U. NOTTINGHAM (UK) — A computer-based tool could help speed early diagnosis of patients suffering from two of the most common forms of cancer, saving up to 5,000 lives a year.
By highlighting potentially worrying combinations of symptoms and risk factors, the algorithm is able to successfully identify gastroesophageal cancer and lung cancer at an earlier stage.
The results, published in two studies in the British Journal of General Practice, showed that the 10 percent of patients that the algorithm predicted as most at risk of developing one of the two diseases accounted for 77 percent of all the gastroesophageal and lung cancers diagnosed over the following two years.
“Earlier diagnosis of cancer is a major challenge and we hope this new research will help doctors identify patients for earlier referral and investigation,” says Julia Hippisley-Cox, professor of clinical epidemiology and general practice at the University of Nottingham.
Evidence suggests that simply raising awareness of symptoms and speeding up diagnosis could save 5,000 lives a year without any new advances in medicine.
The researchers developed two simple web calculators—one for lung cancer and and another for gastroesophageal cancer—which are designed for use by doctors but a simpler version could also be made available on the Internet to raise awareness among the general public and to prompt patients with high risk factors or symptoms to seek advice from their doctor.
Missed symptoms
Lung cancer is the most common cancer worldwide, with 1.3 million new cases diagnosed every year. It has one of the lowest survival rates because two-thirds of patients are diagnosed too late to be successfully treated, presenting a huge challenge for family physicians because the symptoms can be common and non-specific.
While smoking is a well-known risk factor, evidence suggests that other factors including age, social deprivation, and chronic obstructive airways disease also have an important part to play.
The presence of the disease can be indicated by red flag symptoms such as new onset of coughing, coughing up blood, weight loss, loss of appetite and anemia. Currently, doctors focusing on just one of these symptoms without taking into account other risk factors are likely to miss 80 percent of current lung cancer cases.
Similarly, gastresophageal cancer is one of the most common cancers worldwide. Earlier diagnosis could improve treatment options and improve five-year survival.
Alarm symptoms for gastresophageal cancer include vomiting blood, difficulty swallowing, appetite loss, weight loss, or abdominal pain, However, focusing on one symptom alone in diagnosis can mean that up to 40 percent of cases are missed. Other underlying factors which could also alert doctors to patients at risk include heavy smoking.
Red flags
The study aimed to develop and test the success of a computer algorithm that would incorporate both the symptoms and underlying risk factors of patients to flag those in need of urgent investigation or referral.
It used 375 general practices in the UK already using the QResearch database system—a not-for-profit partnership between the University of Nottingham and GP systems supplier EMIS—to collect anonymous patient information.
It included patients aged 30 to 84 years who were free from a diagnosis of the cancers at the start of the study. For lung cancer it ruled out those patients who had previously seen their doctors within the previous 12 months with symptoms of coughing up blood, loss of appetite, or weight loss, while the gastresophageal study looked at those patients at the beginning of the study free from difficulty swallowing, vomiting blood, abdominal pain, appetite loss, or weight loss.
The study then identified those patients with the highest associated risk factors for the cancers to predict which were most likely to develop the disease and then validated their results by looking at which patients had been diagnosed with the cancers at the end of two years.
The study found that the new algorithm worked so well it could identify 10 percent of the population in which around 77 percent of all new cases of cancer arose over the two years.
The algorithm could be incorporated into existing computer records to alert doctors to patients who are potentially at a higher risk of developing the diseases.
More news from University of Nottingham: http://communications.nottingham.ac.uk/News.html
