Unplanned hospital admissions: defining a set of primary care sensitive admissions and predicting patients at risk.

Admissions to hospital are an increasing problem for the NHS. Unplanned admissions to hospital are those which are not planned or not from a waiting list. They represent over a third of hospital admissions (4,158,734 emergency admissions in 2003/4). Unplanned admissions are expensive, create uncertainty for those responsible for planning and delivering services and are distressing for patients and their families. This programme includes two linked research studies: The first will tell policy makers and clinicians about the most common reasons for unplanned hospital admission and how costly these admissions are. The research will also explore admissions for conditions that could be treated effectively in primary care and, in consultation with doctors, produce a list of admissions that could be prevented by care outside hospital. This information will help to target future spending and provision of care. The second study will develop and test statistical models, or methods, which will assist primary care clinicians to identify patients at high risk of unplanned admission. This will use primary care medical records and will help both the patient‘s usual GP and nurses, and other colleagues who may not know a patient, to identify people at risk.

Unplanned hospital admissions: defining a set of primary care sensitive admissions and predicting patients at risk

Admissions to hospital are an increasing source of pressure on NHS resources. Unplanned admissions to hospital are those which are not planned or not from a waiting list. They represent 35.5% of hospital admissions (4,158,734 emergency admissions in 2003/4). Unplanned admissions are expensive in terms of resource use, create uncertainty for those responsible for planning and delivering services and are distressing for patients and their families. There is no evidence on the use of routinely available data from GP records to predict those patients at highest risk from unplanned hospital admission. Recent projects in the UK have used either recent hospital admissions or resource intensive assessment procedures to identify potentially at risk individuals. This programme includes three discrete but linked studies:

Study 1. Describing unplanned hospital admissions and identifying primary care sensitive admissions (PCSA).
A descriptive study of routine data on unplanned admissions at population (PCT) level. Will use HES data to identify rates of unplanned admissions including PCSAs (standardised for age and sex) across PCTs. Costs attributable to admission HRGs will also be obtained. Previously identified associated factors including sociodemographic variables will be included in the analysis. The study will use univariable and multivariable logistic regression to identify a proposed set of PCSAs after modeling to control for predictive variables. Assessment of face and content validity in questionnaire based two stage Delphi process to validate proposed set of PCSAs.
This information will define a set of PCSAs which will assist with the targeting of interventions to prevent admission.

Study 2. Development of a prediction rule for identifying patients at risk of PCSA.
Population based retrospective case-control study of patients admitted to hospital as PCSAs. Cases identified from routine NHS data. Primary care medical record data extracted from GP clinical databases. Univariable and multivariable regression modeling will be used to derive a prediction rule.

Study 3. Validation of a prediction rule for identifying patients at risk of PCSA.
Population based retrospective cohort study of patients. Longitudinal follow up for predictor variables and admission to hospital. Data from GP research database. Testing of predictive models and positive predictive values for risk factors identified in study 2.
The clinical risk tool will assist primary care clinicians to identify patients at high risk of PCSA.