Article Text

Download PDFPDF

P244 Accuracy of Admission Diagnosis of Pneumonia
  1. Lc Price1,
  2. Ms Anwar2,
  3. K Srikanthan3,
  4. A Bercusson2,
  5. E Williams1,
  6. E Prior1,
  7. A Owusu-Ageyi3,
  8. H Umpleby1,
  9. M Berry1
  1. 1St Mary's Hospital, London, UK
  2. 2Hammersmith Hospital, London, UK
  3. 3Charing Cross Hospital, London, UK


Background The inclusion of ‘new infiltrates’ on the admission chest radiograph (CXR) is part of the BTS audit tool (1), but it has been reported that up to 37% of inpatients coded (using the ICD-10 coding system) as pneumonia did not have CXR consolidation (2). We assessed our local audit population for potential reasons for this pneumonia miscoding.

Methods Patients selected by coding for the 2012/13 BTS Pneumonia Audit at hospitals within Imperial Healthcare Trust (Charing Cross (CX), Hammersmith (HH) and St Mary’s (SMH) Hospitals) underwent a notes review.

Results Of all patients clinically coded as community-acquired pneumonia (CAP), 88/176 (50%) had a diagnosis of CAP compatible with the audit criteria, with infiltrates on the admission CXR–in 15 (39%), 25 (61%) and 48 (50%) of cases at CX, HH and SMH respectively.

Of the patients found not to have CAP by the current BTS audit criteria (n = 88), 47/88 (53%) had an abnormal admission CXR not showing CAP. The main abnormalities in these admission CXRs were pulmonary oedema (in 30%), COPD/bronchiectasis (27%), malignancy (13%), interstitial lung disease (ILD) (7%) and pleural effusions (7%). In the 88 ‘non-audit criteria CAP’ patients, in combination with symptoms, inflammatory markers (WCC 10.4 (1.7–33) x109 and CRP 84.8 (3.6–381) mg/L), and the CXR series, the likely diagnosis was felt to be LRTI (35%), CAP (17%), HAP (13%), COPD (10%), pulmonary oedema (9%), malignancy (4.5%), UTI (3%), and ILD, asthma and effusions (all 2%), in the ‘non-audited CAP’ patients.

Of those remaining 41 patients with a normal admission CXR excluded from the audit, 24 patients (59%) had a repeat CXR within 72h, of which 5 (21%) then showed CAP. An additional 12 patients with normal admission CXRs had CT scans performed within 72h, 6 of which (50%) detected consolidation. Thus, of all the patients with a normal admission CXR, 11 (27%) had CAP that was missed by solely looking at the admission CXR.

Conclusion We confirm earlier findings that coding diagnoses are insufficiently accurate to judge quality of care, but also show that current audit criteria exclude a significant proportion of cases.


  1. WS Lim, et al. Thorax 2011;66:548–549

  2. Ruickbie SV, et al. Thorax 2012;67(Suppl 2):A69

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.