Multiple prior comparisons of the performance of the U.S. health care system with competitor countries have found that it consumes a much larger share of national income to achieve health and quality of care that are no better, and are in some cases worse, than other industrialized countries and some rapidly developing countries. Business Roundtable commissioned Mercer Health and Benefits to construct an annual study that reflects the impact of these international differences in health system performance on the global competitiveness of U.S. employers and their workforces.
Mercer selected a national expert advisory panel, and subject to their guidance, created such a study. The resulting Health Care Value Comparability (HCVC) Study uses a standardized 100-point scale to combine differences on annual comparisons of the United States to competitor countries along 19 of the most employer-relevant health spending and health measures. Mercer used the HCVC Study to compare the United States with a basket of five U.S. trading partners (the “G-5”) that are relatively affluent and separately with a basket of three large rapidly developing countries (“BIC”). Internationally reported measurements available for use in both studies are imperfect, suffer a two-year reporting lag, and do not fully capture all elements of a health care industry’s national value to employers and their workforces. Accordingly, confidence in conclusions drawn from annual changes in the HCVC Study will grow as the number of measurement periods and measures increase.
After weighting component measures for employer and worker relevance and transforming them into a study that equally weighted workforce health and health care spending, the resulting study score suggests that the value to U.S. employers and workers of the U.S. health system was 23 percent below that of the G-5 countries’ health systems. The bulk of the U.S. value shortfall was attributable to much higher spending in the United States to attain a level of workforce health and care quality that trails the G-5 by roughly 10 percent across 17 measures.
Compared with the G-5 countries, the HCVC Study score deteriorated by a quarter percentage point based on measures reported in 2008, and by a half percentage point during the year prior. If this trend is confirmed over the next several years, it suggests that the United States is not making progress against the G-5. Compared with the basket of three developing countries, the study also showed no progress over the same three-year period.
In addition to tracking the annual change in the HCVC Study, Mercer constructed a supplementary study of annual change in factors likely to predict the pace of U.S. health care value improvement over the next 10 years, the HCVC Forecast (HCVC-F) Study. These include adoption of performance accelerants such as electronic health records and performance-sensitive provider payment methods. Using a constant set of such measures, the HCVC-F Study rose 0.6 percentage points during the most recent measurement period. This compares unfavorably with the year prior, when the HCVC-F Study rose by 6.9 points.
These disappointing movements in the HCVC and HCVC-F Studies add to the urgency felt by all major U.S. stakeholders to stimulate faster improvement in the value of America’s health industry. Employers, employees and government purchasers must do more to reward high-value care. Clinical teams and their suppliers must learn and energetically apply the discipline of systematic process re-engineering that underpins our most globally competitive industries. We can do better. The global competitiveness of U.S. employers and workers depends on it.
Building a Health Care Value Comparability Study
At the request of Business Roundtable, Mercer Health & Benefits convened expert panels of health economists, health service researchers and clinicians to guide Mercer’s development of a new, intuitive, annually releasable study of the U.S. health care system’s performance. The purpose of such a study is to enable Business Roundtable members and other key stakeholders to track annual U.S. health care system progress compared with competitor countries.
In releasing this study, Mercer acknowledges the generous and thoughtful guidance of its expert panelists. They participated on three different project work groups. While panelists’ views substantially converged, this report does not necessarily represent their individual views or the positions of their organizations. When panelists’ views were not unanimous, Mercer’s lead author and panel coordinator, Arnold Milstein, M.D., M.P.H., took responsibility for issue resolution. The report’s expert panelists included:
- Harvard University, Richard M.J. Bohmer, M.D., M.P.H.
- Harvard University, W. Dale Compton, Ph.D.
- Purdue University, William H. Dow, Ph.D.
- UC Berkeley, Alan Garber, M.D., Ph.D.
- Stanford University, Paul B. Ginsburg, Ph.D.
- Center for Studying Health System Change, Dana Goldman, Ph.D.
- RAND, Ronald C. Kessler, Ph.D.
- Harvard University, Robert Mecklenburg, M.D.
- Virginia Mason Medical Center, Len Nichols, Ph.D.
- New America Foundation, Michael O’Grady, Ph.D.
- Independent Consultant, Neeraj Sood, Ph.D.
- RAND, John Toussaint, M.D.
- ThedaCare Center for Healthcare Value, Arnold Milstein, M.D., M.P.H.
- Panel Coordinator, Mercer Health & Benefits and
- UCSF Institute for Health Policy Studies
Scientific consensus has not yet been reached on a direct measure of health care system impact on employer and worker global competitiveness. However, its impact can be indirectly estimated via international comparisons of measures of health spending and measures of health and care quality, which are the desired consequences of such spending. These measures, in turn, can be specifically selected and weighted for their relevance to employers and their workforces. Based on guidance from the expert panelists, Mercer pursued this approach.
The resulting Business Roundtable Health Care Value Comparability Study is composed of two performance domains: (1) health spending and (2) workforce health. Each domain is composed of measures that are readily available and periodically updated for both the United States and a group of five G-8 countries (the “G -5 group”). These five countries are major U.S. trading partners that, like the U.S., are affluent and benefit from mature, comprehensive health care systems: Canada, Japan, Germany, the United Kingdom and France. Recognizing that tomorrow’s U.S. competitors may include today’s most rapidly developing countries, we applied the HCVC Study method separately to a basket of three such countries (the “BIC group” – Brazil, India and China).
Composing the Health Spending Domain
For Business Roundtable’s purpose, Mercer sought an internationally comparable measure that would combine employer and employee per capita health care spending. At a minimum, we sought a measure that encompassed health insurance premiums or premium equivalents paid by employees and employers, out-of-pocket spending by employees, and tax payments by employers and employees used by the government to fund health benefit plans (e.g., Medicare, Medicaid and the VA) and health care (e.g. Federally Qualified Health Centers) for individuals other than active workers and their dependents enrolled in employer-sponsored health benefit plans. Because the data to construct this type of comprehensive measure is not available across all countries, we composed the domain from the two closest measures available. One of these measures is employer-paid health benefits paid per hour in manufacturing, the most export-exposed U.S. industry (Chart 1) on page 5.
Because the measure shown in Chart 1 fails to account for all types of employee and employer health care spending, the health spending domain also incorporates a second, broader spending measure, GDP-adjusted per capita health spending, shown in Chart 2 on page 6. Via a GDP adjustment, no country’s health care system is held accountable for the higher per capita health care spending associated with national prosperity. These figures are much lower than non-adjusted per capita health care spending figures because the amount of health spending that is attributable to GDP has been removed. The national per capita health care spending measure includes all streams of health care spending such as, direct employer and employee spending, as well as spending financed through tax payments by employees and employers. The data underlying this measure does not allow removal of out-of-pocket spending by patients who are not employees or dependents of employees.
An expert advisory panel of economists weighted these two spending measures based on their suitability as proxies for actual employer and employee total spending on health. The panel assigned the first measure an average weighting of 0.38 and the second measure an average weighting of 0.62, reflecting their view that national per capita health care spending more closely reflects what employers and workers actually spend. Since spending data for BIC countries is only available for the second measure, the BIC comparison with the United States is based solely on GDP-adjusted per capita health care spending.
While the expert panel judged this to be the best representation of employer and employee spending based on readily available data, it does not account for some health care system costs and some differences in spending patterns across countries. For example, it does not capture the estimated impact of each country’s general taxation level on its economic growth, or the impact of taxation to fund health spending specifically. It also does not account for the unfunded fiscal liabilities of future Medicare costs in the United States or their impact on U.S. interest rates and future U.S. tax rates. One limitation specific to Brazil, India and China is that while, on average, health spending figures for these countries are low, health spending by employers competing globally, such as car manufacturers, may be significantly higher. Finally, it does not account for the cost of “job-lock” from each country’s employee health benefits approach.
Composing the Workforce Health Domain
The workforce health domain of the HCVC Study consists of 17 employer-relevant national measures of health and health care. Examples of employer-relevant measures in the Workforce Health Domain include working-age adult mortality rate and health-related absenteeism in the workplace. The full set of component measures and computation methods are detailed in Appendix B. Though internationally comparable figures measuring workforce-specific health and health care are limited, the 17 measures represent the best of those available. When a country failed to report in a year, we projected the measure based on its prior trend. Each measure was weighted by an expert clinician-researcher team for its employer and worker relevance. The weighting criteria were (a) the contribution of a favorable score on the measure to worker productivity, and (b) the degree to which the health care system can affect the measure. The set of health measures available for the BIC basket (9 of the 17 measures) is smaller than the set of measures available for the G-5 basket.
Our approach to composing this domain has limitations. First, perceived health differs across countries and cultures; accordingly, the two perception-based measures may not reflect true differences in health. Second, the study’s measure of self-reported absence from work may partly reflect national sick leave policies rather than health status. Third, the study’s two measures of medical error are reported by health care providers and may be under-reported. Fourth, there is a risk of inaccuracy when measures for years when data are not reported are projected based on a measure’s preceding trend line. Finally, it is not possible to capture or adjust for non-health care determinants of health that may affect 14 of the study’s measures that reflect health status.
Converting Domain Scores Into Quantitative Comparisons of the U.S. With the Two Comparator Baskets
Scoring on each component measure in the health spending and workforce health domains consists of a simple ratio that compares performance of the U.S. with the G-5 group average and then separately with the BIC basket average. For each measure, we identify a benchmark value, defined as the most favorable average spending or health score between the two groups (either the United States or competitor basket). Using the numerically larger of the two values as a denominator and the smaller as the numerator, we derive a comparison percentage score for the less well-performing comparator and assign a 100% comparison score to the better performing comparator (see Appendix B for greater detail). Weights developed by the expert panel within each domain were then applied to each basket’s comparison score for each measure, and a “combined weighted comparison score” for each domain was calculated for the United States versus the G-5 and versus the BIC group.
Charts 3 and 4 on page 8 portray combined weighted comparison scores for the health spending and workforce health domains respectively for the three most recent years for which data is available. They confirm what prior studies have shown: while spending much more than either basket of competitors, the United States does not attain the best workforce health. G-5 employers and employees spend about 63% of what the U.S. spends on health care, but U.S. workforce health trails the G-5 by about 10%. The United States compares more favorably with BIC countries on workforce health measures, lagging by 5%. However, the BIC countries spend only 15% of what the United States is spending. The fact that the U.S. has a lower health score than the three BIC countries reflects two factors: (1) the limited number of workforce health measures on which to compare the United States with developing countries; and (2) the drastic degree by which the United States underperforms the BIC group on the obesity measure (prevalence of BMI >=25 kg/m2). This underscores the importance of Business Roundtable’s intent to use the new HCVC Study to track changes in relative value, rather than relative value per se.
Discerning Change in the HCVC Study
The U.S. workforce health domain score was then averaged with the spending domain score to create the HCVC Study versus the G-5 and versus BIC for the three most recent years for which comparative measures are available (Chart 5). The HCVC Study for the United States versus the G-5 is trending downward, at 78.1% for 2004, 77.5% for 2005, and 77.3% for 2006 The Business Roundtable Healthcare Value Comparability Study versus the BIC basket is stable, at 54.46% for 2004, 54.64% for 2005, and 54.39% for 2006.
We summarize recent annual change scores in the HCVC Study via Chart 6 on page 10. The annual change scores of The Business Roundtable Healthcare Value Comparability Study for the United States vs. the G-5 group are -0.55% from 2004 to 2005 and -0.28% from 2005 to 2006. This indicates that the contribution of the U.S. health care system to the global competitiveness of employers and workers relative to G-5 countries is declining, though at a decreasing rate. This trend is primarily due to U.S. spending figures increasing in comparison to G-5 spending, without corresponding increases in workforce health. The annual HCVC Study change scores for the United States vs. the BIC basket are +0.19% from 2004 to 2005 and -0.26% from 2005 to 2006, suggesting no consistent trend over the same time period.
Because annual studies of available measures are vulnerable to noise in measurement, future study releases will portray change scores over longer periods, in addition to the annual change score. Major movement in the HCVC requires fundamental U.S. health care industry changes, likely to take longer than a year to materialize and detect. HCVC Study annual change scores are best interpreted as an early signal of general direction that will sharpen with increases in measurement intervals, available measurements and quality control in international data collection.
Forecasting the Intermediate-Term Direction of the HCVC: The HCVC - Forecast Study
In addition to measuring the direction of the HCVC, it is useful to track a composite of current factors reflecting a national health industry’s innovations in value, since these may forecast the direction of the HCVC Study during the next 10 years. Mercer constructed a supplementary study, the HCVC-Forecast (HCVC-F) Study, to serve this purpose. Since the data reflecting such factors are not collected across countries, the HCVC-F Study applies to the United States only. Because we are interested in predicting future trends in the HCVC Study, we used the most recent data available for each measure, despite the drawback of combining data from inconsistent years. The reporting periods are labeled T1, T2 and T3 and the measurement year for each data point is detailed in Chart 7 on page 11. Measures comprising this study were weighted by a panel of experts in clinical performance improvement, as described on the next page Details of study construction, including specification of measurement sources, are available in Appendix D.
Measures in the HCVC-Forecast Study include: the Agency for Healthcare Research and Quality’s (AHRQ) improvement rate; the AHRQ trends in U.S. hospital efficiency; the Center for Disease Control and Prevention’s (CDC) measure of electronic medical record use in M.D. offices; the E-Health Initiative’s (EHI) measure of electronic clinical data flow within the Health Care Industry; the U.S. Patent Office’s annual domestic share of U.S.-issued health care patents; and the three-year moving average of approved new drug applications and biologic license applications by the F.D.A. Beginning with its inception in 2007, we also included the National Business Coalition on Health’s measure of the enrollment-weighted percent of health insurance plans that reward high quality and efficient providers via higher payments.
The HCVC-Forecast Study differs from the HCVC Study in three respects. First, it is not composed of current performance measures; rather it is composed of measures of predictors of improvement in health and/or health spending over the next 10 years. Second, since its component measures are not collected and reported internationally, comparison with the G-5 and BIC groups is not possible. Thus, the HCVC-F predicts health care value improvement in the United States, without regard to the rate that value may improve concurrently in other countries. A final difference is that, with one exception, it is populated by year-to-year percentage point changes in each predictor. In contrast, the HCVC Study is populated by measures of performance in a particular year, rather than by annual percentage point changes in measures. The HCVC-F includes one measure that is itself a change measure, the AHRQ Quality Improvement Rate. Accordingly, this measure is included directly in the HCVC, rather than converted to a measure of percentage point change. Change scores comprising the HCVC-F are then combined to form a weighted sum, using weights determined by the expert panel.
The expert panel assigned relative weights to each change score based upon its likely impact on health care value (increase in health and/or decrease in health spending) to U.S. employers and workers throughout the next 10 years. Change scores are assigned a positive value when the direction of change reflects improvement and a negative value when it reflects deterioration. The weighted change scores between measurement periods for each measure in the HCVC-F Study are shown in Chart 8.
Because a measure of the percent of health insurance plans that reward high quality and efficient providers via higher payments was only available for two reporting years, we list it separately on the bottom of Charts 7 and 8. Despite considerable enthusiasm among policymakers for more performance-sensitive payments to U.S. health care providers, on an enrollment-weighted basis, only 36% of U.S. commercial health insurers used payment methods that rewarded higher value hospitals and physicians in 2008.
Chart 9 shows that the weighted average change score using constant measures for the HCVC-F Study increased 0.61 percentage points over the last two reporting years, indicating a slight improvement in factors likely to raise the value of U.S. health care to U.S. employers and their workforces over the next 10 years. This reflects positive change in a number of measures, including electronic medical record use and electronic health data flow, but unfavorable change in the measures of new drug application and biologic license approvals from the F.D.A. Of note is a slowing in AHRQ’s rate of improvement score across multiple dimensions of the U.S. health care industry performance.
This small set of “leading indicators” is insufficient to populate a robust forecast. Mercer and Business Roundtable will continue to encourage other entities to make available for inclusion additional annually collected measures. Changes in two such measures will be available beginning in 2009 from the National Business Coalition on Health: the enrollment-weighted percentage of U.S. health insurance plans that offer incentives for enrollees to select high quality and efficient providers, and the percent of plans that offer incentives for enrollees to participate in health and care management activities. These measures and measures of their enrollment-weighted uptake by U.S. employers will be incorporated into the HCVC-F in 2009. Their inclusion reflects Business Roundtable’s view that improvement in U.S. health care value depends not just on better value stewardship by health industry leaders. It also depends on better health care value stewardship by employers, government purchasers and their insurers in both measuring health care value and incentivizing faster value improvement by U.S. clinicians and their suppliers.
For Appendix A-D, please view the PDF.
i Nichols, LM and S Axeen. “Employer Health Costs in a Global Economy: A Competitive Disadvantage for U.S. Firms.” New America Foundation, May 2008. Sources for these figures come from: International Social Security Association, Social Security Programs Throughout the World, 2005 and 2006 (for percent of payroll due to sickness and maternity benefits); U.S. Bureau of Labor Statistics, International Comparisons of Hourly Compensation Costs for Production Workers in Manufacturing, November 2006 (for hourly wage rates by country); 2008 Employer Costs for Employee Compensation (for U.S. employer contributions to health insurance in manufacturing).
ii Angrisano, C, Farrell, D., Kocher, B., Laboissiere, M., Parker, S. McKinsey Global Institute. “Accounting for the Cost of Health Care in the US.” January 2007.
iii The 2007 National Healthcare Quality Report can be found at: http://www.ahrq.gov/qual/nhqr07/nhqr07.pdf. The core measures are described on pages 14-16.
iv The measure is described on page 127 of the 2007 National Healthcare Quality Report. Hospital cost efficiency is examined using stochastic frontier analysis, which can account for differences among providers in outputs, case mix, or quality of care. This analysis uses data from the American Hospital Association Annual Survey, the Medicare Cost Reports, and quality indicators from the application of the AHRQ Quality Indicator software to data from the Healthcare Cost and Utilization Project. Here, the efficiency of a hospital is defined as the ratio of best-practice costs to total observed costs. For example, given the types and quantities of outputs a hospital produces, the input prices it pays, its case mix, its quality, and the characteristics of its market, a theoretical best-practice hospital might incur expenses amounting to $90 million. A comparison hospital in an identical situation with total expenses of $100 million would have an estimated cost efficiency of 90%. Cost efficiency estimates have been converted to study numbers with a base of 100 for the year 2000 as a way to place less emphasis on the specific magnitude of estimated cost efficiency than on its general trend.
v Hing ES, Burt CW, & Woodwell DA. “Electronic Medical Record Use by Office-Based Physicians and Their Practices: United States, 2006.” Advance Data From Vital and Health Statistics. CDC Number 393, October 26, 2007.
vi Survey results from e-Health Initiative can be found on the Web site: http://www.ehealthinitiative.org/HIESurvey/.
vii U.S. Patent Office Web site: http://www.uspto.gov/# and F.D.A. website: http://www.fda.gov/cder/rdmt/default.htm.