This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

Endoscopic surveillance of Barrett's oesophagus to detect malignancy in an early and curable stage.

Screening, diagnosis, and treatment.

Cost-effectiveness analysis.

Patients with Barrett's oesophagus.

Hospital. The economic study was carried out in the United States.

Effectiveness and resource use data for breast cancer surveillance were collected during 1994. Effectiveness and resource use data for Barrett's oesophagus were collected during the period 1980 to 1995. 1995 prices were used.

The estimate for final outcomes was derived from a single study and from a review of previously published studies.

Costing was undertaken retrospectively on the same patient sample as that used in the effectiveness analysis.

For breast cancer, 12,537 mammograms were performed and 267 patients underwent needle-localisation biopsies. For Barrett's oesophagus, 149 patients were seen of whom 13 had adenocarcinoma when first seen. Power calculations were not used to determine the sample size.

This was a retrospective case series carried out in a single centre: cancer detection rates for both groups were determined by retrospective record review of patients followed at the authors' hospital. Patients with benign Barrett's oesophagus were followed for 510 patient years. Loss to follow up was not reported.

The analysis of the clinical study was based on intention to treat. The primary health outcome used in the analysis was the number and percentage of cancers detected and 'cured' patients.

For breast cancer 50 of the 267 needle-localisation biopsies proved to be malignant, thus yielding a positive biopsy rate of 19% and a rate of occult malignancy detection of 0.4%. It was calculated that 43 of the 50 patients (86%) would be cured.

The 136 patients with benign Barrett's oesophagus underwent 354 endoscopic examinations with biopsy; seven adenocarcinomas were detected during the follow-up period. All seven patients underwent oesophagogastrectomy; five were alive without disease at 1, 2, 4, 6 and 7 years postoperatively, and two had died. From the literature it was assumed that 76% (5/7) were cured.

The authors' reported that surveillance endoscopy has been shown by this study and others to reliably detect cancer in an early and curable stage, resulting in lives saved.

Operative mortality, five year survival after surgery for adenocarcinoma in Barrett's oesophagus detected by surveillance endoscopy, and cure rate for screening mammography were assessed via a literature review.

Not stated.

Not stated.

Not stated.

Not stated.

Six studies.

Mean values were calculated based on the results reported in the studies.

Not stated.

The mean five year survival after surgery for adenocarcinoma in Barrett's oesophagus was 76% and operative mortality was 0%. Cure rate for screening mammography was 86 %.

Cancers detected, cancers cured, and life-years gained were the outcome measures used in the economic analysis.

Quantities and costs were analysed separately. Costs were not discounted in the baseline analysis, although the time frame of the analysis was more than 2 years. The estimation of quantities was based on actual data and on the authors' assumptions regarding the proportion of patients likely to undergo various forms of therapy. Patient charges for 1995 were used as cost estimates. Only direct costs related to detection and therapy aimed at cure were included. Costs for breast cancer detection included mammography charges for the entire group, and the cost of excisional biopsy for those found to have positive mammograms. The cost of curing a cancer included detection costs and the cost of subsequent cancer therapy. Biopsy costs for patients with positive mammograms but with benign biopsies were included. The cost of therapy, which included adjuvant chemotherapy for both groups, was based on the assumption that half of those found to have cancer would undergo lumpectomy and radiation therapy and that the other half would undergo mastectomy. Surveillance and therapy for adenocarcinoma in Barrett's oesophagus included endoscopy and biopsy charges for the entire screened group and the cost of oesophagogastrectomy for those found to have cancer, which included the diagnosis of high grade dysplasia. No adjuvant therapy was included in the charges.

Not included.

US dollars ($).

The sensitivity analysis substitutes for the average values used in the initial calculations, the extreme values of the incidence of adenocarcinoma in Barrett's, mammographic cancer detection rates, and cure rates for oesophageal resection. Discount rates were varied from 0 to 7%.

For breast cancer, 50 cancers (0.4%) were detected and the cure rate was 86%. Given a relative risk of 0.74 (9.4 cancer deaths in 50 patients compared with the 7 deaths assumed for the surveillance group) and a life expectancy gain of 18 years per patient, the 2.4 cancer deaths averted by a surveillance strategy would yield 43.2 life-years gained.

For Barrett's oesophagus, 7 cancers were detected and 76% of them were cured. Assuming that surveillance increased the cures from 1.4 of seven to 5.3 of seven patients (a relative mortality risk of 0.26) then the number of life years gained is 62.7. The side-effects of treatment were not considered in the economic analysis. Benefits were not discounted.

For breast cancer, at $125 per mammogram and including the surgical biopsy cost of $4,339 each for 267 patients with positive mammograms, a total of $2,725,638 was expended. Assuming that therapy would be divided equally between lumpectomy and radiation ($21,437) and mastectomy ($12,800), a total cost of $3,581,563 would be incurred. The cost of no surveillance was $1,079,125. For Barrett's oesophagus the total cost for 354 endoscopies with biopsy would be $265,500 (at a unit cost of $750), and the cost of resecting seven patients would be $151,200 (unit cost $21,600). This represents a total cost of $416,700. The cost of no surveillance was $156,450.

For breast cancer the cost per case detected was $54,513 and the cost per cure would be $83,292. The cost per life year gained would be $57,926. For Barrett's oesophagus assuming that all seven patients underwent resection at a cost of $21,600 per case, and that five patients, or 76% were cured, the total cost per cure would equal $83,340 and the cost per life-year gained would be $4,151. The only assumption in the sensitivity analysis that led to a cost-effectiveness advantage for mammographic surveillance was that treatment of all 50 mammographically detected tumours resulted in cures in women who would otherwise have died within 1 year of breast cancer (cost per life year saved of $3,233 for mammography and $5,490 for endoscopy).

The cancer incidence of patients with Barrett's oesophagus was 1,370 cases per 100,000 population per year, amounting to a yearly cancer risk of 1.37%. This is substantial, especially for young patients, and warrants surveillance, the cost of which is adequately recovered in years of life saved by early intervention. The cost-effectiveness of Barrett's oesophagus surveillance results from the high cancer incidence, the ease of cancer detection at an early stage, and the marked difference in surgical outcome between surveillance-detected and clinically detected cancers.

The reason for the choice of no screening as a comparator is clear.

The sample size was not determined using power calculations. Some of the effectiveness evidence was derived from the literature, but it is not clear whether the review of the literature was systematic. Additionally, no justification was provided for the use of mean values of the studies' findings included in the review. As a result, the results of this study should be treated with caution.

Resource quantities were reported separately from prices and adequate details of the methods of quantity and cost estimation were given. Costs might not be generalisable to other settings or countries.

Sensitivity analyses were used to account for the uncertainties in the data. Appropriate comparisons were made with other studies.

None stated.