Flow diagram of the study population. LTBI indicates latent tuberculosis infection; 4-RIF, 4-month regimen of rifampin therapy; and 9-INH, 9-month regimen of isoniazid therapy.
Percentage of patients completing treatment by month for isoniazid and rifampin groups. Error bars represent 95% confidence intervals. 4-RIF indicates 4-month regimen of rifampin therapy; and 9-INH, 9-month regimen of isoniazid therapy.
Page KR, Sifakis F, Montes de Oca R, Cronin WA, Doherty MC, Federline L, Bur S, Walsh T, Karney W, Milman J, Baruch N, Adelakun A, Dorman SE. Improved Adherence and Less Toxicity With Rifampin vs Isoniazid for Treatment of Latent TuberculosisA Retrospective Study. Arch Intern Med. 2006;166(17):1863-1870. doi:10.1001/archinte.166.17.1863
Treatment of latent tuberculosis infection (LTBI) is an important aspect of tuberculosis control in the United States, but the effectiveness of this strategy is compromised by poor adherence to the recommended 9-month isoniazid regimen. In this study, we compared treatment completion and clinically recognized adverse drug reactions in patients prescribed 9 months of isoniazid therapy or 4 months of rifampin therapy for LTBI.
Retrospective chart review of patients who received LTBI treatment at a public health clinic.
A total of 770 patients were prescribed 9 months of isoniazid therapy, and 1379 patients were prescribed 4 months of rifampin therapy. The percentages of patients who completed 80% or more of their prescribed treatment were 52.6% and 71.6% in the isoniazid and rifampin groups, respectively (P<.001). In multivariate logistic regression analysis, treatment regimen was independently associated with treatment completion (adjusted odds ratio for treatment completion, 2.88 for rifampin group vs isoniazid group; 95% confidence interval, 2.27-3.66). Clinically recognized adverse reactions resulting in permanent treatment discontinuation occurred in 4.6% and 1.9% of patients in the isoniazid and rifampin groups, respectively (P<.001). Clinically recognized hepatotoxicity was more common in the isoniazid group (1.8%) than in the rifampin group (0.08%, P<.001).
Compared with a 9-month isoniazid regimen, a 4-month rifampin regimen was associated with a higher percentage of patients completing treatment and a lower percentage of patients with clinically recognized adverse reactions. Additional studies are warranted to determine efficacy and effectiveness of rifampin therapy for LTBI.
Treatment of latent tuberculosis infection (LTBI) is an essential component of the tuberculosis (TB) elimination strategy in the United States.1,2 Isoniazid is the preferred drug for LTBI treatment.3,4 In a setting of good adherence, the protective efficacy of isoniazid is approximately 90%.5 However, because of poor treatment completion rates, reported effectiveness is usually lower, ranging from 25% to 92%.5 Furthermore, high rates of isoniazid resistance in certain geographic areas can diminish the protective efficacy of isoniazid therapy for LTBI.4,6,7 The currently recommended isoniazid duration is 9 months; clinical studies indicate that the maximum protective benefit of isoniazid therapy for LTBI is achieved after approximately this duration.3,8
A 4-month regimen of rifampin monotherapy (4-RIF) has been recommended as an “acceptable alternative” to a 9-month regimen of isoniazid (9-INH), but 4-RIF has not been extensively studied or widely used.3 From a biologic perspective, there is rationale for the use of rifampin for LTBI treatment, since rifampin has potent activity against Mycobacterium tuberculosis, especially against mycobacterial subpopulations with short bursts of metabolic activity.9 Animal10 and clinical7,11- 13 studies involving relatively small numbers of patients indicate that rifampin therapy may be safe and effective for LTBI.
In the late 1990s, Prince George's County Health Department, Cheverly, Md, expanded its use of 4-RIF for LTBI treatment because of perceived low completion rates with isoniazid therapy for LTBI and concerns about hepatotoxicity (hepatotoxic effects) associated with a 2-month regimen of pyrazinamide plus rifampin. We performed a retrospective cohort study of 2255 consecutive patients who began either 9-INH or 4-RIF over a 5-year period. The primary objective was to determine and compare the percentage of patients who completed the prescribed treatment under routine TB program conditions. We also determined and compared the percentages of patients with adverse drug reactions.
We performed a retrospective study of all individuals who were prescribed either 9-INH or 4-RIF for LTBI from January 1, 1999, to January 31, 2004, at Prince George's County Health Department. A list of all patients was obtained from a computerized registration database maintained by the TB control program. Information was obtained by medical record review, abstracted onto a data form, and entered into a study database.
Patients were evaluated for LTBI treatment and monitored for adverse effects according to published guidelines.3 Monthly clinical monitoring, performed by a nurse, consisted of a symptom review using a list of standard questions. Only patients with risk factors for hepatotoxicity underwent routine measurement of transaminase levels.3 There were 4 main prescribing clinicians. Generally, clinicians presented patients with the option of 9-INH, described as the standard first-choice regimen, or 4-RIF, described as an alternative shorter-duration regimen that had not been as extensively evaluated for LTBI treatment. Exceptions included individuals who had (1) contraindications to one of the drugs, (2) family members already receiving one of the drugs, or (3) been exposed to a patient with drug-resistant TB; these individuals were prescribed the most appropriate regimen. The dosages of isoniazid and rifampin therapy were in accordance with guidelines.3 Rifampin was prescribed for daily use. Isoniazid was typically used daily, but a minority of 9-INH recipients received directly observed, twice-weekly therapy. Patients who relocated and therefore had TB care transferred to another TB program were not included in the data analysis. Institutional review boards of the Johns Hopkins University School of Medicine, Baltimore, Md, and the Maryland Department of Health and Mental Hygiene, Baltimore, approved this study.
To estimate treatment effectiveness, medical records of active TB cases at Prince George's County Health Department were reviewed. Also, the TB case registries for Maryland and Washington, DC, from 1999 through mid-2005 were cross-checked with study patient names and birth dates. To maximize sensitivity and accuracy, a computer program was used to adjust for typographical errors or minor variations in name spelling and date of birth and for reversals of first and last names.
Primary outcomes were the percentage of patients in each treatment group who completed treatment and the percentage in each group who had an adverse reaction that resulted in a clinician's recommendation to permanently discontinue the prescribed regimen. The LTBI treatment regimen was defined as the first prescribed regimen. For the primary analysis, patients were considered to have completed treatment if they took 80% or more of the prescribed doses within 43 weeks for 9-INH or within 20 weeks for 4-RIF.13 Hepatotoxicity was defined as alanine aminotransferase (ALT) and/or aspartate aminotransferase (AST) levels (1) greater than or equal to 3 times the upper limit of normal with hepatitis symptoms or (2) greater than or equal to 5 times the upper limit of normal without symptoms. Asymptomatic individuals with ALT and/or AST levels that were 3 to fewer than 5 times the upper limit of normal were classified as having elevated transaminase levels. The term clinically recognizedtoxicity reflects the fact that our study captured data only for patients who returned to the clinic; some patients who did not return to the clinic may have had medication-related toxicity that was not captured in the study. Patients diagnosed as having active TB after starting LTBI treatment were classified as treatment failures. Data obtained from medical records included sex, age, race/ethnicity, region of origin, prescribed LTBI treatment regimen, reason for LTBI therapy, tuberculin skin test size, prescribing clinician, adverse drug reactions, completion of treatment, reason for not completing treatment, measurement of baseline and follow-up ALT and/or AST levels (if performed); and human immunodeficiency virus (HIV) serostatus.
Given the lack of randomization, factors that could be associated with treatment completion were included in the analysis. The distribution of these factors among the 2 treatment groups was evaluated to adjust for potential confounders in the multivariate analysis of completion rates. The t test was used to compare means, and Pearson χ2 to test differences in treatment assignment in categorical variables. The Fisher exact test was used to compare variables with small cell sizes (n<5). Association between treatment completion and demographic and clinical factors was assessed using the χ2 test. Association between treatment completion and treatment assignment was assessed via logistic regression analysis, thus accounting for potential confounders resulting from lack of randomization. All factors that were significant (P≤.05) in univariate analysis were entered in the multivariate model. Variables with more than 2 categories were coded as a group of dichotomous indicator variables. The final model was derived by removing insignificant (P>.05) variables as determined by the likelihood ratio test. All indicator variables in a group were removed together if they were insignificant. Adjusted odds ratios (AORs) and corresponding 95% confidence intervals (95% CIs) were calculated for variables that remained in the final model. Association between adverse drug reactions and treatment regimen was assessed using the χ2 test. Analyses were performed using SAS version 9.0 (SAS Institute, Cary, NC).
A total of 2255 patients were included in this study; 843 patients were prescribed 9-INH, and 1412 patients were prescribed 4-RIF. The flow diagram of the study population is shown in Figure 1. Seventy-three patients (8.6%) in the 9-INH group and 33 patients (2.3%) in the 4-RIF group had care transferred to another TB control program owing to change of residence and were excluded from the analysis. Data analysis was performed on all remaining patients (770 treated with 9-INH and 1379 treated with 4-RIF). Baseline characteristics of the study population are shown in Table 1. Baseline characteristics were similar between the individuals who were transferred out and the individuals who were included in the analysis. Compared with the patients prescribed 9-INH, those prescribed 4-RIF were older and more likely to be male, to be born outside North America or Europe, to be a recent contact or tuberculin skin test converter, and to have self-described race/ethnicity as black. The percentage of individuals with pretreatment AST and/or ALT levels greater than the upper limit of normal was slightly higher in the 9-INH group than in the 4-RIF group (5.7% vs 4.3%, P = .02). Individuals with documented HIV infection or unknown HIV status were more likely to be prescribed 9-INH. Finally, a higher percentage of patients in the 9-INH group received directly observed treatment (5.1% vs 1.1%, P = .001).
As shown in Table 2, the percentages of patients completing treatment according to the study definition were 52.6% (405/770) in the isoniazid group and 71.6% (987/1379) in the rifampin group (95% CI for difference, 14.7%-23.2%; P<.001). Factors significantly associated with treatment completion by univariate analysis were included in a multivariate regression logistic model. Treatment regimen was the factor most strongly associated with treatment completion (AOR, 2.88 for rifampin vs isoniazidgroups; 95% CI, 2.27-3.66) in multivariate analysis after adjustment for age, region of origin, race/ethnicity, treatment regimen, pretreatment liver function test results, and adverse reactions (Table 2). Region of origin was also associated with treatment completion, with higher odds of completion for certain groups born outside of North America or Europe. Not surprisingly, hepatotoxicity and other adverse reactions were associated with treatment discontinuation (AOR for completion, 0.03; 95% CI, 0.00-0.27 for hepatotoxicity; and AOR, 0.21; 95% CI, 0.15-0.30 for other adverse drug reactions).
The percentages of patients who were prescribed treatment but failed to complete 1 month of treatment were similar in both groups (12.7% [98/770] 9-INH vs 13.6% [188/1379] 4-RIF; 95% CI for difference, 3.8%-2.1%; P = .56). Moreover, the percentages of patients who dropped out during months 3 or 4 were similar between the treatment groups.
In the 9-INH group, 56 patients (7.3%) took 80% or more of prescribed medication over a period of more than 43 weeks (“clinical completers”). In the 4-RIF group, 51 patients (3.7%) took 80% or more of prescribed medication over a period of more than 20 weeks. When these clinical completers were combined with treatment completers who met the study definition, the overall percentages of patients who completed treatment were 59.9% (461/770) and 75.3% (1038/1379) in the 9-INH and 4-RIF groups, respectively (95% CI for difference in percentagess, 11.3%-19.5%; P<.001). Of the patients prescribed 9-INH, 64% (493/770) completed at least 6 months of treatment (P<.001 for comparison with 4-RIF completers). Figure 2 shows the percentages of patients who completed each treatment month. Interestingly, between the isoniazid and rifampin groups, there was no difference in the percentages of patients who completed 4 months of treatment (73.9% [569/770] 9-INH vs 75.3% [1038/1379] 4-RIF; 95% CI for difference in percentages, 5.3%-2.4%; P = .50).
As shown in Table 3, patients prescribed 9-INH were more likely to have a recognized adverse reaction than patients prescribed 4-RIF (11.3% vs 8.3%; P = .02). Furthermore, the percentage of patients who were advised by a clinician to permanently discontinue their prescribed regimen because of an adverse reaction was higher in the isoniazid group than in the rifampin group (4.6% vs 1.9%; 95% CI for difference, 1.1%-4.3%; P<.001); this difference remained significant if only the first 4 months of isoniazid therapy for LTBI were included in the analysis (4.0% [27/670] for 9-INH vs 1.9% [23/1229] for 4-RIF; 95% CI for the difference, 0.6%-3.6%; P = .008). Analysis of adverse reactions by category showed no significant difference between the 2 treatment regimens in the development of rash, gastrointestinal upset, thrombocytopenia, or fatigue.
However, as shown in Table 3, the percentage of patients with clinically recognized hepatotoxicity was significantly higher in the 9-INH group than in the 4-RIF group (1.8% vs 0.08%; 95% CI for difference, 0.9%-1.9%; P<.001). The difference remained significant if only the first 4 months of isoniazid therapy for LTBI were included in the analysis (1.0% [7/670] for 9-INH vs 0.08% [1/1229] for 4-RIF; 95% CI for difference, 0.3%-1.2%; P = .003). The mean age of the patients who developed hepatotoxicity was 39.7 years (median age, 44 years; age range, 16-66 years). Nine (69%) of 13 patients were female; 7 (54%) of 13 patients were Hispanic; and no patients had documented HIV infection. Pretreatment measurement of ALT and/or AST levels had been performed in 11 of the 12 isoniazid recipients who developed hepatotoxicity, and in 3 of these 11, the ALT and/or AST level was higher than the upper limit of normal. The patient who developed hepatotoxicity while taking rifampin did not have documented pretreatment ALT and/or AST levels. Even if all patients with measured pretreatment ALT and/or AST levels higher than the upper limit of normal were excluded from the data analysis, the frequency of recognized hepatotoxicity remained significantly higher in the 9-INH group than in the 4-RIF group (1.3% vs 0.08%, respectively; P = .001). The prescribed regimen was permanently discontinued in all of the patients with hepatotoxicity. One patient who received isoniazid was hospitalized, but there were no fatalities in either group. There was no difference between the treatment groups in the percentage of patients with elevated transaminase levels during treatment. Four (31%) of the 13 patients with elevated transaminase levels during treatment were advised to permanently discontinue the prescribed regimen. Among 103 patients with pretreatment ALT and/or AST levels higher than the upper limit of normal, 3 (7%) of 44 in the 9-INH group and 1 (2%) of 59 in the 4-RIF group developed elevated transaminase levels during treatment (P = .31); 3 (7%) of 44 in the 9-INH group and none of 59 in the 4-RIF group developed hepatotoxicity (P = .07); and 4 (9%) of 44 patients in the 9-INH group and none of 59 in the 4-RIF group were advised to permanently discontinue therapy owing to hepatotoxicity or elevated transaminase levels (P = .03).
Of 2149 study participants included in the data analysis, only 1 with subsequent active TB was identified. This patient had been prescribed 4-RIF and 1 year later developed cervical lymphadenopathy. Cultures from her lymph node aspirate yielded M tuberculosis, susceptible to rifampin. She had reported taking rifampin as prescribed for 4 months, was healthy without known immunocompromise, and had no known risks for reinfection with M tuberculosis. There was no statistically significant difference in treatment failure frequencies between 4-RIF and 9-INH groups (1/1379 vs 0/770, respectively; P = .46).
Our study expands the reported experience using 4-RIF for LTBI treatment. To our knowledge, this is the largest study comparing 9-INH with 4-RIF. After adjusting for potential confounders, we found that the treatment regimen was the factor most strongly associated with treatment completion. Specifically, the use of 4-RIF increased the odds of LTBI treatment completion by almost 3-fold. Also, compared with the 9-INH group, smaller percentages of patients in the 4-RIF group had permanent treatment discontinuation at the recommendation of a health care provider, clinically recognized hepatotoxicity, or any clinically recognized adverse drug reaction. Differences between groups in the percentages of patients with hepatotoxicity or treatment discontinuation owing to an adverse reaction were not solely attributable to a longer duration of isoniazid exposure, since the differences remained significant if only the first 4 months of isoniazid therapy for LTBI was considered.
These results have important clinical implications. Medication adherence has a significant impact on the outcome of LTBI treatment.1,3,5 The wide range in reported isoniazid protective efficacy is generally attributed to differences in adherence.3,5 To facilitate treatment completion, emphasis has been placed on the development of shorter treatment regimens. During clinical studies, a 2-month course of rifampin plus pyrazinamide was as efficacious as 9-INH in the treatment of LTBI among HIV-infected individuals.14 However, subsequent use of the rifampin-pyrazinamide regimen under routine conditions was associated with unacceptably high rates of hepatotoxicity, prompting the Centers for Disease Control and Prevention to issue a warning against the use of this combination for LTBI treatment.15,16 A clinical trial comparing 9-INH with a 12-week regimen of once-weekly isoniazid plus rifapentine is currently being conducted by the Centers for Disease Control and Prevention, but the results will not be available for several years. Cost is a major barrier to the initiation of additional prospective studies comparing the efficacy of LTBI treatments. Furthermore, efficacy in the setting of a clinical trial does not necessarily translate into effectiveness under routine TB program conditions. We found an inverse relationship between LTBI treatment duration and treatment completion. This finding provides a strong rationale for identification of short LTBI regimens and emphasizes the importance of ascertaining effectiveness and efficacy of 4-RIF in preventing TB.
To our knowledge, rifampin LTBI treatment has been evaluated in 2 randomized clinical trials, 1 retrospective cohort study, and 1 prospective cohort study; a combined total of approximately 500 individuals received rifampin monotherapy in these published studies.7,11- 13 A double-blind, randomized, placebo-controlled study compared a 3-month course of rifampin, a 3-month course of rifampin plus isoniazid, a 6-month course of isoniazid, and no treatment in 679 men with silicosis in Hong Kong.11 In this study, 165 evaluable patients received rifampin monotherapy, which was as effective in preventing progression to active disease as isoniazid-based regimens and was associated with fewer hepatotoxic effects. Over a 5-year follow-up period, high percentages of patients in each arm progressed to active TB in the setting of silicosis but, importantly, there was no acquired rifampin resistance among the 15 rifampin monotherapy recipients who subsequently developed TB and had rifampin susceptibility determined. An open-label study of 116 patients randomized to either 4-RIF or 9-INH showed significantly higher completion rates in the 4-RIF group (91% vs 76%) and a trend toward fewer adverse drug reactions.13 Polesky et al7 reviewed cases in which rifampin was used for LTBI treatment during an outbreak of isoniazid-resistant TB among the homeless in Boston, Mass, in 1984. Rifampin therapy, prescribed for 6 to 12 months, was effective in preventing active TB among skin test converters, and the incidence of adverse effects was similar to that among patients receiving either isoniazid or isoniazid plus rifampin. Villarino et al12 prospectively followed up 157 high-school students who were prescribed daily rifampin for approximately 6 months in the setting of exposure to isoniazid-resistant M tuberculosis. Two individuals (1.3%) required permanent discontinuation of rifampin therapy, in one instance because of elevated transaminase levels, and no cases of active TB were identified during the 2-year study period.7,12
Like the studies described above, we found a low incidence of clinically recognized serious adverse effects among 4-RIF recipients, even though the rifampin recipients in our study were generally older than those described by Villarino et al12 and Menzies et al.13 In our study, the percentage of 9-INH recipients with clinically recognized hepatotoxicity was generally similar to that reported by others, indicating that the low incidence of recognized rifampin-associated hepatotoxicity may be generalizable to other populations. The percentage of patients who completed 4-RIF treatment (71.6%) in our study was lower than that reported by Menzies and colleagues (91%) in the context of a clinical study that was designed to assess adherence. However, 71.6% compares favorably with reported completion rates for isoniazid therapy under routine TB control program conditions: typically approximately 60% for 6 months of isoniazid therapy.17- 19
The strengths of our study are the relatively large sample size and the diverse study population, which may be similar to the population served by many urban and semiurban North American TB programs. Patient diversity may increase the chances of detecting adverse drug reactions that may not occur in carefully selected populations enrolled in clinical trials.20,21 Furthermore, treatment adherence in the setting of a clinical trial may not reflect adherence in the general community.22,23 Along these lines, the percentage of patients who completed at least 6 months of isoniazid therapy in our study (63%) is remarkably like those reported by LoBue and Moser18 (64%) and Nolan et al20 (64%) in their studies of routine LTBI treatment in public health clinic settings, making it plausible that the 4-RIF completion rates in our study are generalizable to other settings.
There were several important limitations of our retrospective study. First, patients were not randomized to the treatment groups. Our analysis adjusted for differences in baseline characteristics and possible confounders between the 2 treatment groups, but it remains possible that unaccounted factors contributed to the results. Specifically, we cannot fully exclude the possibility that unknown variables, especially those related to treatment regimen selection, influenced the results of our study. However, similar monthly dropout rates during the first 4 months for both groups argue against a strong effect from an unknown variable. Second, because LTBI treatment was carried out under routine TB control program conditions, screening for baseline liver disease and monitoring for hepatotoxicity during treatment, while performed in accordance with published guidelines, were not uniform across all patients. Third, we cannot exclude the possibility that the patients who were unavailable for follow-up defaulted because of adverse drug effects that were not reported to the clinic and therefore not captured in our study; therefore, we were unable to definitively determine the incidence of adverse reactions. Fourth, as there were few patients with documented HIV infection in our cohort, our results may not be applicable to settings with high HIV prevalence. This issue is relevant, since rifampin has drug-drug interactions with many HIV nonnucleoside reverse transcriptase inhibitors and protease inhibitors. Rifampin's numerous drug-drug interactions complicate its use in some individuals. Finally, our effectiveness data were limited by the follow-up period (range, 1-6 years) and the method of case ascertainment. Given the population mobility and potential for name changes, we cannot ensure that all treatment failures were captured in our query.
In conclusion, in this retrospective cohort study of LTBI treatment under routine TB program conditions, compared with 9-INH, 4-RIF was associated with a higher percentage of patients completing treatment and a lower percentage of patients with clinically recognized treatment-related toxicity. Our findings support the use of 4-RIF for treatment of LTBI. Prospective studies or registries are warranted to further clarify 4-RIF efficacy, effectiveness, and cost-effectiveness as well as the risk for development of acquired rifampin resistance in the setting of treatment failure.
Correspondence: Susan E. Dorman, MD, Department of Medicine, Johns Hopkins University School of Medicine, 1550 Orleans St, CRB II, Room 1M-06, Baltimore, MD 21231 (firstname.lastname@example.org).
Accepted for Publication: May 23, 2006.
Author Contributions: Dr Dorman had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None reported.
Funding/Support: Dr Dorman is supported by National Institutes of Health grant K23AI51528. Dr Page is supported by National Institutes of Health grant T32AI07608.
Role of the Sponsor: The National Institutes of Health had no role in study design or conduct; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript.
Acknowledgment: We thank Gillian VanBlerk, MD, for her thoughtful patient care, Frederick Corder, MD, for facilitating the study, Mark Sherwood and Sidney Clemons for expert assistance with computer database management, Patricia Bey, MD, and Rene Washington, MD, for assistance with chart abstraction, Ernestine Rice and Vida Kwanza for assistance with medical records, and Darryl Hardge for assistance with queries of TB active case registries in Washington, DC.