Factors Associated With Exceeding Predicted Hospital Stays in Pediatric Trauma Patients.

Background: The Center for Medicare and Medicaid Services (CMS) calculates predicted length of stay (LOS) for specific diagnosis-related groups and ties it to reimbursement. The Injury Severity Score (ISS) and the presence of non-accidental trauma (NAT) have been associated with longer hospital stays in pediatric trauma patients. To what extent do those and other factors in the pediatric trauma patient impact the CMS-predicted LOS is unknown. This study aims to identify factors in our patients where practice change may reduce LOS.

Methods: A retrospective chart review was conducted of 1005 pediatric trauma patients admitted to our urban, academic Level 1 Trauma Center from January 2018 to December 2020, who met inclusion criteria. Patient data were analyzed with a significance given of P < .05.

Results: A total of 177 (17.6%) patients exceeded their CMS-predicted LOS. These patients had greater admissions to the ICU (56.1% vs 13.8%; P = .000), confirmed-NAT (11.6% vs 3.3%; P = .000), child protective service involvement (33.1% vs 21.3%; P = .001), intubations (36.0% vs 5.4%; P = .000), ventilator days (3.66 vs 0.23; P = . 000), transfusions (26.7% vs 3.0%; P = .000), and discharges to inpatient rehab (21.5% vs 1.6%; P = .000). They presented with a higher ISS (16.0 vs 7.0; P = .000). Etiologies associated with excessive LOS included motor-vehicle-crashes (22.1% vs 12.9%; P = .001), motor-vehicle vs pedestrian (10.5% vs 6.5%; P = .032), and suspected abuse (14% vs 5.1%; P = .000). Patients with excessive LOS had a higher likelihood of brain injury (48.8% vs 21.3%; P = .000) and internal organ injury (22.7% vs 6.9%; P = .000). They had more imaging studies (2.01 vs 1.51; P = .000), including more CT (68.6% vs 49.3%), MRI (22.7% vs 6.8%), and ultrasound (37.2% vs 22.6%) use. Patients who exceeded LOS underwent more surgical procedures (1.2 vs 0.6; P = .000), with a longer time from admission to operating room (1.92 vs 0.76 days; P = .001) and longer operative times (113.1 vs 40.0 minutes; P = .000). They more frequently required intervention by general surgery (5.8% vs 2.1%), neurosurgery (16.9% vs 1.6%), or multiple surgical teams (15.1% vs 3.7%). They had lower rates of intervention by other surgical subspecialties.

Conclusions: A proportion of admitted pediatric trauma patients exceeded their CMS-predicted LOS, influencing hospital reimbursement. Our study identifies factors associated with excessive LOS related to patient demographics, trauma etiologies, and inpatient courses. These factors, particularly concerning patient care, should be considered to improve LOS predictions and to reduce actual LOS.