among I-IIV negative DU.
Introduction Pneumonia is observed more often among HIV infected persons compared to non-HIV infected persons and the incidence a m o n g HIV infected persons increases with a declining n u m b e r of CD4 + cells in the asymptomatic phase of HIV infection. 14 For this reason recurrent bacterial pneumonia has been included in the AIDS definition, both in the United States and Europe. Among injection drug users, whether HIV infected or not, the incidence of pneumonia is higher compared to the incidence a m o n g for example homosexual men. s'6 Possibly, behavioural and social factors increase the risk
* Address correspondence to: I. J. B. Spijkerman, Municipal Health Service, Department of Public Health and Environment, P.O. Box 20244, 1000 HE Amsterdam, The Netherlands. This study was supported by: the Netherlands Foundation for Preventive Medicine (grant no. 28-1258). 0163-4453/96/030181 + 06 $12.00/0
for pneumonia among drug users (DU). However, studies on risk factors for pneumonia among DU are sparse. Here we report on the incidence and risk factors for pneumonia among DU in Amsterdam stratified by HIV serostatus.
Methods In December 1985, we began a prospective study to assess the prevalence and incidence of HIV infection, to determine risk factors for HIV infection, and to evaluate the natural history of HIV infection among DU. The study design has been described extensively in previous records. 1'4'S'7 Enrolment in the study is ongoing. Participants are recruited at low-threshold methadone posts and at a sexually transmitted diseases clinic for drugusing prostitutes. Participants are asked to return every 4 months for repeated interviews, blood sampling, and © 1996 The British Society for the Study of Infection
182
Risk Factors for Pneumonia
physical examination but participants occasionally return after longer periods.
Present study The present study was designed to determine the incidence and risk factors for pneumonia. The medical history is taken by a physician. At every visit participants were asked whether they suffered from pneumonia. If the answer was positive, standardized questions were asked about complaints, diagnostics and treatment. All cases of pneumonia which were diagnosed by a physician and treated with antibiotics are considered as reported pneumonia. Reported episodes of bronchitis or upper respiratory tract infections are not included. When a participant reported to be hospitalized for pneumonia, the medical specialist in attendance was contacted to obtain the reason for hospitalization. For the analysis of hospitalized cases of pneumonia, only verified cases were included, which showed consolidation on a chest X-ray. We have analysed self-reported cases of pneumonia and hospitalized cases of pneumonia separately. At intake pneumonia is reported over the period between 1980 and intake. At follow-up visits pneumonia is reported over the period since the previous visit. To reduce possible recall bias we have excluded intake visits and included only visits for which the interval between visits was less than 1 year. In our cohort, lymphocyte counts have been determined routinely since April 1989. Because the CD4 count was shown to be an important risk factor for pneumonia 4' 8,9 only visits after April 1989 were included in the analysis. Follow-up ended February 1994.
housing (yes/no), engaging in prostitution (yes/no), previous history of pneumonia since 1980 (yes/no), injecting drugs (yes/no), frequency of injecting (once per day), drug mainly injected (heroin, cocaine, heroin and cocaine, rest), smoking heroin (yes/no), smoking cocaine (yes/no), alcohol use (4 units, >4 units), use of tranquilizers (yes/no) or barbiturates (yes/no), dose of methadone (0, 1-35 mg, 3 6 - 7 0 mg, >70 mg), history of zidovudine use (yes/no), history of PEP prophylaxis (yes/no) and Quetelet index (defined as weight/length2: <19, 19-21, >21). When participants had used bronchodilator drugs at any time during their follow-up, since April 1989, they were considered to have an asthmatic constitution during their entire follow-up period (yes/no). Because seroconversion is associated with pneumonia, 5 the variable "recent seroconversion" is taken into account, which compares the first HIV positive visit after an HIV negative visit to all further HIV positive visits from seroconverters and prevalent HIV positive participants.
Laboratory tests Sera were tested for antibodies to HIV in a commercial ELISA (Vironostika Teknika, Organon, Oss, The Netherlands). HIV infection was confirmed by immunoblotting and/or competitive ELISA. Peripheral blood mononuclear cells were isolated from heparinized venous blood by density gradient centrifugation on Ficoll-Hypaque. Absolute numbers of CD4 + and CD8 + lymphocytes were determined by cytofluorometry.4
Statistics Definition of variables Preliminary analysis showed that pneumonia may influence the behaviour in the period between two visits in which the participant suffered from pneumonia. A participant who has been seriously ill may, for example, reduce the frequency of injecting and may receive daily methadone in higher doses during his/her illness or hospitalization, than participants without pneumonia. For this reason we have used behavioural and demographic variables from the visit preceding the visit on which the episode of pneumonia was reported. Immunological markers may also be disturbed by the occurrence of pneumonia and therefore we also used immunological data from the visit preceding the episode of pneumonia. Independent variables studied were: sex, age ( < 3 3 , >_ 34), HIV status, nationality (Dutch, foreign), number of CD4 cells (<200, 2 0 0 - 5 0 0 , >500 x 106/mm3), steady
Incidences were calculated for self reported and hospitalized cases of pneumonia by dividing the number of pneumonia by the sum of the total follow-up period, stratified to the level of a specific variable. Analysis was performed separately for HIV positive and HIV negative participants. With Poisson regression the univariate relative rates were calculated. To determine independent predictors, all univariately significant variables (P<0.05) were entered in Poisson regression models and backward selection was used. Interaction between the variables in the final model was checked. Because the number of hospitalized episodes of pneumonia among HIV positive DU was small and because we were interested whether the Rate Ratios for a specific variable differed for self reported and hospitalized episodes, we have forced the same variables, which were multivariately associated with self reported episodes of pneumonia, into the multivariate model of hospitalized pneumonia.
G. H. C. Mientjes et al. Table 1. General characteristics of drug users in the study by HIV serostatus HIV positive HIV negative (n=203) (n=437) Male sex (%) Mean number of visits (S.D.) Mean interval between visits in months (S.D.) Dutch nationality (%) Prostitution previous period (%) Injected previous period (%) First injection>lO years ago (%) Having permanent housing (%) CD4+cell count<500.109/1 (%)
123 (60.6) 257 (58.8) 7.3 (4.4) 6.7 (4.4) 4.7 (1.6) 138 (68.0) 56 (27.6) 171 (84.2) 111 (55.2) 149 (73.4) 108 (57.1)
4.8 (1.6) 336 (76.9)* 132 (30.2) 255 (59.1)* 141 (32.6)* 345 (78.9) 25 (6.0)*
* P<0.05 compared to HIV positive participants.
Because for each participant more than one observation could be included in the analyses, observations m a y be non-independent. To adjust for this possible intraperson correlation of data, Rate Ratios (RR) and 95% confidence intervals were not only determined with Poisson regression, but also by generalized estimation equations (GEE) using a Poisson link function. ~°'u These repeated measures analyses were performed both with independent and with exchangeable correlation structures.
Results A total of 203 HIV positive and 437 HIV negative drug users were in follow-up in the period between April 1989 and December 1993 and were eligible for the present study. The general characteristics at the first visit after April 1989 are summarized in Table I. The total followup period for HIV positive subjects was 578 years and for HIV negative subjects 1171 years. HIV positive subjects were more likely to have injected in the preceding period and to have had a previous history of pneumonia but were less likely to have smoked drugs compared to HIV negative subjects. During follow-up 111 episodes of pneumonia were reported by 65 HIV positive DU (35 DU: 1 episode; 19 DU: 2 episodes; 7 DU: 3 episodes; 3 DU: 4 episodes and 1 DU: 5 episodes). Of these 111, 29 episodes of pneumonia required hospitalization (21 DU: 1 episode and 4 DU: 2 episodes). HIV negative DU reported 55 episodes (38 DU: 1 episode; 5 DU: 2 episodes; 1 DU: 3 episodes and 1 DU: 4 episodes). Of these 55, nine episodes of pneumonia required hospitalization (9 DU: once). The incidence of self reported pneumonia a m o n g HIV positive DU was 0.19 per person-year and a m o n g HIV negative DU 0.05 per person-year.
183
Of the 29 hospitalized episodes among HIV positive DU the causative microorganism was identified by sputum culture in 20 cases. Haemophilus influenzae was identified 10 times, Streptococcuspneumonia four times, Haemophilus influenzae and Streptococcus sp. were cultured three times, Enterobacter sp. twice and Streptococcus sp. once. Univartate Poisson regression showed that among HIV positive participants, current injecting, an asthmatic constitution, a previous history of pneumonia, recent seroconversion, a lower Quetelet Index and a lower CD4 + cell count were significantly and positively associated with self reported pneumonia (Table II). In the same group we found that asthmatic constitution, recent seroconversion and nationality were also associated with hospitalized episodes of pneumonia. Among HIV negative participants, female sex, heroin smoking, asthmatic constitution, a previous history of pneumonia and a lower Ouetelet Index were significantly and positively related with self reported pneumonia (Table II). Due to small numbers of hospitalized pneumonia among HIV negative DU, these were not analyzed. Table III shows the results of the multivariate Poisson regression for self-reported and hospitalized pneumonia a m o n g HIV infected DU. Among HIV positive DU injecting drugs, recent seroconversion, a lower level of CD4 + cell count, asthmatic constitution and a previous history of pneumonia was independently associated with self reported pneumonia. When we forced the same variables, which were multivariately associated with self reported episodes of pneumonia, into the multivariate model of hospitalized pneumonia, we found that recent seroconversion, asthmatic constitution and a lower CD4 + cell count were associated with hospitalized episodes of pneumonia. Because nationality was significantly associated with hospitalized cases in the univariate analysis and did not substantially alter the RR for the other variables in the model, we present the multivariate model for hospitalized episodes of pneumonia including nationality (Table III). Among HIV negatives self reported cases were multivariately associated with smoldng heroin, asthmatic constitution, a previous history of pneumonia and a lower Quetelet Index (Table IV). The results of the analysis with Generalized Estimation Equations (see methods) revealed only minor differences from the results obtained with Poisson Regression (data not shown).
Discussion As was shown in several studies, HIV positive DU have a nearly 4-fold higher risk of pneumonia than HIV negative DU 1'2.12 and this risk increases with lower levels
184
Risk Factors for Pneumonia
TabLe II. Incidence and univariate rate ratios for pneumonia among drug users by HIV serostatus Reported cases HIV-positive
Sex
Male Female No Yes No Yes No Yes No Yes No Yes >500 200-500 <200 <19 19-21 >21 Dutch Foreign
Current injecting Heroin smoking Asthmatic constitution Ever pneumonia before Recent seroconversion* CD4 count Ouetelet Index Nationality
INC
RR
18.4 19.5 9.5 21.6 19.4 17.3 15.9 49.0 10.6 29.0 18.2 55.6 13.0 20.1 26.5 22.9 18.8 11,6 17.0 22,7
1,00 1.06 1.00 2.28 1.00 0.89 1.00 3.07 1.00 2.74 1.00 3.03 1.00 1,55 2,05 1,00 0,82 0,52 1.00 1.34
(95% CI)
(0.72-1.55) (1.28-4.06) (0.58-1.37) (1.97-4.80) (1.84-4.09) (1,25-7.69) (0.98-2.45) (1.18-3,54) (0.53-1,26) (0.30-0.87) (0.91-1.97)
Hospitalized cases HIV-positive INC
RR
4.9 4.9 2.9 5.5 4.9 4.9 4.1 13.7 3.6 6.5 4.3 44.4 3.4 4.6 8,8 5.9 3,8 4,5 3.3 8.7
1.00 0.99 1.00 1.89 1.00 1,O0 1,00 3.36 1.00 1.79 1.00 10.O1 1.00 1.37 2.30 1.00 0.64 0.77 1.00 2.7
Reported cases HIV-negative
(95% CI)
(0.47-2.09) (0.66-5.43) (0,44-2.28) (1.44-7,38) (0.85-3.74) (3.57-105.4) (0,55-3.42) (0.81-6.56) (0.26-1.26) (0.31-1.93) (1.30-5.60)
INC
RR
(95% CI)
3.5 7.7 3.8 6.4 3.7 7.4 4.2 14.8 3.2 10,6 -----6.8 5.9 2.9 5.5 4.3
1.00 2.18 1.00 1.70 1.00 2,00 1,O0 3.52 1.00 3.34 -----1.O0 0.87 0.43 1.00 0.79
(1.27-3.74) (0.97-2.97) (1.17-3.41) (1,94-6.36) (1.96-5.68) -----(0.45-1.66) (0.21-0.87) (0.39-1,61)
Univariate rate ratios (RR) and 95% Confidence Intervals (95% CI) were calculated with Poisson regression. Behaviour, CD4 count and Quetelet Index were measured at the visit preceding the report of an episode of pneumonia. Recent seroconversion is the first HIV positive visit after an HIV negative visit. * The first HIV positive visit is compared to all further HIV positive visits from seroconverters and prevalent HIV positive participants (methods).
Table III. Risk factors for reported (n = 111) and hospitalized (n = 29) pneumonia in multivariate Poisson regression among HIV infected drug users Reported episodes
RR
95%
1.87 3.62 2.84 1.00 0.82 0.41
(1.06-3.19) (1.91-6.85) (1.60-5.03)
Hospitalized episodes
RR
95% CI
RR
2.13 3.92 2,72 1.00 1.67 2.23
(1.16-3.92) (1.52-10.2) (1.73~t.27)
2.28 (0.67-7.77) 25.1 (6.26-10.1) 3.29 (1.36-7.92)
(1,25-3.97) (1,25-3.97)
Ever pneumonia before 2.43 Foreign nationality
(1.60-3.68)
5.61 (1.51-20.9) 5.61 (1,51-20.9) 1.70 (0.76-3.79) 3.12 (1.45-6.71)
Current injecting Recent seroconversion Asthmatic constitution CD4 count>500 CD4 count 200-500 CD4 count<200
Table IV. Risk factors for reported pneumonia (n=55) in multivariate Poisson regression among HIV negative drug users
95% CI
RR=rate ratio, 95% CI= 95% Confidence Interval. All variables measured at the last visit preceding the episode of pneumonia. For the analysis of the hospitalized cases all variables significant in the analysis of the reported episodes were included, together with nationality.
of C D 4 cells. 4' s W e f o u n d t h a t a s t h m a t i c c o n s t i t u t i o n is a r i s k f a c t o r for p n e u m o n i a , w h i c h is w e l l I m o w n . G h o d s e a n d H y l e s 13 h a v e s h o w n t h a t , i n a s a m p l e of 2 2 7 6 o p i a t e a d d i c t s , t h e u s e of h e r o i n m a y inflict a s t h m a t i c r e a c t i o n s , e s p e c i a l l y i n w o m e n . W e also f o u n d t h a t a s t h m a t i c c o n s t i t u t i o n w a s m o r e p r e v a l e n t i n w o m e n . Of all w o m e n
Heroin smoking Asthmatic constitution Ever pneumonia before Quetelet Index<19 Quetelet Index 19-21 Onetelet Index>21
(0.42-1.57) (0.20-0.84)
RR=rate ratio, 95% CI= 95% Confidence Interval. All variables measured at the last visit preceding the visit at which the episode of pneumonia was reported.
in the study, 12.7% had an asthmatic constitution, comp a r e d to 4 . 5 % of t h e m e n ( P = 0 . 0 0 0 1 4 ) . H o w e v e r , s i n c e w e h a v e classified a s t h m a t i c c o n s t i t u t i o n i n a v e r y r o u g h way (anyone who received anti-asthmatics during any t i m e d u r i n g f o l l o w u p w a s classified as h a v i n g a n a s t h matic constitution during their entire follow-up period) t h e v a l i d i t y of t h i s f i n d i n g is difficult t o assess. B e y o n d t h a t , it is also p o s s i b l e t h a t DU w i t h p n e u m o n i a a r e m o r e i n c l i n e d to r e c e i v e a n t i - a s t h m a t i c s t h a n DU w i t h o u t pneumonia. Therefore the relation between pneumonia and asthmatic constitution may be biased. The rate ratio, h o w e v e r , is h i g h a n d a t l e a s t g i v e s a n i n d i c a t i o n t h a t
G. H. C. Mientjes et al. HIV positive drug users with asthma m a y be at increased risk for pneumonia. A previous history of pneumonia was also strongly associated with pneumonia. This is consistent with reports indicating that pre-existing lung disease is associated with pneumonia. 14 Our earlier observation that DU who seroconverted are at increased risk to acquire pneumonia shortly after seroconversion was confirmed in the present study, s The risk factors for pneumonia a m o n g HIV positive subjects did not differ substantially between self reported and hospitalized episodes, though recent injecting and a previous history of pneumonia were not significantly associated with hospitalized episodes. This was probably due to the smaller n u m b e r of hospitalized episodes, as the rate ratios did not differ substantially. Nationality was associated only with hospitalized pneumonia and not with self-reported cases. This is probably caused by the fact that non-Dutch participants are more often not insured and have less access to primary health care. They also live more often under worse social conditions because they have less access to housing facilities and do not receive welfare. Pneumonia a m o n g these non-Dutch participants might often be diagnosed at a later stage of the disease when hospitalization is more likely to be necessary. A new finding is that current drug use-related behaviour was associated with pneumonia. Among HIV positive subjects injecting drugs, as opposed to not injecting drugs, gave a rate ratio of 2.13 for reported and 2.28 for hospitalized episodes of pneumonia. Since recurrent pneumonia is included in the revised AIDS definition, this means that current behaviour m a y influence the incidence of AIDS. The exact cause for the increased risk to acquire pneumonia among injectors is not clear. The higher risk of current injectors could be due to the (not measured) worse social conditions of active injectors. A risk factor for pneumonia a m o n g HIV negatives was smoking heroin. This was already suggested by physicians who daily work with addicts (van Brussel, personal communication). Possibly smoking heroin decreases the cough reflex and ciliary activity. TM ~s Recently a study in the U.S.' has shown that smoking cocaine or m a r i h u a n a increases the risk of pneumonia in HIV positive DU. 12 Smoking heroin appeared to be no risk factor a m o n g HIV positives, probably because they smoked drugs less frequently. There are some possible sources of bias in the incidence of pneumonia, as we measured it. At first we have included only pneumonia which was reported by participants. Therefore, pneumonia from which participants had died were not included. Since only two HIV positive
185
participants and two HIV negative participants died of pneumonia in the study period (data not shown) this can not bring about a serious bias. The low mortality due to bacterial infections has been reported before. ~'16. i7 In an earlier study, 1 we found that mortality caused by pneumonia is not a large problem in Amsterdam, in contrast to the inner cities in the U.S.A. 1.-2° Second, since only one episode of pneumonia can be reported per visit the already high incidence of pneumonia m a y even be an underestimation, especially when the interval between visits is long. We have tried to avoid this by censuring visits with an interval longer than 1 year. Third, self reported events m a y give rise to recall bias. For instance participants m a y have forgotten they suffered from pneumonia, but m a y also report pneumonia from which they had suffered in a previous period. However the interviewers were attentive to double reporting at two consecutive visits. The incidence of self reported pneumonia is m u c h higher than the incidence of hospitalized pneumonia (in HIV positive DU 26% of the episodes resulted in hospitalization, in HIV negative DU 16%). The difference in incidence between self reported and hospitalized pneumonia is probably a reflection of the organization of the Dutch health care system in which most Dutch drug users have close contact with methadone programmes and/or family practitioners. Physicians will be inclined to treat DU, who are thought to suffer from pneumonia on the basis of complaints and physical examination, without further diagnostics because they know this group is poorly motivated to further diagnostics (e.g. X-ray). Therefore severe bronchitis m a y have been diagnosed as pneumonia. In conclusion, our study shows that there is a distinct population of DU who is at increased risk for pneumonia. The identified risk groups are: HIV positive DU, DU with a previous history of pneumonia and DU who have an asthmatic constitution. Among HIV positive DU, those who inject drugs and those with lower CD4 levels are at increased risk. Among HIV negative DU also heroin smoking and a lower Quetelet Index are risk factors, in addition to asthmatic constitution and a previous history of pneumonia. Pneumococcal vaccination should be focused on the risk groups identified. In several countries vaccination against pneumococcal infections for HIV positive individuals is recommended, a~ By this study it is also possible to identify risk groups among HIV negative DU. We will offer pneumococcal vaccination to our HIV positive participants, as well as to HIV negative participants who are at increased risk, such as DU with an asthmatic constitution or a history of pneumonia.
186
Risk Factors for Pneumonia
Acknowledgements The authors wish to thank the nurses B. Scheeringa-Troost, J. Bax and A. van Dijk, and Dr H. Fennema, the medical specialists and the Drug department of the Municipal Health Service in Amsterdam for their contributions to the study.
References 1 Mientjes GH, Ameijden van EJC, Hoek van den JAR, Coutinho RA. Increasing morbidity without rise in non-AIDS mortality among HIV-infected intravenous drug users in Amsterdam. AIDS 1992; 6: 207-212. 2 SelwynPA, FeingoldAR, HartelDetal. Increasedriskofpneumonia in HIV-infected intravenous drug users without AIDS. AIDS 1988; 2: 267-272. 3 Selwyn PA, Hartel D, Wasserman W, Drucker E. Impact of the AIDS epidemic on Morbidity and Mortality among intravenous Drug Users in a New York City Methadone Maintenance Program. Am ] Public Health 1989; 79: 1358-1362. 4 Mientjes GHC, van Ameijden EJC, van den Hoek JAR, Goudsmit J, Miedema F, Coutinho RA. Progression of HIV infection among injecting drug users: indications for a lower rate of progression among those who have frequently borrowed injecting equipment. AIDS I993; 7: 1363-1370. 5 Mientjes GHC, van Ameijden EJC, Weigel HM, van den Hoek JAR, Coutinho RA. Clinical symptoms associated with seroconversion for HIV-1 among misusers of intravenous drugs: comparison with homosexual seroconverters and infected and non-infected misusers of intravenous drugs. Br Med J 1993: 6: 371-373. 6 Greenberg AE, Thomas PA, Landesman SH et al. The spectrum of HIV-1 related disease among outpatients in New York City. AIDS 1992; 60: 849-859. 7 Van de Hoek JAR, Coutinho RA, van Haastrecht HJA, van Zadelhoff AW, Goudsmit J. Prevalence and risk factors of HIV infection among drug users and drug-using prostitutes in Amsterdam. AIDS 1988; 2: 55-60.
8 Selwyn PA, Alcabes P, Hartel D et al. Clinical manifestations and predictors of disease progression in drug users with Human Immunodeficiency Virus infection. N Enfll I Med 1992; 327: 16971703. 9 Munoz A, Vlahov D, Solomon L e t aI. Prognostic indicators for development of AIDS among intravenous drug users. J AIDS 1992; 5: 694-700. 10 Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73: 13-22. 11 Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986; 42: 121-130. 12 Caiaffa WT, Vlahov D, Graham NMH et al. Risk factors for pneumonia in HIV seropositive injecting drug users. Ninth International Conference on AIDS. Berlin, June 7-11 1993 (Abstr PO-C15-2930). 13 Ghodse AH, Myles IS. Asthma in Opiate addicts. J Psychos Res 1987; 31: 41-44. 14 Roth Y, Aharonson EF, Teichthhl H, Baum GL, Priel Z, Modan M. Human in vitro nasal and tracheal ciliary beat frequencies. Ann Ot Rh La 1991; 100: 378-384. 15 Karlsson JA, Lanner AS, Persson CG. Airway opiate receptors mediate inhibition of cough and reflex bronchoconstriction in guinea pigs. ] Phar Exp Yher 1990; 252: 863-868. 16 Zaccarefli M, Gattari P, Rezza Get al. Impact of HIV infection on non-AIDS mortality among Italian injecting drug users. AIDS 1994; 8: 345-350. 17 Perucci CA, Davoli M, Rapiti E, Abeni D, Forastiere F. Mortality of intravenous drug users in Rome: a cohort study. Am f Public Health 1991; 81: 1307-1310. 18 Stoneburner RL, Des Jarlais DC, Benezra D et al. Spectrum of Severe HIV-1-Related Disease in Intravenous Drug Users in New York City. Science 1988; 242: 916-918. 19 Centers for Disease Contxol. Increase in pneumonia mortality among young adults and the H1V epidemic, New York City, United States. MMWR 1988; 37: 593-596. 20 Drucker E, Webber MP, McMaster P, Vermund SH. Increasing Rate of Pneumonia Hospitalizations in the Bronx: A sentinel indicator for Human Immunodeficiency Virus. Inter l Epidemiol 1989; 18: 926-933. 21 Immunization Practices Advisory Committee (ACIP). Pneumococcal poly saccharide vaccine. MMWR 1989; 38:64-68 and 73-76.
Incidence and Risk Factors for Pneumonia in HIV infected and Non-infected Drug Users G. H. C. Mientjes, I. J. B. Spijkerman*, E. J. C. van Ameijden, J. A. R. van den Hoek and R. A. Coutinho Municipal Health Service, Department of Public Health and Environment, Amsterdam, The Netherlands Accepted for publication 1 November •995
Objective: To study the incidence and risk factors for pneumonia in a cohort of HIV infected and non-infected drug users
(DU). Design: A prospective epidemiological study. Setting and patients: Injecting and non-injecting DU who attended the Municipal Health Service in Amsterdam for followup visits in the stu@. Main results: 203 HIV infected and 437 non-infected DU were followed for a total of 1749 person-years. HIV infected DU reported 111 episodes of pneumonia, which required hospitalization in 29 cases, and HIV negative DU reported 55 episodes, which required hospitalization in nine cases. The incidences among HIV positive and HIV negative DU were O. 19 and 0.05 per person-year respectively. With multivariate Poisson regression current injecting (RR 2.13), recent seroconversion (RR 3.92), asthmatic constitution (RR 2.72), CD4+ ceil count between 200-500 (RR 1.67 compared to >500), CD4+ cell count less than 200 (RR 2.23 compared to >500) and a previous history of pneumonia (RR 2.43) were independently associated with self-reported pneumonia among HIV infected DU. Among HIV negative DU heroin smoking (RR 1.87), asthmatic constitution (RR 3.62) and a previous history of pneumonia (RR 2.84) were independently associated with self-reported pneumonia. Also a higher Quetelet Index (OI) appeared to be protective (QI>_21 RR 0.42, QI 19-21 RR 0.82 compared to 0I<19) among HIV negative DU. Risk factors for reported and hospitalized cases of pneumonia were comparable among HIV positive DU. Conclusions: HIV infected DU are at increased risk for pneumonia and the rate increases with lower CD4 cell counts. Also behavioural characteristics, such as injecting drug use and smoking heroin, and clinical history variables, such as a history of pneumonia or an asthmatic constitution, are risk factors for pneumonia among DU. Pneumococcal vaccination should not only be focused on HIV positive DU but also on the identified risk groups among I-IIV negative DU.
Introduction Pneumonia is observed more often among HIV infected persons compared to non-HIV infected persons and the incidence a m o n g HIV infected persons increases with a declining n u m b e r of CD4 + cells in the asymptomatic phase of HIV infection. 14 For this reason recurrent bacterial pneumonia has been included in the AIDS definition, both in the United States and Europe. Among injection drug users, whether HIV infected or not, the incidence of pneumonia is higher compared to the incidence a m o n g for example homosexual men. s'6 Possibly, behavioural and social factors increase the risk
* Address correspondence to: I. J. B. Spijkerman, Municipal Health Service, Department of Public Health and Environment, P.O. Box 20244, 1000 HE Amsterdam, The Netherlands. This study was supported by: the Netherlands Foundation for Preventive Medicine (grant no. 28-1258). 0163-4453/96/030181 + 06 $12.00/0
for pneumonia among drug users (DU). However, studies on risk factors for pneumonia among DU are sparse. Here we report on the incidence and risk factors for pneumonia among DU in Amsterdam stratified by HIV serostatus.
Methods In December 1985, we began a prospective study to assess the prevalence and incidence of HIV infection, to determine risk factors for HIV infection, and to evaluate the natural history of HIV infection among DU. The study design has been described extensively in previous records. 1'4'S'7 Enrolment in the study is ongoing. Participants are recruited at low-threshold methadone posts and at a sexually transmitted diseases clinic for drugusing prostitutes. Participants are asked to return every 4 months for repeated interviews, blood sampling, and © 1996 The British Society for the Study of Infection
182
Risk Factors for Pneumonia
physical examination but participants occasionally return after longer periods.
Present study The present study was designed to determine the incidence and risk factors for pneumonia. The medical history is taken by a physician. At every visit participants were asked whether they suffered from pneumonia. If the answer was positive, standardized questions were asked about complaints, diagnostics and treatment. All cases of pneumonia which were diagnosed by a physician and treated with antibiotics are considered as reported pneumonia. Reported episodes of bronchitis or upper respiratory tract infections are not included. When a participant reported to be hospitalized for pneumonia, the medical specialist in attendance was contacted to obtain the reason for hospitalization. For the analysis of hospitalized cases of pneumonia, only verified cases were included, which showed consolidation on a chest X-ray. We have analysed self-reported cases of pneumonia and hospitalized cases of pneumonia separately. At intake pneumonia is reported over the period between 1980 and intake. At follow-up visits pneumonia is reported over the period since the previous visit. To reduce possible recall bias we have excluded intake visits and included only visits for which the interval between visits was less than 1 year. In our cohort, lymphocyte counts have been determined routinely since April 1989. Because the CD4 count was shown to be an important risk factor for pneumonia 4' 8,9 only visits after April 1989 were included in the analysis. Follow-up ended February 1994.
housing (yes/no), engaging in prostitution (yes/no), previous history of pneumonia since 1980 (yes/no), injecting drugs (yes/no), frequency of injecting (
Laboratory tests Sera were tested for antibodies to HIV in a commercial ELISA (Vironostika Teknika, Organon, Oss, The Netherlands). HIV infection was confirmed by immunoblotting and/or competitive ELISA. Peripheral blood mononuclear cells were isolated from heparinized venous blood by density gradient centrifugation on Ficoll-Hypaque. Absolute numbers of CD4 + and CD8 + lymphocytes were determined by cytofluorometry.4
Statistics Definition of variables Preliminary analysis showed that pneumonia may influence the behaviour in the period between two visits in which the participant suffered from pneumonia. A participant who has been seriously ill may, for example, reduce the frequency of injecting and may receive daily methadone in higher doses during his/her illness or hospitalization, than participants without pneumonia. For this reason we have used behavioural and demographic variables from the visit preceding the visit on which the episode of pneumonia was reported. Immunological markers may also be disturbed by the occurrence of pneumonia and therefore we also used immunological data from the visit preceding the episode of pneumonia. Independent variables studied were: sex, age ( < 3 3 , >_ 34), HIV status, nationality (Dutch, foreign), number of CD4 cells (<200, 2 0 0 - 5 0 0 , >500 x 106/mm3), steady
Incidences were calculated for self reported and hospitalized cases of pneumonia by dividing the number of pneumonia by the sum of the total follow-up period, stratified to the level of a specific variable. Analysis was performed separately for HIV positive and HIV negative participants. With Poisson regression the univariate relative rates were calculated. To determine independent predictors, all univariately significant variables (P<0.05) were entered in Poisson regression models and backward selection was used. Interaction between the variables in the final model was checked. Because the number of hospitalized episodes of pneumonia among HIV positive DU was small and because we were interested whether the Rate Ratios for a specific variable differed for self reported and hospitalized episodes, we have forced the same variables, which were multivariately associated with self reported episodes of pneumonia, into the multivariate model of hospitalized pneumonia.
G. H. C. Mientjes et al. Table 1. General characteristics of drug users in the study by HIV serostatus HIV positive HIV negative (n=203) (n=437) Male sex (%) Mean number of visits (S.D.) Mean interval between visits in months (S.D.) Dutch nationality (%) Prostitution previous period (%) Injected previous period (%) First injection>lO years ago (%) Having permanent housing (%) CD4+cell count<500.109/1 (%)
123 (60.6) 257 (58.8) 7.3 (4.4) 6.7 (4.4) 4.7 (1.6) 138 (68.0) 56 (27.6) 171 (84.2) 111 (55.2) 149 (73.4) 108 (57.1)
4.8 (1.6) 336 (76.9)* 132 (30.2) 255 (59.1)* 141 (32.6)* 345 (78.9) 25 (6.0)*
* P<0.05 compared to HIV positive participants.
Because for each participant more than one observation could be included in the analyses, observations m a y be non-independent. To adjust for this possible intraperson correlation of data, Rate Ratios (RR) and 95% confidence intervals were not only determined with Poisson regression, but also by generalized estimation equations (GEE) using a Poisson link function. ~°'u These repeated measures analyses were performed both with independent and with exchangeable correlation structures.
Results A total of 203 HIV positive and 437 HIV negative drug users were in follow-up in the period between April 1989 and December 1993 and were eligible for the present study. The general characteristics at the first visit after April 1989 are summarized in Table I. The total followup period for HIV positive subjects was 578 years and for HIV negative subjects 1171 years. HIV positive subjects were more likely to have injected in the preceding period and to have had a previous history of pneumonia but were less likely to have smoked drugs compared to HIV negative subjects. During follow-up 111 episodes of pneumonia were reported by 65 HIV positive DU (35 DU: 1 episode; 19 DU: 2 episodes; 7 DU: 3 episodes; 3 DU: 4 episodes and 1 DU: 5 episodes). Of these 111, 29 episodes of pneumonia required hospitalization (21 DU: 1 episode and 4 DU: 2 episodes). HIV negative DU reported 55 episodes (38 DU: 1 episode; 5 DU: 2 episodes; 1 DU: 3 episodes and 1 DU: 4 episodes). Of these 55, nine episodes of pneumonia required hospitalization (9 DU: once). The incidence of self reported pneumonia a m o n g HIV positive DU was 0.19 per person-year and a m o n g HIV negative DU 0.05 per person-year.
183
Of the 29 hospitalized episodes among HIV positive DU the causative microorganism was identified by sputum culture in 20 cases. Haemophilus influenzae was identified 10 times, Streptococcuspneumonia four times, Haemophilus influenzae and Streptococcus sp. were cultured three times, Enterobacter sp. twice and Streptococcus sp. once. Univartate Poisson regression showed that among HIV positive participants, current injecting, an asthmatic constitution, a previous history of pneumonia, recent seroconversion, a lower Quetelet Index and a lower CD4 + cell count were significantly and positively associated with self reported pneumonia (Table II). In the same group we found that asthmatic constitution, recent seroconversion and nationality were also associated with hospitalized episodes of pneumonia. Among HIV negative participants, female sex, heroin smoking, asthmatic constitution, a previous history of pneumonia and a lower Ouetelet Index were significantly and positively related with self reported pneumonia (Table II). Due to small numbers of hospitalized pneumonia among HIV negative DU, these were not analyzed. Table III shows the results of the multivariate Poisson regression for self-reported and hospitalized pneumonia a m o n g HIV infected DU. Among HIV positive DU injecting drugs, recent seroconversion, a lower level of CD4 + cell count, asthmatic constitution and a previous history of pneumonia was independently associated with self reported pneumonia. When we forced the same variables, which were multivariately associated with self reported episodes of pneumonia, into the multivariate model of hospitalized pneumonia, we found that recent seroconversion, asthmatic constitution and a lower CD4 + cell count were associated with hospitalized episodes of pneumonia. Because nationality was significantly associated with hospitalized cases in the univariate analysis and did not substantially alter the RR for the other variables in the model, we present the multivariate model for hospitalized episodes of pneumonia including nationality (Table III). Among HIV negatives self reported cases were multivariately associated with smoldng heroin, asthmatic constitution, a previous history of pneumonia and a lower Quetelet Index (Table IV). The results of the analysis with Generalized Estimation Equations (see methods) revealed only minor differences from the results obtained with Poisson Regression (data not shown).
Discussion As was shown in several studies, HIV positive DU have a nearly 4-fold higher risk of pneumonia than HIV negative DU 1'2.12 and this risk increases with lower levels
184
Risk Factors for Pneumonia
TabLe II. Incidence and univariate rate ratios for pneumonia among drug users by HIV serostatus Reported cases HIV-positive
Sex
Male Female No Yes No Yes No Yes No Yes No Yes >500 200-500 <200 <19 19-21 >21 Dutch Foreign
Current injecting Heroin smoking Asthmatic constitution Ever pneumonia before Recent seroconversion* CD4 count Ouetelet Index Nationality
INC
RR
18.4 19.5 9.5 21.6 19.4 17.3 15.9 49.0 10.6 29.0 18.2 55.6 13.0 20.1 26.5 22.9 18.8 11,6 17.0 22,7
1,00 1.06 1.00 2.28 1.00 0.89 1.00 3.07 1.00 2.74 1.00 3.03 1.00 1,55 2,05 1,00 0,82 0,52 1.00 1.34
(95% CI)
(0.72-1.55) (1.28-4.06) (0.58-1.37) (1.97-4.80) (1.84-4.09) (1,25-7.69) (0.98-2.45) (1.18-3,54) (0.53-1,26) (0.30-0.87) (0.91-1.97)
Hospitalized cases HIV-positive INC
RR
4.9 4.9 2.9 5.5 4.9 4.9 4.1 13.7 3.6 6.5 4.3 44.4 3.4 4.6 8,8 5.9 3,8 4,5 3.3 8.7
1.00 0.99 1.00 1.89 1.00 1,O0 1,00 3.36 1.00 1.79 1.00 10.O1 1.00 1.37 2.30 1.00 0.64 0.77 1.00 2.7
Reported cases HIV-negative
(95% CI)
(0.47-2.09) (0.66-5.43) (0,44-2.28) (1.44-7,38) (0.85-3.74) (3.57-105.4) (0,55-3.42) (0.81-6.56) (0.26-1.26) (0.31-1.93) (1.30-5.60)
INC
RR
(95% CI)
3.5 7.7 3.8 6.4 3.7 7.4 4.2 14.8 3.2 10,6 -----6.8 5.9 2.9 5.5 4.3
1.00 2.18 1.00 1.70 1.00 2,00 1,O0 3.52 1.00 3.34 -----1.O0 0.87 0.43 1.00 0.79
(1.27-3.74) (0.97-2.97) (1.17-3.41) (1,94-6.36) (1.96-5.68) -----(0.45-1.66) (0.21-0.87) (0.39-1,61)
Univariate rate ratios (RR) and 95% Confidence Intervals (95% CI) were calculated with Poisson regression. Behaviour, CD4 count and Quetelet Index were measured at the visit preceding the report of an episode of pneumonia. Recent seroconversion is the first HIV positive visit after an HIV negative visit. * The first HIV positive visit is compared to all further HIV positive visits from seroconverters and prevalent HIV positive participants (methods).
Table III. Risk factors for reported (n = 111) and hospitalized (n = 29) pneumonia in multivariate Poisson regression among HIV infected drug users Reported episodes
RR
95%
1.87 3.62 2.84 1.00 0.82 0.41
(1.06-3.19) (1.91-6.85) (1.60-5.03)
Hospitalized episodes
RR
95% CI
RR
2.13 3.92 2,72 1.00 1.67 2.23
(1.16-3.92) (1.52-10.2) (1.73~t.27)
2.28 (0.67-7.77) 25.1 (6.26-10.1) 3.29 (1.36-7.92)
(1,25-3.97) (1,25-3.97)
Ever pneumonia before 2.43 Foreign nationality
(1.60-3.68)
5.61 (1.51-20.9) 5.61 (1,51-20.9) 1.70 (0.76-3.79) 3.12 (1.45-6.71)
Current injecting Recent seroconversion Asthmatic constitution CD4 count>500 CD4 count 200-500 CD4 count<200
Table IV. Risk factors for reported pneumonia (n=55) in multivariate Poisson regression among HIV negative drug users
95% CI
RR=rate ratio, 95% CI= 95% Confidence Interval. All variables measured at the last visit preceding the episode of pneumonia. For the analysis of the hospitalized cases all variables significant in the analysis of the reported episodes were included, together with nationality.
of C D 4 cells. 4' s W e f o u n d t h a t a s t h m a t i c c o n s t i t u t i o n is a r i s k f a c t o r for p n e u m o n i a , w h i c h is w e l l I m o w n . G h o d s e a n d H y l e s 13 h a v e s h o w n t h a t , i n a s a m p l e of 2 2 7 6 o p i a t e a d d i c t s , t h e u s e of h e r o i n m a y inflict a s t h m a t i c r e a c t i o n s , e s p e c i a l l y i n w o m e n . W e also f o u n d t h a t a s t h m a t i c c o n s t i t u t i o n w a s m o r e p r e v a l e n t i n w o m e n . Of all w o m e n
Heroin smoking Asthmatic constitution Ever pneumonia before Quetelet Index<19 Quetelet Index 19-21 Onetelet Index>21
(0.42-1.57) (0.20-0.84)
RR=rate ratio, 95% CI= 95% Confidence Interval. All variables measured at the last visit preceding the visit at which the episode of pneumonia was reported.
in the study, 12.7% had an asthmatic constitution, comp a r e d to 4 . 5 % of t h e m e n ( P = 0 . 0 0 0 1 4 ) . H o w e v e r , s i n c e w e h a v e classified a s t h m a t i c c o n s t i t u t i o n i n a v e r y r o u g h way (anyone who received anti-asthmatics during any t i m e d u r i n g f o l l o w u p w a s classified as h a v i n g a n a s t h matic constitution during their entire follow-up period) t h e v a l i d i t y of t h i s f i n d i n g is difficult t o assess. B e y o n d t h a t , it is also p o s s i b l e t h a t DU w i t h p n e u m o n i a a r e m o r e i n c l i n e d to r e c e i v e a n t i - a s t h m a t i c s t h a n DU w i t h o u t pneumonia. Therefore the relation between pneumonia and asthmatic constitution may be biased. The rate ratio, h o w e v e r , is h i g h a n d a t l e a s t g i v e s a n i n d i c a t i o n t h a t
G. H. C. Mientjes et al. HIV positive drug users with asthma m a y be at increased risk for pneumonia. A previous history of pneumonia was also strongly associated with pneumonia. This is consistent with reports indicating that pre-existing lung disease is associated with pneumonia. 14 Our earlier observation that DU who seroconverted are at increased risk to acquire pneumonia shortly after seroconversion was confirmed in the present study, s The risk factors for pneumonia a m o n g HIV positive subjects did not differ substantially between self reported and hospitalized episodes, though recent injecting and a previous history of pneumonia were not significantly associated with hospitalized episodes. This was probably due to the smaller n u m b e r of hospitalized episodes, as the rate ratios did not differ substantially. Nationality was associated only with hospitalized pneumonia and not with self-reported cases. This is probably caused by the fact that non-Dutch participants are more often not insured and have less access to primary health care. They also live more often under worse social conditions because they have less access to housing facilities and do not receive welfare. Pneumonia a m o n g these non-Dutch participants might often be diagnosed at a later stage of the disease when hospitalization is more likely to be necessary. A new finding is that current drug use-related behaviour was associated with pneumonia. Among HIV positive subjects injecting drugs, as opposed to not injecting drugs, gave a rate ratio of 2.13 for reported and 2.28 for hospitalized episodes of pneumonia. Since recurrent pneumonia is included in the revised AIDS definition, this means that current behaviour m a y influence the incidence of AIDS. The exact cause for the increased risk to acquire pneumonia among injectors is not clear. The higher risk of current injectors could be due to the (not measured) worse social conditions of active injectors. A risk factor for pneumonia a m o n g HIV negatives was smoking heroin. This was already suggested by physicians who daily work with addicts (van Brussel, personal communication). Possibly smoking heroin decreases the cough reflex and ciliary activity. TM ~s Recently a study in the U.S.' has shown that smoking cocaine or m a r i h u a n a increases the risk of pneumonia in HIV positive DU. 12 Smoking heroin appeared to be no risk factor a m o n g HIV positives, probably because they smoked drugs less frequently. There are some possible sources of bias in the incidence of pneumonia, as we measured it. At first we have included only pneumonia which was reported by participants. Therefore, pneumonia from which participants had died were not included. Since only two HIV positive
185
participants and two HIV negative participants died of pneumonia in the study period (data not shown) this can not bring about a serious bias. The low mortality due to bacterial infections has been reported before. ~'16. i7 In an earlier study, 1 we found that mortality caused by pneumonia is not a large problem in Amsterdam, in contrast to the inner cities in the U.S.A. 1.-2° Second, since only one episode of pneumonia can be reported per visit the already high incidence of pneumonia m a y even be an underestimation, especially when the interval between visits is long. We have tried to avoid this by censuring visits with an interval longer than 1 year. Third, self reported events m a y give rise to recall bias. For instance participants m a y have forgotten they suffered from pneumonia, but m a y also report pneumonia from which they had suffered in a previous period. However the interviewers were attentive to double reporting at two consecutive visits. The incidence of self reported pneumonia is m u c h higher than the incidence of hospitalized pneumonia (in HIV positive DU 26% of the episodes resulted in hospitalization, in HIV negative DU 16%). The difference in incidence between self reported and hospitalized pneumonia is probably a reflection of the organization of the Dutch health care system in which most Dutch drug users have close contact with methadone programmes and/or family practitioners. Physicians will be inclined to treat DU, who are thought to suffer from pneumonia on the basis of complaints and physical examination, without further diagnostics because they know this group is poorly motivated to further diagnostics (e.g. X-ray). Therefore severe bronchitis m a y have been diagnosed as pneumonia. In conclusion, our study shows that there is a distinct population of DU who is at increased risk for pneumonia. The identified risk groups are: HIV positive DU, DU with a previous history of pneumonia and DU who have an asthmatic constitution. Among HIV positive DU, those who inject drugs and those with lower CD4 levels are at increased risk. Among HIV negative DU also heroin smoking and a lower Quetelet Index are risk factors, in addition to asthmatic constitution and a previous history of pneumonia. Pneumococcal vaccination should be focused on the risk groups identified. In several countries vaccination against pneumococcal infections for HIV positive individuals is recommended, a~ By this study it is also possible to identify risk groups among HIV negative DU. We will offer pneumococcal vaccination to our HIV positive participants, as well as to HIV negative participants who are at increased risk, such as DU with an asthmatic constitution or a history of pneumonia.
186
Risk Factors for Pneumonia
Acknowledgements The authors wish to thank the nurses B. Scheeringa-Troost, J. Bax and A. van Dijk, and Dr H. Fennema, the medical specialists and the Drug department of the Municipal Health Service in Amsterdam for their contributions to the study.
References 1 Mientjes GH, Ameijden van EJC, Hoek van den JAR, Coutinho RA. Increasing morbidity without rise in non-AIDS mortality among HIV-infected intravenous drug users in Amsterdam. AIDS 1992; 6: 207-212. 2 SelwynPA, FeingoldAR, HartelDetal. Increasedriskofpneumonia in HIV-infected intravenous drug users without AIDS. AIDS 1988; 2: 267-272. 3 Selwyn PA, Hartel D, Wasserman W, Drucker E. Impact of the AIDS epidemic on Morbidity and Mortality among intravenous Drug Users in a New York City Methadone Maintenance Program. Am ] Public Health 1989; 79: 1358-1362. 4 Mientjes GHC, van Ameijden EJC, van den Hoek JAR, Goudsmit J, Miedema F, Coutinho RA. Progression of HIV infection among injecting drug users: indications for a lower rate of progression among those who have frequently borrowed injecting equipment. AIDS I993; 7: 1363-1370. 5 Mientjes GHC, van Ameijden EJC, Weigel HM, van den Hoek JAR, Coutinho RA. Clinical symptoms associated with seroconversion for HIV-1 among misusers of intravenous drugs: comparison with homosexual seroconverters and infected and non-infected misusers of intravenous drugs. Br Med J 1993: 6: 371-373. 6 Greenberg AE, Thomas PA, Landesman SH et al. The spectrum of HIV-1 related disease among outpatients in New York City. AIDS 1992; 60: 849-859. 7 Van de Hoek JAR, Coutinho RA, van Haastrecht HJA, van Zadelhoff AW, Goudsmit J. Prevalence and risk factors of HIV infection among drug users and drug-using prostitutes in Amsterdam. AIDS 1988; 2: 55-60.
8 Selwyn PA, Alcabes P, Hartel D et al. Clinical manifestations and predictors of disease progression in drug users with Human Immunodeficiency Virus infection. N Enfll I Med 1992; 327: 16971703. 9 Munoz A, Vlahov D, Solomon L e t aI. Prognostic indicators for development of AIDS among intravenous drug users. J AIDS 1992; 5: 694-700. 10 Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73: 13-22. 11 Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986; 42: 121-130. 12 Caiaffa WT, Vlahov D, Graham NMH et al. Risk factors for pneumonia in HIV seropositive injecting drug users. Ninth International Conference on AIDS. Berlin, June 7-11 1993 (Abstr PO-C15-2930). 13 Ghodse AH, Myles IS. Asthma in Opiate addicts. J Psychos Res 1987; 31: 41-44. 14 Roth Y, Aharonson EF, Teichthhl H, Baum GL, Priel Z, Modan M. Human in vitro nasal and tracheal ciliary beat frequencies. Ann Ot Rh La 1991; 100: 378-384. 15 Karlsson JA, Lanner AS, Persson CG. Airway opiate receptors mediate inhibition of cough and reflex bronchoconstriction in guinea pigs. ] Phar Exp Yher 1990; 252: 863-868. 16 Zaccarefli M, Gattari P, Rezza Get al. Impact of HIV infection on non-AIDS mortality among Italian injecting drug users. AIDS 1994; 8: 345-350. 17 Perucci CA, Davoli M, Rapiti E, Abeni D, Forastiere F. Mortality of intravenous drug users in Rome: a cohort study. Am f Public Health 1991; 81: 1307-1310. 18 Stoneburner RL, Des Jarlais DC, Benezra D et al. Spectrum of Severe HIV-1-Related Disease in Intravenous Drug Users in New York City. Science 1988; 242: 916-918. 19 Centers for Disease Contxol. Increase in pneumonia mortality among young adults and the H1V epidemic, New York City, United States. MMWR 1988; 37: 593-596. 20 Drucker E, Webber MP, McMaster P, Vermund SH. Increasing Rate of Pneumonia Hospitalizations in the Bronx: A sentinel indicator for Human Immunodeficiency Virus. Inter l Epidemiol 1989; 18: 926-933. 21 Immunization Practices Advisory Committee (ACIP). Pneumococcal poly saccharide vaccine. MMWR 1989; 38:64-68 and 73-76.