Cover © International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Internet: www.worldbank.org; Telephone: 202 473 1000 This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in this work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or other partner institutions or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 4.0 International (CC BY 4.0) unported license https://creativecommons.org/licenses/by/4.0/. Under the Creative Commons Attribution license, you are free to copy, distribute and adapt this work, including for commercial purposes, under the following conditions: Attribution – Please cite the work as follows: Optimizing Investments in Mozambique’s Tuberculosis Response: Results of a Tuberculosis Efficiency Study. 2019. Washington DC: World Bank. License: Creative Commons Attribution CC BY 4.0 Translations – If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in its translation. All queries on rights and licenses should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington DC, 20433, USA; fax: 202-522-2625; email: pubrights@ worldbank.org. OPTIMIZING INVESTMENTS IN MOZAMBIQUE’S TUBERCULOSIS RESPONSE: RESULTS OF A TUBERCULOSIS EFFICIENCY STUDY National TB Program, Mozambique Pereira Zindoga, Jorge Jone, Ivan Manhica World Bank Lung Vu, Elina Pradhan, Nicole Fraser-Hurt, Nejma Cheikh, Humberto Cossa, Ronald Mutasa, Zara Shubber, Marelize Görgens, David Wilson, Theo Hawkins (cover and content design) University College London (Optima Consortium for Decision Sciences partner) Tom Palmer, Gerard Joseph Abou-Jaoude, Ibrahim Abubakar, Lara Goscé, Hassan Haghparast-Bidgoli (author), Jolene Skordis Burnet Institute Romesh Abeysuriya, David Kedziora, Rowan Martin-Hughes, David P. Wilson This page is for collation purposes only. CONTENTS TABLE OF TABLE OF CONTENTS ACRONYMS............................................................................................................................................. ix EXECUTIVE SUMMARY KEY MESSAGES..................................................................................................................................... xi EXECUTIVE SUMMARY ................................................................................................................... xiii Current status and future projections of the TB epidemic in Mozambique ....................................... xiii TB care cascades.................................................................................................................................. xiii INTRODUCTION Optimized allocations............................................................................................................................xiv Scale-up scenarios.................................................................................................................................xiv 1 Recommendations .................................................................................................................................xv 1 INTRODUCTION................................................................................................................................. 1 1.1 Overview of the TB epidemic ....................................................................................................... 1 POLICY QUESTIONS & METHODOLOGY 1.2 Rationale for allocative efficiency analysis................................................................................... 4 2 2 POLICY QUESTIONS AND METHODLOGY.................................................................................. 7 2.1 Policy questions............................................................................................................................. 7 2.2 Methodology.................................................................................................................................. 8 3 RESULTS ............................................................................................................................................15 RESULTS 3.1 Demographic trends..................................................................................................................... 15 3 3.2 Past and future TB epidemic trends............................................................................................ 17 3.3 Impact of meeting national and international TB care targets on the TB epidemic?................. 22 3.4 What is impact of reaching 90% ART coverage on future HIV incidence?............................... 26 CONCLUSION 3.5 How to optimally allocate currently available resources for TB treatment?.............................. 27 4 4 CONCLUSIONS...................................................................................................................................37 5 DISCUSSION OF KEY FINDINGS...................................................................................................39 6 RECOMMENDATIONS.....................................................................................................................43 KEY FINDINGS ANNEXES................................................................................................................................................45 REFERENCES......................................................................................................................................... 60 5 ANNEXES RECOMMENDATIONS 1 Optima TB model features and key definitions at a glance........................................................ 47 2 Tuberculosis model structure ...................................................................................................... 49 6 3 Population sizes............................................................................................................................ 51 4 Births and background (non-TB) mortality................................................................................. 51 5 TB epidemiological parameters................................................................................................... 52 6 Number of notified pulmonary TB cases by age and drug resistance type (2017)..................... 53 ANNEXES v 7 Calibration methodology uncertainties in disaggregating tuberculosis incidence by HIV status............................................................................................................... 53 8 Epidemiological assumptions...................................................................................................... 54 9 Programmatic data: screening and diagnostics........................................................................... 54 10 Sensitivity of screening /testing methods.................................................................................... 55 11 TB Treatment interventions: target groups, unit costs, volume, total spend and outcome......... 55 12 Component costs of TB treatment regimens (USD)................................................................... 56 13 Cost of MDR-TB regimen (short course)..................................................................................... 56 14 Cost of MDR-TB regimen (long course)..................................................................................... 56 15 Cost of XDR-TB regimen............................................................................................................ 57 16 Constraints for program funding................................................................................................. 57 17 Number of notified pulmonary TB infections per population group (2017)............................... 57 18 Modelled TB incidence per 100k by population group............................................................... 58 19 Modelled latent infection and deaths (2017)................................................................................ 58 20 Defining new and relapse cases in the model.............................................................................. 58 21 Defining treatment outcomes in the model................................................................................. 58 22 Optimal allocation of Mozambique’s TB expenditure – screening............................................ 59 23 Optimal allocation of Mozambique’s TB expenditure – treatment and prevention (100% of current spending)....................................................................................... 59 FIGURES 0.1 TB care cascade ......................................................................................................................... xiv 0.2 Optimized allocations of the current TB funding ...................................................................... xv 1.1 Trends in TB notification and detection rate................................................................................. 2 1.2 Treatment outcomes for DS-TB..................................................................................................... 3 1.3 Allocative efficiency in the TB response....................................................................................... 4 2.1 The Optima approach to TB modelling......................................................................................... 8 3.1 Projected demographic trends in Mozambique for children aged 0–14 years (2002–35)........... 15 3.2 Projected demographic trends in Mozambique for HIV-negative and HIV-positive adults aged 15 years and older (2002–35).............................................................. 16 3.3 Total population .......................................................................................................................... 16 3.4 Prevalence of pulmonary TB per 100 thousand ......................................................................... 18 3.5 Projected future TB incidence (a–c) ........................................................................................... 19 3.6 Projected future DR-TB (total cases) .......................................................................................... 20 3.7 Model-derived total latent prevalence in Mozambique, 2005–35 (% of total population)......... 20 vi | CONTENTS CONTENTS TABLE OF 3.8 TB care cascade (2017; all cases) ................................................................................................ 21 3.9 MDR-TB treatment cascade (2017)............................................................................................. 22 3.10 TB Care Cascade by HIV status.................................................................................................. 22 EXECUTIVE SUMMARY 3.11 Modelled impact of reaching 90% diagnosis rate by 2025 on the number of people with active pulmonary TB (TB prevalence) (2001–35)............................................... 24 3.12 Modelled impact of reaching 90% diagnosis rate by 2025 on the annual number of pulmonary TB-related deaths (2001–35)................................................................... 24 3.13 Total modelled number of people with DR-TB........................................................................... 25 INTRODUCTION 3.14 Projected impact of meeting NTP care cascade targets on the number of people with pulmonary a. DR-TB and b. XDR-TB (2015–35)................................... 25 1 3.15 Estimated impact of reaching a. different ART coverage levels and b. 90% ART coverage levels on the annual number of new active pulmonary TB infections (2017–35)............................................................................................ 26 POLICY QUESTIONS & METHODOLOGY 3.16 Importance of ART coverage for future a. TB-related deaths and b. TB incidence in HIV-negative adults...................................................................................... 27 2 3.17 Overview of estimated TB expenditure (2017)........................................................................... 28 3.18 Optimizing Mozambique’s TB program funding allocations..................................................... 29 3.19 Optimal reallocation of current TB screening and prevention expenditure to minimize active pulmonary TB prevalence between 2017 and 2035 in Mozambique............... 31 RESULTS 3.20 Optimal reallocation of current TB treatment expenditure to minimize 3 active pulmonary TB prevalence between 2017 and 2035 in Mozambique ............................... 32 3.21 Estimated number of people with active pulmonary TB ........................................................... 32 3.22 Estimated number of annual TB-related deaths ......................................................................... 33 CONCLUSION 3.23 Programs funded under different amounts of spending.............................................................. 33 4 3.24 Impact of different amounts of expenditure on TB prevalence ................................................. 34 3.25 Impact of different amounts of expenditure on TB-related mortality........................................ 35 KEY FINDINGS A.2.1 Schematic diagram of the health state structure of the model.................................................... 49 A.7.1 Uncertainties in disaggregating tuberculosis incidence by HIV status...................................... 53 5 A.7.1 Uncertainties in disaggregating tuberculosis incidence by HIV status...................................... 53 RECOMMENDATIONS TABLES 1.1 Mozambique: NTP budget by source of financing (2017)............................................................. 4 6 2.1 Mozambique: total estimated expenditure on the TB programmatic response included in this analysis by area (2017)......................................................................................... 9 2.2 Model parameterisation............................................................................................................... 10 2.3 National and international TB care cascade targets.................................................................... 13 ANNEXES CONTENTS | vii 3.1 Model estimates of number and prevalence of active pulmonary TB infections by sub-population (2017)................................................................................................................... 17 3.2 Model estimates of active pulmonary TB incidence, latent infections and TB-related deaths, by sub-population (2017).............................................................................. 17 3.3 Modelled pulmonary TB incidence per 100,000 in Mozambique, by sub-population (2017, 2025 and 2035)......................................................................................... 19 3.4 Scenario parameters: Improved TB screening/testing................................................................ 23 3.5 Scenario parameters: Improved treatment outcomes.................................................................. 23 3.6 Estimated unit cost per person (treatment).................................................................................. 29 3.7 Current and optimal allocations of 2017 TB spending, by intervention (in million USD)......... 30 3.8 Estimated impact of reaching 90% coverage of screening for key risk groups.......................... 31 3.9 Impact of different amounts of expenditure on TB outcomes ................................................... 35 viii | CONTENTS CONTENTS TABLE OF ACRONYMS ART Antiretroviral therapy EXECUTIVE SUMMARY BCG Bacillus Calmette-Guérin CHW Community health worker DOT Directly observed treatment INTRODUCTION DR Drug resistant DS Dug susceptible 1 HIV Human immunodeficiency virus LTBI Latent TB infection POLICY QUESTIONS & METHODOLOGY MDR Multi-drug resistant NSP National Strategic Plan 2 NTP National Tuberculosis Programme PEPFAR U.S. President's Emergency Plan for AIDS Relief PHC Primary health care RESULTS PLHIV People living with HIV 3 RR Rifampicin resistant TB Tuberculosis UN United Nations CONCLUSION USD United States Dollar 4 XDR Extensively drug-resistant Xpert GeneXpert MTB/RIF, detecting DNA sequences specific for M. tuberculosis and rifampicin resistance KEY FINDINGS WHO World Health Organization 5 RECOMMENDATIONS 6 ANNEXES CONTENTS | ix This page is for collation purposes only. CONTENTS TABLE OF KEY MESSAGES 1. Epidemiological projections using the Optima TB model indicate a declining future trend EXECUTIVE SUMMARY in TB incidence, TB prevalence and TB-related deaths in Mozambique. However, the total number of new cases is projected to remain fairly stable due to the rapid increase in population size 2. The country is currently working on the 2020–25 strategic plan with a goal of reaching TB diagnosis rate of 90%. Our model suggests that reaching this 90% diagnosis target INTRODUCTION by 2025 and sustaining it would, by 2035 reduce the total number of TB cases and TB-related deaths by up to 80% and reduce DR-TB by up to 50% 1 3. An optimized allocation of resources could have a substantial impact on key TB indicators. Relative to the current allocation, an optimized allocation of spending could, in 2025: – Reduce the annual number of active TB cases by 19% POLICY QUESTIONS & METHODOLOGY – Reduce the annual number of TB-related deaths by 18% – Reduce the rate of TB incidence per 100,000 by 11% 2 4. Under an optimized allocation of current TB expenditure, TB treatment would receive more funding (in relative to the current condition) and would absorb approximately 28% of total TB spending in Mozambique compared with the current 19%, enabling the NTP to treat the additional TB cases identified through accelerated case finding 5. RESULTS Latest available data suggests that the proportion of TB notifications that are clinically 3 diagnosed is high (60%; 2017). This hinders accurate assessment of the progress made in improving case detection rate in Mozambique. Expanding the use of rapid and sensitive drug resistance testing methods such as GeneXpert would strengthen accurate TB diagnosis for both DS and DR-TB. The effectiveness, cost, and operational aspects of universal GeneXpert use in TB suspects could be investigated through implementation CONCLUSION science research for decision support. 4 6. An optimized allocation of resources would decrease funding for screening and diagnosis programs from 25.4 million (2017 level) to 23.9 million USD by relocating resources to cheaper and more effective case findings activities: screening PLHIV at outpatient visits, expanding initiatives such as cough monitors, reducing community KEY FINDINGS outreach case findings (e.g., general house to house screening), increase coverage of programs targeting key high-risk groups such as prisoners, and doubling coverage of 5 household contact tracing. In contrast, an optimized allocation would increase funding for TB treatment programs from 10.6 million to 11.9 million USD to ensure that the cases found with this combination of efficient diagnosis programs can also get treatment. RECOMMENDATIONS 7. Although mainly managed by the HIV program, ART provision is a key factor in the TB response as treatment reduces the probability of progression from latent to active TB. For 6 example, scaling up to 90% ART coverage by 2035 is projected to reduce the number of new TB cases among adult PLHIV in 2035 by 35% (compared to an ART coverage of 70%) ANNEXES xi This page is for collation purposes only. CONTENTS TABLE OF Mozambique is one of the 30 highest TB burden countries in the world Reaching and sustaining a 90% diagnosis rate by 2025 would reduce the total number of TB cases and TB-related deaths by up to 80% and reduce DR-TB by up to 50% by 2035. EXECUTIVE SUMMARY Mozambique can cut TB prevalence and TB deaths by 20%, and TB incidence by 11% by allocating resources optimally. Increasing investments in active case finding programs are essential to improve the estimated case detection rate of 52%. INTRODUCTION EXECUTIVE SUMMARY 1 CURRENT STATUS AND FUTURE PROJECTIONS OF THE TB POLICY QUESTIONS & METHODOLOGY EPIDEMIC IN MOZAMBIQUE Mozambique is one of the 30 highest tuberculosis (TB) burden countries in the world with an estimated 2 551 new TB cases per 100,000 population, a low TB detection rate of just above 50%, and a low DR-TB diagnosis and treatment success rate (38%) (Global TB Report, 2018). Overall, historical data from the National Tuberculosis Programme (NTP) shows an increasing trend in TB case detection rate (36% in year 2010 to 53% in 2017) coupled with a high treatment success rate for drug susceptible tuberculosis (DS-TB) (90%). However, it is still far from the 2020 global target of identifying and placing 90% of all RESULTS TB patients on treatment, indicating challenges for the country to reach NTP targets for 2025 as well as 3 End-TB targets for 2035. Epidemiological projections using the Optima TB model indicate a declining future trend in TB incidence, TB prevalence and TB-related deaths in Mozambique. At the time of this analysis, the country was working on the 2020–25 Strategic Plan with a goal of reaching 90% TB diagnosis rate. CONCLUSION Our model outputs suggest that reaching and sustaining a 90% diagnosis rate by 2025 would reduce the total number of TB cases and TB-related deaths by up to 80% and reduce DR-TB by up to 50% 4 by 2035. Furthermore, reaching 90% ART coverage (at 55% in 2017) by 2035 would contribute to a reduction of new TB cases of 35%. To reach the (global) END-TB targets, the country must (i) increase TB detection rates, and at the same time, (ii) improve ART coverage and (iii) reduce HIV infection rates. Mozambique has already achieved high coverage of HIV services in TB clinics and among TB KEY FINDINGS patients with 97% of TB patients registered aware of their HIV status and 95% of TB patients living with HIV on ART. Overall, ART coverage was at 55% in 2017 (UNAIDS, 2017). 5 TB CARE CASCADES We estimated that 51% of active TB cases got diagnosed, using routine data available in 2016/17 RECOMMENDATIONS (Figure 0.1). Among the diagnosed cases, 93% initiated treatment, and of those initiated treatment, 86% completed treatment. As a result, despite high treatment success rates, only about 41% of all new active TB cases successfully completed treatment. 6 ANNEXES xiii Figure 0.1 TB care cascade 150k 140k 130k 120k New active TB cases 2017 (k = thousands) 100k 90k 80k 70k 51% 60k 93% 50k 86% 40K 30k 20K 10k 0 Active TB Diagnosed Initiated Completed treatment (47%) treatment (41%) Source: Populated Optima TB model, Mozambique. Note: TB = tuberculosis. OPTIMIZED ALLOCATIONS Assuming same amount of funding as for 2017 (USD 36.8 million) remains available each year up to 2035, the optimized budget allocation differs from current allocations in several areas, including: • Increasing annual funding for outpatient screening from USD 10.8 million to USD 13.2 million, including regular TB screening for all PLHIV at outpatient clinics (currently at 74% coverage) • Prioritizing investments in case-finding programs for key risk groups, such as prisoners and health workers • Increasing funding for treatment programs to accommodate increased number of notifications SCALE-UP SCENARIOS f Mozambique can cut TB prevalence and TB deaths by 20%, and TB incidence by 11% by allo- cating resources optimally. Specifically, this can be done by (i) doubling the rate of household contact tracing for notified cases, (ii) screening all PLHIV during their routine outpatient visits, and (iii) focusing on the community outreach activities among key populations such as prisoners, cross-border miners and community health workers. f Compared to current resource allocation across programs, the optimized resource allocation scenario indicates a need to significantly increase active case finding programs. Increasing invest- ments in active case finding programs are essential to improve the estimated case detection rate of 52%. In addition to allocative efficiency arguments, there is also an equity argument for funding active case finding programs, as it means that populations targeted by outreach activities would receive care that would otherwise not be available to them. f Changes across screening, diagnosis and prevention interventions: An optimized allocation of resources would result in increased funding for screening, diagnosis and intervention programs xiv | EXECUTIVE SUMMARY CONTENTS TABLE OF from 23.9 million to 25.4 million USD and increase funding for treatment from 10.6 million to 11.9 million USD. f Shifts within treatment interventions: In an optimized intervention mix, TB treatment would receive more funding and would absorb approximately 28% of total TB spending in Mozambique EXECUTIVE SUMMARY compared with the current 19%. Figure 0.2 Optimized allocations of the current TB funding INTRODUCTION 1 POLICY QUESTIONS & METHODOLOGY 2 RESULTS 3 Source: Populated Optima TB model, Mozambique. CONCLUSION Note: BCG = Bacille Calmette-Guerin; DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. 4 RECOMMENDATIONS Based on the key findings, we make the following recommendations: 1 Intensify contact-tracing of notified cases KEY FINDINGS The number of traced contacts should increase by 50–100% the current coverage (estimated 1.32 contacts traced per 1 notified case). In addition, the yield data of contact-tracing of 5 notified cases for Mozambique varied highly and should be monitored to inform future program planning and similar exercise. 2 Screen all PLHIV at each outpatient visit RECOMMENDATIONS An estimated 74% of new cases enrolled in HIV treatment were screened for TB in 2017. Given high TB prevalence in PLHIV and low costs, screening all HIV patients regularly for 6 TB as per global recommendation, will likely improve TB case finding and save costs. 3 Expand active case finding programs for high-risk populations Health care workers, prisoners, and cross-border miners should be screened annually, as these groups are at higher risk of TB infection. ANNEXES EXECUTIVE SUMMARY | xv 4 Increase yield from outpatient screening activities About 75% of notified cases are found in the outpatient setting in Mozambique. As only 60% of cases are currently clinically diagnosed (2017), the NTP should aim to increase the proportion of TB cases that are bacteriologically confirmed by increasing coverage of rapid diagnostic tests such as GeneXpert. Without increasing bacteriological confirmation rate, it is difficult to assess true progress in the TB response in Mozambique. We recommend small-scale implementation science research project to assess the effectiveness, cost, and operational aspects of an implementation of universal use of GeneXpert. 5 Closely monitor the pilot of MDR regimens to inform future MDR treatment recommendations The NTP is planning to move away from all injectable based regimens starting in late 2019. The oral short course regimen is currently being studied in several high volume MDR-TB sites in the Maputo City area. By the time of this analysis, data from this research was not yet available. Final discussions about regimen and protocol should be based on operational research data available from studies such as the ongoing on in Maputo city. 6 Expand ART coverage, aiming to reach UNAIDS/WHO’s 95-95-95 targets by 2030 (new global targets) Increasing ART coverage has a significant impact on TB incidence amongst PLHIV. Continued expansion of ART care, including adherence support to keep people in care and achieve and maintain viral suppression, will significantly reduce the number of new active TB infections. 7 Continue ambulatory-focused care for both DS-TB and DR-TB patients The country should continue this approach and avoid unnecessary hospitalisation. This approach reduces costs without affecting outcomes, provided directly observed treatment (DOTS) is in place. 8 Collect data on community case finding programs to inform decision making Community case finding programs contribute around 25% of notified cases in Mozambique. Currently, several community case-finding programs are underway with fragmented implementation and lack of data on cost, coverage, and impact. We recommend the NTP collects data, especially cost data, of these pilot projects in order to inform cost-effectiveness estimates and guide future policy recommendations. 9 Maximize the collection and use of TB routine data to inform programming and policies The quality and availability of cost and coverage data in Mozambique provided a challenge for this analysis and results in uncertainty in model parameters. Key data sources include tracking of TB-specific expenditures, reporting of how TB cases are identified (by intervention modality) and keeping better records of often fragmented implementation of activities. Therefore, monitoring and evaluation systems should be streamlined, and spending and coverage data should be collected for all TB programs (NTP led and non-NTP led) 10 More funding is needed for TB programs The TB response in Mozambique is not on track to meet the 2025 milestones or 2035 End-TB targets—a revised target or timeline and more funding for the TB program are needed. xvi | EXECUTIVE SUMMARY CONTENTS TABLE OF TB has been diagnosed in the country predominantly through passive case-finding programs, meaning that TB is only diagnosed after a person seeks healthcare. Active case finding in Mozambique has been expanding and community-based efforts now account for around 25% of detected cases. The primary objective of the national TB response therefore remains to increase EXECUTIVE SUMMARY the case detection rate, which was estimated at just 53%. Allocative efficiency modelling focused on a national level analysis to provide guidance on how to use existing resources and identify TB care models towards maximizing the impact of the TB response in Mozambique. INTRODUCTION 1 INTRODUCTION 1 M ozambique is located in Southern Africa bordering Tanzania, Malawi, Zambia, Zimbabwe, POLICY QUESTIONS & METHODOLOGY South Africa, and eSwatini. Mozambique’s estimated population is about 28 million (UN Population Division, 2017) and the majority of its population lives in rural areas. Despite 2 recent economic growth, it is estimated that about half of the population lives under the poverty line. Main economic activities in the country include agriculture, fishing tourism, mining and oil industry. The whole country has about 1,600 health facilities and 398 laboratories located in a vast territory of 800.000 km2. 1.1 OVERVIEW OF THE TB EPIDEMIC RESULTS 3 Mozambique is amongst the 30 high burden countries for TB, TB/HIV and MDR-TB. The Global TB Report 2018 estimated that Mozambique has (i) a TB incidence of 551 TANZANIA MALAWI Challenges cases per 100,000 population, (ii) MDR-TB incidence ZIMBABWE Case for optimization CONCLUSION ESWATINI of 30 per 100,000 population, (iii) treatment coverage S. AFRICA of just 52%, and (iv) 40% TB patients living with HIV Estimated TB incidence of 4 (WHO, Global TB report 2018). There is a gap between 551 cases per 100,000 the estimated TB burden in the community and the case population detection by the National Tuberculosis Program (NTP), for both drug susceptible TB (DS-TB) and multi-drug MDR-TB incidence of 30 per KEY FINDINGS resistant TB (MDR-TB). It is estimated that 48% of 100,000 population incident TB cases are undetected each year and only 5% 5 of MDR-TB cases receive a correct diagnosis (WHO, Treatment coverage Global TB report, 2018). Additionally, treatment success of just 52% rates for MDR-TB are low at 48%. Among new TB cases, an estimated 3.7% are MDR whereas among previously 40% TB patients living RECOMMENDATIONS treated cases, about 20% are MDR. The NTP has been with HIV able to achieve and maintain the treatment success rates 6 at 90% for DS-TB (2018 Global TB Report). Estimated 48% of TB incidence 5% TB case finding programs undetected each year TB has been diagnosed in the country predominantly through passive case-finding programs, meaning that Only 5% of MDR-TB cases TB is only diagnosed after a person seeks healthcare. An receive a correct diagnosis ANNEXES 1 estimated 75% of cases were notified in this way. Passive case-finding generally results in diagnosis at a later stage of disease than diagnosis through active case-finding. Ongoing efforts are being made to train healthcare workers to screen people who are attending for non-TB related health conditions. These healthcare workers are called “cough monitors” or cough officers. A more proactive approach to TB diagnosis at this level could facilitate more diagnoses at an earlier stage of disease and reduce the time of TB transmission prior to diagnosis. Figure 1.1 Trends in TB notification and detection rate 180,000 100% 159,000 163,000 160,000 150,000 154,000 140,000 140,000 140,000 130,000 130,000 80% Number of TB cases TB detection rate 120,000 53% 60% 100,000 46% 80,000 38% 38% 39% 40% 36% 37% 86,515 40% 60,000 73,470 59,075 61,576 40,000 47,459 52,642 53,585 20% 46,174 20,000 0 0% 2010 2011 2012 2013 2014 2015 2016 2017 Year Total TB cases noti ed Incidence estimate by WHO Detection rate Source: NTP TB desk review, 2018. Note: Detection rate increased from 36% (2010) to 53% (2017); TB = tuberculosis; WHO = World Health Organisation. Active case finding in Mozambique has been expanding and community-based efforts now account for around 25% of detected cases. This involves both contact tracing of notified cases, and other community-level interventions such as active house-to-house case finding in TB hotspots. However, there are opportunities for further expansion. For example, currently, just 1.3 contacts are traced per notified case on average, whereas the average household size is 4.4 (WHO LTBI Dataset). Other smaller active case finding programs focus on key populations of prisoners, miners, health workers and PLHIV. It is estimated that TB screening is carried out in 74% of newly registered HIV patients. All new enrolees should be screened for TB, and there should be regular routine TB screening for all HIV patients. Similarly, there are an estimated 50,000 healthcare workers in Mozambique, of which approximately only 27% were screened for TB in 2017 (NTP Desk Review, 2018). Although screening for TB has been conducted annually in prisons, this also does not cover all prisoners—it was estimated that only a small proportion of prisoners (15%) received a bacteriological exam for TB in 2017. For miners, the World Bank is supporting screening programs in two regions—Gaza and Maputo. However, the total cases found from these programs was small (544 cases in 2017). In all of the above screening settings, there remain issues of bacteriological confirmation and lab capacity. Around 60% of pulmonary TB notifications were clinically diagnosed in 2017, a proportion which has increased from around 51% in 2014. This is despite increasing use of GeneXpert, which was used 167,162 times in total for monitoring and diagnosis in Mozambique (2017). Although use of GeneXpert is increasing, there are no records of total number who were screened, received a lab test, types of lab test used, and positivity rates. 2 | INTRODUCTION CONTENTS TABLE OF TB Treatment The Mozambique NTP has achieved a treatment success rate of 90% among new and relapsed DS-TB cases (WHO, 2018). The primary objective of the national TB response therefore remains to increase the case detection rate, which was estimated at just 53% (WHO 2018). Treatment of TB in Mozambique EXECUTIVE SUMMARY is delivered in outpatient settings, with patients receiving inpatient care only for the treatment of adverse events or complications even for drug resistant cases. The use of community health workers has improved access to treatment but access still remains uneven. DOT has been implemented nationwide, with patients being required to attend outpatient clinics daily for the first two months of the treatment regimen, after which they attend twice per month. Improving access to supervised treatment through INTRODUCTION community-based initiatives therefore remains a priority. Figure 1.2 Treatment outcomes for DS-TB 1 CABO 6% 4% 4% 4% 2% 5% DELGADO 10% 6% 8% 7% 3% TETE 9% POLICY QUESTIONS & METHODOLOGY NIASSA NAMPULA 95% 93% 94% 2 ZAMBÉZIA 91% 91% 90% 89% 90% 83% 85% 85% 85% MANICA SOFALA Not evaluated Treatment failure rate RESULTS INHAMBANE Death rate 3 GAZA Lost-to-follow-up rate Treatment success rate Maputo City to City aputo lgado bane Gaz a Sof ala ia béz ampu la e a a Tet Niass anic IONA L MAPUTO apu M o De nham Zam N M NA T M Cab I CONCLUSION Source: NTP TB desk review, 2018. 4 Treatment outcomes for DR-TB remain sub-optimal; only 48% of MDR-TB cases and 38% of XDR-TB cases beginning second line treatment in 2015 successfully completed treatment. A transition from the use of injectables to newer, oral drugs, such as bedaquiline, as recommended by WHO, could improve KEY FINDINGS outcomes, but would also incur higher drug costs. Most MDR-TB patients are currently treated with a regimen of 18 months. A smaller cohort of patients are receiving the short-course oral regimen of 9 5 months. Finally, Mozambique has started to implement integrated psychosocial and adherence support packages and counselling for DR-TB patients. These packages include support such as reimbursement of transportation costs and nutritional support. RECOMMENDATIONS TB program expenditure The vast majority of the National TB Program (NTP) budget in Mozambique is funded by international donors, primarily the Global Fund (53%) and the World Bank (31%). Domestic funds cover 5% of total 6 expenditure, with the remaining 11% funded by other sources. Based on data provided by the individual funding sources, it is estimated that the total NTP budget in 2017 amounted to approximately USD 27 million in Mozambique. ANNEXES INTRODUCTION | 3 Table 1.1 Mozambique: NTP budget by source of financing (2017) Funding source 2017 spending (USD) % share Global Fund 14,176,960 53% World Bank 8,206,204 31% Domestic 1,361,600 5% Other sources 2,920,512 11% Grand Total 26,665,276 Source: NTP; latest data source was 2017. 1.2 RATIONALE FOR ALLOCATIVE EFFICIENCY ANALYSIS To achieve the 2030 SDG target and the UN/WHO’s goal of ending TB by 2035, it is crucial to determine which models of TB care and treatment should be prioritized, especially in the context of limited resources. Mozambique’s NTP budget was an estimated 27 million USD, of which 95% was internationally funded, indicating the high reliance on donor funding of the Mozambique TB response. In addition, as the country is moving on from the National Strategic Plan (NSP) for 2015–19 to the NSP 2020–25, an understanding what targets can be achieved under a given resource envelope is essential. This allocative efficiency study aims to support Mozambique in assessing advancements towards the strategic targets and provide inputs into the country’s decision-making on strategic TB investments to attain the 2035 End TB targets. Any resource-constrained efforts to improve health outcomes are inevitably faced with the need to allocate resources judiciously to achieve better results with the available resources. Moreover, any additional resources allocated in a resource-constrained context as is the Mozambique’s TB program, also merits optimized allocation across case finding strategies, diagnosis and treatment regimens as well as management and surveillance activities. This makes strategic decisions in prioritization of which programs to fund (allocative efficiency) and how they should be implemented (implementation efficiency) critical to maximize health outcomes. Figure 1.3 Allocative efficiency in Make the best possible TB investment decisions the TB response Support for demand and delivery of services to the The concept of allocative efficiency best feasable standards: refers to the maximization the right services of health outcomes with the in the right places least costly mix of health interventions. Implementation in the right way efficiency can be enhanced by for the right clients a range of measures relating at the right cost to service delivery modalities, at the right time management arrangements, unit costs in procurement and service For the greatest TB and health impact delivery, and several other areas. Cascade models have been ...while moving early and urgently to institutionalise and sustain services successfully applied in different Source: World Bank. Note: TB = tuberculosis 4 | INTRODUCTION CONTENTS TABLE OF contexts to identify breakpoints in health care, compare service delivery models and identify effective interventions to address patient drop-off in the service delivery continuum. Further to the request from the Government of the Republic of Mozambique and with the understanding of these two dimensions of improving TB efficiency, detailed consultations were held EXECUTIVE SUMMARY with program managers and experts in the National TB Program and the Ministry of Health. From these discussions, it was determined that the allocative efficiency modelling should focus on a national level analysis, aiming to: 1. Understand the current and future (predicted) TB epidemic and care cascade; and 2. Provide guidance on how to use existing resources and which models of TB care can maximize INTRODUCTION impact of the TB response in Mozambique. 1 POLICY QUESTIONS & METHODOLOGY 2 RESULTS 3 CONCLUSION 4 KEY FINDINGS 5 RECOMMENDATIONS 6 Photo: National Tuberculosis Programme. Used with permission. ANNEXES INTRODUCTION | 5 This page is for collation purposes only. CONTENTS TABLE OF To assess how incremental changes in spending affect TB epidemics and determine an optimized funding allocation, the optimisation model parameterises relationships between the cost of TB interventions, the coverage level attained by these interventions, and the resulting outcomes. EXECUTIVE SUMMARY Costs of all treatment programs were estimated using a ‘bottom-up’ approach, including outpatient care as well as hospitalisation. 2 POLICY QUESTIONS AND INTRODUCTION METHODLOGY 1 T his section outlines the study questions posed and the accompanying analyses conducted and POLICY QUESTIONS & METHODOLOGY presented in this report. Additional details are available in Annexes 1, 2, 3, 4, 5 (Technical summary of Optima TB and Data inputs into the model). 2 2.1 POLICY QUESTIONS To support Mozambique in allocating TB resources, the analyses presented in this report set out to answer five key policy questions developed together with key stakeholders in the initial planning and methodology workshop. These are: RESULTS 1. What is the status of the TB epidemic and TB care cascade for Mozambique using the most 3 recent available data? What are the current estimated numbers of active TB infections, latent TB infections, TB incidence, TB prevalence and TB-related deaths by population group and resistance type: CONCLUSION i. Children aged 0–14 years ii. Adults 15+ years not living with HIV 4 iii. Adults 15+ years living with HIV iv. By resistance type What is the status of the current TB care cascade? KEY FINDINGS 2. What is the projected future trend of the TB epidemic in Mozambique under current levels of budget and assuming status-quo programming? 5 What are the future numbers of active TB infections, latent TB infections, TB incidence, TB prevalence and TB-related deaths up to 2035 if current programs are implemented with constant coverage in: RECOMMENDATIONS i. Children aged 0–14 years ii. Adults 15+ years not living with HIV 6 iii. Adults 15+ years living with HIV iv. By resistance type ANNEXES 7 3. What is the projected impact on the TB epidemic of meeting key national and international targets? i. TB incidence ii. TB prevalence iii. TB deaths 4. What is the projected future trend of the country’s TB epidemic with optimized allocation of currently available resources? How can the TB care cascade be improved? i. Is the projected future trend of the country’s TB epidemic with optimized allocation of currently available resources? ii. What are the key interventions for addressing break points in the cascade and what is the evidence for their effectiveness? iii. Which steps of the cascade should be prioritized in resource allocation and programming? 5. What resources are required to achieve key targets of the national TB response? 2.2 METHODOLOGY Collaboration and stakeholder involvement The analysis was a collaboration between the Government of the Republic of Mozambique, the Mozambique National Tuberculosis Program, the World Bank, and University College London as part of the Optima Consortium of Decision Sciences (OCDS). Focal Points were assigned within each organisation to implement the analyses and coordinate contributions. A group of experts and key informants was brought together in two workshops to provide input into the policy questions and analytical framework, share data and expertise, and review the outputs. Epidemiological, program and cost data were collected in a joint effort using an adapted Excel-based Optima TB data entry spreadsheet. Input data, model calibration and cost-coverage-outcome relations were reviewed and validated by the in-country study group. The team also consulted with government experts and other in-country partners on preliminary results. Optima TB model To carry out the analyses, the team used Optima TB, a mathematical model of TB transmission and disease progression integrated with an economic and program analysis framework. (Figure 2.1). Figure 2.1 The Optima approach to TB modelling BURDEN OF DISEASE PROGRAMMATIC RESPONSES OBJECTIVES AND CONSTRAINTS Epidemic model Identify interventions Strategic goals Data synthesis Delivery modes Economic constraints Calibration projection Costs and e ects Ethical and logistic constraints SCENARIO ANALYSIS OPTIMIZATION PROJECTED HEALTH AND ECONOMIC OUTCOMES Source: Optima Consortium for Decision Science. Optima TB incorporates evidence on biological transmission probabilities, detailed infection progression and population mixing patterns, in a compartmental mathematical model, which 8 | POLICY QUESTIONS AND METHODOLOGY CONTENTS TABLE OF disaggregates populations into different model compartments including susceptible, vaccinated, early latent, late latent, undiagnosed active TB, diagnosed active TB, on treatment and recovered populations. In addition, compartments are further disaggregated by drug resistance types: drug susceptible (DS), multi-drug resistant (MDR) and extensively drug resistant (XDR). These EXECUTIVE SUMMARY compartments change in size based on yearly transition rates. A detailed illustration of the compartmental model structure is included in Annexes 1–5. A national TB prevalence survey is currently underway in Mozambique. In the absence of survey results, Optima TB was calibrated primarily based on data on TB case notifications and WHO estimated TB incidence (2018 Global TB report) with guidance from NTP. The model was calibrated INTRODUCTION to closely match the yearly number of notified TB cases, as well as estimates of key TB indicators such as active-TB incidence and prevalence and latent TB prevalence. Parameters with high levels of 1 uncertainty, such as force of infection were adjusted to closely match notifications, as well as other indicators including TB incidence and prevalence. To assess how incremental changes in spending affect TB epidemics and determine an optimized funding allocation, the model parameterises relationships between the cost of TB interventions, the POLICY QUESTIONS & METHODOLOGY coverage level attained by these interventions, and the resulting outcomes (cost-coverage-outcome relations). These relationships are specific to the place, population, and intervention being considered. 2 Using the relationships between cost, coverage, and outcome in combination with Optima TB’s epidemic model, it is possible to calculate how incremental changes in the level of funding allocated to each intervention will impact the overall epidemic indicators. Furthermore, by using a mathematical optimization algorithm, Optima TB is able to determine an optimized allocation of funding across different TB interventions. Additional details of the Optima TB model and the Mozambique application RESULTS are included in Annexes 1–5. 3 Cost estimates Total expenditure in the TB response was also estimated. In addition to the core TB budget (26.67 million in 2017), this includes funding from other sources that support the management of TB/HIV cases and TB prevention among HIV patients. For example, a bottom-up costing of expenditure on CONCLUSION TB-preventive treatment for PLHIV suggests that this intervention costs 12 million USD annually, and is funded by the HIV department. Similarly, ambulatory care for TB patients and BCG vaccinations for 4 children are funded by the Ministry of Health. Estimated total expenditure across program areas and funding streams amounted to around 39 million USD in 2017, nearly double the actual TB budget. Table 2.1 Mozambique: total estimated expenditure on the TB programmatic response included in this KEY FINDINGS analysis by area (2017) Program areas 2017 spending (USD) % share 5 Screening and diagnosis 21,778,771 55% Prevention 4,480,459 12% RECOMMENDATIONS Treatment 10,524,949 27% Management, HR and other 2,514,886 6% Total 39,299,066 100% 6 Source: Optima TB analysis estimates. Note: HR = human resources Although the NTP budget (Table 2.1) is USD 27 million, the modelled programs include expenditure on ANNEXES TB from other sources. This applies to an estimated USD 3.4 million on preventive therapy (funded by POLICY QUESTIONS AND METHODOLOGY | 9 the HIV program), estimated USD 11 million of state funding for ambulatory care and other services, and an estimated USD 0.9 million on the BCG vaccination program. As data on TB spending by expenditure category was not available, estimates were made used using a bottom-up costing approach. Both total expenditure and unit costs for interventions are therefore subject to some uncertainty In addition to the total expenditure reported above, a detailed breakdown of expenditure on drugs, diagnostics and other supplies was provided by the NTP. All of these sources facilitated the estimation of program costs for inclusion in the analysis in this report. Analytical framework Model parameters are summarised in Table 2.2 and detailed in Annex 5. All DR-TB treatment programs include financial support to improve patient adherence. Table 2.2 Model parameterisation Parameterization in Category the Optima model Description / Assumptions Children Male and Female Children aged 0-14 Populations (0–14 years) defined in the model Adults (15+) Male and Female Adult Population aged 15+ (HIV-negative) Adults (15+ HIV+) Male and Female Adult Population aged 15+ (HIV-positive) Ambulatory- Current treatment delivery for DS-TB implemented in focused DS Mozambique, 6 months of treatment with daily outpatient treatment visits for the first two months followed by twice-monthly outpatient visits thereafter. Ambulatory- Treatment delivery for MDR-TB, 18 months of treatment with focused MDR daily outpatient visits for the first two months followed by treatment (long twice-monthly outpatient visits thereafter. Treatment regimen Program course) includes bedaquiline. Severe MDR cases, re-treatments and expenditure adverse drug reactions may require some hospital care. areas defined in the model and Ambulatory- Treatment delivery for MDR-TB, 9 months of treatment with included in focused MDR daily outpatient visits for the first two months followed by optimization treatment (short twice-monthly outpatient visits thereafter. Oral treatment analysis course) regimen includes bedaquiline or delamanid. Very sick MDR cases, re-treatments and adverse drug reactions may require some hospital care. Ambulatory- Treatment delivery for XDR-TB, 24 months of treatment with focused XDR daily outpatient visits for the first two months followed by treatment twice-monthly outpatient visits thereafter. Very sick XDR cases, re-treatments and adverse drug reactions may require some hospital care. Table 2.2 continued... 10 | POLICY QUESTIONS AND METHODOLOGY CONTENTS TABLE OF Table 2.2  Model parameterization (continued) Parameterization in Category the Optima model Description / Assumptions BCG Vaccination Vaccination with Bacillus Calmette-Guérin targeting new- EXECUTIVE SUMMARY borns within the 0–14 population Preventive treatment (contacts 6-month Isoniazid Preventive Therapy given to children aged of active TB - under 5 identified through contact tracing of people with under 5) active TB INTRODUCTION Preventive 6-month Isoniazid Preventive Therapy given to adults living treatment (PLHIV) with HIV 1 Outpatient screening Includes the cost of complete symptom screen delivery, as (excluding PLHIV) well as the cost of available TB-testing for diagnosis Program POLICY QUESTIONS & METHODOLOGY expenditure HIV outpatient Includes the cost of complete symptom screen delivery, as areas defined screening well as the cost of TB-testing for diagnosis in the 2 Active case finding Includes the cost of complete symptom screen delivery, as model and (health workers) well as the cost of available TB-testing for diagnosis included in Active case finding Includes outreach costs in addition to the cost of complete optimization (miners) symptom screen delivery, as well as the cost of available TB- analysis testing for diagnosis RESULTS Active case finding Includes outreach costs, in addition to the cost of complete 3 (contact tracing) symptom screen delivery, as well as the cost of available TB- testing for diagnosis Active case finding (community Includes outreach costs in addition to the cost of complete CONCLUSION outreach at symptom screen delivery, as well as the cost of available TB- hotspots) testing for diagnosis 4 Active case finding Includes outreach costs in addition to the cost of complete (prisons) symptom screen delivery, as well as the cost of available TB- testing for diagnosis The components of Some program areas have not been optimized but instead were KEY FINDINGS TB spending that fixed at agreed amounts. This was done for different reasons: were not included due to an unclear relationship between an intervention and 5 in the optimization its effect on TB incidence, morbidity or mortality, or because Expenditure analysis: there was no detail on what the expenditure was for. ART areas not coverage was held fixed as the budget is very large and, RECOMMENDATIONS optimized once on ART, treatment should never be withdrawn. The importance of ART coverage was instead explored in scenario analyses (see results). 6 Management, HR and other costs Fixed at USD 2,514,886 Table 2.2 continued... ANNEXES POLICY QUESTIONS AND METHODOLOGY | 11 Table 2.2  Model parameterization (continued) Parameterization in Category the Optima model Description / Assumptions Expenditure ART for adult Fixed at USD 165,606,756 (Note that this is not included in the areas not PLHIV TB budget estimate) optimized 2001 Year of model initiation, start year for data entry 2017 Base year Years and time horizons 2025 Milestone year for End TB Strategy and target year for achievement of Stop TB partnership targets 2035 Target year for End TB Strategy Baseline As per authors’ Total spending on modelled TB programs in 2017 as per this scenario expenditure study’s expenditure analysis (estimated approximately USD funding analysis 36.8 million, excluding management, HR and other costs.) Source: World Bank. Note: ART = antiretroviral therapy; BCG = Bacille Calmette-Guerin; DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. Costs of all treatment programs listed above were estimated using a ‘bottom-up’ approach, including outpatient care as well as hospitalisation. Although treatment is delivered in the outpatient setting in Mozambique, DR-TB cases often require hospitalisation, for example due to adverse effects of treatment regimens. Data from one province was used to inform the length of hospitalisation. In Maputo province, 58% of DR-TB cases received some inpatient treatment, with an average length of stay of 35 days. An average cost per ambulatory interaction was also derived and applied to both screening programs and to outpatient treatment following the initial hospitalisation period. Based on spending per person reached with an intervention, cost-coverage-outcome relations were developed. Calibrations and cost-coverage outcome relations were produced in collaboration with in-country experts and are further explained in Annexes 1, 2, while unit costs are shown in Annexes 9–15. Modelling the impact of the HIV epidemic In recognition of the importance of the HIV epidemic in the high-prevalence and incidence setting of Mozambique, the most recent Spectrum modelling estimates for Mozambique (conducted by AvenirHealth and PEPFAR partners) from 2000 to 2021 were used to inform ART coverage, HIV prevalence and HIV incidence. Several Optima TB model parameters are impacted by HIV. Co-infection directly impacts nine model parameters whereby each parameter was influenced differently depending on co-infection rates and ART coverage (see Appendices on Optima TB and on TB epidemiological parameters for further details): 1. Mortality rates (excluding TB-related deaths) 2. Mortality rates (including TB-related deaths) 3. Susceptibility to TB infection 4. TB infectiousness 5. Departure rate from early TB latency 6. Probability of latent TB infection versus active TB infection 7. Proportion of new active TB cases with different smear/strain combinations 12 | POLICY QUESTIONS AND METHODOLOGY CONTENTS TABLE OF 8. Rate of TB diagnosis (by population which may be targeted differently) 9. Proportion of TB treatment outcomes for each smear/strain combination Strategic TB targets used in the analysis The national 2025 targets and global 2025 STOP TB targets both aim to improve diagnosis rates and EXECUTIVE SUMMARY treatment success rates. The targets used in the modelling analyses are shown in Table 4. Table 2.3 National and international TB care cascade targets Impact of improved Baseline (‘current care cascade conditions’, 2017) NSP targets (2025) STOP TB target (2025) INTRODUCTION DS-TB care Diagnosis 52% 90% 90% 1 Treatment success 90% 90% 90% MDR-TB care Diagnosis 5% 80% 90% POLICY QUESTIONS & METHODOLOGY Treatment success 48% 70% 90% 2 XDR-TB care Diagnosis 5% 80% 90% Treatment success 38% 70% 90% Sources: WHO Mozambique TB country profile; Mozambique National Strategic Plan 2015–19 STOP-TB. Note: MDR and XDR diagnosis rates are calculated on the basis of number of cases diagnosed and treated for DR-TB (i.e., RESULTS not those who receive a TB diagnosis but receive treatment for DR-TB). The model’s “diagnosis rate” was calculated using 3 notified as a proportion of total prevalence and not incidence in the absence of data on treatment initiation, an arbitrary pre-treatment loss to follow up of 2% was assumed in consultation with local experts. Limitations of the analysis As with any mathematical modelling analysis it is necessary to make assumptions about data that are CONCLUSION not routinely collected or available, and about some of the expected relationships between variables. These assumptions necessarily imply certain limitations: 4 Active TB prevalence:  This parameter includes diagnosed and undiagnosed active TB cases and is of key importance in TB modelling. Routine data on TB notifications formed the basis for estimating this parameter. WHO estimates of total TB prevalence in 2000 formed the baseline estimate for prevalence in the model, while prevalence for the following years is estimated based on yearly transition rates in KEY FINDINGS the model. Prevalence is also disaggregated across populations based on reported notifications of TB cases. This means that prevalence may be underestimated in populations with lower diagnosis rates. 5 Empirical data from the TB prevalence survey will improve the accuracy of future TB modelling in Mozambique. TB expenditure:  There was limited data on the coverage and costs of key TB interventions in RECOMMENDATIONS Mozambique, affecting the estimation of TB expenditure. At the same time, TB spending data was reported in broad expenditure areas only, while this analysis uses discrete TB interventions. Unit costs for interventions were subject to some levels of uncertainty. While data triangulation of intervention 6 cost, coverage, and program expenditure used all available information from different funding sources, it had data limitations. Implementation efficiency:  The analysis included considerations of implementation efficiency in a limited way only, as detailed modelling of implementation efficiency was beyond the scope of the study. ANNEXES POLICY QUESTIONS AND METHODOLOGY | 13 For instance, reduced drug prices (leading to lower unit costs, better efficiency and cost-effectiveness) were not modelled, although treatment regimens were carefully costed by component cost. Lower unit costs can influence resource allocation recommendations. Intervention effectiveness:  Allocative efficiency modelling depends critically on the availability of evidence-based parameters for the effectiveness of individual interventions. Although these estimates were derived from global systematic literature reviews where possible, they may vary in specific countries and populations. In particular, the quality of implementation and levels of adherence may vary by context and population. All interventions and spending categories for which effectiveness parameters could not be obtained were treated as fixed spending in the mathematical optimization. Non-TB benefits:  Effects outside of TB indicators, such as the non-TB benefits of different TB treatment modalities are not considered in these analyses. Given the range and complexity of interactions among interventions and their non-TB benefits, the model did not consider wider health, social, human rights, ethical, legal, employment-related or psychosocial implications; but acknowledges that they are important aspects to be considered in planning and evaluating TB responses. Photo: National Tuberculosis Programme. Used with permission. 14 | POLICY QUESTIONS AND METHODOLOGY The Optima estimate of TB prevalence is much higher than previous WHO 2014 CONTENTS TABLE OF estimate of 150,000 Assuming constant conditions of TB intervention coverage and outcomes, and that ART coverage in Mozambique will reach 90% by 2035, the projected TB incidence rates per 100,000 are on a downward trajectory, decreasing by an average of approximately 2.5% per year between 2017 and 2035. The incidence of TB in children remains far lower than in the adult populations. TB incidence is projected to reduce significantly in PLHIV as ART coverage increases. EXECUTIVE SUMMARY Treatment for HIV significantly reduces the probability of progression from latent to active TB. It is expected that Mozambique would achieve a 90% target for ART coverage by 2035. INTRODUCTION 3 RESULTS 1 3.1 DEMOGRAPHIC TRENDS POLICY QUESTIONS & METHODOLOGY The following key observations can be made: 2 Children:  The number of children in Mozambique has been increasing. UN Population Division projections suggest that the number of people under 15 years old in Mozambique will increase from around 13.4 million in 2017 to 19.2 million in 2035. Children are vaccinated with BCG at birth in Mozambique (99% in 2016). For modelling purposes, the analysis is based on the assumption of 50% efficacy of vaccination at birth (Mangtani et al. 2014). RESULTS 3 Figure 3.1 Projected demographic trends in Mozambique for children aged 0–14 years (2002–35) 20m Population age 0−14 18m CONCLUSION 16m Number of people 14m 4 12m 10m 8m KEY FINDINGS 6m 5 4m 2m RECOMMENDATIONS 0m 2005 2010 2015 2020 2025 2030 2035 Year 6 Source: Optima TB model output Adults:  UN Population Division projections suggest that the total number of people age 15 years and over in Mozambique will increase from around 14.3 million in 2017 to 28 million in 2035. Currently, an estimated 2 million adults are living with HIV in Mozambique (UNAIDS). To ensure alignment with ANNEXES 15 existing HIV projections in Mozambique, SPECTRUM data on future HIV prevalence was used in this analysis. These projections suggest adult prevalence of 1.8 million in 2035. Future HIV prevalence and incidence remain highly uncertain and will depend on future coverage of ART. In 2017, adult coverage of ART was 55%. Figure 3.2 Projected demographic trends in Mozambique for HIV-negative and HIV-positive adults aged 15 years and older (2002–35) a.  15+ population b.  15+ HIV-positive population 30m 2.0m 1.8m 25m Number of HIV-positive population age 15+ 1.6m Number of population age 15+ 1.4m 20m 1.2m 15m 1.0m 0.8k 10m 0.6k 5m 0.4k 0.2k 2m 0 0 2005 2010 2015 2020 2025 2030 2035 2005 2010 2015 2020 2025 2030 2035 Year Year Source: Optima TB model output. Total population:  It is estimated by the UN that between 2017 and 2035, the total population of the country will increase by about 19.5 million (+70%) and would reach 48 million in 2025. Figure 3.3 Total population 50m 48 million 40m Population size 30m 28.5 million Between 2017 and 2035, the total population of the country will increase by about 19.5 million (+70%) to 20m in 2025 10m 2017 0m 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima TB model output. 16 | RESULTS CONTENTS TABLE OF 3.2 PAST AND FUTURE TB EPIDEMIC TRENDS What is the projected future trend of the TB epidemics in Mozambique under current budget levels and assuming status-quo programming? This section outlines the epidemic trajectory for DS-, MDR- and XDR-TB across the three EXECUTIVE SUMMARY sub-populations in Mozambique. Estimates for the 2017 base year of the Mozambique analysis Given that 2017 was used as the base year for the analysis, Table 3.1 and Table 3.2 below present Optima TB estimates of active TB prevalence, incidence, latent infections and TB-related deaths by sub-population for 2017. INTRODUCTION Table 3.1 Model estimates of number and prevalence of active pulmonary TB infections by sub- population (2017) 1 Active TB Active DS-TB Active MDR-TB Active XDR-TB Active TB Population cases cases cases cases prevalence 0-14 years 32,826 32,165 651 9 0.24% POLICY QUESTIONS & METHODOLOGY 15+ years HIV- 163,538 155,036 8,323 179 1.15% 15+ years HIV+ 197,435 186,992 10,228 214 10.26% 2 Total 393,799 374,193 19,938 403 1.32% In percent 100% 95.0% 4.8% 0.1% – Source: Optima TB model output. Note: DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; XDR = extensively drug-resistant. RESULTS 3 Table 3.2 Model estimates of active pulmonary TB incidence, latent infections and TB-related deaths, by sub-population (2017) Population Incidence per 100,000 Latent TB cases TB-related deaths 0–14 years 144 976,635 2,052 CONCLUSION 15+ years 542 6,032,670 20,880 4 15+ years HIV+ 2,592 583,964 37,406 Total 491 7,593,269 60,339 Source: Optima TB model output. Note: TB = tuberculosis. KEY FINDINGS Past trends in Mozambique’s TB epidemic 5 Historical TB notifications data for Mozambique were used to calibrate the model and assess past epidemic trends. Although the number of notifications and the estimated case detection rate in Mozambique increased rapidly, WHO estimates suggest that TB incidence also increased. Accounting for pulmonary TB only, there were: RECOMMENDATIONS • 79,658 notified TB cases in 2017, of which 1.25% were DR-TB • 0,140 notified TB cases in 2015, of which 1.40% were DR-TB 6 • 66,187 notified TB cases in 2013, of which 1.52% were DR-TB ANNEXES RESULTS | 17 Past TB epidemic trends for the period 2002 to 2017 show significant differences across sub-populations included in the analysis. Results are presented for children aged 0 –14, adults aged 15 and over (HIV-negative) and adults aged 15 and over (HIV-positive) (Table 3.2). TB prevalence estimates over time Actual TB prevalence in Mozambique is currently unknown. In 2014, WHO estimated TB prevalence of approximately 150,000 [80,000– 243,000]. In this analysis, Optima TB estimate is much higher than previous WHO estimate (about 352,000 for year 2017). Figure 3.4 shows the temporal trend of TB prevalence per 100,000 population over 30 years. Figure 3.4 Prevalence of pulmonary TB per 100 thousand 1,200 Prevalent number of TB cases per 100k 1,000 800 600 400 200 0 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima model output. Note: TB = tuberculosis. Future TB incidence projections Future projections for TB incidence, assuming TB intervention coverage and outcome conditions as per 2015, are shown below in Table 3.3, Figure 3.5 (a–c), and Figure 3.6 for 2025 and 2035. Given the importance of ART coverage on TB incidence in this context, in consultation with local experts, a figure of 90% adult ART coverage by 2035 (compared to 55% of adult PLHIV in 2017; UNAIDS) was used to inform a realistic target for the purposes of this analysis. ART coverage has a significant impact on modelled TB incidence. For PLHIV on ART, the probability of progressing from latent to active TB is assumed to be the same as for the general population. For PLHIV not on ART, the probability of progression is significantly higher. Assuming constant conditions of TB intervention coverage and outcomes, and that ART coverage in Mozambique will reach 90% by 2035, the projected TB incidence rates per 100,000 are on a downward trajectory, decreasing by an average of approximately 2.5% per year between 2017 and 2035 (Figure 3.6). The incidence of TB in children remains far lower than in the adult populations. TB incidence is projected to reduce significantly in PLHIV as ART coverage increases (Table 3.3 and Figure 3.5 (a–c). 18 | RESULTS CONTENTS TABLE OF Table 3.3 Modelled pulmonary TB incidence per 100,000 in Mozambique, by sub-population (2017, 2025 and 2035) Sub-population TB incidence 2017 TB incidence 2025 TB incidence 2035 0-14 years 144 156 127 EXECUTIVE SUMMARY 15+ years 519 499 406 15+ years HIV+ 2,752 1,356 545 Total 495 401 303 Source: Optima model output. INTRODUCTION Note: TB = tuberculosis. 1 Figure 3.5 Projected future TB incidence (a – c) a. Projected new TB cases (total cases) b. Projected new TB cases per 100,000 population POLICY QUESTIONS & METHODOLOGY 200,000 700 Number of adjusted incident TB cases per 100k 600 2 Number of incident TB cases 150,000 500 400 100,000 300 200 RESULTS 50,000 3 100 0 0 2005 2010 2015 2020 2025 2030 2035 2005 2010 2015 2020 2025 2030 2035 Year Year CONCLUSION c. Projected TB cases/100,000 population among adults 15 years and older living with HIV 70,000 4 Number of incident TB cases per 100k population 60,000 50,000 40,000 KEY FINDINGS 30,000 5 20,000 10,000 0 RECOMMENDATIONS 2005 2010 2015 2020 2025 2030 2035 Year 6 Source: Optima model output. Note: k = thousands; TB = tuberculosis. ANNEXES RESULTS | 19 Figure 3.6 Projected future DR-TB (total cases) 6,000 5,000 Number of incident DR-TB cases 4,000 3,000 2,000 1,000 0 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima model output. Note: TB = tuberculosis. Incidence of drug-resistant TB is projected to follow the same overall trend. The WHO estimate 3.7% [2.5%–5.2%] of new TB cases are MDR/RR-TB (2017). These projections assume constant TB care coverage and outcomes to 2035 and ART coverage increasing to 90% by 2035. Temporal trends in latent TB infections The actual prevalence of latent TB in Mozambique is unknown. Our analysis, based on observed active TB infections in Mozambique, estimated latent TB prevalence of around 34% in Mozambique for 2016 (Figure 3.7). This is consistent with published national estimates of between 31% and 38% latent TB prevalence in Mozambique, with a best estimate of 34% (Houben and Dodd 2016). The projections for future years are inherently uncertain due to probable changes in disease burden and program coverage. Figure 3.7 Model-derived total latent prevalence in Mozambique, 2005-35 (% of total population) 0.40 0.35 0.30 Latent TB prevalence 0.25 0.20 Model (calibrated) Data (Houben and Dodd best 0.15 estimate with con dence intervals 0.10 0.05 0 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima TB model output. Note: Data for comparison represents range of estimate from Houben and Dodd (2016); TB = tuberculosis. 20 | RESULTS CONTENTS TABLE OF What is the current status of the TB epidemic and TB care cascade for Mozambique using the most recent available data? Figure 3.8 shows outcomes from each stage in the TB care cascade based on total modelled pulmonary TB incidence (2018). It shows that TB diagnosis remains suboptimal with an estimated 51% of all active EXECUTIVE SUMMARY pulmonary TB incidence diagnosed. As a result, despite high treatment success rates, only about 41% of all new active TB cases attain treatment success. Improved screening and diagnosis are therefore a priority for Mozambique’s TB response. Figure 3.8 TB care cascade (2017; all cases) INTRODUCTION 150k 140k 1 130k 120k New active TB cases 2017 (k = thousands) 100k POLICY QUESTIONS & METHODOLOGY 90k 80k 2 70k 51% 60k 93% 50k 86% 40K RESULTS 30k 3 20K 10k 0 Active TB Diagnosed Initiated Completed treatment (47%) treatment (41%) CONCLUSION Source: Optima TB model output. 4 Note: This cascade is probabilistic rather than cohort-based. It shows the probability of final outcomes from each stage in the cascade, e.g., for people with active TB, how many of those will be diagnosed prior to natural recovery or succumbing to a TB-related death. This cascade is based on modelled incidence of TB for 2018 and rates of flow through the stages of the cascade. The green color in each column indicates total cases, diagnosed, initiated treatment, and completed treatment. KEY FINDINGS Figure 3.9 shows the 2018 care cascade for MDR-TB. Results show a low diagnosis rate of about 19% (of new active MDR-TB). This leads to a very low percentage of cases with treatment success (12%). 5 Increased coverage of drug susceptibility testing, either using GeneXpert or other diagnostic testing is essential to improve the care cascade for MDR-TB. Figure 3.10 show outcomes from each stage in the care cascade based on total modelled pulmonary RECOMMENDATIONS TB incidence (2018) in adults disaggregated by HIV status. Our finding suggests that PLHIV have a significantly higher undiagnosed TB mortality rate—40% of total incident TB in adult PLHIV died without receiving a TB diagnosis—compared to around 20% amongst HIV-negative adults. Improved 6 screening and timely diagnosis in PLHIV are therefore a priority for preventing deaths amongst coinfected cases. ANNEXES RESULTS | 21 Figure 3.9 MDR-TB treatment cascade (2017) 5,000 4,000 New active MDR-TB cases 2017 3,000 2,000 1,000 19% 17% 12% 0 Active MDR-TB Diagnosed Initiated Completed (19%) treatment (17%) treatment (12%) Source: Optima TB model output. Note: MDR-TB = multi-drug resitant TB; TB = tuberculosis. Figure 3.10 TB Care cascade by HIV status a. HIV+ Active TB Treatment cascade b. HIV- Active TB Treatment cascade 80k 70k New active TB cases 2017 (k = thousands) 60k 60k New active TB cases 2017 (k = thousands) 50k 50k 40k 40k 54% 50% 30k 30k 43% 20k 44% 40% 20k 35% 10k 10k 0 0 Undiagnosed Diagnosed Initiated Completed Undiagnosed Diagnosed Initiated Completed Active TB treatment treatment Active TB treatment treatment Source: Optima TB model output. Note: TB = tuberculosis. 3.3 IMPACT OF MEETING NATIONAL AND INTERNATIONAL TB CARE TARGETS ON THE TB EPIDEMIC? A scenario analysis was performed to understand the impact of meeting international care cascade targets set for 2025 on key TB indicators. For each scenario, there is a time frame for programmatic change to occur, which is the time period over which programmatic targets are achieved, and another 22 | RESULTS CONTENTS TABLE OF time frame for tracking impact, which is the time period for which the effect of these achievements is estimated. For example, in the 2025 target scenario, coverage targets are achieved by 2025 and the impact of achieving and sustaining 2025 coverage levels is tracked up to 2035. Testing and treatment scenarios to meet 2025 STOP-TB targets EXECUTIVE SUMMARY This group of scenarios models the impact of meeting 2025 NSP and STOP TB targets separately for: • TB screening/testing • TB treatment outcomes These effects are then considered simultaneously to assess what impact on key TB indicators can INTRODUCTION be obtained by meeting 2025 STOP TB targets. Although the 2025 STOP TB targets also aim for treatment initiation rates amongst diagnosed cases of 100%, there was insufficient data on treatment 1 initiation rates in Mozambique to conduct a meaningful scenario analysis. Improved TB screening/testing What is the impact of reaching 2025 targets for case detection? The parameters modified in the model POLICY QUESTIONS & METHODOLOGY to assess the effect of the scenario are summarised in Table 3.4. 2 Table 3.4 Scenario parameters: improved TB screening/testing Current NTP 2025 STOP-TB 2025 Improved TB screening/testing conditions (2017)* Targets Targets Case detection for DS-TB 52% 90% 90% Case detection for MDR-TB 5% 80% 90% RESULTS Case detection for XDR-TB 5% 80% 90% 3 Source: WHO (2018); STOP-TB. Note: * = MDR and XDR diagnosis rates are calculated on the basis of number of cases diagnosed and treated for DR-TB (i.e., not those who receive a TB diagnosis but receive treatment for DR-TB). The model’s “diagnosis rate” was calculated using notified as a proportion of total prevalence and not incidence. In the absence of data on treatment initiation, an CONCLUSION arbitrary pre-treatment loss to follow up of 2% was assumed in consultation with local experts; DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; XDR = extensively drug-resistant.. 4 Improved treatment outcomes Table 3.5 presents various targets related to improved treatment outcomes in the TB care cascade. Table 3.5 Scenario parameters: Improved treatment outcomes KEY FINDINGS Current NTP 2025 STOP-TB 2025 5 Improved treatment outcomes conditions (2017) Targets Targets Treatment success rates for DS-TB regimens 90% 90% 90% Treatment success rates for MDR-TB regimens 48% 70% 90% RECOMMENDATIONS Treatment success rates for XDR-TB regimens 38% 70% 90% Source: WHO (2019); STOP-TB. 6 Note: DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; XDR = extensively drug-resistant. Figure 3.11 presents the impact of meeting and sustaining the NTP and STOP-TB care cascade target of 90% diagnosis rate by 2025 on all active pulmonary TB prevalence. Although treatment success rate in Mozambique is high (90% for new and relapsed cases, WHO 2017), the estimated case detection is low ANNEXES RESULTS | 23 (52%; WHO, 2017). Therefore, improving case detection is a priority for Mozambique’s TB response – both NTP targets and Stop-TB targets aim for 90% diagnosis rate by 2025. We modelled the impact of meeting the 2025 targets of diagnosis rate and our results show significant reductions in the total number of active TB cases and TB-related deaths of up to 80% relative to current conditions by 2035. Figure 3.11 Modelled impact of reaching 90% diagnosis rate by 2025 on the number of people with active pulmonary TB (TB prevalence) (2001–35) 400k Total number of active pulmonary TB infections 350k 300k 250k Current conditions (k = thousands) 200k 150k 90% diagnosis 100k 50k 0 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima TB model output. Note: TB = tuberculosis. Similarly, Figure 3.12 presents the impact of meeting the 90% diagnosis rate by 2025 on annual TB-related deaths. Meeting and sustaining the proposed care cascade would yield significant reductions in the total number of deaths, of up to 80% relative to current conditions by 2035. Figure 3.12 Modelled impact of reaching 90% diagnosis rate by 2025 on the annual number of pulmonary TB-related deaths (2001–35) 60k Annual number of TB-related deaths 50k 40k Current conditions (k = thousands) 30k 20k 90% diagnosis 10k 0 2005 2010 2015 2020 2025 2030 2035 Year Source: Optima TB model output. Note: TB = tuberculosis. Figure 3.13 presents the impact of meeting the STOP-TB care cascade targets for all types of drug-resistant TB. Simultaneously, meeting the proposed targets can reduce 50% new DR-TB infections by 2035. 24 | RESULTS CONTENTS TABLE OF Figure 3.13 Total modelled number of people with DR-TB 25k Total number of active pulmonary DR-TB infections (k = thousands) Current conditions 20k Current diagnosis, 90% success EXECUTIVE SUMMARY 15k 90% diagnosis, current outcomes 10k 90% diagnosis, 90% success 5k 0 2001 2013 2025 2035 INTRODUCTION Year Source: Optima TB model output. 1 Note: DR-TB = drug resistant tuberculosis. Figure 3.14 presents the impact of meeting and sustaining the NTP 2025 care cascade targets for drug-resistant TB. These targets show similar trends to those observed under the Stop TB scenarios. POLICY QUESTIONS & METHODOLOGY As the targets for diagnosis and treatment success are lower than the Stop TB targets, the impact on DR-TB prevalence is correspondingly less. Simultaneously meeting NTP targets for diagnosis and 2 treatment of MDR-TB by 2025 could reduce MDR-TB prevalence in 2035 by 50% relative to current conditions. Figure 3.14 Projected impact of meeting NTP care cascade targets on the number of people with pulmonary a. DR-TB and b. XDR-TB (2015–35) RESULTS a. Modelled number of people with MDR-TB 3 25k MDR-TB infections (k = thousands) Total number of active pulmonary Current conditions 20k Current diagnosis, 70% success 15k 80% diagnosis, current outcomes 10k CONCLUSION 80% diagnosis, 70% success 5k 4 0 2001 2013 2025 2035 Year b. Modelled number of people with XDR-TB KEY FINDINGS 1,000 Total number of active pulmonary 800 5 80% diagnosis, current outcomes XDR-TB infections 600 80% diagnosis, 70% success Current conditions 400 Current diagnosis, 70% success RECOMMENDATIONS 200 0 2001 2009 2017 2025 2033 6 Year Source: Optima TB model output. Note: MDR-TB = multi-drug resistant tuberculosis; XDR-TB = extensively drug-resistant tuberculosis. ANNEXES RESULTS | 25 3.4 WHAT IS IMPACT OF REACHING 90% ART COVERAGE ON FUTURE HIV INCIDENCE? Treatment for HIV significantly reduces the probability of progression from latent to active TB. In consultation with local experts, it is expected that Mozambique would achieve a 90% target for ART coverage by 2035. Scaling up to 90% ART coverage by 2035 is projected to reduce the number of new TB cases among adult PLHIV in 2035 by over 50% compared to current ART coverage (55%). Additionally, the timeframe for ART scale-up is important for reducing TB incidence. Reaching 90% ART coverage more quickly reduces incident TB in PLHIV very rapidly (Figure 3.15 b). This highlights the importance of expanding access to treatment for HIV to as many people as possible and as soon as possible in order to minimize TB incidence in Mozambique. Figure 3.15 Estimated impact of reaching a. different ART coverage levels and b. 90% ART coverage levels on the annual number of new active pulmonary TB infections (2017–35) a. Impact of different scale-up assumptions by 2035 among HIV positive adults 60,000 Number of new PLHIV active pulmonary 50,000 70% by 2035 80% by 2035 infections age 15+ 40,000 90% by 2035 30,000 20,000 10,000 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Year b. Impact of timing of reaching 90% coverage among HIV positive adults 60,000 Number of new PLHIV active pulmonary 50,000 90% by 2022 90% by 2025 infections age 15+ 40,000 90% by 2030 30,000 90% by 2035 20,000 10,000 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 Year Source: Optima TB model output. Note: PLHIV = people living with HIV. Although the effect is far smaller compared to the effect of ART coverage among HIV positive adults, increased ART also indirectly reduces active TB incidence in the HIV-negative population, as a result of reduced transmission probability. 26 | RESULTS CONTENTS TABLE OF Figure 3.16 Importance of ART coverage for future a. TB-related deaths and b. TB incidence in HIV- negative adults a. TB-related deaths, adults 15+ EXECUTIVE SUMMARY 30k TB-related number of deaths per year 25k Model (70% ART by 2035) (k = thousands) 20k 15k Model (80% ART by 2035) INTRODUCTION 10k Model (90% ART by 2035) 5k 1 0 2018 2020 2022 2024 2026 2028 2030 2032 2034 Year POLICY QUESTIONS & METHODOLOGY b. TB incidence adjusted for notified EPTB, adults 15+ 120k 2 Number of incident cases (k = thousands) 100k 80k Model (70% ART by 2035) 60k RESULTS Model (80% ART by 2035) 3 40k Model (90% ART by 2035) 20k 0 CONCLUSION 2018 2020 2022 2024 2026 2028 2030 2032 2034 Year 4 Source: Optima TB model output. Note: ART = antiretroviral therapy; EPTB = extrapulmonary TB; TB = tuberculosis. 3.5 HOW TO OPTIMALLY ALLOCATE CURRENTLY KEY FINDINGS AVAILABLE RESOURCES FOR TB TREATMENT? 5 The analysis presented in this section addresses the core questions of this allocative efficiency study, looking at the entire TB response and determining how resources could be allocated to maximize health outcomes. RECOMMENDATIONS • How can the TB care cascade be improved and resource allocation be optimized? • How is Mozambique’s TB epidemic projected to change if we optimize the allocation of current funding available? 6 • How close will Mozambique get to international targets with an optimal allocation of funding? As outlined in above section, current TB spending and allocation patterns in Mozambique are projected to lead to a steady decline in TB incidence rates. The scope of this section is therefore to explore whether greater reductions in key indicators can be achieved by optimally re-allocating TB spending. ANNEXES RESULTS | 27 Optimized allocations of resources are only optimal relative to a specific set of objectives and within a given time frame. The optimization analysis was performed for a combination of three objectives: • Avert more new infections • Further reduce prevalence • Prevent additional TB deaths Overview of TB budget and spending in Mozambique Total national spending on TB prevention and care was estimated at ~USD 36.8 million in 2017 (Table 2.1), including: • NTP funding for diagnosis and treatment of pulmonary TB • Estimated expenditure: USD 3.4 million on preventive therapy for PLHIV (funded by the HIV program) • Estimated state funding through Ministry of Health hospital care and other services: USD 11 million • Estimated BCG vaccination: USD 0.9 million • Estimated expenditure on ART for PLHIV was USD 166 million (not included in this exercise) Figure 3.17 Overview of estimated TB expenditure (2017) 35m Ambulatory XDR Treatment Ambulatory MDR Treatment (short course) 30m Ambulatory MDR Treatment (long course) Ambulatory DS Treatment 25m USD per year (m = millions) Active case nding (prisons) Active case nding (community outreach) 20m Active case nding (miners) Active case nding (health workers) 15m HIV outpatient screening Outpatient screening (excluding PLHIV) 10m Preventive treatment (PLHIV) Preventive treatment (contacts of active TB, under 5yrs) 5m BCG vaccination 0m Current spending Source: Optima TB model output. Note: As data on TB spending by expenditure category was not available, estimates were made using a bottom-up costing approach; Both total expenditure and unit costs for interventions are therefore subject to some uncertainty; BCG = Bacille Calmette-Guerin; DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. Optimizing Mozambique’s TB program funding allocations Figure 3.18 and Table 3.6 show the overall optimized allocation of expenditure to minimize TB incidence, prevalence and deaths. In this analysis it was assumed that the same USD 36.8 million that were available for TB-related interventions in 2017 would remain available each year up to 2035. The 28 | RESULTS CONTENTS TABLE OF optimized budget allocation differs from current allocations in several areas, including: • Increasing annual funding for outpatient screening from USD 10.8 million to USD 13.2 million, including regular TB screening for all PLHIV at outpatient clinics (currently at 74%) • Prioritizing investment in case-finding programs for key risk groups, such as prisoners and health EXECUTIVE SUMMARY workers • Increasing funding for treatment programs to accommodate increased number of notifications Unit costs estimated using bottom up costing approach INTRODUCTION Table 3.6 Estimated unit cost per person (treatment) Unit cost (USD) 1 Preventive TB treatment (contacts of active TB - under 5) 8.1 Preventive TB treatment (PLHIV) 19.7 Ambulatory-focused DS treatment 87 POLICY QUESTIONS & METHODOLOGY Ambulatory-focused MDR treatment (long course) 3,493 Ambulatory-focused MDR treatment (short course) 1,573 2 Ambulatory-focused XDR treatment 13,198 Figure 3.18 Optimizing Mozambique’s TB program funding allocations 35m RESULTS Ambulatory XDR Treatment 3 Ambulatory MDR Treatment (short course) 30m Ambulatory MDR Treatment (long course) Ambulatory DS Treatment 25m Active case nding (prisons) CONCLUSION Active case nding (community outreach) USD per year 20m Active case nding (contact tracing) 4 Active case nding (miners) 15m Active case nding (health workers) HIV outpatient screening 10m Outpatient screening (excluding PLHIV) KEY FINDINGS Preventive treatment (PLHIV) Preventive treatment 5m 5 (contacts of active TB, under 5yrs) BCG vaccination 0m Optimized 100% Current spending RECOMMENDATIONS USD 36.8m USD 36.8m Source: Optima TB model output. Note: 2017 = base year (current allocation); Optimized budget: It was assumed expenditure of USD 36.8 million available for 6 TB-related programs in 2017 would remain available on an annual basis up to 2035; BCG = Bacille Calmette-Guerin; DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. ANNEXES RESULTS | 29 Table 3.7 Current and optimal allocations of 2017 TB spending, by intervention (in million USD) Optimized Current Program 100% (USD) spending (USD) Change (USD) Active case finding (contact tracing) 2,800,000 1,700,000 1,100,000 ▲ Active case finding (health workers) 130,000 38,000 92,000 ▲ Active case finding (miners) 370,000 250,000 120,000 ▲ Active case finding (community outreach) 3,100,000 8,900,000 -5,800,000 ▼ Active case finding (prisons) 560,000 120,000 440,000 ▲ Ambulatory DS treatment 7,900,000 6,700,000 1,200,000 ▲ Ambulatory MDR treatment (long course) 680,000 3,100,000 -2,420,000 ▼ Ambulatory MDR treatment (short course) 2,600,000 92,000 2,508,000 ▲ Ambulatory XDR treatment 710,000 670,000 40,000 ▲ Outpatient screening (excluding PLHIV) 9,500,000 8,900,000 600,000 ▲ HIV outpatient screening 3,700,000 1,900,000 1,800,000 ▲ Preventive treatment (contacts of active TB - under 5) 220,000 210,000 10,000 ▲ Preventive treatment (PLHIV) 3,500,000 3,400,000 100,000 ▲ Source: Optima TB model output. Note: DS = drug susceptible; HR = human resources; MDR = multi-drug resistant; PLHIV = people living with HIV; TB = tuberculosis; USD = United states dollar; XDR = extensively drug-resistant. Shifts within screening and diagnosis interventions Gaps in diagnosis represent a major break point in the TB care cascade in most countries and finding the “missing cases” is a key challenge for Mozambique’s TB program. An optimized allocation of resources would reduce total funding for screening and diagnosis, mainly due to reallocating resources to cheaper activities, including: (1) screening of PLHIV at each outpatient visits, (2) increasing funding for other outpatient screening such as cough monitors, (3) increasing coverage of programs targeting key high-risk groups such as prisoners, (4) doubling coverage of household contact tracing, and (5) reducing the house-to-house screening activities (Table 3.7 and Figure 3.19). Screening and diagnosis would then consume about 51% of total TB spending. Shifts within treatment programs In an optimized intervention mix, TB treatment would receive more funding and would absorb approximately 32% of total TB spending in Mozambique compared with the current 29%. This is due to the increased number of cases diagnosed as a result of increased funding for active case finding programs. • An optimal allocation increases annual funding for treatment programs as a result of a greater number of people diagnosed • Mozambique already treats all patients in the ambulatory setting, reducing costs, impacting on patients’ lives and the risk of nosocomial TB transmission • Both long-course and short-course regimens for MDR-TB in Mozambique now use bedaquiline (in-line with WHO guidance) 30 | RESULTS CONTENTS TABLE OF • Short-course regimens for MDR-TB are currently used for just a small number of patients • Although short-course regimens will not be suitable for all people with MDR-TB, increased coverage of the short-course regimen could significantly reduce costs • Future policy decisions on the choice of DR-TB regimens in Mozambique should be closely EXECUTIVE SUMMARY monitored and made based on local evidence when results for each cohort become available Figure 3.19 Optimal reallocation of current TB screening and prevention expenditure to minimize active pulmonary TB prevalence between 2017 and 2035 in Mozambique USD 25.5m INTRODUCTION 25m USD 23.9m Active case nding (prisons) 1 20m Active case nding (community outreach) USD per year (in millions) Active case nding (contact tracing) 15m Active case nding (miners) Active case nding (health workers) POLICY QUESTIONS & METHODOLOGY 10m HIV outpatient screening Outpatient screening (excluding PLHIV) 2 5m Preventive treatment (PLHIV) Preventive treatment (contacts of active TB, under 5yrs) 0m Optimized 100% Current spending Source: Optima TB model output. RESULTS Note: 2017 = base year (current allocation); Optimized budget: It was assumed expenditure of USD 36.8 million available for 3 TB-related programs in 2017 would remain available on an annual basis up to 2035; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. Table 3.8 Estimated impact of reaching 90% coverage of screening for key risk groups Potential CONCLUSION additional 4 Estimated yield from Estimated Estimated % current Target reaching Estimated Target population screened for Estimated spending for target additional population size TB 2017 yield (USD) screening coverage cost (USD) Adult PLHIV KEY FINDINGS (on ART) 1,100,000 48% 21,021 1,880,392 90% 18,579 1,661,948 5 Health workers 50,000 27% 395 38,030 90% 932 89,692 Household contact tracing 270,836 42% 6,224 1,657,168 90% 7,646 1,893,130 RECOMMENDATIONS Miners 70,000 29% 506 252,400 90% 1,069 533,357 Prisoners 80,000 15% 776 123,221 90% 3,755 596,493 6 Source: Optima TB model output. Notes: ART = antiretroviral therapy; PLHIV = people living with HIV; TB = tuberculosis; USD = United states dollar. ANNEXES RESULTS | 31 Figure 3.20 Optimal reallocation of current TB treatment expenditure to minimize active pulmonary TB prevalence between 2017 and 2035 in Mozambique 14m 12m USD 11.9m USD 10.6m 10m USD millions per year 8m Ambulatory XDR Treatment 6m Ambulatory MDR Treatment (short course) 4m Ambulatory MDR Treatment (long course) Ambulatory DS Treatment 2m 0 Optimized 100% Current spending Source: Optima TB model output Note: DS = drug susceptible; HR = human resources; MDR = multi-drug resistant; XDR = extensively drug-resistant. Improved TB outcomes under optimized resource allocation scenarios As shown in Figure 3.21, an optimized allocation of resources could have a substantial impact on key TB indicators. Relative to the current allocation, an optimized allocation of spending could: • Reduce the number of active TB infections in 2035 by 19% • Reduce the number of TB-related deaths per year in 2035 by 18% • Reduce the rate of TB incidence per 100k in 2035 by 11% Figure 3.21 Estimated number of people with active pulmonary TB 1.2k 1.0 Prevalent TB cases per 100k (k = thousands) 0.8 Current 0.6 Optimized 0.4 0.2 0 2018 2020 2022 2024 2026 2028 2030 2032 2034 Year Source: Optima TB model output. Note: Total annual expenditure is assumed constant at USD 36.8 million until 2035. Relative to the current allocation, an optimized allocation of spending could also reduce the annual number of TB-related deaths by 18% (by 2035) (Figure 3.22) 32 | RESULTS CONTENTS TABLE OF Figure 3.22 Estimated number of annual TB-related deaths 80k 70k TB-related deaths per year 60k EXECUTIVE SUMMARY (k = thousands) 50k 40k Current 30k Optimized 20k 10k INTRODUCTION 0 2018 2020 2022 2024 2026 2028 2030 2032 2034 1 Year Source: Optima TB model output. Note: Total annual expenditure is assumed constant at USD 36.8 million until 2035. POLICY QUESTIONS & METHODOLOGY Optimized allocations under different amounts of spending and their impact Figure 3.23 shows the optimal allocation at different spending levels. Screening of PLHIV in the 2 outpatient setting is continuously prioritized under various amounts of spending relative to current spending (80%, 100%, 120%, 150%, 200%). The pattern of optimized treatment expenditure remains consistent across spending levels, with a shift towards expanding active case finding programs as the budget increases. Preventive TB treatment for PLHIV remains outside the optimal allocation mix even at 200% of budget. With reduced expenditures, funding for community case finding programs are RESULTS no longer funded, with cheaper and more targeted programs prioritized such as screening of health 3 workers. Figure 3.23 Programs funded under different amounts of spending 70m Ambulatory XDR Treatment CONCLUSION Ambulatory MDR Treatment (short course) 60m Ambulatory MDR Treatment (long course) 4 Ambulatory DS Treatment 50m Active case nding (prisons) Active case nding (community outreach) 40m Active case nding (contact tracing) USD per year Active case nding (miners) KEY FINDINGS 30m Active case nding (health workers) HIV outpatient screening 5 20m Outpatient screening (excluding PLHIV) Preventive treatment (PLHIV) 10m Preventive treatment (contacts of active TB, under 5yrs) RECOMMENDATIONS 0 BCG vaccination Optimized Optimized Current Optimized Optimized Optimized 80% 100% spending 120% 150% 200% 6 Source: Optima TB model output. Note: BCG = Bacille Calmette-Guerin; DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi- drug resistant; PLHIV = people living with HIV; XDR = extensively drug-resistant. ANNEXES RESULTS | 33 Impact of different amounts of expenditure on TB outcomes (Figure 3.24 and 3.25). Reductions in TB expenditure to 80% of current levels, if optimally allocated, would still result in reduced prevalence and TB-related deaths as a result of reallocation to more cost-effective interventions. The optimized allocation of current expenditure is projected to yield significant gains, and there are significant gains from increasing the budget further, enabling further increases in the case detection rate through increased expenditure on active case finding programs. However, a larger increase in spending has diminishing marginal returns of impact. Furthermore, based on model outputs, it is not feasible to meet End-TB 2035 targets for reductions in TB incidence and TB-related deaths with any budget amount with the modelled interventions. Modelled TB incidence for 2017 was 491 per 100,000 population, meeting the End-TB target would require reducing the incidence rate to 49.1 per 100,000 by 2035. Expenditure of twice the 2017 level is estimated to reduce TB incidence to 147 cases per 100,000 population by 2035, hence missing the END-TB target (Figure 3.24). Figure 3.24 Impact of different amounts of expenditure on TB prevalence 1.2k 1.0k Prevalent TB cases per 100k Model (optimized 80%) 80 (k = thousands) Model (optimized 100%) 60 Model (current spending) Model (optimized 120%) 40 Model (optimized 150%) 20 Model (0ptimized 200%) 0 2018 2020 2022 2024 2026 2028 2030 2032 2034 Year Source: Optima TB model output. Note: TB = tuberculosis. Impact of different amounts of expenditure on TB-related mortality (Figure 3.25). • The number of modelled TB-related deaths for 2015 was around 54,000–meeting the End-TB target would require reducing the number of TB deaths to around 2,700 by 2035. • When the allocation is optimized, reducing the current budget by 20% would still result in a comparable number of TB-related deaths in 2035 • Budget increases buy additional impact–expenditure of twice the 2017 level is estimated to reduce the number of TB-related deaths to 8,500 by 2035 • Increasing expenditure on a finite set of programs will likely result in diminishing marginal returns. Therefore, consideration could be given to new prevention and treatment modalities 34 | RESULTS CONTENTS TABLE OF Figure 3.25 Impact of different amounts of expenditure on TB-related mortality 80k 70k TB-related deaths per year EXECUTIVE 60k SUMMARY Model (optimized 80%) (k = thousands) 50k Model (optimized 100%) 40k Model (current spending) 30k Model (optimized 120%) 20k Model (optimized 150%) INTRODUCTION 10k Model (0ptimized 200%) 0 1 2018 2020 2022 2024 2026 2028 2030 2032 2034 Year Source: Optima TB model output. POLICY QUESTIONS & METHODOLOGY Note: TB = tuberculosis. 2 Table 3.9 Impact of different amounts of expenditure on TB outcomes Optimized Optimized Current Optimized Optimized Optimized 80% 100% spending 120% 150% 200% TB prevalence in population RESULTS 0 –14 years 47,385 41,790 44,498 36,461 29,304 19,776 3 15+ years 225,528 202,901 214,684 180,439 148,631 103,404 PLHIV 60,020 54,446 78,144 47,802 40,011 33,002 Total 332,933 299,136 337,325 264,701 217,946 156,183 TB related deaths CONCLUSION 0 –14 years 2,809 2,587 2,723 2,377 2,079 1,625 4 15+ years 26,037 23,549 24,815 21,081 17,504 12,380 PLHIV 9,632 8,744 12,473 7,675 6,374 5,226 Total 38,477 34,880 40,011 31,132 25,957 19,232 KEY FINDINGS TB incidence in population 0 –14 years 24,181 23,079 24,884 22,001 20,562 18,218 5 15+ years 98,171 96,384 99,870 94,366 91,577 86,119 PLHIV 34,216 32,928 32,426 31,721 29,656 25,586 RECOMMENDATIONS Total 156,568 154,391 157,179 148,088 141,795 129,924 Source: Optima TB model output. 6 Note: PLHIV = people living with HIV; TB = tuberculosis. ANNEXES RESULTS | 35 This page is for collation purposes only. CONTENTS TABLE OF Relative to the current allocation (2017), an optimized allocation of current spending could, by 2035 reduce the number of active TB infections by 19%, TB-related deaths per year by 18% and TB incidence rate (per 100,000 population) by 11%. Expenditure of twice the 2017 level is estimated to reduce TB incidence to 147 cases per 100,000 population by 2035, hence missing the END-TB target. EXECUTIVE SUMMARY Increasing expenditure on a finite set of programs will likely result in diminishing marginal returns. Therefore, consideration could be given to new prevention and treatment modalities. INTRODUCTION 4 CONCLUSIONS 1 This section summaries key findings in responding to each of the policy questions. POLICY QUESTIONS & METHODOLOGY Policy question 1 What is the current status of the TB epidemic and TB care cascade for Mozambique using 2 the most recent available data? • TB incidence is amongst the highest in the world (551 cases per 100 000 in 2017) and 40% of TB cases are also HIV-positive • Diagnosis remains the most significant breakpoint in the TB care cascade for Mozambique: only RESULTS an estimated 50% of new active TB cases were diagnosed in 2018 and mortality prior to diagnosis 3 is estimated to be particularly high among PLHIV (44%, vs. 21% in HIV-negative adults) • As a result, despite high treatment success rates, only about 41% of all new active TB cases completed TB treatment (2018), emphasizing the importance of finding more cases Policy question 2 CONCLUSION What is the projected future trend of the TB epidemic in Mozambique under current level of budget and assuming status-quo programming? 4 • Assuming no changes in TB outcomes and TB program coverage (status quo), pulmonary TB incidence is projected to remain high, although decreasing, from 2017 (495 per 100,000) to 2035 (303 per 100,000) KEY FINDINGS • This is driven by the projected growth of the Mozambique population, estimated at a 70% increase in this time period 5 • While the TB incidence per 100 thousand population is decreasing, the number of new TB cases is projected to remain fairly constant at around 150,000 annually due to the rapid demographic growth RECOMMENDATIONS • New TB cases among PLHIV are projected to decrease as a result of ART coverage (anticipated to scale up to 90% by 2035) • New TB cases in the HIV-negative adult population are projected to continue to increase – from 6 ~76,000 in 2017 to ~113,000 by 2035 ANNEXES 37 Policy question 3 What is the projected impact on the TB epidemic if meeting key national and international targets? • Both NSP targets and Stop-TB targets aim for 90% diagnosis rate by 2025. Meeting and sustaining this target for 2025 is projected to yield significant reductions in the total number of active TB cases and TB-related deaths of up to 80% relative to current conditions by 2035 • Meeting NSP targets for diagnosis and treatment of DR-TB by 2025 could reduce DR-TB prevalence in 2035 by 50% relative to current conditions. With most DR-TB cases that are diagnosed with TB improving coverage of drug susceptibility testing is crucial to reaching this target • Although mainly managed by the HIV program, ART provision is a key factor in the TB response as treatment reduces the probability of progression from latent to active TB. For example, scaling up to 90% ART coverage by 2035 is projected to reduce the number of incident TB cases among adult PLHIV in 2035 by around 35% compared to ART coverage of 70% Policy question 4 What is the projected future trend of the country’s TB epidemic with optimized allocation of currently available resources? Relative to the current allocation (2017), an optimized allocation of current spending could, by 2035 reduce: • The number of active TB cases by 19% • The number of TB-related deaths per year by 18% • TB incidence rate (per 100,000 population) by 11% Policy question 5 What resources are required to achieve key targets of the national TB response? • This analysis estimated that about USD 36.8 million was spent in 2017 by government on the TB response, including resources from dedicated TB funding streams and from the wider health sector • The 2017 level of spending, even if allocated optimally, is unlikely to achieve End-TB targets of 95% reduction in deaths and 90% reduction in TB incidence by 2035 • Budget increases buy additional impact–doubling the 2017 spending level is estimated to reduce TB incidence to 147 cases per 100,000 population by 2035 • However, large increases of current spending have diminishing returns and it is not feasible to meet End-TB 2035 targets with any budget amount with the modelled interventions but instead requires additional and new tools 38 | CONCLUSION CONTENTS TABLE OF Increased tracing intensity would add to the yield of notified cases in Mozambique, and our optimization analysis indeed suggested that the effort of household contact tracing should be doubled. Screening PLHIV for active TB at outpatient clinics represents a relatively inexpensive and high yield way to increase the number of diagnosed cases in Mozambique. EXECUTIVE SUMMARY Expanding the use of rapid testing methods such as GeneXpert will reduce the number of false negatives and increase the notification rates for both DS and DR TB. More data should be collected on the diagnostic yield of different case finding strategies, as well as their costs, in order to inform local best practice in Mozambique. INTRODUCTION 5 DISCUSSION OF KEY FINDINGS 1 M ozambique is one of the 30 highest TB burden countries in the world with an estimated 551 POLICY QUESTIONS & METHODOLOGY new TB cases per 100,000 population, a low TB detection rate of just above 50%, and a low DR-TB diagnosis and treatment success rates (38%). Overall, data from NTP suggests 2 high treatment success rates for DS-TB (90%) and improvements in TB case detection rates over time. However, it is still far from the global target detection rates, indicating challenges for the country to reach NTP targets for 2025 as well as End-TB targets for 2035. More efforts and resources are needed for Mozambique to reduce the TB burden. Our modelling exercise shows a declining trend in the risk of TB infections and the risk of TB related RESULTS deaths, thanks to the rapidly improved HIV treatment coverage and TB services. However, due to the 3 rapidly-increasing population size, the absolute number of TB cases continues to increase, especially among the HIV negative population. At the time of analysis, Mozambique was working on the National TB Strategic Plan 2020-2025 and our analysis shows that reaching a 90% diagnosis rate by 2025 would reduce the total number of TB cases and TB-related deaths of up to 80%, and reduce DR-TB of up to 50% (relative to current condition, by 2035). In addition, reaching 90% ART coverage by 2035 CONCLUSION would contribute to a reduction of new TB cases of 35%. To reach END-TB targets, the country must 4 increase TB detection rates, and at the same time, improve ART coverage and continue to prevent new HIV infections. The country has achieved high coverage of HIV services in TB clinics with 97% of TB patients registered aware of their HIV status and 95% of TB patients living with HIV on ART. Ongoing, regular TB screening and diagnosis among HIV patients remain critical, as is patient support for IPT which has been reported to suffer from low uptake and poor completion rates (Rodriques & KEY FINDINGS Lisboa, 2019). 5 With the current level of budget and TB spending (2017), however, the country can cut TB prevalence and death by nearly 20%, and TB incidence by 11% by allocating resource optimally. Specifically, this can be done by doubling the average number of contacts traced per notified case, screening all PLHIV during their routine outpatient visits, and focusing on the community outreach activities among key RECOMMENDATIONS populations such as prisoners, cross-border miners and health workers. Currently, an estimated 1.3 contacts per active TB case are traced. Increased tracing intensity would add to the yield of notified 6 cases in Mozambique, and our optimization analysis indeed suggested that the effort of household contact tracing should be doubled. A study on systematic contact tracing of MDR patients in a rural area demonstrated the value of comprehensive contact investigations in finding additional cases and getting them on treatment early (Pires Machai, 2019).. However, the most effective way of conducting household contact tracing remains uncertain and is beyond the scope of this analysis. ANNEXES 39 In the optimal resource allocation scenario (compared to the status quo), both community case finding and active case finding amongst miners received less funding according to model outputs, but this was based on limited data on coverage and yield of these programs in Mozambique. Instead, we used global literature estimates for our modelling analysis. Generally, the delivery of case finding services in Mozambique remains fragmented, with multiple implementing partners operating to provide services. A study on active case finding approaches in the communities (using lay counsellors) and at health facilities (with cough monitors) demonstrated the importance of using combined approaches to find cases while ensuring high linkage to care rates of TB suspects identified in the community (Polana et al. 2019). Our optimization analysis suggested that increasing funding for outpatient TB symptom screening, through expanding initiatives such as cough monitors, is a recommended resource shift. In any case, more data should be collected on the diagnostic yield of different case finding strategies, as well as their costs, in order to inform local best practice in Mozambique. Screening PLHIV for active TB at outpatient clinics represents a relatively inexpensive and high yield way to increase the number of diagnosed cases in Mozambique. In 2017, around 74% of new cases enrolled in HIV treatment were screened for TB. The total number of PLHIV in care screened for TB is unknown. On the basis of the estimated active case finding yield in this setting (Shapiro et al, 2013), we estimated that around half of PLHIV on treatment were screened for TB in 2017 for the purposes of this analysis. Although there is some uncertainty in this estimate, all PLHIV should nevertheless be regularly screened for TB in the outpatient setting. Other high-risk groups include prisoners and health workers. Both of these groups are relatively easy to find and should be screened for TB at least annually. The Challenge TB Project also concluded from an exercise extending HCW screening to community health workers that this is an essential occupational group to be included in regular HCW screening campaigns to find cases among them early and treat appropriately (Abdula, 2019). Tackling low TB detection rates, especially for DR-TB, also requires laboratory capacity, which is beyond the scope of this analysis. Data and expert opinions in the country indicate low laboratory capacity, resulting in a high rate of TB patients identified by clinical symptoms (60% in 2017). Concerns were raised by local experts over the quality of sputum samples. A video created by TB Reach three years ago that has been translated into Portuguese and five local languages in Mozambique on best practices in collecting sputum could be shown to nurses, community health workers and presumptive TB cases. A pilot could be run in certain districts to see if using this video improves sputum quality and subsequent yield. Increased coverage of bacteriological testing would also require adequate sample transportation systems to ensure sputum quality. Such issues of implementation capacity should be carefully considered in any scale-up of laboratory testing. Additionally, the country currently has about 100 GeneXpert machines and has planned to deploy more. Expanding the use of rapid testing methods such as GeneXpert will reduce the number of false negatives and increase the notification rates for both DS and DR TB. However, the overall impact of scaling up GeneXpert is difficult to estimate due to limited data available in Mozambique. We recommend that the NTP conducts operational research to investigate universal GeneXpert coverage in areas where it was previously unavailable, while also considering the necessary placement conditions. This would provide valuable information to the country on the roll-out of the technology, including its cost-effectiveness and impact. Several studies have also highlighted the importance of training on GenXpert use to address under-utilization of existing GenXpert test capacity (Jaintila et al., 2019) and the high rates of unsuccessful Xpert MTB/RIF results and associated financial costs (Teixeira et al., 2019). During the course of this analysis, we were requested by local stakeholders to explore the impact of community-based interventions such as the use of CHWs and traditional healers in case finding. 40 | KEY FINDINGS CONTENTS TABLE OF However, due to insufficient data, we were not able to include this scenario in our analysis. Based on NTP data, community case finding programs already contribute around 25% of notified cases, playing a role in making progress towards case detection targets. It may be that that the increased rates of clinical diagnosis of TB may be due to the expanded use of CHWs and traditional healers who often EXECUTIVE SUMMARY work in rural areas with limited access to smear and Xpert testing. While accelerated TB case finding is essential in Mozambique, suitable capacity needs to be built to correctly diagnose cases, including resistance typing, so that the correct regimens can be prescribed and patients attain treatment success. These uncertainties can be answered by a small-scale implementation research study. INTRODUCTION 1 POLICY QUESTIONS & METHODOLOGY 2 RESULTS 3 CONCLUSION 4 KEY FINDINGS 5 RECOMMENDATIONS 6 ANNEXES Photo: National Tuberculosis Programme. Used with permission. KEY FINDINGS | 41 This page is for collation purposes only. CONTENTS TABLE OF EXECUTIVE SUMMARY INTRODUCTION 6 RECOMMENDATIONS 1 Based on the key findings, we make the following recommendations: POLICY QUESTIONS & METHODOLOGY 1. Continue focusing on contact-tracing of notified cases As discussed above, the number of traced contacts should increase by 50–100% relative 2 to current conditions. While all index patients should be followed up with contact tracing, those sputum-smear positive, those with (suspected) DR-TB, and those with children or immunocompromised people among household contacts should have greatest priority for comprehensive tracing of all contacts. In addition, the yield data of contact-tracing of notified cases for Mozambique is uncertain and should be monitored. RESULTS 2. 3 Screen all PLHIV for TB symptoms at each outpatient visit About 74% of new HIV cases enrolled in ART were screened for TB in 2017. Given a high TB prevalence in PLHIV, screening all HIV patients regularly for TB as per global recommendation will likely increase yield. 3. CONCLUSION Expand active case finding programs for key high-risk populations Health care workers, prisoners, and cross-border miners should be screened annually as 4 these groups are at higher risk of TB and are relatively easy to find. 4. Increase yield of notified cases from outpatient screening and reduce clinical diagnosis About 75% of notified cases are found in the outpatient setting in Mozambique. As 60% of cases are clinically diagnosed (2017), increased funding should aim to increase the KEY FINDINGS proportion of TB cases that are bacteriologically confirmed, by increasing coverage of 5 tests such as GeneXpert. Without increasing bacteriological confirmation rate, it is difficult to assess true progress in the TB response in Mozambique. We recommend small-scale implementation science research to assess the effectiveness, cost, and operational aspects of the introduction and universal use of GeneXpert. RECOMMENDATIONS 5. Continue expanding ART coverage and ART adherence support Increasing ART coverage has a significant impact on TB incidence amongst PLHIV and 6 continued expansion of ART care will significantly reduce TB infection. ANNEXES 43 6. Closely monitor the pilot of MDR regimens to inform future MDR treatment recommendations The NTP is planning to move away from all injectable based MDR regimens from late 2019 onward. The oral short course regimen has been provided in several high volume MDR-TB sites in the Maputo City area. By the time of this analysis, data from these research studies was not yet available. Final discussions about investments in these short course regimen should be based on the outcomes from these studies. 7. Continue ambulatory-focused care for both DS-TB and DR-TB patients The country should continue this approach and avoid unnecessary hospitalisation as long as it reduces costs without affecting outcomes, and provided directly observed treatment (DOT) and adherence support are in place. Community health workers are well placed to support TB patients during treatment in addition to their important role in contact tracing. 8. Collect data on community case finding programs to inform decision making As discussed above, community case finding programs contribute to around 25% of notified cases in Mozambique. Currently, several community case-finding programs are underway with fragmented implementation and lack of data on cost, coverage, and impact. We recommend the NTP collects data, especially cost data, of these pilot projects in order to inform cost-effectiveness estimates and guide future policy recommendations. 9. Maximize the collection and use of TB routine data to inform programming and policies The quality and availability of cost and coverage data in Mozambique provides a challenge for any allocative efficiency analysis, and results in uncertainty in epidemic and programmatic model parameters. Key data sources include TB-specific expenditures, reports on how TB cases are identified (by intervention modality), and better records of often fragmented implementation of service delivery activities. Therefore, monitoring and evaluation systems should be streamlined, and spending and coverage data should be collected for all TB interventions and programs (NTP led and non-NTP led). 10. Increase funding for TB program The TB response in Mozambique is not on track to meet the 2025 milestones or 2035 End-TB targets, more funding for the TB program is needed as additional investment will provide more impact and close the gap to reaching TB strategic targets. 44 | RECOMMENDATIONS CONTENTS TABLE OF EXECUTIVE SUMMARY INTRODUCTION ANNEXES 1 The Optima mathematical modelling suite was designed to support decision-makers in prioritization, POLICY QUESTIONS & METHODOLOGY resource allocation and planning to maximise health impact. Optima-HIV was the most widely used component of the Optima modelling suite. A more detailed summary of the model and methods is 2 provided elsewhere. Optima TB is a mathematical model of TB transmission and disease progression integrated with an economic and program analysis framework. Optima uses TB epidemic modeling techniques and incorporates evidence on biological transmission probabilities, detailed disease progression and population mixing patterns. Optima TB is a compartmental model, which disaggregates populations RESULTS into different model compartments including susceptible, vaccinated, undiagnosed early or late 3 latent-TB, diagnosed early or late latent-TB, on treatment early or late latent-TB, undiagnosed active TB, diagnosed active TB, on treatment and recovered active-TB populations. In addition, active-TB compartments are further disaggregated by drug resistance type into drug susceptible (DS), multi-drug resistant (MDR) and extensively drug resistant (XDR). Annex 2 summarises the main features of Optima TB. CONCLUSION 4 KEY FINDINGS 5 RECOMMENDATIONS 6 ANNEXES 45 This page is for collation purposes only. CONTENTS TABLE OF < Annex 1 OPTIMA TB MODEL FEATURES AND KEY DEFINITIONS AT A GLANCE Disaggregation by EXECUTIVE SUMMARY smear-status and drug-resistance • Both smear positive and negative; DS-TB, MDR-TB, XDR-TB New vs. relapse cases • The WHO definition for incident TB cases includes both new and relapse cases. In the model, incident TB cases correspond to the following transitions between compartments: INTRODUCTION • New cases: these are represented by the number of progressions to active TB from early and late latent-TB compartments. ‘New’ also 1 includes recurring episodes of TB from the recovered compartment following re-infection • Relapse cases: these correspond to all unsuccessful treatments in the model, which include failure, relapse, LTFU and re-treatments POLICY QUESTIONS & METHODOLOGY Latent TB • Multiple compartments for latent TB infection (LTBI) • Cannot skip latent state for disease progression 2 • States include undiagnosed, on treatment, and completed treatment • Accounts for re-infection and latent care-status using a secondary latent TB pathway. Cases previously treated for LTBI, or vaccinated individuals, can transition to the active TB pathway in the case of reinfection RESULTS Vaccination, immunity • Vaccination explicitly included in model 3 and resistance • Patients that spontaneously clear from infection Treatment • States for undiagnosed, diagnosed, diagnosed but not on-treatment, on-treatment, and recovered patients for different types of drug- resistance CONCLUSION • Failed or defaulted treatment can acquire drug resistance Treatment outcomes • Treatment success includes ‘cured’ and ‘treatment completion’, as per 4 the WHO • Treatment failure in the model includes ‘loss to follow-up’ during treatment, ‘treatment failure’, and ‘not evaluated’ • Death during TB treatment is not included in treatment failure, but is KEY FINDINGS considered separately Population structure, key • Age-structured populations can be user defined 5 populations and People • Ability specify additional key populations with defined transition living with HIV rates to/from general population groups • HIV positive populations represented as separate key population RECOMMENDATIONS Optima TB is based on a dynamic, population-based TB model (Annex 2). The model uses a linked system of ordinary differential equations to track the movement of people among health states. 6 The overall population is partitioned in two ways: by population group and by TB health state. TB infections occur through the interactions among different populations. Each compartment in Annex 2 corresponds to a single differential equation in the model, and each rate (Annex 2 arrows) corresponds to a single term in that equation. The analysis interprets empirical estimates for model parameter values in Bayesian terms as previous distributions. The model then ANNEXES ANNEXES | 47 must be calibrated: finding posterior distributions of the model parameter values so that the model generates accurate estimates of notified TB cases, TB incidence, TB prevalence, the number of people on treatment, and any other epidemiological data that are available (such as TB-related deaths). Model calibration and validation normally should be performed in consultation with governments in the countries, in which the model is being applied. 48 | ANNEXES CONTENTS TABLE OF Annex 2 TUBERCULOSIS MODEL STRUCTURE Figure A.2.1 Schematic diagram of the health state structure of the model EXECUTIVE SUMMARY PRE INFECTION Recovery Completed RECOVERY TB RELATED DEATH Relapse treatment (active) Susceptible Natural Recovery Susceptible Vaccinated (diagnosis restricted) On treatment On treatment On treatment On treatment On treatment On treatment SP-DS SP-MDR SP-XDR SN-DS SN-MDR SN-XDR INTRODUCTION LATENT INFECTION 1 Early latent Diagnosed Diagnosed Diagnosed Diagnosed Diagnosed Diagnosed Early latent Early latent (diagnosis untreated on treatment SP-DS SP-MDR SP-XDR SN-DS SN-MDR SN-XDR restricted) POLICY QUESTIONS & METHODOLOGY Late latent Late latent Undiagnosed Undiagnosed Undiagnosed Undiagnosed Undiagnosed Undiagnosed Late latent (diagnosis untreated on treatment SP-DS SP-MDR SP-XDR SN-DS SN-MDR SN-XDR restricted) 2 ACTIVE INFECTION ACTIVATION SMEAR SMEAR POSITIVE NEGATIVE RESULTS Source: Prepared based on model structure 3 Note: Each compartment represents a single population group with the specified health state. Each arrow represents the movement of numbers of individuals between health states. All compartments except for “susceptible” and “vaccinated” represent individuals with either latent or active TB. Death can occur for any compartment, but TB related mortality varies between compartments. SN-DS = smear negative drug susceptible; SP-DS = Smear-positive drug susceptible; SP-MDR =smear positive-multi-drug resistant; SN-MDR =smear negative-multi-drug resistant; SN-XDR = smear nega- CONCLUSION tive-extensively drug-resistant; TB = tuberculosis. TB Resource Optimization and Program Coverage Targets 4 Optima TB is able to calculate allocations of resources that optimally address one or more TB-related objectives (for example, impact-level targets in a country’s TB national strategic plan). Because this model also calculates the coverage levels required to achieve these targets, Optima TB can be used to inform TB strategic planning and the determination of optimal program coverage levels. KEY FINDINGS The key assumptions influencing resource optimization are the relationships among (1) the cost of 5 TB programs for specific target populations, (2) the resulting coverage levels of targeted populations with these TB programs, and (3) how these coverage levels of TB programs for targeted populations influence screening and treatment outcomes. Such relationships are required to understand how incremental changes in spending (marginal costs) affect TB epidemics.1 RECOMMENDATIONS To perform the optimization, Optima uses a global parameter search algorithm, which is an adaptive stochastic descent algorithm. The algorithm is similar to simulated annealing in that it makes stochastic 6 1 A traditional approach is to apply unit cost values to inform a linear relationship between money spent and coverage attained. This assumption is reasonable for programs such as an established treatment program that no longer incurs start-up or initiation costs. However, the assumption is less appropriate for diagnostic programs. Most programs typically have initial setup costs, followed by a more effective scale-up with increased funding. However, very high coverage levels have saturation effects because these high levels require increased incremental costs due to the difficulty ANNEXES of diagnosing more people as the yield of diagnostic interventions declines. ANNEXES | 49 downhill steps in parameter space from an initial starting point. However, unlike simulated annealing, the algorithm chooses future step sizes and directions based on the outcome of previous steps. For certain classes of optimization problems, the team has shown that the algorithm can determine optimized solutions with fewer function evaluations than traditional optimization methods, including gradient descent and simulated annealing. Uncertainty Analyses Optima uses a Markov chain Monte Carlo algorithm for performing automatic calibration and for computing uncertainties in the model fit to epidemiological data. With this algorithm, the model is run many times (typically, 1,000–10,000) to generate a range of epidemic projections. Their differences represent uncertainty in the expected epidemiological trajectories. The most important assumptions in the optimization analysis are associated with the cost-coverage and coverage-outcome curves.2 2 All available historical spending data and achieved outcomes of spending, data from comparable settings, experience, and extensive discussion with stakeholders in the country of application can be used to inform these ranges. All logistic curves within these ranges then are allowable and are incorporated in Optima uncertainty analyses. These cost-coverage and coverage-outcome curves thus are reconciled with the epidemiological, and biological data in a Bayesian optimal way, thereby enabling the calculation of unified uncertainty estimates. 50 | ANNEXES CONTENTS TABLE OF Annex 3 POPULATION SIZES Population name Value Year Source or Assumption 0–14 years old 13,410,016 2017 EXECUTIVE SUMMARY UN Population Division; 15+ years old 14,258,818 2017 UNAIDS 15+ years old HIV+ 2,000,000 2017 Annex 4 BIRTHS AND BACKGROUND (NON-TB) MORTALITY INTRODUCTION Population name Value Year Source or Assumption 1 Annual number of births 1,122,995 2016 Provided by country Annual non-TB death rate, 0–14 years old 0.72% 2017 Insitute for Health Metrics and Evaluation, Global Burden of POLICY QUESTIONS & METHODOLOGY Disease study 2016.3 Annual non-TB death rate, 15+ years old 0.53% 2017 Insitute for Health Metrics and Evaluation, Global Burden of 2 Disease study 2016. Annual non-TB death rate, 15+ years old 1.24% 2017 Insitute for Health Metrics and HIV+ Evaluation, Global Burden of Disease study 2016. RESULTS Note: TB = tuberculosis. 3 CONCLUSION 4 KEY FINDINGS 5 RECOMMENDATIONS 6 3 All available historical spending data and achieved outcomes of spending, data from comparable settings, experience, and extensive discussion with stakeholders in the country of application can be used to inform these ranges. All logistic curves within these ranges then are allowable and are incorporated in Optima uncertainty analyses. These cost-coverage and coverage-outcome curves thus are reconciled with the epidemiological, and biological data in a Bayesian optimal ANNEXES way, thereby enabling the calculation of unified uncertainty estimates. ANNEXES | 51 Annex 5 TB EPIDEMIOLOGICAL PARAMETERS Latest year or default Full Name Population value Source or Assumption Vaccination Rate Annual number of births 98.7% Provided by country Early Latency Houben et al. 2016 (appendix of TIME Departure Rate model) - 0.1%/year reactivation rate All populations 0.2001 (0.01– 0.25). Late Latency Andrews et al. 2012 - risk of progression to Departure Rate* All populations 0.003 active. Probability of Early-Active vs. Early-Late LTBI Andrews et al. 2012 - risk of progression to Progression* All populations 0.177 active. Infection Vulnerability All populations 0.5 Mantgani et al., 2013 (protective efficacy of Factor (Vaccinated vs. BCG found to range from 0-80%). A value Susceptible) of 0.5 was used for populations aged 0–14, and no protection (i.e., 1) was used for all populations older than 14 years old. Infection A value of '1' is the default, but this is likely vulnerability factor to be significantly higher in vulnerable (relative population populations such people living with HIV. susceptibility) All populations 1.0 Values between 1–11 were used in calibrations Smear positive (SP) TB Infectiousness* All populations 1.0 A value of '1' is the default Smear negative (SN) TB Infectiousness (Compared to SP-TB) All populations 0.22 Behr et al.1999 Active Infection Rate This value is representative of a global (Active Recovered)* All populations 0.02 average Smear positive TB natural recovery rate All populations 0.03 Tiemersma et al. 2011 Smear negative TB natural recovery rate All populations 0.16 Tiemersma et al. 2011 Smear positive untreated-TB death rate All populations 0.12 Tiemersma et al. 2011 Smear negative untreated-TB death rate All populations 0.02 Tiemersma et al. 2011 Source: Note: * = Parameters with the least confidence/available literature, and chosen across different studies to be adjusted to calibrate the model. Not all of these apply to the calibration process in Mozambique; Notified cases disaggregated by age and resistance-type were provided by the country; BCG = Bacille Calmette-Guerin; TB = tuberculosis; LTBI = latent TB infection. 52 | ANNEXES CONTENTS TABLE OF Annex 6 NUMBER OF NOTIFIED PULMONARY TB CASES BY AGE AND DRUG RESISTANCE TYPE (2017) Number of Number of Number of EXECUTIVE SUMMARY notified notified notified Total number Population DS-TB cases MDR-TB cases XDR-TB cases of notified cases 0–14 years old 10,398 36 1 10,435 15+ years old 40,625 551 18 41,194 15+ years old HIV+ 27,642 375 12 28,029 INTRODUCTION Notes: DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. 1 Annex 7 CALIBRATION METHODOLOGY UNCERTAINTIES POLICY QUESTIONS & METHODOLOGY IN DISAGGREGATING TUBERCULOSIS INCIDENCE BY HIV STATUS 2 • Mozambique NTP did not have notifications data disaggregated by HIV status • It was therefore necessary to use WHO data to estimate the proportion of TB patients in Mozambique who are also HIV-positive • As there appear to be some inconsistencies in WHO data for earlier years, values before 2010 were RESULTS not used in the model calibration 3 • Instead, a baseline value of 61% (IHME estimate for 2001) was used, and a linear increase each year until 2010 (63.6% WHO) was assumed Figure A.7.1   Uncertainties in disaggregating tuberculosis incidence by HIV status CONCLUSION 120,000 4 100,000 HIV+ 80,000 Number of people 60,000 HIV– KEY FINDINGS 40,000 5 20,000 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 RECOMMENDATIONS Year 6 ANNEXES ANNEXES | 53 Annex 8 EPIDEMIOLOGICAL ASSUMPTIONS Epidemiological parameter Assumption Smear-Status Smear status results for DS-TB were based on notifications data for 2001-2017 reported by the Mozambique National Tuberculosis program Treatment Based on those reported by the WHO. These are disaggregated by resistance type Outcomes and HIV-status but not by agein the absence of population-specific data. Active TB Used the WHO active TB incidence estimates, and assumed that pulmonary incidence notified cases are a best indicator of the specific burden within population groups. These were used to disaggregate the WHO estimate into sub-population pulmonary TB incidence. Active/Latent-TB Used Houben and Dodd (2016) latent-TB prevalence estimates as a basis for Prevalence initialization estimates and comparison with model outputs. Although WHO estimates for active TB prevalence were also used in the same way, these do not appear to be reliable as they report estimated prevalence equal to estimated incidence for the most recent year in which estimates are available (2014). There is considerable uncertainty in actual prevalence of active TB in Mozambique until the results of the prevalence survey that is currently underway are available. Annex 9 PROGRAMMATIC DATA: SCREENING AND DIAGNOSTICS Unit cost/ person screened Number of Total estimated Source or Intervention (USD), 2017 screens, 2017 spending (EUR) Yield assumption Outpatient screening (excluding PLHIV) 3.58 525,525 8,916,890 4.00% Shapiro et al., 2013 PLHIV outpatient Kranzer, 2010; screening 1.07 8,298,182 1,880,392 0.45% Kranzer, 2012 Active case finding (health workers) 2.84 13,399 38,030 2.95% Local data Active case finding (contact tracing ) 14.57 113,776 1,657,168 5.88% Shapiro et al., 2013 Active case finding (community outreach) 12.01 741,867 8,910,671 1.85% NTP desk review Active case finding (miners) 12.47 20,237 252,400 2.50% Shapiro et al., 2013 Active case finding (prisons) 10.00 12,327 123,221 6.29% Shapiro et al., 2013 Note: All unit costs were derived by the authors of this report using data provided by country colleagues; PLHIV = people living with HIV. 54 | ANNEXES CONTENTS TABLE OF Annex 10 SENSITIVITY OF SCREENING /TESTING METHODS Screening or testing method Sensitivity Source or assumption Full symptom screen 84% Van’t Hoog et al., 2013 EXECUTIVE SUMMARY Sputum Smear microscopy 61% Van’t Hoog et al., 2013 Xpert MTB/RIF 89% Van’t Hoog et al., 2013 Clinical diagnosis 24% Van’t Hoog et al., 2013 INTRODUCTION Annex 11 TB TREATMENT INTERVENTIONS: TARGET 1 GROUPS, UNIT COSTS, VOLUME, TOTAL SPEND AND OUTCOME Unit cost/ Total POLICY QUESTIONS & METHODOLOGY course of Patients estimated Treatment treatment covered, Source or spending Treatment Source or program (USD) 2017 assumption (USD) success assumption 2 Ambulatory Number of Current treatment DS treatment 86.53 77,091 notified cases 6,670,684 88.0% outcomes Ambulatory Number of MDR notified cases. WHO Review of RESULTS treatment NTP data on the evidence on 3 (long course) 2,563.00 884 regimen coverage. 2,266,387 69.3% Bedaquiline, 2016 Ambulatory Number of MDR notified cases. WHO Review of treatment NTP data on the evidence on CONCLUSION (short course) 794.00 59 regimen coverage. 46,586 69.3% Bedaquiline, 2016 Ambulatory 8,021.00 51 Number of 4 XDR notified cases. Pym et al., 2016; treatment NTP data on WHO report on regimen coverage. 409,093 66.4% Bedaquiline, 2016 Note: DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people KEY FINDINGS living with HIV; USD = United states dollar; XDR = extensively drug-resistant. 5 RECOMMENDATIONS 6 ANNEXES ANNEXES | 55 Annex 12 COMPONENT COSTS OF TB TREATMENT REGIMENS (USD) Drug Patient Treatment regimen Inpatient Outpatient Monitoring support Total unit programs cost* costs* costs* costs* costs* cost* Ambulatory-focused DS treatment 22.57 0.00 48.30 15.66 0.00 86.53 Ambulatory-focused MDR treatment (long course) 3,154.94 46.90 68.19 150.39 72.41 3,492.84 Ambulatory-focused MDR treatment (short course) 1,250.75 46.90 52.56 150.39 72.41 1,573.02 Ambulatory-focused XDR treatment 12,846.95 54.57 73.87 150.39 72.41 13,198.19 Note: * = all costs in USD; DS = drug susceptible; HR = human resources; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = United states dollar; XDR = extensively drug-resistant. Annex 13 COST OF MDR-TB REGIMEN (SHORT COURSE) MDR short-course regimen Cost (USD) Bedaquiline 528.60 Clofazimine 304.91 Cycloserine 140.35 Levofloxacin 35.40 Linezolid 241.50 Total 1,250.75 Note: MDR = multi-drug resistant. Annex 14 COST OF MDR-TB REGIMEN (LONG COURSE) MDR long-course regimen Cost (USD) Bedaquiline 1,120.82 Clofazimine 677.58 Cycloserine 311.88 Levofloxacin 78.66 Linezolid 966.00 Total 3,154.94 Note: MDR = multi-drug resistant. 56 | ANNEXES CONTENTS TABLE OF Annex 15 COST OF XDR-TB REGIMEN XDR regimen Cost (USD) Bedaquiline 1,120.82 EXECUTIVE SUMMARY Clofazimine 1,626.19 Delamanid 8,379.36 Cycloserine 561.38 Linezolid 1,159.20 INTRODUCTION Total 12,846.95 Note: XDR = extensively drug-resistant 1 Annex 16 CONSTRAINTS FOR PROGRAM FUNDING POLICY QUESTIONS & METHODOLOGY The following minimum and maximum funding amounts for specific programs were included to match constraints on program funding 2 Minimum coverage (%) Maximum coverage (%) BCG Vaccination 98.7% of newborns – Preventive treatment (contacts 100% of all identified contacts of active of active TB - under 5) TB aged 5 and under – RESULTS Treatment for DR-TB cases 100% of diagnosed DR-TB cases to 3 receive treatment for DR-TB – Note: BCG = Bacille Calmette-Guerin; TB = tuberculosis; DR = drug resistant. CONCLUSION Annex 17 NUMBER OF NOTIFIED PULMONARY TB 4 INFECTIONS PER POPULATION GROUP (2017) Population DS-TB MDR-TB XDR-TB Total notified 0–14 years 10,398 36 1 10,435 KEY FINDINGS 15+ years 40,625 551 18 41,194 5 15+years HIV+ 27,642 375 12 28,029 Total 78,665 962 31 79,658 Note: DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; XDR = extensively drug-resistant. RECOMMENDATIONS 6 ANNEXES ANNEXES | 57 Annex 18 MODELLED TB INCIDENCE PER 100K BY POPULATION GROUP Population 2017 2022 2035 0–14 years 144 164 127 15+ years 519 534 406 15+years HIV+ 2,752 1,783 545 Total 495 447 303 Annex 19 MODELLED LATENT INFECTION AND DEATHS (2017) Annual number of Population Number of latent TB cases TB-related deaths per year 0–14 years 1,108,440 1,863 15+ years 8,180,868 18,412 15+years HIV+ 1,203,562 30,094 Total 10,492,870 50,369 Note: TB = tuberculosis. Annex 20 DEFINING NEW AND RELAPSE CASES IN THE MODEL The WHO definition for incident TB cases includes both new and relapse cases In the model, incident TB cases correspond to the following transitions between compartments: • New cases:  these are represented by the number of progressions to active TB from early and late latent-TB compartments. ‘New’ also includes recurring episodes of TB from the recovered compartment following re-infection • Relapse cases:  these correspond to all unsuccessful treatments in the model, which include failure, relapse, LTFU and re-treatments Annex 21 DEFINING TREATMENT OUTCOMES IN THE MODEL • Treatment success includes ‘cured’ and ‘treatment completion’, as per the WHO definition • Death during TB treatment is not included in treatment failure, but is considered separately • Treatment failure and ‘loss to follow-up’ during treatment are included as separate outcomes in the model 58 | ANNEXES CONTENTS TABLE OF Annex 22 OPTIMAL ALLOCATION OF MOZAMBIQUE’S TB EXPENDITURE – SCREENING (100% OF CURRENT SPENDING) EXECUTIVE SUMMARY 2017 Spending Optimized Program Name (USD) spending (USD) Change Outpatient screening (excluding PLHIV) 8.917 17.833 8.916 PLHIV outpatient screening 1.880 3.746 1.865 Active case finding (health workers) 0.038 0.128 0.090 INTRODUCTION Active case finding (miners) 0.252 0.191 -0.062 1 Active case finding (contact tracing ) 1.657 3.554 1.896 Active case finding (community outreach) 8.911 2.093 -6.818 Active case finding (prisons) 0.123 0.720 0.597 POLICY QUESTIONS & METHODOLOGY Total Screening 21.779 28.264 Note: PLHIV = people living with HIV. 2 Annex 23 OPTIMAL ALLOCATION OF MOZAMBIQUE’S TB EXPENDITURE – TREATMENT AND PREVENTION (100% OF CURRENT SPENDING) 2017 Spending Optimized RESULTS Program Name (USD) spending (USD) Change 3 Preventive TB treatment (contacts of active TB - under 5) 1.540 2.456 0.916 Preventive TB treatment (PLHIV) 12.026 0.000 -12.026 Ambulatory-focused DS treatment 6.671 9.851 3.181 CONCLUSION Ambulatory-focused MDR treatment (long course) 2.266 3.607 1.341 Ambulatory-focused MDR treatment (short course) 0.248 0.040 -0.208 4 Ambulatory-focused XDR treatment (short course) 0.409 0.682 0.272 Total treatment and prevention 24.067 17.542 Note: DS = drug susceptible; TB = tuberculosis; MDR = multi-drug resistant; PLHIV = people living with HIV; USD = KEY FINDINGS United states dollar; XDR = extensively drug-resistant. 5 RECOMMENDATIONS 6 ANNEXES ANNEXES | 59 REFERENCES Abdula (2019). Intensifying TB screening among community health workers: experience of project Challenge TB, Mozambique. EP-16-250-02, World Conference on Lung Health of the International Union Against TB and Lung Disease, India, Oct/Nov 2019 Akolo, C., Adetifa, I., Shepperd, S. and Volmink, J., 2010. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane database of systematic reviews, (1). Andrews, Jason R., Stephen D. Lawn, Corina Rusu, Robin Wood, Farzad Noubary, Melissa A. Bender, C. Robert Horsburgh, Elena Losina, Kenneth A. Freedberg, and Rochelle P. Walensky. 2012. “The Cost-Effectiveness of Routine Tuberculosis Screening with Xpert MTB/RIF Prior to Initiation of Antiretroviral Therapy: A Model-Based Analysis.” AIDS (London, England) 26 (8): 987–95. https://doi.org/10.1097/QAD.0b013e3283522d47. Badje, A., Moh, R., Gabillard, D., Guéhi, C., Kabran, M., Ntakpé, J.B., Le Carrou, J., Kouame, G.M., Ouattara, E., Messou, E. and Anzian, A., 2017. Effect of isoniazid preventive therapy on risk of death in west African, HIV-infected adults with high CD4 cell counts: long-term follow-up of the Temprano ANRS 12136 trial. The Lancet Global Health, 5(11), pp.e1080-e1089. Behr, M. A., M. A. Wilson, W. P. Gill, H. Salamon, G. K. Schoolnik, S. Rane, and P. M. Small. 1999. “Comparative Genomics of BCG Vaccines by Whole-Genome DNA Microarray.” Science (New York, N.Y.) 284 (5419): 1520–23. Diacon, Andreas H., Alexander Pym, Martin P. Grobusch, Jorge M. de Los Rios, Eduardo Gotuzzo, Irina Vasilyeva, Vaira Leimane, Koen Andries, Nyasha Bakare, and Tine De Marez. 2014. “Multidrug-Resistant Tuberculosis and Culture Conversion with Bedaquiline.” New England Journal of Medicine 371 (8): 723–32. Graves, S.K., Augusto, O., Viegas, S.O., Lederer, P., David, C., Lee, K., Hassane, A., Cossa, A., Amade, S., Peleve, S. and Zindoga, P., 2019. Tuberculosis infection risk, preventive therapy care cascade and incidence of tuberculosis disease in healthcare workers at Maputo Central Hospital. BMC infectious diseases, 19(1), p.346. Hoog, Anna H. van’t, Frank Cobelens, Anna Vassall, Sanne van Kampen, Susan E. Dorman, David Alland, and Jerrold Ellner. 2013. “Optimal Triage Test Characteristics to Improve the Cost-Effectiveness of the Xpert MTB/RIF Assay for TB Diagnosis: A Decision Analysis.” PLOS ONE 8 (12): e82786. https://doi.org/10.1371/journal.pone.0082786. Hoorn, Rosa van, Ernesto Jaramillo, David Collins, Agnes Gebhard, and Susan van den Hof. 2016. “The Effects of Psycho-Emotional and Socio-Economic Support for Tuberculosis Patients on Treatment Adherence and Treatment Outcomes – A Systematic Review and Meta-Analysis.” PLOS ONE 11 (4): e0154095. https://doi.org/10.1371/journal.pone.0154095. Houben, R.M., Sumner, T., Grant, A.D. and White, R.G., 2014. Ability of preventive therapy to cure latent Mycobacterium tuberculosis infection in HIV-infected individuals in high-burden settings. Proceedings of the National Academy of Sciences, 111(14), pp.5325-5330. Houben, Rein M. G. J., and Peter J. Dodd. 2016. “The Global Burden of Latent Tuberculosis Infection: A Re-Estimation Using Mathematical Modelling.” PLOS Medicine 13 (10): e1002152. https://doi. org/10.1371/journal.pmed.1002152. Insitute for Health Metrics and Evaluation. n.d. “Global Burden of Disease Study 2016.” Accessed November 8, 2018. http://ghdx.healthdata.org/gbd-2016. 60 | ANNEXES CONTENTS TABLE OF Jaintilal, A Teixeira Chongo, S Guilengue, KM Morais, S Lobo, HE Coelho Hamene, L De Morais, M Urrego, K Azam (2019). The impact of sensitisation and use of TB diagnostic algorithms to optimise Xpert MTB/RIF utilisation in Maputo City, Mozambique. PS -14-649-31, World Conference on Lung Health of the International Union Against TB and Lung Disease, India, Oct/ EXECUTIVE SUMMARY Nov 2019 Johnson, J.L., Okwera, A., Hom, D.L., Mayanja, H., Kityo, C.M., Nsubuga, P., Nakibali, J.G., Loughlin, A.M., Yun, H., Mugyenyi, P.N. and Vernon, A., 2001. Duration of efficacy of treatment of latent tuberculosis infection in HIV-infected adults. Aids, 15(16), pp.2137-2147. Karumbi, Jamlick, and Paul Garner. 2015. “Directly Observed Therapy for Treating Tuberculosis.” The INTRODUCTION Cochrane Database of Systematic Reviews, no. 5: 1. Kerr, Cliff C., Robyn M. Stuart, Richard T. Gray, Andrew J. Shattock, Nicole Fraser-Hurt, Clemens 1 Benedikt, Markus Haacker, et al. 2015. “Optima: A Model for HIV Epidemic Analysis, Program Prioritization, and Resource Optimization.” Journal of Acquired Immune Deficiency Syndromes (1999) 69 (3): 365–76. https://doi.org/10.1097/QAI.0000000000000605. POLICY QUESTIONS & METHODOLOGY Kibret, Kelemu Tilahun, Yonatan Moges, Peter Memiah, and Sibhatu Biadgilign. 2017. “Treatment Outcomes for Multidrug-Resistant Tuberculosis under DOTS-Plus: A Systematic Review and 2 Meta-Analysis of Published Studies.” Infectious Diseases of Poverty 6 (January). https://doi. org/10.1186/s40249-016-0214-x. Lutge, Elizabeth E., Charles Shey Wiysonge, Stephen E. Knight, David Sinclair, and Jimmy Volmink. 2015. “Incentives and Enablers to Improve Adherence in Tuberculosis.” The Cochrane Database of Systematic Reviews, no. 9: 1. RESULTS Mangtani, Punam, Ibrahim Abubakar, Cono Ariti, Rebecca Beynon, Laura Pimpin, Paul E. M. Fine, 3 Laura C. Rodrigues, et al. 2014. “Protection by BCG Vaccine against Tuberculosis: A Systematic Review of Randomized Controlled Trials.” Clinical Infectious Diseases: An Official Publication of the Infectious Diseases Society of America 58 (4): 470–80. https://doi.org/10.1093/cid/cit790. Pires Machai (2019). Systematic contact investigation to find missing cases and increase treatment CONCLUSION adherence and success for MDR-TB patients in rural Morrumbala District, Zambezia Province, Mozambique. PS -23-753-01, World Conference on Lung Health of the International Union 4 Against TB and Lung Disease, India, Oct/Nov 2019 Polana, C Saeze, P Mambo, A Balate, T Ferreira, A Jaramillo, R Frescas, J Cardoso, H Muquingue (2019). Yield in active TB case finding in facility-based and community-based approaches in KEY FINDINGS Mozambique. OA-12-395-01, World Conference on Lung Health of the International Union Against TB and Lung Disease, India, Oct/Nov 2019 5 Pym, Alexander S., Andreas H. Diacon, Shen-Jie Tang, Francesca Conradie, Manfred Danilovits, Charoen Chuchottaworn, Irina Vasilyeva, et al. 2016. “Bedaquiline in the Treatment of Multidrug- and Extensively Drug-Resistant Tuberculosis.” The European Respiratory Journal 47 (2): 564–74. RECOMMENDATIONS https://doi.org/10.1183/13993003.00724-2015. Rodrigues, M Lisboa (2019). Low rate of completion of isoniazid preventive therapy and associated risk factors in Beira, Mozambique: retrospective cohort study. SOA-02-1013-31, World Conference on 6 Lung Health of the International Union Against TB and Lung Disease, India, Oct/Nov 2019 Samandari, T., Agizew, T., Nyirenda, S., Tedla, Z., Sibanda, T. and Mosimaneotsile, B., 2012, March. TB incidence increase after cessation of 36 months isoniazid prophylaxis in HIV+ adults in Botswana. In 19th Conference on Retroviruses and Opportunistic Infections (pp. 5-8). ANNEXES ANNEXES | 61 Shapiro, Adrienne E., Ripa Chakravorty, Tokunbo Akande, Knut Lonnroth, and Jonathan E. Golub. 2013. “A Systematic Review of the Number Needed to Screen to Detect a Case of Active Tuberculosis in Different Risk Groups.” World Health Organization Google Scholar. Stop TB Partnership. 2015. “The Paradigm Shift 2016-2020. Global Plan to End TB.” http://www. stoptb.org/assets/documents/global/plan/GlobalPlanToEndTB_TheParadigmShift_2016-2020_ StopTBPartnership.pdf. Teixeira Chongo, D Jaintilal, K Azam, L De Morais, HE Coelho Hamene, I Pinto (2019). Unsuccessful Xpert MTB/RIF test results in Mozambique: causes and cost implications. PS -27-795-01, World Conference on Lung Health of the International Union Against TB and Lung Disease, India, Oct/ Nov 2019 Tiemersma, Edine W., Marieke J. van der Werf, Martien W. Borgdorff, Brian G. Williams, and Nico J. D. Nagelkerke. 2011. “Natural History of Tuberculosis: Duration and Fatality of Untreated Pulmonary Tuberculosis in HIV Negative Patients: A Systematic Review.” PLOS ONE 6 (4): e17601. https://doi.org/10.1371/journal.pone.0017601. UNDESA. 2017. “World Population Prospects: The 2017 Revision.” 2017. https://population.un.org/ wpp/. WHO. 2013. “Global Tuberculosis Report 2013.” World Health Organization. http://apps.who.int/iris/ handle/10665/91355. World Health Organization. 2015. “The WHO End TB Strategy.” Geneva: World Health Organization. World Health Organization. 2016. “Review of Available Evidence on the Use of Bedaquiline in the Treatment of Multidrug-Resistant Tuberculosis.” http://www.who.int/tb/publications/2017/ GDGreport_Bedaquiline/en/. World Health Organization. 2018a. “Latent TB Infection : Updated and Consolidated Guidelines for Programmatic Management.” http://www.who.int/tb/publications/2018/latent-tuberculosis-in- fection/en/. World Health Organization. 2018b. Tuberculosis Surveillance and Monitoring in Europe 2018 2016 Data. World Health Organization. 2019. “Global TB Burden.” 2019. http://www.who.int/tb/country/data/ download/en/. 62 | ANNEXES 1 2 3 4 5 6 TABLE OF EXECUTIVE INTRODUCTION POLICY QUESTIONS RESULTS CONCLUSION KEY FINDINGS RECOMMENDATIONS ANNEXES CONTENTS SUMMARY & METHODOLOGY ANNEXES | 63