What is the sensitivity of CXR for lung cancer? & Are outcomes different for patients who have a ‘true’ positive’ compared to those who have a ‘false negative’ CXR?


Year:

Session type:

Stephen Bradley, Richard Neal, Martyn PT Kennedy, Matthew EJ Callister, Luke NF Hatton, Bethany Shinkins, Luke TA Mounce, William T Hamilton

Abstract

Background

Despite remaining a first line test, there is remarkably little evidence for the accuracy of chest x-ray (CXR) in the detection of symptomatic lung cancer or the possible adverse consequences of false negative CXR. 

Study objectives were to:

  • determine sensitivity of CXR for lung cancer
  • compare stage at diagnosis and survival between those who had true positive and false negative CXR

Method

All patients diagnosed with lung cancer in Leeds Teaching Hospitals between 2008 and 2016  who had a CXR requested by their GP in the year before diagnosis were categorised into ‘positive’ and ‘negative’ groups based on the first CXR report in that year.

 

Analysis was conducted with respect to CXR result, time to diagnosis and stage at diagnosis. 

Results

4738 patients were diagnosed with lung cancer of whom 2129 (44.9%) had at least one GP requested CXR in the year before diagnosis. 

The index CXR was positive in 1753 (82.3%) and negative in 376 (17.6%).

Median duration from index CXR to diagnosis was 43 days (IQR 27-78) for those with a ‘positive’ result compared to 204 (IQR 105-285) days to those who had a ‘negative’ CXR. 

634 (29.8%) patients were diagnosed at stage I or II, of whom 508 (29.0%) had a ‘positive’ index CXR and 126 (33.5%) had a negative index CXR.  There was no statistically significant association between CXR result and stage at diagnosis, X2 (1, N=2124) = 2.92, p = 0.09

Conclusion

This is by far the largest such investigation.  Although sensitivity was slightly higher than reported in smaller studies, cancer was not detected in 18%.  Non-detection was associated with longer time to diagnosis but not adverse stage.  The results could reflect the limits of observational data -undetected cancers may have been less advanced or not present at all at the time of CXR. Although it has been suggested that CT should replace CXR as the first line test for lung cancer the benefits that could be expected from such an approach remain unknown.      

Impact statement

The benefits of replacing CXR with CT should not be assumed without a prospective study of both modalities, encompassing accuracy, cost-effectiveness and downstream implications.