Estimating the costs and resources required to achieve rabies elimination through mass dog vaccination in Kenya

Murungi, Maurice, Muloi, Dishon, Saitoti, Ezra, Fevre, Eric and Thomas, Lian (2026) Estimating the costs and resources required to achieve rabies elimination through mass dog vaccination in Kenya. [Data Collection]

Data Catalogue DOI: 10.17638/datacat.liverpool.ac.uk/3116

Description

This dataset contains the programme data, model inputs and simulation code underlying a costing study of the mass dog rabies vaccination programme in Machakos County, Kenya, 2020 to 2024. The study combined observed programme expenditure and vaccination records with a Monte Carlo simulation model to estimate the cost and workforce required to reach 70% vaccination coverage of the county's owned dog population. The dataset combines three sources of information: 1. Programme records (campaign_summary and programme_expenditure sheets), giving the annual number of animals vaccinated, target populations and geographic scope for each campaign phase, and total programme expenditure aggregated by cost category. 2. Model inputs (model_inputs, scenario_definitions and unit_cost_sources sheets), giving the point estimate, range and distribution of every parameter in the simulation, the parameters changed in each scenario, and the programme records from which each unit cost was derived. 3. Simulation code (machakos_cost_model.R), an R script that reproduces all results, tables and figures in the publication from the model inputs. The data were collected in Machakos County, Kenya, from a vaccination programme implemented by Veterinaires Sans Frontieres Germany and the One Health Research, Education and Outreach Centre in Africa at the International Livestock Research Institute, funded by the German Federal Ministry for Economic Cooperation and Development, in collaboration with the County Government of Machakos. STUDY DESIGN A retrospective cost analysis was conducted from the public sector perspective using the implementing partner's annual financial reports and the programme's vaccination registers for 2020 to 2024. Each expenditure item was coded to one of eleven cost categories. Unit costs derived from these records, together with the estimated owned dog population, target coverage, team throughput, campaign duration and vaccine wastage, were assigned probability distributions and combined in a Monte Carlo simulation of 50,000 iterations to estimate the total cost, cost per dog and number of teams required for a county-wide campaign. Five scenarios were simulated: the base case, a donated-vaccine scenario, and three optimised scenarios with campaign durations of 90, 60 and 360 days. Sensitivity to each input was assessed using partial rank correlation coefficients. Each row in the campaign_summary sheet corresponds to one campaign year. Each row in the model_inputs sheet corresponds to one model parameter. Each row in the simulation output files produced by the script corresponds to one iteration of the model. Model equations: D = P x C; T = ceiling((D / CD) / VT); total cost (KES) = (a + b) + [c(1 + W) + d + e(1 + W) + f] x D + [g + h x CD + 0.5 x i x CD + j] x T; cost per dog = total cost / D. KES are converted to USD at a fixed rate of 129. Symbols are defined in the model_inputs sheet. DATA STRUCTURE The dataset package consists of three linked components: 1. Excel workbook (machakos_rabies_cost_model_data.xlsx) with seven sheets: - model_inputs: one record per model parameter, with symbol, point estimate, minimum, maximum, distribution, unit and how the cost is scaled in the model - scenario_definitions: the parameters changed in each of the five scenarios - campaign_summary: one record per year, 2020 to 2024, with phase, months active, geographic scope, target dog population, dogs, cats and donkeys vaccinated, proportion of target vaccinated, and team structure - programme_expenditure: total expenditure by cost category, 2020 to 2024, in KES - pilot_expenditure_2020: expenditure of the 2020 pilot campaign by category, in KES - unit_cost_sources: for each unit cost, the observed values in programme records, the years and record type, and how the point estimate and range were set - data_dictionary: definition of every column in every sheet and in the simulation output files 2. R script (machakos_cost_model.R), which reads the model inputs, runs all scenarios and the sensitivity analysis, and writes the simulation outputs. Requirements: R 4.1 or later; packages mc2d, sensitivity, dplyr, ggplot2. Place the script in a folder named code with a folder named data alongside, set the working directory to code, and run source("machakos_cost_model.R"). Outputs are written to a folder named output. The script uses a fixed random seed (123), so the outputs are identical on every run. 3. Simulation outputs, generated by the script: per-iteration results for each of the five scenarios (50,000 rows each), summary statistics, median cost by component and scenario, and partial rank correlation coefficients with bootstrap confidence intervals. These are not deposited as files because they are fully reproducible from the script and the model inputs; their columns are defined in the data_dictionary sheet. Model parameters are linked to the unit cost sources and to the code by the parameter symbol (P, C, VT, W, CD, T, a to j). Scenario definitions are linked to the model inputs by the same symbol and to the simulation outputs by the scenario identifier (base, donated, opt90, opt60, opt360).

Keywords:

dog, ecology, rabies, kenya, gps

Date Deposited:

25 Sep 2026 09:33

Last Modified:

25 Sep 2026 09:35

DOI:

10.17638/datacat.liverpool.ac.uk/3116

Geography:

Kenya, Africa

URI:

https://datacat.liverpool.ac.uk/id/eprint/3116