Skip to 0 minutes and 12 seconds Tunde: Selecting the right screening test for your screening programme is the most fundamental step in setting up your screening service. Screening programmes identify individuals without symptoms. And the right screening test will have to be able to identify those with the disease and those without the disease correctly. We need to consider all of these fundamental steps in order to set up a reliable screening programme at a local level. Prior to selecting the screening test, there are several key things that we need to consider. First of all, we need to consider what equipment is already in place and what equipment needs to be purchased.
Skip to 0 minutes and 57 seconds Then we have to decide if the equipment is going to be used by one user or by multiple users. And is the equipment going to be at a static site or at a mobile screening facility. The next step is to make sure that a reliable power supply is available and power shortages will not damage the equipment and the equipment is maintained regularly. For the test to be reliable, we need to have trained personnel in place and have adequate access to training. Lastly, for the screening test to be acceptable It must not take too long to perform, it must not cause pain or harm and it should only cause minimum discomfort and no long-lasting effects.
Skip to 2 minutes and 0 seconds There are several test methods available to carry out diabetic retinopathy screening. They fall into two broad categories, non-photographic– also known as ophthalmoscopic– and photographic. The non-photographic methods are direct ophthalmoscopy, binocular indirect ophthalmoscopy, and slit-lamp biomicroscopy. Fundus photographic methods use mydriatic and non-mydriatic cameras.
Skip to 2 minutes and 27 seconds Non-photographic methods provide a view of the retina but share some disadvantages. They are subjective. Quality depends on the examiner’s observation and interpretation. They don’t give a hard record– a photograph– for quality assurance audit. And the personnel performing the tests need considerable training.
Skip to 2 minutes and 51 seconds Direct ophthalmoscopy provides a small field of view with a clear view of the optic disc and macular area of the retina, which is needed to identify the key signs of sight-threatening diabetic retinopathy. The main disadvantages of direct ophthalmoscopy are, the view is two-dimensional– as the examiner only uses one eye– and it consistently fails to meet the minimum required 80% sensitivity and 90% specificity measures. For this reason, direct ophthalmoscopy should not be used in a screening programme.
Skip to 3 minutes and 25 seconds In binocular indirect ophthalmoscopy, a trained examiner sees a three-dimensional, but inverted image. This method requires pupil dilation and is best performed with the patient reclined. It provides an adequate view in magnification, but it’s not easy to see the details of any damage to the macula, maculopathy. No screening programme currently uses this method.
Skip to 3 minutes and 50 seconds Slit-lamp biomicroscopy is a common method in the eye clinic. It uses a good slit lamp and condensing lenses to magnify the view through a dilated pupil. The advantages of slit-lamp biomicroscopy are, it gives a three-dimensional view of the retina; a teaching arm can be added for good teaching sessions; in the hands of trained personnel, it can achieve both high sensitivity and specificity. The major disadvantage is that most slit lamps aren’t portable and can only be used in fixed locations.
Skip to 4 minutes and 24 seconds Photographic methods use 35-millimeter film, Polaroid instant film, or digital images. Retinal cameras can capture various fields of view, retinal depth through stereoscopy, and be in color or grayscale.
Skip to 4 minutes and 40 seconds Across most programmes, trained camera operators can achieve the required 80% sensitivity and 90% specificity for a screening test. All images are graded by medical or non-medical trained graders.
Skip to 4 minutes and 56 seconds Digital images are easier to acquire, store, and transfer. And they can be reviewed at the time with the person being screened. Generally, people also find the lower-intensity flash more comfortable. The quality of the digital image significantly impacts a grader’s ability to accurately detect any abnormalities. For example, as cataracts affect image clarity, more ungradable images may be found in an older population. There are several methods available to capture digital photographs. Mydriatic digital cameras require dilated pupils. Two-field stereoscopic imaging using mydriatic digital photography has been found to have a good sensitivity and specificity of around 93% and 87% respectively. The rates of ungradable images is about 4%, although this will vary across different populations.
Skip to 5 minutes and 49 seconds A significant drawback of this method is the risk of mydriasis-induced acute angle closure glaucoma attack, especially in Asian populations.
Skip to 6 minutes and 2 seconds Non-mydriatic digital photography taken by a fundus camera does not require pupil dilation. Drawbacks of this method include higher technical difficulties for media opacities, small pupils, and in obtaining stereoscopic views. All non-mydriatic cameras take better quality images if the pupil is dilated, and this reduces the ungradable rate. This is especially important if corneal haze or cataracts are a problem in the population.
Skip to 6 minutes and 34 seconds Some newer imaging systems use lasers to scan the retina and provide an ultra-wide field image, up to 200 degrees instead of 45 degrees. The lasers are non-contact and do not need a dilated pupil, which is better for the patient. They can also penetrate through opacities such as corneal haze and early cataract. Studies using this method have found it to have a high agreement with the gold standard seven-field photography.
Skip to 7 minutes and 1 second Handheld and mobile phone cameras are likely to be the future in hard-to-reach areas and populations. Currently, not many of these cameras take images good enough to see the retinal details required for screening. On the other hand, many take good enough images of the optic disc and the immediate area– such as the macula– to detect sight-threatening changes for urgent treatment. These cameras should be on the watch list for all programmes. They may change our view of how screening is conducted once they reach the necessary quality for good sensitivity and specificity. Automated image analysis. In 2018, the US Food and Drug Administration approved the first artificial intelligence diagnostic system for detection of diabetic retinopathy in primary care.
Skip to 7 minutes and 45 seconds As an autonomous system, it makes an assessment and interpretation of each image without the need of a clinician. This new technology may provide a significant transformation for DR screening.
Skip to 8 minutes and 2 seconds Additional practical requirements for photographic screening methods include locating the camera in a fairly dark room– especially for non-mydriatic cameras– protecting the camera from damage, power surges, and outages, and ensuring it has a stable electricity supply, and establishing maintenance contracts and training for cleaning the camera.
Skip to 8 minutes and 25 seconds Tunde: Technology moves faster than our understanding of it at the moment. We need to make sure that we future-proof our screening programmes and be more mindful of these new technologies. But these need to be as reliable, valid, and sensitive and specific as the old technologies were in order to make sure that the best quality images are obtained. In summary, test methods for diabetic retinopathy screening range from ophthalmoscopy to fundus photography. Methods suitable for a screening programme need to achieve a sensitivity and specificity of more than 80% and 90% respectively and have a technical failure rate of less than 5% of images being ungradable.
Skip to 9 minutes and 16 seconds To achieve high coverage, it’s also important to consider the appropriateness of the selected test for the local setting. For example, does the equipment need to be portable. Fundus photography relies on specifically trained personnel to screen and grade images in most programmes.
What screening tests are available?
The ideal diabetic retinopathy (DR) screening test has sensitivity of at least 80%, a specificity of at least 95% and a technical failure rate of less than 5% of images being ungradeable. The gold standard photography method for the detection of DR, as defined by the Early Treatment Diabetic Retinopathy Study (ETDRS) group, is stereoscopic color fundus photography in 7 standard fields (30°). This method is also useful for identifying DME and subtle retinal neovascularisation. However, both patients and the photographers find the 7 fields method time-consuming, uncomfortable and labour intensive.
The International Council of Ophthalmologists’ diabetic eye care guidelines state, “Currently, the two most sensitive methods for detecting DR are retinal photography and slit-lamp biomicroscopy through dilated pupils. Both depend on interpretation by trained eye health professionals.” (ICO, 2017)
Professor Tunde Peto, clinical lead of Northern Ireland’s DR screening programme, introduces the video on this step which describes the test methods currently available to carry out DR screening. These features inform programmes’ decisions about which test is best suited for their specific setting. For example:
- Fundus photography creates a permanent record, and for that reason, it is the preferred method for retinopathy assessment. However, well-trained observers can identify DR without photography and there are many situations in which that will be the examination of choice.
- Using any instrument requires training and competence but more skill is needed to carry out indirect ophthalmoscopy and slit-lamp biomicroscopy than fundus photography. Newer, semi-automatic non-mydriatic fundus cameras can be easy to use.
- Media opacities in the eye - in the cornea, lens, or within the anterior or posterior chambers - will lead to photograph or view degradation and each one must be reviewed by trained personnel.
- The use of artificial intelligence to automate grading decisions is on the horizon, but this approach may remain out of reach for many health systems.
As you watch this video, consider what the most suitable DR screening method is for your setting. What are the main reasons for your choice?
© London School of Hygiene & Tropical Medicine CC-BY-NC-SA 4.0