A quantitative content analysis about DUIC was performed on public web pages using the Google Canada search engine.

To be included, the web page had to (1) have information related to cannabis and driving, (2) be available in English, (3) be accessible with no fee, (4) have text to analyze, and (5) be available at the time of analysis. Web pages were excluded if (1) the page became no longer available during analysis and (2) the web page only contained images.

Web pages were identified through the Google search engine. Google was chosen because it is the dominant search engine in Canada, holding 91.98% of market shares [29], and one study showed that 89.8% of people preferred using Google [30]. A private search through incognito mode was used to avoid the search history from biasing results. Six separate Google searches were performed using the terms outlined in Textbox 1, and the first 20 URLs were collected from each search. Neutral search terms were used to ensure the collected web pages were not biased in one direction. The first 20 URLs were collected, as most people consider no more than the first 20 web pages when performing an internet search [15,31]. For our study, one researcher (SS or MJ) extracted web page addresses with Google Chrome (version 99.0.4844.51) [32]. The search was first completed in October 2021 using the first 4 search terms and then repeated fully in April 2022 after 2 new search terms were added.

Web pages were organized into categories based on categories used in similar studies that assessed the quality of health-related information on the internet [33]. These categories included digital media, commercial web pages, government organizations, health organizations, nonprofit foundations, peer-reviewed materials, and “other.”

Descriptive statistics was performed with the mean (μ), standard deviation (σ), and total sample size (n) being reported for the average QUEST score of all web pages and by category. To assess correlations between QUEST scores and readability scores (ie, GFI, FRES, FKGL, and SMOG), a Pearson 2-tailed test was performed [33]. To assess the QUEST score for the presence of the HON code, an unpaired 1-tailed t test was performed, testing if the HON code was present on web pages with higher QUEST scores [33].

This study was exempted from ethical approval because it does not involve human participants.

A total of 120 web pages were identified for analysis (Multimedia Appendix 2). Of the 120 web pages, 34 were removed as duplicate web pages, and 4 were removed as they did not meet the eligibility criteria, leaving 82 web pages included in the study (Figure 1). Of these, 40% (33/82) of web pages were categorized as digital media, 20% (16/82) as commercial web pages, 13% (11/82) as government organizations, 12% (10/82) as health organizations, 10% (8/82) as nonprofit foundations, 4% (3/82) as peer-reviewed content, and 1% (1/82) as “other.” Multimedia Appendix 3 presents the web pages included in the data analysis.

The range of the QUEST scores was between 7 and 27, with the average Quest score being 17.4 (SD 5.6) out of a total of 28 (Table 3). Average QUEST scores by category showed that the peer-reviewed category had the highest quality with a score of 26.3 out of 28, and government web pages scored the lowest at 10.0/28.

The HON code was only present on 4 (5%) web pages, and they were found in the categories labelled as commercial (n=2), nonprofit (n=1), or health organization (n=1). There was no significant difference (P=.2) between the presence of a HON code on a website and the QUEST score without a HON code. Multimedia Appendix 4 presents the data from the full quality evaluation for each web page.

The average readability scores were 9.7 (SD 2.3) for FKGL, 11.4 (SD 2.8) for GFI, 12.2 (SD 1.9) for SMOG index, and 49.9 (SD 12.3) for FRES. Assessing the readability scores for all web pages in relation to the universal readability score, 19.5% (16/82) of the web pages were universally readable by the FKGL score (score <8 considered universally readable), 16% (13/82) by the FRES score (score >60 considered universally readable), 11.1% (9/82) by the SMOG index (score <10 considered universally readable), and 9.8% (8/82) by the GFI score (score <8 considered universally readable). None of the web pages in the peer-reviewed or other categories were considered universally readable by any readability scoring tool (Table 4). Multimedia Appendix 5 presents the readability scores for each web page.

A Pearson 2-tailed test showed a significant positive correlation between the QUEST score and the FKGL (r=0.41; P<.001), GFI (r=0.28; P=.01), and SMOG (r=0.34; P=.002) readability scores. A negative correlation was found between the QUEST score and the FRES score (r=–0.40; P<.001).

Of the 79 web pages that were eligible to be reviewed for accuracy, 23 web pages discussed information related to the timing of the effects from cannabis consumption; 31 web pages were related to the prevalence of DUIC; 53 were related to the effects of cannabis impairment on driving ability; 51 were related to the risk of collision; and 49 had information related to detection by law (Figure 2). From those, 96% (22/23) had accurate information on the effects from cannabis consumption; 97% (30/31) of the web pages had accurate information about the prevalence of DUIC; 92% (49/53) of the web pages presented accurate information on the effects of cannabis impairment on driving ability; 80% (41/51) of the web pages had accurate information on the risk of collision; and 71% (35/49) of the web pages presented accurate information on detection by law. Sample excerpts from web pages and accuracy categorization are included in Multimedia Appendix 6.

This study assessed the quality, readability, and accuracy of information about DUIC found on the internet using the Google search engine. Our findings showed that peer-reviewed papers had the highest quality of information; however, these web pages were not considered universally readable. The difficulty with comprehension may lead to misinterpretation and inaccurate expectations [52,53].

Surprisingly, our research indicated that government web pages were rated as having the lowest quality, contrary to the general perception that government sources would contain high-quality information. This low rating was likely attributed to the tone of the presented text, given much of the information on government websites was one-sided, used strong words such as “will,” and did not discuss limitations of the information presented. This low quality could be due to the fact that government websites often presented information on laws and regulations and did not provide references to other information. This is unfortunate, as government web pages are typically viewed as an accurate source of information as indicated by various academic guides for evaluating information sources [54,55].

Readability for the public was problematic for most pages, with less than 20% of all web pages considered readable based on the FKGL, GFI, SMOG, and FRES readability tools. The majority of the content was written at a higher level of reading, which would often be used in academic settings or postsecondary education. The paucity of web pages written at levels that were considered universally readable was consistent with other health information topics on the internet (eg, general surgical procedures [56] and total joint arthroplasty [57]), suggesting that this could be a wider issue than solely information on cannabis [33,47,58]. Kruger et al [58] also suggests that significant efforts are still needed to provide accurate cannabis-related information on the internet for the health and safety of individuals and society [58].

The readability could be contributing to the misperceptions and behaviors; however, further studies assessing the interpretation of high-quality information with low readability scores could be beneficial. Associations and health advocacy groups should consider the health literacy of the public [59] when creating content to educate the public on DUIC. In addition, a more active form of education for the public could be beneficial as opposed to the passive information presented on a web page.

Our research shows that 80% of the information available about DUIC and its risks for accidents was accurate. However, although most information on DUIC was accurate, it was the lack of complete information that was most concerning. Of the 79 web pages that were analyzed for information about DUIC, 48% (n=38) either had no information on the risk of collision or had mixed or inaccurate information. Misperceptions surrounding cannabis particularly do not recognize the increased risk of accidents associated with DUIC, which highlights the need for comprehensive and accurate information [6,10], as many people turn to the internet to find information about cannabis [8,60] and about health in general [61].

Contrary to our finding that many web pages generally presented accurate information regarding DUIC, Lau et al [20] found that around 80% of the internet claims were inaccurate when investigating the information related to cannabis health benefits. This may suggest that the evidence regarding DUIC is less debated compared to suggested health benefits of cannabis; still, DUIC behaviors persist despite the presence of easily accessible accurate information [5,6]. Studies have shown that both adolescents and adults have a low risk perception of cannabis [62] and feel they are in control of their driving after cannabis consumption [63]. This is problematic given the evidence that cannabis can significantly impair motor coordination, judgment, and reaction time [64,65], increasing the risk of motor vehicle accidents [66].

This study has a number of limitations. Although we have used what appears to be the most appropriate tools to evaluate web page information, there are no best practices for conducting this type of research. Second, the QUEST tool does not have a target quality score or a threshold of acceptable quality, and therefore, we can only make relative comparisons with the web pages included in this study. Finally, we made the assumption that peer-reviewed content was accurate and excluded those sites from the accuracy assessment. However, there is no guarantee that all peer-reviewed materials are fully accurate. Fortunately, only 3 web pages fell into this category, so this would have minimal impact on the overall analysis.

Most of the identified web pages on Google Canada search engine provided accurate information about DUIC; however, the information was incomplete, the readability was generally low, and the quality of information varied depending on the source. Health organizations should consider health literacy of the public when creating content to help prevent misinterpretation and perpetuate prevailing misperceptions surrounding DUIC. Delivering high-quality, readable, and accurate information in a way that is comprehensible to the public is needed to support informed decision-making.