Cost-Effectiveness of Korea’s National Cervical Cancer Screening Program

The annual incidence of cervical cancer worldwide has increased by 0.6%, from 378,000 cases in 1980 to 454,000 cases in 2010 (Forouzanfar et al., 2011). It is one of the most common cancers in developing countries, and it also represents a major health issue in Korea and Japan, economically developed Asian nations (Konno et al., 2008; Razak et al., 2013). In Korea, cervical cancer accounted for 9.8% of new cancer cases in 2002, although the age-standardized incidence has steadily declined from 19 per 100,000 women in 1993-1995 to 15 per 100,000 women in 1999-2002 (Chung et al., 2006; Shin et al., 2007). In Japan, the Papanicolaou (Pap) smear, the conventional cytological method for cervical cancer mass screening, is used in an organized cervical cancer screening program (Konno et al., 2008). In Korea, a National Cancer Screening Program (NCSP) including the Pap smear was established in 1999, and, since 2002, coverage has included all National Health Insurance (NHI) beneficiaries and Medical Aid recipients (Cho et al., 2013). Several studies have reported beneficial outcomes associated with cervical cancer screening. For instance, in


Introduction
The annual incidence of cervical cancer worldwide has increased by 0.6%, from 378,000 cases in 1980 to 454,000 cases in 2010 (Forouzanfar et al., 2011).It is one of the most common cancers in developing countries, and it also represents a major health issue in Korea and Japan, economically developed Asian nations (Konno et al., 2008;Razak et al., 2013).In Korea, cervical cancer accounted for 9.8% of new cancer cases in 2002, although the age-standardized incidence has steadily declined from 19 per 100,000 women in 1993-1995 to 15 per 100,000 women in 1999-2002 (Chung et al., 2006;Shin et al., 2007).
In Japan, the Papanicolaou (Pap) smear, the conventional cytological method for cervical cancer mass screening, is used in an organized cervical cancer screening program (Konno et al., 2008).In Korea, a National Cancer Screening Program (NCSP) including the Pap smear was established in 1999, and, since 2002, coverage has included all National Health Insurance (NHI) beneficiaries and Medical Aid recipients (Cho et al., 2013).
Several studies have reported beneficial outcomes associated with cervical cancer screening.For instance, in

Cost-Effectiveness of Korea's National Cervical Cancer Screening Program
Eun Cho 1 , Moon Hae Kang 2,3 , Kui Son Choi 4 , MiNa Suh 4 , Jae Kwan Jun 4 , Eun-Cheol Park 3,5 * one Japanese prefecture, screening participation increased from 0.2% in 1961 to 30.4% in 1991, and cervical-cancer related mortality fell from 12.1 per 100,000 in 1961 to 4.0 per 100,000 in 1994 (Sato et al., 1998).Similarly, a cohort study found that Korean women who had been screened more than once had a significantly reduced risk of invasive cervical cancer or carcinoma in situ compared to unscreened women (Jun et al., 2009).Studies of the cost-effectiveness of Pap screening strategies have also been conducted in Asian countries.However, most of them have used hypothetical economic evaluation models rather than assessing empirical data (Koong et al., 2006;Woo et al., 2007;Chen et al., 2011;Zhao et al., 2012).
The purpose of the present study was to use empirical data to investigate the life years saved (LYS) by cervical cancer diagnosis through the Korean NCSP.In addition, the cost of administering the NCSP was examined in terms of LYS.

Study population and data sources
The target population was Korean women 30 years or older who were invited to participate in cervical cancer screening provided by the NCSP between 2002 and 2007.The NCSP's cervical cancer screening initially included all Medical Aid Program recipients, and, in 2002, it expanded to include all NHI beneficiaries as well (Jung et al., 2010;Han et al., 2012).However, because of the separation of administrative management for NCSP, the NCSP database used in this study included only a subset of the individuals covered according to screening years.For the years 2002-2004, only Medical Aid Program recipients were included in the NCSP database records.All Medical Aid Program recipients and National Health Insurance (NHI) beneficiaries were included in the NCSP database in 2005-2006. Finally, in 2007, Medical Aid Program recipients and NHI beneficiaries in the bottom 50% by income were included in the database.
The NCSP invited women to undergo cervical cancer screening at their even-numbed ages.Using NCSP records, individuals who had received cervical cancer screening were identified and assigned to the 'screened group.'Women who did not undergo screening were assigned to the 'non-screened group'.
Cervical cancer diagnoses in the study population were identified using Korean Central Cancer Registry (KCCR) data.The detected cancers were classified by cervical intraepithelial neoplasia (CIN) stage and spread: CIN 1/2, local (CIN 3), regional, and distant.Cervical cancerrelated mortality was determined by merging databases from the Korea National Health Insurance Corporation and the National Statistical Office.Cost data relevant to screening participation were obtained directly, from the internal accounts of screening units, and indirectly, from published studies and national statistics.

Cost-effectiveness analyses
Cost-effectiveness outcomes were examined from the perspective of the payer, the NCSP.The information used in the cost-effectiveness models is presented in Table 1.In the COST I model, which dealt with direct costs, expenditures related to screening and follow-up examinations were considered.Follow-up examination costs were included only for patients for whom the screen yielded a false positive, as follow-up examinations for true-positive cancer patients can be viewed as a part of the treatment course instead.
The costs associated with traveling to attend screening or follow-up examinations may also be considered screening costs, so the COST II model consisted of the data included in COST I as well as these indirect costs.Because screening and travel costs occurred only if people utilized NCSP cervical cancer screening, the COST I and COST II models were only applied to individuals who had been screened.All costs were inflated to 2009 values using the National Consumer Indices (Statistics Korea, 2012).
To measure the effectiveness of the national cervical cancer screening program, we examined 5-year survival rates and LYS, starting from the year of diagnosis.In determining LYS, people with no mortality record during the 5-year follow-up period were assumed to live until the last year of their life expectancy (Korean Statistical Information Service: KOSIS, 2012).
Individuals in both the screened and non-screened groups were divided into 5-year age groups according to their age at screening or diagnosis.Both effectiveness outcomes and costs were presented per 100,000 people, stratified by age group, to facilitate comparison of the screened and non-screened groups.Outcome information for the total sample was age-adjusted to match the age distribution in Korea as a whole (KOSIS, 2012).
To determine cost-effectiveness of NCSP cervical cancer screening, cost and effectiveness outcomes associated with cervical cancer diagnoses in the screened and non-screened groups were compared.The sums obtained in COST I and COST II were divided by the number of cervical cancer cases diagnosed via screening.Then, the incremental costs attributable to the screening program were evaluated based on the LYS from 2002-2007.In addition, the incremental costs needed to detect additional early-stage cervical cancer cases via screening were examined by considering diagnoses of CIN 1/2 or local-stage cervical cancer.Data management and analyses were performed using Microsoft Excel™ and SAS 9.2 software.

Study population
According to our database, the average cervical cancer screening participation rate in 2002-2007 was 20.2% (Table 2).Women 55-59 years old had the highest rate of participation, at roughly 30%.Women in their early 30s and women over 75 years of age had the lowest NCSP participation rates (Table 2).In 2002, the age-adjusted rate of cervical cancer diagnosis via the NCSP was 146 per 100,000 screening participants (Table 3).The detection rate peaked in 2003-2004, at 186 cancers per 100,000 screenings.Then, the NCSP detection rate gradually decreased; by 2007, it had fallen to 103 cervical cancer diagnoses per 100,000 screenings.The incidence of cervical cancer diagnoses among non-NCSP participants was 49.5 per 100,000 in 2002-2007, accounting for roughly 30% of cervical cancers diagnosed in Korea (Table 3).Overall, the rate of cervical cancers diagnosed was more than twice as high in the screened group.During 2003 and 2004, the NCSP rate of cervical cancer diagnosis was more than triple that in the non-screened group (Table 3).

Incremental clinical and cost outcomes
Among cervical cancer patients screened by the NCSP in 2002-2007, 4.9% died within 5 years.At 16.6%, the 5-year mortality for patients diagnosed with cervical cancer though a route other than the NCSP was more than triple that (Table 3).Although the 5-year mortality rate among screened cancer patients was over 10% in 2003-2004, by 2007 it had dropped to 2.8%.Likewise, the mortality rates among non-screened cervical cancer patients in 2005 and 2006, which were 15.3% and 12.7%, respectively, were half those seen in the previous years (Table 3).Mortality in this group did increase slightly in 2007, however, to 16.2%.
Based on the expenditures associated with screening listed in Table 1, direct and indirect costs per cervical cancer diagnosis were estimated using the COST I and COST II models (Table 4).In 2002-2007, the average direct cost of screening was 7,323,019 Korean Won (KW) (equivalent to 6,498 US Dollars, USD; exchange rate May 2013 1 USD=1,127 KW) per cervical cancer diagnosis, while the average total cost was roughly two times greater (17,721,662 KW;15,725 USD).
Based on a 5-year follow-up of all patients diagnosed with cervical cancer, the number of incremental LYS by screening was 2.31 years (Table 5).The number of LYS was greatest for the age groups 65-69 years (3.18 per diagnosis) and 50-54 years (3.14 per diagnosis).By contrast, in women in their 30s, the incremental LYS was less than 1 year (Table 5).
Using the COST I model, the incremental cost-effectiveness ratio (or ICER, obtained from the ratio of incremental cost:LYS) was 3,175,535 KW/LYS (2,818 USD/LYS) per cervical cancer patient (  5).
In 2002-2007, the NCSP for cervical cancer was associated with a 16.3% increase in early-stage detection relative to the non-screened group (Table 6).The difference in the proportion of early-stage cancers diagnosed was roughly 20% in patients over the age of 50.By contrast, among women aged 30-34 years, the increase in the early-stage detection rate associated with screening was only 4.0%.
For each 1% increase in the early-stage cervical cancer detection rate via screening, the associated direct costs were 418,901 KW (372 USD, Table 6).When both direct and indirect costs were considered, the ICER for the same increase was 1,000,138 KW (887 USD).The age groups 55 years and over had ICER estimates below the age-adjusted average for the sample as a whole (Table 6).

Discussion
The 5-year survival rate for cervical cancer patients in 1993-2002 was reported to be 78.7%(Jung et al., 2007).This estimate is significantly lower than that identified in our research (95.1% in the screened group; 83.4% in the non-screened group), which focused on patients diagnosed from 2002-2007.It is likely that the introduction of the NCSP in 1999 and the gradual expansion of beneficiaries have partially contributed to increased cervical cancer survival rates.Indeed, our results reveal a trend of decreasing 5-year mortality rates, with the exception of the baseline year 2002 (Table 3).The 5-year mortality estimate for cervical cancers identified in 2005-2007 was less than half the estimate for cancers identified in 2003-2004.We found that the average incremental direct screening cost was 3,175,535 KW/LYS (2,818 USD/LYS).When travel costs associated with screening were considered, the incremental cost doubled.From the perspective of the payer, the NCSP, the costs associated with adding one year to a cervical cancer patient's life seem reasonable.In its early years, the absence of empirical data on the cost-effectiveness of screening was considered one of the NCSP's major challenges (Kim et al., 2011).By demonstrating superior cost-effectiveness and survival outcomes, our research reveals that Korea's NCSP for cervical cancer is beneficial.
When ICER values were compared according to age group, women aged 30-34 years required the greatest incremental costs per LYS.This age group also had the fewest LYS (0.20) per diagnosed patient (Table 5).It is possible that women in their early 30s had never undergone cervical cancer screening before attending the NCSP; thus, if they did have cervical cancer, it may have been more likely to reach an advanced stage relative to the cancers detected in other age groups.In addition, with the exception of women aged 80 years and older, the participation rate in the NCSP was lowest in women aged 30-34 years (6.3%, Table 2).This suggests that women in their early-to-mid 30s who underwent cervical cancer screening may be different from the average woman in this age group.Indeed, the increase in early-stage cancers detected via screening was lowest for the 30-34 year-old age group (4.0%).By contrast, ICER estimates per LYS were lowest for patients older than 50 years.In this age range, 20-35% of cervical cancers were detected at an early stage (Table 6), which contributed to the favorable ICER values.
Recently, the prevalence of human papillomavirus (HPV) infection has increased in Korea, resulting in a continued, high burden of cervical cancer (Konno et al., 2008).Although two HPV vaccines are available in Korea, the cost of including them in the National Immunization program is prohibitive.Recent research on the implementation of wide-scale HPV vaccination in Asian countries has predicted beneficial clinical and cost-effectiveness outcomes (Ezat and Aljunid, 2010;Yamamoto et al., 2012).Nevertheless, from the payer's perspective, the expense associated with universal HPV vaccination may seem overwhelming (Praditsitthikorn et al., 2011;Sharma et al., 2012).For this reason, continuous efforts to expand participation in cervical cancer screening are still required.Increasing the uptake of cervical cancer screening is especially important because even though the ICER of adding vaccination to the Pap smear screening would be an cost-effective option, women of low socioeconomic status are less likely to participate in screening or to be able to afford the vaccination (Konno et al., 2008;Park et al., 2011;Praditsitthikorn et al., 2011;Yamamoto et al., 2012).The cost-effectiveness of vaccination in combination with frequent or infrequent screening by the NCSP needs to be evaluated, in light of young women's low screening participation rate and  Recently, the HPV-DNA screening test has attracted public attention, especially in developing countries, as a potentially more cost-effective alternative to the Pap smear (Levin et al., 2010;Flores et al., 2011).It may not be feasible for all women to receive regular Pap smears throughout their lives.Moreover, although the Pap smear is inexpensive, it is less accurate than the HPV-DNA test; therefore, screening at wider intervals with the highlysensitive HPV-DNA test may be more cost-effective than the current system (Gravitt et al., 2010;Levin et al., 2010;Chen et al., 2011;Flores et al., 2011;Shi et al., 2011).While HPV-DNA tests are not covered under the current NCSP, previous studies comparing screening methods have reported that the sensitivity of the Pap smear was improved markedly in combination with the HPV-DNA test (Kim et al., 2013).Further research examining the cost-effectiveness of single and combination test screening should be conducted to determine the most practical cancer screening strategy.
To the best of our knowledge, this is the first study to use empirical data to investigate differential outcomes for cervical cancer cases detected through Korea's NCSP vs other means.However, our study has several limitations.First, we did not consider treatment costs.Although costeffectiveness analysis was conducted from the perspective of the NCSP, a substantial portion of the cost of cervical cancer treatment is paid by NHI, a resource available to the public.Treatment costs for cervical cancer are not expected to differ between the screened and non-screened groups, however, because patients' access to care is identical in terms of NHI coverage.
Second, we did not consider intangible costs and benefits, such as anxiety or discomfort associated with undergoing screening.In addition, we did not examine the utility assigned to a reduced risk of dying from cervical cancer via attending the screening program.Indeed, a Japanese study has shown that the most common reasons for screening refusal are inconvenience, inadequate time, cost, and feelings of shame associated with the procedure (Konno et al., 2008).Although we assume these factors are relatively minor, utility considerations could cause the cost-effectiveness of the NCSP to deviate from our results.Utility assessments should be reviewed thoroughly in future studies.
Third, although the NCSP is open to all Koreans, participation may be influenced by socioeconomic status or education level.Several studies of Korea's national gastric cancer screening program have reported that people with higher incomes, more education, and supplementary private health insurance tend to undergo screening at higher rates (Kim et al., 1994;Hahm et al., 2011;Lee et al., 2011;Park et al., 2011).These results suggest that the superior outcomes of patients diagnosed with cervical cancer by the NCSP may be influenced by the higher socioeconomic status of screening participants.This potential disparity implies that cervical cancer screening participation deserves attention from health social workers.
In conclusion, this analysis of Korea's national cervical cancer screening program demonstrates that, in 2002-2007, a higher rate of cancers was identified in the screened population, and the 5-year mortality was lower for cases that had been screened as well.The incremental direct costs related to screening were between 3,175,535 KW (2,818 USD) and 7,581,679 KW (6,727 USD) per LYS over a 5-year observation period, and, when evaluated in the light of Korea's per capita gross domestic product (32,272 USD in 2012), these estimates appear very costeffective.

Table 1 . Costs Relevant to Participation in Korea's National Cervical Cancer Screening Program
(6,390+17,970+ (d) (e): Division of Medical Information and Technology, (b) biopsy+(c) histopathologic exam+ 19,690+16,880+ Yonsei University Health System, Seoul (d) consultation+(e) specialty consultation fee 50%*5,648) Cost II Model: Cost I+travel costs *All unit costs were inflated to 2009 values; aThe cost for the specialty consultation fee was multiplied by 50% for this analysis, under the assumption that half of the participants with false-positive screening results received a specialty consultation and the remaining half were retested by a general physician; KW, Korean Won DOI:http://dx.doi.org/10.7314/APJCP.2013.14.7.4329  Cost-Effectiveness of Cervical Cancer Screening in Korea

Table 4 . Total Direct and Indirect Screening Costs per Cervical Cancer Diagnosis in the Screened Group, 2002-2007
Cost I includes direct screening costs for cervical cancer.Cost II includes direct plus indirect (i.e.travel) costs.Total costs were age-adjusted and inflated to 2009 values *

Table 3 . Number of Cervical Cancer Cases Detected per 100,000 Korean Women and 5-year Mortality
*The total frequency values were age-adjusted.Cervical cancers in the screened

Table 5 . Incremental Life Years Saved (LYS) and Incremental Cost-Effectiveness Ratios (ICERs) for Cervical Cancer Screening: 2002-2007
*Total LYS and ICERs were age-adjusted.The LYS outcomes were derived from the 5-year period following diagnosis.All cost estimates were adjusted to 2009 values