Purpose:
Predictive values are important when selecting a screening method. Positive predictive values (PPVs) depend strongly on disease prevalence. Therefore studies including a sample of clinical glaucoma patients would yield falsely high PPVs. A population based sample would yield a more valid prevalence and thereby also a more valid PPV. Our aim is to describe the diagnostic performance of time domain Stratus and spectral domain Cirrus OCT in a population based sample.
Methods:
Out of 4718 persons over 50 years old, a random sample of 307 individuals was invited. The 170 individuals that responded underwent a comprehensive ophthalmic examination. Of the 138 individuals that were finally included, one eye per individual was analysed. Retinal nerve fiber layer thickness (RNFLT) was measured with the Fast RNFLT protocol of Stratus OCT (Carl Zeiss Meditec, Dublin, CA, USA) and the glaucoma analysis package of Cirrus OCT (Carl Zeiss Meditec, Dublin, CA, USA). Specificity, sensitivity and PPVs for mean RNFLT, quadrants and the clock hour sectors, as well as AROCs for the two latter RNFLT parameters, were calculated using the significance limits p< 5% and p< 1% as cut-off for positive findings.
Results:
Nine glaucoma patients were identified (prevalence of 6.5%). Mean age was 72 (range: 61-83) and 64 years (range: 48-81) for glaucoma and healthy individuals respectively. Specificity, sensitivity and PPVs for mean RNFLT, quadrants and the clock hour sectors with the corresponding AROCs can be seen in the table.
Conclusions:
Both Stratus and Cirrus OCT performed well. For all parameters and both cut-off levels, p<5% and p<1%, Stratus OCT yielded better specificity and positive predictive values, both desirable properties of a method to be used for screening. However, the higher sensitivity of Cirrus OCT is preferable in clinical settings where priority is given to early detection of glaucoma.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical • nerve fiber layer