advertisement

---

Editors Selection IGR 12-3

Risk factors: Risk calculator

Comment by Luca Rossetti on:

13052 Validation of a predictive model to estimate the risk of conversion from ocular hypertension to glaucoma, Medeiros FA; Weinreb RN; Sample PA et al., Archives of Ophthalmology, 2005; 123: 1351-1360

See also comment(s) by Franz Grehn •

Find related abstracts

---

The assessment of risk factors for development of glaucomatous damage has become more and more important in the management of a patient with ocular hypertension (OH). From a practical point of view, the clinician has the problem of integrating the information from each risk factor in order to obtain a global estimate of the actual risk of the individual patient with OH before considering any treatment. To help solving this problem, a predictive model based on assessment of risk factors and able to tell the clinicians which patients have a higher chance of developing damage would be highly desirable. Furthermore, the incorporation of such a model into clinical practice might increase consistency in the standards of treatment and, more in general, the quality of care for the patients with OH. The study 'Validation of a predictive model to estimate the risk of conversion from ocular hypertension to glaucoma' by Medeiros et al. (953), presents the validation of a predictive model based on published results from the OHTS and tested on a cohort of 256 eyes of 126 untreated OH patients from the Diagnostic Innovations in Glaucoma Study (DIGS) carried out at the Hamilton Glaucoma Center, UCSD.

Age as a risk factor should be balanced by life expectancy

Two models based on the published average survival time, regression coefficients and means of risk factors from the OHTS, were developed and applied to 126 untreated OH patients from the DIGS. Of these patients, 25% developed glaucoma at five years (55% with an optic disc end-point, 32% with a visual field and 13% with both) with an average probability of conversion of 11.6%. The hazard ratios for all risk factors were very similar to the ones from OHTS except for diabetes. The models were also good in discriminating those who will develop glaucoma from those who will not. A simple risk scoring system to predict the risk for conversion from OH to glaucoma in five years was also developed. From this, a 60 year-old patient, with a baseline IOP of 25 mmHg, with a CCT of 600 µm, a vertical C/D of 0.3, a PSD of 1.25 and absence of diabetes has a predicted five-year risk of glaucoma development between one to five percent. Of course, all attempts to build a risk calculator 'providing numbers' may suffer from a series of drawbacks and limitations. The choice of the level of risk for which treatment should be considered is arbitrary and potentially dangerous: we all know that the risk of falsely labelling a patient as glaucomatous may have a negative impact on her/his quality of life due to diagnosis (especially if it is a wrong diagnosis) and, more important, to treatment. So which is the level of risk deserving treatment? This, of course, must be left to the clinician's judgment. If on the one hand the application of statistics to help solving clinical problems may result in a favourable rationalization of the standards of care, on the other hand such procedures should be considered cautiously. Also the best predictive models may have some limitations. In this particular case,

Quantification of risk helps predicting conversion from ocular hypertension to glaucoma

risk assessment is based just on age (which probably should be somehow adjusted for life expectancy) IOP and CCT (as C/D and PSD might represent the disease and, as mentioned in the paper, the presence of these two variables do not significantly change the performances of the model). CCT apart, we make our decisions on the basis of untreated IOP, which, no doubt, is really good sense. As far as the risk scoring system is concerned, an unbalanced scenario comes to my mind: an old patient (77 years) with a cornea of 485 µm, an IOP of 25 mmHg, a C/D of 0.3 and a PSD of 1.00 has the same risk of a patient aged 48, with 28 mmHg of IOP, 595 µm of CCT a C/D of 0.7 and a PSD of 1.95. Risk models are useful as far as they are applied with good sense.

---

• ← Previous
• Next →

---