An experts’ opinion-based comparison and benet cost analysis of post-mortem versus tuberculin skin test surveillance systems, Mpumalanga, South Africa

Tuberculosis (TB) is a global health concern caused mostly by Mycobacterium bovis (M. bovis) and Mycobacterium tuberculosis (M. tuberculosis) in animals and humans respectively. As part of TB control strategies, most governments instituted test and slaughter policies for bovine TB (bTB) eradication with varied level of success. Using the SurvCost ® we evaluated the postmortem surveillance (PMS) system as an alternative to the tuberculin skin test (TST). Experts’opinions survey was used to collect information on the perceived level of acceptability of PMS and TST, successes and challenges of both surveillance systems, economic and budget data. Benet cost analysis of both systems were evaluated and the comparative economic benet of PMS over TST was determined.


Abstract
Background Tuberculosis (TB) is a global health concern caused mostly by Mycobacterium bovis (M. bovis) and Mycobacterium tuberculosis (M. tuberculosis) in animals and humans respectively. As part of TB control strategies, most governments instituted test and slaughter policies for bovine TB (bTB) eradication with varied level of success.

Methods
Using the SurvCost ® (http://www.cdc.gov/idsr/survcost.htm), we evaluated the postmortem surveillance (PMS) system as an alternative to the tuberculin skin test (TST). Experts'opinions survey was used to collect information on the perceived level of acceptability of PMS and TST, successes and challenges of both surveillance systems, economic and budget data. Bene t cost analysis of both systems were evaluated and the comparative economic bene t of PMS over TST was determined.

Results
TST implementation was challenging due to poor logistics, procurement challenge, poor feedback, inconsistency in testing and poor return rate for retesting. Experts agreed that PMS was cheaper but almost impracticable due to late detection and probable poor compliance rate but farmers were more open to PMS than TST. Personnel cost remains the largest part of the surveillance cost (47.8% of total costs). TST and PMS systems can be up to 4.40 and 5.96 times more bene cial that not tackling bTB respectively and PMS is 1.35 more cost bene cial that TST.

Conclusion
While TST is empirical, compliance by farmers was poor due to the associated inconveniences. In the alternative, PMS was convenient for farmers but experts believed that adherence will be poor unless increased manpower is available. We advocated for a blended approach between the two systems.
Improved eld surveillance and detailed economic data should bene t future economic assessment.
Trial registration Not applicable.

Background
Mycobacterium bovis (M. bovis) is the causal agent for bovine tuberculosis (bTB), an infectious and chronic disease of livestock, wildlife and humans [1,2]. M. bovis is a signi cant pathogen due to its conservation threat (wildlife infections), economic implication (cattle disease), and zoonotic importance.
Bovine TB remains a worldwide problem, therefore it is imperative to intensify control and preventive measures in livestock aimed at its eradication [3].
The TST is recognised universally and used for preliminary diagnosis in bTB control programmes [4].
Other tests include antibody ELISAs and Bovigam®™ (a commercial IFN-g assay), used as ancillary tests in eradication and control of bTB [4]. In South Africa the TST is accepted as the standard eld test for diagnosis in bTB control programmes. The advantages of the TST and reasons for its extensive use are low costs, relatively better accessibility of tuberculin puri ed protein derivative (PPD), history of successful use, and the lack of better alternative methods to detect bTB [5]. Despite these advantages the skin test has many known limitations, including the di culties in eld administration and interpretation of results, the need for a supplementary visit to measure the skin response and associated noncompliance by farmers, a low degree of standardisation, and imperfection associated with test accuracy [5,6]. As such, its eld application appears unpopular among the eld veterinarians and their paraveterinary staff [7].
Eradication of bTB from cattle in some countries has been unexpectedly protracted, and this has raised questions about the effectiveness of skin testing, particularly when the incidence of disease in the population is low or where there is the potential for contact with environmental mycobacteria. Given the above, recent research has focused on developing alternative, innovative, and complementary testing procedures [2,8].
It has been shown that in countries with low bTB, disease prevalence or disease-free status, meat inspection is effective as a diagnostic and surveillance tool [4]. The importance and impact of meat inspection as a diagnostic and preventive tool cannot be overemphasized. In addition to providing epidemiological information on life threatening zoonotic diseases of meat-borne origin, such as bovine TB, brucellosis, and other toxicoinfections (such as salmonellosis), it also ensures standards for hygienic and wholesome meat free from infections and other toxicoses of inorganic sources are met and upheld [9].
In 2015, Musoke and colleagues performed a bTB surveillance study on cattle in the Mnisi Community, based in the Bushbuckridge District Municipality, Mpumalanga province, South Africa [10]. They found a low prevalence (approx. 0.34%, excluding suspect animals) of bTB in the cattle population. This low prevalence in cattle may justify the use of point-of-slaughter postmortem evaluation in communal areas for diagnosis and surveillance. VanderWaal et al. [11] studied different surveillance strategies such as PMS and targeted testing as alternatives to routine skin testing in low TB prevalence settings. They found that targeted surveillance was more cost effective and reduced sampling effort by 40% without increasing the incidence of bTB.
Currently, in the Mnisi Community in the Bushbuckridge Municipality of Mpumalanga province, the TST surveillance in cattle is minimally applied due to high costs and unwillingness to conduct the di cult TST by technicians (personal communication with professionals). This is not an unusual phenomenon. O'Reilly & Daborn [1] have noted that the TST is not always performed as recommended because of management conditions that make it di cult to perform. Furthermore, bTB TST evaluation has focused on cattle [12; other livestock species such as goats and sheep have been neglected though they are susceptible to M. bovis [13].
In some parts of Africa, bTB detection has depended on slaughter surveillance as the most economical and e cient method for the detection of infected cattle [14]. Meat inspection at abattoirs is thus considered as a pivotal and the utmost obligatory method for the detection of bTB or other mycobacterial infections [14]. In view of the foregoing and considering that such PMS systems, which are seen as cheaper and easier to perform, are adapted to rural communities at the point of personal slaughter and not just at the abattoirs, we sought to investigate if this would be a more cost effective and e cacious detection method for TB surveillance versus the current bTB TST in the Mnisi Community which has achieved poor compliance in TST? Also, would the veterinarians and para-veterinarians consider this procedure a less laborious and easily adaptable method? Outcome can serve as a proof of concept to geographical terrains with similar situations.

Expert opinion survey
To get a representative opinion of the acceptability of the proposed postmortem surveillance system, a questionnaire directed at government o cials (veterinary directors, state veterinarians and technicians) from all nine provinces of South Africa was developed. Questionnaires were administered by email or by telephonic interview. It included open-ended and closed questions, which focused on the current budget for TST surveillance, challenges faced with the surveillance program, opinion on an incentive-based PMS system and a panel of questions on logistic and other issues that support the application and administration of the current programme.

Bene t cost analysis
Budget data were obtained from the State Veterinary services. Additional prevailing economic data were retrieved either through eld survey or personal interviews with the stakeholders in the industry. Population and other demographic data as well as prevalence of TB in Mnisi, Mpumalanga or South Africa were obtained through literature review and from the website of Statistics South Africa (www.statssa.gov.za). Using budget information from State Veterinary services, all costs (recurrent, operational and capital expenditure) were calculated under different cost heads (Supplementary data).
An annual summary of all costs was generated as outputs, and an integrated disease surveillance cost for TB was produced based on eld data using the SurvCost ® , an Excel-based analysis tool (http://www.cdc.gov/idsr/survcost.htm). Integrated Disease Surveillance and Response System (IDSR) is a system whereby all diseases of interest in an area through passive or active surveillance are reported together using the same human, capital and infrastructure resources already available to the area [15].
SurvCost ® uses this system to estimate surveillance costs of diseases of interest. Outputs were generated in tables and graphs. Comparative costs and effectiveness were evaluated using the bene t cost analyses of overall estimated surveillance costs using TST and PMS.

Results
Expert opinion survey Ten experts that included directors of veterinary services, state veterinarians and animal health technicians (AHT) in the South African Veterinary Services participated in the expert opinion survey. A response rate of 67% was obtained. In view of the sensitivity of budget matters, and stringent rules regarding non-disclosure, it was a challenge to obtain some detailed budget information from technical experts. While some feigned ignorance or referred the researcher to higher authorities, others ignored the question completely. However, detailed information was obtained from two provinces and this formed the basis for comparison and validation of economic data.
Ninety percent of all experts had a bTB surveillance program operational in the areas covered by their operations. The TST is most used as recommended by DAFF for South African Veterinary Services [16].
The main challenges of the TST reported by participants include among others: poor logistics in the implementation of the herd testing program, centralised and red tape in the procurement system for reagents (tuberculin), lack of timeous feedback to central agency (DAFF) and no consistency in frequency of testing.
Overall, 90% of all experts suggested that a PMS system is impracticable and not good for adoption.
Reasons advanced for opposing the PMS were: it detects infection very late and so may enhance spread, farmers would use their own resources to report pending slaughter and this might result in low compliance, slaughter rates are low, there will be too many homesteads to follow up, attitude of farmers varies and there are illiterate farmers who may not take the strict implementation seriously, manpower and transport would be a challenge, and that some mortalities may not be reported.
The experts agreed that the TST is not being implemented effectively due to the above cited challenges, but do not see PMS as a viable and practical alternative.

Bene t Cost Analysis
Using SurvCost ® , an integrated disease surveillance and response system implemented for TST will cost R1,783, 242 per annum for the Mnisi Community with an average cost of R84.33 per animal. Speci cally, 35.4% of the cost will be for personnel, 16.0% for o ce support, 7.5% for transport, 30.9% for test sundries (i.e tuberculin ,syringes, gloves), 0.2% for treatment and 9.9% as capital cost (Table 1, Figures1  and 2). For a similar programme implemented for PMS, the overall cost will reduce to R1,320,361 per annum with an average cost of R62.44 per animal. In this case, 47.8% of the cost will be for personnel, 21.7% for o ce support, 15.6% for transport, 1.3% for test sundries, 0.3% for treatment and 13.4% as capital cost (Table 1, Figures 1 and 2).
In both graphs, (Figures 2 & 3, Supplementary material), personnel costs were the highest by proportions.
In TST surveillance, the costs of tuberculin added signi cantly to the test sundries cost while in PMS, the test sundries costs decrease proportionately to approximately 1% since there was no need for the skin test. However, the cost for transport increased proportionately in PMS due to costs incurred in door to door communal slaughter postmortem inspections. Overall, in this study, the PMS model appears cheaper than the TST model.
The 2017 estimates of cattle population in Mnisi will be 21,145 and at a 3.1% prevalence of bTB in cattle, approximately 656 heads of cattle will be infected (Musoke et al., 2015); at a market value of approximately R12 000 per cow, a hypothetical 100% rejection of infected cattle population will result in losses of R7,872,000. Even, at 50% or 25% rejection rates, assumed losses will amount to R3,936,000 or R1,968,000 respectively ( Table 2). Through the integration of TST and PMS, the bene t cost ratio will be 4.41 or 5.96 respectively. While it is evident that conducting the annual surveillance is better in both respects using TST or PMS, the latter (PMS) was at least 1.35 fold economically cost bene cial compared with TST (Table 2). Below a 25% rejection rate of the infected animals, the surveillance system becomes non-bene cial for implementation in the Mnisi Community. Similarly, surveillance using TST or PMS based on visceral (lungs and livers) rejection alone cannot justify the investment as inputs far outweigh the expected bene t. (Table 2). Additional bene ts including the deduction in risk of transmission to other animals and prevention of 339 potential human cases were not quanti ed in this analysis.

Discussion
The experts con rmed that cattle are the most tested animals and the TST is most used test but agreed that the test is under-implemented and not effectively carried out. A review of the implementation strategies and reasons for non-compliance will need further evaluation. It will also be necessary to include all susceptible animals in such a surveillance programme. Furthermore, key challenges have been identi ed and efforts will need to be engaged to overcome these issues. Because outbreaks of other state-controlled diseases (foot and mouth disease, rabies, and brucellosis) divert attention from bTB surveillance, a comprehensive or integrated surveillance system that incorporates all the potential endemic diseases may need to be implemented. On the low level of compliance with second time visits by farmers, intensive awareness campaigns may achieve some degree of success. However, because farmers will prefer PMS, and experts will want a continuation of the TST, a blended approach combining these two surveillance systems may need to be assessed as no stakeholder can be ignored with regards to effective surveillance. It will be important for experts to also pay attention to gathering detailed eld level data for future epidemiological evaluations. In the present survey, only 40% of the experts have veri able prevalence data. It should be understood that the diptanks present excellent opportunity for the collation of such epidemiological data.
This study provides economic justi cation for the implementation of surveillance against bTB in the rst instance, and showed evidence that PMS is cheaper than the TST.
It is approximately six times and four times more bene cial to perform the PMS and the TST surveillances respectively than to allow the spread of bTB in the Mnisi Community without any control. Mwacalimba, Mumba & Munyeme [22] have reported that taking into consideration only the monetary value of a cow at point of sale, there is a positive cost-bene t to the control of bTB. If a broader approach is considered such as the impact of TB on human health and tourism in TB affected wildlife, there is no doubt that costs associated with TB control are minimal compared to the bene ts of eradicating TB [22]. In this analysis, even for 25% level of losses, it is bene cial to conduct surveillance against TB; however, lower level losses do not justify surveillance in this respect. This may partially explain why countries with low TB prevalences or disease-free status undertake meat inspection only as a cheap alternative for TB surveillance system [4].
It is agreed that this bene t cost analysis has certain limitations because it does not take into account the broader bene ts of performing surveillance such as reducing zoonotic disease risk, human loss in quality of life and productivity when sick, loss in animal production, milk and meat yields, loss in tourism and conservation when wildlife species are affected. These may serve as additional reasons to justify surveillance in the Mnisi Community. Mpumalanga has a human TB incidence rate of 0.4% and this translates to 339 human infections in Mnisi Community. The control of TB in animals could potentially prevent a proportion of these human cases [18][19][20].

Conclusion
Finally, both models have their own merits and challenges. The PMS model would need increased manpower and transport resources while the TST has not been effective due to di culty in implementation and low level of compliance. As the TST currently achieves low effectiveness in South Africa, merging the two surveillance systems remains a viable surveillance option. The most effective and cheapest way of controlling zoonotic diseases is to control it at the animal source, and this study may be used as a baseline for future studies to nd more effective ways to control bTB in resource-poor or rural communities, and enhance the 'End TB' strategy [23].