Background: Colonoscopy bowel preparation failure is common, costly, and clinically harmful. Prediction models can identify patients at high risk for preparation failure, but they are rarely used. Goals: To investigate the clinical and economic effects of using a more intensive preparation upfront (a "targeted" strategy) for patients identified as high risk for preparation failure. Study: We developed a Markov decision analytic model to compare usual care with a targeted strategy. Usual care consisted of a 4 L preparation, followed by a 2-day preparation in the event of preparation failure. The targeted strategy consisted of a 4 L preparation for low-risk patients, and upfront 2 days preparation for high-risk patients. Base-case values were gathered from literature review. Under each strategy, we calculated days spent preparing for, or undergoing, colonoscopy (patient burden), and cost per patient (payer perspective). Sensitivity analyses were performed. Results: In the base case, the targeted strategy resulted in a similar patient burden compared with usual care (2.56 vs. 2.51 d, respectively). However, it substantially reduced cost per patient ($1254 vs. $1343) by reducing the number of colonoscopies. In sensitivity analyses, the targeted strategy reduced cost across the entire plausible range of risk of preparation failure and prediction model sensitivity and specificity. The targeted strategy resulted in less patient burden than usual care when (1) preparation failure risk exceeded 20%; (2) prediction sensitivity exceeded 73%; or (3) prediction specificity exceeded 76%. Conclusions: Targeted bowel preparation is likely to reduce costs associated with repeat colonoscopy with minimal effect on patient burden.
- personalized medicine