Explanations for variations in hospital expenditures among four large California counties

In 2020, the United States spent about $4.1 trillion or $12,530 per person on health care [1]. At approximately 19.7% of its Gross Domestic Product (GDP), the US had the highest percentage among the Organization for Economic Co-operation and Development countries [1]. As the number of individuals 85 years and older is expected to grow nearly 200 percent by 2060, continued growth in health care expenditures are expected [2]. In addition, the growing number of individuals with pre-diabetes, diabetes, and obesity may increase health care spending. About 44.1 million individuals are projected to have diabetes with annual spending related to diabetes alone expected to be $336 billion in 2034 (in constant 2007 USD) [3]. Between 2012 and 2017, the economic costs of diabetes increased by 26% [4]. Recent estimates suggest that the number of adults with diabetes was 22.3 million (9.1%) in 2014, but will increase to 39.7 million (13.9%) by 2030 [5]. Despite the increase in diabetes and obesity, costs of health care continue to differ substantially in defined geographic regions, even among communities with comparable disease burden.

When examining the population in the United States by insurance groups, Medicaid and Medicare users only make up 19.8% and 14.2% of the population (data from CMS,gov), respectively [6]. Both estimates include about 12.5 million people who are enrolled in both programs. In 2019, about 17% of the Medicare population received Medicaid and about 14% of the Medicaid population also got Medicare benefits [7]. Almost half of the country’s population, about 49.6%, have private insurance policies from employers [6]. However, most policy analyses concentrate on the minority who are insured by government programs. Publicly available data on individuals with private insurance policies has been limited. As a result, health care policymakers use data from the Center for Medicare and Medicaid Services (CMS) to make recommendations for the entire population, even individuals with private insurance policies. Although this is changing as claims data for employer-sponsored health care plans become available, it is still the norm to use databases on Medicare and Medicaid data such as the Dartmouth Atlas of Healthcare [8]. The Atlas has documented variations in the distribution and use of medical resources using Medicare and Medicaid data for over 20 years and has become the gold standard for US health care data.

Recent analyses highlight substantial differences in payment between public and private insurers. Schulman and Milstein showed that US Medicare and Medicaid payments to hospital were, on average lower than the hospital costs [9]. Conversely, private insurers were paying at rates that were more than 140% of hospital costs [10]. In other words, private insurers subsidize public payers. These private insurers pay between 2 and 5 times the rate paid by Medicare and Medicaid.

In this paper, we compare healthcare costs in the four major metropolitan health care markets in California. We focus on this single large state because California has a population larger than most countries (nearly 40 million people), is ethnically and racially diverse, and has environmental variability. Limiting the analysis to one state reduces the extraneous influence of different Medicaid policies across states and the influence of variability in insurance regulatory rules.

Data for this analysis came from several sources. To estimate hospital prices, we used data on self-insured employers, state-based all-payer claims databases, and health plans that have been assembled by RAND [11]. The hospital data were from 2016–2018. However, we believe these indicators remain relatively stable. Relative prices were estimated from the allowed amount paid by the private plan as a percentage of what Medicare pays (by pre-negotiated contract) the same hospital for the same services.

Hospital price data were categorized by county group based on zip code. Statistical tests and descriptive analyses, and graphic displays were completed using R. The U.S. Census Bureau 2020 [12] and the California Health Interview Survey (CHIS) [13] were used to better understand the demographic makeup of each county. Other cost data from Kaiser Family Foundation (KFF) [6] were used to calculate 2019 Medicaid and Medicare spending per enrollee.

Data from the California Department of Health Care Services were used to examine Medicare Advantage enrollment data by county [14]. State- and county-level enrollment numbers are from August 2021. To examine how competitive the different health care markets are and the concentration of hospitals in each of the various markets, we used the Herfindahl–Hirschman Index (HHI) from the Yale Tobin Center for Economic Policy [15].

A general linear model analysis created an identifier for each county by aggregating zip codes falling within county geographic boundaries. Each county was then contrasted with the remaining zip codes in the state.

Interactive data maps on the Dartmouth Atlas website show price-adjusted Medicare reimbursements per enrollee were higher in LA ($14,361.41) in comparison to San Diego ($10,966.41), San Francisco ($9,214.26), and Sacramento ($9,494.30) counties [16]. The RAND data includes hospital price information from self-insured employers, state-based all-payer claims databases, and health plans. Prices reflect the negotiated allowed amount paid per service, including amounts from the health plan and the patient, with adjustments for the intensity of services provided. We specifically examined what private insurers pay hospitals in relation to Medicare contracted rates for the relative price for inpatient and outpatient services. We expected LA to have a similar or higher ratio when comparing the amount private insurers pay hospitals in relation to Medicare, as seen in the Dartmouth Atlas data. The analysis revealed that the ratio was lower in LA (2.15) than San Diego (2.93), San Francisco (3.00), and Sacramento (2.77) counties (Fig. 1). The general linear model analysis demonstrated that the ratio of private payer to Medicare payments for inpatient and outpatient services was significantly lower in Los Angeles in comparison to the rest of California (95% CI 1.26–93.1, p = 0.04), but was non-significant for the other three counties.

We considered four alternative explanations for why the ratio of what private insurers pay hospitals in relation to Medicare for the relative price for inpatient and outpatient services is lower in LA (2.15) than San Diego (2.93), San Francisco (3.00), and Sacramento (2.77) counties. The hypotheses included:

Not only is the population in LA is approximately three times as large as the population in San Diego, 6 times as large as Sacramento, and 11.5 times as large as San Francisco, but the demographic composition of LA is distinct. In comparison to the other counties, LA has a higher percentage of Hispanics and a lower percentage of white, non-Hispanic residents (Table 1).

Considering differences in education, Los Angeles had a similar percentage of individuals with only a high school or an equivalent degree as San Diego and Sacramento counties and a lower percentage than San Francisco. These differences in educational attainment education are associated with income differentials. Median income in LA is $71,358, which is much lower than the median income of San Francisco and San Diego, $119,136 and $82,426, respectively. LA also has the highest percentage of persons below poverty at 14.2%. LA has a higher percentage of individuals without health insurance at 9.2% compared to San Diego at 7.6%, Sacramento at 5.5%, and San Francisco at 3.6%.

Dartmouth Medicare FFS data show that health care costs in LA are higher in comparison to other counties in California. However, RAND data indicate they are lower. In this analysis, we explored four explanations for why the ratio of what private insurers pay hospitals in relation to Medicare is lower in LA than San Diego, San Francisco, and Sacramento counties. Even though LA has the highest uninsured rate among these counties, private insurers are paying hospitals significantly more in San Diego, San Francisco, and Sacramento counties. This observation disconfirms the widely held belief that private insurers pay more in communities where a larger portion of the population is uninsured.

The Dartmouth Atlas Project is built on FFS claims data. Since MA may remove healthier enrollees, FFS is left with more expensive seniors. However, enrollment in MA was not higher in LA. When examining the counties by current health insurance coverage, LA had the highest percentage of individuals enrolled in Medicaid and the lowest percentage of individuals enrolled in Medicare. However, in LA, Medicare costs were lowest and Medi-Cal costs were highest.

Ultimately, none of these first three hypotheses explained the lower ratio of private to Medicare costs in LA. However, the HHI suggested that the LA market is more competitive. Although LA might be the most expensive county when hospitals do not compete on price, such as Medicare FFS, private insurers are better able to negotiate price when there is an oversupply of hospitals. These results are consistent findings in an unpublished white paper by Kronick and Neyaz that identified and ranked the largest and smallest private insurance to Medicare ratios among hospitals in California [17]. They reported four of the ten lowest ratios are in LA, while only one hospital in LA is among those with the top ten largest ratios [17]. These findings underscore the need for more research that includes private health care claims. Concentrating on the Medicare Fee for Service population, as in the Dartmouth Atlas Project data, may obscure the more complete picture that emerges from data representing the full range of payers.

In summary, datasets based on Medicare do not provide an accurate representation of hospital spending for the general US population, but rather for the limited group of Medicare FFS enrollees. Although Dartmouth Atlas of Health Care data suggest that hospital costs are highest in LA, analysis of RAND data indicates that private insurers are paying hospitals less in relation to Medicare contracts in LA than San Diego, San Francisco, and Sacramento counties. The hypotheses investigated: differences in demographics, enrollment rates in MA, and the number of individuals enrolled in Medicare and Medi-Cal, failed to explain why LA hospital costs are lower for those not in Medicare FFS. However, the HHI indicates that high market competition among hospitals could explain this difference. Future research is required to understand this relationship. In addition, we need publicly available data to untangle the significant regional differences in California hospital costs.