P75. Pathway analysis in spine surgery: a model for evaluating length of stay

Abstract

BACKGROUND CONTEXT Spine surgeries utilize significant hospital resources, with length of stay (LOS) being a major cost driver. Predicting inpatient hospitalization is essential for the operational success of a hospital. Previous studies identifying pertinent factors that impact LOS in spine surgery are limited. PURPOSE To identify pre-, intra-, and post-operative factors associated with LOS in patients undergoing spine surgery. STUDY DESIGN/SETTING Retrospective review of spine surgery patients at a Canadian tertiary academic teaching hospital. PATIENT SAMPLE A total of 415 spine surgery patients. OUTCOME MEASURES LOS (calculated from the date of surgery to the date of discharge, with no accommodation for partial days). METHODS Institutional research ethics board (IRB) approval was obtained. Patients undergoing surgery by two orthopedic spine surgeons between October 2014 to October 2016 were identified using billing codes. Chart data were extracted on 16 surgical and nonsurgical factors by two independent reviewers. Factors collected included age, sex, BMI, ASA class, insurance claim, transfusion, preop medication use (antidepressant, neuroleptic and narcotic), surgical procedure (sub-grouped into 10 categories: single level transforaminal lumbar interbody fusion (TLIF), 2-level TLIF, discectomy, laminectomy, posterior c-spine, minimally-invasive stabilization, anterior lumbar interbody fusion (ALIF), anterior cervical decompression and fusion (ACDF), major deformity, and other (hematoma, removal of hardware, incision and drainage), Charlson Comorbidity Index, preop hemoglobin, surgical time, initial versus revision surgery, intraop analgesic dose, and LOS. Multiple quasi-poisson regression was used to identify factors associated with LOS. RESULTS There were 378 patients with full data included in the final analysis. Average age of the cohort was 52 years (16.3 sd, range 16-88) and 51% were female. Thirty percent (n=123) of the procedures were 1-level TLIF, while 6% (n=25) were 2-level TLIF. Other prominent procedures included discectomy (18%), laminectomy (15%) and posterior c-spine (7%). The average LOS for the entire cohort was 5.9 ±11.0 days and was right skewed: 58 subjects (14%) had 0 days in hospital and 59 subjects (14%) stayed longer than 1 week. A poisson model was built to predict LOS with surgical procedure while controlling for the other study variables. Compared to patients undergoing one- or two-level laminectomy, those that had single level or two-level discectomy stayed less than half as long (IRR=0.44, 95% CI: [0.24, 0.80]). Posterior c-spine patients (IRR=4.92, 95% CI: [2.92, 8.28]), major deformities (IRR=3.07, 95% CI: [1.85, 5.10]), ACDFs (IRR=2.24, 95% CI: [1.21, 4.15]), and 1-level TLIFs (IRR=1.50, 95%CI: [1.03, 2.16]) all had significantly longer hospital stays. Amongst the other study variables, ASA class 4 patients had a significantly longer stay (IRR=4.3), and patients that received a transfusion also stayed longer (IRR=2.7). All other factors evaluated were not found to be independently predictive of LOS. CONCLUSIONS The LOS within this cohort is strongly correlated with the procedure performed. Medical comorbidities and other factors did not significantly impact LOS except in very ill patients. Future pathway improvement should focus on procedure-specific, post-operative rehabilitation protocols and appropriate pre-operative optimization of severe medical illness. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs.