This study used a cross-sectional website evaluation design. A total of 429 web pages were collected from the health care section of the City of Helsinki website at the page level. Accordingly, the analysis was restricted to the health care section of the City of Helsinki website, which was selected as the primary object of analysis. Representative screenshots of the City of Helsinki health care website are provided in Figure 1 to illustrate the website interface and page layout.

In this study, public reporting websites were conceptualized as textual information resources. It should be noted that a substantial proportion of the City of Helsinki web pages consisted of short headings and hyperlink labels.

Internal web pages within the City of Helsinki health care website were included in the analysis. Hyperlink labels appearing on these web pages were retained as part of the visible text because they constitute information presented to users. In contrast, hyperlinks directing users to external websites were recorded but not followed or included as analytical units.

Furthermore, because readers do not naturally filter out redundant or navigational content when accessing information on web pages, all textual elements were retained for analysis. Therefore, no minimum text length threshold (eg, excluding texts containing fewer than 100 words) was applied, allowing all visible English-language textual content to be included in the readability analysis.

All eligible visible textual content within each web page was aggregated into a single web page–level text corpus before readability assessment. Therefore, headings and hyperlink labels were not evaluated as independent short text units. Readability was then assessed at the web page level using 7 established readability indexes.

This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cross-sectional studies. The completed STROBE checklist is provided in Checklist 1.

All websites were assessed in November 2025. Data collection was conducted using Python (version 3.9.6; Python Software Foundation), with packages including Requests, Beautiful Soup, Fake-useragent, urllib.parse, and openpyxl.

First, all textual content was scraped from the website. To ensure accuracy and minimize errors, author LD reviewed the dataset. The study then established a Microsoft Excel sheet containing the URLs of each website (Multimedia Appendix 1).

Subsequently, all 429 extracted URLs, together with the textual content from each web page, were exported to a Microsoft Excel file. The web page was defined as the primary unit of sampling. Subsequent verification of the URLs revealed that some represented anchor URLs pointing to different sections within the same web page. Duplicate textual descriptions across different numbers were identified, revealing that 183 URLs contained duplicated text (Multimedia Appendix 2).

Nevertheless, website users encounter these web pages individually during navigation and do not manually remove duplicate content. After identifying duplicated textual content and identical readability scores, 246 unique textual records were retained for statistical analysis.

The text associated with each URL was recorded as the corresponding description. Only English-language web pages were included because the readability indexes used in this study were validated primarily for English-language text and rely on English-specific linguistic features such as word length, syllable count, and familiar word lists. In addition, PDF documents, images, and other nontextual content (eg, external links) were excluded from the readability analysis.

6162

Following data extraction, web page records with identical textual content and identical readability scores were identified and deduplicated using a custom Python script (Multimedia Appendix 3). Readability indexes and text characteristics were then summarized using descriptive statistics.

Readability-related text characteristics were extracted for each web page, including the total number of characters, words, and sentences; mean sentence length (words per sentence); total syllables; number of polysyllabic words; and the proportion of polysyllabic words among all words (Multimedia Appendix 4).

Readability is commonly defined as the ease with which a text can be read and understood [61]. In health literacy and patient education research, it is often described more precisely as “the study of matching reader and text,” so that the reading level of the material corresponds to the reader’s ability to read with understanding [62].

Readability was assessed as an indicator of linguistic complexity that may influence users’ cognitive processing of public reporting information. Existing conceptual frameworks for public reporting websites include readability alongside dimensions such as accuracy, completeness, design, usability, and accessibility [63]. Automated website assessment frameworks operationalize these dimensions across domains such as accessibility, content, marketing, and technology [64].

Numerical scores were generated automatically using predefined computational algorithms. No subjective human scoring was involved. Before the readability assessment, web page text was preprocessed by removing noncontent elements, including scripts, HTML markup, stylesheets, content from external websites, and PDF documents where applicable.

Within the Nibbler website assessment framework, readability is evaluated as an indicator within the “Accessibility” and “Content” categories [65-67]. Given that the conceptualization and operationalization of readability vary across disciplines, this study used 7 established readability measures to capture complementary aspects of text difficulty. These measures differ in their underlying linguistic features and weighting schemes, allowing the consistency of readability assessments to be examined across alternative operationalizations.

Table 1686569,704171727365

The primary outcomes were the 7 readability indexes (FRES, FKGL, GFI, SMOG, CLI, ARI, and DCRS). Descriptive statistics, including the mean, SD, and minimum and maximum values, were then calculated and are presented in the Results section.

Using the readability measures as shown in Table 1, readability was assessed using the Python Textstat library, which was used to calculate 7 established readability measures: the Flesch Reading Ease Score (FRES) [68], Flesch-Kincaid Grade Level (FKGL) [65], Gunning fog index (GFI) [69,70], Simple Measure of Gobbledygook (SMOG) [41], Coleman-Liau index (CLI) [71], automated readability index (ARI) [72], and Dale-Chall readability score (DCRS) [73]. These indexes estimate the reading difficulty of health-related materials and have been extensively applied in previous studies evaluating online health information [65].

Although some measures rely on overlapping variables, they differ in how these variables are weighted and interpreted. The FRES estimates reading ease, whereas the FKGL, GFI, SMOG, and CLI provide approximate US grade-level estimates. The ARI similarly provides a grade-level estimate based on character and sentence length, whereas the DCRS incorporates lexical familiarity through a predefined familiar word list. Thus, the 7 measures were treated as complementary indicators of readability rather than as 7 independent outcomes.

For interpretation, higher FRES scores indicate greater reading ease, whereas higher scores on the other grade-level or difficulty-oriented measures indicate greater reading difficulty. The US Department of Health and Human Services sixth-grade reading level recommendation corresponds to a FRES score of 80 or higher and FKGL, GFI, SMOG, and CLI scores of 6.9 or lower [74-76].

This study analyzed only publicly available website content and did not involve human participants, personal data, or any intervention. Therefore, formal ethics approval was not required. According to the Finnish National Board on Research Integrity guidelines, ethical review is generally not required for research based solely on publicly available data [77].

This study found that patient-facing public reporting web pages consistently exceeded the readability levels recommended for patient education materials, suggesting that transparency alone does not guarantee that health care information is understandable or usable by patients. Although public reporting websites make health care performance information publicly available, the linguistic complexity of the content indicates a potential challenge for patient-facing communication. This distinction is particularly important when patients seek public reporting information to compare health care providers, understand health care performance, or inform their health care decisions. When information is publicly available but difficult to understand, formal transparency may not translate into meaningful access or effective use.

The magnitude of the observed readability difficulty further underscores this concern. The mean GFI score of 14.43 corresponds to a readability level estimated at approximately 14 years of formal education, or a college-level reading requirement, whereas the SD of 2.43 indicates moderate variation in readability across web pages [79]. The FRES results similarly indicated a college-level reading requirement. These findings substantially exceed the grade 6 to 8 readability level recommended by the American Medical Association and the National Institutes of Health for written health materials, particularly those intended for individuals with limited health literacy [80,81]. Thus, when patients turn to public reporting websites to obtain information relevant to their health care choices, the relatively high reading difficulty observed in this study may represent a potential challenge for comprehension and subsequent use of the information [82]. Rather than simply increasing the amount of information available to patients, patient-centered public reporting should therefore consider whether users can understand and meaningfully apply the information provided.

The consistency of the findings across the 7 readability measures further strengthens this interpretation. Although these measures differ in their underlying variables, weighting schemes, and interpretations, they consistently indicated a relatively high level of reading difficulty. FRES and FKGL both incorporate sentence length and syllables per word but differ in how their scores are interpreted, whereas the GFI emphasizes complex words, the SMOG focuses on polysyllabic words, the CLI and ARI rely on character-based features, and the DCRS incorporates lexical familiarity. The convergence of these measures suggests that the observed difficulty is not attributable to a single readability formula but reflects a broader pattern of linguistic complexity in patient-facing public reporting content. The relatively high GFI scores indicate substantial linguistic complexity across many web pages, potentially including specialized medical terminology that may be difficult for lay users, such as patients and family members, to process [82]. Despite differences in their underlying formulas and weighting schemes, the 7 indexes therefore provide converging evidence supporting the robustness of the overall findings.

These findings also have implications beyond the immediate readability of patient-facing materials. Health literacy involves individuals’ ability to access, understand, evaluate, and use health information, whereas data literacy encompasses the ability to interpret and meaningfully use data in a given context [83]. In increasingly digital health environments, effective use of public reporting information may require both forms of literacy. Within public reporting systems, this process involves multiple groups, including health care professionals who generate or interpret performance information, developers who translate underlying data into digital interfaces, and patients and other users who ultimately interpret and apply the information presented to them. Public reporting websites therefore function not merely as channels for patient education but also as shared information environments in which health literacy and data literacy are enacted. Ensuring that complex performance information is communicated in accessible language may consequently support more equitable participation in the interpretation and use of health care data.

High-quality information is valuable only when it can be readily understood and meaningfully used by individuals with different levels of health literacy [84]. There is previous evidence suggesting that individuals with higher educational attainment tend to prefer access to raw data, whereas those with lower educational attainment rely more heavily on intuitive representations, such as color coding and graphical displays [85-87]. The present study did not examine differences in users’ educational backgrounds or health literacy. Instead, the high readability difficulty observed in the public reporting web pages draws attention to the linguistic accessibility of patient-facing information and the need to consider variation in users’ information needs and capabilities when designing and communicating public reporting information. Future research could examine whether combining appropriately simplified textual information with visualizations and other user-oriented presentation formats can improve users’ comprehension, interpretation, and use of public reporting information.

Previous studies have emphasized that narrative-based communication may strengthen users’ understanding and interpretation of public reporting systems by improving the transparency, credibility, and perceived relevance of publicly reported health care information [88-90]. Building on this literature, our findings suggest that contextualized communication alone may not be sufficient if the language used to convey the information remains difficult for the intended audience to understand. Readability should therefore be considered complementary to narrative and other communication strategies aimed at improving transparency and credibility. In the context of patient-provider shared decision-making, public reporting systems may be more effective when information is not only transparent and contextually explained but also presented in a language that patients can readily understand and apply to their health care decisions.

Increasing transparency of health care performance reporting should be regarded not only as making health care performance data publicly available but also as ensuring that these data are presented in ways that facilitate public understanding. Accordingly, government agencies responsible for public reporting websites may consider incorporating readability evaluation into routine website quality assessment. Automated readability assessment tools could be integrated into website maintenance workflows to identify pages requiring revision and improve the consistency of patient-facing information.

These considerations may become applicable as digital health information is incorporated into AI-enabled health information services. Publicly available web pages may serve as information sources for systems that retrieve, summarize, or generate health-related information. Improving the quality, consistency, and readability of public reporting web pages may therefore have implications beyond direct patient use by improving the quality of information available to downstream digital and AI-enabled health information services. However, this potential benefit should be interpreted cautiously as further research is needed to determine whether improvements in web page readability translate into greater accuracy, interpretability, or reliability of AI-generated health information.

Improving the quality of public reporting websites should be considered alongside the broader processes through which health care information is stored, structured, validated, and published. The public reporting data presented on websites represent only a subset of the information maintained within health care databases [91,92]. Website data generally consist of structured and unstructured content, with semistructured formats such as XML functioning as an intermediary layer for data validation, exchange, and standardized representation [93]. Optimizing XML-based data structures during database development can improve the accuracy and consistency of public reporting data, thereby enhancing transparency [94].

At the same time, redevelopment of existing information systems typically requires substantial time, financial investment, and organizational resources, making ecosystem redesign impractical in many health care settings [95]. Under these constraints, optimizing intermediate data structures (eg, XML) represents a practical strategy for improving key data processing workflows and enhancing the consistency and standardization of publicly reported information, thereby supporting more transparent and reliable public reporting. However, improvements in the underlying data infrastructure should complement rather than replace efforts to improve the readability and user-centered presentation of information. Ultimately, effective public reporting requires both reliable underlying data and communication practices that enable patients, health care professionals, and other users to understand and meaningfully use those data.

Previous research suggests that the way in which information is structured and visually presented can influence how users attend to and respond to publicly reported health care information [96-100]. Websites commonly use visual elements such as icons, symbols, and color to communicate information, whereas the quantity and presentation of information may also affect users’ ability to process and use it [101-103]. Studies of public reporting and health care information presentation have further examined users’ responses to quality ratings, visual displays, and other forms of information presentation [96-98]. These findings highlight the importance of considering not only what information is presented but also how it is structured and communicated to users. We provide an initial examination of the readability of the underlying information presented on patient-facing public reporting web pages, highlighting a potential issue at the data and content level that warrants further investigation.

From a practical perspective, data development processes could therefore incorporate annotations and contextual information within the underlying code to better accommodate users. Such annotations could provide greater clarity regarding the meaning and interpretation of data elements, which may contribute to more readable and understandable information when the data are subsequently presented on web pages. Similarly, web page design should consider how colors, symbols, and other visual elements are likely to be interpreted by users. Selecting and applying these elements in ways that are intuitive and readily interpretable may further support the accessibility and usability of patient-facing public reporting information.

This study has several limitations. First, the readability analysis was performed only on web pages with textual content. Although no minimum text length threshold was applied, manual inspection indicated that web pages containing fewer than 100 words were mainly composed of headings, navigation labels, short notices, or introductions. Therefore, these short text segments were retained for descriptive statistical analysis. Future website research should perform word segmentation and apply appropriate text exclusion criteria once sufficient textual data are available to ensure more robust readability assessment.

Second, the readability results were based on automated algorithmic scores rather than direct user testing. During the manual review of all extracted texts, we observed that the public reporting system presented relatively limited performance data but included a dedicated feedback form for reporting website-related issues. As user heterogeneity in health literacy, educational level, language proficiency, and familiarity with health care terminology exists [42,104], future studies could investigate dashboard-based public reporting systems with user testing that integrate audiovisual elements to enhance information presentation [105,106].

Finally, the data were obtained from the English-language health-related pages of a single municipal website in Finland. Therefore, the findings primarily reflect this specific website and its linguistic and operational context, and caution is warranted when transferring the findings to other public reporting environments. Future research could examine public reporting websites across different organizational, operational, and language settings. In particular, comparative analyses of Finnish- and English-language content could determine whether readability differs across language versions of the same website.

This study evaluated the readability of the City of Helsinki public reporting website from the perspective of health care information design. The findings consistently demonstrated that the website content exceeded the readability levels recommended for patient education materials and exhibited considerable variability across web pages.

These findings suggest that the effectiveness of public reporting depends not only on the transparency of health care data but also on how such information is encoded, organized, and communicated to end users.

Improving readability and narrative-based communication may enhance users’ understanding of public reporting information and, ultimately, promote more informed health care decision-making. Future research should further examine multilingual public reporting websites and investigate how different information presentation strategies influence users’ comprehension and engagement across diverse health care contexts.