Unexplained Variations in Patient Admissions: Part I

There is an unexplained geographic variation in how often patients are admitted to inpatient hospital care for a given diagnosis, with significant variation identified independent of patient age, insurance coverage, or clinical severity of illness.

There is an equally important means by which to measure this variation: how often preference-sensitive procedures are performed.

The decision to perform some procedures is straightforward: for example, almost all hip fracture patients undergo surgery. For other procedures, such as spinal surgery for symptomatic spinal stenosis or elective coronary artery stenting, the intervention is one of several treatment options.

This sort of care is called preference-sensitive as the decision to undergo the procedure is dependent on choices made by patients and physicians in the face of competing sets of risks, benefits, and tradeoffs.

Taking the example of spine surgery for lumbar stenosis, looking at age-, sex-, and race-controlled data for Medicare fee-for-service patients, analysis by The Dartmouth Atlas Series has found several- fold variation across the country in how often lumbar decompression or fusion procedures are performed. Some regions have rates 4 to 10 times higher than others.

Another area of wide unexplained variation is catheter-based intervention for coronary artery disease. A study by Daniel Matlock and colleagues looked at the rates of catheter-based intervention in Medicare fee-for-service patients across 32 hospital referral regions in 12 states.

They found that urgently needed procedures such as treatment of acute coronary syndromes did not vary very much. However, elective procedures such as treatment of stable symptomatic coronary disease varied significantly.

For both angiography alone and angiography with stenting, rates of age-, sex-, race-, and income-adjusted procedure performance varied more than twofold.

Interestingly, performing the same analysis for Medicare Advantage patients demonstrated an even larger, three- to four-fold variation.

Importantly, the magnitude of variation seen in studies such as those I’ve described is felt to far exceed variation attributable to differences in severity of illness. As discussed previously for medical hospital-based decision-making, reducing this sort of variation is crucial to attempts to improve outcomes and reduce harms and costs in our healthcare system.

Among other tools, utilization of evidence-based criteria sets and procedure rate benchmarks, such as those produced by MCG, is central to this effort.