Medication Data from Nationally Representative Provider and PopulationBased Surveys

1
Medication Data from Nationally Representative
Provider- and Population-Based Surveys

• Lisa L. Dwyer, MPH
• Saeid Raofi, MS Pharmacy
• Karen A. Lees, MPH
• Ryne Paulose, PhD
• National Center for Health Statistics
• 2006 Data Users Conference (Session 50)
• Washington, D.C.
• July 12, 2006

2
Background

• NCHS is the Nations principal health statistics
  agency
• compile statistical information to guide actions
  and policies to improve the health of our people
• provide public use files of survey data to the
  public
• Congress
• researchers
• health planners

3
Background

• Our health statistics allow us to
• document the health status of the population
• monitor trends in health status and health care
  delivery
• support biomedical and health services research
• provide information to guide and evaluate health
  policy decisions and programs

4
(No Transcript)
5
(No Transcript)
6
Background

• NCHS surveys that have collected medication data
• National Health Care Survey (NHCS)
• National Ambulatory Medical Care Survey (NAMCS)
• National Hospital Ambulatory Medical Care Survey
  (NHAMCS)
• National Nursing Home Survey (NNHS)
• National Hospital Discharge Survey (NHDS)
• National Health and Nutrition Examination Survey
  (NHANES)

7
Background

• National Health Care Survey
• family of mostly provider-based surveys
• collects information about health care
  facilities, their services, and their patients
• National Health and Nutrition Examination Survey
• population-based survey
• consists of a household interview, medical/dental
  examinations, and lab tests

8
Objectives

• To describe how the National Center for Health
  Statistics (NCHS) collects medication data across
  its surveys
• To describe how our data can be used to generate
  national estimates
• To discuss the future direction of NCHS surveys

9
Prescription Medications

• Drugs and their associated costs are at the
  forefront of national health care debates.
• According to figures reported by CMS,
  prescription drug expenditures increased at a
  much faster rate than the total health care
  expenditure for most of 1995-2004.
• Access to and affordability of drugs for the
  elderly were major drivers behind the Medicare
  Part D Drug Benefit implementation.

10
Health Care Expenditures
Source Centers for Medicare Medicaid
Services www.cms.hhs.gov/NationalHealthExpendData
/
11
Drug Utilization

• This increase in cost is driven, in part, by an
  increase in utilization.
• The national ambulatory health care surveys show
  that the number of drugs mentioned per visit
  increased between the 10-year period, 1993/1994
  and 2003/2004.
• Previous study reports that medication use is
  highest among the institutionalized elderly. This
  population continues to increase.

12
Increase in Drug Mention Rates
Source 1993-1994, 2003-2004 NAMCS and NHAMCS
13
Collection and Processing of Drug Information in
National Ambulatory Medical Care and National
Hospital Ambulatory Medical Care Surveys
14
Drug Data Collection in National Health Care
Surveys

• I will focus on the National Ambulatory Health
  Care surveys, NAMCS and NHAMCS, which have
  collected drug data the longest.
• The system developed for the processing and
  coding of the collected drug data for NAMCS and
  NHAMCS will be used for processing of the data in
  other surveys as well.
• I will also give a detailed description of this
  processing and coding system.

15
NAMCS and NHAMCS Background

• NAMCS
• Fielded 1973-1981, 1985, 1989-present
• Began collecting drug data in 1980
• NHAMCS
• Fielded annually since 1992
• Began collecting drug data in 1992

16
Items Collected

• Patient characteristics
• Age, sex, race, ethnicity
• Visit characteristics
• Source of payment, continuity of care, reason for
  visit, diagnosis, treatment, medications ordered
  or provided
• Provider characteristics
• Physician specialty, hospital ownership

17
NCHS Common Methodology

• National probability sample surveys
• Complex sample designs
• Common definitions, data items, sampling frames
• Medical diagnoses coded to ICD-9-CM
• High response rates
• Data processed by private contractor

18
NAMCS and NHAMCS Sample Design

• NAMCS
• 3-stage sample
• PSUs
• physicians
• visits during 1 week
• NHAMCS
• 4-stage sample
• PSUs
• hospitals
• ED/OPD clinics
• visits during 4 weeks

19
Generating National Estimates from Samples

• Statistics from the NAMCS and NHAMCS are derived
  by a multistage estimation procedures that
  produce essentially unbiased national estimates.
• The basic components of estimation are
• Inflation by reciprocals of the sampling
  selection probabilities
• Adjustment for nonresponse
• Weight smoothing
• A calibration ratio adjustment

20
Sample Weight

• The estimation procedure produces a single
  weight, called Patient Visit weight, for each
  NAMCS, OPD, and ED record.
• This weight is used for both visits and drug
  mentions.
• Weight must be applied or estimates of totals,
  percents and effects will be incorrect.

21
Definition of Drug Mentions

• A drug mention is the providers entry of drugs
  (prescription or over the counter),
  immunizations, allergy shots, anesthetics,
  chemotherapy, and dietary supplements that were
  ordered, supplied administered or continued
  during the visit.

22
Drug Data Processing

• Since 2003, the provider can list up to eight
  drug mentions on the survey form. From 1995 to
  2002 the provider could enter up to six drug
  mentions and before then up to five mentions.
• Each drug mention will be associated with a drug
  code at data entry stage.
• Drugs not in the database will be assigned a new
  unique code.

23
Adding Drug Characteristics

• Upon completion of visit files, the following
  drug characteristics are added to visit files for
  each drug mention
• Generic name
• Therapeutic class
• Ingredients
• Composition
• Control status
• Rx or OTC

24
Drug Coding and Characterization Example
25
Utility of Drug Characteristics

• Drug characteristics can be used to create
  summary reports based on therapeutic class,
  active ingredients, etc.
• They can be used in combination with patient and
  visit characteristics to study pharmacotherapy in
  specific disease areas.
• They can be used in combination with physician
  characteristics in studies looking at prescribing
  behavior.

26
Example Therapeutic classes with the highest
mention rate in 2003-2004
27
Example Mentions of Antihypertensive Drugs for
Ages 55-64 from 1999-2002
28
Therapeutic Classification System Through 2004

• Since 1985, the FDAs NDC therapeutic
  classification has been used
• Limitations of this system
• Only has one level of sub-classification
• FDA has discontinued this product

29
Adoption of Multum Lexicon as the Therapeutic
Classification System

• Starting with 2005 data, Multum therapeutic
  classification system will be used for
  classifying NAMCS and NHAMCS drug data.
• This system has two level of sub-classification.
• It is regularly updated.

30
Example Classification of Paroxetine by the two
classification systems

• NDC system
• 0600 central nervous system
• 0630 antidepressants
• Multum Lexicon system
• 242 psychotherapeutic agents
• 249 antidepressants
• 208 SSRI antidepressants

31
Using NAMCS/NHAMCS public use files for analyzing
drug data
32
Ambulatory Care Data Structure
33
File Structure

• Flat ASCII files for each setting and year
• Use file layout to read the data
• Input and format code available for
• SAS
• STATA
• SPSS
• Can use SETS (but no sampling variance estimates)

34
Visit File Layout
35
Ambulatory Health Care Data
http//www.cdc.gov/nchs/about/major/ahcd/ahcd1.htm
36
Drug Database System
37
Example of Drug Lookup Function

• By brand name PAXIL
• BY GENERIC NAME PAROXETINE
• http//www2.cdc.gov/drugs/

38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42

• For more information on the NAMCS and NHAMCS,
  please visit
• http//www.cdc.gov/nchs/about/major/ahcd/ahcd1.htm
  

43
Medication Data Collected in the 2004 National
Nursing Home Survey
44
2004 National Nursing Home Survey

• Nationally representative sample survey of U.S.
  nursing homes
• services/programs
• staff
• residents
• Conducted periodically since 1973-74
• 1977, 1985, 1995, 1997, 1999, 2004

45
2004 National Nursing Home Survey

• Taken out of the field after the 1999 survey for
  a major redesign.
• Put back into the field in 2004
• computerized data collection
• many new content items, including collection of
  medication data
• supplemental survey on nursing assistants, NNAS

46
2004 National Nursing Home Survey

• Two-stage probability survey design
• nursing home facility
• residents (up to 12 current residents)

47
2004 National Nursing Home Survey

• Sampling frame
• Centers for Medicare and Medicaid Services
  Provider of Services file of U.S. nursing homes
• state licensing lists compiled by private
  organization
• total of 16,628 nursing homes in frame

48
2004 National Nursing Home Survey

• Eligibility criteria
• licensed by State as a nursing facility
• certified and non-certified facilities
• three or more beds

49
2004 National Nursing Home Survey

• Survey items
• medications taken 24 hrs before facility
  interview
• standing or routine medications, or PRNs
• up to 25 medications
• medications taken regularly but not 24 hrs before
  facility interview
• up to 25 medications
• reason medications were prescribed

50
2004 National Nursing Home Survey

• Medication data
• found in medication administration records
• did not collect dosage, frequency, route
• collected during in-person interview at facility
• entered into CAPI system by interviewer
• processed like NAMCS/NHAMCS data

51
2004 National Nursing Home Survey

• Medication data collected
• prescription and nonprescription medications
• generics
• supplements
• vitamin/mineral, herbal, nutritional

52
2004 National Nursing Home Survey

• Drug characteristics appended
• generic name
• ingredients
• therapeutic classes
• composition status
• prescription status
• DEA status

53
2004 National Nursing Home Survey

• Data collected in 2004 NNHS are organized into
  three independent files
• Facility
• Resident
• Prescribed medication

54
2004 National Nursing Home Survey

• Resident File
• age
• sex
• race
• marital status
• admission diagnosis
• current primary and secondary diagnoses
• services/treatments received
• activities of daily living (ADLs)
• vaccination status
• expected source(s) of payment

• Facility File
• bed size
• ownership
• services
• per diem rates
• special programs
• staffing

55
2004 National Nursing Home Survey

• The Prescribed Medications (PM) file includes
• medication codes
• ICD-9 codes
• drug characteristics

56
2004 National Nursing Home Survey
Link data files using a randomly assigned ID
PM Data File
Resident Data File
Analytic File


(164,000 KB)
(13,000 KB)
Warning Great analytic potential but very large
file with over 13,000 records and over 1000
variables per record.
57
2004 National Nursing Home Survey
Preliminary Results.

• New data set provides information on
• 1.5 million current residents (weighted estimate)
• 71 female, 29 male
• mean age 81 (standard error 0.24)
• 86 White, 12 Black, 2 Other

58
2004 National Nursing Home Survey

• Resources available to data users
• Tab delimited ASCII file of PM data
• Long-term Care Drug Database
• Data dictionary document
• Users manual
• SAS, SPSS, and STATA input statements

59
2004 National Nursing Home Survey

• Things to consider when analyzing NNHS data
• complex sample survey design
• multiple stages of selection
• sampling weights are required
• point estimate
• standard error
• statistical software that takes the sample design
  into account

60
2004 National Nursing Home Survey

• Guidelines for Reporting Estimates
• Check sample size and standard error.
• Calculate the relative standard error (RSE).
• If sample size lt 30, then the value of the
  estimate should not be reported.
• If sample size is 30?59, or greater than 59 and
  the RSE ? 30, then the estimate can be reported
  but should not be considered reliable.
• If sample size ? 60 and the RSE lt 30, then the
  estimate is considered reliable.

61
2004 National Nursing Home Survey
Preliminary Results.

• Example Mean number of medications per resident
• Total population Mean 8.73, SE Mean 0.07
• Male population Mean 8.52, SE Mean 0.11
• Female population Mean 8.81, SE Mean 0.07

62
2004 National Nursing Home Survey
Preliminary Results.

• RSE (S.E. of point estimate/point estimate)
  100
• RSE for Total population (0.07/8.73) 100
  0.80
• RSE for Male population (0.11/8.52) 100
  1.29
• RSE for Female population (0.07/8.81) 100
  0.79

63
2004 National Nursing Home Survey

• Other examples of how data can be used
• to analyze how medications are used and if used
  for off-label indications
• to examine the differences in medication use
  among subpopulations
• to explore which medications were taken by
  residents receiving hospice/palliative/end-of-life
  care
• to determine the top therapeutic classes taken by
  nursing home residents

64
Top Therapeutic Classes Taken by Residents
Preliminary Results.
65

• For more information on the NNHS, please visit
• http//www.cdc.gov/nchs/nnhs.htm

66
Collecting Medication Data in the National
Hospital Discharge Survey Results from a Pilot
Study
67
National Hospital Discharge Survey

• Conducted annually since 1965
• Produces nationally representative data on
  characteristics of patients discharged from
  Non-Federal, short-stay hospitals

68
National Hospital Discharge Survey

• National probability sample
• Short-stay, non-Federal hospitals
• Three stage design
• Geographic units (PSUs)
• Hospitals
• Discharges

69
National Hospital Discharge Survey

• Hospitals included
• General hospitals
• Childrens general hospitals
• Hospitals with an average length of stay of less
  than 30 days
• Hospitals excluded
• Federal hospitals
• Military and VA hospitals
• Hospitals in institutions (such as prisons)
• Hospitals with fewer than 6 beds

70
National Hospital Discharge Survey

• Sample Size
• Approximately 500 hospitals sampled per year
• Over 300,000 discharges sampled per year
• Data Collection
• 55 manual
• 45 automated
• States, commercial firms, individual hospitals

71
National Hospital Discharge Survey

• Data are abstracted from the patients medical
  record

• Data are edited and weighted to produce national
  estimates

72
National Hospital Discharge Survey

• Patient Data
• Age
• Sex
• Race
• Expected source of payment
• Admission source and type
• Discharge status

• Hospital Data
• Bed size
• Ownership
• Geographic region

73
National Hospital Discharge Survey

• Medical Data
• Diagnoses principal and up to six secondary
• Surgical, diagnostic, or therapeutic procedures
  up to four
• Coded according to the International
  Classification of Diseases (ICD-9-CM)

74
National Hospital Discharge Survey

• Weight
• Inverse of the probability of selection
• Adjustments for non-response
• Population weighting ratio adjustment

75
National Hospital Discharge Survey and Uniform
Bill-92 (UB-92)

• Objective of UB-92
• To standardize and increase the submission of
  electronic claims
• UB-92 limits the information available for the
  NHDS to that which is necessary for billing
• Unable to modify the variables collected in the
  NHDS

76
NHDS Pilot Study

• To examine whether pharmaceutical data can be
  added to the manual or primary data collection
  part of NHDS
• Two-phase study conducted in 34 hospitals in
  three areas of the country
• 791 discharges from 2003
• Registered Health Information Technicians (RHIT)
  collected data
• Collected the names of all medications listed as
  administered in the medical record for that
  discharge

77
NHDS Pilot Study

• Medications
• Total of 10,839 medications collected
• 74 were illegible or indeterminate (lt1)
• Range 0 to 63
• Mean 13.61, Median 13.00
• 3 had no medications listed

78
Average number of medications overall and by
gender
79
Average number of medications administered by age
Overall 13.6 medications
80
Average number of medications by length of stay
Overall 13.6 medications
81
Top Therapeutic Classes
82
Top Generic Drugs Administered
83

• For more information on the NHDS, please visit
  our webpage
• http//www.cdc.gov/nchs/nhds.htm
• For more information on the pilot study or the
  NHDS redesign, please contact me at
• Karen Lees, MPH
• Email KLees_at_cdc.gov
• Phone [phone]

84
NHCS Future Steps

• Adoption of Multum therapeutic classification
  system beginning with 2005 data
• 2007 National Home and Hospice Care Survey
• 2008 National Survey of Residential Care
  Facilities
• 2006 National Survey of Ambulatory Surgery
• NHDS Redesign
• Contract currently let with RAND
• Options being evaluated currently
• Anticipate new NHDS collecting data in 2010