This Article Abstract Full Text (PDF) Online Supplement All Versions of this Article: 200407-905OCv1 170/10/1124    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Chang, K. C. Articles by Tam, C. M. Search for Related Content PubMed PubMed Citation Articles by Chang, K. C. Articles by Tam, C. M.

A Nested Case–Control Study on Treatment-related Risk Factors for Early Relapse of Tuberculosis

Tuberculosis and Chest Service, Department of Health; Tuberculosis and Chest Unit, Grantham Hospital; and Department of Community and Family Medicine, School of Public Health, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong SAR, China

Correspondence and requests for reprints should be addressed to Dr. Kwok Chiu Chang, Yaumatei Chest Clinic, 2/F, Yaumatei Jockey Club Polyclinic, 145, Battery Street, Kowloon, Hong Kong, China. E-mail: ymtcc{at}dh.gov.hk

	ABSTRACT

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
This nested case–control study aimed at evaluating treatment-related risk factors of relapse of tuberculosis under a service program of directly observed treatment. Out of 12,183 patients with pulmonary tuberculosis who completed treatment within 1 year, 113 relapsed within 30 months after commencement of therapy. The overall 30-month relapse rate was 0.9% (95% confidence interval [CI] 0.8–1.1%). On matching 113 cases with 226 control subjects in a conditional logistic model, thrice-weekly treatment increased the risk of relapse in comparison with daily treatment (odds ratio 3.92, 95% CI 1.78–8.63), whereas prolonging both intensive phase and overall treatment by 50% or more protected against relapse (odds ratio 0.24, 95% CI 0.08–0.70). When pretreatment culture was positive and cavitation was absent, the 30-month relapse rate for standard thrice-weekly regimen was 1.1% (95% CI 0.6–2.0%). The corresponding rates in the presence of cavitation were 7.8% (95% CI 4.0–14.6%) for standard thrice-weekly regimen; 3.3% (95% CI 1.9–5.5%) for standard daily regimen; 0.5% (95% CI 0.1–2.6%) for extended thrice-weekly regimen; and 0.4% (95% CI 0.1–0.9%) for extended daily regimen. Further studies are required to reduce the risk of relapse under program settings.

Key Words: logistic models • recurrence • risk factors • therapeutics • tuberculosis

Relapse of tuberculosis refers to the situation in which a patient becomes and remains culture-negative while receiving antituberculosis drugs but develops active tuberculosis again after completion of treatment (1). The reported rates of relapse after standard 6-month treatment for drug-susceptible pulmonary tuberculosis were approximately 1 to 2% at 24 months (2), and 3.4% over 5 years (3). Host factors such as sex (4), disease factors (3–8), and treatment-related factors (4, 9–14) possibly interact to cause relapse. Despite its infrequent occurrence, relapse is the most informative measure of the efficacy of antituberculosis regimens (5). A better understanding of the risk factors of relapse is conducive to optimizing the current treatment and paving the way for future development.

Although many clinical trials on antituberculosis chemotherapy have been published, the most effective means of decreasing the likelihood of relapse for patients at increased risk has not yet been determined (1). This may be attributed to a scarcity of studies that have assessed the interaction of multiple host factors, disease factors, progress factors, and treatment-related factors in a comprehensive and systematic manner.

This study, approved by the Ethics Committee of the Department of Health, aimed at better delineating the role of treatment-related risk factors for relapse amidst multiple covariates under the service setting of directly observed treatment in Hong Kong. Thrice-weekly and daily 6-month short-course regimens are both standard regimens in Hong Kong, although there is a tendency to use daily regimens in hospitals and intermittent regimens in outpatient clinics. All patients are followed up clinically, radiologically, and if necessary, bacteriogically with sputum examination every 6 months, or as clinically indicated, for at least 2 years after commencement of treatment. Medical practitioners are statutorily required to notify tuberculosis to the Tuberculosis and Chest Service of the Department of Health. The notification database has been computerized with the identity card number as the unique identifier. It is therefore possible to ascertain the clinical status of previously treated patients after the regular follow-up period.

While treatment-related risk factors of relapse may be best studied by randomized controlled trials (15), it is often difficult to conduct a trial with sufficient statistical power to address an uncommon treatment outcome when the effect size is modest. A nested and matched case–control study may be the second best design for studying relapse if confounding in the service setting is properly addressed. Matching of basic confounding factors already characterized in a large cohort may increase the statistical power of detecting the influence of other clinically relevant factors that are not easily characterized on a large scale.

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
From a computerized registry of tuberculosis patients treated in government chest clinics, all pulmonary tuberculosis patients who commenced treatment from January 1 1998 to December 31 2000, and completed treatment within one year were identified. The cohort was followed through the territory-wide tuberculosis reporting system for relapse up to 30 months after commencement of treatment. Each case was matched individually with two control subjects nested within the cohort by sex, age, year in which treatment was commenced, chest clinic, and pretreatment sputum culture status. Medical records of all cases and control subjects were retrieved for verification and data capture. Active pulmonary tuberculosis was defined as tuberculosis confirmed by the isolation of Mycobacterium tuberculosis in sputum or bronchial aspirate, or compatible clinico-radiographic and/or histologic findings together with an appropriate response to antituberculosis treatment. Relapse was defined as the circumstance in which a patient became or remained culture-negative while receiving antituberculosis drugs, but at some point after completion of therapy, either became culture-positive again or experienced clinical and/or radiographic features consistent with active tuberculosis, including an appropriate response to retreatment. Patients with multidrug-resistant tuberculosis were excluded.

Statistical Analysis
A basic conditional logistic model of relapse was set up with three groups of covariates (16). The first group included three treatment-related factors (treatment regimens, dosing frequency, and treatment prolongation) that were forced in regardless of p values. The second group consisted of baseline host and disease factors that were significantly associated with the allocation of treatment regimens, dosing frequency, or treatment prolongation. They were identified among all matched control subjects by univariate analysis (using the chi-square test with or without Yates' correction and the Fisher's exact test) followed by forward stepwise logistic regression analysis. The third group comprised baseline host and disease factors that were significantly associated with relapse by univariate conditional logistic regression analysis (16). These three groups of covariates were all forced into the basic logistic model. Forward stepwise conditional logistic regression using the maximum likelihood ratio test was then used to test if the basic logistic model would be changed significantly by other treatment-related factors (directly observed treatment at the clinic/adherence during intensive and continuation phases) and factors evaluated after 2 to 3 months of treatment (persistence of positive sputum smear/culture). Significant factors thus identified were then forced into the basic logistic model to derive an overall conditional logistic model of relapse. The level of statistical significance for univariate analysis was set at 0.1, whereas that for the conditional logistic model was 0.05. The p values for entry and removal during stepwise selection were 0.05 and 0.1, respectively. Multicollinearity was considered before performing logistic regression analysis. Interaction was considered after identifying the main effects. SPSS version 10 (Chicago, IL) was used.

Thirty-month relapse rates were calculated for clinically important subgroups within the cohort. It was assumed that the distribution of the treatment-related variables could be extrapolated from the control subjects to the cohort after stratification by the matched variables. Sampling fractions for different strata were used to compute the number of subjects in each subgroup.

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
Study Subjects
A total of 18,749 patients with tuberculosis were identified from the computerized tuberculosis registry. Only 12,946 patients completed antituberculosis treatment for pulmonary tuberculosis with or without extrapulmonary tuberculosis within 1 year. After excluding 763 subjects with incomplete data on sex, age, or pretreatment sputum culture, 12,183 subjects were finally included in the study cohort.

A total of 113 cases that relapsed within 30 months after commencement of treatment were identified from the study cohort by searching the tuberculosis registry and crosschecking with the territory-wide tuberculosis reporting system. There were 44 cases in 1998, 35 cases in 1999, and 34 cases in 2000. They were under the care of 17 chest clinics. Eighty-four cases (74.3%) were male. The mean age upon diagnosis of tuberculosis was 47.8 ± 19.1 years (SD). The pretreatment culture status was positive by sputum bacteriology in 81 cases (71.7%) and by bronchial aspirate bacteriology in three cases (2.7%). The mean time to relapse (counted from commencement of treatment) was 14.9 ± 5.5 months (SD), with 77 cases (68.1%) relapsing within 18 months, 30 cases (26.6%) from 18 to 24 months, and 6 cases (5.3%) from 24 to 30 months. Table 1 summarizes the sites of initial episodes of tuberculosis and relapse. Relapse involved the lung in 94 patients and extrapulmonary sites in 26 patients, including seven with both pulmonary and extrapulmonary lesions. The most common extrapulmonary sites of relapse were lymph nodes (involving 15 cases) and pleura (involving 9 cases).

Univariate analysis by the chi-square test showed no significant differences between the 113 cases and the 12,070 subjects in the rest of the cohort by the year in which treatment was commenced (p = 0.586), sex (p = 0.206), quintessential age groups (p = 0.256), and pretreatment sputum culture (p = 0.104). The chi-squared goodness-of-fit test for the distribution of relapses among the 17 clinics was not interpretable because the expected frequencies were fewer than two in four cells (11.1%) and fewer than five in seven cells (19.4%). Significant differences could be demonstrated when clinics were recoded under three main geographic districts, namely, Hong Kong Island, Kowloon, and the New Territories (p = 0.013). A two-by-two contingency table showed a significantly higher proportion of relapses among clinics on the Hong Kong Island (p = 0.004). However, the difference became insignificant after stratification by age groups (p > 0.1).

Univariate conditional logistic regression analysis showed that history of opiate use, coexisting extrapulmonary tuberculosis, higher radiographic extent and cavitation on initial chest radiograph, body weight below 50 kg, thrice-weekly treatment throughout, and persistence of positive culture after 2 to 3 months of treatment were associated with increased risks of relapse (Table 2).

Conditional Logistic Model of Relapse Among all Cases
Table 3 shows the baseline host and disease factors that might affect the allocation of three principal treatment-related factors. Table 4 shows the results of multiple conditional logistic regression analysis among all matched cases and control subjects. On matching 113 cases with 226 control subjects in a conditional logistic model, thrice-weekly treatment increased the risk of relapse in comparison with daily treatment (odds ratio [OR] 3.92, 95% confidence interval [CI] 1.78–8.63), whereas prolonging both the intensive phase and overall treatment protected against relapse (OR 0.24, 95% CI 0.08–0.70). Extension of either the intensive phase or overall treatment alone did not confer statistically significant protection. Other risk factors included coexisting tuberculous lymphadenitis, cavitation on initial chest radiograph, and conditions predisposing to active tuberculosis.

Subgroup Analysis
Table E2 in the online supplement shows the conditional logistic model among cases with positive pretreatment culture. In addition to thrice-weekly treatment throughout and prolongation of treatment in both phases, the extent of directly observed treatment at the clinic during intensive phase was found to be significantly associated with relapse. An increase of 10% of directly observed treatment reduced the risk of relapse by 15.8% (95% CI 0–28.5%). Furthermore, a suggestive linear trend of increasing protection was shown with the weakest protection provided by prolongation of the intensive phase alone and the strongest by extension of both the intensive phase and overall treatment. Cavitation on initial chest radiograph and conditions predisposing to active tuberculosis were consistently identified as risk factors.

Risk of Relapse
The overall 30-month relapse rate was 0.9% (95% CI 0.8–1.1%). Assuming that the matched control group was representative of the whole cohort in the distribution of disease characteristics and dosing schedules after stratification by the matched variables, the 30-month relapse rate for patients with positive pretreatment culture in the absence of cavitation on initial chest radiograph was 1.1% (95% CI 0.6–2.0%) for the standard thrice-weekly regimen. In the presence of cavitation, the corresponding rate for the standard thrice-weekly regimen was 7.8% (95% CI 4.0–14.6%), which became 3.8% (95% CI 1.5–9.3%) when intensive phase alone was prolonged, 0.9% (95% CI 0.3–3.4%) when overall treatment was prolonged, and 0.5% (95% CI 0.1–2.6%) when both were prolonged. The corresponding rates for the standard daily regimen were 3.3% (95% CI 1.9–5.5%), 0.9% (95% CI 0.4–2.3%), 2.6% (95% CI 1.2–5.5%), and 0.4% (95% CI 0.1–0.9%), respectively.

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION REFERENCES

 
This is probably the first observational study that has shown a lower treatment efficacy of thrice-weekly antituberculosis regimens in comparison with daily treatment. Consistent findings were shown by sensitivity and subgroup analyses.

There was good evidence that intermittent treatment after an initial daily phase as short as 2 weeks produced results equal to those of daily treatment (17). However, these studies might not have sufficient statistical power to detect a difference when the odds ratio was modest and the prevalence of the outcome measure was low.

A better therapeutic efficacy of daily treatment is not surprising for two reasons. First, most of the risk factors of relapse are ultimately related to the bacterial load of the lesions (5). Second, M. tuberculosis replicates approximately once daily when it is actively dividing. Faster sterilization is expected when actively dividing bacteria are killed daily, even though postantibiotic effects of intermittent treatment may continue to inhibit bacteria (18). Our study showed no significant difference in the risk of relapse between daily treatment throughout and daily treatment in only one phase of treatment. If this is not a type II error caused by the small proportions of subjects who were allocated the latter dosing schedules, it may be logical to prescribe daily treatment throughout intensive phase, followed by thrice-weekly treatment during continuation phase when the bacterial population comprises mainly "persisters" that replicate infrequently (19). Such a dosing schedule may also minimize any inconvenience incurred by directly observed treatment.

Our study reaffirmed the protective effect of prolonging treatment. Two randomized controlled trials using modern short-course regimens demonstrated the protective effect of extending the total treatment duration among patients with silico-tuberculosis and patients coinfected with tuberculosis and human immunodeficiency virus (HIV) (11, 13). These at-risk groups have a much higher risk of relapse than the usual immunocompetent patients. This might have allowed the differences to be detected readily with a randomized controlled trial design. In this nested case–control study, we also showed that extension of both intensive phase and overall treatment by 50% and above protected against relapse. Similar effect was seen in subgroup analysis among initially culture-positive patients, but not among culture-negative ones. Although we failed to demonstrate the protective effect of extension of either the intensive phase or overall treatment alone, type II error could not be completely excluded.

Directly observed treatment also appeared to play an important role in this study, even though its role has been most difficult to substantiate in randomized controlled trials. In subgroup analysis among subjects with positive pretreatment culture, directly observed treatment at the clinic during intensive phase significantly reduced the risk of relapse. The observation of such an effect among a cohort with apparently good adherence suggests that failure to take supplied medications may be more common than expected.

In addition to the identification of treatment-related risk factors of relapse, we also defined their roles amidst other independent risk factors. First, we reaffirmed that cavitation on initial chest radiograph was an independent risk factor of relapse. A randomized controlled trial reported a 21% rate of relapse among subjects with cavitation on chest radiograph and a positive culture upon completion of 2 months of therapy versus 2% for patients with neither factor (6). Second, we found by sensitivity analysis a significant association between relapse and extensive disease on chest radiograph, which had been reported by randomized controlled trials (4, 6). Third, we showed that relapse was significantly associated with conditions predisposing to active tuberculosis, among which diabetes mellitus was the most common in our study. This finding is consistent with the higher risk of relapse among patitents with silico-tuberculosis (11) and subjects coinfected with tuberculosis and HIV (13), although there were only five cases of silico-tuberculosis and no subjects coinfected with tuberculosis and HIV in our study. The absence of subjects coinfected with tuberculosis and HIV was expected because the prevalence of HIV infection among patients with tuberculosis in chest clinics was low. Unlinked anonymous urine testing conducted in chest clinics from 1998 to 2000 showed that approximately 0.5% of patuents with tuberculosis were HIV-seropositive (20). Fourth, we identified a significant association between relapse and coexisting tuberculous lymphadenitis. A previous metaanalysis reported a relapse rate of 3.3% (95% CI 1.7–5.5%) over a mean follow-up period of 31 months after treatment (21). However, only one out of the 12 cases of relapse in the lymph node in our study was confirmed by culture. Some of these cases might be attributed to a dysregulated immune system rather than true relapse due to inadequate sterilization. Paradoxical enlargement of existing lymph nodes and emergence of new lymph nodes have been reported in up to 30 percent of patients with tuberculous lymphadenitis after commencement of treatment (22).

We studied the matched variables by comparing their distribution between cases and the rest of the cohort. There were no significant differences in their distribution by sex, age, and pretreatment sputum culture. The association between male sex and relapse has been reported in only one published clinical trial (4). Two other studies showed no significant association between sex and relapse (5, 6). We did not show an association between advanced age and relapse, which was identified by an earlier study (5); neither did the Tuberculosis Trials Consortium reproduce this association (6). We found a significant difference in the relapse rates among clinics encoded under three main geographic districts of Hong Kong, but the difference became insignificant after stratification by age groups.

Although it might not be necessary to match cases and control subjects in the absence of significant difference in the distribution of the matching variables, we believe that individual matching of cases and control subjects by the basic confounders might still increase the statistical power by inducing concordance among subjects within matched units (23).

An important source of confounding came from any systemic bias in the allocation of treatment regimen, dosing frequency, and prolongation of treatment. We have tried to identify most of these confounding factors by studying the matched variables within the entire cohort (sex, age, pretreatment culture, and clinic) and examining the associations between possible confounders and treatment-related factors among the control subjects, which should reflect the general situation in Hong Kong. All identified confounders were controlled either by matching or multiple conditional logistic regression analysis.

Misclassification bias might also affect the validity of our study. Although misclassification bias might have arisen out of incomplete statutory reporting, the likelihood of incomplete reporting of relapse cases was much lower in government chest clinics than the other settings because of the sound infrastructure. Misclassification bias might also have occurred owing to premature loss to follow-up after completing treatment, for example, emigration, and mislabeling cases as control subjects. However, this was not likely because a major proportion of control subjects completed 30 months of follow-up, the overall relapse rate was low, and the sampling fraction was small. Misclassification bias might also occur from mislabeling reinfection as relapse. We did not perform genotyping to distinguish between the two scenarios within our service settings. However, such misclassification would be unlikely in view of the short time between relapse and treatment completion, a low annual risk of infection estimated around 0.6% per annum in Hong Kong (20, 24), and the even smaller proportion of reinfected patients developing clinically manifest disease. Even if misclassification did occur, it would be expected to dilute any contrast between cases and control subjects, thus leading to an underestimation of the true effect.

We failed to find a significant association between relapse and persistence of positive sputum culture after completing 2 months of treatment, which was previously shown to be an independent risk factor of relapse (6). Although multicollinearity was considered before performing logistic regression analysis, weak correlation between persistently positive culture and treatment-related factors, though statistically insignificant, might still be sufficient to reduce the statistical power in detecting such an influence. Missing data in 19% of matched subjects with positive pretreatment culture might also have contributed.

With the widespread application of intermittent regimens to promote directly observed treatment in many high-incidence areas, there is a need to reexamine the optimal mode of drug administration. Based on our estimates of 30-month relapse rates, the standard thrice-weekly regimen with fewer administered doses might still be cost-effective in the absence of cavitation on initial chest radiograph. However, among patients with positive pretreatment culture and cavitation, meaningful reduction of the 30-month relapse rate could be achieved by a daily regimen, and the risk might be further reduced by prolonging treatment. Further studies are required to work out the best strategy that maximizes both patient adherence and treatment outcome under program settings.

	Acknowledgments

 
The authors thank all colleagues of the Tuberculosis and Chest Service for their important contribution in collecting data for the computerized tuberculosis registry.

	FOOTNOTES

 
This article has an online supplement, which is viewable in this issue's table of contents at www.atsjournals.org

Conflict of Interest Statement: K.C.C. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; C.C.L. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; W.W.Y. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.C.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; C.M.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.

Received in original form July 13, 2004; accepted in final form September 9, 2004

	REFERENCES

TOP ABSTRACT METHODS RESULTS DISCUSSION 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.[Free Full Text]
2. British Thoracic Association. A controlled trial of six months chemotherapy in pulmonary tuberculosis. Second report: results during the 24 months after the end of chemotherapy. Am Rev Respir Dis 1982;126:460–462.[Medline]
3. Hong Kong Chest Service/ British Medical Research Council. Five-year follow-up of a controlled trial of five 6-month regimens of chemotherapy for tuberculosis. Am Rev Respir Dis 1987;136:1339–1342.[Medline]
4. Tam CM, Chan SL, Kam KM, Goodall RL, Mitchison DA. Rifapentine and isoniazid in the continuation phase of a 6-month regimen. Final report at 5 years: prognostic value of various measures. Int J Tuberc Lung Dis 2002;6:3–10.[Medline]
5. Aber VR, Nunn AJ. Factors affecting relapse following short-course chemotherapy. Bull Int Union Tuberc 1978;53:276–280.[Medline]
6. Benator D, Bhattacharya M, Bozeman L, Burman W, Cantazaro A, Chaisson R, Gordin F, Horsburgh CR, Horton J, Khan A, et al. Tuberculosis Trials Consortium. Rifapentine and isoniazid once a week versus rifampicin and isoniazid twice a week for treatment of drug-susceptible pulmonary tuberculosis in HIV-negative patients: a randomized clinical trial. Lancet 2002;360:528–534.[CrossRef][Medline]
7. Mitchison DA. Assessment of new sterilizing drugs for treating pulmonary tuberculosis by culture at 2 months. Am Rev Respir Dis 1993;147:1062–1063.[Medline]
8. Zierski M, Bek E, Long MW, Snider DE Jr. Short-course (6-month) cooperative tuberculosis study in Poland: results 30 months after completion of treatment. Am Rev Respir Dis 1981;124:249–251.[Medline]
9. A comparative study of daily followed by twice- or once-weekly regimens of ethambutol and rifampicin in the retreatment of patients with pulmonary tuberculosis: second report. Tubercle 1976;57:105–113.[CrossRef][Medline]
10. Third East African/British Medical Research Councils study. Controlled clinical trial of four short-course regimens of chemotherapy for two durations in the treatment of pulmonary tuberculosis: second report. Tubercle 1980;61:59–69.[CrossRef][Medline]
11. Hong Kong Chest Service/Tuberculosis Research Centre. Madras/British Medical Research Council. A controlled clinical comparison of 6 and 8 months of antituberculosis chemotherapy in the treatment of patients with silicotuberculosis in Hong Kong. Am Rev Respir Dis 1991;143:262–267.[Medline]
12. Johnson JL, Okwera A, Vjecha MJ, Byekwaso F, Nakibali J, Nyole S, Milberg J, Aisu T, Whalen CC, Mugerwa RD, et al. Risk factors for relapse in human immunodeficiency virus type 1 infected adults with pulmonary tuberculosis. Int J Tuberc Lung Dis 1997;1:446–453.[Medline]
13. Perriens JH, St Louis ME, Mukadi YB, Brown C, Prignot J, Pouthier F, Portaels F, Willame JC, Mandala JK, Kaboto M, et al. Pulmonary tuberculosis in HIV-infected patients in Zaire: a controlled trial of treatment for either 6 or 12 months. N Engl J Med 1995;332:779–784.[Abstract/Free Full Text]
14. Bock NN, Sterling TR, Hamilton CD, Pachucki C, Wang YC, Conwell DS, Mosher A, Samuels M, Vernon A. The Tuberculosis Trials Consortium, Centers for Disease Control and Prevention, Atlanta, Georgia. A prospective, randomized, double-blind study of the tolerability of rifapentine 600, 900, and 1200 mg plus isoniazid in the continuation phase of tuberculosis treatment. Am J Respir Crit Care Med 2002;165:1526–1530.[Abstract/Free Full Text]
15. Horowitz RI. The experimental paradigm and observational studies of cause-effect relationships in clinical medicine. J Chronic Dis 1987;40:91–99.[CrossRef][Medline]
16. Campbell MJ. Statistics at square two: understanding modern statistical applications in medicine. London: BMJ Books; 2001. pp. 51–55.
17. Cohn DL, Catlin BJ, Peterson KL, Judson FN, Sbarbaro JA. A 62-dose, 6-month therapy for pulmonary and extrapulmonary tuberculosis: a twice-weekly, directly observed, and cost-effective regimen. Ann Intern Med 1990;112:407–415.
18. Chan CY, Au-Yeang C, Yew WW, Hui M, Cheng AF. Postantibiotic effects of antituberculosis agents alone and in combination. Antimicrob Agents Chemother 2001;45:3631–3634.[Abstract/Free Full Text]
19. Mitchison DA. Mechanisms of the action of drugs in the short-course chemotherapy. Bull Int Union Tuberc 1985;60:36–40.[Medline]
20. Hong Kong Tuberculosis and Chest Service. Annual Report of the Tuberculosis and Chest Service for 2000. Hong Kong: Department of Health; 2001.
21. van Loenhout-Rooyackers JH, Laheij RJ, Richter C, Verbeek AL. Shortening the duration of treatment for cervical tuberculous lymphadenitis. Eur Respir J 2000;15:192–195.[Abstract]
22. Geldmacher H, Taube C, Kroeger C, Magnussen H, Kirsten DK. Assessment of lymph node tuberculosis in northern Germany: a clinical review. Chest 2002;121:1177–1182.[Abstract/Free Full Text]
23. Greenland S, Morgenstern H. Matching and efficiency in cohort studies. Am J Epidemiol 1990;131:151–159.[Abstract/Free Full Text]
24. Crofton J, Horne N, Miller F. Clinical tuberculosis. London and Basingstoke: Macmillan Education Ltd; 1992.

This article has been cited by other articles:

				E. S. Guy and A. Mallampalli Review: Managing TB in the 21st century: existing and novel drug therapies Therapeutic Advances in Respiratory Disease, December 1, 2008; 2(6): 401 - 408. [Abstract] [PDF]

				I. M. Rosenthal, M. Zhang, D. Almeida, J. H. Grosset, and E. L. Nuermberger Isoniazid or Moxifloxacin in Rifapentine-based Regimens for Experimental Tuberculosis? Am. J. Respir. Crit. Care Med., November 1, 2008; 178(9): 989 - 993. [Abstract] [Full Text] [PDF]

				J. F. Djoba Siawaya, N. B. Bapela, K. Ronacher, N. Beyers, P. van Helden, and G. Walzl Differential Expression of Interleukin-4 (IL-4) and IL-4{delta}2 mRNA, but Not Transforming Growth Factor Beta (TGF-{beta}), TGF-{beta}RII, Foxp3, Gamma Interferon, T-bet, or GATA-3 mRNA, in Patients with Fast and Slow Responses to Antituberculosis Treatment Clin. Vaccine Immunol., August 1, 2008; 15(8): 1165 - 1170. [Abstract] [Full Text] [PDF]

				H. S Cox, M. Morrow, and P. W Deutschmann Long term efficacy of DOTS regimens for tuberculosis: systematic review BMJ, March 1, 2008; 336(7642): 484 - 487. [Abstract] [Full Text] [PDF]

				P. Nahid, L. C. Gonzalez, I. Rudoy, B. C. de Jong, A. Unger, L. M. Kawamura, D. H. Osmond, P. C. Hopewell, and C. L. Daley Treatment Outcomes of Patients with HIV and Tuberculosis Am. J. Respir. Crit. Care Med., June 1, 2007; 175(11): 1199 - 1206. [Abstract] [Full Text] [PDF]

				W. W. Yew and C. C. Leung Update in Tuberculosis 2006 Am. J. Respir. Crit. Care Med., March 15, 2007; 175(6): 541 - 546. [Full Text] [PDF]

				K. C. Chang, C. C. Leung, W. W. Yew, and C. M. Tam Standard anti-tuberculosis treatment and hepatotoxicity: do dosing schedules matter? Eur. Respir. J., February 1, 2007; 29(2): 347 - 351. [Abstract] [Full Text] [PDF]

				K. C. Chang, C. C. Leung, W. W. Yew, S. L. Chan, and C. M. Tam Dosing Schedules of 6-Month Regimens and Relapse for Pulmonary Tuberculosis Am. J. Respir. Crit. Care Med., November 15, 2006; 174(10): 1153 - 1158. [Abstract] [Full Text] [PDF]

				W. J. Burman, S. Goldberg, J. L. Johnson, G. Muzanye, M. Engle, A. W. Mosher, S. Choudhri, C. L. Daley, S. S. Munsiff, Z. Zhao, et al. Moxifloxacin versus Ethambutol in the First 2 Months of Treatment for Pulmonary Tuberculosis Am. J. Respir. Crit. Care Med., August 1, 2006; 174(3): 331 - 338. [Abstract] [Full Text] [PDF]

				I. M. Rosenthal, K. Williams, S. Tyagi, C. A. Peloquin, A. A. Vernon, W. R. Bishai, J. H. Grosset, and E. L. Nuermberger Potent Twice-Weekly Rifapentine-containing Regimens in Murine Tuberculosis Am. J. Respir. Crit. Care Med., July 1, 2006; 174(1): 94 - 101. [Abstract] [Full Text] [PDF]

				P. Nahid, M. Pai, and P. C. Hopewell Advances in the diagnosis and treatment of tuberculosis. Proceedings of the ATS, January 1, 2006; 3(1): 103 - 110. [Abstract] [Full Text] [PDF]

				I. M. Rosenthal, K. Williams, S. Tyagi, A. A. Vernon, C. A. Peloquin, W. R. Bishai, J. H. Grosset, and E. L. Nuermberger Weekly Moxifloxacin and Rifapentine Is More Active Than the Denver Regimen in Murine Tuberculosis Am. J. Respir. Crit. Care Med., December 1, 2005; 172(11): 1457 - 1462. [Abstract] [Full Text] [PDF]

				B. Nemery, W. W. Yew, R. Albert, C. Brun-Buisson, W. MacNee, F. J. Martinez, D. C. Angus, and E. Abraham Tuberculosis, Nontuberculous Lung Infection, Pleural Disorders, Pulmonary Function, Respiratory Muscles, Occupational Lung Disease, Pulmonary Infections, and Social Issues in AJRCCM in 2004 Am. J. Respir. Crit. Care Med., March 15, 2005; 171(6): 554 - 562. [Full Text] [PDF]

				A. A. Vernon and M. F. Iademarco In the Treatment of Tuberculosis, You Get What You Pay for... Am. J. Respir. Crit. Care Med., November 15, 2004; 170(10): 1040 - 1042. [Full Text] [PDF]

This Article Abstract Full Text (PDF) Online Supplement All Versions of this Article: 200407-905OCv1 170/10/1124    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Chang, K. C. Articles by Tam, C. M. Search for Related Content PubMed PubMed Citation Articles by Chang, K. C. Articles by Tam, C. M.