|
Retrospective cost-effectiveness analysis of screening mammography |
Stout N K, Rosenberg M A, Trentham-Dietz A, Smith M A, Robinson S M, Fryback D G |
|
|
Record Status 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. Health technology Several strategies for screening mammography and the subsequent treatment of breast cancer (BC) were examined. The primary strategy was screening mammography as implemented in the USA from 1990 to 2000. This consisted of screening all women aged 40 years or older annually or biennially. Sixty-four alternative mammography screening scenarios were considered, each with a particular fixed screening schedule and varying by the age at first screen (40, 45, 50 or 55 years), by the age at the last screen (65, 70, 75 or 80 years), and by the screening interval (1, 2, 3 or 5 years).
Study population The study population comprised a hypothetical cohort of women aged 40 years or older. Different starting ages were considered, depending on the screening strategy.
Setting The setting was primary care. The economic study was carried out in the USA.
Dates to which data relate The effectiveness data were derived from studies published between 1978 and 2006. No dates for resource use were reported. The costs were derived from sources published in 1995 and 2000. The price year was 2000.
Source of effectiveness data The effectiveness evidence came from a synthesis of published studies and authors' assumptions.
Modelling A discrete-event simulation model of BC epidemiology was used to compute the costs and benefits of the alternative screening mammography strategies in a hypothetical cohort of women. The time horizon of the model was lifetime. The model incorporated secular trends in BC risk, screening use and treatment dissemination. This model was developed at the University of Wisconsin as part of the National Cancer Institute's Cancer Intervention and Surveillance Modeling Network. Most details on the decision model were available online. The four main components of the model were natural history of BC, BC detection, BC treatment and BC mortality. Five main health states appear to have been considered. These were healthy, BC diagnosed and treated in situ or localised stage, BC diagnosed and treated in regional stage, distant-stage BC and death. BC was assumed to progress in disease stage according to a stochastic Gompertz-type growth model. BC could be detected by screening mammography or routine clinical detection (self and/or clinical breast examinations). Treatment was based on adjuvant therapy with chemotherapy and/or tamoxifen. The effectiveness of the treatment was a function of age at diagnosis and tumour characteristics. Death could be due to BC or other causes.
Outcomes assessed in the review The outcomes estimated from the literature were:
BC incidence and progression,
BC mortality,
the sensitivity and specificity of mammography,
the detection rates with routine clinical detection,
the effectiveness of treatment (adjuvant therapy),
all-cause mortality, and
quality of life (QoL) weights.
Study designs and other criteria for inclusion in the review It was not stated whether a systematic review of the literature was undertaken to identify the primary studies. Much of the epidemiological data came from the Surveillance, Epidemiology, and End Results (SEER) dataset. The sensitivity and specificity of mammography were estimated from observational studies. All-cause mortality was derived from US life tables. QoL data for healthy women were obtained from the 2000 wave of the Medical Expenditure Panel Survey, a nationally representative survey of health care usage, insurance and costs in the USA. In this survey, QoL scores were estimated using the EuroQol EQ-5D instrument.
Sources searched to identify primary studies Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Observational studies were used to represent the screening strategy historically implemented in the USA.
Number of primary studies included The clinical estimates were derived from 17 primary studies and 2 electronic sources.
Methods of combining primary studies The sensitivity and specificity of mammography were estimated by model calibration of data obtained from observational studies.
Investigation of differences between primary studies Results of the review Most clinical parameters used in the model were not reported. The sensitivity of mammography ranged from 60 to 90%, while its specificity ranged from 88 to 90%.
QoL weights were specified for four health states (healthy, BC diagnosed and treated in the in situ or localised stage, regional BC and distant-stage BC). These utility weights were reported for several age groups (5-year groups from age 30 to more than 85 years).
The utility weights decreased with increasing age. For example, for women aged 30 to 34 years, the QoL weights were:
0.856 for healthy,
0.770 for in situ or localised BC,
0.642 for regional BC,
0.513 for distant BC,
0.852 for a mammogram with a negative result, and
0.841 for a mammogram with a positive result.
At the other extreme, for women aged 85 years or more, the QoL weights were:
0.59 for healthy,
0.531 for in situ or localised BC,
0.442 for regional BC,
0.354 for distant BC,
0.587 for a mammogram with a negative result, and
0.58 for a mammogram with a positive result.
Only mammograms with positive or negative results were included in the sensitivity analysis.
Methods used to derive estimates of effectiveness Some assumptions were made to obtain utility weights for the different health states.
Estimates of effectiveness and key assumptions For in situ or localised BC the utility weight was assumed to be 90% of that for a healthy individual. For regional cancer it was assumed to be 75%, and for distant BC, 60%. Also, a 100% participation rate for all screening strategies was assumed in the base-case.
Measure of benefits used in the economic analysis The summary benefit measure used was the number of quality-adjusted life-years (QALYs). These were estimated by combining survival data and QoL estimates in the decision model. An annual discount rate of 3% was applied.
Direct costs The perspective chosen for the cost analysis was not explicitly stated, although it might have been that of a third-party payer as only direct medical costs were included. Specifically, the analysis considered the costs associated with the screening mammogram, diagnostic follow-up for a positive or abnormal screening result, diagnosis from clinical detection, and treatment. The unit costs were not presented separately from the quantities of resources used for all items. The costs came from published studies, while the source of resource use was unclear. Discounting was relevant, as long-term costs were considered, and an annual rate of 3% was applied. The price year was 2000 and the costs were adjusted to 2000 values using the medical care component of the Consumer Price Index.
Statistical analysis of costs The costs (and also QALYs and cost-utility ratios) were reported as mean values with 95% confidence intervals (CIs).
Indirect Costs The indirect costs were not included in the economic evaluation.
Sensitivity analysis A univariate sensitivity analysis was performed to assess the robustness of the estimated cost-utility ratios to variations in the level of population participation and assumptions about QoL weights. The authors stated that plausible ranges of values were used.
Estimated benefits used in the economic analysis In the cohort of approximately 95 million women in the USA, the expected QALYs (in millions) would be:
945.8 with no screening;
947.5 with the actual screening strategy;
946.5 with screening starting at age 55 years, stopping at age 70 years, and an interval of 5 years;
946.8 with screening starting at age 55 years, stopping at age 70 years, and an interval of 3 years;
947.2 with screening starting at age 50 years, stopping at age 75 years, and an interval of 3 years;
947.4 with screening starting at age 45 years, stopping at age 75 years, and an interval of 3 years;
947.7 with screening starting at age 50 years, stopping at age 75 years, and an interval of 2 years;
948.0 with screening starting at age 45 years, stopping at age 75 years, and an interval of 2 years;
948.4 with screening starting at age 40 years, stopping at age 80 years, and an interval of 2 years;
949.1 with screening starting at age 45 years, stopping at age 75 years, and an interval of 1 year;
949.4 with screening starting at age 45 years, stopping at age 80 years, and an interval of 1 year; and
949.6 with screening starting at age 40 years, stopping at age 80 years, and an interval of 1 year.
Cost results In the cohort of approximately 95 million women in the USA, the expected costs (in billions) would be:
$103 with no screening;
$166 with the actual screening strategy;
$121 with screening starting at age 55 years, stopping at age 70 years, and an interval of 5 years;
$130 with screening starting at age 55 years, stopping at age 70 years, and an interval of 3 years;
$144 with screening starting at age 50 years, stopping at age 75 years, and an interval of 3 years;
$151 with screening starting at age 45 years, stopping at age 75 years, and an interval of 3 years;
$160 with screening starting at age 50 years, stopping at age 75 years, and an interval of 2 years;
$170 with screening starting at age 45 years, stopping at age 75 years, and an interval of 2 years;
$187 with screening starting at age 40 years, stopping at age 80 years, and an interval of 2 years;
$223 with screening starting at age 45 years, stopping at age 75 years, and an interval of 1 year;
$237 with screening starting at age 45 years, stopping at age 80 years, and an interval of 1 year; and
$252 with screening starting at age 40 years, stopping at age 80 years, and an interval of 1 year.
Synthesis of costs and benefits Incremental cost-utility ratios were calculated to combine the costs and benefits of the alternative screening strategies. Dominated strategies (those which were less effective and more expensive) were excluded, which left 11 of the 64 alternative strategies. The authors pointed out that the current screening strategy was dominated by some alternatives, but its incremental cost per QALY compared with no screening was approximately $37,000.
The incremental cost per QALY was:
$27,000 with screening starting at age 55 years, stopping at age 70 years, and an interval of 5 years;
$28,000 with screening starting at age 55 years, stopping at age 70 years, and an interval of 3 years;
$31,000 with screening starting at age 50 years, stopping at age 75 years, and an interval of 3 years;
$31,000 with screening starting at age 45 years, stopping at age 75 years, and an interval of 3 years;
$34,000 with screening starting at age 50 years, stopping at age 75 years, and an interval of 2 years;
$34,000 with screening starting at age 45 years, stopping at age 75 years, and an interval of 2 years;
$47,000 with screening starting at age 40 years, stopping at age 80 years, and an interval of 2 years;
$49,000 with screening starting at age 45 years, stopping at age 75 years, and an interval of 1 year;
$53,000 with screening starting at age 45 years, stopping at age 80 years, and an interval of 1 year; and
$58,000 with screening starting at age 40 years, stopping at age 80 years, and an interval of 1 year.
The sensitivity analysis showed that the results of the analysis were sensitive to assumptions about population participation. With a rate of participation of 50% (it was 100% in the base-case), the current screening strategy approached the efficiency frontier and hence was cost-effective.
The second sensitivity analysis on the variations in QoL values showed that the inclusion of small decrements to QoL associated with screening participation had a negative affect on the strategies with more frequent screening tests.
Authors' conclusions The current screening strategy for the detection of breast cancer (BC) in the USA was cost-effective in comparison with no screening. However, the analysis revealed that there is room for improvement, although a more efficient use of resources implies less intensive and less inclusive screening strategies, which might not be acceptable to the general public.
CRD COMMENTARY - Selection of comparators The rationale for the choice of the comparators was clear and was appropriate given the objective of the study. Several screening strategies were compared, including the current strategy in the USA. You should decide whether these are valid comparators in your own setting.
Validity of estimate of measure of effectiveness The effectiveness data used to populate the decision model were obtained from published studies. However, it was unclear whether these studies were identified selectively since the methods and conduct of a systematic review were not reported. With the exception of SEER data, the information on the design of the primary studies was limited and details on patient samples and follow-up were not given. However, the authors chose observational data to estimate the incidence of BC and the accuracy of mammography, in order to reflect actual US data, and to include retrospectively the costs and effectiveness of the actual screening programme in the USA. This choice was appropriate given the objective of the study. The issue of uncertainty was addressed in the sensitivity analysis, although this was restricted to only two clinical inputs (participation rate and QoL values).
Validity of estimate of measure of benefit QALYs were the most appropriate benefit measures because they capture the impact of the interventions on both quality of life and survival, which are the most relevant dimensions of health for women with BC. The use of QALYs allows comparisons to be drawn with the benefits of other health care interventions. The instrument used to derive utility was reported for healthy individuals, but other BC health states were based on authors' assumptions, although the utility weights used were similar to some published weights. Discounting was applied, as recommended by US economic evaluation guidelines.
Validity of estimate of costs The perspective adopted in the study was not explicitly stated, although it might have been that of a third-party payer. The authors stated that the inclusion of indirect costs would have reduced the amount of money required for increasing participation, and thus would have favoured all screening strategies. Details of the unit costs and quantities of resources used were not presented. A detailed breakdown of the cost items was not given and some costs were presented as macro-categories, which will hinder the replication of the analysis in other settings. The cost estimates were specific to the study setting and no alternative costing approach was used in the sensitivity analysis. Statistical analyses were carried out and CIs around the cost estimates were calculated. This represents a strength of the cost analysis. The price year was reported, which will assist any reflation exercises in other time periods.
Other issues The authors stated that comparisons with the results from other studies were difficult, owing to differences in underlying assumptions. However, in general, the current analysis corroborates the conclusions of other studies that demonstrated the cost-effectiveness of screening mammography over no screening. The issue of the generalisability of the study results to other settings was not addressed. As only few sensitivity analyses were carried out, the external validity of the study was low. The analysis revealed the importance of including decrements in utility associated with screening mammography in the modelling exercise.
Implications of the study The study results appear to support the cost-effectiveness of current BC screening strategies. The implementation of a more efficient screening option would have a large impact in terms, for example, of increased participation. This aspect could have strong budgetary implications for health authorities. The authors noted that future studies should differentiate among women according to the presence of risk factors.
Source of funding Supported by the Agency for Healthcare Research and Training, the National Cancer Institute, and the Harvard Center for Risk Analysis.
Bibliographic details Stout N K, Rosenberg M A, Trentham-Dietz A, Smith M A, Robinson S M, Fryback D G. Retrospective cost-effectiveness analysis of screening mammography. Journal of the National Cancer Institute 2006; 98(11): 774-782 Other publications of related interest Mandelblatt J, Saha S, Teutsch S, et al. The cost-effectiveness of screening mammography beyond age 65 years: a systematic review for the US Preventive Service Task Force. Ann Intern Med 2003;139:835-42.
Lindfors KK, Rosenquist J. The cost-effectiveness of mammography screening strategies. JAMA 1995;274:881-4.
US Preventive Service Task Force. Screening for breast cancer: recommendations and rationale. Ann Intern Med 2002;137:344-6.
Indexing Status Subject indexing assigned by NLM MeSH Adult; Aged; Breast Neoplasms /economics /mortality /radiography; Cost-Benefit Analysis; Direct Service Costs /statistics & Female; Humans; Mammography /adverse effects /economics; Mass Screening /adverse effects /economics /methods; Middle Aged; Practice Guidelines as Topic; Quality of Life; Quality-Adjusted Life Years; Retrospective Studies; Sensitivity and Specificity; United States /epidemiology; Women's Health; numerical data AccessionNumber 22006001422 Date bibliographic record published 31/12/2006 Date abstract record published 31/12/2006 |
|
|
|