Review 1: "C-reactive Protein-Based Tuberculosis Triage Testing: A Multi-Country Diagnostic Accuracy Study"

RR\ID Evidence Scale rating by reviewer:

• Strong. The main study claims are very well-justified by the data and analytic methods used. There is little room for doubt that the study produced has very similar results and conclusions as compared with the hypothetical ideal study. The study’s main claims should be considered conclusive and actionable without reservation.

***************************************

Review: This is a well-written paper by a highly renowned TB diagnostic consortium. It is novel and merits publication. I have a few detail comments, but would also like to invite the authors to speculate a bit and perhaps revise the limitations section.

Unfortunately, supplementary materials could not be found and reviewed.

Methods /Procedures:

• It reads as if the lab technicians processing sputum were blinded to CRP results, but I would rather expect the performers of the index test to be blinded to reference test results.

• Liquid mycobacterial culture: Reference is given, but the culture method should at least be named here – since it is the reference standard. Specifically, please describe how non-tuberculous mycobacteria were detected to avoid “false positive” reference standard results.

Results: 

• People with newly diagnosed diabetes are mentioned, please also mention the method for this in the methods section.

• Beta-coefficient: along with the numbers, the units should be given for continuous variables, e.g. ß coefficient for an increase in 1 BPM. 

Discussion:

• The sentence “These geographic differences in CRP sensitivity, which have also been reported by prior studies, correlated with observed geographic differences in mycobacterial load” makes sense when sensitivity is considered – more severely ill patients tend to have higher CRP apparently – but on specificity, this should be reconsidered as higher TB load in cases would not be expected to cause false positivity in non-TB cases, which is the meaning of specificity. Other factors could be a genetic disposition to higher inflammatory response in Africans, or a different spectrum of non-TB pathogens causing cough, ones that would be more likely to cause an increase in CRP.

• The statement: “Alternatively, for people with known HIV infection, CRP could be used to screen unselected PWH or PWH who screen-positive by screening strategies more sensitive than cough ≥2 weeks (i.e., WHO 4-part symptom screen, CXR), and in these scenarios, CRP can be expected to have higher specificity than reported here.” is problematic, as the ROC regression suggests that in people with more mild disease, CRP sensitivity might be lower.

Limitations:

• I would be most interested to hear some thought, rather than being prescriptive on a change. One limitation of the study is suggested by the relationship between bacterial load and specificity – in a population where TB is ruled out, such as the one used for specificity, there should not be such a relationship. The correlation that was apparently found, suggests that there are indeed a few individuals with TB in this population, not detected by the reference standard but by CRP testing; which leads to a lower specificity – however this would then be falsely low. In a population where bacterial load tends to be higher, such cases then may be more likely to be CRP positive. An alternative explanation could of course be that there are other non-TB factors that are collinear with higher TB bacterial burden in a population.

• It is however difficult to conceive a better reference standard for such a large study – one could think of a follow-up visit to see how symptoms resolved without TB treatment, as in children, but this would affect feasibility. One way to address this in this paper would be to search for literature on the sensitivity of the reference standard used, against a much more elaborate workup. Such limitations to the reference standard might be well placed in a revised limitations section, and am curious about the authors’ thoughts.