Abstract
Purpose :
To detect glaucoma progression, a comprehensive examination that includes perimetry and imaging should ideally be performed frequently and regularly. However, the associated costs can be high. We evaluated different examination modalities, and the parameters obtained by them, to examine their cost-effectiveness in detecting glaucoma progression.
Methods :
One eye of 167 glaucoma patients followed prospectively was analyzed. Automated perimetry and optical coherence tomography were performed at 6-month intervals. Parameters analyzed included mean deviation (MD), Bruch's membrane opening-minimum rim width (BMO-MRW) and peripapillary retinal nerve fibre layer thickness (RNFLT). To ascertain disease stage (mild, moderate or severe) at each visit, a hierarchical cluster analysis was performed with MD, BMO-MRW and RNFLT. The effectiveness of each parameter in detecting progression was estimated with Hidden Markov Models, which assumes the unobserved true disease stage can only be diagnosed by other parameters. Examination costs were obtained from previous literature (Blumberg et al., 2014). Cost-effect ratios (CER, i.e. cost per unit effectiveness) were used to compare parameters.
Results :
Subjects were followed for a median (range) of 4 (2-6) years. Figure 1 shows the disease stages of all visits depending on the parameter values. With MD, on average (95% CI), 8 (6-12) examinations were needed to detect progression from mild to moderate, with 40% accuracy. In contrast, with BMO-MRW and RNFLT, only 4 (3-5) examinations were needed to detect progression from mild to moderate, with 80% accuracy. However, the average number of examinations to detect progression from moderate to severe was similar between three parameters (MD: 5 (3-9), with 60% accuracy; BMO-MRW: 6 (3-10), with 50% accuracy; RNFLT: 7 (4-10), with 45% accuracy). To detect progression from mild to moderate, CERs in $US/unit accuracy were 13.5, 2 and 2.2 with MD, BMO-MRW and RNFLT, respectively. That is, to increase the accuracy of progression detection (mild to moderate) by 1%, costs $US 13 with MD and only $US 2 with BMO-MRW or RNFLT.
Conclusions :
Following up structural parameters to detect glaucoma progression from mild to moderate was more cost-effective compared to following up with MD. However, the cost-effectiveness of the three parameters was not significantly different in detecting progression from moderate to severe glaucoma.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.