Clinical health outcome: A health state of a patient resulting from health care.
Cluster: In this study, various regression models were used to test the influence of a specific “cluster” or group of independent or predictor variables on dependent or outcome variables.
Coefficient of variation (CV): The ratio of the standard deviation of a random variable to its mean. The committee uses the CV to compare the degree of variability in Medicare and commercial populations with respect to health care spending and use. Because both the numerator and the denominator of this variable are in the same units, the magnitude of the CV does not depend on the units in which it is measured (e.g., dollars or thousands of dollars).
Cohort: A group of persons who have at least one clinical characteristic in common. This study defined 15 cohorts based on clinical conditions. A single person may appear in one or more cohorts.
Control model: A statistical model that includes all independent predictor variables, except those an investigator especially wants to understand. In this study, the control regression model is adjusted, or controls for length of time beneficiaries are in plans and year of analysis. The effect of other predictors can be calculated by comparing the estimates of models adjusted for clusters 1 through 10 to the control model. In effect, the control model is used to eliminate variation that is of no interest.
Correlation coefficient (Pearson’s and Spearman’s): A measure of the relationship between two variables, indicating how the direction and magnitude of change in one is accompanied by change in another. It varies between +1 (perfect, positive association meaning as one variable increases the other variable increases) and –1 (perfect, negative association meaning as one variable increases the other decreases). It does not, however, indicate that there is a causal relationship between the variables.
Efficiency: Production and allocation of goods and services that generate the most utility for a given set of resources or inputs.
Health care cost: The actual costs of production.
Mean: The average of a group of values.
Median: The value that separates the highest 50 percent of scores on a variable from the lowest 50 percent.
Medicare Part A: Also known as the Hospital Insurance (HI) program, Part A covers inpatient hospital services, skilled nursing facility, home health, and hospice care.
Medicare Part C (Medicare Advantage): Also known as the Medicare Advantage program, Part C allows beneficiaries to enroll in a private plan, such as a health maintenance organization, preferred provider organization, or private fee-for-service plan, as an alternative to the traditional fee-for-service program. These plans receive payments from Medicare to provide Medicare-covered benefits, including hospital and physician services, and in most cases, prescription drug benefits.
Medicare Part D: Part D, the outpatient prescription drug benefit, was established by the Medicare Modernization Act of 2003 (MMA) and launched in 2006. The benefit is delivered through private plans that contract with Medicare: either stand-alone prescription drug plans (PDPs) or Medicare Advantage prescription drug (MAPD) plans.
Outliers: Values for a variable so extreme that they may have undue influence on the resulting values of certain statistics. They can distort the mean value, but the median value remains unaffected.
Percentile: The value of a variable, below which a certain percentage of data points or observations fall. For example, if Medicare spending at the 90th percentile is $10,000 per person, then 90 percent of all observations would be expected to have spending less than $10,000 per person.
Price: The amount paid by insurers and beneficiaries to a provider for health care services.
Quality: The degree to which health care services for individuals and populations increase the likelihood of patient-desired health outcomes and are consistent with current professional knowledge.
Quintile: One-fifth of a sample or population based on division in intervals of a particular variable. In the Acumen Growth Analysis, hospital referral regions (HRRs) were classified into quintiles based on expenditure levels in 1992, such that the same number of HRRs are included in each quintile. Quintiles are generally presented in order from top to bottom.
Regression analysis: A statistical technique for predicting the value of an dependent variable Y as a function of one or more predictor variables X. The resulting predicted value is the expected value used to calculate the residual.
Residual: The difference between the actual observation (e.g., actual spending in a hospital referral region [HRR]) and the expected value (e.g., expected spending) based on a set of predictor variables. For example, adjusting for mortality in an HRR for its age and sex mix, the residual is the difference between actual mortality and predicted mortality.
Total health care spending: What medical providers and suppliers are paid for their services and products, reflecting both price and utilization of health care services.
Utility: Consumer satisfaction or use.
Utilization: The volume or amount of health care services consumed within a given time period.
Value: The excess (or shortfall) of overall health benefit and/or well-being produced net of health care cost.