Search strategy: We attempted to identify all relevant studies without language restriction for the review. The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, Issue 4, 2008), Cochrane Database of Systematic Reviews were searched. We combined the MEDLINE search (December 2008) with the highly sensitive search strategy for identifying controlled trials, as designed by Dickersin, et al.(8). The search terms used were: tuberculosis (text or MESH heading), children (text or MESH heading), therapy (text or MESH heading) with limits of age: 0-16 yrs, Randomized Controlled Trial. All the citations were screened from titles. Of those considered to be relevant, abstracts were screened. After screening of abstracts, full articles were obtained if considered relevant. References of all the studies were hand searched for potential inclusion. If a full article was not available, the authors were contacted.

Inclusion criteria: Randomized controlled trials (RCTs) in children aged 16 years or below with pulmonary/extrapulmonary tuberculosis, which compared intermittent and daily regimen (containing rifampicin), in hospital or ambulatory settings and recorded the outcome of cure/significant improve-ment (symptomatic relief and/or radiologic clearing at the completion of treatment course). Studies without any separate data for children were excluded.

Data extraction: Two reviewers independently extracted data pertaining to age, sex, history of contact with tuberculosis patient, demonstration of acid fast bacilli (AFB) by microscopy or culture, cure/significant improvement (symptomatic relief and/or radiologic clearing at the completion of treatment course), completion of treatment, relapse rate, adverse effects, and death.

Quality assessment: We used the previously validated Jadad five point scale(9), to assess randomization (zero to two points), double-blinding (zero to two points), and withdrawals and dropouts (zero to one point).

Intermittent short course therapy: Any rifampicin-containing multiple drug regimen, administered twice or thrice a week for a maximum of nine months; initial daily dosing phase not exceeding one month.

Daily short course chemotherapy: Any rifampicin-containing regimen given daily throughout (or five times a week in DOTS) for a maximum of nine months.

Pulmonary TB: Children with TB involving at least lungs (include disseminated tuberculosis with pulmonary involvement).

Extra pulmonary TB: Extra pulmonary tuberculosis involving pleura, lymph nodes, abdomen, bones and joints, disseminated, intestines, larynx, CNS.

Previously untreated patients: Patients who did not receive antituberculous drugs in the past.

Smear positive: AFB demonstrated by Ziehl-Neelsen stain on gastric lavage/ other fluid by direct microscopy.

Culture positive: Mycobacterium tuberculosis identified on culture from gastric aspirate/ sputum or other body fluid.

Treatment completed: Children completing regimen for assigned period.

Interrupted treatment: Children who interrupted treatment for 2 months (8 weeks) of chemotherapy due to various reasons (other than death).

Cure or significant improvement: A child who became free of clinical symptoms and/or showed significant radiological improvement at the end of assigned regimen.

Relapse rates: A child who showed cure/ significant improvement with assigned regimen but had recurrence of symptoms during follow up of up to 2 years.

Death: A child who died during the chemotherapy.

Baseline characteristics of patients: There were no statistically significant differences in distribution by age, sex distribution, nutritional status, history of contact with an infective case of tuberculosis, BCG vaccination status, positive tuberculin test and identification of acid fast bacilli between children enrolled to receive either intermittent or daily therapy.

Type of tuberculosis: A total of 439 children (of 466) had pulmonary TB(PTB). The distribution of cases of PTB in two groups were similar (P=0.67). One study enrolled ten cases of cavitary tuberculosis(17). Only one study enrolled 27 children with lymph-node tuberculosis with similar distribution between two groups (P=0.37)(19).

Cure/significant improvement: Per protocol analysis revealed lower cure rates among children getting twice weekly regimen as compared to daily regimen (OR 0.27, 95% CI 0.15-0.51 (Fig. 2). Results of intention to treat analysis suggested that there was a trend towards lower cure rates in twice weekly intermittent therapy as compared to daily therapy (OR 0.66; 95% CI 0.23-1.84). Results of sensitivity analysis revealed that the cure rates were more in daily treatment except when study by Ramachandran, et al.(17) is excluded, but it did not reach statistical significance (Table II). In study by Ramachandran, et al.(17), when only those who did not need extended regimen were considered cured, the pooled analysis shows trend towards better cure rates in daily regimen (OR 0.53, 95% CI 0.23-1.21), though it did not reach statistical significance.

TABLE II



Results of Sensitivity Analysis

Secondary outcomes: Table III gives the details of the outcomes of included studies. In the study by Kansoy, et al.(18), three children in daily treatment group were excluded because of poor compliance. In the study by Kumar, et al.(19), 13 children were excluded: 10 children, who belonged to migrant farm laborers, dropped out of the trial after a variable period of 2 to 4 months (interrupted treatment); all of them had pulmonary TB. Three children, two of whom had died soon after completing 2 months of therapy (reason not established) and one child who was diagnosed to have Mycobacterium avium intracellulare infection were excluded from analysis. Sixty-three children completed the study and had outcome determined. In the study by Te Water Naude, et al.(20), treatment records of seven children were described as lost for assessment of 4 weekly adherence data. However, the outcomes for these children have been included for analysis and the authors have stated that "it is unlikely that cases of relapse would not have come to our notice"(20).

TABLE III



Per Protocol Analysis for Treatment Outcome

The event rates for adherence, relapse, drug related side effects and deaths were very low but there was no significant difference between the two groups for these variables (P>0.05). Only one study(19) reported data on children who interrupted treatment and there was no difference between the groups (P=0.56).

Out of four studies, three scored 2 on Jadad’s 5 point scale while one study scored 3 points suggesting that these studies were not of very good quality.

We identified and analyzed four RCTs including 466 children comparing intermittent, twice weekly therapy with daily therapy; with mainly pulmonary tuberculosis. There was no RCT comparing the efficacy of thrice weekly regimen with daily treatment in childhood tuberculosis. Analysis of the pooled data revealed that daily therapy was superior to twice weekly intermittent therapy for children.

The major problem in assessment of treatment outcome in childhood tuberculosis is difficulty in defining outcome as documentation of conversion from AFB positive to AFB negative state (that is the gold standard for tuberculosis in adults) is extremely difficult as very small proportion of patients are positive in the beginning. In such a scenario, pediatricians have to rely on the clinico-radiologic markers of improvement, this may lead to some subjectivity. We tried to define outcome in the beginning and did multiple analysis (sensitivity analysis, per protocol, intention to treat, and defining cure when the patient did not require extension of treatment) to avoid bias in the results. All analyses showed trend towards better outcome in daily treatment group as compared to twice weekly regimen.

The finding of inferiority of the intermittent regimen in our review may be due to the different regimens for treatment. Two out of the four studies had given twice weekly therapy during intensive phase which is no longer recommended(24). In all the studies, the therapy in intermittent regimen was directly observed, as recommended(25); even though there was loss to follow up. A single missed dose in an intermittent regimen represents a larger fraction of the total number of treatment dose than in a daily regimen increasing the risk of treatment failure. Adherence to therapy and hence, default rates may be influenced by other factors (like overcrowding in household, default in the first month in children with tuberculosis). One of the studies used a two drug regimen of isoniazid and rifampicin for 9 months for daily therapy(17); this will be considered as inadequate by current standards.

The main highlight of the review is that there are no RCTs in children comparing thrice weekly with daily regimen. The available studies comparing daily with twice weekly regimen lack uniformity in diagnosis and assessment of outcome. To overcome heterogeneity in the studies, we performed sensitivity analysis that revealed results in the same direction except when study by Ramchandran, et al.(17) was excluded. Even if it is presumed that in study by Ramachandran, et al.(17), even those who did not need extension of treatment were cured, the pooled analysis shows trend towards better cure rates in daily regimen. The study has some limitations, the results may not be valid for extrapulmonary tuberculosis as majority of the patient were having pulmonary TB. The data on relapse, adherence and interruption were limited to few patients, therefore, no valid conclusion can be drawn from this review for these outcomes.

Contributors: SKK planned the study, extracted data, conducted analysis and drafted manuscript and will act as the guarantor. PRM searched literature, extracted data, and drafted manuscript. RL conducted analysis and drafted manuscript. SS contributed to literature search and manuscript writing.

Funding: None.

Competing interest: None stated.