McMahon, Timothy T.*; Irving, Elizabeth L.†; Lee, Claudia
A relatively new phenomenon has encroached upon the ophthalmic landscape. This new issue is the ability for patients to purchase spectacles from online vendors via the Internet. This new point of sale potentially bypasses professional measurement of interpupillary distance (PD) and the fitting of the spectacle frame to the patient's face and head. Online sites that provide methods and instructions for patients to measure their own PD now exist. To our knowledge, there are no data on the accuracy and repeatability of these PD measurements when measured by patients themselves or by an untrained friend. The range of PDs found in a Chinese population has been described and provides bounds for what might be considered a reasonable PD.1 In addition, centration errors in premade reading spectacles have been described.2 This study showed that a sizeable portion of these spectacles did not meet the accepted standards for spectacle fabrication. The American National Standards Institute has produced a continually reviewed set of standards for ophthalmic lens fabrication. This includes the accuracy of the PD as measured by the clinician or optician.3
The purpose of this study was to determine the accuracy of self-measurement of PDs using the methods found online in a naïve group of participants compared with PDs measured by a trained examiner.
PARTICIPANTS AND METHODS
Fifty-three individuals were recruited to participate in five experiments. Attempts were made to recruit participants in pairs as some experiments required two people. Inclusion criteria were healthy eyes by self-report and willingness to sign informed consent and participate in the study. Exclusion criteria included the presence of active or past ophthalmic disease by self-report, a history of facial or ophthalmic injury, or strabismus. Participants were also excluded if they were optometrists, optometry students, or other ophthalmic professionals.
PDs were measured in five different ways in the following order. (1) One author (CL) measured each subject's distance PD using a millimeter rule following common and accepted procedures.4,5 (2) The author then measured each subject's PD via a commercially available pupillometer (BRT-II Digital PD Meter, China). (3) Participants then measured their own PD using a millimeter rule employing the procedure found online (http://goggles4u.com/pd.asp. Last accessed September 25, 2011). (4) Each pair of participants measured the other's PD using a PD rule. (5) Finally, each subject measured their own PD using an online Application (App) (http://www.sunmoresystems.com/Sunmore_Systems/Pupil_Meter.html. Last accessed September 25, 2011). At this site, there is a link to a free “Pupil Meter Lite” program which was downloaded to an IPod. The same IPod was used for all participants.
Before conducting the experiments, the project was reviewed by, and received ethics clearance through, the Office of Research Ethics at the University of Waterloo. The Office of Research Ethics at the University of Waterloo adheres to the Declaration of Helsinki regarding human research. All subjects were educated and informed of the procedures that would be performed and written consent was obtained to participate in the study.
For all experiments, the average of two measurements was used to assess accuracy and the difference in the two measurements was used to evaluate repeatability. In some cases, instructions from the source suggests taking three measurements; however to keep the process consistent, only two measurements were taken for each experiment.
PD Rule by Trained Examiner
The trained examiner had the participant fixate his/her right eye on the examiner's left eye. Using a millimeter PD rule placed parallel to the floor on the bridge of the participant's nose, the top edge of the rule was located midway along the vertical meridian of the pupil and the location on the rule associated with the nasal limbus was identified. The subject was then asked to fixate the left eye on the examiner's right eye. The examiner then noted the location of the temporal limbus on the rule. The difference in these two measurements was the distance PD in millimeters.
Pupillometer by Trained Examiner
One examiner (CL) explained what the pupillometer was and then took measurements. This measurement was used by the trained examiner only and served as the gold standard or referent for these experiments. The instrument was placed on the bridge of the nose of the participant and following the manufacturer's instructions, the examiner located the first Purkinje corneal reflex to assess binocular distance PD. Two measurements were taken for each participant.
The pupillometer was calibrated according to manufacturer's recommendation and instructions. Calibration indicated that readings were off by 1 mm consistently. Therefore, all data from the pupillometer were adjusted by 1 mm accordingly during analysis.
PD Rule by Self (Subject)
In this experiment, participants were given a printout of instructions from a website (Goggles4u.com/pd.asp). The participants were then instructed to follow a four-step process reproduced here verbatim. Initially, participants were instructed to face a plane mirror affixed to a wall. (1) “Place a simple mm ruler on your nose so that the starting point of the ruler is exactly at your left or right eye pupil. Keep the scale straight; (2) make sure to look straight; (3) stand in front of the mirror or ask one of your friend (sic) to read the scale to measure the distance between the pupils; (4) repeat this at least three times to me (sic) more accurate.” Two repetitions were actually made for this experiment.
PD Rule by a Friend
In this experiment, one participant (friend) measured another participant's PD following the same procedure as above but instead of looking at one's self in the mirror, the participant looked at the examiner. Two measurements were taken and averaged.
PD Measured by Self Using an Online App
Using the “Pupil Meter Lite” from Sunmor Systems (East Sussex, UK), participants were instructed to follow the online instructions. Using an IPod camera, the participants followed the following instructions:
(Begin Verbatim quotes from the help section of the application)
Pupil Meter Instructions
1. You will need your iPhone and a standard credit/bank card. Make sure you have good lighting, nice and bright, for a clear picture.
2. Position yourself approximately 2 feet (arms length) away from the iPhone, eyes level to it, and hold card under your nose, black strip facing the camera.
3. Keep iPhone and card level. Centre subject's head within frame and take photo to begin measuring.
4. In the captured image tap on your credit card strip, to detect reference area. Adjust green edges to fit the card strip exactly. Centre left/right sights over pupil centers to measure your p.d.
5. Use the Zoom buttons for greater accuracy.
6. Forward button to save your p.d. reading, with image, to your photo library or to email on. (End Verbatim quote)
For each experiment, one examiner was present, answered questions, and ascertained whether the subject completed the task according to the instructions. Data that were collected from measurements that were not done according to the instructions were recorded but excluded from analysis and the examiner reinstructed the subject and had them repeat the measurements.
Four labs were contacted that accept online orders for spectacles and were requested to provide upper and lower limits to PD measurements that would be accepted. Three labs responded; the mean and range for the lower (narrower) PDs accepted were 52 mm and 50 to 54 mm. The mean and range for the upper (larger) PDs were 76.3 mm and 74 to 80 mm. PD measurements that were outside the mean limits were excluded from further analysis.
Accuracy and repeatability were determined for all conditions tested. Accuracy was assessed using the approach recommended by Bland and Altman.6 The average of the two measurements taken by the trained examiner, using the pupillometer, was considered the gold standard or referent in this study. Repeatability was determined using the technique described by Bland and Altman.6 For this assessment, the two individual measurements were compared with the mean of the two.
Data from 52 participants were analyzed for this study. One enrolled subject was not used due to a confirmed diagnosis of strabismus. Although two viable measurements were obtained from all 52 participants, it should be noted that frequently the initial measurements needed to be discarded due to gross subject errors. In 11 of 52 (21%) cases, participants performed the measurements incorrectly and obtained false data, necessitating further instruction on the technique by the examiner and asking the subject to repeat the measurement. Of the total occurrences of bad data collection, four participants actually measured what appeared to be pupil size instead of PD (PD range 2.8 to 4.5 mm). In three cases, PDs >75 mm were recorded (range 120 to 298 mm). In three cases, participants misread the ruler leading to error. In the remaining cases, the subject failed to follow the instructions and did not obtain a measurement.
Examiner Pupillometer vs. Examiner PD Rule
Fig. 1A demonstrates a minimal bias of 0.6 mm and a 95% limits of agreement (95% LoA) of −0.7 to 1.9 mm. The bias suggests that the mean difference between the two types of measurements yields a larger PD when measured by the pupillometer than the PD rule. Despite this, both measurements fell within the ANSI (American National Standards Institute) Z80.1-2010 standard7 for unwanted prism in spectacles.
Fig. 2A presents the data in a different but equally informative manner. The graph shows the frequency distribution of average difference in PD measurements for the two measurement types in 1 mm bins by the number of participants in each bin. As can be seen, virtually all participants were within ±2 mm. The shaded portion represents those cases that would violate ANSI Z80.1-2010 unwanted prism standard for greater than ±2.5 mm of decentration for spectacle lens powers >2.75 D or 0.67 prism diopters. In this comparison, only one subject's measurement would fall outside the standard.
Self-Measurement (Mirror) vs. Examiner Pupillometer
Fig. 1B displays an increased spread of the data comparing measurements obtained via the pupillometer and self-measurement using a mirror. Again there was a modest bias of 0.5 mm where the pupillometer gave slightly larger average PDs. The difference between the two measurement types varied greatly however. The mean 95% LoA ranged from −5.2 to 6.1 mm. This difference implies the presence of large errors in PDs which could lead to sizeable amounts of unwanted horizontal prism.
In Fig. 2B, the data are displayed for the self-measurement using a mirror as in Fig. 2A for the PD rule. The distribution, though clustered within ±2 mm, encompasses a wide range. In some cases, the difference between the measures is larger than 10 mm.
Friend Measurement vs. Examiner Pupillometer
Fig. 1C shows that there is a meaningful difference in the average measurements between these two measurement techniques. Having a naïve friend measure a PD using a PD rule, even with verbal instructive support, yielded on average a 2.0 mm smaller PD than the pupillometer. The 95% LoA also was large ranging from −3.8 to 7.8 mm.
For Fig. 2C, not only is the spread of data significant but the peak values lie outside the ANSI Z80.1-2010 standards. The most frequent difference in PD (mode) between the “friend” and “pupillometer” was 2 to 3 mm smaller for the “friend” than that recorded by the examiner with a pupillometer.
Self-Measurement with App vs. Examiner Pupillometer
Fig. 1D reveals an even larger mean bias of 3.2 mm, indicating that the IPod App yielded smaller PDs than the gold standard by >3 mm. The spread was also quite large with a 95% LoA ranging from −3.1 to 9.6 mm.
Finally, Fig. 2D shows a similar spread and trend data as in Fig. 2C, demonstrating that the majority of cases fail to meet the ANSI Z80.1-2010 standard, and the peak frequency of the difference in PD between the two measures is 2 to 3 mm smaller for the App than that recorded with a pupillometer.
Repeatability was assessed by comparing the first measurement to the second for the pupillometer and then referencing the spread to the mean or bias. Fig. 3A shows the results for the pupillometer measurements. The mean bias was −0.03 mm, with a 95% LoA of −0.8 to 0.7 mm, well within the ANSI standard.
Examiner PD Rule
Fig. 3B shows the variation of the two measures for the examiner PD rule measurements. The bias is small at 0.1 mm. The 95% LoA ranged from −1.0 to 1.2 mm and was also well within the ANSI standard.
Fig. 3C reveals a slightly larger but possibly meaningless bias of 0.6 mm. The spread is noticeably increased with a 95% LoA of −3.6 to 4.8 mm.
The 0.10 mm bias is small with a 95% LoA of −3.7 to 3.9 mm. The spread of the data suggests that having a friend take PD measurements is slightly more repeatable than measuring one's own PD but remains much less repeatable than when a trained examiner measures the PD.
Self-Measurement with App
Fig. 3E displays the repeatability data for the IPod App. Although the bias is small at −0.1, the 95% LoA is quite large at −6.6 to 6.5 mm.
The frequency with which patients are obtaining spectacles via internet, although small at present, is increasing. The Vision Council of America conducted surveys in 2007 and 2010 estimating the prevalence of internet spectacle sales.3,8 In 2007, 1.7% of all spectacle sales in the United States were purchased from the internet and by 2010 the surveys indicated that this number increased to 2.8%, a 64.7% increase.
Fig. 4 displays the amount of induced prism associated with optical center displacement (y axis) for a given spherical lens power (x axis). The solid lines plotted represent the prism induced for varied decentrations and optical powers. The ANSI standard is also plotted. There are two conditions for the standard that account for the strange shape of the function. Using the power method for lens powers <2.75 D, a maximum tolerated prism is ≤0.67 prism diopters. Over 2.75 D of lens power, the standard shifts to a maximum of <2.50 displacement error.7 The dashed lines represent 95% LoA in the positive direction for the different PD measurement types. It is clear that for very low-powered corrections, large PD errors would not lead to clinically significant induced prism. However, for moderate- to high-powered corrections, errors well within the 95% LoA for the self-measurements could result in induced prism exceeding 10 prism diopters.
The advent of obtaining spectacles via the internet, bypassing ophthalmic professionals, is a new matter of concern. Concerns about patients avoiding healthcare providers and seeking glasses on their own are varied and intensifying. These concerns range from patient safety and loss of income by optometrists, opticians, and ophthalmologists to patient perspectives of personal rights and ease of access to these products. A number of internet providers do not require a valid prescription from a licensed provider but patients do need to select a frame and do need to obtain their distance and sometimes their near PD measurements to complete the online order. Normally, this measurement is determined by the eye care practitioner or dispensing optician. Typical accuracy and repeatability when obtaining PD measurements is within ±1 to 2 mm for a PD rule compared with a pupillometer.4,9–11
It should be noted that there is the potential for experimental bias as the trained examiner measured PDs using both a PD rule and a pupillometer and was not masked to the results. Because the PD rule was used before the pupillometer, if there was a bias it would likely be toward a similar reading. It is important to realize that we used only the pupillometer as the referent or gold standard in this study.
Beyond the issue of obtaining a reliable PD measurement is the notion that the quality of spectacles obtained through internet sales is questionable. Citek et al.12 recently showed that the error rates in spectacle corrections, prism deviation, and impact testing were remarkably poor from a sample of 308 lenses from 154 pairs of glasses. In their effort, 28.6% of lenses failed at least one measure of optical performance using ANSI standard Z80.1-2010 as a reference. 22.7% failed impact testing standards. Overall, 44.8% of spectacles failed at least one measurement. This poor showing is concerning from a safety perspective in particular and from a consumer satisfaction point of view as well. In our study, we selected naïve participants, gave them instructions on how to perform each measurement technique, and recorded their measurements for each. The trained examiner (CL) was a second-year optometry student and not a seasoned optometrist. Using an optometry student would most likely affect the accuracy and repeatability results by decreasing accuracy and repeatability to some degree compared with a highly experienced optometrist using a pupillometer. Even so, the results using both the pupillometer and PD rule for the trained examiner were quite good and more accurate and repeatable than all the other measures and within the ANSI Z80.1-2010 standard.
In the real world, patients seeking spectacles would not have the advantage of coaching, instruction, and tutelage. We therefore contend that the result for self-determination of PD measurements in this study is actually better than what could be expected without such help. Even with this instructive assistance, the self-measurements by patients or naïve patients measuring each other's PDs were poor. Again approximately 20% of participants would have provided useless PD information in the online order had errors not been detected during the measuring process. Even when participants performed the measurements as instructed, 95% LoA expanded from six- to ninefold compared with data obtained from a trained examiner. Given the poor results obtained with the current self-measurement techniques available, patients should be advised to obtain PD measurements from trained personnel.
Timothy T. McMahon
Department of Ophthalmology and Visual Sciences
University of Illinois College of Medicine
1855 West Taylor Street (M/C 648)
Chicago, Illinois 60612