Randomized healthservices study of human papillomavirus-based management of low-grade cytological abnormalities†
The study was primarily designed by the principal investigators (Drs. L. Dillner, J. Dillner and Törnberg), with input from all authors. Dr. Kemetli maintained the database and analysed the data under supervision of Dr. Törnberg. Drs. Elfgren, S. Andersson, Persson and Rylander were responsible for the clinical management of the patients. Drs. Bogdanovic and P. Andersson performed the HPV testing, under the supervision of Dr. Grillner. Ms. Carlsten-Thor assisted with study coordination. Drs. L. Dillner and J. Dillner primarily drafted the manuscript, to which all authors contributed. Dr L. Dillner, as the corresponding author, had full access to all the data in the study and had final responsibility for the decision to submit the manuscript for publication. The final version of the manuscript was seen and approved by all authors.
Human papillomavirus (HPV)-based management of women with borderline atypical squamous cells of undetermined significance (ASCUS) or mildly abnormal cervical intraepithelial neoplasia (CINI) cervical cytology has been extensively studied in the research setting. We wished to assess safety and health care resource use of a real-life health care policy using HPV triaging. All 15 outpatient clinics involved in the organized population-based screening program in Stockholm, Sweden screening program were randomized to either continue with prior policy (colposcopy of all women with ASCUS/CINI) or to implement a policy with HPV triaging and colposcopy only of HPV-positive women. The trial enrolled the 3,319 women who were diagnosed with ASCUS (n = 1,335) or CINI (n = 1,984) in Stockholm during 17th March 2003 to 16th January 2006. Detection of high-grade cervical lesions (CINII+) and health care cost consumption was studied by registry linkages. The proportion of histopathology-verified CINII+ was similar for the two policies (395 of 1,752 women (22.5%; 95% Confidence interval [CI]: 20.6–24.6%) had CINII+ diagnosed with HPV triaging policy, 318 of 1,567 women (20.3%; 95%CI: 18.3–22.4%) had CINII+ with colposcopy policy). Sixty-four percent of women with ASCUS and 77% of women with CINI were HPV positive. HPV-positivity was age-dependent, with 81% of women below 35 years of age and 44% of women above 45 years of age testing HPV-positive. HPV triaging was cost-effective only above 35 years of age. In conclusion, a real-life randomized healthservices study of HPV triaging of women with ASCUS/CINI demonstrated similar detection of CINII+ as colposcopy of all women.
Cytological screening identifies women with cervical lesions that confer an increased risk of cervical cancer.1 Further assessment and/or treatment will prevent progression to invasive disease. Women with high-grade cytological lesions should be referred directly for further exploration.2 However, several different policies for management of women with minor cytological lesions exist.2, 3 Follow-up policies after atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intra-epithelial lesions (LSIL) vary from conservative repeat cytology4–6 to direct referral for colposcopy and biopsy.7, 8 Referral to the colposcopy clinic with subsequent biopsy and endocervical sampling for histopathological examination results in substantial costs for the health care system9 and often creates feelings of anxiety and discomfort for the women concerned.10, 11
Low-grade cervical lesions commonly regress and do not require treatment.12 Although most women with an ASCUS smear result do not have significant disease, a substantial proportion of them do harbour histopathologically confirmed high-grade cervical intraepithelial neoplasia (CINII+) lesions.13–15 For a population of women participating in screening in the United States, it was estimated that more than one third of CINII+ lesions were discovered on follow-up of a previous smear with ASCUS.15 Referral of all women with minor cytological lesions for colposcopy is a common approach in many settings in the United States,3 is a recommended practise in the European Union4 and has been the routine policy used in Stockholm, Sweden for many years.
Given the overwhelming accumulated evidence concerning the etiological role of oncogenic human papillomavirus (HPV) infections in the development of cervical cancer and CINII+,4 HPV testing has been extensively studies as a triage method to select women at increased risk of cervical cancer, thus justifying referral for colposcopic exploration.13, 14, 16–19 Systematic reviews of research studies have clearly documented that HPV triaging is superior to repeat cytology in follow-up of ASCUS smears20 and that HPV triaging is equivalent to repeat cytology in follow-up of LSIL smears.20
A major 3-way randomized trial comparing HPV triaging, repeat cytology and colposcopy of all women ASCUS-LSIL Triage Study (ALTS) found that repeat cytology was inferior for managing ASCUS smears in women <30 years of age and that HPV triaging and colposcopy of all women were equivalent in terms of safety.21 HPV triaging is recommended policy for triaging of ASCUS, but not LSIL, smears in both Europe and the United States.22, 23
Generalizing data from the research setting to the real-life screening policy situation may be difficult. For example, diagnostic practises will vary between countries24 as well as between the routine setting and the expert panels used for diagnosis in the research setting. There may be differences in the adherence to a study protocol in the research setting and to the recommended policy in the routine setting. Finally, the research setting does not study neither the acceptability of a policy to the health care provider (payer) nor the acceptability to the entire general population targeted by a screening program.
Randomized healthservices studies are a method for evaluating and comparing the real-life effect of different health care policies in the routine setting. The design is not yet commonly used, except in public health and in evaluation of the effect of screening policies.25, 26
We wished to compare the safety and real-life use of health care resources by the two different management policies that have in previous research not been found to be inferior, namely colposcopy of all women and HPV triaging with referral only of HPV-positive women.
Material and Methods
The organized Swedish cervical cancer screening program invites women aged 23–49 years for screening at 3-yearly intervals and women aged 50–60 years at 5-yearly intervals. Invitations are based on the population register, but comparison is first made with cytology registers covering the population to exclude women who have recently taken a smear outside the program. All women resident in the Stockholm County that on their organized, invitational smear had the cytological diagnosis ASCUS or CINI between 17th March 2003 and 16th January 2006 were included in the study. Sweden uses the old U.S. cytologic nomenclature,27 in which CINI is also a cytological, not only a histopathological diagnosis. The new cytological term LSIL is a combination of LSIL and koilocytotic atypia. In the system by Koss, a diagnosis of koilocytosis can be made in addition to any other diagnosis. As only 0.2% of smears are reported as normal, with koilocytosis as an ancillary diagnosis in Sweden, the cytological diagnosis of CINI used in Sweden is essentially similar to the LSIL nomenclature used in other countries.
The two policies compared were (1) Colposcopy of all women, that referred all women with ASCUS or CIN1 for colposcopy and biopsy (previous policy) and (2) HPV-based triaging, referring all women with ASCUS or CINI for a new visit with HPV-test (new policy), using the FDA-approved Hybrid Capture 2 test (DiGene, Gaithersburg, MD). Sampling was made using the DiGene specimen collection kit, following the instructions of the manufacturer. The HPV test was performed in a certified virus diagnostics laboratory, following the instructions of the manufacturer. HPV-positive women were referred for colposcopy and biopsy, whereas HPV-negative women were referred for a repeat cytology visit 12 months later. The repeat cytology among HPV-negative women with ASCUS/CINI has not been used in some of the major HPV triaging trials,28 but an extra safety measure was considered advisable when going from the more carefully controlled research setting to the routine setting. For both the old and the new policy, all women who had been referred to colposcopy were also scheduled for two follow-up visits with cytology, 12 and 24 months after the colposcopy. These follow-up smears were scheduled also if the colposcopy had been normal and no biopsies were taken.
The objectives of the randomized healthservices study were to assess whether the new policy was equally safe as the old policy and to determine whether it was cost-efficient. The primary outcome measure of safety compared between the two policies was the CIN2+ detection rate. The secondary outcome measure was the use of health care resources.
The two policies were submitted for consideration to the cervical screening expert group of the Swedish Association for Obstetrics & Gynecology, to the Obstetrics & Gynecology expert group of the Stockholm County and to the heads of the Obstetrics & Gynecology departments in all 15 ObGyn clinics in Stockholm County. After the assertion by these expert groups and officials that the two policies were both in accordance with evidence-based medicine, the randomized health care policy evaluation plan was approved by the Ethical Review Board in Stockholm. As mandated by the Ethical Review Board, information about the policy switch, the randomization and the follow-up was included as a separate information leaflet that was given to all women who participated in the organized cervical screening in the Stockholm County (about 180,000 leaflets).
All 15 ObGyn clinics in Stockholm County were randomized to colposcopy of all women (old policy) or to HPV triaging (new policy). For 14 clinics there was a pairwise block randomization, where the clinics were paired according to (i) the size of the catchment area population (to increase the probability that the two policies would include a similar number of women) and (ii) the CINII+ incidence in the catchment area population (low variance in the incidence of the outcome increases the statistical power of cluster-randomized trials).29, 30 The 15th and smallest clinic was randomized without a paired clinic. Randomization was performed using computer-based random numbers by an independent entity (the Cancer Registry of Stockholm) and the eight clinics randomized to HPV triaging all switched policy on 2003-03-17.
All women were followed up until 2007-12-10 by linkages with the regional cytology and pathology registries using the unique personal identification numbers of the women. The registries contain data on all smears and also biopsies taken in the greater Stockholm area, not only those taken in the organized program. When several histopathology reports had been issued for specimens taken on the same date, they assumed to result from the same colposcopy visit. The reasons why more than one histopathology report had been issued were not investigated, but the most likely explanation is that several different types of tissues specimens had been taken.
The hybrid capture II test (DiGene Inc, Gathersburg, MD) was performed as recommended by the manufacturer. The samples were taken using the DiGene specimen collection kit. After denaturation, the sample was hybridized to the high-risk RNA probe cocktail, recognising HPV-genotypes 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68. Samples were scored as positive when the FDA-recommended cutoff value for test-positive results was equivalent to or exceeded 1.0 RLU (Relative Light Unit: equivalent to 1 pg HPV DNA per 1 ml of sampling buffer). For each test, the ratio of low/high calibrator probe results was calculated, and the percentage of borderline results (0.7–2.0 RLU) was assessed. If the ratio was below 2.5 and/or the percentage of borderline results above 10%, the test was repeated. Quality controls also included, for each 96 well plate, two positive controls of 104 and 105 HPV16-positive SiHa cervical carcinoma cells (containing one HPV genome copy per cell) and one negative control of 104 HPV-negative C33A cervical carcinoma cells.
Health care costs
The actual costs paid by the health care provider to clinics and laboratories were used. The county council of the region is the provider of health care. Hospitals in the region are funded by the amount of health care delivered (e.g., the number of colposcopies performed by Ob/Gyn clinics and the number of histopathology reports issued by pathology laboratories). The price lists used in the region at the time of study are detailed in Tables 3 and 4. The costs of ethical review board permission, information leaflets and for study management are not included, as these are considered to be part of the budgeted costs for evaluation and quality assurance of the screening program. Provided equal safety, the policy that had lower costs was considered to be most cost-efficient.
The trial was powered to detect a 2-fold difference in detection of CIN2+, as estimated using n = 15.8*(S2c + S2w/m)/d2, where S2c is interclinic variance, S2w is intraclinic variance, m is number of women per clinic and d2is the magnitude of the detectable difference.29 As clusters were large, matching was broken in analysis.30 Analyses were by “intention to treat,” which means that costs and CINII+ detection rates are based on all women (both attending and nonattending). All women could be followed up with registry linkages for the entire duration of the study.
Proportions were compared with Fischer's exact test and Clopper Pearson 95% confidence intervals (CI) for binominal proportions were calculated using SPSS (Chicago, IL). Testing with Cochrane-Armitage trend test used SAS (Cary, NC). As the point of the intervention was to reduce the number of follow-up tests required, we considered women as having been followed up until the last date of the registry linkage rather than to the last follow-up test.
There were 1,752 women with ASCUS/CINI in the catchment populations of the HPV triaging clinics, compared with 1,567 women belonging to the all colposcopy clinics, corresponding to an imbalance in number of women of 12% (Fig. 1). Although all women managed by the HPV triaging policy were referred to HPV testing, 152 women (8.7%) (62 women with an ASCUS smear and 90 women with a CINI smear) were never tested, because of relocation, logistic failure or for unknown reasons.
Altogether 72% of the women were HPV-positive, with HPV-positivity being more common among women with CINI (77%) than ASCUS (64%) (p < 0.0000001; Table 1). HPV-positivity was strongly age-dependent, both among women with CINI and among women with ASCUS (p for trend <0.0001; Table 1).
Table 1. HPV positivity in relation to age
During the follow-up until 2007-12-10, 713 women had a histopathological diagnosis of CINII+ (Table 2). The proportion of women being diagnosed with CINII+ was similar in both populations, with a slight tendency for a higher proportion in the HPV triaging population (22.5%, 95% CI: 20.6–24.6%) than in the all colposcopy population (20.3%, 95% CI: 18.3–22.4%) (odds ratio [OR]:1.14; 95% CI: 0.96–1.36) p > 0.1 (Table 2), despite the fact that only the women testing HPV-positive (66% of the women in the HPV triaging population) were referred to colposcopy and biopsy (Fig. 1). The attendance rate among women referred to colposcopy was higher in the HPV triaging arm (91%) than in the all colposcopy arm (81%) (Fig. 1), and the actual proportion of women who had (at least 1) colposcopy with biopsy in the HPV triaging arm was therefore 60% (Fig. 1). The proportions of histopathologically diagnosed CINII+ were similar for both polices also when considering whether the index smear had ASCUS or CINI (ASCUS: 21.2% (95%CI: 18.3–24.4%) using all colposcopy and 21.8% (95% CI 18.6–25.3%) using HPV triaging. CINI: 19.5% (95% CI: 16.9–22.4%) using all colposcpy and 22.9% (95% CI:20.5–25.5% using HPV triaging)) (Table 2).
Table 2. Number of women with CIN II+ diagnoses in histopathology after ASCUS/CIN I in cytology in relation to the health care policy used
In the HPV triaging arm, there were 27 women who were diagnosed with CINII+ in histopathology in spite of not having been referred because of HPV-positivity (Table 2). Sixteen of these women were among the 152 women where the HPV test was never performed. Most of these women had been directly referred to colposcopy. The extra cytology visit scheduled 1 year later for women with HPV-negative ASCUS/CINI was only attended by 243 of 443 (55%) of the HPV-negative women (Fig. 1). The controls smears for these 243 women found 13 ASCUS, 8 CIN1 and 3 CINII+ cytologies. Six CINII+ (two CINIII and four CINII) histopathological diagnoses were made as a result of the extra cytology visit. Three of these women with CINII in histopathology had benign smears at the control smear, suggesting that their extra cytology visit had included a colposcopy. Four women had a follow-up visit already within 6 months of the index smear (not recommended in the policy). Finally, one HPV-negative woman did not attend the extra cytology visit, but came for organised screening 3 years later, when a second ASCUS smear resulted in CINIII in histopathology.
The total numbers of colposcopies with biopsies that were actually performed was only marginally lower in the triaging clinics (1,917) than in the all colposcopy clinics (1,967) (Table 3). As there were 12% more women in the triaging arm, (1,752 women in the triaging arm, compared to 1,567 in the colposcopy arm), this corresponds to a 13% decrease of number of colposcopies with biopsy (1,917 colposcopies for 1,752 women compared to 1,967 colposcopies for 1,567 women). The total number of cervical histopathological reports that was paid for was actually somewhat higher in the HPV triaging arm (Table 3), corresponding to a mere 5% decrease in histopathologies (2,766 histopahtologies for 1,752 women, compared to 2,596 histopathologies for 1,567 women). The total actual costs increased by 21% in the HPV triaging arm (from 0.63 million US Dollars to 0.77 million US Dollars) (Table 3). The cost in the triaging arm is expected to be lower in the future, because the lower number of women referred to colposcopy results in a reduced amount of women scheduled for follow-up. All women referred to colposcopy are also scheduled for two follow-up visits with cytology, 12 and 24 months after the colposcopy (in both arms). These follow-up smears are scheduled also if the colposcopy had been normal and no biopsies were taken. The number of women scheduled for follow-up was reduced with 32% in the triaging arm (1,967 colposcopies, each one eliciting follow-up smears, in the colposcopy arm compared with the adjusted number of colposcopies in the triaging arm [1,712]) (Table 3). However, after considering the future follow-up costs also, the HPV triaging arm remained more expensive (Table 3).
Table 3. Costs incurred to date in SEK (10 SEK = 1 Euro)
As the HPV prevalence was strongly age-related, the proportion of women referred for colposcopy is reduced with increasing age in the HPV triaging policy, which reduces the costs of the policy with HPV triaging with increasing age. The break-even where the costs of the two policies are approximately the same occurs for a policy with HPV triaging only for women above or equal to 35 years of age, as detailed in Table 4. The CINII+ detection rates were somewhat lower among women above or equal to 35 years of age (232 of 1231 women [18.8%] compared with 481 of ,2088 women below 35 years of age [23.0%]), but the proportion of women diagnosed with CINII+ was similar for the two policies also in this age group (HPV triaging: 123 of 629 women (19.6%); All colposcopy: 109 of 602 women (18.1%); OR: 1.10, 95% CI: 0.82–1.48; p > 0.5).
Table 4. Costs incurred to date in SEK (10 SEK = 1 Euro) women >34 years
This study has provided a real-life evaluation on the actual costs used when switching to a policy with HPV-based triaging of low-grade cervical abnormalities and found that a policy using colposcopy of all women with ASCUS/CINI below 35 years of age, but HPV triaging above 35 years of age was the least costly policy in practise. Furthermore, the relative safety of the two policies appeared to be similar, as there was an equal proportion of CINII+ diagnoses made for the two policies.
Several of the theoretical concerns that had motivated this study found support in the actual data obtained. The cytological diagnostic practises vary between countries, making it difficult to apply the results from studies performed in other countries to our population. Specifically, a formal exchange of slides found that most of the slides diagnosed as ASCUS in the USA and UK are reported as normal in Sweden.24 Indeed, the proportion of women with ASCUS that were HPV-positive was considerably higher in this study than in most reports from USA or the UK, which resulted in a relatively lower cost-efficiency of HPV triaging. However, our major concerns were related to the differences between the research study setting and the real-life health care setting. The research setting does not include the entire population, only a select subset of participating subjects. Furthermore, the support, dedication and resources invested in a research study enable close monitoring of adherence to a study protocol and extraordinary quality control procedures (e.g., re-reviewing of cytological and histopathological slides by several senior experts) that are not possible in the routine setting, where guidelines and policy documents can be issued but where the management of the patient is decided by each physician. Real-life acceptability of a policy may be quite different both for the medical personnel, for the patients as well as for the health care provider that pays the costs in the health care setting but not in the research setting. Furthermore, as most research studies are blinded, the effects that the information of the test results may induce are not included in the overall effects measured.
The randomized HPV triaging trial ALTS found that HPV-triaging of LSIL smears was not cost-efficient as virtually all LSIL smears were HPV-positive if the diagnosis had been confirmed by the expert panel re-review.21 However, as senior expert panel re-review of all smears is not possible in routine practise, it is rather the proportion of HPV-positive LSIL smears in routine diagnosis that is of relevance. Furthermore, previous research studies of HPV triaging of LSIL have either been focused on younger women or not reported age-specific data for LSIL. As it is well known that there are lower HPV prevalences in LSIL, with increasing age, we felt that it would be necessary to perform a real-life evaluation of HPV triaging of both ASCUS and CINI smears. (LSIL terminology includes CIN1 and koilocytosis, but koilocytosis in otherwise normal smears is an exceedingly rare diagnosis in Sweden (0.2% of smears) and is classified as normal cytology). Indeed, HPV prevalences in CINI smears were lower in our real-life study, using routine cytology, than in most studies in the research setting that use the expert re-review cytology.
The break-even for lower costs of the HPV triaging policy was found to be at an HPV prevalence of about 70%, which would correspond to >35 years of age for ASCUS and >40 years of age for CINI—an essentially negligible difference in assigning the optimal cutoff for use of HPV-triaging.
Our concern about whether the health care system would be adhering to policy guidelines or whether new guidelines might induce new clinical management behaviour also found some support: Even though 32% fewer women were referred with the HPV triaging policy, the number of colposcopies was reduced by only 13% and the number of histopathologies by only 5%. Possible reasons for this effect include those induced by the fact that the clinics and the women knew that the smear was HPV-positive, for example, more aggressive efforts to contact women known to also be HPV-positive, increased likelihood of women to attend when they know they are HPV-positive and more careful colposcopical examination with increased likelihood of biopsy-taking. There may also be effects induced by the fact that the patient burden was reduced for the clinics in the HPV triaging arm, for example, shorter waiting time between diagnosis and appointment (likely to improve attendance rate) and more time available per referred woman for the colposcopists. The improved attendance rate seen with HPV triaging is relevant to the issue of safety and might be related to the nonsignificant tendency for HPV triaging to increase the CINII+ detection rate, in spite of the fact that a lower proportion of women were referred.
The HPV triaging policy evaluated included an extra “safety” smear 1 year after having tested HPV-negative. Some studies in the research setting have used safety smears in HPV-negative women21 but other studies have not found this to be required.28 We were concerned that widespread introduction of a new test in clinical routine may initially have lower quality (e.g., in sampling or sample handling) than a carefully controlled research trial. The decision to include an extra safety smear was preceded by some debate and incomplete acceptance in the medical community of the necessity of this extra smear is a possible explanation for the low attendance rate for this smear. However, the present data finding 6 of 243 women (2.7%) with CINII+ in this group (about three times excess yield compared with general population screening) argues that this extra smear is indeed required in the routine setting.
A major cause of increased cost in the HPV triaging arm was the fact that the HPV test was taken at an extra visit to a physician. Reflex HPV testing of the actual index smear sample was not possible as Stockholm was not using liquid-based cytology (LBC), because of high costs and an uncertain evidence base on its advantage.31 However, since the start of our study, LBC has become more affordable and a benefit has been found in some studies.32 A switch to LBC would greatly improve on the cost-efficiency of HPV-based triaging, by obviating the extra visit for an HPV test.
The decision to have the HPV-test taken by a physician was motivated in part by a perceived need to address questions from the women about HPV, the meaning of the HPV test positivity and about the study. However, the feed-back from participating clinics has indicated that this was not necessary and that HPV test taking by midwives would have been sufficient, which would have reduced costs.
Strengths of the study include the fact that it is not merely generalizable to a real-life setting- it is the real-life setting. The only major deviation from routine health care was the existence of an ethical review board permission for the randomization of the policies. There was no nonparticipation in the study (all women with ASCUS/CINI in cytology were included in the follow-up) and no research funds were used. By comparison, the British trial of management of borderline and low-grade abnormal smears (TOMBOLA) is a trial of a select cohort of 4,474 women33 (participation rate 52%) is funded by a research grant of several million pounds.
Another major strength is the fact that the follow-up of all actual smear-taking and biopsy-taking in the two arms was essentially complete for the entire population. Registration of all smears and histopathologies, based on unique personal identifiers, is already a part of the organised screening program. The required resources in coordination, database management et cetera were already available and a part of the required duties of the office of the organised program to follow-up on the performance of the program.
The major weakness of our study is the fact that randomization involved only 15 clusters, resulting in imbalances in study size and likely imbalances in local clinical and diagnostic practises. Studies in the research setting with individual randomization would have eliminated these weaknesses, but would have introduced a series of more severe problems, such as not being population-based and not being generalizable.
The economic analysis in this study is focussed on actual costs, disregarding discounting. Actually paid costs were preferred as the two policies compared were running concurrently. The future follow-up is scheduled for 2 years, which would have discounted the follow-up costs for these smears with 1 year or 2 years, respectively. As the future projected costs were <10% of all costs, discounting of future costs would have had negligible effects on total costs. Therefore, we preferred to present the actual figures without manipulation, for maximal transparency.
In summary, this study has demonstrated that carefully controlled and evaluated implementation of routine preventive services can contribute significantly to the scientific knowledge base. There are improvements particularly in generalizability, but the large size of the study also enabled conclusions to be drawn that would have been unreasonably expensive/impossible to answer with studies in the research setting. In particular, the evidence that HPV-based triaging of women with CINI in cytology is cost-effective above 40 years of age has not been shown before, presumably because of small numbers and/or studies not being population-based. We found that HPV-based triaging was safe, in terms of equal CINII+ detection rates, for all ages and for both ASCUS and CIN1. With the current costing, the HPV-based triaging was less expensive only for women older than 35 years of age. However, the demonstrated safety of HPV-based triaging could support HPV-based triaging also in younger ages, provided that the costs change (e.g., if reflex testing on LBC samples becomes possible and/or the price of the HPV test is reduced).
We thank Tiina Malmström for database assistance. We also thank our enthusiastic contact persons at all the different hospitals and clinics in the Stockholm county especially: Ilona Barnard, Agneta Fredriksson, Meseret Mengisto, Ricardo Munoz, Carl Preastiin, Stanislav Rodau and Gunilla Torpare. The study had no research grants or commercial sponsorships. The HPV tests were paid at ordinary market value by the regional health care provider.