Clin Invest Med 1996; 19 (4): 271-278
(Original manuscript submitted July 31, 1995; received in revised form Apr. 23, 1996; accepted Apr. 30, 1996)
Copyright 1996, Canadian Medical Association
Design: Case review.
Participants: All known patients with TB and HIV infection in British Columbia, in whom TB was diagnosed between 1990 and 1994. This group was compared with those in whom TB was diagnosed between 1984 and 1990.
Outcome measures: Patients' demographic characteristics and risk factors for HIV infection, site of TB, occurrence of drug-resistant TB, treatment and outcome.
Results: Forty-four patients with HIV infection and TB were identified, of whom 16% were women, whereas none of those diagnosed from 1984 to 1990 were women, and 14 (32%) were aboriginal Canadians, compared with only 3 (8%) of those diagnosed from 1984 to 1990 (p * 0.01 for both). Forty patients had identifiable risk factors for HIV infection. A smaller proportion of the recent group than of the previous group were homosexual men (excluding those for whom risk factors were not known, 17/33 men [52%] in 1990 to 1994 v. 36/39 [92%] in 1984 to 1990), and a larger proportion were intravenous drug users (22/40 [55%] in 1990 to 1994 v. 8/39 [21%] in 1984 to 1990, p <0.01 for both). Since 1984 an increasing proportion of patients with TB diagnosed each year have also had HIV infection (linear trend p <0.001). Drug resistance was not found in any cultures taken at the time of diagnosis; however, rifampin resistance developed 7 months after therapy was initiated in one patient. Of the 40 patients who started therapy, 24 had directly observed therapy. Nine patients died while receiving therapy, and four died without receiving any antituberculous therapy. TB was the cause of death, or a contributing factor to death, in five cases.
Conclusions: Significantly more intravenous drug users, aboriginal Canadians and women are now presenting with HIV-related TB in British Columbia. All HIV-positive patients need to be evaluated for TB, and HIV infection must be considered in assessing all newly diagnosed TB cases and in screening contacts of active cases, especially if patients have risk factors for HIV infection.
[Table of Contents]
Résumé
Objectifs : Identifier les sujets porteurs concomitants d'une infection à VIH et de tuberculose. Analyse des tendances récentes de prévalence et des facteurs de coinfection.
Conception : Revue rétrospective de cas.
Sujets : Tous les patients connus atteints de tuberculose et d'infection à VIH en Colombie-Britannique et chez lesquels la tuberculose avait été diagnostiquée entre 1990 et 1994. Ce groupe a été comparé avec les sujets dont le diagnostic avait été porté entre 1984 et 1990.
Variables mesurées : Caractéristiques démographiques, facteurs de risques pour l'infection à VIH, localisation de la tuberculose, présence de tuberculose à souches résistantes, traitement et évolution.
Résultats : Quarante-quatre sujets atteints d'infection à VIH et de tuberculose concomitante ont été identifiés. Alors que de 1984 à 1990, il n'y avait aucun sujet féminin et que seulement 3 (8 %) sujets étaient des Autochtones, 16 % étaient des femmes et 32 % étaient des Autochtones dans le groupe plus récent (p <0,01 dans chaque cas). Quarante sujets avaient des facteurs de risques identifiables pour l'infection à VIH. Une proportion plus faible du groupe plus récent était homosexuelle (17/33 hommes [52 %] entre 1990 et 1994 contre 36/39 [92 %] de 1984 à 1990), et une plus grande proportion était constituée de toxicomanes consommateurs de drogues injectables (22/40 [55 %] de 1990 à 1994 contre 8/39 [21 %] de 1984 à 1990, p <0,01 dans les deux cas). Depuis 1984, sur une base annuelle, une proportion croissante de sujets chez lesquels la tuberculose a été diagnostiquée sont aussi porteurs d'infection à VIH (tendance linéaire p <0,001). Aucune souche résistante n'a été observée dans les cultures obtenues au moment du diagnostic. Par contre, la résistance au rifampin a été notée 7 mois après le diagnostic chez un sujet avec tuberculose persistante. On a fait le suivi direct du traitement chez 24 des 40 sujets qui débutèrent un traitement anti-tuberculeux. Neuf décès survinrent sous traitement, et quatre décès survinrent sans aucun traitement anti-tuberculeux. La tuberculose était la cause du décès ou un facteur contributif chez cinq sujets.
Conclusions : En Colombie-Britannique, un nombre significativement plus grand de toxicomanes consommateurs de drogues injectables, d'Autochtones et de femmes sont maintenant porteurs de tuberculose associée à l'infection à VIH. Tous les sujets porteurs d'infection à VIH doivent être évalués pour la tuberculose, et l'infection à VIH doit être envisagée dans l'évaluation de tout nouveau patient atteint de tuberculose de même que dans l'évaluation des personnes qui ont été en contact avec les sujets porteurs de maladies actives, particulièrement si les sujets ont des facteurs de risques pour l'infection à VIH.
We examined the cases of coinfection with HIV and TB reported in British Columbia since 1990 and compared them with cases from 1984 to 1990 that we had previously studied[6] to test our impression that a significant change had occurred in the demographic profile of this subgroup of patients with TB.
[Table of Contents]
Methods
TB is a reportable disease in British Columbia, and all cases are recorded in a registry maintained by the Division of Tuberculosis Control of the British Columbia Ministry of Health. Since distribution of antituberculous drugs is controlled directly by the Division of Tuberculosis Control, all treated cases are sure to be registered. AIDS is also a reportable disease in British Columbia, but it is reported anonymously. HIV infection per se is not reportable. Annual reports from the Divisions of Tuberculosis Control and Sexually Transmitted Diseases/AIDS Control were reviewed to obtain the number of cases of each disease reported annually.
The patient information form for each TB case diagnosed in British Columbia includes a list of factors that predispose to TB, including HIV infection and AIDS. Therefore, information on HIV-related TB was collected systematically during the study period. We reviewed the registry for all new cases of active TB from 1990 through 1994. We excluded cases from 1984 to 1990 that we had previously reviewed.[6] However, some of the cases diagnosed in 1990 are included in this report because they were not yet registered when we conducted our earlier review. We analysed the following data, collected on the standardized forms, for all cases: demographic characteristics, medical history including risk factors for HIV infection, presenting signs and symptoms of TB, results of microbiologic tests, TB treatment, adverse reactions and outcome. This group was then compared with the 40 cases identified from 1984 to 1990 with the use of similar methodology.[6]
Cases of active TB were defined according to the Classification and Reporting of Tuberculosis in Canada,[7] revised in January 1990.[8] Unless otherwise stated, "pulmonary TB" includes pleural disease.
The data were analysed with the use of SAS software (version 6.01, SAS Institute, Cary, NC) in conjunction with EpiInfo (version 5.00, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta). Fisher's exact test and the chi-sqared test were used to evaluate differences between groups. A p value of less than 0.05 was considered significant.
[Table of Contents]
Results
We identified 44 people (37 men and 7 women) with HIV infection and TB diagnosed between 1990 and 1994 (Table 1). Their mean age was 38 years (standard deviation 9.1, range 23 to 65 years). Of these people, 31 (70%) were born in Canada, 8 (18%) were from countries with a high prevalence of TB and 2 were from countries with a low prevalence of TB. For three of the people, the country of birth was unknown. Twenty of the people were white, 14 were aboriginal Canadians, 4 were of Hispanic descent, 3 were of African descent and 3 were of Asian descent. In terms of high-risk groups, 22 were intravenous-drug users (IVDUs), 17 were homosexual men, 8 had multiple heterosexual partners (including prostitutes) and 1 had had a blood transfusion. For four of the people, risk factors were unknown. There was an overlap between the IVDUs and other groups: eight people had two risk factors.
When the recent group was compared with the cases previously reported, there were significant differences in sex, in the proportion of aboriginal Canadians and in risk factors for HIV infection (sexual orientation and intravenous drug use) (p <0.01 for all differences, Fig. 1). When cases were evaluated by year of diagnosis there was a significant linear trend: an increasing proportion of newly diagnosed TB cases involved people who also had HIV infection (p <0.001). From 1990 to 1994, 1543 TB cases were reported, of which 58 (3.8%) also had documented HIV infection. (This figure includes 14 patients in whom TB was diagnosed in 1990 who were included in our earlier review.[6]) The proportion of newly diagnosed AIDS cases in which the patients also have TB cannot be assessed accurately. Because TB develops after AIDS is diagnosed, TB status is not recorded in the AIDS registry. Also, patients with HIV infection and pulmonary TB but no other opportunistic infections would not have been considered AIDS cases before the case definition of AIDS was changed in 1993.
Three patients had a documented history of TB. Two of these patients had received adequate treatment for primary TB during childhood, and both had had a reaction to a tuberculin skin test of 20 mm or more in the mid-1980s.
Sixteen patients (36%) had a history of alcoholism, and 3 had documented alcoholic hepatitis. Nine additional cases of hepatitis, including four cases of hepatitis B, were reported among the patients.
Two of the seven female patients were pregnant during treatment. One delivered a healthy infant at term; the infant is receiving zidovudine orally and is being tested for HIV infection. The second woman delivered at term; however, her infant died shortly after birth and had HIV infection and disseminated cytomegalovirus infection.
Seventeen patients (39%) had pulmonary disease alone, and an additional six (14%) had both pulmonary and extrapulmonary disease. Twenty-one (48%) had extrapulmonary disease alone, including 15 cases of miliary disease. There were six cases of pleural disease, of which three were associated with pulmonary parenchymal disease and two with disease in extrapulmonary sites.
Forty-two of the patients underwent initial evaluation in British Columbia, of whom 41 had bacteriologically proven disease. The 42nd patient, the only patient identified by contact tracing, had a negative result of a culture of his pleural fluid. The two remaining patients had moved to British Columbia from Ontario and New York, respectively, after diagnosis and initiation of therapy, and their initial culture results were unavailable. Twenty-six of the patients had a positive result of a sputum culture, and 11 (42%) of these patients also had a positive result of a smear. Four additional patients had a positive result of a culture of a sample obtained by bronchoscopy, and four patients who had a negative result of a sputum smear had a positive result of a smear of specimens obtained by bronchoscopy. Additional positive results of cultures were obtained from samples of pleural fluid or biopsy, lymph nodes (in 11 cases), peritoneal fluid, intestinal biopsy, urine, cerebrospinal fluid (in one case), blood and bone marrow.
In one case, an organism resistant to rifampin was obtained from a necrotic lymph node. The patient had received 7 months of therapy, but the patient's compliance had been poor. Initial lymph-node sampling from a different region yielded a strain of Mycobacterium tuberculosis sensitive to all of the drugs tested. The patient died 6 months later, while still receiving treatment. Analysis of the two specimens by restriction-fragment-length polymorphism showed the organisms to be identical. No other cases of drug resistance were documented.
Of the 40 patients who started treatment, 16 (40%) had completed treatment and 11 (28%) were being treated in British Columbia at the time of data analysis. Nine patients (23%) died before treatment was completed, and two patients left the province while still receiving therapy. Two patients failed to complete therapy, one because of drug toxicity and one because of noncompliance. Four patients died before receiving any antituberculous treatment.
Twenty-four patients (60%) received directly observed therapy (DOT); of these, 14 had good or excellent compliance (taking least 80% of prescribed doses). None of the patients on DOT had a relapse, whereas 2 of the 16 patients on unsupervised therapy had a relapse. However, the statistical significance of this difference was not significant. One patient who had a relapse initially received isoniazid, rifampin and pyrazinamide. Rifampin was discontinued after 2 months because of an associated rash, ethambutol was added and 9 months of therapy were completed. However, the patient died of disseminated TB 5 months after completion of therapy. The other patient who had a relapse had pleurisy and a positive result of a culture of specimens obtained from bronchial washings and pleural fluid. He completed 6 months of therapy, with an initial three-drug regimen of isoniazid, rifampin and pyrazinamide, and a 4-month follow-up of isoniazid and rifampin; however, he had a relapse involving miliary TB 8 months later. He responded to a new course of therapy, and TB cultures continued to show that the strains were sensitive to all of the drugs tested. One patient on supervised therapy had persistent disease.
Adverse drug reactions were reported among 17 (44%) of the patients receiving treatment, and changes in therapy were required in 11 cases (27.5%). The types of reactions in these 11 cases were elevated aspartate aminotransferase level (one case), nausea (three cases), rash (three cases), paresthesia (two cases) and other or unspecified reaction (five cases). In the earlier group, 12 of the 36 patients receiving treatment had adverse reactions that led to a change in therapy.
Overall, 17 patients (39%) had died by the completion of data collection. In five of these deaths, TB was reported as the cause of death or a contributing factor to death. One patient who did not receive antituberculous therapy died 1 month after specimens were obtained for mycobacteriologic tests. The result of the smears was negative, and only the culture result, received after the patient had died, had a positive result. Two patients with miliary TB died while taking treatment, one after 2 days of treatment and one after 4 months. Two patients with pulmonary TB each received 9 months of unsupervised therapy; one had a relapse and died from disseminated TB, as described earlier, and the other still had active disease at the time of death.
[Table of Contents]
Discussion
Since the first case of AIDS in British Columbia was reported in 1984, there have been 84 reported cases of HIV-associated TB in the province. Distribution of antituberculosis drugs is directed by the Division of Tuberculosis Control in British Columbia; therefore, all TB cases are reported. However, on the basis of record-linkage studies of the Canadian Mortality Database,[4] it is estimated that cases of AIDS are underreported by approximately 15%. The 3.8% overlap of AIDS and TB cases in British Columbia is similar to the 4.0% rate reported in the United States;[9] however, in a 1991 survey of TB clinics in the United States conducted by the Centers for Disease Control and Prevention, the median prevalence of HIV seropositivity was 8.2%,10 indicating an even greater overlap in the populations with these two infections. In contrast, in England and Wales only 0.42% of TB patients were also on the AIDS registry in 1988.[11]
Active pulmonary TB has been considered an AIDS-defining disease since 1993; therefore, all HIV-positive patients diagnosed with active TB meet the criteria for AIDS. They represent 4.4% of the 980 cases of AIDS reported in British Columbia from 1990 to 1993. In Quebec, where the demographic profile of HIV and TB coinfection is quite different, 4.5% of all reported cases of AIDS have also been diagnosed as having TB.[12]
In our study, recently identified patients with combined HIV infection and TB are more likely to be IVDUs, aboriginal Canadians or women and less likely to be homosexual men than in the earlier group identified. These trends reflect the changing demographic profile of new AIDS cases in Canada[4] and the higher rates of TB in groups with increasing rates of HIV infection. The potential for overlap in dually infected patients is particularly great among those who are aboriginal. Even in the absence of HIV-related disease, in 1993 the rate of TB in aboriginal Canadians in British Columbia was five times higher than the rate in the rest of the population (Division of TB Control, British Columbia Ministry of Health: unpublished data, 1993). Similarly, even before the AIDS epidemic, IVDUs in New York City were found to have rates of TB at least six times higher than other patients treated in the same hospital.[13]
We cannot exclude the possibility that serologic tests for HIV have been carried out more frequently among patients with newly diagnosed TB during recent years. However, given the clear change in risk and ethnic groups affected, as well as the change in the sex of patients affected, we feel that changes in testing could not account for the large increase that we have reported. The demographic changes we have shown have implications for local TB control strategies, since an increasing proportion of the TB cases reported annually in British Columbia also involve HIV infection. A rapid increase in the seroprevalence of HIV among IVDUs[14] is also possible, leading to concern that the overall prevalence of TB in British Columbia may begin to increase.
Reported treatment-completion and cure rates among high-risk groups vary widely depending on the treatment strategies employed.[15-18] Of our patients, 60% received DOT, whereas three of the four patients who had a relapse or persistent disease received unsupervised therapy. No drug resistance was observed in the initial specimens cultured; however, rifampin resistance developed in one patient who had poor compliance with therapy.
DOT has an important role in improving outcomes among patients with TB and HIV coinfection, as shown by the results of other studies of treatment and outcome. One study showed that, of 178 TB patients discharged from Harlem Hospital in New York City, 89% failed to complete standard outpatient therapy. Of this group, 68% were homeless, and 80% of those tested were HIV-positive.[15] Another study followed a group of patients in which 41% were homeless and 21% were IVDUs.[16] Compliance was assessed by pill count and urine check. Patients on self-administered therapy who were poorly compliant were transferred to DOT and were given food and shelter vouchers as incentives. Those who failed DOT left the area and could not be traced. Thanks to these measures, antituberculous treatment failed owing to poor compliance in only 4 of 89 patients with HIV infection. In a TB outbreak in a Seattle shelter for homeless men, only 6 of 23 men failed therapy (experiencing persistent disease or relapse) after DOT was recommended.[17] In five of these six, isoniazid resistance developed. The inducement to participate in DOT in this setting was continued access to shelters in Seattle. DOT has also been used to treat more than 90% of cases in Tarrant County, Texas, since late 1986, and has resulted in significant (p <0.001) reductions in drug resistance and relapse.[18] Our data, although descriptive, support the need to consider DOT in all cases, as recommended elsewhere.[19,20] Recent data from Zaire suggest that an additional 6 months of treatment with isoniazid and rifampin may decrease relapse rates in patients with HIV infection and TB, although overall death rates in the two groups studied did not differ.[21] Additional study is needed to determine whether longer antituberculous therapy regimens would benefit patients with HIV infection.
Two of the seven women in the recent group we identified were pregnant. Since newborns are very susceptible to TB, antituberculous therapy must be delivered effectively to pregnant women in order to prevent infection of their children.[22] Knowledge of whether the patient is pregnant is also important in selection of the drugs for therapy. Streptomycin is the only first-line drug contraindicated in pregnancy. There are limited data on the safety of pyrazinamide; however, it is not contraindicated in pregnancy.[22] According to recent evidence, therapy with zidovudine should be considered for new mothers with HIV infection during pregnancy and delivery to decrease the likelihood of HIV transmission to the infant.[23] Zidovudine therapy was started at 14 weeks' gestation in one of the pregnant subjects of our study. Unfortunately, therapy was not successful in preventing vertical transmission.
Six of the patients in this study reported prior close contact with people with active TB; however, only two of these patients were identified through the index case. One of these patients started treatment after having been identified and evaluated as a contact. The other was not available for tuberculin skin testing, and he had symptoms within 2 months. No recent tuberculin skin-test results were given for the other reported contacts. Since isoniazid prophylaxis in patients with HIV and TB coinfection has been shown to decrease the incidence of active TB,[3,24] some of the six cases resulting from contact with a patient with TB could probably have been prevented if the patients' skin-test reaction had been assessed and isoniazid prophylaxis offered.[20]
All reports of positive results of tests for HIV
in British Columbia are accompanied by an explanation of the importance of screening for TB in HIV-positive patients. However, in a cohort of HIV-positive men followed in Vancouver, less than 50% had received tuberculin skin testing during annual physician visits.[25] Patients with HIV infection may experience a lower rate of other opportunistic infections and survive longer if they do not have active TB.[26]
Because of the increased risk of active TB in patients with HIV and TB coinfection,[27,28] it is crucial that the immune status of contacts of active TB cases be assessed, especially if the patients are members of a group at a high risk of HIV infection. If contacts with HIV infection have a negative result of tuberculin skin testing, evaluation for anergy is indicated. In fact, some experts recommend that HIV-positive contacts of infectious pulmonary cases of TB should receive prophylaxis regardless of tuberculin skin-test results,[29] although this recommendation is not based on controlled studies.
There is growing concern about the ability of traditional contact-tracing methods to identify infected contacts. In studies in the United States and Switzerland,[30,31] analysis by restriction-fragment-length polymorphism uncovered many linked cases that were not previously linked by conventional contact-tracing techniques. People with HIV infection are a special concern, since they are at an increased risk of rapid progression to active disease.[27] The disproportionate number of cases of HIV-associated TB diagnosed in 1990 suggests that, in Vancouver, there was a cluster of TB cases among people with HIV infection; however, isolates of the TB strains involved are not available for DNA fingerprinting. One of us (J.M.F.) and a colleague recently demonstrated such clustering in among aboriginal Canadians who did not have HIV infection.[32]
[Table of Contents]
Conclusions
The demographic profile of patients with TB who also have HIV infection has changed significantly since our earlier report and now includes more women, aboriginal Canadians and IVDUs. The proportion of new TB cases associated with HIV infection continues to increase. We observed a rate of treatment failure and relapse of 9% and a disproportionately high rate among those who did not receive DOT. Resources must be targeted appropriately to screen for TB infection in groups at a high risk of coinfection, to offer chemoprophylaxis when indicated and to treat TB in people with HIV infection and in other high-risk groups. Evaluation for TB infection upon diagnosis of HIV infection or during follow-up is critical, since isoniazid prophylaxis can prevent active TB in people with HIV infection.[3,24] DOT is indicated for all cases because it decreases the likelihood of drug resistance and treatment failure. People who have multiple medical problems or who are taking other drugs daily may be at an increased risk of noncompliance with antituberculosis therapy and should be considered a high-priority group for DOT. Such measures are particularly important if we are to prevent multiple-drug-resistant TB in Canada. Contact tracing should include assessment for HIV risk factors, and HIV testing should be offered when indicated. Information about HIV seropositivity among patients with TB and their contacts will help ensure that appropriate treatment is given.
[Table of Contents]
References
The full text may also be ordered from the Canada Institute for Scientific and Technical Information (CISTI) or Institute for Scientific Information (ISI).