Risk-Adjustment
CareScience employs a new generation of risk assessment methodologies developed, tested, and published by the University of Pennsylvania’s Corporate Hospital Rating Project ("CHRP"). CHRP developed a regression-based model to analyze raw outcomes for measured severity and create unbiased, time-stable estimates of provider performance. The model operates at the patient level on the basis of the following sets of variables:

1. Clinical Factors

• Chronic disease/comorbidity
  (cancer)
• Principal diagnosis
• Procedures
• Urgency of admission

2. Demographic Factors

• Age
• Household income
• Gender

3. Patient Selection Factors

• Travel distance to facility
• Payor class
• Mode of arrival at facility
• Transfer disposition

Standardized Outcomes
Measures of standardized outcomes for each provider are estimated from patient-level data by isolating hospital or physician-specific effects in an overall model for each outcome. The statistical technique identifies the marginal effect of patient characteristics and other severity measures on the probability of an adverse outcome. The model estimates the probable provider effect by comparing predicted outcome rates to actual patient outcomes.

Standardized Complication Rates
Complications are defined as events that occur after admission to a hospital. The presence of complications does not necessarily indicate negligence or therapeutic misadventure. Complications are distinguished from comorbidities by applying an expert-panel-based model validated on extensive chart reviews and incorporating Care Management System risk-assessment methodologies.

Standardized Morbidity
This measure identifies the percentage of patients with at least one “severe” complication, defined as having a significant risk of temporary or permanent impairment, unscheduled ICU admissions, and/or a significant increase in length of stay. Standardized morbidity is based on risk assessment and used to separate patient determinants of morbidity from those related to the facility or physician.

Standardized Mortality
This measure is the risk-based predicted rate of mortality for a hospital or physician, as standardized against the patients in the entire comparison database, using the Care Management System methodologies. Standardized mortality is used to separate patient determinants of mortality from those related to the facility or physician.

Standardized LOS
This measure is the risk-based predicted length of stay for a hospital or physician as standardized against the patients in the database using the Care Management System methodologies. Standardized length of stay is used to separate patient determinants of length of stay from those related to the facility or physician.

Standardized Charges (Cost)
This is a summary measure of the risk-assessed treatment intensity. Standardized charges separate price effects from volume (as seen in average charges) so that comparisons of relative treatment intensity can be made. Where hospitals have cost estimates incorporated, these are included in the analysis. CareScience also has the capability of estimating costs for its clients when other costing data are not available.

The standardization model used in Care Management System differs from other "actual vs. expected" methods in multiple important ways:

• The model measures the impact of the process of care by allowing the provider to exert a separate effect from patient characteristics.
• Standardized outcomes (outcome risks) are computed by using patient characteristics to estimate expected outcomes, which are compared to actual outcomes for each provider.
• The model generates standard errors that permit a determination of the precision of the measures.
• The assessment method used in the models presumes that physicians or hospitals compared in the model perform equally until proven otherwise.
• The model controls for patient selection bias, resulting in a fairer evaluation of hospitals or physicians that draw or are chosen by the sickest patients.
• Each model is run separately for each diagnosis and outcome. As a result, CareScience effectively utilizes more than 5,000 different "severity assessment" equations that are specific to a particular disease (e.g. CHF or diabetes) and outcome measure (e.g. mortality or LOS).