Postoperative complications in high-risk surgical patients - predictors, risk factors, and outcomes following major surgery study (PROFS study NCT02626546): validation of three prediction modelsShow others and affiliations
2017 (English)In: Acta Anaesthesiologica Scandinavica, ISSN 0001-5172, E-ISSN 1399-6576, Vol. 61, no 8, p. 1056-1056Article in journal, Meeting abstract (Other academic) Published
Abstract [en]
Background: Postoperative complications increase the risk of death 2–3 years postoperatively. Prediction of complications can support clinical decisions. Before clinical use of any prediction model, validation is reasonable. We aimed to validatethree models: Revised Cardiac Risk Index (RCRI), ARISCAT and POSSUM.
Methods: The PROF S-study was performed in four Swedish university hospitals. Patients werere cruited between 2015-11-01 and 2016-02-15. Inclusion criteria were adults, ASA classification ≥3, major/complex upper and lower gastrointestinal, urogenital or orthopedic surgery. Complications were screened on days 3, 7 and 10 by the postoperative morbidity survey (POMS). Only patients with grade≥2 (Clavien-Dindo classification) were accounted for. Study outcomes were cardiovascular (RCRI model) and pulmonary (ARISCAT model) complications, and a composite of the POMS domains (POSSUM model). Discrimination was evaluated by C-statistics (area under receiver operator characteristic curve; AUC ROC).
Results: The number of patients included was 1089. Thirteen patients were excluded due to wrong inclusion, and another three were lost to follow-up. Presen ce of malignancy was 41%. Patient characteristics and outcomes are displayed in Table 1. The RCRI underestimated the risk for cardiovascular complications, and discrimination was low (AUC ROC 0.64; 95% CI 0.59–0.68). The prediction by the ARISCAT model was fair (AUC 0.72; CI 0.69–0.76). The POSSUM model had poor /fair discrimination (AUC 0.70; CI 0.67–0.73).
Conclusions: The ARISCAT model predicted pulmonary complications with fair discrimina-tion and so could be used as decision support. Parameters with significant odds ratios of the RCRI and POSSU M models might be used as complements of clinical judgement.
Place, publisher, year, edition, pages
John Wiley & Sons, 2017. Vol. 61, no 8, p. 1056-1056
National Category
Anesthesiology and Intensive Care
Identifiers
URN: urn:nbn:se:oru:diva-59287DOI: 10.1111/aas.12941ISI: 000407231100158OAI: oai:DiVA.org:oru-59287DiVA, id: diva2:1136857
2017-08-292017-08-292020-12-01Bibliographically approved