Hospital-based evidence on cost-effectiveness of brucellosis diagnostic tests and treatment in Kenyan hospitals

Abstract

Hospitals in Kenya continue to use the Febrile Antigen Brucella Agglutination Test (FBAT) to diagnose brucellosis, despite reports showing its inadequacy. This study generated hospital-based evidence on the performance and cost-effectiveness of the FBAT, compared to the Rose Bengal Test (RBT).Twelve hospitals in western Kenya stored patient serum samples that were tested for brucellosis using the FBAT, and these were later re-tested using the RBT. Data on the running time and cost of the FBAT, and the treatment prescribed for brucellosis, were collected. The cost-effectiveness of the two tests, defined as the cost in US Dollars ($) per Disability Adjusted Life Year (DALY) averted, was determined, and a basic sensitivity analysis was run to identify the most influential parameters. Over a 6-month period, 180 patient serum samples that were tested with FBAT at the hospitals were later re-tested with RBT at the field laboratory. Of these 24 (13.3%) and 3 (1.7%) tested positive with FBAT and RBT, respectively. The agreement between the FBAT and RBT was slight (Kappa = 0.12). Treatment prescribed following FBAT positivity varied between hospitals, and only one hospital prescribed a standardized therapy regimen. The mean $/DALY averted when using the FBAT and RBT were $2,065 (95% CI $481-$6,736) and $304 (95% CI $126-$604), respectively. Brucellosis prevalence was the most influential parameter in the cost-effectiveness of both tests. Extrapolation to the national level suggested that an estimated $338,891 (95% CI $47,000-$1,149,000) per year is currently spent unnecessarily treating those falsely testing positive by FBAT. These findings highlight the potential for misdiagnosis using the FBAT. Furthermore, the RBT is cost-effective, and could be considered as the mainstay screening test for human brucellosis in this setting. Lastly, the treatment regimens must be harmonized to ensure the appropriate use of antibiotics for treatment.

Fig 1. The model structure used for the comparative cost-effectiveness analysis of the Febrile Antigen Brucella Agglutination Test and the Rose Bengal Test.

Fig 2. The relative frequency of results for the cost-effectiveness, in terms of $/DALY averted, for the Febrile Antigen Brucella Agglutination Test and Rose Bengal Test in western Kenya (Scenario 1).

Fig 3. The relative frequency of results for the cost-effectiveness, in terms of $/DALY averted, for the Febrile Antigen Brucella Agglutination Test and Rose Bengal Test in northern Kenya (Scenario 2).

Fig 4. The relative frequency of results for the cost-effectiveness, in terms of $/DALY averted, for the Febrile Antigen Brucella Agglutination Test and Rose Bengal Test nationally (Scenario 3).

Fig 5. Tornado Graph of Spearman Rank Correlation Coefficient values for different input parameters illustrating those most influencing the cost-effectiveness of the Febrile Antigen Brucella Agglutination Test in western Kenya.

Fig 6. Tornado Graph of Spearman Rank Correlation Coefficient values for different input parameters illustrating those most influencing the cost-effectiveness of the Rose Bengal Test in western Kenya.