Facilitating consumers to find, assess, and understand health information and to make effective decisions based on that information is the impetus for research on health literacy [1]. Further, the increasing availability of online and digital health information motivates a similar need to study digital or eHealth literacy [2]. eHealth literacy is “the ability to seek, find, understand, and appraise health information from electronic sources and apply the knowledge gained to addressing or solving a health problem” [2]. Digital media (eg, internet, mobile apps) have the potential to create new opportunities and streamline information for consumers (eg, tailoring, progressive disclosure). However, they also have the potential to create additional challenges for consumers trying to find and use health information, so the design of the system and how the information is written require careful consideration and study.

One example of consumers requiring health information is when they take medications. Approximately, 4 in 10 Canadians (40.5%) between the ages of 6 and 79 years take at least 1 prescription medication and, unsurprisingly, people are more likely to take medications the older they are [3]. Given the widespread use of prescription medications, consumers should understand and remember information about the medications they take in order to maximize the therapeutic benefits and minimize the risks. Moreover, memory (ie, recalling or recognizing information) and comprehension (ie, understanding information) are factors proposed to affect therapy adherence [4,5]. In this study, we are emphasizing the importance of recalling information about the medication (eg, administration, storage, side effects) as opposed to remembering the particular time to use a medication—a distinct area warranting research. Essentially, by providing consumers with medication information, we are trying to help them understand how to take the medication, what to avoid, what to watch out for, etc. A systematic review of written medication information indicated that consumers appreciate and use this information and it may improve medication adherence [6].

Methods of communicating medication information to consumers need to be carefully studied, designed, and deployed. Relying solely on verbal communications for medication information is not prudent because memory is generally poor. Specifically, consumers only remember 20% to 60% of information that health care professionals discuss verbally immediately after the interaction [7-9]. Therefore, it is important to supply complementary and supplementary information to consumers to improve comprehension, memory, and ideally adherence and therapeutic benefits while minimizing risks. Moreover, merely providing long text-only handouts may not encourage consumers to read, understand, and remember the information.

It is important to explore materials that offer more than simply text to determine the impact of visuals and potentially narration to create more appealing and robust representations of medication information. Many studies have shown that multimedia benefits learning, and there are principles guiding how multimedia can be most effectively applied [10-12]. For example, it may be worthwhile exploring the use of data visualizations for communicating the likelihood of side effects rather than merely relying on vague terms such as “possible” or “common.” There are a variety of worthwhile avenues for exploration to improve medication information beyond what is currently available. This study used a common consumer resource for medication information (ie, consumer medication information [CMI]) and systematically transformed it using multimedia (ie, added images to text, replaced text with narration) to determine the effect of incorporating multimedia on memory, perceptions of comprehensibility, utility, and design quality, as well as overall preference.

CMI attempts to address the need for medication information that can be subsequently referenced. CMI, for the purposes of this study, is the term used for the text-based paper information sheet(s) typically given to consumers at Canadian pharmacies when a prescription is filled for the first time. Although there is guidance for CMI, it is not regulated by Health Canada and unfortunately, as a result, there are often considerable disparities between CMI sourced from different pharmacy chains [13].

CMI contains typical information about what the medication is used to treat and its common dosage, but it may not match the individual consumer’s actual prescription or condition. CMI conveys a variety of general information about the medication including the following: dispensing pharmacy (eg, name, address, phone number), consumer’s name, prescriber’s name, date, brand and chemical (or generic) names of the medication, drug identification number (DIN), conditions that the medication is usually used to treat, how the medication is typically administered, potential side effects, important information about the medication, and how to store the medication.

Many posit that, as currently designed and delivered, CMI and other similar types of medication information offer limited value to users. Findings from a review on written medication information suggests that its value is currently limited because of language complexity, poor visual presentation, lack of tailoring, and use of words rather than numbers to convey risk of side effects [14]. Others have argued that medication information is often difficult to read and not suitable for consumers, especially older people [15] or those with limited health literacy [13,16-18]. Moreover, medication information may not be adequately addressing user information needs by failing to provide answers to questions consumers want to know about their medications [6]. Although we are generally seeing a shift from hard copy materials to digital options or replacements, this has not yet been observed with respect to CMI. However, when this shift inevitably occurs, we should be prepared with evidence to inform the design and deployment of these materials to optimize consumers’ learning and use of them.

Multimedia is an approach to information design that has yet to be systematically applied and investigated for its potential benefits in disseminating health information to consumers [19]. Multimedia research is motivated by evidence that combining multiple methods of communication to convey information is more successful than relying on a single method. Thus, the definition of multimedia is “presenting words (such as printed text or spoken text) and pictures (such as illustrations, photos, animation, or video)” [12]. Domains such as education, entertainment, advertising, and more recently health care have embraced the benefits of multimedia [20]. Additionally, investigations of the potential benefits of multimedia for communication of health information [21], and even medication information specifically [22], for consumers have begun. However, these studies have largely overlooked the body of research done in multimedia learning and therefore the materials developed may not be as effective as possible [19].

Mayer [10] developed the cognitive theory of multimedia learning (CTML) to integrate the evidence and depict how people process multimedia presentations. Effects consistently observed and reported in multimedia learning studies have been organized into a set of multimedia principles that are used to both (1) describe why particular cognitive phenomena occur and (2) guide multimedia design to ensure it is done most effectively [10-12]. Thus, it is important to leverage existing evidence-based multimedia principles for the design of new multimedia health information to optimize its efficacy [19].

Given its demonstrated benefits in other domains, multimedia is a promising method of enhancing understanding and memory of medication information. There are many multimedia principles and new ones are continuously being developed [12]. However, this study only explored the following multimedia and modality principles: (1) people learn better from words and images than words alone [10], and (2) people learn better from narration and images than from written words and images [10].

There are emerging studies that are attempting to improve CMI and other medication information for consumers. However, there were 4 primary factors that were not adequately addressed in other studies that motivated this study: (1) the failure to isolate the effect of multimedia, (2) the limited use of multimedia in stimuli, (3) the exploration of possible multimedia effects for younger people with adequate health literacy, and (4) the dearth of studies examining narration.

First, most previous studies that explored potential opportunities to improve different types of medication information have manipulated multiple aspects of design and content simultaneously. Moreover, most of the recent research seeking to improve medication information for consumers has concentrated on modifying both its content and its layout. There is evidence that various layout redesigns (eg, 2 columns, segmented sections, modelled after over-the-counter drug facts boxes [23]) improve consumers’ perceptions of medication information, such as ratings of comprehensibility [24], utility, or design quality, or all 3 [25], as well as ease of locating information [26], attractiveness, readability [27], attitude toward the materials, and intention to read it [28]. In addition to increasing consumers’ subjective ratings, layout redesigns have also bolstered different aspects of performance, such as locating information more quickly and effectively [24,29], as well as improving comprehension [24,27-30]. However, a major shortcoming of these studies is that the redesigned layouts was paired with changes in the length of the materials. Thus, the content was not controlled and instead were also modified in conjunction with layout. Therefore, comparisons were often between lengthier (control or current practice) and briefer stimuli, which confounded their results. For example, one study [30] compared a 4-page medication guide with a 1-page redesign. Thus, it is not necessarily surprising that consumers understood the shorter materials better, as there was less information that could potentially distract them or exceed their cognitive processing capabilities. Similarly, studies that have added multimedia to medication information typically made modifications to content as well [28,31]. For example, in addition to adding icons to represent dosing schedule, one study also increased the font size, lowered readability scores, and shortened and reorganized the content [31]. Again, the impact of multimedia cannot be distinguished from the effects of other modifications to the stimuli.

Investigations such as those above are valuable because they demonstrate that design and content changes can improve perceptions (eg, ratings of comprehensibility, utility, design quality, attractiveness, readability, attitudes and intentions) and performance (eg, comprehension, memory, information location) of medication information. However, by changing multiple aspects of the stimuli simultaneously, their methods preclude attributing gains to individual factors (eg, multimedia, length, readability, organization, layout). In contrast, this study used the same content for all 3 formats to determine if multimedia affected memory, perceptions, and/or preferences. That is, the exact same words and sequence of words were used to describe a medication, regardless of whether its presentation format (ie, text, text + images, or narration + images). This control allowed for the potential effect of multimedia to be isolated.

Second, studies exploring the impact of multimedia on medication information have generally limited the use of images to complement text to a narrow component of medication information, such as dosing schedules [31,32], directions and precautions [33], or only a few symbols and an image of the medication itself [28]. Thus, to address this shortcoming in the existing literature around use of multimedia medication information, this study included images throughout the entire presentation (eg, indications, side effects).

Third, most studies have focused their efforts on improving medication information using multimedia for particular groups of people who may inherently have more difficulty processing this information and therefore may have the most to gain. Specifically, multimedia medication information has been explored for older people [31,34] and people with limited health literacy [33,35]. However, it is also worthwhile to determine if multimedia benefits people who do not belong to these groups.

Fourth, no studies were identified that have explored the use of narration for medication information specifically. In response, the proposed study created a format of CMI using narration to convey information in lieu of text with complementary images.

This study examined memory, perceptions, and preferences by investigating the following 7 research questions: (1) Is there evidence of a multimedia effect for CMI on memory (ie, does adding images to text impact memory for CMI)?, (2) Is there evidence of a modality effect for CMI on memory (ie, does using narration instead text accompanied by images impact memory for CMI)?, (3) Are there differences in how participants perceive the CMI formats in terms of comprehensibility?, (4) Are there differences in how participants perceive the CMI formats in terms of utility?, (5) Are there differences in how participants perceive the CMI formats in terms of design quality?, (6) Do most participants’ share a favorite CMI format?, and (7) Do most participants’ share a least favorite CMI format?

We used an objective approach to investigating participants’ memory and a subjective approach to determining their perceptions and preferences regarding CMI in 3 formats: text, text + images, and narration + images. We tested participants’ memory by having them respond to free recall questions for each CMI format. We determined perceptions by having participants rate the CMI in terms of comprehensibility, utility, and design quality. Finally, participants ranked the 3 formats from most to least favorite to indicate overall preference.

The number of participants needed to achieve a significant difference between conditions in this study was estimated based on findings from a meta-analysis comparing the effectiveness of static images versus animations [36]. Höffler and Leutner [36] found that the mean weighted effect size was 0.44 for declarative knowledge (ie, memory) in 40 studies. Thus, to calculate the number of participants for the proposed study, we adopted a critical effect size of 0.45, significance level of 0.05, and power of 0.8. Using the aforementioned parameters for 1-tailed tests, as memory hypotheses were directional, a sample size of 28 participants was suggested [37]. However, given the counterbalancing, we needed a number divisible by 6, and to be even more conservative (ie, run 1 more participant in each sequence than suggested), 36 participants were recruited for this study.

To advertise the study, the investigators sent out a call for participants through the University of Victoria’s School of Health Information Science listserv and hung posters on campus to advertise the study. Each participant received a gift card worth Can $20 (US $15.38) as compensation for their time.

Participants were excluded by self-report from the study for any of the following reasons: (1) they had a medical or health professional background (eg, nurses, pharmacists, doctors), (2) they were not proficient in the English language, or (3) they had compromised visual or auditory acuity that was not effectively compensated for by assistive devices (eg, glasses, hearing aids).

Two participants were identified as outliers due to their age (ie, >3 SD from the mean age) and replaced with 2 new participants to maintain equal numbers of participants in each sequence.

We gave participants hard copies of the text format on 8.5×11-inch paper to emulate the current dispensing practice of CMI at Canadian pharmacies. We displayed the remaining 2 formats (ie, text + images and narration + images) on an Apple Macbook Air laptop computer with a 13.3-inch colour display. The text + images format was displayed on a single webpage (scrolling required). Participants were shown the narration + images format as an embedded YouTube video. To keep the exposure timing consistent, participants were only able to watch the video once from start to finish. We recorded the computer screen and audio using QuickTime media player (Apple Inc), even when the computer was not involved (eg, when participants were studying text format) and made an additional audio recording using a digital recorder.

The experiment was conducted in a quiet office. Participants were seated comfortably at a desk and the experimenter sat alongside him or her with the experimental materials that were not currently in use (eg, text format).

A summary of the participants’ characteristics (demographic, educational, and medication related) can be found in Table 1. The mean age of the participants was 23.6 years (SD 3.8; range 18-35). Most participants in this study were female (26/36, 72%), identified as Caucasian (23/36, 64%), and reported English as their first language (31/36, 86%). All of the participants were students. The majority of participants were currently enrolled in school full-time (30/36, 83%). Participants were students of various faculties, but the 3 most common faculties were science (9/36, 25%), social sciences (8/36, 22%), and human and social development (7/36, 19%).

Participants reported using several different resources for medication information. The most commonly reported medication resources were physicians (27/36, 75%). An equal number of participants reported consulting pharmacists (16/36, 44%) and electronic resources (16/36, 44%) for information about medications. Many participants (16/36, 44%) reported not taking any prescription medications daily; however, over one-third (13/36, 36%) of participants reported taking 1 medication daily. Nearly one-half (17/36, 47%) of the participants reported following medication instructions completely.

According to Weiss and colleagues’ marking framework [39], most participants (30/36, 83%) were likely to have adequate health literacy. Six participants (17%) were classified as possibly having limited health literacy. However, no participants had a high likelihood of limited health literacy. Interestingly, using Monkman and colleagues 4-category framework [44] to classifying self-perceptions of eHealth literacy using eHEALS [40] scores, only a minority (8/36, 22%) of participants had high eHealth literacy scores [44]. The majority of participants reported only moderate (21/36, 58%) self-perceptions of eHealth literacy [44]. Concerningly, 7 participants (19%) reported low self-perceptions of eHealth literacy [44]. However, no participants lacked self-perceived eHealth literacy skills [44]. Interestingly, there was no correlation between participants’ scores on the NVS [39] and the eHEALS [40], calling into question the extent of the relationship between health literacy and eHealth literacy or the respective measures used [44].

First, to negate any practice effects (eg, studying and rehearsing answers specific to recall questions), participants’ memory in the first condition was examined. A 1-way, between-subjects ANOVA yielded no indication of CMI format affecting memory (F_2,33=0.19, P=.830). Mean number of items remembered on participants’ first attempt with the memory task was 12.00 (95% CI 9.64-14.36; range 5-17) for the text format, 11.25 (95% CI 9.39-13.11; range 7-17) for the text + images format, and 11.75 (95% CI 10.24-13.26; range 9-16) for the narration + images format. Second, to minimize the effect of individual differences (eg, some participants having better memories), participants’ memory in all 3 conditions was compared. Again, a 1-way, repeated-measures ANOVA determined there was no significant effect of CMI format on memory (F_2,70=0.1, P=0.901). The mean number of items remembered in the memory task for all participants was 12.44 (95% CI 11.05-13.84; range 5-25) for the text format, 12.53 (95% CI 11.28-13.78; range 6-21) for the text + images format, and 12.75 (95% CI 11.71-13.79; range 7-218) for the narration + images format.

In summary, there was no evidence to support either the multimedia principle or the modality principle. That is, participants remembered approximately the same amount of information regardless of whether the CMI was presented as text, text + images, or narration + images in the first condition and across all 3 conditions.

All 3 Friedman tests comparing participants’ perceptions of the 3 CMI formats were significant. Specifically, the Friedman tests yielded comprehensibility (χ²₂=26.5, P<.001), utility (χ²₂=8.21, P=.016), and design quality (χ²₂=35.69, P<.001). Post hoc analyses with Wilcoxon signed-rank tests with a Bonferroni correction applied resulted in a significance level set at P<0.017. These pairwise comparisons indicated that participants rated the text + images format higher than both the text format and the narration + images format in terms of comprehensibility and design quality (Table 2). Further, narration + images was also rated significantly higher than the text format on these 2 dimensions. Despite the significant utility omnibus test, differences between the pairwise comparisons did not reach the threshold for significant differences (Table 2).

This study sought to determine whether multimedia CMI impacted memory, perceptions, and/or preferences for CMI. A summary of the findings to the specific research questions posed at the onset of this experiment can be found in Table 3. The use of multimedia (ie, images, narration) in CMI did not appear to have any influence on memory in this experiment. Despite the lack of evidence to support any differences in memory between the CMI formats, there were observable differences in participants’ perceptions of and preferences for the 3 CMI formats (see Table 3). Specifically, the text + images format was rated the highest in terms of design quality and comprehensibility and was also selected by the participants’ most frequently as their most favorite CMI format.

Participants remembered approximately the same amount of information, regardless of what CMI format they saw and thus there was no evidence to support the multimedia or modality principles in this study. If the multimedia and modality effects were observed, the expected pattern of results would have been that participants remembered the most in the narration + images condition, followed by the text + images condition, and the least in the text condition. The results from this study suggest that the CTML [10-12] does not apply to CMI, at least with respect to memory performance as an index of learning. Similarly, King et al [33] failed to show significant effects of multimedia on memory for medication information. Although their study limited their test stimuli to medication directives (ie, directions and precautions) [33], this study used multimedia to complement as much of the written content in CMI as possible. Additionally, this study also investigated whether narration had an impact on CMI memory, which failed to generate differences either.

Do these findings (or more accurately lack thereof) insinuate that developing multimedia materials for CMI and consumer health information is a poor investment? Despite the lack of evidence to support previous assertions promoting the importance of multimedia in consumer health information [19], multimedia may still in fact be very valuable in consumer health communications. There are several reasons why multimedia consumer health information warrants continued investigation: memory ceiling effects; memory, not understanding, as an index of learning; CMI is a fragmented description, not a narrative process explanation; multimedia benefits some more than others; and multimedia improves perceptions and people prefer it.

There were several limitations that may affect the transferability and generalizability of the results of this study. Opinions and performance of young, educated, generally healthy adults, such as those in this sample of participants, may not be representative of other groups of consumers, or consumers as a whole. This study used a convenience sample, which resulted in a predominantly female sample who had higher than expected rates of prescription medication use compared with national averages [3]. Additionally, due to the stimuli exclusion process to enhance equivalency, all pills were excluded. However, pills are likely the most frequently prescribed, dispensed, and used medications. The NVS [39] has only been validated using paper administration, not online administration as in this study. Further, subscales from the adapted CIRF [42] inspired the single-item perception measures of comprehensibility, utility, and design quality. However, collapsing multiple ratings into single-item measures resulted in them being inherently less detailed and made it difficult to determine with any certainty to what extent individual factors influenced these perceptions. Finally, given the time limitations, participants were only exposed to the information in the narration + images condition once verbally, whereas—depending on their reading rates—they may have been able to revisit information in the other 2 conditions (ie, text, text + images) more than once.

There are several valuable conclusions to be drawn from this study. Like other consumer health information, effort has been exerted to develop CMI and human resources are continuously invested into dispensing them to consumers in hope that they will help educate people on the benefits and minimize the potential consequences of risks associated with medications. However, merely providing materials to consumers does not ensure that they will use them and indeed usage rates of medication information tend to be low. For example, medication information reading rates in a similar sample of university students (N=306; mean age 23.6 years) found that 37% of participants reported reading CMI always or often, and an alarming 32% participants reported reading it rarely or never [28]. Thus, if making these materials more appealing to users increases the likelihood of them being used, that would be a worthwhile investment. Therefore, it would be worthwhile to investigate whether incorporating multimedia into CMI has a positive impact on reading rates of these materials.

CMI also creates challenges around disentangling comprehension from memory. Future work would benefit from determining methods to examine comprehension independently from memory and information localization. Arguably, memory is important in circumstances when CMI is unavailable, whereas information localization and comprehension take precedence when CMI is available. Unfortunately, the current practice of distributing CMI as a hard copy often renders them unavailable. However, this situation will likely be remedied when digital methods of CMI distribution are adopted.

This study focused on only a narrow aspect of eHealth literacy competencies, but other facets of eHealth literacy could be explored using CMI. Specifically, we developed this experiment on the premise that consumers received medication information, as is currently the typical practice in North America. Therefore, the focus on this study was whether participants would remember different aspects of the information to simulate addressing or solving a health problem (eg, experiencing a side effect, missing a dose) rather than the acts of seeking, finding, and appraising health information from electronic sources. Thus, there are many other aspects of citizens’ actual medication information use that warrant exploration. For example, do people use paper copies of CMI or online resources instead? What online resources do citizens prefer? If CMI was digitized, how would citizens like to receive it (eg, in a mobile app, on a pharmacy website, by email)? Moreover, at what point in the prescription process would citizens want digital CMI? What factors would impact the usage rates and efficacy of digitized CMI?

Although multimedia is a potentially valuable tool for consumer health information, the conditions in which benefits are observed may be limited to specific people, specific stimuli, or other specific contexts. For example, in this study with a sample of younger, adequately health literate people, no improvements in memory for a specific type of health information (ie, CMI) were observed. That does not preclude benefits of multimedia for other types of multimedia consumer health information for older people and/or people who have limited health or eHealth literacy, who may arguably be helped more by multimedia materials. Moreover, despite the lack of objective improvements as a result of multimedia, subjective improvements (ie, peoples’ perceptions and preferences) for multimedia CMI were significantly enhanced. Although ideally we would have observed improvements in both subjective and objective measures, we cannot discount the importance of peoples’ opinions of consumer health information. Multimedia consumer health information warrants more investigation with respect to what impacts it has on which specific subjective and objective measures and under what conditions (eg, stimuli topics, characteristics of the sample). If evidence suggests that performance and perceptions of certain groups of people are affected variably by multimedia information, it may further motivate argument for tailored health information that aligns with individuals’ information needs.