Dartmouth Atlas Project data - onetomapanalytics/Meta_Data GitHub Wiki

Dartmouth Atlas Project data

General description

  1. Database primary purpose - Collect and analyze data to reveal disparities in healthcare resources and usage across the United States. The project provides valuable information to policymakers, analysts, and the media to improve their understanding of the healthcare system's efficiency and effectiveness. This data serves as a foundation for ongoing efforts to improve health and healthcare systems throughout the country.
  2. Overall data type - Health facilities, hospital expenditures, demographics
  3. Dataset type - Longitudinal
  4. Data source - Claims. The main source is Medicare data provided by the Centers for Medicare and Medicaid Services (CMS), but other data sources include the U.S. Census, the American Hospital Association, the American Medical Association, and the National Center for Health Statistics.
  5. Data level - Population level, hospital level
  6. Geographic location of the data collection sites - United States
  7. Sponsor, manager, or home institution - The Dartmouth Institute for Health Policy and Clinical Practice
  8. Date range - Care for chronically ill: 2008-2019; Medicare mortality rates: 1999-2019; Medicare reimbursements: 2010-2019; Post-discharge events: 2008-2017; Medical discharge rates: 1992-2015; Primary care access and quality: 2008-2015; Surgical discharge rates: 1992-2015
  9. Geolocation data - ZIP code, city, state, hospital service areas (HSAs), hospital referral regions (HRRs)
  10. Dates - year
  11. Hospital identifiers - Hospital and system name
  12. Physician identifiers - provider ID
  13. Longitudinal tracking - Track providers
  14. Financial variables - Payments per visit, average co-payment, reimbursements
  15. Clinical areas of interest - All

Applicable methods

  1. Exploratory analysis (1, 2)
  2. Inferential tests (3)
  3. Association methods, such as linear regression models (4), logistic regression models (5, 6)
  4. Difference-in-differences (3)
  5. Decomposition methods (7)
  6. Propensity scores (8)
  7. Structural equation modeling (9)

High-impact designs

  1. Compare the performance of distinct approaches for defining groups of hospitals (10)

  2. Evaluate the degree of regionalization of care in the US by characterizing the activity of hospital systems in different states (11)

  3. Examine the association between local healthcare intensity and drug death rates (12)

  4. Explore health care system factors associated with regional variation in systemic overuse of health care resources (13)

  5. Examine how patients' location of death relates to health care utilization and spending for surviving spouses (14)

  6. Assess the association between health care intensity in the region where physicians practice and their ability to practice high-value care (15)

  7. Evaluate hospital characteristics and economic conditions of communities surrounding hospitals with and without stroke centers (16)

  8. Explore potential imprinting on clinical rotations by describing high- and low-value behaviors among medical students and examining relationships with regional healthcare intensity (17)

  9. Describe practices that joined the Comprehensive Primary Care Plus model and compare hospital service areas with and without CPC+ practices (18)

  10. Determine the source of recent changes in end-of-life Medicare expenditures (19)

Data dictionary

To access Dartmouth Atlas Project data dictionary, click here

Variable categories

  1. Patient demographics (i.e., sex, race)
  2. Medical and surgical discharge rates (discharges per 1,000 Medicare enrollees)
  3. Primary care access and quality (e.g., number of diabetics, number of mammograms, leg amputation)
  4. Hospital and physician capacity (e.g., number of acute care beds, registered nurses, FTE hospital employees)
  5. Post-discharge events (e.g., ambulatory visits, emergency room visits, readmission)
  6. Medicare reimbursements (e.g., total per enrollee, price-adjusted reimbursements, medical equipment)
  7. Mortality (e.g., total and non-HMO mortality)
  8. Care for chronically ill (e.g., death, spending, co-payment)

Linkage to other datasets

  • Linkages can be established for any dataset that might have geolocation data (i.e., ZIP code) or hospital identifiers (i.e., hospital name, system, or ID)