Comments on Werner et al., “C “Consumer response tto public bli reporting: Changes in market share in nursing g homes” Alan Zaslavsky Dept of Health Care Policy H Harvard dM Medical di l S School h l Do consumers respond to reported quality information? • Heroic efforts for at least 15 years to show consumer response p – Epstein & Schneider 1998, Pennsylvania cardiac surgery reports: <1% of patients knew and used their surgeon’s rating – Guadagnoli et al al. 2000: CAHPS reports read more by consumers who are considering switching plans; impact unknown. unknown – etc. Typical difficulties • Quality info distributed simultaneously, no comparison p g group p • No quality info available prior to reporting • Hard H d tto conclusively l i l show h effects ff t off reporting, distinct from response to quality through other channels of info Werner et al al. strategy • Reconstruct quality measures before and after reporting p g • Difference in differences (before/after by low/high quality) in linear model • Outcome: market shares to various nursing homes (NH) – Market Market-level level fixed effects Technical concerns (1) • Market M k t shares h a problematical bl ti l metric ti – Especially in linear model – HSA #1 with 2 NH: before=50%, 50%, after=55% in good,45% in bad – HSA #2 with 10 NH: before=10% each, after=15% in good, 5% in bad??? • Alternative specifications of outcome – log(market share) →proportional effects – Occupancy rate Technical concerns (2) • Estimation of measure of residual demand – Not in p paper p draft – Role in model not clear – Why needed for diff diff-in-diff in diff analysis? • Falsification test with small NH – No public quality report – Test has low power (small samples/home) Things I would like to know! • Practical significance: How many were moved to higher g q quality y nursing g homes? • Associations of “noticed” measures with overall (“front ( front page”) page ) quality summary summary. Some NH-specific analytic concerns • Complications of capacity/occupancy – Certification of beds for Medicare, Medicaid – Competition for private/commercial pay – Post-acute versus long-term g • Community-dwelling vs long-term postacute Nursing home interpretive issues • Quality measure interpretation – – – – – Outcomes, not processes. Adequacy of risk adjustment??? Subjectivity of pain measure; upcoding?? Walking might be most relevant for many many. 5-star reporting system. • Who makes the decision? Who uses the information? – Families, patients have little input – Discharge planners have major role • More use of quality info (38%) than by family members (12%) – Complex incentives incentives, uncertain agency • What did they actually see? Long-stay Short stay Short-stay Broader issue about quality reporting •Why? •To To whom? Rationales for quality reporting (1) Improve quality by providing information to consumers for informed choice Consumers migrate g to higher-quality g q yp providers Market pressure for quality improvement Standard competitive p economic model assumes: Competitive markets with low switching costs Availability of appropriate information Information processing capacity at minimal cost Unconflicted agency of intermediaries Improvement, not just selection (→ disparities?) Ideology of “Consumerism” Rationales for q quality y reporting p g ((2)) 1. Improve quality by providing information to consumers for informed choice 2. Other motivational aspects Appeal to provider professionalism Accreditation Regulatory/administrative sanctions (or threat) P4P: monetary incentives 3. Value of information to facilitate QI Realistic self self-assessment assessment of weaknesses eaknesses Regulatory/administrative direct action Conclusions • Naïve “consumerism” model may be hard to justify – Empirically or as policy • There are other rationales for quality measurement and reporting – … ranging g g from national p progress g assessment to QI implementation • Public reporting of quality data may be an ethical and political necessity – Applaud efforts to improve and better understand reporting ti • Thanks to – David Grabowski – David Stevenson for helpful discussions