Update Notes
MEPS
HC-010A: 1996 Prescribed Medicines File
Update #8: 2/02/18
A problem was discovered with the linking between the MEPS Prescribed Medicines files and the Cerner Multum file that resulted in some incorrect therapeutic classes being assigned.
The Multum Lexicon Addendum files (HC-068) contain the corrected therapeutic classes. Analysts should use caution when using the Cerner Multum therapeutic class variables for analysis
and should always check for accuracy, because the Multum classification has been changed by the addition of new classes and subclasses, and by changes in the hierarchy of classes.
See the documentation for the Multum Lexicon Addendum files for examples.
The Multum Lexicon Addendum files (HC-068) also contain an additional variable not in the 1996
through 2012 MEPS Prescribed Medicines files: the generic name of the drug most commonly used by prescribing physicians.
Update #7: 11/22/13
STATA Programming Statements have been added.
Update #6: 06/28/12
A problem was discovered with the linking between the MEPS Prescribed Medicines files and the Cerner Multum
file that resulted in some incorrect therapeutic classes being assigned. In particular, some diagnostic tests
and medical devices were inadvertently assigned to be in a therapeutic class when they should not have been.
Specifically, from 1996-2002, some diabetic supplies were assigned to be in TC1S1=101 (sex hormone), and from
2003 through 2010 some diabetic supplies were assigned to be in TC1S1=37 (toxiods). In addition, starting
in 2006, NDC 00169750111 should have been assigned to TC1=358 and TC1S1=99. Analysts should use caution when
using the Cerner Multum therapeutic class variables for analysis and should always check for accuracy, because
there may be additional problems not described in this user note.
Update #5: 10/26/12
SPSS Programming Statements have been added.
Update #4: 06/06/01
For approximately 9,000 records on the Prescribed Medicines File
(HC-010A), health conditions were imputed based upon the type of
medication reported. Therefore, condition data included on these
records will not necessarily match the conditions identified on the
1996 Condition File (HC-006R) for these respondents.
Because these prescribed medicine records were imputed, they cannot
be linked to the Condition File using the Condition Link File (File
1 of HC-010I). These records can be identified on the Prescribed
Medicines File where PCIMPFLG=0. This problem will not exist for
the 1997 Prescribed Medicines File.
Update #3: 3/16/01
This release updates data previously
released on the 1996 Prescribed Medicines File(HC-010A),
Differences between this release and the earlier release are the
following:
1) MEPS staff discovered an inaccuracy
in the number of times a household reported purchasing or otherwise
obtaining a prescription
drug in a particular round for a small percentage of household-reported
medications in the Spring 2000 release. This inaccuracy was due
to an instrument design flaw which caused interviewer error, and
in isolated cases, resulted in mis-reported large numbers of prescription
refills for a medicine in a given round. For some cases, it seems
that the year that the person started taking the drug was recorded
in the field that gives the number of times that the person purchased,
or otherwise obtained the drug, during the round, as well as in
the field that provides the year the person started taking the
medicine. For example in the round a drug was first mentioned,
when a person reported having first started taking the drug in
1996, a "96" was entered in the field for the year the
person first started taking the drug. For the problem cases, a "96" also
appeared in the preceding field indicating the number of times
the drug was reported as being purchased or otherwise obtained
during the round. Outlier values where this situation (and similar
instances) occurred were determined by comparing the number of
days a respondent was in the round and the number times the person
reported having purchased or otherwise obtained the drug in the
round, and were determined in consultation with an industry expert.
For the cases where this error occurred, a new value for the number
of times a person reported purchasing or otherwise obtaining the
drug for that round was imputed.
2) In the Spring 2000 release, the prescribed medicine events
in which a household respondent did not know/remember the number
of times a certain prescribed medicine was purchased or otherwise
obtained in a particular round were inadvertently excluded from
the file. For those events, all the data related to that drug event,
including the number of times a respondent purchased or otherwise
obtained the drug in a particular round, was imputed from a donor
drug event with similar characteristics.
3) Based on the two issues described above, as well as additional
editing rules, the variable that contains the number of times a
prescription was purchased or otherwise obtained by a person in
a round was edited. Round 3 drugs, which spanned 1996 and 1997,
had to be reallocated between 1996 and 1997. Because of this reallocation
of Round 3 events, it is possible that records (and persons) included
on the Spring 2000 release may not be included on this release.
4) This release DOES include the variable SAMPLE and DOES NOT
include the variable FREEFLG. The SAMPLE variable indicates if
a respondent reported receiving a free sample of a prescription
medicine in the round (0=no, 1=yes). Therefore, SAMPLE=1 for all
acquisitions of a certain prescription medicine for a person during
a round that the respondent reported having received a free sample
of that prescription during the round. In the Spring 2000 release,
FREEFLG (which is NOT included on this release) indicated if a
record on the file had been designated as a free sample. After
reviewing the data and questionnaire in great detail, it was determined
that it was not appropriate to make this assumption.
5) Some of the estimation examples (see Section 4 of documentation)
had to be revised because the variable FREEFLG is not included
on this release. (Please see #4 above for details.) The variable
DIABFLG was substituted for the variable FREEFLG in the estimation
examples, as necessary.
Update
#2: 8/01/00
An
updated version of File 1 of the 1996 Prescribed Medicines
File (HC-010A) was released on 8/1/00. This file is identical
to the previous version
with the exception of the condition variables (NUMCOND, evICDiX,
evPROiX and evCCCiX), which were modified for a subset of
records. At the present time users can obtain the correct
set of linked condition variables for this file by using
the updated condition-event link file provided as File 1
of the Appendix File, HC-010I, with the condition file, HC-006.
See the HC-010I documentation for information on linking
event files to the condition file.
Update
#1: 4/28/00
Public
data users should note the following changes to the
1996 Prescribed Medicines File (HC-010A) since the preliminary
release of this file in December 1999 are:
1. Prescribed medicines event identifiers are now
unique for all purchases as well as other types of
acquisitions including free samples.
2. RXBEGYR (formerly RXBEGYY) and RXBEGMM have been
recoded and will differ slightly from the preliminary
release.
3. Condition and clinical classification codes have
been ordered so that up to three valid codes are listed
for each prescribed medicine event. Valid condition
codes are followed by invalid codes, when appropriate.
4. Detailed linking examples, including SAS code,
are included in the forthcoming Appendix File (May
2000) which will enable analysts to perform various
linking activities including prescribed medicines to
conditions (and vice-versa) and prescribed medicines
to other events (and vice-versa).
|