Background Hospital Acquired Pneumonia (HAP) is defined as lung infection in a non-intubated patient with new infiltrates on chest X-ray, >48 hours after hospital admission. Prediction scores exist for Community Acquired Pneumonia (CAP); no such scores exist in HAP. We aimed to identify features which are predictive of mortality in HAP.

Methods All cases coded as HAP in Heart of England Foundation NHS trust in 2013 trust were identified (293 cases). For each of these cases the chest X-ray (including radiologists report) was reviewed; if X-ray did not show infiltrates consistent with pneumonia, cases were excluded leaving 153 cases for whom case notes were reviewed. Cases were excluded if diagnosis of HAP was made <48 hours after admission leaving 136 cases. Data was collected regarding demographics, co-morbidities, investigations, observations, mortality during admission and within 12 months.

Univariate analysis was conducted to identify features associated with mortality. Multivariate analysis was completed using identified associated features.

Results Sixty-four cases (47.0%) died during admission; and a further 32 within 12 months (70.5%).

Demographics: Mean age was 81.6 years (range 52–98); mean number of co-morbidities was 5 (range 0–11). Mean haemoglobin was 110.9 g/dL. The mean white cell count (WCC) was 13.68 × 109/L (range 1.87–51.7 × 109/L). Mean urea was 10.5 mmol/L (range 1.9–6.1 mmol/L)

Univariate analysis: Table 1 shows the results of the univariate analysis.

Multivariate analysis: Only combination of raised urea and raised or low WCC were significantly associated with mortality (p = 0.024). Adding features of age, observations and co-morbidities did not improve prediction of mortality.

Conclusion Prediction of mortality in HAP is more complex than in CAP. On multivariate analysis, raised urea and raised or low WCC were predictive of mortality. Other features including age, number of comorbidities and observations at the time of diagnosis were not associated with mortality. This perhaps reflects our elderly cohort, with the majority having multiple co-morbidities, with very small numbers aged <65 years or with few co-morbidities. Further work with a larger dataset is ongoing.