An Evaluation of the 2035 WHO End TB Targets in 40 High Burden Countries

Tuberculosis (TB) is the leading global cause of death from a single infectious agent, with more than 10 million new cases annually. As a part of its efforts to control TB, the World Health Organisation (WHO) adopted the End TB Strategy in 2014 to reduce TB incidence by 90% between 2015 and 2035, with intermediate targets every five years. We examined TB incidence data from 2000 to 2018 for 40 high burden countries (HBCs) from the WHO published statistics, contextualized and assessed trends over time, and projected incidence to 2035 for each country. Two recurrent patterns accounted for 26 of the 40 HBCs: linear decrease (n = 14) or a peak in the 2000s followed by decline (n = 12). As uncontrolled HIV is the greatest risk factor for TB, we assessed trends in TB in HIV infected people and TB in HIV uninfected people separately in 15 Sub-Saharan African countries with high HIV prevalence. The projections of current trends were compared against the reductions required to meet the WHO End TB targets. Of the 25 countries without a high burden of TB, only 5 are on track to meet the End TB targets: Ethiopia, Laos, Myanmar, Russia, and South Korea. Of the 15 high HIV burden countries, 6 are on track to meet the End TB targets: Eswatini, Kenya, Lesotho, South Africa, Tanzania, and Zimbabwe. Overall, we predict 62 million excess cases of TB between 2020 and 2035 in the 29 HBCs projected to miss the WHO End TB targets. In high HIV burden countries, new programs aimed directly at TB will be required to maintain momentum. Moreover, our projections are based on data prior to the COVID-19 pandemic; the disruption of the pandemic is overwhelmingly likely to interrupt vital TB services and increase TB incidence. We anticipate that these findings will help orientate countries to their progress towards the End TB goals and inform the level of investment required to meet these important targets for a TB-free world.


Introduction 42
Mycobacterium tuberculosis, the causative agent of tuberculosis (TB), claims more lives than any other 43 infectious organism. In 2018 alone, there were 1.4 million deaths and 10.0 million new cases of TB 44 worldwide, with the 30 highest burden countries (HBCs) bearing 87% of the incidence. 1 Twenty-five of the 45 30 countries are classified as low or lower-middle income by the World Bank. 46 47 With HIV/AIDS, 2,3 smoking, 4 diabetes mellitus, 5 malnutrition, 5,6 and air pollution 7 among the risk factors, 48 TB is a global health concern of many faces. 1 Particularly, HIV-positive individuals are 20 to 30 times more 49 likely to develop active TB infection. 2,3 In Sub-Saharan Africa, where the prevalence of HIV/AIDS is the 50 highest globally, up to 70% of TB patients are estimated to be coinfected with HIV. 3 As coinfection is known 51 to increase mortality, TB is an especially pressing epidemic in this region. The burden of TB further extends 52 beyond mortality to severe productivity and treatment costs that exacerbate the difficult economic climates 53 in many HBCs. In 2018, TB cost the global economy $1 billion US dollars in missed workdays from 54 sickness, $11 billion from an average loss of 15-TB deaths, and $9.2 billion for 55 treatment. 8 56 57 To support its vision of a world free of TB, the World Health Organisation (WHO) developed the End TB 58 Strategy, which includes ambitious targets and milestones for 2020, 2025, 2030 and 2035, and a high-59 level strategy to achieve them. The targets include: 95% reduction in TB deaths between 2015 and 2035, 60 90% reduction in TB incidence in the same time span, and no families facing catastrophic costs as a result 61 of TB. 9 Accurate assessment of TB incidence is thus a prerequisite and critical part of the End TB Strategy. 62 The WHO estimations of incidence for each country is tailored according to the information available, which 63 depends on the strength of public health system and surveillance measures. While accurately 64 assessing TB incidence in high burden countries without strong public health systems is challenging, it is 65 reassuring that a study from the Global Burden of Disease consortium, using different methods than the 66 WHO, found a similar overall TB incidence in 2015 at 10.2 million cases compared to WHO 10.4 67 million. 10 there was a cumulative reduction in global incidence by 6.3%, far short of the 20% milestone set by the 72 End TB Strategy for 2020. 1 The WHO G T R 2019 estimates that only 11 of the 30 73 HBCs are on track to meet the 2020 milestone for incidence: Cambodia, Ethiopia, Kenya, Lesotho,74 Myanmar, Namibia, Russia, South Africa, Tanzania, Zambia, and Zimbabwe. 1 Other studies have 75 assessed this first milestone with grim results. A CDC report concluded that only the Europe region, which 76 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint includes no countries on the HBC list, is likely to meet the 2020 incidence target. 12 Another study by Pan 77 et al. predicts that only 2 regions out of 21 (Sub-Saharan Africa and Eastern Europe) and a mere 11 78 countries of the 195 analysed are on track to meet the 20% decrease in incidence; of these, 2 countries, 79 South Africa and Russia, are evaluated in this present study. 13 80 81 This study aims to characterise the current trends in TB incidence in HBCs and place them into the context 82 of the incidence End TB milestones and targets through 2035. Our projections extend beyond the WHO 83 analyses and previously published reports, which currently only project to 2020, and provide detailed 84 country-level breakdowns and analysis. We believe that this longer-term view, while subject to 85 considerable uncertainty, will enable stakeholders of HBCs to more effectively advocate for the resources 86 necessary to meet these targets. 87 88 89 90 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint confidence interval (CI). All trend lines and linear models fit to the entire time frame can be seen in 124 Supplementary Figure 6. 125 126 We fit a linear trend to the incidence data for each of the 40 countries in our study using the lm function. 127 We found that 5 HBCs had no change in incidence values at all, which we interpreted as the WHO 128 treating further reports from these countries as unreliable data. We examined 21 HBCs with an R 2 < 0.95 129 in more detail. Some countries displayed a linear trend in TB incidence for a more recent part of the time 130 series; a linear model was thus fitted to this sub-section. We identified pivotal TB control efforts since 131 2000 and their implementation years with literature searches and considered these when selecting the 132 years of TB incidence data to include in the linear model. Incidence estimates for 2020-2035 were inferred for each HBC using the predict function in R and the 136 model for that country. Additionally, for high HIV burden countries in Sub-Saharan Africa, the trends in 137 TB/HIV incidence and non-HIV TB incidence were modelled separately and combined to give an overall 138 estimate. The CI ranges for future projections were calculated by taking the average ratio of the high or 139 low CI to the estimated incidence over the 2014 to 2018 period of the WHO data. This average ratio was 140 then used to project the CI forward alongside the projected estimates through 2035 for each country. The 141 minimum value for projected and target incidence was set to 10 per 100,000 people, the threshold for the 142 Comparison of the projected incidence to the End TB target incidence 148 For each country, the End TB target incidence was modelled by multiplying the 2015 reported incidence 149 by 0.8 for 2020, 0.5 for 2025, 0.2 for 2030 and 0.1 for 2035. These milestones were used to precisely fit 150 a polynomial regression of the 5 th degree using the glm function in R, from which target values for each 151 year were interpolated. These targets were then compared against the predicted TB incidence values, 152 and the net difference between the two from 2020 through 2035 was calculated. A country was 153 concluded to miss the End TB target if the lower CI of its projected number of cases exceeded its target 154 number of cases in 2035. 155 156 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint The absolute numbers of TB cases for both the projected and End TB target scenarios were calculated 157 for the years 2020 to 2035 by multiplying the projected or target incidence per 100,000 people by the 158 World Bank projected population size divided by 100,000. End TB goals were conducted as explained with the overall TB incidence data, resulting in three separate 177 analyses: overall TB, TB in HIV-infected people (TB/HIV), and TB in HIV-uninfected people. Countries that 178 had been projected to be on track to meet targets for overall TB, but failed to meet targets in either of these 179 two sub-analyses by HIV status, were concluded to miss End TB targets. As TB in people without HIV 180 infections was not directly reported by WHO, and we calculated it by subtracting the incidence in people 181 with HIV from the total, there were no confidence intervals available for this estimate, or for the projection. 182 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. Trends in TB incidence in high burden countries 184 The reported TB incidence of 14 of the 40 analysed HBCs was well-described by a linear model (R 2 > 0.95) 185 with a negative slope (Table 1). This reflects a remarkably steady decline in TB incidence in these 186 countries: Botswana, Cambodia, Cameroon, Chad, China, Ethiopia, Ghana, India, Indonesia, Laos, 187 Thailand, Vietnam, Zambia, and Zimbabwe ( Figure 1). While the consistency of decline was similar among 188 these countries, the rates of decrease (in incidence per 100,000) varied between 0.57 * in Chad 189 and 37.25 * in Zimbabwe (full results in Table 1). This wide range is likely driven by the varying 190 epidemiology of TB in these countries (HIV associated or not), as well as the overall stage of TB control, 191 with faster absolute decreases being achieved in countries with higher incidence. 192 193 There were 12 HBCs characterised by a distinctive TB incidence trend of a peak during the 2000s ( Figure  194 2). All countries are Sub-Saharan African with a high prevalence of HIV: Angola, Cameroon, Congo,195 Eswatini, Kenya, Lesotho, Malawi, Namibia, Sierra Leone, South Africa, Tanzania and Zimbabwe. The HIV 196 prevalence in these countries ranges from 1.5% in Namibia to 23.6% in Kenya (average = 11.7%). As  suggests that data quality deteriorated at this time, leading the WHO to assume constant TB incidence 213 onwards. 16 While a definite cause cannot be determined for this unexpected trend of increasing incidence, 214 one possibility is that Angola, Liberia, and Mozambique have experienced civil wars until 2002, 2003, and 215 1992, respectively, with ongoing conflict pervasive in these regions. The poor data quality from 3 of these 216 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint 4 countries, recently too low to even be included, and the wide CIs associated with previous estimates 217 indicate that interpretations of incidence from these countries warrant caution. 218 219 Bangladesh, the Central African Republic, Nigeria, North Korea and Papua New Guinea reported the same 220 TB incidence from 2000 to 2018 (Supplementary Figure 2). One explanation for this, suggested by the 221 2009 WHO report on TB surveillance, is an assumption that incidence is constant in the absence of 222 sufficient data collected by a country. 16 Thus, the lack of change in estimated TB incidence indicates the 223 lack of effective and/or consistent reporting in these 5 HBCs from 2000 to 2018. For instance, Bangladesh 224 has had considerable changes in case finding efforts leading to fluctuations in notification rates, and thus 225 its TB incidence has been assumed to be flat until more data becomes available. 16 The lack of sufficient 226 surveillance to infer incidence in these countries is worrying, considering that they have a combined is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint Uganda has seen its rate of decline in TB incidence slow between 2000 and 2018, partly due to increasing 253 incidence of TB in HIV uninfected people (Supplementary Figure 3 & Figure 6). 254

255
The trend of TB incidence in South Korea was particularly noteworthy. Although South Korea is not on the 256 WHO high burden list globally, it has the highest incidence of TB among the 36 OECD (Organisation for 257 Economic Co-operation and Development) countries. 21  Observed Therapy Short-course (DOTS) treatments in 95% of its health centres and anti-TB medication 290 at no cost. 25 The steps South Korea has taken to control its TB incidence have been detailed above and 291 similarly entail interventions that increase both the quantity and quality of TB treatment. 292

293
The remaining 20 low HIV burden countries are projected to miss the targets (Table 2, Figure 4) with a 294 combined excess of 58.5 million cases of TB. More than half of these excess cases are accounted for by 295 only 3 countries: India, Indonesia and the Philippines. Progress toward reducing TB incidence in these 296 countries has been hampered by decentralised health systems, public-private mixed models of health 297 service leading to patients lost along the diagnosis-treatment pathway, and poor surveillance reflected in 298 large confidence intervals or unchanging incidence values in the WHO statistics. Four countries (Brazil,299 Cambodia, China, and Thailand) will come close to meeting the targets, but the End TB goal in 2035 is 300 not within the lower bound of their projected incidence (Table 2). Despite narrowly missing the targets, 301 Brazil is projected to achieve effective elimination by 2035 (incidence of 10 cases per 100,000 people), 302 and China will have only 13 cases per 100,000 people, which would still represent a major global health CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint There are 9 high HIV burden countries currently projected to miss the End TB targets (Table 3, Figure 6). 323 In an initial analysis of the overall trend (i.e. without separating HIV infected and HIV uninfected), 3 of 324 these countries (Botswana, Namibia and Zambia) had been projected to be on track to meet the End TB 325 targets. This speaks to the slower control of TB in HIV uninfected individuals in these countries 326 compared to great progress in the TB/HIV syndemic. Of particular note is Botswana, where 10,290 fewer 327 cases of TB are projected compared to End TB targets overall. This counterintuitive result is likely 328 because Botswana is projected to dramatically exceed the End TB targets until TB in HIV infected people 329 is effectively eliminated in 2021, after which progress toward targets will slow dramatically. Sierra Leone 330 is the only country to display a trend of increasing incidence of TB in HIV infected people, while the 331 Congo and Uganda both show decreases in overall TB and TB/HIV but an increase in the incidence of 332 TB in HIV uninfected people. Greater national and international focus on the TB programmes in these 333 countries is required to ensure that these alarming trends are reversed, especially as they have thus far 334 been hidden within the values of overall incidence. One weakness of this study is the underlying assumption of a linear trend to TB incidence. On balance, 345 we believe this is a reasonable assumption given that 14 countries had an R 2 > 0.95 when their 346 incidence data from 2000 to 2018 were fit to a linear model. However, it still remains a simplification, 347 especially for countries for which only a subset of the time series was used to fit the linear model. In 348 addition, while we took projected changes in population sizes into account when calculating the absolute 349 number of projected cases, changes in risk factors such as smoking, diabetes mellitus, malnutrition, 350 poverty, poor air quality, and age profiles of the population were not used to directly model the projected 351 TB incidence. Nevertheless, HIV, the most significant risk factor and comorbidity, was evaluated. Another 352 limitation is that as MDR-TB has different epidemiological trends to non-MDR-TB, countries on the WHO 353 list for MDR-TB, 1 but not overall TB, were excluded. Similar analyses may be of interest for these 10 354 countries (Azerbaijan, Belarus, Kazakhstan, Kyrgyzstan, Peru, Moldova, Somalia, Tajikistan, Ukraine, 355 and Uzbekistan) to determine their progress towards the WHO End TB targets by 2035. As noted above, 356 there are numerous countries with low quality data indicating that the current incidence estimates do not 357 reflect the reality of TB incidence in the population. These countries have still been included in our 358 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint analyses, as to exclude them may falsely suggest that there is no problem with TB in those countries. infectious agents whose coinfection likely leads to worse prognosis than either one alone. 29-31,33 378 379 Indirectly, the ongoing pandemic redirects resources and attention away from TB, as well as from HIV 380 and malaria. Restrictions in international commerce that limit access to necessary medications, shifts in 381 manufacturing of TB diagnostic tests to COVID-19 tests, and reallocation of financial and human 382 resources towards COVID-19 are only a few of many concerns for the future of TB control. A June 2020 383 survey by the Global Fund, which provides 69% of all international financing for TB, found that 78% of TB 384 and 85% of HIV programmes face disruptions in care delivery due to COVID-19. 35 The similarity of initial 385 symptoms between COVID-19 and TB further exacerbates the situation, fuelling stigmatisation and 386 reluctance of health workers to deliver comprehensive treatment. 387

388
Modelling studies carried out since COVID-19 emerged predict a regression of global TB control, 389 including significant increases in TB mortality, ranging from 13% in 2020 to 20% through 2025. 36,37 One 390 study from the Stop TB Partnership and USAID (United States Agency for International Development) 391 predicts 6.3 million excess TB cases globally between 2020 and 2025 from COVID-19, assuming a 3-392 month lockdown and a 10-month recovery, as an underestimate. 38 With an end to the pandemic 393 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint uncertain, decades-long gains in TB control may be reversed as a repercussion of the pandemic unless 394 strong corrective actions are taken. 395

Conclusions 396
Our projections reveal that 29 of the 40 analysed countries will not meet the WHO End TB targets if 397 current trends continue, with an excess of 61,792,010 cases in these countries between 2020 and 2035 398 compared to if the targets were met. Seven of the 11 countries projected to meet targets are in Sub-399 Saharan Africa. Three high HIV burden countries that had appeared on track to meeting targets when 400 only overall incidence was projected were revealed to miss targets when TB incidence in HIV uninfected 401 people were modelled separately. This reveals that as the number of people with uncontrolled HIV 402 diminishes, the rate of decrease in TB incidence is likely to slow. The ongoing COVID-19 pandemic is 403 also expected to disrupt the projected trajectories of TB incidence. Therefore, our prediction that 11 of 404 the 40 countries analysed will meet the End TB targets may still be a significant overestimate, and more 405 directed programmes for TB control should be concentrated in these regions with consideration for 406 country-specific contexts and successful measures in peer countries. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. .  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 2, 2020. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020. .  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020.   . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 2, 2020. . https://doi.org/10.1101/2020.10.02.20175307 doi: medRxiv preprint