Practices eligible for the MOS survey were those identified as an office–based provider by a respondent in the MEPS HC and identified by that respondent as his or her USC with whom they had a visit in 2016. The 2016 data year MOS survey was fielded in March 2017 through October 2017.
The purpose of the MOS instrument was to collect information not captured in the MPC concerning the organization of medical practices and resources available to practices identified as a household person's USC. The MOS was designed to complement MPC data with additional information on the characteristics of providers. The MOS collected data in the following topical areas: organization characteristics, health information technology, case management and use of clinical data, and financial arrangements. MOS data can be linked to MEPS sample respondents in the 2016 Full Year Consolidated Data File to enable analyses at the person level using characteristics of provider practices.
After contacting the selected office–based providers, MOS DCSs introduced the MOS survey and identified an appropriate MOS respondent. Although in some instances the MOS respondent was the MPC point of contact (POC), the data collection protocols anticipated scenarios in which the MOS respondent would be a different person than the MPC respondent and unfamiliar with the MPC or its history in that practice. Materials were developed to introduce the MOS to either the MPC POC or a different POC for the MOS. The MOS questionnaire was designed to be completed by either telephone, online, or by mailing or faxing a completed questionnaire.
Table 2 presents provider level sample sizes and completion rates by provider type, while Table 3 presents pair level sample sizes and completions rates by provider type. These two levels measure different concepts. The provider completion rate is a measure of provider cooperation. The pair completion rate is a measure of pair coverage and is used to determine how many pairs complementary provider data is obtained.
For data collection purposes, a provider was classified as a participant if health care event data were obtained for any of the provider's sampled patients during calendar year 2016. Each event reported by the provider was required to have, at a minimum, a date of service and a procedure or service. Individual providers were identified by a unique provider ID. As Table 2 illustrates, 86 percent of the 6,170 eligible hospital providers and about 87 percent of the 12,903 eligible office–based providers participated in the survey. Furthermore, 83 percent of the 323 eligible HMOs, which can be a hospital or an office–based provider, participated in the survey. Fifty–five percent of SBD providers, 91 percent of pharmacy providers, and 85 percent of home care providers also participated in the survey. Overall refusal rates ranged from 0 percent for HMOs, institutions, and pharmacy providers to 3.6 percent for SBD providers. Refusals are categorized based on speaking with the POCs and having a definitive response on their desire to not participate. Providers classified as "other nonresponse" were primarily those who could not be located, who could be located, but had no record of providing care to the sampled patient, or no definitive contact was made before the end of the data collection period. Overall, non–participation rates ranged from 9 percent for pharmacies and institutions to 42 percent for SBDs. Note that when comparing the original sample with the final eligible sample, the original sample includes providers that were later determined to be out of scope for the MEPS MPC, or providers that eventually were merged or split.
Table 3 presents sample sizes and completion rates by provider type at the pair level. Completion rates at the pair level were mostly similar to or lower than the provider level: 85 percent for hospitals, 86 percent for office–based physicians, 77 percent for HMOs, 54 percent for SBDs, 85 percent for pharmacies, and 84 percent for home care agencies.
Completion rate calculations include completes and partial completes. Completes include no events within the provider/pair that are missing critical items and partial completes include providers/pairs for which at least one event is missing a critical item and at least one event is not missing any critical items.
Table 4 presents sample sizes and completion rates for the MOS at the practice and pair level. At the practice level, the MOS completion rate was 76 percent. At the pair level, the completion rate was 77 percent.
Over the last decade, there have been significant developments in the United States health care delivery system, including the consolidation of major health systems and increased use of electronic health records. These developments have affected the way in which health care information is recorded, maintained, and released. This section describes how the MPC has evolved to respond to these developments, while maintaining the ability to compare data year to year.
Adapting the MPC instruments to CAI (Computer-Assisted Interviewing) from PAPI (Paper-and-Pencil Interviewing)
During the 2009 data year collection cycle, a computer–assisted interviewing system was developed by the MEPS MPC team for both interviewing and record abstraction, which was a change from the previous MEPS MPC paper–and–pencil interviewing methodology. This integrated data collection system is flexible and supports the effort to recruit providers by telephone and to interview medical records and billing staffs of medical facilities or providers. The system is also used to abstract information from medical records and patient accounts from providers that send hardcopy records. Data for all provider types are collected in this electronic data collection system, which was used and refined during the 2016 data year collection cycle. The integrated system also produces reports for the MEPS MPC team and develops data files for the matching process for developing medical expenditure estimates.
Since the MEPS was designed in the mid–1990s, health care delivery in the United States has evolved in many ways. One change has had a large impact on MPC operations: the trend toward consolidation of health care facilities. This has occurred both by individual physicians moving into group practices (Welch et al., 2013) and by independent hospitals being acquired by health care systems (Tsai & Jha, 2014). There has also been consolidation among group practices as well as hospital systems acquiring physician group practices. Consolidation can complicate the identification of an appropriate POC within the provider facility. In the MPC sample design, the frame of events, the basic data collection unit, is nested within a hierarchy of patient–provider pairs and contact groups.
The source of the event level use data–the household respondent–is generally accurate at distinguishing a provider as an office–based physician or a hospital. However, the contact information supplied often includes the name of the physician but may not include the group's practice name, name of the hospital, or the name of the health care system that owns the hospital. Poor information about the linkages among providers and their institutional affiliations can result in errors as providers are sorted into contact groups for the purposes of MPC data collection. Errors can result in multiple contacts to the same contact group because it was not identified that two physicians belonged to the same group. Errors can also result from "overgrouping," where physicians are mistakenly added to a contact group. Overgrouping can result in erroneous disavowals.
These experiences have led the MPC team to be careful about the sorting of events and pairs into contact groups for initial contacts to physician offices, hospitals, and health care systems. A stage of this sorting occurs during sample preparation. The MPC team has also introduced various automated approaches to matching providers and improving accuracy in the formation of contact groups. Adoption of the National Provider Identifier (NPI) in the 2010 MPC has helped to mitigate the problem, but has not completely solved it. A single provider may bill through any of several NPIs and provider facilities often have separate NPIs for different units within the facility. And, despite the best efforts of the field interviewers in the HC, a portion of providers identified by household respondents are not matched to NPIs in the directory available on the field laptops. To address this situation, the MPC team has developed matching algorithms based on address, telephone numbers, and other data elements that might indicate that providers belong in the same contact group.
In addition to accepting hardcopy records, the MEPS MPC now offers several electronic options to POCs to reduce the burden of data transfers. The MPC provides the option to deliver data via CDs containing PDFs of records. Other data transfer choices have also been made available: During the 2013 cycle, the MEPS MPC team developed documented procedures and instructions for providers to use a secure File Transfer Protocol (FTP) site and secure email. During the 2016 data year collection cycle, we continued to offer both FTP and secure email routinely to our largest providers. Requests from other POCs for FTP or secure email were also accommodated.
The MEPS MPC also accepts electronic data files that contain the requested data fields. In the 2009 through 2014 data year collections, two pharmacy contact groups provided data electronically, and one hospital/office–based provider POC provided electronic records. With the more widespread adoption of electronic health records, we anticipate the frequency of submission of electronic data files to increase and expand to other provider types in future years of the MPC. For the 2016 data year collection cycle, the MEPS MPC developed an application to allow POCs to securely download Authorization Form packets from and upload records to a MEPS MPC website.
Key changes to the MOS from the first data collection completed during the 2015 data collection cycle compared to the 2016 data collection cycle are described below.
Staffing and training of the MOS DCSs differed during the 2016 data collection when compared to the 2015 data collection. During the 2015 data collection, MOS was a new survey for all MPC DCSs; during the 2016 data collection, veteran MOS staff from the previous data collection were available to work during the new cycle. A more streamlined training was able to focus on instrument changes and allowed for a quicker data collection start. Throughout data collection, staff working on office–based providers were identified to replenish MOS staffing levels as needed; the connections between office–based providers and MOS data collection made this a natural transition for many staff.
Based on review of the data from the 2015 data collection, AHRQ revised the order of some of the MOS survey questions before the start of the 2016 cycle. However, minimal wording changes were made to the questions, so that consistency could be maintained between the two data collection cycles.
Data obtained from providers in the MPC are critical to the development of MEPS expenditure estimates. Since insurance providers negotiate reimbursement rates with providers that are significantly lower than the total established list price that appears on a billing statement, expenditures in the MEPS are defined as the sum of payments from all payer sources including out–of–pocket payments. While most household respondents can report with some degree of accuracy how much they pay out of pocket for medical care, including prescribed medicines, they do not always know the total payments made on their behalf by third–party payers. Providers generally have more complete information on reimbursement arrangements (capitation versus fee–for–service) and how much was paid by payer source for care delivered to household respondents.
In general, the methodology used to develop 2016 MEPS medical expenditure estimates was based on merging medical events reported in the HC with data from the MPC using a probabilistic matching procedure. For all medical events except prescribed medicines, the first stage of the MEPS expenditure estimation methodology involved matching the provider–reported expenditure data in the MPC to the household–reported medical event in the HC using dates and detailed information on conditions and procedures that were collected in both the HC and MPC. Because of the matching, each medical event in the HC either had expenditure data from both the MPC and HC, from the MPC or HC only, or from neither source (i.e., missing payment data). Household–reported event type may be different from provider–reported event type, so some cross–event type matches were permitted. However, "event type" on the final files released to the public was defined according to the HC classification. A hierarchical approach was used to develop complete expenditure data. When a match was found for a particular medical event, expenditure data from the MPC were substituted for household–reported information. In certain cases, if MPC data were not available, and complete household payment data were available, the household data were used. A series of logical edits were applied to both the HC and MPC data to correct for several issues with the reported data (e.g., outliers, misreported data). For more information on the types of issues encountered and the logical edits applied, see "Section 2.5.11 – Utilization, Expenditures and Source of Payment Variables" of the documentation for the 2016 Full Year Consolidated Data File (HC–192) at the following link: https://meps.ahrq.gov/mepsweb/data_stats/download_data_files_detail.jsp?cboPufNumber=HC-192.
Where MPC or HC payment data were not available, either due to item nonresponse or because the payment was not tied to a specific visit (as in capitation arrangements), payment components (e.g., private insurance, Medicare, Medicaid and/or out–of–pocket payments) were imputed using a predictive mean matching procedure. This procedure uses regression models (based predominantly on events with completely reported expenditure data assigned to the donor pool) to predict total expenses for each event. Then, for each event with missing payment information, a donor event with the closest predicted payment having the same pattern of expected payment sources as the event with missing payment was used to impute the missing payment component values. Data collected in the MPC were used as donors whenever possible when imputing expenditures because those data are generally regarded as more accurate than information collected in the HC.
For prescribed medicines, the general approach to matching data was to merge drug information collected from MPC pharmacy providers to HC drug data. To improve the quality of matches, all HC drugs were assigned codes (Generic Product Identifier or GPI) from a proprietary database (the Master Drug Data Base or MDDB) based on the medication names provided by the household. These codes were also assigned to the prescriptions in the MPC by using the NDC (when available) or medication names reported by the pharmacy providers. Software was developed that merged MPC drug data to the HC drug data by matching drug events from each file on a variety of characteristics (e.g., person id, GPI, potential payment sources, age, sex, health status, geographic location, and medication names). Selected editing was done prior to the matching to correct data inconsistencies in both data sets and to fill in missing data and correct outliers on the MPC file. For HC events that did not have corresponding data in the MPC, the MPC data were used as the imputation source.
The MEPS MPC is a voluntary survey designed to supplement and validate health care expenditure and source of payment data collected in the MEPS HC. The MEPS MOS is also a voluntary survey designed to expand current MPC data collection activities to include information on the organization of the practices of office–based care providers identified as a USC in the MEPS HC and having been seen by the HC respondent in 2016. This report, an update and enhancement to MEPS Methodology Report #23: Design, Methods, and Field Results of the Medical Expenditure Panel Survey Medical Provider Component–2006 Calendar Year Data, focuses on the MPC data collection effort that collected information from medical providers on health care events that took place between January 1, 2016, and December 31, 2016. The report describes the MPC sample design, survey methodology, procedures for data collection, sample sizes, and response rates as well as the relationship between the HC survey and the MPC survey. Additionally, it includes information on the MOS survey which was new in the 2015 MPC data year collection and was fielded again in the 2016 MPC. Both the 2015 and 2016 MOS were funded in part by the Robert Wood Johnson Foundation.
Additional information on MEPS is available from the MEPS Project Director at the Center for Financing Access and Cost Trends, Agency for Healthcare Research and Quality, 5600 Fishers Lane, Mailstop 07W41A, Rockville, MD 20857; MEPSProjectDirector@ahrq.hhs.gov.
Cohen, J. Design and Methods of the Medical Expenditure Panel Survey Household Component. MEPS Methodology Report No. 1. AHCPR Pub. No. 97–0026. Rockville, MD: Agency for Health Care Policy and Research, 1997. https://meps.ahrq.gov/data_files/publications/mr1/mr1.shtml
Cohen, S. Sample Design of the 1996 Medical Expenditure Panel Survey Household Component. MEPS Methodology Report No. 2. AHCPR Pub. No. 97–0027. Rockville, MD: Agency for Health Care Policy and Research, 1997. https://meps.ahrq.gov/data_files/publications/mr2/mr2.shtml
Cohen, S. Design Strategies and Innovations in the Medical Expenditure Panel Survey. Medical Care. July 2003: 41(7) Supplement: III–5–III–12.
Ezzati–Rice, T.M., Rohde, F., Greenblatt, J. Sample Design of the Medical Expenditure Panel Survey Household Component, 1998–2007. Methodology Report No. 22. March 2008. Agency for Healthcare Research and Quality, Rockville, MD. https://meps.ahrq.gov/data_files/publications/mr22/mr22.shtml
Machlin, S.R., Dougherty, D.D. Overview of Methodology for Imputing Missing Expenditure Data in the Medical Expenditure Panel Survey. Methodology Report No. 19. March 2007. Agency for Healthcare Research and Quality, Rockville, MD. https://meps.ahrq.gov/data_files/publications/mr19/mr19.shtml
Machlin, S.R., Taylor, A.K. Design, Methods, and Field Results of the 1996 Medical Expenditure Panel Survey Medical Provider Component. Methodology Report No. 9. May 2000. Agency for Healthcare Research and Quality, Rockville, MD. https://meps.ahrq.gov/data_files/publications/mr9/mr9.shtml
Hill, S.C., Roemer, M., Stagnetti, M.N. Outpatient Prescription Drugs: Data Collection and Editing in the 2011 Medical Expenditure Survey. MEPS Methodology Report No. 29. March 2014. Agency for Healthcare Research and Quality, Rockville, MD. https://meps.ahrq.gov/data_files/publications/mr29/mr29.shtml
Stagnitti, Marie N., Beauregard, Karen, and Solis, Amy. Design, Methods, and Field Results of the Medical Expenditure Panel Survey Medical Provider Component (MEPS MPC)–2006 Calendar Year Data. Methodology Report No. 23. November 2008. Agency for Healthcare Research and Quality, Rockville, MD. https://meps.ahrq.gov/data_files/publications/mr23/mr23.shtml
Tsai T.C., Jha A.K. Hospital consolidation, competition, and quality: is bigger necessarily better? Journal of the American Medical Association. 2014;312(1):29–30.
Welch, W.P., Cueller, A.E., Stearns, S.S., Bindman, A.B. Proportion of physicians in large group practices continued to grow in 2009–11. Health Affairs. 2013;32(9):1659–1666.
Zodet, M., S. Chowdhury, S. Machlin, and J. Cohen. 2016. Linked designs of the MEPS Medical Provider and Organization Surveys. In JSM Proceedings, Survey Research Methods Section. Alexandria, VA: American Statistical Association. 1914–1921.
The Medical Expenditure Panel Survey (MEPS) provides nationally representative estimates of health care use, expenditures, sources of payment, and health insurance coverage for the U.S. civilian noninstitutionalized population. The MEPS Household Component (HC) also provides estimates of respondents' health status, demographic and socioeconomic characteristics, employment, access to care, and satisfaction with health care. Estimates can be produced for individuals, families, and selected population subgroups. The panel design of the survey, which includes five rounds of interviews covering two full calendar years, provides data for examining person level changes in selected variables such as expenditures, health insurance coverage, and health status. Using computer assisted personal interviewing (CAPI) technology, information about each household member is collected, and the survey builds on this information from interview to interview. All data for a sampled household are reported by a single household respondent.
The MEPS HC was initiated in 1996. Each year a new panel of sample households is selected. Because the data collected are comparable to those from earlier medical expenditure surveys conducted in 1977 and 1987, it is possible to analyze long–term trends. Each annual MEPS HC sample size is about 14,000 households. Data can be analyzed at either the person or event level. Data must be weighted to produce national estimates.
The set of households selected for each panel of the MEPS HC is a subsample of households participating in the previous year's National Health Interview Survey (NHIS) conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention. The NHIS sampling frame provides a nationally representative sample of the U.S. civilian noninstitutionalized population and reflects an oversample of blacks, Hispanics and, starting in 2006, Asians. The linkage of the MEPS to the previous year's NHIS provides additional data for longitudinal analytic purposes.
Medical Provider Component
Upon completion of the household CAPI interview, and after obtaining permission from the household survey participants, a sample of medical providers are contacted by telephone to obtain information on medical events that household respondents cannot accurately provide. This part of the MEPS is called the Medical Provider Component (MPC) and information is collected on dates of visits, diagnosis and procedure codes, charges, and payments. The Pharmacy Component (PC), a subcomponent of the MPC, does not collect charges or diagnosis and procedure codes but does collect detailed information on prescriptions, including National Drug Code (NDC) and medicine name, as well as date(s) prescriptions were filled and sources and amounts of payment. The MPC is not designed to yield national estimates. It is primarily used as an imputation source to supplement and/or replace household–reported expenditure information.
Medical Organizations Survey
The Robert Wood Johnson Foundation provided support and funding to the Agency for Healthcare Research and Quality (AHRQ) to conduct a new pilot survey of health care providers called the Medical Organizations Survey (MOS) in 2015 and 2016. Designed as an add–on to the MPC, the MOS is intended to collect supplemental information on usual source of care (USC) practice characteristics for MEPS sample persons who saw their USC during the year. The MOS data collection is for a subset of office–based care providers already included in the MPC sample. The design of the MOS is multi–modal including phone, fax, mail, self–administration, electronic transmission, and secure email. The MEPS MOS data file includes a population weight from which national estimates can be derived.
MEPS HC and MPC data are collected under the authority of the Public Health Service Act. Data are collected under contract with Westat (HC) and RTI International (MPC). Data sets and summary statistics are edited and published in accordance with the confidentiality provisions of the Public Health Service Act and the Privacy Act. The National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention provides consultation and technical assistance related to the selection of the MEPS household sample.
As soon as data collection and editing are completed, the MEPS survey data are released to the public in staged releases of summary reports, micro data files, and tables via the MEPS Web site at www.meps.ahrq.gov. Selected data can be analyzed through a summary tables app and through MEPSnet, a set of online interactive tools designed to give data users the capability to statistically analyze MEPS data in a menu–driven environment.
1 A person/provider pair (or pair) represents one sample person visiting a health care provider during the year. A health care provider can be visited by more than one sample person during the year, resulting in many person/provider pairs for one provider. 2 An SBD node is a unique combination of the following data elements which are specific to the original hospital or institutional medical event: patient name, SBD provider name, dates(s) of service, location of service (ER, inpatient, or outpatient), and the hospital or institution at which the services were delivered.