ORIGINAL ARTICLE. Introduction

Transcription

1 ORIGINAL ARTICLE Vasc Fail 2018; 2: A Cost-benefit Analysis of Percutaneous Coronary Intervention and Coronary Artery Bypass Grafting using Reimbursement Data of Japan: A Single-center Pilot Study Sayuri Nonaka, M.P.A.S 1), Susumu Fujii, PhD 1), Megumi Hara, MD 2), Kojiro Furukawa, MD 3), Yutaka Hikichi, MD 4), Eisaburo Sueoka, MD 1), Koichi Node, MD 4) and Shigeki Morita, MD 3)5) Abstract: Background: Although there have been studies comparing percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG), few comparative cost-benefit analyses using reimbursement data have been performed, especially in Japan. Data from the Diagnosis Procedure Combination/Per-Diem Payment System (DPC/PDPS), a reimbursement system in Japan, may be useful in performing a cost-benefit analysis. Methods and Results: Between July 2008 and March 2016, a total of 48,177 patients were admitted to Saga University Hospital. Using DPC/PDPS data, we identified 638 patients without a history of myocardial infarction who underwent PCI (Group PCI, n=462) or isolated CABG (Group CABG, n=176). There were no marked differences in the mortality rate, but the incidence of myocardial infarction was higher in Group PCI. A multivariate logistic regression analysis showed that performing PCI was a significant risk factor for myocardial infarction. The number of admissions was smaller in Group CABG, but the medical cost was higher and the total hospital stay was longer than in Group PCI. However, after three or more PCIs, the difference in medical cost disappeared between PCI and CABG. Conclusions: In our single-center experience, we were able to show that, while there was no marked difference in the survival rate, the incidence of myocardial infarction was higher with PCI than with CABG. PCI required multiple admissions but was less costly than CABG. The feasibility of using the DPC/PDPS data for costbenefit analysis of procedures like PCI and CABG should be confirmed in a multi-center study. Key words: Diagnosis Procedure Combination/Per-Diem Payment System, Percutaneous coronary intervention, Coronary artery bypass grafting Introduction Percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) are the major therapeutic techniques for treating ischemic heart disease. Although numerous studies 1-6) have compared the immediate and medium- to long-term results between the two approaches, few studies 5,6) have compared the cost-effectiveness of PCI and CABG. In this single-center pilot study, we attempted to compare the cost-effectiveness of PCI and CABG using a dataset from the Diagnosis Procedure Combination/Per-Diem Payment System (DPC/PDPS). The DPC/PDPS is a reimbursement system in Japan for patients admitted to acute care hospitals. These data include details of daily medical procedures, surgeries, and prescribed medicines, as well as demographics, medical history, and survival. Although anatomical and/or pathophysiological data are not available in the DPC/PDPS data, these data may neverthe- 1) Saga University Hospital 2) The Department of Preventive Medicine, Saga University 3) The Department of Thoracic and Cardiovascular Surgery, Saga University 4) The Department of Cardiology of Saga University 5) The Clinical Research Center of Kyushu Medical Center, National Hospital Organization Corresponding author: Shigeki Morita, morita@kyumed.jp Received: December 1, 2017, Accepted: January 10, 2018 Copyright 2018 Japan Society for Vascular Failure Vascular Failure 2018; 2(1):

2 NONAKA S et al. less be sufficient for comparing the cost-effectiveness of PCI and CABG. In the present study, we compared the cost-effectiveness of PCI and CABG using DPC/PDPS data and assessed the feasibility of using such data in this manner, as this system covers almost all of the centers where PCI and CABG are performed in Japan. If this approach proved to be feasible, the DPC/PDPS data may be considered a useful platform for developing a nationwide, unbiased, and readily available source for a wide variety of diseases. Patients and Methods Between July 2008 and March 2016, a total of 48,177 patients were admitted to the Saga University Hospital. Of those patients, we identified patients who were diagnosed with ischemic heart disease with chronic angina and underwent PCI or CABG, using DPC codes and K-codes. DPC codes designate diagnoses, such as for acute myocardial infarction and recurrent myocardial infarction, while K-codes reflect procedures, such as K549 for PCI with stent implantation and K-552 for CABG. These codes are revised every other year 7). The data were obtained from Form 1 and the E/F file of the DPC/PDPS data. The detailed process of selecting patients is shown in Figure 1. In short, patients who had an admission code for angina (DPC code of ) with either PCI or CABG as first intervention during the study period were identified (Figure 1, Step 1). We then selected s who did not have myocardial infarction at the time of first PCI/CABG, in two steps. First, using the DPC admission code, we excluded patients who had a history of admission with a diagnosis of acute myocardial infarction (DPC code of ) prior to admission for PCI/CABG. Second, using the International Classification of Disease 10th revision (ICD-10) code listed under the accompanying disease at the time of admission, we excluded patients who had ICD-10 codes for old myocardial infarction at or prior to the admission for the first PCI/ CABG (Figure 1, Step 2). In this manner, we attempted to make the population homogeneous in terms of not having a history of myocardial infarction. To further refine the population, patients who had been admitted for reasons unrelated to myocardial infarction or angina before or after the admission for the first PCI/CABG were excluded (Figure 1, Step 3). This was done because including patients who have diseases unrelated to ischemic heart disease might influence the data, especially the data for the cost analysis. The patients who underwent concomitant interventional procedures were excluded by checking the second diagnosis listed as an ICD-10 code (Figure 1, Step 4). Regulations state that a second diagnosis be listed when an intervention unrelated to the first diagnosis is performed. In this study, the first diagnosis was always related to PCI or CABG. In this manner, patients who underwent concomitant procedures at the time of CABG, such as aortic valve replacement, were excluded. Patients who underwent procedures related to complications of PCI or CABG were not excluded, because the occurrence of complications influences the clinical outcome and medical costs. Finally, 638 eligible patients were identified. We limited enrollment from July 2008 to March 2016, and followed each patient s data through to March 2017; hence, at least one year of follow-up data was available for all patients. The patients were divided into two groups according to the first procedure they had: PCI (Group PCI) or CABG (Group CABG). Group PCI was further divided into three groups, according to the number of PCIs they had undergone: one (PCI-1), two (PCI-2), or three or more (PCI-3). The primary items we examined for the benefit analysis were hospital mortality and the occurrence of new myocardial infarction. In terms of the cost analysis, the number of admissions, total length of hospital stay, and total medical cost were compared. The total medical cost was estimated based on the amount reimbursed by the insurance provider. The amount reimbursed comprised a fee-for-service component and a DPC/ PDPS component. The reimbursement of costs related to intervention and surgery was done on a fee-for-service basis, such as the services provided in the catheterization laboratory or in the operating room. Costs outside the catheterization laboratory or operating room are then reimbursed based on the DPC/PDPS, which means that the amount paid per day is determined by the unique combination of diagnosis, major procedure, and the comorbidities listed at the time of admission, irrespective of the medications given or the tests performed. Aside from the reimbursement data, regulations state that pay-for-service data for the entire hospital stay be submitted, so that the government can compare the amount reimbursed and the actual costs the hospital had to pay. We therefore used the pay-for-service data to breakdown the medical costs, such as those used to pay for medications, tests, and devices/equipment. Statistical analysis With regard to continuous variables between two groups (Groups PCI and CABG), the F-test was used to determine whether the variables followed a normal distribution. If the variables were normally distributed, a t-test was used for statistical comparison. For variables that did not follow a normal distribution, the Mann-Whitney U test was performed. For comparison among the four groups (Groups PCI-1, PCI-2, PCI-3, and CABG), a non-parametric multiple comparison method (Dunn test) was performed. For categorical variables, chi-squared analysis was performed. A p- value<0.05 was considered statistically significant. A commercially available software program (StatFlex; Arctech, Co., Ltd., Osaka, Japan) was used for the analyses. 26 Vascular Failure 2018; 2(1): 25-31

3 Comparison of PCI and CABG using DPC data Patients admitted between July 2008 and March 2016 n=48,177 Step 1 Was PCI or CABG performed for the first time under the code of "Angina (050050)? No n=933 Step 2 At the time of the first PCI or CABG, did have a history of MI? No n=722 Step 3 Was admitted for reasons unrelated to MI or Angina before or after the first PCI/CABG? No n=640 Step 4 Any procedures unrelated to PCI or CABG at the first admission? No n=638 PCI Which procedure did undergo, PCI or CABG? CABG Group PCI n=462 Group CABG n=176 Did undergo CABG after PCI? No How many times did undergo PCI? How many times did undergo CABG? No Did undergo PCI after CABG? Once Once Group PCI-CABG n=8 Group PCI-1 n=275 Twice Group PCI-2 n=127 Twice Group CABG n=170 Group CABG-2 n=0 Group CABG-PCI n=6 Thrice or more Thrice or more Group PCI-3 n=52 Group CABG-3 n=0 Figure 1. The patient selection process using DPC/PDPS data. *PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, MI: myocardial infarction, DPC/PDPS: Diagnosis Procedure Combination/Per-Diem Payment System Results The allocation of patients is shown in Figure 1. The number of patients allocated to Groups PCI and CABG were 462 and 176, respectively. The number of male patients was 331 (72.9%) in Group PCI and 119 (70.9%) in Group CABG. Comparison of the background characteristics of s in the two groups is shown in Table 1. There was no significant difference in the age distribution between Group PCI (mean 69.7 years, standard deviation [SD] 10.5 years, median 71 years) and Group CABG (mean 69.9 years, SD 9.5 years, median 71 years). The percentages of hypertension, hyperlipidemia, and hypercholesterolemia were higher in Group PCI than in Group CABG. The number of patients in Groups PCI-1, PCI-2, and PCI- 3, were 275, 127, and 52, respectively. Comparison of the Vascular Failure 2018; 2(1):

4 NONAKA S et al. Table 1. Patients background characteristics, mortality, and medical costs in Groups PCI and CABG. PCI (n=454) CABG (n=170) P value Age (years; mean, SD, median) (69.7, 10.5, 71) (69.9, 9.5, 71) Male (n; %) % % Comorbidity Hypertension % % < Hyperlipidemia % % Hypercholesterolemia % % < Diabetes % % Heart Failure % % Renal failure % % Hyperuricemia % 5 2.9% Atrial Fibrillation % 8 4.7% Old cerebral infarction 9 2.0% 4 2.4% Carotid artery disease % 7 4.1% Peripheral artery disease % 9 5.3% Mortality 7 1.5% 4 2.4% Medical cost (JPY: 10 6 ; mean, SD, median) (2.0, 1.1, 1.7) (3.7, 1.4, 3.2) < PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, SD: standard deviation, JPY: Japanese yen Table 2. Patients background characteristics, mortality, and medical costs in Groups PCI-1, PCI-2, and PCI-3. PCI-1 (n=275) PCI-2 (n=127) PCI-3 (n=52) P value Age (years; mean, SD, median) (69.6, 11.1, 71) (69.7, 9.9, 71) (70.3, 9.4, 72) Male (n; %) % % % Comorbidity Hypertension % % % Hyperlipidemia % % % Hypercholesterolemia % % % Diabetes % % % Heart Failure % % % Renal failure % 6 4.7% 3 5.8% Hyperuricemia % % 4 7.7% Atrial Fibrillation 9 3.3% 5 3.9% 5 9.6% Old cerebral infarction 3 1.1% 6 4.7% 0 0.0% Carotid artery disease 3 1.1% 3 2.4% 4 7.7% Peripheral artery disease 5 1.8% 6 4.7% 0 0.0% CABG Mortality 4 1.5% 2 1.6% 1 1.9% 4 2.4% Medical cost (JPY: 10 6 ) (mean, SD, median) (1.4, 0.5, 1.3) (2.5, 0.7, 2.4) (4.1, 1.3, 4.0) (3.7, 1.4, 3.2)* *PCI-1 vs. PCI-2, PCI-1 vs. PCI-3, PCI-2 vs. PCI-3, PCI-1 vs. CABG, PCI-2 vs. CABG: P<0.01, PCI-3 vs. CABG: NS PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, SD: standard deviation, JPY: Japanese yen background characteristics of s in Groups PCI-1, PCI-2, and PCI-3 is shown in Table 2. There were eight patients who had PCI first followed by CABG (Group PCI- CABG), and six patients who had CABG first followed by PCI (Group CABG-PCI). These patients were not included in the sub-group analysis because of their small numbers. There were no significant differences in the mortality rates between Groups PCI and CABG (Table 1, 2). The incidence of myocardial infarction after the first intervention was significantly higher in Group PCI than in Group CABG (p<0.001) (Figure 2). There were also more admissions in Group PCI than in Group CABG (Figure 3), but the cumulative length of stay in the hospital was shorter (Figure 4) and the cumulative medical cost smaller in Group PCI (Table 1, 2). There was no marked difference in the medical cost between Group PCI-3 and Group CABG. The breakdown of medical costs is shown in Figure 5. The cost for devices was larger in Group PCI, whereas the cost for procedures was larger in Group CABG. A multivariate logistic regression analysis was performed to examine the factors related to myocardial infarction after PCI or CABG. Male gender and PCI were found to be significant factors (Table 3). Discussion Using DPC/PDPS data, we showed that there was no marked difference in the mortality rates between PCI and 28 Vascular Failure 2018; 2(1): 25-31

5 Comparison of PCI and CABG using DPC data CABG, but patients who were treated with PCI required more admissions and had a higher incidence of newly developed myocardial infarction than those with CABG. The total length of stay was longer with CABG, and the total costs were higher. However, when the comparison was limited to patients who required PCI three times or more, the differences in total cost disappeared. Previous comparative studies have indicated a better survival and lower rates of major adverse cardiovascular events after CABG than after PCI 4). However, our study revealed no marked differences in the survival between the two procedures. This discrepancy may be due to limitations associated with the method of surveillance of deaths in our study. Although we were able to detect deaths during the hospital stay patients who died outside the hospital were deemed to be lost to follow-up and were not counted as deaths. At present, we have no system to register the births and deaths of (n=13) 4.7% (n=13) 10.2% (n=13) 25.0% (n=2) 1.2% PCI-1 PCI-2 PCI-3 CABG Figure 2. The incidence of newly developed myocardial infarction during the study period. Significant differences were found between Group CABG and Groups PCI-1, PCI-2, and PCI-3. *PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting the people in Japan for the purpose of gathering health care information. Although the My Number system was recently introduced to register all people in Japan in order to follow each person s income, the system has yet to be connected to health care information. Combining this system with the DPC/PDPS data may help integrate patients health care data dispersed among hospitals, making it a powerful system for obtaining national health care information. The incidence of newly developed myocardial infarction was higher in PCI patients than in CABG patients, with a high odds ratio of >8.0. It has been shown that PCI carries a risk of sudden stent occlusion, even with the use of drugeluting stents, whereas CABG has been shown to have a better survival, especially when the left internal thoracic artery is anastomosed to the anterior descending branch of the left coronary artery. The difference in the incidence of myocardial infarction between these two approaches may be due to differences in the patency rates between the stent and the graft after each procedure. These findings should be confirmed using original data, such as electrocardiogram findings or cardiac enzyme levels, and not to be based merely on secondary information determined by ICD-10 codes provided s physicians, as was done in this study. A validated method for documenting myocardial infarction should be explored in future studies. The results presented in this study were derived solely from the DPC/PDPS data of Saga University Hospital. Regulations state that these data be uploaded to a semipublic third-party institute. As long as the DPC/PDPS data are in the same format, our algorithm can be applied to the data reported from other institutions using the DPC/PDPS throughout Japan. This implies that similar analyses can be performed for a wide range of patients. In Japan, as of April 2017, 1,664 hospitals and more than 483,000 beds are covered by the DPC/PDPS, accounting for 54% of the total hospital beds in Japan. Furthermore, this system likely cov- Median Figure 3. Total number of admissions due to cardiovascular disease. *PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting Vascular Failure 2018; 2(1):

6 NONAKA S et al. Median Figure 4. Cumulative hospital stay (days) due to cardiovascular disease. *PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting 4,500,000 4,000,000 3,500,000 3,000,000 2,500,000 2,000,000 1,500,000 1,000, ,000 0 PCI-1 PCI-2 PCI-3 CABG Miscellaneous Tests/Labs Medications and intravenous fluids Device / Equipment Fee for beds Fee for PCI /CABG Figure 5. Breakdown of the medical costs. *PCI: percutaneous coronary intervention, CABG: coronary artery bypass grafting, JPY: Japanese yen ers almost all of the hospitals performing PCI and CABG. Using DPC/PDPS data supported by anonymity technology, similar analyses with multiple participating institutions will be possible, facilitating the construction of big data. Most large-scale clinical studies are currently funded by the pharmaceutical or medical device industry. For studies with no funding from an industrial agency, such as ours, the DPC/ PDPS data may be useful for investigating clinical issues not related to commercial products. Several limitations associated with the present study warrant mention. First, the observations in this study were limited to the admissions at Saga University Hospital. Data from other hospitals providing these treatments were not included. This limitation is related to the lack of a system integrating patient data with a unique code or number across institutions. The advent of such a system is eagerly awaited, provided that careful and meticulous preparation has been performed, taking care to protect the privacy of s information using anonymization technology. Second, it was not possible to measure and analyze the costs related to personnel expenses. In addition, the expenses related to the maintenance and utilization of the operating room or catheterization laboratory were not taken into account. Because our purpose was to examine the feasibility of using DPC/ PDPS data for a comparative cost-benefit analysis, evaluating the personnel and maintenance costs was beyond the 30 Vascular Failure 2018; 2(1): 25-31

7 Comparison of PCI and CABG using DPC data Table 3. Multiple logistic regression analysis of factors related to newly developed myocardial infarction after the fi rst PCI/CABG hospitalization. Variable β SE (β) z value p value OR 95% CI Male ~9.74 Age ~1.01 Hypertension ~2.27 Hyperlipidemia ~2.14 Hypercholesterolemia ~2.19 Diabetes ~1.44 Heart Failure ~4.46 Atrial Fibrillation ~6.53 Carotid artery disease ~10.26 Renal Failure ~5.03 PCI ~35.15 Conformity degree index of the regression: AIC= , AUC= OR: odds ratio, 95% CI: 95% confi dence interval, β: regression coefficient, SE (β): standard error (β), PCI: percutaneous coronary intervention, AIC: Akaike s Information Criterion, AUC: Area Under the Curve scope of our study due to the nature of the DPC/PDPS data. This does not imply that personnel and maintenance cost analyses are not important. However, a different approach or framework should be employed to explore these important issues. Third, anatomical and pathophysiological information, such as descriptions of coronary angiography, the ejection fraction, and electrocardiogram findings, were not available in the DPC/PDPS data. Furthermore, as mentioned earlier, the history of myocardial infarction was taken from the ICD-10 diagnosis reported by s physician, instead of examining clinical data (electrocardiography, echocardiography, or myocardial enzymes). Moreover, unexpected findings of higher incidences of hypertension and hyperlipidemia in Group PCI than in Group CABG might be related to the physicians bias of listing accompanying disease, because cardiologists performed the listing for Group PCI, while cardiac surgeons performed the listing for Group CABG. While the numerical values of biochemical and physiological tests are expected to be included in the DPC/ PDPS data at some point, these data are not available in the system at present. Further refinement of the system, including the gathering of a wide range of medical information into the DPC/PDPS database with online availability, is warranted. In conclusion, a cost-benefit analysis using DPC data showed that, except with regard to the incidence of newly developed myocardial infarction, PCI was a more advantageous technique than CABG when performed only once or twice. However, the benefit of PCI over CABG was lost when the technique was performed three times or more. The usage of DPC/PDPS data for analysis is feasible when the application is limited to the data related to survival, comorbidities, length of hospital stay, or reimbursement. Further studies are warranted to utilize the DPC/PDPS data for clinical research in order to perform unbiased and relatively inexpensive analyses. Conflicts of Interest There are no conflicts of interest References 1. Kimura T, Morimoto T, Furukawa Y, Nakagawa Y, Shizuta S, Ehara N, et al. Long-term outcomes of coronary-artery bypass graft surgery versus percutaneous coronary intervention for multivessel coronary artery disease in the bare-metal stent era. Circulation 2008; 118: S Seung KB, Park DW, Kim YH, Lee SW, Lee CW, Hong MK, et al. Stents versus coronary-artery bypass grafting for left main coronary artery disease. N Engl J Med 2008; 358: Serruys PW, Morice M-C, Kappetein AP, Colombo A, Holmes DR, Mack MJ, et al. Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease. N Engl J Med 2009; 360: Kappetein AP, Feldman TE, Mack MJ, Morice MC, Holmes DR, Stahle E, et al. Comparison of coronary bypass surgery with drugeluting stenting for the treatment of left main and/or three-vessel disease: 3-year follow-up of the SYNTAX trial. Eur Hear J 2011; 32: Vieira RDO, Hueb W, Hlatky M, Favarato D, Rezende PC, Garzillo CL, et al. Cost-effectiveness analysis for surgical, angioplasty, or medical therapeutics for coronary artery disease: 5- year follow-up of medicine, angioplasty, or surgery study (MASS) II trial. Circulation 2012; 126: Magnuson EA, Farkouh ME, Fuster V, Wang K, Vilain K, Li H, et al. Cost-effectiveness of percutaneous coronary intervention with drug eluting stents versus bypass surgery for patients with diabetes mellitus and multivessel coronary artery disease: results from the FREEDOM trial. Circulation 2013; 127: Research Institute of Social Insurance. Interpretation of the medical reimbursement score table. 39 th version. Tokyo: p Vascular Failure 2018; 2(1):