advertisement

---

Editors Selection IGR 15-2

Predictive Models for Glaucoma Development

Comment by George Lambrou on:

21940 Predictive models to estimate the risk of glaucoma development and progression, Medeiros FA; Weinreb RN, Progress in Brain Research, 2008; 173: 15-24

Find related abstracts

---

The successful management of any chronic disease relies heavily on the ability of the clinician to predict with some degree of confidence what the future course of the disease will be, with or without therapeutic intervention, based on the assessment of current clinical status and the presence or absence of risk factors ‐ a process known as 'risk assessment'. While risk assessment is part educated guesswork and part clinical expertise of the Health Care Professional, the development of predictive models based on data from large longitudinal studies has contributed significantly to the improvement of therapeutic outcomes ‐ originally in terms of mortality reduction from cardiovascular disease. This success has led to the more recent implementation of risk assessment tools in other disease areas, and namely to the construction of predictive models for the development and progression of glaucoma. The present paper by Medeiros and Weinreb (1763) is a review of the steps involved in such an endeavor, inspired by predictive models for cardiovascular disease. These models are built around the concept that glaucoma is a continuum in which the affected eye ‐ if left untreated for enough time ‐ may evolve from normal to blind, progressively moving through the stages of undetectable disease, asymptomatic disease and functional impairment. The rate of progression through these various stages is influenced by several risk factors (such as IOP, age or corneal thickness). While these risk factors are well-known to clinicians who use them qualitatively in their daily practice, it is not straightforward to combine them quantitatively into an overall assessment of the risk that the condition of a particular patient will deteriorate. Indeed, previous studies have shown that ‐ even with the help of handouts summarizing the main OHTS findings ‐ ophthalmologists tended to vary widely in their estimation of the progression risk for a given eye, usually erring on the side of underestimation.

While predictive models are a useful tool for the management of patients, their use cannot replace the clinical judgment of the healthcare professional

This is where the predictive model described by the authors comes into play. Based on data from OHTS, they have developed a point system and an electronic risk calculator which estimates the probability of an ocular hypertensive patient to develop glaucoma within five years if left untreated. Like all models based on a specific dataset, this conversion-to-glaucoma model needed to be validated before it could be used on other patient populations. The validation was successfully performed using the DIGS (Diagnostic Innovations in Glaucoma Study) data from San Diego. Taking this approach one step further would lead to the development and validation of a predictive model for progression, rather than for onset of glaucoma. This could theoretically be based on studies such as the Early Manifest Glaucoma Trial, followed by validation on an independent glaucoma patient population followed longitudinally.

The authors conclude their article by correctly pointing out that while predictive models are a useful tool for the management of patients, they must be used with circumspection in the context of a holistic approach of the patient's problems, keeping in mind that their use can in no case replace the clinical judgment of the Health Care Professional, when making patient-management decisions.

---

• ← Previous
• Next →

---