Introducing the MEPS Prescribed Medicines Data

By Julia A. Rivera Drew

The Household Component of the Medical Expenditure Panel Survey (MEPS), administered by the Agency for Healthcare Research and Quality (AHRQ), is a short panel survey collecting information for a nationally representative sample of the civilian, noninstitutionalized population. Since 1996, the MEPS has collected information on demographic and socioeconomic characteristics; health status; medical conditions; and health care access, utilization, and expenditures.

Based on information provided by a family respondent about each family member at each interview, AHRQ produces a dataset of all reported fills of prescribed medicines purchased by family members during the calendar year (including refills). For example, if a prescription was filled monthly, there would be 12 records for that specific prescribed medicine (DRUGID) in the annual file. The prescribed medicines data includes information such as the medication name (RXNAME), national drug code (RXNDC), therapeutic classification (MULTC1), when the person began taking the medication (RXBEGMM and RXBEGYR), amounts paid (RXFEXPTOT), and source of payment (RXFEXPSRC).

IPUMS MEPS provides a harmonized and integrated version of the MEPS Household Component data, including data from the prescribed medicines files.

What can I do with the MEPS prescribed medicines data?

The MEPS prescribed medicines data can be used to support investigations of prescription drug utilization and expenditures over time. In terms of expenditures, the data enable researchers to generate estimates of how much money was spent on average per person per prescription medication fill (both within the interview round and the calendar year) or to generate estimates of the total expenditures per person on prescription medications.

Below, we include two illustrations of the types of analyses one can conduct using the harmonized MEPS prescribed medicines data from IPUMS. One example is the proportion of adults aged 15-44 who purchased at least one fill of a prescription medication contraindicated in pregnancy, over time, and by gender (see Figure 1). Prescription medications contraindicated in pregnancy are those demonstrated to pose serious human fetal risk during pregnancy and where the risks of the drug clearly outweigh the potential benefits (a value of PREGCAT=X, see the PREGCAT variable description for more information). The majority of medications taken by women aged 15-44 contraindicated in pregnancy are hormonal contraceptives (MULTC1=97 with drug names (RXNAME) such as “Nuvaring” and “Ortho Evra”).

Figure 1. Proportion of Adults Age 15-44 who Purchased at Least One Prescribed Medication Contraindicated in Pregnancy by Gender, 1996-2021

Line graph showing the proportion of adults age 15-44 who purchased at least one prescribed medication contraindicated in pregnancy by gender from 1996-2021.

Another example of the types of analysis that can be supported by the prescribed medicine data is displayed in Figure 2, which provides the trend in average prescribed medicine expenditures paid by Medicare per prescribed medication fill from 1996-2021.

Figure 2. Average Expenditure per Prescribed Medication Fill (2021 dollars) by Whether Medicare Was the Payer, 1996-2021*

Line graph that shows the average expenditure per prescribed medication fill by whether medicare was the payer from 1996 to 2021

*For expenditures of $1 or more.

I’m interested, how do I get started?

Check out our user note describing how the MEPS prescribed medicines data are offered through IPUMS, and discover what prescribed medicines variables are available from IPUMS MEPS by browsing the list of RX Medicines variables. If you wish to make an extract using prescribed medicines data, you will need to click the “Change Data Structure” button and select the “Hierarchical” option:

IPUMS MEPS Data Extract generator screen with the "Change Data Structure" button outlined in red

IPUMS MEPS Extract Data Structure selection wit "Hierarchical" circled with red

For more information about MEPS and the MEPS prescribed medicines data, check out our user guide and reach out to us at ipums@umn.edu with questions.

2022 ATUS Eating and Health Module Data: New Variables and Updates

By Annie Chen & Sarah Flood

The American Time Use Survey Eating and Health Module, funded by the Economic Research Service, asks a series of questions related to grocery shopping, food preparation, and nutrition. The most recent module was fielded in 2022 during the COVID-19 pandemic and was previously fielded in 2006 to 2008 and 2014 to 2016. The 2022 Eating and Health Module, set to be fielded again in 2023, asks new questions, asks similar questions in different ways than previously fielded modules, and contains additional variables of high interest to researchers.

New Variables in 2022

The 2022 ATUS Eating and Health Module asks a series of new questions related to exercise/physical activity, grocery shopping, meal preparation, and food quality. The food quality questions are especially interesting because they provide researchers with the opportunity to assess relationships between food quality and time use, which hasn’t been possible previously with these data. This is the first time that the ATUS has asked any information about respondents’ food intake on the ATUS diary day. The module is also responsive to changes in shopping behavior during the pandemic, specifically online grocery shopping and grocery delivery/pickup options. The shopping and meal preparation enjoyment questions might allow for comparisons to the ATUS Well-Being Module (fielded in 2010, 2012, 2013, and 2021).

Here’s a list of the new variables in 2022:

  • Exercise
    • EXINT – Amount of increase in breathing or heart rate due to vigorous exercise/activity
  • Food quality
  • Grocery shopping
    • GROSHPAMT – Amount of grocery shopping for household
    • GROSHPENJOY – Enjoyment level of grocery shopping for household
    • ONGROSHPFREQ – Frequency of online grocery purchase
    • ONGROSHPGET – Typical online grocery retrieval
    • ONGROSHPWHY – Main reason for online grocery shopping
    • ONGROSHPWHYNOT – Main reason for not online grocery shopping
  • Meal preparation
    • FASTFDATE – Ate prepared food yesterday
    • PRPMELAMT – Amount of meal preparation in household
    • PRPMELENJOY – Amount of enjoyment for meal preparation in household

Variables in 2022 that are Similar to Previous Years

There are three variables new to the 2022 data that are similar to variables available in the earlier modules. While they are similar, they are different enough that we did not harmonize them with the previous variables. The similar pairs of variables are listed in the table below.

Variable NameVariable Description2006200720082014201520162022
FOODSHOPATUS respondent usually does the food shoppingXXXXXX
GROSHPAMTAmount of grocery shopping for householdX
MEALPREPATUS respondent usually does the meal preparationXXXXXX
PRPMELAMTAmount of meal preparation in householdX
FASTFDPurchased prepared food in last seven daysXXX
FASTFDATEAte prepared food yesterdayX

FOODSHOP asks whether the respondent usually does the grocery shopping while GROSHPAMT asks about how much grocery shopping the respondent does. Figure 1 compares estimates from these two variables by gender. It shows the weighted percent of men and women who reported usually doing the food shopping (FOODSHOP) in 2006-2018 versus the percent of men and women who reported doing “all” (GROSHPAMT=5) of the food shopping and “a lot” or “all” (GROSHPAMT=4 or 5) of the food shopping in 2022. (As a reminder, when analyzing variables from the Eating and Health Module, researchers should use EHWT because compared to WT06, EHWT accounts for module non-response.) The more restrictive estimate from 2022 (GROSHPAMT=5) is quite a bit lower than estimates from previous years. But when we include two grocery shopping categories (GROSHPAMT=4 or 5), estimates for 2022 are much closer to the estimates for FOODSHOP suggesting that there is decent comparability between these questions over time. That said, we cannot say whether the difference in estimates is due to gender convergence in grocery shopping or whether the differences are due to respondents being asked different questions.

Figure 1: Grocery shopping for household by sex. Bar graph that shows the percentage FOODSHOP==1, GROSHPAMT==5, and GROSHPAMT==4 OR 5 by male and female

We conducted a parallel analysis for meal preparation, which is shown in Figure 2. These figures compare the weighted percent of men and women who reported usually doing the meal preparation (MEALPREP) in 2006-2018 versus the percent of men and women who reported doing “all” (PRPMELAMT=5) or doing “a lot” or “all” of the meal preparation (PRPMELAMT=4 or 5) in 2022. As we saw for grocery shopping, the percentages of men and women who report doing a lot or all of the meal preparation are reasonably similar across time despite changes in the exact questions asked to respondents.

Figure 2: Meal preparation for household by sex. Bar graph that shows the MEALPREP, PRPMELAMT==5, and PRPMELAMT==4 OR 5 by male and female

High Interest Variables from the Eating and Health Module

There were also a set of variables in the 2022 Eating and Health Module that were asked previously relating to general health (GENHEALTH), Body Mass Index (BMI), physical activity (EXERCISE), food security (ENOUGHFD), and eating while engaged in other activities (SED_EAT_LN), in addition to other person-level variables.

Announcing IPUMS MICS

By Anna Bolgrien

IPUMS MICS Logo

IPUMS has an exciting new data collection to announce: IPUMS MICS!

IPUMS MICS is the integrated version of UNICEF MICS (Multiple Indicator Cluster Surveys), the largest and most robust source of data on women and children’s well-being across the globe, including countries in Africa, Eastern Europe, Asia, and Latin America. Separate datasets cover women of childbearing age, children aged 0 to 4, children aged 5 to 17, respondent’s birth history, men, household members, and household characteristics.

Currently, IPUMS MICS includes harmonization of data from 202 MICS samples, which represent 88 countries, and cover surveys conducted between 2005-forward. There are over 800 integrated variables currently available on our website. Future releases will expand the sample and variable coverage of IPUMS MICS.

Continue reading…