Cardiac Care – A Case Study in Practice Variation
In 2010, when we published our study on healthcare variation in Michigan, we were able to show considerable geographic variation around the state of Michigan on a variety of procedures and services. We intentionally chose services where the research indicated either a tendency toward over-utilization (relative to evidence-based guidelines) or where the guidelines were unclear.
While we adjusted for age, gender, and severity (to the extent possible) in that report, we did not look at outcomes or risk factors that might be influencing variation. So, although we were able to demonstrate considerable variation that seemed unrelated to any particular patient attribute, some wondered whether or not higher rates of utilization could be explained by other population-specific risk factors, and whether or not they might be producing better outcomes of care.
Our latest study, Variation in Interventional Cardiac Care, was designed to look more closely at variation and possible associated circumstances. And while we still cannot definitively answer questions about cause and effect, we do have more information about associations between various factors that have been attributed to practice variation by researchers, policy makers, and clinicians.
We decided to look at several specific questions about cardiac care. In particular, our focus was on regional variation in Michigan for percutaneous coronary intervention (PCI) – surgical stenting. Meta analyses of the clinical research on indications and outcomes for PCI have showed that for many with stable coronary artery disease, surgical intervention may not be necessary. Many people do just as well with a non-invasive approach: medical intervention. Given that research, we wanted to look at both supply-related questions and clinical issues that might be associated with higher rates of PCI in the population.
One theory about practice variation is that use rates tend to be higher in areas with more facilities, services, and/or specialty practitioners. That conclusion has spawned an adage in health care: supply creates demand (or, as an economist, Milton Roemer, put it many years ago: “a built bed is a filled bed is a billed bed”). We looked at this question in terms of the number of catheterization labs in communities, as well as the number of cardiovascular surgeons, compared to the rate of combined cardiac interventions in those communities (much of which is driven by elective PCI). While we did not find any association between the ratio of cardiovascular surgeons and the rate of elective PCI, we did find an association between the ratio of cath labs and rates of elective PCI. That is, regions with higher ratios of cath labs to population also tended to have higher rates of elective PCIs.
Also of great interest was the lack of association we found between higher regional rates of elective PCI (among patients with stable coronary artery disease) and either cardiac risk factors or cardiac mortality rates. That is, communities with higher rates of elective PCI did not have worse population health status than those with lower rates of elective PCI, nor did they have lower mortality rates. Again, we cannot determine cause and effect from these data, but the lack of any association between variation and cardiac risk factors or mortality rates is quite telling.
So, what are we to make of this latest contribution to the variation research? Our work certainly directionally supports the idea that practice variation is largely reflective of community-wide practice patterns rather than evidence-based, clinical guidelines. And, based on other research, we think these community-wide practice patterns tend to be driven by informal medical cultures rather than by patient preferences. If we are correct, and we really want to reduce unwarranted variation and have an impact on health care spending, then informing those practice communities and changing incentives to better align with the evidence will be fundamental to changing the picture we see today.