Errata: Correction Volume 102, Issue 9, 1748, Article first published online: 18 August 2011
To whom correspondence should be addressed. E-mail: email@example.com
The epidemiological characteristics of breast cancer in Korean women are different from the characteristics reported in Western women. The highest incidence rate occurs in Korean women in their 40s. The purpose of this study was to determine the most cost-effective screening interval and target age range for Korean women from the perspective of the national healthcare system. A stochastic model was used to simulate breast cancer screenings by varying both the screening intervals and the age ranges. The effectiveness of mammography screening was defined as the probability of detecting breast cancer in the preclinical state and the cost was based on the direct cost of mammography screening and the confirmative tests. The age-specific mean sojourn times and the sensitivity of the mammography were applied in the stochastic model. An optimal cost-effectiveness was determined by the incremental cost-effectiveness ratio and lifetime schedule sensitivity. Sensitivity analyses were undertaken to assess parameter uncertainty. The selected cost-effective strategies were: (1) the current biennial mammography screenings for women who are at least 40 years old; (2) biennial screening for women between the ages of 35 and 75 years; and (3) a combination strategy consisting of biennial screening for women aged between 45 and 54 years, and 3-year interval screening for women aged between 40 and 44 years and 55 and 65 years. Further studies should follow to investigate the effectiveness of mammography screening in women younger than 40 years in Asia as well as in Korea. (Cancer Sci 2009; 100: 1105–1111)
Breast cancer has become the fastest-growing cancer in Korean women due to their Westernized lifestyle, accelerated by rapid socioeconomic development.(1) A national biennial screening program including mammography for women at least 40 years of age was launched in Korea in 1996.(2) Approximately 30.8% of women had mammography screening in 2004 or 2005.(3)
Mammography has been used worldwide for the early detection of breast cancer since the 1980s. Many randomized controlled trial studies have shown that mammography screening for women aged between 50 and 65 years is indeed an effective tool for the early detection of breast cancer and for reduction of the breast cancer mortality rate.(4) It was also reported that a screening plan for women in their 40s or for women older than 65 years might be most cost-effective.(5,6) In recent years, there have been several debates as to whether the screening of women older than 70 years of age is cost-effective.(7) The consensus regarding a mammography screening plan in Western countries is to perform it annually or every 2 or 3 years for women who are at least 40 years old.(8) Korean women are also recommended to follow the mammography guidelines developed in the West.
However, several Korean researchers have questioned whether these Western guidelines are appropriate for Korean women because the epidemiological patterns of breast cancer among Korean women are different from those in Western women, as shown in Figure 1. For example, the incidence of breast cancer among Korean women as well as other Asians is less than 50% of the breast cancer incidence in the West although the incidence rate has sharply risen in Asia.(9,10) Also, the peak incidence rate of breast cancer among Korean women has been observed in women between the ages of 45 and 49 years, whereas the incidence rate of breast cancer among women in the West has a linear relationship with age (Fig. 1).(2,11–14) A new strategy for cost-effective screening of women in Korea might be required based on the incidence pattern by age group and both the cost-effectiveness and sensitivity of the mammography.
Based on a previous study that determined an effective screening schedule for the early detection of breast cancer among Korean women, the schedule sensitivity of biennial mammography among Korean women was 50–65%, which is consistent with the best sensitivity estimate (58–66%) of a screening schedule based on the threshold method.(15,16) The study showed that biennial screening is effective in detecting breast cancer early in Korean women between the ages of 40 and 65 years. However, no studies have investigated the cost-effectiveness of screening Korean women by analyzing various age ranges and screening intervals.
There are two major considerations regarding the cost-effectiveness of mammography screenings: (1) the age range; and (2) the screening interval for each age group in question. The purpose of the present study was to determine the most cost-effective mammography strategy for Korean women by comparing the costs of various screening intervals with different age ranges from 35 to 75 years of age within the current cancer screening system in Korea.
Materials and Methods
Model used for the cost-effectiveness analysis. To determine an optimal cost-effective mammography screening, we calculated the ICER. The ICER is defined as the ratio of the changes in costs and effectiveness of a screening strategy to an alternative strategy. By calculating the ICER of each screening strategy, strategies were classified into one of three categories: (1) non-dominated; (2) dominated; or (3) extended dominated.(17) A dominated strategy is defined if it generates worse effects and higher costs than an alternative strategy. Extended dominance occurs when a strategy is less effective and has a higher ICER than an alternative strategy.
The effectiveness of a screening plan was measured in terms of the probability of detecting the breast cancer while in the preclinical state, as proposed by Lee and Zelen.(16) It is basically assumed that the natural history of breast cancer is progressive in the manner of S0→ Sp→ Sc, where S0 means the disease-free state, Sp the preclinical state in which the disease has no symptoms but can be diagnosed, and Sc the clinical state. The goal of an early detection strategy is to diagnose a breast cancer in the preclinical state. The sensitivity of the mammography, the distribution of the sojourn time in the preclinical state, and the age-specific incidence rates were required to estimate the probability. Lee and Zelen made an assumption in their analysis that both the MST and the sensitivity of the mammography were constant values.
However, some studies have shown that both the MST and the sensitivity of the mammography might be verified among different age groups.(18,19) In the present study, we extended upon Lee and Zelen's method by including the age-dependent MST and mammography sensitivity. We assumed that screenings would begin at ages t1 < t2 < . . . < tn. Let us define Dr as the probability of detecting breast cancer in the preclinical state at screenings carried out at age tr. Therefore:
where w(x) is the transition probability from S0 to Sp, which can be calculated by age-specific breast cancer incidence and an assumed distribution of sojourn time in Sp, is the survival distribution of the sojourn time in Sp at age t with tr–1 ≤ t < tr, and βr is the sensitivity of mammography at age tr. The sum of the probability Dr at each screening, , represents the probability of detecting breast cancer in the preclinical state at schedule exams.
Cost in this study was defined as the direct cost of the mammography screening, which was reasonable for low-incidence disease.(16) The costs for false-positive outcomes were included in calculating this cost. However, costs for other adverse effects from mammography screening, such as pain and discomfort from breast compression or patients’ revisits for additional imaging, were not considered in the model. The time horizon for the study was 30–85 years of age. We adopted a perspective of the national healthcare system and discounted future costs and cases found in preclinical stage at an annual rate of 3%.
In addition, the lifetime schedule sensitivity, which is the fraction of cases diagnosed at scheduled exams during an individual's lifetime, is also likely to give an intuitive indication of the effectiveness of a strategy because the detection probability in the preclinical state depends on the incidence rates of breast cancer.(16) In the present study, the lifetime schedule sensitivity is represented as D/(D + I), where D is the expected total number of detected cases upon the scheduled examinations and I is the expected total number of interval cases in the age interval [0, 85].
Based on current recommendations, we generated 40 possible screening strategies by varying both the screening intervals and the age ranges of the women being screened (Table 1). The four ages that were chosen to begin the mammography screenings were 30, 35, 40, and 45 years, and the three ages that were chosen to end the screening were 65, 70, and 75 years. Screening intervals were set as occurring every year, every other year, or every 3 years. Additionally, combination screening strategies were included with different screening intervals according to the age groups, as derived from Lee et al.(15) Combination strategies consisted of biennial screening for women between the ages of 45 and 55 years and then a screening interval of every 3 years for women younger than 45 years or older than 55 years.
Table 1. Mammography screening strategies for cost-effectiveness analysis
Starting age (years)
Stopping age (years)
Starting age (years)
Stopping age (years)
Screening occurs every 2 years for women between the ages of 45 and 54 years, and every 3 years for women younger than 45 years of age or older than 55 years of age:
Screening of M_t_3065 occurs at the ages of 30, 33, 36, 39, 42, 45, 47, 49, 51, 53, 55, 58, 61, and 64 years.
Screening of M_t_3565 occurs at the ages of 35, 38, 41, 44, 46, 48, 50, 52, 54, 56, 59, 62, and 65 years.
Screening of M_t_4065 occurs at the ages of 40, 43, 45, 47, 49, 51, 53, 55, 58, 61, and 64 years.
Screening of M_t_4565 occurs at the ages of 45, 47, 49, 51, 53, 55, 58, 61, and 64 years.
Data and model assumptions. To estimate age-specific breast cancer incidence rates, we utilized two data sets: one was from the Korea Central Cancer Registry data in 2002 and the other was from female population projections in 2002 based on the results of the 2000 population census in Korea. Detailed descriptions are provided in our previous study.(15) Incidence rates were calculated for each age group and ranged from 0 to 4 years of age to more than 85 years of age. The cost estimates of the mammography and the confirmative examination were obtained from the National Health Insurance Corporation of Korea in 2007.(20)
Previous studies have demonstrated that MST in the preclinical state were exponentially distributed.(21,22) Tabar et al. reported that MST for the 40–49-, 50–59-, 60–69-, and 70–79-year age groups were 2.4, 3.7, 4.2, and 4.0 years, respectively.(23) Thus, the assumption for the current study was that the sojourn time in the preclinical state might follow an exponential distribution according to age-specific MST. The MST in the preclinical state ranged from 2 years for women younger than 50 years of age, 3 years for women aged 50–59 years, and 4 years for women older than 60 years of age. The sensitivity of the mammography was assumed to be 0.6 when carried out on women younger than 50 years of age and 0.7 when the women were older than 50 years of age, based on the results of a previous study.(18) To estimate the total cost, we defined the unit cost of the mammography, the unit cost of the confirmative tests, and the specificity of the mammography as US$19.50, US$171.90, and 0.95, respectively.(15,20) To evaluate the robustness of the proposed screening strategies, one-way sensitivity analyses were conducted by changing the MST in the preclinical state, the sensitivities and the specificities of the mammography, costs, and discount rates. The study parameters are shown in Table 2 with their references.
Table 2. Baseline assumptions and ranges tested in the sensitivity analysis
(2,3,4), 2 years (ages <50 years), 3 years (ages 50–59 years), and 4 years (ages ≥60 years).
Table 3 presents the number of cases found per 100 000 screened, costs, costs per cases found, incremental cases found, incremental costs, ICER, and lifetime schedule sensitivity of possible mammography screening strategies for women between the ages of 30 and 75 years. The most extensive strategy, which was an annual mammography screening for women who were between the ages of 30 and 75 years, had the highest lifetime schedule sensitivity (73.1%) and found 196.4 preclinical cases per 100 000 screened. The least extensive screening plan, which consisted of a screening interval of 3 years for women between the ages of 45 and 65 years, had the lowest lifetime schedule sensitivity (31.9%) and detected 72.5 preclinical cases per 100 000 screened. Among the 40 alternative strategies tested, there were 10 strategies that were identified as non-dominated strategies and other strategies were eliminated by either simple or extended dominance. The least extensive strategy was compared to a plan in which no screening was carried out.
Table 3. Cost effectiveness of mammography screening strategies for Korean women
Cases found in the preclinical state per 100 000 persons = detection probability × 100 000.
Incremental cases found in the preclinical state compared with the next least expensive, non-dominated strategy (in bold). The M_3_4565 strategy was compared to a plan without screening.
Incremental costs compared with the next least expensive, non-dominated strategy (in bold). M_3_4565 strategy was compared to a plan without screening.
Incremental cost-effectiveness ratio (ICER) = incremental costs/incremental cases found in preclinical state. E. dominated, extended dominated.
7 250 483
7 250 483
8 506 677
1 256 194
9 290 906
2 040 423
9 556 833
9 861 570
2 611 087
10 842 302
11 078 989
1 217 418
11 610 572
1 749 002
12 089 998
12 102 224
2 240 654
12 843 584
1 235 600
13 459 835
1 357 611
13 704 012
14 015 621
1 913 397
14 538 204
15 010 750
15 603 638
15 886 706
1 871 085
16 853 763
16 960 781
1 074 075
17 624 915
17 898 314
18 417 046
1 456 265
19 662 194
2 701 413
20 289 846
21 272 632
4 311 851
22 793 872
23 437 162
24 660 810
3 388 179
5 732 299
4 459 667
26 150 755
27 170 094
5 897 462
30 315 892
9 043 261
33 028 068
11 755 436
35 144 428
13 871 797
38 288 583
3 144 155
40 999 220
5 854 792
44 386 962
47 529 333
6 530 113
1 088 352
50 238 301
9 239 081
All biennial mammography screenings starting at 40 years of age were not dominated, including M_2_4065, M_2_4070, and M_2_4075. A 2-year interval plan for the 40–65-year-old age group (M_2_4065) had an ICER of $197 257 per 1 case found. A combination screening strategy (M_t_4065) of 3-year intervals (for ages 40–44 years), 2-year intervals (for ages 45–54 years), and 3-year intervals (for ages 55–65 years) was very similar to the 2-year interval plan for ages 40–65 years (M_2_4065) in terms of the number of cases found and the costs. The 3-year interval non-dominated screening plans such as M_3_4565 and M_3_4065 showed fewer cases per 100 000 screened, which was less than 50% of cases from the most expensive strategy (M_1_3075).
Compared to a 2-year interval plan for the 40–75-year age group (M_2_4075), extending biennial screening to a starting age of 35 years (M_2_3575) was the next non-dominated strategy with an ICER of US$291 341 per 1 case found. The non-dominated screening plans above M_2_3575, such as M_1_3565, M_1_3575, and M_1_3075, were not relatively cost-effective because the costs of these strategies were two to three times higher than the cost of the current biennial screening for women older than 40 years of age.
Figure 2 illustrates our expansion path results consisting of the most cost-effective screening strategies. The expanded path graph, which plots the expected number of detected cases against the costs of each strategy, was illustrated based on the ICER presented in Table 3. The graph representing the ICER shows a slow increase to the M_2_3575 strategy, but a steep increase at the M_1_3565 strategy. In other words, the graph visually demonstrates that the strategies above M_2_3575 had large additional costs for increasing effectiveness and were thus less cost-effective strategies. Among the remaining non-dominated plans, M_3_4565 and M_3_4065 found only 36.9% (72.5 cases per 100 000 screened) and 45.5% (89.4 cases per 100 000 screened) of the detected preclinical cases in the most extensive screening of M_1_3075 (196.4 cases per 100 000), respectively, which had the highest sensitivity. Finally, considering the current national healthcare budget limit, ICER, and the relative detection probability in the preclinical stage, the M_t_4065, M_2_4065, M_2_4070, M_2_4075, and M_2_3575 strategies were chosen as the most cost-effective strategies for Korean women.
The sensitivity analyses along with cost-effective strategies were carried out based on the different setting of parameters. The non-dominated strategies selected from each model for sensitivity were consistent with those from the baseline model. Table 4 shows the robustness of cases and costs in the M_t_4065 and M_2_3575 strategies with the different parameters. According to the various values of the parameters, the models had approximately 100% of the cases and cost from the baseline model, except the unit cost of mammography and discount rate. The costs per 100 000 persons for the different values of unit costs of mammography varied from 50 to 150% of the baseline model.
Table 4. Sensitivity analysis of recommended mammography screening strategies for Korean women
The consensus among researchers regarding mammography screening for women in their 50s and 60s is that these screenings significantly reduce the breast cancer mortality rate,(24–27) but there are limitations to these randomized controlled trials in evaluating the effectiveness of mammography screening.(28,29) There is ongoing debate as to whether it is beneficial to carry out mammography screening for women in their 40s because the incidence rate of breast cancer and the sensitivity of mammography in women in this age group are relatively lower than for the older age groups.(5) Despite this debate, all of the major US medical organizations recommend that women over the age of 40 years should undergo mammography annually or biannually because screening could reduce the mortality of breast cancer by 7–23%.(30) Researchers have suggested that biennial mammography for women between the ages of 65 and 74 years would reduce mortality at a reasonable cost, and this is recommended to individuals as their own medical choice.(6)
In Korea, the incidence rate of breast cancer is markedly lower than that in the West, as shown in Figure 1. Breast cancer has an earlier onset in Korean women than in the West, and the highest incidence rate occurs among women in their 40s. Recently, the breast cancer incidence rate has increased sharply in Korea and both the public and private healthcare sectors are paying much more attention to the importance of mammography screening.(31) Although mammography screening for Korean women is expected to reduce the breast cancer mortality rate, the optimal screening modality, in terms of cost-effectiveness, remains unclear. A cost-effective screening schedule in terms of screening interval and age range should be planned carefully based on the epidemiological figures of breast cancer in Korea.
Recent studies on the cost-effectiveness of mammography in Asia have reported that the prevalence rates and peak incidence of breast cancer is among women in their 40s, which is similar to the epidemiological pattern of breast cancer in Korea (Fig. 1). In Hong Kong, researchers conducted a cost-effectiveness study of mammography screening for Chinese women using a state-transition Markov model.(32) The study reported that the most cost-effective screening strategy would be biennial mammography for Chinese women between the ages of 40 and 69 years. The authors also indicated that the biennial screening modality was optimal when they projected the cost-effectiveness using the Markov model based on the QALY saved and life years saved. They claimed that this strategy might not be cost-effective when they used the arbitrary threshold of US$50 000 per QALY. The researchers speculated that the incidence rate among women in Hong Kong was less than half of that in the West, although the breast cancer incidence among Chinese women has increased dramatically in recent years due to Westernization and socioeconomic development.
In another study in Japan, researchers reported that biennial mammography screening with clinical breast examination might be the most cost-effective strategy but less effective than annual mammography screening with clinical breast examination.(18) They also reported the same results for women in their 40s. Mammography screening in Korea might not be as cost-effective based on the threshold of US$50 000 per QALY because the breast cancer incidence rate in Korea is lower than in China and Japan.
The purpose of our study was to identify the most cost-effective screening interval and target age range based on the epidemiological characteristics of breast cancer in Korean women. Because Korea has a mandatory NHI program, the entire population is enrolled in NHI. Mammography screenings sponsored by NHI are free of charge for most of the enrollees. Therefore, we carried out the analysis from the perspective of the national healthcare system because we were primarily concerned with determining direct payments from the government. We didn't measure indirect costs or lost productivity associated with breast cancer in this study.
Cancer screening is aimed at reducing deaths from cancer; therefore, cancer mortality is widely used as the final endpoint for evaluating screening strategies. However, we considered detected cases at the preclinical state as potential surrogate outcomes in screening strategies. The reasons why we couldn't use mortality reduction, QALY, and disability adjusted life years as outcomes of cancer screening were as follows. Endpoints for previous models mostly demand various parameters such as progressive transition probabilities from state to state, fatality rates in each state, and so on. To have adequate statistical estimates, long follow-up clinical trials with a sufficiently large number of subjects would be required.(33) We could not obtain a sufficient amount of relevant data for breast cancer in Korea, such as the various parameters mentioned above. In addition, the evidence for decreased breast cancer mortality following early detection was shown in a series of early detection clinical trials.(33,34) Therefore, using the detected cases at preclinical state might be the best alternative way to measure effectiveness in this study.
The results of sensitivity analyses showed that the M_2_3575 and M_t_4065 strategies were universally cost-effective strategies. All biennial strategies (M_2_4065, M_2_4070, and M_2_4075) were selected as cost-effective strategies except in the case of MST. Our results showed that the current biennial mammography screening for women aged at least 40 years is cost-effective, and biennial screening with the first mammography at the age of 35 years could be an alternative cost-effective screening program in Korea. The results from a study involving Chinese women were congruent with our results in that the first mammography screening should be carried out when a woman turns 35 years old in countries where the highest breast cancer incidence rate is among young women.(32)
In addition, an alternative combination screening plan with biennial screening for women between the ages of 45 and 55 years, and 3-year interval screening for women between the ages of 40 and 45 years and 55 and 65 years (M_t_4065) was also found to be cost-effective. When a mammography interval of 3 years is used for women in their 40s, caution must be exercised as the sojourn time of breast cancer might be shorter in young women.
One limitation of the present study was the lack of data on mammography sensitivity and specificity for different age groups of Korean women. As a result, the analysis had to be carried out with reference data from Japan,(18) which were assumed to be similar to Korean data. It was hypothesized that the participation rate of mammography screenings would be the same for all age groups. In reality, different screening intervals and age groups could affect the participation rate and also the analysis of the cost-effectiveness, depending on the strategies chosen.
Our study was the first study to use a probability model to find cost-effective mammography screening strategies. The analysis that was used to simulate the estimated cost of mammography was based on fewer assumptions and provided more robust results when compared to the existing cost-effectiveness analysis method.
In conclusion, the current biennial mammography screening for women aged at least 40 years is cost-effective. However, based on the relatively low incidence of breast cancer in Korean women and the highest incidence rate in women in their 40s, we cautiously suggest that the age for the first mammography screening could be changed from 40 to 35 years in Korean women, and also a screening schedule with a combination of 2- and 3-year intervals would be considered a cost-effective alternative. Further studies might be needed to support our conclusions and to examine the effectiveness of mammography screening for women younger than 40 years of age. Also, additional research should be conducted to compare the effectiveness of different mammography screening plans used in Asia where the incidence varies and is reportedly different from the West.
This work was supported by a grant from the National R & D Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (0520150-1).