Key points covered in risk adjusting the U.S.

population for poor outcomes related to COVID-19 infection

Infection Models

Positive Tests and mortality
in each 3-digit zip code (ZIP3) region

Total Population

2010 U.S. Census counts projected
forward to 2020 counts (by age and
gender cohorts) for each 3-digit zip
code (ZIP3) region

Risk-Adjustment Methodology

How anonymized patient-level claims data
mapped to CMS HCCs risk-adjusts population
for unfavorable underlying conditions

PurpleLab Claims Repository

~70% census of annual U.S. medical claims
anonymized patient level data (de-identified
according to HIPPA safe harbor rules)


PurpleLab’s medical terminology platform
has ICD9 CM & ICD10 CM codes mapped to
CMS Hierarchical Condition Categories (HCCs)

Risk Adjusted Population

Projected 2020 population counts of Low,
Moderate, High and Severe risk cohorts for
each ZIP3 region (by age and gender cohorts)

Capacity-to-Treat Methodology 

Counts as basis for risk-adjusted model of “demand” for hospitalization, ventilation and risk of mortality relative to “supply” of capacity as measured across 4 measures of capacity: (i) Hospital Total Beds; (ii) Hospital ICU Beds; (iii) Physicians with experience in caring for ventilator dependent patients; and (iv) Respiratory Therapists


Sensitivity analyses for increases in capacity-to-treat. Unretiring physicians (+15%). Adding beds (+10%, 25% and 50%). Adding ventilatory capacity (really adding ICU beds +25%, 50% and 100%).

Key Findings

Insights from the data about at risk parts of the US