診斷性試驗(yàn)Meta分析

1、Revman, Stata, Meta-disc 在診斷試驗(yàn)準(zhǔn)確性(DTA) 系統(tǒng)評(píng)價(jià)中的應(yīng)用文獻(xiàn)數(shù)據(jù)摘HProGRP與NSE對(duì)小細(xì)胞肺癌診斷價(jià)值的meta分析文中提取數(shù)據(jù)陽(yáng)性界值(pg/ml)TPFPFNTN作者國(guó)家研究方法盲法研究對(duì)象金標(biāo)準(zhǔn)病例數(shù)abababababSchnei徳國(guó)前連病2933131122der瞻續(xù)理85885632192Stiebe德國(guó)回病3143944421r顧理4194681883Molina西班前連病80502511754655牙瞻續(xù)理2319011474487Nissan以色前連病16482221618111列瞻續(xù)理298291193Shibay日本連病3549

2、74114622ama續(xù)理94910053435Lamy法國(guó)回是病24531711242399顧理511967570Takada日本回是 連病3276242321顧續(xù)理63323880832Yamagu日本是 連6050&187624444chi續(xù)2096786499Sun中國(guó) 病105021689165理059508Yang中國(guó) 病144644911276理46067325注：表中10個(gè)原始研究均使用酶聯(lián)免疫吸附測(cè)定法檢測(cè)陽(yáng)性界值；TP二真陽(yáng)性數(shù)；FP二假陽(yáng)性數(shù)；FN二假陰性數(shù)；TN二真陰性數(shù)a：ProGRP,b:NSE新建診斷試驗(yàn)準(zhǔn)確性(DTA)系統(tǒng)評(píng)價(jià)模板Recent FilesClos

3、ePJ 5avectri-s咼 Save As.定）My RGVIGV/S. Check Out.曰 Print.ctri-p:&! Published PDF preview.Exitf-ne j Edit B NewFormat占 Open.view loo is Table vyindovj 旦eip,匕 II l OCtrl-O世 Check In.聒 Hew Review V/izard二:： ：: ：?： :：，：，: ：?：:：童呈空因IIIMow Rovicv/ Wizardwmcn stage should tno reviowsiaftirr?maStago:O Iitie

4、 only ProtocolO Full reviewCancelFinish添加所有納入研究TablesD Characteristics of studies n Characteristics of ir 口 Characteristics of e 口 Characteristies of s Characteristics ofoD Summary offindings ta n Additi on al tablesFootnotesQ Add Study 眇 StudyPastecharacteristics ol卿 Studies and references 斗 Data a

5、nd analyses瞬 Properties. Notes.昌 Print.t Help Handbooktnotesdd Study聒 Hew Review V/izard二:： ：: ：?： :：，：，: ：?：:：童呈空因IIIS.udyrd.一Mn鉉 New Study Wizard :-兇. . . . .New Study Wizard阿同How sh ould the study De i dentified?LLJI-1Study ID: Sch neider 2002|Example: Cochran 2007CancelFinishNow Study Wizardin w

6、hich year was the study done*?Year; |2003CancelElnish此處對(duì)每篇文獻(xiàn)QUADAS2質(zhì)量特征進(jìn)行描述，以便探討異質(zhì)性來(lái)源及作表A. Risk Of BiasL，；、 C“、CZ：CR ：F7iL-viiaiaK-wiuui vi mwuuvu ouuuiu rivvviuu ： gGeneralPatient SelectionIndex Test 丨 Reference StandardFlow and TiminaSort By:Study 10User de Study ID use)Year of studyUseroetined or

7、der二 U$r denned 3 Heading:Add Comparativo Dofnain口 TaWes? 口 Characte fistics of stu dies? O Chari de sties of include- i stu c ?漪 Lamy 2000D PaDentSeiodionD Index TestD Reference Siandard0 Row and Timings Molina20095 潔 Nissan2009 s Schneider2003 s Shi Dama 2001 s SDeberi999 s 夠 3un 2005 s 茨 T3Kada 1

8、996 s Yamaguchi 1995 e Yang 2005 口 Characteristics oroxdudod siu數(shù)據(jù)分析里面添加所要研究的待評(píng)價(jià)診斷試驗(yàn)-? m Data and analysesB PaUfihaancMs 口 auduuiEdowPdlMl Sarong、粉 8a dfOXT Wv*e 09尿Epj D iRDm a BVm 尿如 回5： a尿爭(zhēng)3:壬 園沏“ *閔5小 F如2)譏 3$ a casecntioi liuoy ZacmrBs的血 MiftXWS tWUtKnQ0E如 tutfySt w3 VMW sa*if ciurftCMniucsctwG

9、Pmu ?H*-S5C5 and settngconcerns that tho included weienti and seeing do rvot nwteh :he review auosticn?T”：歩f 口 Data tables by 口 Data tables b*y D CovariatesD AnalysesQ Figuresa 鬆 Sources of suppo Feedback AppendicesQ Add TestH2 Renumber TestsQ Notes. 旦 Print.? Help0 HandbookNew Test Wizard ：. 因New T

10、est WizardWhat name should the test havevLTJIOJName:Euli Name:| ProGRPProGRP沁忖也卅抱扌旦診斷CancelEinish匚 User denned 2 Heading:I ? 斗 Data and analysesI ? Data tables by testI 1 ProGRPI 2NSE9 斗 Data and analyses O Data tablesI 1 ProGR| 2 NSE 口 Data tables D Covariates D AnalysesS Figures 鳴 Sources of supf2

11、Y test0 Add Test Data 0Edit Test 或 Delete Test 耆 Rename TestMove Up 竽 Move DownStuoTPFPFWTWscratwtMDs% ci spoony m% a)lUvrr 200035is2290.89 Q 64, OBI0.93(0096(h*na 200941g的2180.47P36.0 務(wù)0.時(shí)0 93.09$Mssan 20091M79415480.7770.08330.87(0 85.090(如 20032963113078 Q 62.0 匆0.950 90.09$3h5-fa 20017411402340.

12、65 p 55 0 740.95(0 02. OQSrS1oer 199911T2M970.8073. 08G0.98(0 03,100fSun 20057322232030.T2P62 0310.90(0 86,024：Faxada 1995BDG474690.53 P S4 0 710.990 97,1 0QYimaouchi 1525e$600.74 P.6e. 0370.91(7!Yano 2005的1?720.73 P.60. Oj 0.89 (SiTost 1 ProGRPTestHSESlxlyTPFPFNTNSen5tni)y(95%CILamy

13、 2000aa382恰07610 W.0 960.WP9l.O90i|M0lna 2009w193.04譏0 “.0刑0.釘P 75.歸Nian 2O091U6167708510 58.0 72)0.92 P 90. 0,沏吶2003id12竹113-04910 32.0 660.90 P 84.0951Sn-ema 200149lOi043(0 M. 0 5310.98P93.0 制aecec 伽11a4務(wù)07510 68.0 820.MP.90.0J9ftjn 200563433818Z08210 52.0 720.61 P 75.0TaKdoa 19 弼冷簽39062(0 53.071J

14、0.9510 92.0的YdXfll 餌519i858056(0 38.073)0.86 P 78.0刖vanfl 20054065-06310 50.07510.80 P 70.08”6 叵Q aFFWl SB3pecmcn/O5%CI)口 Data tables by te st j 1 ProGRPa j 2NSEs 口 Data tables by studyD Covariatesa F.qUn B AFN TNSen8mnty(95% Q)SpocrncRT I9JSC05r“ 20003018W2290C9 |05l. 0 81)093 ID B9.0 9CI4194& 2180

15、47 |0 36. 0 56006(D93.0Qfl2如n 200Q134704168Q?7 |0 ?0. 0 83)087 (0 85.0 201Sctr-der 200329091190 78 |0GZ 0 9(0095(0 90.0 981Shib卒耐 XO17411 40234065 10 55,0 74)0 06(0 X.O Ml-D&Uuli cotorO Siaci cok Dirfautt symbol Stuoy 勺n4 Default tanooSpactflcity ranaaLOW：HlQh：PafamaurLambdae&timirtdPafAfnatarEfitma

16、wEOT_ParamfiurSElEUOflCSfl；)口 0l5plF 95* prwUctKw reJoNe w Analysis WixardNaw ArtolyAifi WizardWhilch type of an alysls to you want to creatorType: Multiple tests analysisI I Analysao pairod data onlyI】Investigate sources or tieteroge neityStudyTPFPFNTNSensitivity (95% Cl)Specificity (95% Cl)Sensiti

17、vity (95% Cl)Specificity (95% Cl)Lamy 20003518162290.69 (0.54, 0.810.93 0.89. 0.96JMohna 2009419462180.47 0.36. 0.580.96 0.93, 0.98Nissan 200913479415480.77 (0.70. 0.830.87 (0.85. 0.90-Schneider 20032968 1190.78 (0.62. 0.90)0.95 (0.90. 0.98)-flShibayama 20017411402340.65 (0.55, 0.740.96 0.92. 0.98JS

18、Heber1999117229970.80 (0.73, 0.86J0.98 0.93, 1.00JHF-Sun 20057322282030.72 (0.62, 0.81J0.90 0.86. 0.94-4HTakada 1996806474690.63 (0.54, 0.71)0.99 0.97, 1.00)1Yamaguchi 19952569600.74 (0.56. 0.870.91 7)917Yang 200546720.73 (0.60. 0.83J0.89 0.80, 0.95J 卜I I I I I1 1 1 1 1 100.2 0.4 0.6 0.8 10

19、0.2 0.4 0.6 0.8 1NQxt aUa nealEinifihFoW p*CAnalyiis： 1 ProGRP“maftyCalcWQ Promeer5 HSROC mo4e prmeft*r5Q firranateporsmeec*?SROCDkrtCMiHJence ond prrdiclKx* 呦如耐 Qis*plaf sunimafy s w。蛤pu, mniEfv polM口 Oi$( (H 町 t5* Goonoence review設(shè)置參數(shù)若用戶對(duì)上圖窗口中的統(tǒng)計(jì)分析顯示的結(jié)果不滿意，可點(diǎn)擊右上角的屬性 按鈕 SR0C圖、森林圖和異質(zhì)性來(lái)源的參數(shù)進(jìn)行設(shè)置，并點(diǎn)擊Ap

20、ply使其生效，見(jiàn)下圖。StudyLamy 2000Molina 2009Nissan 2009Schneider 2003Shibayama 2001TPFP FNTN351S16229419462101347941296B1197A1140234Sensitivity (95% Cl)0.69 (O.b4. O.B1) )0.47 (0.36. 030.77 (0.7O. 0.83O.7 (0.62. O.9OJO.6!S (0.55. 0.7釗Specittclty (95% Cl)0.93 (O.B9. O.96J0.96 (0.93. O.9HJ0.87 (O.Bb. 0.90)O.

21、9b |O.96 0.98)0.96 (0.92 O 98)Sensitivity 95% Cl)Specificity (95% Cl)I -1-b-O 0.2 0.4I0.6 0.8 1I-1-1-1 -15O 0.2 0.-4 0.6 0.8 1亞組分析（假如，原文沒(méi)做）斗 Data and analyses 口 Data tables by testJ 1 ProGRP s I 2 NSE ? j ? NSE對(duì)小細(xì)胞癌診斷 嚴(yán) Lamy 2000 翳 Molin? 2009 翳 Nissan 2009Schneider2003Shibayama 2001 ? I |4ProGRPg)

22、翳 Stieoeri999Sun 2005納 Takada 1996 関 Vamaguchi 1995Yang 2005?弓and analyse as4 口 Data tables by test 卜 I Q ProGRP a | n NSE9 I 3 ProGRP.： 澱 Lamy 2000 尿 Molma 2009 fA Nissan 2009 談 3cTneider 20ProGRPCmm*ProGRPmmiStudyStietxsr 1999Sun 2005r akada 1996 Yamaqucrii 1995 Yang 2OObTPFPFNTNSensitivity 3% Cl)

23、Specialty 3% Cl)Sensitivity (95*% Cl)117229970.80 (0.73, 0.86O.9 (0.93. l.OOJ-7322282030.72 (0.62. O.81J0.90 (O.BG. 0.9-1 J O6474G90.63 (0.54. O.71J0.99 (0.97. 1.OOJ-2569600.74 (O.S6. 0.870.91 7-46917720.73 (0.60. 0.83)0.89 0.80. O.5JI -1-1-1 _IO 0.2 0.4 0.6 O.B 1Specificity (95*b ClO.A 0.6F

24、ifluresSourcoa o1 supportFeeaoackCancelNext finish25%50%75%100%Risk of BiasApplicability Concernsj J UnclearE3 New Figure Wizard . .1 A cumulative barplot of risk of bias and pplicaoility concorns across all studies.QUADAS-2偏倚表（圖）制作New Figure WizardWhich type of figure do you want to create?Figure T

25、ype:O Forest plotQ SROC plotRisk of bias and applicability concerns graphO Risk of s and applicability concerns summaryO Study flow diagram (PRISMA template)O Flow d.iagram (blank)O Other figurePatient SelectionIndex TestReference StandardFlow and Timing25%100%0%O|A summary taoie of review authors*

26、ratinQS； of risk of bias and applicability concerns for each studyNew Figure WirardNew Figure- WiraidWhich type of Tigure do you want to createFigure Type:O Forest plotO iROC PlotO Risk of bias and applicability concerns graph Risk of bias and applicability concerns summaryO Sjudy flow diagram (PRIS

27、MA template)O FIOVJ diagram (blankO Other figurecancelHext EinishRisk of BiasApplicability ConcernsLamyMolinaNissan200020092009phjpjes 卷 aJPSH 密 lsE 8急鬲總衛(wèi) gwlipueMOU P&WES eoja&ey val*PM ssapstfahidSun 2005Takada 1996Yamaguchi1995Yang 2005?Schneider 2003ShiDayama 2001Stieber 1999?7?77High7 Unclear I

28、 Low異質(zhì)性來(lái)源在DTA系統(tǒng)評(píng)價(jià)里面不能直接進(jìn)行似然比、診斷比值比的森林圖以及各指標(biāo)漏斗圖制作，但可以改變四個(gè)表數(shù)據(jù)模式或直接計(jì)算相關(guān)指標(biāo),添加入干預(yù)性系統(tǒng)評(píng)價(jià)模板中進(jìn)行制作及査看異質(zhì)性、發(fā)表偏倚（漏斗圖）。Statal2擬合雙變量混合效應(yīng)模型：midas命令葩 Sta?a/SC 12.0 - ResuitsJ * J ?j 一彳；一CommandOrUlReviewT P XResultsan 4-2Q Command_rc/ / ReviewCtdr3There are no items to_/ / /_/VariablesCtrl *4show.Statioicc/DatiPrope

29、rties3U5%C*M2 RdxlGraphViewerData EditorQrU8De-file EditorVariables ManagerFile Edit Data Graphics Statistic% UserNores: Station, XTA-PC56-4500B6-4eai 77645 OSA http:/wvw . atar .cHeipVariablwTXVariablelabelThere are no rtems to Show.i/v option or -set naxvar-i 5000 maxurjis variablessunning D:Doxnl

30、o d &olaStAraS212.0psaxl.doProperties VariableNameLabellyFormar_ OatM Editor (EcJrt) - UmtitlvclFile Edit View Data Tools 4 丄亠到 了卻廿yeartl2003k-5R5-xaurnoryearTPFPFMTMXschneiacr |2003 |2.02CSZS2sr 1 otiar921SnMolimm200034704:1Z 4G今Ni 2”2009296eIIS5Sint baya*na20017 A1140234e( (_a*ry3000aar2Z9”7-rakca

31、aA工67322zaZO5nYAmAQucr-lX99EsoC474G9oSun2OOG2CaCOIOVorg200546今1772| VariableQ、Filter variQbloc her(2) VMMMbUL b l門(mén) authorVZ yM TPR FPS3 FZ“ TN曰 VariablesNsm*?L b ( (Typo formotValue LabelZot:Add noteyear doublo%10.0g1 計(jì)算所有診斷試驗(yàn)統(tǒng)計(jì)學(xué)指標(biāo)（敏感度、特異度、似然比、診斷比值比等）及異質(zhì)性檢驗(yàn)統(tǒng)計(jì)量:Command*midas tp fp fn tn, es |xl res (

32、allly ex -( (o.04i-a.4S2Cor r* LAA r.uly1 onROC ArrraQX naanA1v* r 1 *no AUROC - O. a Q 92 )1-RT I St -a.4OCXQC.na.CX -at -2.oo.LFcT_p O . OOOYuno2OOGSon20OSVAmwgucIMTaKactoaocLamy2000SShltrnyami*2001Naaan2003MoE.2lj*l Mi x)cc A -03 &rnac E A 24 17Prt. AMt Prot of QI 八O 27QBAT. w*n-Aeutfly vrtno* vA

33、 r 1 aq 11dKN)O . 1 3zPA r AnMerRMC. lmj05% C XO7 03|O 4 1,o 7SC|dpAc1f1a1tyOQ42|O QI 3roPOM X 11 v* t. 1 k 1 haod ICA 11 O1 234fi |ft 24 5rlfi4 04 |NMQ A 1.1 v& AQnoMtia floor A3“5|322 Or41O2|O& AQnoMtio Qddz FCAC. iono071 I2S251 rOd2.繪制敏感度、特異度森林圖:CommandUmidas tp fp In tn, id author) year year es

34、(x) ms (0.75) ford for (dss)SEMSITIVITV Cll0.73 fO GO - OA3|0.74 |0.G O 07|0 G3 I0.&4 O 7l|O 72(0X2 O O1|0 00(0.73 - O OC|O C(O.G& O 7*|O 7A(O.G2 O90|O.77IO.7D O O3j0 47 |0.3C O &Q|0G9(0.G4 oai|O TOHO.G4 O 7C|t*ncyUVT_O 9S 4 ld#(X -= STUDY 146iiPCOirtClTV 3% CMOM (O.M OX)OS (O01 O.7|o VU( (aw7 i.MHO

35、 IM (O M - O.M)O wo (O S 1.MH ooa (O va o ua|OG (OAW 0.00)oa?( (o O.OCHO WO 03 O UO|OS (OMf o.g)O.M(O VI O.M)3.繪制ROC曲線圖:CommandQmidas tp fp fn tn, es (x) plot sroc2xObserved DataSummary Operating Point SENS - 0.70 0.64 0.76SPEC - 0.94 0.91 0.96 SROC CurveAUC - 0.89 0.86 0.82 95*5 Prediction EllipseO

36、.o-I-! -11 .O0.50.0Specificity4.繪制漏斗圖，識(shí)別發(fā)表偏倚:Command4midas tp fp In tn, pubbiasSTATISTICAL TESTS FOR SMALL STUDY EFFECTS/PUBLICATION BIASyb |Coef Std Err 匸 PIzI95% Conf. Interv alBias |-.700537617.62442-0.040.969-41.3425339.94Intercept |3782341.1057513.420.00912324756.33二 146206Log Odds Ratio versus

37、 vsqrt(Effective Sampte Size) )(Deeks) )DLR POSITIVE (95% CO6.57 |3.49 123988(3.87 &49.87 (22.27 - 111.67)7.38( (939.67(10.04 - 156.77)14.46 (7.99-26.1616.33(7.38.3B.30|6 08(4.87-7 59) )11.9( (6.03 - 231|9.42 (5.82-1525) )12.35J8.24 - 18曲|Study Regression Lineii0.040.C60 Ml1/root(ESS)1 10.1

38、0 0.(25.繪制似然比森林圖：Commandmidas tp fp fn tnid (author) year (year) es pf r* X r*cj丄 ccj v A 1 UI * :H 匕 E匕 JoeO sXo31 XXA KQod73.720-40?匕) )0”1 lA ccjX AXX KoodCQ 3O 3 S 3 3I txermnXor2 s工 aq1 AK XX KoodG 7.a0 013工匕sreUJoc3 sX QCJX Xc1X Kood C5 7 ：！：！*7 a 濟(jì) a ClE* K* o xr m X r% eg j w vi 久* c tx E* k

39、 n XmXXor* :X tx r j* e ionOsX ocX AX XKoodG7.37(190工匕irmOoc1X丄cqX X K w X XG7.37O-7GXXrm n Xon 2 xX acj X X CM X X Kood G 7 ： 7 Q 7 4工匕ire 匕 JoexXooX X c X XKoodG 7 ：！：！ 7 Cl 7 4 -12.繪制SROC曲線Command4metandi tp fp fn tn, plot.Specificityn nd XFStudy estimate HSROC curve 95% prediction regionS

40、ummary point 95*X confidonco regionMeta-discl4表2 Meta-Disc軟件的主要功能主要功能說(shuō)明Describing primary resuIts and exploring描述原始結(jié)果和探索異質(zhì)heterogeneity性Tabular resuIts將結(jié)果以表格形式列出Forest以森林圖形式顯示靈敏plots (sensitivity, specificity, LRs, d度、特異度、似然比和診OR)斷比值比ROC plane seatter-plotsROC平面散狀圖Cochran-Q, Chi-Square, Inconsistenc

41、y判斷研究間異質(zhì)性indexF訂tering/subgrouping capacities亞組分析Exporing Threshold effect探討閾值效應(yīng)Spearman correlation coefficientSpearman相關(guān)系數(shù)ROC plane plotsROC平面圖SROC curve under the curve(AUC) and Q擬合SROC曲線、計(jì)算AUC和Q指數(shù)Meta-regression analysis回歸分析，探討異質(zhì)性來(lái)源Univariate and multivariate Moses and（加權(quán)或未加權(quán)）單變量Litteenberg mode

42、l(weight and及多變量Mosesunweight)Litteenberg 模型Statistical polling of indices合并統(tǒng)計(jì)量Fixed effect model固定效應(yīng)模型Random effect model隨機(jī)效應(yīng)模型數(shù)據(jù)錄入i i 回 i事 File Edit Analyze Window HelpRlkllxlNo? Author Schneider SticbcrMolinaNissan Shibayama Lamy Iakada Yamaauchi SuriYangStudy IdTPFPFNTN200335181G1999419462182009

43、1347941548200929681192001741140224200011722997199673Z22820319958064746920052569020054691772 2冋-汕紅I晶I電I越I 1 1 n 1 OptionsMWPooling methodConfidence IntervalC Mantel-Haenszel (FEM)C 3095 r 99G D crS imonion -La ird (REM)I 0vcrdipcrsion correction參 Meta-DiScData - D:診斷性吞統(tǒng)評(píng)價(jià)實(shí)戰(zhàn)met“discl4dscI參 File Edit A

44、nalyzh| Window HelpTabular ResultsI金I呈淳No.AuthorThreshold AnalysisFPFNTN1SchneiderPlots.3518162StieberM eta- regressi o n.419463Molina13479414NissanFilter Studies.29685Shibayama7411406LamyOptions.117229Handling studies： with empty cellsC Drop studiesG Add 1 /2 to all cellsFile EditWindow.勺越碩空.；| Gra

45、phics | ROC Regression modek |Meta分析1.探索閾值效應(yīng)豪曰* Edit Analyze Window HelpD|!S -Tabular ResultsIu日垂臺(tái)No.AuthorThreshold AnalysisFPFNTN1SchneiderPlots.3518162232StieberMeta-regression.419462183MolinNissanFilter Studies.29G81195Shibaama7411402346nLamyT Options.117229cc97cccAnalysis of Diagno

46、stic ThreslioldSpearman correlation coezicienx* 0.321 p-value 0.365 (Legit (TPR) vs irogxu ( FPRJMoses model (D = a + bSWaxgh.t cd xar Q ssxon Invexso Vaxiaracoi)VarCoef f S匸d. ErrorTp-va luea2.82S0.3438.09S0.0000bC 1) -0.3830.16S2.31S0.0493Tau-scjuared esxmace = o.o35Z (Convergence xs actiieved aze

47、er 15 icerations)Sestricred Maximum Likelihood, estimation (RZML)INo. studies =10FiXter OFFAdd 1/2 to all cells of the studies witli zero2合并效應(yīng)量、探討異質(zhì)性Analyze Window HelpI . Il(? II de II鼻 File Edit Ana lyze Window HelpDllQTabular ResultsThreshold AnalysisPlots.AnalysisMeta-regression.Filter Studies.O

48、pti on&ai-rnan rrvr-iSuiftiaary Sensi t ivi tySxviayI SenCozx IcervalTP/TP+FN) TN/(TN+FP)S chine i-der35/5110.36229/2470.541-0.809St xob-o r10.4710.262-O . SQL41/87218/227Molina10.760 .96-0.82G124/17554Q/637Nissan10.7840 .18-0.90229/37119/125Shi-bay ama10.490.554-0.73674/114234/245Lamy1 0.8010.727-0

49、.863117/14697799Takada10.7230 .25-0.80773/101203/22SYamaguchi10.300.540-0.71480/12749/475Sun107350.556-O.87X25/3460/66Yang10.7300 .03-0.83446/637Z/8LPooled Sen10.990.669-O .729HecerogeneXcy c2i-square-d = 37.52 (cl. r.= 勺) p = 0.000 Incaris istency (I-square ) = 7G . 0 %No. studies =10,FXXL-er OFFAd

50、d 1/2 to all cells of the studies wxch zeroSummary SpecificityStudy | Spe 95% Conf Iterva1TP/ (TP+FN) TN/(TN+FP)Schneider|-35/51 229/247Stieber|-41/87 218/227Molina-548/627134/175Nissan-29/37 119/125Shibayama-74/114 234/245Lamy-117/14697/99Takada-73/101 203/225Yamaguchi-80/127 469/475Sun-25/34 60/66

51、Yang-46/63 72/81Pooled Spe|-Heterogeneity chi-squared 二 二 9) p 二Inconsistency (I-square) 二 %No. studies 二 10.Filter OFFAdd 1/2 to all cells of the studies with zeroSummarv Positive Likelihood Ratio (Fixed effects model)rlexghtScudy1匚R十95% Conf.Iterval%ScHn.Q xdex7&. 91一 15.249Scxebez15.811

52、久886G. 035一 23.411Molina140.166.0774.868一 7 .S87Nxssan13.L916.3297.345一 36.299Ghb*a yoma18.1414.4897.990一 2G.1G2Emmy12.7839.6810.037一 15.77Takada115.887.3924.882-11.193YamagAichi1Z -9549.86922.29一 111.67Suirx14.758.0883.674一 17.806Vano*19-17C. &713.4012.000(FEM) pooledLR+110.2378.693-12.056Hete r og

53、eneit y=hi sqiaxecL 47.2QScHna Idoic1o . SS&0-225O . SO9S 匸 XQ1QX110-2.0O . BSXo . as丄O . &73MoXXnA120 - X9O.26SO -205O-3：6XZ：L = n on1哎&9OZZ7O - XZ3o msnxta yamc1as. “O . 3S7O Z9GO 4：7ZLamyi3.76O .203O 140.91Takeda1ao. &2O .307O. 2?40.42 2Y mxx-iA g u ubi 丘Iae. 71O . S7SO . 290-4.7 0Son13 - aaO . 29XO X5O . S1 SXjrxg1& 32.O - 304.O - 2OX0-4.5