HOSPITAL-BASED CARE COORDINATION INTERVENTIONS: EVALUATION OF POST-DISCHARGE UTILIZATION THROUGH CAUSAL INFERENCE METHODS

Background: The Johns Hopkins Hospital (JHH) and Bayview Medical Center (JHBMC) deployed new risk screening and care coordination programs to improve patient transitions from acute to community settings. Referrals for these interventions were made through non-randomized means, complicating unbiased program evaluation. Methods: Prospective multicenter observational study comparing 82,786 patients referred to post discharge home care coordination intervention against standard post discharge follow-up from January 1, 2016 to September 31, 2019. First, evaluating similarities or differences in patient, community, and hospital risk factors for patients exposed and not exposed to home care. Followed by measuring the association between socioeconomic disadvantage (Area Deprivation Index) and referral, receipt of the home care intervention, and the post-discharge readmission outcomes. Lastly, comparing generalized estimating equations logistic regression and propensity score weighting approaches in controlling underlying patient differences while estimating treatment effect of the home care intervention. Results: Research Aim 1 - Patient and hospital-level characteristics influencing higher odds of referrals to a home care coordinator included: patients aged 65 years and older, females, African-Americans, managed care insurance coverage, higher counts of comorbidity indicators, longer hospital length of stay, positive ESDP risk screen, and higher baseline risk of readmission. Comparatively, African-American patients were associated with higher odds of conversion and follow-up contact by a home care coordinator. Research Aim 2 - The Area Deprivation Index is significantly associated with a patient’s likelihood of being referred and converting with home care services, and odds of unplanned readmissions within 30-days of discharges. Research Aim 3 – Propensity score adjustment methodology controlled for patients in patient, hospital, and community factors between exposed and non-exposed groups (patients referred, but did not receive home care) yielding a slight protecting effect for home care follow-up against unplanned 30-day readmission (aOR = 0.98, p < 0.05). Conclusions: These findings demonstrate the utility of the Area Deprivation Index to measure socioeconomic deprivation and association with hospital interventions like converted home care services and patient outcomes like unplanned 30-day readmissions. Propensity score modelling demonstrated home care has a statistically significant but minor association in lower odds of readmission, sensitive to unmeasured confounding