Improving Census Bureau Estimates of Health Insurance Coverage

Contracting Office Address Department of Commerce, U. S. Census Bureau, Suitland, Acquisition Division Room 3J448-B Washington, DC, 20233 THIS IS AN INTENT TO SOLE SOURCE ANNOUNCEMENT PURSUANT TO FAR 6.302-1. THIS CONSTITUTES THE ONLY WRITTEN NOTICE, NO OTHER NOTICE OR SOLICITATION SHALL BE ISSUED AS A RESULT OF THIS NOTICE. The NAICS Code for this acquisition is 611310 Colleges, Universities and Professional Schools with a Size Standard of $7.0M. The potential contract value is $250,000.00. Background The U.S. Census Bureau's Current Population Survey (CPS) is the most commonly used source of data for understanding health insurance coverage and its social and economic covariates at the state level. CPS has been the benchmark survey for state policy analysis and evaluation of health insurance coverage for more than two decades. While the CPS has been the prominent source of health insurance coverage estimates for states over time, the sample size and collection of methodological changes over time have made the data difficult for policy analysts to use. The American Community Survey (ACS) began collecting health insurance information in 2008 and is a promising data source that adds many new dimensions to the picture of the uninsured in this country by providing local level health insurance data. This robust data source is critical to help states understand the impact of health reform laws such as the Children's Health Insurance Program Reauthorization Act (CHIPRA) in 2009 and the Patient Protection and Affordable Care Act (PPACA) of 2010. The ACS has begun producing health insurance coverage estimates for the country, states and local areas on an annual basis with the first data release in August 2009. With just one year of estimates to date, the ACS data are still new to the policy community and many questions need to be explored. The Census Bureau can best respond to these questions by further examining the data relative to other sources that provide health insurance coverage in order to respond to inquiries and provide a needed reporting mechanism for CHIPRA and PPACA. Due to burgeoning health reform efforts at the state level, the demands for credible data and effective translation of these data to the local policy context are particularly critical. States need a reliable, central resource for basic research, data dissemination and translation of the complex methodological issues underlying the CPS and the ACS. The State Health Data Assistance Center (SHADAC), under the Regents of the University of Minnesota contract YA1323-08-CN-0057, has played an important role in evaluating the first set of health insurance estimates produced from the ACS. SHADAC's evaluation of the ACS has been a viable source of health insurance estimates at the national and sub-national level. SHADAC also developed and recommended a set of edits that improved the data quality of the ACS estimates and also increased their utility as a vehicle for examining issues related to health insurance and health policies. New needs based on CHIPRA and PPACA point to the importance of utilizing the robust sample size of the ACS. States are not yet comfortable with this data source, and many do not have the capacity to analyze such a large data set. The Census Bureau sees a need to provide assistance in helping the states to access and understand the ACS estimates. The Census Bureau knows of only one source, SHADAC under the Regents of the University of Minnesota, that has the capabilities to facilitate the sharing of information among states; help policy makers understand the unique characteristics of the uninsured; and provide assistance to states estimating health insurance coverage rates at a state level. Objective The Census Bureau requires a contractor to perform services to assess the state-level health insurance estimates for 2008, 2009 and 2010. This work will focus on understanding discrepancies in reported coverage between the ACS and the other major sources of state-level estimates (for example, survey estimates from the CPS and National Health Interview Survey (NHIS)), and to provide technical direction for state health policy analysts as they utilize the ACS for populations covered or eligible for coverage under CHIPRA and PPACA. Task Requirements The contractor shall perform the following tasks: 1. ACS Mode Analysis: Expand the mode analysis already underway with the ACS Public Use Microdata Sample (PUMS) file. Mode analysis has been performed for other survey questions and it is important to have this studied and documented for health insurance coverage. This analysis will address some of the debate about the use of state-specific program names in the health insurance question and inform the direct purchase analysis described below in Section 2.0. Furthermore, the research will examine any bias that may be tied to mode that may be relevant to health policy analysis. Weighted and unweighted estimates by mode will be examined. The analysis will require the use of the internal ACS file at the contractor's site. 2. ACS Direct Purchase Analysis: Conduct analyses of direct health insurance purchase responses in the ACS using three approaches, as described below in 2.1-2.3. These three approaches will be used because ACS direct health insurance purchase prevalence appears high, but there is no gold standard for the "true" number of people with direct purchase insurance. Thus, varied approaches will be explored in order to search a wide array of possible explanations. 3. Characteristics: Examine the characteristics of people with direct purchase health insurance across datasets (ACS, CPS, and NHIS) for the nation. Consider state level analysis for a select group of states for which CPS and NHIS have sufficient samples (NHIS analysis would involve using internal data). The selection of states will be agreed upon by the Census Bureau's COTR and the contractor prior to this analysis. Examine direct health insurance purchase estimates among policy relevant subpopulations such as children, people in poverty, and minority groups. 4. ACS Specific State Level Analysis: Compare health insurance coverage estimates among states that have implemented major reform initiatives, such as an insurance exchanges (Massachusetts, for example) and Medicaid expansion programs (Wisconsin, for example). This analysis will focus on coverage estimates in states with low uninsurance rates, looking particularly at direct purchase estimates versus employer-sponsored and public program estimates. Where applicable, analysis will focus on subpopulations that have been directly targeted by policy initiatives, such as children and low-income populations to examine the impact of these initiatives on direct health insurance purchase reporting. 5. ACS Question Design: Consult with the Census Bureau and recommend a strategy to test changes to ACS health insurance questions in light of the direct purchase evaluation and implementation of federal health reform by the states. The changes to ACS health insurance questions to consider may include moving direct purchase to the bottom of the question list, adding qualifiers to the question sub-parts, and adding an item or altering text to clarify purchase through an exchange. 6. Coverage of Young Adults: Analyze state-level health insurance coverage estimates for young adults in both the ACS and CPS to establish baseline information (establish current health insurance estimates prior to enactment of changes so the impact of changes will be measureable) in light of the expansion of dependent coverage due to the implementation of federal health reform. Investigate how employers provided coverage changes for this group over time; comparing 2008, 2009 and 2010 estimates. 7. Coverage Error: Rates of health insurance coverage vary by geography, in particular urban versus rural areas. One possible reason for discrepancies between surveys is that a survey sample may not be exactly representative of the U.S. population, either for the whole country or a particular area within the U.S. This analysis will investigate the possibility that biases in the ACS or CPS sample affects health insurance coverage estimates and, if evidence supports this, how children and low-income subpopulation estimates are impacted. Measurement of students residing in college dormitories will also be investigated. The analysis plan will restrict the ACS to the sampling geography used by the CPS and utilize both descriptive and controlled comparison of estimates, including a specific assessment of direct purchase. This is an important unanswered question for researchers and health policy analysts to provide an understanding of differences between the CPS and ACS. 8. Linking Project: ACS-Medicaid Statistical Information System (MSIS): Recently a set of papers were released on the Census Bureau's website that discussed Medicaid under-reporting by examining survey data matched to Medicaid administrative data (MSIS). This work was the result of a partnership between SHADAC and several government agencies, and was known as "SNACC" based on the first initials of the participating organizations (for more information, see http://www.census.gov/did/www/snacc/ ). This analysis has not been expanded thus far to the ACS. The contractor shall work with the Census Bureau to expand this analysis to the ACS by assisting in the determination of which SNACC tables should be produced based on the ACS, consider additional analysis such as studying partial benefits, perform an analysis of the linked data, and assist with writing a final report. 9. Time Series Harmonization. Through the University of Minnesota's Integrated Public Use Microdata Series (IPUMS) website, SHADAC has produced a series of CPS datasets that are much more harmonized (consistent over time) than those available through the Census Bureau as described below in Sections 10 through 12. 10. ACS: SHADAC has worked with the Census Bureau to develop a series of ACS coverage edits that were incorporated into the 2009 ACS. However, the 2008 ACS PUMS was released prior to the incorporation of these edits, so ACS public file users cannot take advantage of these improvements. The contractor shall apply these coverage edits to the 2008 ACS PUMS file. This revised file will be distributed through the IPUMS website. These data and estimates will provide researchers with a consistent time series for analysis because the Census Bureau will not re-release the 2008 data with the changes. Census Bureau's work with the states has found that consistent trend data are very important for policy decisions and reporting requirements. The contractor shall also produce a short working paper explaining the new coverage edits applied to the 2009 PUMS file and how data users can undo the coverage edits if pre-coverage edit data are needed for specific research. 11. CPS: The Census Bureau will be implementing CPS edit improvements in 2011. These improvements were based on the results of the analysis conducted by SHADAC and Census Bureau. These CPS edit changes will be incorporated into the enhanced time series developed by the contractor and disseminated through the IPUMS website. 12. ACS Health Insurance Unit: Develop a policy-relevant Health Insurance Unit (HIU) variable for data users and researchers to understand health insurance policy and how they affect families. A consistent and well-documented method for developing a HIU will be beneficial for public data users and will promote comparative analysis in the field of health services research. There are different methods to create a HIU and thus are proposing developing one consistent and well-documented method that will help analysts better understand the potential impact of health insurance policy changes on families and individuals. Place of Performance The place of performance shall be at the contractor's site. The period of performance shall consist of a Base Period and four (4) Option Periods. The government anticipates awarding a Time and Materials contract using Part 15 Negotiated Competitive Procurement procedures. THE GOVERNMENT KNOWS OF ONLY ONE SOURCE: Regents of the University of Minnesota, 200 Oak Street, SE, Minneapolis, MN 55455-2009. Interested parties may submit responses that demonstrate its ability to provide services in accordance with this announcement, which will be considered within 15 days of the posting of this notice. As a result of analyzing responses to this synopsis of intent to sole source, the Contracting Officer may determine that the action may be competed. A determination by the Government not to compete this requirement based upon responses to this notice is solely within the discretion of the Government. Information received will be considered solely for the purpose of determining whether to conduct a competitive procurement. For questions concerning this procurement, please contact Yolanda J. Mitchell, Contract Specialist, at +1 (555) 000-0000.