5. Second step is usually to calculate
the correlation coefficient (r)
r = 0.94 (p < 0.001)
因為H0 (兩種flow meter的結果不會有線性相關)成立的機
率很小 推翻H0是很安全的 H1成立
兩種flow meter的結果是線性相關的!
但是,兩種結果的高度相關性不代表一致性好
Perfect correlation: if the points lie along any straight
line
Perfect agreement: only if the points in the fig 1 lie
along the line of equality
一致性好,相關性一定好; 相關性好,一致性不一定好
Lancet, 1986; i: 307-310
6. Third step: Measuring agreement
the Bland–Altman plot
Lancet, 1986; i: 307-310
Limit of
agreement : (LoA)
8. Material and methods
24 healthy volunteers
Caucasians
no Hx/symptoms of heart disease or other chronic
disease
Mean age: 44 years (range: 25–60)
Male 71%, female 29%
Mean BMI 25 kg/m2 (range: 19–32)
FP-RNA+GBPS (same day) and MRI: within 1
month
21. Some points I agree with
RVEF is of interest in:
pulmonary hypertension
chronic pulmonary diseases enlargement
congenital heart diseases
The three most used modalities to assess RVEF:
FP-RNA, GBPS, cMRI
“2D echocardiography as we believe it is generally
not considered a quantitative method with regard to
the right ventricle”
“Planar equilibrium multiple ECG-gated isotope
ventriculography is not useful for RVEF as it is not
possible to separate the right ventricle and its
results show a very high variability”
28. In summary
Measurements of RVEF differ significantly
between MRI and FP or GBPS
And, different normal limit of RVEF
MRI: 0.49-0.72
FP: 0.44-0.66
GBPS: 0.40-0.69
The LoA between the methods are wide
Not interchangable!
不一致,但是否相關呢?
未知or不易得知
真值
間接方法測量
新方法測
量
MRI
FP-RNA GBPS
•But MRI being the gold standard of RVEF?
?