诊断试验的系统综述.ppt
http://www.100md.com
参见附件(198KB)。
诊断试验的系统综述
张博恒
复旦大学肝癌研究所
复旦大学临床流行病学培训/
循证医学中心
Bzhang@zshospital.com
诊断试验评价
? 准确性
? 敏感度、特异度和似然比
? ROC曲线
? 比数比(DOR)
? 重复性
诊断试验的准确性
诊断试验的准确性(ROC)
诊断试验的准确性(DOR)
诊断试验准确性的系统综述
系统综述的步骤：
? 提出问题
? 检索并选择研究
? 对纳入的研究进行评价
? 收集提取资料
? 分析并形成结果 (meta-分析)
? 结果的解析
? 系统综述的修正与更新
诊断试验准确性的系统综述
? 提出问题
?给予一定的限制
? Example:AFP 诊断原发性肝癌的准确性？
? AFP 的阳性率(敏感度？)
?原发性肝癌：HCC ， CCC ， mixed
? HCC ：HBV-related, HCV-related
? HBV-related: small, big
?太局限：应用性vs太广：异质性
诊断试验准确性的系统综述
? 检索并选择研究
?纳入、排除标准
?多方位的检索
?电子，手工，联系
?关键词的选择
?选择的时候可以稍宽要求
?敏感性分析
电子数据库
Retrospective
? Index Medicus (Medline)
? Excerpta Medica (Embase)
? HealthStar (prev HealthPlan)
? PsychLit
? CINAHL
? CancerLit
? ToxBase
Prospective
? Science Citation Index
诊断试验准确性的系统综述
? 对纳入的研究进行评价(critical appraisal)
? 设计
? 随机
? 金标准
? 盲法
? 病谱
? 样本量
critical appraisal关心的4个问题
? Why did they start?Worth reading on?
? What did they do?Validity?
? What did they find? Results?
? What does it mean?Relevance?
诊断试验设计
诊断试验设计
病谱偏倚
诊断试验准确性的系统综述
? 收集提取资料
?研究的时间
?病人的基本情况
?病谱
?诊断试验方法
?阈值
诊断试验准确性的系统综述
? 分析并形成结果 (meta-分析)
?阈值一致的，或可以忽略的
?阈值不一致
低剂量螺旋CT筛查肺癌的评价:系统综述
Background and rationale
? Lung cancer is the leading cause of death from malignant diseases
? The poor prognosis: difficulty in detecting early stage disease and the low cure rate for advanced disease.
? 5-year survival: stage IA cancers (T1N0M0)67% to 80% vsstage IIIB NSCLC <5%
CXR +/- sputum cytology
6 RCT +1 non-RCT (245 610)
? no mortality reduction from lung cancer
? frequent CXR screening might even be harmful
LDCT
? Sone S et al. Mass screening for lung cancer with mobile spiral CT scanner. Lancet 1998;351:242-245
? Henschke CI, et al. Early lung cancer action project: overall design and findings from baseline screening. lancet, 1999;354:99-105
The preliminary results:
? LDCT can detect lung cancers at an average size of 1.5cm or less.
? Approximately 80% of these lung cancers are Stage IA.
? Only 15-20% are Stage IA in current medical practice.
研究目的
? To analyse the performance of screening test by using LDCT for lung cancer.
? Assuming: no over-diagnosis existed, and screening for lung cancer with LDCT lead to a significant reduction of lung cancer mortality rate.
Identification of studies
Time: January 1990 and September 2002 in English-language literature.
Sources: Current Contents, Medline (from 1990- September 2002), Premedline (1990 - 2002), EMBASE (from 1982 to September 2002), CINAHL (from 1982 to 2002), the Cochrane Controlled Trials Register, and Cancerlit (from 1966 to 2002).
Medline keywords : "lung cancer neoplasms" or "lung carcinoma" or "lung cancer", "screen$", "CT" or "computed tomography"
Contacting researchers, experts, the reviewers, / examining the bibliographies /Conference proceedings
Selection of studies
(1) study reported LDCT screen for lung cancer;
(2) the absolute numbers of true-positive, false-negative, false-positive and true negative observations were available or derivable from the data presented;
(3) not a duplicate study of same patient group.
(4) all participants in study received long-term follow-up, and subjects with abnormal LDCT result were followed up with a definitive diagnostic investigation.
Quality assessment
? The quality of studies of screening test for inclusion in the meta-analysis was assessed following the checklist suggested by Cochrane.
Extract data
? Data were extracted using a structured form by one reviewer (BZ) and checked by the other reviewer (MA).
Statistical analysis
? sensitivity and specificity, 95% CI,LR+/-,diagnostic odds ratio (DOR), % stage I cancers.
?
? Pooled estimate of LR and DOR: Fixed-effects model and D&L random-effects model.
? Smoothed fitting to the SROC curve and AUC
? TNM for NSCLC. SCLC: limited-stage / extensive-stage
? All data analyses were performed using STATA.
Results
Stage at diagnosis of screen-detected lung cancer in prevalence screening
Littenber和Moses方法
? 计算每个研究的D和S
Littenber和Moses方法
诊断试验的系统综述评价
? How were the papers identified
? How was the quality of papers assessed
? How were the results summarised
? It the topic well defined
? Was publication bias taken into account
? Was missing information sought
? Were the detailed study designs reviewed
? Was the heteogeneity of effect investigated
? Are there other findings which merit attention
? Are the conclusions justified
诊断试验的系统综述
张博恒
复旦大学肝癌研究所
复旦大学临床流行病学培训/
循证医学中心
Bzhang@zshospital.com
诊断试验评价
? 准确性
? 敏感度、特异度和似然比
? ROC曲线
? 比数比(DOR)
? 重复性
诊断试验的准确性
诊断试验的准确性(ROC)
诊断试验的准确性(DOR)
诊断试验准确性的系统综述
系统综述的步骤：
? 提出问题
? 检索并选择研究
? 对纳入的研究进行评价
? 收集提取资料
? 分析并形成结果 (meta-分析)
? 结果的解析
? 系统综述的修正与更新
诊断试验准确性的系统综述
? 提出问题
?给予一定的限制
? Example:AFP 诊断原发性肝癌的准确性？
? AFP 的阳性率(敏感度？)
?原发性肝癌：HCC ， CCC ， mixed
? HCC ：HBV-related, HCV-related
? HBV-related: small, big
?太局限：应用性vs太广：异质性
诊断试验准确性的系统综述
? 检索并选择研究
?纳入、排除标准
?多方位的检索
?电子，手工，联系
?关键词的选择
?选择的时候可以稍宽要求
?敏感性分析
电子数据库
Retrospective
? Index Medicus (Medline)
? Excerpta Medica (Embase)
? HealthStar (prev HealthPlan)
? PsychLit
? CINAHL
? CancerLit
? ToxBase
Prospective
? Science Citation Index
诊断试验准确性的系统综述
? 对纳入的研究进行评价(critical appraisal)
? 设计
? 随机
? 金标准
? 盲法
? 病谱
? 样本量
critical appraisal关心的4个问题
? Why did they start?Worth reading on?
? What did they do?Validity?
? What did they find? Results?
? What does it mean?Relevance?
诊断试验设计
诊断试验设计
病谱偏倚
诊断试验准确性的系统综述
? 收集提取资料
?研究的时间
?病人的基本情况
?病谱
?诊断试验方法
?阈值
诊断试验准确性的系统综述
? 分析并形成结果 (meta-分析)
?阈值一致的，或可以忽略的
?阈值不一致
低剂量螺旋CT筛查肺癌的评价:系统综述
Background and rationale
? Lung cancer is the leading cause of death from malignant diseases
? The poor prognosis: difficulty in detecting early stage disease and the low cure rate for advanced disease.
? 5-year survival: stage IA cancers (T1N0M0)67% to 80% vsstage IIIB NSCLC <5%
CXR +/- sputum cytology
6 RCT +1 non-RCT (245 610)
? no mortality reduction from lung cancer
? frequent CXR screening might even be harmful
LDCT
? Sone S et al. Mass screening for lung cancer with mobile spiral CT scanner. Lancet 1998;351:242-245
? Henschke CI, et al. Early lung cancer action project: overall design and findings from baseline screening. lancet, 1999;354:99-105
The preliminary results:
? LDCT can detect lung cancers at an average size of 1.5cm or less.
? Approximately 80% of these lung cancers are Stage IA.
? Only 15-20% are Stage IA in current medical practice.
研究目的
? To analyse the performance of screening test by using LDCT for lung cancer.
? Assuming: no over-diagnosis existed, and screening for lung cancer with LDCT lead to a significant reduction of lung cancer mortality rate.
Identification of studies
Time: January 1990 and September 2002 in English-language literature.
Sources: Current Contents, Medline (from 1990- September 2002), Premedline (1990 - 2002), EMBASE (from 1982 to September 2002), CINAHL (from 1982 to 2002), the Cochrane Controlled Trials Register, and Cancerlit (from 1966 to 2002).
Medline keywords : "lung cancer neoplasms" or "lung carcinoma" or "lung cancer", "screen$", "CT" or "computed tomography"
Contacting researchers, experts, the reviewers, / examining the bibliographies /Conference proceedings
Selection of studies
(1) study reported LDCT screen for lung cancer;
(2) the absolute numbers of true-positive, false-negative, false-positive and true negative observations were available or derivable from the data presented;
(3) not a duplicate study of same patient group.
(4) all participants in study received long-term follow-up, and subjects with abnormal LDCT result were followed up with a definitive diagnostic investigation.
Quality assessment
? The quality of studies of screening test for inclusion in the meta-analysis was assessed following the checklist suggested by Cochrane.
Extract data
? Data were extracted using a structured form by one reviewer (BZ) and checked by the other reviewer (MA).
Statistical analysis
? sensitivity and specificity, 95% CI,LR+/-,diagnostic odds ratio (DOR), % stage I cancers.
?
? Pooled estimate of LR and DOR: Fixed-effects model and D&L random-effects model.
? Smoothed fitting to the SROC curve and AUC
? TNM for NSCLC. SCLC: limited-stage / extensive-stage
? All data analyses were performed using STATA.
Results
Stage at diagnosis of screen-detected lung cancer in prevalence screening
Littenber和Moses方法
? 计算每个研究的D和S
Littenber和Moses方法
诊断试验的系统综述评价
? How were the papers identified
? How was the quality of papers assessed
? How were the results summarised
? It the topic well defined
? Was publication bias taken into account
? Was missing information sought
? Were the detailed study designs reviewed
? Was the heteogeneity of effect investigated
? Are there other findings which merit attention
? Are the conclusions justified
附件资料：
相关资料：
- 《弹出的书-POP纸艺》(The Pop-Up Book)影印版.pdf
- 《二十四节气养生大全》扫描版.pdf
- 小儿病毒性感染与细菌性感染的初步区别 .doc
- 新生儿与新生儿疾病(Asphyxia & HIE).pdf
- 《Roitt免疫学基础 第十版》(Roitt's Essential Immunology Tenth Edition)影印版.pdf
- 《伯里曼人体结构教学》扫描版.pdf
- 肿瘤转移.ppt
- 《这样猜透女人心·彩图版》扫描版.pdf
- 《唐代饮食》扫描版.pdf
- 呼吸力学监测的临床应用.ppt
- 获得性免疫缺陷综合征.ppt
- 《让青少年热爱科学的故事》扫描版.pdf
- 第08章 细胞凋亡于肿瘤 .doc
- 骨科医生行医指南 .doc
- 《小样，治不好你我还叫兽医？》扫描版.pdf