Navigating Data Uncertainty and Modeling Assumptions in Quantitative Microbial Risk Assessment in an Informal Settlement in Kampala, Uganda.

Decision-makers in developing communities often lack credible data to inform decisions related to water, sanitation, and hygiene. Quantitative microbial risk assessment (QMRA), which quantifies pathogen-related health risks across exposure routes, can be informative; however, the utility of QMRA for decision-making is often undermined by data gaps. This work integrates QMRA, uncertainty and sensitivity analyses, and household surveys in Bwaise, Kampala (Uganda) to characterize the implications of censored data management, identify sources of uncertainty, and incorporate risk perceptions to improve the suitability of QMRA for informal settlements or similar settings. In Bwaise, drinking water, hand rinse, and soil samples were collected from 45 households and supplemented with data from 844 surveys. Quantified pathogen (adenovirus, , and spp./EIEC) concentrations were used with QMRA to model infection risks from exposure through drinking water, hand-to-mouth contact, and soil ingestion. Health risks were most sensitive to pathogen data, hand-to-mouth contact frequency, and dose-response models (particularly ). When managing censored data, results from upper limits of detection, half of limits of detection, and uniform distributions returned similar results, which deviated from lower limits of detection and maximum likelihood estimation imputation approaches. Finally, risk perceptions (e.g., it is unsafe to drink directly from a water source) were identified to inform risk management.