Supporting Material 2

Clarification Statement for 200701-0920-005 June 13, 2007

Steps to Ensure Obtaining Accurate Physician Addresses When Sampling Through the AMA Master List

We frequently utilize the AMA Master List as a source of addresses for physicians and our experience is that the data we receive is very complete and usually accurate. For a recent CDC survey mailed to psychiatrists we sampled 2,004 providers and sent each a survey using Federal Express, followed by a postcard sent through the US postal service. FedEx will return any package that is not deliverable, allowing us to then try and locate a new address using the Internet and telephone directories. In case the FedEx packet was left in a neighboring office, the postcard will be delivered as addressed. For our survey of psychiatrists we could not locate a current address for only 62 providers (3% of the sample).

For the current proposed survey regarding stillbirth pregnancy outcomes we will utilize a similar protocol with FedEx mailings followed by USPS postcards. We will again use the Internet and telephone directories to identify the current address for any returned mailings. If we only have a PO Box as the sampled address (to which FedEx will not deliver) or if we have a home address provided in our sample (to which FedEx cannot leave an unsigned for survey packet since cash will be included) we will either trace to locate an office address or use the USPS Priority Mail instead. Therefore, we do not anticipate losing much of the sample to bad address data.

The AMA Master List does contain names of doctors who are deceased, retired and not currently conducting direct patient care. We include a postcard in our mailing that lists these options so a doctor or someone in the office can indicate this status on the card and return the postcard to us so that we eliminate that doctor from our sample. For our recent psychiatrist sample of 2,004 names, 103 doctors (5%) were retired or did not see patients, and 6 doctors (0.3%) were deceased. Combined with the 62 for whom we could not locate an address, 8.5% of our sample was ineligible.

 

For this survey, because the sample size will be small, after the two survey mailings, if we still have a group of non-responders for which we have no information, we will attempt to call the office and determine status.