全面性新生兒聽力篩檢之隨機成本效用分析

Abstract

研究背景 先天性聽力缺損發生率在世界各國約為千分之 1~3，遠高於其他新生兒先天性疾病發生率。罹患新生兒聽力缺損的病童，若無早期發現及矯治，其語言及非語言溝通能力將受影響，導致學習不良及社會適應障礙，生活品質終其一生將受持續性影響，亦將影響社會財政上的負擔。因此各國開始探討全面性新生兒聽力篩檢之可行性；然而實施全面性大規模篩檢將牽涉龐大之經費支出，如何對全面性新生兒聽力篩檢策略進行成本效益評估，對衛生政策決策者而言十分重要。 研究目的 本論文研究之主要目的是分別以馬可夫決策樹決定型模式及馬可夫決策樹隨機型模式成本效用分析方法，評估三種不同的全面性新生兒聽力篩檢策略。 研究方法 本研究針對臺灣地區新生兒世代共 20 萬人，研究設計以馬可夫決策樹模式針對全面性新生兒耳聲傳射法篩檢、全面性新生兒自動聽性腦幹反射法篩檢及不實施全面性新生兒聽力篩檢三種篩檢策略進行成本效用分析。有關效用及成本分析參數的來源主要來自文獻查證。效用評估乃利用聽力缺損之臨床語言測驗評估進行效用折算。另外考慮參數的不確定性，我們選用適合的統計分佈，運用蒙地卡羅模擬，進行5000 次的電腦模擬。結果的呈現是比較不同篩檢策略之下之增加成本效用比(incremental cost-utility ratio, ICUR)、成本效用接受曲線(acceptability curve)及淨利益值(net benefit)。 研究結果 成本效用分析結果發現不論是以OAE或aABR進行全面性新生兒聽力篩檢，相較於不進行聽力篩檢，兩種篩檢策略均是節省成本的。若以OAE 進行全面性新生兒聽力篩檢，相較於無篩檢，可節省27.5 美元的成本，且可多獲得0.0017的效用值，其增加成本效用比為-4800.89 (-24475.83, 15351.54)。實施aABR 篩檢策略，與無篩檢策略比較，可節省14.25 美元且多獲得0.031 的效用值，其增加成本效用比為-4111.23(-22540.02, 13774.97)。若實施aABR 篩檢策略，與OAE 篩檢策略比較需多花費13.24 美元，但多獲得0.0014 的效用值，其增加成本效用比為6273.2(-15421.2, 28936.5)。以機率性敏感度分析則發現，在成本效用接受曲線分析中，與不篩檢策略相較，aABR 有高達91%的機會是具成本效用的；且aABR 策略較OAE 策略有較高之增加淨利益。 結論 不論以 OAE 或aABR 篩檢工具進行全面性新生兒聽力篩檢，相較於不進行新生兒全面性聽力篩檢，都是節省成本的。本研究結果支持目前臺灣地區所推行利用 aABR 進行全面性新生兒聽力篩檢之策略，以社會觀點而言，基於實證醫學的基礎，本研究支持這個政策應持續執行。

Background Congenital hearing impairment is one of the most prevalent congenital anomaly, afflicting 1 to 3 per 1000 newborn infants. Without early recognition and intervention, patients’ language ability would be delayed, leading to suboptimal school performance and production loss, which further causing significant socio-economic cost. Universal newborn hearing screening (UNHS) has been suggested for early identification worldwide. However, enormous costs would be involved for a large-scale newborn hearing screening program. The grasp of the balance between utility and costs for the implementation of such universal newborn hearing screening is of importance to health decision-maker. Aim The objective was to propose a Markov decision tree model to perform deterministic and probabilistic cost-utility analysis of universal newborn hearing screening in Taiwan. Material and methods A Markov decision tree model was framed with three strategies: UNHS with otoacoustic emission (OAE) test, automatic acoustic brainstem reflex (aABR) test, and no UNHS (natural history model). The target population was all newborn babies in health insurance system in Taiwan. The parameters for analysis were derived from literatures. The utility was evaluated using scores of Clinical Evaluation of Language Fundamentals (CELF) test. By assigning the specific statistical distribution to each parameter, Monte Carlo simulation with 5000 trials was therefore applied to capture the parameter uncertainty. We then present the outcome in terms of incremental cost-utility ratios (ICURs), acceptability curve, and incremental net benefit (INB) under varying prevention strategies. Results Cost-utility analysis reveals UNHS with OAE and aABR were the two cost-saving strategies compared with no UNHS. For a hypothetical population of 200,000 newborns, the incremental cost-utility ratios for OAE and aABR against the no screening strategy were calculated as $-4800.89 (-24475.83, 15351.54) and $-4111.23(-22540.02, 13774.97) for per utility gained, respectively. The incremental cost-utility ratio for aABR against OAE was calculated as $6273.2(-15421.2, 28936.5) for per utility gained. aABR strategy demonstrates a up to 91% higher probability of being cost-utility compared with no screening demonstrated in acceptability curve, and higher incremental net monetary benefit compared with OAE. Conclusion Compared to no screening, UNHS with OAE and aABR are both cost-saving. aABR is the most economic one. These findings, although they are still uncertain, suggest a nationwide universal newborn screening for hearing loss with aABR, from societal viewpoint, can be continued under the evidence-based principle.