Web Calculator Estimates Survival of Bowel Cancer Patients

The new online prognostic tool could help manage expectations and make more informed decisions about treatment options.

The calculator was developed by researchers at the University of Nottingham (UN; Nottingham, UK) and medical software company ClinRisk (UK) using the QResearch database, which gathers patient data from approximately 1500 general practices across England through the clinical IT computer systems of EMIS Health. QResearch is a large consolidated database of anonymous health records, developed by a not-for-profit partnership between the UN and EMIS.

“We are grateful for the continued commitment of practices using our systems and their patients who are helping to improving healthcare through data analytics,” said Shaun O’Hanlon, medical director at EMIS Health, “While having the best information is always important, it is how it is delivered and used that is key.”

Accuracy testing showed that the calculator can predict absolute survival rates. It also allows patients to update their mortality risk based on how long they have survived following a diagnosis of cancer.

“Current methods of estimating survival tend to be unreliable and sometimes patients can be given a fairly misleading and unnecessarily gloomy prognosis based only on the grade and stage of their cancer, only to find that in reality they live much longer than these crude predictions,” said Prof. Hippisley-Cox, “This new calculator, which doctors and patients can access, will offer a far more realistic estimate. We understand that not everyone will want to do this, of course, but some patients are very keen on this approach so it’s an individual choice.”

Current methods of predicting survival are based on simple averages based on age or the grade and stage of the cancer in the wider population. To deliver a far more personalized prognosis, the new tool takes into account additional risk factors, including the patient’s smoking history, body mass index, family history, other illnesses and treatments (e.g. aspirin or statins), as well as other information (e.g. whether they have had surgery or treatments such as chemotherapy for the cancer).

The team used information from more than 44,000 patients from 947 practices to develop separate equations for men and women aged between 15-99 years when diagnosed with bowel cancer. They then tested the equations by using them retrospectively to predict the outcome at 1, 5, and 10 years after diagnosis for 15,214 bowel cancer patients from 305 different GP practices, and for 437,821 colorectal cancer patients from the national cancer registry.

The results indicated that they had devised strong models for prediction of cancer survival outcomes. They were also able to provide conditional survival estimates that showed how mortality risks change over time, particularly important among patients where the initial prognosis was poor due to late-stage disease.

For example, a 38-year-old woman with stage 4 advanced bowel cancer might the calculator and see that her chance of surviving for 5 years without any treatment would be predicted to be only 6%; with surgery it would increase to 23%, and with both treatments it would be 45%. This compares with the published chance of 5-year survival of 66% based solely on her age, or the 8% chance based solely on her cancer stage. If she has both treatments and survives for 1 year following diagnosis, her predicted 5-year conditional survival would increase to 57%.

The study, by Hippisley-Cox J and Coupland C, was published June 15, 2017, in the British Medical Journal (BMJ).

Related Links:
University of Nottingham
ClinRisk