Sensitivity and specificity estimates for the criterion of an IOP ⩾21 mm Hg and the associated positive likelihood ratios (that is, sensitivity/1 − specificity) and PPV* (%) for glaucoma, given an IOP ⩾21 mm Hg. The PPV estimates have assumed a prevalence of undetected glaucoma of 1% and no additional patient risk factors (that is, pretest probability of glaucoma of 1%)
| IOP criterion ⩾21 mm Hg | Sensitivity (95% CIs) | Specificity (95% CIs) | Likelihood ratio (+ve) | Positive predictive value (%) |
|---|---|---|---|---|
| Daubs and Crick5 | 52% (48–56 ) | 49% (43–55 ) | 1.0 | 1.0 |
| Tielsch et al 6 | 51% (44–58) | 84% (83–85) | 3.2 | 3.1 |
| Harper and Reeves7 | 51% (39–63) | 97% (94–100) | 17.0 | 14.7 |
PPV = (prevalence × sensitivity)/((prevalence × sensitivity) + (1 − prevalence × 1 − specificity)).
95% CIs are not given for these PPV estimates, since, for the purposes of this comparison, we have used Bayes's theorem and the sensitivity/specificity at an IOP cut off criterion ⩾21 mm Hg (assuming the same glaucoma prevalence rate in each case), rather than working back from the frequency of study cases.