Assessing the Accuracy of Three National Physician Sampling Frames.

Abstract

T o the Editor: We are writing to clarify several methodological points in BAssessing the Accuracy of Three National Physician Sampling Frames^. In this study, we drew a sample of physicians from the National Plan and Provider Enumeration System and compared the rate of matching on contact information across the SK&A and the AMA’s Preferred Mailing Address data files. We also attempted to contact a random subsample of physicians by phone in order to verify their contact information. We selected the AMA’s PPMA rather than the AMA’s PrimaryOffice Location (POLO) file, because survey researchers frequently use the PPMA as a sampling frame for physician surveys. To our knowledge, the POLO is not in common use for this purpose. Harrington states that we attempted to verify physician addresses in the PPMA by calling the physician office location. This is not correct. For a random subsample of physicians, we attempted to verify the information in each database by contacting the physician at the address and/or phone number listed in the database. That is, if a given physician had three different addresses/phone numbers listed across all three data files (one for PPMA, one for NPPES, and one for SK&A), we attempted to verify each set of contact information. An individual physician could have had up to three correct sets of contact information. We thank the AMA for their cooperation with our study and agree that providing as much information as possible about the array of available physician sampling frames will allow researchers to make informed decisions about this critically important component of research methodology.