Summarising the predictive performance of the SHFM and MAGGIC models

Heart failure (HF) means the heart cannot pump the blood around the body efficiently. People with HF can live comfortably for many years if their symptoms are under control. Achieving this involves receiving the right treatments at the right time, but this doesn’t always happen. One reason is that HF is unpredictable as it may remain stable for a long time in some people and get worse very quickly in others.

Predictive scores can be used to work out how likely it is that a person will have a particular outcome in the future, such as surviving for the next five years. These predictions are tailor-made to the individual, as they are based on their age, the current severity of their disease, their current treatments and the results of various medical tests. Predictive scores could help people with HF discuss, with their doctors, the current risks of their condition becoming worse and how these risks may be reduced by changing their treatments. To be useful, the scores need to be very accurate, and it is not clear which of the various HF predictive scores available is the most accurate.

Our study aims to summarise the accuracy of the Seattle and MAGGIC scores, which are the two predictive scores that have been tested most often on different groups of people with HF. Testing a score on a new group of people is called an external validation study. It is important to test a predictive score on many groups of people because differences in results of studies may just be due to chance, or due to actual differences between the studies. For example, some studies have more people with severe HF than others, and some studies may test the score more rigorously than others.

We will study how well the two scores predict survival at different times in the future and we will investigate the impact that different factors have on accuracy. These factors will include the severity of HF of the people involved in the study, the quality of the study, where and when the study was carried out, and the way in which accuracy was measured.

Our project will show if either score predicts survival accurately enough, and therefore how useful the two scores might be to doctors and patients when talking about survival. This will show us if the scores should be recommended for use in clinical practice or whether new, more accurate scores need to be developed. We are aware that the likelihood of survival is a difficult topic to discuss and we are mindful that the patient perspective is crucial to ensure that using scores such as these are acceptable to patients and their families.

Amount awarded: £66,666