哈爾濱醫(yī)科大學(xué)流行病學(xué)教研室

1、哈爾濱醫(yī)科大學(xué)流行病學(xué)教研室 Department of epidemiology ，Harbin Medical University,篩檢,Screening,疾病自然史（Natural History of Disease）,目 錄,概述 篩檢試驗(yàn)的評價(jià) 篩檢效果的評價(jià),Introduction Evaluation of screening testEvaluation of effectiveness,Outl

2、ine,,篩檢是運(yùn)用快速、簡便的試驗(yàn)、檢查或其他方法，將健康人群中那些可能有病或缺陷，但表面健康的人、同那些可能無病者鑒別開來,Screening is defined as the use of quick and simple testing procedures to identify and separate persons who are apparently well, but who may be at risk of a

3、 disease, from those who probably do not have the disease.,第一節(jié) 概述Introduction,篩檢試驗(yàn)流程圖 Flow diagram of screening test,,,篩查方法的評價(jià)與選擇程序 Procedures of evaluation and selection for screening test,金標(biāo)準(zhǔn) (gold standard),目標(biāo)人群T

4、argetpopulation,,病人Patient,非病人Non-patient,,,,,,,,,,,,,,,待評價(jià)篩查方法Screening to be evaluated,,,,,評價(jià)指標(biāo)Evaluation indicator,＋,－,＋,－,,,篩檢試驗(yàn)Screening test,識別健康人群中未被發(fā)現(xiàn)的某病患者或可疑患者，或高危個(gè)體Screening is the detection of patients,

5、suspected patient and high risk of the disease in healthy population,篩檢試驗(yàn)Screening test,問卷、常規(guī)體格檢查、物理學(xué)檢查、實(shí)驗(yàn)室檢驗(yàn)、分子生物學(xué)技術(shù)Questionnaire, routine physical examination, physical examination, laboratory tests, molecular biolog

6、y techniques,根據(jù)所用的篩檢方法的數(shù)量多少單項(xiàng)篩檢多項(xiàng)篩選,篩檢的分類Categories of screening,Based on the on the quantity of screening methodsSingle screening Multiple screening,根據(jù)篩檢對象的范圍不同人群篩檢 目標(biāo)篩檢,篩檢的分類Categories of screening,Based on th

7、e scope of screening subjects Mass screening Targeted screening,簡單性 廉價(jià)性 快速性 安全性 可接受性,篩檢試驗(yàn)應(yīng)具備五個(gè)特征Five characteristics that screening test should have,Simple Cheap Quick Safe Acceptable,一級預(yù)防 二級預(yù)防疾病的自然史流

8、行病學(xué)檢測,篩檢的目的,Primary prevention Secondary prevention Natural history of disease Epidemiological surveillance,Purposes of screening,篩檢的疾病是當(dāng)?shù)匾粋€(gè)重大的公共衛(wèi)生問題具備有效的治療或預(yù)防方法,應(yīng)用篩檢的原則 Principles for the application of screening,

9、The disease being screened should be a major problemAcceptable treatment or prevention methods should be available for individuals with diseases discovered in the screening process.,有進(jìn)一步確診的方法與條件該病的自然史明確,A suitable and

10、effective test or examination for the disease should be available The natural history of the disease should be adequately understood,應(yīng)用篩檢的原則 Principles for the application of screening,該病具有較長的潛伏期或領(lǐng)先時(shí)間預(yù)期有良好的篩檢效益,The di

11、sease should have a recognizable course, with early and latent states of the disease being identifiableExpected good efficiency of screening should be available,應(yīng)用篩檢的原則 Principles for the application of screening,倫理學(xué)問題

12、Ethical Issues,尊重個(gè)人意愿 Respect for person’s will 有益無害 Useful and harmless 公正 Fairness,一、篩檢試驗(yàn)的評價(jià)方法 （Evaluation methods）確定金標(biāo)準(zhǔn) (Determine Gold Standard)選擇研究對象 (Chose Study Subjects)樣本量的估計(jì) (Estimate Sample

13、 Size) 整理評價(jià)結(jié)果 (Sort out evaluation result),第二節(jié) 篩檢試驗(yàn)的評價(jià)Evaluation of Screening test,金 標(biāo) 準(zhǔn),所謂“金標(biāo)準(zhǔn)”是指當(dāng)前臨床醫(yī)學(xué)界公認(rèn)的診斷疾病的最可靠方法,Gold Standard,The best available test that is universally recognized by clinical medical circle

14、s is called the GOLD STANDARD .,病例組 納入各種類型的病人 對照組 最好選擇需要與研究疾病鑒別的其他病人，即所選擇的對照組與病例組具有許多相似的條件,研究對象 Study Subjects,Case group Include appropriate spectrum of patients Control group Need to be distinguished t

15、o the disease studied, that is the control has co-morbid conditions with case,樣 本 量 Sample Size,影響樣本量的因素待評價(jià)篩檢試驗(yàn)的靈敏度待評價(jià)篩檢試驗(yàn)的特異度顯著性水平（α）允許誤差（d）,Factors determine sample sizeSensitivity of screening to be evaluated S

16、pecificity of screening to be evaluated Significance level（α）Allowable error（d）,當(dāng)待評價(jià)篩檢試驗(yàn)的靈敏度和特異度均接近50％Both sensitivity and specificity of screening to be evaluated approach 50%,,,n: the number of abnormal (diseased)

17、 or normal (healthy) patients in the study.δ: admissible error. P: the estimation of sensitivity and specificity for testing. ua : the u value for the cumulative probability equal to a/2.,當(dāng)待評價(jià)篩檢試驗(yàn)的靈敏度和特異度小于20%或大于80%

18、Both sensitivity and specificity of screening to be evaluated less than 20% or greater than 80%,,,,,本來有病，被篩檢為陽性,本來有病，被篩檢為陰性,本來無病，被篩檢為陽性,本來無病，被篩檢為陰性,評價(jià)結(jié)果 Evaluation Results,評價(jià)結(jié)果 Evaluation Results,真實(shí)性（validity） 可靠性（rel

19、iability） 預(yù)測值（predictive value）,二、篩檢試驗(yàn)的評價(jià)指標(biāo),Evaluation Index,靈敏度（sensitivity）假陰性率（false negative rate）,真實(shí)性（validity）,False Negative Rate,Likelihood of a negative result when patient actually has disease,False Negativ

20、es,Sensitivity,,,,Likelihood of a negative result when patient actually has disease,False Negative Rate,False negative rate increases with decreased sensitivity,False Negative Rate,Likelihood of a negative result when pa

21、tient actually has disease,特異度（specificity）假陽性率（false positive rate）,真實(shí)性（validity）,False Positive Rate,Likelihood of a positive result when patient does not have the disease,False Positive Rate,Likelihood of a positive

22、 result when patient does not have the disease,False Positive Rate,Likelihood of a positive result when patient does not have the disease,False positive rate increases with decreased specificity,正確指數(shù)（Youden’s index）

23、 ＝（特異度＋靈敏度）－1 ＝1－（假陽性率＋假陰性率）,真實(shí)性（validity）,陽性似然比(positive likelihood ratio，＋LR）陰性似然比(negative likelihood ratio，－LR）,真實(shí)性（validity）,真實(shí)性（validity）,一致率（agreement rate），符合率,一致率,真實(shí)性（validity）,取值范圍：-1和

24、＋1之間,Range: between -1 and +1,Kappa值 = 實(shí)際一致性/非機(jī)遇一致性觀察一致性(Observation consistency) 機(jī)遇一致性(Opportunity consistency) 非機(jī)遇一致性(Non-opportunity consistency)實(shí)際一致性(Actral consistency),真實(shí)性（validity）,標(biāo)準(zhǔn)差與變異系數(shù) coefficient of vari

25、ance，CV符合率（agreement rate）,Agreement rate,可靠性（reliability）,影響可靠性的因素,Determinants of reliability,☉B(tài)iological variations of subjects☉Variations in different observers☉Variations in different lab conditio

26、n,受試對象生物學(xué)變異觀察變異實(shí)驗(yàn)室條件,陽性預(yù)測值(positive predictive value) 陰性預(yù)測值(negative predictive value),預(yù)測值（predictive value）,靈敏度、特異度、患病率與預(yù)測值的關(guān)系,Relationship between sensitivity, specificity, prevalence and predict value,,,sensitivi

27、ty and specificity in different blood sugar level,,,blood sugar SN SP blood sugar SN SP（mg/100ml） （﹪） （﹪） （mg/100ml） （﹪） （ ﹪）,,80 100.

28、0 1.2 150 64.3 96.1 90 98.6 7.3 160 55.7 98.6 100

29、97.1 25.3 170 52.9 99.6 110 92.9 48.4 180 50.0 99.8 120

30、88.6 68.2 190 44.3 99.8 130 81.4 82.4 200 37.1 100.0 140 7

31、4.3 91.2,,,,,,,Relationship between sensitivity and specificity,靈敏度與特異度的關(guān)系,,,,,非病人,病人,某項(xiàng)診斷標(biāo)準(zhǔn)的數(shù)值,截?cái)帱c(diǎn)Threshold,,三、篩檢試驗(yàn)截?cái)嘀档拇_定Determination of screening cut-off point,,,理想情況Ideal situation,人數(shù)Number,Normal,Di

32、sease,,,,非病人Normal,病人Abnormal,某項(xiàng)診斷標(biāo)準(zhǔn)的數(shù)值,Cut-off,,,,三、篩檢試驗(yàn)截?cái)嘀档拇_定Determination of screening cut-off point,實(shí)際情況Actual situation,人數(shù)Population,Screening tests use an arbitrary cut-off to separate two populations (normal

33、 vs. abnormal),,,,,a,c,b,d,a,c,d,b,Where cut-off is placed determines sensitivity and specificity,Screening tests use an arbitrary cut-off to separate two populations (normal vs. abnormal),,,Where cut-off is placed dete

34、rmines sensitivity and specificity,Normal,Abnormal,Few False NegativesHigh Sensitivity,,,- result,+ result,Screening tests use an arbitrary cut-off to separate two populations (normal vs. abnormal),,,Where cut-off is p

35、laced determines sensitivity and specificity,Normal,Abnormal,Few False PositivesHigh Specificity,,,- result,+ result,Screening tests use an arbitrary cut-off to separate two populations (normal vs. abnormal),,,Where cu

36、t-off is placed determines sensitivity and specificity,Normal,Abnormal,Trade off between False – and False +,,,- result,+ result,Screening tests use an arbitrary cut-off to separate two populations (normal vs. abnormal

37、),,,Test with High Sensitivity,Normal,Abnormal,A negative result allows you to rule-out whatever you are testing,,,- result,+ result,SnNout,Screening tests use an arbitrary cut-off to separate two populations (normal vs

38、. abnormal),,,Test with High Specificity,Normal,Abnormal,A positive result allows you to rule-in whatever you are testing,,,- result,+ result,SpPin,Screening tests use an arbitrary cut-off to separate two populations (n

39、ormal vs. abnormal),用真陽性率和假陽性率作圖得出的曲線，可用于確定篩檢試驗(yàn)的截?cái)嘀怠?A receiver operating characteristic (ROC), or simply ROC curve, is a graphical is a graphical plot of the sensitivity, or true positives, vs. (1 ? specificity), or fa

40、lse positives, for a binary classifier system as its discrimination threshold is varied.,受試者工作特性曲線 Receiver operator characteristic curve, ROC,特異度,血糖測定診斷糖尿病的ROC曲線,ROC curve for diagnosis of diabetes using blood glucose,

41、第三節(jié) 篩檢效果評價(jià) Evaluation of effectiveness,一、收益選擇患病率高的人群選用高靈敏度的篩檢試驗(yàn)采用聯(lián)合試驗(yàn)串聯(lián)并聯(lián),YieldTo select the population with high prevalence rate To select the test with high sensitivity To design a combine test Series testP

42、arallel test,二、篩檢的生物學(xué)效果評價(jià)Biological effectiveness evaluation,病死率、死亡率、生存率,篩檢組Screened group,未篩檢組Unscreened group,,比較,,Fatality rate、mortality rate、survival rate,三、篩檢的衛(wèi)生經(jīng)濟(jì)學(xué)效果評價(jià)Health-economic effectiveness evaluation,成

43、本效果分析 Cost-effectiveness analysis 成本效益分析 Cost-benefit analysis 成本效用分析 Cost-utility analysis,領(lǐng)先時(shí)間偏倚病程長短偏倚志愿者偏倚,Lead time bias Length bias Volunteer bias,第四節(jié) 篩檢中的偏倚及其控制Bias and Control,領(lǐng)先時(shí)間偏倚 （lead ti

44、me bias）,領(lǐng)先時(shí)間：篩檢診斷時(shí)間和臨床診斷時(shí)間之差,Lead time: Interval between the diagnosis of a disease at screening and the usual time of diagnosis (by symptoms),,,Diagnosis by screening,Diagnosis via symptoms,,Lead Time,Assumes surviv

45、al is time between screen and deathDoes not take into account lead time between diagnosis at screening and usual diagnosis.,,,Diagnosis by screening in 1994,,Deathin 2008,,Survival = 14 years,認(rèn)為延長的生存時(shí)間是篩檢診斷時(shí)間與死亡時(shí)間之差 沒

46、有考慮篩檢診斷時(shí)間和臨床診斷時(shí)間之間的領(lǐng)先時(shí)間,領(lǐng)先時(shí)間偏倚 （lead time bias）,,,,Diagnosis by screeningin 1994,Usual time of diagnosis via symptoms in 1998,,Lead Time 4 years,,Deathin 2008,,True Survival = 10 years,,,,,Survival = 14 years,,,領(lǐng)先時(shí)間

47、偏倚 （lead time bias）,病程長短偏倚 （length bias）,OBiological onset of disease,YSymptomsBegin,DDeath,PDisease detectable via screening,,,,篩檢Screening,時(shí)間Time,惡性程度高Aggressive,惡性程度低Less aggressive,,,病程長短偏倚 （length bias）,