Introduction
The Agency for Healthcare Research and Quality (AHRQ) Medical Expenditure Panel Survey (MEPS) is a set of data on U.S. households about their healthcare expenditures. It includes data on the individual / household demographics, socioeconomic status, insurance coverage, and healthcare expenditures. Healthcare expenditures include data on health-related spending, medical conditions, prescriptions, and utilization (e.g., number of office-based visits). MEPS draws upon a nationally representative subsample from the National Health Interview Survey, which is conducted by the National Center for Health Statistics. Hence, MEPS provides researchers with the ability to generate estimates for the representative U.S. population.
MEPS Data
MEPS data are located on their website in their data files page. You can find data from 1996 to the most recent available year (during the writing of this tutorial, 2020 was the latest release).
The MEPS data files include the Full-Year Consolidated Data files, which is the calendar-year summary of the different longitudinal panels. The Full-Year Consolidated Data files contain information on the annual healthcare expenditures by the type of care; it contains data on spending, insurance coverage, health status, patient satisfaction, and several health conditions. The Full-Year Consolidated Data files also contains information from several surveys (e.g,. Diabetes Care Survey).
Load MEPS data into R
MEPS data can be downloaded onto your local storage and read into a
statistical software program such as Stata or R. But you can also
communicate directly with the AHRQ MEPS website to load your data rather
than having to download it. We will load the Full-Year Consolidated Data
file from 2020, which is named HC-224
. To find out the name
of the file, you will need to go MEPS data files site and click on the
Full-Year Consolidated Data files. In this page (Figure 2), you can see
the data file with the code HC-224
, which is the Full-Year
Consolidated Data file for 2020. When we enter this into our R code, we
will use the file name h224
.
You will need to download and install the MEPS
package. The MEPS
package will provide tools for you to
load and manipulate the MEPS Data files. You will need to have R
devtools
package installed.
## Install the devtools package
# install.packages("devtools") ## You only need to install this once
# library("devtools") ## You will need to reload the MEPS package each time you restart R
# install_github("e-mitchell/meps_r_pkg/MEPS") ## This will install the MEPS package for R
There are two methods to load MEPS data into R.
Method 1 requires that you know the file name. In this example, the
MEPS 2020 Full-Year Consolidated Data file is named h224
.
We will use the read_MEPS
function to load the MEPS data
onto R.
When using Method 2 to load the MEPS data, we don’t need to know the
file name, but we need to know the year and the data type. For example,
for the Full-Year Consolidated Data file, we use the
year = 2020
and type = "FYC"
option. For this
method, we will also use the read_MEPS
function to the MEPS
data onto R.
The tolower
function is used to change all the variable
names from upper case to lower case. MEPS defaults the column names to
upper case. I like to change this to lower case because it’s easier for
me to type.
### Load the MEPS package
library("MEPS") ## You need to load the library every time you restart R
#### Method 1: Load data from AHRQ MEPS website
hc2020 = read_MEPS(file = "h224")
#### Method 2: Load data from AHRQ MEPS website
hc2020 = read_MEPS(year = 2020, type = "FYC")
## Change column names to lowercase
names(hc2020) <- tolower(names(hc2020))
There are over 1400 variables in the MEPS 2020 Full-Year Consolidated
Data file. We can reduce this to the essential variables using the
subset
function. This will generate a smaller data frame
that we will call keep_meps
. The variables that we want to
collect are the subject unique identifier (dupersid
), the
survey weights (varpsu
, varstr
,
perwt20f
), and the total healthcare expenditures for 2020
(totexp20
).
### Keep the subject's unique ID, survey weights, and total expenditures
keep_meps <- subset(hc2020, select = c(dupersid, varpsu, varstr, perwt20f, totexp20))
head(keep_meps) ## View the first six rows of the data frame
## # A tibble: 6 × 5
## dupersid varpsu varstr perwt20f totexp20
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 2320005101 1 2079 8418. 459
## 2 2320005102 1 2079 5200. 564
## 3 2320006101 1 2028 2140. 140
## 4 2320006102 1 2028 2216. 4673
## 5 2320006103 1 2028 4157. 410
## 6 2320012102 2 2069 1961. 2726
Since MEPS uses a complex survey design, these weights are needed to estimate standard errors that are reflective of the representative sample of the U.S. population. We’ll learn how to apply these survey weights to the MEPS data files in a future tutorial.
Conclusions
Loading MEPS data into R allows us to perform analysis easily and
quickly. In this tutorial, you learned how to load MEPS data into R
directly from the MEPS website. However, you can also download the MEPS
data onto your local storage and use the setwd
command to
set the working directory.
In future tutorials, we’ll learn how to apply the survey weights and perform descriptive analyses using the MEPS data files.
Acknowledgements
There are a lot of tutorials on how to use MEPS data with R. I found the AHRQ MEPS GitHub page to be an invaluable resource.
This is a work in progress, and I may update this in the future.