Accountable Care Organizations: Early Results and Future Challenges

In interpreting lessons from these evaluations, several questions are worth keeping in mind. If a new payment system is correlated with changes in medical spending, is this explained by underlying changes in prices or in quantities? Since medical spending is the product of prices of services and quantities (volume) of services, an intervention that affects spending must affect either the prices or the volume of care. In the Medicare program, where prices are standardized, a global budget contract that works off of the underlying physician fee schedule would only affect spending through volume. In the private insurance sector, however, an ACO contract may affect spending through both volume and prices, since variations in prices across providers creates an opportunity for savings if care is obtained through a less expensive provider.

Separate from its relationship to medical spending, which is measured through the claims submitted by providers, what is the connection between a new payment system and total payouts from the insurer to the provider? An ACO contract contains a variety of incentives to providers that may generate additional payments from the insurer (most notably shared savings and quality bonuses). These non-claims payments may partially or entirely offset savings obtained through medical claims, making them an important dimension in the evaluation of the contract. Yet they are also different from medical claim dollars in a meaningful way. Changes in medical spending reflect underlying physician (or patient) behavior—what care is delivered and how much of it is delivered—whereas non-claims payments reflect the incentive structure of the contract.

On the quality dimension, it is worth noting whether a new payment system has similar effects on process and outcome measures. Process measures, which have been widely used by health plans, are operationally similar to additional items on a fee schedule, whereby the delivery of a service effectively generates a payment. Clinical outcome measures, such as blood pressure or cholesterol, and patient experience measures, on the other hand, cannot be fulfilled by simply checking off a box. Therefore, these measures may represent quality in a more meaningful way.

Early Results from Medicare ACOs

Thirty-two provider organizations entered the Pioneer ACO program in 2012, with about 669,000 Medicare beneficiaries attributed to these organizations. According to CMS, medical spending growth for ACO beneficiaries was 0.3% in the first year, compared to 0.8% for similar beneficiaries outside of these organizations [18].This generated a gross savings of $87.6 million in 2012, of which $33 million were returned to the Medicare trust fund. These savings were driven by 13 ACOs, with another 17 ACOs not reporting statistically significant changes in spending and 2 ACOs garnering losses with spending above the target of about $4 million total. Lower rates of admissions and readmissions largely explained the savings. A separate analysis comparing these ACOs to their local markets estimated a higher year 1 savings of $147 million dollars, driven by 8 ACOs whose savings ranged from $396 to $1224 per beneficiary per year [19].

Pioneer ACOs also earned over $76 million for quality. In the first year, quality bonuses were awarded for the reporting of quality measures rather than for performance, and all 32 ACOs successfully reported. According to CMS, Pioneer ACOs on average performed better than fee-for-service Medicare beneficiaries on 15 clinical quality measures for which comparison data were available, including blood pressure control and cholesterol control for diabetic patients. A complete analysis of quality performance is not yet available.