Predicting Risk of Infection After Rhinoplasty with Autogenous Costal Cartilage: A Cohort Study

Overview

The study developed and validated a nomogram, a scoring chart predicting risk, for infection after rib cartilage rhinoplasty. The primary group was 672 patients assessed between October 2018 and December 2020. A separate validation group of 118 consecutive patients from January to June 2021 tested it.

Sections of note

• Cohort study; 672 patients in the development group and 118 in the validation group; level of evidence IV.
• Twenty one candidate features were reduced to 10 predictors using LASSO regression.
• The ten factors were age, number of previous nose operations, length of hospital stay, operation time, history of nasal trauma, animal contact after surgery, smoking, drinking and spicy food within one month after surgery, and history of nasal infection.
• The model showed a C-index of 0.987, 95 percent confidence interval 0.978 to 0.996, with internal validation of 0.967.
• Calibration was good and the area under the curve was 0.987.
• In the validation group the C-index was 0.935, 95 percent confidence interval 0.910 to 0.960.
• Decision curve analysis indicated the tool was clinically helpful.
• The authors state this is the first such tool for rib cartilage rhinoplasty.

What it means for a patient

• Several of the strongest predictors are behaviours in the month after surgery, including smoking, alcohol, spicy food and animal contact.
• Previous nasal operations and previous nasal infection also raised predicted risk.
• A discrimination score of 0.987 is unusually high for a clinical prediction model, which warrants caution about overfitting.
• Both groups came from the same centre, so external validation elsewhere has not been done.

Why this paper matters

Infection after rib grafting can destroy the graft and require reconstruction, and no established way of identifying high risk patients existed. This study builds and internally validates one. Whether it performs as well in other populations remains untested.

Terms

• Nomogram: a chart converting several patient factors into a predicted risk.
• Autogenous costal cartilage: rib cartilage taken from the patient's own body.
• LASSO regression: a statistical method that selects the most informative variables.
• C-index: a measure of how well a model separates those who develop an outcome from those who do not.
• Calibration: how closely predicted risks match observed rates.
• Decision curve analysis: a method assessing whether using a model improves clinical decisions.