eco question

Health Economics ECON 5860 PROF. KURT LAVETTI

Small Area Variations in the Use of Medical Care

2

Small Area Variations

 A large amount of variation in the use of medical care and in total medical spending occurs across small-market areas

 What explains this variation?

3

Spending Variations 4

Source: Dartmouth Atlas of Health Care

What are some explanations why spending might be so different in different geographic areas?

Spending Variations 5

6

Categorizing Care

 Dartmouth Atlas classifies the variations in care into three categories  Underuse of effective care  Misuse of preference-sensitive care  Overuse of supply-sensitive care

7

Underuse of Effective Care

 Services that are of proven value and have no significant trade-offs  Example: Eye exams for diabetics  55 percent get recommended care (McGlynn)

 Receipt of recommended care in Medicare is inversely correlated with spending

 Possible explanations  Too many doctors with none clearly in charge and

responsible for managing organization of care  Financial incentives in accountability are lacking

8

9

Misuse of Preference- Sensitive Care

 Misuse of care that involves trade-offs affecting quality and/or length of life where consumers tend to have strong (but potentially uninformed) preferences  Example: Hip and knee replacements  Variation in breast cancer treatments

 Possible Explanations  Practice style hypothesis: Local medical practice

styles or “medical culture” has a strong influence on treatment choice rather than recommended guidelines

 Imperfect information about “best” medical decisions

 Not supply driven

10

11 Geographic differences in use of “optional” care, after conditioning on characteristics of population and physician supply

Supply-sensitive care and moral hazard  Supply-sensitive care refers to health services whose use

depends greatly on the supply or availability of that service  Example: a doctor’s reliance on MRI technology may depend on

how accessible an MRI machine is to him

 Length of stay in a hospital could depend on how much vacancy there is at the time

 Hypothesis: doctors with greater access to resources will tend to overprescribe care

Supply-sensitive care

 If demand for care is sensitive to supply, then we would expect hospitals with more resources to have larger expenditures

 Dartmouth Atlas shows evidence of a positive correlation between # of hospital beds and # of hospital discharges  Suggestive evidence that the variation is at least in

part due to technology overuse

14

• Strong correlation between total hospital capacity and utilization of hospital care

• Very weak correlation between hospital capacity and certain medical care that can’t easily be induced--like treating a broken hip

15

Attempts to Quantify Variation

 Difficult to precisely quantify what causes this variation in medical utilization patterns

 Finkelstein et al (2016) estimate that differences in patient health and patient preferences (including misuse of preference-sensitive care) explain 47% of the geographic variation in spending  Remaining 53% is due to place-specific factors,

including differences in underuse of effective care and overuse of supply-sensitive care

16

To Understand Behavior Look to Incentives—what Affects Provider Incentives?

 Financial Incentives  Do physicians change the medical advice they give to patients

in response to changes in prices?

 Intrinsic Motivation  What happens when physician ratings are made publicly

available?

 Does the desire to have a high rating affect the way doctors treat patients?

 Is this good?

17

Provider Incentives: Clemens and Gottlieb (2011)

18

• Study how physicians respond to exogenous change in financial incentives

• Medicare reimbursements are based in part on geographic adjustments (reimbursement = base price*adjustment factor)

• Eg geographic adjustment factor is higher in NYC than in Columbus because it costs more to deliver care in NYC)

• In 1996 there 210 geographic regions, and in 1997 to simplify things reduced to 89 regions

• Physicians in many areas had exogenous shock to reimbursement rate from Medicare

Clemens and Gottlieb (2014) 19 Price adjustment regions before change

Clemens and Gottlieb (2014) 20 Price adjustment regions after change

Clemens and Gottlieb (2014) 21 Change in prices caused by geographic adjustment policy

Clemens and Gottlieb (2014) 22

Interpretation: Each dot is an elasticity estimate. Results suggest if prices increase 1%, the long-run aggregate quantity of medical care supplied increases by about 1.5% (elasticity=1.5)

Clemens and Gottlieb (2014)

 Exogenous change in prices  2% increase in payment leads to a 3% increase in

care provision  Elective procedures respond more strongly  Price increases lead practices to buy more

expensive equipment like MRI machines  No impact on health

23

How Does Intrinsic Motivation of Physicians Affect Medical Care?

 What happens when information about physician performance is available?

 When physicians are compared to their peers, if physicians are intrinsically motivated there may be a supply response to information about quality

 Provision of quality information will also affect demand consumers

24

Measuring Provider Quality

 Most common measures are outcome and process  The Center for Medicare and Medicaid Services

(CMS) published hospital level mortality rates beginning in 1987

 Oldest and most widely studied, report cards for surgeons performing Coronary Artery Bypass Grafting (CABG) in New York and Pennsylvania

 CABG reporting now in NJ, CA, MA, UK, etc.  Market based information

 U.S. News and World Report  Increasingly available online physician ratings (healthgrades.com, Angie’s List)

25

Evolution of Quality Report Cards

 Longest running reporting efforts are in CABG  1994-95

26

Risk-Adjusted Expected Fatality Rate

95% Confidence Interval of Fatality Rate

Evolution of Quality Report Cards 27

Average Above Average

Individual Surgeon Report Cards

Risk Adjustment and Selection

 When patients vary in “difficulty” and providers know that their outcomes are being monitored, there is an incentive to select only the easiest patients to treat

 Quality measures require risk adjustment:  Estimate the probability of a bad outcome based on

observable patient attributes  Predict the likelihood that a provider would have been

successful given the patients characteristics  Adjust outcomes for the relative difficulty of patients

treated

29

Provider Response 30

Provider Quality and Consumer Choice  Survey evidence:

 Consumers have difficulty understanding report cards

 Physicians under-use report cards when making referrals to other physicians

 Statistical studies:  Consumers respond to report card information

 But healthier individuals more likely to use report cards!

31

Survey Evidence

 Schneider and Epstein (1996) survey both physicians (left) and patients (right) in Pennsylvania

32

Welfare Effects of Report Cards

 Dranove et al (2003)  Increase in total CABG volume driven by work on

less severe patients  Improvements in matching  Increased spending overall  Worse outcomes for more severe patients with

little gain on healthier  More skilled surgeons prefer to avoid very complicated

cases, so severe patients are worse off after report cards are released

 Conclude that report cards decreased social welfare overall

Welfare Effects of Report Cards

• Report cards increased hospital spending by 3.95 percentage points

• For sicker patients, report cards increased:

• Rate of being readmitted to hospital with heart failure by 2.3 percentage points

• Probability of death by 0.7 percentage points

Attendance Bonus Question

 In the assigned reading “The Cost Conundrum,” what US state does the author study?

35

Spending Variations 36

Physician Behavior from a Physician’s Perspective

 Atul Gawande article “The Cost Conundrum”  Why is McAllen so expensive?  Spending of roughly $15,000/Medicare enrollee  More than twice the national average  Reasons? Mechanisms?  Gawande suggests the fix should focus on

 Reimbursement and incentives for physicians  Improving research on best practices and

guidelines to reduce medical gray area  focus on organizational structures that promote

accountability and discourage overutilization

37