A Primer on Understanding Health Care Cost Trends:

Originally published by the Center for Studying Health System Change (HSC). HSC was a nonpartisan policy research organization funded principally by the Robert Wood Johnson Foundation.

A Primer on Understanding Health Care Cost Trends: The Story Behind the Numbers

Issue Brief No. 05 | December 1996

Virtually every study showed that rates of increase in health care costs had declined throughout the 1990s, but the bottom-line figures often differed from one study to the next. There were many reasons for these discrepancies, including which data sources were used and how researchers analyzed them. No ideal data source existed; the best available sources varied considerably in accuracy, timeliness, and -- most importantly -- whether the information actually matched the questions being asked. This Issue Brief provided guidance on understanding what the numbers really represented.

What Is Being Measured?

When a government or employer survey reported that the rise in health care costs was slowing, the first questions to ask were: What exactly is this study measuring? All health care costs? Whose health care costs? These questions mattered because in nearly all cases, the data had been prepared and adjusted to address a specific policy or research question.

For instance, when analyzing Medicare data to gauge progress in controlling costs, researchers typically adjusted for general inflation and changes in the population served. This was done to isolate changes resulting from cost control efforts from factors like the aging of the population. But if the goal was planning Medicare financing for the 21st century, demographic trends would be critical, and researchers would not want to strip them out.

Health care data could also be examined by their component parts -- the amount spent on hospital care, physician services, and drugs. These breakdowns could reveal important trends in the quantity and price of services, offer insight into what was driving overall cost trends, and provide a basis for projecting future costs.

Employer surveys used premium costs as a stand-in for health care costs. Most gathered data on total premium cost -- including both the employer and employee share. Premium cost trends were valuable for employers but were often a poor indicator of underlying health care cost trends.

How Data Vary by Source

The three major types of data used to construct cost trends were provider costs, claims incurred by insurers, and premiums paid by employers. Ideally, data would be timely, cover the entire population, reflect resources used in providing services rather than what was paid for them, and provide insight into the performance of different plan types. In practice, each major source had strengths and weaknesses.

Provider cost data scored well on several dimensions. They were the most timely, largely because providers logged cost or expense information as it occurred. They also covered the entire population, including the uninsured. A weakness was that hospitals were the main source of information, with limited good data available from other providers such as home health services and clinical laboratories. This problem was expected to grow as inpatient services continued to decline as a share of health care spending.

Claims data from insurers provided complete spending information for covered services and could separate expenditures into price and quantity components. However, proprietary restrictions meant researchers could not obtain claims data from all insurers. Health maintenance organizations (HMOs) recorded visits by encounters rather than claims (if they recorded them at all), and encounter data generally could not be combined with claims data. This made claims data an indicator for only a nonrepresentative and shrinking segment of the health care system. Changes in benefit structure were also hard to capture in claims data.

It was worth noting that the terms costs and expenditures, though often used interchangeably, were conceptually different. Costs reflected the resources devoted to health care that were not available to produce other goods and services. Expenditures were what purchasers paid for health services or what providers received. The two diverged whenever payments exceeded or fell short of the actual resources that went into providing services.

Premium data from employers enabled researchers to track trends by type of health plan, which was useful for understanding the mechanisms behind changing cost trends. For example, were the changes driven by a shift in enrollment from fee-for-service plans to HMOs? Or were both types of plans experiencing a slowdown in premium increases?

On the downside, premium data were not as current as they appeared. Premiums were set months in advance based on claims data and generally remained in effect for a year. Premiums also reflected a three-year underwriting cycle in which premiums caught up to real claims experience. If they had been set too high, premiums might remain flat for two years; if too low, they might spike. In other words, the current premium was unlikely to reflect current underlying claim costs. Additionally, employer data excluded people covered by Medicare, Medicaid, and the uninsured, making them unrepresentative of the nation as a whole.

Examples of Health Care Cost Databases

The gold standard and best-known database on health care costs was the National Health Accounts (NHA), prepared by the Health Care Financing Administration (HCFA) of the Department of Health and Human Services. It used information from providers and insurers and was produced annually, though data for calendar year 1994 were not released until late spring 1996.

Milliman & Robertson's Health Cost Index (HCI) drew data from provider surveys on the major components of health spending: hospitals, physicians, and prescription drugs. The proprietary data were collected monthly and published quarterly, with only a three-month lag time.

The Employment, Hours, and Earnings (EHE) data series published by the Bureau of Labor Statistics at the U.S. Department of Labor covered payroll data for all nonfarm establishments broken down by industry code. The series for production workers in health services establishments could be used to track health care costs. Although it covered only labor costs, the data were high quality and published monthly with a lag of less than two months.

Several leading accounting and benefits consulting firms conducted annual surveys of employers' health plans. These surveys, designed to meet employer needs, included data on premiums and benefits. Many attempted to maintain a core panel of respondents and encouraged participation by providing comparative analyses.

A Matter of Interpretation

The limitations of health care cost data were compounded when people used them for purposes they were never intended for. Employer surveys, for example, were not designed to track cost trends in the broader economy. They were designed to help employers understand what they were spending on health insurance and what benefits they were providing to their employees, especially relative to other employers. They served as a competitive tool for business, not as a measure of national health care expenditure trends.

While generically similar, employer surveys differed for a number of reasons, including their sample composition. Both employer surveys referenced in the Issue Brief showed a significant decline in premium costs for all plan types -- fee-for-service as well as managed care -- but the specific figures were not the same. KPMG Peat Marwick and Hay-Huggins not only drew from different firms but also used different sampling methods: Peat Marwick used a stratified random sample while Hay-Huggins drew from its clients with some additional firms added for representativeness.

Further differences emerged when changes in benefit structure were considered. Hay-Huggins actuaries performed an analysis for the Center for Studying Health System Change examining benefit structure changes by plan type from 1993 to 1995. They found the benefit structure of fee-for-service plans virtually unchanged while HMO benefits had increased somewhat. When benefits were adjusted for, HMO premium increases were between 0.5 and 1.0 percent lower per year.

Survey results could also differ based on how the change in premium was calculated. One employer survey that showed a lower rate of increase based its calculations on the average premium across all plans. But since people were switching from traditional plans to managed care plans with lower premiums, the trend in the average premium would naturally be lower than a weighted average of the trends for each plan type. Whether lower cost trends came from shifts among plan types versus within plan types had real implications for how people viewed the future of health care costs. The outlook was more encouraging if each plan type was slowing its rate of increase, rather than if premiums were growing just as fast but more people were simply switching to cheaper plans.

Reading Behind the Headline

A recent data release from the Department of Labor (DOL) carried the headline: Growth in employers' health costs slows. But did this actually mean cost controls were working? The data behind the headline came from the Employment Cost Index, which included a component for health insurance costs borne by employers.

Several reasons were cited for the slowing growth in employer health insurance costs, but many had little to do with broader cost control efforts. Instead, they reflected the shifting of costs from employers to employees -- such as employees paying a greater share of the premium -- and a drop in the proportion of employees receiving employer-sponsored health insurance, either because employers stopped offering it or because employees (usually those in low-paying jobs) were less likely to enroll, especially for dependent coverage.

This did not mean the DOL data were wrong or misleading. They were not. But data could not serve all purposes, and much of what was presented to researchers and the public had not been developed to portray overall health care cost trends.

This Issue Brief is adapted from "Tracking Health Care Costs" by Paul B. Ginsburg and Jeremy D. Pickreign, which appeared in the Fall 1996 issue of Health Affairs.

Sources and Further Reading

Ginsburg, P. B. and Pickreign, J. D. "Tracking Health Care Costs." Health Affairs, Fall 1996. Center for Studying Health System Change, Issue Brief No. 05, December 1996.