- 學位論文
產前唐氏症危險度估計及篩檢之經濟評估
Risk Estimation of Down’s Syndrome and Economic Evaluation for Its Screening
摘要
唐氏症 (21號色體三染色體症)，是最常見的染色體異常，也是最常被確認的先天性智能發育遲緩的原因，唐氏症的發生率在每1000個活產兒為1.2-1.7個，盛行率在每1000個活產兒及死嬰中為1.5個，因為沒有治療的方法，產前診斷及中止妊娠是目前唯一的預防方式． 唐氏症的發生原因及危險因素已被廣泛地討論，在一般族群中，高齡孕婦是主要危險因素．1970年代起，開始推廣高齡孕婦接受羊膜穿刺及胎兒染色體分析來產前診斷唐氏症兒，然而用年齡(≧35歲) 為標準的篩檢方法最多只能診斷出30%的唐氏症兒(族群中5%的高齡孕婦)，用孕婦血清之生化標記值做產前唐氏症兒篩檢的方法是1980年代發展出來的，其中甲型胎兒蛋白，人類絨毛性腺激素及動情三醇是懷孕中期最常用的生化標記，懷孕中期用各種不同的生化標記加上孕婦年齡，在不同的危險值判定標準(1:200-1:300)及篩檢偽陽率之下可篩檢出35-70%的唐氏症兒． 在台灣，唐氏症兒的發生機率是0.118%，用兩種生化標記(甲型胎兒蛋白及人類絨毛性腺激素)的孕婦血清篩檢方法在1992年引進台灣，目前已被孕婦普遍的接受，血清篩檢的影响已表現在年輕孕婦的唐氏症兒活產對死產的比率下降之中． 然而，目前的血清篩檢危險度估計方法複雜難懂，最好能發展出一種較簡單的方法去估計危險度，現行統一的危險度標準，在小於35歲的年輕孕婦血清篩檢敏感度不夠好，改變危險度標準，可以得到較好的篩檢敏感度，此外，經濟評估方面，最常用的双重及三重生化標記篩檢方法，其成本效益尚有爭論，血清篩檢對孕婦的好處也缺乏由產婦觀點的評估． 這個研究主要的目的如下： (1)應用邏輯思迴歸發展一個較簡單的計分方法做血清篩檢的危險度 計算． (2)依照第(1)部分的模式，用Receiver Operating Characteristics 曲線決定年輕孕婦 (小於35歲) 血清篩檢的最 佳危險度判斷標準． (3)用第(1)、(2)部分的模式，做不同的篩檢方法的平均及邊際成本 效益分析． (4)調查孕婦對篩檢的付費意願及影響因素． 此研究對象是1993年10月到2002年6月之間到一個醫學中心做唐氏症篩檢的孕婦，有懷孕相關資料及各種血清標記值，作付費意願調查時，是在門診發問卷． 應用孕婦年齡及生化標記值，我們建立了單變項及多變項的邏輯思迴歸模式，用Spiegelhalter-Knill-Jones (S-KJ) 模式建立一個簡單的計分方法去估計危險度，計算出不同危險度判定下的敏感度及精確度，用Receiver Operating Characteristics (ROC) 曲線決定年輕孕婦的最佳危險度判斷標準．在成本效益分析中，應用邏輯思迴歸算出的敏感度及精確度，加上各種篩檢的成本來算出平均及邊際的成本效益比值．再用不同的成本，敏感度及羊膜穿刺接受度計算平均及邊際的成本效益比值的變化．另外，用單變項及多變項線性迴歸分析模式分析影響付費意願的因素． 主要結果如下： (1)用S-KJ模式建立的簡單計分方法，可以有效的估計孕婦懷有唐氏 症兒的危險度． (2)用ROC曲線分析得到的年輕孕婦的最佳危險度判定值為1：499， 在此標準下，偽陽性為17.8%，敏感度為90.0%． (3)加上動情三醇的三重生化標記的血清篩檢比雙重生化標記的篩檢 成本效益更好． (4)孕婦願意付平均1911元作血清唐氏症篩檢，這個數值比目前的篩 檢費高很多．付費意願與家庭收入及孕婦年齡呈正向相關． 由本研究可以得知，應用邏輯思迴歸去估計危險度的母血唐氏症篩檢是有效的，用S-KJ 模式可以建立一個簡單的計分方法去估計危險度，用ROC曲線分析可以得到年輕孕婦的最佳危險度判定值，三重生化標記篩檢比雙重生化標記篩檢更有效且成本效益更好，家庭收入較多及高齡孕婦願意付較多費用作唐氏症篩檢．
並列摘要
Down’s syndrome (trisomy 21) is the most common chromosomal anomaly worldwide, and the most commonly recognized genetic cause of mental retardation. The incidence of Down’s syndrome (DS) is 1.2-1.7 per 1000 live born, and the total prevalence was about 1.5 per 1000 live birth and fetal deaths. Because there is no treatment for this condition, prevention with early prenatal diagnosis and termination of the pregnancy is the only available method. Etiology and risk factors have been discussed from many aspects. Advanced maternal age is the only dominant risk factor in general population. Prenatal diagnosis with invasive procedures and cytogenetic analysis has been offered to pregnant women based on maternal age (≧ 35y/o) since 1970’s in many developed countries. However, less than 30% DS fetuses would be identified by this screening policy in a population with 5% pregnant women above 35 years old. The maternal serum screening for DS with biomarkers was introduced in 1980s. The alpha-fetoprotein (AFP), human chorionic gonadotrophin (hCG), and unconjugated estriol (uE3) were the most commonly used markers in the second trimester. With variable combination of biochemical markers with maternal age at second trimester, the detection rates of serum screening ranged from 35 to 70% at different cut-off points (1:200-1:300) and false positive rates. In Taiwan, the reported incidence of Down’s syndrome is 0.118%. Maternal serum screening with double test (AFP and hCG) was introduced in Taiwan in 1992, and was generally undertaken by pregnant women now. The impact was presented in the decreased ratio of live-birth to stillbirth of Down’s syndrome fetuses in young women thereafter. However, Risk estimation for Down’s syndrome in serum screening with maternal age and multiple serum biomarkers is usually complicated and computationally intensive. It is necessary to construct a relatively simple and flexible scoring system. The detection rate of maternal serum screening in younger pregnant women is relatively low. It can be improved by changing the cut-off point. Double test and triple test have been the two most commonly used methods in mid-trimester. The cost-effectiveness with adding uE3 is controversy. Much discussion about performance of the screening has been provoked, with little known about how women benefit from it. We intended to adopt WTP to assess the value pregnant women attach to the benefits related to screening for Down’s syndrome. The major aims of this thesis are: (1)To develop a risk estimation method of serum screening using logistic regression with simple scoring system. (2)To determine the optimal cut off risk for maternal serum screening in young women (<35y/o) with Receiver Operating Characteristics Curve using method in (1). (3)To perform a cost-effectiveness analysis for comparison of different serum screening methods using method in (1) and (2). (4)To identify the willingness-to-pay for screening of pregnant women and the influencing factors. The study population was pregnant women who visited a medical center in Taipei for prenatal care during October 1993 to June 2002. For each woman, basic data about this pregnancy and measurements of biomarkers were collected. For willingness-to-pay evaluation, questionnaires concerning willingness-to-pay for Down’s syndrome screening were collected at out-patient department of this center. Univariate and multivariate logistic regression models were established with maternal age, and biomarker levels. A simple scoring system with Spiegelhalter-Knill-Jones (S-KJ) approach was developed. Sensitivity and specificity at different cut off points were calculated. Receiver Operating Characteristics (ROC) curves for different models were plotted to determine the optimal cut-off value. For cost-effectiveness analysis, costs of serum screening, amniocentesis and termination of pregnancy were determined. The average and incremental cost-effectiveness ratios were calculated for double and triple test. Sensitivity analyses for cost-effectiveness at different costs, detection rates, and amniocentesis up-taken rates were performed. For analysis of willingness-to-pay for antenatal screening for Down’s syndrome, univariate and multivariate linear regressions were performed to identify factors influencing the value of willingness to pay. The major results include: Firstly, a predictive model for Down’s syndrome pregnancy based on maternal age and serum levels of alpha-fetoprotein and human chromic gonadotrophin was developed using logistic regression. A simple scoring system for risk estimation of maternal serum screening was developed with S-KJ approach. The efficiency of this scoring system was validated. Secondly, the optimal cut off risk estimated from Receiver Operating Characteristics curve was 1:499 for women under 35y/o, with a sensitivity of 90.0% and false positive rate of 17.8%. Thirdly, adding uE3, the triple test was more cost-effective than double test in this series. Fourthly, the average amount of willingness-to-pay for serum screening of Down’s syndrome of pregnancy women was estimated to be 1911 NTD, which was more than the current fee for screening. In a multivariate linear regression analysis, the willingness to pay was positively correlated to family income and maternal age. In conclusion, predictive model for Down’s syndrome based on maternal age and serum levels of AFP and hCG was developed using logistic regression. The S-KJ scoring system is a simple, and efficient method for risk estimation. According to ROC curve analysis, the optimal cut-off risk for young pregnant women could be estimated. With higher sensitivity, triple test is more cost-effective than double test. Pregnant women with more family income and advanced age are willing to pay more for screening for Down’s syndrome.