Weight loss as a feature of cancer in primary care

Background Unexpected weight loss (UWL) is a non-specific symptom posing a diagnostic challenge to general practitioners (GPs). It can be associated with several conditions including cancer. There are no international or national clinical guidelines to support GPs to select patients with UWL for cancer investigation. Aim The overall aim is to provide the evidence necessary to allow GPs to identify which patients with UWL should undergo testing for cancer. Methods Six studies are included in this thesis. Firstly, an evidence synthesis of studies investigating the diagnostic accuracy of UWL as a diagnostic feature of cancer in primary care patients. Following this, a series of five linked studies using electronic health records data from the UK: i) a cohort study about weight recording in NHS primary care from 2000 to 2017 to establish whether weight records could be used to define weight loss; ii) an internal validation study of weight related coding to determine which codes can be used most reliably to define UWL; iii) a time-to-event matched cohort analysis of the association between UWL and cancer to ascertain the time period of increased risk for a cancer diagnosis following a presentation with UWL, for cancer overall and by cancer site and stage; iv) a cross-sectional diagnostic accuracy study investigating the predictive value of risk factors, symptoms, signs and blood test results in patients with UWL; v) the derivation and validation of prediction models to estimate cancer risk in patients with UWL, leading to the creation of simple clinical risk scores to stratify cancer risk using gender, age, risk factors, symptoms and blood tests. Results Weight recording is not a routine activity in UK primary care. It occurs too infrequently to define UWL in epidemiological studies. Validated UWL codes represent ≥5% weight loss over six months. Coded UWL is a strong predictor of cancer diagnosis within three months of presentation to primary care. The commonest cancers in people with UWL were pancreas, cancer of unknown primary, gastro-oesophageal, lymphoma, hepatobiliary, lung, bowel and renal tract. Men aged 50 years and older who have smoked and people with co-occurring clinical features or blood test abnormalities were at greatest risk. Clinical features typically associated with specific cancer sites are markers of several cancer types when they occur with UWL. The simple clinical risk scores showed excellent discrimination, calibration and clinical utility. A risk score that prioritised ruling-out cancer investigation (sensitivity 97.5%) corresponded to 35 people with UWL being referred for investigation for each person with cancer referred and 1,730 people spared referral for each person with cancer not referred. Conclusion A GP’s decision to code UWL is predictive of several but not all cancers. Risk factors, co-occurring clinical features, and blood test results can be used to select patients with UWL for urgent investigation. These findings will inform future clinical guidelines.