Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
Abstract
Background and Aims
Acute kidney injury (AKI) is a common and serious condition associated with longer length of hospital stay (LOS) for affected patients; leading to increased morbidity, emotional distress and higher healthcare costs. Implementation of AKI care quality improvement interventions have consistently been shown to reduce patient LOS, suggesting its measurement may serve as a valuable AKI care quality indicator for hospitals. In this study, we set out to explore, for the first time, unwarranted centre variation in case-mix adjusted LOS for patients who develop hospital acquired AKI (HA-AKI) across England.
Method
Analysis was undertaken using a routinely collected national database of patients with biochemically detected AKI, linked with hospitals administrative data. 250,504 HA-AKI episodes were studied in total, across 103 hospitals between 01/01/2017 – 31/12/2018. LOS was defined as the number of days between first AKI alert and discharge. A negative binomial model was used to generate predicted LOS for patients with HA-AKI at each hospital trust using age, sex, diagnosis group, comorbidity score, AKI severity, admission method, month of AKI alert and critical care attendance. Variation in “observed LOS – predicted LOS” values across centres were then compared using a funnel plot. Additional analyses were undertaken to investigate potential associations between centre LOS and readmission rates as well as variation in LOS amongst patients in the most common diagnosis groups.
Results
The overall median LOS following HA-AKI alert was 8 days (Interquartile range, 3-17). There was significant variation seen in “observed - predicted LOS” values across hospitals, with 55/103 trusts classed as outliers (99.8% control limits). When analysis was limited to patients with more severe AKI (stages 2 or 3) at presentation this number fell to 10/103 (99.8% control limits). Risk factors for longer LOS included male sex, advancing age, higher comorbidity score, more severe AKI at presentation, emergency admission and critical care attendance. Patients admitted due to fracture neck of femur were at highest risk of longer LOS and exhibited considerably more variation in “observed - predicted LOS” values compared to patients with congestive heart failure, pneumonia or septicaemia. Just over 1/5 of patients with HA-AKI went on to have an emergency re-admission within 30 days of discharge, with no association observed between a trusts “observed – predicted LOS” values and readmission rates (Figure).
Conclusion
There was considerable centre-variation in LOS for patients with HA-AKI across England. Further interrogation of patient and centre-level factors underlying the observed variation is now necessary to better understand why this variation exists and to inform development of future targeted quality improvement interventions to address this.