Accuracy of volume measurements by a clinical spirometer in multiple veterinary hospitals

Spirometry devices, which are components of many anaesthesia machines, are commonly used to assess lung mechanics during anaesthesia. Spirometry calibration usually adheres to manufacturer recommendations without established guidelines. Although more accurate and less variable than inbuilt spirometry in certain General Electric anaesthesia ventilators, near-patient spirometry lacks adequate evaluation.

We assessed near-patient spirometers' performance using Pedi-lite and D-lite flow sensors. Certified 1 L calibration syringes were used on 67 monitors located in 14 veterinary hospitals. Three consecutive inspired and expired volume values displayed by the monitors for each volume of the calibration syringe were recorded. Volumes studied were 50, 100, 150, 250, 300 mL for Pedi-lite and 150, 300, 450, 500, 750 mL for D-lite. Measured and targeted volumes were averaged, agreement error calculated. Accuracy was assessed plotting agreement errors against calibration volumes. A linear mixed-effects model was used to obtain linear regression between the error and the calibration volume. Mean, differential and proportional bias, limits of agreement, claimed accuracy and 10% clinical tolerance were calculated and displayed. Differences among monitors were evaluated using the Friedman rank sum test, differences between inspired and expired volumes using the Wilcoxon signed-rank.

Inter-monitor variability for inspired and expired volume readings using both sensors was high; intra-monitor variability was low. The error magnitude was independent of volumes evaluated. Using Pedi-lite, only a minority of measurements met manufacturer's specification or a 10% clinical tolerance; both inspired and expired volumes were significantly underestimated. Using D-lite, superior performance was demonstrated for volumes between 300 and 750 mL (mean biases close to zero and the majority of measurements meeting manufacturer's specifications and clinical tolerance). The difference between measured inspired and expired volumes with both sensors was significant.

These results support caution when interpreting clinical measurements of lung volumes and mechanics in anaesthetised patients when using these sensors. This is particularly important in smaller patients where lung volumes are below 300 mL. Trends should be reliable.