THE MACULAR ASSESSMENT PROGRAM: OUTCOME PREDICTION MODELS FOR AMD

Title THE MACULAR ASSESSMENT PROGRAM: OUTCOME PREDICTION MODELS FOR AMD
Author, Co-Author Mira Acs, Michael Kaplan, Dorothy Barrie
Topic
Year
2013
Day
Program Number
135026
Room
Room 6C
Affiliation
Alcon Research, Ltd
Abstract PURPOSE:
The Macular Assessment Program (MAP) was developed to raise the awareness of risk for Age-related Macular Degeneration (AMD) and the need for screening certain individuals. Key risk factors identified and associated with level of risk and diagnosis of AMD will help to establish Outcome Prediction Models for AMD.

METHODS:
290 optometrists from across Canada participated, each assessing 20 of their patients’ who they considered at risk of AMD. 4,953 patients were assessed for risk of AMD. The AMD assessment questionnaire addressed the non-modifiable and modifiable risk factors, enabling the optometrists to establish a category of risk for each patient; LOW, MODERATE or HIGH.

RESULTS:
All Risk Factors in the AMD assessment questionnaire were considered as inputs for 2 multivariate logistic regression models: • Dry and Wet AMD Diagnosis – Estimate the probability of AMD and the relative probability of the 3 dry AMD stages and wet AMD, based on risk factors • Optometrists’ assessment of AMD risk – Use model to interpret how physicians used risk factors to assess risk Both models identified risk factors as significant in predicting patient outcomes with respect to AMD • 10 of 16 risk factors were included in the AMD diagnosis model (Table 1) • 11 of 16 risk factors were included in the Optometrist risk assessment model (Table 2) The significant factors were ranked according to statistical significance of impact (p-value) • Description for each significant factor indicates the direction of impact on expected AMD outcome Table 1: AMD diagnosis model. Table 2: Optometrist risk assessment model. The 2 models showed statistical agreement for the majority of risk factors (Table 3) • 3 risk factors showed notable differences between models with respect to significance ranking: Activity Level; Blood Pressure; Smoking. Table 3: Model Comparison. The proportion of patients that fell into each stage of diagnosis was calculated (Table 4) • Patients classified by optometrists as: o High risk - spread across the five diagnosis o Moderate risk - majority negative or Stage 1 AMD o Low risk - most in negative AMD diagnosis group Table 4: Risk Level vs Diagnosis.

CONCLUSIONS:
Based on easily determined patient attributes, key risk factors can be identified that establish an Outcome Prediction Model for the risk and diagnosis, including the stage, of AMD.

ADDITIONAL COMMENTS:
4 Graphic elements will be included: - 4 Tables

Affiliation of Co-Authors Alcon Research, Ltd, Alcon Research, Ltd
Outline