Table 5.

Data Display When There are Multiple Examiners*

	Examiner (n = number examined)	Sensitivity (n = number with condition)	Specificity (n = number without condition)	Likelihood ratio positive (95% confidence interval)	Likelihood ratio negative (95% confidence interval)
Finding 1	Examiner 1 (n = 100)	0.67 (n = 30)	0.79 (n = 70)	3.1 (1.9–5.2)	0.42 (0.25–0.71)
	Examiner 2 (n = 80)	0.65 (n = 23)	0.91 (n = 57)	7.4 (3.0–18)	0.38 (0.22–0.67)
Summary finding 1	Range	0.65–0.67	0.79–0.91	3.1–7.4	0.38–0.42
	Summary measures	0.66 (95% CI 0.52–0.79)	0.84 (95% CI 0.74–0.93)	4.0 (2.4–6.6)	0.40 (0.27–0.59)

*We used the data from Table 2for the first 2 symptoms, as if they were for the same symptom but by different examiners. Note that a summary estimate can be used to show the variability in performance between examiners. With only 2 examiners, it may be preferable to report the summary measure as the range. In this hypothetical example, the data are clinically and statistically homogenous, so we show a random-effects summary measure to demonstrate how the data could be displayed. Different summary measure calculations could lead to slightly different, although not clinically important, changes in the estimates (e.g., the Maentel–Haenzel chi-square for the LR+ is 4.4 with a 95% CI of 2.7 to 7.0).