October 2017
Agency for Healthcare Research and Quality
Center for Financing, Access, and Cost Trends
5600 Fishers Lane
Rockville, MD 20857
(301) 427-1406
TABLE OF CONTENTS
A. Data Use Agreement
B. Background
1.0 Household Component
2.0 Medical Provider Component
3.0 Survey Management
C. Technical and Programming Information
1.0 General Information
2.0 Data File Information
3.0 Linking Instructions
4.0 Other Considerations
5.0 Further Information
Individual identifiers have been removed from the micro-data contained in these files. Nevertheless,
under sections 308 (d) and 903 (c) of the Public Health Service Act (42 U.S.C. 242m and 42 U.S.C. 299 a-1), data collected
by the Agency for Healthcare Research and Quality (AHRQ) and/or the National Center for Health Statistics (NCHS) may not
be used for any purpose other than for the purpose for which they were supplied; any effort to determine the identity of
any reported cases is prohibited by law.
Therefore in accordance with the above referenced Federal Statute, it is understood that:
- No one is to use the data in this data set in any way except for statistical reporting
and analysis; and
- If the identity of any person or establishment should be discovered inadvertently, then
(a) no use will be made of this knowledge, (b) the Director Office of Management AHRQ will be advised of this
incident, (c) the information that would identify any individual or establishment will be safeguarded or destroyed,
as requested by AHRQ, and (d) no one else will be informed of the discovered identity; and
- No one will attempt to link this data set with individually identifiable records from any
data sets other than the Medical Expenditure Panel Survey or the National Health Interview Survey.
By using these data you signify your agreement to comply with the above stated statutorily based
requirements with the knowledge that deliberately making a false statement in any matter within the jurisdiction of any
department or agency of the Federal Government violates Title 18 part 1 Chapter 47 Section 1001 and is punishable by a
fine of up to $10,000 or up to 5 years in prison.
The Agency for Healthcare Research and Quality requests that users cite AHRQ and the Medical
Expenditure Panel Survey as the data source in any publications or research based upon these data.
Return to Table of Contents
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
socio-economic 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 5 Rounds of
interviews covering 2 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 15,000 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.
The NHIS sampling frame provides a nationally representative sample of the U.S. civilian noninstitutionalized population
and reflects an oversample of blacks and Hispanics. In 2006, the NHIS implemented a new sample design, which included
Asian persons in addition to households with black and Hispanic persons in the oversampling of minority populations. MEPS
further oversamples additional policy relevant subgroups such as low income households. The linkage of the MEPS to the
previous year’s NHIS provides additional data for longitudinal analytic purposes.
Return to Table of Contents
Upon completion of the household CAPI interview and obtaining permission from the household survey
respondents, a sample of medical providers are contacted by telephone to obtain information that household respondents
can not accurately provide. This part of the MEPS is called the Medical Provider Component (MPC) and information is
collected on dates of visit, 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 drug detail
information, including National Drug Code (NDC) and medicine name, as well as date 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/replace household reported expenditure information.
Return to Table of Contents
MEPS HC and MPC data are collected under the authority of the Public Health Service Act. Data are
collected under contract with Westat, Inc. (MEPS HC) and Research Triangle Institute (MEPS 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) provides consultation and technical assistance.
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:
www.meps.ahrq.gov. Selected data can be analyzed through MEPSnet, an online
interactive tool designed to give data users the capability to statistically analyze MEPS data in a menu-driven environment.
Additional information on MEPS is available from the MEPS project manager or the MEPS public use
data manager at the Center for Financing Access and Cost Trends, Agency for Healthcare Research and Quality, 5600 Fishers
Lane, Rockville, MD 20857 (Ph: 301-427-1406).
Return to Table of Contents
Due to the complex survey design of the MEPS‑HC, special methods must be used to calculate the
standard errors of MEPS‑HC estimates. To facilitate the calculation of design-based standard errors, MEPS‑HC
annual public use datasets contain stratum and PSU variables which can be utilized by the survey procedures that implement
the Taylor series linearization method of variance estimation. There is also a public use file, HC‑036, which provides
a standardized set of pooled linkage variance estimation units over all years of MEPS‑HC so that estimates can be made
with datasets created by pooling over multiple years of annual MEPS‑HC data.
Although useful, the linearization method is limited in the number of survey estimators for which
variances can be calculated, including population totals, simple proportions and regression parameters. It is not possible,
for instance, to calculate the variances of a median or the ratio between two medians using a Taylor series expansion.
For these types of estimators, users may calculate a proper design based standard error using either the PSU bootstrap
or balanced repeated replication (BRR) method of variance estimation.
This dataset, HC‑036BRR, contains the information necessary to construct the BRR replicate samples
that are necessary to calculate the BRR variances. It contains the unique person level identifier (DUPERSID and PANEL)
of every MEPS respondent appearing in any of the 1996‑2015 annual full year samples: HC‑012 (1996), HC‑020 (1997),
HC‑028 (1998), HC‑038 (1999), HC‑050 (2000), HC‑060 (2001), HC‑070 (2002), HC‑079 (2003),
HC‑089 (2004), HC‑097 (2005), HC‑105 (2006), HC‑113 (2007), HC‑121 (2008), HC‑129 (2009),
HC‑138(2010), HC‑147 (2011), HC‑155 (2012), HC‑163 (2013), HC‑171 (2014), and HC‑181 (2015).
It also contains a set of 128 flags (BRR1—BRR128), each of which is
coded 0 or 1 to indicate whether the person should or should not be included in that particular replicate sample. These
flags should be used in conjunction with the sample weights from the full year sample files to construct the BRR replicate
weights needed to calculate BRR variances.
Return to Table of Contents
Released as an ASCII data file (with SAS®, STATA®, and SPSS® user statements) and in SAS Transport
version, the HC‑36BRR file contains 354,558 records. These records contain the standard MEPS person‑level ID
variables (DUID, PID, DUPERSID and PANEL), as well as the 1996‑2015 replicate indicator variables BRR1‑BRR128.
There is a record for each person who appears on any of the 1996‑2015 MEPS full‑year
person level public use files: HC‑012, HC‑020, HC‑028, HC‑038, HC‑050, HC‑060,
HC‑070, HC‑079, HC‑089, HC‑097, HC‑105, HC‑113, HC‑121, HC‑129, HC‑138,
HC‑147, HC‑155, HC‑163, HC‑171, and HC‑181. These twenty datasets have a combined total of 655,643 records.
However, because each person may appear in one or two of these datasets the number of unique persons (354,558) is fewer
than the combined total number of records on the annual files.
Return to Table of Contents
The following steps should be taken to create a file containing persons from any one or more of the
twenty years of MEPS HC data.
- Create a dataset for each year containing the person and/or event level records of all
persons to be included in the analysis. Keep the unique person identifier (DUPERSID and PANEL), the person level
sampling weight, any classification variables (e.g., sex, race/ethnicity) and response variables (e.g., total
expenditure amount, number of prescription drug purchases, etc) to be used in the data analysis.
- Reconcile the discrepancies in variable names. For all years, most variable names on the
annual public use files contain a 2-digit year suffix. For instance, in the 1997 consolidated person level
file (HC‑020) the panel variable is called PANEL97, the total annual expenditure amount variable is called
TOTEXP97 and the sampling weight variable is called WTDPER97. But in the 2003 dataset (HC‑079) these same
variables are named PANEL03, TOTEXP03 and PERWT03F, respectively, and in the 1996 dataset (HC‑012) the total
expenditure and sampling weight variables are named TOTEXP96 and WTDPER96, respectively, and the panel variable
is missing (users should assign a value of 1 for each record in HC‑012). As illustrated below, the variable
names must be made consistent before pooling the data. Note: starting in 2005, the panel variable is called
simply PANEL (no year suffix).
![Reconciling the discrepancies in variable names before pooling the
data as follow: 1996 (HC‑012) - calculated PANEL, WTDPER96, and TOTEXP96; 1997 (HC‑020) - PANEL97, WTDPER97
and TOTEXP97; 2003 (HC‑079) - PANEL03, PERWT03F and TOTEXP03; to variable names that are consistent - PANEL,
PERWT and TOTEXP, respectively.](../equations/images/linkvariables.gif)
- Create a pooled analysis dataset by combining the individual year datasets by row;
that is, append the records from the 1996 dataset with those from the 1997 and 2003 datasets.
- Attach the BRR replicate flags to the pooled analysis dataset by column; that is, merge
the variables BRR1‑BRR128 from this HC‑036BRR file to the pooled analysis dataset by DUPERSID and
PANEL keeping all records in the pooled analysis dataset and only those records in HC‑036BRR dataset
that match. Depending on the software being used to manage the datasets, the pooled analysis dataset may need
to be sorted by DUPERSID and PANEL prior to merging.
- To calculate a standard set of 128 BRR replicates, multiply each BRR replicate flag by
2 and by the sample weight (PERWT, if using the example above). That is, BRR1wt = BRR1 * 2 * PERWT and BRR2wt =
BRR2 * 2 * PERWT, …, BRR128wt = BRR128 * 2 * PERWT. This method creates a set of balanced replicates
whereby half the sample in each replicate will have a replicate weight equal to two times sample weight (if the
BRR flag is 1) or 0 (if the BRR flag is 0). Users interested in implementing Fay’s BRR method may chose
different multipliers than 0 and 2 against which to factor the sample weights in each replicate. For instance,
they may chose to multiply the sample weights by 0.5 if the BRR flag is 0 and multiply them by 1.5 if the BRR
flag is 1.
Return to Table of Contents
When working with pooled data, analysts should consider whether they need to adjust the survey
weights from the annual public use files to account for the reprojection of survey estimates to a multi-year time
period. The survey weights provided in the 1996 annual dataset (HC‑012) project the HC‑012 sample to
the US population in 1996, and the survey weights in the 1997 dataset (HC‑020) project the HC‑020
sample to the US population in 1997. When combining two years of annual MEPS data (e.g., 1996 and 1997), these
single year weights over represent the population in the new two year period (1996 and 1997) by a factor of 2.
Likewise, when combining three years of MEPS data, the single year weights over represent the new three year
population by a factor of 3.
This over representation will only affect the estimates of totals but not the estimates of proportions.
That is, all estimates of total expenditures and their standard errors will be twice as high as they should be if using
the annual weights on the annual public use files pooled over two years without adjustment; these same estimates will be
three times too high if pooling over three years. Ratio estimates, such as the mean expenditure or the percent of
expenditures paid out of pocket, will not be too high when using the annual weights after pooling several years of MEPS
datasets together. Users wishing to estimate totals have two options to account for the multi year period: they may factor
the weights before they make any estimates or they may factor the estimates themselves (they should not do both).
- To illustrate the first method (factoring the weights), users who pool two years of MEPS
data should divide the sampling weight (variable PERWT if following the example above) by 2 prior to constructing
the BRR replicate weights. They would divide the sampling weight by three if pooling three years of MEPS data
together. With this adjustment to the sampling weight, all estimates of totals (and their standard errors) will
reflect the new multi year period. Estimates of proportions (and their standard errors) will also be correct after
this adjustment.
- To illustrate the second method (factoring the estimates themselves), users would make the
estimates of totals (and optionally their proportions) with the annual weights as is. They would then factor the
estimates of the totals (as well as their standard errors) by the number of years that were pooled. If the estimates
were made with two years of data, the totals and their standard errors would be divided by 2, if they were made
with three years of data, the totals and their standard errors would be divided by 3. Users would only adjust the
estimates and standard errors of totals, not those of proportions.
Return to Table of Contents
For any question regarding the HC‑036BRR file or pooling of data, please contact
Sadeq Chowdhury by email at: sadeq.chowdhury@ahrq.hhs.gov or
Fred Rohde by email at: frederick.rohde@ahrq.hhs.gov.
Return to Table of Contents
|