Introduction Modified Early Warning Scores (MEWS), calculated from patients' vital signs, are used in hospital to identify patients who may benefit from admission or intensive care: higher MEWS indicates greater clinical risk. We aimed to evaluate MEWS to support paramedics' decisions to transport patients to hospital or treat and leave them at home.

Methods We used an interrupted time series design. We trained 19 volunteer paramedics to use MEWS to support decisions to transport or treat and leave at home. We used linear regression to evaluate differences in weekly transportation rates (percentage of patients attended and transported to hospital) and revisit rates (percentage of patients attended, treated at home and subsequently revisited within 7 days), comparing trends in rates 17 weeks prior (pre-MEWS) and 17 weeks post implementation of MEWS. Auto-calculated scores retrospectively applied to all data provided pre-MEWS and were compared with paramedic calculated scores post-MEWS.

Results Of the 4140 patients attended, 2208 were excluded owing to missing values (n=1897), recording errors (n=21) or excluded clinical complaints (n=290). From the remaining data (n=1932) there were no significant differences in transportation rates (pre=55±6%; post=63±11%) by catering for the existing increasing trends where the confidence intervals of the regression slopes overlap (pre=0.15; 95%CI −0.51 to 0.80 vs. post=0.54; −0.58 to 1.65). Similarly, there were no significant difference in revisit rates (pre=4±4%; post=2±4%) catering for the similar trends (pre=−0.13; −0.53 to 0.27 vs. post=0.08; −0.33 to 0.49). Paramedic scores were incorrect 39% of the time (n=622).

Conclusion MEWS had a minimal effect on transportation or revisit rates. Scores were frequently not calculated or recorded, or incorrectly calculated. Opportunities for ongoing training, clinical support and feedback were limited. A larger study, ensuring adequate ongoing support, is recommended before implementing MEWS on a wider scale.