INT J TUBERC LUNG DIS 11(8):868–875 © 2007 The Union
Adverse drug reactions associated with first-line anti-tuberculosis drug regimens F. Marra,*† C. A. Marra,*‡ N. Bruchet,* K. Richardson,*‡ S. Moadebi,† R. K. Elwood,*† J. M. FitzGerald*§ * University of British Columbia, Vancouver, † BC Center for Disease Control, Vancouver, ‡ Center for Health Evaluation and Outcome Sciences, Vancouver, § Center for Clinical Epidemiology and Evaluation, Vancouver, British Columbia, Canada SUMMARY BACKGROUND:
Standard treatment of active tuberculosis (TB) consists of isoniazid (INH), rifampin (RMP), pyrazinamide (PZA) and ethambutol (EMB). Although this regimen is effective in treating active TB, it is associated with many adverse drug reactions (ADRs) and poses a significant challenge to completion of treatment. O B J E C T I V E S : To examine the incidence of major ADRs and risk factors associated with first-line anti-tuberculosis medications. M E T H O D S : This study evaluated patients receiving treatment for active TB from a population-based database (2000–2005). The nature of the ADRs, likelihood of association with the study medications and severity were evaluated. R E S U L T S : A total of 1061 patients received treatment, of whom 318 (30%) had at least one major ADR. The
overall incidence of all major ADRs was 7.3 events per 100 person-months (95%CI 7.2–7.5): 23.3 (95%CI 23.0– 23.7) when on all four first-line drugs, 13.6 (95%CI 13.3– 14.0) when on RMP, INH and PZA, and 2.4 (95%CI 2.3–2.6) when on INH and RMP. Adjusted hazard ratio (HR) revealed that combination regimens containing PZA, females, subjects aged 35–59 and 60 years, baseline aspartate aminotransferase 80 U/l and drug resistance were associated with any major event. C O N C L U S I O N S : First-line anti-tuberculosis drugs are associated with significant ADRs. There are several risk factors associated with the development of ADRs, including exposure to regimens containing PZA. K E Y W O R D S : first-line medications; tuberculosis; adverse events
THE RECOMMENDED first-line regimen for the treatment of active tuberculosis (TB) consists of isoniazid (INH, H), rifampin (RMP, R), ethambutol (EMB, E) and pyrazinamide (PZA, Z).1,2 These medications are associated with significant adverse drug reactions (ADRs), which can make the successful treatment of active TB in patients with drug-sensitive disease a challenge.1,3,4 Common ADRs reported from field trials of first-line anti-tuberculosis drugs include skin rash and pruritus, hepatitis, nausea/vomiting, thrombocytopenia, influenza-like illness, arthralgia and neuropsychiatric symptoms.5 An ADR can result in significant morbidity, leading to the withdrawal of first-line medication and its substitution with a less effective and often more poorly tolerated second-line medication. This substitution will prolong the duration of treatment and possibly affect adherence to the regimen. Severe ADRs increase hospitalizations among out-patients and lead to an increase in the number of clinic visits. Predicting who will be at an increased risk for ADRs
to first-line TB therapy can assist in identifying patients who require closer monitoring to prevent potential morbidity, hospitalization and mortality. Female sex, age 60 years, Asian birth and human immunodeficiency virus (HIV) infection have been suggested to be associated with an increased incidence of ADRs to first-line TB medications.6 A history of either hepatitis B or C has been linked to the development of severe ADRs to first-line drugs,7,8 although other studies have not confirmed this finding.9 Although ADRs to anti-tuberculosis medications have been documented from clinical trials and other observational studies, we were unable to locate any population-based studies. A population-based approach is preferable as it eliminates selection bias. Another potential weakness of current studies is that they fail to account for the entire mix of drugs that TB patients are exposed to, and instead determine risk factors for ADRs on a drug-by-drug basis. Clearly, this approach does not reflect reality, and methods should account for the combination of medications
Correspondence to: Carlo Marra, Assistant Professor, Faculty of Pharmaceutical Sciences, University of British Columbia, Center for Health Evaluation and Outcomes Sciences, St Paul’s Hospital, 620-1081 Burrard Street, Vancouver, BC V6Z 1Y6, Canada. Tel: (1) 604 806 3215. Fax: (1) 604 806 8778. e-mail:
[email protected] Article submitted 10 January 2007. Final version accepted 6 May 2007.
Adverse drug reactions to TB medications
that patients ingest and the timing of their exposure, as anti-tuberculosis medications are often stopped and re-started.10 The present study examines the incidence of major ADRs associated with combinations of first-line antituberculosis medications and determines the patientrelated factors associated with an increased risk of developing an ADR in a population-based cohort of patients with active TB.
METHODS Patient population This study evaluated the medical records of all patients who received treatment for active TB in British Columbia between 2000 and 2005. All individuals who are identified to have active TB are eligible to receive treatment through a publicly-funded program. All records are kept within a centralized database, the Public Health Information System (iPHIS). All mycobacteriology is completed at a provincial reference laboratory. This ensures that all cases of active TB in the province are identified and their treatment followed to conclusion. Viral hepatitis serology was determined by linking the TB dataset with the hepatitis registry which includes all hepatitis laboratory tests done in the province. This study was approved by the University of British Columbia Behavioral Ethics Committee. Active TB was confirmed by the isolation of Mycobacterium tuberculosis from culture or by the presence of a clinical presentation compatible with active TB. Information regarding patient age, sex, ethnicity, type and location of disease, sensitivity patterns, intolerance to previous and current TB medication, HIV infection and baseline aspartate aminotransferase (AST) values were extracted. Baseline AST was measured prior to the initiation of treatment or as the first value entered into the patient’s medical record (within the first 2 weeks of treatment initiation). In addition, the presence of hepatitis A, B or C co-infection, the duration of treatment and dosing regimen of the first-line TB medications and the incidence and nature of ADRs were also recorded for each patient. Drug-resistant TB was defined as resistance of M. tuberculosis to one or more first-line anti-tuberculosis medications. Outcome measures and definitions Patients were evaluated at TB control clinics or public health offices throughout the province at least once a month by a physician and/or a public health nurse. During these visits, information was gathered in a standardized fashion regarding the patient’s tolerance to the TB medications and this information was entered into the iPHIS database. In the assessment of the study outcomes, an ADR to the TB medications was recorded as such when both the attending physician and
869
the nurse practitioner concurred that, at the time of patient assessment, the ADR occurred as a result of the TB medications. These ADRs were categorized and further analyzed for severity and the likelihood of association between ADR and individual study medications according to the criteria described below. These are standard criteria for assessing adverse reactions, and we have used these criteria in our pharmacoepidemiological evaluation of levofloxacin.11 We used the following definitions for adverse events: 1) central nervous system events included vertigo, tinnitus, fever, seizures, visual disturbances, paresthesias, headache and confusion; 2) cardiovascular events were classified as palpitations or shortness of breath; 3) gastrointestinal (GI) events were characterized as those resulting in nausea/vomiting/GI upset, anorexia/ weight loss, dyspepsia or abdominal pain; 4) hepatitis was defined as an AST more than three times the upper limit of normal (ULN) in the presence of gastrointestinal symptoms or an AST 5 ULN in the absence of symptoms; 5) musculoskeletal events were characterized as weakness, fatigue and joint pain; 6) dermatologic events included rash, pruritus and swelling; and 7) hematological events were those that affected platelet or neutrophil counts. The severity of the ADR was classified as mild, moderate or severe. Only moderate and severe ADRs were included in the analysis. ADRs were described as moderate when the signs and symptoms returned to normal upon the discontinuation of TB medications and minor treatment was initiated. Severe ADRs were those that resulted in extended hospital stay and/or required significant treatment or resulted in the discontinuation of the TB medication. The likelihood of the association of an adverse event with a specific medication was defined as definite, probable, possible or unlikely, using the following five criteria: 1) known adverse drug reaction; 2) temporal relationship; 3) adverse reaction disappeared with dose reduction or discontinuation of the study medication; 4) symptoms could not be explained by any other known condition or predisposition of the patient; and 5) the symptoms reappeared upon re-challenge or laboratory tests showed higher than normal drug levels or metabolic disturbances, which explained the symptoms. The level of association was defined as definite when all five criteria were satisfied, probable when the first four criteria were satisfied, possible when the first three criteria were satisfied or unlikely when relevant information could not be obtained, the temporal sequence was atypical or other conditions or dispositions were considered far more likely to have caused the symptoms. Only those ADRs that were considered to be major events were included in the data analysis defined as a moderate or severe reaction that was definitely, probably or possibly associated with the TB medications according to the criteria.
870
The International Journal of Tuberculosis and Lung Disease
Statistical analysis In the primary analysis, the incidence of major ADRs to the first-line anti-tuberculosis regimens and patientrelated factors associated with the time to the first major event were determined. In addition, incidences were further stratified by each combination of firstline agents and each ADR category. Incidences were calculated as events per 100 person-months, with 95% confidence intervals (CIs) estimated. Cox proportional hazards regression analysis was used to estimate the hazard ratios (HR) and their 95%CIs by modeling the time until both the first ADR and all ADRs whilst taking into account important confounders. Treatment with the first-line anti-tuberculosis drug regimens was treated as time-dependent binary covariates indicating time on the drug regimen to account for the multiple starting and stopping patterns of these drugs throughout therapy. Multiple event analysis was performed where follow-up time is broken into segments defined by the events,12 and the robust sandwich estimate of Lin and Wei13 used for the covariance matrix to account for the correlation within patients due to multiple events. Time-dependent covariates were fitted to the models to test the proportional hazards assumptions.14 Statistical significance in this study was defined as P 0.05. SAS Version 9 (Statistical Analysis System, Cary, NC, USA) and R version 2.3.1 (The R Foundation for Statistical Computing, Vienna, Austria) were used for all statistical analysis.
RESULTS During the 5-year study period, a total of 1061 patients received first-line treatment for active TB (Table 1). The mean duration of treatment was 243 days (standard deviation [SD] 112) and the mean age was 50 years (SD 21). Half of the participants were female (50%) and the most common ethnic group was East Asian (40%). The majority of the patients were being treated for pulmonary TB (69%) and had a normal AST 40 U/l at baseline. More than half were exposed to the HRZE, HRZ and HR combination regimens. At least one major ADR was seen in 318 (30%) patients during the entire course of therapy. The total number of ADRs is stratified by type in Table 2. Each patient experiencing an ADR had on average more than two episodes. The most common ADRs were hepatitis (28% of all events), gastrointestinal disturbances (19%), rash (15%), weakness/fatigue (7%) and joint pain (6%). The number of patients with at least two events was 163 (51% of those having ADRs). The incidence of all major ADRs was 7.3 events per 100 person-months (95%CI 7.2–7.5). When patients were taking the four-drug regimen, the incidence of major ADRs was 23.3 per 100 person-months (95%CI 23.0–23.7): 13.6 (95%CI 13.3–14.0) when on RHZ; 2.4 (95%CI 2.3–2.6) when on HR; 5.2 (95%CI 4.9–5.4) when on other first-line drug regi-
Table 1
Baseline demographics
Characteristic Number of patients Total number of person-months studied Mean (SD) Number of person-months on first-line drugs Mean (SD) Sex Female Male Age, years, mean (SD) Age categories 35 years 35–59 years 60 years Ethnicity Asian (Oriental) South-east Asian White Aboriginal Other Unknown Mean weight, kg (SD)* Site of disease Pulmonary Other Unknown Baseline AST, U/L 40 40–79 80 Unknown HIV Positive Negative Unknown Hepatitis infection Any hepatitis infection Hepatitis A Hepatitis B Hepatitis C Patient exposure to first-line drug regimens† HRZE HRZ HR Other first-line drug regimens‡ Second-line drugs Person-days on each first-line drug regimen HRZE HRZ HR Other first-line drug regimens‡ Second-line drugs First-line drugs H R Z E
n (%) 1 061 8 800.5 8.3 (4.2) 8 476.9 8.0 (3.7) 528 (49.8) 533 (50.2) 49.9 (20.9) 272 (25.6) 433 (40.8) 356 (33.6) 419 (39.5) 298 (28.1) 129 (12.2) 104 (9.8) 62 (5.8) 49 (4.6) 58.0 (13.8) 735 (69.3) 325 (30.6) 1 (0.1) 849 (80.0) 102 (9.6) 30 (2.8) 80 (7.5) 58 (5.5) 467 (44.0) 536 (50.5) 93 (8.7) 4 (0.4) 25 (2.4) 69 (6.5) 661 (62.3) 572 (53.9) 699 (65.9) 427 (40.2) 255 (24.0) 32 463 38 806 107 167 79 575 9 847 1 044 (98.4) 1 031 (97.2) 989 (93.2) 771 (72.7)
* Weights were not recorded for 508 patients. † Due to the starting and stopping patterns of individual drugs, patients were often exposed to multiple regimens. ‡ Include H or RZE; HRE; E or H or RZ; H or RE. SD standard deviation; AST aspartate aminotransferase; H isoniazid; R rifampicin; Z pyrazinamide; E ethambutol.
mens; and 0.6 (95%CI 0.5–0.7) when on second-line drug regimens (Table 3). The incidence of ADRs stratified by the three most common ADRs (hepatitis, gastrointestinal disorders and rash) and by each drug is shown in Table 3.
Adverse drug reactions to TB medications
Table 2
Characterization of ADRs ADRs Likelihood of association with first-line drugs*
Patients experiencing each ADR n (%)
Total n
Possible n (%)
Probable n (%)
Definite n (%)
All adverse events†
318 (30.0)
646
177 (27.4)
334 (51.7)
135 (20.9)
Type of adverse event† Hepatitis GI (nausea/vomiting/GI upset) Rash Weakness/fatigue Joint pain Fever Pruritus Headache Vertigo/tinnitus Visual disturbances Paresthesias Anorexia/weight loss Abdominal pain Swelling Palpitations Others‡
148 (13.9) 106 (10.0) 80 (7.5) 39 (3.7) 34 (3.2) 25 (2.4) 22 (2.1) 19 (1.8) 18 (1.7) 18 (1.7) 14 (1.3) 10 (0.9) 9 (0.8) 8 (0.8) 5 (0.5) 7 (0.7)
179 124 94 46 36 28 23 22 21 18 15 10 9 9 5 7
45 (25.1) 36 (29.0) 18 (19.1) 15 (32.6) 6 (16.7) 7 (25.0) 6 (26.1) 7 (31.8) 7 (33.3) 12 (66.7) 2 (13.3) 4 (40.0) 4 (44.4) 3 (33.3) 1 (20.0) 3 (42.9)
100 (55.9) 57 (46.0) 51 (54.3) 20 (43.5) 27 (75.5) 14 (50.0) 14 (60.9) 8 (36.4) 8 (38.1) 4 (22.2) 10 (66.7) 5 (50.0) 5 (55.6) 4 (44.4) 4 (80.0) 4 (57.1)
34 (19.0) 31 (25.0) 25 (26.6) 11 (23.9) 3 (8.3) 7 (25.0) 3 (13.0) 7 (31.8) 6 (28.6) 2 (11.1) 3 (20.0) 1 (10.0) 0 (0) 2 (22.2) 0 (0) 0 (0)
* Assigned the likelihood of that of the first-line drug most likely associated with the ADR. † Patients or events add up to more than the total, as people could experience more than one event. ‡ Characterizes less than 1% of patients and includes seizures (three ADRs in three patients), shortness of breath (two ADRs in two patients), dyspepsia (one ADR) and neutrophilia (one ADR). ADR adverse drug reaction; GI gastrointestinal.
Table 3
Incidence of major ADRs
Treatment and side effects Overall All Hepatitis GI Rash HRZE All Hepatitis GI Rash HRZ All Hepatitis GI Rash HR All Hepatitis GI Rash Other first-line drug regimens† All Hepatitis GI Rash Second-line drugs All Hepatitis GI Rash
n
Personmonths
1061
8800
661
572
699
427
255
Persons n (%)
Events n
Incidence*
95%CI
318 (30.0) 148 (13.9) 106 (10.0) 80 (7.5)
646 179 124 94
7.3 2.0 1.4 1.1
7.2–7.5 1.9–2.1 1.3–1.5 1.0–1.1
154 (14.5) 54 (5.1) 52 (4.9) 42 (4.0)
249 57 55 44
23.3 5.3 5.2 4.1
23.0–23.7 5.2–5.5 5.0–5.3 4.0–4.3
108 (10.2) 46 (4.3) 34 (3.2) 20 (1.9)
174 47 34 23
13.6 3.7 2.7 1.8
13.3–14.0 3.5–3.8 2.5–2.8 1.7–1.9
59 (5.6) 34 (3.2) 13 (1.2) 8 (0.8)
86 37 13 8
2.4 1.1 0.4 0.2
2.3–2.6 1.0–1.1 0.3–0.4 0.2–0.3
81 (7.6) 32 (3.0) 19 (1.8) 17 (1.6)
135 38 21 19
5.2 1.5 0.8 0.7
4.9–5.4 1.3–1.6 0.7–0.9 0.6–0.8
1 (0.1) 0 (0.0) 1 (0.1) 0 (0.0)
2 0 1 0
0.6 0.0 0.3 0.0
0.5–0.7 NA 0.2–0.4 NA
1067
1275
3521
2614
324
* Incidence is expressed as events per 100 person-months. † Include H or RZE; HRE; E or H or RZ; H or RE. ADR adverse drug reaction; CI confidence interval; GI gastrointestinal; H isoniazid; R rifampin; Z pyrazinamide; E ethambutol; NA not applicable.
871
872
The International Journal of Tuberculosis and Lung Disease
Figure Smoothed hazard functions illustrating the rate of ADRs across treatment time. ADR adverse drug reaction.
Panel A of the Figure displays the smoothed rate of first ADRs across treatment time. First ADRs were more likely earlier in treatment, and very few first ADRs occurred after 150 days. This plot approximates the hazard function modeled within the Cox regression time to first event analysis. Table 4 shows the adjusted HRs for the time to first event. The following factors were independently associated with any major side effect: 1) baseline AST 80 U/l vs. 40 U/l (HR 2.8, 95%CI 1.7–4.7); 2) combination regimen of HRZE (HR 2.3, 95%CI 1.5–3.5); 3) the combination regimen of HRZ (HR 1.8 95%CI 1.2– Table 4
2.8); 4) females (HR 1.6, 95%CI 1.3–2.0); 5) age 35– 59 and 60 years relative to 35 years (HR 1.6, 95%CI 1.2–2.2, HR 2.0 95%CI 1.5–2.8, respectively); and 6) drug resistance (HR 1.6 95%CI 1.1– 2.3). None of these variables had significant interactions with the logarithm of treatment time, and are thus assumed to conform to the proportional hazards assumptions (i.e., they possess the given risk ratios across the entire treatment period). For example, the elevated risk of ADRs associated with exposure to HRZE was constant across the entire treatment period despite the fact that most first events occurred
Adjusted hazard ratios for risk factors associated with time to first ADR Any major adverse event (n 318)
First-line drug regimen* Other/none (reference) HRZE HRZ HR Any prior ADR* Female sex (vs. male) Age, years 35 (reference) 35–59 60 Asian ethnicity Base AST, U/l 40 (reference) 40–79 80 Hepatitis infection (vs. none or unknown)† Drug-resistant TB Pulmonary TB HIV-positive (vs. negative or unknown)
Hepatitis (n 148)
GI (n 106)
Rash (n 80)
HR (95%CI)
P value
HR (95%CI)
P value
HR (95%CI)
P value
HR (95%CI)
P value
1.0 2.3 (1.5–3.5) 1.8 (1.2–2.8) 0.6 (0.3–1.0)
0.01
0.01
1.0 2.2 (1.0–5.0) 1.6 (0.7–3.8) 0.9 (0.3–2.5) 1.6 (0.7–3.7) 1.5 (0.9–2.3)
0.14
0.01
1.0 2.7 (1.3–5.7) 2.1 (1.0–4.4) 0.9 (0.4–2.2) 1.7 (0.8–3.3) 2.0 (1.3–3.0)
0.01
1.6 (1.3–2.0)
1.0 2.9 (1.6–5.2) 2.1 (1.2–3.8) 1.0 (0.5–1.9) 1.2 (0.7–2.0) 1.8 (1.3–2.5)
1.0 1.6 (1.2–2.2) 2.0 (1.5–2.8) 1.1 (0.9–1.4)
0.01
1.0 1.5 (0.9–2.5) 2.5 (1.5–4.1) 1.3 (0.9–1.8)
0.01
1.0 1.4 (1.0–2.0) 2.8 (1.7–4.7)
0.01
1.0 2.5 (1.6–3.8) 5.7 (3.2–10.1)
0.01
1.0 1.8 (1.0–3.3) 4.8 (2.4–9.3)
0.01
1.0 0.7 (0.3–1.9) NA
0.01
1.0 (0.6–1.5) 1.6 (1.1–2.3) 0.8 (0.7–1.1)
0.91 0.03 0.16
1.6 (0.9–2.7) 2.0 (1.2–3.5) 0.9 (0.6–1.2)
0.13 0.01 0.37
0.8 (0.3–1.9) 1.2 (0.6–2.7) 0.5 (0.4–0.8)
0.55 0.62 0.01
0.5 (0.2–1.9) 2.4 (1.2–4.7) 0.6 (0.4–0.9)
0.33 0.01 0.02
0.9 (0.5–1.6)
0.73
1.3 (0.6–2.6)
0.45
0.7 (0.2–2.3)
0.53
1.3 (0.4–4.6)
0.71
0.51
0.56 0.01
0.16
1.0 1.7 (1.0–3.0) 2.3 (1.3–4.2) 1.0 (0.7–1.6)
0.15 0.01 0.02
0.82
1.0 1.2 (0.7–2.2) 1.7 (0.9–3.2) 1.2 (0.8–1.9)
0.22 0.11 0.18
0.38
* Time-dependent covariate. † Hepatitis infection hepatitis A, B or C. ADR adverse drug reaction; GI gastrointestinal; HR hazard ratio; CI confidence interval; H isoniazid; R rifampin; Z pyrazinamide; E ethambutol; AST aspartate aminotransferase; NA not applicable; TB tuberculosis; HIV human immunodeficiency virus.
Adverse drug reactions to TB medications
within the first 150 days. Variables associated with the time to hepatitis development and gastrointestinal ADRs were similar to those associated with any major ADR. In addition, for the gastrointestinal disturbances and rash analyses, pulmonary TB (HR 0.5 and 0.6, respectively) was found to be protective. Panel B of the Figure displays the smoothed rate of all ADRs across treatment time. This function has a similar shape to the hazard function for the first ADRs. Very few ADRs occurred after 200 days of treatment. All of the results shown in the simple time to event analyses were seen in the multiple event analyses (Table 5). Thus, baseline AST 80 U/l, combination regimens containing PZA, females, age 35–59 and 60 years, and drug resistance were also associated with any major side effect. However, the multiple event analysis also showed that having a prior ADR (HR 2.5 95%CI 1.9–3.3) was also significantly associated with the occurrence of a major ADR. Again, each of these variables conformed to the proportional hazards assumptions and were thus assumed to possess the given risk ratio across the entire treatment period. The results were similar for the occurrence of hepatitis, gastrointestinal events or rash, with the exception that a prior ADR of a different type was protective for an ADR, but a prior event of the same type was significantly associated with the occurrence of that event. Table 5
873
DISCUSSION This is the first, population-based study to examine major ADRs to first-line anti-tuberculosis medications that simultaneously accounted for the risks of all medications. Past research has examined the risk of each anti-tuberculosis medication independently without considering the range of agents to which each patient is exposed concurrently. We found that when anti-tuberculosis medication regimens were considered simultaneously as time-dependent covariates, regimens containing PZA were consistently associated with the time to first ADR. We confirmed this finding by conducting an analysis using the individual first-line agents as time-dependent covariates (data not shown). Consistently, PZA was the only drug found to have significantly elevated hazard ratios for all major ADRs, hepatitis, gastrointestinal events and rashes. Although it is possible that the higher incidence of ADRs observed with PZA may partially be explained by the fact that most exposures to PZAcontaining regimens occur early in the treatment of TB, the analysis with individual first-line agents revealed that PZA provided excess risk throughout the whole treatment period and that other drugs given early in therapy (e.g., INH and RMP) were not significantly associated with ADRs.
Adjusted hazard ratios for risk of ADRs using multiple event analysis Any major adverse event (n 456)
First-line drug regimen* Other/none (reference) HRZE HRZ HR Any prior ADR* Prior ADR of a different type* Prior ADR of same type* Female sex (vs. male) Age, years 35 (reference) 35–59 60 Asian ethnicity Base AST, U/L 40 (reference) 40–79 80 Hepatitis infection (vs. none or unknown)† Drug-resistant TB Pulmonary TB HIV positive (vs. negative or unknown)
Hepatitis (n 179)
GI (n 124)
Rash (n 94)
HR (95%CI)
P value
HR (95%CI)
P value
HR (95%CI)
P value
HR (95%CI)
P value
1.0 2.7 (1.9–3.8) 2.1 (1.5–2.9) 0.9 (0.6–1.3) 2.5 (1.9–3.3)
0.01
1.0 3.1 (1.8–5.6) 2.5 (1.4–4.4) 1.3 (0.7–2.3)
0.01
1.0 2.3 (1.1–4.5) 1.8 (0.9–3.5) 0.8 (0.4–1.8)
0.01
1.0 1.8 (0.8–3.7) 1.3 (0.6–2.9) 0.7 (0.3–1.8)
0.19
0.7 (0.4–1.2) 2.0 (1.1–3.5) 1.7 (1.3–2.4)
0.17 0.02 0.01
0.5 (0.3–0.9) 2.5 (1.4–4.7) 1.9 (1.3–2.8)
0.02 0.01 0.01
0.5 (0.2–0.9) 3.8 (1.9–7.7) 1.5 (1.0–2.2)
0.02 0.01 0.07
1.0 1.7 (1.1–2.7) 2.5 (1.6–3.9) 1.3 (1.0–1.8)
0.01
1.0 1.5 (0.9–2.6) 2.0 (1.2–3.5) 1.2 (0.8–1.7)
0.04
1.0 1.4 (0.8–2.3) 1.9 (1.1–3.3) 1.3 (0.9–2.0)
0.04
0.01
1.5 (1.2–1.8)
0.01
1.0 1.5 (1.2–2.0) 2.1 (1.6–2.8) 1.1 (0.9–1.4)
0.01
1.0 1.3 (1.0–1.8) 1.8 (1.3–2.5)
0.01
1.0 2.3 (1.5–3.4) 3.9 (2.3–6.5)
0.01
1.0 1.7 (1.0–2.9) 3.6 (2.2–6.1)
0.01
1.0 0.7 (0.3–1.6) NA
0.41
0.9 (0.6–1.3) 1.6 (1.2–2.2) 0.9 (0.7–1.1)
0.51 0.01 0.24
1.3 (0.8–2.3) 2.0 (1.2–3.4) 0.9 (0.6–1.2)
0.30 0.01 0.48
0.7 (0.3–1.5) 1.4 (0.7–2.9) 0.6 (0.4–0.9)
0.38 0.37 0.01
0.4 (0.1–1.5) 1.9 (1.1–3.3) 0.7 (0.4–1.0)
0.17 0.02 0.05
1.2 (0.7–2.1)
0.57
1.4 (0.7–2.8)
0.34
0.6 (0.2–1.6)
0.28
0.8 (0.2–3.0)
0.74
0.18
0.09
0.38
0.20
* Time-dependent covariate. † Hepatitis infection hepatitis A, B or C. ADR adverse drug reaction; GI gastrointestinal; HR hazard ratio; CI confidence interval; H isoniazid; R rifampin; Z pyrazinamide; E ethambutol; AST aspartate aminotransferase; NA not applicable; TB tuberculosis; HIV human immunodeficiency virus.
874
The International Journal of Tuberculosis and Lung Disease
Other important findings were the large number of ADRs experienced and the nature of these events (most were hepatic, gastrointestinal or skin-related) which occurred within the first 100 days. We also confirmed the findings of others in that female sex, increased age, increased baseline AST, and drug resistance were associated with increased occurrence of ADRs.15 Using our multiple event analysis, we found that previous ADRs of the same type were associated with a higher risk of recurrence, whereas previous events of a different type were protective. Yee et al. have previously reported the incidence of serious ADRs to first-line medications. In their study of 430 patients, 46 serious events occurred involving 37 patients (incidence of 1.48 per 100 person-months).6 This was a clinic-based rather than a population-based study, and it therefore had a much lower number of events than in our study. Their study also used slightly different definitions for ADRs from ours. For example, we included ADRs that were classified as ‘possible’, ‘probable’ or ‘definite’. The definitions employed by Yee et al. would have fit our criteria of only ‘probable’ or ‘definite’, and thus the 30–50% of events classified as ‘possible’ would not have been evaluated. Yee et al. compared drug risks by comparing drugspecific incidence rates (i.e., the number of ADRs attributed to the drug per person-time on the drug). This may have exaggerated differences between INH, RMP and PZA, as PZA is often only given early in treatment, and thus person-time is low. However, in our analysis, we compared the drug risks by evaluating exposure to combinations of medications using timedependent variables within the Cox regression analysis, which fully took into account the actual timing during treatment of the multiple stops and re-starts of medication. As PZA-associated hepatotoxicity is thought to be dose-dependent,16,17 most guidelines suggest using lower doses, at 20–25 mg/kg, for the treatment of active TB.1 In 2000, the US Centers for Disease Control and Prevention published the first report on cases of fatal hepatitis associated with the use of RMP/PZA for the treatment of latent TB infection, a phenomenon that remains largely unexplained.18 Since that time, other observational studies, including ours, have found that the incidence of hepatotoxicity is higher with PZA than other anti-tuberculosis medications. The analysis in the present study did not detect any association between viral hepatitis and ADRs. Other investigators have also looked at this association with hepatitis B and C; the results vary, with some finding an increased incidence of hepatotoxicity,6–8,19–21 while others did not.9 Our study may have underestimated the risk of both viral hepatitis and HIV on the occurrence of ADRs. Although we linked the populationbased TB registry with the provincial hepatitis registry, it is possible that some TB patients may have been positive for viral hepatitis but were not tested. Simi-
larly, we could not confirm that all patients had been tested for HIV status. However, we believe that the risk for this latter misclassification is limited, as most TB patients with significant risk factors for HIV would likely be tested in our health care environment. Another concern regarding the rate of ADRs reported in our study is the potential impact on completion rates. Given the challenge of completing therapy, even in patients who tolerate treatment, the current results suggest that there is a further reason to use the DOTS strategy in all patients. One assumes, based on other diseases, that patients who experience adverse events will likely have reduced adherence and thus unsupervised therapy may be associated with a delay in identification of an adverse event and/or a failure to continue on a recommended regimen. Such incomplete treatment is likely to contribute to less effective cure rates and, if selectively initiated by patients, facilitate the development of drug resistance. In summary, we found that the currently available first-line anti-tuberculosis drugs are associated with significant ADRs that are often serious in nature. As many of these drugs were developed several decades ago, little research has been done to develop drugs with more favorable harm-benefit profiles. This has been due in part to the declining incidence of TB in the industrialized world, resulting in less favorable profit margins for innovative pharmaceuticals. There has been some resurgence in interest in TB research into diagnostic (i.e., gamma interferon assays) and vaccine strategies, but more focus needs to be paid to the development of effective, safe and affordable pharmaceutical treatments. References 1 American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America. Treatment of Tuberculosis. Am J Respir Crit Care Med 2003; 167: 603–662. 2 Long R, ed. Canadian tuberculosis standards. 5th ed. Ottawa, Canada: Canadian Lung Assoc and Health, 2000. 3 Combs D L, O’Brien R J, Geiter L J. USPHS tuberculosis shortcourse chemotherapy trial 21: effectiveness, toxicity and acceptability: the report of final results. Ann Intern Med 1990; 112: 397–406. 4 Jindani A, Nunn A J, Enarson D A. Two 8-month regimens of chemotherapy for treatment of newly diagnosed pulmonary tuberculosis: international multicentre randomized trial. Lancet 2004; 364: 1244–1251. 5 Forget E J, Menzies D. Adverse reactions to first-line antituberculosis drugs. Expert Opin Drug Saf 2006; 5: 231–249. 6 Yee D, Valiquette C, Pelletier M, et al. Incidence of serious side effects from first-line antituberculosis drugs among patients treated for active tuberculosis. Am J Respir Crit Care Med 2003; 167: 1472–1477. 7 Schaberg T, Rebhan K, Lode H. Risk factors for side effects of isoniazid, rifampin and pyrazinamide in patients hospitalized for pulmonary tuberculosis. Eur Respir J 1996; 9: 2026–2030. 8 Ungo J R, Jones D, Ashkin D, et al. Antituberculosis druginduced hepatotoxicity: the role of hepatitis C virus and the human immunodeficiency virus. Am J Respir Crit Care Med 1998; 157: 1871–1876.
Adverse drug reactions to TB medications
9 Hwang S J, Wu J C, Lee C N, et al. A prospective clinical study of isoniazid-rifampicin-pyrazinamide-induced liver injury in an area endemic for hepatitis B. J Gastroenterol Hepatol 1997; 12: 87–91. 10 Essebag V, Platt R W, Abrahamowicz M, Pilote L. Comparison of nested case-control and survival analysis methodologies for analysis of time-dependent exposure. BMC Med Res Methodol 2005; 5: 5. 11 Marra F, Marra C A, Moadebi S, et al. Levofloxacin treatment of active tuberculosis and the risk of ADRs. Chest 2005; 128: 1406–1413. 12 Andersen P K, Gill R D. Cox’s regression model counting process: a large sample study. Ann Stat 1982; 10: 1100–1120. 13 Lin D Y, Wei L J. The robust inference for the proportional hazards model. J Am Stat Assoc 1989; 84: 1074–1078. 14 Hosmer D W, Lemeshow S. Applied survival analysis: regression modeling of time to event data. New York, NY, USA: John Wiley, 1999. 15 Horsburgh C R Jr. Priorities for the treatment of latent tuberculosis infection in the United States. N Engl J Med 2004; 350: 2060–2067. 16 McDermott W, Ormond L, Muschenhein C, Deuschle K,
17
18
19
20
21
875
McCune R M, Tompsett R. Pyrazinamide-isoniazid in tuberculosis. Am Rev Tuberc 1954; 69: 319–333. Campagna M, Calix A A, Hauser G. Observations on the combined use of pyrazinamide (aldinamide) and isoniazid in the treatment of pulmonary tuberculosis. Am Rev Tuberc 1954; 69: 334–350. Centers for Disease Control and Prevention. Fatal and severe hepatitis associated with rifampin and pyrazinamide for the treatment of latent tuberculosis infection—New York and Georgia 2000. MMWR Morb Mortal Wkly Rep 2001; 50: 289–291. Wong W M, Wu P C, Yuen M F, et al. Antituberculosis drugrelated liver dysfunction in chronic hepatitis B infection. Hepatology 2000; 31: 201–206. Pan L, Jia Z S, Chen L, et al. Effect of anti-tuberculosis therapy on liver function of pulmonary tuberculosis patients infected with hepatitis B virus. World J Gastroenterol 2005; 11: 2518– 2521. Lee B H, Koh W J, Choi M S, et al. Inactive hepatitis B surface antigen carrier state and hepatotoxicity during anti-tuberculosis chemotherapy. Chest 2005; 127: 1304–1311.
RÉSUMÉ C O N T E X T E : Le traitement standard de la tuberculose (TB) active comporte l’isoniazide (INH), la rifampicine (RMP), le pyrazinamide (PZA) et l’éthambutol (EMB). Bien que ce régime soit efficient pour le traitement de la TB active, il s’accompagne de nombreuses réactions indésirables à l’égard des médicaments (ADR) qui constituent un obstacle significatif à l’achèvement du traitement. O B J E C T I F S : Examiner de l’incidence des principaux ADR et des facteurs de risque associés aux médicaments antituberculeux de première ligne. M É T H O D E S : Cette étude a évalué les patients recevant un traitement pour TB active dans une base de données basée sur la population (2000–2005). La nature des ADR, leur risque d’association avec les médicaments étudiés et leur gravité ont fait l’objet d’une évaluation. R É S U L T A T S : Sur 1061 patients traités, 318 (30%) ont
connu au moins un ADR. L’incidence globale de l’ensemble des ADR majeurs a été de 7,3 incidents par 100 personnes-mois (IC95% 7,2–7,5) : de 23,3 (IC95% 23,0– 23,7) en cas de quadrithérapie de première ligne ; de 13,6 (IC95% 13,3–14,0) en cas de trithérapie RMP, INH, PZA ; de 2,4 (IC95% 2,3–2,6) en cas de bithérapie INH et RMP. Les risques relatifs ajustés ont démontré que les régimes de combinaisons comportant le PZA, le sexe féminin, les groupes d’âge de 35–59 et 60 ans ainsi que des taux de base d’AST 80 U/l et la résistance aux médicaments ont été en association avec n’importe quelle complication majeure. C O N C L U S I O N S : Les médicaments antituberculeux de première ligne sont associés à des ADR significatifs. Il existe plusieurs facteurs de risque associés à l’apparition d’ADR, parmi eux l’exposition à des régimes contenant le PZA. RESUMEN
M A R C O D E R E F E R E N C I A : El tratamiento de referencia de la tuberculosis (TB) activa consiste en isoniacida (INH), rifampicina (RMP), pirazinamida (PZA) y etambutol (EMB). Si bien esta pauta es eficaz en el tratamiento de la TB activa, se asocia con frecuentes reacciones adversas a los medicamentos (ADR), las cuales constituyen un obstáculo considerable a la compleción del tratamiento. O B J E T I V O S : Estudiar la incidencia de ADR graves a los medicamentos antituberculosos de primera línea y los factores de riesgo asociados. M É T O D O S : En el presente estudio, se estudiaron los pacientes que recibieron tratamiento por TB activa entre 2000 y 2005, a partir de una base de datos poblacional. Se evaluaron el tipo y la gravedad de las ADR y su probabilidad de asociación con los medicamentos estudiados. R E S U L T A D O S : De los 1061 pacientes que recibieron tratamiento, 318 (30%) presentaron al menos una ADR. La
incidencia global de ADR graves fue 7,3 por 100 mesespersona (IC95% 7,2–7,5) ; esta incidencia fue 23,3 (IC95% 23,0–23,7) cuando se administraron los cuatro medicamentos de primera línea ; 13,6 (IC95% 13,3–14,0) cuando se administraron RMP, INH y PZA ; y 2,4 (IC95% 2,3– 2,6) cuando se administraron solo INH y RMP. El cociente ajustado de riesgos instantáneos reveló que los siguientes factores estaban asociados con la presencia de una ADR : las pautas terapéuticas combinadas que contienen PZA, el sexo femenino, los grupos de edad entre 35 y 59 años y de 60 años, una concentración sérica inicial de aspartatoaminotransferasa 80 U/l y la farmacorresistencia. C O N C L U S I Ó N : Los medicamentos antituberculosos de primera línea se asocian con la aparición de ADR notables. Existen varios factores de riesgo asociados con la aparición de ADR, entre ellos la exposición a pautas terapéuticas que contienen PZA.