Analyzing NCHS Drug Data: NHANES and N/HAMCS

1. Analyzing NCHS Drug Data: NHANES and N/HAMCS Amy B. Bernstein, Sc.D. Presented at the AcademyHealth Annual Research Meeting San Diego, California June 6, 2004 National Center for Health Statistics

2. Topics • Definition of “drug” • Issues in drug utilization analyses • Coding issues • Examples of trend analyses

3. Definition of “Drug” NHANES • A drug is defined as a unique combination of generic ingredients. N/HAMCS • Recorded on visit record and abstracted verbatim

4. NHANES Drug Data • All drugs used in the past month at time of survey interview • Only prescription drugs are included, with a few exceptions • Example: --penicillin --penicillin, clavulanate potassium • More closely approximates “prevalence” of use

5. N/HAMCS Drug Data • Up to six drugs recorded (until 2003)—possible biases for drugs that are not salient to the physician or patient (e.g., PRN drugs) • Both prescription and non-prescription drugs are included • No information is available on compliance or use • Approximates prescribing patterns of drugs associated with medical care visits

6. Issues in Drug Utilization AnalysesNHANES Strengths: • Nationally representative and population-based • Examination, laboratory and questionnaire data on conditions, biochemical markers, nutrition, health status, and other items • Respondent-reported sociodemographic data (e.g., race and ethnicity are collected from respondent)

7. Issues in Drug Utilization AnalysesNHANES Possible Limitations: • Small sample size for less frequently prescribed drugs and small population subgroups • No trade names that help to determine therapeutic use on public use file • Respondents may not report use of some drugs

8. Issues in Drug Utilization AnalysesN/HAMCS Strengths: • Nationally representative • Physician/hospital characteristics • Conditions (from medical record—but limited number) • Selected procedures and tests • Relatively large sample size of visits

9. Issues in Drug Utilization AnalysesN/HAMCS Possible Limitations: • Limited sociodemographic data • Race/ethnicity data are reported by provider, not patient • Limited information on episodes or continuity of care • No data on compliance or actual utilization • Censoring of both drugs and diagnoses • Drugs are recorded verbatim from visit records, with possible misspelling

10. Issues in Drug Utilization AnalysesN/HAMCS Factors influencing N/HAMCS counts of drugs: • Person must have visited a physician or OPD • The more visits made for a specific condition requiring a specific drug, the greater the count of that drug on the N/HAMCS

11. Coding Drug DataInto Therapeutic Classes: Why? Therapeutic classes are useful in determining trends in treatment because: • There may be competing drugs that are used for the same condition • Guidelines may suggest a class of drugs for treatment of a condition, not a specific drug • There are too many individual drugs to analyze each individually! • Ideally, one wants to group “like” drugs together

12. FDA’s National Drug Code (NDC) Therapeutic Classes • Used by both N/HAMCS and NHANES on public use files • Code set is nonproprietary and downloadable from the FDA website • Identifies each of 20 major drug classes • Two-digit categories are general and represent all sub-categories (e.g., 03, Antimicrobial agents) • Specific four-digit categories represent the breakouts of the general category (e.g., 0346, Penicillins).

13. Coding IssuesWith FDA’s National Drug Code (NDC) Therapeutic Classes Some therapeutic categories are based on indication, while others are based on chemical composition or chemical mechanism --Examples: NDC Class 1374/Anticonvlusants or NDC Class 1945/Cough and Cold remedies or NDC Class 0512/Beta Blocker

14. FDA’s Therapeutic Drug Class Codes • When using any therapeutic drug class codes, need to review the drugs that are included under that category. • Ask yourself- “Is this the list of products I am interested in?” • When using NAMCS/NHAMCS or NHANES, look at FDA’s therapeutic drug classes and each drug under those classes. • Is the drug class coding sufficient for your research? • If not, manually review drugs and recategorize using the generic name codes

15. Coding Issues • Only generic ingredients provided on NHANES public use file, and some N/HAMCS drugs are reported as generic drugs (e.g., “aspirin”) • “Main reason for use” is collected and coded into ICD-9-CM classification • Some drugs have the same ingredients but different strengths, or different routes of administration that help determine therapeutic use Examples: Asthma drugs

16. Coding IssuesChanges Over Time • N/HAMCS 1980-2001: one NCD therapeutic class on PUF for each drug recorded • N/HAMCS 2002-2006: up to three NCD therapeutic classes on PUF for each drug recorded • NHANES 1988-94: three NDC therapeutic classes on PUF for each drug reported • NHANES 1999-2000: six NDC therapeutic classes on PUF for each drug reported

17. Coding IssuesChanges Over Time • Approved indications for drugs change over time (added or subtracted) • “Major” uses for drugs with multiple therapeutic uses change over time • Drugs may be replaced by other similar drugs • Codes and categories are periodically revised • Codes do not reflect “off-label” use

18. Coding IssuesChanges Over Time • Because therapeutic indications change over time • Analysts can merge the most current classifications to drug data from previous years • Otherwise drugs may be classified differently in different data years • This is less of an issue when analyzing specific drugs

19. Drugs Prescribed, Administered or Provided During Physician Office or OPD Visits, by NDC Therapeutic Drug Class, 1995-96 and 2001-02 SOURCE: Centers for Disease Control and Prevention, National Center for Health Statistics, National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey.

20. Coding IssuesMultiple Therapeutic Indications • Do you allow drugs with multiple approved indications to be double counted? • Example: • Aspirin has multiple therapeutic uses. • More of an issue when comparing or ranking categories than for analyses within a given therapeutic category

21. Aspirin: 1724  Antiarthritics   1722  Non-Narcotic analgesics   1728  Antipyretics Acetaminophen 1722  Non-Narcotic Analgesics   1728  Antipyretics Ibuprofen 1724  Antiarthritics  1722  Non-Narcotic Analgesics  1727  NSAID Toradol 1720  General Analgesics  1722  Non-Narcotic Analgesics  1727  NSAID Celebrex 1724  Antiarthritics 1727  NSAID Meloxicam 1724  Antiarthritics 1727  NSAID Coding IssuesDrugs with Overlapping Classifications in FDA NDC Therapeutic Classes

22. Antidepressant Use in N/HAMCS and NHANES Sources: National Health and Nutrition Examination Surveys, National Ambulatory Medical Care Surveys and National Hospital Ambulatory Medical Care Surveys

23. Cholesterol Drugs Prescribed, Administered Or Provided During Physician Office And Outpatient Department Visits, By Age Group: United States, 1995-2001 Statin only 80 years and over Other anti-cholesterol drugs 65-79 years 40-54 years 65-79 years SOURCE: Centers for Disease Control and Prevention, National Center for Health Statistics, National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey.

25. Coming Soon….. Watch for Health, United States, 2004, with Special Feature on Drugs! • Overall use • Asthma drugs • Psychotherapeutic drugs used by children • Antidepressant use by adults • Anti-cholesterol drugs • NSAIDs

26. Prescription Drug Use in Past Month, by Race and Ethnicity, 1999-2000 Source: Centers for Disease Prevention and Control, National Center for Health Statistics, National Health and Nutrition Examination Survey

27. Percent of Asthma Visits with Long-Term Control Drugs Prescribed, Ordered or Provided, 1995-2002 SOURCE: Centers for Disease Control and Prevention, National Center for Health Statistics, National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey.