Technical Details

Methodology for Calculating Health Care Costs Back to the Top

This website displays health care costs for common episodes of care. An episode of care is defined as the combination of all health care services associated with a specific treatment or procedure from beginning to end. Grouping these services and procedures together sheds light on the total cost of care.

DATA SOURCE

This analysis uses data from the Maryland Medical Care Data Base (MCDB), a database run by the Maryland Health Care Commission (MHCC) that contains administrative and claims data on Maryland residents insured through public and private commercial health plans. This database captures information on utilization and payment for health care services rendered in all care settings – including hospitals, physician offices, ambulatory surgery centers, pharmacies, and long term care facilities.

The Wear the Cost website uses utilization and payment data on Maryland’s privately insured population, which was obtained from commercial insurers through MCDB’s quarterly data collection process. Data included in analysis is limited to the data from fully-insured health plans that include medical and pharmacy claims for all enrollees. Next year, the website will be updated to include Maryland’s Medicare population, using data provided by the Centers for Medicare and Medicaid Services (CMS). MHCC will further update the website as more years of data become available, to allow for year-over-year trend analyses.

CALCULATING EPISODES OF CARE

To calculate the episodes of care, MCDB claims data were run through the PROMETHEUS Analytics© tool, an episode grouper developed by HCI3 (now part of Altarum Institute). This tool uses a proprietary algorithm to group procedure and diagnosis codes into comparable buckets, or “episodes of care,” and then calculate the combined costs of these episodes. Using this methodology, average costs per episode were calculated for acute care hospitals in Maryland, and were risk adjusted for patient illness burden.

The methodology identifies facility and professional claims with trigger codes for the episodes, and then determines all episode related costs in a look back and look forward timeframe. Only completed episodes are included in the analysis; consequently, procedures occurring in the last quarter of the calendar year are excluded from the annual analysis. Additionally, the Prometheus grouper also applies guardrail filters when defining the unit of measurement which is a completed episode. These filters are:
Upper and lower age limit – Only episodes for patients within the age limit specified for each episode are included.
Lower and upper episode cost limit – The grouper includes low and high trim points that determine whether an episode is an outlier. Any episode that is below or above the outlier trim points is excluded from the analysis.
Coverage gap – Members with coverage gap are excluded from the analysis and only members that have full coverage are included.
Orphan Episodes – Only episodes that are attributable are included. This means episodes that do not have a facility claim either inpatient or outpatient are excluded as orphan episode.

The cost is calculated using the allowed amounts for all services related to the episode, including prescription drugs. Only services relevant to the episode are included. Risk-adjustment is done using multi-part regression models based on patient demographics, historical risk factors, and episode severity, with separate models for typical and potentially avoidable complications. Results of the Maryland analysis were benchmarked by Altarum against results they have from the neighboring states of Pennsylvania and New Jersey, and were found to be similar.

IDENTIFYING POTENTIALLY AVOIDABLE COMPLICATIONS (PACS)

Episode costs were divided into two categories: those related to expected components of a patient’s care (such as office visits, physician consultation and follow-up, surgery, and physical therapy), and those that were due to potentially avoidable complications (PACs). PACs were defined as events that could harm patients and that could be avoided – such as those due to guidelines not put into practice or followed, mistakes made in the clinical setting, and other system failures. Episode costs for expected care and for PACs were then compared across Maryland hospitals to shed light on key differences in cost and quality.

VERIFYING THE DATA QUALITY

MHCC employed a multi-step approach to ensuring data quality for this analysis. Claims data in MCDB was subject to several levels of data quality testing, both upon initial receipt from data submitters and on an ongoing basis. In addition, episode of care calculations run through the PROMETHEUS Analytics tool were carefully checked by staff from MHCC and its analytic vendors. Finally, MHCC shared episode cost calculations and data displays with each hospital prior to publishing its findings, and gave them the opportunity to provide feedback. These data quality standards and practices help to ensure that the information on this website is valid and credible.

Technical Resources Back to the Top

Additional technical information on the methodology for grouping and analyzing episodes of care is available on Altarum’s website at the links below:

For questions related to the methodology, please contact us.