The Accuracy and Precision of Gravity-Driven Infusions

When infusion pumps are unavailable in the prehospital setting, clinicians must estimate fluid flow by counting drops in a drip chamber. We aimed to assess the accuracy and precision of gravity-driven infusions and to estimate the magnitude of inaccuracy and imprecision attributable to human and non-human factors. We recruited 55 paramedics to participate in structured, experimental, simulated study under controlled laboratory conditions, in which each participant set nine drip rates as accurately as possible on a standardized setup. Measurements of flow were made visually (drip rate) and volumetrically (delivered fluid). Volumetric measurements were used to represent overall error, visual measurements were used to represent human error, and the difference was used to represent non-human error. We measured accuracy (percentage error) and precision (mean absolute deviation) using the natural log of error factor to represent the multiplicative decomposition of human and non-human error additively. Mixed effect models were used to assess accuracy and precision across targeted drip rates. Inaccuracy and imprecision varied by targeted drip rate. Overall inaccuracy ranged from -45.3% (95%CI, -52.7% to -36.8%) to 5.8% (95%CI, -8.5% to 22.3%). Human inaccuracy ranged from -38.8% (95%CI, -46.0% to -30.8%) to 44.1% (95%CI, 27.1% to 63.4%). Non-human inaccuracy ranged from -26.7% (95%CI, -30.1% to -23.2%) to -6.8% (95%CI -11.1% to -2.3%). Imprecision was primarily driven by human factors. The degree of human variation between trial participants ranged from 24.2% (95%CI, 15.3% to 33.7%) to 43.7% (95%CI, 33.4% to 54.9%) across targeted rates. The targeted drip rate of 1gtt/s was more accurate than the pool of all other targeted rates (absolute error 31.1% versus 48.5%). Targeted drip rates of whole numbers were more accurate than targeted drip rates of fractional numbers (absolute error 40.7% versus 54.1%). Targeted drip rates below 1gtt/s were more accurate than rates above 1gtt/s (absolute error 35.3% versus 63.1%). Gravity-driven fluid delivery exhibits inaccuracy and imprecision, which varies by targeted rate. While potentially modifiable human factors may improve the accuracy of this technique, non-human error contributes substantially to inaccuracy. There is a need to assess whether this inaccuracy and imprecision adversely affect patient outcomes.