email article Use of suicide risk prediction models can be cost-effective for U.S. healthcare systems, even with relatively low accuracy, researchers suggested. In an analysis designed to determine accuracy thresholds needed for cost-effectiveness -- with a specificity of 95% or higher -- cognitive behavioral therapy (CBT) reached that benchmark at a sensitivity of at least 35.7% (95% CI 23.1%-60.3%), while "active contact and follow-up" (ACF) did so at a sensitivity of at least 17% (95% CI 7.4%-37.3%), reported Eric Ross, MD, of Massachusetts General Hospital in Boston, and colleagues. Many statistical models have been developed in the hopes of predicting suicide risk, but it's not known how accurate these models need to be in order to warrant clinical implementation.