Predictive values of lung cancer alarm symptoms in the general population: a nationwide cohort study

We aimed to firstly determine the 1-year predictive values of lung cancer alarm symptoms in the general population and to analyse the proportion of alarm symptoms reported prior to diagnosis, and secondly analyse how smoking status and reported contact with general practitioners (GPs) regarding lung cancer alarm symptoms influence the predictive values. The study was a nationwide prospective cohort study of 69,060 individuals aged ≥40 years, randomly selected from the Danish population. Using information gathered in a survey regarding symptoms, lifestyle and healthcare-seeking together with registry information on lung cancer diagnoses in the subsequent year, we calculated the predictive values and likelihood ratios of symptoms that might be indicative of lung cancer. Furthermore, we analysed how smoking status and reported contact with GPs regarding the alarm symptoms affected the predictive values. We found that less than half of the patients had reported an alarm symptom six months prior to lung cancer diagnosis. The positive predictive values of the symptoms were generally very low, even for patients reporting GP contact regarding an alarm symptom. The highest predictive values were found for dyspnoea, hoarseness, loss of appetite and for current heavy smokers. The negative predictive values were high, all close to 100%. Given the low positive predictive values, our findings emphasise that diagnostic strategies should not focus on single, specific alarm symptoms, but should perhaps focus on different clusters of symptoms. For patients not experiencing alarm symptoms, the risk of overlooking lung cancer is very low.