B C Papare.pdf

Cost Analysis of the National Breast and Cervical Cancer Early Detection Program Selected States, 2003 to 2004

Donatus U. Ekwueme, PhD1

James G. Gardner, MSPH1

Sujha Subramanian, PhD2

Florence K. Tangka, PhD1

Bela Bapat, MA2

Lisa C. Richardson, MD, MPH1

1 Division of Cancer Prevention and Control, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Con-trol, Atlanta, Georgia.

2 RTI International, Waltham, Massachusetts.

BACKGROUND. The National Breast and Cervical Cancer Early Detection Program

(NBCCEDP) was established by the U.S. Congress in 1990. In recent years, there has been an emphasis on ascertaining the NBCCEDP’s costs of delivering screen-ing and diagnostic services to medically underserved, low-income women. The objective of this report was to address 3 economic questions: What is the cost per woman served in the program, what is the cost per woman served by pro-gram component, and what is the cost per cancer detected through the program?

METHODS. The authors developed a questionnaire to systematically collect activ-ity-based costs on screening for breast and cervical cancer from 9 participating programs. The questionnaire was developed based on well established methods of collecting cost data for program evaluation. Data were collected from July

2003 through June 2004.

RESULTS. With in-kind contributions, the cost of screening services to women in

9 programs was estimated at $555 per woman served. Without in-kind contribu-tions, this cost was $519. Among the program components, screening and coali-tions/partnerships accounted for the highest and lowest cost per woman served, respectively. The median cost of screening a woman for breast cancer was $94, and the cost per breast cancer detected was $10,566. For cervical cancer, these costs were $56 and $13,340, respectively.

CONCLUSIONS. Costs per woman served, screened, and cancers detected are needed for programs to accurately determine the resources required to reach and screen eligible women. With limited program resources, these cost estimates can provide useful information to assist programs in planning and implementing cost-effective activities that could maximize the allocation of program resources.

Cancer 2008;112:626–35. Published 2007 by the American Cancer Society*

KEYWORDS: cost analysis, National Breast and Cervical Cancer Early Detection Program, breast cancer, cervical cancer, screening.

I n the U.S., both breast cancer and cervical cancer impose consid-erable health and economic burdens on society and on indivi-duals who are diagnosed with these diseases. According to data from the U.S. Cancer Statistics in 2003, an estimated 181,646 Ameri-can women were diagnosed with breast cancer, and 41,619 women died from it; for cervical cancer, the estimates were 11,820 women diagnosed and 3919 deaths.1 Many of these deaths are preventable with screening using mammography for breast cancer and the Papa-nicolaou (Pap) test for cervical cancer.

Despite their acceptability and widespread availability, these screening methods may not be affordable to low-income women.

Because uninsured, low-income women may not be able to afford

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.

Address for reprints: Donatus U. Ekwueme, PhD, Division of Cancer Prevention and Control, National Center for Disease Prevention and Health Promotion, Centers for Disease Control, 4770 Buford Highway, MS K-55, Atlanta, GA 30341; Fax: (770) 488-4639; E-mail: dce3@cdc.

gov

We sincerely thank the selected programs that participated in this phase of the project. In addi-tion, we thank Mary C. White, Donald K. Black-man, and Susan True for their support of this project. We also thank the National Breast and Cervical Cancer Early Detection Program editorial office for their review of an earlier draft of this article and Amber Moore and Caren Kramer for their assistance in developing the site-visit instrument.

Received May 1, 2007; revision received August 17, 2007; accepted August 21, 2007.

*This article is a US Government work and, as such, is in the public of main in the United States of America.

Published 2007 by the American Cancer Society* DOI 10.1002/cncr.23207 Published online 21 December 2007 in Wiley InterScience (www.interscience.wiley.com).

screening, they may lose the benefits of early detec-tion and treatment that, in the case of cervical can-cer, also may include prevention.

In response to the lack of affordability of screen-ing and follow-up and the disproportionate burden of breast and cervical cancer borne by uninsured, low-income women, in 1990, the U.S. Congress passed the Breast and Cervical Cancer Mortality Pre-vention Act, (Public Law 101–354, Title XV, Public

Health Service Act).2 This law established the

National Breast and Cervical Cancer Early Detection

Program (NBCCEDP), which is administered by the

U.S. Centers for Disease Control and Prevention

(CDC). The NBCCEDP is organized and implemented through cooperative agreements with health depart-ments in the 50 states, the District of Columbia, 4

U.S. territories, and 13 American Indian/Alaska

Native tribal organizations. The NBCCEDP provides support and assistance for screening medically underserved, low-income women ages 18 years to

64 years for cervical cancer and ages 40 years to

64 years for breast cancer. The NBCCEDP also pro-vides diagnostic testing and follow-up services for women whose screening tests are abnormal. From its inception through 2004, the NBCCEDP has provided

6.5 million screening examinations to 2.7 million women and has diagnosed 26,000 breast cancers and

1700 cervical cancers.3

The NBCCEDP, as the largest organized cancer screening program in the U.S., seeks to maximize the benefits of breast and cervical cancer early detection for the nation’s uninsured, low-income women not only by seeking to reach and screen hard-to-reach eligible women. The NBCCEDP uses a comprehensive approach to ensure that all women benefit from the program through educational activities, coalitions/ partnerships, and quality assurance and improve-ment in cancer screening and diagnostic services.

Furthermore, the program strives to reach and screen as many eligible women as possible. However, given the current climate of increasing healthcare costs and limited resources, there is a continuing emphasis on economic costs of delivering screening and diag-nostic services to eligible women.

In 2003, we undertook a multiyear research pro-ject to systematically address economic issues related to the NBCCEDP. In this report, we present the results of a cost analysis of 9 selected programs. We addressed 3 economic-related questions: 1) What is the cost per woman screened and the cost per woman served? 2) What is the cost per woman served by program component? and 3) What is the cost per cancer detected through the program? Cost per woman screened is the cost of screening a woman who received NBCCEDP-funded screening procedures, which includes either mammography and a clinical breast examination (CBE) or the Pap test. Cost per woman served is the cost of screening a woman in the program plus the cost of diagnostic follow-up of abnormal results for a woman who was screened outside the program but who was referred to the program at the diagnostic stage.

Given these research questions, we estimated 1) the cost per woman served with and without in-kind

(nonfederal) contributions; 2) the cost per woman served per each program component; 3) the cost per woman served for clinical services; 4) the cost per woman screened for breast and cervical cancer, respectively; and 5) the cost per breast and cervical cancer detected. Cost data on screening and diagnos-tic services will be essential in assisting policy and decision makers in allocating program resources equitably across the individual programs.

MATERIALS AND METHODS

Program Selection and Data Collection Procedures To address the questions stated above, we used both qualitative and quantitative data4 to select 9 pro-grams. The details of data collection methods that were used for this cost analysis have been described previously (unpublished results). Briefly, we visited 4 programs. These programs were selected by their structure for delivering screening services (eg, centra-lized, decentralized, or mixed), by the volume of women screened, and by location. We used the same questionnaire for all 4 sites. We conducted a group interview with key staff members, including the pro-gram director and financial manager.

Using qualitative information obtained from the site visits, we developed a questionnaire to systema-tically collect activity-based costs. The questionnaire was developed based on well established methods of collecting cost data for program evaluation.5–8 Cost data were collected on personnel, consultants, con-tracts, materials, supplies, administration/overhead

(eg, telephone, rent, website), and other consum-ables. We collected cost data on the following activ-ities or program components: program management, screening and diagnostic procedures, case manage-ment, tracking and follow-up, data management, public education and outreach, professional educa-tion, coalitions/partnerships, quality assurance and improvement, and surveillance and evaluation.

The questionnaire was administered to 9 pro-grams, which were identified by a combination of random selection and purposive sampling.9 The pro-gram selection process is outlined in Figure 1. We

Cost Analysis of the NBCCEDP/Ekwueme et al. 627 began with 51 programs that represented the 50 states and the District of Columbia; then, we excluded the 4 programs we had visited. The remain-ing programs (n 5 47 programs) were stratified into

3 categories: centralized programs, decentralized pro-grams, or mixed programs.

Centralized programs offer clinical services in satellite offices, but all other program activities are performed centrally, such as tracking and case-man-agement services. Decentralized programs contract with local and regional health departments, private hospitals, or other healthcare facilities to fully man-age and provide screening and diagnostic services and other program activities. Programs with a mixed service delivery structure have both centralized and decentralized features.

We randomly selected 6 programs within these 3 strata (n 5 18 programs). From each of the 6 pro-grams, we used purposive sampling to select 3 pro-grams in each stratum (n 5 9 programs). To ensure diversity across all of the programs, we used the fol-lowing criteria to select 9 programs for data collec-tion: lessons learned from the site visits, number of women served, number of women screened, distribu-tion of screening (ie, the proportion of breast and cervical cancer screenings), type of funding source, financial data from the program, geographic location, and size of the area served. Data were collected on these 9 programs from July 2003 through June 2004.

Quality Control To ensure the quality of our data, first, we conducted a thorough pilot study; then, we conducted several rounds of follow-up telephone calls with the 9 pro-grams to ensure that the data we received were accu-rate and complete. Overall, we observed that the information reported by the programs was accurate;

when there were inconsistencies, we consulted with the program.

Measures of Effectiveness The effectiveness measures were obtained from the

NBCCEDP surveillance database. Details of this data-base have been described elsewhere.10,11 To measure effectiveness, we used 3 variables: 1) the number of women screened for breast or cervical cancer, 2) the number of women served, and 3) the number of cancers detected. The number of women screened equaled those who received NBCCEDP-funded screen-ing procedures, which included either mammography and a CBE or the Pap test. The number of women served included those screened in the program plus those screened outside the program who were referred to the program for diagnostic follow-up of abnormal screening results. The number of breast cancers detected included both carcinoma in situ (CIS) and invasive cancer. The number of cervical cancers included both cervical intraepithelial neoplasia (CIN2/

CIN3/CIS) and invasive cancer.

Data Analysis and Calculation of Intermediate Outcome Measures In this study, we estimated only intermediate out-come measures, which reflect relatively near-term effects of the program on the population being screened or served.8,12 These intermediate measures included the cost per woman screened or served and the cost per cancer detected. One of the advantages of this approach is that it helps to guide resource allocation among the programs. The data collected were entered into a series of Microsoft Excel (Micro-soft Corporation, Redmond, Wash) spreadsheets and were analyzed using that platform. The analysis was conducted from a program perspective, suggesting that we included only the direct costs of providing preventive cancer screening services. Because the study was conducted during a 1-year period, costs and outcomes were not discounted.

We calculated several intermediate outcome measures from the 9 programs. First, we calculated the cost per woman screened for breast cancer (or cervical cancer) by using the reported cost of screen-ing plus the cost of office visits divided by the num-ber of women screened. Second, we calculated

FIGURE 1. The program-selection process (*SHAPE/* MERGEFORMAT). The asterisk indicates that 17 programs administered by tribes and territories were excluded, generally because they were small. That is, each of these programs served or screened <4% of the number of women served in the program per year.

628 CANCER February 1, 2008 / Volume 112 / Number 3 average costs per woman served with and without in-kind contributions by using the average total pro-gram cost with and without these contributions divided by the average total number of women served. We defined in-kind contributions as those contributions that strictly represented the value of resources donated to each program through other sources (ie, opportunity cost). Third, to ascertain the level of compliance with legislative requirements that

�60% of NBCCEDP funds must be used for direct clinical services,13 we calculated the cost of clinical services per woman served by dividing the reported average total cost of clinical services by the number of women served. We defined the cost of clinical ser-vices to include the cost of screening and diagnostic follow-up, referral for treatment and the cost of case management.13 Fourth, we calculated costs by pro-gram component per woman served; we also calcu-lated the percent distribution of the total program cost represented by each component. Fifth, we calcu-lated the average cost per breast cancer detected and, separately, calculated these costs by CIS and invasive cancer only. Finally, we calculated average cost per cervical cancer detected and, separately, the average cost per CIN2/CIN3/CIS only and per case of invasive cervical cancer.

Sensitivity Analysis Because the selected programs are heterogeneous, the estimated costs may not be representative of the costs incurred by programs around the U.S. in deli-vering screening and diagnostic follow-up services to a woman. To assess how the heterogeneity of these programs affected the results of the cost analysis, we performed sensitivity analyses using the Monte Carlo simulation method.14–16 We constructed probability distributions for each of the intermediate outcome measures discussed above. For example, we fitted normal distributions to the cost per woman served with and without in-kind contributions, triangular distributions to the cost per woman screened for breast and cervical cancer, and log-normal distribu-tions to the respective cost per breast and cervical cancer detected. Using the fitted probability distribu-tions, we performed simulations using @Risk soft-ware (version 4.5; Palisade Corporation, Newfield, NY), an Excel add-in that employs Monte Carlo sam-pling methods. In total, 3500 simulation runs were performed. On each simulation run, a value for each intermediate outcome measure (eg, cost per woman screened for breast cancer) was drawn from its asso-ciated distribution, and this value was stored as an output. Using the output of the 3500 simulation runs as data, descriptive statistics were calculated, includ-ing the mean, standard deviation, median, and per-centile intervals for all outcome measures. We report the results of the simulation as medians with 25th and 75th percentile intervals.

RESULTS

The median number of women served was 6287

(25th–75th percentile, 2926–10,515 women), and the median total program cost without in-kind contribu-tions was $3.3 million (25th–75th percentile, $1.4–

4.5 million) across the 9 programs (Table 1). With in-kind contributions, the median total program cost was $4.9 million (25th–75th percentile, $1.9–5.3 mil-lion) or 33% higher than the cost without these con-tributions. The median cost per woman served without in-kind contributions was $519.16 (25th–75th percentile, $352.16–532.34); and, with in-kind contri-butions, it was $554.57 (25th–75th percentile, $503.05–661.11). The median cost of clinical services per woman served was $247.59 (25th–75th percentile, $208.42–289.78). The median number of women screened per program was 5880 (25th–75th percen-tile, 2818–9877 women) for breast cancer and 3724

(25th–75th percentile, 2278–7159 women) for cervical cancer (Table 2). The cost per woman screened was

$94.43 (25th–75th percentile, $72.90–120.31) for breast cancer and $55.64 (25th–75th percentile, $43.60–62.01) for cervical cancer.

The median number of breast cancers detected

(CIS plus invasive carcinoma) was 75 (25th–75th per-centile, 27–123 breast cancers); for cervical cancer

(including CIN2/CIN3/CIS and invasive carcinoma), this number was 26 (25th–75th percentile, 12–51 cer-vical cancers) (Table 3). The median costs per cancer detected were $10,566 (25th–75th percentile, $8417–

12,714) for breast cancer and $13,340 (25th–75th per-centile, $1596–25,084) for cervical cancer.

The results of the multivariate sensitivity analy-ses are summarized in Figures 2 and 3. The median costs per woman served by the NBCCEDP compo-nent are shown in Figure 2. Screening accounted for the highest cost ($220.65; 25th–75th percentile, $172.53–268.75) (Fig. 2), which represented 48.2% of the cost per woman served. Public education/out-reach was the second highest with an estimated cost of $62.52 (25th–75th percentile, $23.34–101.02). Coa-litions/partnerships had the lowest cost of $1.76

(25th–75th percentile, $0.44–3.07). The estimated me-dian costs per woman served or screened, which are shown in Figure 3, varied slightly from the baseline values reported in Tables 1 and 2. Compared with the baseline, this variation ranged from a 0.3% decrease for the cost per woman served for clinical

Cost Analysis of the NBCCEDP/Ekwueme et al. 629 services to a 9.9% decrease for the cost per woman served without in-kind contributions. Conversely, the estimated median cost per woman served with in-kind contributions was 9.5% higher in the sensi-tivity analysis in contrast to the estimated value reported at baseline. The estimated costs per woman screened for breast and cervical cancers were 6.9% higher (cervical cancer) and 10.1% higher

(breast cancer) in the sensitivity analysis compared with the value reported at baseline. These variations in the estimated cost per woman served (with and without in-kind contributions), screened, and clini-cal services reflect the heterogeneity that exists among these programs.

DISCUSSION

Using 9 programs in the NBCCEDP, we calculated a median cost per woman served of $555 when the value of in-kind contributions was included; without in-kind contributions, the median cost per woman served was $519. The cost of providing direct clinical

TABLE 1

The Number of Women Served and the Mean and Median Total Costs for Women Served From Selected Programs, 2003–2004

Program

Total no.

of women served

Without in-kind contributions, $* With in-kind contributions, $y

Total program cost

Total cost for clinical services

Cost per woman served

Cost per woman served for clinical services{

Total cost from in-kind contributions

In-kind contributions per woman served

Cost per woman served

Total program cost

1 1285 974,810 449,012 758.61 349.43 433,731 337.53 1096.14 1,408,541

2 6287 3,267,851 1,952,209 519.78 310.52 1,632,993 259.74 779.52 4,900,844

3 17,234 4,455,695 2,506,505 258.54 145.44 415,790 24.13 282.67 4,871,485

4 3859 1,358,981 629,448 352.16 163.11 525,000 136.05 488.20 1,883,981

5 9915 3,862,332 2,066,450 389.54 208.42 1,419,100 143.13 532.67 5,281,432

6 19,374 6,312,989 4,628,839 325.85 238.92 3,433,153 177.20 503.05 9,746,142

7 2926 1,519,066 847,910 519.16 289.78 415,330 141.94 661.11 1,934,396

8 2152 1,285,450 578,591 597.33 268.86 49,543 23.02 620.35 1,334,993

9 10,151 5,403,829 2,513,276 532.34 247.59 225,600 22.22 554.57 5,629,429

Mean 8131 3,160,111 1,796,916 472.59 246.90 950,027 140.55 613.14 4,110,138

SD 6594 1,980,693 1,355,452 156.76 67.87 1,072,763 109.40 226.43 2,769,008

Median 6287 3,267,851 1,952,209 519.16 247.59 433,731 141.94 554.57 4,871,485

25th–75th percentile 2926–

10,151

1,358,981–

4,455,695

629,448–

2,506,505

352.16–

532.34

208.42–

289.78

415,330–

1,419,100

24.13–

177.20

503.05–

661.11

1,883,981–

5,281,432

SD indicates standard deviation.

* Costs were obtained by using National Breast and Cervical Cancer Early Detection Program (NBCCEDP) funds from the Centers for Disease Control and Prevention (CDC), which may include state government matching funds.

y Costs were obtained by using NBCCEDP funds from the CDC plus additional funds from state government and other sources. Other sources of funds were provided in the form of in-kind contributions.

{ Clinical services included screening, diagnostic follow-up, and case management.

TABLE 2

Mean and Median Total Number of Women Screened, Follow-up Visits for Abnormal Results, and Mean and Median Total Cost of Screening and Diagnostic Follow-up: Selected Programs, 2003–2004

Variable

Breast screening Cervical screening

Mean 6 SD Median (25th–75th percentile) Mean 6 SD Median (25th–75th percentile)

No. of women screened 7233 � 5448 5880 (2818–9877) 4745 � 3282 3724 (2278–7159)

No. of follow-up visits for abnormal results 1363 � 1132 1240 (302–2125) 416 � 549 178 (91–418)

Cost per office visit, $* 35.47 � 11.85 39.91 (26.53–41.44) 35.47 � 11.85 39.91 (26.53–41.44)

Total costs of office visits, $ 265,857 � 231,957 243,692 (60,884–394,765) 172,513 � 140,088 154,339 (44,161–264,269)

Total cost of screening, $ 445,434 � 324,034 487,588 (158,972–689,426) 74,129 � 35,150 79,322 (64,766–84,804)

Total cost of screening and office visit, $ 711,291 � 543,403 859,034 (216,022–933,118) 246,642 � 169,572 230,931 (99,330–347,897)

Cost per woman screened, $ 105.09 � 47.65 94.43 (72.90–120.31) 59.75 � 30.05 55.64 (43.60–62.01)

SD indicates standard deviation.

* The mean number of office visits per woman was 1.15 for both breast and cervical cancers; the mean number of office visits per woman was 1.61 for breast cancer and 2.65 for cervical cancer.

630 CANCER February 1, 2008 / Volume 112 / Number 3 services per woman served was $248. The cost of screening a woman for breast cancer using mam-mography was $94, and the cost per cancer detected was $10,566. For cervical cancer, these costs were

$56 and $13,340, respectively. In an era of limited resources, these cost estimates can provide useful in-formation for program planning and budget alloca-tion. Lack of cost information is often a limiting factor in reaching and screening women who are eligible for the NBCCEDP.

It often is difficult to compare cost studies, because they use different types of costing methods, depending on the objective and the viewpoint of the study.8–12 The current study was conducted from the program’s viewpoint, which implies that only direct costs incurred by the programs in providing cancer screening services were included in the study. We reviewed the literature to extract data from studies that also used the same viewpoint to generate aver-age costs for breast and cervical cancer screening.

Overall, we observed that our estimated average costs for breast and cervical cancer screening did not differ substantially from other estimates reported in the literature. For example, breast screening studies have reported that average costs of screening ranged from $64 to $172 (in 2004 dollars).17–22 For cervical screening, reported estimates (in 2004 dollars) ran-ged from $32 to $114.18,23–28

Few studies have been reported on the cost per cancer detected for public health programs.29–31

These studies estimated that the cost per breast can-cer detected ranged from $12,558 to $18,238 when adjusted to 2004 dollars. In general, these estimates are comparable to the cost estimates reported in the current article, because those studies were conducted in different settings, timeframes, and populations.

For example, the studies by Hurley et al. and Gerard et al. were conducted in Australian mammography programs30,31 with a different healthcare delivery sys-tem and perhaps different eligibility criteria to qua-lify for their programs. For cervical cancer, we could not find a comparable cost per cancer detected for a similar public health program. This reveals that, although substantial literature has been published on the cost-effectiveness of breast and cervical cancer, very few studies have been conducted in public health programs for targeted populations. Another reason could be that these studies are not reporting the cost of cancer as an outcome measure.

Because the NBCCEDP was established by the

U.S. Congress to provide free or low-cost, compre-hensive breast and cervical cancer screening and diagnostic follow-up of abnormal results to medically underserved, low-income women, the NBCCEDP is grounded in public law with regulations that must be complied with by participating states, district, terri-tories, and tribal organizations. One such legislative regulation states that no less than 60% of program funds must be used for direct clinical services. These clinical services include screening and diagnostic fol-low-up, referral for treatment, and case manage-ment.13 In the current study, we observed that 58.2% of the program resources (excluding in-kind contri-butions) are allocated to clinical services, which is very close to the legislative requirement.

The remaining 40% of the program’s resources are used to support other activities that are essential to the success of clinical services. These activities include data management, public education and out-reach, professional education, coalitions/partner-ships, quality assurance and improvement, and surveillance and evaluation. The support of these activities by policy makers is intended to address fac-tors other than direct financial barriers that may have an impact on the ability of uninsured low-income women to receive cancer screening services. Behav-ioral intervention studies have identified these factors include both patient and provider behavior.32–38 For instance, studies have identified that the degree to which individual patients seek cancer screening services is influenced by usual source of care, public information and education, outreach, physician rec-ommendation, and implementation of interventions to increase general knowledge of prevention among health providers.39–45 The NBCCEDP allocates a proportion of its resources to public education and outreach campaigns to increase program awareness among eligible women and to encourage them to use

TABLE 3

The Estimated Median Number of Cancers Detected and Cost per Cancer Detected: Selected Programs, 2003–2004

Variable

Median (25th–75th percentile)

Breast screening Cervical screening

Average no. of cancers detected

All 75 (27–123) 26 (12–51)

CIS only 14 (7–21) —

CIN2/CIN3/CIS only — 23 (10–47)

Invasive cancer only 61 (16–105) 2 (1–4)

Average cost per cancer detected ($)

All 10,566 (8417–12,714) 13,340 (1596–25,084)

CIS only 46,201 (31,283–66,876) —

CIN2/CIN3/CIS only — 14,304 (1937–26,675)

Invasive cancer only 14,259 (10,662–18,958) 63,768 (35,081–92,447)

CIS indicates carcinoma in situ; CIN, cervical intraepithelial neoplasia.

Cost Analysis of the NBCCEDP/Ekwueme et al. 631 program services. In the current study, we observed that selected programs spent 12.1% of their total cost

(or $62.52 per woman served) on this activity and that they spent another 5.7% (or $26.47 per woman served) on professional education. Quantifying the amount of the program resources allocated to differ-ent program activities is vital to the success of the

NBCCEDP, and it provides accountability on how resources are used to influence screening behavior.

There are some caveats in our analysis. First, although we used a combination of random selection and purposive sampling9 to select 9 programs, the average (or median) costs reported here may not be generalizable beyond the 9 programs we selected for this study. This is because of substantial variations that exist among the programs nationally. However, we performed a detailed literature search and con-ducted appropriate sensitivity analyses to assess the consistency of our findings. We plan to conduct more detailed, empiric investigation to explore the sources of variation between average cost per woman served or cost per cancer detected on the number of women served or the number of cancers detected.

Describing the true underlying factors may help pub-

FIGURE 2. Sensitivity analysis: the estimated distribution of median cost per woman served by program component. Note that the distribution of cost per program component excluded in-kind contributions. That is, the estimated cost per program component was obtained by using National Breast and Cervical

Cancer Early Detection Program funds from the Centers for Disease Control and Prevention, which may have included state government matching funds. Ranges represent 25th and 75th percentile intervals. The total median cost per woman served was $473.67 (range, $271.70–674.40).

FIGURE 3. Sensitivity analysis: the estimated median cost per woman served or screened from the selected programs. Note that ranges represent

25th and 75th percentile intervals. The dagger indicates that clinical services included screening, diagnostic follow-up, and patient management; the dou-ble dagger indicates that estimated costs were obtained by using the

National Breast and Cervical Cancer Early Detection Program (NBCCEDP) funds from the Centers for Disease Control and Prevention (CDC), which may have included state government matching funds; and the section sign indi-cates that estimated costs were obtained by using NBCCEDP funds from the

CDC plus additional funds from state government and other sources.

632 CANCER February 1, 2008 / Volume 112 / Number 3 lic health officials understand why a given program may not be able to allocate its limited resources more efficiently to screen as many eligible women as possible. Second, we analyzed only 1 year of data.

We know that program funding and other sources of resources vary from year to year, depending on the activities planned in that year. This year-to-year vari-ation may have affected our calculated costs of deli-vering screening services. An ideal study would be to analyze economic cost data from multiple years, which is part of Phase 2 of this project, which cur-rently is being conducted. A third concern is that selected programs (within the 9 we analyzed) could not separate the cost of office visits for breast screen-ing from those for cervical screening; thus, we assumed that these costs are identical. This assump-tion may have caused us to underestimate or overes-timate the costs of screening for these cancers;

however, our review of the literature indicated that the impact of this limitation on average costs may be minimal. For instance, Goldie et al. reported the cost of an office visit for cervical screening ranged from

$11 to $50,28 a range that includes our estimate of

$35.47. Fourth, many of the programs could not dis-tinguish the cost of CBE from that of mammography.

The programs, however, informed us that the cost of

CBE is very small and, in most women, is included in the costs for office visits. Two studies have reported that the cost of CBE ranges from $8.16 to

$24.74 (adjusted to 2004 dollars).17,18 Because we were unable to separate the cost of CBE from mam-mography, the average cost of breast screening we report may be a combination of mammography and

CBE. Finally, we reported the cost per cancer detected as an intermediate outcome measure, but a program such as the NBCCEDP, which is designed to detect early-stage cancer, would save more life-years than programs that detect cancers at later stages.

Therefore, reporting intermediate outcomes for this program may provide practical information on the direct cost of detecting a case of cancer in medically underserved, low-income women. However, it may not provide complete information on the benefits of the NBCCEDP. Hence, readers should note that this study is not a cost-effectiveness analysis, but a cost analysis.

These caveats notwithstanding, we believe that the results presented in this report provide a good picture of the overall allocation of NBCCEDP resources in the selected programs. We also believe that our findings may help to inform decision mak-ing in the NBCCEDP once all programs have been surveyed. To our knowledge, our study provides the first systematic cost analysis in the selected programs and is unique, in that it attempts to determine the real costs of providing comprehensive screening and diagnostic follow-up services to low-income, unin-sured women.

Future Directions This study addressed short-term economic costs in 9 programs. Currently, the CDC is conducting Phase 2 of this project in all 68 programs operating in the

U.S. Phase 2 addresses long-term economic issues related to the evaluation of the NBCCEDP. These issues include developing cost-assessment tools for the better collection of data on economic cost and developing cost-effectiveness models to estimate the full costs and benefits of the NBCCEDP. Phase 2 also includes developing econometric methods that would relate outcomes and inputs in allocating pro-gram resources based on the effectiveness and effi-ciency of each program; conducting threshold and scenario analyses to determine the amount of resources required to screen 25%, 50%, or 75% of the eligible women nationally; estimating the cost of treating a woman diagnosed with cancer in the

NBCCEDP; describing the methods used by pro-grams to deliver screening and diagnostic follow-up services; documenting the extent to which these pro-grams use a common set of service delivery struc-tures; and determining how different types of service delivery structures affect the number of women screened or served. It is our hope that addressing these economic issues will provide invaluable infor-mation that program directors and managers can use to make informed decisions on how to allocate

NBCCEDP resources more efficiently. It also will pro-vide a framework for establishing the minimum number of NBCCEDP-eligible women who can be screened given the resources available.

REFERENCES

1. US Cancer Statistics Working Group. United States Cancer

Statistics: 2003 Incidence and Mortality. Atlanta, Ga: U.S.

Department of Health and Human Services, Centers for

Disease Control and Prevention and National Cancer Insti-tute; 2006. Available from URL: www.cdc.gov/cancer/npcr/ npcrpdfs/US_Cancer_Statistics_2003_Incidence_and_Mortality.

pdf Accessed April 13, 2007.

2. Ryerson AB, Bernard VB, Major AC. National Breast and

Cervical Cancer Early Detection Program: 1991–2002

National Report. Atlanta, Ga: US Department of Health and

Human Services, Centers for Disease Control and Preven-tion; 2005. Available from URL: http://www.cdc.gov/cancer/ nbccedp/Reports/NationalReports/index.htm Accessed on

January 20, 2007.

3. Centers for Disease Control and Prevention. The National

Breast and Cervical Cancer Early Detection Program. Avail-able from URL: http://www.cdc.gov/cancer/nbccedp/about.

htm Accessed January 20, 2007.

Cost Analysis of the NBCCEDP/Ekwueme et al. 633

4. Greene JC, Caracelli VJ, Graham WF. Toward a conceptual framework for mixed method evaluation designs, Educ

Eval Policy Anal. 1989;11:255–274.

5. French MT, Roebuck MC, McLellan AT. Cost estimation when time and resources are limited: the Brief DATCAP.

J Subst Abuse Treat. 2004;27:187–193.

6. French MT, Dunlap LJ, Zarkin GA, McGeary KA, McLellan

AT. A structured instrument for estimating the economic cost of drug abuse treatment.The Drug Abuse Treatment

Cost Analysis Program (DATCAP). J Subst Abuse Treat.

1997;14:445–455.

7. Salome HJ, French MT, Miller M, McLellan AT. Estimating the client costs of addiction treatment: first findings from the client drug abuse treatment cost analysis program

(Client DATCAP). Drug Alcohol Depend. 2003;71:195–

206.

8. Haddix AC, Corso PA, Gorsky DR. Costs. In: Haddix AC, Teutsch SM, Corso PA, eds. Prevention Effectiveness: A

Guide to Decision Analysis and Economic Evaluation, 2nd

ed. New York, NY: Oxford University Press; 2003:53–76.

9. Kerlinger FN. Foundations of Behavioral Research, 3rd ed.

New York, NY: Holt, Rinehart, and Winston; 1986.

10. Eheman CR, Benard VB, Blackman D, et al. Breast cancer screening among low-income or uninsured women: results from the National Breast and Cervical Cancer Early Detec-tion Program, July 1995 to March 2002 (United States).

Cancer Causes Control. 2006;17:29–38.

11. Benard VB, Eheman CR, Lawson HW, et al. Cervical screen-ing in the National Breast and Cervical Cancer Early Detec-tion Program, 1995–2001. Obstet Gynecol. 2004;103:564–

571.

12. Gold MR, Siegel JE, Russell L.B, Weinstein MC. Cost-Effec-tiveness in Health and Medicine. New York, NY: Oxford Uni-versity Press; 1996.

13. Centers for Disease Control and Prevention, National

Breast and Cervical Cancer Early Detection Program. Poli-cies and Procedures Manual. Atlanta, Ga: Centers for Dis-ease Control and Prevention; 2005.

14. Critchfield GC, Willard KE. Probabilistic analysis of deci-sion trees using Monte Carlo simulation. Med Decis Mak-ing. 1986;6:85–92.

15. Doubilet P, Begg CB, Weinstein MC, Braun P, McNeil BJ.

Probabilistic sensitivity analysis using Monte Carlo simula-tion. A practical approach. Med Decis Making. 1985;5:157–

177.

16. Dittus RS, Roberts SD, Wilson JR. Quantifying uncertainty in medical decisions. J Am Coll Cardiol. 1989;14(3 suppl

A):23A–28A.

17. Benoit RM, Naslund MJ. An economic rationale for pros-tate cancer screening. Urology. 1994;44:795–803.

18. O’Malley AS, Lawrence W, Liang W, et al. Feasibility of mo-bile cancer screening and prevention. J Health Care Poor

Underserved. 2002;13:298–319.

19. Rubin E, Mennemeyer ST, Desmond RA, et al. Reducing the cost of diagnosis of breast carcinoma: impact of ultra-sound and imaging-guided biopsies on a clinical breast practice. Cancer. 2001;91:324–332.

20. Lindfors KK, Rosenquist CJ. The cost-effectiveness of mammographic screening strategies. JAMA. 1995;274:881–

884.

21. Salzmann P, Kerlikowske K, Phillips K. Cost-effectiveness of extending screening mammography guidelines to include women 40 to 49 years of age. Ann Intern Med. 1997;127:

955–965.

22. Mandelblatt JS, Schechter CB, Yabroff KR, et al. Toward optimal screening strategies for older women. Costs, benefits, and harms of breast cancer screening by age, biology, and health status. J Gen Intern Med. 2005;20:487–

496.

23. Helms LJ, Melnikow J. Determining costs of health care services for cost-effectiveness analyses: the case of cervical cancer prevention and treatment. Med Care. 1999;37:652–

661.

24. Maxwell GL, Carlson JW, Ochoa M, Krivak T, Rose GS, Myers ER. Costs and effectiveness of alternative strategies for cervical cancer screening in military beneficiaries.

Obstet Gynecol. 2002;100:740–748.

25. Mandelblatt JS, Lawrence WF, Womack SM, et al. Benefits and costs of using HPV testing to screen for cervical can-cer. JAMA. 2002;287:2372–2381.

26. Kulasingam SL, Myers ER. Potential health and economic impact of adding a human papillomavirus vaccine to screening programs. JAMA. 2003;290:781–789.

27. Fetters MD, Lieberman RW, Abrahamse PH, Sanghvi RV, Sonnad SS. Cost-effectiveness of Pap smear screening for vaginal cancer after total hysterectomy for benign disease.

J Low Genit Tract Dis. 2003;7:194–202.

28. Goldie SJ, Kim JJ, Wright TC. Cost-effectiveness of human papillomavirus DNA testing for cervical cancer screening in women aged 30 years or more. Obstet Gynecol.

2004;103:619–631.

29. Schweitzer ME, French MT, Ullmann SG, McCoy CB. Cost-effectiveness of detecting breast cancer in lower socioeco-nomic status African American and Hispanic women through mobile mammography services. Med Care Res Rev.

1998;55:99–115.

30. Hurley SF, Livingston PM, Thane N, Quang L. Mammo-graphic screening: measurement of the cost in a popula-tion based programme in Victoria, Australia. J Epidemiol

Commun Health. 1994;48:391–399.

31. Gerard K, Salkeld G, Hall J. Counting the costs of mam-mography screening: first year results from the Sydney study. Med J Aust. 1990;152:466, 469–471.

32. Bonfill X, Marzo M, Pladevall M, Marti J, Emparanza JI.

Strategies for increasing women participation in commu-nity breast cancer screening. Cochrane Database Syst Rev.

2001;1:CD002943.

33. Legler J, Meissner HI, Coyne C, Breen N, Chollette V, Rimer

BK. The effectiveness of interventions to promote mam-mography among women with historically lower rates of screening. Cancer Epidemiol Biomarkers Prev. 2002;11:59–

71.

34. Marcus AC, Crane LA. A review of cervical cancer screening intervention research: implications for public health pro-grams and future research. Prev Med. 1998;27:13–31.

35. Meissner HI, Breen N, Coyne C, Legler JM, Green D, Edwards BK. Breast and cervical cancer screening interven-tions: an assessment of the literature. Cancer Epidemiol

Biomarkers Prev. 1998;7:951–961.

36. Snell JL, Buck EL. Increasing cancer screening: a meta-analysis. Prev Med. 1996;25:702–707.

37. Stone EG, Morton SC, Hulscher ME, et al. Interventions that increase use of adult immunization and cancer screening services: a meta-analysis. Ann Intern Med. 2002;

136:641–651.

38. Yabroff KR, Mandelblatt JS. Interventions targeted toward patients to increase mammography use. Cancer Epidemiol

Biomarkers Prev. 1999;8:749–757.

634 CANCER February 1, 2008 / Volume 112 / Number 3

39. Breen N, Wagener DK, Brown ML, Davis WW, Ballard-Barbash

R. Progress in cancer screening over a decade: results of can-cer screening from the 1987, 1992, and 1998 National Health

Interview Surveys. J Natl Cancer Inst. 2001;93:1704–1713.

40. Fajardo LL, Saint-Germain M, Meakem TJ 3rd, Rose C, Hill-man BJ. Factors influencing women to undergo screening mammography. Radiology. 1992;184:59–63.

41. Mandelblatt JS, Gold K, O’Malley AS, et al. Breast and cer-vix cancer screening among multiethnic women: role of age, health, and source of care. Prev Med. 1998;28:418–425.

42. Friedman LC, Woodruff A, Lane M, Weinberg AD, Cooper

HP, Webb JA. Breast cancer screening behaviors and inten-tions among asymptomatic women 50 years of age and older. Am J Prev Med. 1995;11:219–223.

43. Haggerty J, Tamblyn R, Abrahamowicz M, Beaulieu MD, Kishchuk N. Screening mammography referral rates for women ages 50 to 69 years by recently licensed family phy-sicians: physician and practice environment correlates.

Prev Med. 1999;29:391–404.

44. Frazier EL, Jiles RB, Mayberry R. Use of screening mammography and clinical breast examinations among black, Hispanic, and white women. Prev Med. 1996;25:118–

125.

45. Lane DS, Zapka J, Breen N, Messina CR, Fotheringham DJ.

A systems model of clinical preventive care: the case of breast cancer screening among older women. For the NCI

Breast Cancer Screening Consortium. Prev Med. 2000;31:

481–493.

Cost Analysis of the NBCCEDP/Ekwueme et al. 635