Found a study. Looks at least worth looking into.
Photoscreening for Amblyogenic Factors
------------------------------------------------------------------------
Wanda L. Ottar, OC(C),COMT;William E. Scott, MD; and Sandra I. Holgado, MD
ABSTRACT
Background: The Medical Technology Inc (MTI) Photoscreener is a new eccentric photoscreener that is being marketed as a device for the detection of amblyogenic factors in preverbal children. The purpose of this study was to evaluate the accuracy of the MTI Photoscreener in the target population of young, healthy children.
Methods: One thousand and three healthy children between the age of 6 months and 59 months were photoscreened with the MTI Photoscreener. Nine hundred and forty nine children were included in the study and their results were compared with a complete ophthalmologic examination with cycloplegia.
Results: The sensitivity of the MTI Photoscreener was determined to be 81.8% with a specificity of 90.6%. The overall agreement rate was 88.8%. The positive and negative predictive values were 68.9% and 95.2%, respectively. All cases of strabismus and media opacities were detected.
Conclusion: The MTI Photoscreener is an accurate and reliable device designed to detect amblyogenic factors in young children. The camera offers promise as a useful mass-screening tool.
INTRODUCTION
Amblyopia is a common childhood disorder affecting 3% to 5% of the population. 1, 2 The need for early detection of amblyopia and amblyogenic factors is widely recognized. 3, 4 However, most current vision screening methodologies are not effective in screening preverbal children.5
Interest has increased in photoscreening as a method for detecting amblyogenic factors in young children. The methodology is simple: a flash photograph of the subject's eyes is taken. The light reflected from the retina is analyzed to detect refractive errors, strabismus, and/or media opacities. Two types of photoscreeners based on the relationship between the flash source and the optical axis of the camera have been described.6-11 The on-axis system has a coaxial camera and flash source. The off-axis system has a flash source slightly off the optical axis of the camera. Several studies comparing the two systems have found that the off-axis system provides more information with fewer photographs and is better suited for non-cycloplegic refractive screening.6, 12, 14
An off-axis instant film photoscreener, The Medical Technology Inc (MTI) Photoscreener, was commercially manufactured after several studies were conducted with the prototype, Eyecor camera.1, 5-17 The purpose of this study was to measure the sensitivity, specificity and accuracy of the MTI Photoscreener in detecting strabismus, media opacities, and refractive errors in a large population of healthy children between the ages of 6 months and 59 months.
MATERIALS AND CONSTRUCTION
Camera
The MTI Photoscreener was designed and tested to have the same sensitivity on model eye testing as its prototype, the Eyecor camera. The specifications of the Eyecor camera have been previously described.15-17 Some additional modifications were made, including automatic rotation of the flash and a sliding film that places both the horizontal and vertical photographs on the same instant photograph. A 9-D lens with a back focal distance of 5 inches results in a 1:1 magnification. Each photograph was examined for the presence of strabismus, media opacities, and refractive errors. The diagnosis was determined from examination of the photographs as discussed below (see procedure).
A pupil-crescent measurement tool supplied with the camera was used to measure the pupil and crescent sizes in all photographs. The measurement tool is designed with 3 mm through 9 mm offset openings and complete hole openings. Each template is marked with one-mm lines. A small millimeter ruler also is located along the side (Fig 1).
<Picture>
Figure 1: Pupil-crescent measurement tool.
Subjects
One thousand and three "healthy" children between the ages of 6 months and 59 months were examined. Approximately two thirds of the children were photoscreened during routine screening at Public Health and Women, Infants and Children's (WIC) clinics in Iowa and Illinois. The remaining one third were screened in the private offices of pediatricians in Iowa and Texas. Children with a history of an eye problem or who had been previously examined by an ophthalmic professional were excluded. Children with known systemic disease also were excluded. The gender distribution was 51% male and 49% female. Of the patients, 81.3% were Caucasian, 7.8% African American, 5.8% Hispanic, and 2.9% Asian.
Procedure
The purpose of the study was explained, and informed consent was obtained from the legal guardian of each subject. Each child was photographed by one of the authors (S.H. or W.O.) with the MTI Photoscreener in a darkened environment. Photographs were immediately retaken if they were unfocused or if they demonstrated poor subject fixation or a pupil size less than 4 mm. The camera is designed for measurement of the bright crescent to estimate the refractive error. The pupil and bright crescent size were measured to the nearest 0.5 mm with the pupil-crescent measurement tool included with the camera. Myopic crescents were located on the same side as the flash (superiorly and to the left) (Fig 2), and hyperopic crescents are located on the opposite side of the flash (inferiorly and to the right) (Fig 3). By comparing the difference in crescent size, anisometropia (Fig 2) and astigmatism (Fig 3) can be detected. The size of the crescent is related to the amount of refractive error. A two-mirrored system reverses the eyes so that the eye that is to the left is the patients right eye and the eye that is to the right is the patients left eye just as if the examiner were facing the patient. The measurements of each patient were documented and the photographs were scored based on the pre determined photoscreening criteria (Table 1). Each photograph was graded (S.H.) in a masked fashion and identified by a code number prior to the ophthalmic examination. Inter-observer variability was not addressed.
A complete ophthalmic examination including a cover test, mobility evaluation, and pupil assessment was performed by a pediatric ophthalmologist (S.H.) or an orthoptist (W.O.) on each child. Two cycloplegic drops were instilled 5 minutes apart in each child. Children under 12 months of age were given .5% cyclopentolate. All children 12 months or older were dilated with 1% cyclopentolate. If the child presented with dark irides, age-appropriate cyclopentolate combined with 2.5% phenylephrine was installed. Approximately 30 minutes after each installation, a cycloplegic refraction and fundus examination were preformed by a pediatric ophthalmologist (S.H.). Criteria for clinical failure are listed in Table 2. The term failure is used to describe subjects who do not meet the established photoscreening or eye examination criteria. All children who failed the ophthalmic examination were counseled on the need for follow up or treatment.
<Picture>
Figure 2: Photoscreening picture of a subject with a difference in crescent location between eyes. This is an example of anisometropia. Upper photograph--right eye: 3.0-mm myopic crescent in a 5-mm pupil; left eye: 3.0-mm hyperopic crescent in a 5.5-mm pupil. Lower photograph--right eye: 3.0-mm myopic crescent in a 5.5-mm pupil; left eye: 1.0-mm hyperopic crescent in a 6.0-mm pupil.
<Picture>
Figure 3: Photoscreening picture of a subject with a difference in crescent size between the upper and lower photograph. This is an example of astigmatism. Upper photograph--right eye: 0.5-mm hyperopic crescent in a 6-mm pupil; left eye: 2.0-mm hyperopic crescent in a 6.0-mm pupil; Lower photograph--right eye: 3.0-mm hyperopic crescent in a 6.0-mm pupil; left eye: 4.5-mm hyperopic crescent in a 6.0-mm pupil.
<Picture><Picture><Picture>
Analysis of Data
Standard analytical procedures for assessing screening and diagnostic tests were used18 (Table 3). The results obtained from the ophthalmic examination with cycloplegic refraction were compared to the results obtained from photographic analysis. The sensitivity, specificity, overall agreement rate, prescreening probability, and positive and negative predictive values were determined.
RESULTS
Nine hundred forty-nine patients were included in the study. The average age was 28.7 months. Fifty-four patients were excluded for reasons including small pupil diameter (29 patients), inability to obtain retinoscopy (19 patients), poor mydriasis (3 patients), or poor cooperation during photoscreening (3 patients). Mild off-center fixation was seen in 53 photographs (5%), and poor focus in 11 photographs (1%). Interpretation of these photographs did not result in any false positive or false negative results. The average pupil size was 5.38 mm. Anisocoria of 0.5 mm was noted in 140 patients (14.3%) and 1.0 mm in 65 patients (6.6%).
The distribution of refractive errors based on cycloplegic retinoscopy are documented in Table 4. If a patient presented with compound hyperopic astigmatism, he or she was included in both the astigmatism, as well as the hyperopic category. Spherical equivalent calculations were used to determine the level of hyperopia and myopia. The photograph was categorized based on the eye with the higher refractive error. The majority of patients presented with hyperopia. An attempt was made to determine the confidence interval for each type of refractive error; however, due to the small sample size it could not accurately be determined.
On clinical evaluation, 192 subjects failed the eye examination. Table 5 represents the distribution of eye exam failures when each subject is given only one diagnosis. The order in which the subjects were classified is strabismus, media opacitites, hyperopia, anisometropia, astigmatism, and myopia. Therefore, if a subject presented with an esotropia and hyperopia they would only be documented as having esotropia. The majority of patients (97.9%) failed due to significant refractive error. The distribution of eye exam failures changes slightly when each subject is classified in all of their respective categories (Table 6). For example, the previously mentioned hyperopic esotrope would be documented in both the esotropia and hyperopia categories. Table 6 represents all patients who received a cyclopledgic refraction. That includes the 949 subjects included in the study as well as the 29 subjects excluded because of < 4mm pupillary diameter. Although these subjects were excluded, a complete ophthalmic examination was performed.
The results of standard statistical analysis are listed in Table 7. The ability of the photoscreener to correctly identify the presence of an amblyogenic factor was 81.7% (sensitivity). The MTI Photoscreener demonstrated an even greater ability to correctly identify the absence of any amblyogenic factor with a specificity of 90.6%. The overall agreement of detecting abnormal and normal results was 88.8%. Using the postscreening data, the prescreening probability rate of all amblyogenic factors in this population was determined to be 20.2%.
<Picture>
<Picture>
<Picture>
False Negative Results
There were 192 subjects who were failed on the dilated exam (Table 5). Thirty-five of 192 eye exam failures were diagnosed incorrectly with photoscreening, resulting in a negative predictive value of 95.2%. The remaining 157 subjects were identified correctly with the photoscreening camera. The majority of the false negative patients (25/35) had hyperopia of +3.00 D to +5.00 D on cycloplegic retinoscopy. One patient was found to have clinically significant anisometropia (+1.50 D sphere), and one had undetected myopia (-1.00 OD, -1.50 +0.50 X 90 OS). Eight patients failed the eye exam because of an astigmatism (range +1.25 D to +2.00 D). The camera detected all cases of strabismus and media opacities.
False Positive Results
Seventy-one patients (7.5%) were failed on photoscreening, but passed the cycloplegic eye examination. Fifty-two patients (73.5%) exhibited myopic crescents, but were found to have hyperopia of +0.50 D to +2.50 D on cycloplegic refraction. Thirteen patients (18.3%) presented with myopic crescents of 1 mm or more on the photograph, but were found to have less than 1 D of myopia on cycloplegic examination. Five patients showed significant hyperopia on photoscreening, but were found to have less than 2.75 D of hyperopia on cycloplegic retinoscopy.
DISCUSSION
The MTI Photoscreener is a useful screening tool for early detection of amblyogenic conditions, such as anisometropia, astigmatism, high hyperopia, and high myopia. Because only one media opacity and three strabismus cases appeared in the entire series, the sensitivity and positive predictive values for these amblyogenic factors is not known. A large series of patients is required to determine these values accurately. The rechargeable battery with optional AC/DC adaptor permits the camera to be used in almost all situations. The subjects must be photographed in a darkened room to assure proper fixation of the illuminated targets on the subjects forehead, a disadvantage of this camera; however, the darkened room also helps to maximize pupillary dilation, and pupillary diameter of 4 mm is necessary for the most accurate photographic interpretation. Also, if the room is too bright, the examiner cannot see the targets to focus properly.
Several investigators have reported the importance of adequate pupil size in the detection of pathology using photoscreening cameras.15-17 The manufacturer warns of possible problems with photographic interpretation when pupillary diameters are less than 4 mm. For this reason, we excluded all patients with a pupillary diameters are less than 4 mm, after three attempts to increase pupillary dilation by decreasing room illumination. Only 29 (3%) subjects were excluded for this reason. Anisocoria of 0.5 mm was noted in just 14.3% of the photographs. Other authors have reported an incidence of anisocoria as high as 50%.15 Small amounts of anisocoria are common and are not diagnostic.
We have reported a sensitivity rate of 81.8% and a specificity rate of 90.6%. These rates are comparable to other camera systems whose sensitivity rates are reported to be 60% to 94%, with specificity rates of 62% to 94%.6, 17, 19-21 The prescreening probability for amblyogenic factors in our population was 20.2%. It is difficult to compare this with other reported prescreening probability rates of 42% and 63%6, 15 because these studies are conducted on high pathology subjects. Our figure is more representative of the general population.
Many reports exist regarding the accommodative ability of children.22-25 The photoscreening camera is used in a darkened room while the child is fixating on a series red lights to enable documentation of the nonaccommodative state. This technique is usually successful in detecting unaccommodated hyperopia; however, 33 patients, 25 with hyperopia and eight with astigmatism, accommodated on the camera and masked their refractive error. As with other photoscreening devices that are used without cycloplegia, 6, 7, 9, 26, 27 hyperopia is missed when the child accommodates on the camera. We think that the levels of hyperopia in these patients are not amblyogenic because an accommodative esotropia was not present, and patients demonstrated the ability to accommodate thus preventing the development of ametropic amblyopia.
Fifty-two patients (39 with hyperopia < 2.75 D and 13 with myopia < 1.00 D) were diagnosed incorrectly with myopia > 1.00 D. This again can be attributed to accommodating on the photoscreener. A mildly myopic or hyperopic patient would normally present with a small crescent; however, when they accommodate on the camera, a myopic crescent is created.
Lack of standardization of failure criteria for preverbal children limits comparison of sensitivity and specificity rates among different studies using other camera systems. Recently, a group of ophthalmic professionals met and developed preliminary screening criteria for preverbal children (W.L.O., W.E.S., unpublished meeting minutes, March 1995). If the data presented are analyzed using these criteria (Table 8), the sensitivity rate falls to 75.4%, with a specificity of 89.8%, a false negative rate of 50.5%, and a false positive rate of 93.4%. The decline in sensitivity is due to the stringent criteria established for anisometropia > 1.00 D. In a study on anisometropic amblyopia, Kutschke et al determined that only six patients in a population of 124 with anisometropia presented with anisometropic amblyopia of +1.25 D or less. Every one of these patients presented with a manifest strabismus. Therefore, it can not be determined whether the amblyopia was a result of the level of anisometropia or the manifest strabismus. Eight patients had anisometropic amblyopia with +1.50 D of hyperopia. Only three of these patients were determined to have strabismus. Therefore, if the preliminary criteria were changed to fail anisometropia >= 1.50 diopters, the sensitivity increases to 76.8%. The sensitivity further increases to 83.5% when the failure criteria for anisometropia is changed to > 1.50 D.
Each type of refractive error for the 978 healthy children was subdivided into age groups (Table 4). In order to raise the low confidence level for each group, a significantly larger population needs to be screened. Preliminary estimates of statistical power suggest approximately 10000 children should be screened to achieve a high confidence level for each type of refractove error. This would require a multicenter, collaborative effort.
Photoscreening is a useful method for early detection of amblyogenic factors. The main objective is that it be targeted to the population most at risk and most benefited by early intervention. Each photoscreening system must adopt similar pass/fail criteria and compare it to cycloplegic eye exams to determine the sensitivity and specificity rates. The preliminary vision screening failure criteria recently established should be adopted as the standard.
This is the largest documented sample of normative data in which the results of a photoscreening device have been compared to complete cycloplegic examinations. The MTI Photoscreener has demonstrated appropriate sensitivity and specificity rates and is easily performed, has a low cost, and clearly defined test results making it easy to use and interpret.
<Picture>
<Picture>
REFERENCES
1.Atkinson J, Braddick OJ, Durden K, et al. Screening for refractive errors in six to nine month old infants by photorefraction. Brit J Ophthalmol. 1984;68:105-112. 2.Hsu-Winges C, Hamer R, Norcia A, et al. Polaroid photorefractive screening of infants. J Pediatr Ophthalmol Strabismus. 1989;26:254-260. 3.American Association for Pediatric Ophthalmology and Strabismus. Eye care for the children of America. J Pediatr Ophthalmol Strabismus . 1991;28:64-67. 4.American Academy of Pediatrics Committee on Practice in Ambulatory Medicine. Vision screening and eye examination in children. Pediatrics. 1986;77:918-919. 5.Ehrlich MI, Reinecke RD, Simons K. Preschool vision screening for amblyopia and strabismus: programs, methods, guidelines, 1983. Surv Ophthalmol. 1983;28:145-163. 6.Kennedy RA, Sheps SB. A comparison of photoscreening techniques for amblyogenic factors in children. Can J Ophthalmol. 1989;24:259-264. 7.Abramov I, Hainline L, Duckman R. Screening infant vision with para-axial photorefraction. Optom Vis Sci. 1990;67:538-545. 8.Bobier WR, Braddick OJ. Eccentric photorefraction: optical analysis and empirical measures. Amer J Opt Physiol Optics. 1985;62:614-620. 9.Kaakinen KA, Kaseva HO, Teir HH. Two-flash photorefraction in screening of amblyogenic refractive errors. Ophthalmology. 1987;94:1036-1042. 10.Deutsch J, Smellie T, Tovey J. Photorefraction: two methods and their clinical applications. J Audiovis Media Med. 1990;13:124-128. 11.Atkinson J, Braddick O. The use of isotropic photorefraction for vision screening in infants. Acta Ophthalmol (Copenh). 1982;157(suppl):36-45. 12.Morgan K, Johnson WD. Clinical evaluation of a commercial photorefractor. Arch Ophthalmol. 1987;105:1528-1531. 13.Hamer R, Norcia A, Day S, et al. Comparison on- and off-axis photorefraction with cycloplegic retinoscopy in infants. J Pediatr Ophthalmol Strabismus. 1992;29:232-239. 14.Howland H, Braddick O, Atkinson J, et al. Optics of photorefraction. Orthogonal and isotropic methods. Trans Opt Soc Amer. 1983;73:1701-1708. 15.Freedman H, Preston K. Polariod photoscreening for amblyogenic factors. An improved methodology. Ophthalmology. 1992;99:1785-1795. 16.Morris C, Kutschke P, Morris R, et al. Polariod photoscreening for amblyogenic factors in high pathology and normal population groups. In press. 17.Drack A, Stewart S, Scott WE, et al. Photoscreening for amblyogenic factors in preschool population. J Pediatr Ophthalmol Strabismus. In press. 18.Macpherson H, Braunstein J, LaRoche G. Utilizing basic screening principles in the design and evaluation of vision screening programs. American Orthoptics Journal. 1991;41:110-121. 19.Day S, Norcia A. Photographic detection of amblyogenic factors. Ophthalmology. 1986;93:25-28. 20.Kaakinen K, Renta-Kemppainen L. Screening of infants for strabismus and refractive errors with two-flash photorefraction with and without cycloplegia. Acta Ophthalmol (Copenh). 1986;64:578-582. 21.Morgan K, Johnson W. Clinical evaluation of a commercial photorefractor. Arch Ophthalmol. 1987;105:1528-1531. 22.Howland H, Howland B. Photorefraction: a technique for study of refractive state at a distance. J Opt Soc Am A. 1974;64:240-249. 23.Maino J, Cibis G, Cress T, et al. Non-cycloplegic versus cycloplegic retinoscopy in preschool children. Ann Ophthalmol. 1984;16:880-882. 24.Mutti D, Zadnik K, Egashira S, et al. The effect of cycloplegia on measurements of the ocular components. Invest Ophthalmol Vis Sci. 1994;35:515-527. 25.Howland H, Dobson V, Sayles N. Accommodation in infants as measured by photorefraction. Vision Res. 1987;27:2141-2152. 26.Norcia A, Zadnik K, Day S. Photorefraction with a catadioptric lens: improvement on the method of Kaakinen. Acta Ophthalmol (Copenh). 1986;64:379-385. 27.Maslin K, Hope C. Photoscreening to detect potential amblyopia. Aust NZ J Ophthalmol. 1990;18:313-318.
<Picture>Return to Medical Technology &Innovations Inc. Home Page |
