Predicting fluid responsiveness in whom? A simulated example of patient spectrum influencing the receiver operating characteristics curve

Abstract

The influence of patient spectrum on the sensitivities and specificities of diagnostic methods has been termed spectrum bias or spectrum effect. Receiver operating characteristics curves are often used to assess the ability of diagnostic methods to predict fluid responsiveness. As a receiver operating characteristics curve is a presentation of sensitivity and specificity, the purpose of the present manuscript was to explore if patient spectrum could affect areas under receiver operating characteristics curves and their gray zones. Relationships between stroke volume variation and change in stroke volume in two different patient populations using simulated data. Simulated patient populations with stroke volume variation values between 5 and 15 or 3 and 25% had median (2.5th-97.5th percentiles) areas under receiver operating characteristics curves of 0.79 (0.65-0.90) and 0.93 (0.85-0.99), respectively. The gray zones indicating range of diagnostic uncertainty were also affected. The patient spectrum can affect common statistics from receiver operating characteristics curves, indicating the need for considering patient spectrum when evaluating the abilities of different methods to predict fluid responsiveness.

Keywords: Diagnostic techniques and procedures; Population characteristics; Sensitivity and specificity.