2505151

Hospital-Physician Gainsharing

R01 HS018481-01

Jonathan Ketcham, Ph.D.

Arizona State University

September 28, 2010

2505152

Background and Relevance

• Costs are high, value uncertain, and incentives are misaligned due to fragmentation.

• Reforming payment to improve value: from individual to group incentives?

– Unknown whether, when and why group incentives work.

– Challenge for policymakers and managers.

• Exploration by both CMS (demonstration programs) and the private sector (gainsharing).

2505153

Background and relevance

Gainsharing: a group-based incentive, where a physician receives payment if the group overall reduces its costs in predetermined areas relative to its historical baseline.

Preliminary work– Reduced cost per patient

– No negative effects on access, quality

– Subset of patients and time

– Need to understand the mechanisms

2505154

Project aims: To determine…

1. If gainsharing lowered costs through utilization, prices, or both, and how any price reductions were achieved.

2. If gainsharing promoted coordination and standardization of physicians’ treatment decisions.

3. How physicians varied in their responses to gainsharing.

2505155

General Approach

1. Develop a theoretical model

2. Empirically analyze data

3. Develop and estimate a structural model

2505156

Context and Specific Approach

• Proprietary data from coronary catheterization labs

• Analyze all percutaneous coronary intervention (PCI) patients.

• 25 programs at 13 hospitals, plus ≈ 140 non-gainsharing hospitals, 2001-2009.

– Before and during the programs a difference-in-difference strategy.

– Exploring changes in changes (Imbens and Athey)

2505157

Goodroe Healthcare Solutions’ CathSource Data

• Every drug and device chosen by every physician for every patient.

• Hospitals’ prices for those devices, net of rebates.

• Rich risk adjustment, per ACC’s NCDR requirements.

• Timing and targets of gainsharing programs, participating physicians, and their practice affiliations.

– Either price targets alone, or price and utilization targets together.

2505158

Data in the analysis presented today

• 2001 through mid-2007.

• 13 gainsharing programs at 6 hospitals

• 161 physicians from 35 groups treated 58,399 patients under the programs.

• Physician group sizes ranged from 1 to 17 physicians.

• Data from 123 hospitals who did not participate in gainsharing programs but have the software.

• Analysis

– Two highest cost device categories: drug-eluting stents (DES) and bare metal stents (BMS)

– Per patient cost and quantity (risk adjusted)

– Per device price

• Overall

• and within-product to eliminate substitution effect

– Standardization on brand, and on prices at level of

• physician

• group

• hospital

2505159

Insights from the Model

• Group incentives:– Align MD and hospital incentives because of hospitals tiered

contracts with device companies.

– Give MDs reasons to share information and monitor each other

• Savings generated by– Lower utilization

– Lower prices

• Greater hospital bargaining standardization of prices

• Greater contract compliance standardization of brand

• Substitution

• Actual standardization on brand may not occur, depending on the strength of physicians’ preferences.

25051510

Results for Cost and Quantities per Patient

• Price targets alone: $315 drop (-14.2%) for DES, $42 drop (-10.6%) for BMS.

• Adding quantity targets: $444 drop for DES, but $87 increase for BMS due to substitution to them.

– 7.7% reduction in DES use offset by 24% increase in BMS use.

• $358 total reduction in stent cost per PCI patient under either type of targets

25051511

Results for Prices per Stent

• Overall: – DES $120 (4.7%) reduction

– BMS $107 (10%) reduction

• Within-product:– BMS $86 reduction

– DES $123 reduction

Primarily from bargaining and compliance, not substitution, particularly for DES.

25051512

Results for Standardization

• Measured by – hospital-specific brand concentration (an HHI)

– Variation In prices

• For BMS, standardization on brand, and convergence of prices at MD, group and hospital levels.

• For DES, no standardization on brand, but convergence of prices.

– Gainsharing promoted price competition among manufacturers.

25051513

Primary Conclusions

• Theoretically, group incentives have advantages.

• Gainsharing lowered costs. – Greatest for the most costly groups.

– No differences by group size or composition (preliminary).

• Reductions came primarily from lower prices, some from lower utilization.

– Actual standardization in one category.

– But convergence of manufacturers’ prices in another.

– Effects depend on the strength of MD preferences.