A symptomatic approach to tuberculosis screening for high-risk groups in Malaysia: Cost-effectiveness and budget impact analysis

Highlights • Symptomatic Approach for TB screening among high-risk groups was more cost-effective than the existing approach.• Cost of conducting Sputum Acid Fast Bacilli (SAFB) was the key driver in determining which approach is more cost-effective.• Switching from existing approach to symptomatic approach would results in cost-savings of MYR 65.5 million, but resulted in 4473 TB cases undetected in 5 years period.


Introduction
Tuberculosis (TB) is caused by Mycobacterium tuberculosis, and one of the oldest infectious diseases known to infect the human [1]. It remains one of the major public health challenges especially in the lowand middle-income countries despite the advancement in treatment and management. Recent data shows that it has persistently killed millions of people annually since it was first recorded in the 1960s and has since ranked as one of the leading causes of morbidity and mortality by a single infectious agent [2]. In 2019, around 10.0 million people developed TB, of which 1.2 million TB deaths occurred among HIV-negative people and 208,000 deaths among people living with HIV (PL HIV) [3]. From the total global TB burden, more than half of the infections occur in the Western Pacific region.
Ending TB epidemic was set as one of the Sustainable Development Goals (SDGs) target to be achieved by 2030 [4]. In reaction to this, World Health Organization (WHO) through the WHO End TB Strategy has urged for reduction of TB deaths and TB incidence rate by 95 % and 90 % respectively, by 2035 [4,5]. Pillar 1 of WHO End TB Strategy, of which to provide integrated, patient-centred care and prevention, focus must be given to early detection, treatment and prevention to all. Hence, by ensuring equal access to healthcare and better engagement with the patient as well as community. There is no "one size fits all" approach in managing TB, therefore, each country has to adapt based on their diverse country setting.
While TB can infect every-one, certain groups of population such as PL HIV, healthcare workers, and those living in institutional setting have higher risk of contracting with TB infection. TB infection occur upon exposure to an infectious individual, spreading through an airborne. While TB can affect various body sites, TB is primarily affecting the lungs [6]. Hence, the mainstay of TB detection focuses on Sputum for Acid Fast Bacilli (SAFB) and Chest X-ray (CXR). There are continuous debates on various TB screening approaches due to the presence of asymptomatic cases [6]. Generally, TB screening can be done either through targeted screening of high-risk groups or mass population screening [7]. Nevertheless, studies have shown that the high-risk groups have higher TB incidence than the general population [8,9]. Hence, WHO has suggested for a more systematic approach to TB screening compared to the mass population screening. Even though symptom screening was shown to have sensitivity around 69 %, the addition of CXR would results in around 2.5-fold increase in TB case detection and an increase the sensitivity as much as 11 % [10].
Malaysia has achieved quite remarkably in terms of controlling TB infection. Malaysia TB notification rate has remained below 100 cases per 100,000 population for the past few years, however, it is still one of the leading causes of morbidity and mortality among infectious diseases [2]. Nevertheless, TB infections still posed as major threats in institutionalized setting especially in prisons with incidence rates about ten times of general population. Whereas, 6.3 % of TB cases undergone HIV screening turned out as HIV-positive [11]. To achieve higher case detection rate, MOH Malaysia has introduced The Malaysian National Tuberculosis Programmes (NTPs) by expanding the prior TB control strategy, giving focus on early case detection among those having TB symptoms and high-risk groups, at the same time providing quality laboratory services, develop training modules and guidelines, conducting routine training to staffs, plus inter agency collaborations [11].
As one of the main strategies for Ministry of Health (MOH) Malaysia to control TB infection, TB screening among the high-risk groups include those close contact to TB cases, immunocompromised patients, an elderly and others involves both the asymptomatic and symptomatic [12]. According to the TB screening guideline for the high-risk groups, MOH Malaysia, TB screening method is decided upon whether the person is symptomatic or asymptomatic. A person without any symptom is subjected for CXR, while the those with TB symptoms such as cough, loss of appetite and loss of weight will require both CXR and SAFB [12]. Despite that, studies showed that TB screening among the asymptomatic individual using CXR, often yields unreliable results [13]. For example, a study among the asymptomatic in Malaysia revealed a low yield, of which PL HIV was the highest with 25 %. The lowest yield was among close contact of TB case with 4.4 % [14]. Hence, creating debates on the value of screening the asymptomatic.
To achieve the WHO goal for TB in 2035, countries must provide full commitment in terms of funding for tuberculosis care, prevention and treatment. With the budget constraints and increasing health expenditure, prioritization of TB screening strategy is necessary. This includes the decision on whether to screen only those presented with symptoms or both the asymptomatic and symptomatic high-risk groups [15]. An unsystematic and poorly targeted TB screening approach will not only add unnecessary burden to the healthcare systems, but also results in high operational cost. [15,16]. In response, WHO has suggested for a more systematic approach to TB screening [17]. TB screening strategy should take into account the measure of cost-effectiveness while focusing on the high-risk groups [15]. At the same time, Budget Impact Analysis (BIA) should be incorporated into the Cost-Effectiveness Analysis (CEA) for better interpretation of the result, which will allow the policy makers to make better decision. Prior studies have revealed that less than 5 % of global health CEA include BIA in their study [18].
The government through MOH Malaysia has invested lots of resources to control TB infection, especially through screening of the highrisk groups, of which the approach includes screening both the asymptomatic and symptomatic. However, there are lack of documented studies that measure the economic efficacy and the long-term impact of this TB screening approach. With the increasing pressure to end TB epidemic, new programmes were introduced such as screening and treatment of latent TB as well as the introduction of the novel Xpert MTB/RIF diagnostic technology to improve TB case detection. These new and existing programmes give extra pressures to MOH in terms of budget allocations. Hence, this study focuses on measuring the costeffectiveness and the budget impact of symptomatic approach to TB screening compared to the existing approach (screening both the asymptomatic and symptomatic) among the high-risk groups in Malaysia. This study is conducted from the health care provider perspective (MOH, Malaysia).

Methods and materials
This paper is a subset to the study on CEA of High-Risk Groups Tuberculosis Screening in Malaysia, which aimed to measure the relative cost-effectiveness and estimates the five years budget impact of symptomatic TB screening compared to the existing approach [19]. To answer the study objective, two hypotheses scenario were developed, namely existing approach and symptomatic approach. Existing approach is defined as current TB screening approach, of which both symptomatic and asymptomatic high-risk groups were screened. Symptomatic approach was the alternative hypothesis approach, of which only the symptomatic high-risk groups were screened for TB. Data were gathered from multiple sources for the analysis, including the Disease Control Division, MOH, TB Information system (TBIS 204S) as well as Sabah and Sarawak State Health Departments. First, a decision tree model was developed for CEA to compare the two scenarios. The findings from CEA were subsequently introduced into BIA. The outcome of BIA was presented in term of number of TB cases detected and total costs for both scenarios over the span of 5 years, from 2018 to 2022. Costs data were valued in year 2018 using Malaysia Ringgit (MYR). The Willingness to Pay (WTP) threshold of three times GDP per capita was used [19][20][21]. Based on 2018 Malaysia GDP per capita of ~ MYR 40,000, WTP threshold was capped at MYR120,000, where MYR 1.00 ~ USD 0.25 [22,23].

Clinical data
The number of TB cases detected were identified by assessing the number of TB screening and the number of TB cases identified through TB screening procedure. A three years of TB screening data in the period of 2016 to 2018 from Sabah and Sarawak were examined to increase the reliability of the data. These data were extracted from TBIS 204S. Only 11 high-risk groups were included in the study selected based the TB screening guideline for the high-risk groups for MOH, Malaysia. Excluded were those without results, diagnosed by specialists, diagnosed by other modalities, and TB screening of close contact. From the total list provided, only 65,400 cases were included. The estimated effectiveness data are shown in Table 1.
By using similar data, the probabilities data were estimated. These were the probabilities of those undergone TB screening presented with symptoms (being symptomatic) and the probabilities of each high-risk group among the symptomatic or asymptomatic cases. The probabilities data are shown in Table 2.

Costs
Two TB screening procedures were CXR and SAFB. The TB screening related costs from MOH perspective were calculated based on the capital (such as machineries plus their yearly maintenance and calibrations), personnel (such as yearly salary and allowances for staffs involved) and consumables (such as include the X-ray film, chemicals, personal protective equipment and others) involved. A mix of step down and Activity Based Costing (ABC) were implemented for costs estimation. Expert opinion was used to allocate the costs. The allocation of capital and personnel costs were based on duration of the respective screening procedure. The cost of screening asymptomatic case equals to the costs of one CXR procedure, while the costs of conducting TB screening among the symptomatic was the summation of both costs for CXR and SAFB. (Table 3).

Cost-Effectiveness
The CEA was evaluated over one-year time horizon, with existing approach as the comparator. TreeAge Pro Healthcare version 2019 was used to develop a decision tree model for the CEA, of which populated with the cost, probabilities and effectiveness data (Tables 2-4). The decision tree model is depicted as in Fig. 1. The symptomatic approach was compared to the existing TB screening approach. The existing approach consisted of symptomatic screening and asymptomatic screening. Each would then branch into the eleven high-risk groups. The  incremental cost effectiveness ratio (ICER) for each scenario was defined as the ratio of cost per screening (in MYR) to the TB cases detected.
where A is existing approach; B is symptomatic TB screening The decision tree model for cost-effectiveness analysis was based on the assumption that each screening procedure was conducted following similar procedure. Hence, eliminating the variation in terms of resources used such as personnel, machineries and consumables. There was also no variation of cost and resources used by different facilities. The model also assumed that each screening procedure strictly follows the TB screening guideline by MOH [12].

Sensitivity analysis
The uncertainty of costs, number of TB cases and probabilities data were investigated using Deterministic sensitivity analysis (DSA) and Probabilistic Sensitivity Analysis (PSA). In DSA, changes of ICER value were observed with various range of data. This would allow the robustness of the model to be explored. A set of minimum and maximum values were identified and introduced into the DSA. The outcome of DSA was depicted as Tornado Diagram, of which key drivers for the ICER can be determined.
Whereas, PSA used Bayesian method to assess the model outcome [24,25]. To allow variation of data, a suitable statistical distribution was assigned (Tables 2-4). Monte Carlo of 1,000 simulations were executed. The results are shown using Cost-effectiveness plane scatter diagram and Cost-Effectiveness Acceptability Curve (CEAC) allowing the costeffectiveness level to be depicted at various level of WTP.

Budget impact analysis
A BIA was performed using Microsoft Excel. It is to estimate the expenditures and outcome of TB screening approach from provider (MOH) perspective over the span of 5 years, from 2018 to 2022. The BIA measure the direct costs of screening for the two strategies; existing TB screening approach, while the second was the symptomatic approach to TB screening. The estimated total number of TB screening for year 2018 was set as the population to receive TB screening. Data obtained from TBIS 204S revealed the total number of TB screening done in 2018 was CXR Chest X-ray, SAFB Sputum for Acid Fast Bacilli, ¶ The costs data are varied by ± 25 %. 322,841 screening, for all the eleven high risk groups. The cost input for BIA was the cost per screening for each approach, estimated from the CEA. Meanwhile the outcome measured was the number of TB case detected. The incremental (or decremental) effect on cost (cost saving) and event (undetected or missed TB cases) were also included in the BIA. In order to measure the impact of actual monetary value (MYR) at each point in time, the cost used was not discounted [26].

Cost-Effectiveness
The result of CEA is shown in Table 4. The cost-effective plane in Fig. 2 (Fig. 2). Symptomatic approach to TB screening was a strategy with much better outcome even though with a higher cost per screening compared to the existing approach. Fig. 3 shows the results of DSA for the existing approach versus the symptomatic approach. Despite changes made to the cost, probability and effectiveness data, the ICER value did not falls below zero. This indicates that the symptomatic approach would remain relatively dominant compared to existing approach. The costs of conducting SAFB was found as the key driver in determining the ICER value. Higher the cost of conducting SAFB would results in higher ICER value, while lower SAFB cost would results lower ICER. Fig. 4 depicts the PSA results of symptomatic approach versus existing approach. The cost-effectiveness plane shows the result of 1,000 Monte Carlo simulations. It is presented as scatter plot in a graph form with y-axis represents of incremental cost (per TB screening), while the x-axis represents incremental effectiveness (number of TB cases detected). Generally, the symptomatic approach was showed to be more expensive and more effective compared to the existing approach, with the plot mostly occupied the northeast area of the graph. The CEAC demonstrates that 100 % of the iterations resulted in symptomatic approach being more cost-effective of the two strategies almost throughout of all WTP values, up to MYR 240,000. Table 5 shows five years BIA for different TB screening strategies from the year 2018 to 2022. Existing TB screening approach will result in estimated cost of MYR 13.6 million to 17.2 million yearly, while symptomatic approach to TB screening would costs around of MYR 2.0 to 2.5 million yearly (Fig. 5). In total, existing TB screening approach will result in 1,714,007 screening throughout the 5 years, which will cost an estimated total of MYR 76.6 million. This strategy results in an estimated total of 12,169 TB case detected over five years period. Whereas, symptomatic approach to TB screening will result in total of 185,456 screening over five years, at the same time costs around MYR 11.2 million over the same period. However, this strategy only results in a total detection of total 7,696 TB case. Thus, switching from the existing strategy to the screening approach focusing on the symptomatic will result in costs saving as much as MYR 65.5 million. However, this will also result in 4,473 TB cases not being detected (TB cases missed).

Discussion
The results showed that the symptomatic approach was more costeffective compared to the existing approach in terms of high-risk groups TB screening. This is in concordance with the past studies showing that TB case detection can be improved through symptomatic screening such as eliciting the history of cough among the patients [7]. On average, the existing approach would cost around MYR 5,932.78 to detect one TB case. For comparison, the cost was significantly lower through the symptomatic approach with around MYR 1,363.62 per TB case detected. Hence, it would be much more expensive screening both the asymptomatic and asymptomatic to get a similar outcome. This study also found that ICER value was greatly affected by the cost of conducting SAFB. As the cost of conducting sputum AFB becomes more expensive, the ICER value would become much higher, and vice versa. Therefore, to keep the ICER lower, the cost of conducting SAFB must be kept as low as possible.
Prior studies have discussed the presence of TB symptoms (such as productive cough, chest pain, shortness of breath, haemoptysis, fever, night sweats, and weight loss) among high-risk groups. Nonetheless, high-risk groups with TB such as the PL HIV hardly presented with any typical symptom, while some even presented with unspecific symptoms. Moreover, recent studies also found that a small proportion of these high-risk groups presented with minimal symptoms or asymptomatic, especially those living in countries with a high disease burden of TB [27]. For example, CXR for HIV patients infected with TB is less likely to showed cavitation compared to the TB patient not infected with HIV. In addition, about 22 % of HIV patients infected with TB showed to have normal CXR findings [28]. Despite that, many reports and studies suggested the inclusion of systematic screening as part of the screening procedure for the TB screening programme [17,29]. The presence of TB symptoms would further improve the detection of TB through screening methods.
In CEA, the decision of whether one strategy is better or worse compared to the alternative greatly depends on the ICER. Switching from the existing approach to the symptomatic approach of TB screening will result in higher cost but also higher outcomes. The ICER value of MYR 424.71 reflects that additional one TB case detection requires an additional cost of MYR 424.72 by switching from the existing approach to the symptomatic approach. This value is much lower compared to the WTP threshold of MYR 120,000, and thus, suggests the symptomatic approach is cost-effective. Since the number of symptomatic cases affects the number of screening done, therefore, the proportion of active TB cases presented with TB symptoms are very crucial in deciding the strategy worth in terms of the value and outcome to the provider. Higher the number of cases presented with a symptom would result in more cost-effectiveness of the screening programme.
As much as the result of this study supports the implementation of a symptomatic approach to TB screening, it seems to contradict prior studies, which suggested that the undiagnosed TB patients in the community have a lack of typical symptoms and majorities being asymptomatic. This might indirectly influence the outcome of the TB screening programme, resulting in much lower TB case detection [30]. Consequently, even though it will be a cost-saving option to implement a TB screening only among the symptomatic, this will also result in some active TB cases being undetected in the community due to the low symptomatic yield among the community. Hence, unnecessary neglect of this group of people would result in difficulty in controlling the spread of TB in the community.
The BIA suggested that switching to a symptomatic approach to TB screening would result in an estimated total cost-saving as much as MYR 65.5 million over the five years. This also means that about 4473 TB cases will be undetected over the same period. As much as CEA is concerned, switching from an existing approach to the symptomatic approach would have opposite results. This, however, assuming that the number of screening of symptomatic approaches can match those of the existing strategy. Due to the low symptomatic yield among the high-risk groups, it is impossible to match those numbers, unless there are other methods or algorithms available that can increase the yield of symptomatic approach to TB screening. This can either be achieved by modifying or loosening the criteria for TB symptom definition or creating a new algorithm for symptomatic TB detection [17]. While not all high-risk groups need a full-blown TB symptom, some might need a much milder TB symptom. The conflicting results of CEA and BIA showed in this study are often noticeable in other economic evaluation studies as well, especially in real practice [31]. Therefore, both measures need to be taken into consideration before any decision is made.
The main strength of this study is the inclusion of diverse high-risk groups in the analysis. Hence, providing the overall view on TB screening strategies and at the same time can be used for comparison purposes. The inclusion of BIA as a supplement to the CEA allows the projection of cost and the consequence for five years as well as acts as a valuable decision guide for the policymakers [31]. This will provide a greater view of the impact on budget and outcome throughout five years period. For policymakers, it grants them the leverage in making the decision not only based on the cost-effectiveness measures but also based on the real impact towards the programme if such strategy is implemented [32]. In addition, this study also incorporated input from the MOH, who is behind the TB screening programme in Malaysia. This input is crucial in translating the analysis to local settings.
The policymaker can use the results of this study in deciding to improve the TB screening programme. While TB screening is known as the foundation in TB elimination programme, the budget constraints received for health programme requires policymakers to re-think the strategy for TB screening programme [33]. At the upper management level, the BIA provides the policymakers the estimates on costs incurred as well as the benefit gained or lost by adopting either of the strategy [31,32]. Hence, it depends on the policymakers to decides which strategy is suitable based on the budget availability.
National Strategic Plan for Tuberculosis Control for Malaysia (2016-2020) has outlined few targets to be achieved by the year 2020. One of those is to achieve a TB notification rate (all cases) of 100 per 100,000 population [34]. To achieve this, a symptomatic approach to TB screening alone is not adequate, of which according to the BIA would only result in detection of 1450 to 1632 TB cases (which is around 5 per 100,000 population using the estimated population of 32.7 million in 2020 by Department of Statistic Malaysia) [35]. Hence, a carefully planned strategy needs to be put in place to achieve this target. While it is not a misguided practice to implement a symptomatic approach to TB screening and use the remaining cost-saving to patch up the TB programme with other proven practices such as latent TB screening and treatment, there is no hard blanket that can fit all circumstances.
Notwithstanding the above, this study is not without limitations. Firstly, this study used the number of TB cases detected as a measure of effectiveness rather than the most common measures such as Disability Adjusted Life Year (DALY) and Quality Adjusted Life Year (QALY) used by similar studies. Hence, not suitable for comparison purposes. Nonetheless, few studies used the number of TB cases detected as their effectiveness measure [7,19,36]. Using general effectiveness measures such as TB cases detected may not reflect the true burden of the disease. Hence, it may underestimate or overestimate the cost-effectiveness value since an elderly and young adult would have different QALY and DALY values. Furthermore, since this study only used TB data from Sabah and Sarawak, the results might not be representative of the TB burden in the country. Research in the future should aim at the modelling of TB screening programme by taking into account various factors and systematic screening. In addition, research on screening and management of latent TB is also suggested to complement the result of this study.

Conclusions
In conclusion, policymaker should weigh the inevitable trade-off between saving costs and achieving their goals in the national TB control programme to increase TB case detection. While screening only the symptomatic will results in more than half of TB cases detected, the subsequent marginal number of TB case detection will be much more expensive for the policymakers. A cost-effectiveness study might help in providing a piece of hard evidence but the results of BIA should also be taken into consideration in projecting the impact of such strategy, in the long run, making sure that the decision is in line with the objectives of national TB programme in eliminating TB.

Availability of data and materials
All data generated or analysed during this study are included in this published article.

Consent for publication
Not applicable.

Ethical statement
This study was conducted according to the guidelines of declaration of Helsinki and did not include any identifiable human data. This study was registered under National Medical Research Register, MOH Malaysia (NMRR-19-3443-51729) and has attained ethics approval from Medical Research and Ethics Committee (MREC), MOH Malaysia. This study used secondary data analysis and had received informed written consent from Disease Control Division, MOH as well as Sarawak and Sabah State Health Department.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding
This study was funded by MOH Malaysia (NMRR-19-3443-51729).  Table 2. ɸ The number of asymptomatic was calculated by deducting the total number of screening with the number of symptomatic. ¶ 7.1 TB case detected per 1000 screening from Table 4. Ŧ 41.5 TB case detected per 1000 screening from Table 4 Inflation rate of 3.0 % was used. The funder had no role in the conception, data collection, analysis and write up of this study.