Bias in patient assessments of general practice: general practice assessment survey scores in surgery and postal responders.

Patient-based measures of the quality of primary care are increasingly important. However, their effective use requires bias to be minimised. Scores on the General Practice Assessment Survey (GPAS) differ according to whether patients are surveyed in the surgery or by post. It is not clear whether these differences relate to the mode of administration or to the types of patients who complete the scale in postal and surgery samples. Regression indicates that the bias reflects both effects and should be considered when GPAS scores are being interpreted.