The Effect of Anti-Fraud Enforcement on Medicare Costs and Quality
The Medicare program may be subject to fraud and abuse by providers due to the difficulty of observing the health conditions of and treatment received by beneficiaries. Examples of abusive or fraudulent behavior include billing for services not provided, coding patients as having a more severe illness in order to obtain a higher reimbursement, and providing services that were not medically necessary. Reliable estimates of the magnitude of the problem are difficult to obtain, but the Department of Health and Human Services has estimated improper Medicare fee-for-service payments at $12 to $23 Billion, or roughly 7 to 14 percent of all reimbursements.
In Detecting Medicare Abuse (NBER Working Paper 10677), David Becker, Daniel Kessler, and Mark McClellan investigate the effects of anti-fraud enforcement on Medicare costs and quality.
The authors note that most previous studies rely on a review of medical records to estimate the magnitude of Medicare fraud and the types of claims and providers associated with abuse. But because of their cost, such studies rely on small sample sizes and may not yield precise estimates. Such studies also shed little light on the questions of whether anti-fraud enforcement efforts reduce the frequency of abusive behavior, whether the effects of enforcement differ by patient or provider type, and whether enforcement affects patient outcomes.
To answer these questions, the authors assemble data for a large sample of beneficiaries who experienced one of six illnesses particularly vulnerable to abuse during 1994-1998 from the Medicare Provider Analysis and Review file. This is matched to data on death records, hospital characteristics and state-level Medicaid enforcement expenditures; the authors argue that the latter are a good proxy for Medicare enforcement efforts due to the extensive administrative overlap between the agencies responsive for policing both programs. The rich data allows the authors to control for any unobservable characteristics of patients that vary at the zip code level, such as initial health conditions.
The authors find no significant effect of enforcement expenditures on hospital expenditures. However, as the authors note, if states with high level of abuse tend to invest more in enforcement, this would tend to bias the estimated effect towards zero, so this is not a useful estimate of the likely effect of additional enforcement efforts.
The authors do find significant differences in the effects of enforcement across types of patients and hospitals. When enforcement efforts are greater, hospitals reduce expenditures by more for healthy, young, male patients than for infirm, older, female patients. One explanation for this finding is that hospitals realize that the former group is more likely to be able to rehabilitate at home with the aid of a spouse.
The authors also report that expenditures are more responsive to enforcement for patients admitted to for-profit hospitals (and to a lesser extent to non-profit hospitals) than to public hospitals. This provides some support for the view that for-profit hospitals are more responsive to incentives in ways that are sometimes socially harmful.
Another notable finding is that expenditures are more responsive to enforcement for patients admitted to hospitals that participate in a physician-hospital organization or provide skilled nursing care than those that do not. This is consistent with the concerns that physician-hospital organizations provide a vehicle for hospitals to disguise illegal compensation to physicians for referrals and that there may be substantial wasteful use of skilled nursing care.
Finally, the authors report that there is no evidence of systematic or substantial effects of enforcement efforts on health outcomes. How-ever, they caution that their data does not allow them to measure all dimensions of health, such as the rapidity or completeness of recovery from illness.
This research was supported by the Center for Social Innovation at the Stanford Graduate School of Business and the National Institute on Aging (Grant #P30-AG12810 and #R01-AG17756).