Font Size:

Linkage File for 2021 MEPS and 2017-2020 NHIS Public Use Files

1.0Overview

The Medical Expenditure Panel Survey (MEPS) Household Component (HC) uses the National Health Interview Survey (NHIS) as its sampling frame. Each year a new MEPS-HC panel is established by drawing the sample from the previous year's NHIS responding sample. The MEPS-HC design is based on an overlapping panel design. Usually, two consecutive panels are overlapped in a MEPS full year (FY) file. However, the 2021 FY file included four overlapping panels, Panels 23, 24, 25, and 26, to increase the sample size to compensate for low response rates and data collection difficulties due to the coronavirus 2019 (COVID-19) pandemic. Details of the 2021 panel design are provided in the documentation for the 2021 Full Year Population Characteristic File.¹

Due to the 2019 NHIS redesign, from each household only one adult and one child (if any children live in the household) were included in the NHIS sample.² As a result, potentially only one adult and one child (if any), in each MEPS household could be linked to the previous data year's NHIS public use file. However, prior to the 2019 NHIS redesign, all persons in an NHIS household were included in the NHIS PUF, which made it possible to potentially link nearly all persons in a MEPS panel.³

As illustrated in Figure 1 below, the 2021 MEPS full-year public use files (PUFs) cover the calendar year and contain data from Rounds 7, 8, and 9 of MEPS Panel 23 (which uses the 2017 NHIS as its sampling frame), Rounds 5, 6, and 7 of the MEPS Panel 24 (which uses the 2018 NHIS as its sampling frame); Rounds 3, 4, and 5 of MEPS Panel 25 (which uses the 2019 NHIS as its sampling frame); and Rounds 1, 2, and 3 of MEPS Panel 26 (which uses the 2020 NHIS as its sampling frame). Therefore, MEPS Panels 23, 24, 25, and 26 can be linked to 2017, 2018, 2019, and 2020 NHIS PUFs, respectively.

Figure 1. Mapping of MEPS Year, Panels, and Rounds to NHIS Years

This image illustrates that 2020 data was collected in Rounds 6 and 7 of Panel 23, and Rounds 3, 4, and 5 of Panel 24, and Rounds 1, 2, and 3 of PAnel 25.

PUFs containing NHIS data for a given calendar year are available from the National Center for Health Statistics (NCHS).

Users who need to augment the MEPS data with information from NHIS can do so with the linkage file described in the following sections.

2.0Linkage File Description

The MEPS and NHIS linkage file, NHMEP21X.DAT, allows the data user to merge any of the person-level 2021 MEPS full-year public use data files with the 2017, 2018, 2019, and 2020 NHIS person-level PUFs (Person, Sample Adult, and Sample Child).

The NHIS person identifiers changed in 2019. Prior to 2019, each family (FMX) has been considered a separate case, and unique person identifiers have been Household Serial Number (HHX), Family Sequence Number (FMX), and Person Sequence Number (FPX). Beginning in 2019, only a sample adult and, where available, a sample child were included from each household. Therefore, the identifiers in 2019 and later are HHX and record type (RECTYPE) that specifies sample adult, sample child, or not sampled for NHIS.

The linkage file contains 28,336 person-level records and eight variables. In the linkage file, a record exists for each of the MEPS 2021 full-year persons. Each record contains the MEPS sample person ID (DUPERSID) and the corresponding NHIS sample person IDs (HHX, FMX, FPX, and RECTYPE). The linkage file can be linked to any of the person-level MEPS 2021 full-year public use data files using the variable DUPERSID. The linkage file can be linked to the NHIS 2017 or 2018 person-level data files by HHX, FMX, FPX, and SRVY_YR and to the NHIS 2019 or 2020 sample adult and sample child data files by HHX, RECTYPE, and SRVY_YR.

When a MEPS sample person does not link to NHIS, HHX is set to 999999, FMX is set to 99, FPX is set to 99, SRVY_YR is set to 9999, RECTYPE is set to 99, and LINKFLAG is set to 0.

3.0Linkage File Record Counts

Of the 7,016 MEPS Panel 23 persons, 6,267 persons link to the 2017 NHIS data; 6,123 of the 6,696 Panel 24 persons link to the 2018 NHIS data; 3,164 of the 6,213 Panel 25 persons link to the 2019 NHIS data; and 4,173 of the 8,411 Panel 26 persons link to the 2020 NHIS data. A total of 8,609 persons in the four panels do not link to either 2017, 2018, 2019, or 2020 NHIS data. For Panels 23 and 24, these unlinked cases include newborns; newly in-scope persons; and a small number of cases where the NHIS identified a household as responding, but when fielded in MEPS, it was determined to actually be a nonresponding household. As mentioned above, starting with Panel 25, unlinked cases may also be household members who are neither the sample adult nor sample child. Table 1 below summarizes the linkages.

**Table 1 - Linkage File Record Counts**
2021 MEPS Full-Year Data	Total 2021 MEPS Persons	Linked to 2017, 2018, 2019, or 2020 NHIS PUF (total observations in NHIS PUF)	Not Linked to NHIS
2021 MEPS Full-Year Data	Total 2021 MEPS Persons		Not Linked to NHIS	Panel 23 persons (2017 NHIS)	7,016	6,267 (78,132)	749
Panel 24 persons (2018 NHIS)	6,696	6,123 (72,831)	573
Panel 25 persons (2019 NHIS)	6,213	3,164 (41,190)	3,049
Panel 26 persons (2020 NHIS)	8,411	4,173 (37,358)	4,238
Total	28,336	19,727 (229,511)	8,609

4.0Linkage File Record Layout

Table 2 is the record layout for the person-level MEPS-NHIS linkage file (NHMEP20X.DAT).
Variable	Columns	Type	Label and value range^*
DUPERSID	1 - 10	Character	MEPS encrypted person ID (range = 2320005101 - 2689507104)
HHX	11 - 17	Character	NHIS household serial number (range = 000017 - H070028)
FMX	18 - 19	Character	NHIS family number (range = 01 - 06 )
FPX	20 - 21	Character	NHIS person number (range = 01 - 14)
LINKFLAG	22 - 22	Numeric	Linkage status between MEPS and NHIS (1 or 0)
PANEL	23 - 24	Numeric	MEPS panel number (23, 24, 25, or 26)
SRVY_YR	25 - 28	Numeric	NHIS survey year (2017, 2018, 2019, or 2020)
RECTYPE	29 - 30	Numeric	Record Type (10 or 20)

^* Values may be missing based on NHIS survey year or linkage status.

Below is the input statement to convert the linkage file (NHMEP21X.DAT) to a SAS dataset.

DATA XX.NHMEP21X; INFILE "C:\TEMP\MEPS\NHMEP21X.DAT"; INPUT DUPERSID $1-10 HHX $11-17 FMX $18-19 FPX $20-21 LINKFLAG 22 PANEL 23-24 SRVY_YR 25-28 RECTYPE 29-30; RUN;

5.0Linking Instructions for SAS Users

The following is one way of adding NHIS person-level variables to the MEPS person-level file. Input files are MEPS HC-228 (2021 Full-Year Population Characteristic File), the 2017 NHIS person-level data file, the 2018 NHIS person-level data file, the 2019 NHIS sample adult and sample child data files, the 2020 NHIS sample adult and sample child data files, and the linkage file NHMEP21X.DAT.

Create eight SAS datasets as follows:
- Convert MEPS HC-228 (ASCII, SAS transport file, or SAS V9 file) to a SAS dataset named FY2021 (n = 28,336).
- Convert the linkage file NHMEP21X.DAT to a SAS dataset named NHMEP21X (n = 28,336).
- Convert the 2017 NHIS Person file to a SAS dataset named NHIS2017 (n = 78,132). Make sure the SAS dataset includes HHX, FMX, FPX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
- Convert the 2018 NHIS Person file to a SAS dataset named NHIS2018 (n = 72,831). Make sure the SAS dataset includes HHX, FMX, FPX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
- Convert the 2019 NHIS Sample Adult file to a SAS dataset named NHIS2019A (n = 31,997). Make sure the SAS dataset includes HHX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
- Convert the 2019 NHIS Sample Child file to a SAS dataset named NHIS2019C (n = 9,193). Make sure the SAS dataset includes HHX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
- Convert the 2020 NHIS Sample Adult file to a SAS dataset named NHIS2020A (n = 31,568). Make sure the SAS dataset includes HHX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
- Convert the 2020 NHIS Sample Child file to a SAS dataset named NHIS2020C (n = 5,790). Make sure the SAS dataset includes HHX, RECTYPE, SRVY_YR, and other variables that are to be added to the MEPS full-year dataset.
Sort FY2021 by DUPERSID. Concatenate NHIS2017, NHIS2018, NHIS2019A, NHIS2019C, NHIS2020A, and NHIS2020C into one dataset named NHISALL (n = 229,511). Sort NHISALL by HHX, FMX, FPX, RECTYPE, and SRVY_YR.
Merge FY2021 (n =28,336) with NHMEP21X (n = 28,336) by DUPERSID. Name the output dataset MEPS (n = 28,336). Then sort MEPS by HHX, FMX, FPX, RECTYPE, and SRVY_YR.
Merge MEPS (n = 28,336) with NHISALL (n = 229,511) by HHX, FMX, FPX, RECTYPE, and SRVY_YR. Keep records only in MEPS (n = 28,336). Name the output dataset MEPS21NH (n = 28,336).

Sample SAS Code for Adding NHIS Variables to the MEPS Dataset.


  LIBNAME MEPS "C:\TEMP\MEPS";    /*MEPS 2021 Full-Year PUF, MEPS-NHIS Link, output file*/

  LIBNAME NHIS "C:\TEMP\NHIS";    /*NHIS 2017 and 2018 Person Files and NHIS 2019 and 2020 Sampled Adult and Child Files*/ 

  

  PROC FORMAT; 

    VALUE AGE

    .='.'

    0-HIGH='>=0';

  RUN; 

  

  PROC SORT DATA=MEPS.FY2021;

    BY DUPERSID;

  RUN;

  

  DATA NHISALL;

    SET NHIS.NHIS2017 (KEEP=HHX FMX FPX RECTYPE SRVY_YR AGE_P /*other NHIS variables*/)

      NHIS.NHIS2018 (KEEP=HHX FMX FPX RECTYPE SRVY_YR AGE_P /*other NHIS variables*/)

        NHIS.NHIS2019A (KEEP=HHX RECTYPE SRVY_YR AGEP_A RENAME=(AGEP_A=AGE_P) /*other NHIS variables*/)

        NHIS.NHIS2019C (KEEP=HHX RECTYPE SRVY_YR AGEP_C RENAME=(AGEP_C=AGE_P) /*other NHIS variables*/)

        NHIS.NHIS2020A (KEEP=HHX RECTYPE SRVY_YR AGEP_A RENAME=(AGEP_A=AGE_P) /*other NHIS variables*/)

        NHIS.NHIS2020C (KEEP=HHX RECTYPE SRVY_YR AGEP_C RENAME=(AGEP_C=AGE_P) /*other NHIS variables*/);


  RUN;

  

  PROC SORT DATA=NHISALL;

    BY HHX FMX FPX RECTYPE SRVY_YR;

  RUN;

  

  DATA MEPS;

    MERGE MEPS.FY2021 MEPS.NHMEP21X (KEEP=DUPERSID HHX FMX FPX RECTYPE SRVY_YR LINKFLAG);

    BY DUPERSID;

  RUN;

  

  PROC SORT DATA=MEPS;

    BY HHX FMX FPX RECTYPE SRVY_YR;

  RUN;

  

  DATA MEPS.MEPS21NH;

    MERGE MEPS (IN=A) NHISALL;

    BY HHX FMX FPX RECTYPE SRVY_YR;

    IF A;

  RUN;

  

  TITLE1 "MEPS 2021 FY data with NHIS variables";

  PROC FREQ DATA=MEPS.MEPS21NH;

    TABLES LINKFLAG*SRVY_YR*AGE_P/LIST MISSING;

    FORMAT AGE_P AGE.; 

  RUN;

Sample Stata Code for Adding NHIS Variables to the MEPS Dataset


  cd "c:temp"

  

  log using stata21.log, replace

  

  use "meps\h228", clear

  rename *, lower

  sort dupersid

  

  tempfile fy2021

  save `fy2021', replace

  

  use "nhis\nhis2017", clear

  append using "nhis\nhis2018"

  append using "nhis\nhis2019a"

  append using "nhis\nhis2019c"

  append using "nhis\nhis2020a"

  append using "nhis\nhis2020c"

  rename *, lower

  sort hhx fmx fpx rectype srvy_yr

  tempfile nhisall

  save `nhisall', replace

  

  infix str dupersid 1-10 str hhx 11-17 str fmx 18-19 str fpx 20-21 linkflag 22 panel 23-24 srvy_yr 25-28 rectype
  29-30

  using "meps\nhmep21x.dat", clear

  sort dupersid

  tempfile link

  save `link', replace

  

  use `fy2021'

  merge 1:1 dupersid using `link'

  drop _merge

  sort hhx fmx fpx rectype srvy_yr

  tempfile meps

  save `meps', replace

  

  merge m:1 hhx fmx fpx rectype srvy_yr using `nhisall'

  keep if _merge != 2 /*drop cases where a record was found in the NHIS PUFs but not in MEPS*/

  keep dupersid hhx fmx fpx rectype srvy_yr linkflag /*edit this line to add any other desired nhis variables*/

  save "meps\meps21nh", replace

  

  describe

  tab srvy_yr linkflag, missing

  

  log close

6.0Further Information

For any questions regarding the linkage file, please contact May Chu at May.Chu@ahrq.hhs.gov. MEPS public use data files can be downloaded free of charge from the MEPS website at https://meps.ahrq.gov. NHIS public use data files can be obtained by contacting NCHS by telephone (301‑458‑4636) or through their website, https://www.cdc.gov/nchs.

Footnotes

¹ https://meps.ahrq.gov/data_stats/download_data/pufs/h228/h228doc.shtml

² For details of the 2019 redesign, see https://www.cdc.gov/nchs/nhis/2019_quest_redesign.htm.

³ A small number of persons may not be linkable between NHIS and MEPS because they may have left or joined the household between when NHIS was fielded and when MEPS was fielded.