Effect of oximetry error on the diagnostic value of the Qp/Qs ratio.

As a quantitative assessment of the magnitude of shunting, the ratio of pulmonary to systemic blood flow (Qp/Qs) plays an important role not only in the oximetric diagnosis of intracardiac and great-vessel shunts but also in the treatment of the patient. However, the oxygen saturation measurements used to compute the Qp/Qs ratio contain errors due to physiological variability and measurement error of the oximeter used to analyze the blood samples. We have developed a mathematical model to describe the variability that oximetry errors contribute to the uncertainty in the Qp/Qs ratio. Using this model, we compute the probability of making an inappropriate recommendation regarding corrective surgery when a particular value of the ratio is the criterion for surgery, e.g. a Qp/Qs ratio >2. This report also contains a spreadsheet that readers can use to analyze their own oximetry data by computing confidence intervals for the Qp/Qs ratio. The results presented here support the following conclusions. First, because the Qp/Qs ratio is calculated from saturation measurements at four different sites, oximetry errors make the Qp/Qs ratio less effective at detecting the presence of a shunt than the conventional step-up method that depends on samples from only two sites. Second, although oximetry errors are equally likely to cause the calculated Qp/Qs ratio to overestimate the true Qp/Qs ratio as to underestimate it, the overestimations on average have greater magnitudes than the underestimations. Third, in comparison with an oximeter that has 2.5% measurement error, using an oximeter with 1% or less error greatly reduces the uncertainty in the Qp/Qs ratio and thus increases the probability of reaching the right decision regarding corrective surgery. Fourth, the variability in apparent Qp/Qs ratios is also greatly diminished by taking multiple blood samples from each of the four requisite sites and averaging them before calculating the Qp/Qs ratio. Although increasing the number of blood samples from each site can compensate for the error of an oximeter, this approach can be impractical, particularly if the oximeter error is 2.5% or greater.