Background
The burden of cervical cancer in China is substantial, with an estimated ~27,000–130,000 cases in 2010, projected to increase to 42,000–187,000 in 2050, in the absence of any preventative intervention [
1]. Although there may be considerable heterogeneity within China, the evidence is consistent with a higher burden of disease in some rural areas [
1]. Although most women do not have access to screening, a cervical cancer screening initiative has been established for rural women, involving visual inspection with acetic acid (VIA) and cytology. This initiative, which began in 2009, is planned to eventually cover up to “10 million” rural women aged 35–59 years in China [
2]. Cost-effectiveness analyses (CEA) of various cervical cancer prevention strategies, including the use of primary screening for the presence of the human papillomavirus (HPV) with a rapid-throughput technology (
careHPV, Qiagen MD, USA) in rural China have previously been conducted [
3‐
5].
Cost estimation is a central feature of health economic analyses [
6] and has a direct impact on the CEA outcomes. In principle, the results of costing studies from different approaches (micro-costing verses hospital charge estimates) and perspectives (societal verses health provider) may vary as a function of the costing study methodology. In many previous health economic studies related to cervical cancer prevention in less developed settings, cost data have been reported as part of broader cost-effectiveness analyses [
3,
5,
7‐
9]. Some costing studies conducted in other developing countries have reported information on cervical screening and diagnosis; but very little quantification of the costs related to treatment for cervical pre-cancer and cancer has been performed [
10‐
12]. A few local groups in China have estimated cervical cancer treatment expenses, mainly based on charge records from hospitals in urban areas [
13‐
16]. The most precise costing approach, the micro-costing method [
6], has not been used extensively in studies in China, either in relation to costing studies for cervical cancer prevention, or more broadly for other diseases.
The objectives of the current study were to use a micro-costing approach, and from a societal perspective, to estimate the aggregate costs of various cervical screening tests that are potentially feasible in low resource settings [
5], as well as diagnostic procedures and treatment in rural China. This analysis was conducted as a part of the “National Key Technology R&D Program in the 11th Five-Year Plan” project (“11
th-5” project), which is a multi-centre epidemiologic and health economic study project supported by the Chinese Ministry of Science and Technology. More than 12,500 women have been screened through this project, using cytology or HPV DNA-based screening for urban areas, and visual inspection or rapid lower-cost HPV DNA testing for rural areas. The “11
th-5” project was conducted as a research component of a larger visual inspection based screening initiative (Early detection and early treatment of cervical cancer program, EDETCC) [
17].
Discussion
The present analysis provides a range of calculated aggregated costs for cervical screening, diagnostic procedures and the treatment of cervical precancerous lesions and invasive cervical cancer in rural China. We found that the scale on which future cervical screening initiatives will be conducted (i.e. the volume of women screened) will have a major impact on the diagnostic costs in this setting, and we have quantified this effect in detail. The diagnostic costs are comparable to screening costs in situations involving high-volume screening, but greatly increased in lower-volume settings, which is a key consideration for the scale-up phase of new screening programs. The findings of this study can be applied to inform future budget planning and cost-effectiveness evaluations of planned potential large-scale cervical screening programs in rural China.
When considering the
careHPV test, we found that the self-sampling method is slightly less expensive than the physician-sampling method, as expected, due to savings in clinician time. However, the overall savings due to implementing self- rather than clinician-sampling were relatively modest, of the order of $0.50 or less; this modest saving is because the unit cost of HPV testing (i.e. the cost at which the test is supplied by the manufacturer, assumed to be $5 in this analysis) has more significant impact on the aggregated costs than the associated labour cost does in this setting. As expected, we found that the costs of visual inspection screening tests are lower than those related to
careHPV screening, because a large proportion (up to 40%) of the aggregated test cost is directly related to labour. However, this advantage in affordability terms should be interpreted in context of the most recent data on the relative efficacy of these screening tests. For example, in a recent study in rural China it was found that the accuracy of HPV test for the detection of CIN2+ is substantially better than for VIA [
18]; furthermore, a recent randomised trial in rural India found that there was no significant reduction in the rate of cervical cancer death over eight years after a single round of VIA screening compared to no intervention, whereas a single round of HPV screening was associated with a mortality reduction of approximately 50% [
19]. Cost-effectiveness analyses, therefore, are needed to fully represent the full picture of the relative benefits and costs of visual inspection based screening compared to HPV-based screening [
5].
The aggregated costs of visual inspection screening and HPV screening from other less developed countries (including India, Kenya, Peru, South Africa and Thailand) have previously been estimated using various costing methods such as quantity-and-price approaches. It is difficult to directly compare the findings of these prior studies with our results, since these costs were estimated in different countries and currencies and for the year 2000 or 2002 [
7,
9,
11]. However, our estimation of the aggregate cost of VIA is substantially lower than that from a recent study conducted in Ghana, in which resource utilisation data were collected from four screening centres; in that study the costs of VIA were found to vary from US$5-15 (at 2009 prices) [
12]. In prior studies, labour costs have tended to comprise a major part of VIA costs; for example, personnel costs comprised more than half of the aggregate costs in India, Thailand, Kenya and Ghana [
7,
9,
11,
12], whereas in our base case analysis in China we found that this proportion was approximately one third of the total cost. For
careHPV screening in this rural Chinese setting we found that direct costs related to supplies, equipment, and laboratory processes had a more substantial impact on the aggregated costs than other screening costs, which is consistent with previous work in other developing countries, although it should be borne in mind that in some prior studies the specific use of the
careHPV technology was not assumed [
7,
10,
11].
The importance of screening scale has also been addressed by the recent Ghanaian study which found that the number of women screened per provider was one of the most important determinants of costs [
12]. Our baseline estimate of aggregate test costs was based on high screening volumes, and in this situation we found that the total cost of colposcopy is lower than that of
careHPV screening. Colposcopy may have relatively low sensitivity compared to HPV DNA screening, but it has been found that colposcopy was more sensitive than visual inspection screening in rural China [
27,
33]. This raises the possibility that, in a high volume screening situation in this setting, colposcopy may be a viable alternative to HPV testing as a screening test; this has been suggested in the context of other settings [
20]. However, this would require further extensive clinical and cost-effectiveness evaluation in this setting before implementation. Our baseline results were derived assuming the annual number of screened women in a program organised at the county level was over 11,000. As an example, if this number of women were screened in Xiangyuan County in rural Shanxi Province, this would represent an estimated population coverage in women aged 30–59 years of approximately 24%. Therefore, under the base case assumption of a high-volume screening initiative, one screening team could implement close to population-wide coverage in the target age group if screening were conducted over a 4–5 year period. Although, it is quite difficult to reach high coverage within organised programs unless effective individual invitation policies are developed, potentially 60% participation on a per-woman basis has been shown to be achievable in this setting [
5]. By contrast, if only 2000 women were screened (corresponding to an approximate annual population coverage level of 4% in the target age group), the aggregate cost of the colposcopy examination would be tripled, and would exceed the total cost of
careHPV. Similarly, the relative cost of histopathology would increase considerably in lower volume screening situations. This is because in this rural setting, the cost of labour is relatively low compared to capital equipment costs. These findings imply that consideration of the scale and scale-up timing of new cervical screening programs in rural China is very important, and may influence considerations of the most appropriate screening test to be implemented.
The current study also estimated the costs related to cervical precancer and cancer treatment in this rural Chinese setting. We found that the direct medical costs of LEEP, CKC and simple and radical hysterectomy varied from $61–$544, depending on the procedure and whether conducted at county or prefecture level, and that direct non-medical expenditure varied from $83 to $494 for pre-cancer/cancer treatment. These findings may be comparable to those presented in a prior estimate which took a broad perspective across 25 countries (including China) estimated the direct medical cost of cervical cancer treatment for China on average as ~1,500 international dollars (I$) in 2005 for cancer stage-I and I$1,700 for stage II–IV cancer, by leveraging available data in select other countries and extrapolating to China based on gross domestic product per capita and some other indicators [
8]. The costs obtained from the prefecture hospital ($281–$544) were generally found to be higher than those incurred in the county hospital ($61–$133). Given the relatively low annual net income for rural residents in China (~$700 per capita in 2008) [
24], these findings suggest that the treatment of cervical pre-cancer and cancer, particularly treatment in prefecture hospitals, leads to a considerable economic burden for local patients. This raises the possibility that the overall treatment cost could potentially be reduced by treating more patients at county rather than prefecture hospitals; although it is not clear if the cost differentials in treatment would be maintained in a situation of high volume screening with a consequently enlarged referral population for treatment. To develop and strengthen the health care capacity of county hospitals and to enhance their quality control processes would be an important objective for the local health care system.
The micro-costing approach we used in this analysis built on an earlier assessment conducted in 2005 [
3], by adding more elements to cost the complete pathway from screening and diagnosis to treatment, and by implementing a full quality control process. We found that CKC is more expensive than LEEP ($120 for CKC, $61 for LEEP), but the earlier study reported very similar values for the two procedures ($87.13 for CKC, $87.58 for LEEP) [
3]. Overall, treatment costs from the current study were generally lower than those of the earlier study [
3]. Table
5 provides summary information for both studies. Limited data are available on cervical cancer treatment cost estimates based on hospital charge records [
14‐
16] (Table
5). Our micro-costing estimates were found to be relatively lower than those reported by local groups [
14‐
16]. For example, two previous analyses conducted in provincial hospitals reported an average treatment expense per patient with cervical cancer as ~ $1,390 (1997–2001, N = 106) [
15], and as ~ $1,464 (2006) and ~ $1,185 (1996–2006, N = 1,564) [
16]; and a county-hospital-based cost-benefit study reported values based on expert opinion of ~ $585 for CKC and $1,756 for radical hysterectomy [
14]. Our previous work using audits of hospital charges to estimate direct medical costs also found that the average fee per cervical cancer patient was ~ $575 at county level and ~ $1,550 at prefecture level [
4,
5] (Table
5).
Table 5
Comparison of reported cost estimates in relation to cervical cancer for China
The current study | 2008–2009 (English) | Shanxi, Prov. | Rural | Purely costing study | Micro-costing approach (county and prefecture level) | Societal (direct medical and non-medical costs) | Screening, diagnosis and treatment | Indicated in Tables 1, 2, 3, and 4 of the manuscript |
Shi et al 2011, Canfell et al 2011 [ 4, 5] | 2008 (English) | Shanxi, Prov. | Rural | Cost-effectiveness analysis | Micro-costing for screening, diagnosis and part of treatment (LEEP and simple hysterectomy); Checking audits of hospital charges to estimate cancer treatment costs | Societal (direct medical and non-medical costs) | Screening, diagnosis and treatment | VIA screening: $3.55 (mobile), $4.30 (program-based); VILI screening: $0.40; careHPV: $9.20 (self-sampling), $10.34 (provider-sampling); LEEP: $55.95; Cancer treatment: $628 (FIGO I); $1953 (FIGO II); $1810 (FIGO III); $663 (FIGO IV) |
| 2005 (English) | Shanxi, Prov. (Rural) | Rural | Cost-effectiveness analysis | Micro-costing (county and national level) | Societal (direct medical and non-medical costs) | Screening, diagnosis and treatment | Direct medical costs at a county level: Colposcopy: ~$8.21; Biopsy: ~$15.46; LEEP: ~$87.58; CKC: ~$87.13; Simple hysterectomy: ~$354; Local cancer: ~$387; Regional/distant cancer: ~$1,739 |
| 2005 (English) | 25 Asia Pacific regions, including China | Not indicated | Broad cost- effectiveness analysis | Leveraged available data in select other countries and extrapolate to China based on several indicators, including gross domestic product per capita | Societal | Screening, diagnosis and treatment | Average direct medical cost of cervical cancer treatment: ~I$1,500 for cancer stage I, ~I$1,700 for cancer stage II–IV |
| 2006 (Chinese) | A county-level site (Liuyang) in Hunan Prov. | Rural | Cost-benefit analysis | Standard hospital charges (released by the local Office of Health of Hunan Prov.) for screening and diagnosis; expert opinion for treatment | Medical cost only | Screening, diagnosis and treatment | Pap smear screening: ~$4; Colposcopy: ~$13; Biopsy (one punch): ~$18; CKC: ~$585; Radical hysterectomy: $1,756; Mainly chemo/ radiotherapy: ~$3,658 |
| 1996–2006 (Chinese) | Lanzhou City, Gansu Prov. | Urban | Time-trend analysis of in-patient hospitalization fee (cervical cancer and breast cancer) | Hospital bill audits review (N = 1,564) | In-patient fee only | Cancer treatment only | Average treatment expense per patient with cervical cancer: ~$1,464 (2006) and ~ $1,185 (1996–2006) |
| 1997–2001 (Chinese) | Wuhan City, Hubei Prov. | Urban | Time-trend analysis of in-patient hospitalization fee (cervical cancer only) | Hospital bill audits review (N = 106) | In-patient fee only | Cancer treatment only | Average treatment expense per patient with cervical cancer: ~$1,390 (1997–2001) |
| 1996–2001 (Chinese) | Beijing City | Urban | Analysis of in-patient hospitalization fee and clinical outcomes (Hysterectomy only) | Hospital bill audits review (N = 4,180) | In-patient fee only | Treatment only | Average treatment expense of various hysterectomy types: ~$736 - ~ $1,094 |
This analysis has several limitations. Firstly, the study was conducted in a county-level and a prefecture–level hospital, but in actuality, some cervical cancer patients from rural areas travel to provincial hospitals for treatment, and we were not able to quantify costs in this setting. Secondly, we did not take into account the costs of minor or major related complications, and this issue potentially has greater impact on the treatment cost of cancer patients in more advanced cancer stages. Thirdly, as previously described, we have not included all the overhead costs and post-treatment follow-up costs in the final aggregate cost calculation. Fourthly, the treatment costs of the current study are presented by treatment type, rather than by cancer stage (used by most cervical cancer prevention cost-effectiveness studies), although these can be converted to costs by stage if local data on treatment patter-of-care by stage are available. Fifthly, it is not possible in a direct costing study such as this to directly assess the impact of a new screening program on the future quality of cancer treatment. Sixthly, the absence of costs related to cytology is also one of the limitations of the current study. Furthermore, it should be noted that the screening and diagnosis costs presented in the current study, should be regarded as conservative (low-end) estimates, considering the high-volume screening we assumed in the baseline estimate.
Two recently published cost-effectiveness evaluations have utilised preliminary data for screening and diagnostic costs which were obtained from the present costing study. The first of these evaluated the cost-effectiveness of different screening technologies and approaches in rural Shanxi Province [
5]. Assuming high screening volumes, it was found that the cost-effectiveness of primary
careHPV screening compares favourably to that for visual inspection screening methodologies [
5]. The data have also been used in analysis of the cost-effectiveness of combined screening and HPV vaccination approaches in rural China [
4]. In the absence of detailed data on HPV vaccine costs in this setting, this analysis took a threshold approach to identify the cost per vaccinated girl (CVG) at which HPV vaccination would become a cost-effective alternative or addition to screening women with
careHPV. We found that strategies involving vaccination would be cost-effective at CVGs of US$50–54 or less, but at CVGs > $54, screening-only strategies would be more cost-effective [
4].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JFS, JFC, KC and YLQ were responsible for the study concept and design. JFS, RL, JFC, XXF, JFM, YZZ, FHZ, LM, ZFL and YN collected the data from field studies. JFS, KC, JFC, XXF and YLQ analysed and interpreted the data. JFS and KC drafted the manuscript. JFC, XXF, JBL and YLQ revised the manuscript. All authors read and approved the final manuscript.