Patient Navigation Associated with Decreased 30-Day All-Cause Readmission

Abstract

Abstract Background: Oncology patient navigation (PN) programs have been developed to improve outcomes and reduce disparities. Limited data exist to describe the effect of PN on important clinical outcomes, such as readmission, post cancer diagnosis. Methods: We conducted a retrospective cohort study of adults (≥18 yrs) diagnosed with first primary cancer from Jan 2013-Nov 2014 at a multi-site academic community based cancer institute. “Nearest-neighbor with caliper” propensity-score (PS) matching was used to match PN to similar not navigated (NN) patients. Patients had ≥3 mo follow-up post cancer diagnosis. 30d all-cause readmission (ACR) was any inpatient admission within 30d of discharge from index hospitalization (IH). IH was the first inpatient admission ≤12 mo post cancer diagnosis. When IH ended in transfer to another acute care facility, ACR was calculated from final discharge from acute care. Deaths during IH and discharges against medical advice were excluded from ACR analysis. Multivariable random effects and conditional logistic regression models evaluated associations between PN and length of hospital stay (LOS) and ACR. Results: 10532 patients were eligible (2592 PN; 7940 NN). 4324 PS-matched patients (2162 PN/NN) were included, with balanced demographic (age, sex, race, ethnicity, insurance, marital status, employment, rurality) and tumor characteristics (site, stage, grade, vascular invasion, metastases) and overall health (comorbidity index, inpatient admission 12mo before cancer diagnosis) between groups. There were 4156 total inpatient admissions ≤12 mo post cancer diagnosis. 1190 PN and 958 NN had ≥1 inpatient admission (55% vs 44% P < 0.01). Controlling for principle diagnosis, procedures performed, and charges, LOS was slightly shorter in PN than NN (log LOS β = −0.05 [5% shorter stay] P = 0.05). Among those with eligible IH, 17% PN and 21% NN were readmitted ≤30 d. Controlling for principle diagnosis, procedures performed, and LOS during IH, PN had lower odds of ACR (OR = 0.66 95% CI, 0.48–0.92). Conclusion: In a large, diverse PS-matched cohort, PN patients had shorter LOS and lower odds of ACR. Results suggest PN improves care transition and clinical outcomes for cancer patients after inpatient admission. Additional analyses will explore subgroup differences.