Skip to main content

Cochrane Library

Summary

Accessiblü conducted a high-level accessibility evaluation of the Cochrane Library, accessed via the University of Washington’s library portal, to assess its usability for individuals with disabilities. The review was conducted using the JAWS 2025 screen reader on Windows 11 with Google Chrome, keyboard-only navigation, and manual inspection, supplemented by automated scanning with axe DevTools, for conformance with select WCAG 2.2 AA success criteria.

The Cochrane Library is a globally recognized collection of databases containing high-quality, independent evidence for healthcare decision-making. The platform hosts Cochrane Reviews, systematic reviews and meta-analyses authored by the Cochrane Collaboration, along with Cochrane Protocols, clinical trial records, and Clinical Answers. The evaluation covered four key page types: the main library homepage, a search results page for the query “ovarian cancer,” a detailed review article page for a systematic review on prostate cancer screening, and the Advanced Search page.

The Cochrane Library demonstrates several features that serve as a good foundation for accessibility, including a skip-to-content link, meaningful landmark regions (banner, navigation, main, and content information), a well-organized filter panel on search results, and clearly structured review content organized under recognizable sections such as Abstract, Methods, Results, and Authors’ Conclusions. The PICO framework used on review pages provides structured information that benefits all users, including those using assistive technology.

That said, the evaluation identified a number of areas across all four pages that may create challenges for users relying on screen readers, keyboard navigation, and other assistive technologies. The most significant barriers center on interactive controls that are programmatically coded as the wrong element type, missing status messages when search results update or load, keyboard focus management issues on the search page, and data tables on review articles that lack proper header markup. Addressing these areas would meaningfully improve the experience for researchers, clinicians, and library patrons who rely on assistive technologies to access Cochrane’s evidence-based health information.

Key Findings

Across all four pages tested, screen reader users encountered a recurring pattern: interactive elements intended to open menus, trigger content updates, or switch between content types were programmatically coded as simple links or static text rather than buttons, tabs, or other appropriate roles. This meant JAWS announced them with no indication that activating them would change page state. Navigation items across the site, the content-type tabs on search results, the collapse/expand controls on review articles, and the language selector buttons all fall into this pattern.

Beyond role mismatches, the search experience creates two significant barriers: a partial keyboard trap that occurs when the search autocomplete activates, cycling focus back to the top of the page when users attempt to Tab forward, and the complete absence of status messages when search results load or filters are applied. A screen reader user who submits a search query or applies a date filter receives no confirmation that anything changed.

On individual review pages, in-text citation links are not visually distinguishable from surrounding body text in a way that is independent of color, which affects users with color vision deficiencies. Data tables presenting clinical findings lack proper header markup, making structured navigation by row and column impossible for screen reader users.

Top 3 Issues Identified

  1. Interactive Controls Coded with Incorrect Roles Across All Pages

    • Brief Description: Navigation menu items (Cochrane Reviews, Searching for Trials, etc.) are coded as links but open dropdown submenus, requiring button semantics with aria-haspopup and aria-expanded attributes. The Cochrane Reviews / Protocols / Trials / Editorials / Clinical Answers tabs on search results behave as tab panels but are coded as links, and collapse/expand controls use non-interactive elements with aria-label attributes placed on <i> tags, which is a prohibited ARIA use. 
    • Impact: Screen reader users cannot determine the function or state of these controls without activating them, adding significant overhead to every navigation and search task. Every page on the platform is affected
    • WCAG Success Criteria: 4.1.2 Name, Role, Value (A); 1.3.1 Info and Relationships (A)

  2. Search Field Creates Keyboard Partial Trap with No Status Messages 

    • Brief Description: When a search is submitted or filters are applied, JAWS receives no programmatic notification. Focus returns to the top of the page and no live region announces the result count. The search autocomplete creates a partial keyboard trap: attempting to Tab away from the search field cycles focus back to the top of the page rather than moving forward, requiring Escape key intervention that is not communicated to the user. 
    • Impact: Keyboard-only and screen reader users must navigate the entire page structure multiple times to locate updated results, with no confirmation that their interactions registered correctly. This affects every search and filter operation. 
    • WCAG Success Criteria: 2.1.2 No Keyboard Trap (A); 4.1.3 Status Messages (AA); 2.4.3 Focus Order (AA)

  3. Review Article Data Tables Lack Header Markup and Links Lack Visual Differentiation

    • Brief Description: Summary of findings tables and evidence tables on review pages have no table header (<th>) elements and no scope attributes, making structured cell-by-cell navigation impossible. In-text citation links throughout review articles are distinguished from surrounding body text only by color (dark navy on black), with no underline or other non-color indicator. 
    • Impact: Researchers and clinicians with visual impairments cannot navigate evidence tables effectively. Users with color vision deficiencies cannot identify inline citations. These barriers affect the platform’s primary content delivery function. 
    • WCAG Success Criteria: 1.3.1 Info and Relationships (A); 1.4.1 Use of Color (A)

Disabilities Impacted

Blind and Low-Vision Users

  • Issues: Screen reader users encounter controls announced with the wrong role throughout the platform, missing ARIA states on navigation menus and content-type tabs, a partial keyboard trap in the search field, no live region announcements when search results update, inaccessible data tables on review pages, and untitled iframes on the search results page.
  • Impact: These barriers compound across a typical research workflow. A blind researcher navigating to the Cochrane Library, searching for a topic, applying filters, and opening a review article encounters role mismatches, missing feedback, and navigation confusion at nearly every step. The data tables containing clinical evidence are not fully navigable.

Users with Motor Disabilities

  • Issues: Navigation menus that function as dropdowns but are coded as links require mouse interaction to use intuitively. The partial keyboard trap in the search field requires Escape key knowledge to resolve. Collapse/expand controls on review article sections require repeated interaction to determine their state.
  • Impact: Keyboard-only users, switch device users, and users of voice control software will encounter unpredictable focus behavior during search operations and must Tab through considerable page structure after each search interaction.

Neurodiverse Users

  • Issues: The heading hierarchy on the homepage jumps from H1 to H2 to H4 without an H3, creating structural confusion. The absence of status messages after dynamic content updates means users receive no confirmation that their actions worked. The non-standard date entry format on the date range filter is not communicated to users, and error messages for invalid entries are not announced programmatically.
  • Impact: Users who benefit from predictable page structure, explicit action feedback, and clearly communicated input requirements will find the search and filter experience frustrating. The lack of status messages is particularly challenging for users who need explicit confirmation that their interactions have registered.

Users with Color Vision Deficiencies

  • Issues: In-text citation links on review article pages are distinguished from surrounding body text only by color (dark navy #002d64 on black body text), yielding a contrast ratio of approximately 1.55:1 between the link and surrounding text. No underline, border, or other non-color visual indicator is present.
  • Impact: Users with protanopia, deuteranopia, or other forms of color vision deficiency may not be able to identify inline citations as interactive links within review article text, limiting their ability to explore referenced studies and supporting evidence.