SECOND-LOOK ENDOSCOPY FOR BLEEDING PEPTIC ULCER DISEASE: A DECISION- AND COST-EFFECTIVENESS ANALYSIS

Purpose: Following application of therapeutic endoscopic methods for treatment of bleeding peptic ulcer disease (PUD), a follow-up or second-look endoscopy (SLE) may be performed 1-2 days later.  SLE may decrease the risk of recurrent PUD bleeding; however, it is not routinely recommended because it has no clear effect on the need for surgery or on mortality, in large part because clinical trials have been underpowered for these outcomes.

Method: Using literature-based probabilities and Medicare reimbursed costs, we created a decision model comparing routine SLE (rSLE) vs. no SLE for patients with bleeding PUD. In the model, an initial episode of rebleeding was re-treated endoscopically, while a second rebleeding episode was treated surgically. For outcomes, we measured rates of rebleeding, need for surgery, hospital mortality, and hospital costs, and we calculated costs to avoid one of each outcome (expressed as the number needed to treat [NNT]), along with ICERs. When costs were uncertain, we chose costs that would bias the model against rSLE. 

Result: In the base case, rSLE reduces: rebleeding from 16% to 8.2%; the need for surgery from 3.1% to 2.7%; mortality from 1.08% to 0.94%; but not hospitalization cost, which increases from $12,069 to $12,572.  Respective NNTs for rebleeding, surgery, and mortality are 12.8, 251, and 719, respectively. Incremental costs of rSLE to prevent 1 rebleed, 1 surgery, and 1 hospital death are $6,449, $125,750, and $314,375, respectively. Threshold analysis revealed that the rebleeding threshold required to neutralize the need for surgery and mortality is 14%, and is 20% to neutralize the cost difference.  If rSLE were 100% effective in preventing rebleeding, then the rebleeding threshold required for cost neutrality is 8.6%. One-way sensitivity analyses revealed that base case findings for surgery and mortality were sensitive to the probabilities of: rebleeding after index endoscopy; rebleeding after rSLE; continued bleeding; and repeat use of therapeutic endoscopy.

Conclusion: Although this analysis did not consider comorbidity from recurrent bleeding, effect of rSLE on length of hospital stay, or use of adjuvant therapy with proton pump inhibitors, the results suggest that rSLE is not indicated following therapeutic endoscopy for bleeding PUD.  However, if the risk for rebleeding exceeds 20%, then SLE reduces the risk of rebleeding at no additional cost.