We used the Delphi method to identify factors indicative of high intensity care (Table 1). We then classified patients into three intensity groups based on probability of receiving procedures and length of stay. We created a logistic regression model with intensity (high vs. medium/low) as the outcome and patient-level covariates, (age, sex, race, urban/rural, comorbidities, injury severity score (ISS), injury mechanism, distance to nearest trauma center). Post-analysis linear predictions determined IoC scores for patient characteristics.
We used 2013 New York State Medicare Provider Analysis and Review claims from The Centers for Medicare and Medicaid Services. Also incorporated were Master Beneficiary Summary File demographics and address level information from the Vital Status File. Included beneficiaries were 65 years old, treated in New York state, had blunt mechanism of injury with ISS 9, and survived to hospital admission.
A total of 18,486 unique, index ED admissions were analyzed. Males were predicted to have higher IoC than females; intensity was higher for non-white beneficiaries vs white. Male, non-white, 65-84, and urban has the highest probability of IoC based on combined demographic characteristics. For injury factors, ISS >15, non-fall injury mechanisms, and living closer to a trauma center increased predicted IoC.
We conclude that many non-modifiable factors contribute to the IoC patients receive after sustaining a blunt traumatic injury. These differences may serve as starting points when evaluating hospital-level practices regarding patient care. Defining variability in IoC may help hospitals understand how they compare to peer hospitals. Future efforts will investigate the impact of IoC on outcomes to inform discussions on optimal levels of IoC.
We are presenting this research at the Academy Health annual meeting in June 2018 in Seattle, WA. Click here to download a copy of the poster.
This study is funded by grant R01HS023614 from the Agency for Healthcare Research and Quality.