慢性B型肝炎特定之健康照護費用：以Kaplan-Meier Sample Average法處理右設限長期追蹤資料

The Health Care Expenditures Specific to Chronic Hepatitis B:A Kaplan-Meier Sample Average Estimator for Right Censored Long-term Follow-up Data

廖芷嫻(Chih-Hsien Liao)

指導教授 ： 楊銘欽

國立臺灣大學/公共衛生學院/健康政策與管理研究所/博士(2017年)

https://doi.org/10.6342/NTU201700242

摘要

研究背景：疾病成本分析可應用在健康科技評估，特別在慣用全球決策分析模型在地調整適用的健康照護體系中，政策制定者需要具當地代表性、可信賴、符合模型假設的成本參數。本研究旨在以我國B型肝炎病毒的慢性併發症為例，以Kaplan-Meier Sample Average法（KMSA）處理右設限長期追蹤資料，以進行更為精確的疾病成本分析。
材料與方法：本研究採用中央健康保險署觀點，以健保給付的直接醫療費用估計疾病成本。分析全民健康保險申報資料，從中定義2005年到2014年間，全國肝移植、肝癌、失代償性肝硬化、代償性肝硬化、慢性B型肝炎和其他B型肝炎的個案。使用新發個案為基礎的分析方法追蹤病人罹患併發症的時間與對應的疾病成本。應用Kaplan-Meier Sampling Average法處理10年追蹤期間的設限資料，自發病時間起呈現逐月的疾病成本，包括總成本法（完整個案樣本）與遞增成本法（配對後個案與對照樣本）兩套估計值，以供比較。10年累積疾病成本分別以0%、3%或5%年折現率調整。為展示疾病成本在成本效果分析的應用，分析結果經存活條件機率換算為年成本參數，此參數未經折現。另外建構KMSA預測模型，依模擬對象的性別、年齡及發病年度，預測對應的年成本參數。在健保支付點值一點一元的前提假設下，所有估計值皆以新台幣的名目價值為單位。
研究結果：六項併發症疾病成本的高峰皆出現在罹病初期，第一年成本在10年累計成本的佔率分別為：肝移植79%、肝癌41%、失代償性肝硬化41%、代償性肝硬化32%、慢性B型肝炎19%及其他B型肝炎34%。在年折現率為3%的情境中，平均每人的B型肝炎慢性併發症的10年累積總成本（遞增成本）分別為：肝移植元2,159,302元（1,885,138元）、肝癌355,814元（218,922元）、失代償性肝硬化261,297元（147,105元）、代償性肝硬化164,750元（11,740元）、慢性B型肝炎89,829元（68,418元），及其他B型肝炎129,852元（100,886元）。其中，代償性肝硬化的遞增成本遠低於總成本，兩項估計值的比值為3%。以肝臟移植為例，應決策分析模型需要，將疾病成本換算為年成本參數：第一年的年成本參數為1,764,269元，第10年則為47,989元；在接受肝移植手術的男性40歲個案，第一年的年成本參數為1,974,946元，第5年則為76,766元。
結論：本研究是應用KMSA將疾病成本分析轉譯到成本效果分析的前驅研究，經由分析臨床實證資料，確認B型肝炎慢性併發症的疾病成本與罹病時間高度相關。這個關係在成本效果分析中特別重要，因為折現因子與罹病個案數皆隨時間而異。其他慢性疾病，與B型肝炎慢性併發症相似，其疾病成本與罹病時間可能高度相關，應優先採用新發個案為基礎的疾病成本分析，以幫助於政策制定者充分瞭解健康科技所致的經濟影響。

關鍵字

疾病成本分析 ； 成本效果分析 ； 慢性B型肝炎 ； 代償性肝硬化 ； 失代償性肝硬化 ； 肝癌 ； 肝移植

並列摘要

Background: Cost-of-illness analyses can be useful in Health Technology Assessment, especially for the healthcare setting where local adaptation is the major approach to generate economic evidence for decision making. This study aims to demonstrate how to engage a Kaplan-Meier Sample Average (KMSA) Estimator for right censored long-term follow-up data to estimate cost-of-illness more precisely, with application to chronic complications of hepatitis B virus (HBV) infection in Taiwan.
Methods: From the perspective of the National Health Insurance Administration (NHIA), this study defined cost-of-illness (COI) as the direct medical expenditures covered by the NHI. Utilizing the NHI claims datasets, nationwide patients with chronic HBV complications, including liver transplantation (LT), hepatocellular carcinoma (HCC), decompensated cirrhosis (DCC), compensated cirrhosis (CC), chronic hepatitis B (CHB), and other hepatitis B (OHB) between 2005 to 2014 were identified. An incidence-based approach was taken to trace progression patterns and corresponding COI attached to each complication. The KMSA method was implemented to analyze 10-year COI with censored data. Time-dependent KMSA estimators were generated from disease onset until the 120th month, adopting either total cost mode (complete case sample) or incremental cost mode (matched case-control sample). The 10-year cumulative COI was discounted annually at 0%, 3% and 5%, respectively. To demonstrate the application of COI estimators in cost-effectiveness analyses, annual cost parameters were executed by adjusting conditional survival probabilities given time, but not by discounting factors. KMSA predicted models were constructed for various ages, gender, onset year, comorbidity cohorts. All estimators were presented in nominal New-Taiwan dollars (NT$) on the assumption of no adjustment for point value.
Results: For all six complications, the peak of COI was shown in the beginning of disease onset. The share of first-year COI over 10-year cumulative COI were 79% in LT, 41% in HCC, 41% in DCC, 32% in CC, 19% in CHB, and 34% in OHB. At 3% annual discount rate, the 10-year cumulative total COI (incremental COI) per patient were NT$ 2,159,302 (NT$ 1,885,138) in LT, NT$ 355,814 (NT$ 218,922) in HCC, NT$ 261,297 (NT$ 147,105) in DCC, NT$ 164,750 (NT$ 11,740) in CC, NT$ 89,829 (NT$ 68,418) in CHB, and NT$ 129,852 (NT$ 100,886) in OHB, respectively. For CC patients, the incremental COI over total COI was relative low (3%) compared with other complications.
An example applying COI in cost-effectiveness analysis is as follow: The annual cost parameters of LT ranged from NT$ 1,764,269 in the first year to NT$ 47,989 in the 10th year. Given a hypothetical cohort consisted of 40 year old male who underwent LT, the predicted annual cost parameters will be NT$ 1,974,946 in the first year and NT$ 76,766 in the 5th year.
Conclusions: This is a pilot study for using KMSA to estimate COI for cost-effectiveness analysis. Based on real-world evidence, we concluded that the COI of chronic HBV complications were highly associated with onset time. This association is important in cost-effectiveness analysis because both discounting and number of applied cases declined with time. Other chronic diseases may be similar to chronic HBV complications in respect of the association between COI and onset time. Among these diseases, incidence-based COI is recommended to provide information for decision making.

並列關鍵字

cost-of-illness analyses ； cost-effectiveness analysis ； chronic hepatitis B ； compensated cirrhosis ； decompensated cirrhosis ； hepatocellular carcinoma ； liver transplantation

參考文獻

Akobundu, E., Ju, J., Blatt, L., Mullins, C. D. (2006). Cost-of-illness studies : a review of current methods. Pharmacoeconomics, 24(9), 869-890.

Antonanzas, F., Rodriguez-Ibeas, R., Juarez, C., Hutter, F., Lorente, R., Pinillos, M. (2009). Transferability indices for health economic evaluations: methods and applications. Health Econ, 18(6), 629-643. doi:10.1002/hec.1397

Beasley, R. P., Hwang, L. Y., Lin, C. C., Leu, M. L., Stevens, C. E., Szmuness, W., Chen, K. P. (1982). Incidence of hepatitis B virus infections in preschool children in Taiwan. J Infect Dis, 146(2), 198-204.

Bloom, B. S., Bruno, D. J., Maman, D. Y., Jayadevappa, R. (2001). Usefulness of US cost-of-illness studies in healthcare decision making. Pharmacoeconomics, 19(2), 207-213.

Boulenger, S., Nixon, J., Drummond, M., Ulmann, P., Rice, S., de Pouvourville, G. (2005). Can economic evaluations be made more transferable? Eur J Health Econ, 6(4), 334-346. doi:10.1007/s10198-005-0322-1