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.

A two-yearly, population-based mammographic screening programme in women aged 50 to 69 years.

Screening.

Cost-effectiveness analysis.

Women aged 50-69 years.

Secondary care. The economic study was carried out in Norway.

Effectiveness and resource use data were mainly obtained from the Norwegian Mammography Project (NMP) annual report published in 1996 (the inclusion of women in the project started in January 1996). Some missing effectiveness and resource use data were obtained from reports published between 1983 and 1997. The price year was 1997.

Effectiveness data were derived from a single study, a literature review, and the author's assumptions.

Costing was retrospectively performed mainly on the same patient sample as that used in the effectiveness analysis. Some missing data were obtained from other studies or reports.

Power calculations were not used to determine the sample size. The study sample consisted of 46,329 women who entered the screening programme. Initially 60,147 subjects were invited giving a compliance rate of 77%.

This was a retrospective cohort study, carried out in 4 counties. The duration of follow-up in the NMP programme was not explicitly specified. Loss to follow-up was not reported.

The principle used in the analysis of effectiveness (intention to treat or treatment completers only) was not explicitly reported. The health outcome measures were the rate of recall mammography (subdivided into recall due to clinical findings, technical difficulties, and/or suspected cancer on mammography), cases of breast cancer revealed, the rate of false positives, the size of tumours, the proportion of breast cancer patients having axillary lymph node metastases, the proportion of breast cancer patients undergoing breast conserving strategy (BCS) including postoperative radiotherapy, and a modified radical mastectomy (MRM).

The rate of recall mammography was 6% (subdivided into recall due to clinical findings, 0.4%; technical difficulties, 0.9%; and/or suspected cancer on mammography, 4.7%). 337 cases of breast cancer were revealed giving a detection rate of 0.7%. In terms of the rate of false positives, five out of six suspected cases ended up not having breast cancer. The size of tumours was less than or equal to 15 mm, while 18% of breast cancer patients had axillary lymph node metastases. This suggested that the NMP target of a 30% reduction in breast cancer mortality would be achievable. Of the 337 cases of breast cancer detected and treated, 42% had a breast conserving strategy (BCS) including postoperative radiotherapy (2-opposite tangential field technique, 50 Gy in 25 fractions), the others were treated with a modified radical mastectomy (MRM) and an axillary lymph node dissection.

A 30% reduction in breast cancer death was calculated from this survey.

A model was used to estimate the costs and future benefits of screening although the details of the model employed were not given. The description of the non-screening strategy was based on national data on assessment and treatment.

The following outcomes were assessed from a review of the literature:

frequency of stage I and II breast cancers, median life expectancy for women aged 60 and 69, proportion of breast cancer cases occurring in women aged over 60 years, proportion of breast cancers occurring in premenopausal women and corresponding rate of axillary lymph node involvement, proportion of postmenopausal women having 4 or more affected lymph nodes, median life expectancy for patients saved from the death from breast cancer by screening, median delay time for the diagnosis of breast cancer in the no-screening strategy, median survival when contracting an incurable breast cancer, annual number of women dying in Norway because of breast cancer, frequency of MRM in the no-screening strategy, frequency of adjuvant hormonal therapy (AHT) in the screening and non-screening situation, proportion of patients with breast cancer fulfilling the criteria for adjuvant chemotherapy, frequency of adjuvant chemotherapy (CMF-nine cycles) in the screening and no-screening situation, risk of a new breast cancer in the BCS group, mortality rate due to other causes of death during 10 year follow-up, from 60 to 70 years, and proportion of patients going through 10 years follow-up.

Not reported.

Not reported.

Not reported.

Not reported.

A total of 5 studies was directly included in the review, and 3 more studies were indirectly incorporated in the review.

Outcomes were mainly based on individual studies.

Not reported.

The frequency of stage I disease was approximately 57% and for stage II disease was approximately 28%. Median life expectancy for women aged 60 was 23.4 years and for those aged 69 was 16 years. Two-thirds of breast cancer cases occur in women aged over 60. Premenopausal women accounted for about 30% of all breast cancers and 40% had axillary lymph node involvement. Among postmenopausal women about 5% had 4 or more affected lymph nodes. The median life expectancy for patients rescued from mortality due to breast cancer by screening was estimated to be 20 years compared to 5 years for the no-screening strategy. This indicates 15 life years gained per prevented breast cancer death, which was comparable to other European studies. The median delay time for the diagnosis of breast cancer in the no-screening strategy was 2 years. The median survival when contracting an incurable breast cancer was estimated to be 3 years. The annual number of women who die in Norway because of breast cancer was about 800, corresponding to an incidence rate of 38% for this malignancy. The frequency of MRM in the no-screening strategy was about 15.5% versus 7% in the screening strategy. 16% were calculated to undergo adjuvant hormonal therapy (AHT) if screened and 30% if those not screened. About 35% of patients aged under 55 years who have breast cancer fulfil the criteria for adjuvant chemotherapy. A total of 7% were calculated to undergo adjuvant chemotherapy (CMF-nine cycles) in the screened group and 13% if not screened. The risk of a new breast cancer in the BCS group was about 1% for at least 10 years. The mortality rate due to other causes of death during 10 year follow-up, from 60 to 70 years, was 8.3%. Whereas 80% of patients in the no-screening situation were considered going through 10 years follow-up, the corresponding figures when screening was used were raised because of reduced mortality to 86%.

Assumptions about effectiveness were made by the authors.

The frequency of BCS in the treatment of breast cancer in the no-screening strategy (after a 2-year delay in diagnosis) was 25%. The frequency of TI tumours less than 20 mm was assumed to be 70% in the screening scenario and 45% in the no-screening scenario.

Life years saved were estimated as the benefit measure, based on epidemiological data from different reports, modelling, and assumptions made by the authors. A 10 year follow-up period was considered.

Costs were discounted. Some quantities were reported separately from the costs. Cost components were reported separately. The cost analysis covered the costs of screening, breast cancer surgery (breast conserving surgery (BCS) or mastectomy), breast cancer radiotherapy (BCS or mastectomy), adjuvant therapy (hormonal and chemotherapy), and follow-up. The perspective adopted in the cost analysis was that of the health care system. The source of resource use data was the NMP report and other reports published between 1983 and 1997. 1997 price data were used. The treatment cost of patients dying from cancer in a no-screening situation was not considered because of lack of availability of sufficient data.

Costs were discounted. Quantities were reported separately from the costs. The cost items were reported separately. The indirect cost calculation covered the production losses caused by absence from work (1/3 day off work) caused by screening for breast cancer. The calculations were based on the mean income of Norwegian workers in 1996 and a rise of 4% in income from 1996 to 1997. The source of the data was Statistics Norway. 1997 price data were used.

Norwegian Kroner (NOK). A conversion to UK pounds sterling () was performed at a rate of NOK11.00 = 1.00

A one-way sensitivity analysis was conducted based on a change in the gain in life years. Threshold analysis was performed based on two cut-off values employed in the literature (12,000 and 24,000).

Based on the epidemiological data, 15 life years were saved per prevented breast cancer death (based on a median life expectancy of 20 years for a rescued patient versus 5 years for a patient in a no-screening situation). The total benefit estimated from the screening programme amounted to 408 years based on 38 lives saved. The discount rate applied was 5% with a 10 year follow-up period.

The discount rate was 5%. The total cost for the screening programme totalled 4,928,840 compared to 1,435,996 in the no-screening option, leading to a cost of 75.40 per patient screened and 58.10 per patient invited.

Incremental cost per life year (LY) saved was calculated as the main cost-effectiveness measure, leading to a cost per life year of 8,561. The cut-off points for the life years saved to meet the 24,000 and 12,000 cut-off values employed in the literature were nine and four years respectively. The cost per cancer detected was 10,365.

Mammography screening in Norway seems to be cost-effective.

The reason for the choice of the comparator is clear.

Whilst the methodology adopted was appropriate, the internal validity of the benefit results cannot be guaranteed due to the retrospective nature of the study design adopted in the single study, plus the lack of a comprehensive literature review (although it was reported that the data related to the no-screening strategy were based on national statistics), and absence of a reported quality assessment of the primary studies included in the review. No specification was provided for the model used to estimate the future benefits associated with each strategy.

Quantities were reported in general forms separately from the costs. Adequate details of the methods of cost estimation were given. The cost results may not be generalisable to other settings due to the lack of sensitivity analysis on costs. Given the inclusion of indirect costs, it appears that the societal perspective was taken in the cost analysis rather than that of the health care system as reported by the authors.

The author's conclusion appears to be justified given the fact that uncertainty surrounding the benefit results was addressed by conducting sensitivity analysis. The issue of generalisability to other settings or countries was not addressed although appropriate comparisons with other studies were made. The disutility resulting from psychological distress caused by a false positive mammography was not incorporated in the benefit calculations of the screening programme.

The author urges that time has come to encourage the widening of national screening programmes for breast cancer.

None stated.