Articles

Two 8-month regimens of chemotherapy for treatment of

Lancet 2004; 364: 1244–51
International Union Against
Tuberculosis and Lung Disease,
68 Boulevard Saint-Michel,
75006 Paris, France
(A Jindani FRCP,
Prof D A Enarson FRCP); and
Medical Research Council
Clinical Trials Unit, London, UK
(Prof A J Nunn MSc)
Correspondence to:
Prof D A Enarson
denarson@iuatld.org
newly diagnosed pulmonary tuberculosis: international
multicentre randomised trial
A Jindani, A J Nunn, D A Enarson
Summary
Background A WHO-recommended 8-month regimen based on ethambutol and isoniazid was evaluated in a
randomised clinical trial against a 6-month standard regimen.
Methods 1355 patients with newly diagnosed smear-positive pulmonary tuberculosis were randomly assigned one of
three regimens: daily ethambutol, isoniazid, rifampicin, and pyrazinamide for 2 months, followed by ethambutol
and isoniazid for 6 months (2EHRZ/6HE); the same drugs but given three times weekly in the initial intensive phase
(2[EHRZ]3/6HE); or the same initial intensive phase as the first regimen, followed by 4 months of daily rifampicin
and isoniazid (2EHRZ/4HR). Follow-up was to 30 months after the start of chemotherapy. Sputum was regularly
examined by microscopy and culture. Unfavourable outcome was defined as failure during treatment or relapse
afterwards. Analyses were by intention to treat.

Findings At 2 months, a significantly higher proportion of patients assigned the daily intensive phase than of those
assigned the three-times-weekly regimen were culture negative (700/828 [85%] vs 333/433 [77%], p=0·001).
12 months after the end of chemotherapy, the proportions of unfavourable outcomes were 36 of 346 (10%) with
2EHRZ/6HE, 48 of 351 (14%) with 2(EHRZ)3/6HE, and 17 of 347 (5%) with 2EHRZ/4HR. Both 8-month regimens
were significantly inferior to the control 6-month standard regimen (difference between control and 2EHRZ/6HE
5·5% [95% CI 1·6 to 9·4]; between control and 2(EHRZ)3/6HE 8·8% [4·5 to 13·0]). Adverse effects leading to
interruption of treatment for 7 days or longer occurred in 28 patients (12 2EHRZ/6HE, five 2[EHRZ]3/6HE, 11
2EHRZ/4HR).

Interpretation The results of this study must be taken into account in recommendations on management of new
cases of smear-positive tuberculosis.

Introduction for 2 months followed by 6 months of thioacetazone and

Short-course regimens of chemotherapy for pulmonary
tuberculosis lasting either 6 months or 8 months are
now used in almost all national tuberculosis
programmes. These regimens are all based on an initial
intensive phase, in which four drugs are given for at
least 2 months, and a continuation phase, with two
drugs given for 4 months or 6 months.
The currently used initial intensive phase, which in all
cases includes isoniazid, rifampicin, and pyrazinamide,
is based on an extensive series of clinical trials and
the postulates of Mitchison and Dickinson of different
bacterial populations, each with differing susceptibility
to rifampicin or pyrazinamide, depending on their
metabolic state.1 The fourth drug can be either
streptomycin or ethambutol. The continuation phase is
either 4 months of rifampicin and isoniazid or 6 months
of ethambutol and isoniazid or, less commonly,
6 months of thioacetazone and isoniazid.
The DOTS strategy promoted by WHO was developed
by the International Union Against Tuberculosis and
Lung Disease (IUATLD) together with national
tuberculosis programmes. It used a daily regimen of
streptomycin, isoniazid, rifampicin, and pyrazinamide
isoniazid for patients with newly diagnosed smear-
positive disease. This regimen was chosen to reduce the
probability of promoting additional drug resistance in
any patient with organisms already resistant to either
isoniazid or rifampicin.
During the past decade, two changes have led to a
reappraisal of the recommended regimens. First, the
emergence of the epidemic of HIV infection and AIDS
contributed to a massive increase in the number of
cases of tuberculosis worldwide. Second, the emergence
of rifampicin-resistant organisms has been noted
worldwide.2 To prevent the acquisition of rifampicin
resistance, both the IUATLD and WHO recommend
that regimens including rifampicin should be prescribed
only if they can be administered directly so that the
drugs are observed to have been swallowed.3,4
In addition, HIV-infected patients are at increased risk
of severe, in some cases fatal, dermatological toxic
effects from thioacetazone.5 The danger posed by
injections in transmission of HIV has also been
recognised. These factors have resulted in a change in
the WHO recommendations; streptomycin has been
replaced with ethambutol in the initial intensive phase

1244 www.thelancet.com Vol 364 October 2, 2004

For personal use. Only reproduce with permission from Elsevier Ltd

of regimens for patients with newly diagnosed smear-
positive disease, and thioacetazone has been replaced
with ethambutol in the continuation phase of the
8-month regimen.6 However, the 8-month ethambutol
regimen had never been tested against the daily
rifampicin-based 6-month regimen, which is known to
be highly effective with relapse rates of 5% or less in
many clinical trials.7
The objectives of this randomised controlled trial were
to assess, in patients with newly diagnosed smear-positive
pulmonary tuberculosis, whether the bacteriological
results obtained with the two 8-month regimens of
chemotherapy based on ethambutol and isoniazid are
equivalent to those of the 6-month regimen based on
isoniazid and rifampicin.
Methods
Design and procedures
This international, multicentre trial was carried out
within the framework of the Clinical Trials Division
of the IUATLD. It was implemented jointly with the
national tuberculosis control programmes of the
participating countries, in centres where there was access
to a reliable bacteriology laboratory to carry out both
smear and culture examinations. Drug-susceptibility
tests to isoniazid and rifampicin were done in most of the
participants.
Patients with newly diagnosed, smear-positive
pulmonary tuberculosis were randomly assigned one
of three regimens: daily ethambutol, isoniazid,
rifampicin, and pyrazinamide for 2 months, followed by
daily ethambutol and isoniazid for a further 6 months
(2EHRZ/6HE); ethambutol, isoniazid, rifampicin, and
pyrazinamide three times weekly for 2 months followed
by daily ethambutol and isoniazid for 6 months
(2[EHRZ]3/6HE); or a control regimen consisting of daily
ethambutol, isoniazid, rifampicin, and pyrazinamide for
2 months followed by daily rifampicin and isoniazid
for 4 months (2EHRZ/4HR). The doses were according
to WHO and IUATLD recommendations (table 1). When
patients had completed the 2-month initial intensive
phase of treatment, they started the continuation phase,
irrespective of the results of bacteriological examinations.
There was no attempt to conceal treatment allocation
after randomisation from patients, researchers, or health-
care staff.
The randomised allocation sequence was generated by
computer operated by an independent person based at
the IUATLD. Participating centres were supplied with a
batch of sealed and serially numbered opaque envelopes
each containing the treatment card of the allocated
regimen. These were regularly checked during site visits
to ensure that they had not been tampered with. The
names of eligible patients were entered sequentially into
a register to determine the study number allocated and
the allocation envelope to be opened. Patients were
admitted to hospital or attended the treatment facility
www.thelancet.com Vol 364 October 2, 2004
For personal use. Only reproduce with permission from Elsevier Ltd
Articles
Drug Number of tablets
25–39 kg 40–55 kg >55 kg
Daily intensive phase, 2EHRZ
Rifampicin (150 mg) and isoniazid (100 mg) combined tablet 2 3 4
Ethambutol (400 mg) 1·5 2 3
Pyrazinamide (400 mg) 2 3 4
Intermittent intensive phase, 2(EHRZ)3
Rifampicin (150 mg) and isoniazid (100 mg) combined tablet 2 3 4
Isoniazid (100 mg) 1 1 2
Ethambutol (400 mg) 1·5 2 3
Pyrazinamide (400 mg) 2 3 4
Continuation phase, HE*
Ethambutol (400 mg) and isoniazid (150 mg) combined tablet 1·5 2 3
Continuation phase, HR*
Rifampicin (150 mg) and isoniazid (100 mg) combined tablet 2 3 4
*Based on patient’s bodyweight at start of continuation phase.
Table 1: Doses of drugs
daily so that ingestion of the drugs could be directly
observed for the first 2 months. Thereafter, the patient
was given a month’s supply of the drugs to be taken
under the supervision of a relative or other person who
was designated the treatment monitor and who had
agreed to undertake this function. The drugs were
supplied, in bulk, by the Clinical Trials Division.
The study protocol was reviewed and approved by the
Ethics Review Committee of the IUATLD as well as the
medical ethics committee, or an equivalent body, of
each of the participating countries before the start of
the study.
Data forms were sent to the Clinical Trials Division
office in Paris, where data were entered by use of EpiInfo
6.04d. Data from the two Nepalese centres were entered
in Nepal. Data files were transmitted electronically from
Paris and Nepal to the Medical Research Council
Clinical Trials Unit in London for analysis.
Eligibility criteria
Criteria for entry to the study were: age 15–65 years;
two sputum samples positive for tubercle bacilli on
direct smear microscopy; less than a month of previous
antituberculous chemotherapy; and a firm home address
readily accessible for visiting in case of failure to attend.
Each patient provided informed consent to participate in
the study and to undergo HIV testing.
Patients were not eligible if they were so ill they
were thought unlikely to survive the initial weeks of
treatment. Other exclusion criteria were: extra-
pulmonary tuberculosis; other diseases likely to
prejudice the response to, or assessment of, treatment
(eg, diabetes, liver disease, nephritis, blood disorders,
epilepsy, peripheral neuritis); pregnancy; and psychiatric
illness or alcoholism.
Follow-up
Two sputum samples were collected for smear and
culture examination before treatment, at 2 months, and at
the end of treatment. A smear examination only was
1245
 Articles

attendances at months 3, 6, and 12 after the end of

treatment were not introduced until February, 2000.
1355 randomised
Whenever possible, one pretreatment culture and one
culture obtained at failure or relapse were stored at –20°C.
After counselling, patients were tested for HIV by local
456 assigned 466 assigned 433 assigned procedures. The result was communicated to the patient
2EHRZ/6HE 2(EHRZ)3/6HE 2EHRZ/4HR
at his or her request. Post-test counselling was given in
these instances, and HIV-infected patients were
54 lost to 56 lost to 50 lost to provided with clinical care within the health services
follow-up follow-up follow-up where they were being treated.
14 died 4 died 8 died
28 not seen 30 not seen 17 not seen The Data and Safety Monitoring Committee reviewed
12 no result 22 no result 25 no result the study data in April, 2001, October, 2001, and April,
2002. On each occasion the committee recommended
402 in end-of- 410 in end-of- 383 in end-of- that the follow-up should continue.
treatment treatment treatment
analysis analysis analysis
Statistical methods
The study was designed as an equivalence trial on the
12 not available 17 not available 18 not available assumption that the control 2EHRZ/4HR regimen
at end of at end of at end of would have a combined failure/relapse rate of no more
treatment treatment treatment
but assessed but assessed but assessed than 5%. The study was designed to test the null
later later later hypothesis that either of the two 8-month regimens was
at least 5% inferior to the control regimen, or that both
68 lost to 76 lost to 54 lost to were. Two regimens would be considered equivalent if
follow-up follow-up follow-up the 95% CI for the treatment difference fell wholly
6 died 7 died 3 died
44 not seen 49 not seen 44 not seen within the interval ±5%. Since we were not trying to
18 no result 20 no result 7 no result prove that the 8-month regimens were better than the
control regimen, a one-sided test was appropriate.8
346 in follow-up 351 in follow-up 347 in follow-up Enrolment of 326 patients to each regimen would have
analysis analysis analysis been sufficient to demonstrate equivalence with
90% power and 5% significance; to allow for a 20% loss
to follow-up this number was increased to 408. To
Figure 1: Trial profile increase the precision of the estimates of outcomes, we
decided to increase the total size of the trial further, to
carried out at month 5. Two sputum samples for culture 1500 patients. However, after 1355 patients had been
were collected at 3 months, 6 months, and 12 months entered the Data and Safety Monitoring Committee
after the scheduled end of treatment and at 24 months recommended that enrolment should be closed, mainly
and 30 months after the start of chemotherapy; the because most centres had already achieved their
enrolment targets.
2EHRZ/6HE (n=456) 2(EHRZ)3/6HE (n=466) 2EHRZ/4HR (n=433) The outcome of treatment was deemed favourable if a
sputum culture was negative at or after the time point of
Demography and anthropometry
Male 305 (67%) 302 (65%) 282 (65%) the assessment provided the patient had not already
Female 151 (33%) 164 (35%) 151 (35%) been classified as having an unfavourable response. The
Mean (SD) age, years 31·4 (11·7) 29·9 (10·8) 30·5 (11·1) outcome was deemed doubtful if fewer than 20 colonies
Mean (SD) weight, kg 51·0 (9·5) 49·7 (9·0) 50·4 (8·8)
were present on the sputum culture. Unfavourable
HIV status
Positive 45 (10%) 45 (10%) 37 (9%) outcome was defined as failure or relapse. Failure was
Negative 294 (64%) 308 (66%) 290 (67%) defined as a culture of 20 or more colonies at month 6 or
Not available 117 (26%) 113 (24%) 106 (24%) 8, or a change of treatment by the local investigator
Pretreatment culture
Confirmed positive 388 (85%) 408 (88%) 379 (88%)
owing to treatment failure. Relapse was defined as a
Negative or unconfirmed 68 (15%) 58 (12%) 54 (12%) culture of 20 or more colonies at any point after the end
Sensitivity test of treatment or, in the absence of culture confirmation,
Total tested 277 (61%) 324 (70%) 279 (64%) initiation by the local investigator of treatment for
Fully sensitive 245 293 241
Resistant to H alone 23 24 29
relapse. Patients were not classified as having failure or
Resistant to R alone 1 1 3 relapse if a culture of 20–100 colonies was followed by
Resistant to HR 8 6 6 negative cultures and the patient had not been re-treated.
H=isoniazid, R=rifampicin, E=ethambutol, Z=pyrazinamide. Data are number of patients unless otherwise stated. The analyses were based on intention to treat,
including all assessable randomised patients. Two-tailed
Table 2: Pretreatment characteristics of all randomised patients
significance tests were used throughout. The primary

1246 www.thelancet.com Vol 364 October 2, 2004

For personal use. Only reproduce with permission from Elsevier Ltd
 Articles

outcomes were the proportion of patients with negative 2EHRZ/6HE 2(EHRZ)3/6HE 2EHRZ/4HR
cultures at 2 months and the proportions with negative Total tested* 424 433 404
cultures 12 months after completion of chemotherapy Culture negative 365 (86%) 333 (77%) 335 (83%)
(ie, at 18 months or 20 months after the start of Culture positive 59 (14%) 100 (23%) 69 (17%)

chemotherapy, depending on the regimen). The 2-month
culture result was chosen to compare the rates of sputum
conversion for the three-times-weekly and daily intensive
phases. Secondary outcomes were the proportion of
failures at the end of chemotherapy and the proportion
of patients with adverse events necessitating withdrawal
of their chemotherapy for 7 days or longer. Proportions
were compared and 95% CI for differences, unadjusted
odds ratios, and odds ratios adjusted for centre calculated
by use of STATA (version 8). Adjustments were also
made for HIV status, but the results were very similar to
those after adjustments for centre and are therefore not
presented. Time to an adverse outcome was analysed
with Kaplan-Meier plots and Cox’s proportional-hazards
regression analysis.
Other analyses were done on subgroups of patients
according to HIV status and bacteriological drug
susceptibility; although these analyses were not specified
in the protocol, the intention to undertake them was
implicit in the collection of these data.
The characteristics of patients by centre are given
in webtable 1 (http://image.thelancet.com/extras/04art
3075webtable1.pdf) and rates of loss to follow-up and
status by centre 12 months after the end of treatment are
given in webtable 2 (http://image.thelancet.com/extras/
04art3075webtable2.pdf).
Role of the funding sources
Substantial amounts of the drugs were donated by
Hoechst Marion Roussel, Italy; FATOL, Arzneimittel,
Germany; and BRACCO, SpA, Italy. Funding in cash or
kind was obtained from Ministère des Affaires Etrangères,
Direction du Développement et de la Coopération
Technique, France; the Norwegian Heart and Lung
*Patients seen more than 14 days before or after the 2-month point are classified as
having no result.
Table 3: Culture results at 2 months
patients were male, and the average age was 30 years.
HIV test results were available for 1019 patients; 12% of
those tested were HIV infected, all but one of them from
the African centres; the proportion of HIV-infected
participants ranged from none to 33% in the eight
centres. Culture confirmation of Mycobacterium
tuberculosis was obtained in 1175 (87%) of patients.
Susceptibility test results were available for 880 (65% of
the patients); in 76 (9%) of those tested the mycobacteria
were resistant to isoniazid alone and in 20 (2%) they
were resistant to both rifampicin and isoniazid.
The analysis of the 2-month culture result was limited
to participants for whom sputum samples were collected
within 2 weeks of the due date, 93% in each group. The
proportion with negative cultures at 2 months was
significantly lower for patients assigned the three-times-
weekly intensive phase than for those assigned the daily
intensive phase (table 3). The difference was 7·6% (95%
CI 3·0 to 12·3, p=0·001; unadjusted odds ratio 1·64
[1·23 to 2·20]; odds ratio adjusted for centre 1·80 [1·32
to 2·46]). The proportions with negative smear results at
2 months were 665 (78%) of 857 patients receiving daily
treatment and 348 (78%) of 446 patients receiving three-
times-weekly treatment.
There were 27 deaths during chemotherapy; 17 were
in HIV-infected patients who died from complications
of AIDS (ten 2EHRZ/6HE, three 2[EHRZ]3/6HE, and
2EHRZ/6HE 2(EHRZ)3/6HE 2EHRZ/4HR

Association; Norwegian Agency for Development Total assessed 402 410 383
Cooperation; US Agency for International Development; Favourable status
Culture negative 357 (89%) 374 (91%) 356 (93%)
Trustees of the Royal Free Hospital, London, UK; and the Doubtful status
Kuratorium Tuberkulose in der Welt e.V. None of these Total 26 (7%) 14 (3%) 15 (4%)
organisations had any influence on the design or 1–9 colonies 25 12 15
interpretation of the trial, writing of the report, or the 10–19 colonies 1 2 0
Unfavourable status
decision to submit it for publication. The corresponding Total 19 (5%) 22 (5%) 12 (3%)
author had full access to all the data in the study and 20 or more colonies 9* 16 7
had final responsibility for the decision to submit for Treatment changed, smear 8 6 3
confirmed
publication.
Treatment changed, not 2 0 2
confirmed
Results Not assessed†
1355 patients were enrolled from eight centres (706 Total 54 56 50
No result 12 (22%) 22 (39%) 25 (50%)
from Africa, 649 from Asia) between March, 1998,
Not seen 28 (52%) 30 (54%) 17 (34%)
and December, 2001 (figure 1). 456 were assigned the Died 14 (26%) 4 (7%) 8 (16%)
2EHRZ/6HE regimen, 466 2(EHRZ)3/6HE, and 433 *Including one patient with a heavily positive culture at 2 months who died 8 weeks
2EHRZ/4HR. The pretreatment characteristics are later. †Percentages are based on the total not assessed in each group.
given in table 2. The characteristics of the patients in
Table 4: Status at the end of treatment
the three regimens were broadly similar. Two-thirds of

www.thelancet.com Vol 364 October 2, 2004 1247

For personal use. Only reproduce with permission from Elsevier Ltd
 Articles

2EHRZ/6HE 2(EHRZ)3/6HE 2EHRZ/4HR The data on status of patients 12 months after the
Total assessed 346 351 347 end of chemotherapy given in table 5 are based on
Favourable status those assessable at 18 months for the control regimen
Total 290 (84%) 292 (83%) 316 (91%) and 20 months for the 8-month regimens. 42 patients
Culture negative 251 262 272
Later culture negative 39 30 44
were not assessed because they had died, 46 did not
Doubtful status have bacteriological results, and 223 failed to attend.
Total 20 (6%) 11 (3%) 14 (4%) Two patients who died shortly after relapsing have
1–9 colonies 19 11 13 been included as having unfavourable outcomes.
10–19 colonies 1 0 1
Unfavourable status
All the other patients who died after the end of
Total 36 (10%) 48 (14%) 17 (5%) treatment were culture negative when last seen except
20 or more colonies 21* 38 11 for one HIV-infected patient assigned 2EHRZ/6HE,
Treatment given, smear confirmed 10 8 4 who had growth of one colony at month 8 and died
Treatment given, not confirmed 5 2 2
Not assessed†
3 months later.
Total 110 115 86 The proportions with unfavourable outcomes were
No result 19 (17%) 20 (17%) 7 (8%) 10% for the 2EHRZ/6HE regimen, 14% for 2(EHRZ)3/
Not seen 71 (65%) 84 (73%) 68 (79%)
6HE, and 5% for 2EHRZ/4HR. Of the 101 patients with
Died 20 (18%) 11 (10%) 11 (13%)
unfavourable outcomes, 17 of 36 in the 2EHRZ/6HE
The table shows results at 18 months for the 6-month regimen and 20 months for the
8-month regimens unless the patient was already classified as having had treatment
group, 26 of 48 in the 2(EHRZ)3/6HE group, and five of
failure or relapse. *Including one patient with a heavily positive culture at 2 months who 17 in the 2EHRZ/4HR group were classified as relapses.
died 8 weeks later. †Percentages are based on the total not assessed in each group. The differences in frequency of unfavourable outcomes
Table 5: Status 12 months after the scheduled end of treatment between the regimens and the relevant odds ratios are
given in table 6.
An analysis comparing the results of the culture at
four 2EHRZ/4HR). One patient with multidrug- 2 months with the status 12 months after the end of
resistant disease who was assigned 2EHRZ/6HE had a treatment showed a strong and consistent relation
heavily positive culture at 2 months; he died 8 weeks between a positive culture at 2 months and an
later and has been classified as a treatment failure. The unfavourable outcome (Mantel-Haenszel estimate of odds
remaining ten patients (five 2EHRZ/6HE, one ratio stratified by regimen 5·61 [3·34 to 9·43]; p=0·95, test
2[EHRZ]3/6HE, and four 2EHRZ/4HR) were all HIV for homogeneity).
negative. There was no evidence that active If the analysis was limited to patients who were
tuberculosis contributed to their deaths; seven were known to have mycobacterial strains sensitive to
culture negative when last seen, and the 2-month both isoniazid and rifampicin before treatment,
culture of the sample from another grew fewer than the numbers with unfavourable outcomes were 18
five colonies. Of the remaining two (both (9%) of 194 assigned 2EHRZ/6HE, 23 (10%) of 226
2EHRZ/6HE), one died of an infection 7 days after assigned 2(EHRZ) 3/6HE, and seven (4%) of 190
starting antituberculosis treatment, the other defaulted assigned 2EHRZ/4HR. The difference between the
very early in treatment and was reported dead ethambutol regimens was 0·9% (–4·8 to 6·6); the
7 months later. odds ratio adjusted for centre was 1·17 (0·60 to
At the end of chemotherapy, three patients each 2·28). The odds ratio for the comparison between
with one culture of 10–19 colonies were classified as the two 8-month regimens combined and the
having doubtfully favourable status because the second control regimen was 2·86 (1·23 to 6·64) after
culture was negative. The proportions of patients adjustment for centre.
with unfavourable outcomes were 5% of 402 assigned Among patients with single drug resistance to isoniazid
2EHRZ/6HE, 5% of 410 assigned 2(EHRZ)3/6HE, and at trial entry, one (4%) of 23 patients in the control group
3% of 383 assigned 2EHRZ/4HR (table 4). Only four compared with five (38%) of 13 in the 2EHRZ/6HE
(two 2EHRZ/6HE, two 2EHRZ/4HR) of the 49 patients group and six (27%) of 22 in the 2(EHRZ)3/6HE group
who had their treatment changed did not have had an unfavourable response (p=0·02, Fisher’s exact test
bacteriological confirmation of failure by culture or for comparison of the two 8-month regimens combined
direct smear. and the control regimen).

Comparison Difference in proportion, % p Unadjusted odds ratio Odds ratio adjusted for
(95% CI) (95% CI) centre (95% CI)

2EHRZ/6HE vs 2EHRZ/4HR 5·5 (1·6 to 9·4) 0·006 2·25 (1·25 to 4·07) 2·25 (1·22 to 4·15)
2(EHRZ)3/6HE vs 2EHRZ/4HR 8·8 (4·5 to 13·0) 0·0001 3·08 (1·74 to 5·43) 3·11 (1·74 to 5·58)
Two 8-month regimens combined vs 2EHRZ/4HR 7·2 (3·8 to 10·5) 0·0002 2·66 (1·56 to 4·53) 2·71 (1·57 to4·67)
2(EHRZ)/6HE vs 2(EHRZ)3/6HE 3·3 (–1·6 to 8·1) 0·18 1·36 (0·86 to 2·16) 1·44 (0·89 to 2·32)

Table 6: Difference between regimens in proportions with an unfavourable status 12 months after the end of treatment

1248 www.thelancet.com Vol 364 October 2, 2004

For personal use. Only reproduce with permission from Elsevier Ltd
 Articles

Of the patients with multidrug-resistant disease

assessable 12 months after the end of chemotherapy, 15 2(EHRZ)3/6HE

Proportion with unfavourable outcome (%)

three of five assigned the control regimen and ten of 12
assigned the 8-month regimens had unfavourable
outcomes. All five patients (one 2EHRZ/6HE, one
2(EHRZ)3/6HE, and three 2EHRZ/4HR) with single
rifampicin resistance had favourable outcomes.
A total of 68 African patients who were HIV infected
were available for assessment 12 months after the end of
chemotherapy; one (5%) of 19 assigned the 6-month
control regimen had an unfavourable outcome
compared with 13 (27%) of 49 assigned the 8-month
regimens (p=0·09, Fisher’s exact test). The proportions
of HIV-negative African patients with unfavourable
outcomes were three (2%) of 138 assigned the control
regimen and 25 (9%) of 275 assigned the 8-month
regimens (p=0·008).
A total of 579 patients with favourable or doubtfully
favourable response 12 months after the end of treatment
were assessable 30 months after randomisation. 13 had
had relapses (seven [4%] of 174 assigned 2EHRZ/6HE,
seven [4%] of 196 assigned 2[EHRZ]3/6HE, and one of 209
assigned 2EHRZ/4HR).
The results of the analysis of time to an unfavourable
outcome are plotted on a Kaplan-Meier graph in figure 2.
Of particular note is the difference in long-term trend
between the 8-month and 6-month regimens. With the
8-month regimens, there was continuing accumulation
of unfavourable outcomes. With the 6-month regimen,
almost all of these outcomes occurred during the first
18 months. The hazard-ratio estimates from the Cox’s
regression analysis are 1·26 (SE 0·26) for the comparison
of the two 8-month regimens, 2·09 (SE 0·58) for the
comparison of the 2EHRZ/6HE and 2EHRZ/4HR
regimens, and 2·78 (SE 0·74) for the comparison of
2(EHRZ)3/6HE and 2EHRZ/4HR regimens.
28 patients experienced adverse events that led to a
change of treatment or an interruption of treatment of
7 days or longer (table 7). The most frequent adverse
event was jaundice. Loss of visual acuity led to the
termination of ethambutol from the allocated regimens
in four patients. Symptoms and signs subsided on the
interruption of treatment. There were no deaths
attributable to adverse events.
Discussion
The main objective of this trial was to find out whether
an 8-month regimen of chemotherapy including a
continuation phase of isoniazid and ethambutol is
equivalent in efficacy to a 6-month control regimen with
isoniazid and rifampicin in the continuation phase, for
previously untreated pulmonary tuberculosis. The
second aim was to investigate the possibility of reducing
the burden of direct observation in the initial intensive
phase by giving treatment three times weekly rather
than daily without reducing the 2-month bacterial
conversion rate. This intermittent intensive phase was
10
2EHRZ/6HE
2EHRZ/4HR
5
0
0 0·5 1·0 1·5 2·0 2·5
Number at risk
Time from start of treatment (years)
2EHRZ/6HE 456 348 227
2(EHRZ)3/6HE 466 366 231
433 350 248
2EHRZ/4HR
Figure 2: Time to unfavourable outcome by regimen
chosen because there was good reason to believe that
it would be highly effective. In a study in Hong Kong,
the same intermittent initial intensive phase resulted
in culture negativity rates of 90% at 2 months in
164 patients with fully susceptible organisms.9 We
expected, therefore, that such an intensive phase
followed by an effective two-drug combination could
result in a cure rate of 95% or more. Neither of these
expectations was fulfilled in our trial.
This study was specifically designed to evaluate
treatment recommendations proposed by WHO for
routine use under programme conditions. The study was
meant to provide one piece of evidence (a comparison
of effectiveness) necessary for consideration of policy
recommendation. Although such a comparison is a vital
component in policy formulation, it is only one of several
factors that must be taken into account. The conditions
of a clinical trial are not those of routine practice.
For example, the study protocol, in an attempt to
compare the direct effect of the medications on the
efficacy of treatment, did not include the extension
of the initial intensive phase of treatment for patients
whose sputum continued to be positive on smear, as is
the routine recommendation in practice. This
discrepancy would be expected to have more effect on
the 8-month than on the 6-month treatment regimens.
Side-effect 2EHRZ/6HE 2(EHRZ)3/6HE 2EHRZ/4HR
Hepatic 7 1 2
Gastric 2 0 5
Cutaneous 1 2 2
Neurological 0 0 1
Ocular 2 1 1
Thrombocytopenic purpura 0 1 0
Treatment changed or interrupted 12 5 11
for 7 days or longer
Table 7: Adverse events resulting in a change of treatment or
interruption of 7 days or more

www.thelancet.com Vol 364 October 2, 2004 1249

For personal use. Only reproduce with permission from Elsevier Ltd
 Articles
Moreover, this study did not attempt to address a key
argument in the selection of a treatment regimen
without rifampicin in the continuation phase for cases of
tuberculosis never previously treated—namely, the
greater safety of the latter in preventing the emergence of
drug resistance. The production of cases of tuberculosis
due to multidrug-resistant organisms is a serious
problem, no matter how rare; patients who have
unfavourable responses to treatment but whose
organisms remain susceptible to the key medications can
still be treated successfully with the recommended re-
treatment regimen, but those with multidrug-resistant
organisms cannot. To address this issue would have
necessitated a very much larger study and would not
have been feasible. Moreover, a clinical trial can never
accurately evaluate the operational characteristics of
routine services, where poor-quality services might have
the most influence in promoting drug resistance. These
issues need to be resolved in formulation of policy
choices in the treatment of tuberculosis by the com-
munities affected as well as by international technical
agencies.
There were several potential sources of bias in this
trial. Because the design was open label, physicians with
a prejudice against one or other of the regimens might
have been tempted to change treatment for failure or
relapse on less solid evidence in those regimens. There
is no evidence that this bias occurred. Losses to follow-
up could have concealed patients with unfavourable
outcomes; the rate of loss to follow-up was substantial
and varied widely across the centres, being highest in
the centres with high frequencies of HIV infection.
However, the rates were very similar across the three
regimens. Incomplete data on drug-susceptibility tests
could have led to an imbalance in the characteristics
of the regimens, which could have influenced the
differences in outcomes. We therefore separately
analysed the results for patients with susceptibility-test
data.
The thrice-weekly treatment was less effective than
daily treatment as measured by culture conversion rates
at 2 months with a suggestion of a less favourable
outcome overall, although the difference in outcome
from the 8-month daily regimen was negligible for
patients with organisms known to be susceptible at
baseline (0·9% [–4·8 to 6·6]). The 6-month 2EHRZ/
4HR regimen, with an unfavourable outcome of only 5%
in all assessable patients, was substantially superior to
the 8-month 2EHRZ/6HE regimens, with unfavourable
outcomes of 10% after the daily initial phase and 14%
after the intermittent initial phase. The superiority of
the 6-month regimen was also evident among those
patients whose organisms were initially drug sensitive or
resistant to isoniazid alone. The results in HIV-infected
patients, although not achieving statistical significance,
also suggested that the 8-month regimens were inferior
to the 6-month regimen in these patients.
1250
For personal use. Only reproduce with permission from Elsevier Ltd
The response of patients with drug-resistant
organisms before treatment is of particular interest.
With the 6-month regimen, the proportion of patients
with unfavourable outcomes was not affected by initial
isoniazid resistance, as has been found in other clinical
trials. Thus, isoniazid has no sterilising activity in a
regimen that includes rifampicin throughout. In the
8-month regimens however, initial isoniazid resistance
led to a substantially greater proportion of patients
with unfavourable outcomes (31% vs 10% in those with
sensitive organisms). These proportions are similar
to 31·3% with initially resistant and 7·7% with sensitive
organisms in treatment regimens of SHRZ/TH
(streptomycin, isoniazid, rifampicin, pyrazinamide,
thioacetazone, isoniazid),10 which shows that in such
regimens, isoniazid continues to have a sterilising role.
Thus, the greater benefits of the 6-month regimen in
patients with organisms initially resistant to isoniazid
alone, the commonest form of initial resistance,
contribute substantially to its greater overall efficacy.
Patients with multidrug-resistant strains did uniformly
badly, since neither isoniazid nor rifampicin would be
available. However, the few patients with organisms
initially resistant to rifampicin alone did well.
Why was the 8-month regimen with the EH
continuation phase much less successful than might
have been expected? First, the duration of 8 months
was based on two East African studies11,12 with a small
number of patients; most of their chemotherapy was
given under direct supervision in hospital. In the
first study, there were no relapses among 81 patients
in an 8-month 2SHRZ/TH regimen,11 and in the
second there was a relapse rate of only 3% among
123 patients receiving the 8-month 2SHRZ/H
regimen.12 In our study, streptomycin, which probably
has some sterilising activity, was replaced by
ethambutol, which seems to have none. Furthermore,
there is evidence that ethambutol antagonises the
activity of other drugs as shown in an in-vitro study on
the bactericidal action of various drug combinations.13
Regression analyses of an in-vivo early bactericidal study
showed that there is significant antagonism of sterilising
activity due to ethambutol during the first 2 weeks of
treatment.14 There is also limited evidence from clinical
trials to support this idea; in one trial, the addition of
ethambutol to an SHRZ regimen increased the relapse
rate from 1% to 4%,15 and in another, the substitution of
ethambutol for streptomycin increased the relapse rate
from 1·7 % to 3·1%.16 However, the differences were not
statistically significant in either study.
Side-effects leading to an interruption of treatment
were few with all three regimens, and there were no
unexpected side-effects. Most deaths were related to
HIV infection. The highest proportion of deaths was
in the group assigned the daily 8-month regimen, in
contrast to the findings of a study in Malawi, in which a
significantly higher mortality rate was encountered in
www.thelancet.com Vol 364 October 2, 2004

patients assigned an intermittent regimen than in those
assigned daily treatment.17
In conclusion, the results of our study, which was
conducted in centres spread over a wide geographical
area, have raised important issues about the regimens
that should be recommended for first-line treatment of
tuberculosis. They also show the importance of well-
designed clinical trials in the assessment of treatment
regimens and the hazards of making recommendations
in the absence of controlled data.
Contributors
Amina Jindani coordinated the development of the protocol,
recruitment, and conduct of the trial in the centres through regular site
visits. Andrew Nunn set up the database and carried out the analyses.
Donald Enarson was responsible for the overall supervision of the trial.
Amina Jindani, Andrew Nunn, and Donald Enarson were involved in
drafting the article.
Study centres and investigators
German/Nepal Tuberculosis Project, Kathmandu, Nepal
(Bhabhana Shrestha, Bhagwan Maharjan, Urmilla Raut,
Sanu Tandukar).
Tuberculosis/Leprosy Project, Nepalganj, Nepal (K R Sharma,
Hari Kafle, Indra Devgiri, Mangal Tharu, Jag Bahadur Chaudhury,
Chandra Prasad Lamichhane, Ratna Bhattrai).
Tianjin Tuberculosis Institute, China (Wang Jie-siu, Zhao De fu,
Li Shang lun, Wu Chun ling, Ha Xiao sheng, Wang Bing hua,
Pan Xue).
Henan Tuberculosis Centre, Zheng Zhou, China (Du Chang mei,
Liu Chuanyu, Jiao Xiaolei, Yan Guo rui).
Kai Feng Anti-TB Institute, China (Huang Yu ping, Chen Ming chao,
Song Zi wei, Zhang Chang qing, Qu Ya hong, Song Jin ying,
Du Peng cheng, Zhao Ting ting).
Ping Ding Shan Anti-TB Institute, China (Wang Xiao wei, Jia Li jia,
Zhang Guo feng, Guo Xin mei).
An Yang Anti-TB Institute, China (Zhang Dong sheng, Liu qi,
Feng Guang ming, Zhang Xue yong).
Luo Yang Anti-TB Institute, China (Gou Sheng min, Dong Shu fang,
Du Jian xun, Bai Mai luan, Li Jin long).
Programme National Contre la Tuberculose, Cotonou, Benin
(Martin Gninafon, Ferdinand Kassa, Monteiro Pio Germain,
Blanche Tanimono, Alain Fatinde, Theodora Santos, Marie Adande,
Germain Tonoudo, Nicolas Adjovi).
Lutte Nationale Contre la Tuberculose, Conakry, Guinea
(Oumou Bah-Sow, Soumah Soriba, Mamadou Dian Barry,
Coumbassa Mariama Tata Barry, Alimou Diallo, Lansana Mady Camara,
Abdourahmane Barry, Boubacar Diallo).
Kibong’oto National Tuberculosis Hospital, Sanya Juu, Tanzania
(Leonard Uiso, Godfrid Nkunda).
Kilimanjaro Christian Medical Centre, Moshi, Tanazania (Noel Sam,
Aloyce ole Solulu, Mramba Nyindo, Gamaliel Kisyombe).
National Tuberculosis Control Programme, Maputo, Mozambique
(Paula Perdigao, Armando Novela, Maungate.Salomao, Mercedes
Fernandes).
Data and Safety Monitoring Committee
Janet Darbyshire (chair), Günnar Böman, Catherine Hill, and
Anne Horgheim.
Conflict of interest statement
We declare that we have no conflict of interest.
www.thelancet.com Vol 364 October 2, 2004
For personal use. Only reproduce with permission from Elsevier Ltd
Articles
Acknowledgments
We thank Janet Darbyshire, Denis Mitchison, and Andrew Vernon for
constructive suggestions on an earlier draft.
References
1 Mitchison DA, Dickinson JM. Bactericidal mechanisms in
short-course chemotherapy. Bull Int Union Tuberc 1978; 53:
254–59.
2 WHO/International Union Against Tuberculosis Global Project on
Anti-Tuberculosis Drug Resistance Surveillance. Anti-tuberculosis
drug resistance in the world. Geneva: WHO, 1997:
WHO/TB/97.229.
3 Enarson DA, Rieder HL, Arnadottir T, Trebucq A. Management
of tuberculosis: a guide for low income countries, 5th edn. St-Just-
La-Pendue: Compogravure Impression, Broachage Imprimerie,
2000.
4 WHO. WHO report on the tuberculosis epidemic. Geneva: WHO,
1995: WHO/TB/95.
5 Nunn P, Kibuga D, Gathua S, et al. Cutaneous hypersensitivity
reactions due to thiacetazone in HIV-1 seropositive patients treated
for tuberculosis. Lancet 1991; 337: 627–30.
6 WHO. Treatment of tuberculosis: guidelines for national
programmes. Geneva: WHO, 2003: WHO/CDS/TB/2003.313.
7 British Medical Research Council Tuberculosis Units, 1946–1986.
Studies on the treatment of tuberculosis. Int J Tuberc Lung Dis
1999; 3 (suppl): S231–79.
8 Machin D, Campbell MJ, Fayers PM, Pinol APY. Statistical tables
for the design of clinical trials. Oxford: Blackwell Scientific
Publications, 1997.
9 Hong Kong Chest Service/ British Medical Research Council.
Controlled trial of 4 three-times-weekly regimens and a daily
regimen all given for 6 months for pulmonary tuberculosis: results
up to 24 months. Tubercle 1982; 63: 89–98.
10 Mitchison DA. Role of individual drugs in the chemotherapy of
tuberculosis. Int J Tuberc Lung Dis 2000; 4: 796–806
11 Third East African/British Medical Research Council Study.
Controlled trial of four short-course regimens of chemotherapy for
two durations in the treatment of pulmonary tuberculosis: second
report. Tubercle 1980; 61: 59–69.
12 East and Central African/British Medical Research Council Study.
Controlled clinical trial of 4 short-course regimens of
chemotherapy (three 6-month and one 8-month) for pulmonary
tuberculosis: final report. Tubercle 1986; 67: 5–15.
13 Dickinson JM, Aber VR, Mitchison DA. Bacterial activity of
streptomycin, isoniazid, rifampin, ethambutol and pyrazinamide
alone and in combination against Mycobacterium tuberculosis.
Am Rev Respir Dis 1977; 116: 627–35.
14 Jindani A, Doré CJ, Mitchison DA. The bactericidal and sterilising
activities of antituberculosis drugs during the first 14 days.
Am J Respir Crit Care Med 2003; 167: 1348–54.
15 Hong Kong Chest Service/British Medical Research Council. Five-
year follow-up of a clinical trial of three 6-month regimens of
chemotherapy given intermittently in the continuation phase in the
treatment of pulmonary tuberculosis. Am Rev Respir Dis 1987; 136:
1339–42.
16 British Thoracic Society. A controlled trial of 6 months’
chemotherapy in pulmonary tuberculosis: final report, results
during the 36 months after the end of chemotherapy and beyond.
Br J Dis Chest 1984; 78: 330–36.
17 Harries AD, Gausi FK, Kwanjana JH, Nyirenda TE,
Salaniponi FML. Is oral intermittent initial phase anti-tuberculosis
treatment associated with higher mortality in high HIV-prevalence
areas in sub-Saharan Africa? Int J Tuberc Lung Dis 2001; 5:
483–85.
1251