NYS SPARCS Public 2014 Dataset: Let's Look at Septicemia, What Can We Learn?
Lisa Szydziak, MS
Northwell Health - Southside Hospital
Simon Sheather, PhD
University of Kentucky
New York State Department of Health provides publicly available de-identified data for the general public and researchers at www.health.ny.gov/statistics/gov referred to as SPARCS data. SEPTICEMIA & DISSEMINATED INFECTIONS cases representing a high volume APRDRG in the 2014 SPARCS dataset were investigated. A statistical analysis enables a comparison of facilities in terms of charges and costs adjusting for race, gender, age, length of stay (LOS), admit type, disposition, Severity of Illness (SOI), Risk of Mortality (ROM), payer, procedure, high cost, admit day of week and discharge day of week. In modeling charges/costs, a single model was found not to fit and models differed across 2 groups: "Patients that die in 1 day" and "Patients that do not die in 1 day". A Logistic Model was produced for examining the probability of death in "Patients that do not die in 1 day" adjusting for gender, age, LOS, admit type, disposition, SOI, ROM, facility, payer, procedure and discharge day of week. Modeling of publicly reported data is very powerful, it could support policy makers, administrators and patients in understanding the complexity of healthcare costs and quality.