Pooling of sputa as a diagnostic strategy for diagnosis of TB when resources are constrained
Barriers to initial TB test uptake include high per test costs, stock outs of consumables, and weak supply chain systems. Periods of resource constraint exacerbate these challenges, further limiting access to testing services. Sputum pooling, in which samples from several individuals are mixed and tested together, is a strategy to improve testing efficiency and reduce costs in resource-constrained environments. This strategy has been used in other infectious diseases (e.g., blood bank screening for infectious hepatitis) and can lead to reductions in cost test, reagent and commodity costs, machine time, and laboratory workload.
Principles of pulmonary TB detection on pooled sputa
In practice, testing pooled sputum for TB is a two-step process. For LC-aNAAT testing, up to four sputum samples that are identified for pooling are brought together, labeled with a unique identifier for their pool, and individually treated with sample processing reagents according to manufacturer instructions. A portion of each liquefied sputum sample is then added, one at a time to the pool, and the pooled sample is tested. All the individual and pooled sputum samples are stored under manufacturer-recommended conditions until test results are available.
If a pooled sample tests negative, then no further testing is performed as it is presumed that all of the individual samples that contributed to the pool are negative for TB detection. If a pooled sample tests positive, then the individual samples that contributed to the pool must be re-accessed and tested individually to identify which of them is positive. One or more of the individual samples in the pool may be positive for TB.
When resource constraints do not allow for the testing of individual samples: In adults and adolescents with signs and symptoms for pulmonary TB or who screen positive for pulmonary TB, LC-aNAATs on (up to 4) pooled sputa may be used as the initial diagnostic strategy for diagnosing TB rather than LC-aNAATs on individual samples (conditional recommendation, high certainty of evidence).
Remarks
- The product for which eligible data were assessed and met the class-based performance criteria for LC-aNAATs for this recommendation was Xpert MTB/RIF Ultra.
- Evidence was available for both induced and expectorated sputum.
- This recommendation applies to pooling of individual specimens from up to four individuals based on a majority of evidence for this pooling ratio.
- Pooling may result in reduced costs and increased cost-effectiveness for testing programs. Savings were reduced and cost-effectiveness was not documented in settings of test positivity above 24% (commodity costs alone, modeled evidence).
- This recommendation does not apply to people living with HIV (limited evidence), children (insufficient evidence), or individuals at increased risk of drug resistance due to potential for reduced equity and increased harm. These and other priority populations may be prioritized for individual testing according to local risk groups.
- Diagnostic accuracy of the pooling strategy was reduced when sputum was collected in community settings (sensitivity range of 25-96%, 4 studies) compared with facility-based sputum collection (sensitivity range of 84-100%, 7 studies) and the reductions were greater than those associated with individual testing. The reasons for this variability were not directly assessed but may include bacterial load or other factors.
Given relative reduced accuracy and high user value on accurate testing services, the panel suggested that pooling should be reserved for situations where resource constraints do not allow for the testing of individual samples. However, even in these situations, the GDG determined that people known to be living with HIV, children, individuals at increased risk of drug resistance and other priority populations should not be tested with the pooled testing strategy but should instead be prioritized for individual testing due to the potential for reduced equity and increased harm among individuals at increased risk of TB morbidity and mortality.
Overview of the evidence and judgements
The available evidence on accuracy of the pooled diagnostic strategy included seven studies (7 countries, 14 741 participants), where 83% of the sputum samples (12 287 participants) were pooled at the 1:4 ratio. Accuray evidence was only available for Xpert MTB/RIF Ultra and swabs of sputum were not assessed.
The economic evidence was based on a systematic review (8 published studies and 2 unpublished reports) specific to use of LC-aNAATs on pooled respiratory sample for TB diagnosis. Evidence on user perspectives of pooling as a diagnostic strategy for LC-aNAATs was available from in-depth interviews with TB programme officers, implementers, lab managers and researchers, a focus group discussion with people with lived experience of TB, an online survey of national TB programme officers, implementers, lab managers and researchers and unpublished evidence submitted through the WHO open call for data.
The GDG judged LC-aNAAT accuracy of the pooling strategy to be very accurate, with a summary sensitivity of 84.8% (95% CI 81.3 to 87.8) and a summary specificity of 98.1% (95% CI 97.4 to 98.7) based on high certainty of evidence. Sensitivity of the pooling strategy was lower than individual testing (reduction in sensitivity at the 1:4 pooling ratio: -3.1%, 95% CI -7.0% to +0.8%) with greater reductions observed in community-based settings (sensitivity reduction of 3-33%, 4 studies) than in facility-based settings (sensitivity range of 0-4%, 7 studies).
The strategy was found to have the potential to significantly reduce costs, particularly in low-prevalence or low-positivity settings. Cost-minimization studies consistently demonstrated cost savings utilizing the pooling strategy compared with individual testing. The cost per person tested using individual tests ranged from $7.97 to $11.33, while the estimated cost per test in the pooled testing approach ranged from $3.04 to $9.95. These studies assumed no loss in effectiveness or cases diagnosed among the pooled versus individual approaches. The GDG judged the strategy to lead to moderate savings and determined it would be cost-effective. The GDG also suggested that equity would be increased because it would enable testing more people with LC-aNAATs, and that the strategy would probably be acceptable and feasible.