Image by Sara CC 2.0 via Flickr https://flic.kr/p/fawUQ

An equitable healthcare system, where each patient receives equal treatment regardless of socio-economic status (SES) is an important public policy goal. A key question is whether market-based reforms aimed at increasing efficiency have had adverse consequences for equity. To inform this debate, we conduct a study of equity in the United States, comparing the private and public systems.

We find clear evidence of socio-economic status disparities in healthcare spending for privately insured people in the United States. Some striking results are that, for women under 65 diagnosed with breast cancer, spending by the health care system is $7,033 higher for college educated women than for high school dropout women. Similarly, for men under 65 diagnosed with non-specific chest pain, medical spending on the college educated exceeds that of the high school graduates by $5,968.

Remarkably, we find that once people reach age 65 and are covered by public Medicare insurance, socio-economic status disparities in spending largely vanish. This may serve as a cautionary tale for those in Australia who advocate a system more closely resembling that of the United States.

Methods

The United States has a mixed public/private system of health care financing where most people under age 65 purchase private insurance through an employer, while most people aged 65 and over have access to the publicly funded Medicare system. Our key aim is to determine whether the level of equity differs significantly across these two types of insurance systems.

For this purpose, we compare the levels of healthcare spending across patients with different levels of socio-economic status, conditional on type of insurance and medical need. We proxy socio-economic status by education level, which is a relatively fixed individual characteristic and is highly correlated with earnings ability and lifetime income. Medical need is captured by respondents’ detailed medical conditions at the 3-digit ICD-9 level.

We note that low income individuals who are constrained by out-of-pocket costs demand less medical care. But the key question is whether healthcare system spending varies with socio-economic status, controlling for ability to pay (that is, net of out-of-pocket spending, and controlling for insurance type).

We use data from the Medical Expenditure Panel Survey (MEPS) 2000-14 which is collected by the US Department of Health and Human Services. After excluding those younger than 23 and those with missing observations, the subsamples used in our analysis contain 132,327 women under age 65; 114,491 men under 65; 30,509 women who are 65+; and 22,193 men who are 65+.

We estimate regression models of individual-level total healthcare spending on a large set of covariates that predict spending (medical conditions, demographics, health, and health insurance). The coefficients on the ICD-9 medical diagnosis codes in these regressions can be interpreted as the average increase in medical spending that occurs if a person is diagnosed with that condition, relative to the baseline level of spending on a “similar” person who does not have the condition, but who has the same values of the control variables.

We estimate separate regressions for people under and over 65, for men and women, and for different education groups (less-than high school, high school, college).

To infer the role of ability-to-pay in accounting for any observed differences in total spending across education groups, we re-estimate the same regressions using individual-level, out-of-pocket medical spending as the dependent variable.

In the absence of disparate treatment based on socio-economic status, individuals with the same ICD-9 medical diagnosis codes should have the same level of total healthcare spending, regardless of education. Differences in ICD-9 code coefficients across education groups are interpreted as evidence of socio-economic bias in treatment when these are not accounted for by out-of-pocket spending differences.

Results

A major finding is that total spending on women younger than 65 diagnosed with breast cancer differs greatly by education level. Conditional on our extensive set of controls, a breast cancer diagnosis leads to an increase in annual medical spending of $10,865 among college women, compared to only $3,832 among women with a less-than high school education. Importantly, differences in out-of-pocket spending on breast cancer across education groups explain only 10 per cent of the difference in total spending. Thus, differences in ability to pay play little role in explaining the socio-economic status spending gradient on breast cancer.

For women aged 24 to 64, we also find significantly positive socio-economic spending gradients for seven of the other top 30 high-cost conditions: normal pregnancy, esophageal disorders, depression, disc disorders, joint disorders, anemia, and consultation without sickness. For instance, in the case of normal pregnancy, spending on college women is $1,630 greater than for high school dropout women. Only 16 per cent of this difference is accounted for by the difference in out-of-pocket spending.

For men aged 24 to 64, a striking result is the large education gradient in spending on “non-specific chest pain.” Conditional on our extensive controls, a diagnosis of “non-specific chest pain” leads to an $8,292 increase in annual medical spending on college educated men, compared to only $2,324 for men with a high school education level, a difference of $5,968. This difference is generated entirely by the healthcare/insurance system itself, in the sense that we find no differences in out-of-pocket spending between education groups.

In sharp contrast to the results for men and women under age 65, we find no clear evidence of positive socio-economic spending gradients on medical conditions for those aged 65 and over covered by Medicare.

We do find that system spending is generally higher for better educated men aged 65+ among those in poor health, conditional on medical conditions and the other controls. But it turns out that this is explained by the prevalence of private insurance among men 65+: many retired people in the United States have private health insurance as a retirement benefit provided by their former employers; such benefits are more common for men. When we delete those with private insurance coverage from the sample, and re-estimate the regressions, we no longer find any significant socio-economic status spending gradients for either men or women aged 65+.

Conclusion

We examined how the United States healthcare system allocates resources to patients from different socio-economic groups who present the same medical conditions. We found that in the working age population that is predominantly covered by private, employer-sponsored health insurance, spending is increasing with education for several important medical conditions that account for a large share of total costs. Differences in ability to pay out of pocket go only a short way in explaining this gradient. But we find that socio-economic disparities in health system spending are greatly mitigated by the public Medicare system for those in the 65+ population.

There are several possible explanations for our findings, including discrimination in healthcare in the population not eligible for Medicare. Patient’s socio-economic status has been found to be associated with physicians’ perceptions of patients’ abilities and behavioral tendencies, which may affect recommended treatment. Socio-economic groups could also differ in their ability to navigate the insurance system to obtain the care they desire. All of these channels would imply a strong case for policy intervention.

Socio-economic spending differences could also arise from differences in preferences or beliefs towards healthcare, leading to heterogeneous demand across socio-economic groups. Or socio-economic status may alter the effectiveness of specific health treatments. Medical providers may also price discriminate, such that high socio-economic status individuals pay more for the same treatment.

More research is needed to understand the root causes of socio-economic disparities and the implications in terms of system equity and efficiency. However, our results suggest that market-based health care reforms might lead to losses in equity.

 

This blog post is summarized from Capatina, E, Keane, M & Maruyama, S 2018, ‘Socio-economic Disparities in U.S. Healthcare Spending: The Role of Public vs. Private Insurance‘, Discussion Papers 2018-03, School of Economics, The University of New South Wales.

Like or share this piece:

Leave a comment

Your email address will not be published. Required fields are marked *

*