Application of Optovue Trend Analysis in the Management of Patients
Written by Larry J Alexander OD FAAO Saturday, 27 December 2014
Change or trend analysis is another tool to apply in the management of patients with ocular conditions that may affect the optic nerve, the retinal nerve fiber layer and the ganglion cell complex. It must be remembered that this is not the absolute tool for the decision making process but rather another tool in the management armamentarium.
Disclaimer: APPLICATION OF IMAGING IS BUT ONE TOOL TO ASSIST THE DOCTOR IN MANAGEMENT OF THE PATIENT. NO SINGLE TOOL CAN PROVIDE THE DIAGNOSIS AND MANAGEMENT. The content provided herein is for educational and informational purposes only; nothing provided herein should be intended to create a physician-patient relationship or an offer for a contract and does not constitute medical diagnosis or advice. The opinion relating to a review of any issue presented or discussed is the sole responsibility of the healthcare provider or medical advisor. The doctor or patient must assume all liability, including, but not limited to, any direct, indirect, incidental, consequential, or any other monetary or other damages, or liabilities arising out of or relating in any way to the content of the information provided herein.
In all diagnosis and management of disease, the physician is attempting to perform testing that pushes the likelihood of the diagnosis in a positive direction. The entire process relies on statistical probability using both Sensitivity which applies to the ability of a test to give a positive result when the test is applied to a patient known to have the disease and Specificity which applies to the ability of a test to give a negative result when given to a person known to not have the disease (a normal). To simplify even further, when the doctor orders a test to determine an answer, the doctor is trying to improve the likelihood of the diagnosis. In so doing a number of issues arise including false positives (specificity or what is there is not really there) and true positives (sensitivity or what is there is there). All of this comes together under the application of Baye’s Theorum. A test that does not push toward a posterior probability of 1.0 is of limited value. Remember that a posterior probability of 0.5 can be achieved simply by flipping a coin.
Definition of Trend Analysis
In the application to SDOCT testing change analysis affords the physician an estimation of the rate of change of the overall retinal nerve fiber thickness and the overall ganglion cell complex thickness once there are three or more recorded patient visits. There are, however, no limits to the total visits and all events become a part of the analysis. Poor quality scans should be deleted to minimize pollution of the analysis.
The analysis utilizes a simple linear regression model, which is designed to explain the relationship between two or more variables using a straight line. The data collected in the SDOCT scans is applied to a simple linear regression analysis that generates three specific items of information applied to the rate of change:
- Estimate of the Rate of Change…The Slope of the Line
- The 95% Confidence Interval of the Rate…The Range of the Slope
- The P-value of the Significance of the Slope…What is the Likelihood That the Change is Statistically Significant
The Scan Components
The analysis appears on the following report, which shows four RNFL and four GCC scans. By default, two scans from the earliest visits and two from the latest visits are show but these may be modified or deleted from the image display section shown on the left. On the upper right there is a comparison table of the first and last visit as related to the normative database. The RNFL and GCC rates of change are displayed on the bottom section of the report.
A = Up to 4 GCC images…Default First Two Visits + Last Two Visits
B = Up to 4 RNFL images…Default First Two Visits + Last Two Visits
C = Raw Data of GCC and RNFL Comparing First Visit + Last Visit Including FLV % and GLV %
D = TSNIT of RNFL Overlaying the Normative Database Distribution
E = Statistical Analysis of Progression for RNFL on the Left and GCC on the Right
Analysis of the Trending Plots
A break down of the three factors related to the rate of change with clinical application. One must consider the relevance of any of these statistics in relationship to all other patient characteristics paying special attention to the initial starting point for the averages and patient age. For example, starting with an average RNFL thickness of 100 microns, which would represent a reasonably healthy reading, one must realize that all RNFL readings will thin naturally over time. Having said that, a thinning rate of 2 microns per year in this patient would have much less overall impact than the same thinning rate in a patient starting with an average RNFL thickness of 75. Over 10 years the first patient would be down to 80 microns average while the second patient would be at a very significant level of 55 microns average. Analysis of any of the trending plots must be within the framework of the patient’s unique characteristics.
- Estimate of the Rate of Change…The Slope of the Line is a hard number that indicates the rate of change in average thickness in microns over the period of one year. It is presented as that hard number of, for example, 1.5 microns/Yr. In addition to that the number the p-Value of the number is highlighted in light purple (1.5 microns/Yr) if the value is 0.1 and dark purple (1.5 microns/Yr) with white letters if the value is 0.05.
- The 95% Confidence Interval of the Rate…The Range of the Slope. The more narrow the range, the more reliable the slope estimation. In the following the GCC has a much tighter range than the RNFL, which is logical when considering the method of acquisition of the of these two parameters. If the range includes zero, it means that the estimated slope is not statistically significantly different from zero.
- The P-value of the Significance of the Slope…What is the Likelihood That the Change is Statistically Significant. If life were perfect and the patient did not have thinning with age or with disease, the slope would be zero. The p-Value represents whether the estimated slope is statistically different from zero (smaller p value means it is less likely the true slope is zero). Both the slope and the p-value are highlighted in light purple with black text when the p-value reaches 0.1; both the slope and the p-value are highlighted in dark purple with white text when the p-value reaches 0.05.
An Example of Clinical Interpretation
As a clinician analyzing the above scan one could conclude:
- The GCC thickness map shows no apparent visual progression
- The RNFL Analysis shows some variability but does demonstrate some areas of concern based on the normative database.
- The data table in the upper right hand corner demonstrates progressive thinning of both the GCC and RNFL with a progression in the FLV % and GLV %
- The TSNIT demonstrates the same variability in the RNFL as the RNFL Analysis
- The RNFL rate of change is -1.00 micron/Yr with a Confidence Interval that is pretty wide at -5.46 to 3.47 and a p value of 0.43. All in all the Confidence Interval calls into question the variability observed in the RNFL Analysis and the TSNIT.
- The GCC rate of change is -1.93 microns/Yr and is highlighted in light purple indicating the approach toward statistical significance. The Confidence Interval is tight at -3.87 to 0.01 indicating minimal variability. In this case the p value is 0.055 highlighted in light purple indicating an approach toward significance.
- The overall analysis should put the clinician on alert that the trend of decrease in the GCC is approaching significance. Any clinical decision must be based on all factors involved with this particular patient.
As a contrast:
As a clinician analyzing the above scan one could conclude:
- The RNFL rate of change is -3.09 micron/Yr (highlighted in dark purple) with a Confidence Interval that is pretty narrow at -4.59 to -1.60 and a p value of 0.015 (highlighted in dark purple). All in all the Confidence Interval demonstrates minimal variability and adds reinforcement to the significance of the reduction.
- The GCC rate of change is -3.30 microns/Yr and is (highlighted in dark purple) indicating the approach toward statistical significance. The Confidence Interval is tight at –6.05 to -0.55 indicating minimal variability. In this case the p value is 0.055 (highlighted in dark purple) indicating an approach toward significance.
- The overall analysis should put the clinician on alert that the trend of decrease in the RNFL and GCC has reached statistical significance. Any clinical decision must be based on all factors involved with this particular patient.
Another Example of Clinical Interpretation
As a clinician analyzing the above scan one could conclude:
- The GCC thickness map shows statistically significant thinning compared to the normative database
- The RNFL Analysis shows statistically significant thinning OD>OS compared to the normative database.
- The disc areas are identical and within the normal ranges
Adding the Trend Analysis for OD
As a clinician analyzing the above scan one could conclude:
- The GCC thickness map shows no apparent visual progression.
- The RNFL Analysis shows some variability but does demonstrate some areas of concern based on the normative database.
- The data table in the upper right hand corner demonstrates progressive thinning of both the GCC and RNFL with a progression in the FLV % and GLV %...however the progression outwardly appears to be minimal.
- The TSNIT demonstrates the same minimal variability in the RNFL as the RNFL Analysis
- The RNFL rate of change is -0.36 micron/Yr with a Confidence Interval that is pretty tight at -0.93 to -0.21 and a p value of 0.18. All in all the Confidence Interval demonstrates minimal variability and adds reinforcement to the significance of the reduction.
- The GCC rate of change is -0.47 microns/Yr. The Confidence Interval is tight at -1.16 to – 0.22 indicating minimal variability. In this case the p value is 0.15.
- The overall analysis should indicate an initial dramatic thinning of both the GCC and RNFL, but a non-statistically significant progression of thinning over the years with a very high Confidence Level. Any clinical decision must be based on all factors involved with this particular patient.
The fellow eye is represented below.
As a clinician analyzing the above scan one could conclude:
- The GCC thickness map shows definite visual progression.
- The RNFL Analysis shows some variability but does demonstrate some areas of concern based on the normative database as well as an apparent visual progression.
- The data table in the upper right hand corner demonstrates progressive thinning of both the GCC and RNFL with a progression in the FLV % and GLV %.
- The TSNIT demonstrates apparent minimal variability in the RNFL measurements.
- The RNFL rate of change is -0.22 micron/Yr with a Confidence Interval that is pretty tight at -0.82 to 0.37 and a p value of 0.41. All in all the Confidence Interval demonstrates minimal variability and adds reinforcement to the conclusions.
- The GCC rate of change is -0.93 microns/Yr (highlighted in dark purple) indicating the approach toward statistical significance.. The Confidence Interval is tight at -1.62 to – 0.24 indicating minimal variability. In this case the p value is 0.01 (highlighted in dark purple) indicating the approach toward statistical significance.
- The overall analysis should indicate an initial thinning of both the GCC and RNFL, but a non-statistically significant progression of thinning of the RNFL over the years with a very high Confidence Level. Likewise, there appears to be a statistically significant thinning of the GCC over the years. Any clinical decision must be based on all factors involved with this particular patient.
Summary
If the slope (Rate of Change) is deemed reliable (95% Confidence Interval), it could be used to estimate the status of the measurements for the patient over coming years. As stated before this becomes even more important when related to patient age and the initial measurement. For example, a rate of -1.5 microns/year could mean loss of 15 microns in 10 years in the RNFL or the GCC thickness. Should the initial measurements for the RNFL be 70 microns (estimated at 55 microns) versus 100 microns (estimated at 85 microns) measured at age 50, this then becomes a very significant clinical issue.
For a point of reference for the clinician, the following may be applied. Based on the OCT normative database, the estimated age related loss of average RNFL and average GCC is less than 0.2 microns/Yr. It is important to realize, however, that an individual’s age-related loss may have a different rate than the average value. Future studies will address this in more detail. However, if a much higher rate of change is detected in an eye, further clinical evaluation may be necessary. The clinician should pay particular attention to the intraocular symmetry or asymmetry of these analyses as disease usually presents with asymmetry. As previously stated all of these findings must be related to the gestalt of the patient’s data and medical and physical characteristics.
Related Articles
- The Value Of Pre-emptive SDOCT Testing In The Prevention Of Vision Loss
- The Importance Of SDOCT In Differential Diagnosis
- The Power of Focal Loss Volume (FLV%) and Global Loss Volume (GLV%) in the Diagnosis and Management of Glaucoma as Measured by Visual Field Progression
About the Author(s)
Larry J Alexander OD FAAO
Dr. Alexander (1948-2016) was a 1971 graduate of Indiana University School of Optometry. He served in the US Navy then served as a Professor at the University of Alabama Birmingham School of Optometry. Larry contributed to a number of chapters in textbooks and has published three editions of Primary Care of the Posterior Segment, as well as contributed to the professional literature. He also lectured extensively in the area of ocular and systemic disease. His areas of special interest included dysfunctional tear syndrome, glaucoma and macular degeneration. His lessons are the basis for this site and he will be dearly missed.