• 作者：Anthony Park Morton MD ; MS ; FFARACS ; MSc Appl^a ; ^c ; ^{amor5444@bigpond.net.au"" rel=""nofollow} ; Susan Erica Smith BSc (Hons)^a ; Daniel V. Mullany FANZCA ; FJICM^b ; Andrew J.B. Clarke FRACS^a ; Douglas Wall FRACS^a ; Peter Pohlner FRACS ; GCM^a
• 关键词：Statistical process control ; Coronary Artery Bypass Grafts ; Cardiac surgical patients ; Outcomes monitoring
• 刊名：Heart, Lung and Circulation
• 出版年：2011
• 出版时间：May 2011
• 年：2011
• 卷：20
• 期：5
• 页码：312-317
• 全文大小：203 K

Objective

Methods

Analysis of Cardiac Surgical Register database using tabulations, funnel plots and random-effects (Bayesian shrinkage) analysis for aggregated data. Combined CUSUM and cumulative observed minus expected (modified VLAD) charts and combined CUSUM and cumulative funnel plots used for individual observation sequential data and binomial control charts and generalised additive models (GAMs) for quarterly sequential data. Risk adjustment employed re-calibrated EuroSCORE.

Results

There were 2575 procedures with an unadjusted in-hospital mortality rate of 1.17 % . Mean age was 65 years and 21 % of patients were female; 43.6 % were elective procedures. Median ventilation time was 10 hours and median length of stay in intensive care (ICU) was 23 hours. Return to theatre for bleeding occurred in 3 % of cases. Return to theatre for surgical site infection occurred in 0.4 % of cases; 4 % were re-do procedures. Permanent stroke or neurological deficit occurred in 1 % , perioperative myocardial infarction in 0.8 % , arrest in 1.2 % , renal failure in 1.6 % and ICU return in 2.3 % of cases.

Conclusions

Complication rates and mortality were comparable with similar units. Use of random-effects (Bayesian shrinkage) analysis for aggregated data is encouraged together with generalised additive models (GAMs) and combined CUSUM and cumulative observed minus expected (modified VLAD) charts for sequential data.