Article Text

P166 The frequency of chest radiographs prior to the onset of lung cancer symptoms
  1. MPT Kennedy,
  2. O Walkowiak,
  3. MEJ Callister
  1. Leeds Teaching Hospitals NHS Trust, Leeds, UK


Introduction Previous data has shown wide variation in the frequency of CXRs requested by GPs that is not explained by case mix factors. The relationship between threshold for CXR request and lung cancer characteristics at diagnosis is unknown.

Aim To analyse the frequency of CXRs prior to the development of lung cancer symptoms according to stage at presentation.

Method Retrospective review of an electronic database of lung cancer patients, excluding small cell, from 2010–2013. The dates of all CXRs in the three years before the first appointment with the lung cancer team were recorded. The frequency of CXRs was compared using Mann-Whitney U test with normal approximation.

Results 1750 patients were included. 589 had early stage disease (I/II) and 1161 had late stage disease (III/IV). The frequency of CXRs from 36 to 6 months prior to diagnosis is shown in Figure 1 according to stage at diagnosis. Patients subsequently diagnosed with early stage cancer had significantly more CXRs performed during this period compared to late stage patients (1.70 vs 0.92, p < 0.001).

Abstract P166 Figure 1

CXRs per patient 6 month period prior to diagnosis

The number of CXRs performed during the 6 months leading up to diagnosis was much higher for both groups, but remained significantly higher for early vs late stage disease (1.32 vs 1.15, p = 0.009). This will include many CXRs which triggered referral to the lung cancer team.

The number of patients with no CXR in the three years before diagnosis (excluding the six months immediately before diagnosis) was higher in late stage (64.6%) than in early stage (45.0%).

Conclusion Differences are apparent in the frequency of CXRs many months prior to lung cancer diagnosis. We speculate that these changes are seen at a time when symptoms attributable to the lung cancer are unlikely to have been present. Instead, this may reflect differences in patient behaviour (threshold for seeking review for respiratory symptoms) and GP behaviour (threshold for requesting CXR). There may be a degree of incidental detection of more indolent early stage cancers (i.e. length-time bias).

A patient-GP relationship that results in increased frequency of CXR seems to be associated with increased detection of early stage disease.

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.