Background Delays in referral of symptomatic patients for chest X-ray (CXR) has been proposed as one explanation for poorer survival from lung cancer in UK compared to other developed nations. Such delays may reflect patients not seeking medical advice for persistent respiratory symptoms, or not being referred for CXR following presentation to primary care. We sought to determine the relationship between GP CXR referral rates and patient characteristics at the time of lung cancer diagnosis.

Methods Data was collected for the number of CXRs ordered by GP practises in Leeds between 2008 and 2010 (corrected for list size). Patient characteristics at presentation with lung cancer (stage and performance status), treatment and outcome (one year survival) were compared over the same three year period. GP practises were grouped into quintiles by CXR referral rate (according to population served), and lung cancer outcomes compared between quintiles.

Results CXR referral rates varied from 6.7 to 62.3 CXRs per 1,000 population per year. Data was collected for 1,394 patients diagnosed with lung cancer. The number of lung cancer diagnoses per CXR quintile were 230, 276, 258, 322, 308 from lowest to highest CXR quintile respectively. The proportion of patients with early stage disease at presentation (stage I and II) did not vary by CXR quintile (29.6%, 26.1%, 28.7%, 26.7%, 25.6% from lowest to highest CXR rates). Similarly there were no differences in the distribution of performance status, the proportion of patients undergoing any anti-cancer treatment and one year survival between the quintile groups. When the cohort was split into 2 groups by CXR rates, no differences in lung cancer outcomes were seen between GP practises (number of lung cancers 660 for low CXR group vs. 734 for high CXR group).

Conclusion Despite demonstrating large variability in CXR rates, our small single-centre study does not show a link between referral rates and lung cancer outcomes. This study is underpowered to detect small changes in survival or stage shift, but suggests large differences in outcome are unlikely to result from this variability in practise at primary care level.