- 學位論文
個人化高血壓防治之成本效益分析
Cost-effectiveness Analysis of Personalized Prevention for Hypertension
摘要
前言:高血壓是導致心血管疾病、腦血管疾病和慢性腎臟疾病及其相關併發症和死亡風險的主要原因之一。不論是用藥物或非藥物介入的方式去提昇病識或以群族基礎去進行介入計畫的重要性均無庸置疑。雖然這種策略過去都曾被提出，但近年發展出來的個人化醫療模式，其可結合藥物或非藥物介入的策略以進行高血壓防治，故漸漸受到關注，但仍不足，且缺乏相關的經濟評估，其因部分因為缺乏流行病學上資訊及高血壓自然病史(正常高血壓前期第一期高血壓第二期高血壓)及各階段的相對應的影響因子，以致過去未有相關文獻去探討在不同危險族群下之個人化高血壓防治策略的經濟評估。 研究目的：本論文研究乃利用社區實證資料 1)了解男性及女性在年輕(<50歲)及年長(>=50歲)高血壓前期、第一期及第二期高血壓的盛行率及發生率； 2)利用上述資料量化高血壓四階段自然病史，包含高血壓前期回復至正常血壓的回復率及從正常血壓往前至高血壓前期、第一期及第二期高血壓的轉移速率，並求算至第一期及第二期高血壓的淨力； 3)將各階段別的危險因子考慮至上述各個轉移速率，並計算至第一期及第二期高血壓的淨危險分數； 4)利用上述危險分數將族群進行危險分層； 5)依據不同危險分層，建議不同介入策略，包括生活型態改變及強化病識感介入 6)利用以上結果進行單一預防策略(包含大規模族群策略、高血壓前期介入及合併二者)及個人化預防策略相對於一般高血壓照護的經濟評估。 材料及研究方法: 7)本研究對於高血壓流行病學及用以估計高血壓疾病自然史之資料來自於基隆社區整合式篩檢計畫。前者資料起迄為1999年至2015年，後者資料則為1999年至2002年之數據。 8)本研究利用四階段馬可夫廻歸模型所估算之每個疾病轉移廻歸係數以計算不同疾病進程之危險分數。而進展到高血壓前期或其他高血壓階段的淨風險分數是透過三個惡化轉移的廻歸係數減去從高血壓前期回歸到正常的廻歸係數來計算的。根據這種四階段馬可夫廻歸模型的結果繪製從高血壓前期，高血壓一期，最後到二期高血壓之個人化動態曲線。 9)本研究以馬可夫分析決策模型進行確定性和機率成本效益分析，以模擬單一預防策略和個人預防化策略的成本和效果。經濟評估為基於單一保險人(single payer)之觀點並考量折現率為3%。 結果: 社區資料顯示女性停經前高血壓期的盛行率及發生率均較同年齡層的男性為低，然而停經後兩者則不再有差異，四階段高血壓疾病自然病史估計結果證明了上述現象是因為女性在停經前從高血壓前期回復至正常血壓的轉移速率(0.7527, 95% CI: 0.6652, 0.8401)高於男性(0.3763, 95% CI: 0.3400, 0.4125)，而從高血壓前期往第一期高血壓的轉移速率則較低(女：0.4525, 95% CI: 0.3680, 0.5370，男：0.6626, 95% CI; 0.5646, 0.7606))，但這樣性別的差異50歲之後就不再明顯。考量危險因子之後，我們可進一步計算各個階段的危險分數，例如男性當中由正常血壓至高血壓高期轉移的危險分數為： 0.5557×exp{0.2121×(if 40≤age≤49)+0.3759×if(50≤age≤59)+0.5950×(if60≤age≤69)+0.7319×(if 70≤age≤79-0.2134×(ifhigh education)+0.6230×(if BMI≥25)+0.2140×(if waist≥90 cm)+0.1967×(if AC≥110 mg/dL)+0.1187×(if TC≥200 mg/dL)-0.3880×(if current smoker)-0.1440×(if with HTN family history) +0.0879×(if UA≥7 mg/dL)} 同理，從高血壓前期至第一期高血壓、第一期至第二期高血壓、高血壓前期回復至正常血壓的其他三種轉移亦以類似方式發展的風險分數。 經濟評估分析顯示，單一預防策略相對於一般高血壓照護在增值成本效益比（ICER）分別為每增加一個生活品質調人年在大規模族群策略的$103,282，針對高血壓高期介入的$78,019及結合前述兩項的$183,633。而利用危險分層進行的個人化預防策略則是-$4,170，亦即個人化預防策略相對於一般高血壓照護為優勢策略。從機率性成本效益分析的結果顯示，若願付成本在$300,000、$200,000及$400,000以上時，分別可使大規模族群策略、高血壓高期介入及合併二者的單一預防策略相對於一般照護具有90%以上具成本效益的機率，而個人化預防策略則是具成本效益則是不論付費意願為何之下幾可篤定。個人化預防策略相對於任一種單一預防策略也機乎均為優勢策略。 結論: 本研究運用四階段高血壓自然病史個人化模型發展單一或個人化預防策略之健康經濟決策模型，比較單一預防策略，個人化高血壓防治策略為節省成本策略。此結果對於個人化精準醫療之高血壓預防決策具相當助益。
並列摘要
Introduction Hypertension is one of major causes responsible for the risk of cardiovascular disease, cerebrovascular diseases, and chronic kidney disease and their related complications and deaths. Prevention of hypertension through different pharmacological and non-pharmacological intervention based on awareness program or population-wide approach cannot be over-emphasized. While population-wide approach through life-style modification and high-risk approach through pharmacological intervention have been envisaged in previous studies, the recently proposed personalized medicine to combine both approaches for prevention of hypertension has increasingly gained attention and has been scarcely addressed. There is lacking of economic appraisal of personalized preventive strategies partly because of insufficient information on epidemiological profiles and the disease natural history of hypertension (normal pre-hypertension stage 1 hypertension stage 2 hypertension) with the incorporation of state-specific covariates and partly because of not considering how to assign various preventive strategies according to different risk groups. Aims The objectives of this thesis, using the empirical data from community-based study, are therefore to 1)elucidate prevalence and incidence of pre-hypertension, stage 1 hypertension, and stage 2 hypertension by gender and age group (< 50 years and ≥ 50 years); 2)apply the previously developed four-state disease natural history Markov model to quantify the risk of regression from pre-hypertension to normal, the progression to pre-hypertension, stage 1 and stage 2 hypertension, and the net force of progression to stage 1 and stage 2 after considering regression and progression; 3)superimpose state-specific risk factors into each transition to develop four sets of risk score for the four corresponding transitions (normal pre-hypertension, pre-hypertension stage 1 hypertension, stage 1 to stage 2 hypertension, and pre-hypertension normal) by using the previously developed Markov regression model and also the net risk score for progression to stage 1 or stage 2; 4)classify risk groups by decile distribution of four sets of risk score and also the net risk score; 5)assign various intervention strategies from life style modification, awareness, and poly-pill prescription according to various risk groups; 6)develop Markov analytical decision model with various intervention strategies in comparison with no intervention based on (1)-(4) to do economic appraisal for universal strategies and personalized preventive strategies with the combination applied to different risk groups. Material and methods 7)Data used for delineating epidemiological profiles and estimating the transition parameters governing four-state disease natural history of hypertension were derived from Keelung community-based integrated screening (KCIS) program. The former was based on data from 1999 to 2015 whereas the latter was merely based on data from 1999 to 2002. 8)The disease natural history model based on the previously four-state Markov regression model was used to calculate the risk score using regression coefficients of each transition for the classification of the underlying population into risk groups in decile distribution. The net risk score for progression to stage 1 or stage 2 was calculated by the regression coefficients of three progressive transitions minus that of regression from pre-hypertension to normal. Personalized dynamic curves of the evolution from pre-hypertension, through stage 1 hypertension, and finally to stage 2 hypertension were plot on the basis of the results of this four-state Markov regression model. 9)Analytical Markov decision model was develop to model cost and effectiveness of universal strategy and personalized strategy using deterministic and probabilistic cost-effectiveness analysis. The perspective was based on single-payer viewpoint and the discount rate of 3% was used. Results The estimated prevalence and incidence of pre-hypertension, stage 1, and stage 2 hypertension found females had lower prevalence and incidence of prehypertension than males before menopause but both estimates were identical for both males and females after menopause. The parameters estimated from the disease natural history of four-state Markov model by gender indicated the lower incidence and prevalence for females before menopause was attributable to a higher likelihood of regression from pre-hypertension to normal and lower chance of progression toward stage 2 hypertension with the order of annual transition rate being 0.7527 (95%CI: 0.6652, 0.8401) and 0.3763 (95%CI: 0.3400, 0.4125), respectively for females compared with the corresponding figures being 0.4525 (95%CI: 0.3680, 0.5370) and 0.6626 (95%CI; 0.5646, 0.7606) for males, respectively. However, such a disparity in gender difference disappeared for the old age group. The risk score equation for progression from normal to prehypertension among male based on regression coefficients is expressed as follow 0.5557×exp^{0.2121×(if 40≤age≤49)+0.3759×if(50≤age≤59)+0.5950×(if60≤age≤69)+0.7319×(if 70≤age≤79-0.2134×(ifhigh education)+0.6230×(if BMI≥25)+0.2140×(if waist≥90 cm)+0.1967×(if AC≥110 mg/dL)+0.1187×(if TC≥200 mg/dL)-0.3880×(if current smoker)-0.1440×(if with HTN family history) +0.0879×(if UA≥7 mg/dL)} The risk score equations are similar for other 3 transitions. Regarding economic appraisal, the incremental cost-effectiveness ratios (ICER) were $103,282, $78,019, and $183,633 per quality adjusted life year (QALY) gained for universal population-wide approach, pre-hypertension prevention, and combined the two, respectively, and -$4,170 per QALY gained for personalized preventive strategy compared with usual care. The probabilities of being cost-effective reached 90% for universal approach when the ceiling ratio of willingness to pay were $300,000, $200,000, and $400,000 for population-wide approach, pre-hypertension prevention, and combined strategy, and almost sure for personalized preventive strategy. The corresponding analysis for personalized strategy in comparison with universal approach was still cost-saving. Conclusions Personalized four-state disease natural history of hypertension model was developed to support health economic decision model for economic appraisal of universal and personalized preventive strategy. The main finding is that personalized preventive strategy is cost-saving. The results are very useful for health decision-making based on precision preventive medicine of hypertension.