Acute care in medical inpatients: risk factors, care models and outcomes

Efficient and effective management of medical patients poses a major challenge to clinicians, administrators and policy makers. Significant achievements in efficiency have largely absorbed the impact of population ageing over the past two decades, but an emphasis on efficiency might compromise functional outcomes and increase readmission rates in this vulnerable patient group. AIMS: The aim of this research was to improve understanding of the characteristics and outcomes of general medical patients, to inform improvements in care. Specific aims were to: describe the characteristics and outcomes of a large, representative cohort of general medical patients; measure the impact of an interdisciplinary care model on outcomes, with a focus on length of stay, functional outcomes, and readmissions; determine whether beneficial outcomes were confined to particular patient subgroups; and investigate risk factors for adverse functional outcomes and readmissions. METHODS: The principal study was a pseudo-randomised controlled trial of interdisciplinary care for general medical patients, enrolling all consecutive general medical patients over 24 weeks. The intervention included greater allied health staffing, consistent interdisciplinary care teams, consistent team communication and explicit discharge planning. Control patients received usual care from a consistent medical team with allied health referral. Baseline data were collected prospectively at admission from nursing and medical assessments. Length of stay, discharge destination and diagnosis were collected at the time of discharge. Functional status was obtained from nursing reports of preadmission, admission and baseline dependency in six activities of daily living. Readmissions and cost-weight (a severity surrogate) were obtained from the hospital information systems. Mortality at 6 months was obtained from the state-wide death registry. Characteristics and outcomes were described for all patients. Outcomes were compared between the intervention and control groups. Subgroup analyses explored the impact of age, frailty and residential aged care. Findings regarding functional status and readmissions were then examined in greater detail using multivariate analyses, adjusting for recognised confounders identified through systematic literature reviews. RESULTS: 1538 patients were enrolled, 746 in the control and 792 in the intervention group. Two thirds were aged over 65 years, of whom 19% came from residential aged care (RAC) and 65% met frailty criteria. Mean length of stay was 7.5 days, 5% of patients died in hospital, 15% died within 6 months of discharge and 30% were readmitted to hospital. Nearly 30% had pre-existing impairments in one or more activity of daily living, 42% had new functional deficits at the time of admission, 5% had new or further decline in hospital, and 23% still had greater functional impairments than usual at the time of hospital discharge. The intervention arm demonstrated a trend to reduction in length of stay (0.5 days; 95% CI -0.2, 3.2; p=0.18) and a significant reduction in in-hospital functional decline (3.2% versus 5.5%; p=0.04), consistent with previous studies. Unexpected findings were a reduction in in-hospital mortality (3.9% versus 6.4%; p=0.03) and a low rate of in-hospital decline. Older age, frailty markers and RAC origin were strongly associated with poorer outcomes. Marked mortality benefits from the intervention were seen in the RAC subgroup. Amongst patients aged over 65, we observed a high rate of functional decline occurring before hospital admission (pre-hospital decline) and a low rate of in-hospital decline. Predictors of pre-hospital in-hospital and post-hospital functional changes differed. In order to promote improved functional recovery in older patients, the interdisciplinary care model was refined to include a more pro-active approach to early functional rehabilitation. This functional independence model was implemented in a pilot controlled trial, and demonstrated greater functional improvement in the intervention group (median increase in modified Barthel index 8.5 versus 3.5 in the control group, p=0.025). Using confounders identified in a systematic literature search, multivariate techniques were used to explore predictors of hospital readmissions and examine alternative approaches to dealing with deaths in readmission analysis. This analysis confirmed a previous hospital admission as a powerful predictor of future readmission. A prospective observational study of 142 older medical patients who had already had two recent hospital admissions revealed that chronic diseases, malnutrition and depression were significant predictors of further readmission. This may inform future post-hospital interventions. CONCLUSIONS: These studies advance our understanding of the complex interplay between characteristics, care models and outcomes in medical patients. They highlight some of the methodological challenges involved in health services research, including the potential interactions between outcomes, which may influence study results and interpretation. They contribute to design of future studies by defining risk factors to consider as confounders and/or potential intervention targets. Importantly, the principal study has resulted in a major permanent change in clinical practice at the RBWH, has informed the design of other general medical services and has improved research capacity. Our ongoing multidisciplinary research program will continue to improve the quality of care for a large and vulnerable group of patients