The role of high and low hydraulic conductivity (K) regions in heterogeneous, stratified and nonstratified flow fields and the subsequent effect of rate dependent geochemical reactions are investigated with regards to mobilized arsenic from CO2 leakage at a Carbon Capture and Storage (CCS) site. Following the methodology of previous work, human health risk is used as an endpoint for comparison via a two-stage or nested Monte Carlo scheme, explicitly considering joint uncertainty and variability for a hypothetical population of individuals. This study identifies geo-hydrologic conditions where solute reactions are either rate limited (non-reactive), in equilibrium (linear equilibrium assumption, LEA, is appropriate), or are sensitive to time-dependent kinetic reaction rates. Potential interplay between multiple parameters (i.e. positive or negative feedbacks) is shown utilizing stochastic ensembles. In particular, the effect of preferential flow pathways and solute mixing on the field-scale (macrodispersion) and sub-grid (local dispersion) is examined for varying degrees of stratification and regional groundwater velocities. Results show effective reaction rates of kinetic ensembles are dissimilar from LEA ensembles with the inclusion of local dispersion, resulting in an additive tailing effect of the solute plume, a retarded peak time, and an increased cancer risk. This discrepancy between kinetic and LEA ensembles is augmented in highly anisotropic media, especially at intermediate regional groundwater velocities. The distribution, magnitude, and associated uncertainty of cancer risk are controlled by these factors, but are also strongly dependent on the regional groundwater velocity. We demonstrate a higher associated uncertainty of cancer risk in stratified domains is linked to higher aquifer connectivity and less macrodispersion in the flow field. This study has implications in CCS site selection and groundwater driven risk assessment modeling.