Modelling Innovative Diagnostic Tools for Tuberculosis

Transcription

1 Modelling Innovative Tools for Tuberculosis Ivor Langley Basra Doulla, MoH, Tanzania Hsien-Ho Lin, NTU, Taiwan Kerry Millington Bertie Squire HaCRIC11 - SEPTEMBER 2011

2 Modelling Innovative Tools for Tuberculosis WHERE?.the archetypal disease of poverty World Health Organisation (2010), Stop, 2010/11 Tuberculosis Global facts 2

3 Agenda BACKGROUND THE PROBLEM & OPPORTUNITY? THE APPROACH RESULTS CONCLUSIONS NEXT STAGES 3

4 Modelling Innovative Tools for Tuberculosis WHY? More than two billion people, equal to one third of the world s population, are infected with bacilli. One in every 10 of those people will become sick with active. A total of 1.7 million people died from in 2009, equal to about 4700 deaths a day. is a disease of poverty with the majority of deaths in the developing world, affecting mostly young adults. This despite their being good and effective drug treatment (6months) World Health Organisation (2010), Stop, 2010/11 Tuberculosis Global facts 4

5 What are the Problems? ACCESS TO DIAGNOSIS Case detection is only around 45% in the highest burden countries Costly for patients to access diagnosis in developing countries Kemp JR, Mann G, Simwaka BN, Salaniponi FM, Squire SB (2007), Can Malawi's poor afford free tuberculosis services? 5 Patient and household costs associated with a tuberculosis diagnosis in Lilongwe. Bull World Health Organ. 2007;

6 What are the Problems? SPEED OF DIAGNOSIS Minimum of 4 visits to a diagnostic centre to receive diagnosis Leading to high default rate during diagnosis ~15% Multi-Drug Resistant (MDR-) diagnosis takes up to 3-4 months is contagious and spreads through the air. If not treated, each person infects people/yr Receive Treatment Medicine At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Smear Positive ing Every 2 wks for Medicines Diagnosed At end of intermediate phases if smear negative Smear Negative Receive Diagnosis Treatment Monitoring No Found Cure Provide Sputum Sample 2 Becomes Sick with cough Health Clinic Suspect Provide Sputum Sample 1 Squire SB, Belaye AK, Kashoti A, Salaniponi FM, Mundy CJ, Theobald S, et al. (2005), 'Lost' smear-positive pulmonary The tuberculosis Liverpool School cases: of Tropical where Medicine are they and why did we lose them? Int J Tuberc Lung Dis. 2005; 9(1):

7 What are the Problems? ACCURACY OF DIAGNOSIS Sputum Smear Microscopy only detects around 30 70% of cases Accuracy is much worse for HIV+ patients (typically 40% of cases) Secondary techniques such as chest x-ray and short course antibiotics treatment are likely to have high false positive rate 25-50% is a leading killer among people living with HIV Hargreaves NJ, Kadzakumanja O, Whitty CJ, Salaniponi FM, Harries AD, Squire SB (2001). 'Smear-negative' pulmonary tuberculosis The Liverpool in School a DOTS of Tropical programme: Medicine poor outcomes in an area of high HIV seroprevalence. IJTLD. 2001; 5:

8 What is the opportunity? New Technologies The introduction and scale-up of new tools for the diagnosis of Tuberculosis () in developing countries has the potential to make a huge difference to the lives of millions of people living in poverty e.g. Solid Culture DST by Proportion Method LED Fluorescence Microscopy Liquid Culture MGIT Line Probe Assay HAIN Xpert M/RIF Small The Liverpool PM, Pai School M. (2010), of Tropical Tuberculosis Medicine diagnosis - time for a game change. N Engl J Med. 2010; 363(11):

9 Modelling Innovative Tools for Tuberculosis OBJECTIVE?. to facilitate effective uptake of the most appropriate diagnostic technology and algorithms in developing countries.. to do this we must address the key questions of policy makers and funders. 9

10 Policy Makers and Funders Questions? Impact Assessment Framework* Patients Health System Community EFFICACY - How well does it work? How many more patients with will be diagnosed? How many more patients will be cured? EQUITY - Who benefits and why? Which patients will benefit? - The poor, HIV+? HEALTH SYSTEM - Operational effects? What will be the effect on staffing, Infrastructure, procurement, bottlenecks and capacity? SCALE-UP - Impacts of national rollout? How will it effect access to care and patient costs? How much will it cost to Implement, maintain, & run? Is it sustainable? How will it effect transmission? & MDR- incidence? prevalence? POLICY - How does it compare to other technologies? How cost effective is it compared to other approaches? Now and in the future? * Mann G, Squire SB, Bissell K, Eliseev P, Du Toit E, Hesseling A, et al. (2010), Beyond accuracy: creating a The Liverpool comprehensive School of Tropical evidence Medicine base for diagnostic tools. Int J Tuberc Lung Dis. 2010; 14(12):

11 Our Approach to these Questions? Some questions could be and are being addressed as part of Laboratory Tests Demonstration Trials Randomised controlled trials Modelling is required to complement these trials to particular address questions of Effects on Patient Outcomes and Health System Infrastructure Impacts of national scale-up Cost Effectiveness Context related operational issues Transmission impacts We decided to use Discrete Event Simulation linked to Transmission Modelling WHY? 11

12 Key Features of Discrete Event Simulation? FEATURE Core Data Driven REQUIREMENT With a limited set of core data available for diagnostic centres in the developing world the models need to be able to run on this limited data with key assumptions discussed with experts and tested through sensitivity analysis. Flexible So that models can be used to understand a variety of diagnostic approaches and across a wide range of different settings in the developing world they must be easily and quickly reconfigurable. Visual In order to provide policy makers with confidence in the model accuracy, its outputs, and to assist in validation the model must be visual and present a simple but realistic schematic image of the health system being modelled. Powerful In order to link to longer term transmission modelling and to speed model runs there must be sufficient processing power to simulate up to 10 years within 1 hour of modelling time. Stochastic To model variables such as patient arrivals, power outages, and default probabilities the modelling tool must have the ability to model uncertainty based on user defined and statistical probability distributions. Interactive What-If scenarios must be quickly and interactively testable. Portable In order to demonstrate and validate the model on location in remote rural diagnostic centres the model must be portable. Transferable In order to give personnel within National Programmes (NTP) the opportunity to experiment and evaluate appropriate tools and impacts for themselves in individual diagnostic centres it must be possible to transfer executable versions of the model for ongoing evaluation within the NTP 12

13 Why this approach rather than the more usual Decision Analysis? A. MODEL OPERATIONAL BOTTLENECKS The approach takes account of the complex interactions in the health system between patients, staff and processes. The resultant bottlenecks affect the outputs e.g. patient default. B. VISUAL- The approach gives a visual representation that enables policy makers to engage with the modelling and assist in validation - its not a black box. C. MATCHES THE W.H.O. OUTPUT REQUIREMENTS OF TRIALS The models produce all the detailed outputs to an Excel spreadsheet specifically required by the WHO for monitoring implementation of Xpert M/RIF D. COMPREHENSIVE - The approach includes patient, health system, and transmission impacts. These are linked e.g. improvements in operational performance affect time to diagnosis which affects incidence, which in turn affects the operation, patient access, and health system costs. E. MULTI-PURPOSE The models can be used to model individual diagnostic centres or country scale-up. The models can therefore inform country level policy decisions or assist in implementation in specific districts 13

14 Modelling Approach? - Design A detailed model was developed using the WITNESS Simulation Software. Patients are modelled through each process from Health Clinics into a and through to Treatment where appropriate. 14

15 Modelling Approach? - Design Sputum samples are modelled through the lab process. The lab model shows a schematic of the process and includes microscopy and Xpert M/RIF with flows defined for each sample by rules defined in an input data files Lab technicians and assistants as well as clinical staff as modelled as resources 15

16 Modelling Approach? - Design Samples from patients where there is a risk of drug resistance are transported to the Central Reference Lab for Drug Sensitivity Testing. 16

17 LAB TESTS COMPLETED TOTAL NEW SUSPECTS RETREAT TREATMENT MONITORING TOTAL Micrsoscopy GeneXpert Micrsoscopy GeneXpert Micrsoscopy NEW SUSPECTS RETREAT TREATMENT MONITORING TOTAL Year Positive Negative Positive Negative Positive Negative Positive Negative Positive Negative Micro GX Total Micro GX Total Micro Micro GX Total RIF Sensitive RIF Resistant RIF Sensitive RIF Resistant RIF Sensitive RIF Resistant RIF Sensitive RIF Resistant RIF+% RIF+% RIF+% QTR % % 0.4% % 0 0.0% % 0.0% % % % % 0.4% % QTR % % 0.4% % 0 0.0% % 0.0% % % % % 0.4% % QTR3 0 LAB 0 TESTS COMPLETED % % 0.3% % 0 0.0% % 0.0% % % % % 0.3% % QTR4 0 TOTAL % % 0.2% % 0 0.0% % 0.0% % % % % 0.2% % QTR NEW SUSPECTS RETREAT TREATMENT 0 MONITORING 0.0% % 0.2% % 0 0.0% % 0.0% % TOTAL % % % 0.2% % QTR Micrsoscopy GeneXpert Micrsoscopy 0 0 GeneXpert Micrsoscopy 0.0% % 0.2% NEW SUSPECTS 18.7% 0 0.0% % 0.0% 6354 RETREAT 100.0% % TREATMENT MONITORING 1.6% % 0.2% TOTAL 11.9% QTR3 1 0 Year 0 Positive Negative Positive 770 Negative 0 0 Positive 7774 Negative 23 0 Positive Negative Positive 0 0.0% Negative Micro 18.8% 0.2% GX 18.8% 0 Total 0.0% 7797 Micro 100.0% 0.3% 7797 GX 100.0% Total 1.7% Micro 1.7% Micro 20.2% 0.2% GX 11.8% Total QTR RIF Sensitive RIF Resistant 785 RIF Sensitive 0 RIF Resistant RIF Sensitive 0 RIF Resistant 0 RIF Sensitive 4222 RIF Resistant % % 0.2% % RIF+% 0 0.0% % 0.3% % RIF+% ve 1.8% rate % % 0.2% % RIF+% QTR1 2 0QTR % % 0.0% 0.2% % 19.2% 0.4% % 17.3% % 0.0% 0.2% % 100.0% 0.0% % 1.7% % 13.0% % 13.0% 0.2% % 11.6% 0.4% % QTR2 2 0QTR % % 0.0% 0.2% % 19.2% 0.4% % 18.7% % 0.0% 0.2% % 100.0% 0.0% % 1.6% % 5.2% % 5.2% 0.2% % 11.4% 0.4% % QTR3 2 0QTR3 0 LAB 0 0 TESTS COMPLETED % % 0.0% 0.2% % 19.1% 0.3% % 19.0% % 0.0% 0.2% % 100.0% 0.0% % 1.6% % 2.6% % 2.6% 0.2% % 11.3% 0.3% % QTR4 2 0QTR4 0 TOTAL % % 0.0% 0.2% % 19.1% 0.2% % 18.7% % 0.0% 0.2% % 100.0% 0.0% % 1.7% % 1.8% % 1.8% 0.2% % 11.2% 0.2% % QTR1 3 0QTR NEW SUSPECTS RETREAT % TREATMENT MONITORING 18.9% 0.0% 0.2% % 18.9% 0.2% % 18.8% % 0.0% 0.2% % 100.0% 0.0% % TOTAL 1.7% % 1.8% % 1.8% 0.2% % 11.2% 0.2% % QTR2 3 0QTR Micrsoscopy GeneXpert Micrsoscopy GeneXpert % Micrsoscopy 19.0% 0.0% 0.2% % 19.0% 0.2% NEW SUSPECTS 0.0% 18.7% % 0.0% 0.2% % 100.0% 0.0% RETREAT 100.0% 1.7% % 1.6% TREATMENT MONITORING 20.3% 1.6% 0.2% % 11.1% 0.2% TOTAL 11.9% QTR3 3 0QTR Year Positive Negative Positive Negative Positive Negative Positive Negative 0 0.0% Positive % 0.0% Negative 0.2% Micro % 19.0% 0.2% GX 0.0% 18.8% Total 100.0% 0.0% 0.2% 7797 Micro % 100.0% 0.3% GX 100.0% 1.6% Total 1.6% 1.7% Micro 20.3% 1.7% 0.2% Micro % 11.1% 0.2% GX 11.8% Total QTR4 3 0QTR RIF Sensitive RIF Resistant RIF Sensitive 0 0 RIF Resistant RIF Sensitive 0 0 RIF Resistant RIF Sensitive RIF Resistant 0 0.0% % 0.0% 0.2% % 19.0% 0.2% % 19.1% RIF+% % 0.0% 0.2% % 100.0% 0.3% % RIF+% 1.6% ve 1.6% 1.8% rate % 1.8% 0.2% % 11.0% 0.2% % RIF+% QTR1 4 0QTR1 20 0QTR % % 0.0% % % 0.0% 19.0% 0.2% % 0.0% 19.2% 0.4% % 0.0% 17.3% 0.4% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.6% % 1.6% 1.7% % 1.7% 13.0% 0.2% % 13.0% 11.0% 0.2% % 11.6% 0.4% % QTR2 4 0QTR2 20 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.2% 0.4% % 0.0% 18.7% 0.4% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.6% % 1.6% 1.6% % 1.6% 5.2% 0.2% % 5.2% 10.9% 0.2% % 11.4% 0.4% % QTR3 4 0QTR3 20 0QTR3 0 LAB TESTS COMPLETED % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.1% 0.3% % 0.0% 19.0% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.7% % 1.7% 1.6% % 1.6% 2.6% 0.2% % 2.6% 10.9% 0.2% % 11.3% 0.3% % QTR4 4 0QTR4 20 0QTR4 0 TOTAL % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.1% 0.2% % 0.0% 18.7% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.6% % 1.6% 1.7% % 1.7% 1.8% 0.2% % 1.8% 10.9% 0.2% % 11.2% 0.2% % QTR1 5 0QTR1 30 0QTR NEW SUSPECTS RETREAT % % 0.0% TREATMENT 0.2% MONITORING % 0.0% 18.9% 0.2% % 0.0% 18.9% 0.2% % 0.0% 18.8% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.7% % TOTAL 1.7% 1.7% % 1.7% 1.8% 0.2% % 1.8% 10.9% 0.2% % 11.2% 0.2% % QTR2 5 0QTR2 30 0QTR Micrsoscopy GeneXpert Micrsoscopy GeneXpert % % 0.0% % Micrsoscopy % 0.0% 19.0% 0.2% % 0.0% 19.0% 0.2% NEW SUSPECTS 100.0% 0.0% 18.7% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.7% RETREAT 100.0% 1.7% 1.7% % 1.7% 1.6% TREATMENT 0.2% MONITORING % 1.6% 10.9% 0.2% % 11.1% 0.2% TOTAL 11.9% QTR3 5 0QTR3 30 0QTR Year Positive Negative Positive Negative Positive Negative Positive % Negative % 0.0% Positive 0.2% % 0.0% Negative 19.0% 0.2% Micro % 0.0% 19.0% 0.2% GX 100.0% 0.0% 18.8% 0.3% Total % 0.0% 100.0% 0.2% 7797 Micro % 100.0% 0.3% 1.7% GX 100.0% 1.7% 1.6% Total 20.3% 1.6% 1.7% 0.2% Micro % 1.7% 10.8% 0.2% Micro % 11.1% 0.2% GX 11.8% Total QTR4 5 0QTR4 30 0QTR RIF Sensitive RIF Resistant RIF Sensitive RIF Resistant RIF Sensitive RIF Resistant RIF Sensitive 0 0.0% RIF Resistant % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.0% 0.2% % 0.0% 19.1% RIF+% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.3% 1.7% % RIF+% 1.7% 1.6% ve 20.3% 1.6% 1.8% rate 0.2% % 1.8% 10.8% 0.2% % 11.0% 0.2% % RIF+% QTR1 6 0QTR1 40 0QTR QTR % % 0.0% % % 0.0% % 0.2% % 0.0% 0.0% 19.0% 0.2% % 100.0% 0.0% 19.2% 0.4% 0.3% % 0.0% 17.3% 100.0% 0.4% % 0.0% 100.0% 0.2% % % 100.0% 0.0% 1.7% 1.6% % 20.2% 1.6% 1.7% 0.2% % 1.7% 13.0% 10.8% 0.2% % 13.0% 11.0% 0.2% % 11.6% 0.4% % QTR2 6 0QTR2 40 0QTR QTR % % 0.0% % % 0.0% % 0.2% % 0.0% 0.0% 18.9% 0.2% % 100.0% 0.0% 19.2% 0.4% 0.2% % 0.0% 18.7% 100.0% 0.4% % 0.0% 100.0% 0.2% % % 100.0% 0.0% 1.7% 1.6% % 20.2% 1.6% 1.6% 0.2% % 1.6% 5.2% 10.8% 0.2% % 5.2% 10.9% 0.2% % 11.4% 0.4% % QTR3 6 0QTR3 40 0QTR QTR % % 0.0% % % 0.0% % 0.2% % 0.0% 0.0% 18.9% 0.2% % 100.0% 0.0% 19.1% 0.3% 0.3% % 0.0% 19.0% 100.0% 0.3% % 0.0% 100.0% 0.2% % % 100.0% 0.0% 1.7% 1.7% % 20.2% 1.7% 1.6% 0.2% % 1.6% 2.6% 10.8% 0.2% % 2.6% 10.9% 0.2% % 11.3% 0.3% % QTR4 6 0QTR4 40 0QTR QTR % % 0.0% % % 0.0% % 0.2% % 0.0% 0.0% 18.9% 0.2% % 100.0% 0.0% 19.1% 0.2% 0.2% % 0.0% 18.7% 100.0% 0.3% % 0.0% 100.0% 0.2% % % 100.0% 0.0% 1.7% 1.6% % 20.3% 1.6% 1.7% 0.2% % 1.7% 1.8% 10.8% 0.2% % 1.8% 10.9% 0.2% % 11.2% 0.2% % QTR1 7 QTR1 5 0QTR1 30 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 18.9% 0.2% % 0.0% 18.8% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.7% % 1.7% 1.7% % 1.7% 1.8% 0.2% % 1.8% 10.9% 0.2% % 11.2% 0.2% % QTR2 7 QTR2 5 0QTR2 30 0QTR % % 0.0% % % 0.0% 19.0% 0.2% % 0.0% 19.0% 0.2% % 0.0% 18.7% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.0% 1.7% % 1.7% 1.7% % 1.7% 1.6% 0.2% % 1.6% 10.9% 0.2% % 11.1% 0.2% % QTR3 7 QTR3 5 0QTR3 30 0QTR % % 0.0% % % 0.0% 19.0% 0.2% % 0.0% 19.0% 0.2% % 0.0% 18.8% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.3% 1.7% % 1.7% 1.6% % 1.6% 1.7% 0.2% % 1.7% 10.8% 0.2% % 11.1% 0.2% % QTR4 7 QTR4 5 0QTR4 30 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.0% 0.2% % 0.0% 19.1% 0.3% % 0.0% 100.0% 0.2% % 100.0% 0.3% 1.7% % 1.7% 1.6% % 1.6% 1.8% 0.2% % 1.8% 10.8% 0.2% % 11.0% 0.2% % QTR1 8 QTR1 6 0QTR1 40 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 19.0% 0.2% % 0.0% 19.2% 0.3% % 0.0% 100.0% 0.4% % 100.0% 0.2% 1.7% % 1.7% 1.6% % 1.6% 1.7% 0.2% % 1.7% 10.8% 0.2% % 11.0% 0.2% % QTR2 8 QTR2 6 0QTR2 40 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 18.9% 0.2% % 0.0% 19.2% 0.2% % 0.0% 100.0% 0.4% % 100.0% 0.2% 1.7% % 1.7% 1.6% % 1.6% 1.6% 0.2% % 1.6% 10.8% 0.2% % 10.9% 0.2% % QTR3 8 QTR3 6 0QTR3 40 0QTR % % 0.0% % % 0.0% 18.9% 0.2% % 0.0% 18.9% 0.2% % 0.0% 19.1% 0.3% % 0.0% 100.0% 0.3% % 100.0% 0.2% 1.7% % 1.7% 1.7% % 1.7% 1.6% 0.2% % 1.6% 10.8% 0.2% % 10.9% 0.2% % QTR4 8 QTR4 6 0QTR4 40 0QTR % % 0.0% % % 0.0% 19.0% 0.2% % 0.0% 18.9% 0.2% % 0.0% 19.1% 0.2% % 0.0% 100.0% 0.3% % 100.0% 0.2% 1.7% % 1.7% 1.6% % 1.6% 1.7% 0.2% % 1.7% 10.8% 0.2% % 10.9% 0.2% % QTR1 9 QTR1 7 QTR1 5 0QTR % % 0.0% 0.2% % 18.9% 0.2% % 18.9% % 0.0% 0.3% % 100.0% 0.2% % 1.7% % 1.7% % 1.7% 0.2% % 10.9% 0.2% % QTR2 9 QTR2 7 QTR2 5 0QTR % % 0.0% 0.2% % 19.0% 0.2% % 19.0% % 0.0% 0.3% % 100.0% 0.2% % 1.7% % 1.7% % 1.7% 0.2% % 10.9% 0.2% % QTR3 9 QTR3 7 QTR3 5 0QTR % % 0.0% 0.2% % 19.0% 0.2% % 19.0% % 0.0% 0.3% % 100.0% 0.2% % 1.7% % 1.6% % 1.6% 0.2% % 10.8% 0.2% % QTR4 9 QTR4 7 QTR4 5 0QTR % % 0.0% 0.2% % 18.9% 0.2% % 19.0% % 0.0% 0.3% % 100.0% 0.2% % 1.7% % 1.6% % 1.6% 0.2% % 10.8% 0.2% % QTR1 10 QTR1 8 QTR1 6 0QTR % % 0.0% 0.2% % 18.9% 0.2% % 19.0% % 0.0% 0.3% % 100.0% 0.4% % 1.7% % 1.6% % 1.6% 0.2% % 10.8% 0.2% % QTR2 10 QTR2 8 QTR2 6 0QTR % % 0.0% 0.2% % 18.9% 0.2% % 18.9% % 0.0% 0.2% % 100.0% 0.4% % 1.7% % 1.6% % 1.6% 0.2% % 10.8% 0.2% % QTR3 10 QTR3 8 QTR3 6 0QTR % % 0.0% 0.2% % 18.9% 0.2% % 18.9% % 0.0% 0.3% % 100.0% 0.3% % 1.7% % 1.7% % 1.7% 0.2% % 10.8% 0.2% % QTR4 10 QTR4 8 QTR4 6 0QTR % % 0.0% 0.2% % 19.0% 0.2% % 18.9% % 0.0% 0.2% % 100.0% 0.3% % 1.7% % 1.6% % 1.6% 0.2% % 10.8% 0.2% % QTR1 9 QTR1 7 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR2 9 QTR2 7 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR3 9 QTR3 7 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR4 9 QTR4 7 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR1 10 QTR1 8 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR2 10 QTR2 8 QTR % % 0.2% % 0 0.0% % 0.2% % % % % 0.2% % QTR3 10 QTR3 8 QTR % % 0.2% % 0 0.0% % 0.3% % % % % 0.2% % QTR4 10 QTR4 8 QTR % % 0.2% % 0 0.0% % 0.2% % % % % 0.2% % QTR1 9 QTR1 7 QTR2 9 QTR2 7 QTR3 9 QTR3 7 QTR4 9 QTR4 7 QTR1 10 QTR1 8 QTR2 10 QTR2 8 QTR3 10 QTR3 8 QTR4 10 QTR4 8 QTR1 9 QTR2 9 QTR3 9 QTR4 9 QTR1 10 QTR2 10 QTR3 10 QTR4 10 Modelling Approach? - Design EXCEL WORKBOOK Data is output to an Excel Workbook every Quarter in line with the World Health Organisation requirements for trials of diagnostics 17

18 Modelling Design - Linking to Transmission Modelling for Scale-Up Patient & Health System Impacts Time to diagnosis Default Rates Cure Rates Incidence Transmission Impacts Lin HH, Langley I, Mwenda R, et al. (2011), A modelling framework to support the selection and implementation of new tuberculosis diagnostic tools. Int J Tuberc Lung Dis 15(8): , doi: /ijtld

19 Modelling Approach? Data Collation & Model Calibration Directed by sponsor to test approach in Tanzania. One of the 22 Highest burden countries Measure Count Population 44million New cases per year 80k p.a. cases Treated 64k p.a. HIV+% in Incidence 47% Treatment Cure Rate 87% Health Facility Count Central Reference Lab 1 Zonal Laboratories 2 Regions 27 Districts 168 Laboratories >900 Sources Basra Doulla, Ministry of Health, Dar es Salaam, Tanzania 19

20 Modelling Approach? Data Collation & Model Calibration Minimum Data Requirements for each diagnostic district Parameter Example Data from Tanzania Current Tools ZN Microscopy & X-ray Laboratory Staff 2 lab technicians cases per year New Smear +ve 560 New Smear -ve 450 Retreatment 93 HIV+ rate in Cases 42.0% Smear +ve Rate - New Suspects 11.4% Assumptions were validated with experts, published literature, and could be subject to sensitivity analysis e.g. Accuracy (Sensitivity and Specificity) of the alternative tests suspect and treatment default rates processes and treatment times Costing information 20

21 OPTION 1 - BASE CASE ZN Microscopy At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Treatment Monitoring 30mins Becomes Sick with cough Receive Treatment Medicine Smear Positive 11.4% ing Every 2 wks for Medicines At end of intermediate phases if smear negative 10.0% Diagnosed Smear Negative Receive Diagnosis No Found Cure 78.6% 30mins 30mins Provide Sputum Sample 2 Health Clinic Suspect Provide Sputum Sample 1 21

22 OPTION 2 LED Fluorescence Microscopy At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Treatment Monitoring 20mins Becomes Sick with cough $1,100 per micro Receive Treatment Medicine Smear Positive 13.4% ing Every 2 wks for Medicines At end of intermediate phases if smear negative 7.9% Diagnosed Smear Negative Receive Diagnosis No Found Cure 78.7% 20mins 20mins Provide Sputum Sample 2 Health Clinic Suspect Provide Sputum Sample 1 $1.50 per sample 22

23 OPTION 3 Xpert M/RIF Microscopy Receive Treatment Medicine At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Smear Positive 20.2% ing Every 2 wks for Medicines At end of intermediate phases if smear negative 1.6% Diagnosed Smear Negative Receive Diagnosis Treatment Monitoring No Found 20mins Cure 78.2% Becomes Sick with cough Health Clinic 131mins Suspect Provide Sputum Sample 1 $17,000 per machine $10.70 per sample FIND (2010), FIND negotiated prices for Xpert M/RIF and Country list, (Accessed June 2011, about/whatwedo/ successes/find-negotiated-prices/xpert_mtb_rif.html 23

24 OPTION 4 Xpert for HIV+ / LED for HIV- At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Treatment Monitoring Becomes Sick with cough HIV- Receive Treatment Medicine Smear Positive ing Every 2 wks for Medicines Diagnosed At end of intermediate phases if smear negative Cure No Found Health Clinic Suspect Provide Sputum Sample 1 $1,100 per micro $1.50 per sample Smear Negative Receive Diagnosis Provide Sputum Sample 2 At end of intermediate phases if smear Positive Test For Drug Resistance and put on MDR - Treatment if found Treatment Monitoring Becomes Sick with cough HIV+ Receive Treatment Medicine Smear Positive ing Every 2 wks for Medicines Diagnosed At end of intermediate phases if smear negative Smear Negative Cure No Found Health Clinic Suspect Provide Sputum Sample 1 $17,000 per machine $10.70 per sample Receive Diagnosis 24

25 OUTPUTS Example Patient Outcomes - for 1 diagnostic district in Tanzania Time to Diagnosis (DAYS) ZN LED Xpert Xpert/LED -39% -29% -9% Default rates ZN LED Xpert Xpert/LED Default, 15.6% Default, 13.9% Default, 10.7% Default, 10.7% Complete, 84.4% Complete, 86.1% Complete, 89.3% Complete, 89.3% 25

26 OUTPUTS Patient Outcomes of Patients Cured each Year ZN LED +5%% Xpert +16% Xpert/LED +11% ,000 1,200 26

27 OUTPUTS Example Health System Outcomes of Sputum Samples Analysed p.a. Running Costs ($k p.a.) ZN ZN LED LED +0.2% Xpert Xpert +49% Xpert/LED Xpert/LED +27%

28 PATIENT OUTCOMES ZN XPERT/LED MICROSCOPY M/RIF AROUND 69.5% 77.2% 73.6% 66.6% OF PATIENTS/SUSPECTS SEEKING DIAGNOSIS ARE CORRECTLY DIAGNOSED AND TREATED FOR 28

29 OUTPUTS Cost Effectiveness Analysis Incremental Cost Effectiveness can be used to compare interventions Difference to Base Case A LED Micro. B Xpert Full C Xpert Part Incremental DALY averted/ yr 399 1, (95% Confidence Intervals) ( ) ( ) ( ) Incremental costs/ Additional Cure $0.8 (95% Confidence Intervals) ( ) Incremental costs/ DALY averted (95% Confidence Intervals) Sensitivity of Incremental costs/ DALY averted to Xpert cost per test $16.9 (max) rather than $10.7 (min) $0.08 ( ) $0.08 ( ) $353 ( ) $38.3 $38.3 ( ) $67.8 ( ) +77% $282 ( ) $30.3 ( ) $48.4 ( ) +60% Disability Adjusted Life Year (DALY) averted is used to measure the success of the intervention in reducing the number of years of life lost and the number of years lived with disability. The incremental cost divided by the DALY averted gives an incremental cost ratio 29

30 Conclusions Discrete Event Simulation of options for diagnostics in the developing world:- is a practical and effective approach provides comprehensive information can be linked to transmission models engages policy makers in validation and questioning enables complex interactions to be taken into account can take longer to develop than some other approaches, but once developed can easily be adapted to other contexts 30

31 What Next? Model impacts of Scale-up across Tanzania Implement the model in the Ministry of Health in Tanzania for modelling individual diagnostic centres Apply the modelling approach to other contexts Malawi Brazil High MDR- settings Lin HH, Langley I, Mwenda R, et al. (2011), A modelling framework to support the selection and implementation of new tuberculosis diagnostic tools. Int J Tuberc Lung Dis 15(8): , doi: /ijtld

32 Acknowledgment USAID YaDiul Mukadi Tanzania NTLP Dr Saidi Egwaga Basra Doulla Malawi MOH Reuben Mwenda Lanner Group Geoff Hook Liverpool School of Tropical Medicine Bertie Squire Gillian Mann Kerry Millington Harvard School of Public Health Ted Cohen Megan Murray 32