Cost-Effectiveness Evaluation of a Novel Integrated Bite Case Management Program for the Control of Human Rabies, Haiti 2014–2015

Haiti has the highest burden of rabies in the Western hemisphere, with 130 estimated annual deaths. We present the cost-effectiveness evaluation of an integrated bite case management program combining community bite investigations and passive animal rabies surveillance, using a governmental perspective. The Haiti Animal Rabies Surveillance Program (HARSP) was first implemented in three communes of the West Department, Haiti. Our evaluation encompassed all individuals exposed to rabies in the study area (N = 2,289) in 2014–2015. Costs (2014 U.S. dollars) included diagnostic laboratory development, training of surveillance officers, operational costs, and postexposure prophylaxis (PEP). We used estimated deaths averted and years of life gained (YLG) from prevented rabies as health outcomes. HARSP had higher overall costs (range: $39,568–$80,290) than the no-bite-case-management (NBCM) scenario ($15,988–$26,976), partly from an increased number of bite victims receiving PEP. But HARSP had better health outcomes than NBCM, with estimated 11 additional annual averted deaths in 2014 and nine in 2015, and 654 additional YLG in 2014 and 535 in 2015. Overall, HARSP was more cost-effective (US$ per death averted) than NBCM (2014, HARSP: $2,891–$4,735, NBCM: $5,980–$8,453; 2015, HARSP: $3,534–$7,171, NBCM: $7,298–$12,284). HARSP offers an effective human rabies prevention solution for countries transitioning from reactive to preventive strategies, such as comprehensive dog vaccination.


Appendix A. Economic and epidemiological input for the program evaluation
Appendix A shows the economic and epidemiological inputs used to evaluate the Haiti Animal Rabies Surveillance Program (HARSP). Table A1 shows the costs of capital and operation during the implementation of HARSP in Petionville, Carrefour, and Croix-des-Bouquets, Haiti. Table A2 shows the annual capital and operational costs of the animal rabies diagnostic facility costs, and Table A3 shows the annual costs of training the personnel involved in the implementation of the HARPS program. We used a 3% discount rate to estimate the costs of capital investments. Notes: *We used constant dollars (no inflation) and a real discount rate. 1 †These estimates show the equivalent annual cost for the capital outlay assuming that the resale value is zero. The "clinical" life of an equipment is usually less than its physical life, so were being conservative in our estimate.

Implementation of HARSP: costs of surveillance, diagnostics, and training
‡The total costs of surveillance, including personnel, transport, diagnostic tests, and related costs were included as operational costs. We included the costs per dog euthanized, home observation of dogs that depend solely on the total number of dogs investigated (i.e., drugs for sedation, euthanization, and the costs of disposal). § For sedation, Xylazine: Ketamine = 2: 10. To produce complete anesthesia, 2 ml / 10 kg IM; for lethal injection (euthanasia), 2mEq/mL; dose: 2mmol/kg or 1mL/kg intracardiac. 2,3 ¶If owned, estimate the cost per m 2 for an office in the same area. Office rental and utilities costs were considered separately for labs and rabies surveillance. If the office space was shared, the value was multiplied by the share of time dedicated to rabies. #The cost per worker day was based on the monthly salary of a veterinary technician (Centre) of US$275. We considered that 10% of these costs correspond to fringe benefits. **Annual costs of personal protective equipment and equipment for laboratory work, assuming that they wear out in a year. Notes: *We used constant dollars (no inflation) and use a real discount rate. 1 †These estimates show the equivalent annual cost for the capital outlay assuming that the resale value is zero. The "clinical" life of an equipment is usually less than its physical life, so we were conservative in estimating the years of useful life. ‡ If owned, estimate the cost per m 2 for an office in the same area. Office rental and utilities costs were considered separately for labs and rabies surveillance. If the office space was shared, the value was multiplied by the share of time dedicated to rabies. Notes: The costs of training were prorated through the time frame of the project evaluation (with no adjustment for inflation or discounting since the investment was on the first year). *The cost per day per participant was based on the salary/wage and fringe benefits of participants in the workshop; the figure shows a weighted average of the costs per worker involved in the animal surveillance program. †We assumed that the classes took place in the facilities where the program was implemented, thereby the costs of renting a classroom were zero. ‡We estimated that a teacher would spend at least the equivalent to classroom days + field days + 2 days of settling in. §Teacher's wage was estimated based on an annual salary of $80,000, assuming training from a CDC employee or equivalent, and 260 (=52*5) annual work days.
¶Per Diem allowances for Petionville, Haiti. Lodging: $155.00, meals and IE $110.00. 4 #The price of air-tickets were estimated at market value. 5 Table A4 shows a summary of the epidemiological data from HARSP used in the effectiveness evaluation. The comparison scenarios for patients with suspected exposure to rabies in the intervention area of the program, including the situation in Haiti before the implementation of HARSP (no bite case management scenario, NBCM), the HARSP program, and the recommendations and guidelines for rabies treatment by HARSP (HARSPr) and the World Health Organization (WHOr) are shown in Table A5.

Time frame for analysis Years 5 5 Definition
Notes: HARSP denotes Haiti Animal Rabies Surveillance Program. *The categories of dog rabies infection (i.e., confirmed, probable, suspected, and negative) were used as a reference. HARSP and HARSPr were the only scenarios in which this information would be known; under WHOr and in the no bite case management scenario (NBCM), all patients would be treated as "suspected" rabid exposure. †The types of contact are following WHO (2013) 6 PEP recommendations. Category I: touching or feeding animals, licks on the skin. Category II: nibbling of uncovered skin, minor scratches or abrasions without bleeding, licks on broken skin. Category III: single or multiple transdermal bites or scratches, contamination of mucous membrane with saliva from licks, exposure to bat bites or scratches. Category I requires no treatment, Category II requires immediate vaccination, and Category III requires immediate vaccination and RIG.
‡We are assuming that HARSP detected all potential human rabies cases in the study area. §The probability that the offending dog had rabies was estimated based on empirical data, as described in Wallace et al., 7 Figure 2. Probability = (confirmed + probable dogs) / total dog investigations. We used this probability for suspected dogs in the HARSP sample. ¶ Probability of developing rabies in the absence of PEP if bitten by a rabid dog. #The number of dogs placed under observation in 2014 reflect 444 dogs that were observed for 14 days because they had bitten a human, and 9 dogs that were put in observation for 6 months because they were bitten a probably rabid dog. In 2015 there were 1,186 dogs under observation for 14 days, and 3 dogs under observation for 6 months.

Fatal rabies infections
Per the four programmatic scenarios (NBCM, HARSP, HARSPr, and WHOr), we estimated the total fatal human rabies infections as: Where subscript i stands for a patient who was in contact with HARSP or local health clinics, i.e., is part of the study sample, j represents the type of exposure (i.e., confirmed, probably, suspected, negative).

FRyear = fatal rabies infections (year = 2014, 2015)
ERij = patient i with type of bite exposure j was exposed to rabies, i.e., P(ERij) is the probability that a patient i with bite exposure type j was exposed to rabies (confirmed = 100%, probable = 75%, suspected = 6.3%, negative = 0%) MCij = patient i with type of bite exposure j seeks medical care, i.e., P(MCij) is the probability that a patient i with bite exposure j seeks formal medical care

Cost and cost-effectiveness of the program
Here we show the results from the evaluation of the program based on the share of the costs of PEP regimes that is paid for by the government (0%, 50%, and 100%). Figure B1 shows that the total costs of the program (2014 US dollars, thousands) vary substantially based on the share of PEP costs that the government pays. Figure B2 shows the cost-effectiveness of the program as total costs per death averted and Figure B3   recommendations and guidelines for implementation of the program and rabies treatment. ¶The age distribution to estimate years of life lost was based on the age distribution of Haiti, assuming that Haiti has the same incidence rates of rabies by age as Tanzania. 9 Figure B4. Two-way sensitivity analysis of the total fatal human rabies infections in the area of implementation of HARSP in 2015 by share of patients who seek medical care (%), and probability that a person bitten by a dog was exposed to rabies, A: 1%, B: 6.3% (estimate from HARSP), and C: 36%. † Notes: The figure shows results for 2015, results for 2014 are shown in the main manuscript. Figure 2A

Estimates extrapolated to the West Department
Combining literature and modeling, a study of the burden of rabies estimated 130 annual deaths from rabies in Haiti. 10 Assuming that the disease burden of rabies is proportional to the human population, we would have expected 14 deaths in the three communes where HARSP was implemented in 2014 (population : 11 1,085,817), which is what we obtained in the NBCM scenario. Using the same criteria, we would expect an average of 48 annual deaths in the complete West Department (population: 11