UHN study suggests mortality rate is underestimated in CIHI model for specialised cardiac centres

by • September 6, 2018 • Feature Slider, Feature-Home, Featured-Slides-HomeComments Off on UHN study suggests mortality rate is underestimated in CIHI model for specialised cardiac centres252

Researchers from the Peter Munk Cardiac Centre (PMCC) at UHN suggest that the model used by the Canadian Institute for Health Information (CIHI) significantly underestimates mortality in specialized heart surgery centres. The study’s findings show that CIHI’s model does not encapsulate all the medical problems that patients have when at elevated risk for surgery.

The study, published in CMAJ Open, looked at the outcomes of 1,635 cardiac operations performed at the PMCC between 2013 and 2016. It compared the observed mortality within 30 days of surgery to the predicted mortality rates estimated with either the CIHI administrative data model or the clinical data model used by the Society of Thoracic Surgeons (STS).

Of the 1,635 patients that had heart surgery at the PMCC, 32 died within 30 days of surgery. These results were in line with the 1.96 per cent mortality predicted by STS model – which uses comprehensive data to describe how sick patients are before surgery – for this group of patients. In contrast, the CIHI Cardiac Care Quality Indicator estimated that the mortality rate for these 1,635 patients would be 1.03 per cent.

“The failure to include these seven medical conditions causes the CIHI database to underestimate predicted mortality after heart surgery in high-risk patients.” – Dr. Barry Rubin.

PMCC researchers noted that the STS model captures seven medical conditions that predict worse outcomes after heart surgery that are not captured in the CIHI model.  These risk factors include whether the patient had heart failure, an abnormal heart rhythm, a recent heart attack, very low blood pressure (shock), needed recent CPR or a mechanical heart pump to live, or had kidney failure. These serious conditions forecast a worse outcome after heart surgery, and are not included in the CIHI database.

“The predicted mortality of patients undergoing heart surgery based on the STS model is similar to what we actually observed,” says Dr. Barry Rubin, medical director of the PMCC and one of the authors of the study. “The failure to include these seven medical conditions causes the CIHI database to underestimate predicted mortality after heart surgery in high risk patients.”

According to Dr. Douglas Lee, senior scientist at the PMCC and lead author of the study, mortality prediction models – either based on clinical or administrative data – use risk adjustment to account for how sick patients are before surgery at different hospitals. This is necessary as outcomes may vary if leading academic institutions such as the PMCC operate on sicker patients.

“The CIHI and STS models aim to predict outcomes based on the medical complexity and acuity of the patient. In general, academic hospitals take on higher risk cases compared to community hospitals, and good risk adjustment models should factor that in when predicting mortality rates,” explains Dr. Lee.

Incorrect data may lend the impression that there is a higher level of mortality then there should be at the PMCC or other academic centres. This could have the inadvertent effect of causing the highest risk surgery patients to defer potentially-life saving heart surgeries.

CIHI has a legislative mandate to publicly release Cardiac Care Quality Indicator data and has done so since October 2017. Clinicians and researchers have been concerned that the CIHI model may underestimate how sick cardiovascular surgery patients are at PMCC.

Incorrect data may lend the impression that there is a higher level of mortality then there should be at the PMCC or other academic centres. This could have the inadvertent effect of causing the highest risk surgery patients to defer potentially-life saving heart surgeries.

The STS model considers vital medical conditions to accurately predict how high-risk patients will do after surgery. The CIHI model does not accurately account for complexity of patients, which is recorded in databases specifically designed for the measurement of surgical quality, such as STS.

“Clinical data-based models like the STS collect much more detailed patient information, but are also more costly to maintain,” says Dr. Rubin. “Administrative models like CIHI’s continue to play a very important role in assessing quality of care across Canada. We will continue to work in collaboration with CIHI to improve the accuracy of quality report cards that can be used as valid evaluation tools for Canadian hospitals.”

The authors caution that there are limitations to the study, as it was completed in a single centre during a three-year period, observing 1,341 isolated coronary bypass grafts, 143 isolated aortic valve replacements and 151 combined procedures. PMCC researchers declared there were no conflict of interests but would suggest validation of these findings in other academic centres over a longer time interval that would include a larger subset of cardiac operations. This study was supported by the Canadian Institutes of Health Research.

Comments are closed.