Key points are not available for this paper at this time.
OBJECTIVE: Surveillance of Barrett's patients is recommended, to detect dysplasia and early cancer. The reported risk for developing cancer varies substantially, however. Our previous analysis used an average cancer incidence of 1/75 patient-years (PY). Recent reports suggest that the risk may range from 1/251 to 1/208 PY in combined series of patients with long segment Barrett's esophagus (LSBE, >3 cm), and short segment Barrett's esophagus (SSBE), and up to 1% annually in patients with SSBE. Our goal was to consider these new estimates of cancer risk in a cost-utility analysis of surveillance of patients with Barrett's esophagus. METHODS: Using our previously published model, we incorporated an average of the recent estimates of cancer risk (0. 4% annually, 1/227 PY), and our primary data on quality of life after esophagectomy. We included actual variable (direct) costs and used a discount rate of 5%. From the perspective of an HMO, the model evaluates surveillance every 1-5 yr and no surveillance, with esophagectomy performed if high grade dysplasia is diagnosed, and calculates the incremental cost-utility ratios for each strategy. RESULTS: The results suggest that, at our baseline, annual cancer risk surveillance every 5 yr is the only viable strategy. More frequent surveillance costs more and yields a lower life expectancy. The incremental cost-utility ratio for surveillance every 5 yr is 98, 000/quality-adjusted life year (QALY) gained, comparable to the incremental cost-effectiveness ratios of accepted practices (heart transplantation and screening for tuberculosis in selected populations, 160, 000/LY gained and 216, 000/LY gained, respectively). CONCLUSIONS: Surveillance of Barrett's patients should extend life, with incremental cost-utility ratios that compare favorably with some accepted medical practices. Policy makers can compare the cost of surveillance to that of other accepted practices to determine their willingness to fund surveillance.
Provenzale et al. (Sun,) studied this question.