Prediction of Operative Mortality with Valve Replacement Surgery

Abstract & Commentary

Synopsis: The information in this paper provides benchmarks for assessing your institution’s surgical results and a tool for assessing the relative risk in individual patients.

Source: Edwards FH, et al. J Am Coll Cardiol. 2001; 37:885-892.

Risk assessment is a critical component of cardiothoracic surgery, but there is limited information about the prediction of risk in valve replacement surgery. Therefore, Edwards and associates analyzed the Society of Thoracic Surgeons national cardiac surgery database for the outcome of patients with isolated aortic or mitral valve replacement surgery and valve replacement surgery plus coronary artery bypass surgery (CABG). The database was started in 1986 and now contains more than 1 million patients from 487 hospitals in 47 states and Canada. For the purpose of developing a risk model, 92,536 patients with either isolated aortic valve replacement (32,968), isolated mitral valve replacement (16,105), CABG plus aortic valve (32,528), or CABG plus mitral valve (10,925), operated on between 1994 and 1997 were analyzed. After development of the model, a validation sample of 51,492 patients operated on between 1998 and 1999 were used to test the model.

In general, patients with isolated valve replacement surgery were younger and had less CAD risk factors. Also, operative mortality and morbidity were greater for valve surgery combined with CABG vs. valve surgery alone. In addition, operative mortality and morbidity were higher for mitral vs. aortic surgery. Operative mortality was lowest for isolated aortic valve surgery and highest for mitral valve replacement plus CABG (see Table). Also, ceratin complications were more common with one type of surgery vs. the others.

Perioperative Complications by Type of Surgery

Operative mortality (%) 4 6 7 13
Stroke (%) 2 2 3 4
Reoperation for bleeding (%) 4 5 6 6
Renal failure (%) 4 5 6 13
Infection (%) 0.5 0.3 0.7 0.8
Hospital stay (days) 11 13 13 16

Patient subgroup analysis showed that diabetes approximately doubles the risk of operative mortality. Renal dialysis increases risk 3-4 fold to 17-37%; peripheral vascular disease almost doubles the risk to 7-21%; and the presence of 3 vessel CAD increases risk to 10-16%. The presence of prior stroke, immunosuppressive therapy, and previous cardiac surgery increases operative mortality risk less than two fold. Systemic hypertension and chronic obstructive lung disease did not appreciably increase risk.

The validation sample analysis showed that the predicted operative mortality rate from the model was close to the observed rate: the overall rate for isolated valve replacement was 4.8% predicted vs. 4.7% observed; the overall rate for CAGB plus valve replacement was 8.6% predicted vs. 8.2% observed. Hameed et al concluded that they have developed a statistical model that accurately predicts operative mortality after valve replacement surgery.

Comment by Michael H. Crawford, MD

Risk prediction models have been well worked out for CABG surgery, but little data exist for valve surgery. Thus, this report from the STS database is of interest. The strength of this database is that it includes a wide variety of public, government, and private hospitals. Consequently, Edwards et al suggest that this data can be used to form a national benchmark for valve replacement surgery. On the other hand, they do not believe their data can be used as a national standard, because of regional differences in practice and referral patterns. However, the data are useful for discussing the potential risk with patients if you do not have such data at your own institution.

Several things stand out in the results that are of interest to the clinician. Mitral valve replacement plus CABG surgery has the greatest operative mortality, the highest complication rate, and the greatest length of stay. The subgroup with the highest operative mortality is patients on renal dialysis (3-4 fold increase). Prior stroke and previous cardiac surgery have a modest effect on mortality (less than a 2-fold increase). Chronic lung disease has little influence on mortality. These results are useful for assessing risk in a particular patient. Thus, the information in this important paper provides benchmarks for assessing your institution’s surgical results and a tool for assessing the relative risk in individual patients.