This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

The study assessed the cost-effectiveness of Xpert MTB/RIF assay for rapid diagnosis of tuberculosis (TB) in low- and middle-income countries in comparison with conventional diagnosis based on smear microscopy and clinical diagnosis. The authors concluded that Xpert was a cost-effective strategy for rapid diagnosis of TB from the perspective of the health care system. The study used a valid methodological framework that was appropriate for the study setting. The authors’ conclusions appear robust.

Cost-effectiveness analysis, cost-utility analysis

The study assessed the cost-effectiveness of Xpert MTB/RIF assay for rapid diagnosis of tuberculosis (TB) in low- and middle-income countries in comparison with conventional diagnosis based on smear microscopy and clinical diagnosis.

Three diagnostic strategies were compared. The conventional strategy consisted of two sputum microscopy examinations followed (in smear-negative individuals with suspected TB) by clinical diagnosis that might include chest X-ray and antibiotic trial. Two alternative strategies were considered: Xpert in addition to smear (two smear examinations, if negative followed by Xpert on a single sputum specimen) and Xpert replacement of smear (single sputum specimen tested by Xpert instead of smear examination for all individuals with suspected TB).

India, South Africa and Uganda. Primary and secondary care.

Analytical approach:

Decision tree model in a hypothetical cohort of patients. Lifetime horizon. Health service perspective.

Effectiveness data:

Clinical data were derived from a selection of known relevant sources that included published meta-analyses of clinical trials, cohort studies, systematic reviews and World Health Organization (WHO) databases. Local data were used when available. All data on the accuracy (sensitivity and specificity) of diagnostic tests (a key input of the model) were derived from a demonstration study that reported implementation of the diagnostic tests at nine facilities. Some assumptions were needed.

Monetary benefit and utility valuations:

Disability weights were derived from a WHO publication (Global Burden of Disease)

Measure of benefit:

Disability-adjusted life-years (DALYs) and the number of cases detected were used as the summary benefit measures. An annual rate of 3% was applied to DALYs.

Cost data:

The economic analysis included costs associated with diagnostic procedures (which included building, overhead, staff, equipment and consumables, quality control and maintenance, and calibration inputs) and treatment (drugs, outpatient visits and hospitalisations). Diagnostic costs were derived from the same demonstration study as the accuracy data. Treatment costs were estimated using drugs costs from Global Drug Facility and the Management Sciences for Health International Price Tracker and unit costs for outpatient visits and hospitalisations from WHO-CHOICE. These estimates were validated using literature-based sources and a mid point between values found in the literature and actual estimates found in the authors’ settings. Costs were in US dollars ($). The price year was 2010. A 3% annual discount rate was applied when relevant.

Analysis of uncertainty:

One- and two-way sensitivity analyses were carried out on several inputs of the model using published ranges of values. Accuracy of diagnostic tests was investigated in several ways. A probabilistic sensitivity analysis was carried out using a Monte Carlo simulation to generate confidence intervals around model outcomes.

In India (low HIV prevalence and low multidrug-resistance prevalence), total costs and DALYs were $513,698 and 17.133 with conventional testing, $664,191 and 19.887 with Xpert in addition to smear, $709,248 and 20,019 with Xpert replacement of smear. In comparison with conventional testing, the incremental cost per DALY avoided was $55 with Xpert in addition to smear and $68 with Xpert replacement of smear.

In South Africa (high HIV prevalence and high multidrug-resistance prevalence), total costs and DALYs were $1,084,698 and 15.805 with conventional testing, $1,594,276 and 20,420 with Xpert in addition to smear, $1,758,467 and 20,702 with Xpert replacement of smear. In comparison with conventional testing, the incremental cost per DALY avoided was $110 with Xpert in addition to smear and $138 with Xpert replacement of smear.

In Uganda (high HIV prevalence and low multidrug-resistance prevalence), total costs and DALYs were $544,499 and 22,182 with conventional testing, $643,172 and 24,570 with Xpert in addition to smear, $670,137 and 24,611 with Xpert replacement of smear. In comparison with conventional testing, the incremental cost per DALY avoided was $41 with Xpert in addition to smear and $52 with Xpert replacement of smear.

Compared to Xpert in addition to smear, the incremental cost per DALY avoided with Xpert replacement of smear was $361 in India, $564 in South Africa, and $276 in Uganda. All these figures were well below the cost-effectiveness threshold based on gross domestic product (GDP) per capita.

Expected TB cases detected and incremental cost per TB case detected were reported.

The base case results were robust to variations considered in the sensitivity analyses. In the epidemiological setting of Uganda, small variations were observed but at a willingness to pay threshold of $490 the probability of Xpert replacement of smear versus Xpert in addition to smear remained high (around 75%).

The authors concluded that Xpert was a cost-effective strategy for rapid diagnosis of TB in low- and middle-income countries.

Interventions:

The selection of comparators appeared appropriate for the specific setting of the study. Sputum culture was considered to be the gold standard.

Effectiveness/benefits:

It appeared that a selective approach was used to derive clinical inputs for the model, as no information on a literature review was provided. Data sources were not extensively described. Most evidence on efficacy of the diagnostic tests came from the real-world experience of these procedures in a multi-setting study. Other data were retrieved from valid sources such as meta-analyses of clinical trials and large WHO databases. Extensive sensitivity analysis was conducted on key clinical inputs and focused on test accuracy. Both benefit measures appeared to be appropriate for capturing the impact of the disease on patients’ health. DALYs are widely used in developing countries as they allow the assessment of the whole burden of disease. Disability weights were derived using the validated WHO methodology.

Costs:

Costs categories included in the study reflected the perspective adopted by the authors. Sources of economic data included published studies and data retrieved at local institutions and this appeared to be methodologically valid. Some unit costs were presented but most data were given as macro-categories. Costs were varied and treated stochastically in the probabilistic sensitivity analysis. Other details such as the price year and discount rates were given.

Analysis and results:

The study results were reported clearly and presented as diagrams. Expected costs and benefits were appropriately combined using an incremental approach. The GDP criterion was used to consider a maximum acceptable threshold for cost-effectiveness ratios. Deterministic and probabilistic sensitivity analyses were performed to examine the robustness of base case findings. The methods and results of these analyses were reported clearly. The authors stated that the main limitations of the analysis were related to a lack of data for some clinical inputs and the need for assumptions; these were generally conservative against the Xpert option. Study findings should be transferable to settings with similar epidemiological and economic levels.

Concluding remarks:

The study used a valid methodological framework that was appropriate for the study setting. The authors’ conclusions appear robust.