Isoniazid Prophylactic Therapy to Prevent TB in Children

109 42
Isoniazid Prophylactic Therapy to Prevent TB in Children


We included only randomized controlled trials (RCTs) involving children aged ≤ 15 years, regardless of their HIV status, comparing INH (4–20 mg/kg) with placebo or no prophylaxis. The studies had to have been published in English in peer-reviewed journals. Publications were of any date through October 2012. Studies were restricted to RCTs to provide the strongest evidence for the efficacy of INH.

Search Methods and Data Collection

RCTs were identified through a search of PubMed, Embase, Aidsonline, Google Scholar and Cochrane database electronically and by reviewing the references of identified articles (Figure 1). The following search terms were used: isoniazid, INH, prophylaxis, preventive therapy, children, tuberculosis, TB, HIV as listed in titles and abstracts, and randomized controlled trial as publication type. The search included all English articles through October 2012.

(Enlarge Image)

Figure 1.

Diagram of systematic search (Flow Diagram).

Two reviewers searched and evaluated trials for inclusion in the analysis, and data were collated and checked by both reviewers before inclusion. Discrepancies were resolved by consensus.

The quality of each RCT included in this analysis was graded by use of a validated score that included the following items (10): randomization of participants, double-blind evaluation, and a description of participants who withdrew from or dropped out of the trial. The scoring gives one point for each item present. If randomization is concealed and the method of double-blind evaluation is appropriate, the study is assigned one additional point for each of these, yielding a score of 0–5 points. Studies that scored three or more points were considered to be high quality studies, while those with a score below three were considered to be low quality studies. To supplement this score we used the Cochrane criteria to grade concealment of treatment allocation as: adequate (A), unclear (B), inadequate (C), or allocation was not used (D).

Statistical Analysis

Most studies presented their results using risk ratios (RRs). For those that included both adults and children, we disaggregated the raw data and calculated RRs and the corresponding 95% confidence intervals for the children separately. Analysis was performed using the intention-to-treat principle.

The summary RR estimates were calculated using both the fixed effects inverse variance weighting method and the DerSimonian-Laird random effects method. Heterogeneity among studies was assessed using the general variance-based method. If heterogeneity was present, confidence intervals in the fixed effects model were adjusted to account for between-study variance using the method presented by Shore et al.. The advantage of this approach is that it still weights by precision, while also taking heterogeneity into account.

Heterogeneity of INH efficacy between studies was investigated using the I statistic. An I value of >50% was considered an indicator of substantial heterogeneity (values of I equal to 25%, 50%, and 75% were deemed to represent low, moderate, and high heterogeneity, respectively). In addition, significant heterogeneity was deemed to be present when the Chi test statistic was greater than the degrees of freedom. To explore heterogeneity among studies, we performed subgroup analyses based on the HIV status of the children, duration of administration of INH prophylaxis (< or ≥ six months), age of participants (arbitrary cut-off set at 5 years of age), primary and secondary prophylaxis, TB endemicity where the study was conducted, and quality of the study. Primary prophylaxis here refers to administration of INH for prevention of TB infection while secondary prophylaxis refers to administration of INH to prevent TB disease. Sensitivity analyses were also carried out to assess the influence of individual studies on the summary effect estimate.

Publication bias was assessed by plotting the logarithm of the RR for each study by its standard error (SE) and explored using a funnel plot. Publication bias was also assessed using the Egger and Begg tests. An absence of publication bias would be indicated by an even dispersion of effect sizes around the pooled effect estimate. All meta-analyses risk estimates, graphs and plots were performed using STATA, version 12.1.

Subscribe to our newsletter
Sign up here to get the latest news, updates and special offers delivered directly to your inbox.
You can unsubscribe at any time

Leave A Reply

Your email address will not be published.