In this randomized crossover study, healthcare providers with prior experience using pulse oximetry received 2 sets of 10 mock screening scenarios and were asked to interpret the results of each scenario as 鈥減ass,鈥?鈥渇ail,鈥?or 鈥渞etest.鈥?Participants were randomized to use either the paper algorithm or computer-based tool for the first set of 10 scenarios and the alternative method for the second set. We used Wilcoxon rank sum tests to compare the accuracy of interpretation using the 2 methods.
The 102 participants answered 81.6% of the scenarios correctly when manually interpreting the algorithm vs 98.3% correct when using the computer-based tool (P < .001). These differences were most pronounced for the 鈥渇ail鈥?scenarios (65.4% manual vs 96.7% computer, P < .001) and the 鈥渞etest鈥?scenarios (80.7% manual vs 98.7% computer, P < .001), but were also significant for the 鈥減ass鈥?scenarios (94.1% manual vs 99.0% computer, P < .001).
Use of a manual algorithm for the interpretation of results in screening for critical congenital heart disease with pulse oximetry is susceptible to human error. Implementation of a computer-based tool to aid in the interpretation of the results may lead to improved accuracy and quality.