Impact of radiofrequency ablation for patients with varicose veins on the budget of the German statutory health insurance system

Treatment options for varicose veins can be divided into interventional and non-interventional procedures. Conservative, non-interventional treatments such as compression therapies with medical stockings and pantyhose can be used in every stage of the disease. However, they neither remove the varice nor hinder its development. In contrast, non-interventional treatments aim at alleviating the symptoms [1].

The interventional treatments can be categorized in sclerotherapy, endovenious thermal therapy and surgery. Surgical treatment of varicose veins includes high ligation (crossectomy) and saphenous vein stripping, with or without phlebectomy [20] and has been performed since the early 20th century. Until recently, high ligation and stripping was the standard treatment for patients with varicose veins demonstrating improvements in quality of life as well as reductions in symptoms and reoperation rates compared with high ligation and phlebectomies only [21‐23]. However, various complications can occur following varicose vein surgery. Even though serious complications such as deep vein thrombosis (1 in 200 patients) or pulmonary embolism (1 in 600 patients) are rare [24], minor complications occur frequently. For instance, reported rates of wound complications including infection, haematoma and abscess formations vary from 3-10% [24, 25].

Sclerotherapy techniques aim at inducing endothelial and vein wall damage in a controlled fashion by injecting toxic liquids or foam in the varice. This results in the obliteration of the varicose vein. Sclerotherapy is considered as the gold standard treatment for leg telangiectasias, venulectasias, and reticular veins [26]. A disadvantage is that the treatment of saphenous veins with non-foam sclerosing agents is associated with high recurrent rates [27, 28]. However, recently, treatment with ultra-sound guided foam sclerotherapy (UGFS) has shown better results [29‐31].

Minimally invasive endovenous thermal treatments include endovenous laser ablation (EVLA) and radiofrequency ablations (RFA). The aim of endovenous thermal ablation is the irreversible obliteration of the varice. For the EVLA procedure, a thin laser fiber is inserted in the vein under duplex ultrasound control and heated by laser energy to cause thermal damage of the vein wall. The ultrasound guided RFA technique uses radiofrequency energy which is directed through a small cathether inserted through a tiny insertion in the vein to heat up the varice and damage its vein wall. In different randomized trials, the first-generation RFA device VNUS ClosurePlus and EVLA showed comparable effectiveness to surgery; however, RFA was associated with reduced pain and improved QoL as well as a faster recovery after treatment [32‐42].

In the following section, we will describe the methods used to explore the potential financial consequences of an introduction of ClosureFast in the German SHI.

The basic structure of the model was adapted from a recently published cost-effectiveness model of the National Institute for Health and Clinical Excellence (NICE) [44, 45], illustrated in Figure 1. (For the following description of the model structure see NICE [44, 45].)

The time horizon of the model was five years with a cycle length of one month. Patients enter the model receiving initial interventional treatment for varicose veins (state: “1st treatment episode”). Following the completion of the treatment episode, patients move to the state “1st varices free episode”, in which they do not require further treatment. They remain in this state until a varicose vein recurs. Hence, they transit to the state “physical symptoms of recurrent varicose veins (1)”. A defined proportion of patients receive a 2nd interventional treatment while the others are treated with conservative care in form of a compression therapy until the end of the model’s time horizon (or death). Patients who undergo a second interventional treatment move on to the state “2nd varices free episode”, where they remain until they experience a 2nd clinical recurrence. In case of a 2nd recurrence, patients only receive conservative care until the end of the model’s time horizon (or death). Transition to death is possible from all states of the model. The annual risk of dying is the same in every health state. The monthly all-cause death probability was calculated on the basis of official life tables [46] (Table 1).

To calculate the budget impact, every month, a defined cohort had to enter the model receiving initial interventional treatment according to given market shares. Considering the time horizon of five years (2013–2017), the model included 60 patient cohorts. However, since recurrence rates play an important role in the treatment of recurrent varicose veins (see segment effectiveness) patients with an initial or 2nd interventional treatment of varices before 2013 should not be neglected. Hence, we added another 48 cohorts (four years) to the model and followed-up each of these cohorts for five years (see Figure 1).

German literature data exists only for the annual number of varicose vein surgeries performed in an inpatient setting. Göckeritz estimates that surgeries still account for about 90% of all interventional treatments for varicose veins in Germany [17]. Under the assumptions that the small number of patients treated with endovenous thermal ablations within the SHI can be neglected (expert opinion) and all top-up treatments are UGFS (see Table 1), about 95% of all interventional non-top-up treatments have to be surgical and 5% sclerosing procedures. Furthermore, about 60% of varicose vein surgeries are undertaken in an outpatient setting according to experts’ estimates.

If ClosureFast is introduced in the SHI benefit catalogue, the market shares will shift – in particular at the expense of surgery market shares. On the basis of expert opinion, we assume that ClosureFast accounts for 10% of all procedures in the year of its introduction and has an annual market growth of five percentage points each following year (Base Case assumption). The increase of the use of ClosureFast is assumed to reduce the market share of surgeries equally. The medical experts do not assume a reimbursement of EVLA by the SHI in the near future; therefore, EVLA treatment was not included in the budget impact analysis (Table 2).

In Germany, inpatient varicose vein surgeries are coded as OPS 5–385 within the G-DRG system. Based on the numbers of coded surgical procedures which were derived from the German Federal Statistical Office [47] as well as the above stated expert opinions on market shares, we calculated the size of the monthly patient cohorts. Table 3 shows the different steps of the calculation process in detail.

The effectiveness of the different treatment procedures was modeled via recurrence data. “Success” of interventional treatment was not considered in the model, as patients receive top-up treatments (additional treatment in case of complications, i.e. vein has not been occluded or obliterated) until the treatment episode is complete [44, 45] (NICE).

A recurrence describes varicose veins which occur in the same area of the treated varice, independently of the type of the initial treatment. Noppeney et al. distinguish between three causes of recurrent varices: technical failure of the initial treatment, neovascularisation and progression of the venous disease [55]. About 20% of varicose surgeries are attributed to recurrences [49‐53]. Differences in the initial treatment, the method of measuring recurrences and duration of follow-up make a comparison of recurrence rates difficult [56]. After surgical treatment of varicose veins and a follow-up period of three to eleven years, recurrence rates of 26-62% have been reported [57, 58]. In addition, data indicates that the rate of recurrences increases over time [59].

To gather data on recurrence rates as well as adverse events of all relevant treatment options for varicose veins, we performed a structured literature search in 44 databases using the database search tool of the German Institute of Medical Documentation und Information (DIMDI). Additionally, we conducted a hand research. The search strategy targeted meta-analyses of RCTs for surgery and UGFS as well as RCTs for ClosureFast.

The systematic literature review identified five meta-analyses [32, 44, 60‐62] which compared recurrence data as well as adverse events of RFA and surgery. Two of these studies [32, 44] included a comparison between foam sclerotherapy and surgery. Only one RCT was found which reported recurrence rates as well as adverse events of ClosureFast in comparison with surgery and UGFS [18]. The study was also included in two of the identified meta-analyses [44, 62]. In total, six RCTs [18, 36, 63‐66] presenting relevant data on recurrence rates have been identified via the five meta-analyses. Table 7 of the Appendix presents an overview of the reported recurrence rates. Only Belcora et al. found a statistically significant difference in recurrences [63].

In the Base Case analysis, we used the effectiveness data reported by NICE. Using a network meta-analysis approach, the authors estimated a one-month recurrence probability of 0.008331 for surgery, 0.005833 for endovenous thermal ablations and 0.009141 for UGFS. However, only a combined probability for RFA and EVLA was calculated [44]. Since none of the other identified RCTs and meta-analyses found a statistically significant difference between RFA and surgery, we made the conservative assumption that ClosureFast had the same effectiveness as surgery.

Adverse events were not included in the analysis. The complications reported in different trials varied a lot and different measurement methods were used. According to a multidisciplinary Guideline Development Group, which currently develops the NICE guidelines for the diagnosis and management of varicose veins, adverse event profiles of the different interventions are similar to the extent that they can be neglected in health economic models [44, 45].

Resource and cost data regarding the treatment of varicose veins as well as the resulting side effects are only available in the literature to a limited extent. Therefore, we have identified relevant treatment procedures and resource usage via expert opinion and official handbooks. Hence, we have evaluated the resource use from the perspective of the SHI in Germany, taking also into consideration patient co-payments as well as discounts for medications given by the manufacturer and pharmacies as required by legal obligations in Germany. For this, the current German recommendations [67, 68] for the valuation of resource usage were applied to evaluate the costs from the perspective of the SHI. In the Base Case analysis, we assumed that costs of interventional treatments in 2012 and costs of compression therapy in 2011 (no newer data currently available) would stay stable over time. Considering the increasing costs of inpatient treatment of varicose veins in recent years, this is a conservative assumption, since rising prices of inpatient procedures would have a greater effect on overall surgery costs due to its higher inpatient ratio compared with ClosureFast. However, price assumptions were varied in sensitivity analyses, e.g. by using the mean annual price inflation rate of varice treatments between 2005 and 2012, in order to take the uncertainty regarding future price developments into account. The historic development of surgery and UGFS prices are shown in Table 8 of the Appendix.

In Germany, inpatient stays are reimbursed via lump sums (diagnosis related groups, DRG). In these payments, in general all expenses of the hospital, incl. medication costs, are included. Relevant DRG were identified via the German DRG-Catalogue [69] using the official German DRG Definition Handbook from 2012 [70] and evaluated from the perspective of the German SHI [67‐70].

Regarding the outpatient setting, we applied the official German Uniform Valuation Scheme (EBM) [71] and identified relevant OPS-codes for outpatient operations for surgery as well as physician visits. Again, the above mentioned recommendations for the evaluation were used [67, 68]. For general outpatient physician visits of SHI-insured persons, it is important to mention that the doctors are reimbursed via lump sum payments per quarter, independent from the number of visits of an individual patient per quarter.

The proportion of patients treated in an outpatient and inpatient setting, respectively, was identified via expert opinion, since no official figures or literature data was available. Due to the uncertainty of the proportions, extensive sensitivity analyses were performed.

Table 4 summarizes the costs and effectiveness as well as the setting of all interventional treatments for varicose veins which were included in the BIA.

Various key inputs of the model are based on expert opinions (e.g. outpatient ratio of treatments) or on limited evidence (e.g. recurrence rates of interventional treatments for varicose veins) which may result in high uncertainty of model outcomes. To test the robustness of the model, we conducted three types of sensitivity analyses. First, we performed one-way sensitivity analyses to assess the impact of variations in values of every input parameter on the results. In order to test how the results react to a simultaneous variation of several inputs, we assumed four reasonable alternative scenarios of which two scenarios (ClosureFast + and ClosureFast++) were based on input values more likely to favor a “World with ClosureFast” and two other scenarios based on input values more likely to favor a “World without ClosureFast” (ClosureFast- and ClosureFast--). The alternative scenarios are summarized and described in Table 9 of the Appendix. Furthermore, we calculated the budget impact of the Base Case and the 4 alternative scenarios for three different market uptakes of ClosureFast:

Finally, uncertainty was assessed using a probabilistic sensitivity analysis (PSA). The PSA was carried out as a Monte Carlo simulation with 10,000 iterations, simultaneously drawing random numbers for most model inputs from the distributions listed in Table 10 of the Appendix.

The discount rate was set to 0% in all investigated scenarios (Base Case as well as sensitivity analyses), since discounting is – in contrast to cost-effectiveness models – no typical procedure in BIA [73].

According to the results of the Base Case, the introduction of ClosureFast for the treatment of varicose veins saves costs of about 19.1 Mio. € over a time period of five years in Germany. Detailed results are shown in Table 5. Since we assumed equal effectiveness between ClosureFast and surgery, the inpatient and outpatient treatment costs as well as the proportion of inpatient procedures of ClosureFast and surgery, respectively, are the factors which affected the results the most. While the higher outpatient treatment costs of ClosureFast” increases total costs on the one hand, the smaller proportion of inpatient procedures reduces overall costs on the other hand. All in all, the substitution of outpatient for inpatient treatments overcompensate the higher costs of ClosureFast in the outpatient setting. Over a time horizon of five years, a total of 1,623,749 interventional treatments for varicose veins are performed in a “World without ClosureFast”. Of these, about 38% are inpatient procedures causing about 70% of total interventional treatment costs. In a “World with ClosureFast”, according to the results, about 32% of interventional treatments for varicose veins are performed in an inpatient setting causing about 60% of the total interventional treatment costs.

Results of the one-way sensitivity analyses show that the model outcome is highly sensitive to variations in interventional treatment prices (inpatient as well as outpatient) and the proportion of outpatient treatments. For instance, a 10% increase in the ClosureFast outpatient price leads to additional costs of about 32 Mio € in a “World with ClosureFast”. On the other hand, a 10% price reduction of the ClosureFast outpatient treatment results in additional savings of the same amount. Variations in the values of other input parameters (monthly probability of recurrence, probability of requiring top-up treatment and probability of receiving a 2nd interventional treatment) only had minor effects on the model outcome. Results of all performed one-way sensitivity analyses are shown in Table 11 of the Appendix.

The results of the scenario analyses are presented in Table 6. As one might expect, the introduction of ClosureFast saves more costs in the scenarios with input values favoring a “World with ClosureFast” compared with the Base Case. However, in the two other scenarios (ClosureFast-, ClosureFast--), the introduction of ClosureFast is not cost saving anymore and causes up to 300 Mio € of additional costs for the SHI in five years. A variation of the market uptake does not change the results qualitatively; solely the total costs as well as the cost difference vary between a “World with ClosureFast” and a “World without ClosureFast”.

According to the results of the PSA, illustrated in Figure 2, a “World with ClosureFast” saves costs with a probability of about 59% over a time horizon of five years. With a probability of 25% cost savings of over 75 Mio € can be realized in a “World with ClosureFast”. However, with the same probability (25%) the introduction of ClosureFast is likely to cause additional costs of at least 37 Mio € in the SHI. All in all, it has to be noted that there is high uncertainty regarding the results, which reflects the limited evidence of several key input parameters.