TOTAL COST-EFFECTIVENESS OF MAMMOGRAPHY SCREENING STRATEGIES

Abstract

Background

Given improvements in breast cancer screening technology and treatment over the past decade, we sought to evaluate the cost-effectiveness f breast cancer screening in Canada.

Data and Methods

Using the Canadian breast cancer incidence, mammography performance and cost data, we adopted the Wisconsin CISNET breast cancer simulation model to the Canadian population. We estimated the costs, life years gained (LYG), quality-adjusted life years (QALYs) for 11 mammography screening scenarios varying by age and frequency. The cost-effectiveness analysis was conducted from the Canadian societal perspective. Incremental cost effectiveness ratios were presented. Sensitivity analyses assessed the robustness of model conclusions.

Results

The overall societal cost associated with No Screening was $4.8 M ($4,875 per woman) and ranged between $7.6 to $16.0M for other active screening scenarios over a lifetime time horizon. Stepwise incremental cost-effectiveness analysis showed that triennial screening (50–69 years) was the most cost effective at $83,070 per LYG and $94,762 per QALY. Biennial ($87,420 per LYG and $97,006 per QALY) and annual ($184,654 per LYG and $226,278 per QALY) scenarios had higher incremental ratios.

Interpretations

Both the benefits (life years gained, QALY) and costs of screening rise almost linearly with the number of lifetime screens per women. The decision on how to screen in Canada is influenced by the willingness to pay and the rate of recalls for further examinations after positive screens.

Introduction

With its single payer publicly funded health system, the implementation of a population-wide screening program in Canada has significant budget implications because of the utilization of substantial health system resources. Decisions regarding whether to screen, who to screen, what modalities should be used, and how frequently screening should occur are best made when there is an understanding of the trade-offs between improved health outcomes, potential harm, and financial costs of the intervention.

The economic value of mammography screening is of particular importance in countries with socialized health systems or single payer systems. Mammography screening recommendations, in particular, are continually being updated and modified over different countries and over time^1–3. The age range, frequency, effectiveness and cost-effectiveness for population mammography screening programs continue to be hot topics of debate^1–3.

While a number of studies have examined the costs for the treatment of breast cancer from the Canadian perspective, none to our knowledge have examined or incorporated the full cost of screening in Canada^4–7. Moreover, a number of breast cancer natural history models to project the impact of different mammography screening scenarios have been developed^8–13. However, none of the models have evaluated the value of mammography to society, which is an important issue in a country such as Canada, where screening for breast cancer has been offered through organized government funded screening programs. The objective of this work, therefore, was to evaluate the costs, outcomes and cost-effectiveness of mammography over different screening scenarios in women from the societal perspective, using a validated computer breast cancer natural history simulation model¹⁴.

Data and Methods

Model

We adapted the University of Wisconsin Breast Cancer Epidemiology Simulation (UWBCS) Model¹⁴ to the Canadian context. Developed under the US National Cancer Institute-funded Cancer Intervention and Surveillance Modeling Network (CISNET) program¹⁵, the model is a discrete-event simulation model originally designed and calibrated to predict US breast cancer incidence and mortality trends from 1975–2000 and then extended to match observed incidence and mortality until 2010. The details of the model logic and components have been described elsewhere and are available at www.cisnet.cancer.gov/breast/¹⁶.

In a companion paper, we describe the model assumptions and input data for the Canadian version and present model predictions related to health outcomes, while here, we describe inputs specific to resource use and analyze the cost-effectiveness of screening.¹⁷ We examined 11 screening scenario, consisting of annual, biennial, and triennial regimens across different age bands (starting at 40 or 50 and ending at 69 or 74years) for women plus a no screening scenario. The results presented here are based on calculations for a 1960 birth cohort.

Efficacy

Clinical model inputs are described in full in the original model¹¹ and details of the Canadianized modified inputs are provided in the companion paper¹⁷

Resources

Resources associated with screening, diagnosis, and cancer management were identified based on the Canadian publicly funded health system environment and included mammography, clinic visits, physicians, diagnostic procedures, and treatment (surgery, radiation and medication). To account for the societal perspective, lost productivity related to screening, screening results, and diagnosis and the cost of premature death were included. Once identified, quantities of resources utilized during the management of women were determined through guidelines, reports, peer-reviewed literature, and expert opinion (Table 1). In the base-case, we assumed that 100% of eligible participants would be screened using digital mammography. We assumed that all positive screens and all detection of suspicious findings outside of screening incurred a non-invasive workup cost and a subset of these incurred invasive workup. It was assumed that all women took time off from their workplace related to mammography and in the first year after a diagnosis. Productivity loss associated with premature death due to breast cancer was also counted.

Table 1.

Resource Utilization and Cost Input Parameters

Variables	Definitions and Descriptions	Resource Utilization (RU)	Costs (2012CAN$)
Screening and Diagnosis
Breast Cancer Screening program	Resource Utilization (RU): Assume all women eligible would be screened- Cost: Includes mammography physician based on the Ontario Health Insurance Plan (OHIP)29	100% of women eligible for screening	$183.00 per screen
Non-invasive Workup	RU: Assume all women with a positive mammogram Cost: Includes work-up mammography, radiology, physician clinic visit based on OHIP29	100% of women who have a positive screening exam	$445.95 per work-up
Invasive Work-up-Needle Biopsy	RU: Of the women who get a non-invasive workup, 14.7% subsequently receive an invasive procedure based on the Ontario Breast Screening Program (OBSP)30 Cost: physician clinic visit, needle biopsy procedure and pathology based on OHIP29	82.1% of women who receive an invasive work-up.	$745.46 per work-up
Invasive Work-up-Excisional	RU: Of the women who get a non-invasive workup, 14.7% receive an invasive procedure based on OBSP30; Cost: physician clinic visits, excision, pathology based on OHIP29	17.8% of women who receive an invasive work-up.	$1,652.44 per work-up
Treatment
Chemotherapy	RU: Proportion of women receiving chemotherapy; Cost: Mean value of 1st, 2nd and 3rd generation chemotherapies based on expert opinion, Cancer Care Ontario (CCO)31, and Sunnybrook Pharmacy Department32.	Women receiving chemotherapy with invasive cancer-see details below Women receiving chemotherapy with DCIS=0%-see details below	$7,376.10 per course
Trastuzumab	RU: Incidence of Trastuzumab-assume all women who are HER2+ received Trastuzumab; Cost: Assume that treatment Costs are based on 8 cycles; Includes chemotherapy costs recommended for Trastuzumab (paclitaxel); includes health care personnel costs and physician clinic visits associated with administration. Based on ASCO abstract33, CCO31, and Sunnybrook Pharmacy Department32	14% with invasive cancer 0% with DCIS	$29,709 per course
Tamoxifen	RU: Assume utilization of tamoxifen for appropriate population over 10 year time horizon. Cost: annual cost (excludes markup+dispensing)+physician clinic visit 4 times per year. Based on the Ontario Drug Benefit Formulary (ODBF)34	100% for invasive cancer based in eligible women 100% for DCIS in eligible women	$383.40 per annum
Aromatase Inhibitors (AI)	RU: Assume utilization of AI for appropriate population over 10 year time horizon; Cost: annual cost of letrozole (excludes markup and dispensing) + physician clinic visits 4 times per year. Based on the ODBF34.	100% for invasive cancer in eligible women 0% for DCIS	$822.40 per annum
Treatment Cost per Appropriate Cohort
ER+, <50 years, DCIS, annual cost	RU: tamoxifen only (over 10 year time horizon); Cost: tamoxifen only based on the ODBF34	100%	$383.40
ER+, <50 years, Invasive, annual cost	RU: tamoxifen (over 10 year time horizon)+chemotherapy; Cost: tamoxifen (over 10 year time horizon)++chemotherapy based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100%	$7,759.50
ER+, <50 years, Invasive, annual cost with Trastuzumab	RU: tamoxifen (over 10 year time horizon)+chemotherapy+Trastuzumab; Cost: tamoxifen (over 10 year time horizon)+chemotherapy+Trastuzumab. Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100%	$37,462.50
ER+, >=50 years, DCIS, annual cost	RU: tamoxifen (over 10 year time horizon)+only; Cost: tamoxifen only based on ODBF34	100% of women in this cohort received this regimen	$383.40
ER+, >=50 years, Invasive, annual cost	RU: chemotherapy+AI (over 10 year time horizon); Cost: chemotherapy+AI (over 10 year time horizon). Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100% of women in this cohort received this regimen	$8,198.50
ER+, >= 50 years, Invasive, annual cost with Trastuzumab	RU: chemotherapy+AI (over 10 year time horizon)+Trastuzumab; Cost: chemotherapy +AI (over 10 year time horizon)+Trastuzumab. Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100% of women in this cohort received this regimen	$37,901.50
ER-, <50 years, DCIS, annual cost	No drug therapy. Based on expert opinion.	100% of women in this cohort received this regimen	$0
ER-, < 50, Invasive, annual cost	RU: chemotherapy; Cost: chemotherapy. Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100% of women in this cohort received this regimen	$7,376.10
ER-, < 50 years, Invasive, annual cost with Trastuzumab	RU: chemotherapy+Trastuzumab; Cost: chemotherapy+Trastuzumab Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100% of women in this cohort received this regimen	$37,079.10
ER-, >=50 years, DCIS, annual cost	No drug therapy. Based on expert opinion	100% of women in this cohort received this regimen	$0
ER-, >=50 years, Invasive, annual cost	RU: chemotherapy; Cost: chemotherapy. Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department31 and expert opinion.	100% of women in this cohort received this regimen	$7,376.10
ER-, >=50 years, Invasive, annual cost with Trastuzumab	RU: chemotherapy+Trastuzumab; Cost: chemotherapy+Trastuzumab Based on the ODBF34, CCO31, Sunnybrook Pharmacy Department32 and expert opinion.	100% of women in this cohort received this regimen	$37,079.10
Procedures
Radiation for invasive cancer	RU: Proportion of women with breast cancer who received radiation therapy. Based on Mittmann et al35. Cost: 25 fractions * $138 (1996) or 188.39 (2012). Based on Bank of Canada19, Earle et al36, and OHIP29.	67%	$5,014.05 per radiation
Surgery for invasive cancer	RU: Proportion of women with breast cancer receiving surgery based on Mittmann et al37l.	90%	Costs stratified by lumpectomy and mastectomy
Surgery-lumpectomy for invasive cancer	RU: Proportion of all women receiving surgery; Cost: OCCI 2012. Based on Ontario Case Cost Initiative (OCCI)38.	63% The joint report by the Canadian Institute for Health Information (CIHI) and the Canadian Partnership Against Cancer show the annual rates for the two treatments vary widely from one province to another.	$4,937.06 per surgery
Surgery-mastectomy for invasive cancer	RU: Proportion of all women receiving surgery Cost: OCCI 2012. Based on OCCI38.	37% The joint report by the Canadian Institute for Health Information (CIHI) and the Canadian Partnership Against Cancer show the annual rates for the two treatments vary widely from one province to another.	$6,956.77 per surgery
Radiation for DCIS	RU: Proportion of women with DCIS who received radiation therapy Cost: 25 fractions * $138 (1996) or 188.39 (2012). Based on Bank of Canada19, Earle et al36, and OHIP29.	50%	$5,014.05 per radiation
Surgery-lumpectomy for DCIS	RU: Proportion of DCIS women receiving surgery, expert opinion40 Cost: OCCI 2012. Based on OCCI38, Canadian Institute for Health Information (CIHI) Report39 and expert opinion40	67%	$4,937.06 per surgery
Surgery-mastectomy for DCIS	RU: Proportion of all DCIS receiving surgery Cost: OCCI 2012. Based on OCCI38, CIHI Report39, and expert opinion40	33%	$6,956.77 per surgery
Indirect Costs
Time off for screening	RU: Assume 4 hours for women based on travel, appointment and waiting time; Cost: $22.87/hour for women. Based on Statistics Canada 201241.	100%	$91.48 per time off for screening
Lost productivity in first year	RU: Assumption, all women working; women lost 27% of annual income due to breast cancer treatment based on Lauzier 200842; Cost: $22.87/hour for women. Based on Statistics Canada 201241	100%	$12,838.18 per annum
Premature death costs	RU: Assumption, all women who died Cost: $22.87/hour for women* 40 hours * 52 weeks. Based on Statistics Canada 201241

RU=resource utilization

All women received either mastectomy or lumpectomy with and without radiation. Chemotherapy was assigned by stage of disease. Hormonal treatment was assigned by hormone receptor status. Traztuzumab was assigned by HER2 status. Women with ductal carcinoma in situ (DCIS) did not receive trastuzumab (Table 1).

Costs

Canadian unit costs ($2012CAN) were applied to each of the resources utilized and modeled ($1.01US= $1CAN based on Dec 31, 2012)¹⁸. Non-2012 costs were converted to 2012 values using the Consumer Price Index¹⁹. A number of cost sources were used, including provincial drug formularies, national statistics programs, costing programs, health resource management, and the published literature. Capital or institutional costs of the equipment were not included in this analysis. Lost productivity costs were based on the literature and the average Canadian wage per women (Table 1).

Utility Values

We used age-specific population health preference values²⁰ derived from US Medical Expenditure Panel Survey data applying the EuroQoL EQ-5D instrument using US scoring²¹. For women with newly diagnosed breast cancer, we assumed decrements to quality of life based on stage lasting for one year post-diagnosis after which a women would return to her appropriate age-specific value. For a women diagnosed with Stage IV breast cancer, the decrement was applied to her remaining lifetime. For screening-related health states, disutilities of 0.006 for one week and 0.105 for five weeks were applied for a screening mammogram and positive screening respectively²².

Analyses

When estimating the lifetime costs of screening and management for each screening scenario and no screening to calculate the overall cost to society and for premature death for each of the different screening scenarios where no discount rates were used to determine overall lifetime costs. Outcomes were LYG and QALY. We compared the costs and outcomes from each screening scenario in an incremental stepwise cost-effectiveness analysis as follows: a scenario was considered to be efficient when there was no alternative scenario with improved outcomes for a lower or at the same cost or when there was no combination of two other screening scenarios that improved outcomes for the same costs. Incremental cost-utility ratios (ICURs) were computed as the difference in cost divided by the difference in outcomes only for the efficient scenarios. We also determined the marginal (or average) cost-effectiveness of each screening relative to the no screening scenario. All costs and health outcomes for the incremental analyses were discounted at a 5% rate. We carried out univariate sensitivity analyses in which we varied the inputs for cost and utilization of key parameters such as screening rates, screening sensitivity and specificity, health preference, treatment costs, cost of medications and discount rate (Table 2).

Table 2.

Univariate Sensitivity Analyses

Variable	Base Case Value	Sensitivity Analysis Value	Source and Comments
Cost of mammography	$183.00 per screen	$100.00 per screen	Assumption-Reduction in cost
Percent of women eligible for screening who missed screening	0%	50%	In Ontario, 60% of women over age 50 participate in regular screening, but about 43% do this through the organized program. The retention rate for women continuing screening after a previous screen is approximately 85%43.
Screening rates for population	100% first screen	No subsequent screens	Assumption: Impact of repeat screening/retention
Sensitivity	Model calibrated as per empirical data44;45	100%	Assumption: Ideal scenario
Specificity	as per empirical data	100%	Assumption: Ideal scenario
Treatment	Surgery, radiation, chemo, hormonal	0% treated	Assumption applied to all women in model
Adjuvant chemotherapy	100% of cohort received adjuvant chemotherapy	75% of cohort received chemotherapy	Assumption: not all patients would receive chemotherapy
Utility values	Model values	+/− 25%	Assumption: change in utility values
Disutility	Model value	No disutility	Assumption: change in utility values
Discount Rate	Model rate	0%	Assumption applied to all women in model

Results

From the societal perspective, overall costs (undiscounted) for annual screening for 1,000 women ranged from $11.3 (50 to 69 years) to $16.0M (40 to 74 years). In contrast, costs for biennial and triennial screening for 1,000 women were between $8.4–$11.2M (50–69; 50–74 years) and $7.6–$8.3M (50–60; 50–74 years), respectively. The overall cost associated for the no screening scenario in 1,000 women was $4.8 M ($4,875 per woman) over a lifetime time horizon (Table 3). Active screening itself was a system cost driver and represented a significant portion of the overall cost to the health system and the societal cost burden. The ratio of the cost of screening to overall cost was directly proportional to the aggressiveness of the overall screening scenario. Treatment costs were slightly higher for active screening scenarios when compared to No Screening. Screening costs were higher proportion-wise when compared to treatment and procedure costs for more aggressive scenarios. Lowering the age of screening to 40 from 50 years of age, added approximately $2–3M for 1,000 women ($2,000–$3,000 lifetime cost per woman). Adding 5 years to the upper limit of 69 years, also added approximately $1M per 1,000 women ($1,000 per woman) to the total cost. If the cost of premature death was included in the societal cost, it would increase the overall cost by an additional $7.4M to $10.8M depending on the screening scenario, with more aggressive screening being associated with fewer lives lost lives and thus lower costs associated with premature deaths.

Table 3.

Component costs for each Screening Scenario

Screening Scenario	Screening Costs	Diagnostic and Clinical Workup Costs	Surgical and Radiation Costs	Adjuvant Medication Costs1	Indirect Costs2	Overall Societal Cost (excluding	Additional Cost of Premature Death
No Screening	$0	$83,936	$1,220,608	$1,713,473	$1,856,569	$4,874,587	$12,671,495
Digital Triennial 50–69	$1,573,325	$61,405	$1,406,127	$1,821,409	$2,724,921	$7,587,187	$10,085,986
Digital Triennial 50–74	$1,911,210	$53,721	$1,477,251	$1,864,335	$2,964,934	$8,271,452	$9,778,164
Digital Biennial 50–69	$2,175,956	$55,863	$1,437,689	$1,807,948	$3,019,322	$8,496,778	$9,261,094
Digital Biennial 50–74	$2,672,157	$46,018	$1,524,021	$1,849,832	$3,350,174	$9,442,203	$8,866,323
Digital Annual 50–69	$4,127,472	$47,245	$1,487,568	$1,713,469	$3,904,655	$11,280,409	$7,847,010
Digital Annual 40–49	$2,484,365	$75,590	$1,276,581	$1,705,534	$2,848,195	$8,390,266	$10,795,828
Digital Annual 50–74	$4,908,452	$37,013	$1,575,399	$1,724,528	$4,364,947	$12,610,339	$7,408,591
Digital Annual 40–49, Biennial 50–69	$4,551,101	$49,334	$1,479,841	$1,788,107	$3,985,553	$11,853,937	$7,552,892
Digital Annual 40–49, Biennial 50–74	$5,027,968	$39,529	$1,567,832	$1,832,766	$4,316,621	$12,784,716	$7,130,482
Digital Annual 40–69	$6,540,531	$40,605	$1,530,563	$1,694,982	$4,864,284	$14,670,965	$6,180,986
Digital Annual 40–74	$7,314,735	$30,529	$1,618,139	$1,707,003	$5,322,003	$15,992,409	$5,794,587

¹Adjuvant medication costs included adjuvant chemotherapy, hormonal therapies (e.g., tamoxifen and aromatase inhibitors), and traztuzumab.,

²Based on lost productivity

For the cost-effectiveness analysis, screening scenarios for women between 50 and 69 were dominant over scenarios where the stopping age for screening was 74 years. Annual screenings for women between 40–69 and 40–74 were the considered the dominant treatment when compared to annual scenarios using narrower age ranges.

When compared to the no screening scenario, all marginal cost-effectiveness (CE) ratios for the societal perspective were under $140,000/LYG and $155,000/QALY. The lowest incremental ratio was calculated for the least frequent screening with the tightest age band (triennial screening for 50–69 year olds) $79,035/LYG and $94,762/QALY. The most aggressive scenario (annual screening for 40–74 year olds) was associated with the highest incremental ratio $132,932/LYG and $154,187/QALY (Table 4; Figure 1).

Table 4.

Cost-effectiveness Different Screening Scenarios from the Societal Perspective^*

Screening Scenario	Total Societal Cost (per 1,000 women)	Total QALYs (per 1,000 women)	Incremental Cost per QALY	Average Cost per QALY relative to no screening scenario
No Screening	$1,387,948	14,059	–	Basecase
Triennial 50–69	$2,511,486	14,071	$94,762	$94,762
Triennial 50–74	$2,652,005	14,072	dominated	$99,670
Biennial 50–69	$2,885,565	14,075	$97,006	$95,313
Biennial 50–74	$3,083,486	14,076	dominated	$100,912
Annual 50–69	$3,083,486	14,082	$159,858	$114,778
Annual 40–49	$4,156,667	14,073	dominated	Higher cost; worse outcome
Annual 50–74	$3,970,099	14,083	$226,278	$120,521
Annual 40–49, Biennial 50–69	$5,572,348	14,088	dominated	$146,333
Annual 40–49, Biennial 50–74	$5,767,102	14,089	dominated	$146,951
Annual 40–69	$6,668,051	14,094	$216,828	$150,938
Annual 40–74	$6,943,065	14,095	$262,828	$154,187

^*Cost and QALYs discounted at 5%.

Figure 1.

Incremental Cost-Effectiveness Plane for Societal Perspective. Value expressed per 1,000 women alive at age 40.

Sensitivity Analyses

When comparing active screening scenarios to No Screening in the marginal analyses, we found that the major cost drivers were no treatments (including drug, surgery, or radiation) following a diagnosis and no subsequent screening (Table 5). The model was generally insensitive to the changes in missed screening and no subsequent screening. But showed more favourable incremental ratios when there was a reduction in the proportion of individuals receiving adjuvant chemotherapy, when screening costs were decreased and when specificity and sensitivity. Not providing treatment (no medication, radiation, surgery), progression had the biggest impact on the ICER. The model was sensitive to modifications in utility values, showing more favourable ICURs when the utility values were increased by 25%, thereby showing greater incremental benefits between the active screening scenarios and No Screening. In contrast, less favourable or higher ICURs were modeled when the utility values were decreased by 25% as a result of smaller differences in the benefit. Lastly, changing the discount rate to 0% decreased the ICURs substantially.

Table 5.

Univariate Sensitivity Analysis: Societal Perspective-Screening Scenarios compared to No Screening where the outcomes are cost per quality adjusted life year (QALY) (discount=5%).

Screening Scenarios	Base Case	Utility Values +25%	Utility Values −25%	Specificity 100%	Sensitivity 100%	50% missed screenings	Disutility excluded	Screening Cost =$100	No Treatments (including drug, surgery or radiation)	No subsequent screening	75% receive adjuvant chemotherapy	Discount rate =0%
No Screening	–	–	–	–	–	–	–	–	–	–	–	–
Digital Triennial 50–69	$94,762	$74,849	$126,349	$78,780	$85,837	$95,995	$107,858	$77,384	$1,963,362	$93,624	$86,364	$41,575
Digital Triennial 50–74	$99,700	$78,812	$132,932	$83,793	$91,707	$100,339	$113,732	$81,684	$2,017,095	$93,624	$89,826	$46,703
Digital Biennial 50–69	$95,313	$75,461	$127,083	$80,679	$83,935	$94,904	$107,993	$76,713	$1,616,400	$93,624	$87,140	$41,984
Digital Biennial 50–74	$100,912	$79,954	$134,549	$86,120	$89,992	$99,722	$114,702	$81,461	$1,870,599	$93,624	$90,735	$47,597
Digital Annual 50–69	$114,778	$91,076	$153,037	$99,421	$101,858	$103,318	$132,970	$89,701	$1,240,537	$93,624	$103,592	$52,077
Digital Annual 50–74	$120,522	$95,675	$160,695	$105,048	$107,224	$108,112	$140,148	$94,347	$1,410,235	$241,662	$107,059	$57,858
Digital Biennial 40–74	$126,612	$100,640	$168,816	$103,579	$107,306	$120,253	$154,502	$100,680	$5,968,018	$241,662	$116,064	$51,194
Digital Annual 40–49, Biennial 50–69	$146,333	$116,540	$195,111	$121,352	$122,534	$135,003	$184,737	$114,221	$5,834,291	$241,662	$135,983	$53,166
Digital Annual 40–49, Biennial 50–74	$146,950	$117,061	$195,934	$123,180	$124,491	$136,750	$184,756	$114,985	$5,968,397	$241,662	$135,473	$55,855
Digital Annual 40–69	$150,937	$120,211	$201,250	$125,147	$128,592	$131,972	$190,762	$116,971	$4,081,165	$241,662	$137,604	$58,900
Digital Annual 40–74	$154,187	$122,810	$205,583	$128,097	$131,177	$134,090	$195,009	$119,646	$4,726,306	$241,662	$138,757	$63,233

Discussion

This is the first economic analysis to evaluate on the cost-effectiveness of mammography screening from a Canadian societal perspective. Since health care funders and policy decision makers require a demonstration of the value of new interventions/strategies through a comparison of costs and benefits, we conducted a lifetime cost-effectiveness and cost-utility analysis for various relevant screening scenarios for this perspective. Early diagnosis of breast cancer through screening mammography can save a woman’s life, however, there are costs, limitations, and possible harms associated with the screening tests and treatments. We tried to determine the value of mammography screening from both the cost and benefit end.

The UWBCS model was selected for the analysis because it allowed the simulation of the growth of a distribution of breast cancers within a cohort of women and the consideration of the individual effects of various detection scenarios and treatment regimens on mortality or other outcomes. The strength of the UWBCS model was that it was validated against empirical data from the US. Modified for use in the Canadian context, it performed quite well in predicting breast cancer incidence in the absence of screening¹⁷. In addition, the model used modern empirical data on the sensitivity and specificity of screening mammography versus age and breast density to describe the screening process. Canadian data on the utilization of therapies and costs were used and no assumptions on the mortality reduction of screening were applied explicitly in the model.

The main overall cost driver for the active screening scenarios was the frequency of screening. We also showed that there was a significant reduction in the cost of premature death with more frequent screening, with an inverse relationship between the cost of screening and the cost of premature death.

Our stepwise incremental cost-effectiveness analysis showed that broader age bands and more frequent screening were dominated indicating that tighter age bands and less frequent screening were optimal when cost-effectiveness rather than lifesaving was considered. The exception is annual screening from 40 to 74 years, which is not dominated but has high ICERs.

In our discounted marginal cost-effectiveness analysis, we showed that all active screening scenarios were more effective than the no screening scenario. All ICERs for active screening scenarios when compared to No Screening scenario generally fell below the well-accepted thresholds of $100,000 per LYG ratio and most scenarios fell below the $150,000 per QALY. The marginal incremental ratios generated from our screening lifetime model are generally in line with other ICERs used in oncology decision-making ($/QALY)²³. Moreover, the incremental ratios are well below the willingness to pay threshold of >$300,000/QALY reported by approximately half of Canadian and American Oncologists defined “good value for money” at incremental ratios been $50,000–$100,000/LYG, with another third using >$100,000/LYG as a good value definition²⁴. A recent editorial on incremental ratios questioned the wisdom of the existing thresholds defining²⁵. A number of studies have evaluated the cost-effectiveness of screening scenarios. Most have been conducted from a US health system perspective and focused on different risk factors such as early and late age and genetic profile^8;10;26;27. However, none have examined the impact of lost productivity from the societal perspective or for that matter a Canadian lens.

Annual scenarios had higher marginal ratios than less frequent scenarios but they were also associated with higher benefits (LYG, QALY). In this analysis, societal costs were based on lost productivity around attending mammography screening mammogram (1/2 day), positive (true and false) screens (5 weeks) and diagnosis of invasive cancer (one year). Any changes in time horizons would affect the overall cost. The more aggressive the screening scenario, the more cancers and false positives are detected, which leads to more lost productivity leading to higher incremental costs and higher incremental ratios. Since both the LYG and QALYs and costs rise together almost linearly with the number of lifetime screens per woman, the decision on how to screen is mainly related to willingness to pay and avoiding recalling too many women for further examinations after positive screens. Thus a screening tool, which would achieve higher specificity, would be attractive.

A comparison of the incremental ratios for all the active screening scenarios compared to no screening showed a relatively tight range of ratios within $40,000–$60,000 of one another. Increasing the age of screening from 69 to 74 years, marginally increased the incremental ratios due to additional screening costs, but improved outcomes. Lowering the age of screening in women from 50 to 40 years, resulted in increased incremental ratios, mostly due to the increased screening costs but also yielded more LYGs and QALYs.

If examined by age band (50–69 and 50 to 74 years), the modelled incremental ratios for annual, biennial and triennial screening scenarios compared to No Screening each showed very similar incremental ratios. When choosing a screening scenario based on the value assessments, one should also consider the improvement in LYGs and QALYs associated with more frequent screening.

For the univariate sensitivity analysis, based on the analyses for the societal perspective (discount=5%), where the different screening scenarios were compared to No Screening for both LYGs and QALYs, the model was generally robust. The ICER model was generally insensitive to changes in missed screening, no subsequent screening and reduction in the proportion of individuals receiving adjuvant chemotherapy. The model was sensitive to decreased screening costs, improved specificity and sensitivity, leading to more favourable ICERs. Not providing treatment (no medication, radiation, surgery) thereby expediting progression had the biggest impact on the ICER increasing it to over $500,000 per LYG. Modifying the same parameters for the ICUR analysis, showed similar results to the ICER analysis but also showed that the model was sensitive to modifications in utility values, predicting more favourable ICURs when the utility values were increased by 25%, thereby showing greater incremental benefits between the active screening and No Screening scenarios. In contrast, less favourable or higher ICURs were generated when the utility values were decreased by 25% as a result of smaller differences in the benefit leader to higher ICUR. Most significantly, ICUR values fell substantially when the no discount rate was applied. Discounting significantly affected the long-term effectiveness outcomes over the model time horizon, while the costs were up front.

As with every model, there are limitations. The model itself is subject to limitations outlined in previous work¹⁷. We assumed that 100% of eligible women were screened in this model whereas in reality, compliance is markedly lower. In Ontario, the most heavily populated province in Canada, screening rates were found to be 61% (2010–2011) of women 50 to 74 years through an organized screening program²⁸. When we decreased the screening rate to 50% in this model, incremental ratios remained similar to those of our basecase, mainly due to the fact that along with the costs for screening decreasing, so did the number of invasive cancers detected affecting LYG and QALYs.

We did not include the cost of premature death in the incremental evaluations to avoid the possibility of double counting. However, given the significance of premature death to society, as well as its costs, we anticipate that more frequent screening would significantly decrease the costs associated with premature death due to breast cancer.

The work done here may be helpful in informing the issue regarding the most appropriate screening scenario for a population. We showed that the greatest single cost contributor in a screening program is the screening mammography itself. This is greater than the costs of therapy in the health system perspective. The more screens that a woman receives in her life, the greater the financial cost to the health care system and to society. Because both the life savings and costs rise together almost linearly with the number of lifetime screens per women, the decision on how to screen is mainly related to willingness to pay and avoiding recalling too many women for further examinations after positive screens. Future models will consider the impact of different screening technologies on costs and outcomes.

Figure 2.

Incremental Cost-Effectiveness Plane for Societal Perspective. Values expressed per 1,000 women.