STS: Risk score predicts rehospitalization after heart surgery

Identifying the factors that determine whether a patient will need rehospitalization following discharge after heart surgery is a huge and unresolved problem. Risk models for the rate of hospital readmission following cardiothoracic surgery have historically performed poorly. Perhaps that’s because the models often fail to include factors with the strongest impact on readmissions. Most of the factors that appear to drive readmissions seem to be out of the direct control of hospital staffs, such as a lack of support for patients once they leave the hospital. Socioeconomic factors like this have usually not been included in risk models.

Dr. David M. Shahian

The C statistic (area under the receiver operating characteristic curve) for the model reported by Dr. Kilic was 0.66, very close to the 0.648 that my colleagues and I reported in 2014 for a risk model of 30-day hospital readmission following isolated coronary artery bypass grafting that used data from more than 162,000 Medicare patients who underwent surgery during 2008-2010 (Circulation. 2014 July 29;130[5]:399-409). This means that both models accounted for roughly two-thirds of the variability in readmission rates, which makes our model as well as Dr. Kilic’s model marginal in its ability to identify patients at greatest risk. Similar limitations exist for the other reported models for assessing the readmission risk following heart surgery.

One strength of the model reported by Dr. Kilic was its inclusion of patient factors that developed following the start of the index admission, such as postoperative acute renal failure.

Dr. David M. Shahian is a professor of surgery at Harvard Medical School and associate director of the Codman Center for Clinical Effectiveness in Surgery at the Massachusetts General Hospital, both in Boston. He made these comments during the discussion of Dr. Kilic’s report. Dr. Shahian had no relevant financial disclosures.