Research Article

Noninferiority and Equivalence Evaluation of Clinical Performance among Computed Radiography, Film, and Digitized Film for Telemammography Services

Table 5

Equivalence tests (δ = 0.05) for TN, TP, FN, and FP by paired devices.

VariableCompared devices⁢(1-2α)% confidence interval for equivalence testing
() device() device LowerUpper

TNLUMIXICR −0.023 0.011−4.31<0.001Ha 0.023
LUMIXCR −0.020 0.008−5.23<0.001Ha 0.020
LUMIXFILM −0.032 0.000−3.43<0.001Ha 0.032
ICRCR −0.015 0.015−5.48<0.001Ha 0.015
ICRFILM −0.026 0.006−4.10<0.001Ha 0.026
FILMCR −0.001 0.021−5.98<0.001Ha 0.021

TPLUMIXICR 0.000 0.032−3.43<0.001Ha 0.032
LUMIXCR −0.025 0.009−4.04<0.001Ha 0.025
LUMIXFILM −0.051 −0.011−1.590.06Ho 0.051
ICRCR −0.042 −0.007−2.450.007Ha 0.042
ICRFILM −0.069 −0.025−0.230.41Ho 0.069
FILMCR −0.003 0.047−1.820.03Ha 0.047

FNLUMIXICR −0.032 0.000−3.43<0.001Ha 0.032
LUMIXCR −0.009 0.025−4.04<0.001Ha 0.025
LUMIXFILM 0.011 0.051−1.590.06Ho 0.051
ICRCR 0.007 0.042−2.45<0.001Ha 0.042
ICRFILM 0.025 0.069−0.230.41Ho 0.069
FILMCR −0.047 0.003−1.820.03Ha 0.047

FPLUMIXICR −0.011 0.023−4.31<0.001Ha 0.023
LUMIXCR −0.008 0.020−5.23<0.001Ha 0.020
LUMIXFILM 0.000 0.032−3.43<0.001Ha 0.032
ICRCR −0.015 0.015−5.48<0.001Ha 0.015
ICRFILM −0.006 0.026−4.10<0.001Ha 0.026
FILMCR −0.021 0.001−5.98<0.001Ha 0.021

Each comparison was calculated from 980 observations (70 cases, seven radiologists, and two devices). TN: true negative proportion, TP: true positive proportion, FN: false negative proportion, FP: false positive proportion, SE: standard error of the mean, Ho: null hypothesis, Ha: alternative hypothesis for testing equivalence, α: significance of the test (0.05), δ: difference of the means allowed to achieve equivalence, : test for difference of compared devices, that is, = (Difference ()δ)/SE, : retained hypothesis equivalence at δ level (“Ha” indicates equivalence achieved and “Ho” indicates failing to reject the null hypothes).
Ho: difference ()δ = 0.
Ha: difference ()δ < 0.