A new method called CAD-DA has been developed to test anomaly detection results after adapting to different environments. CAD-DA can control the chance of wrongly identifying anomalies at a set level. This method addresses the challenge of adjusting for domain adaptation effects to ensure accurate results. By using conditional Selective Inference, CAD-DA can handle these effects. This is the first method capable of conducting valid statistical inference in domain adaptation scenarios. The CAD-DA method has been tested on both artificial and real datasets, showing promising performance.