We estimated the incremental cost and impact on diagnosis and treatment uptake of national rollout of Xpert MTB/RIF technology (Xpert) for the diagnosis of pulmonary TB above the cost of current guidelines for the years 2011 to 2016 in South Africa.
We parameterised a population-level decision model with data from national-level TB databases (n = 199,511) and implementation studies. The model follows cohorts of TB suspects from diagnosis to treatment under current diagnostic guidelines or an algorithm that includes Xpert. Assumptions include the number of TB suspects, symptom prevalence of 5.5%, annual suspect growth rate of 10%, and 2010 public-sector salaries and drug and service delivery costs. Xpert test costs are based on data from an in-country pilot evaluation and assumptions about when global volumes allowing cartridge discounts will be reached.
At full scale, Xpert will increase the number of TB cases diagnosed per year by 30%–37% and the number of MDR-TB cases diagnosed by 69%–71%. It will diagnose 81% of patients after the first visit, compared to 46% currently. The cost of TB diagnosis per suspect will increase by 55% to USD 60–61 and the cost of diagnosis and treatment per TB case treated by 8% to USD 797–873. The incremental capital cost of the Xpert scale-up will be USD 22 million and the incremental recurrent cost USD 287–316 million over six years.
Citation: Meyer-Rath G, Schnippel K, Long L, MacLeod W, Sanne I, Stevens W, et al. (2012) The Impact and Cost of Scaling up GeneXpert MTB/RIF in South Africa. PLoS ONE 7(5): e36966. https://doi.org/10.1371/journal.pone.0036966
Editor: Mark Patrick Nicol, University of Cape Town, South Africa
Received: December 13, 2011; Accepted: April 10, 2012; Published: May 31, 2012
Copyright: © 2012 Meyer-Rath et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Funding for this study was provided by the South Africa Mission of the US Agency for International Development (USAID) under the terms of Associate Cooperative Agreement No. 674-A-00-09-00018-00 under Country Research Activity Leader Award No. GHS-A-00–03-00020 (GMR, KB, LL, WM and SR) and under Agreement No. 674-A-00-08-0007-00 to Right to Care (IS). KB received additional training support from Fogarty International Center/ National Institutes of Health, ICOHRTA AIDS/TB Grant No. U2RTW007373. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
South Africa bears a large share of the global burden of HIV and tuberculosis (TB) co-infection, with a TB prevalence of 795/100,000 in 2010 . Recent studies have shown that up to 70% of TB suspects tested for HIV are HIV co-infected, with TB being the most common cause of mortality in HIV infected persons . South Africa also has a high burden of multi-drug resistant TB (MDR-TB), with more than 7,000 cases diagnosed in 2010 .
In this context, conventional TB diagnostic technologies that have been used for decades, such as smear microscopy, are no longer reliable, because 24% to 61% of HIV-positive tuberculosis patients are smear-negative . Significant hope for turning the tide of the TB epidemic thus lies with the recent development of rapid molecular assays. One of these, the GeneXpert System (Cepheid, Sunnyvale, CA) using the cartridge-based Xpert MTB/RIF (Xpert) assay, allows for rapid detection of Mycobacteria tuberculosis (MTB) and a rapid screen for rifampicin (RIF) resistance , . Once a sputum sample has been collected from a patient, results are available in about 2 hours, without the requirement of highly trained laboratory personnel or additional biosafety measures .
In a multi-centre, prospective evaluation that included two South African sites, Xpert was found to be highly specific (99.2%) and highly sensitive (98.2% in smear-positive patients and 72.5% in smear-negative, culture-positive patients) for MTB . Similar results have been obtained in other studies , . In December 2010 the World Health Organization strongly recommended the use of Xpert, endorsing it as “the initial diagnostic test in individuals suspected of MDR-TB or HIV/TB”- in other words, most TB suspects in South Africa .
In March 2011, the South African National Department of Health announced a rapid, nationwide scale up of access to Xpert, to be achieved within a 2–3 year period. In conjunction with the South African National Health Laboratory Service (NHLS), it launched a pilot program that placed Xpert platforms in 25 smear microscopy laboratories across the country, with throughputs ranging from 16 to more than 400 tests per day and all instruments interfacing electronically with the centralised laboratory information system. By July 2011, over 50,000 samples had been processed, with a 3.4% error rate . Based on this successful pilot, existing Xpert-enabled laboratories are now being upgraded to allow complete migration from smear microscopy to Xpert for TB diagnosis. Next, Xpert will be placed in all laboratories in nine designated high case-load districts. Finally, Xpert instruments will be placed at all other existing smear microscopy laboratories, fully replacing smear microscopy for the diagnosis of pulmonary TB in South Africa. Smear microscopy capacity for monitoring of TB treatment however will remain in these laboratories.
To help determine the additional budgetary resources required to procure and utilise Xpert as planned, while taking into account savings from the reduction in smear microscopy, the NHLS asked us to estimate the impact of Xpert scale-up on the number of TB cases diagnosed and treated and the incremental cost of the scale-up plan. Here we report the main results of our analysis, which have since been used to guide South Africa’s national policy.
Sources of Data
For this study we developed a model representing the diagnostic process starting with TB suspects, continuing to TB cases, and ending with treatment. The number of TB cases and cost were calculated quarterly for the financial years 2011 to 2016, covering the period from April 2011 to March 2017. Data for the model came from a random sample of all patients entered into the national-level NHLS TB specimen database in 2010 (n = 1,329,664) and a random sample of all patients entered into the national-level Electronic TB Register in 2010 (ETR, n = 286,741). Access to the two databases had been granted by the management of the NHLS and of the TB unit in the South African Department of Health, respectively. The databases had been anonymised for our use and did not contain any patient identifiers. We also used results of South African Xpert implementation studies ,  and other literature –. Additional inputs on the performance of the Xpert test (RIF resistance rate and test failure rate) were based on NHLS data from the Xpert pilot phase during March and April 2011. Information on loss rates in suspects and baseline bacteriological coverage came from the Quarterly TB Statistics 2010 collected by the NDoH .
Data from the NHLS Database
The national-level database of the National Health Laboratory Service provided model assumptions on the proportion of suspects who were smear microscopy-positive and those who were culture-positive (by smear status). The NHLS database contains the results of all laboratory tests done in the public sector in South Africa, with multiple entries for the same patient linked using the patient’s name, first line of address, and date of birth. We requested a subset of the database containing information on tests for TB (smear microscopy, TB culture, line-probe assay, and drug-susceptibility testing) of all patients who had an entry during the 2010 calendar year, as well as all TB test data on the same patients for 2004 to 2009. Due to the size of the original database we used a random sample of 100,000 patients (7.5% of all patients) for our analysis, representing all provinces and all types of clinical settings. Since the analysis was restricted to the diagnosis of TB in ambulatory care settings (and the South African guidelines prescribe a different approach to diagnosing patients in inpatient settings), we limited our analysis to the 83,977 subjects who contributed samples from outpatient clinics.
For positivity rates we used the first smear and/or culture performed in 2010 only. Since the database contained no information on whether tests were done for diagnostic or for treatment monitoring purposes, we deleted from the database everyone with a first entry in 2009 in order not to capture the treatment monitoring smears and cultures done during 2010 for patients initiated on treatment in 2009.
Data from the Electronic TB Register
The Electronic TB Register (ETR) provided model assumptions on the proportion of patients with positive, negative, and unknown HIV status; the proportion of patients with a TB history; and proportions of smear and culture positive tests stratified by HIV and TB history status. The ETR is a register of all patients diagnosed with TB who are notified and initiated on treatment at the clinic level. The register starts at the clinic level as a book with one entry per notified patient. Copies of patient-level demographic and diagnostic information and treatment outcomes are sent to the district TB office several times a month, where they are entered into an electronic database. We downloaded information on all patients registered during the 2010 calendar year from all 9 provinces except KwaZulu Natal, for which data had been captured in a different format. We restricted our analysis to patients who had not moved or been transferred into the programme while already on treatment in order to avoid double-counting patients, and to patients with pulmonary TB. The remaining sample size was n = 57,688.
The database contained information for all patients on whether diagnosed cases were new cases or re-treatment cases and for some patients on HIV status. We used the proportion of re-treatment cases out of all cases to inform our assumption of the proportion of patients with TB history. In order to calculate the proportion of patients with a positive, negative or unknown HIV status we removed all patients with missing HIV information from the analysis (n = 32,951). For the analysis of proportions of smear and culture positive tests stratified by HIV status we regarded patients with an unknown HIV status (5.87%) as HIV positive, as suggested in the South African TB diagnostic guidelines . Information on pre-treatment microscopy status was missing in 21.38% of patients.
Scenarios and Algorithms
The model follows quarterly cohorts of TB suspects through up to three diagnostic visits under either a baseline scenario (current guidelines) or an Xpert scenario, under which the Xpert technology is scaled up either by the end of 2012 (accelerated scale-up) or by the end of 2013 (gradual scale-up). Both scale-up scenarios start from an Xpert coverage of 16% in June 2011, representing the existing level of Xpert capacity after the pilot phase. Patients with a positive Xpert MTB diagnosis are assumed to be initiated on treatment according to their resistance status at the same visit as they receive the positive result. Under both scenarios, we count patients as diagnosed even if they fail to return for the visit at which the positive result would be communicated to them, though they would not be assumed to initiate treatment, and we assume no loss to follow up after treatment initiation since no data is available yet on whether and how loss to treatment would differ between diagnostic scenarios. The diagnostic algorithms used in the baseline and Xpert scenarios, which reflect South Africa’s current and proposed new guidelines for TB diagnosis, are illustrated in Figure 1 and the treatment options for diagnosed patients in Figure 2.
LPA denotes line probe assay; DST, drug-susceptibility testing. Adult patients with suspected pulmonary MTB infection undergo a number of TB diagnostics at each of up to three consecutive diagnostic visits. Under the Baseline scenario, diagnostics are differentiated by whether or not patient have a history of TB treatment, and, for smear-negative patients, by HIV status. Under the Xpert algorithm, diagnostics are differentiated by HIV status for patients with a negative Xpert result only. Under both algorithms patients with a positive culture undergo further testing by line probe assay and, if this shows a resistance to RIF and/or INH, by drug-susceptibility testing for second line TB drugs. Under the Xpert scenario, patients with a positive Xpert result and RIF resistance (MTB+/RIF+) undergo further sputum microscopy and culture +/− LPA +/− DST for confirmation of the MDR-TB result and exclusion of XDR-TB. For smear microscopy, two sputa are collected; for an Xpert test, a single sputum is used. All tests are done on spot sputum samples.
R denotes rifampicin, H, isoniazid, Z, pyrazinamide, E, ethambutol, S, streptomycin, OFX, ofloxacin, KM, kanamycin, ETO, ethionamide, PZA, pyrazinamide, TRD, terizidone. Patients diagnosed with pulmonary TB are initiated on TB treatment according to their drug resistance status and, for the Baseline scenario, by their TB history. In the Xpert scenario, all patients without resistance are treated as New cases; regimen 2 is no longer used.
Model Inputs and Assumptions
Number of suspects.
Numbers of suspects (patients and/or contacts of TB cases with a positive TB symptom screen) were calculated using data on the general population aged 15 years and above from the Actuarial Society of South Africa AIDS Model  and an assumption of a prevalence of TB symptoms of 5.5% based on the Provincial Quarterly TB Reports . In the main analysis, this percentage increases by 10% every year, in line with the targets for South Africa’s ongoing Intensified Case Finding campaign . In sensitivity analysis we consider rates of 0% and 6.5%, a rate suggested by the WHO Stop-TB “Planning and Budgeting for TB Control” model for South Africa . The smear positivity rate is set at 9.89% of suspects at baseline, based on the 2010 NHLS TB database, and, together with the culture positive rate of smear negatives, decreases in both the baseline and Xpert scenarios as a function of the growth in suspects, allowing us to model a stable epidemic. This rate of decrease is calculated aswhere P(ss+) is the proportion of patients who are smear-positive, y is the model year, and gr is the rate of growth in suspects (10%, 0%, or 6.5% according to growth scenario). (The same calculation applies to the proportion of smear-negative patients who are culture-positive).
The current bacteriological coverage rate (the proportion of suspects who have a documented smear microscopy result, based on the Provincial Quarterly TB Progress Reports ) of 85% is held constant in the baseline scenario but increases to 100% in the Xpert scenario to reflect a higher use of laboratory diagnosis as a result of the faster turn-around and greater specificity achieved by Xpert.
We calculated the cost of TB diagnosis and treatment under the baseline and the Xpert scenarios from the government perspective, including the cost of outpatient visits, equipment, drugs, laboratory and radiological tests, infrastructure, training, and overhead. We used expert opinion and public-sector salary data to estimate the duration and cost of clinic visits, 2010 NHLS charges for all laboratory costs except the Xpert test, public-sector radiology costs, and public-sector drug costs and standard treatment algorithms for presumptive antibiotic treatment. TB treatment drug costs (including for mono-resistant and MDR-TB treatment) were calculated using August 2011 drug tender costs and the current South African TB guidelines . All non-drug outpatient costs are based on previously reported estimates , adjusted for inflation to 2011 ZAR and converted to USD at 1 USD = 7.94 ZAR . We included only outpatient treatment costs for MDR-TB since during the projection period a new strategy that replaces inpatient with outpatient care for MDR-TB patients is planned to be rolled out . Costs are reported in 2011 USD and presented undiscounted and inclusive of value-added tax (VAT).
The per-test cost of the Xpert technology was calculated in a separately-reported analysis , based on a bottom-up cost analysis of the pilot phase, estimates of the number and type of instruments required given diagnostic sample volumes at NHLS smear microscopy labs, and assumptions about when global volumes allowing discounts on the international price of Xpert MTB/RIF cartridges will be reached and how much of that global test volume will be borne by South Africa. Instrument costs were annualised over an expected useful life of the modules of four years; when varying the value to 3, 5 and 8 years, we found the cost per Xpert test to be insensitive to this assumption. The separately-reported analysis  estimated that in the period 2011 to 2013, the average cost per Xpert test performed will start at USD 33 and then decline to USD 25 between 2014 and 2016 when higher volumes and the resulting global volume discounts are reached. More detail on the cost per test can be found in Table 2.
Unless otherwise stated, all Xpert scenario results reported reflect accelerated scale-up, 10% growth in suspects, and South Africa procuring 50% of global Xpert test volumes . In sensitivity analysis we considered variation in five parameters: a) 0% and 6.5% growth rate in suspects; b) South Africa’s share of global volumes at 90%; c) an impact of full Xpert coverage on smear positivity and culture positivity rates of suspects as a result of a reduction in transmission, based on the results of a previous model , d) South Africa accessing Xpert cartridges at the volume-discounted price of USD 10.72 ahead of global test volumes reaching 3.4 million in December 2011, and e) an additional 4 months of inpatient care per patient for MDR-TB, valued at the national average cost per inpatient-day equivalent of $168.28.
The study was approved by the Human Research Ethics Committee of the University of the Witwatersrand and the Institutional Review Board of Boston University Medical Center.
Number of Suspects
Table 3 reports the projected numbers of suspects and patients diagnosed with TB under the baseline and accelerated Xpert scenarios. As a result of the 10% increase in suspects or contacts undergoing TB diagnosis expected from the ongoing Intensified Case Finding campaign , along with some population growth, the number of suspects requiring TB diagnosis increased in the model from 1.9 million per year to 3.2 million per year. Fifty-six percent of suspects were assumed to be HIV positive, based on data from the Electronic TB Register.
Number of Patients Diagnosed with TB
By 2013, the first year of full Xpert coverage, 30% more patients were diagnosed in the Xpert scenario than in the baseline scenario. This difference increased annually until 2016 due to the decreasing prevalence of smear-positive TB that resulted from intensified case-finding and the higher sensitivity of the Xpert algorithm for smear-negative TB. In addition, the Xpert algorithm led to a 69% increase in the number of patients diagnosed with drug resistance. The proportion of diagnosed patients with or without drug resistance who were initiated on treatment increased from 75% to 81%.
Tests Used and Timing of Diagnosis
Under the baseline scenario 43% of patients were diagnosed by smear microscopy, 51% by culture, and 6% clinically. This breakdown changed dramatically under the Xpert scenario, in which 85% were diagnosed by Xpert, 12% by culture, 3% clinically, and only 0.05% by smear microscopy. Figures 3 and 4 show the distribution of diagnoses across visits. As can be seen, under the baseline scenario, 44% of all patients were diagnosed by their second clinic visit, three to five days after the first visit, and another 39% by visit 3, six to seven weeks after the second visit (Figure 3). At full Xpert coverage in 2013, 82% of patients were diagnosed by visit 2, three to five days after their initial presentation, and 91% by visit 3. The difference in timing of diagnosis was even more pronounced for MDR-TB patients, with 0% of patients diagnosed by their second visit under the baseline scenario and 82% under the Xpert scenario (Figure 4).
Percentage of all patients diagnosed by visit 2 (assumed 5 days after first visit), by visit 3 (assumed 4–6 weeks after first visit), and thereafter (accelerated scale-up, 10% growth in suspects).
Percentage of patients with drug-resistant TB diagnosed by visit 2 (assumed 5 days after first visit), by visit 3 (assumed 4–6 weeks after first visit), and thereafter (accelerated scale-up, 10% growth in suspects).
While Xpert can largely replace smear microscopy for TB diagnosis, it cannot yet be used as a substitute for microscopy in treatment monitoring, since a DNA-based molecular test such as Xpert is unable to distinguish between live and dead mycobacteria. As a result, all patients diagnosed at their first Xpert test as MTB+ have to undergo additional smear microscopy for the purpose of obtaining a baseline for treatment monitoring. The WHO recommendation for the use of Xpert, moreover, currently only applies to sputum samples, and not to smears done from other materials for the diagnosis of extrapulmonary TB. The need for some smear microscopy capacity thus remains, though the number of smear microscopy tests decreased by 61–67% at full Xpert coverage. Similarly, the number of diagnostic cultures declined by 21–23%, and the number of line-probe assays by 44–55%. The number of drug-susceptibility tests required, in contrast, increased by 52–65% at full Xpert coverage.
Total Cost, Incremental Cost, and Cost Per Case
The incremental capital cost of introducing Xpert technology (including instruments, additional laboratory building space, security, and training) between 2011 and 2016 was estimated to be USD 22 million. The incremental recurrent cost (including cartridges, staff, transport, and quality assurance) over this time period varied between USD 287 million (gradual scale-up) and USD 316 million (accelerated scale-up). Capital cost did not differ between scale-up scenarios, but recurrent cost for the gradual scale-up scenario was lower in the first two years owing to the smaller number of machines placed and overall lower testing capacity. The resulting incremental cost for the Xpert roll-out per year was between USD 30 million and USD 64 million (accelerated scale-up) and between USD 20 million and USD 64 million per year (gradual scale-up).
Table 4 presents the resulting cost of the TB diagnostic programme (including laboratory cost, clinic visits, and clinical diagnosis) and of the diagnostic and outpatient treatment programme combined. At full Xpert coverage (from 2013 onwards) the Xpert technology increased the total cost of the TB diagnostic programme by between 53% and 57%, or USD 48 million to USD 70 million, per year. The cost per suspect tested increased by USD 21 to USD 22 per year, or between 53% and 57%. The cost per patient diagnosed with TB increased by between USD 46 and USD 52 per year, or 15% to 17%.
The outpatient cost of the TB treatment programme, assuming full coverage with treatment, increased by USD 55–62 million per year from 2013 onwards, or by 34–37%. This increase was the result of a rise in the number of patients initiating treatment due to lower loss to care during the diagnostic process. The cost of the full TB diagnostic and treatment programme per patient diagnosed and treated increased by USD 53–58 per year, or 7–8% (Table 4).
In sensitivity analysis we found the annual diagnostic and treatment cost to be sensitive to the assumption of growth in the number of suspects (under full Xpert coverage, reduction in cost of 7.7% to 20% for 0% growth, and by 2.5% to 7.3% for 6.5% growth) and to assumptions about an impact of full coverage with Xpert technology on transmission via a reduction in the smear and culture positivity rates in suspects (reduction in cost of 3.7% to 11.2%) (Table 4). Results were not sensitive to changes in the assumption about South Africa’s share of the total Xpert test volume from 50% to 90%. Including an approximation of inpatient cost for MDR-TB, however, doubled the total cost of the Xpert scale-up due to the large number of additional cases of MDR-TB diagnosed by Xpert.
The accelerated scale-up scenario and an assumption of 50% volume share led to price reductions being realised earlier and to lower overall cost. Accessing the cartridge price that is conditional on 3.4 million accumulative cartridges sold ahead of time, i.e., in December 2011, saves 26% of total diagnostic and treatment cost over the first 3 years.
Based on the model we developed for the South African National Department of Health, we estimate that the introduction of Xpert technology will substantially increase the number of TB cases diagnosed, MDR-TB cases identified, and patients started on appropriate treatment. It will also accelerate diagnosis, with 82% instead of 44% of patients diagnosed after the first visit. While these results are clear improvements in the performance of the TB diagnostic programme, they come at a cost, with the annual cost of the diagnostic programme increasing by 53–57% annually at full Xpert coverage, or USD 48–70 million per year, and the annual cost of the treatment programme increasing by 34–37%. The projected total cost of TB diagnosis and treatment in 2011 under the accelerated Xpert scenario, USD 293 million, is 35% more than the USD 218 million estimated as the total public-sector TB budget for 2011 , although it comprises only 2% of the total public health budget of South Africa for that year . Potential cost savings could come from the country’s guaranteeing to procure certain test volumes in order to access volume-dependent cartridge price discounts ahead of time, simultaneously ensuring access to Xpert technology at the lower price for all other countries.
The prospect of new molecular technology for the diagnosis of TB and MDR-TB with a potential for implementation at point of care gave rise to a number of papers urging that the operational difficulties of using this technology be taken into account alongside the standard cost and effectiveness considerations , –. Although it is impossible to foresee all the challenges that large-scale implementation of Xpert will generate, we attempted to incorporate these concerns into our model by capturing impacts and costs for up to three consecutive visits per patient and including patient and sample loss rates and test failure rates at every step. We also attempted to reduce bias introduced by trial-specific cost and selection by basing most epidemiological parameters on national-level operational data, rather than clinical trial results.
The analysis presented here is limited in several ways. First, our epidemiological model inputs are based on national-level databases which are designed for routine patient care and programme monitoring. Although this allows us to use data from a far larger patient sample than is available in clinical trials and observation studies, mismatching by the linking mechanism in the NHLS database or missing data in both databases could have introduced bias into our analysis. Since the resulting parameter values are similar to those reported by the national TB programme  we do not believe this bias to be large. Second, our analysis is restricted to the full and incremental cost of the new diagnostic algorithm only. As such, it does not fully capture the potential benefits of the Xpert technology, such as longer survival and better quality of life, or its opportunity cost. Third, since the baseline scenario reflects current guidelines for TB diagnosis, the frequency and cost of clinical diagnosis and of treatment initiation without laboratory diagnosis and the potential cost-saving if Xpert diagnosis prevents these patients from being treated might be underestimated. Fourth, we do not include loss to follow-up from treatment, which could be higher in the Xpert scenario as patients who would previously have been lost to follow-up during a longer diagnostic process might now be lost during the early stages of TB treatment. Fifth, as with any model of an intervention that has not yet been brought to scale, our results assume that the impacts and costs seen in the pilot phase of Xpert implementation in South Africa and in small demonstration projects will continue to reflect the impacts and costs when nationwide scale-up is achieved. While we have taken some potential variation into account in our sensitivity analyses, close monitoring of the scale-up will be needed to verify many of the assumptions we have made. Sixth, the algorithm for Xpert diagnosis might change in the years to come as a result of the experience generated during the roll-out. Total diagnostic costs may be reduced once greater confidence in the technology allows for a less conservative algorithm for the management of HIV-positive, Xpert negative patients to be implemented, or as a result of increasing test volumes if Xpert becomes the diagnostic of choice for paediatric TB suspects and suspects with extrapulmonary TB as well. Finally, the role of MDR-TB treatment in driving the total cost of the South African TB program is only partly explored in the model. Our results indicate that costs would nearly double if the existing inpatient model of MDR-TB treatment is not replaced with an outpatient model of care, a policy that has been approved but not yet widely implemented in South Africa.
Limitations notwithstanding, it is clear from our model that nationwide scale-up of Xpert MTB/RIF technology in South Africa will substantially increase the cost of the national TB diagnosis and treatment programmes. It will also vastly increase the number of TB cases diagnosed, MDR-TB cases identified, and patients initiating TB treatment. The importance of this potential benefit in helping to turn the tide against TB in South Africa has been recognised in policy, with the Minister of Health committing in May 2011 to roll Xpert MTB/RIF technology out “to every district in the next six months and to every facility in the next 18 months” . As new data are made available over the course of the national scale-up, we will update the model presented here to continue to guide implementation decisions.
We are grateful to David Mametja, Lindiwe Mvusi, Raymond Mabope for their review of model assumptions and results, to Nevilla Somnath and Sicelo Dlamini for access to the Electronic TB Register, to Sue Candy and Michelle Potgieter for providing data from the National Health Laboratory Service database, to Linda Erasmus and Gerrit Coetzee for their input, and to Mark Nicol for analysing additional, unpublished data from the South African Xpert implementation study to produce central model parameters.
Conceived and designed the experiments: GMR KS SR WS YP. Performed the experiments: GMR KS. Analyzed the data: GMR KS LL WM. Contributed reagents/materials/analysis tools: WS IS SP YP. Wrote the paper: GMR KS LL WM WS IS SP YP SR. Built model: GMR. Parametrised model: GMR KS. Reviewed model outputs: KS IS WS SR YP.
- 1. World Health Organization (2011) Global tuberculosis control. Report. World Health Organization2011Global tuberculosis control.Report
- 2. Scott LE, McCarthy K, Gous N, Nduna M, Van Rie A, et al. (2011) Comparison of Xpert MTB/RIF with Other Nucleic Acid Technologies for Diagnosing Pulmonary Tuberculosis in a High HIV Prevalence Setting: A Prospective Study. PLoS Med 8(7): e1001061.LE ScottK. McCarthyN. GousM. NdunaA. Van Rie2011Comparison of Xpert MTB/RIF with Other Nucleic Acid Technologies for Diagnosing Pulmonary Tuberculosis in a High HIV Prevalence Setting: A Prospective Study.PLoS Med87e1001061
- 3. Getahun H, Harrington M, O’Brien R, Nunn P (2007) Diagnosis of smear-negative pulmonary tuberculosis in people with HIV infection or AIDS in resource-constrained settings: informing urgent policy changes. Lancet 369: 2042–49.H. GetahunM. HarringtonR. O’BrienP. Nunn2007Diagnosis of smear-negative pulmonary tuberculosis in people with HIV infection or AIDS in resource-constrained settings: informing urgent policy changes.Lancet369204249
- 4. Boehme CC, Nabeta P, Hillemann D, Nicol MP, Shenai S, et al. (2010) Rapid molecular detection of tuberculosis and rifampin resistance. N Engl J Med 363(11): 1005–1015.CC BoehmeP. NabetaD. HillemannMP NicolS. Shenai2010Rapid molecular detection of tuberculosis and rifampin resistance.N Engl J Med3631110051015
- 5. Boehme CC, Nicol MP, Nabeta P, Michael JS, Gotuzzo E, et al. (2011) Feasibility, diagnostic accuracy, and effectiveness of decentralised use of the Xpert MTB/RIF test for diagnosis of tuberculosis and multidrug resistance: a multicentre implementation study. Lancet 377: 1495–1505.CC BoehmeMP NicolP. NabetaJS MichaelE. Gotuzzo2011Feasibility, diagnostic accuracy, and effectiveness of decentralised use of the Xpert MTB/RIF test for diagnosis of tuberculosis and multidrug resistance: a multicentre implementation study.Lancet37714951505
- 6. World Health Organization (2010) Tuberculosis Diagnostics: Xpert MTB/RIF Test. WHO Endorsement and Recommendations. Fact sheet. World Health Organization2010Tuberculosis Diagnostics: Xpert MTB/RIF Test.WHO Endorsement and Recommendations. Fact sheet
- 7. Erasmus LK, Coetzee GJ, Stevens WS (2011) Scale up of Xpert MTB/RIF from the national laboratory perspective: issues and challenges. Int J Tuberc Lung Dis 15: 11, Suppl 3S61.LK ErasmusGJ CoetzeeWS Stevens2011Scale up of Xpert MTB/RIF from the national laboratory perspective: issues and challenges.Int J Tuberc Lung Dis1511, Suppl 3S61
- 8. Matsoso LG, Veriava Y, Poswa X, V Gabashane, JM Ratabane, et al. (2010) Validation of a rapid tuberculosis PCR assay for detection of MDR-TB patients in Gauteng, South Africa. South Afr J Epid Inf 25(2): 12–15.LG MatsosoY. VeriavaX. PoswaGabashane VRatabane JM2010Validation of a rapid tuberculosis PCR assay for detection of MDR-TB patients in Gauteng, South Africa.South Afr J Epid Inf2521215
- 9. Parsons LM, Somoskövi A, Gutierrez C, Lee E, Paramasivan CN, et al. (2011) Laboratory Diagnosis of Tuberculosis in Resource- Poor Countries: Challenges and Opportunities. Clin Microbiol Rev 24(2): 314–350.LM ParsonsA. SomosköviC. GutierrezE. LeeCN Paramasivan2011Laboratory Diagnosis of Tuberculosis in Resource- Poor Countries: Challenges and Opportunities.Clin Microbiol Rev242314350
- 10. Chihota VN, Grant AD, Fielding K, Ndibongo B, van Zyl A, et al. (2010) Liquid vs. solid culture for tuberculosis: performance and cost in a resource-constrained setting Int J Tuberc Lung Dis; 14(8): 1024–1031.VN ChihotaAD GrantK. FieldingB. NdibongoA. van Zyl2010Liquid vs.solid culture for tuberculosis: performance and cost in a resource-constrained setting Int J Tuberc Lung Dis;14810241031
- 11. Sinanovic E, Floyd K, Dudley L, Azevedo V, Grant R, et al. (2003) Cost and cost-effectiveness of community-based care for tuberculosis in Cape Town, South Africa. Int J Tuberc Lung Dis 7(9): S56–S62.E. SinanovicK. FloydL. DudleyV. AzevedoR. Grant2003Cost and cost-effectiveness of community-based care for tuberculosis in Cape Town, South Africa.Int J Tuberc Lung Dis79S56S62
- 12. TB Control Programme, National Department of Health, Republic of South Africa (2010) Provincial Quarterly TB Progress Reports, Quarter 1 to 4: TB Control Programme, National Department of Health, Republic of South Africa2010Provincial Quarterly TB Progress Reports, Quarter 1 to4
- 13. National Department of Health, Republic of South Africa (2009) National Tuberculosis Management Guidelines. National Department of Health, Republic of South Africa2009National Tuberculosis Management Guidelines
- 14. World Health Organization (2010) Roadmap for rolling out Xpert MTB/RIF for rapid diagnosis of TB and MDR-TB. World Health Organization2010Roadmap for rolling out Xpert MTB/RIF for rapid diagnosis of TB and MDR-TB
- 15. Actuarial Society of South Africa AIDS Demographic model 2003lite. Available: http://aids.actuarialsociety.org.za. Accessed 28 April 2009. Actuarial Society of South Africa AIDS Demographic model 2003lite. Available: http://aids.actuarialsociety.org.za. Accessed 28 April 2009.
- 16. National Department of Health, Republic of South Africa (2011) National Strategic Plan for HIV and AIDS, STIs and TB, 2012–2016: National Department of Health, Republic of South Africa2011National Strategic Plan for HIV and AIDS, STIs and TB,2012–2016
- 17. WHO - StopTB Planning and Budgeting for TB Control Model. Available: www.who.int/tb/dots/planning_budgeting_tool/en/. Accessed 28 July 2011. WHO - StopTB Planning and Budgeting for TB Control Model. Available: www.who.int/tb/dots/planning_budgeting_tool/en/. Accessed 28 July 2011.
- 18. XE currency conversion rates. Available: www.xe.com/ucc/. Accessed 10 November 2011. XE currency conversion rates. Available: www.xe.com/ucc/. Accessed 10 November 2011.
- 19. National Department of Health, Republic of South Africa (2011) Multi-Drug Resistant Tuberculosis - A Policy Framework on Decentralised and Deinstitutionalised Management for South Africa. National Department of Health, Republic of South Africa2011Multi-Drug Resistant Tuberculosis - A Policy Framework on Decentralised and Deinstitutionalised Management for South Africa
- 20. Meyer-Rath G, Schnippel K, Long L, MacLeod W, Sanne I, et al. (2011) The incremental cost and diagnostic impact of rolling out the GeneXpert MTB/RIF platform in the public sector of South Africa. 42nd Union World Conference on Lung Health 2011, Lille, France. G. Meyer-RathK. SchnippelL. LongW. MacLeodI. Sanne2011The incremental cost and diagnostic impact of rolling out the GeneXpert MTB/RIF platform in the public sector of South Africa.42nd Union World Conference on Lung Health 2011, Lille, France
- 21. UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases/Foundation for Innovative New Diagnostics (2006) Diagnostics for tuberculosis: global demand and market potential. UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases/Foundation for Innovative New Diagnostics2006Diagnostics for tuberculosis: global demand and market potential
- 22. Dowdy DW, O’Brien MA, Bishai D (2008) Cost-effectiveness of novel diagnostic tools for the diagnosis of tuberculosis. Int J Tuberc Lung Dis 12(9): 1021–1029.DW DowdyMA O’BrienD. Bishai2008Cost-effectiveness of novel diagnostic tools for the diagnosis of tuberculosis.Int J Tuberc Lung Dis12910211029
- 23. National Treasury, Republic of South Africa (2010) Budget Review 2010. Treasury NationalRepublic of South Africa2010Budget Review 2010
- 24. Sohn H, Minion J, Albert H, Dheda K, Pai M (2006) TB diagnostic tests: How do we figure out their costs? Expert Rev Anti Infect Ther 7(6): 723–733.H. SohnJ. MinionH. AlbertK. DhedaM. Pai2006TB diagnostic tests: How do we figure out their costs?Expert Rev Anti Infect Ther76723733
- 25. Dowdy DW, Cattamanchi A, Steingart KR, Pai M (2011) Is Scale-Up Worth It? Challenges in Economic Analysis of Diagnostic Tests for Tuberculosis. PLoS Med 8(7): e1001063.DW DowdyA. CattamanchiKR SteingartM. Pai2011Is Scale-Up Worth It? Challenges in Economic Analysis of Diagnostic Tests for Tuberculosis.PLoS Med87e1001063
- 26. Mann GH, Thomson R, Jin C, Phiri M, Vater MC, et al. (2011) : The role of health economics research in implementation research for health systems strengthening. Int J Tuberc Lung Dis 15(6): 715–721.GH MannR. ThomsonC. JinM. PhiriMC Vater2011: The role of health economics research in implementation research for health systems strengthening.Int J Tuberc Lung Dis156715721
- 27. Health Budget Vote Policy Speech presented at the National Assembly by Minister of Health, Dr A Motsoaledi, 31 May (2011) Health Budget Vote Policy Speech presented at the National Assembly by Minister of Health, Dr A Motsoaledi, 31 May2011Available: www.info.gov.za/speech/DynamicAction?pageid=461&sid=18751&tid=34232. Accessed 08 July 2011. Available: www.info.gov.za/speech/DynamicAction?pageid=461&sid=18751&tid=34232. Accessed 08 July 2011.