Response to Glasgow

We appreciate Dr. Glasgow’s (1) interest in our article (2) and the opportunity to respond to his comments. He seems to impute a biased worldview to our decision to drop smoking cessation counseling from chart audits subsequent to the baseline audit. First, we can assure him that we were not disinterested in the issue of smoking, as evidenced by the fact that this was one of the areas identified up front as important by the primary care physicians who invited us into their community. Data collection limitations alone drove the decision to drop this measure from subsequent analyses. Our sole source of data for the study was the charts kept by a group of independent solo practitioners, each of whom had a different system for keeping patient information. The auditors had to extract a wealth of information about both physician and patient behavior from these charts in a highly labor-intensive process. For each measure of adherence to guidelines, we had to ascertain both a numerator (the number of patients whose charts showed that they received the recommended care) and a denominator (the number of eligible patients). In all guideline areas except smoking cessation counseling, including the example Glasgow gives, the denominator was relatively easy to ascertain (e.g., all diabetic patients, all diabetic patients <75 years of age, all diabetic patients using insulin). However, since only smokers would be eligible for smoking cessation counseling, we needed a good estimate of the denominator (the number of smokers), which was not available in our dataset. As can be seen in Table 1 of our article (2), documentation of current smoking was only present for 23 patients in the baseline audit. This 8% rate is far below estimates of smoking in the state of Indiana, and was felt to be too unstable a denominator to allow meaningful follow-up of interventions targeted at the numerator.

Second, Dr. Glasgow feels that group-based educational sessions on predetermined topics were doomed to fail. Our sessions for the lay public were highly interactive, well attended, and linked temporally to the physician sessions. Not all studies have shown that group education performs less well than individualized instruction (3), but we agree that there are more potent patient education interventions than those we used in our study. However, interventions such as nurse-based self-management training or proactive phone counseling are costly to initiate and sustain. In closed systems such as Kaiser Permanente or the Veterans Affairs, in which the payor theoretically will recoup the savings that ensue from improved patient self-management, there is an incentive to fund such programs (although, interestingly, they are still rare). Unfortunately, in the amorphous health care system we studied, which is not atypical for much of the U.S., no organized force exists to develop and fund such interventions. We sought to make our interventions less costly and translatable to current systems that do not have funds for nurse case managers or one-on-one education with each high-risk subject.

We agree that there is a need to ask hard translational questions and to search for innovative solutions to the “quality chasm.” Solutions will vary depending upon the resources available in the local environment. It is evident that these are not simple problems or they would have been solved by now.