Missing data are ubiquitous in clinical trials. Methods for handling the problem of missing data inevitably depend on missing mechanism assumptions that are often not testable. Therefore, sensitivity analyses become necessary and should always be conducted, especially when the primary analysis result appears significant. The method of delta-adjustment tipping point analysis is widely employed in practice, but it has limitation that the magnitude of delta depends on the scale of the measurement. In this paper, we propose a modified tipping point analysis which uses an exponential decay model to represent the deviation from primary analysis assumption. None the less, we propose a method based on delta approximation method to directly derive the decay rate tipping point. A simulation study is conducted to verify the method and access the features of the decay model. A rare blood disease trial with three biomarker endpoints is studied to demonstrate the practical use of the decay model. The sensitivity analysis conducted using decay model indicated that the primary analysis result was not robust.