This operational study evaluated the performance of the MODS assay among drug-resistant TB suspects living in a high HIV-prevalence setting. MODS detected M. tuberculosis
and associated drug resistance with high sensitivity and shorter time to positivity compared with reference standard culture and DST methods. Given the expanding global prevalence of MDR-TB/HIV and continued need for an affordable, accurate, and rapid point-of-care test, these findings have implications for other limited-resource settings 
Zimbabwe has among the highest TB incidence per capita (603/100,000) in the world, 
with approximately 70% of active TB cases occurring among individuals co-infected with HIV. 
Although HIV prevalence has declined since the 1990s, 16% of the adult population remains HIV-infected. 
The World Health Organization estimates the prevalence of MDR-TB in Zimbabwe among patients with a prior history of TB treatment to be 8.3% (95% CI, 3–20%), 
though these data were collected in 1995 and the current extent of drug resistant-TB in the country is unknown. That prevalence of MDR-TB has increased in the country in the context of severe economic destabilization, challenges to tuberculosis control, and population displacement has been suggested, 
though supporting evidence is thus far lacking.
Expanded capacity to perform DST in high burden settings is a critical need. In countries where mycobacterial culture is not routinely utilized, failure of one or more regimens of TB drugs is typically a prerequisite for referral for DST. Thus, 12 or more months often elapse from clinical presentation to MDR-TB confirmation. Given high early mortality 
and the potential for ongoing transmission, 
expedited diagnosis and early institution of effective therapy is life-saving and a critical public health mandate. Although debate exists as to best scale-up option for DST in resource limited settings, 
the high accuracy, low cost, ability to discern both isoniazid and rifampicin resistance, relative ease of operational implementation and short turnaround time should make MODS a strong consideration.
Consistent with the single other study assessing MODS diagnostic accuracy among TB suspects in a high-HIV prevalence region, 
we found somewhat lower sensitivity for M. tuberculosis
detection than that reported from other settings. 
In our study, most false-negative specimens either required prolonged incubation prior to positivity or were considered false-negative due to MODS contamination. Although culture contamination was not dissimilar to that reported by other investigators, 
contamination of liquid mycobacterial cultures is a known challenge for routine laboratories in sub-Saharan Africa. 
Sensitivity for detection of isoniazid and rifampicin resistance was similar to previously reported studies, 
though negative predictive value was lower due to the high prevalence of drug resistance noted among MDR-TB suspects in this high burden setting; negative predictive value would be marginally higher when testing new patients without history of prior TB treatment (i.e., those at lower risk for drug resistance). Further, the sensitivity for detection of isoniazid resistance could be increased through use of a lower MIC (0.1ug/ml) cutpoint in MODS. 
Of note, diagnostic accuracy in studies of drug susceptibility testing is dependent upon choice of denominator for analysis. With a denominator including patients with reference standard culture-positive disease (as opposed to a denominator including specimens culture-positive by both index test and reference standard), MODS sensitivity and negative predictive value for detection of drug resistance would be marginally lower.
Although upfront costs are higher relative to standard, noncommercial MODS (approximately $5.00 per test versus $1.48 for standard MODS), 
it has been anticipated that use of the TB MODS Kit™ (Hardy Diagnostics, Santa Maria, CA USA) will improve biologic security, attention to published standard operating procedures, and adherence to quality assurance systems. While validation data reported by the manufacturer are excellent, 
diagnostic accuracy studies by independent investigators are necessary and are underway. In the current study, no meaningful difference in diagnostic accuracy was noted between the “in-house” noncommercial MODS assay and the commercial kit, though power for this determination was limited.
A strength of our study is its operational, real-world nature. However, threats to internal or external validity include the following. First, as in many settings, routine DST of retreatment TB cases in Harare, Zimbabwe is codified in policy though not yet standard practice due to resource limitations, and our sample must be regarded as one of convenience. Further, we were unable to ensure standardization of specimen collection and processing for routinely collected samples, and cannot rule out the possibility that some false-negative results may have been due to suboptimal quality in these areas; similarly, data was incomplete and collected retrospectively for some individuals. Second, power to detect meaningful differences in our HIV-stratified analysis was limited. However, that our point estimates are similar to a recent adequately powered study from a setting of similar HIV prevalence 
lends confidence to our results. Third, we were unable to undertake comprehensive microbiologic, molecular, and epidemiologic investigation into discordant cases. Our use of a reference standard including both solid and liquid culture methods provided a rigorous definition of true positive results. Last, diagnostic accuracy is a surrogate for patient-important outcomes such as time to treatment initiation and mortality.
In conclusion, MODS detected M. tuberculosis and M. tuberculosis drug resistance with high sensitivity and more rapid time to positivity compared with standard culture and DST methods. Further, no detectable differences in diagnostic accuracy were noted for HIV-infected patients. Prompt treatment of patients with MDR-TB and screening of their contacts will be essential to prevent further spread of drug-resistant M. tuberculosis; that this will occur within the context of continued socioeconomic stabilization and improved health service delivery is hoped for and anticipated. Studies focused on patient-important outcomes, along with valid sampling methods to generate accurate estimates of the prevalence of MDR-TB in modern-era Zimbabwe, are urgently needed.