Natalie Flaks-Manov

Predicting 30-day readmissions with preadmission electronic health record data.

Background:Readmission prevention should begin as early as possible during the index admission. Early identification may help target patients for within-hospital readmission prevention interventions. Objectives:To develop and validate a 30-day readmission prediction model using data from electronic health records available before the index admission. Research Design:Retrospective cohort study of admissions between January 1 and March 31, 2010. Subjects:Adult enrollees of Clalit Health Services, an integrated delivery system, admitted to an internal medicine ward in any hospital in Israel. Measures:All-cause 30-day emergency readmissions. A prediction score based on before admission electronic health record and administrative data (the Preadmission Readmission Detection Model—PREADM) was developed using a preprocessing variable selection step with decision trees and neural network algorithms. Admissions with a recent prior hospitalization were excluded and automatically flagged as “high-risk.” Selected variables were entered into multivariable logistic regression, with a derivation (two-thirds) and a validation cohort (one-third). Results:The derivation dataset comprised 17,334 admissions, of which 2913 (16.8%) resulted in a 30-day readmission. The PREADM includes 11 variables: chronic conditions, prior health services use, body mass index, and geographical location. The c-statistic was 0.70 in the derivation set and of 0.69 in the validation set. Adding length of stay did not change the discriminatory power of the model. Conclusions:The PREADM is designed for use by health plans for early high-risk case identification, presenting discriminatory power better than or similar to that of previously reported models, most of which include data available only upon discharge.

Functional status before and during acute hospitalization and readmission risk identification.

BACKGROUND Recent efforts to prevent readmissions are increasingly focusing on early identification of high-risk patients. OBJECTIVE To test whether information on functioning during hospitalization contributes to the ability to accurately identify older adults at high risk of readmission beyond their baseline risk. DESIGN Prospective cohort study. SETTING Internal medicine wards at 2 medical centers. PATIENTS Five hundred fifty-nine community-dwelling older adults (aged ≥70 years) discharged to their homes. MEASUREMENTS Data on unplanned 30-day readmissions were retrieved from electronic health records. Data on at-admission activities of daily living (ADL) and in-hospital ADL decline were collected using validated questionnaires. Multivariate logistic regression was used to model the association between functioning and readmission controlling for known risk factors. RESULTS Higher in-hospital ADL decline was significantly associated with readmission (odds ratio for each 10-point decrease in ADL = 1.32, 95% confidence interval = 1.02-1.72) but did not contribute to the overall discrimination of the model, as compared with the at-admission data (C statistic = 0.81 for each model). Identifying high-risk (10th highest percentile) patients by the at-admission model did not detect 7/55 (12.7%) of patients who would have been categorized as high risk if risk identification was postponed to the discharge date and included data on in-hospital ADL decline. CONCLUSIONS The study highlights the ability to identify patients at high risk for readmission already early in the index hospitalization using data on functioning, nutrition, chronic morbidity, and prior hospitalizations. Nonetheless, at-discharge functional assessment can detect additional patients whose readmission risk changes during the index hospitalization. Journal of Hospital Medicine 2016;11:636-641.

Health information exchange systems and length of stay in readmissions to a different hospital

BACKGROUND Readmission to a different hospital than the original discharge hospital may result in breakdowns in continuity of care. In different-hospital readmissions (DHRs), continuity can be maintained when hospitals are connected through health information exchange (HIE) systems. OBJECTIVE To examine whether length of readmission stay (LORS) differs between same-hospital readmissions and DHRs, and whether in DHRs the LORS differs by the availability of HIE. DESIGN A retrospective cohort study of all internal medicine 30-day readmissions in 27 Israeli hospitals between January 1, 2010 and December 31, 2010. SETTING Clalit Health Services-Israels largest integrated healthcare provider and payer. POPULATION Adult Clalit members (aged 18 and older) with at least 1 readmission during the study period. METHODS A multivariate marginal Cox model tested the likelihood for discharge during each readmission day in same-hospital readmissions (SHRs), DHRs with HIE, and DHRs without HIE. RESULTS Of the 27,057 readmissions, 3130 (11.6%) were DHRs and 792 where DHRs with HIE in both the index and readmitting hospital. Partial continuity (DHRs with HIE) was associated with decreased likelihood of discharge on any given day compared with full continuity (SHRs) (hazard ratio [HR] = 0.85, 95% confidence interval [CI]: 0.79-0.91). Similar results were obtained for no continuity (DHRs without HIE) versus full continuity (HR = 0.90, 95% CI: 0.86-0.94). The difference between DHRs with and without HIE was not significant. CONCLUSIONS The prolonged LORS in DHRs versus SHRs was not mitigated by the existence of HIE systems. Future research is needed to further elucidate the effects of actual use of HIE on length of DHRs. Journal of Hospital Medicine 2016;11:401-406.

Predictors of Persistent Carbapenem-Resistant Enterobacteriaceae Carriage upon Readmission and Score Development

BACKGROUND Carriers of carbapenem-resistant Enterobacteriaceae (CRE) are often readmitted, exposing patients to CRE cross-transmission. OBJECTIVE To identify predictors of persistent CRE carriage upon readmission, directing a risk prediction score. DESIGN Retrospective cohort study. SETTING University-affiliated general hospital. PATIENTS A cohort of 168 CRE carriers with 474 readmissions. METHODS The primary and secondary outcomes were CRE carriage status at readmission and length of CRE carriage. Predictors of persistent CRE carriage upon readmission were analyzed using a generalized estimating equations (GEE) multivariable model. Readmissions were randomly divided into derivation and validation sets. A CRE readmission score was derived to predict persistent CRE carriage in 3 risk groups: high, intermediate, and low. The discriminatory ability of the model and the score were expressed as C statistics. RESULTS CRE carrier status persisted for 1 year in 33% of CRE carriers. Positive CRE status was detected in 202 of 474 readmissions (42.6%). The following 4 variables were associated with persistent CRE carriage at readmission: readmission within 1 month (odds ratio [OR], 6.95; 95% confidence interval [CI], 2.79-17.30), positive CRE status on preceding admission (OR, 5.46; 95% CI, 3.06-9.75), low Norton score (OR, 3.07; 95% CI, 1.26-7.47), and diabetes mellitus (OR, 1.84; 95% CI, 0.98-3.44). The C statistics were 0.791 and 0.789 for the derivation set (n=322) model and score, respectively, and the C statistic was 0.861 for the validation set of the score (n=152). The rates of CRE carriage at readmissions (validation set) for the groups with low, intermediate, and high scores were 8.6%, 38.9%, and 77.6%, respectively. CONCLUSIONS CRE carrier state commonly persists upon readmission, and this risk can be estimated to guide screening policy and infection control measures.