Bundled payments for chronic conditions may be tricky

One of the goals of health care reform has been to replace fee-for-service schemes with new payment models, such as bundled payments, based on the argument that it can make care provision more efficient and, possibly, cheaper.

So far, those bundled payment efforts have been limited to surgical procedures and other one-time care episodes. But what about chronic conditions, which affect a large number of Americans and are expensive to treat?

A new study says it's too early to consider bundled payments for chronic diseases, largely because various systems disagree on how to diagnose patients with a chronic disease and how to determine if certain episodes of care are chronic disease-related. This is especially difficult when comparing patients of different severity or those with multiple chronic diseases.

Published in the latest Medicare & Medicaid Research Review, the study looked at a group of patients who received care between 2003 and 2006 that met the Healthcare Effectiveness Data and Information Set (HEDIS) criteria for diabetes or coronary artery disease (CAD). The patients' care was analyzed using two common episode classification algorithms, Episode Treatment Groups (ETGs) and the Prometheus model, to estimate an appropriate episode-based payment.

The two algorithms varied significantly on the number of patients included and the value of care actually related to diabetes or CAD. For example, HEDIS identified 1,580 patients as having diabetes with a four-year payment mean of $67,280. But ETGs counted only 1,447 of the patients with a four-year payment mean of $12,731. Prometheus counted 1,512 diabetes patients, but eliminated 317 of them as having more than one chronic condition, and the remainder had a four-year payment mean of $23,250.

For CAD, HEDIS identified 1,644 patients with a four-year total payment mean of $65,661. ETG reduced the number to 983 patients with a four-year payment mean of $24,362, but Prometheus identified 1,135 and ran 948 through the model for a four-year payment mean of $26,536.

The researchers concluded that policy makers will have to have more in-depth discussions on how chronic care patients are identified and their care is classified before introducing bundled payments for chronic conditions.