Expanding insurance coverage: Financial and quality spillovers in

local health care markets

JA Pagán9/15/2009

Acknowledgment

• AHRQ’s Minority Research Infrastructure Support Program.

• AHRQ’s M-RISP Program funded the Health Services Research Initiative at The University of Texas-Pan American.

• Grant Number R24HS017003.

Issues• Health care reform debate is confusing.

• Everyone wants to know: “What’s in for me?”

• Understanding potential spillovers of insurance/ uninsurance on local health care markets is key.

• Empirical analyses based on data from the Community Tracking Study → Potential/realized access and assessment of health care providers.

• Policy modeling and implications.

The uninsured population• 45 million nonelderly uninsured in 2009.

• 54 million nonelderly uninsured by 2019.

• 65% are from low-income families (<200% FPL).

• 35% of Latinos, 29% of Native Americans and 20% of African Americans are uninsured compared to 12% of whites.

Sources: Congressional Budget Office, 2009; Kaiser Family Foundation (2006) The Uninsured, A Primer.

Rationale for providing accessRationale for providing access

• Good but not fully convincing argument:

“It is the right thing to do”

• Enlightened self-interest argument:

“It makes you and me better off”

Local communities and uninsuranceLocal communities and uninsurance

Lower revenue forhealthcare providers

Increased public/Private spending

Financial instability ofProviders/institutions

Reduce charity/uncompensated care

Economic effects

Access to care

Population health

Public health services

Unmet medical needs by community Unmet medical needs by community ranking and insurance statusranking and insurance status

17.1%

18.8% 18.6%17.2%

5.6% 6.1%7.2%

8.3%

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

16.0%

18.0%

20.0%

Low (8%)

Mid-low(12%)

Mid-high(16%)

High (22%)

% n

ot g

ettin

gne

eded

car

e

Uninsured

Insured

Pagán JA, Pauly MV. (2006). “Community-level uninsurance and the unmet medical needs of insured and uninsured adults,” Health Services Research, 41(3): 788-803.

• IOM framework: High uninsurance results in higher uncompensated care (bad debt and charity care).

• End result: Insured people pay for this (10% higher premiums; higher prices; cost-shifting).

• This is a pecuniary community spillover (that is, a financial spillover that works its way through prices).

• Local pecuniary spillovers on the insured are likely to be small when money flows from the outside (e.g., Medicare DSP, state funds).

Spillovers and vulnerability: Spillovers and vulnerability: the case of community uninsurance the case of community uninsurance

(Pauly and Pagán, Health Affairs 2007)(Pauly and Pagán, Health Affairs 2007)

• Occur when the uninsured demand a lower quality and quantity of health care than the insured.

• Preference externalities: “distinct groups of

consumers who have substantially different preferences [from others]… bring forth products with more appeal to themselves but less appeal to others” (Waldfogel, 2003).

• Caveat: Local health care markets must not be perfectly segmented.

Non-pecuniary or real spilloversNon-pecuniary or real spillovers

• Uninsured demand lower quality care.

• Even if the insured have differential access to health care (e.g., better facilities to go to), there are high fixed costs for quality differentiation.

Non-pecuniary, real spillovers in Non-pecuniary, real spillovers in partially insured marketspartially insured markets

• Community 1: Uninsured never use charity care and can only pay for low quality care.

→ Only quality spillover.

• Community 2: Uninsured get charity care and do not skimp on quality.

→ Only pecuniary spillover.

Example: Two communities of same size and Example: Two communities of same size and number of uninsurednumber of uninsured

• By patronizing health care suppliers that do not provide care to the uninsured or who do not engage in cost shifting.

• This can only happen in large enough markets that can be segmented (e.g., in certain regions and for low-fixed-cost services).

• Quality spillovers are more likely in specialty care, care requiring expensive equipment (e.g., scanning), etc.

How can the insured reduceHow can the insured reduce these negative effects? these negative effects?

Community uninsurance and mammography Community uninsurance and mammography (J of Clinical Oncology April 2008)(J of Clinical Oncology April 2008)

• Women ages 40-69 from 2000-2001 CTS HS (n=12,595).

• Prob of mammography screening falls by 1.3% for every 10% increase in community uninsurance.

• Report covers: (1)Health insurance

coverage trends(2)Health effects(3) Community effects

2009 IOM Report

Pauly & Pagán: Spillovers of uninsurance in communities

• 2003 CTS Household Survey → Participants clustered in 60 communities (48 large metro areas, 3 small metro areas and 9 non-metro areas).

• Working age population (N = 31,935).

• Health care measure = f(Community uninsurance, Charity care, X, C).

Dependent variables

• Potential and realized health care access: (1) Respondent has a place to go when sick(2) Visited doctor(3) Visited doctor for routine preventive care

• Assessment of health care providers: (1) Satisfaction with choice of PCP(2) Satisfaction with health care received(3) Trust doctor

Independent variables (distinguishing pecuniary and real spillovers)

• Community uninsurance rate → Proportion uninsured.

• Charity care → Proportion of uninsured respondents in a community with no cost-related problems obtaining medical care.

• Charity care proxy is correlated with different sources of free care (hospitals, doctors and FQCHCs; Herring, 2005).

Had a place to go when

sick

Had a doctor’s

visit in the past year

Had a visit for routine preventive

care

Has seen specialist in the last 12

months

% % % %

Community uninsurance

Percent uninsured in community (Baseline=15.15%)

92.01 84.98 64.79 98.25

Percent uninsured in community (New=25.15%)

87.96 83.37 62.56 97.94

Percentage point change -4.05 -1.61 -2.23 -0.31

Charity care

Percent of uninsured population with no cost-related access difficulties (Baseline=67.71%)

92.18 85.05 64.91 98.26

Percent of uninsured population with no cost-related access difficulties (New=77.71%)

92.05 85.08 64.65 98.18

Percentage point change -0.13 0.03 -0.26 -0.08

Simulated Effects of Changes in Community Uninsurance and Charity Care on Potential and Realized Health Care Access

Very satisfied with PCP

choice

Very satisfied

with health care

Trust doctors Very satisfied with choice of

specialist seen

% % % %

Community uninsurance

Percent uninsured in community (Baseline=15.15%)

62.68 53.73 72.14 74.54

Percent uninsured in community (New=25.15%)

55.62 51.03 70.76 71.82

Percentage point change -7.06 -2.70 -1.38 -2.72

Charity care

Percent of uninsured population with no cost-related access difficulties (Baseline=67.71%)

63.10 53.92 72.22 74.67

Percent of uninsured population with no cost-related access difficulties (New=77.71%)

61.34 52.68 72.05 74.59

Percentage point change -1.76 -1.24 -0.17 -0.08

Simulated Effects of Changes in Community Uninsurance and Charity Care on Assessment of Health Care System Providers

Conclusion/caveats

• Model predictions depend on: – the degree of market segmentation, – the amount of charity care (or below cost care), and– the source of resources to cover the cost of charity or

below cost care.

• Empirical evidence points to both real and pecuniary spillovers.

• The theory is value neutral → Insurance-related quality spillovers could be good or bad.

• Different approaches to cover the uninsured will lead to different effects on the insured.

• Ex 1: Increase support for safety net providers that cater to the uninsured (e.g., community health centers) → little positive real spillover on the insured.

• Ex 2: Tax credits for health insurance would have a positive real spillover on the insured.

PunchlinePunchline

Agent-based model Agent-based model (D Damianov, JA Pagán)(D Damianov, JA Pagán)

• Objective: Understand consequences of various health care reform proposals.• Approach: Modeling incentives to purchase insurance and use services based on the quality

of the service provided, price, and risk/preferences. • Model useful to understand complex social dynamics and learning between micro/macro-

level processes.• Agents respond to their social context, especially to the actions of other members in their

community (local health care market).

FutureFuture• Health care reform plans are constantly changing: How do we muddle

through and get our message across?

• In making the individual, business or social case for reform, the answer to the “What’s in for me?” question should always consider the broader community/market effects.

Thanks!Thanks!