|
Two view mammography at incident screens: cost effectiveness analysis of policy options |
Johnston K, Brown J |
|
|
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 Use of two view mammography at incident (subsequent) screens in breast screening programmes.
Economic study type Cost-effectiveness analysis.
Study population Women undergoing mammography the age range for which, for women in the UK, is 50-64.
Setting Secondary care. The economic study was carried out in the UK.
Dates to which data relate Effectiveness and resource use data corresponded to women screened between 1 April 1996 and 31 March 1997. Some of the resource use data were obtained from studies published in 1995 and 1996. The fiscal year was 1996/97.
Source of effectiveness data Effectiveness data were derived from a single study.
Link between effectiveness and cost data Costing was retrospectively performed mainly on the same patientsample as that used in the effectiveness analysis.
Study sample Power calculations were not used to determine the sample size. The number of programmes implementing the policy of one view single reading was 17 with a total of 133,902 women screened; for one view double reading (recall if one reader suggests), the number was 23 with a total of 194,502 women screened. The corresponding figures for the policies of one view double reading (consensus), and one view double reading (arbitration) were 14 and 6 with respective totals of 110,755 and 64,520 women screened. A total of 23 (with a total of 211,483 women screened) and 4 programmes with the policies of one view double reading (complex recall policy) and two views at incident screens were excluded from the study. All programmes used two view mammography at the prevalent screens.
Study design This was a retrospective cohort study, carried out on a nation-wide scale. The duration of the follow-up was until the end of reading procedure. No loss to follow-up was reported.
Analysis of effectiveness The principle used in the analysis of effectiveness was intention to treat. The outcome measures were incident invasive cancer (<15 mm and >=15 mm). The data were collected from returns (KC62 forms)completed by individual programmes and supplemented by information gathered via a questionnaire addressing the number of readers and views used by individual programmes. The standardized detection ratio (SDR), the ratio of the observed to the expected number of invasive cancers, was calculated as an indirect standardization measure. Poisson regression was used to adjust for confounding by age. The comparison was made with respect to a "base level" of a single reading.
Effectiveness results The observed cases of invasive cancers <15 mm at the incident screens with one view mammography were:
policy of single reading, 192;
double reading (recall if one reader suggests), 383;
double reading (consensus), 235;
double reading (arbitration), 163.
The respective values for standardised detection rate (SDR) statistic were 0.68, 0.93, 1.00, and 1.18.
The corresponding values for the observed cases of invasive cancers >=15 mm at the incident screens were 198, 307, 168, and 107. The respective values for the SDR statistic were 0.86, 0.91, 0.88, and0.95.
The differences in the cases of invasive cancers >=15 mm were not found to be significant.
The small (<15mm) SDR in order of increasing detection above one view/single reading were as follows (for two view programmes, the source prevalent screening data):
one view/single reading, 0.68;
one view/double reading (recall if one suggests), 0.93;
two view/single reading, 0.97;
one view/double reading (consensus), 1.00;
two view/double reading (recall if one reader suggests), 1.05;
two view/double reading (consensus), 1.12;
one view/double reading (arbitration), 1.18;
and two view/double reading (arbitration), 1.28.
Clinical conclusions The results from this study (the original study by Blanks et al, 1998). suggest that those programmes using double reading, and specifically double reading with arbitration, detect considerably more small invasive cancers than programmes using a single reading.
Measure of benefits used in the economic analysis The additional invasive cancers (<15 mm) detected per 10,000 women screened was the estimated measure of benefit based on the original results from the study by Blanks et al (1998).
Direct costs Costs were not required to be discounted due to the short time frame of the study. Some quantities were reported separately from the costs. The unit costs of screens and costs of assessment were reported separately. The cost analysis covered the costs of staff time (taking, processing, and reading of films and administering results), consumable items (films and chemicals), overhead, and capital. The perspective adopted in the cost analysis was that of the health service. Some resource use data were obtained from the literature (1995 and 1996). The source of unit cost data was a discussion paper published in 1996. The hospital and community pay and price index was used to upgrade all costs to the fiscal year chosen for the study. The date to which the price data referred was 1996-97.
Sensitivity analysis A set of one-way sensitivity analyses was performed by considering alternative values for the number of minutes required to take and process the second view, the proportion of films requiring consensus or arbitration, recall rate, and the number of cancers detected.
Estimated benefits used in the economic analysis The additional cancers detected (95% confidence intervals) per 10,000 women screened for 43 programmes were:
one view double reading (recall if one reader suggests) switching to two views, double reading (recall if one reader suggests), 6 (95%CI: 3 - 8);
one view double reading (consensus) switching to two views, double reading (consensus), 6 (95% CI: 3 - 10);
one view double reading (arbitration) switching to two views, double reading (arbitration), 6 (95% CI: 3 - 10).
The additional cancers detected for 17 programmes using one view/single reading for a change to an alternative policy from the baseline of one view single reading were:
11 (95% CI: 4 - 18) for one view/double reading (recall if one reader suggests);
12 (95% CI: 4 - 23) for two views/single reading;
14 (95% CI: 6 - 23) for one view/double reading (consensus);
16 (95% CI: 7 - 26) for two views/double reading (recall if one reader suggests);
19 (95% CI: 9 - 31) for two views/double reading (consensus);
21(95% CI: 12 - 34) for one view/double reading (arbitration);
and 26 (95% CI: 14 - 31) for two views/double reading (arbitration).
After excluding the inefficient policies for 17 programmes using one view/single reading, the corresponding recalculated values for a change to policies of one view/double reading (recall if one reader suggests), one view/double reading (arbitration), and two views/double reading (arbitration) were 11, 10, and 5, respectively. After excluding the inefficient options again, the values amounted to 21 for changes to one view/double reading (arbitration), and 5 for two views/double reading (arbitration).
Cost results The additional costs per 10,000 women screened for 43 programmes using one view/double reading were 39,538, 40,296, and 39,917 in the cases of switching from the baseline one view/double reading with recall policies of recall if one reader suggests, consensus, and arbitration to corresponding two views policies, respectively.
The additional costs for 17 programmes using one view/single reading for a change to an alternative policy from the baseline of one view single reading were:
4,034 for one view/double reading (recall if one reader suggests);
35,504 for two views/single reading;
4,792 for one view/double reading (consensus);
43,572 for two views/double reading (recall if one reader suggests);
45,089 for two views/double reading (consensus);
4,413 for one view/double reading (arbitration);
and 44,330 for two views/double reading (arbitration).
After excluding the inefficient policies for 17 programmes using one view/single reading, the corresponding recalculated values for a change to policies of one view/double reading (recall if one reader suggests), one view/double reading (arbitration), and two views/double reading (arbitration) were 4,034, 379, and 39,917, respectively. After excluding the inefficient options again, the values were 4,413 for changes to one view/double reading (arbitration), and 39,917 for two views/double reading (arbitration).
Synthesis of costs and benefits An incremental cost-effectiveness ratio was calculated by dividing additional costs by additional invasive cancers detected. The incremental cost-effectiveness ratios for 43 programmes using one view/double reading were 6,589, 6,716, and 6,652 in the cases of switching from the baseline one view/double reading with recall policies of recall if one reader suggests, consensus, and arbitration to corresponding two views policies, respectively.
The corresponding values for 17 programmes using one view/single reading for a change to an alternative policy from the baseline of one view single reading were:
366 for one view/double reading (recall if one reader suggests);
2,959 for two views/single reading;
342 for one view/double reading (consensus);
2,723 for two views/double reading (recall if one reader suggests);
2,373 for two views/double reading (consensus);
210 for one view/double reading (arbitration);
1,705 for two views/double reading (arbitration).
After excluding the inefficient policies for 17 programmes using one view/single reading, the corresponding recalculated values for a change to policies of one view/double reading (recall if one reader suggests), one view/double reading (arbitration), and two views/double reading (arbitration) were 366, 38, and 7,983, respectively. After excluding the inefficient options (relative to the next best option) again, the values were 210 for changes to one view/double reading (arbitration), and 7,983 two views/double reading (arbitration), respectively. The sensitivity analysis established the relative robustness of the results.
Authors' conclusions The cost-effectiveness of two view mammography at incident screens depends on the film reading policy. A policy of two view mammography at incident screens in England and Wales would be efficient only if programmes using single reading moved to double reading.
CRD COMMENTARY - Selection of comparators A justification was given for the choice of the comparator. It was the policy of choice in England and Wales at the time of the study. You, as a database user, should consider whether this is a widely used health technology in your own setting.
Validity of estimate of measure of benefit The internal validity of the estimates of benefit can not be reasonably guaranteed due to the observational nature of the study, a fact acknowledged by the authors of the original study.
Validity of estimate of costs Some quantities were reported separately from the costs and adequate details of methods of cost estimation were given. Cost results may not be generalizable to other settings or countries.
Other issues The authors' conclusions seem to be justified given that the uncertainties surrounding the parameters of the study were addressed in sensitivity analysis. The issue of generalisability to other settings or countries was not systematically addressed. Appropriate comparisons with other studies were not made.
Implications of the study The authors indicate that the implementation of two view mammography at incident screens in programmes in England and Wales would cost 2.9 million and require 13.4 whole time equivalent radiologists. Given limited resources, priority should be given to introducing double reading in the subset of programmes currently using single reading as this requires fewer additional radiologists and is more cost-effective. A constraint on available radiologists may suggest that attention should be focused instead on alternative reading modalities, such as automated or radiographer reading, but these would need to be evaluated formally in terms of their effectiveness and cost effectiveness.
Source of funding Funded by the NHS breast screening programme.
Bibliographic details Johnston K, Brown J. Two view mammography at incident screens: cost effectiveness analysis of policy options. BMJ 1999; 319: 1097-1102 Other publications of related interest Blanks R G, Wallis M G, Moss S M. A comparison of cancer detection rates achieved by breast cancer screening programmes by number of readers, for one and two view mammography: results from the UK National Health Service breast screening programme. Journal of Medical Screening 1998;5:195-201.
Indexing Status Subject indexing assigned by NLM MeSH Breast Neoplasms /economics /prevention & Cost-Benefit Analysis; England; Female; Health Policy; Health Priorities; Humans; Mammography /economics /methods; Mass Screening /economics /organization & Middle Aged; Program Evaluation; Sensitivity and Specificity; Wales; administration; control AccessionNumber 21999008299 Date bibliographic record published 31/03/2000 Date abstract record published 31/03/2000 |
|
|
|