Symptoms and risk factors to identify women with suspected cancer in primary care: derivation and validation of an algorithm

Background Early diagnosis of cancer could improve survival so better tools are needed. Aim To derive an algorithm to estimate absolute risks of different types of cancer in women incorporating multiple symptoms and risk factors. Design and setting Cohort study using data from 452 UK QResearch (R) general practices for development and 224 for validation. Method Included patients were females aged 25-89 years. The primary outcome was incident diagnosis of cancer over the next 2 years (lung, colorectal, gastro-oesophageal, pancreatic, ovarian, renal tract, breast, blood, uterine, cervix, other). Factors examined were: 'red flag' symptoms including weight loss, abdominal pain, indigestion, dysphagia, abnormal bleeding, lumps; general symptoms including tiredness, constipation; and risk factors including age, family history, smoking, alcohol intake, deprivation, body mass index (BMI), and medical conditions. Multinomial logistic regression was used to develop a risk equation to predict cancer type. Performance was tested on a separate validation cohort. Results There were 23 216 cancers from 1 240 864 females in the derivation cohort. The final model included risk factors (age, BMI, chronic pancreatitis, chronic obstructive pulmonary disease, diabetes, family history, alcohol, smoking, deprivation); 23 symptoms, anaemia and venous thrombo-embolism. The model was well calibrated with good discrimination. The receiver operating curve statistics were lung (0.91), colorectal (0.89), gastro-oesophageal (0.90), pancreas (0.87), ovary (0.84), renal (0.90), breast (0.88), blood (0.79), uterus (0.91), cervix (0.73), other cancer (0.82). The 10% of females with the highest risks contained 54% of all cancers diagnosed over 2 years. Conclusion The algorithm has good discrimination and could be used to identify those at highest risk of cancer to facilitate more timely referral and investigation.