Using health administrative data to model associations and predict hospital admissions and length of stay for people with eating disorders

BackgroundEating disorders are serious mental illnesses requiring a whole of health approach. Routinely collected health administrative data has clinical utility in describing associations and predicting health outcome measures. This study aims to develop models to assess the clinical utility of health administrative data in adult eating disorder emergency presentations and length of stay.MethodsRetrospective cohort study on health administrative data in adults with eating disorders from 2014 to 2020 in Sydney Local Health District. Emergency and admitted patient data were collected with all clinically important variables available. Multivariable regression models were analysed to explore associations and to predict admissions and length of stay.ResultsEmergency department modelling describes some clinically important associations such as decreased odds of admission for patients with Bulimia Nervosa compared to Anorexia Nervosa (Odds Ratio [OR] 0.31, 95% Confidence Interval [95%CI] 0.10 to 0.95; p = 0.04). Admitted data included more predictors and therefore further significant associations including an average of 0.96 days increase in length of stay for each additional count of diagnosis/comorbidities (95% Confidence Interval [95% CI] 0.37 to 1.55; p = 0.001) with a valid prediction model (R-2 = 0.56).ConclusionsHealth administrative data has clinical utility in adult eating disorders with valid exploratory and predictive models describing associations and predicting admissions and length of stay. Utilising health administrative data this way is an efficient process for assessing impacts of multiple factors on patient care and predicting health care outcomes.

Automated summary

This study aims to develop models to assess the clinical utility of health administrative data in adult eating disorder emergency presentations and length of stay. The results show that health administrative data can describe associations and predict admissions and length of stay. Utilizing health administrative data in this way is an efficient process for assessing impacts of multiple factors on patient care and predicting health care outcomes.