Existing Predictive Models for Postoperative Pulmonary Complications Perform Poorly in a Head and Neck Surgery Population

C. Burton Wood, Justin R. Shinn, Andrew B. Rees, Priyesh N. Patel, Robert E. Freundlich, Derek K. Smith, Matthew D. McEvoy, Sarah L. Rohde

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


Postoperative pulmonary complications (PPCs) are common following major surgical procedures. Risk stratification tools have been developed to identify patients at risk for PPCs. While otolaryngology cases were included in the development of common predictive tools, they comprised small percentages in each tool. It is unclear how these tools perform in patients undergoing major head and neck surgery with free flap reconstruction. This retrospective review studied all free flap reconstructions in head and neck surgery over a 12-year period at a single institution in the southeastern US. Baseline demographic and medical information were included for each case. All cases were reviewed for development of major PPCs, including pneumonia and respiratory failure. The cohort underwent risk stratification using the ARISCAT and Gupta pulmonary risk indices. Performance of these predictive models for head and neck surgery was determined through receiver-operator curve comparison. 794 patients were identified with a median age of 62 years (IQR 41–83). Sixty-five percent were male. Forty-three (5.4%) developed pneumonia, 23 patients developed respiratory failure (2.9%), and 38 patients developed both (4.8%), resulting in a total PPC proportion of 13.1% (n = 104). Both ARISCAT and Gupta pulmonary risk indices demonstrated low discrimination to predict PPCs in head and neck free flap reconstruction, with areas under the curve of 0.60 and 0.65, respectively. Two major indices for prediction of postoperative pulmonary complications do not accurately identify risk in patients undergoing major head and neck surgery. Further studies are needed to develop predictive tools for PPCs in this high-risk population.

Original languageEnglish (US)
Article number312
JournalJournal of Medical Systems
Issue number10
StatePublished - Oct 1 2019
Externally publishedYes


  • Microvascular free flap reconstruction
  • Post-operative complications
  • Predictive algorithms
  • Risk stratification

ASJC Scopus subject areas

  • Medicine (miscellaneous)
  • Information Systems
  • Health Informatics
  • Health Information Management


Dive into the research topics of 'Existing Predictive Models for Postoperative Pulmonary Complications Perform Poorly in a Head and Neck Surgery Population'. Together they form a unique fingerprint.

Cite this