Avoiding Unnecessary Medical Interventions

Background

Every year there are news reports about the pressures facing our healthcare systems – in the UK and globally. Pressures come from a variety of sources including an ageing population and increasing prevalence of chronic conditions. Admitting people to hospital can lead to lengthy hospital stays, increased frailty and loss of independence, and risk of hospital acquired infections.

The Challenge

The questions GPs face is: When is the right time to admit patients to hospital? How can we avoid adding more pressure and avoidable health complications to a system already under strain?

Many GP Surgeries across the UK have implemented software which identifies patients at high-risk of emergency admission to hospital, by calculating a "risk score" for every individual patient, based on previous admissions, underlying conditions and medicaitons. This intervention – called predictive risk stratification - allows GPs to identify people who may benefit from an early intervention in order to prevent unplanned (emergency) admissions to hospital. 

Man with a cow

At the point at which the Welsh Government were due to implement predictive risk stratification software (PRISM) across all GP surgeries in Wales, Professor Snooks and her team won funding to run a trial to evaluate implementation of the software. The aim of the trial was to ascertain the impact of this intervention on patient care, costs and health outcomes.

The method

Using the Secure Anonymised Health Data (SAIL Databank) Professor Snooks and her team ran a clinical trial across 32 GP surgeries in the Swansea University Health Board Area, one of the largest trials ever run in the UK, involving 230,000 people registered with a participating GP in the Health Board Area.

This was done by randomly allocating clusters of GP surgeries to the week of the year when they would receive the software this is called a Stepped Wedge Trial. The anonymised health data, and self reported quality of life outcomes for the individuals registered to different GP practices across the health board were monitored for the following year to see what effects the intervention would have on hospital admissions, use of other services, health outcomes and costs.

The Results

PRISM implementation increased use of health services: emergency hospital admission rates by 1%; emergency department (ED) attendance rates by 3%; outpatient visit rates by 5%; the proportion of days with recorded GP activity by 1% and time in hospital by 3%.  NHS costs per participant increased by £76 per year.

Welsh Predictive Risk Service

Image from Welsh Predictive Risk Service Software

Screen grab of Welsh Predictive Risk Service Screen

The impact

The Welsh Government paused and then halted roll-out of the software to GP Surgeries in Wales due to commissioning and then results of the PRISMATIC trial. Currently only 14% of GP surgeries in Wales report access to predictive risk stratification tools compared to 80% across the rest of the UK. Based on trial results, non-implementation of predictive risk software in Wales has avoided approximately 30,000 additional emergency admissions to hospital and 76,000 additional days spent in hospital each year. Avoided admissions and other contacts have saved approximately £200 million per year in Wales

Key messages

Health interventions do not always have effects that are intended. It is important to carry out robust evaluation of new technologies and treatments in pragmatic trials in order to understand effects in practice. Working closely with policy makers in Wales allowed the PRISMATIC trial team to ensure that findings have had an immediate impact in the real world and that in this case, health policy in primary care in Wales is based on the highest quality evidence

United Nations Sustainable Development Goals

Swansea University Research Themes