Prediction of serious complications in patients admitted to a surgical ward

Br J Surg. 2002 Jan;89(1):94-102. doi: 10.1046/j.0007-1323.2001.01963.x.


Background: Prediction of complications is an essential part of risk management in surgery. Knowing which patients are at high risk of developing complications will contribute to the quality and cost reduction of surgery.

Methods: All patients admitted to a general surgical ward during a 1-year interval were followed until 30 days after discharge. Complications and data on potential risk factors were recorded prospectively. Relative risks were calculated for each risk factor and predictive values for the development of a serious or minor complication were computed using logistic regression analysis. The predictive values of different combinations of variables were expressed as receiver operating characteristic curves.

Results: Of 3075 patients, 375 suffered one or more serious complications and 319 experienced a minor complication. A model was developed for prediction of serious complications, consisting of 11 variables, with an area under the curve (AUC) of 0.79 (95 per cent confidence interval (c.i.) 0.76 to 0.81). The prognostic value of the model for minor complications (seven variables) was lower (AUC 0.68 (95 per cent c.i. 0.65 to 0.71)).

Conclusion: Serious complications in patients admitted to a surgical ward can be predicted using a model consisting of 11 variables. The risk score can be used in the decision-making process before surgery.

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Aged, 80 and over
  • Area Under Curve
  • Female
  • Forecasting
  • Hospitalization / statistics & numerical data*
  • Humans
  • Male
  • Middle Aged
  • Netherlands
  • Postoperative Complications / etiology*
  • Risk Assessment
  • Risk Factors
  • Severity of Illness Index
  • Surgical Procedures, Operative