Published ahead of print on June 1, 2004, doi:10.1164/rccm.200404-478OC
© 2004 American Thoracic Society
Treatment Completion and Costs of a Randomized Trial of Rifampin for 4 Months versus Isoniazid for 9 MonthsRespiratory Epidemiology and Clinical Research Unit, Montreal Chest Institute, and Division of Clinical Epidemiology, Royal Victoria Hospital, McGill University, Montreal, Quebec, Canada Correspondence and requests for reprints should be addressed to Dick Menzies, M.D., M.Sc., Montreal Chest Institute, 3650 St-Urbain, Room K1.24, Montreal, PQ, H2X 2P4 Canada. E-mail: dick.menzies{at}mcgill.ca
There is little published information regarding treatment completion, safety, and efficacy of rifampin administered daily for 4 months—a recommended alternative to 9 months of isoniazid for therapy of latent tuberculosis infection. In an open-label randomized trial at a university-affiliated respiratory hospital, consenting patients whose treating physician had recommended therapy for latent tuberculosis infection were randomized to daily self-administered rifampin for 4 months or daily self-administered isoniazid for 9 months. Of 58 patients randomized to rifampin, 53 (91%) took 80% of doses, and 50 (86%) took more than 90% of doses within 20 weeks compared with 44 (76%) and 36 (62%) who took 80 and 90%, respectively, of doses of isoniazid within 43 weeks (relative risks: 80% of doses, 1.2 [95% confidence interval: 1.02, 1.4]; 90% of doses, 1.4 [1.1, 1.7]). Adverse events resulted in permanent discontinuation of therapy for two (3%) patients taking rifampin, and for eight (14%) patients taking isoniazid. Three patients developed drug-induced hepatitis—all were taking isoniazid. Total costs of therapy were significantly higher for isoniazid. In conclusion, completion of therapy was significantly better with 4 months of rifampin and major side effects were somewhat lower. Further studies are needed to assess the safety and efficacy of the 4-month rifampin regimen.
Key Words: latent tuberculosis infection • treatment of latent tuberculosis infection • tuberculosis prevention A major activity of many tuberculosis (TB) control programs in industrialized countries is the identification of persons with latent TB infection (LTBI), who have increased risk of development of active TB. Treatment of such individuals can provide individual (1–6) and public health (4–8) benefits. The current recommended standard therapy is 9 months of isoniazid (9INH). This has an efficacy of more than 90% (9) if taken properly. However, because of the long duration, only 64–67% of patients (10, 11) or fewer (12) complete therapy under routine program conditions. Another limitation of INH is the occurrence of drug-induced hepatitis. Although rare, this can be fatal (13–17). The incidence of this complication appears to have diminished over the past three decades (10, 18, 19), but nevertheless this remains an important disadvantage of INH therapy. As well, the long duration, and need for close monitoring because of the possibility of serious adverse events, make this regimen relatively costly. Because of these problems, there has been considerable interest in finding shorter and better tolerated regimens for the treatment of LTBI (20). In 2000, two alternative regimens were recommended—2 months of daily rifampin and pyrazinamide (2RIF–PZA) and 4 months of daily rifampin (4RIF) (8, 21). After more widespread use, the 2RIF–PZA regimen was associated with serious hepatotoxicity (22–24) and deaths (22, 25). As a result this is now recommended only for carefully selected high-risk individuals such as human immunodeficiency virus (HIV)-infected persons (26), in whom its efficacy and safety have been demonstrated (27). This leaves 4RIF as the only currently recommended alternative to 9INH for treatment of LTBI in the great majority of HIV-negative persons. However, there is little published information regarding treatment completion, safety, or efficacy. In one randomized trial, the efficacy and tolerability of 3 months of rifampin alone were somewhat better than those of 6INH among older Chinese males with pulmonary silicosis (28). In two case series, 6 months of rifampin was well tolerated in homeless persons in Boston (29), and in high school students in California (30). We have undertaken a single-center randomized trial to test the hypothesis that completion of therapy would be significantly better with 4RIF than with 9INH. We also compared the safety and tolerability of the two regimens. Some results from this study have been presented in abstract form (31).
Setting, Study Population, and Randomization An open-label randomized controlled trial was conducted at a university-affiliated respiratory hospital. Patients were considered eligible if they were 18 years of age or older and had a documented tuberculin skin test that met the criteria for a positive test by Canadian standards (21), and their treating physician initially recommended 9INH for LTBI. Contacts of INH-resistant cases (32) and patients with hypersensitivity to rifamycins or who were taking therapy with potential interactions with rifampin, without any available acceptable alternative (e.g., antiretroviral therapy), or who refused alternatives (e.g., oral contraceptives), were ineligible. Eligible patients who signed informed consent forms were randomized to 4 months of daily RIF (10 mg/kg, up to 600 mg/day) or 9 months of daily INH (5 mg/kg, up to 300 mg/day), using an Internet-accessible computerized program that also verified eligibility. Randomization was stratified by risk of TB (high if patient was HIV infected, had close contacts with active TB, or had fibronodular changes on chest X-ray; and low to moderate for all others), because compliance may be different in these risk groups (33–35).
Randomization of 58 patients per arm provided 80% power to detect significantly better treatment completion with 4RIF (
Data Gathering The primary outcome was the percentage of prescribed doses taken, measured with an electronic device in the pill container cap, which recorded the date and time of bottle opening (medication event-monitoring system [MEMS] device; Aprex [a division of Aardex], Fremont, CA). Secondary outcomes included adverse events resulting in permanent discontinuation of therapy. Drug-induced hepatitis was defined as liver transaminase (alanine transaminase) levels more than three times the upper limits of normal with symptoms, or transaminase levels more than five times the upper limits of normal without symptoms (8). Health care system costs were based on health care use after randomization (37). Institutional costs were based on actual costs at the Montreal Chest Institute in 2003, and professional fees on the 2003 reimbursement schedule of the Régie de l'Assurance Maladie du Québec. Medication costs were based on pills dispensed, pharmacist fees, and global TB drug facility prices (38). Costs for the electronic monitors were not included.
Data Analysis
Characteristics of nonparticipants and participants, and of participants randomized to the two arms, were compared on the basis of
Between January 21 and October 1, 2002, 227 patients were recommended to take 9INH for LTBI by their treating physicians and referred to the study for screening. As shown in Figure 1 , 18 of those screened were not eligible, and 93 refused. Most of those who refused did so because they were not interested in participating in research, or preferred to take the standard therapy as initially recommended by their treating physician. As shown in Table 1 , the characteristics of those who refused and of the patients randomized to the two treatment arms were not significantly different. Information about nonparticipants was limited to that shown in Table 1. However, among participants other characteristics, not shown in Table 1, that were similar in the 9INH and the 4RIF groups included the following: cigarette smoking, alcohol use, time since arrival in Canada (if foreign born), travel to TB-endemic countries, past TB exposure, presence of symptoms, and other medical illnesses including liver and kidney diseases, and hepatitis B and hepatitis C infection.
As shown in Table 2 , of the 58 randomized to 4RIF, 53 (91%) completed therapy, and 50 (86%) took more than 90% of doses within the allowed time, compared with 44 (76%), and 36 (62%), respectively, of patients allocated to 9INH (both differences significant, p < 0.05). Major adverse events, resulting in permanent discontinuation of therapy by their treating physician, were somewhat more frequent among patients allocated to 9INH. These events were significantly associated with age over 35 years and a history of liver disease, but were not associated with sex, country of origin, or years since arrival in Canada (data not shown in tabular form). Drug-induced hepatitis developed in three patients, after 74, 105, and 137 doses of isoniazid. None of these three patients had abnormal liver tests before treatment. Conversely, of the eight with baseline abnormal transaminases, none developed hepatitis. Minor symptoms—which did not result in any change of therapy—were similar for the two regimens except for a significantly greater occurrence of fatigue with 4RIF (15 versus 4%, p < 0.001).
As seen in Table 3 , the proportion of doses taken each month was generally higher with rifampin, although significantly higher only in Month 2. Most dropouts occurred early. Therefore, among the patients remaining on therapy after Month 2, the proportion of doses taken was high.
Overall costs of follow-up were significantly greater for patients allocated to 9INH (Table 4) , although the majority of costs were related to routine follow-up visits. No patients were hospitalized during follow-up. Nonroutine costs—for unscheduled clinic visits, consultations, and emergency room visits, and related laboratory investigations—were approximately twice as high per patient randomized to 9INH than to 4RIF.
In this study, significantly more patients randomized to 4RIF completed an adequate course of therapy, with significantly lower follow-up costs, and somewhat less frequent major adverse events. This small trial did not have adequate power to assess safety or efficacy. These results should not be interpreted to mean that 4RIF can replace 9INH for routine treatment of latent TB infection. It may seem obvious that a much shorter regimen should result in better treatment completion. However, in randomized trials completion of 2RIF–PZA was no better than 6INH (23, 39), although it was better than 12INH (27). And in two case series, only 57% (40), or 68% (41), of patients completed the 2RIF–PZA regimen, which cannot be considered optimal. Evaluation of the efficacy of a shorter LTBI regimen is a major undertaking. To initiate this would be justified only if completion of that regimen was significantly better than with the current standard of 9INH. This is why the primary objective of the present study was to assess treatment completion. Having now demonstrated superior completion rates, the next step should be a study with adequate power to evaluate safety. Once this is completed, then a decision can be made to launch a large-scale efficacy study. An important potential limitation of the study design was the absence of blinding. Awareness of the regimen may have influenced reporting by patients, and recording by providers, of symptoms. The resultant bias may have accounted for the differences seen. Bias also may have influenced decisions by physicians to discontinue therapy permanently. Although drug-induced hepatitis was diagnosed on the basis of objective laboratory measurements, judgment regarding the severity of the other adverse events such as nausea and vomiting, or fatigue, is more subjective, and therefore potentially subject to bias by the treating physician. However, an unblinded study was justified because the primary study outcome—treatment completion—was likely strongly influenced by the duration of therapy. In addition, the results of an unblinded study may be more applicable to routine practice. This also permitted assessment of the impact on patient compliance of the greater number of pills, and discoloration of body fluids of 4RIF. Patient compliance may have been overestimated, if patient behavior was influenced by the knowledge that every opening of the pill container was recorded electronically. However, this was considered essential to achieve accurate measurement of the primary outcome. It may also be argued that actual ingestion of the medication was not measured, because the electronic device recorded only when the bottle was opened. However, regularity of bottle opening, measured with the same device, has been correlated with efficacy of treatment of hyperlipidemia (42) and diabetes (43). The study had a number of strengths. The process of routine care was disrupted as little as possible, to minimize the effect of participation in a randomized trial, which may itself result in greater patient compliance. As a result the estimates of treatment completion may be more realistic. In support of this, the completion rate of 9INH in this trial fell within the range of previous measures of treatment completion under routine conditions at this hospital, that is, 50–60% with 12INH (33, 34) and 80% with 6INH (35). In initial randomized trials with INH and 2RIF–PZA, incidence of major adverse events was substantially lower (27, 44) than documented subsequently when prescribed by many providers in routine practice (13, 17, 22, 25). Why incidence has been lower in randomized trials is unknown, but we speculate this may have reflected selection into these trials of persons at lower risk for adverse events. In the present study only patients in whom rifamycins were clearly contraindicated were excluded. By including, as much as possible, all other patients, the estimates of adverse events and compliance should be more realistic and relevant to routine practice. For example, this was why patients with abnormal baseline lung function tests were considered eligible for this trial. In these patients the decision to prescribe 9INH was made without consideration of this trial, by the treating physician who had judged that the benefits of treatment outweighed the increased risks. Only after this clinical decision had been made, and discussed with the patients, were they considered eligible for this trial. This method of recruitment may explain why a substantial number of eligible subjects refused. However, their demographic and clinical characteristics were similar to those of the participants, reducing the likelihood that nonparticipation caused bias in the estimates of compliance and adverse events. The results of this study confirm that patient acceptance and compliance with 4RIF are excellent. The next step is a trial with adequate power to assess safety and tolerability in a group of patients representative of those likely to receive this therapy in future. Some may argue that the safety and tolerability of rifampin is well known from treatment of millions of persons with active TB. However, a similar assumption proved incorrect for the 2RIF–PZA regimen—once introduced widely into clinical practice for treatment of LTBI, it was associated with unacceptable rates of hepatotoxicity and death. Therefore a careful assessment of the safety of the 4RIF regimen is needed before undertaking an efficacy study. If the safety and tolerability of 4RIF are equivalent to or better than those of 9INH, then a large-scale efficacy study would be justified.
The authors thank the staff of the tuberculosis clinic, particularly Jen Lee, Isabelle Parisien, Isabelle Rocher, and Denis Francis.
Supported by a grant from the Medical Research Council of Canada (FRN#44154). Dr. Menzies received a scientist award from the Medical Research Council of Canada and currently holds an FRSQ Chercheur Boursier Senior Award. Dr. Schwartzman holds a Chercheur Boursier Clinicien Award from the FRSQ, and Dr. Brassard holds a New Investigator award from the CIHR. Conflict of Interest Statement: D.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; M.-J.D. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; B.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; P.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; K.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form April 8, 2004; accepted in final form May 30, 2004
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
= 0.05 and one-sided test), expecting 70% completion of 9INH, and 90% with 4RIF (
2 tests for categorical variables, and Student t tests for continuous variables (
