Cross-sectional survey of HIV infection among patients with tuberculosis in Nairobi, Kenya

Tuhercle and Lmp Lhease (I 992) 13.45-S I

Cross-sectional survey of HIV infection among patients with tuberculosis in Nairobi, Kenya P. Nunn*f$, C. Gicheha”, R. Hayes*, S. Gathuat, Omwega*, J. Were*, K. McAdam$

R. Brindle*t§,

D. Kibugat,

T. Mutiet,

R. Kamunyi**,

M.

*Kenya Medical Research Institute, Nairobi, Kenya, flnfectious Diseases Hospital, Kenyatta National Hospital, Nairobi, Kenya, SLondon School of Hygiene and Tropical Medicine, London, UK, aPublic Health Laboratory, Oxford, UK and **National Public Health Laboratories, Tuberculosis Laboratory, Nairobi, Kenya

S UMMA R Y. Evidence from many countries suggests an association of human immunodeficiency virus (HIV) infection and tuberculosis of major public health significance. In order to begin assessing the impact of HIV on tuberculosis in Kenya, we have determined the HIV-l seroprevalence among tuberculosis patients and compared the clinical characteristics of tuberculosis in HIV-positive and HIV-negative patients in two crosssectional studies at the Infectious Disease Hospital (IDH) and the Ngaira Avenue Chest Clinic (NACC), Nairobi, Kenya. The diagnosis in 92% of all patients with pulmonary tuberculosis was confirmed by culture. The remainder were diagnosed on histological, clinical or radiological grounds. HIV seroprevalence among tuberculosis patients at IDH was 26.5% (52/196) compared to 9.2% (18/195) at NACC (P c 0,901). There was no association between numbers of streptomycin injections in the previous 5 years and HIV infection. Positive sputum smear rates in HIV-positive patients were slightly lower than in HIV-negative patients at both study sites (71% vs 83% at IDH and 73% vs 82% at NACC) but the difference was not significant. Only Mycobuctetium tuberculosis was isolated. Miliary disease was not associated with HIV infection. Persistent diarrhoea, oral candidiasis, generalized itchy rash, herpes zoster and generalized lymphadenopathy were all associated with HIV infection, but 46% (95% CI:38-54%) of all HIV-positive patients had none of the clinical features listed in the WHO Clinical Criteria for the Diagnosis of AIDS, apart from fever, cough and weight loss. Stevens-Johnson Syndrome was reported in 7/52 (13%) patients with HIV infection, and in 4/144 (3%) patients without (RR 4.85, 95% CI: 1.45-15.88). The management of HIV-associated tuberculosis is complex and requires more resources than have previously been given to tuberculosis treatment services. Les constatations faites dans plusieurs pays suggerent une association entre le virus de 1’immunodCficience humaine (VIH) et la tuberculose, posant un probleme majeur de Sante publique. Pour entreprendre une evaluation de l’impacte du VIH sur la tuberculose au Kenya, a CtC determinee la seroprevalence du VIH-1 parmi les malades atteints de tuberculose, et comparees les particularites cliniques de la tuberculose VIH+ et VIH- dans 2 etudes transversales a l’Hbpita1 des Maladies Contagieuses (I’IDH) et B la Ngaira Avenue Chest Clinic (NACC), B Nairobi, Kenya. 92% de tous les malades atteints d’une tuberculose pulmonaire ont un diagnostic confirm4 par culture. Les autres cas ont CtCdiagnostiques sur des don&es histologiques, cliniques ou radiologiques. La seroprevalence du VIH parmi les malades de I’IDH atteints de tuberculose Ctait de 265% (52096) cornparke a 9,2% (B/195) au NACC (P < 0,001). 11 n’y avait aucun rapproachement entre le nombre d’injections de streptomycine pendant les 5 am&s precedantes et l’infection par le VIH. Le taux de frottis positifs des crachats des malades seropositifs etait legerement inferieur B celui des malades seronegatifs dans les deux lieux d’etude (71% contre 83% a I’IDH et 73% contre 82% au NACC), cependant la difference n’est pas significative. Mycobacterium tuberculosis seul a ete isole. Une miliaire n’etait pas associee a l’infection par le VIH. Une diarrhee persistante, une candidose buccale, un rash prurigineux generali&, un zona et une lymphadenopathie gCnCralisCes’associaient toutes 1 l’infection par le VIH, mais 46% (9S%CI 38%-54%) de tous les malades seropositifs n’avaient aucune des caractdristiques cliniques conformes aux criteres Cliniques de I‘OMS pour le Diagnostic du SIDA, a part la fievre, la toux et R kS UMi.

Correspondence to: Dr Paul Nunn, c/o The School of Public HealthBEHS, 101 Haviland Hall, University of California, Berkeley, CA 94720. USA. 45

46

Tuber&

and Lung Disease

l’amaigrissement. Un syndrome de Stevens-Johnson a dtCconstate dans 7/52 (13 % ) des malades seropositifs, et dans 41144 (3%) des malades seronegatifs (RR 4,85, 95% CI: 1,45-15,SS). La surveillance de la tuberculose associee au VIH est complexe et necessite davantage de ressources que celles mises jusqu’ici a la disposition des services pour le traitement de la tuberculose. R ES U A4 E N. Evidencias provenientes de numerosos paises sugieren una asociacidn entre la infecci6n por el

virus de la inmunodeficiencia humana (VIH) y la tuberculosis, hecho que tiene una importancia considerable en terminos de salud publica. Para comenzar nuestra evaluation de1 impact0 de1 VIH sobre la tuberculosis en Kenia, determinamos la seroprevalencia VIH-1 en pacientes tuberculosos y comparamos las caracteristicas clinicas de 10s cases de tuberculosis VIH positivos y VIH negativos en 2 estudios transversales efectuados en el Hospital de Enfermedades Infecciosas (HEI) y en el dispensario de enfermedades respiratorias Ngaira Avenue (DERNA), en Nairobi, Kenia. Un 92% de 10s pacientes con tuberculosis pulmonar fueron confirmados mediante cultivo. El resto fue diagnosticado en base a criterios histolijgicos, clinicos o radiologicos. La seroprevalencia VIH entre 10s pacientes tuberculosos de1 HE1 fue de 26,5% (52/196) comparado con un 9,2% (lW195) en el DERNA (P < 0,001). No hubo asociaci6n entre el numero de inyecciones de estreptomicina en 10s 5 silos precedentes y la infecci6n VIH. Las tasas de frotis de esputo positivos en 10s pacientes VIH positivos fueron levemente inferiores a las de 10s pacientes VIH negativos en ambos lugares de estudio (71% versus 83% en el HE1 y 73% versus 82% en el DERNA), pero la diferencia no fue significativa. Solamente se aisl6 Mycobacteriun tuberculosis. Una enfermedad miliar no estaba asociada con la infecci6n VIH. Una diarrea persistente, una candidiasis oral, una erupcidn cut&tea pruriginosa, un herpes zoster y una linfadenopatia generalizada, estaban asociadas con la infecci6n VIH, pero un 46% (95% IC: 38-54%) de todos 10s pacientes VIH positivos no presentaba ninguno de 10s elementos clinicos enumerados en 10s Criterios Clinicos de la OMS para el diagnostico de1 SIDA, salvo fiebre, tos y perdida de peso. El Sindrome de Stevens-Johnson fue observado en 7/52 (13%) pacientes con infecci6n VIH y en 4/144 (3%) sin infecci6n (RR 4,85 ; 95% IC: 1,45-15,88). El manejo de la tuberculosis asociada a VIH es complejo y requiere mb recursos que 10s otorgados hasta la fecha a 10s servicios de tratamiento de la tuberculosis.

INTRODUCTION The association

of human immunodeficiency virus (HIV) infection and tuberculosis has been reported from the USAl Europe’ and Africa63 7. In areas with a high prevalence of infection with both HIV and M. tuberculosis, the incidence of tuberculosis appears to be rising.8,9 Among drug users in New York, it appears that HIV causes reactivation of latent infection with M. tuberculosis’0 and, if this is the case elsewhere, a secondary epidemic of tuberculosis in Africa is a distinct possibility. ’ ’ It is clear that HIV is already having a profound effect on the workload of tuberculosis treatment centres and control programmes in Africa. As the first step in a programme to assess the impact of HIV on tuberculosis in Kenya, we have carried out a cross-sectional survey of patients undergoing antituberculosis chemotheraphy at two sites in Nairobi, Kenya: the Infectious Disease Hospital (IDH), and the City Commission Ngaira Avenue Chest Clinic (NAC). The aims of the study were to determine the HIV seroprevalence among these two groups of tuberculosis patients, to examine the role of streptomycin injections in the transmission of HIV, and to compare the clinical presentation and diagnosis of HIV-associated tuberculosis with tuberculosis in HIVnegative patients. The frequency of cutaneous complications during treatment was also assessed in the two groups.

METHODS Survey procedure, IDH and NACC Between November 1988 and March 1989, 211 consecutive inpatients and outpatients undergoing treatment for tuberculosis at IDH were asked to take part in the study. Patients could be recruited at any stage of their treatment. 15 refused (11 did not wish to have blood taken, 2 declined because of a previous positive HIV test result and 2 gave no reason). Following informed consent, the remaining 196 patients were examined by means of a questionnaire and a brief physical examination and a blood sample was taken. Numbers of previous streptomycin injections were determined from the patients’ files or clinic cards. Full-size chest radiographs taken at the time of presentation were assessed by a respiratory physician (PN) without prior knowledge of the HIV result. Radiographs were divided horizontally by eye into three equal zones per lung field and the total number of zones involved by parenchymal disease was recorded. The presence of a rniliary pattern, lymphadenopathy and pleural or pericardial effusions or cavitation was also recorded. Cavitation was defined as radiolucent areas greater than 1 cm in diameter surrounded by consolidation or fibrosis, or with a wall at least 2 mm thick. 2 17 outpatients were invited to participate in a similar study between May and August 1989 at NACC. 22 refused because they did not wish to have blood taken. The remaining 195 were included in the study. Similar

HIV infection and tuberculosis

data were collected as at IDH. Chest radiographs were mostly miniature films (10 x 10 cm) and were assessed by 2 health workers (PN and CG) without a magnifying apparatus, and without knowledge of the HIV status. Doubtful films were reviewed and a consensus was reached. IDH is a referral hospital with about 100 tuberculosis beds. Patients are referred from the main Kenyatta National Hospital (KNH) for treatment, or from provincial hospitals, City Commission clinics, private hospitals or practitioners for diagnosis and/or treatment. NACC is solely an outpatient facility run by the City Commission Health Department for the diagnosis of chest diseases and the treatment of tuberculosis. In both institutions at the time of the study, routine antituberculosis chemotherapy consisted of 1 month of streptomycin (1 g i.m. daily for 7 days per week, or 6 days weekly at NACC) concurrently with 12-18 months of oral thiazina (a combined preparation of thiacetazone 150 mg and isoniazid 300 mg). If adverse effects occurred, all drugs were stopped until the reaction had subsided, and challenge doses of one drug at a time were administered daily according to standard practice.” Rifampicin, ethambutol and pyrazinamide were available if either resistance or hypersensitivity to one or more drugs were observed. The diagnosis of tuberculosis was ‘confirmed’ if either acid-fast bacilli (AFB) were seen on microscopical examination of sputum, a positive culture for M. tuberculosis was obtained, or caseating granulomata or AFB were seen on histological examination of a biopsy specimen. ‘Unconfirmed’ cases of tuberculosis had either radiological appearances consistent with tuberculosis (parenchymal shadowing with or without cavitation) and were clinically resistant to broad-spectrum antibiotics but did respond to antituberculosis treatment, or had clinical signs suggestive of extrapulmonary tuberculosis together with a response to antituberculosis treatment.

Laboratory

methods

HIV serology HIV-l infection was determined by a repeatable positive result on both of two ELISA tests (DuPont HTLV-III test, DuPont de Nemours, Wilmington, DE 19898, USA and Wellcozyme, Wellcome Diagnostics, Dartford, UK). Positive tests were confirmed by the presence of antibodies directed against both envelope (gp1201160 or gp41) and core (~24) antigens in a Western blot (Sorin Biomedica SpA, Saluggia (VC), Italy). HIV-2 infection has not been reported in Kenya. Thus, HIV-l infection is referred to as HIV-positivity or seropositivity throughout the text. Mycobacterial microscopy and culture In the case of patients presenting to IDH or NACC initially, sputum or other relevant clinical material was

in Nairobi

47

routinely sent to the National Public Health Laboratories (NPHL) Tuberculosis Laboratory for both light microscopy after Ziehl-Neelsen staining, and culture (see below), before treatment was begun. At other health institutions only sputum microscopy with Ziehl-Neelsen staining was performed. Patients who came to IDH or NACC within 1 month of start of treatment generally had repeat specimens sent to the NPHL TB Laboratory. Decontamination of sputum was carried out by mixing equal volumes of 4% sodium hydroxide and sputum and shaking for 15 min. After centrifugation, the supernatant was poured off and the deposit neutralized with 10% hydrochloric acid and inoculated onto Liiwenstein-Jensen medium. Both glycerol and pyruvate-containing media were used. Incubation was at 37” for up to 8 weeks. Screening for ‘atypical’ mycobacteria was performed by observing growth at 25”C, colonial morphology, pigment production, and by growth on Lowenstein-Jensen medium containing glycerol and 500 mg/l paranitrobenzoic acid. Results of mycobacterial microscopy and culture were obtained from case records and/or from the NPHL TB Laboratory.

Statistical analyses Chi-squared, the Mantel-Haenszel (MH) chi-squared test, Fischer’s exact test (2-sided), and logistic regression were used. Where a test other than chi-squared with 1 d.f. was used, it is mentioned in the text. All confidence intervals (CI) were calculated at the 95% level.

RESULTS Seroprevalence

data and risk factors

Concordance was obtained between the two ELISA tests used in all but 2 patients (both from NACC) who were negative on Wellcozyme but positive on the DuPont ELISA. These were both Western blot negative and are, therefore, defined as HIV-negative. All remaining positive ELISA tests were confirmed by Western blot. The prevalence of HIV seropositivity at IDH was 25% (521196) compared to 9% (18/195) at NACC (Relative risk (RR) = 2.87, CI: 1.75-4.73). HIV seropositivity varied with age but the difference was not statistically significant (Table 1) (X2 = 6.85, 3 d.f., P = 0.07). The peak prevalence occurred in the 25-44 year age group. At IDH, more females were infected with HIV than males (Mantel-Haenszel (MH) Relative Risk, allowing for age, 0.87, CI: 0.97-2.5, P = 0.09), while at NACC there was no difference. Widowed or divorced tuberculosis patients had a higher HIV-seroprevalence than married or single people (Table 2) and this remained so when allowing for age, sex, and whether patients attended IDH or NACC (clinic). Likewise, residence within Nairobi city boundary was associated with a higher level of HIV-

48

Tubercle and Lung Disease

Table 1. HIV seropositivity sex

of IDH and NACC patients by age and

Age

NACC No. HIV+/total (%) Men Women

Total HIV+/total (%)

15-24 25-34 354 45+

2/19(10) 12/28(32) 9/32(28) l/20 (5)

7/25(28) 11131(35) 6/14(43) 3/13(23)

O/23 (0) 10/63(16) 2/24(8) l/27(4)

l/24(4) 3/20(15) l/9(1 1) O/5(0)

10/91(11) 36/152(24) 18/79(23) 5/65(8)

Total

24/109*(22)

27/83(32)

13/137(9)

5/58(9)

69/387(18)

Site

PulmonaryonIy*

*Age not known for 4 patients, 2 men (1 HIV-positive)

Table 2. Association between HIV-seropositivity

and 2 women.

and potential risk

factors Variable

Number of Patients

Odds Ratio (95% confidence intervals) for HIV-seropositivity adjusted* for age, sex, clinic

P

Marital status Married Single Widowed/divorced

224 130 36

1 0.94 (0.46-1.95) 3.67 (1.56-8.67)

0.025

Residence? Nairobi Other

257 129

1 0.44 (0.23-0.84)

0.01

67 77 131 113

1 1.11 (0.46-2.68) 0.83 (0.35-1.93) 0.84 (0.33-2.11)

0.8

38 40 268 39

0.82 (0.32-2.14) 1.45 (0.62-3.34)

0.7

(year)

Streptomycin injections in past 5 years (including previous and current courses) O-9 l&27 28-30 31+

to anatomical

site of

Confirmed Unconfirmed MiGaly (radiographically defined)

Exbqndmomuy Lymphatic Pleural Other

Total * t t 8

NACC HIV+ HIV-

HIV+/total(%)

109 103 6

16 13 3

154 143 11

50/313 43/289 7/24

(16) (15) (30)

2

9

0

2

2113

(15)

16

26

2

21

18165

(28)

5t 107 3

2* 24@ 35

1 1 0

6 10 5

6/14 11/45 3/14

(42) (24) (21)

52

144

18

177

70/391

(18)

IDH HIV+

HIV-

34 30 4

Excludes miliary disease. 2 patients had both pleural and lymphatic tuberculosis. 1 patient had both pleural and lymphatic tuberculosis. 2 patients had both pleural disease and tuberculosis elsewhere, spinal tuberculosis and 1 with abdominal tuberculosis.

1 with

pared to 110/133 (83%) of the HIV-negative patients. Corresponding figures for NACC were 1 l/15 (73%) and 127/154 (82%) respectively (MH, RR, allowing for clinic, 0.87, CI:O.74-1.03, P = 0.09). At IDH 36/47 (77%) of HIV-positive patients with sputum culture results available were culture-positive, while 112019 (94%) of HIV-negatives were culture-positive (RR = 0.81, CI: 0.69-0.96). Corresponding figures for NACC were 12/12 (100%) and 125/133 (94%) (RR = 1.06, CI:1.02-1.11, Fisher’s exact test, P = 1.0). The smear-negativity rate among culture-positives was higher among HIV-positive than HIV-negative patients (7/33 (21%) vs 16/115 (14%) at IDH and 4/12 (33%) vs 23/125 (8%) at NACC) but the difference was not significant (MH RR, allowing for clinic, 1.63, CI:O.90-2.96, P = 0.18).

:.08 (0.46-2.57)

*Adjusted Odds Ratio and significance tests obtained by logistic regression. t 4 prisoners excluded. Data were unobtainable on some patients, so not all sections add up to total.

seropositivity. Multivariate analysis shows marital status and residence were independently related to HIV seroprevalence. There was no significant relationship between years of education or numbers of streptomycin injections in the previous 5 years (including the current course) and HIV seroprevalence (Table 2).

Diagnosis of tuberculosis Among the cases 92% (CI 89-95%) bacteria other than 35 of 49 (71%) had sputum smears

according

tuberculosis

IDH No. HIV+/total (%) Men Women

Education
Table 3. HIV seroprevalence

with pulmonary tuberculosis alone, were confirmed (Table 3). No mycoM. tuberculosis were isolated. HIV-positive patients from IDH who performed were smear-positive com-

Clinical characteristics At IDH, 16/42 (38%) patients with extrapulmonary disease were HIV-positive, while 34/143 (24%) with pulmonary disease alone had HIV infection (RR = 1.6, CI:O.99-2.6). Five out of 7 of those with lymphatic tuberculosis at IDH were HIV-seropositive, compared to 47/189 (25%) of all other IDH patients (RR = 2.87, CI:1.69-4.88, Fisher’s exact test P = 0.015). Milky disease was not significantly associated with HIV infection; only 2/13 (15%) patients with a milky pattern on chest radiography were HIV-seropositive compared to the 68/378 (18%) of the remaining patients (RR = 0.86, CI:O.23-3.12, Fisher’s exact test, P = 1.0). Clinical findings significantly more common among HIV-seropositives at IDH (Table 4) were persistent diarrhoea, oral candidiasis, generalized itchy skin rash, current or past herpes zoster and palpable lymph nodes present in more than one extrainguinal site. However, 17/52 (33% CI:19.9-45.4%) had none of these symptoms or signs and neither did they have Kaposi’s sarcoma, dissem-

HIV infection and tuberculosis

Table4. Number Clinical finding

of HIV-seropositive

and seronegative

IDH HIV+ n=52(%)

HIVn=144(%)

symptoms* Weight loss Fevenl month Diarrhoetilmontb

42 (81) 37 (71) 13 (25)

144 (79) 91 (63) 12 (8)

Cough>1 month

43 (83)

114 (79)

8 (15) 9 (17) 6 (12)

1 (1) 9 (6) 0 (0)

Lymphadenopathy

22 (42)

22 (15)

0.0001

Kaposi’s sarcoma

2 (4)

0 (0)

0.076

Signst Candidiasis Itchy rash Herpes zoster

Complications of therapy At IDH, a history of skin rash following the start of antituberculosis therapy was obtained in IO/52 (19%) HIV-seropositive patients and 16044 (11%) HIVseronegative patients (RR = 1.73, CI:O.84-3.57), but a history suggestive of Stevens-Johnson Syndrome (a skin rash involving most of the body together with the involvement of either oral, genital or ocular mucosa) was obtained in 7/52 (13%) patients with HIV infection and in 4044 (3%) patients without (RR = 4.85, CI:1.45-15.88, Fisher’s exact test, P = 0.01). Almost all patients were treated with standard chemotherapy, so it was not possible to implicate a particular drug in this study. At NACC, only 6 patients reported a skin rash (all HIV-negative) and 1 had a history suggesting Stevens-Johnson Syndrome (HIV-positive).

P

NACC HIV+ n=18(%)

HIVn=177(%)

0.96 0.39 0.004

3 (17) 1 (6) 0

13 (7) 6 (3) 0

0.17$ 0..50* _

0.73

2 (11)

12 (7)

0.62%

0.0001~ O.CM 0.001~

1 (6) 1 (6) 0 (0)

0 (0) 4 (2) 0 (0)

0.09* 0.39* _

1 (6)

1 (1)

0.18$

0 (0)

0 (0)

(IDH) or time of recruitment

inated herpes simplex infection, or cryptococcal meningitis; that is, they bad none of the features specific for HIV infection used in the WHO Criteria for the Clinical Diagnosis of AIDS.13 At NACC, the proportion of HIVseropositives with no clinical evidence of HIV infection was 15/18 (83%, CI:66-100%). Owing to the fact that at IDH the symptoms reported were those at the time of presentation and the signs those found at recruitment, while at NACC both symptoms and signs were those at time of recruitment, the symptoms of the two groups cannot be compared. However, if we assume that the signs at recruitment were also present at presentation, the WHO criteria were met in 33/52 HIV-positives at IDH, giving a sensitivity of 63%, but were also found in 81/144 HIV-negatives, yielding a specificity of only 44%, and a positive predictive value of 29%. At IDH, 8 (29%) of the 28 patients with a history of previous treatment of tuberculosis were HIVseropositive, compared to 44 (26%) of the 168 with no such history. At NACC too few patients had had previous treatment to allow analysis.

49

patients with selected clinical findings

P

* Symptoms reported at time of presentation T Signs observed at time of recruitment. * Fisher’s exact test.

in Nairobi

-

(NACC).

Chest radiography 154 patients with confirmed pulmonary tuberculosis, with or without concomitant extrapulmonary disease, at IDH and 135 at NACC had chest radiographs taken at the time of presentation available for analysis (Table 5). HIV-seropositive patients had significantly fewer zones affected (P= 0.04at IDH and 0.02 at NACC) and were more likely to have disease in either middle or lower zones, with relative sparing of the upper zones. Cavitation was less common among HIV-infected patients (MH RR, allowing for clinic, 0.80, CI: 0.62-1.02, P = 0.06), and lymphadenopathy was more common (MH RR, allowing for clinic, 2.06, CI: 1.16-3.68, P = 0.03).

DISCUSSION The seroprevalence of HIV infection was significantly greater at IDH than at NACC. Both seroprevalence rates were significantly higher than the 2% quoted for blood donors in Nairobi” during a similar period. These results further strengthen the association of HIV infection and tuberculosis in Kenya. The higher rate for IDH compared to NACC probably reflects the fact that IDH is a referral centre for tuberculosis patients from all over Kenya, and attracts patients with difficult problems, some of which are likely to be HIV-related. Similar results were found in a study in Kinshasa, Zaire, where HIVseropositivity was found in 85/234 (36%) sanatorium patients and, in the same year, in 17% of 509 outpatients.6 If groups of tuberculosis patients are to be used for sentinel surveillance of the spread of HIV infection such factors need to be taken into account in interpreting the results. The actual prevalence of HIV in all cases of tuberculosis may in fact be higher than these figures since smear-negative, culture-negative patients with suspected tuberculosis and HIV infection were not included in this study unless a clinical response to antituberculosis

50

Tubercle and Lung Disease Table 5. Radiological cases only

findings in HIV-seropositive

Finding*

Total zones affected (mean) 0

IDH HIV+ n=34(%)

3.2

and seronegative

P HIVn=120(%)

3.8

patients: confirmed

NACC HIV+ n=14(%)

0.04

2.9

pulmonary

P HNn=121(%)

4.0

0.02

No. of upper zones affected 0 1 2

5 (15) 20 (59) 9 (26)

15 (13) 48 (40) 57 (47)

0.087

6 (40) 7 (47) 1 (7)

23 (18) 40 (31) 58 (46)

0.01t

Cavitation Lymphnodes Pleural effusion Pericardial effusion

22 (67) 9 (27) 7 (21) 1 (3)

95 (80) 18 (17) 18 (15) l(1)

0.24 0.29 0.57 -

5 (36) 2 (14) 0 0

71(59) 4 (3) 6 (5) 0

0.005 0.23 0.86

* Radiographs unavailable T x2 test for trend.

in 23 patients (IDH) and 53 patients (NACC).

therapy was seen together with typical radiographic features or extrapulmonary disease. HIV infection can lead to an atypical radiographic picture and advancing HIV disease could overshadow a response to antituberculosis therapy. It is also important to emphasise that this study was of patients at all stages of their treatment. Thus, if a substantial early mortality occurred in HIV-positive patients, the seroprevalence among new cases would be even higher. The most likely explanation for the association of tuberculosis and HIV infection is the reactivation of latent tuberculous infection brought about by the immunosuppression caused by HIV. This is supported by the study of Selwyn et aLlo which showed an annual incidence of 8% among previously tuberculin skin test positive drug users with HIV infection compared to 0.3% among tuberculin skin test negative drug users. In addition, where tuberculosis is common, the risk of infection is higher and an immunocompromised host is presumably at greater risk of developing disease from an exogenous infection or reinfection. This may account for the less frequent occurrence of upper zone disease and cavitation in the HIV-positive as compared to the HIVnegative patients with pulmonary tuberculosis. In some resource-poor countries, an alternative hypothesis for the HIV-tuberculosis association is that HIV infection could be contracted in tuberculosis patients by the use of poorly sterilized injection equipment for the administration of streptomycin. If this were the case, we would expect to find an association between numbers of streptomycin injections received and HIV infection. No such association was found. Nor was there any relation between length of treatment and HIV-positivity (data not shown). The associations of HIV infection with the age range 25-44 years, residence within Nairobi and with being divorced or widowed are all compatible with the known transmission dynamics of HIV. The apparent lack of an

association between previous treatment and HIV infection does not necessarily mean that HIV infection does not predispose to relapse of tuberculosis. To assess HIV as a risk factor for relapse, the correct comparison would be the HIV prevalence among relapsed cases compared to the rate among those originally treated around the same time but who have not relapsed. In this study we have looked at tuberculosis cases only and can only say that HIV appears to be an equally important risk factor for both new cases of tuberculosis and relapsed cases. It is likely, in any case, that the HIV epidemic is at a comparatively early stage in Kenya and there are insufficient people with both a previous history of tuberculosis and HIV infection to give a substantial number of relapses. Alternatively, the mortality among HIV-positive people due to causes other than tuberculosis may be sufficiently high as to preclude a noticeable increase in the relapse rate. Combining the two study sites, over 70% (CI 59-81%) of the HIV-positive patients in the studies had pulmonary tuberculosis. Previous reports, mainly from the USA,rd have suggested that HIV-associated tuberculosis is more often extrapulmonary than HIV-negative tuberculosis. In this study, we also found HIV infection to be more common in extrapulmonary tuberculosis than in pulmonary tuberculosis, but not significantly so, with the exception of lymphatic tuberculosis. This may be due to the fact that, for historical reasons, smear-positive patients are usually referred from the main Kenyatta National Hospital (KNH) to IDH, while extrapulmonary cases remain at KNH. However, even if we assume that the extrapulmonary component was underestimated in this study, a significant amount of HIV-associated tuberculosis is pulmonary and smear-positive, and thus infectious. This may explain the upsurge in the annual incidence of tuberculosis currently being seen in some African countries.8 Not surprisingly, just over half of all patients at IDH

HIV infection and tuberculosis

and NACC with HIV infection have at least one clinical feature suggestive of such an infection. However, 46% (CI:34-58%) give no clinical indication of HIV infection. These patients can thus only be shown to be infected with HIV by the use of serological tests. This reinforces the importance of educating all personnel involved in health care work in countries known to have a significant level of HIV infection to take precautions against occupational exposure to HIV with all patients. Severe cutaneous adverse reactions were reported more commonly among HIV-positive patients. Nearly 90% of the patients in this study were treated with the standard regimen of streptomycin, isoniazid and thiacetazone, so it was impossible to determine if a single drug was responsible, but it is commonly held, however, that thiacetazone is responsible. It is essential to determine by prospective studies if this is truly the case as a number of African countries are currently using or considering the introduction of a short-course treatment regimen which includes thiacetazone in the continuation phase.15 HIV-associated tuberculosis generally has a more complex presentation with more subsequent complications than HIV-negative tuberculosis. For the increasing number of patients already presenting with HIV-associated tuberculosis, more resources will need to be allocated to treatment centres by national tuberculosis programmes if the credibility and efficacy of tuberculosis treatment services is to be maintained. Acknowledgements We thank all the patients and staff at IDH and NACC; Dr W. Watkins and staff at the Wellcome Trust Research Laboratories (WTRL) for their help; MS Maria McDonnell for typing the manuscript; Jo Morris for statistical assistance and the Director, Kenya Medical Research Institute (KEMRI) for permission to publish these findings. The study was conducted through the KEMRI/WTRL Collaborative Research Programme.

in Nairobi

5I

References 1 Pitchenick A E, Cole C, Russell B W, Fischl M A, Spira T J, Snider D E. Tuberculosis, atypical mycobacteriosis, and the acquired immunodeficiency syndrome among Haitian and nonHaitian patients in South Florida. Ann Intern Med 1984; 101: 641-645. 2 Centers for Disease Control. Tuberculosis and the acquired imrnunodeficiency syndrome - Florida. MMWR 1986; 35: 587-590. 3 Centers for Disease Control. Tuberculosis and AIDS Connecticut. MMWR 1987; 36: 131, 134-135. 4 Pitchenik A E, Burr J, Suarez M, Fertei D, Gonzalez G, Moas C. Human T-cell lymphotropic virus III (HTLV-III) seropositivity and related disease among 71 consecutive patients in whom tuberculosis was diagnosed. Am Rev Respir Dis 1987; 135: 875-879. 5 Soriano E, Vallolas J. Gate11 J M et al. Characteristics of tuberculosis in HIV-infected patients: a case control study. AIDS 1988; 2: 429-432. 6 Colebunders R L, Ryder R W, Nzilambi N et al. HIV infection in patients with tuberculosis in Kinshasa, Zaire. Am Rev Respir Dis 1989; 139: 1082-1085. 7 Elliott A, Luo N, Tembo G et al. Impact of HIV on tuberculosis in Zambia: a cross sectional study. Br Med J 1990; 301: 412415. 8. Centers for Disease Control. Tuberculosis provisional data United States 1986. MMWR 1987; 36: 254-255. 9. Slutkin G, Leowski J, Mann J M. Tuberculosis and AIDS. The effects of the AIDS epidemic on the tuberculosis problem and tuberculosis programmes. Bull Int Union Tuberc Lung Dis 1988; 63: 214. 10 Selwyn PA, Hartel D, Lewis V A, Schoenbaum E, Vermund S, Klein R, Walker A, Friedland G. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N Engl J Med 1989; 320: 545-550. 11. Nunn P P, Kibuga D, Elliott A, Gathua S. Impact of human immunodeficiency virus on transmission and severity of tuberculosis. Tram R Sot Trop Med Hyg 1990; 84 (buppl 1): 9-13. 12. Citron K M. Girline D J. Tuberculosis. In: Weatherall D J. Ledingham J G G, -Warrell D A, eds. Oxford textbook of medicine. 1st ed. Oxford: Oxford Medical Publications, 1983: p 255. 13. World Health Organization/Centers for Disease Control. Case definition for AIDS. Wkly Epidemiol Ret 1986; 61: 69-76. 14. Ministry of Health. National guideline on counselling for HIV infection. Nairobi, Kenya. December 1988; 2. 15. Styblo K. Overview and epidemiologic assessment of the current global tuberculosis situation with emphasis on control in developing countries. Rev Infect Dis 1989; 2 (suppl2): 5339-5346