DISEASE/HEALTH MANAGEMENT WORKING GROUP

The ISPOR BOOK:
Reliability and Validity of Data Sources for Disease and Health Management:

Section 1: Reliability and Validity of Claims and Medication Databases
Section Co-editors: Renée JG Arnold and Iftekhar Kalsekar

Chapter Number	Chapter Subject	Suggested Author(s)
	Overview of the Issues	Renée Arnold (lead); Iftekhar Kalsekar
	Claims databases a. Process and types of claims databases Commercial Plan-based (multiple versus single) Employer-based Health Care Organization (single entity versus system) Public Medicare Different types of files Medicaid Individual states Vendors with multiple-state databases b. Characteristics Overview of the type of files available in claims databases Outpatient files Inpatient files Prescription files Eligibility files 2. Details of the outpatient files: A description of the following fields in outpatient files will be provided and common issues associated with these fields will be discussed: Date of service Diagnosis codes (ICD and others) Procedural codes (CPT codes) Amount billed and amount paid Provider identification 3. Details of the inpatient/hospital files: A description of the following fields in inpatient files will be provided and common issues associated with these fields will be discussed: Date of admission and discharge Diagnosis codes (ICD and others) Procedural codes (CPT codes) DRG codes Amount billed and amount paid Provider identification 4. Details of the prescription files: A description of the following fields in prescription files will be provided and common issues associated with these fields will be discussed: Date of prescription fill Quantity supplied Days supply NDC codes Amount billed and amount paid Pharmacy provider information Prescriber information 5. Details of the eligibility files: A description of the eligibility files will be provided and common methods to determine eligibility within specified time frames will be provided	Iftekhar Kalsekar (lead); Sanjeev Balu; Lisa Mucha
	Scope of claims databases Identifying people with various medical conditions Highly prevalent diseases What to do if the sample size is too small or unusually large Challenges in identifying specialty care (i.e., oncology) Handling eligibility and enrollment data Continuous cohorts across time, or not Continuous eligibility not always optimal Matching or profiling unique individuals Propensity scores or other methods Episodes of care Defined manually or by software package Adherence studies Accurately identifying a prescription Multiple claims on same day Overlapping claims for same drug Overlapping claims for same class of drugs Accurately identifying days supply Defining number of days to define gap in therapy What to do when working with classes where PRN use common MPR calculations Burden of illness studies Incidence/prevalence-based COI Accurate characterization of expenditure Multiple claims for same day, same service Is a 0 value a correction claim or an actual expenditure? Negative expenditures: keep or discard? How to handle outliers Assessment of costs over time Is annualization the best practice? Inflating using medical care component of CPI Assigning cost to the disease Assignment of claims with multiple diagnoses Reliability and validity issues Basic Data Checks Missing data blocks Pregnant males Female prostate exams Geriatric deliveries Shifting age and gender Other evidence of identifier mismatch Enrollment Dual coverage Missing/erroneous effective and/or termination dates Claims Post mortem drug fills and encounters Prenatal infant encounters Duplicate claims Clinical value / claim discrepancies Pharmacy Duplicate fills Date reliability Solutions	Tao Fan (lead); Ifti Kalsekar; Lisa Mucha; Bijal Shah; Sanjeev Balu; Hans Petersen
4.	Using claims to guide decision-making (e.g., Establishment of DM Programs) Introduction Current uses of claims data in decision-making (brief) Pitfalls of claims data Health Policy Identifying populations at risk Shaping health policy Evaluating effectiveness Disease Management Programs Choosing target programs Identifying outcomes of interest Continual program improvement Evaluating effectiveness Pay for Performance – Provider Profiles Setting parameters and performance levels Service/drug utilization Provider profile development Using data to improve patient outcomes Evidence Based Medicine Defining appropriate care Developing consensus-based outcomes Case management and care coordination Monitoring practice guideline use Demonstrating program outcomes Future Directions	Heidi Waters (lead)
5.	Retrospective drug utilization review as a mechanism for generating/evaluating/benchmarking DM data State of the field Definition and scope of a retrospective drug utilization review study Definition and purposes of benchmarking DM data Summary of data sources, diseases evaluated (chronic vs. acute), purpose of studies Approaches used by authors Methodological issues when dealing with Rx databases National Drug Codes (NDC) scheme, ICD-9 codes Multum database Measures (PPPM, mention MPR) Identification of prescription/ supply* Coding accuracy, data problems (missing values, out-of-range, $0 claim, duplicate claims)--to be covered in chapter 3 Specificity of drugs (aspirin vs. antiretrovirals) Biases due to over-the-counter medications Econometric modeling of costs* (here or in the statistical chapter?) Medicaid and now Medicare databases Private providers' perspective Study designs for retrospective DUR studies*: cohort, case-control, nested case-control Continuous eligibility of patients (especially for Medicaid data) Controlling for comorbidities – Charlson index to control for confounding Methodological issues specific to retrospective DUR FDA Approved Drug-Disease Indication- determining approved use Measuring actual use from pharmacy claims data % of drugs actually dispensed average number of drugs per physician encounter Comparing actual use and approved use Comparing actual use and treatment guideline recommendations Interventions based on DUR data – physician education, mailed letters etc. What works and what doesn’t. Future venues	Mireya Diaz (lead); Bijal Shah; Nneka C. Onwudiwe