Pharmacy Data Mark W. Smith, PhD July 13, 2005 Health Economics Teleconference Seminar 1-800-767-1750 access code 45043.

Pharmacy Data

Mark W. Smith, PhDJuly 13, 2005

Health Economics Teleconference Seminar1-800-767-1750 access code 45043

Topics

• Overview of Data Sources

• Access & File Names

• Highlights of Contents

• Guidance for Use

• Non-VA Pharmacy Data

Abbreviations

• AAC: Austin Automation Center• BPA: Blanket Purchase Agreement• CMOP: Centralized Mail-Order Pharmacy• DSS: Decision Support System• MCRR: Medical Care Cost Recovery program• NDE: National Data Extract• PBM: Pharmacy Benefits Management• VAMC: Veterans Affairs Medical Center• VISTA: Veterans Health Information Systems and

Technical Architecture

Overview - 1

• VISTA– Repository of primary data: clinicians enter data directly – Many files comprise pharmacy data “package” – Separate VISTA system at each VAMC

• PBM v3.0 Database – Data from local VISTA systems + new elements– National coverage: all VA pharmacies– Only outpatient prescriptions available currently

• DSS National Pharmacy Extract – Data from local DSS systems + new elements – National coverage: all VA (local) DSS systems– Inpatient and outpatient prescriptions available

Overview - 2

• DSS National Data Extracts – Contains all types of care (inpatient, outpatient,

pharmacy)

– Separate from DSS Pharmacy Extract

– Data from local DSS systems + new elements

– National coverage

Additional Pharmacy Data Sources

• Fee Basis files– Data from non-VA pharmacies paid by VA

– National coverage

• DSS Production Data – Contains finer detail than national DSS extracts

– Pertains to local DSS system only

A Record Represents

• VISTA, PBM database, DSS national pharmacy extract, Fee Basis:– a single prescription or supply for an

individual

• DSS NDE for inpatient/outpatient care: – all prescriptions and supplies for a person

on a particular day

A Record Represents

• DSS production data: – All prescriptions and supplies for a person handled

by a particular production unit on a particular day

– examples of production units: on-site pharmacy, CMOP

Data Format

• VISTA: ASCII

• PBM: You may request SAS, MS Access, or Visual FoxPro

• DSS NDEs at AAC: SAS

• KLF Menu (DSS): Spreadsheet

Access -- Summary

• PBM: Extracts made by PBM staff (www.vapbm.org)

• DSS Rx NDEs: detailed files stored at AAC and accessed through timeshare accounts; some summary data available for free via KLFMenu

• VISTA: Difficult to obtain direct access; easier to request data from local IRMS.N.B.: IRMS staff cannot accept Research funds. They do you a favor by making an extract.

Access -- Reference

• See Table 1 of:MW Smith, G Joseph. Pharmacy Data in the VA Health Care System. Medical Care Research and Review 2003;60(3 Suppl): 92S-123S.

www.herc.research.med.va.gov/Pubs_medcare2.htm

Cost of Obtaining Data

• PBM– Managerial & oversight projects: no charge

– Unfunded pilot studies: usually no charge

– Funded studies: charges for programmer time + optional consulting on study design

– Ask PBM/SHG staff ahead of time

• Other sources– No charge to user; AAC charges billed to VAMC

– KLF Menu: free unless you need to put in your ID # and password

Contents

For PBM and DSS Pharmacy Extract: VIReC research user guide: VHA pharmacy prescription

data. Hines, IL: Veterans Affairs Information Resource Center (VIReC). 2005.

URL:www.virec.research.med.va.gov

For all sources (but somewhat dated): Table 2 in Smith and Joseph Med Care Res Rev article.

Sample of Data Fields

• Medication: drug name, NDC, formulary indicators

• Dispensing: fill date, quantity dispensed, days supplied

• Cost: purchase price (PBM, VISTA) or VA cost including overhead (DSS NDEs, DSS Pharmacy Extract)

Sample of Data Fields

• Patient: SCRSSN; date of birth, gender, age

• Provider: provider ID, provider treating specialty

• Note: Clinical information on related visits/stays can be linked to Rx data using SCRSSN

Co-payments

• VA charges some copayments– Depends on income, disability percentage

– Rules & eligibility levels change year to year

– Rules available on VA internet

• Data sources do not show copayments; they show VA’s expense

• MCRR files could show reimbursement from private insurance, if collected

Unit Costs - 1

• There may be a contract price– Federal Supply Schedule (FSS) – FSS Tier Schedule – Federal ceiling price (“Big 4”) schedule – VA Blanket Purchase Agreement (BPA)

• Price files available on PBM web site (www.vapbm.org)

Unit Costs - 2• There will be discrepancies across sites

– Correct pricing requires *daily* updating of a VISTA price file at each VAMC, which does not occur

– BPAs are specific to individual VAMCs

• National VA formulary may limit use of selected medications – - see PBM website for current formulary, changes

to formulary since 1998, and current rules for particular medications

Choosing a Source: VISTA• Advantages

– Greatest detail on costs, use of care– Access to data not available in extracts

• Disadvantages– Can access data from only the local VAMC– Most often, extracts must be made by IRMS staff

using specialized programs– Requires caution in interpreting differences

across sites

Choosing a Source: PBM

• Advantages– National coverage in one extract– Only source that provides purchase price

(sometimes inaccurate!)– Optional fee-based consulting on pharmacy

data needs & use

Choosing a Source: PBM

• Disadvantages– PBM staff must create the extract– Does not show pharmacy clinic costs

beyond purchase price– Limited clinical and demographic

information

Choosing a Source: DSS National Pharmacy Extract

• Advantages– National coverage – Detailed cost data

• Disadvantages– Limited prescription characteristics– Cost data do not show purchase price

Choosing a Source: DSS Inpatient/Outpatient NDEs

• Advantages– National coverage – Convenient summary cost data by treating

specialty or overall

• Disadvantages– Limited prescription characteristics– Only summary data: no data on individual

prescriptions or supplies

Choosing a Source: DSS Data via KLFMenu

• Advantages– National coverage – Convenient summary cost data – Ease of access and use

• Disadvantages– Only summary data: no data on individual

prescriptions or supplies– Cannot select cases by SCRSSN

Validation Studies: PBM vs. DSS

• Do PBM and DSS data sources contain the same prescription records?

• Study #1: 1,600 patients with hernias in CSP 456. Result: PBM and DSS Pharmacy Extract have >95% concordance in drug names, # scripts, # units dispensed.

Validation Studies: PBM vs. DSS

Study #2: >300 patients with heart disease in CSP 424. Result: PBM and DSS NDE daily summary have very poor concordance in monthly or annual pharmacy costs.

Grouping Prescriptions• DSS sometimes groups two prescriptions into

one record if they are for the same NDC and the same person on the same day

– PBM does not group prescriptions in this way

Other Notes on Pharmacy Data

• KLF Menu provides summary DSS data on pharmacy spending – Pharmacy spending is one element of many DSS

reports

– Level: station, VISN, or nation

– No data on individuals

– Cannot be used to select data on a cohort of individuals

Other Notes on Pharmacy Data

• VA utilization and spending patterns for individual medications is often confidential– E.g.: nationwide prescribing patterns for

branded antipsychotics for patients newly diagnosed with schizophrenia

– Drug manufacturers seek these data to aid in negotiations with VA

– Consult PBM before allowing private firms to see VA pharmacy data

Cautions

• Validation is essential – Fields may have missing or inconsistent

values.– Different sites may complete fields

differently.– Not clear yet whether all prescriptions are

recorded, and how this varies by data source.

Cautions

• Data elements change each year– Written guides become outdated quickly– Crosscheck data elements against printed

information: do you know what each field means?

• E.g.: fill date vs. release date

Validity Checks

• Check data for erroneous values

• Missing values– If possible, fill in values based on consistency checks

(e.g., for gender, age)

– Imputing values adds statistical uncertainty: should account for it (or at least mention if used rarely)

• Inconsistent units– One 50ml bottle could be “50 units” in one record

but “1 unit” in another record

Non-VA Pharmacy Data

• In CSP trials, we typically do not ask patients about non-VA pharmacy– Many VA users get all prescriptions

through VA– VA is likely to be used for most expensive

meds– Over-the-counter and occasional non- VA

prescription use is unlikely to affect total Rx spending much

Non-VA Pharmacy Sources

1. Drug Topics “Red Book” Published annually

Offers “Average Wholesale Price” (AWP), the starting point for Medicaid drug payments

2. Private-sector claims data• Proprietary: must pay for access

Questions on Pharmacy Data?