To systematically review the scientific literature on the effect of antiretroviral therapy on incident tuberculosis, in accordance with PRISMA guidelines. This will be done by comparing rates of tuberculosis by antiretroviral therapy status in adults (= 13 years) using retrospective cohort studies, prospective cohort studies, and randomised controlled trials.
Electronic searches of abstracts and titles will be conducted in PubMed and Embase. The Cochrane Central Register of Controlled Trials, the International Standard Randomised Controlled Trial Number Register, and ClinicalTrials.gov will be searched for future/ongoing studies. We also plan to contact experts in the field to identify unpublished research or ongoing studies. Since Embase searches both the Embase and PubMed databases, it will be used for the systematic searches.
Types of study to be included
Randomised trials, prospective cohort studies, and retrospective cohort studies will be used since their study arms (i.e. participants on ART and participants off ART) are from the same source population.
Condition or domain being studied
HIV-associated tuberculosis. Tuberculosis and human immunodeficiency virus (HIV) present a major combined threat to achieving the Millennium Development Goals. In 2009 there were an estimated 1.2 million incident cases of tuberculosis among the 33.3 million people living with HIV. The 400,000 deaths among incident HIV-positive tuberculosis cases comprised 24% of all tuberculosis deaths and 22% of all HIV-related deaths in 2009.
People living with HIV at least 13 years of age. In the quantitative portion of this systematic review, the outcome will be assessed in different strata of absolute CD4 count. In the paediatric HIV population, absolute CD4 counts exhibit variability, vary by age, and do not have the same diagnostic value as absolute CD4 counts in adults. Therefore for immunological monitoring, clinicians use CD4 percentages in this population. TB diagnosis in children also poses additional challenges. Therefore, the paediatric population has been excluded from this review.
Antiretroviral therapy (defined as three or more drugs for treatment of HIV). Modern antiretroviral therapy uses three or more drugs for treatment of HIV. Use of less than three drugs has been associated with poor durability and the development of antiretroviral resistance. Since IPT may supplement ART’s preventive effects, studies only assessing the effect of ART and IPT on incident tuberculosis will be excluded from the quantitative portion of this review.
Placebo or no drugs. There is considerable debate on appropriate methods to use for determining when to start antiretroviral therapy. One way has been to compare mortality rates in people who started ART at CD4 counts from 351-500 to mortality rates in people who started ART at CD4 counts greater than 500. Another has been to compare mortality rates in CD4 strata, by antiretroviral therapy status. Given that tuberculosis rates differ by CD4 strata, an effective way to assess the effect of antiretroviral therapy on prevention of HIV-associated tuberculosis is by comparing rates of tuberculosis in different CD4 strata by antiretroviral therapy status.
No other inclusion criteria
Incident tuberculosis. There are many ways to diagnose tuberculosis. The gold standard is culture confirmation. Other methods include using acid fast bacilli smears, chest radiography, nucleic acid amplification tests, or clinical symptoms. This review will include studies regardless of diagnosis method used. However, the Newcastle-Ottawa Quality scale will assess for in-study measurement bias by downgrading studies that do not microbiologically confirm TB cases.
Data extraction, (selection and coding)
Author, year of publication, study population, inclusion and exclusion criteria (including what screening for tuberculosis was done prior to ART initiation), intervention and comparator, information on follow-up, outcome data, and study design will be collected in a standardised spreadsheet.
Risk of bias (quality) assessment
The Newcastle-Ottawa Quality Assessment Scale (NOS) will be used to assess bias in cohort studies. This scale rates studies from 0-10 based on nine criteria in the three sources of bias. Each criterion is worth one point except confounding, which is worth two points. Selection bias will be assessed using four criteria: (1) representativeness of the cohort on ART to the average adult on ART in the community from which study participants were drawn, (2) representativeness of the cohort off ART to the cohort on ART, (3) ascertainment of ART use, and (4) demonstration that prevalent tuberculosis was not present at the start of follow up and that cases developed during the first three months of ART are considered prevalent rather than incident episodes. For studies not reporting analyses in CD4 strata, adjustment for baseline CD4 will be used to judge whether appropriate methods have been used to address confounding. For studies reporting analyses in CD4 strata, adjustment for body mass index will be used to assess confounding. Measurement bias will be assessed by three criteria: (1) microbiological conformation of TB cases, (2) adequate follow up to detect ART’s preventive effect on TB (i.e. median follow up of at least one year), and (3) = 30% of participants were lost to follow up during the study.
Strategy for data synthesis
Assessment of publication biases: A funnel plot with the effect measures on the x-axis and the standard error of the log for the effect measures on the y-axis will be created for each of the five CD4 categories. The Egger and Begg tests will be used to test the symmetry of the funnel plots.
Assessment of heterogeneity: Chi-square with a significance level of 0.10 will be used to test for heterogeneity if all studies meeting inclusion criteria, in each of the CD4 categories, present outcome data by ART status. I-squared estimates will be used to test for heterogeneity if all studies meeting inclusion criteria, in each of the CD4 categories, present effect estimates and 95% confidence intervals. I-squared estimates greater than 50% will indicate moderate levels of heterogeneity.
Meta-analysis: Data will be entered on the log scale (generic inverse variance). All effect measures will be entered as ln(effect measure) and standard error as ln(upper limit/lower limit)/3.92. A random effects model will be used for generating summary estimates in each of the five CD4 categories.
Sensitivity analysis: If an I-square value greater than 50% is found in the analysis, its cause will be explored using sensitivity analysis. Differences in TB incidence, TB prevalence, or Human Development Index in the countries where studies were conducted may be causes for heterogeneity.
Analysis of subgroups or subsets
ART is already recommended in people living with HIV with CD4s = 350. Therefore, to make this systematic review and meta-analysis useful to clinicians, policy makers, and programme managers, we plan to have five categories based on CD4 at ART initiation for the quantitative portion of this review: =200, 200-350, 351-500, >500, and not reported.
Contact details for further information
Antiretroviral Treatment and HIV Care
Department of HIV/AIDS
Building D, 1st Floor, room 1005
World Health Organization
Avenue Appia 20
CH-1211, Geneva 27
Organisational affiliation of the review
World Health Organization
Dr Amitabh Suthar,
Details of any existing review of the same topic by the same authors
Anticipated or actual start date
22 March 2011
Anticipated completion date
30 November 2011
Conflicts of interest
Subject index terms status
Subject indexing assigned by CRD
Subject index terms
Antiretroviral Therapy, Highly Active; HIV Infections; Humans; Tuberculosis
Formal screening of search results against eligibility criteria
Risk of bias (quality) assessment
PROSPERO This information has been provided by the named contact for this review. CRD has accepted this information in good faith and registered the review in PROSPERO. CRD bears no responsibility or liability for the content of this registration record, any associated files or external websites.