The 'Big Five'. Hypothesis generation: a multidisciplinary intervention package reduces disease-specific hospitalisations from long-term care: a post hoc analysis of the ARCHUS cluster-randomised controlled trial

Introduction: long-term care (LTC) residents have higher hospitalisation rates than non- LTC residents. Rapid decline may follow hospitalisations, hence the importance of preventing unnecessary hospitalisations. Literature describes diagnosis- specific interventions (for cardiac failure, ischaemic heart disease, chronic obstructive pulmonary disease, stroke, pneumonia- termed ` big five' diagnoses), impacting on hospitalisations of older community- dwellers, but few RCTs show reductions in acute admissions from LTC. Methods: LTC facilities with higher than expected hospitalisations were recruited for a cluster-randomised controlled trial (RCT) of facility-based complex, non-disease-specific, 9-month intervention comprising gerontology nurse specialist (GNS)-led staff education, facility benchmarking, GNS resident review and multidisciplinary discussion of residents selected using standard criteria. In this post hoc exploratory analysis, the outcome was acute hospitalisations for 'big five' diagnoses. Re-randomisation analyses were used for end points during months 1-14. For end points during months 4-14, proportional hazards models are adjusted for within-facility clustering. Results: we recruited 36 facilities with 1,998 residents (1,408 female; mean age 82.9 years); 1,924 were alive at 3 months. The intervention did not impact overall rates of acute hospitalisations or mortality (previously published), but resulted in fewer 'big five' admissions (RR = 0.73, 95% CI = 0.54-0.99; P = 0.043) with no significant difference in the rate of other acute admissions. When considering events occurring after 3 months (only), the intervention group were 34.7% (HR = 0.65; 95% CI = 0.49-0.88; P = 0.005) less likely to have a 'big five' acute admission than controls, with no differences in likelihood of acute admissions for other diagnoses (P = 0.96). Conclusions: this generic intervention may reduce admissions for common conditions which the literature shows are impacted by disease-specific admission reduction strategies.