Background In our randomized trial, the """"""""Re-Engineered Hospital Discharge"""""""" (RED) with ten mutually reinforcing components delivered using a tool called the """"""""After Hospital Care Plan"""""""" (AHCP) reduced the 30 day rehospitalization rate by 30%. The main result is now published in the Annals of Internal Medicine. RED is accepted as a National Quality Forum """"""""Safe Practice"""""""" (SP11) for all patients being discharged from the hospital. We have received many requests inquiring about 1.) The effectiveness of our intervention among various subgroups, 2.) The relative contributions of discharge advocate and the pharmacist's follow up call, and 3.) a prediction model for risk stratification for testing the effects of the intervention on high risk groups. An email we received today states: """"""""I am now the medical director of a Medicaid and Medicare-Medicaid health plan in Michigan. The tool that I especially would be interested in hearing more about is the predictive modeling tool. With limited resources, our case managers have to do a really good job at stratifying the hospitalized members so that they only engage with a limited few that are especially high risk for readmission"""""""" Goal: Perform a complete analysis of the 1,008 discharges of patients enrolled in the Re-Engineered Discharge trial, focusing on the risk, i.e. the probability of a readmission within 30 days after any discharge. We will also estimate the effects of RED for various sub-groups and develop prediction models to identify high risk patients for rehospitalization who are also likely to benefit from the intervention. Methods Since patients may have more than one discharge, the statistical analysis should take into account possible correlation among repeated rehospitalizations for one person. Hence we will treat repeated discharges for a patient as a cluster and estimate a mixed effects logistic regression model using the """"""""lmer"""""""" function in the """"""""lme4"""""""" package developed by Bates et al available in the free statistical software, R, version 2.8.1 . Threshold risk score will be determined using the estimated effect sizes and calculated intervention costs. The best performance model for each sub-group and the risk categories which will benefit will be chosen using net benefit analysis and the software by Tobias Sing, Oliver Sander, Niko Beerenwinkel and Thomas Lengauer (2007). """"""""ROCR: Visualizing the performance of scoring classifiers"""""""". R package version 1.0-2. http://rocr.bioinf.mpi-sb.mpg.de/ Outcomes: We will publish five papers as listed 1.) risk scores and Intervention effect;2.) attributalble contributions of discharge advocate and the pharmacist;3,4 &5.) Gender, Homeless and high utilization as risk factors.
We have developed the Re-Engineered Hospital Discharge (RED), including ten mutually reinforcing components that has been accepted as a National Quality Forum Safe Practice to facilitate the high transition from hospital to home. This project will allow further detailed analyses of our data to determine the effectiveness of the RED. This will also help us to determine to whom to best market the RED toolbox
| Chetty, Veerappa K; Culpepper, Larry; Phillips Jr, Robert L et al. (2011) FPs lower hospital readmission rates and costs. Am Fam Physician 83:1054 |
