• Lung Cancer

We appreciate the paper by O'Dowd et al1 as an interesting contribution to the literature concerning the primary care diagnosis of lung cancer. By concentrating on those patients dying soon after diagnosis, this paper is addressing the issue of access to care of patients with lung cancer who are already gravely ill. The disappointingly low chest X-ray (CXR) referral rates (we calculate obtained in only 40.2% and 50.4% of the patients attending with medium/high frequency and dying before and after 90 days, respectively) may represent irregular recording of CXR requests in The Health Improvement Network (THIN). Electronic research databases store their data in several fields, and CXR requests do not always appear in the main clinical field, sometimes being relegated to ‘hidden text’. In a study of paper and electronic records, Stapley et al found a primary care CXR was obtained in 164/247 (66%) of lung cancer cases;2 a second study examined 3184 cases in the GPRD from 2007 to 2009 inclusive and found a CXR in 1947 (61%), though we believed there was too much missing CXR data to submit the latter paper for publication. Thus, the CXR referral rates in the present study do appear unusually low, especially given that data from practices reporting very low referral rates were excluded.

The more important consideration is whether having a ‘liberal’ primary care CXR policy might facilitate the diagnosis of early-stage and potentially curable lung cancer. We agree with the authors that the association between higher CXR rates at practice level and increased lung cancer mortality likely reflects better identification of lung cancer as a cause of death as a result of having a CXR. We would be interested to know whether the incidence of lung cancer was higher in practices that ordered CXRs more frequently and whether practice-level CXR rates might yet correlate with higher rates of long-term survival in lung cancer.

The study suggests reliance on primary care to identify advanced, let alone early, cases of symptomatic lung cancer is a failing strategy in isolation despite the high frequency with which these patients interact with primary care. Besides the use of a risk calculator, as suggested by the authors, other interventions include empowering patients to request a CXR, as we attempted in the original El Cid study,3 lowering the threshold suggested by the National Institute of Health and Care Excellence (NICE) for ordering a CXR in primary care4 or proactively screening patients with COPD for NICE lung cancer criteria.5