Welch Allyn RETeval-DR, Directions For Use Manual

Page: 57 / 60

• The timing of the electrical response to the 32 Td
∙
s stimulus
• The amplitude of the electrical response to the 16 Td
∙
s stimulus
• The ratio of the pupil area between the 4 Td
∙
s stimulus and the 32 Td
∙
s stimulus
To help ensure accurate results, enter the correct birth date.
Individuals with diabetes who have severe retinopathy typically have pupils that change
size less than the pupils of healthy individuals. If the patient is on medications or has
other conditions that impair the pupil response, extra care must be taken to properly
interpret the RETeval-DR device results, as these individuals are more likely to be
erroneously classified as likely to have vision threatening DR. Further, ensure the
contralateral eye is covered by the patient’s hand to prevent uncontrolled light
stimulation of the contralateral eye from affecting the pupil being measured. Do not use
the DR Assessment Protocol on patients whose eyes are pharmacologically dilated.
The result from the DR Assessment Protocol is interpreted as being low, normal, or high.
Default “Normal” results are within the reference interval 7.1 – 19.9. A result greater
than or equal to 20 (which by default is interpreted as “high” by the device) was
indicative of vision- threatening DR in the study population. When weighted for the
prevalence found in a typical diabetic population, the device has a sensitivity of 83%, a
specificity of 78%, and a negative predictive value (NPV) of 99%, assuming the NHANES
2005-2008 prevalence of 4.4% (Zhang et al. 2010). Using the same cutoff of 20, if CSME
is ignored and only patients with ETDRS 53+ are considered, the sensitivity improved to
87%, the NPV improved to 99.2%, while the specificity remained the same.
The reference interval can be adjusted (see "Device Settings") to trade off sensitivity,
specificity, and negative predictive values as described in the following table for varying
the upper point of the reference interval. The default lower point of the reference interval
is three standard deviations below the mean for the patients without retinopathy. Thus,
only about one person per thousand should fall below the lower limit and be interpreted
as low.
Performance metrics with primary care prevalence
Percent of tested sample
Predictive power of
test
Positive test Negative test
RETeval
cutoff value
Sensitivity Specificity Positive Negative True False True False
17.6 93% 52% 8% 99.5% 4.1% 46% 50% 0.2%
LCL 73% 47% 5% 97% 2.4% 41% 45% 0.0%
UCL 99.9% 57% 13% 100.0% 6.5% 51% 55% 1.3%
18.1 90% 57% 9% 99.2% 3.8% 41% 55% 0.5%
LCL 65% 52% 5% 97% 2.2% 36% 50% 0.1%
UCL 99% 62% 14% 99.9% 6.2% 46% 59% 1.7%
19.1 88% 67% 11% 99.1% 3.8% 31% 64% 0.5%
Directions for use Appendices 53