Skip to main content

The impact of unmeasured clinical variables on the accuracy of hospital report cards: a Monte Carlo study

Austin PC. Med Decis Making. 2006; 26(5):447-66.

Objective — To estimate the baseline risk of arrhythmic death required for prophylactic implantable cardiac defibrillators (ICDs) to result in clinically meaningful survival benefits in the population.

Background — While proven efficacious, the absolute survival impact of ICDs for the primary prevention of sudden cardiac death among patients with left ventricular (LV) dysfunction is highly dependent upon patient's baseline risk of arrhythmic death.

Methods — Using echocardiographic data from a random sample of patients identified from community echocardiographic laboratories, patients with moderate or severe LV dysfunction (ejection fraction < 35%) were linked to administrative databases to characterize baseline mortality risk (median follow-up duration of 4.85 years). Relative efficacy was ascertained from meta-analysis and clinical trial data. The baseline annual risk of arrhythmic death required for prophylactic ICDs to result in clinically meaningful survival benefits in the population was estimated at different ranges of relative efficacy and numbers needed to treat (NNTs) thresholds.

Results — LV dysfunction was a significant independent predictor of adverse outcomes. In total, 35.4% of the patients with moderate to severe LV dysfunction died during the follow-up period. Assuming a base-case relative efficacy of 66%, we estimated that the baseline risk for arrhythmic death required to exert a clinically meaningful NNT threshold of 50 in order to prevent one death (from any cause) was 3% per year or higher.

Conclusions — The survival impact and cost-effectiveness of prophylactic ICDs in the population will depend upon the ability to risk-stratify and identify patients whose baseline risk for sudden cardiac death exceed 3% per year.

View full text

Keywords: Health services research Data evaluation Performance measurement