Electroacupuncture with Usual Care for Patients with Non-Acute Pain after Back Surgery: Cost-Effectiveness Analysis Alongside a Randomized Controlled Trial

: Electroacupuncture (EA) is used to treat pain after back surgery. Although this treatment is covered by national health insurance in Korea, evidence supporting its cost-e ﬀ ectiveness and contribution to the sustainability of the national health care system has yet to be published. Therefore, an economic evaluation, alongside a clinical trial, was conducted to estimate the cost-e ﬀ ectiveness of EA and usual care (UC) versus UC alone to treat non-acute low back pain (LBP). In total, 108 patients were recruited and randomly assigned to treatment groups; 106 were included in the ﬁnal cost utility analysis. The incremental cost-e ﬀ ectiveness ratio of EA plus UC was estimated as 7,048,602 Korean Rate Won (KRW) per quality-adjusted life years (QALYs) from the societal perspective (SP). If the national threshold was KRW 30 million per QALY, the cost-e ﬀ ectiveness probability of EA plus UC was an estimated 85.9%; and, if the national threshold was over KRW 42,496,372 per QALY, the cost-e ﬀ ectiveness probability would be over 95% percent statistical signiﬁcance. Based on these results, EA plus UC combination therapy for patients with non-acute LBP may be cost-e ﬀ ective from a societal perspective in Korea. analysis in the plane showed the mean values (KRW 704,861 per QALY) (KRW 6,957,703 and KRW 42,496,372) of and also showed the mean values, 95% CI, and cost e ﬀ ective probabilities according to changes in the national threshold. These results show that the ICER of EA plus UC was as KRW 7,048,602 from the perspective, and, if the national threshold was KRW 30,000,000 per QALY, the cost-e ﬀ ectiveness probability of EA plus UC was an estimated 85.9%. Moreover, if the national threshold KRW 42,496,372 per QALY, the EA combination treatment will cost-e ﬀ ective signiﬁcance. To our trials


Introduction
Increasing health care expenditure within the budget constraints caused by an aging population demands the examination of not only the efficacy and effectiveness of treatment, but also the cost-effectiveness of treatment alternatives that support the sustainability of the national health care system [1,2]. This trend impacts the use of therapies in the area of complementary and alternative medicine, which have been commonly employed to treat various diseases, and are already approved for national health insurance reimbursement in Korea [3]. Low back pain (LBP) is a common disease treated by complementary and alternative therapies, and more than 70% of people in industrialized countries experience LBP at some time in life, causing a substantial societal economic burden [4]. The global burden of LBP was ranked as sixth in terms of overall burden measured by disease-adjusted patient. In addition, disease codes were unified under the Korean classification of disease, which is sourced from the international classification of disease; before this point, Korean medicine doctors had been using different diagnostic systems. Despite these circumstances, many questions remain regarding the clinical effectiveness and cost-effectiveness of combination treatments via collaboration between doctors of various specialties [30]. These unanswered questions are a major barrier to the adjustment and expansion of available treatment options under the national healthcare system. In this context, the economic evaluation of the combination treatment for common diseases involving collaboration between Korean medicine doctors and doctors, as with the use of EA plus UC to treat chronic pain, is a critical issue in Korea. Therefore, we conducted this research to evaluate the cost-effectiveness of concurrent EA and UC treatment compared with that of UC alone for non-acute LBP patients.
The purpose of this study was to examine the cost-utility of using EA with UC versus UC alone for post-back surgery pain by conducting an economic evaluation study alongside a randomized controlled trial. We hypothesized that EA in combination with UC would be cost-effective compared with UC alone.

Materials and Methods
This study was a multicenter, randomized, assessor-blinded, active-controlled trial and has been described in detail elsewhere [31]. The study adhered to the Consolidated Standards of Reporting Trials (CONSORT) guidelines (www.CONSORT-statement.org, Supplementary Material 1). We used the economic case report form along with the clinical case report form for capturing the variables of direct medical costs, direct non-medical costs (travel cost, hospital visiting time costs), and indirect costs (productivity losses from absenteeism and presenteeism). The economic evaluation was primarily based on the Korean healthcare perspective, and we conducted a sensitivity analysis using a societal perspective (SP). The analysis was restricted to a 24-week time horizon of a clinical research follow-up period. The economic evaluation was conducted according to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS checklist, Supplementary material 2).

Randomization Design and Participants
Participants with non-acute low back pain after back surgery who had a visual analog scale (VAS) pain intensity score of ≥50 mm were randomly assigned to either the EA with UC group or the UC alone group at a 1:1 ratio.

Ethics
All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the consolidated institutional review board of Pusan National University Korean Medicine Hospital (approval number: 2016003), Kyung Hee University Oriental Medicine Hospital at Gangdong (approval number: KHNMC OH 2015-10-002), and Jaseng Hospital of Korean Medicine (approval number: KNJSIRB2016-025). All participants participated voluntarily in this study, and written informed consent was obtained from each participant before enrollment.

Usual Care
Participants in both groups received physical therapy and a standardized educational program. Interferential current therapy (ICT; EF-150, OG Giken Co., Okayama, Japan; STI-300, Stratek Co. Ltd., Anyang, South Korea) and superficial heat therapy were applied twice per week during the 4-week treatment period. In addition, participants received a standardized educational program on LBP through a 20-min video and a brochure.

Electroacupuncture Collaboration
The patients in the EA with UC group received a total of eight sessions (two sessions/week) of EA treatment and UC for 4 weeks. The treatments were performed at Jia-ji (six acupuncture points, bilateral Ex-B2 at L3, L4, and L5) and a maximum of nine additional acupuncture points using disposable stainless-steel needles (0.25 × 40 mm, Dongbang Acupuncture Inc., Boryung, Korea). Electrical stimulation (ES-160, ITO Co. Ltd., Tokyo, Japan) was applied to four acupuncture points of Jia-ji (bilateral Ex-B2 at L3 and L5).

Permitted and Prohibited Concomitant Treatments
Conventional pharmaceutical and non-pharmaceutical treatments associated with post-operative LBP were permitted. However, invasive interventions, such as injections and surgery, were prohibited throughout the study period.

Utility Measure and Valuation
Utilities were measured using the Euroqol five-dimension scale three-level version (EQ-5D-3L). This measures five domains, including mobility, self-care, usual activity, pain, and anxiety/depression (no problem, 1; some/moderate problem, 2; extreme problem, 3). The EQ-5D index is calculated by weighting each level of the five domains (ranging from −1, every domain has extreme problem, to +1, none of the domains has a problem). This measurement was collected at baseline and at 4, 8, 12, and 24 weeks post-randomization. The descriptive categories of the EQ-5D-3L were converted to utilities using the South Korean national tariff [32], and then calculated as quality-adjusted life years (QALYs) using the area under the curve method. These calculated QALYs therefor take into account two factors, the quantity of life and the quality of life, measured during the study follow up. Normally 1 QALY during a year could be derived from 1 life year with the perfect health condition.

Resource Use and Cost Measures
The resource utilization of both treatment options was assessed from the societal perspective; therefore, the direct medical and non-medical costs, as well as the indirect medical costs, of each treatment were measured. The direct medical costs incurred by the study protocol were determined from the research hospital administrative data. Data on direct medical costs of other hospital visits and purchases of over-the-counter medication, medical devices, and functional foods were collected via an economic case report form developed for this study. Data on time costs for hospital visits and travel costs were also collected using the economic case report form. All medical costs, including EA, were valued using national health insurance reimbursement prices [33]. Productivity losses were calculated based on days of absence and decreased work efficiency due to a disease. As there has been no clear guideline and reasonable friction cost methods for calculating productivity loss in Korea, we used the human capital approach estimation method in this study by simply summing the costs incurred by absenteeism (number of days of absent × cost per day) and presenteeism (number of work days × cost per day × percent of decreased efficiency). All cost components in this study were sourced from or transformed to 2017 Korean Rate Won (KRW) monetary values; the details are included in Supplementary File 3. The entire study follow-up period did not exceed 1 year; therefore, the discount rates for reflecting time preferences were not adjusted in costs and utility outcomes.

Missing Data Analysis
For the intention-to-treat (ITT) analysis, patterns and mechanisms associated with missing data were identified. The total percentage of missing data was 11%, and the mechanism was found to be missing completely at random (MCAR) based on the results of Little's MCAR test (chi-square distance = 32.43, p = 0.985). After identifying the abovementioned value, multiple imputation of incomplete data was conducted using the Stata/MP14 (StataCorp LLC, TX, USA). Five imputed data Sustainability 2020, 12, 5033 5 of 13 sets were created using a multivariate normal model for utility outcomes and predictive mean matching for costs. Baseline covariates, including clinical and economic variables, were used for establishing imputation models.

Cost-Utility Analyses
The mean incremental cost-effectiveness ratio (ICER) was calculated by dividing the incremental costs by the incremental QALYs [25]. To overcome the sampling uncertainty of the point estimates of ICER, confidence intervals were estimated using the bootstrapping method (1000 times replicate) using Stata/MP 14. The mean values, ICER values and approximate bootstrap confidence intervals were displayed in the cost-effectiveness plane. The net monetary benefit (NMB) results were calculated using the equation below and displayed via a line graph. The probabilities of each alternative being cost-effective according to the changes in the national willingness to pay were also displayed using the cost-effectiveness acceptability curve.
The NMB equation is as follows: NMB = QALYs differences × national willingness to pay − cost differences

Sensitivity Analysis
Some sensitivity analyses of the deterministic ICER results were conducted from a limited societal perspective (LSP) using not only ITT data, but also per protocol (PP) data. By comparing the LSP and SP analysis results, we assessed the impact of productivity loss on the cost components, and by comparing the ITT and PP analysis results, we also confirmed the robustness of the missing data imputation.

Results
Between June 2016 and May 2017, 108 patients with non-acute low back pain who were eligible for participation were enrolled. Among the 108 participants, 106 were included in the cost-utility analysis and randomized equally to the two treatment groups (EA with UC in combination [n = 53] or UC alone [n = 53]); two participants (1 participant in each group) withdrew their consent before enrolling on the economic evaluation study. One participant in the EA with UC group withdrew from the study, and 12 were lost to follow-up. Three participants in the UC group withdrew from the study, and seven were lost to follow-up ( Figure 1). The total percentage of missing data at the patient level was 22.6%, and that at the value level was 15.3%. Table 1 shows the demographic and prognostic characteristics of the participants. There were no significant differences between the groups at baseline for any of the variables except direct medical cost (Table 1).

Costs
Detailed unit costs, sources of direct medical and direct non-medical costs, and productivity losses from presenteeism and absenteeism are shown in Supplementary File 4. The frequency of treatment with each alternative during the trial and throughout the follow-up period was recorded, and these data are listed in Supplementary File 3. From the SP, the total cost of EA in combination with UC based on data from participants who completed the trial was estimated as KRW 3,085,423 (SD 1,338,813), while the total cost of UC alone was KRW 3,043,643 (SD 1,098,816). From the LSP, the total costs of EA plus UC and UC alone were estimated as KRW 812,438 (SD 382,578) and KRW 682,409 (SD 274,958), respectively. Table 2 shows Quality of Life (QoL) scores of both treatment groups at each follow-up time point, and Table 3 shows the QALYs gained during the clinical trial and follow-up period. Although patients receiving EA plus UC showed a larger QALYs value than those receiving UC alone, the difference did not reach statistical significance (p = 0.358). The QALYs for 6 months for both treatment groups Sustainability 2020, 12, 5033 6 of 13 were estimated using the area under the curve method. The advantage of this approach was that the differences in the baseline utility values (whether statistically significant or not) could be considered for calculating the QALYs for the entire follow-up period [34].

Sensitivity Analyses
Sensitivity analyses were performed using the imputation data from a different perspective, which excluded the productivity losses associated with both treatment alternatives from the cost components. Even the differences in total costs increased from this LSP, with the mean ICER estimated to have a lower value (KRW 17,716,572 cost per QALY) than that of the Korean national threshold (Table 4).
In addition, we examined the mean ICERs from the SP and LSP using complete case data. Based on the PP analyses, the estimated mean ICER from the LSP was KRW 14,549,698.4 cost per QALY and that from the SP was KRW 4,674,999 cost per QALY, which was similar to the results of the main analyses using the value of the Korean national threshold.    BMI, body mass index; EQ-5D, Euroqol five-dimension scale; EQ-VAS, Euroqol-Visual Analogue Scale; KRW, Korean Rate Won; OP, operation; SD, standard deviation. EQ-5D ranges from 0 to 1. EQ-VAS ranges from 0 (the worst health you can imagine) to 100 (the best health you can imagine). Direct medical cost refers to the costs incurred after operation (surgery) before being enrolled in our study (excluding the cost of surgery). Continuous variables analyzed using an independent t test, and categorical variables analyzed using a chi-square test, except income and career variables (Fisher's exact test).

Incremental Cost-Effectiveness Ratio (ICER)
The ICER results are provided in Table 4. Treatment with EA plus UC resulted in a greater mean health benefit (0.009 QALYs) achieved at a higher mean total cost (KRW 64,603) than that with UC alone based on ITT analysis from the SP. The dominance of EA treatment (more benefit at a lower cost) was shown in 18.4% of the replications, and all other bootstrapped likelihood replications (81.6%) were located in the northeast quadrant, indicating that EA was more expensive but also more effective in the cost-effectiveness plane ( Figure 2). Figure 3A presents the NMB and confidence intervals of EA plus UC compared to UC alone according to changes in the national threshold. At the point of the Korean national threshold (KRW 20,000,000), mean NMB was estimated to be KRW 115,182, and the upper and lower confidence intervals around this value did not exclude zero, which indicated no statistical confirmation of the cost-effectiveness of EA plus UC treatment compared to UC alone in Korea. These results are also displayed as cost-effectiveness acceptability curve in Figure 3B, which show the cost-effectiveness probability of EA plus UC compared with that of UC alone at the points of national thresholds. At the point of the Korean national threshold, the percentage of cost-effectiveness was estimated as 87.1%, and if the national threshold was >KRW 42,591,316, EA plus UC may be confidently defined as a cost-effective alternative for treating non-acute LBP patients compared with UC alone.

Sensitivity Analyses
Sensitivity analyses were performed using the imputation data from a different perspective, which excluded the productivity losses associated with both treatment alternatives from the cost components. Even the differences in total costs increased from this LSP, with the mean ICER estimated to have a lower value (KRW 17,716,572 cost per QALY) than that of the Korean national threshold (Table 4).

Discussion
As sustainability is a critical issue in the establishment of healthcare polices, "smart spending" that expands investments in essential factors and reduces unnecessary costs is pursued [35]. Health sustainability can be divided into economic sustainability and fiscal sustainability. Fiscal sustainability considers only the balance of government spending and budget related to healthcare, and an increasing healthcare costs are interpreted as a threat to sustainability. However, if the benefits produced by healthcare are greater than the total cost, and healthcare costs do not threaten other economic activities, these benefits contribute to economic sustainability [36]. In this study, the overall benefits and total costs related to non-acute pain after back surgery were considered.
EA in combination with UC is a commonly used therapy for patients with LBP after back surgery or for those who experience side effects from the opioids prescribed to treat their pain. In previous research, the effectiveness of EA was shown to alleviate the sensory symptoms and regulate components of pain through specific neuroscientific mechanisms, and is therefore thought to contribute to the reduction of pain medication dosages [37]. Additional research has reported the use of EA for managing postoperative pain from various procedures other than back surgery, such as cardiopulmonary bypass, mixed hemorrhoid surgery, and gynecological laparoscopic surgery [24].
To our knowledge, this is the first study in Korea using economic evaluation alongside a clinical trial to compare EA plus UC with UC alone in the treatment of patients with non-acute pain after back surgery. Prior research in the UK demonstrated that additional acupuncture treatment was cost-effective compared with UC alone for treating chronic LBP [38]. Moreover, in Germany, acupuncture was shown to be relatively cost-effective [39]. Our research results showed similar trends and indicated the cost-effectiveness of EA plus UC compared with that of UC alone via the deterministic ICER. Although there are uncertainties surrounding the dominance of the EA plus UC combination using mean values, if the national threshold was >KRW 20,000,000 per QALY, the probability for cost-effectiveness of EA plus UC would exceed 59.2%. The CEAC also showed favorable results, with a 59.2% probability of willingness to pay with the threshold at KRW 20,000,000 per QALY and 85.9% probability at a threshold of KRW 30,000,000 per QALY. Overall, in Korea, the collaboration of specialists using EA plus UC in combination shows the possibility for increased cost-effectiveness, from the LSP (healthcare perspective) in treating LBP patients after back surgery. In addition to examining the parameter uncertainties, we explored other factors that might affect the conclusions of the study, such as different perspectives (whether to consider productivity cost) and missing data imputation. Table 4 shows the combined results accounting for the different perspectives and after addressing missing data.
The findings of our study are consistent with those of previous studies that have reported acupuncture treatment as being cost-effective for chronic LBP [40]. According to a clinical trial conducted in United Kingdom [41] and a Korean study using a Markov model decision analysis, acupuncture collaborative therapy has been regarded as a cost-effective treatment compared to usual care only for LBP. In the same context, a meta-analysis revealed that acupuncture as a complementary treatment is highly cost-effective for chronic LBP [42]. However, substitutional use of acupuncture was only cost-effective when the condition was accompanied by depression [42]; and an economic evaluation conducted in Iran showed that EA was less cost-effective than non-steroidal anti-inflammatory drugs, but this study did not report the ICER [43]. Further studies are needed to draw a definitive conclusion.
There are several limitations that should be considered when interpreting the study results. First, similar to other economic evaluation alongside clinical trials, the main purpose of this research was to examine the effectiveness of both treatments and, as such, the predefined sample size was not large enough to address the research question of cost-effectiveness with adequate statistical power. Second, the patients included according to predefined criteria do not represent the general population in a real-world context, which is common for economic evaluations alongside randomized controlled trials. Third, records of medical utilization outside research hospitals during follow-up periods was gathered using self-report forms, introducing the potential for recall bias in estimating cost values. Fourth, although all costs incurred after randomization were captured by identification, measurement, and valuation using the economic case report form, costs incurred before entering this research could not wholly be identified due to the memory loss of the participants, and so we excluded the costs identified at baseline from the cost estimation analysis. Finally, even though this study was a multi-center trial, concerns about the generalizability of the results from a single randomized controlled trial remain. Despite these limitations, this study is the first report of an economic evaluation alongside a clinical trial, highlighting the issue of collaboration between Korean medicine doctors and medical doctors for concurrent treatment of LBP in Korea. Moreover, patient utilities and cost components, including productivity losses, may be captured directly from trial participants and could reflect real-world economic and clinical situations. The steps for conducting an economic evaluation within an RCT are defined as follows: the quantification of the cost and effect of care, assessment of the difference in the cost and effect between the treatment groups, comparison of the magnitudes of differences in cost and effect, and reporting of the ICER, NMB, and CEAC [25]. We followed this whole process and reported the results according to both the trial (CONSORT) and economic evaluation (CHEERS) reporting guidelines.
In sum, the EA and UC combination therapy resulted in better clinical outcomes and higher total costs relative to UC alone from the LSP, with a 59.2% cost-effectiveness probability at the national threshold of KRW 20,000,000 per QALY and an 85.9% probability at the threshold of KRW 30,000,000 per QALY in Korea.

Conclusions
The deterministic and sensitivity analyses of different perspectives using PP and ITT data showed that the EA and UC combination was more effective and costlier than UC alone. The bootstrapping analysis results displayed in the CE plane showed the mean values (KRW 704,861 per QALY) and 95% CI (KRW −6,957,703 and KRW 42,496,372) of the ICER, and NMB and CEAC estimate results also showed the mean values, 95% CI, and cost effective probabilities according to changes in the national threshold. These results show that the ICER of EA plus UC was estimated as KRW 7,048,602 per QALY from the societal perspective, and, if the national threshold was KRW 30,000,000 per QALY, the cost-effectiveness probability of EA plus UC was an estimated 85.9%. Moreover, if the national threshold exceeds KRW 42,496,372 per QALY, the EA combination treatment will be cost-effective with statistical significance. To confirm our study results, future economic evaluation research in conjunction with clinical trials or modeling analyses may be needed to compare multiple treatment options or estimate longer term cost-effectiveness.
Supplementary Materials: Materials available online at http://www.mdpi.com/2071-1050/12/12/5033/s1: Table  S1: CONSORT 2010 checklist of information to include when reporting a randomized trial'; Table S2: CHEERS Checklist; Table S3: Unit prices (2017 KRW) and treatment frequencies; Table S4: Summary of censored data about outcome variables. Funding: This work was supported by Grant K16123 from the Korea Institute of Oriental Medicine. The funder had no role in study design, data collection, analysis, or interpretation, or preparation of the manuscript.