Estimating personal costs incurred by a woman participating in mammography screening in the National Breast and Cervical Cancer Early Detection Program

Abstract

BACKGROUND.

The National Breast and Cervical Cancer Early Detection Program (NBCCEDP) covers the direct clinical costs of breast and cervical cancer screening and diagnostic follow-up for medically underserved, low-income women. Personal costs are not covered. In this report, the authors estimated personal costs per woman participating in NBCCEDP mammography screening by race/ethnicity and also estimated lifetime personal costs (ages 50-74 years).

METHODS.

A decision analysis model was constructed and parameterized by using empiric data from a retrospective cohort survey of mammography rescreening among women ages 50 years to 64 years who participated in the NBCCEDP. Data from 1870 women were collected from 1999 to 2000. The model simulated the flow of resources incurred by a woman participating in the NBCCEDP. The analysis was stratified by annual income into 2 scenarios: Scenario 1, <$10,000; and Scenario 2, from $10,000 to <$20,000. Sensitivity analyses were conducted to appraise uncertainty, and all costs were standardized to 2000 U.S. dollars.

RESULTS.

In Scenario 1, for all races/ethnicities, a woman incurred a 1-time cost of $17 and a discounted lifetime cost of $108 for 10 screens and $262 for 25 screens; in Scenario 2, these amounts were $31 and from $197 to $475, respectively. In both scenarios, a non-Hispanic white woman incurred the highest cost. The sensitivity analyses revealed that >70% of cost incurred was attributable to opportunity cost.

CONCLUSIONS.

Capturing and quantifying personal costs will help ascertain the total cost (ie, societal cost) of providing mammography screening to a medically underserved, low-income woman participating in a publicly funded cancer screening program and, thus, will help determine the true cost-effectiveness of such programs. Cancer 2008. Published 2008 by the American Cancer Society.

Screening mammography may reduce mortality rates associated with breast cancer by 16% to 30%.1-3 Because of the effectiveness of mammograms in reducing breast cancer mortality, the U.S. Preventive Services Task Force, the National Cancer Institute, the American Cancer Society, and other medical organizations recommend mammography screening every 1 to 2 years for women aged ≥40 years.2, 4-6 Rates of mammography have increased over the past decade, but there is still an estimated 25% gap between screening rates of insured and uninsured women.7 One possible explanation for this gap is that mammography screening is relatively expensive; for example, a recent study by the Centers for Disease Control and Prevention (CDC) that excluded costs incurred by patients reported an average cost of $105 (range, $74-136) per woman screened for breast cancer using mammography.8

To eliminate or reduce financial barriers, the National Breast and Cervical Cancer Early Detection Program (NBCCEDP) was established by the U.S. Congress in 1990 to provide free cancer screening to medically underserved, low-income women.9-11 The NBCCEDP is administered by the CDC through cooperative agreements with health departments in the 50 states, the District of Columbia, 5 U.S. territories, and 12 American Indian/Alaska Native tribal organizations.9, 10

The NBCCEDP covers all costs of providing direct clinical services, which include screening, clinical breast examination, diagnostic follow-up for abnormal results, and case management, as well as the costs for supporting activities, such as program management and evaluation and professional development and recruitment. The program does not cover personal costs, such as opportunity costs and transaction costs. Opportunity costs may include costs associated with travel time, waiting time, time spent receiving screening services, distance traveled to and from screening, loss of leisure time, and loss of productivity in the workplace or at home. Transaction costs are out-of-pocket costs and may include transportation, childcare and/or dependent care, parking, and other expenses related to participation.

The NBCCEDP is limited to women who are at or below 250% of the federal poverty guidelines; thus, any personal costs may be a barrier to participation. These costs, particularly opportunity costs, rarely are considered by policy makers when making decisions on the costs of screening. Consistent with the recommendation of the U.S. Panel on Cost-Effectiveness,12 these costs should be ascertained when estimating the total costs of providing breast cancer screening through the NBCCEDP. The objectives of the current study were 1) to estimate personal costs incurred by a woman aged 50 years to 64 years participating in mammography screening in the NBCCEDP in 4 states by race/ethnicity and 2) to estimate the lifetime personal costs that this woman may incur if screening started at age 50 years and ended at age 74 years. The estimated costs from this study can be combined with program costs to obtain the societal costs of mammography screening in the NBCCEDP.

MATERIALS AND METHODS

Data Sources

Data were obtained from several sources. The Survey of Mammography Rescreening was a retrospective cohort study among women who had a mammogram in 1997 through the NBCCEDP in Maryland, New York, Ohio, or Texas. These programs were chosen because the combined racial/ethnic distributions of their enrollees were most similar to the racial/ethnic distribution of women participating in the NBCCEDP mammography screening program between 1991 and 1998.13 Details of this survey, including the methods, have been described previously.13, 14 Briefly, participants were interviewed by telephone at least 30 months after their 1997 index mammogram (between July 1999 and November 2000). This analysis was restricted to women ages 50 years to 64 years, which is the NBCCEDP priority age group (n = 1870 women). From the survey data, we examined responses to 4 questions: 1) ‘Did you have to take time off from work or volunteer activities to have the mammogram?’; 2) ‘Did you have to take time off from caring for children, grandchildren, or other family members to have the mammogram?’; 3) ‘Did you pay for transportation?’; and 4) ‘Did you pay for parking at the place where you had the mammogram?’ We used SUDAAN statistical software15 to derive point estimates and 95% confidence intervals (95% CIs).

Participants were asked for their household earnings during the past 12 months. Because these responses were not stratified by race/ethnicity, we could not obtain accurate mean annual earnings by these classifications. Accordingly, we requested a special data tabulation from the U.S. Census for households with similar annual earnings stratified by race/ethnicity and age group for the Year 2000.16 We converted the mean annual earnings into hourly wage rates by using the standard of the Bureau of Labor Statistics: 2080 work hours per year.17 We adjusted the estimated wage rate by 22.4% to include fringe benefits and to account for total employee compensation.18

Decision Analysis Model

We constructed a standard decision analysis model to simulate the flow of resources incurred by a woman participating in the NBCCEDP mammography screening program (Fig. 1). The model was constructed using Decision Analysis by TreeAge Pro 2007 (TreeAge Software Inc., Williamstown, Mass), and we parameterized the model by using the data presented in Table 1. Responses to the earnings data (annual income) were dichotomized as <$10,000 (Scenario 1) and from $10,000 to <$20,000 (Scenario 2). In each scenario, we calculated personal costs according to the following racial/ethnic groups: non-Hispanic white, non-Hispanic black, Hispanic, and non-Hispanic other (included Asian, American Indian/Alaska Native, and Native Hawaiian and Other Pacific Islander). All costs were standardized to 2000 U.S. dollars.

Figure 1. Decision analytic model for a woman aged ≥50 years participating in mammography screening in the National Breast and Cervical Cancer Early Detection Program. This is a simplified schematic of the decision tree model that was used to calculate personal costs incurred per woman. Each race/ethnicity was modeled separately. Plus signs indicate that the tree was collapsed, and additional branches were not displayed.

Table 1. Input Variables, Baseline Values, and Ranges Used in Sensitivity Analyses

Variable	Baseline value (Range)				Source
	Non-Hispanic white	Non-Hispanic black	Hispanic	Non-Hispanic other
SMR indicates Survey of Mammography Rescreening; U.S. BLS, U.S. Bureau of Labor Statistics; NBCCEDP, National Breast and Cervical Cancer Early Detection Program; GSA, General Services Administration; U.S. DOT, U.S. Department of Transportation. * SMR, 1999-2000. † NBCCEDP screening and tracking data, April 2006 submission of Minimum Data Elements, 1999-2004. ‡ Assumed a cost of $1.50 (range, $0.50-$3). This assumption was based on responses from focus group research on SMR in which 1 of the focus group members told others that parking tickets could be stamped to reduce the cost from $8 to $3.
Percentage of women by race/ethnicity	41.50 (40.50-42.50)	16 (15.20-17.60)	34 (33.50-36.10)	9.10 (8.60-10.20)	SMR*
Percentage of women who paid for child/dependent care time	5.70 (4.10-7.60)	9.70 (6.50-13.80)	14.80 (11.07-18.30)	12.40 (7.40-19.10)	SMR*
Percentage of women who paid for transportation	10 (7.70-12.80)	31.1 (25.5-37.2)	25.70 (21.80-30)	38.90 (30-48.30)	SMR*
Percentage of women who paid for parking	6.20 (4.40-8.50)	5.70 (3.10-9.60)	6.10 (3.90-9)	12.80 (6.80-21.30)	SMR*
Percentage of women who lost time from work/leisure or other activities	1	1	1	1	Assumed
Percentage of women who used public mass transit	1.80 (0.01-10)	1.80 (0.01-10)	1.80 (0.01-10)	1.80 (0.01-10)
Percentage of fringe benefit	22.40	22.40	22.40	22.40	Grosse 200318
Total time worked per y, h	2080	2080	2080	2080	U.S. BLS 200317
Patient time spent participating in the program, including travel round–trip, waiting, and examination time, h	2.16 (1.38-2.50)	2.16 (1.38-2.50)	2.16 (1.38-2.50)	2.16 (1.38-2.50)	Lairson 2005,19 Secker-Walker 199920
Travel time index	1.39	1.39	1.39	1.39	Schrank&Lomax 200221
Distance traveled round-trip, miles	33.25 (20.40-46.10)	33.25 (20.40-46.10)	33.25 (20.40-46.10)	33.25 (20.40-46.10)	Secker-Walker 1999,20 Guidry 199722
Recall rate for ages 50-64 y, %	17.40 (17.10-17.70)	17.50 (17.10-18)	16.0 (15.70-16.30)	12.10 (11.40-12.80)	NBCCEDP†
Hourly wage for ages 50-64 y, $
Scenario 1: Annual earnings <$10,000	5.88 (5.75-6.01)	5.39 (5.15-5.64)	4.40 (4.09-4.71)	6.18 (5.53-6.83)	U.S. Census Bureau 200416
Scenario 2: Annual earnings $10,000 to <$20,000	11.45 (11.21-11.70)	10.51 (10.03-11.99)	8.57 (7.97-9.17)	12.03 (10.77-13.30)	U.S. Census Bureau 200416
Hourly wage for age ≥65 y, $
Scenario 1: Annual earnings <$10,000	3.04 (2.84-3.25)	3.23 (2.09-4.38)	2.64 (2.06-3.22)	3.14 (2.36-3.91)	U.S. Census Bureau 200416
Scenario 2: Annual earnings $10,000 to <$20,000	5.93 (5.55-6.34)	6.30 (4.07-8.54)	5.15 (4.01-6.28)	6.11 (4.60-7.62)	U.S. Census Bureau 200416
Travel cost per mile, $	0.325	0.325	0.325	0.325	GSA 200028
Hourly cost of child/dependent care, $	11.20 (11-11.41)	11.20 (11-11.41)	11.20 (11-11.41)	11.20 (11-11.41)	U.S. Census Bureau 200416
Parking cost, $	1.50 (0.50-3)	1.50 (0.50-3)	1.50 (0.50-3)	1.50 (0.50-3)	SMR*‡
Cost of transportation per mile, $	0.19 (0.13-0.31)	0.19 (0.13-0.31)	0.19 (0.13-0.31)	0.19 (0.13-0.31)	U.S. DOT 200129
Discount rate, $	0.03 (0-0.05)	0.03 (0-0.05)	0.03 (0-0.05)	0.03 (0-0.05)	Gold 199612
No. of mammograms received per lifetime	10-25	10-25	10-25	10-25	Makuc 200731

Calculation of Costs Incurred per Woman

By using the decision model, the calculation of costs incurred per woman involved 3 steps.12 In the first step, we identified all the resources that a woman participating in a federally funded mammography screening program may incur. These resources were categorized as opportunity and transaction costs.

Then, we measured the amount of each resource used. We obtained the average total time (2.16 hours; range, 1.38-2.50 hours) for travel (round trip), waiting, and examination from the reports by Lairson et al and Secker-Walker et al.19, 20 We adjusted the travel time (round trip) by using the travel time index of 1.39 as reported by Schrank and Lomax.21 This index is the ratio of travel time during peak periods to time during free flow (off traffic). We obtained the distance traveled round trip, which was measured as 33.25 miles (range, 20.4-46.1 miles), from the literature.20, 22 By using estimates from past studies, the percentage of women who used mass transit as the mode of travel for mammography screening was 1.80% (range, 01%-10%).19, 22, 23, 24 We assumed that all women who participated in the screening program, regardless of whether they lost time from work or from leisure activities, incurred some opportunity cost.25, 26 Thus, we accounted for the value of time for all participants regardless of employment status; we measured the value of their time by using the estimated hourly market wage rate as recommended by Luce et al.27 (Table 1).

In the third step, we assigned a monetary value to each of these resources. For travel, waiting, and examination time, we used the mean hourly wage rate stratified by race/ethnicity as presented in Table 1. Travel cost was based on federal mileage reimbursement rates in the Year 2000 of $0.325 per mile.28 The average cost of transit fare per mile was $0.19 (range, $0.13-0.31).29 For women who reported needing child/dependent care to attend mammography screening, we used an hourly wage rate of $11.20 (range, $10.99-11.41), which we obtained from the special tabulation from the Census and adjusted for fringe benefits. Because of the dearth of parking cost data for hospitals and clinics, we assumed a cost of $1.50 (range, $0.50-3.00) based on the parking cost information from the survey of mammography rescreening (SMR) focus group. From the preceding steps, we estimated costs incurred per woman by using the following equation:

• (1)

Recall Rate

Some women who participate in the screening program may be required to return for further evaluation. The recall rate is the proportion of women returning for either immediate diagnostic follow-up or for follow-up within 9 months of the index mammogram, which may be for diagnostic imaging, ultrasound, clinical examination, or biopsy. In this study, we adjusted the estimated cost per woman associated with recall using the recall rate obtained from the NBCCEDP surveillance database from 1999 through 2004, which ranged from 12.1% to 17.5% (Table 1).

Calculation of Lifetime Costs Incurred per Participant

In estimating lifetime personal costs, we made several assumptions. First, we assumed that a woman would receive a mammogram 10 to 25 times between age 50 years and age 74 years.30 Second, the cost incurred per woman at age 50 years is constant up to age 64 years. Third, when a woman aged 50 years reaches age 65 years, she may incur lower personal cost from participating based on the possibility of Medicare coverage and retirement at age 65 years.31 Finally, cost incurred was discounted at an annual rate of 3% to account for the present value (PV) of the lifetime costs incurred (in sensitivity analysis, we varied this rate from 0% to 5%).12 On the basis of these assumptions, the PV of the lifetime costs incurred per woman was estimated by using the following equation:

• (2)

in which subscript r/e denotes each racial/ethnic group, C indicates the cost incurred at ages 50 years through 64 years; T indicates age 64 years; C′ indicates the cost incurred at ages 65 years through 74 years; T′ indicates age 74 years; and r indicates the annual discount rate.

Sensitivity Analyses

We performed both univariate and multivariate sensitivity analyses to assess the impact of each individual variable or of multiple variables on the estimated baseline results. In the univariate analysis, we examined how the estimated personal costs per woman changed when we varied 1 parameter value at a time and held other parameters at their baseline values. We varied the hourly wage rate by assuming that all racial/ethnic groups had uniform wage rates. In addition, we varied the proportion of women who used mass transit; the proportion who paid for child/dependent care, parking, or transportation; parking cost; round-trip travel time; waiting and examination time; distance traveled round trip; and the discount rate.

In the multivariate sensitivity analyses, using the upper and lower bounds of the variables presented in Table 1, we simultaneously varied the values in the opportunity cost variables, which included travel (round trip), waiting, examination, and lost productivity time and the distance traveled (in miles). For an additional check, we performed probabilistic sensitivity analysis in which the values for all the variables in Table 1 were varied simultaneously.32, 33 We constructed probability distributions for each variable in Table 1. We assumed a log-normal distribution for hourly wage rates for each racial/ethnic group, uniform distribution for the discount rate, and, for the remaining variables, we assumed a Program Evaluation Research Technique (PERT) distribution.12, 34 The PERT distribution is a special case of a β distribution, which is defined by the minimum, most likely, and maximum values for each of the variables.12, 34 We used combinations of these probability distributions to perform simulation using Latin hypercube sampling methods assuming that the parameters were independent of each other.34 One thousand independent runs were performed. On each run, the model randomly selected a different value for each variable from its associated distribution. The results from the simulations are presented as means with 5th and 95th percentiles.

RESULTS

Under the baseline analysis, for all racial/ethnic groups, the estimated personal cost incurred was $17.45 per woman with an annual income of <$10,000 (Scenario 1) and $31.19 for someone with an annual income from $10,000 to <$20,000 (Scenario 2) (Table 2). The discounted lifetime cost was $108.47 (10 screens) or $261.60 (25 screens) in Scenario 1 and $196.67 or $474.55, respectively, in Scenario 2. For 1-time screening, a non-Hispanic white woman incurred the highest personal cost in both scenarios: $6.82 (Scenario 1) and $12.60 (Scenario 2). Discounted lifetime costs were $42.36 (10 screens) and $102.16 (25 screens) in Scenario 1 and $78.87 and $190.26, respectively, in Scenario 2. For a 1-time screening, a non-Hispanic woman of other race had the lowest personal cost: $1.69 (Scenario 1) and $2.98 (Scenario 2).

Table 2. Baseline Results of Personal Costs Incurred per Woman by Income and Race/Ethnicity

Race/ethnicity	No. of mammography screenings in a lifetime: cost, $
	1	10*	25†
* Assumed that a woman received a mammogram on average of every 2.5 years from age 50 years to age 74 years. Estimated costs were discounted at an annual rate of 3%. † Assumed that a woman received annual mammography each year from age 50 years to age 74 years. ‡ All races/ethnicities include women with unknown race/ethnicity.
Scenario 1 (annual income <$10,000)
All races/ethnicities‡	17.45	108.47	261.60
Non-Hispanic white	6.82	42.36	102.16
Non-Hispanic black	2.66	16.90	40.78
Hispanic	5.22	32.76	79.05
Non-Hispanic other	1.69	10.38	25.03
Scenario 2 (annual income $10,000 to <$20,000)
All races/ethnicities‡	31.19	196.67	474.55
Non-Hispanic white	12.60	78.87	190.26
Non-Hispanic black	4.74	30.55	73.76
Hispanic	8.77	56.51	136.46
Non-Hispanic other	2.98	18.60	44.88

Sensitivity Analyses

In the univariate sensitivity analyses that assumed a uniform wage rate for all racial/ethnic groups, the overall estimated cost per woman decreased by 8.4% in Scenario 1, but there was no effect in Scenario 2 (Fig. 2). Personal costs incurred by a non-Hispanic black woman decreased by nearly 3% in Scenario 1, but they increased by 6% in Scenario 2. For a non-Hispanic white woman and a non-Hispanic woman of other race, personal costs incurred decreased by >2% in both scenarios. The estimated personal costs incurred by a Hispanic woman increased by >8% in each scenario.

Figure 2. Univariate sensitivity analysis of personal costs incurred per woman, assuming uniform wage rates for all racial/ethnic groups. The category ‘All race/ethnicity’ (asterisk) includes women with unknown race/ethnicity.

In the univariate sensitivity analyses, the discount rate had a large impact on estimated personal costs. For example, in Scenario 1, costs for all races/ethnicities for 10 screens in a lifetime were $94.67 at a 5% discount rate but $138.07 at a 0% discount rate (Table 3). For 25 screens in a lifetime, costs were $222.70 at 5% but $345.18 at 0%. Similar results were observed in Scenario 2 and for each racial/ethnic-specific estimate. Changes in the other set of parameter values (such as cost of parking, round trip travel time, waiting and examination time, and distance traveled) had no discernible effect on baseline results (data not shown).

Table 3. Univariate Sensitivity Analysis of Personal Costs Incurred per Woman by Income and Race/Ethnicity*

Parameter	Cost, $
	All races/ethnicities†	Non-Hispanic white	Non-Hispanic black	Hispanic	Non-Hispanic other
* Variables not reported here had no discernible effect on the results reported at baseline. † All races/ethnicities include women with unknown race/ethnicity.
Scenario 1 (annual income <$10,000)
Discount rate=0%: Screening 10 times in a lifetime	138.07	53.90	21.68	41.86	13.17
Discount rate=0%: Screening 25 times in a lifetime	345.18	134.75	54.19	104.64	32.92
Discount rate=5%: Screening 10 times in a lifetime	94.67	36.98	14.69	28.54	9.08
Discount rate=5%: Screening 25 times in a lifetime	222.70	86.99	34.58	67.17	21.35
Scenario 2 (annual income $10,000 to <$20,000)
Discount rate=0%: Screening 10 times in a lifetime	251.70	100.66	39.42	72.91	23.73
Discount rate=0%: Screening 25 times in a lifetime	629.26	251.64	98.54	182.28	59.33
Discount rate=5%: Screening 10 times in a lifetime	171.16	68.74	26.47	48.97	16.22
Discount rate=5%: Screening 25 times in a lifetime	402.89	161.75	62.36	115.39	38.16

In the multivariate sensitivity analysis, we simultaneously varied parameter values in the opportunity cost variables, which included traveling, waiting, and examination time; lost productivity time; and distance traveled (in miles). In all racial/ethnic groups and in both scenarios, the opportunity cost accounted for >70% of personal costs (Table 4). In racial/ethnic-specific estimates, a Hispanic woman incurred the lowest opportunity cost (72%, Scenario 1; 83%, Scenario 2), whereas a non-Hispanic white woman incurred the highest opportunity cost (89%, Scenario 1; 94%, Scenario 2). The lower and upper bound estimates for all racial/ethnic groups and for all scenarios had a moderate impact on the results (≤30% change from baseline results).

Table 4. Multivariate Sensitivity Analysis on Selected Variables by Race/Ethnicity

Parameter	Cost, $
	All races/ethnicities*	Non-Hispanic white	Non-Hispanic black	Hispanic	Non-Hispanic other
* All races/ethnicities include women with unknown race/ethnicity. † Discounted at an annual rate of 3%. ‡ The percentage of opportunity cost of time included travel, waiting, and examination times, lost productivity time from work or other leisure activities, and distance traveled (measured in miles).
Scenario 1 (annual income <$10,000)
Lower bound: Screening 1 time in a lifetime	10.56	3.86	1.34	2.72	0.74
Lower bound: Screening 10 times in a lifetime†	70.66	25.89	8.62	17.95	4.78
Lower bound: Screening 25 times in a lifetime†	170.84	62.61	20.80	43.37	11.54
Upper bound: Screening 1 time in a lifetime	23.05	9.59	4.21	7.97	2.98
Upper bound: Screening 10 times in a lifetime†	143.38	59.51	27.45	50.60	18.53
Upper bound: Screening 25 times in a lifetime†	345.80	143.53	66.31	122.12	44.69
Percentage of opportunity cost of time‡	83.1%	89.5%	82.6%	72.1%	81.1%
Scenario 2 (annual income $10,000 to <$20,000)
Lower bound: Screening 1 time in a lifetime	19.13	7.25	2.46	4.69	1.35
Lower bound: Screening 10 times in a lifetime†	130.27	49.07	15.99	31.78	8.85
Lower bound: Screening 25 times in a lifetime†	315.14	118.69	38.62	76.88	21.38
Upper bound: Screening 1 time in a lifetime	39.26	16.85	7.02	12.68	4.93
Upper bound: Screening 10 times in a lifetime†	249.33	104.83	46.53	82.29	31.07
Upper bound: Screening 25 times in a lifetime†	601.78	252.85	112.48	198.77	74.97
Percentage of opportunity cost of time‡	90.5%	94.3%	90.2%	83.4%	89.3%

The results of the probabilistic multivariate sensitivity analysis are summarized in Table 5, which illustrates the 5th and 95th percentile intervals for the cost incurred per woman. For the initial (1-time) screening, the estimated costs in all racial/ethnic groups and in both scenarios were quite similar to those reported at baseline, with estimates from the multivariate analysis slightly lower (by 3% in Scenario 2 and by 10% in Scenario 1). Conversely, the discounted personal costs incurred for 25 screens in a lifetime in Scenarios 1 and 2 had mixed results. In Scenario 1 (all racial/ethnic groups), the estimated cost was $8.32 less in Table 5 than at baseline (Table 2); however, in Scenario 2, the cost increased by $13.24 over baseline. In general, the difference in values between the 5th and 95th percentiles was relatively modest.

Table 5. Estimated Monte Carlo Simulation on Personal Costs Incurred per Woman by Race/Ethnicity and by Number of Mammograms Received in a Lifetime

Variable	Cost of initial screening (screening 1 time), $		Discounted lifetime cost of screening 25 times, $*
	Mean cost	5th to 95th percentiles	Mean cost	5th to 95th percentiles
* Estimated cost was discounted at an annual rate of 3%. † All races/ethnicities include women with unknown race/ethnicity.
Scenario 1 (annual income <$10,000)
All races/ethnicities†	15.68	12.88-18	253.28	188.03-326.60
Non-Hispanic white	6.33	5.19-7.29	101.39	75.53-130.28
Non-Hispanic black	2.39	1.95-2.82	39.59	28.88-51.65
Hispanic	4.43	3.61-5.13	73.33	54.24-96.33
Non-Hispanic other	1.50	1.19-1.85	24.07	17.48-31.76
Scenario 2 (annual income $10,000 to <$20,000)
All races/ethnicities†	30.26	4.93-34.67	487.79	366.54-627.06
Non-Hispanic white	12.23	10.03-14.11	195.55	146.19-252.14
Non-Hispanic black	4.61	3.78-5.40	76.12	56.46-100.26
Hispanic	8.53	6.97-9.92	140.90	104.45-182.13
Non-Hispanic other	2.90	2.30-3.53	46.30	33.81-60.98

DISCUSSION

We estimated that a woman with an annual income of <$10,000 will incur a personal cost of $17 for obtaining a mammogram through the NBCCEDP and a discounted lifetime cost as high as $262; for a woman with an annual income from $10,000 to $20,000, these estimates are $31 and $475, respectively. In both income scenarios and in all racial/ethnic groups, we observed that >70% of the cost incurred by a woman was attributed to opportunity cost. This result is consistent with earlier reports for breast and other cancers.20, 22-24, 35

These estimates represent a woman's burden of commitment to this program. Given the modest incomes of women participating in the NBCCEDP, these personal costs could be substantial. From its inception through 2006, the NBCCEDP has accounted for >7.2 million breast and cervical cancer screening and diagnostic services provided to >3 million medically underserved, low-income women; 30,963 breast cancers and 1934 cervical cancers have been diagnosed as well as 45,632 high-grade precursor cervical lesions.10 However, the program screens only 13.2% of women who are eligible for mammography screening in the NBCCEDP.36 The Institute of Medicine Committee on the Early Detection of Breast Cancer has recommended expanding the program to reach at least 70% of the eligible population.37 While policy makers consider expanding the program, they also should develop strategies to offset personal costs incurred by participants.

Previous studies reported that, because of the financial situation of participants in the NBCCEDP, any modest transaction costs or cost sharing through copay or other deductibles may constitute a barrier to participation.38, 39 Other studies have reported that making cancer screening services free may not be sufficient to ensure maximum use.40-42 Our analyses indicate that transaction costs are the smallest component of total personal costs that a woman may incur in mammography screening, but even these may contribute to low rates of mammography use.30, 43, 44 The largest component of personal costs was opportunity costs, which, in this study, were estimated as >70% of personal costs.

Numerous studies have estimated personal costs (both opportunity and transaction costs) incurred in participating in cancer prevention and control services, including screening and treatment.19, 20, 22-25, 35, 45, 46 The current study is unique, because 1) it estimates personal costs incurred by underserved, low-income women ages 50 years to 64 years who participate in a publicly funded cancer screening program; 2) the estimated costs were stratified to ascertain variations in personal costs by racial/ethnic groups; 3) cost estimates were adjusted for recall rate and travel time index; and 4) the study assumed that every woman participating in NBCCEDP has a positive opportunity cost associated with participation. To our knowledge, no previous study included all of these characteristics.

In conducting an economic evaluation of a health intervention, the general recommendation is to use the societal cost, which includes all costs of the intervention regardless of who incurred them or who benefited from the intervention. The results from this study, along with program costs, can be used to calculate the societal cost. For example, in a previously published study, the average cost of screening mammography for breast cancer reportedly was $105.09 (range, $73.95-136.22; in 2004 dollars).8 Assuming 1-time screening, for Scenario 1, the estimated societal cost expressed in 2004 dollars would be $124.23 (range, $85.53-161.50); for Scenario 2, the societal cost would be $139.31 (range, $94.93-179.29). The calculated societal cost can be used by policy makers for program planning and by researchers to estimate societal cost-effectiveness of the NBCCEDP. This example illustrates the usefulness of quantifying the economic costs incurred by the demand-side (patients) in participating in a preventive cancer screening program.

This study has some limitations to consider when interpreting the results. First, we did not account for the cost of the potential side effects associated with screening mammography, which could be mild to severe. For example, if there is a false-positive result, then the woman could experience unnecessary anxiety and distress and possibly receive costly medical procedures to alleviate her emotional distress. However, we believe that the omission of such costs may have minimal effects on personal costs incurred, because the benefits of screening mammography outweigh its potential harms.2 In addition, the survey did not collect information on mode of travel, distance traveled, time traveled, or time spent in mammography clinic. Instead, we obtained data on these variables from the literature in a similar population participating in mammography screening. Consequently, cost estimates reported in this article may not accurately represent the actual costs incurred by a woman participating in the NBCCEDP.

In conclusion, the results presented in this report suggest that capturing and quantifying these opportunity and transaction costs will help ascertain the total cost (ie, societal cost) of providing breast cancer screening to a medically underserved, low-income woman participating in a publicly funded program. In conjunction with other data, personal costs can be used to determine the societal cost-effectiveness of mammography screening in the NBCCEDP.