General Data Notes
To develop this information, SLHI engaged Mercer Government Human Services Consulting to create a healthcare database using data from a variety of sources, including national and regional surveys and actual claims data where available. All of the primary data sources used in creating these pages are known to be valid and reliable, and Mercer provided an additional level of expert review for reasonableness and suitability for use in tracking trends over time. However, as with any source of data, balancing the need to be timely and relevant with the need to be valid and credible can be a challenge. In addition, there is also the possibility of errors or unreported inconsistencies in the primary data sets. Wherever possible, multiple data sources have been reviewed for consistency and to ensure the accuracy of the numbers reported on these pages.
Because our goal is to provide a unique set of Arizona-specific information and trends, Mercer has used accepted actuarial modeling techniques to fill in gaps, make projections and address consistency among sources. These adjustments and projections are based on standard actuarial smoothing and projection techniques, which require the use of assumptions when only limited data is available. While every effort has been made to ensure that assumptions and projections are the most reasonable for the task at hand, due to the inherent variability in health care expenses and the potential for unforeseen external events to influence the healthcare delivery system in the future, it is very likely that actual experience will differ to some degree from these projections. We encourage you to make your own judgment about the extent to which the future will resemble the past. To assist you in that exercise, the following pages provide an explanation of the data source and underlying assumptions used in creating these pages.
Data Sources for Population Numbers
The population focus of these pages is Arizonans who are not eligible/enrolled in Medicare – essentially the “under-65″ population. In addition, dental, behavioral/mental health, long term care and other service costs that are generally not covered by primary medical insurance have been excluded from this study.
Population numbers for all primary market segments are derived from the United States Census Bureau Current Population Survey (CPS). Arizona-specific data was extracted from CPS as of November 2005, which contains statistics through calendar year 2004.
The CPS captures information about all types of health insurance coverage during the prior calendar year, according to the following six categories: 1) Employer, 2) Direct Purchase, 3) Medicare, 4) Medicaid, 5) Military, and 6) Uninsured. Since some individuals may have multiple sources of coverage, the following hierarchy was used to assign households to a primary insurance segment:
Sub-segmenting the Employer Category
Information about current public sector employees and their health insurance coverage is publicly available through their respective government Web sites. Data on the public sector insurance market segment was derived from the State of Arizona, the University system, the public school system, and the larger counties and cities. These figures were used to extrapolate the number of lives likely covered by public sector ESI, and then compared to CPS data about individuals employed in the public sector to ensure the reasonableness of the resulting figures. Historical information was extrapolated from the baseline data and assumes that growth in this sector will be proportionate to growth in the state population.
Adjusting for the Undercounting of Medicaid Households
American Enterprise Institute for Public Policy Research. Nine Million Fewer Uninsured?April 2005.
Hoffman, Catherine and John Holahan. What is the Current Population Survey Telling Us About the Number of Uninsured? Kaiser Commission on Medicaid and the Uninsured. August 2005.
Peterson, Chris L. and Grady, April. Medicaid/SCHIP Enrollees: Comparison of Counts from Administrative Data and Survey Estimates. Congressional Research Service. Updated March 30, 2005.
State Health Access Data Assistance Center (SHADAC), University of Minnesota School of Public Health. Do National Surveys Overestimate the Number of Uninsured? Findings from the Medicaid Undercount Experiment in Minnesota. January 2004.
Callahan, Cathi M. and James W. Mays. Working Paper: Estimating the Number of Individuals in the United States without Health Insurance. Actuarial Research Corporation. March 31, 2005.
Population projections for 2005 – 2010 assume that similar patterns observed in the recent past will continue in Arizona. Year-to-year trends in population often show some volatility that is likely influenced by the survey nature of the data source. Cyclical trends can also be observed in the data; for instance, Medicaid increases in the early years of this decade are likely driven by a slowed economy and the implementation of Proposition 204. Population projections are based on broader trend factors that smooth such year-to-year volatility and seasonal effects.
Data Sources and Notes for Healthcare Costs
The same general approach was used to estimate per capita costs for each of the private health insurance coverage sectors: Large Employers, Small Employers, Individual, and Government. Based on premium data and average benefit plan designs for each segment, out-of-pocket costs that would be incurred by a given population segment were estimated using proprietary actuarial models developed by Mercer.
Costs were allocated to each of the service categories based on findings from a claims data review. The data is from a large private claims database Mercer purchases, and represents 2002 service utilization for the Southwest Region. For years after 2002, the allocation is affected by the service-specific trend factors.
Medical cost projections are based on service-specific cost trends applied to the underlying medical costs. Medical cost projections are based on a variety of sources, including Mercer proprietary trend analyses, published research and analysis of historical trend data. Administrative costs and profits are trended through a formula that incorporates the Consumer Price Index (CPI) (administrative expenses) and trends specific to the medical industry (profit, commission, and contingency).
Out-of-pocket expenses are based on several components. First, co-payments associated with service utilization and medical cost inflation have an effect on out-of-pocket expenses, even in the absence of benefit design change. Second, historically, plan sponsors have reacted to premium increases by increasing cost sharing levels – higher deductible levels and copayments, for example. The effects of each of these trend components was incorporated into a model developed by Mercer to project out-of-pocket expense trends.
Government and Military: 2004 premium, plan benefit design, and contribution data for the State Employees health plan were used as a proxy for the health insurance costs of all public sector employers in the state. This data was then reviewed against TriCare cost data to ensure consistency.
Individual: 2004 premium, plan benefit design, and medical expense (medical loss ratio) data from rate filings filed with the Arizona Department of Insurance were analyzed. The 2004 documents include, and are also the source of data for historical periods.
Large Employers: Information about premium and benefit levels design for employers in the Southwest region with 500 or more employees was obtained from the Mercer National Survey of Employer-Sponsored Health Plans, 2001-2004. Data from this survey was compared to total premiums, relative employer/employee contributions to premium expense and benefit levels for Arizona large employer groups (50 or more employees) reported in the Medical Expenditure Panel Survey – Insurance Component (MEPS – IC), an annual survey conducted by the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services. http://www.meps.ahrq.gov/Data_Pub/IC_Tables.htm.
Medicaid: Estimated health care costs for the Medicaid population are derived from AHCCCS Acute Care financial data from the time periods 10/1/01-9/30/02 and 10/1/02-9/30/03. These data represent cost and utilization experience of multiple population groups, including TANF, SSI without dual Medicare Eligibility, MN/MI, MED, NON-MED, SOBRA FP, and SOBRA MOMS. Because these reports do not include cost data for institutionalized or dual Medicare eligible recipients, Mercer actuaries have made assumptions to estimate non-medical expenses and out-of-pocket costs associated with the AHCCCS population.
Medicare: Medicare recipients and health system costs associated with this population are not included in our analysis at this time, other than the projected enrollment growth in Medicare presented in the Arizona Population by Primary Market Segment chart. Arizonans that are enrolled in both Medicare and Medicaid (“Dual Eligibles”) have been classified as having coverage though Medicare and not through the Medicaid primary market segment.
Small Employers: Premium cost and employer/employee contribution data for Arizona small employer-sponsored plans was obtained from the annual Medical Expenditure Panel Survey – Insurance Component (MEPS-IC) conducted by the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services.http://www.ahrq.gov/data/mepsweb.htm. Out-of-pocket costs are based on a model developed by Mercer, utilizing average plan design and sales information from a statewide insurance brokerage specializing in the small group market. These estimates are based on claims experience of persons enrolled in PPO plans, which are the dominant product in the small employer market.
Uninsured: Per capita healthcare costs for the uninsured population are estimated from 2001 – 2002 Western Region medical expense data reported in the Medical Expenditure Panel Survey – Household Compendium of Tables by the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services.http://www.meps.ahrq.gov/CompendiumTables/TC_TOC.htm.
The values were adjusted to account for additional medical services and uncompensated care that are not captured by the MEPS-HC tables. These actuarial adjustments are based on previously published research and actuarial models used to estimate the medical costs of the uninsured.