TB Treatment

Here  follows the full text of this cited article. The text may not reflect exactly the finally  published version  in the journal due to editing by the journal editor. However, the text  provided here is the final version submitted by us to the journal. Hard copies can be  sourced from Fiona Beswick at beswickf@mrc.ac.za 

Should We Take A History  Of Prior Treatment, And Check Sputum Status At 2-3  Months When Treating Patients For  Tuberculosis?
Dr David Wilkinson BSc MBChB DipPEC DCH DTMH  MSc(Epi) 1,2,3 *
Dr Sudharshanie Bechan MB ChB 1
Mrs Catherine Connolly MPH 1
Dr Elizabeth Standing MB BS MFPHM 4
Dr G Murray Short MB ChB DPH DTMH 4,5

1 Centre for Epidemiological Research in Southern Africa,  Medical Research Council, South Africa.
2 Division of Tropical Medicine, Liverpool School of  Tropical Medicine, UK.
3 Hlabisa Hospital, Hlabisa, South Africa.
4 Pinetown Municipality, Pinetown, South Africa.
5 Regional Office, State Health Department, Durban,  South Africa.

* author for correspondence

Address for correspondence : Centre for  Epidemiological Research in Southern Africa, Medical Research Council, PO Box 187,  Mtubatuba 3937, South Africa.
Tel +27 35 5500607. Fax +27 35 5501436.

Word count: 1188
Running head: Value of previous treatment history and 2  month smear
Key words: tuberculosis, treatment history, sputum smear  examination

Patients with a history of previous treatment for  tuberculosis are typically treated with different drug  regimens from new patients (1),  because they are more likely to have drug resistant disease (1, 2). It is also standard  practice to check sputum status at the end of the initial intensive phase of therapy (2   months for new patients and 3 months for retreated patients), and to extend this phase of  chemotherapy if sputum remains positive (1). A more simple approach to tuberculosis  treatment has been advocated (3). In rural South Africa, using the same drug regimen for  all adults (four drugs given twice weekly for six months) irrespective of previous  treatment history, and not doing smears after diagnosis, consistently high treatment  completion rates (4, 5) and cure rates (6) have been achieved. In an era of much  increased  tuberculosis incidence in sub-Saharan Africa (7, 8), and reduced health system capacity to   treat the resultant caseload (9), secondary in large part to the HIV epidemic (9), we used  historical data  to answer the question: is it really necessary to act on a history of  prior tuberculosis treatment and to check sputum status at 2-3 months?

Between 1975 and 1983 in Pinetown, South Africa, four large  cohort studies were conducted amongst ambulant adults with smear and culture positive  pulmonary tuberculosis, with the aim of testing the  effectiveness of four drug therapy  given twice-weekly from the start of treatment for six months, under direct observation;  85% of patients completed treatment (10). Patients were mainly male migrant  workers. Doses  varied according to the cohort (isoniazid 600-900mg, rifampicin 600mg, pyrazinamide 2-3g,  and streptomycin 1-2g), but within each cohort the same regimen was given irrespective of  treatment history. Sputum status was determined at 2-3 months by microscopy of a  Ziehl-Neelsen stained sputum smear, but the drug regimen was not changed if the smear  remained positive. Cultures were performed on Lowenstein-Jensen (LJ) slants after  digestion-decontamination with 4% NaOH (final concentration 2%) at the King George V  hospital, Durban. The trials (run by GMS and ES) were approved by the Tuberculosis  Research Institute of the South African Medical Research Council and have recently been  analysed and published in full (10).

Of 562 consecutive patients enrolled, mean (sd) age was  34.7 (10.9) years, most (81%) were men and 84% had no history of prior tuberculosis  treatment. As expected, the proportion of mycobacterial isolates resistant to at least one  drug was higher in those with a history of prior treatment (18.2% vs 5.3%; table  1).  However, most patients with resistant isolates (22/34; 65%) had never been treated for  tuberculosis  before. The positive predictive value of previous treatment for tuberculosis,  for a positive sputum smear at 2-3 months, for treatment failure, and for subsequent  relapse was poor, ranging from 5.2-18.3% (Table 1).

To explore the value of examining sputum at 2-3 months data  were analysed irrespective of treatment  history (Table 2). Patients with positive smears  at 2-3 months were more likely to fail therapy (5/40) than were patients with negative  smears (9/325; RR 4.5, 95%CI: 1.6-12.8; p=0.01). However, most of those that did fail  therapy had negative smears at 2-3 months (9/14; 64%). The positive predictive value of a   positive sputum smear at 2-3 months for subsequent treatment failure was only 12.5%. Cured  patients were followed for a mean (sd) of 21 (12) months. Relapse rates were higher in  patients who had positive  smears at 2-3 months (3/31=9.7%) than in patients who had  negative smears at 2-3 months (18/289=6.2%), but this difference was not statistically  significant (p=0.4). Most patients who relapsed had been smear negative at 2-3 months  (18/21).

Despite being treated with the same drug regimen treatment  outcome in these trials did not differ significantly according to previous treatment  history. Most patients who were smear positive at 2-3  months, who failed therapy, or who  relapsed, had not been treated for tuberculosis before. Therefore there seems to be little  value in providing different drug regimens to those treated previously when four  drugs are  given twice weekly under direct observation for six months. 

Although the relative risk of failing therapy was higher in  patients with positive smears at 2-3 months, most treatment failures actually had negative  smears at 2-3 months. Also, despite therapy not being changed at 2-3 months, most patients  with positive smears at that time were cured at six months. This suggests that routine  smears at 2-3 months can safely be omitted when using this regimen. 

Developing countries face a huge and increasing burden of  tuberculosis (7). Innovative and cost-effective  ways of implementing the directly  observed, short-course (DOTS) strategy of the World Health Organisation are now needed.  Our findings suggest that when four drugs (including rifampicin) are given  throughout  treatment, the same drug regimen may be safely given to all patients irrespective of   treatment history. These findings do not imply that current international practice is  wrong, rather that it  may be possible to do away with aspects of it. Simplifying treatment  protocols is likely to ease confusion  and to aid implementation (3), and advocating for  one global 4-drug rifampicin-containing regimen for all patients with all forms of  tuberculosis is equitable and might attract considerable donor support. 

Our findings also suggest that omitting smears at 2-3  months under the conditions described is safe. The reality in some African countries is  that resources to provide a high quality smear microscopy service are severely limited  (11) and the need to do three smears for diagnosis has been questioned (3, 12). Perhaps it  would be best to focus on a microscopy service that reaches as many tuberculosis  suspects  as possible, providing each with one or two high quality smears for diagnosis and a single  high  quality smear at the end of treatment to assess cure. Smears at 2-3 months aim to  predict who will not be cured at 6 months, but our data suggest that they fail to do this  with adequate predictive value. 

In the absence of rifampicin resistance, response to  therapy in patients with initial drug resistance is known to be equivalent to that amongst  patients with fully susceptible bacilli, if at least four drugs  (including rifampicin) are  given for 6 months (13).

When an outcome is uncommon (such as treatment failure in  patients treated for tuberculosis) even if  test sensitivity and specificity is good,  positive predictive value will tend to be poor (14). The positive predictive value is  probably the most best measure in assessing the utility of any test (14). 

How generalisable are our findings? Although the trials  were done in the pre-HIV era the results can probably be applied to HIV infected patients,  as such patients typically respond to chemotherapy in the same way as HIV uninfected  patients do (1, 7), but it will be important to test this prospectively. The conclusions  may not apply to drug regimens that utilise an intensive phase of 4 drugs and a  continuation phase of 2 months (15). 

We conclude that both a previous treatment history and the  result of a smear examination at 2-3 months  have poor predictive value for the more  important outcomes of tuberculosis treatment, namely cure and relapse when 4 drugs are  given twice weekly under direct observation for 6 months. In an era of rapidly rising  tuberculosis incidence and declining capacity to respond, programmes may wish to consider  focusing on ensuring adherence to an effective drug regimen that can be given to all  patients, and to dispense with smears at 2-3 months as they contribute little to overall  treatment outcome. 


  1. Harries AD, Maher D. TB/HIV. A clinical manual.  WHO/TB/96.200. Geneva: World Health Organisation, 1996.
  2. Chan SL. Chemotherapy of tuberculosis. In: Davies PDO,  editor. Clinical tuberculosis. 1st ed. London: Chapman and Hall, 1994:141-56.
  3. De Cock KM, Wilkinson D. Tuberculosis control in  resource-poor countries: alternative strategies in the era of HIV. Lancet 1995;346:675-7.
  4. Wilkinson D. High compliance tuberculosis treatment  programme in a rural community. Lancet 1994;343:647-8.
  5. Wilkinson D, Davies GR, Connolly C. Directly observed  therapy for tuberculosis in rural South Africa, 1991 through 1994. American Journal of  Public Health 1996;86:1094-7.
  6. Wilkinson D, Anderson E, Davies GR, Sturm AW, McAdam  KPWJ. Efficacy of twice-weekly treatment for tuberculosis given under direct observation  in Africa. Transactions of the Royal Society of Tropical  Medicine and Hygiene 1997;91:  87-89.
  7. De Cock KM, Soro B, Coulibaly IM, Lucas SB. Tuberculosis  and HIV infection in sub-Saharan Africa. JAMA 1992;268:1581-7.
  8. Raviglione MC, Snider DE, Kochi A. Global epidemiology  of tuberculosis. Morbidity and mortality of a world-wide epidemic. JAMA 1995;273:220-6.
  9. Gilks CF, Haran D, Wilkinson D. Coping with the impact  of the HIV epidemic - the Hlabisa-Liverpool Link. South African Medical Journal  1996;86:1077-8. 
  10. Bechan S, Connolly C, Short GM, Standing E, Wilkinson  D. Directly observed therapy for tuberculosis given twice weekly in the workplace in urban  South Africa. Transactions of the Royal  Society of Tropical Medicine and Hygiene 1997 (in  press).
  11. Mundy CFJ, Chintolo FE, Gilks CF. The burden of sputum  smear microscopy on laboratory services in Malawi. Royal Society of Tropical Medicine and  Hygiene Third Residential Meeting, Christ's College, Cambridge 18-20 September 1996.
  12. Squire SB, Nyasulu IK, Kanyerere H, Salaniponi FML. Why  bother with three sputum smears for case-finding in tuberculosis control. Royal Society of  Tropical Medicine and Hygiene Third Residential  Meeting, Christ's College, Cambridge 18-20  September 1996.
  13. Mitchison DA, Nunn AJ. Influence of initial drug  resistance on the response to short-course chemotherapy of pulmonary tuberculosis.  American Review of Respiratory Diseases 1986;133:865-6.
  14. Hennekens CH, Buring JE. Epidemiology in Medicine. 1st  ed. Boston/Toronto: Little, Brown, 1987.
  15. Reider, HL. Sputum smear conversion during directly  observed treatment for tuberculosis. Tubercle and Lung Disease 1996; 77: 101-196. 

Table 1. Relationship  between previous treatment history for tuberculosis and initial drug  resistance rates and  treatment outcomes.History of  prior tuberculosis treatment






PPV  (%)

NPV  (%)


n (%) 

n (%) 

(95% CI) 




Isolates resistant to at least one drug

12/66 (18.2 

(5.3) * 





Positive sputum at 2-3 months 2







Failure 3 

3/58 (5.2)






Relapse 4 

5/53 (9.4)

17/293 (5.8 





  * p=0.0008
1. 482 patients had data on prior treatment history and drug susceptibility patterns
2. 432 patients were compliant to 2-3 months and produced a  specimen for examination
3. 396 patients completed treatment and produced a specimen  for examination
4. 346 cured patients were successfully followed up

 The proportion of patients not evaluated for each outcome  was similar in the two groups. 

Table 2. Relationship  between sputum smear status at 2-3 months and treatment outcomes.Sputum smear status at 2-3 months




Sensitivit y

Specificit y

PPV  (%)

NPV  (%)


n  (%) 

n  (%) 

(95%  CI) 





5/40 (12.5)

9/325 (2.8)






3/31 (9.7) 

18/289 (6.2





Data was available on sputum smear status at 2-3 months for  432 patients; data on treatment outcome was available on 365 of these, and data on relapse  was available for 320.


Dr Martie van der Walt
E-mail: vdwalt@mrc.ac.za

Dr Roxanna Rustomjee
E-mail: roxanna.rustomjee@

Prof Valerie Mizrahi
E-mail: mizrahiv@

Prof. Paul van Helden
E-mail: pvh@sun.ac.za


Last updated:

Technical enquiries:

Copyright © 1999-current
SAHealthInfo TM

To SAHealthInfo home