A propensity score is the probability that an individual will be assigned to a condition or group, given a set of baseline covariates when the assignment is made. For example, the type of drug treatment given to a patient in a real-world setting might be non-randomly based on the patient's age, gender, geographic location, and socioeconomic status when the drug is prescribed. Propensity scores are used in many different types of observational studies to reduce selection bias. Subjects assigned to different groups are matched based on these propensity score probabilities, rather than matched based on the values of individual covariates. Although the underlying statistical theory behind the use of propensity scores is complex, implementing propensity score matching with SAS^® is relatively straightforward. An output data set of each subject's propensity score can be generated with SAS using PROC LOGISTIC. And, a generalized SAS macro can generate optimized N:1 propensity score matching of subjects assigned to different groups using the radius method. Matching can be optimized either for the number of matches within the maximum allowable radius or by the closeness of the matches within the radius. This presentation provides the general PROC LOGISTIC syntax to generate propensity scores, provides an overview of different propensity score matching techniques, and discusses how to use the SAS macro for optimized propensity score matching using the radius method.

The hospital Medicare readmission rate has become a key indicator for measuring the quality of health care in the US. This rate is currently used by major health-care stakeholders including the Centers for Medicare and Medicaid Services (CMS), the Agency for Healthcare Research and Quality (AHRQ), and the National Committee for Quality Assurance (NCQA) (Fan and Sarfarazi, 2014). Although many papers have been written about how to calculate readmissions, this paper provides updated code that includes ICD-10 (International Classification of Diseases) code and offers a novel and comprehensive approach using SAS^® DATA step options and PROC SQL. We discuss: 1) De-identifying patient data 2) Calculating sequential admissions 3) Subsetting criteria required to report for CMS 30-day readmissions. In addition, this papers demonstrates: 1) Using the output delivery system (ODS) to create a labeled and de-identified data set 2) Macro variables to examine data quality 3) Summary statistics for further reporting and analysis.

Read the paper (PDF) | View the e-poster or slides (PDF)

This presentation gives you the tools to begin using propensity scoring in SAS^® to answer research questions involving observational data. It is for both those attendees who have never used propensity scores and those who have a basic understanding of propensity scores but are unsure how to begin using them in SAS. It provides a brief introduction to the concept of propensity scores, and then turns its attention to giving you the tips and resources you need to get started. The presentation walks you through how the code in the book 'Analysis of Observational Health Care Data Using SAS^®', which was published by SAS Institute, is used to answer how a particular health care intervention impacted a health care outcome. It details how propensity scores are created and how propensity score matching is used to balance covariates between treated and untreated observations. With this case study in hand, you will feel confident that you have the tools necessary to begin answering some of your own research questions using propensity scores.

Epidemiologists and other health scientists are often tasked with solving health problems but find collecting original data prohibitive for a multitude of reasons. For this reason, it is common to instead use secondary data such as that from emergency departments (ED) or inpatient hospital stays. In order to use some of these secondary data sets to study problems over time, it is necessary to link them together using common identifiers and still keep all the unique information about each ED visit or hospitalization. This paper discusses a method that was used to combine five years worth of individual ED visits and five years worth of individual hospitalizations to create a single and (much) larger data set for longitudinal analysis.