Health economic analyses of latent tuberculosis infection screening and preventive treatment among people living with HIV in lower tuberculosis incidence settings: a systematic review

Introduction: In lower tuberculosis (TB) incidence countries (<100 cases/100,000/year), screening and preventive treatment (PT) for latent TB infection (LTBI) among people living with HIV (PLWH) is often recommended, yet guidelines advising which groups to prioritise for screening can be contradictory and implementation patchy. Evidence of LTBI screening cost-effectiveness may improve uptake and health outcomes at reasonable cost. Methods: Our systematic review assessed cost-effectiveness estimates of LTBI screening/PT strategies among PLWH in lower TB incidence countries to identify model-driving inputs and methodological differences. Databases were searched 1980-2020. Studies including health economic evaluation of LTBI screening of PLWH in lower TB incidence countries (<100 cases/100,000/year) were included. Results: Of 2,644 articles screened, nine studies were included. Cost-effectiveness estimates of LTBI screening/PT for PLWH varied widely, with universal screening/PT found highly cost-effective by some studies, while only targeting to high-risk groups (such as those from mid/high TB incidence countries) deemed cost-effective by others. Cost-effectiveness of strategies screening all PLWH from studies published in the past five years varied from US$2828 to US$144,929/quality-adjusted life-year gained (2018 prices). Study quality varied, with inconsistent reporting of methods and results limiting comparability of studies. Cost-effectiveness varied markedly by screening guideline, with British HIV Association guidelines more cost-effective than NICE guidelines in the UK. Discussion: Cost-effectiveness studies of LTBI screening/PT for PLWH in lower TB incidence settings are scarce, with large variations in methods and assumptions used, target populations and screening/PT strategies evaluated. The limited evidence suggests LTBI screening/PT may be cost-effective for some PLWH groups but further research is required, particularly on strategies targeting screening/PT to PLWH at higher risk. Standardisation of model descriptions and results reporting could facilitate reliable comparisons between studies, particularly to identify those factors driving the wide disparity between cost-effectiveness estimates. Registration: PROSPERO CRD42020166338 (18/03/2020).


Introduction
Nearly a quarter of the world's population has latent TB infection (LTBI), meaning they are infected but do not (yet) have symptoms of tuberculosis (TB) and cannot transmit infection. Without antibiotics, approximately 5% of immunocompetent individuals acquiring LTBI progress to TB disease within the first two years following infection, and another 5% over the remainder of their lifetimes 1,2 . This risk is higher for people living with HIV (PLWH) and may remain elevated even with antiretroviral therapy (ART). While a 2010 systematic review estimated that approximately 30% of co-infected people may eventually develop TB disease, and these subjects were at increased risk of premature death 3 , a UK study found incidence of TB disease during long-term ART to be much closer to background rates 4 . It is therefore important to evaluate the costs and benefits of testing and treatment of LTBI for PLWH, yet little research has been published on the cost-effectiveness of LTBI screening and preventive treatment (PT, also referred to as chemoprophylaxis) for this group.
Earlier detection and PT of LTBI when patients are diagnosed with HIV or when they are receiving HIV care prevents progression to active disease, thereby reducing cost of TB care, TB-related morbidity and may also reduce onward TB infection transmission and costs of contact tracing. As well as these benefits to the patient and to the health system, treatment of patients with LTBI is also an important intervention for TB elimination, particularly for low-incidence countries where the long-lasting benefit of PT will not be mitigated by repeated TB re-exposure within the general population [5][6][7][8][9] . However, there is currently no consensus concerning which individuals to target for LTBI screening/PT: guidelines vary by low TB incidence country.
Many European countries test all HIV clinic attendees, either with the tuberculin skin test (TST) or interferon-gamma release assays (IGRA) 10,11 , while other countries favour a targeted approach. As TB incidence falls in low TB incidence settings, the contribution to active TB of those with reactivation of chronic latent infection increases, but the cost-effectiveness of LTBI screening/treatment falls. Targeting groups at higher risk of infection, for example migrants from endemic regions, may be more feasible and will maximise patient benefit while minimising government spending. For example, the British HIV Association (BHIVA) guidance advises testing with IGRA alone to all PLWH from high/medium TB incidence countries, and only screening those from low TB incidence countries (<40 TB cases/100,000 population) if additional risk factors for TB are present (listed in the guidance) 12 . By contrast, the UK National Institute for Health and Care Excellence (NICE) recommends that all PLWH should be targeted for screening 13 . Given this divergence in guidelines, compliance is reported to be low 14 . A uniform, evidence-based national guideline for the UK is required 15 .
We conducted a systematic review to evaluate whether health economic studies are comparable in their conclusions regarding the cost-effectiveness of LTBI screening/treatment for PLWH or targeting subpopulations of PLWH at higher risk of infection to improve this cost-effectiveness. We focussed on lower TB incidence countries only (<100 TB cases/100,000 population), as this incorporated both low incidence (<40/100,000) countries, which tend to be high-income, plus middle-income countries including Brazil and China, which share more in common with low TB incidence settings in terms of TB control than with high TB incidence settings. We aimed to assess which aspects of these economic evaluations, in terms of both model structure and model inputs, most influence their predictions and where knowledge gaps remain, in order to guide future research to provide the necessary evidence on which to base national guidelines.

Methods
This study was registered on the International Prospective Register of Systematic Reviews (PROSPERO) registration number CRD42020166338 (18/03/2020). It was conducted in accordance with PRISMA guidelines 16 (see Reporting guidelines 17 ).

Selection criteria
To be eligible for inclusion, studies had to: 1) Include an intervention involving screening for LTBI among PLWH aware of their HIV status, and subsequent LTBI diagnosis and treatment. PLWH may or may not be receiving antiretroviral therapy (ART).
3) Report results of a health economic evaluation employing a modelling component. This could include decision tree, Markov, individual-based models or any other type of health economic model structure.

Amendments from Version 1
In response to reviewers' comments and suggestions, we have made the following changes: 1. We have added HIV prevalence, TB incidence and estimates of multi-drug resistance by country to the "Setting" column of Table 1.

2.
We have clarified that the scope of our review includes lower TB incidence countries, defined as <100 cases/100,000 population per year. This has been added to the Abstract Introduction, Methods Selection Criteria section) and a justification of this choice is given in the Introduction.
3. In addition to the CHEERS checklist to assess reporting completeness, we have evaluated the quality of each study using the Gates Reference Case (Table 6) and discussed how well such generic tools can measure this. 4. Figure 1 has been amended slightly, changing reason for exclusion for one study. 5. Extended the discussion of how incorporating secondary TB transmission affects study estimates, in the Discussion section. 6. Table 3: where available, uncertainty intervals for estimates are now included.

REVISED
life-years (DALYs) lost, deaths averted) and a cost component.
Studies were excluded where: 1) The study population was not exclusively PLWH.
2) The intervention involved mass LTBI chemoprophylaxis of all PLWH rather than treatment only following a positive LTBI screening test.
3) The intervention involved screening of TB disease rather than latent TB infection.
Articles for inclusion had to be literature (peer-reviewed full papers or research letters in peer-reviewed journals). Abstracts, presentations, posters, non-research letters and editorials were excluded (these formats provide insufficient details on methods used). Reviews and grey literature were also excluded. No restrictions were placed on the modelled study population in terms of factors such as age, gender, ethnicity, health or treatment status. There was no study exclusion based on choice of comparison groups, but their suitability was assessed as part of the evaluation of study quality. There were no restrictions by date or language of publication.

Search strategy and data extraction
We searched for published studies reporting the costeffectiveness, cost-utility or cost-benefit of screening for LTBI among PLWH in lower TB incidence countries (defined as <100 cases per 100,000 population/year, WHO 2018 estimates 18 ). Ovid Embase, PubMed and Web of Science were searched for articles published between 1 st January 1980 and 30 th September 2020 (date of the most recent search) using terms for cost-effectiveness studies, tuberculosis, screening and HIV (see Extended data 17 for full search terms).
Two reviewers (RFB, CV) independently screened the papers at all levels: title, abstract and full-text. Discrepancies were discussed between the reviewers to reach a consensus, and where necessary, in consultation with co-authors. Bibliographies of articles passing the full-text screening were subsequently reviewed for any additional, relevant papers. A data extraction schedule was developed and used to retrieve information from included studies regarding aspects including: study characteristics (authors, publication year, conflicts of interest and funding statements), setting, characteristics of modelled population, interventions and comparators analysed, year/duration of study, data used for model inputs, model type (e.g., Markov, discrete event simulation), diagnosis methods (including sensitivity and specificity assumptions), latent and active TB positivity rates, LTBI reactivation rate, treatment uptake and completion rates, treatment effectiveness, health economic aspects including model time horizon, perspective adopted (e.g., health service, societal), health and cost discount rates applied, costs included (e.g., costs of screening, costs of treatment), health utilities, and the key results and conclusions of the study (e.g., total incremental costs, QALY/DALYs and incremental cost-effectiveness ratio (ICER) for each screening intervention). We extracted base case cost-effectiveness estimates plus other types of model outcome, and uncertainty bounds and sensitivity analysis methods. Data were extracted independently by two reviewers (RFB, CAD). For the purposes of this analysis, we did not contact authors for clarification because we aimed to evaluate the information that would be available to the reader, particularly policy and decision makers. All data were managed using a Microsoft Excel spreadsheet, and validated by an independent reviewer.

Data analysis
Included studies were summarised according to study design, comparators and overall results. Studies were compared and assessed on the basis of study quality, perspective, design and parameter selection and valuation. Study reporting completeness was assessed using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist 19 and study quality was assessed using the Gates Reference Case for Economic Evaluation 20 (RFB, CV). To aid comparability, costs were inflation-adjusted to 2018 in the local currency and then converted to US$ using consumer price indices and average annual exchange rates, using MS Excel version 2012 21-23 . Forest plots were constructed using R version 4.0.3 to present study ICER (cost/QALY gained or cost/DALY averted) estimates. Cost-effectiveness studies may be more impactful by generating lower ICER values, so structural model assumptions which may particularly affect outputs, and therefore introduce bias, were evaluated. There were too few studies and lack of comparability between studies to employ further analysis by subgroup.

Study characteristics
The main characteristics of included studies and the resulting ICERs are presented in Table 1 and epidemiological factors are summarised in Table 2 26,27,29,32 ; and three evaluated testing schedules that involved both tests 25,29,32 . The remainder evaluated TST only 24,28 . PT regimens modelled were six-month [24][25][26]32 , nine-month 27,30 , and 12-month 28  Majority of input data from Japan. "Low incidence countries" defined as <24 cases per 100,000 people per year "as reported by the reports of the World Health Organization 41 " although in this report low incidence is defined as countries with an incidence rate of <20 cases per 100,000 people per year or <10 cases in total. MDR rate for modelled scenarios reported as 0.012 (range 0-0.1) which represents the proportion of HIV positive pregnant women who had MDR-TB.
f Frequency/schedule of "occasional screenings" scenario not defined.
g "High-risk" is not defined. Kowada reports that the US Centers for Disease Control and Prevention (CDC) states that high-risk women are "those with known or suspected TB contacts, injection drug use, HIV or other immunosuppression, foreign birth, and/or residence in congregate settings in low TB burden countries 42 "which implies that all pregnant PLWH are high-risk.
h System dynamics models are similar to Markov models in being cohort-based but they allow interaction between different model entities e.g., infectious disease transmission models, where interactions between infected and uninfected individuals is important.
i Other risk groups evaluated: non-US-born, diabetics, homeless, and incarcerated 31 . screenings"), keeping each analysis independent and comparing only costs and benefits for each test type used 26 . The author used the most cost-effective testing strategy as the base case for each scenario, so all other ICER values presented were dominated.
One study specified that LTBI screening was undertaken at HIV diagnosis and annually thereafter 30 ; other studies modelled screening of populations in established HIV care 24,25,31 or this was not recorded but is likely also to have been established care [26][27][28][29] . Capocci et al. 2020 stated that the population on which their model was based was offered LTBI screening at their next routine appointment for those in established care, as well as all newly HIV-diagnosed patients 32 .
Screening and treatment parameters. Two-thirds of studies did not report or incompletely reported test sensitivity and specificity values used 24,25,28,[30][31][32] (Table 2). For those studies reporting, TST sensitivity was 43-89% and specificity was 59-92%. IGRA sensitivity was 61%-83% while specificity was consistent at 98-99%. TST specificity is known to vary by BCG inoculation status, but only one study accounted for this (97% specificity for non-BCG-vaccinated individuals, 59% for vaccinated individuals 26 ). A further study stratified specificity by country of origin to reflect this difference implicitly (98% for US-born, 92% for non-US-born 27 ). The remaining study assumed 87% specificity 29 .
Assumed effectiveness of full-course PT with isoniazid (INH) for six months was 62-68% 25,26 (effectiveness assumptions were unclear in the study by Azadi et al. 24 ), while nine-month effectiveness was assumed to be 90% for one study 27 , while a second study assumed differential effectiveness by CD4 count and region of origin (locals versus non-locals) 30 (Table 2). Therefore there is no clear relationship evident between incorporation of secondary TB transmission and cost-effectiveness estimation.
Three studies accounted for multi-drug resistance (MDR) 26,28,30 , all of which assumed around 10-fold higher treatment costs for active TB and two of which assumed higher morbidity and/or mortality 26,30 . In addition, Capocci et al. 2015 stated that they implicitly incorporated the impact of treatment resistance into their treatment effectiveness estimate 25 . Three studies did not explicitly incorporate ART (Table 2). ART use would be expected to reduce cost-effectiveness estimates; it reduces health benefits of the intervention because TB progression rates and active TB-related mortality is vastly reduced for PLWH on ART 45 . The one study parameterised based on the pre-ART era found LTBI screening/PT to be cost-saving 28 . In addition, HIV treatment and care costs continue for life; therefore, for PLWH whose lives are saved by preventing TB-related mortality, these costs continue to accrue over their lifetime. However, of the four studies explicitly incorporating the health impact of ART 24,25,29,30 , only one included HIV care/ART costs in their analysis 29 .  32 but full details of estimation were not reported.

Utility
Main findings. The diversity of model assumptions and parameter values only partly explain the diverse results from these studies. Figure 2 summarises the ICER estimates each included study reported for various LTBI screening/PT strategies, alongside willingness to pay (WTP) estimates quoted or discussed by each study. In general, studies found that at least one screening/PT c Adverse events included were drug-induced liver injury (DILI) only for Sawert et al., 28 Linas et al., 27 Kowada et al. 26 and Tasillo et al. 29  strategy evaluated was cost-effective according to their setting-specific threshold ( Figure 2 and  (Table 3). Overall, the heterogeneity in model assumptions and parameter values we have described make further comparisons between study estimates difficult.

Sensitivity analysis
All studies provided a univariate (one-way) sensitivity analysis using a selection of model parameters, and all but one 27 undertook probabilistic sensitivity analysis (PSA, where all or selected parameters are varied simultaneously within their parametric distribution to produce a range of plausible values for the ICER) (Table 4). However, choice and number of parameters included in analyses varied and were selected subjectively. Systematic presentation of the most influential parameters on model outcomes were attempted by four studies (as a table 25,32 or as a Tornado plot 24,31 , albeit with only three parameters for Azadi et al. 24 ). PSA was generally used to create cost-effectiveness acceptability curves (CEACs) only in earlier studies, showing the strategies by WTP threshold, but more recent studies employed a more systematic, comprehensive approach to SA including presentation of ICER estimates with uncertainty intervals 31,32 .

Quality assessment
Study reporting completeness varied considerably between studies (range 46-88% on CHEERS 25-point checklist,  26,30 as identified by the CHEERS checklist (Table 6). Generally, studies performed poorly on Gates principles which may only recently have been recognised as important for inclusion in cost-effectiveness analyses, such as discussion of equity considerations and budget impact analysis (which is often performed separately to a cost-effectiveness analysis). Heterogeneity, in terms of exploring differential impacts of interventions within subpopulations, was handled differently by studies depending on the research question. This is because some studies treated PLWH as the primary patient population and evaluated respective subgroups (e.g., CD4 count strata 28 , migrant status 26 , country of birth 25,32 ), while others included PLWH as one of several groups at risk for LTBI (e.g., close contacts of TB patients, migrants, vulnerable populations including homeless, drug users and former prisoners, and individuals with medical comorbidities 27,31 ). Given the small number of studies included, we could not conduct any formal subanalysis by study quality, but there was no trend in terms of cost-effectiveness by study quality.

Discussion
To our knowledge, this is the first systematic review of cost-effectiveness of LTBI screening/PT focussing on PLWH in lower TB incidence settings, and it highlights the limited number of studies published. Cost-effectiveness estimates of LTBI screening/PT for PLWH varied widely: taking studies published in the past five years, which should be relatively similar in terms of assumptions such as ART use, cost-effectiveness of strategies screening all PLWH varied from $2828 to $144,929 (n=5, 2018 prices). Included studies have such variation in strategies evaluated, target populations and methods and assumptions used, that it is hard for policy makers to interpret these results, identifying which model inputs are driving these extreme values and how they relate to their own populations, in order to make informed decisions regarding screening strategies. Strategies targeting screening/PT to PLWH at higher risk of LTBI were found to vary markedly in their cost-effectiveness (NICE 2016 strategy: $131,643/QALY gained, BHIVA 2011: $58,297/QALY gained in the UK 32 ), with alternative strategies found to be more cost-effective 32 . These findings should be evaluated in conjunction with estimates of number of LTBI cases missed by each strategy in order to devise revised, coherent national guidelines.
Study quality and reporting completeness were assessed using the Gates Reference Case for Economic Evaluation and the CHEERS checklist, respectively. However the insights gained from these were limited because of the heterogeneity between studies. Furthermore, generic measures of study quality may fail to capture which model assumptions are key and are most likely to bias the outcomes, as they are not specifically designed to evaluate or compare epidemiological models. Development of more precise evaluation tools for these types of analyses, where a range of different models may be used to evaluate costeffectiveness of an infectious disease intervention, will help with model comparison. Such evaluation methods have already been developed for specific model types (infectious disease transmission models) 47,48 .  Difference between "TB risk" and "TB incidence" not stated.
c Systematic one-way sensitivity analysis results presented as a

Title and abstract
1 Title: Identify the study as an economic evaluation or use more specific terms such as "cost-effectiveness analysis", and describe the interventions compared.

13
Estimating resources and costs: Describe approaches and data sources used to estimate resource use associated with model health states. Describe primary or secondary research methods for valuing each resource item in terms of its unit cost. Describe any adjustments made to approximate to opportunity costs.

17
Analytical methods: Describe all analytical methods supporting the evaluation. This could include methods for dealing with skewed, missing, or censored data; extrapolation methods; methods for pooling data; approaches to validate or make adjustments (such as half cycle corrections) to a model; and methods for handling population heterogeneity and uncertainty.

18
Study parameters: Report the values, ranges, references, and, if used, probability distributions for all parameters. Report reasons or sources for distributions used to represent uncertainty where appropriate.
Providing a table to show the input values is strongly recommended.

19
Incremental costs and outcomes: For each intervention, report mean values for the main categories of estimated costs and outcomes of interest, as well as mean differences between the comparator groups.
If applicable, report incremental cost-effectiveness ratios.   Further research is required to provide the evidence base to inform LTBI screening policies. The many methodological facets listed in Table 1 and  Figure 2 illustrates the large differences in thresholds assumed by included studies), demonstrating the variation in what is deemed cost-effective, even when restricted to lower TB incidence settings.
While heterogeneity in model structure and assumptions can hamper comparability, it is still important to consider this diversity to explore the full range of uncertainty and identify which aspects, such as incorporating MDR, or onward TB infection transmission, are most influential and therefore important to include. However, a more standardised approach to presentation of methods and results, including systematic and well-justified sensitivity analyses, will facilitate comparisons between studies so that policy makers can fairly judge the evidence available on which to base LTBI screening guidelines in these settings. Items 15-17 of the CHEERS checklist, relating to model structure, assumptions and methods, only contribute three points to the reporting score but we recommend it should be given more weight as they are crucial to understanding how all model inputs relate to the outputs. Lessons can be learned from other fields to develop a descriptive framework to make future cost-effectiveness analyses more rigorous and comparable 51 .
Assessment of uncertainty is an important aspect of all cost-effectiveness analyses. We found sensitivity analyses conducted by included studies to be highly heterogeneous, and choice of parameters and the ranges through which they were varied were not always rigorously justified, though quality increased over time. To standardise the general reporting of cost-effectiveness analyses, the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) developed the CHEERS reporting checklist [28], which we used to evaluate study quality (Table 4). We recommend further standardisation of cost-effectiveness analyses to mandate inclusion of Tornado plots with justification of each parameter range used. These provide a more effective, objective summary of the most influential parameters driving model output (as long as parameter ranges are well justified) than lengthy descriptions of results in the text, and all parameters should be included rather than just a selection, subjectively chosen (it is also important to identify which parameters have little impact on study outcomes as these should then have less weight in decision making). While PSA was widely used by included studies to create costeffectiveness acceptability curves, we would endorse its use to generate uncertainty ranges for ICER estimates.
Improving clarity will further improve the accessibility of studies. We found a lack of precision in description of model parameters sometimes limited our understanding of how they related to model structure and in turn, model output. For example, authors should be clear whether "TB" refers to TB disease (often referred to as active TB) or latent TB infection, and should always specify units and clarify proportions versus percentages. They should state to which population group or subgroup the specific parameters apply, and for each subgroup created (e.g., patients developing DILI, those with MDR) it should be articulated: 1) what proportion of the cohort is in the subgroup, 2) over what duration they remain in this group and 3) how that affects their costs and health benefits. It should be clear, also, how inputs such as treatment adherence affect therapeutic effectiveness, and therefore influence model outputs.
A contentious issue regarding HIV-associated TB is the downstream costs of HIV care. ART is lifelong; therefore, interventions improving survival for PLWH may appear less cost-effective than for HIV-uninfected individuals. Therefore, it is perhaps unsurprising that only one included study accounted for HIV care costs 29 . Currently, PT for PLWH in low TB incidence (generally higher resource settings) has only a marginal gain in terms of life expectancy (PT nonetheless playing an important role in TB control by reducing morbidity, costs of TB disease treatment and onward TB infection transmission). Therefore, the inclusion or exclusion of ART costs should not be as influential as seen in other contexts [52][53][54] . Nonetheless, it raises important ethical questions regarding the design and interpretation of cost-effectiveness analyses involving increasing the life expectancy of PLWH 52 .
There are limitations to our analysis. Principally, we could not explore factors driving model output in more detail because of the limited number of studies included. While broadening our focus to include higher TB incidence countries would increase these numbers, the very different contexts (TB reinfection rates, mortality rates, ART coverage and costs, among others) means comparisons between studies would be equally challenging. We are also unable to rule out the possibility of publication bias, with potential selective publication of more favourable cost-effectiveness estimates. Only one of the included studies reported a conflict of interest of the authors (receiving personal fees from pharmaceutical manufacturers 30 ), and selection/ omission of model assumptions which would make outcomes more/less favourable (ART costs, secondary transmission) was not uniform across studies. However, Jo et al. selected the four states where more than half of US TB cases occur, so cost-effectiveness of screening is likely to be reduced in states with lower prevalence 31 . These states are also the richest in the US by Gross Domestic Product 55 .
Our study highlights the need for further research evaluating the cost-effectiveness of LTBI screening/PT, employing the highest standards of methods and reporting in order to make useful contributions to the field that can be used by policy makers to inform national guidelines. As TB prevalence hopefully continues to fall across the world, we need to consider targeting strategies which will be cost-effective now and in the future, to provide good value for the resources invested and better health for PLWH.

Data availability
Underlying data All data underlying the results are available as part of the article and no additional source data are required.  13.

Is the review written in accessible language? Yes
Are the conclusions drawn appropriate in the context of the current research literature? Yes No further comments to make.

Is the topic of the review discussed comprehensively in the context of the current literature? Yes
Are all factual statements correct and adequately supported by citations? Yes

Is the review written in accessible language? Yes
Are the conclusions drawn appropriate in the context of the current research literature? Yes people per year). They identified nine studies to include, from the USA, Italy, Japan, Brazil, UK and China.
They also aimed to assess whether studies were comparable in their conclusions for the cost effectiveness of all PLWH vs targeted screening, and which aspects of the studies most influenced the predictions.

Summary of our opinion:
Very clear and well written paper, and could make an important contribution to current policy decisions in the UK.

○
Overall, the evaluation is well structured, with a clear description of study inclusion and exclusion criteria.

○
We would very much like to see the manuscript published after the authors consider making the changes highlighted below ○ Data and methods: Methodologically sound. Reported efforts for reducing bias including multiple abstract reviewers.
Major: A definition of <100 cases per 100 000 people per year was used to define low TB incidence settings. The European framework for Low TB incidence countries defines low TB incidence countries as countries with case notification rates of <10 per 100 000 inhabitants and declining. High risk populations are defined as those with a notification rate of >100 cases per 100 000 population (Broekmans et al. 2002. European framework for tuberculosis control and elimination in countries with a low incidence, ERJ, 19:765-775). Brazil has a TB notification rate of 35 cases per 100 000 people per year and China has a notification rate of 43 cases per 100 000 people per year (WHO Global Tuberculosis Report 2021). Therefore, using the definition from the European TB control framework, neither of these countries would be considered low TB incidence countries. Recommend discussing why this definition was used in the paper, referencing other definitions.  (15):1253-1258). Y Given the amount of work that would be required of the authors to restructure the review to assess the quality of papers included, we recommend updating the ○ ○ Need to add discussion of how the quality of the different studies were assessed and if found to be of low/ high quality, and if/how was this incorporated into the analysis? ○ The discussion section was well laid out. Important topics were covered such as the inclusion of ART costs.
Would also recommend adding discussion of the following: Why did you restrict the analysis to only studies with costs and utility data, and how may excluding costing studies have skewed your results or recommendations?
○ Based on the current evidence, what is your recommendation to policy makers (specifically in the UK) as stated early in the paper, the purpose of the review is to inform their decisionmaking? Major: Very well written paper, clearly and logically outlines the research questions, analysis and discusses the findings. Table 3 lists the key cost components reported in the studies, but it is unclear what the unit of the cost estimate reported was. For example, was the cost of full-course LTBI chemoprophylaxis reported per patient or for all patients assessed in the primary studies? Need to add the unit to the first row of the cost table in Table 3. If the uncertainty intervals were reported in the studies, it would be good to add those to the table too.  Discussion. Suggest replace ", only contribute three points to the quality score but should be given more weight as they are crucial to understanding how all model inputs relate to the outputs", with ", only contribute three points to the quality score but we recommend it should be given more weight as they are crucial to understanding how all model inputs relate to the outputs".

Motivation/research question:
Major: Highly topical research question is currently being considered by policy makers in the UK. Could have been strengthened by also including studies that only reported cost estimates.
○ Suggest adding discussion of this limitation fully in the discussion section of the paper.

Major:
Overall well written and good framing of the research question.
Minor: Statements regarding the prevalence of latent TB infection should be qualified, and critically evaluated. The size of the truly infected population is a current 'hot topic' in TB Add references for the statement that "A quarter of the world's population has latent TB infection (LTBI), meaning they are infected but do not (yet) have symptoms of tuberculosis (TB) and cannot transmit infection." Reviewer Expertise: NF: economics, especially Tuberculosis; RW: epidemiology, especially Tuberculosis