Skip to main content

Table 2 Data generated from ROC curves and logistic regression

From: Can American College of Radiology in-training examination scores be used to predict Canadian radiology licensing examination results? A retrospective study

Year of training*

Threshold ACR score†

OR‡

95% CI for OR

100 × [OR −1] (%) §

ACR 50**

95% CI for ACR 50

PGY2

32

0.91 (P=0.0068)

[0.84, 0.97]

−9.3

4

[0, 13]

PGY3

42

0.93 (P=0.0014)

[0.89, 0.97]

−6.8

4

[0, 15]

PGY4

63

0.94 (P=0.0091)

[0.90, 0.98]

−6.0

8

[0, 23]

PGY5

47

0.93 (P=0.0039)

[0.88, 0.98]

−7.3

8

[0, 20]

  1. PGY=post graduate year, ACR=American college of radiology, OR=odds ratio, CI=confidence interval.
  2. * PGY2 is the first year of 4 consecutive years of radiology-specific specialty training.
  3. † Threshold ACR score at or above which point there is a negligible risk of exam failure. Estimated from the ROC curves.
  4. ‡ Odds ratios were calculated using logistic regression. Statistical significance at P ≤ 0.05.
  5. § 100 × [OR-1] is a formula that gives the percentage change in the odds of failing for every 1 unit increase in the ACR score. A negative value, as in this case, indicates the odds of failing decrease with increases in the ACR score.
  6. ** ACR 50 is the ACR score at which point there was a 50% chance of failure. This was estimated for each year of training using the logistic regression curves.