Table 2 Challenges and recommendations for programmatic implementation of Xpert MTB/RIF testing

Machine utilization

Overestimation of testing numbers and underutilization of machine capacity.

Conduct a needs assessment that includes the current number of people with suspected TB tested at the facility, the need for a referral system and the impact of the proposed testing algorithm on testing numbers.

Two-module machines may be a less expensive alternative in many settings.

Testing algorithms

Lack of national guidance on testing algorithms and use of questionable testing strategies.

To improve yield and reduce cartridge use, consider screening with CXR.

Consider elimination of smear microscopy as a first test to reduce delay, loss to follow-up, and avoid repeat testing of most individuals due to large proportion of smear negative results.

Time to diagnosis

Sputum transport systems and testing algorithms can prolong delay between sputum submission, results and treatment initiation.

Xpert MTB/RIF testing sites should be located as close to the patient as possible to allow for rapid treatment initiation keeping in mind throughput and electrical power limitations.

Referral networks can help utilization rates but can also be costly to maintain.

Procurement

Short shelf life of cartridges, use of machine beyond date of needed calibration, unanticipated extra costs of shipping and customs clearance, local procurement agent not always helpful.

Stagger cartridge shipments to avoid stock outs and expiry.

Plan for module calibration by ordering test kits early.

Plan for extra costs associated with shipment and customs clearance.

Training

Staff rotation and new practices around request forms, specimen transport and clinical decisions for rifampicin resistant results.

Testing can be easily done by well-trained lay workers to support laboratory staff.

The manufacturer has conducted web trainings via videoconference and webinars and has released a web-based training video which some projects used as an adjunct tool for facilitation of trainings.

Infrastructure and power supply

UPS (15 minutes) Cepheid offers is inadequate in many settings. Most laboratories need some infrastructure improvements to allow proper testing.

A standard 800-2000VA inverter and a 12 V/120-200 AH battery provided power for over 6 hours to one four-module GeneXpert machine and a laptop (200 wattage required). As a general rule, a 10AH battery will be able to power a requirement of 100 watts per hour, but more AH are needed when the discharge is over a short period of time (less than 20 hours). These can be procured locally.

Failed tests

Differences between types of failed tests are unclear and available data not always used.

Error results are coded by the machine and should be documented. The .gxx files can provide this information and the types, reason, locations, and technicians associated with the error should be tracked.

High failed test rates in certain projects.

No result test results are often caused by a power failure.

Invalid results are not caused by testing a specimen of insufficient volume or one which contains saliva the way poor sputum quality is defined in a NTP. Rather, they seem to be caused by other problems with sputum. Emphasizing correct sputum collection techniques, including mouth rinsing to remove food or particles which could inhibit the assay, may help reduce Invalid results, as well as improve yield.

Drug resistance

Some confusion about clinical decisions after receiving rifampicin resistant and indeterminate results.

Patients with indeterminate results have TB, but rifampicin resistance cannot be confirmed due to a very low burden of TB bacilli in the specimen. Unless there is documented DR-TB risk, first-line treatment can be started. Follow-up of these patients is warranted.

Trepidation over Xpert MTB/RIF use overburdening DR-TB programs.

The majority of drug-resistant cases in almost all countries will be found among new cases, requiring testing of people with suspected TB rather than patients already in TB treatment.

Patients with rifampicin resistance will be detected in greater numbers and with greater speed than under current conditions, and significant resources and coordination will be required.

Recording and reporting

Supplied GeneXpert Dx software is not appropriate for analyzing patient cohorts, many times failed tests are recorded on paper with a generic error ‘result’, underreporting of errors, lack of clear guidance from national programs on recording and reporting of cases identified by Xpert MTB/RIF.

Dissemination and uptake of WHO guidelines on recording and reporting should be adopted by countries and their NTPs.

Automated reporting mechanisms can improve both the timeliness and accuracy of reporting as well as assist in supply chain management.