Digital Divide: The impact of public sector web accessibility on users with visual impairments

Websites should be accessible to everyone, including those with accessibility or visual requirements. This includes individuals with impaired vision, motor difficulties, cognitive impairments, learning disabilities, deafness or impaired hearing. Almost 9 million people in the UK have accessibility needs, and around 3 million people are colourblind, which can make accessing digital services more difficult.

Accessibility is especially important when it comes to public sector and healthcare websites, when visitors may be in search of vital information. It is estimated that hospital websites alone receive over 6.8 million visits from online searches every month*. That’s millions of people relying on the functionality of the website to provide them with the information they need, in an accessible manner.

In 2018, PSBAR 2018 (Public Sector Accessibility Regulations) was instated requiring public sector bodies to make their website or mobile app accessible by making it ‘perceivable, operable, understandable and robust’. These criteria are often referred to by the acronym POUR. However, this study seeks to investigate the extent to which features that support accessible web access are being implemented, making access to the information on the website available for certain user groups, who may have low vision or colour deficiencies.

The study

We assessed over 1,300 websites across local authorities, social housing and healthcare to assess how accessible these websites were for a user initially trying to access the site. 

The core metrics we evaluated to give a final ‘accessibility score’ were the number of errors on a page, the number of contrast errors and the number of alerts. 

Errors are issues that will impact certain users with disabilities or fail to meet Web Content Accessibility Guidelines (WCAG). 

Alerts are simply elements that may cause issues. 

Contrast errors are instances of text that do not meet WCAG contrast requirements. Colour contrast impacts the readability of a webpage’s content. It is especially important for users who have low vision or for users who have colour blindness.

The report found a gap in the accessibility of pages across all sectors, particularly affecting those with visual impairments or colour vision deficiency, that would rely on screen readers to interact with this kind of digital content.

Key takeaways

• Public sector pages had an average of 9.1 detectable accessibility errors and 8.0 contrast errors, 5.6 times lower that the web average
• Local council websites only had an average of 2.0 errors and 1.7 contrast errors
• The NHS website pages analysed had an average of 5.4 accessibility errors
• Private hospital websites had 3.6 times as many errors as their NHS counterparts, and more than 4 times as many contrast errors
• Of the public sectors analysed, housing association pages performed the worst, with an average of 13.2 errors, potentially leading to inaccessible housing support and making it more difficult to reach out for help with accommodation issues
• Almost three-quarters of the pages analysed had at least one colour contrast issue that could impact low-vision users
• The average ‘accessibility score’ given to websites was 5.83

Accessibility of local council websites

Local council websites are relied on by citizens for key information pertaining to their local area, council services and community, as well as their rights and responsibilities as citizens. A website with poor accessibility would make it more difficult for residents to access this important information.

When looking at each local authority’s websites, there were a number of accessibility issues that arose on the pages analysed, however, there were significantly fewer contrast issues than in other sectors analysed. In fact, there were only an average of 1.7 contrast errors on each of the council pages analysed, meaning the impact for those with visual or colour deficiency issues is less significant. Nevertheless, many of the least accessible council pages demonstrated this issue.

If users are unable to easily access these websites, they may miss out on important information and services. Local council websites are the source for information on paying council tax, contacting the council, waste collection and other important services that may be out of reach for those with accessibility needs. As a consequence, the may miss significant deadlines or opportunities.

The average accessibility score for local council websites was 8.05, better than the average score for all websites analysed, at 5.83. The lowest score (i.e. the least accessible) was Gloucester City Council, scoring only 2.33. However, scores in other sectors went as low as 0.32, so local authorities performed well overall. In fact, over 200 council pages showed no detectable errors at all.

The most common accessibility issues that we encountered on these local authority webpages were:

1. Very low contrast
2. Noscript elements
3. Redundant title text
4. Redundant links
5. Missing form labels
6. Broken ARIA references

Many of these issues are significant as they impact the effectiveness of screen readers. Screen readers are a popular tool for people who are blind or have limited vision that are navigating the web. They convert screen elements like text, buttons and images into speech or braille, allowing users to interact with the digital content. When pages lack the correct fields and descriptions, important information can be lost to screen reader users.

Accessibility of hospital websites

The accessibility of hospital websites is an important issue, as it can impact the ability of people with long-term or temporary disabilities to access important information and services.

When comparing NHS hospitals to private or independent sector hospitals, private hospitals performed significantly worse on average when it came to accessibility barriers. This could perhaps relate to the fact that private websites are not constrained to the same accessibility legal requirements as public sector websites.

However, this does not mean that private healthcare institutions do not have an obligation to their current and potential service users. 

Looking at website homepages, NHS hospitals, on average, had 5.4 accessibility errors per page and 5.6 contrast errors. Private hospitals, on the other hand, saw 20.4 accessibility errors and 22.9 contrast issues on average. This meant that the average accessibility score assigned to private hospitals was 45% lower than that of NHS hospitals.

The most frequently occurring accessibility barriers on NHS websites were:

1. Very low contrast
2. Redundant links
3. Redundant title text
4. Empty links
5. Very small text
6. Suspicious alternative text
7. Linked image missing alternative text
8. Missing/empty form labels

For private hospitals, the most common accessibility issues that arose were:

1. Very low contrast
2. Noscript elements
3. Redundant links
4. Missing form labels
5. Suspicious link text
6. Empty links
7. Missing alternative text
8. Suspicious alternative text
9. Linked image missing alternative text
10. Empty headings
11. Empty form labels
12. Orphaned form labels
13. Missing fieldsets
14. Empty buttons

There was a greater range of issues on private hospital pages than on NHS pages. However, barriers to access such as low contrast and missing text that would assist screen readers occur in both cases. 

Accessibility of social housing websites

Social housing and housing association websites are an important resource for people who live in social housing. They provide information about housing options, tenancy agreements, and other essential services.

According to the latest government statistics, disabled people were more likely to rent social housing than non-disabled people. In the UK in 2019, a quarter of disabled people rented social housing, compared with just 8% of non-disabled people. A lack of digital accessibility can inhibit these individuals’ ability to access key information and participate in their communities.

However, as many housing associations offering social housing are private organisations, they do not yet need to comply with certain regulations for digital accessibility, meaning that potential residents may be struggling further to use these services.

Of the websites analysed for this report, housing association websites had the highest average errors and alerts.

Of all the pages analysed, BPHA’s was the worst performing according to our accessibility score and had the most errors, with 155 errors - almost twice as many as the page with the second most errors.

The most frequently encountered errors and warnings on the worst-performing housing association pages included:

1. Very low contrast
2. Redundant links
3. Missing form labels
4. Suspicious link text
5. Empty links
6. Redundant title text
7. Linked image missing alternative text
8. Noscript elements
9. Skipped heading levels

Many of these are straightforward to fix, they simply need to be made a priority by website owners or developers. Making a website fully accessible benefits all users, improving the user experience for all, not just those with access or assistive technology needs.

Executive summary

This report examines the accessibility of local council, housing association and hospital websites in the UK, focusing on the impact that a lack of accessibility can have on users that have visual impairments. We found that there is a significant gap in the accessibility of some of these websites, with many having a number of errors and alerts that could make them difficult to use for individuals with access requirements. However, this was not the case for all websites and many performed well against standard accessibility criteria.

The report highlights the need for a greater focus on accessibility as well as a need for continued development and maintenance of these essential websites. By making these websites more accessible, public sector organisations can ensure that all potential users can access the information and services they need, in the most simple and accessible manner possible.

Interested in learning more about website accessibility? NEC Digital have share some of the most common website accessibility issues and how to resolve them in the article here: Resolving common website accessibility issues. 

Methodology

We used the WAVE API to analyse the accessibility of web pages using the WAVE processing engine. 

We looked at the number of overall errors, contrast errors and alerts to assess which were the websites with the most accessibility issues. We used the homepage of each site as an indicator of the amount of issues that would be. A web page with many accessibility errors on a core page like the homepage is likely to have accessibility issues across the site that have not been addressed.

To rank the public sector websites, we considered the number of accessibility errors, the number of contrast errors and the number of ‘alerts’. We then used these metrics to rank-order the websites from highest to lowest. 

Error count: Double weight (50% of final score)

Contrast error count: Full weight (25% of final score)

Alert count: Full weight (25% of final score)

We then manually analysed the top 10 pages with the WAVE accessibility evaluation tool to gather the issues affecting each page. The top occurring issues listed in the report are those that occurred in at least four of the top 10 pages for that sector.

The list of housing association websites was taken from (those with linked websites) www.housing.org.uk/about-housing-associations/our-members.

The list of hospital websites was taken from www.nhs.uk/about-us/nhs-website-datasets. Any domains containing .nhs were counted as NHS or public sector hospitals. Those without were deemed ‘independent sector’, or private hospitals.

The list of local council websites was taken from www.local.gov.uk/our-support/guidance-and-resources/communications-support/digital-councils/social-media/go-further/a-z-councils-online.

*Data on the number of searches was collected from SEO tool Ahrefs’ search volume data.

Data gathered in August 2023.