Estimation of the costs of cervical cancer screening, diagnosis and treatment in rural Shanxi Province, China: a micro-costing study

Screening strategies

We considered those cervical screening tests potentially feasible in rural areas in China, including VIA only, VIA combined with visual inspection with Lugol’s iodine (VILI), and self- or clinician-sampling with careHPV [18] for primary HPV screening. Visual screening appears to be a simple, safe, acceptable and inexpensive approach but there are substantial issues related to test accuracy and repeatability of visual screening, and screening with visual inspection was not found to decrease cervical cancer mortality in a large-scale trial in rural India [19]. CareHPV is a recently developed, rapid-throughput lower cost test for HPV, and has been demonstrated to have high sensitivity for cervical high-grade lesions in a rural Chinese setting; this was 90% for cervical specimens and 81% for vaginal specimens, where vaginal sampling represents the potential accuracy of self-sampling [18]. Women with positive screening results were assumed to be referred for further diagnosis, including colposcopy examination, and biopsy and endocervical curettage (ECC) if indicated. Additional file 1: Table S1 provides more details of the screening and diagnostic pathways which were assumed; these were extensively discussed with local key opinion leaders before the commencement of the costing study [4, 5]. Cytology screening was not considered in the current study, which was consistent with the fact that cytology based screening was not recommended by an International Agency for Research on Cancer (IARC) populations with limited resources [20].

Selection of study sites

Data collection for the current study was conducted in a county hospital and a prefecture hospital in rural Shanxi Province, which is in central north China, from April 2008 to April 2009. Shanxi Province has a total population of 33 million and a geographical area of 156,800 km²[21], and the province had a gross domestic product (GDP) per capita of US$2,975 in 2008 [22]. Several important studies and projects related to cervical cancer prevention have been conducted in Shanxi over the last decade, most of the studies being conducted in collaboration with the Cancer Institute of the Chinese Academy of Medical Sciences (CICAMS) [23].

When choosing the sites for the current costing study, the local situation and health systems were considered. There are several administrative levels operating in China, including those at the province, prefecture, county, township and village levels (the smallest unit). County-level hospitals in rural China would potentially be able to provide centralised cervical screening, and some diagnostic and treatment services. For treatment of patients with diagnostically confirmed cervical intraepithelial neoplasia (CIN) or cervical cancer, most of the county hospitals provide only basic surgical facilities, and more comprehensive treatments are available only in hospitals at a higher level (such as prefecture hospitals). Given the particular situation locally, we assumed that a local screening program would be organized to run through a county hospital, and that women would travel to the county hospital for screening with HPV or visual inspection-based screening. For careHPV screening, an additional option was considered; it was assumed that a mobile screening team based at the county hospital would visit a township or village and that women would perform self-sampling for HPV testing [5]. For all screening strategies, we assumed that women with confirmed abnormalities travelled either to a county hospital or a prefecture hospital for further treatment. Therefore, screening and diagnosis cost data were collected in the Women and Children’s Hospital (county-level) in Xiangyuan County, which is one of 13 counties of Changzhi Prefecture in Shanxi, with a population of 250,000 with 47,750 women aged 30–59 years in 2000 [21]. For the collection of treatment cost data, we extended the survey to an affiliated hospital of the Changzhi Medical College (prefecture-level, located in Changzhi City, the capital city of Changzhi Prefecture, which has a population of 3.16 million [21]).

Overview of costing methodology

Direct medical and non-medical costs related to cervical screening, diagnosis and treatment of cervical high-grade lesions and cervical cancer were collected in this study. Figure 1 shows details of the cost collection process. Direct medical costs associated with clinic visits and laboratory tests were identified using observational field work and extensive consultation with experienced local medical staff. The field workflow commenced with interviews with local clinicians, followed by interview with lab technicians, from which lists of items and the quantities consumed were generated. Costs of all individual items and salary data were then collated, along with the collected quantity data, to generate the final aggregated unit costs. For each individual clinical visit and laboratory test, the following items were included in the costing exercise: consumables (quantity used and unit price), drugs (quantity used and unit price), equipment (quantity used, price, years of useful life, number of cases processed annually), and staff (staff category, working time breakdown). Data sources for direct non-medical cost estimates included structured patient interviews and expert opinion. Each item was costed at 2008 prices, which were collated from financial records at each hospital. At least 15 experts or experienced field staff were interviewed at each hospital; these included clinicians, laboratory technicians, and administrative staff. All cost data were originally calculated in Chinese Yuan (CNY) and presented in US dollars (exchange rate: 1 US$ = 6.8304 CNY; 19 May 2009). For equipment depreciation, we used a discount rate of 3% to derive annual costs [3, 10].

When estimating equipment costs, we assumed that the years of useful life was 7–10 years. We used enhanced annual volume and a discount (depreciation) rate of 3% to obtain the average equipment cost per examination per woman. Usually, the costs relating to use of equipment for colposcopy, biopsy, ECC and other diagnostic procedures were considerably more expensive than the local labour cost. For example, the cost of a colposcope used in the current study was ~ $17,600, but the annual net income per resident in rural China was only $697 in 2008 [24]. Based on these assumptions, if a larger number of women are screened and referred for diagnosis, the average equipment cost allocation for each woman would become lower, while the remaining components (labour and supplies) of the aggregated costs would remain unaltered.

Direct medical costs of cervical screening and diagnosis

In addition to considering of the costs associated with clinical visits and laboratory processes, we also considered programmatic costs including data management, advertisements on local radio, dissemination of leaflets, staff training and quality control, and staff transportation. These costs were allocated on a per-woman basis. For visual screening tests, we assumed that the ongoing quality control and training cost was 23% of other clinical costs, based on the average percentage used in a previous study in five developing countries [7]. A lower percentage for the ongoing quality control/ training cost (10%) was assumed for careHPV, colposcopy, biopsy and, ECC procedures (including histology) because of their assumed higher stability and reproducibility when compared to visual inspection. We assumed all screening programs would be implemented at established hospital sites and therefore did not take into account the cost of real estate or hospital buildings, power, water or other aspects of the general infrastructure.

Based on previous field experience from prior studies, we assumed that a maximum of 60 women could be screened on average each day by a well-trained screening team operating at full workload. This corresponds to a total of 11,475 women screened annually (at 75% of the full workload, over 255 working days). However, in order to account for lower volume screening situations or for operational scale-up of large programs, we also examined the impact of alternate lower volume screening assumptions, and assumed that (i) 2,000 and (ii) 6,000 women were screened annually in a sensitivity analysis. These alternative lower volume assumptions were based on previous experience involving a single screening team working on a part time basis since 2004 to implement a screening demonstration project and the EDETCC of visual inspection based screening which was conducted in Xiangyuan County, Shanxi Province.

In clinical field studies in Shanxi Province, the screening test positivity rate has been reported as 10.7% for visual inspection and 14.6% (in women aged 30–54 years) [25] or 14.5% (in women aged 15–59 years) [26] for HPV testing. Based on these data, we used 14.6% test positivity as broadly representative of field performance in Shanxi for all tests. All women with positive primary screening results were assumed to be referred to further colposcopy examination [25]. Based on field studies we also assumed that 55% [27] and 20% [28] of women undergoing colposcopy would then undergo cervical biopsy and ECC, respectively. Thus, the annual numbers assumed in the base case, for women who received colposcopy, biopsy and ECC in the current analysis were 1,675, 925 and 335, respectively. Staff salary data were obtained from the two study hospitals. The unit cost for the careHPV test in the base case was assumed to be US$5 (which is the targeted price by Qiagen for mass screening programs supported by the public sector in low-resource countries [3]), and which was assumed to include test-specific consumables-such as cervical sampler, storage medium and equipment costs. Other cost items related to careHPV processing, including generic consumables and staff costs, were evaluated separately.

Direct medical costs of treatment of cervical precancer/cancer

We also obtained costs related to treatment for cervical precancerous lesions or invasive cervical cancer. The micro-costing approach was used jointly with a clinical pathway approach in order to reduce missing items. It should be noted that the clinical pathways in this context are all from the local setting. As Figure 1 shows, the gynaecological staff members were interviewed to identify the bed days (if any), procedures and events involved in each treatment modality. They were also interviewed to identify the consumption of clinical supplies/drugs, equipment, staff and time, laboratory tests, and other examinations that occurred in each individual day for each treatment type. Interviews were then extended to medical professionals in other departments of the hospital, including the surgery and anaesthesia units, radiotherapy/chemotherapy, pathology and other examinations (such as electrocardiogram, ultrasonic and chest x-ray facilities). Additional file 1: Table S1 gives further details on the clinical pathways for individual treatment types, including bed days, number of examinations/tests involved, type of anaesthesia, and local post-treatment follow-up strategies.

Direct non-medical costs of cervical screening and diagnosis

For screening and diagnosis, the direct non-medical costs included women’s average time in seeking and receiving care and two-way transportation expenses. For screening conducted at the county hospital, women were assumed to take a public bus to the hospital in which the primary screening was conducted. We calculated the average area of 119 counties in Shanxi Province (1,316 km²), and then used half the radius of the (assumed circular) county area ($\text{Length}=(\sqrt{\text{Country Area}/\pi )}/2$) to estimate the average distance between a woman’s home and the county hospital, which was calculated to be 10 km. For the mobile screening program (involving testing with self-sampling careHPV only), women were assumed to collect the vaginal specimens in a local clinic at their home township or village ( Additional file 1: Table S1); in such a situation, we used half the radius of the average township area of Xiangyuan County (105 km²) to estimate the average distance between place of residence and the township/village clinic, which was calculated to be 3 km. Taking into account the average speed of a local public bus of 20 km per hour, this information was used to estimate a woman’s two-way transportation time. A woman’s waiting time for a public bus was estimated based on expert opinion. Women’s waiting time in the hospital/clinic was estimated by experienced staff in the field. Information on women’s time in examination and their out-of-pocket expense for two-way transportation was obtained from experienced investigators involved in previous local projects and informed by inspection of the ongoing screening process. The annual net income per capita for residents in rural China ($697 in 2008) [24] was applied to convert women’s time into equivalent earnings loss, assuming 255 working days and 8 working hours per day. The final average calculated costs per woman in earnings loss were then $2.73 per day or $0.00586 per minute.

Direct non-medical costs of treatment of cervical precancer/cancer

For estimating the direct non-medical costs related to cervical precancer or cancer treatment, we used categories for costs based on the recommendation by World Health Organization, such as women’s time and transportation [29]; but we also included further categories (such as patients and their carer’s additional out-of-pocket expenses on transportation, accommodation and food) and other out-of-pocket expenses (such as women and their carer’s time in seeking/receiving care, and post-treatment recovery time) [6]. A face-to-face interview was conducted with local trained doctors and nurses, and among a convenience sample of women who attended the 11^th-5 project, and among women who were receiving treatment in county or prefecture hospitals. A total of 108 patients with cervical pre-cancer or cancer were interviewed. Thirty of these women were from county hospitals (12 with CIN2, 17 with CIN3, and 1 with invasive cervical cancer), and 78 were from prefecture hospitals (25 with CIN3, 3 with micro-invasive cancer, and 50 with invasive cervical cancer).

Quality control of the costing exercise

We undertook a range of activities to reduce the chance of missing any items in the costing exercise. Firstly, interviews of local clinicians and technicians were restricted only to experienced and senior staff. Secondly, in order to check the information obtained from the interviews, at least one hospital record for each possible treatment type from the local hospitals was checked by CICAMS researchers to confirm inclusion of all relevant items. Thirdly, the preliminary aggregated data were compared to the local prices for various heath care procedures when they were available [30], and outlier results were given additional attention. In addition, the close-to-final aggregate results were sent to local health economists and clinicians for comments. There were multiple interactions between the CICAMS team and the field staff.

Main outcomes

The aggregated unit costs for each screening and diagnostic test, and for each treatment type (according to the level at which treatment was performed – county or prefecture) were calculated. Component costs were calculated in two ways: firstly, broken down by the costs of clinical visit/hospital care, laboratory and programmatic costs; and secondly broken down by costs of drug/supplies, equipment and staff. Women’s out-of-pocket expenses, the related time spent on seeking/receiving care and post-treatment recovery, and the consequent aggregated direct non-medical costs of screening, diagnosis and treatment were also estimated.

In sensitivity analysis, we particularly focused on the role of screening volume, because this is a parameter of major uncertainty and, a priori, was likely to have a major impact on aggregate costs due to the potential for economies of scale at high volumes. We also included a number of other key parameters (discount rate, the annual number of screened women, screening positivity rate, biopsy/ECC rate, and programmatic costs) in one-way sensitivity analysis.

Ethical approval

The costing study was conducted as one component of a larger government-sponsored cervical screening project (EDETCC), which was approved by the Institutional Review Board of the Cancer Foundation of China.

Direct medical costs of cervical screening and diagnosis

Under the base case assumption that screening was conducted at full volume (11,000 women screening per annum in a county), the aggregated direct medical costs of screening were found to be $2.24, $2.64, $7.49 and $7.95 for VIA only, combined VIA/VILI, careHPV (self-sampling) and careHPV (clinician-sampling), respectively. The direct medical costs of diagnostic procedures were found to be $3.90, $5.76, $4.63 and $10.61 for full colposcopy (with no biopsy), biopsy only, ECC only, and biopsy combined with ECC, respectively. Unsurprisingly, the laboratory costs (including test kit cost) were found to comprise a major component (>60%) of the total aggregated cost of careHPV screening and of the pathology tests (biopsy, ECC) (Table 1). It was also found that the programmatic costs (including data management, advertisements, recruitment and staff training) ranked as the largest component of the aggregated cost of visual inspection screening (Table 1). This is explained by relatively low labour, equipment and consumable costs, leading to a low clinical visit cost for this screening test. More information on the component costs, broken down by clinical visit, laboratory and programmatic costs are presented in Table 1. Figure 2 shows a detailed analysis of the costs broken down according to supplies, equipment and staff time. It was found that clinical and laboratory staff labour costs did not comprise a major proportion of the total cost for most procedures: the exception was visual inspection screening for which 29–36% of the aggregated cost was directly related to labour.

	Clinical visit cost	Laboratory cost	Programmatic cost†	Aggregated cost
VIA only	0.96	0.00	1.28	2.24
Combined VIA/VILI	1.28	0.00	1.36	2.64
careHPV (self-sampling)	0.10	5.75	1.64	7.49
careHPV (clinician-sampling)	0.81	5.75	1.39	7.95
Full colposcopy, without biopsy	2.88	0.00	1.02	3.90
Biopsy only	1.55	3.68	0.53	5.76
ECC only	1.14	3.07	0.42	4.63
Biopsy + ECC	1.90	8.18	0.53	10.61

† The programmatic cost included staff transportation (for careHPV self-sampling strategy only), advertising/ recruitment, training/ quality control, and data management.

Figure 3 presents the findings of the sensitivity analysis for the aggregated costs of screening and diagnostic procedures. The aggregate costs for diagnostic procedures were found to be sensitive to assumptions related to the annual number of screened women. When only 6000 women were assumed to be screened annually, the aggregate costs of colposcopy, biopsy only, ECC only, and ECC combined with biopsy, increased to $2.26 $5.61, $7.66, $6.13 and $14.48 (~1.4 times the base case costs), respectively. These aggregate costs further increased to $12.76, $15.61, $12.41 and $30.70 (~3 times the base case costs), respectively, when the number of screened women was assumed to be to 2,000 per annum. Taking colposcopy as an example to estimate the contribution of equipment costs to the aggregated diagnostic costs, when 11,475, 6,000 or 2,000 women were screened annually, the equipment cost per examination per woman was found to be $1.68, $3.22 or $9.66, which comprised 43.2%, 57.4% or 75.7% of the aggregated costs of the colposcopy examination, respectively. By contrast, the aggregate costs of screening procedures were relatively robust to screening volume (<1% variation when assuming 6,000 screened women per annum, and <5% variation assuming 2,000 screened women per annum). The results for the aggregate costs of both diagnostic and screening procedures were relatively robust to changes in the assumed discount rate, screening positivity rate, biopsy/ECC rate, and programmatic costs (Figure 3).

Direct medical costs of treatment of cervical pre-cancer/cancer

In the county hospital setting, the total aggregated direct medical costs for the three available treatments for cervical pre-cancer/cancer were found to be $61.38, $120.85 and $133.42 for loop electrosurgical excision procedure (LEEP), cold-knife conization (CKC), and simple hysterectomy, respectively. In the prefecture hospital setting, LEEP is not generally performed; and the total direct medical costs were estimated as $280.75, $404.73 and $544.39 for CKC, simple and radical hysterectomy, respectively, and were $281.65, $1,064.05, and $125.46 for radiotherapy, neoadjuvant chemotherapy and adjuvant chemotherapy, respectively. The proportional costs related to laboratory expenses in the prefecture hospital setting (<25% for all procedures) were lower than that in the county hospital (>34%). Table 2 shows further information for the components of the treatment costs, broken down into categories of clinical costs and laboratory costs related to cervical precancer or cancer treatment. The proportional breakdown of drugs, supplies, equipment and staff suggests that, compared to the prefecture hospital, the county hospital spent proportionally less on drugs (<14% vs. >20%), but slightly more on equipment (>28% vs. <23%) (Figure 4a). The costs obtained from the prefecture hospital were generally found to be higher than those incurred in the county hospital, for comparable treatment types, and the differences between were mainly derived from more extensive use of clinical supplies, increased labour costs and drugs in the prefecture hospital (Figure 4b).

		Clinical cost	Laboratory cost	Aggregated cost
A.County level
	Loop electrosurgical excision procedure (LEEP)	16.08	45.30	61.38
	Cold-knife conization	75.55	45.30	120.85
	Simple hysterectomy	88.12	45.30	133.42
B. Prefecture level
	Cold-knife conization	228.01	52.74	280.75
	Simple hysterectomy	332.16	72.57	404.73
	Radical hysterectomy	444.35	100.03	544.39
	Simple radiotherapy	251.90	29.75	281.65
	Neoadjuvant chemotherapy	1,051.74	12.31††	1,064.05
	Adjuvant chemotherapy	93.89	31.56††	125.46

† LEEP treatment was not available in the surveyed prefecture hospital.

†† Related to bloods tests and ultrasound examinations.

Direct non-medical costs of cervical screening and diagnosis

Direct non-medical costs of treatment of cervical pre-cancer/cancer