当前位置:文档之家› 医疗器械临床试验统计学要求

医疗器械临床试验统计学要求


Registry
All-cause mortality (%) 14.4 14.9 6.0 0.73 1.41*
*
After adjustment for all differences between treatment groups.
• A final sample size : 600 patients (300 per
group) assuming 5% of loss
N Engl J Med 2011;364:1718-27
Non-inferiority Test for
Primary End Point of 1-Year MACCE
Non-inferiority P value
<0.001
Non-inferior
Zone of non-inferiority Pre-specified margin = 3.5%
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0 %
Upper one-sided 95% CI
• Assumed primary end point of 1-year MACCE
in the CABG group : 13%
• A non-inferiority margin : 7% • A one-sided type I error rate : 0.05 • Power : 80%
1-Year MACCE rate CABG: 6.7%
Pre-specified noninferiority margin: 7%
Difference, 2% 95% CI, -1.6 to 5.6% Non-inferiority p= 0.001
PCI: 8.7%
-2 -1 0 1 2 3 4 5 6 7 8 9 10 Difference (%) of 1-year MACCE rate between (PCI – CABG)
Primary Non-Inferiority Endpoint Met
PRECOMBAT Trial
Design
• DESIGN: A prospective, open-label, randomized trial • OBJECTIVE: To compare PCI with sirolimus-eluting stents and CABG surgery for optimal revascularization of
评价标准 非劣效界值的确定
对照组与安慰 剂的疗效差异
8% vs. 20%
对照组的最低获益程度
-4
-2
0
2
4
试验组与对照 组的疗效差异
1/3 ~ 1/4
非劣效界值= Fraction * (C-P)
-4
-2
0
2
4
NI 界值
PROMISE study
Pamela SD, et al. N Engl J Med 2015; 372:1291-1300
N Engl J Med 2011;364:1718-27
RESOLUTE All Comers Trial Statistical Assumptions
Primary Clinical Endpoint: Target Lesion Failure at 12 Months
(Composite of cardiac death, target vessel MI and clinically driven TLR)
5-Year Mortality Rates for Randomized and Registry Diabetic Patients
Randomized
PTCA CABG p Value RR
All-cause mortality (%) 34.5 19.4 0.0024 1.87
Cardiac mortality (%) 23.4 8.2 0.0002 3.10
1134 1150
1098 1104
1072 1083
1044 1051
2.4% 1.7%
5
1005 1021
从研究目的到研究假设
• 研究目的
在ACS患者中,比较替格瑞洛和氯吡格雷 在All Comers中,对ZES与EES进行比较
• 研究假设
在ACS患者中,替格瑞洛与氯匹格雷相比是否 能够降低复合终点事件的发生(优效)
10 8 6 4 2 0 0
No. at risk Resolute Xience V
1140 1152
Resolute™ ZES (N = 1140) Xience V™ EES (N = 1152)
Log rank P = 0.22
1
2
3
4
Time After Initial Procedure (years)
• 结果 • 16 次试验 (100个中) 未能显示出显著性差异
把握度 (power) 分析
计算机模拟举例 -2
• 假设
试验组死亡率为 20% (x%) 对照组死亡率为 50% (y%) 试验组和对照组的样本量均为 (n) = 100 重复100 (m) 次试验 (随机抽样) 显著性水平为双侧 0.05 检验方法 = 卡方检验
试验设计与证据级别
Coronary Revascularization in Diabetic Patients: A Comparison of the Randomized and Observational Component of the
Bypass Angioplasty Revascularization Investigation (BARI)
• More registry patients were white, more educated, more physically active and reported better quality of life.
Diabetic patients in BARI randomized to PTCA had significantly worse 5-year survival than patients assigned to CABG. This difference was not seen among BARI-eligible patients who opted to select their mode of revascularization.
patients with ULMCA stenosis.
N Engl J Med 2011;364:1718-27
Primary End Point
• A composite of major adverse cardiac or
cerebrovascular events (MACCE) for the 12month period after randomization
体重: 17 - 23kg 血压下降: 20 ± 9 mmHg N=6/group
把握度 (power) 分析
计算机模拟举例 -1
• 假设
试验组死亡率为 20% (x%) 对照组死亡率为 50% (y%) 试验组和对照组的样本量均为 (n) = 50 重复100 (m) 次试验 (随机抽样) 显著性水平为双侧 0.05 检验方法 = 卡方检验
Event rate at 12-months would be 8% and equal in both groups Non-inferiority margin of 3.5% and one-sided type I error of 0.05
• 2300 patients would yield >90% power to detect non-inferiority
Inconclusive?
Inferior
0

Treatment Difference with 95% CI
← New Treatment Better
New Treatment Worse →
Primary Endpoint Analysis of the RESOLUT ALL Comers Trial
Secondary Angiographic Endpoint: In-stent Percent Diameter Stenosis at 13 Months
1.5 lesions would be treated per patient % DS at 13 months would be 16 ± 16% and equal in both groups Non-inferiority margin of 5% and one-sided type I error of 0.05 Attrition rate 20% • 460 patients would yield >90% power to detect non-inferiority
PRIMI: Multivessel vs. Culprit PCI
Wald DS, et al. N Engl J Med 2013; 369:1115-1123
样本量计算
影响因素
效应值 (Effect Size) 的影响
1%
2%
10%
20%
样本量计算 决定样本量的关键因素—”变异”
体重: 3.8 - 42.6kg 血压下降: 20 ± 36 mmHg N=68/group
1,829 patients (19% with DM) with MVD were randomized to CABG or PTCA 2010 eligible patients (17% with DM) did not consent in RCT entered into registry
相关主题