{ "background": "Inadequate water treatment infrastructure in many regions contributes to persistent public health risks. Systematic diagnostic protocols for existing facilities are needed to prioritise interventions and quantify potential risk reductions.", "purpose and objectives": "This study aimed to evaluate a novel engineering diagnostic protocol for rural water treatment systems by measuring its efficacy in reducing microbial risk through a randomised field trial.", "methodology": "A randomised controlled trial was conducted across 40 community water treatment facilities. Sites were randomly assigned to receive either the new diagnostic intervention or to continue with routine practice. The intervention comprised a structured engineering assessment of key treatment processes. Risk reduction was measured via log-reduction in E. coli counts. The primary analysis used a linear mixed-effects model: $Y{ij} = \\beta0 + \\beta1 Ti + uj + \\epsilon{ij}$, where $Y{ij}$ is the log-reduction for facility *i* in district *j*, $Ti$ is the treatment indicator, $uj$ is a district random effect, and $\\epsilon{ij}$ is the error term.", "findings": "Facilities receiving the diagnostic intervention achieved a significantly greater mean log-reduction in E. coli (2.4, 95% CI [2.1, 2.7]) compared to control facilities (1.5, 95% CI [1.2, 1.8]). The model estimated a treatment effect ($\\beta_1$) of 0.9 log units (p < 0.001). The diagnostic most effectively identified failures in coagulation and filtration processes.", "conclusion": "The structured diagnostic protocol proved effective for identifying critical engineering failures and was associated with a statistically significant reduction in microbial risk at the point of treatment.", "recommendations": "The diagnostic protocol should be integrated into routine maintenance schedules for water treatment facilities. Further research should investigate its cost-effectiveness and long-term sustainability.", "key words": "water safety, randomised trial, engineering diagnostics, risk