Gupta and colleagues derived this model using the 2007 American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) data set (N=211,410) and validated it in the 2008 data set (N=257,385), covering adults undergoing a wide range of inpatient operations at 183–211 US hospitals.

Finding	Detail
Outcome	ACS NSQIP-defined postoperative pneumonia within 30 days of surgery.
Event rates	Postoperative pneumonia occurred in about 1.8% of the derivation cohort and 1.5% of the validation cohort. 30-day mortality was markedly higher in those with pneumonia than in those without.
Predictors	Seven preoperative variables: age, ASA class, chronic obstructive pulmonary disease, dependent functional status, preoperative sepsis, smoking within 1 year and procedure category.
Performance	Discrimination was excellent, with C statistics around 0.86 in both derivation and validation cohorts.

The calculator applies best to adult inpatients similar to the ACS NSQIP population and case mix. Contemporary enhanced-recovery pathways, different perioperative practices or much lower-risk day surgery may have lower absolute pneumonia rates, so predicted percentages may overestimate risk in these settings. The model incorporates only preoperative factors and planned procedure type. Intraoperative events and postoperative care bundles can substantially change the realised risk of pneumonia.

Gupta Postoperative Pneumonia Risk