Implementing rapid testing for tuberculosis in Mozambique
James Cowan a, Cathy Michel a, Ivan Manhiça b, Claudio Monivo a, Desiderio Saize a, Jacob Creswell c, Stephen Gloyd a & Mark Micek a
a. Health Alliance International, 1107 NE 45th Street, Suite 350, Seattle, 98105, United States of America.
b. Mozambican National Tuberculosis Control Programme, Maputo, Mozambique.
c. Stop TB Partnership, World Health Organization, Geneva, Switzerland.
Correspondence to James Cowan (email: firstname.lastname@example.org).
(Submitted: 02 April 2014 – Revised version received: 08 November 2014 – Accepted: 11 November 2014 – Published online: 26 November 2014.)
Bulletin of the World Health Organization 2015;93:125-130. doi: http://dx.doi.org/10.2471/BLT.14.138560
People infected with human immunodeficiency virus (HIV) are susceptible to tuberculosis. This has led to an HIV-driven tuberculosis epidemic in sub-Saharan Africa over the past two decades.1 Diagnosing tuberculosis is often difficult given the low sensitivity of smear microscopy. Among people infected with HIV, smear microscopy is even less sensitive due to the lower bacterial load in co-infected individuals.2
Xpert® MTB/RIF (Xpert®; Cepheid, Sunnyvale, United States of America) is a rapid and fully-automated molecular test that uses real-time polymerase chain reaction on sputum samples to simultaneously detect Mycobacterium tuberculosis deoxyribonucleic acid and the genetic sequence indicative of rifampicin resistance.3,4 The test has 88% sensitivity for tuberculosis and 94% sensitivity for rifampicin resistance. Given the high sensitivity, the test is more effective than sputum smear microscopy in diagnosing tuberculosis in HIV-infected people. Therefore the World Health Organization (WHO) has recommended the use of Xpert® for diagnosis of tuberculosis in countries with high prevalence of HIV infection and multi-drug resistant tuberculosis (MDR-TB).5,6 The four-module Xpert® machine and desktop computer retails for 17 000 United States dollars (US$) in low-income countries and each cartridge (one per sample) costs an additional US$ 10.7 This paper describes the introduction of Xpert® in Mozambique.
Mozambique has a high burden of tuberculosis. In 2012, WHO estimated an incidence of 552 per 100 000 people, which corresponds to approximately 140 000 new cases. However, only 34% of these estimated cases are diagnosed and notified to the national tuberculosis programme. Most, (94%), new tuberculosis patients are tested for HIV8 and 58% are found to be co-infected. Tuberculosis diagnosis is done in people reporting a persistent cough, using smear microscopy on sputum samples.8 MDR-TB testing is seldom done since culture of M. tuberculosis is not routinely available.8 New cases typically receive first-line therapy. The treatment success rate is 85% for new sputum smear-positive patients who start treatment at a public treatment centre.9 First-line therapy typically includes isoniazid, rifampicin, pyrazinamide and ethambutol. Second-line therapy typically includes kanamycin, levofloxacin, ethionamide, cycloserine, ethambutol and pyrazinamide.
In 2012, an estimated 3.5% (4900 people) of the new estimated tuberculosis patients had MDR-TB; however, only 213 patients with MDR-TB started appropriate second-line treatment.8 Both in Mozambique and globally, the prevalence of MDR-TB tuberculosis is increasing. Therefore improved MDR-TB diagnostics and access to treatment are priorities for the Mozambican national tuberculosis programme and the global health community.8
To increase the number of people diagnosed and treated with tuberculosis and to improve MDR-TB detection and treatment, the Xpert® implementation project was approved by the Mozambican Ministry of Health, with support from Health Alliance International. We chose to use a two-step testing algorithm for individuals suspected of having pulmonary tuberculosis. All sputum samples were analysed using smear microscopy and new patients with at least one positive sputum smear out of two separate samples were treated with first-line tuberculosis therapy. Patients with two separate smear negative results were retested using Xpert® on one of the original samples.
In the beginning of February 2012, four Xpert® machines were deployed at two district and two urban public hospitals in four districts in Sofala and Manica provinces. We developed a transportation network to transfer smear negative sputum samples from three of the 10 surrounding health centres to the two urban hospitals. Eight remote health facilities routinely transferred samples to the two district hospitals via existing sputum transport networks. This covered approximately 5% of all sputum samples tested in Mozambique.10
Initial training and technical support for clinicians and laboratory staff in the four hospitals was led by Health Alliance International. Prior to Xpert® testing initiation, each site was given a day-long training session to train laboratory technicians to operate the machines and teach clinicians to interpret the results. Initial introductory meetings were also held with district and provincial ministry of health staff and supervisors. Health Alliance International trainers then provided at least weekly supervision visits to each site with focused training for technicians with elevated testing error rates, to troubleshoot problems, to review cartridge procurement and distribution and to undertake initial calibration. The trainers also monitored the testing care cascade to ensure that patients received their test results and were linked to appropriate treatment. Daily Xpert® testing and long-term management – such as technical support, cartridge procurement and routine maintenance – was assumed by the Ministry of Health after appropriate training.
The national tuberculosis programme registers new tuberculosis patients as either sputum smear negative or positive. The system has no capacity to record molecular test results. Therefore, laboratories using the rapid tests were instructed to report patients testing positive with Xpert® as sputum smear negative cases. However, this instruction was not always followed and an estimated 5–10% of sputum smear negative cases that tested positive on Xpert® were reported as sputum smear positive.
The performance of this project was monitored from reports generated from individual Xpert® machines. These reports were collected by hand each month, and information extracted included the number of tests run per quarter, a summary of the results, and lists of patients that tested positive, which were reconciled with the tuberculosis treatment registries. This was combined with quarterly data from the national tuberculosis programme on the reported number of sputum smears analysed and the number of patients started on tuberculosis treatment.
Xpert® was approved for general use by the Mozambican Ministry of Health and the national tuberculosis programme formally approved the implementation of this project. This implementation project was not a formal research study thus the University of Washington institutional review board declined to provide a full review of this proposal.
Implementation and results
From 1 January 2012 to 31 December 2012; 1558/12 509 (12%) people tested at the four hospitals had sputum smear positive results. Of the 10 951 people with sputum smear negative results, 8631 (79%) were tested using Xpert® and 1081 (13%) were positive for tuberculosis. Thus, during this intervention Xpert® testing increased the diagnosis of bacteriologically-confirmed pulmonary tuberculosis by 1081/1558 (69%). Rifampicin resistance was detected in 58/1081 (5%) samples. Overall, 1019 (12%) of tests failed because of machine errors, and invalid or no results due to energy shortages.
We reviewed the records of 445 newly-diagnosed patients from the four hospitals in the second quarter of 2012 and found that 26% [115/445] had not been started on treatment (18% [35/200] of those diagnosed by microscopy and 33% [80/245] diagnosed using Xpert®). For patients with rifampicin resistance, 53% [31/58] were not started on second-line treatment. Three patients with rifampicin resistance were mistakenly initially started on first-line treatment but were converted to second-line treatment.
During 2012, 5076 people were placed on treatment in the four intervention districts, an increase of 632 patients (14.2%) over the baseline yearly average for these districts which was 4444 patients (15 554 patients were treated in these districts during the 3.5 years before implementation).
Despite encouraging outcomes, there were several programmatic and operational challenges in implementing and maintaining the Xpert® machines (Table 1).
Table 1. Responses to technical and logistical challenges encountered when implementing rapid tests for tuberculosis, Mozambique 2012–2013
One computer failed within six months, despite being attached to a surge protector/current stabilizer. Twelve of the original 16 (75%) Xpert® modules broke or failed calibration within 15 months and had to be replaced. In early 2013, Xpert® cartridges and calibration kits were back-ordered, resulting in several days without Xpert® cartridges despite ample lead-time in ordering.11 Updating antivirus software and operating systems and transferring data from the local computer to Cepheid in California, USA, was also difficult. Cepheid depends on Xpert® machines being connected to the internet to be able to diagnose problems remotely – this was not feasible in Mozambique and caused significant delays. Over time, shipments of cartridges and customer service improved as Cepheid employees and our team learned how to work with each other. However, importing consumables to Mozambique and clearing customs remained challenging, expensive and time-consuming.
The lack of a notification system to report positive Xpert® results to the national tuberculosis programme and no ability to automatically upload results electronically to a central secure database made it challenging to assess the impact of this intervention. It was difficult to remotely monitor Xpert® testing and notify key individuals in a timely fashion – such as provincial MDR-TB care coordinators when a patient tested positive for rifampicin resistance. WHO guidance on recording and reporting test results should allow national tuberculosis programmes to better monitor performance.12
The current two-step testing algorithm likely caused delays in tuberculosis diagnosis relative to using Xpert® as a first-line test. Some patients returned after 24 hours for their smear results, but did not get their subsequent Xpert® result. The need for repeat visits to establish a diagnosis may have led to higher pretreatment loss to follow-up. In the beginning, sites did not routinely collect patient contact information so it was impossible to reach these primary defaulters. Initially there was no protocol to routinely reconcile laboratory and treatment registries to ensure that all patients diagnosed with tuberculosis were started on treatment. Efforts are ongoing to improve the entire tuberculosis care cascade and to increase treatment initiation rates.
Our small-scale implementation project allowed us to gain important practical experience about the strengths and limitations of this new diagnostic technique (Box 1). Rapid tests have the potential to improve the diagnosis of tuberculosis and identify drug resistance. However, the tools are expensive and we experienced problems with the machines, data transfer, linkage to tuberculosis treatment initiation, reporting and follow-up. If the use of rapid diagnostic tests for tuberculosis is expanded to the national level, system improvements are needed to prevent these problems from occurring at a larger scale.
Box 1. Summary of main lessons learnt
- Xpert® MTB/RIF (Xpert®) testing increased the number of confirmed bacteriologically positive tuberculosis cases. However, one third of patients diagnosed by Xpert® did not receive tuberculosis treatment, highlighting the need to strengthen the overall tuberculosis care cascade.
- Implementation and maintenance of Xpert® machines are costly and logistically challenging, which creates a need for more affordable and durable molecular testing platforms.
- Of the tuberculosis patients diagnosed with Xpert®, 5% were resistant to rifampicin, highlighting the need to develop a multi-drug resistance tuberculosis programme in Mozambique.
The national tuberculosis programme is committed to effective care of all tuberculosis patients in Mozambique. Improving diagnosis is a part of this commitment and, therefore, rapid tests like Xpert® will be provided to more health centres. However, affordable and durable tuberculosis diagnostic platforms that can handle the stresses of laboratory environments in low-resource settings are still needed. Remote monitoring and evaluation of rapid testing systems may improve with the use of platforms such as GxAlert, XpertSMS and the Cepheid cloud system.13 Finally, while innovation in diagnostic tests is important, this is only one part of the tuberculosis care cascade. When increasing the number of people diagnosed with tuberculosis, attention is needed to ensure they receive the appropriate treatment, care and support.
Health Alliance International is a centre within the Department of Global Health, University of Washington, Seattle, USA.
- Chaisson RE, Martinson NA. Tuberculosis in Africa–combating an HIV-driven crisis. N Engl J Med. 2008 Mar 13;358(11):1089–92. http://dx.doi.org/10.1056/NEJMp0800809 pmid: 18337598
- Ramsay A, Cuevas LE, Mundy CJ, Nathanson CM, Chirambo P, Dacombe R, et al. New policies, new technologies: modelling the potential for improved smear microscopy services in Malawi. PLoS ONE. 2009;4(11):e7760. http://dx.doi.org/10.1371/journal.pone.0007760 pmid: 19901989
- Boehme CC, Nabeta P, Hillemann D, Nicol MP, Shenai S, Krapp F, et al. Rapid molecular detection of tuberculosis and rifampicin resistance. N Engl J Med. 2010 Sep 9;363(11):1005–15. http://dx.doi.org/10.1056/NEJMoa0907847 pmid: 20825313
- Rachow A, Zumla A, Heinrich N, Rojas-Ponce G, Mtafya B, Reither K, et al. Rapid and accurate detection of Mycobacterium tuberculosis in sputum samples by Cepheid Xpert MTB/RIF assay–a clinical validation study. PLoS ONE. 2011;6(6):e20458. http://dx.doi.org/10.1371/journal.pone.0020458 pmid: 21738575
- O’Grady J, Bates M, Chilukutu L, Mzyece J, Cheelo B, Chilufya M, et al. Evaluation of the Xpert MTB/RIF assay at a tertiary care referral hospital in a setting where tuberculosis and HIV infection are highly endemic. Clin Infect Dis. 2012 Nov;55(9):1171–8. http://dx.doi.org/10.1093/cid/cis631 pmid: 22806590
- Tuberculosis diagnostics Xpert MTB/RIF Test October 2013. Geneva: World Health Organization; 2014. Available from: http://www.who.int/tb/publications/Xpert_factsheet.pdf [cited 2014 Nov 21].
- Negotiated prices for Xpert® MTB/RIF and FIND country list [Internet]. Geneva: FIND; 2014. Available from: http://www.finddiagnostics.org/about/what_we_do/ successes/find-negotiated-prices/xpert_mtb_rif.html [cited 2014 Nov 21].
- Global tuberculosis report 2013. Geneva: World Health Organization; 2013.
- Global Health Observatory Data Repository, Mozambique country statistics [Internet]. Geneva: World Health Organization; 2014. Available from: http://apps.who.int/ghodata/?vid=14200&theme=country [cited 2014 Nov 21].
- Relatório de actividades desenvolvidas durante o ano de 2013. Maputo: Ministry of Health; 2014. [Portuguese].
- Creswell J, Codlin AJ, Andre E, Micek MA, Bedru A, Carter EJ, et al. Results from early programmatic implementation of Xpert MTB/RIF testing in nine countries. BMC Infect Dis. 2014;14(1):2. http://dx.doi.org/10.1186/1471-2334-14-2 pmid: 24383553
- Definitions and reporting framework for tuberculosis, 2013 revision. Geneva: World Health Organization; 2013.
- Cowan JF, Manhiça I, Michel C, Gloyd S, Creswell J, Macek C, et al. A brief review of remote monitoring tools for GeneXpert MTB/RIF: Initial implementation of GxAlert in Mozambique. Int J Tuberc Lung Dis. 2014 Nov;18(11 Suppl 1):S512–3.