Hey there, I’m Hazel Kay, the founder of Kay Pictures, and as we celebrate our 40th anniversary this year, I wanted to take you on a trip down memory lane. It all started with a problem in the early 1980s of how to make a better vision test for young children who were unable to name or match shapes or letters.
In 1981, not long out of training, and in my first orthoptic job at Bolton Royal Infirmary, our team was tasked with screening all 3-year-olds. It was a daunting task. The population of the town was 260k with over 4k three years olds to test each year. And many of the children found the letter matching test (SG) we used difficult to complete quickly. It didn’t seem like we had the best solution for screening these children, which ignited an idea for a new vision test.
Bolton Royal Infirmary, 1981
I was very familiar with the history and science employed in letter tests. Before 1862, when Herman Snellen introduced his standardised letter tests, vision testing charts were all over the place, leading to inconsistent results. Snellen’s genius was in creating a consistent measurement system using optical science first noticed by ancient astronomers. They realised that the minimum angle between two adjacent objects, that the average eye is able to see as separate, is 1’ arc. This laid the groundwork for the modern vision testing protocols we rely on today.
Picture vision tests for children too young to name or match letters had been developed as early as the 1950s, using pictures that would have appealed to a small child at that time: a steam train, a sailboat, toy soldiers, a rocking horse. But Snellen’s work was nowhere in sight (see images below). The pictures used were approximately the same height as the Snellen letters, but didn’t follow the Snellen construction, or any standardisation – which means they can’t measure the same thing.
Left: Modernised letter optotypes, adhering to Snellen's principles. Middle and right: Picture optotypes from the same time, matching the size of Snellen's letters but crucially not the measurement, so they can't measure the same thing.
I envisioned pioneering a test – a combination of Snellen’s precise optical science with captivating pictures tailored specifically for children. However, materializing this vision was an intricate and arduous journey. I undertook extensive literature searches to check no one else was working on the same idea already (it seemed so obvious what was needed!). To my surprise, I found nothing.
The types of pictures to use was also a challenge. Picture vision tests hadn’t changed over the last few decades, but England had. Would a child in the 1980s recognise a toy soldier or a horseshoe? And even if they did, were these the best choices I could make? It was better to work it out from scratch. I went back to the library and looked at picture books for babies and toddlers, and spoke with speech and language therapists to understand common first words.
I embarked on extensive research, sketching images, rigorously testing them to ensure both recognition and appropriateness. I eventually landed on a 10×10 grid being required for more complex shapes for picture recognition (compared to 5×5 for letters).
Kay Pictures 10x10 grid for picture optotypes compared to Snellen's 5x5 grid for letter optotypes.
During evenings and weekends, I sketched about 100 pictures, aiming for precision within a grid, each meant to engage and test young minds. Some images posed challenges – fitting the grid while staying recognisable proved tricky. Not all pictures met the grid perfectly, like the umbrella in the image here that missed the mark at the bottom. Uncertain if these imperfections mattered, I pressed on.
Taking my hand-drawn images, I sought feedback from as many two- and three-year-olds as I could find, blending my research into my work routine. The 1980s was a different time to undertake research and I was lucky to explore this with just managerial approval, unlike today’s regulated landscape.
Early experimentation with picture optotypes in a 10x10 grid.
A selection of the pictures achieving 80% or more instant recognition.
Without formal research training, I relied on common sense to avoid biases. The recognition phase was methodical, ensuring only instant, unaided identifications counted toward the results.
Amidst my collection, not all pictures hit the mark. Those unrecognizable to children were discarded based on my results. Ultimately, pictures achieving 80% or more instant recognition were selected.
Moving forward, I carefully drew these chosen images to scale, using graph paper and various fine black pens. Testing these scaled pictures against letter charts revealed a consistent trend – my pictures consistently matched or slightly surpassed letter acuity tests by about one line on the Snellen chart. I was thrilled with the outcome.
Initially, my focus was on creating a reliable and consistent picture test for children, spurred by the shortcomings of existing picture tests in our field. There wasn’t a grand plan; it was about improving patient care. If the hospital approved its use locally and nothing more, I might have stopped there.
However, fate intervened unexpectedly. A chance conversation with a sales representative from Keeler sparked years of dialogue about selling my test. Simultaneously, a suggestion came to submit an article to an academic journal. My initial submission faced rejection due to data queries and the request for a correlation coefficient.
Not knowing any statisticians personally, I sought help from a team at Bolton Royal Infirmary. A meeting with a proud statistician armed with new computer technology secured the correlation coefficient I needed. The revised paper was accepted and published in 1983 in the British Journal of Ophthalmology.
Keeler’s interest continued, but they wanted to produce the test on a chart, contrary to my belief that a book format was essential for engaging and therefore accurately testing young children.
Following this setback and problems with other interested companies changing their terms, I chose to venture out on my own. So, with a generous loan from a fellow orthoptist, I founded Kay Pictures, managing operations from a spare bedroom in our Bolton home. Despite printing challenges, a full time orthoptic job and being pregnant that year, I somehow got through it.
Selling the test involved reaching out to orthoptists across the UK, garnering orders and profits within the first year. Recognising the need for smaller acuity sizes and pre-familiarization with pictures, I expanded the test’s range and added support materials like a recognition booklet and near vision card.
Addressing challenges for non-verbal children led to the development of a reduced set of eight pictures, strategically designed to limit guessing. The evolving landscape prompted periodic reviews and improvements, ensuring the test stayed relevant and effective.
Kay Pictures HQ, Bolton, 1984.
The reduced set of 8 pictures.
In 1999, significant updates reshaped the test, introducing a row of pictures with a crowding box and transitioning from Snellen to LogMAR sizing. These changes stemmed from advancements in vision measurement techniques, emphasizing the importance of crowding in detecting amblyopia and the enhanced engagement of children with picture-based tests compared to letters.
The decision to switch to LogMAR, though not widely adopted in clinical practice at the time, was based on its superior consistency over Snellen charts. All tests were reconfigured for a three-meter testing distance due to research revealing no significant differences between vision assessments at six or three meters. This adjustment catered to maintaining children’s engagement during tests.
I continued working part time as an orthoptist and used the test daily, so knew there was more room for improvement.
In 2008, after more than 25 years, a comprehensive overhaul began. The aim was to modernize the test, utilising graphic design software (no graph paper and marker pens on the kitchen table this time!) to create new optotypes, considering factors like shape similarity, outline precision, and the percentage density of black within images. The introduction of crowding boxes around each optotype aimed to optimise acuity measurements for the very youngest and least able children.
Multiple phases, including recognition studies involving hundreds of children and comparison tests with established vision tests, led to the selection of 6 optimal optotypes for the test. Normative data was collected from children with no vision abnormalities, providing a range of expected acuities across different ages, and showing better testability than letter-based tests.
The 6 Kay Pictures modern optotypes.
We haven’t stopped finding new ways to improve. The evolution continued with the development of the Kay iSight Test Professional app, enabling versatile vision testing and remote patient assessments. Additionally, a free app, Kay Say and Match, has aided parents in preparing their children for vision tests through engaging games using the test pictures.
Paediatric vision testing has come a long way over the decades, from what was available when I began training as an orthoptist, to what I started out making in the 1980s, to what we have today. Hopefully it will continue to improve as we do more research and learn more about what makes our vision testing more accurate, more consistent and more engaging for small children.
Eager to explore the test in action? Delve into this comprehensive video detailing the simplicity and effectiveness of the single crowded Kay Picture Test.