TY - JOUR
AU - , Van Stee, Stephanie K
AB - Abstract Considering the important role of the Internet in health information seeking by consumers, it is critical to examine the health information that is available to them through the Internet. This study contributes to existing knowledge by employing a content analysis to examine visual and textual information on prescription medication websites. A stratified random sample was selected from a list of the 100 most-prescribed medications in the United States. Findings point to under-utilization of audiovisual components on the homepage of prescription medication websites as well as a lack of racial diversity in people pictured. Medications for chronic conditions were more likely to have homepages with a positive emotional tone than those for acute conditions. Further, more depictions of women on homepages predicted a greater number of prescriptions filled. This study includes implications for health education and healthcare professionals, patients and the Food and Drug Administration. Introduction Health-related organizations often use targeted messages, messages designed to appeal to particular recipients with shared characteristics [1], to promote their products and services. Often demographic variables are too broad to use as the sole factor to create an effective message, and diseases are often used to segment audiences in healthcare marketing [1, 2]. Health consumers are unique because their behaviors are impacted by both their condition and ‘consumer-centric’ variables, which allow consumers to identify with a message [3]. As it relates to consumer-related factors, studies of culturally targeted health messages have shown that people’s cultural identities influence message evaluation. For example, culturally framed messages about colorectal cancer screening were related to stronger normative beliefs about the illness for African-American respondents with a stronger racial identity [4]. Similarly, Willis [5] found that direct-to-consumer advertising (DTCA) that reflects the social reality of target audiences is most successful in capturing the attention of audiences due to the consistency of the ‘depiction of shared social identities’ [5]. Similarly, there is a positive correlation between consumer-related factors and trust inspired by the website design [3]. Taking into consideration the needs of the consumer and fair balance in electronic DTCA (eDTCA), this study expands cumulative knowledge related to visual and textual message design strategies in eDTCA. DTCA is increasing in complexity in the context of the digital age, and research on eDTCA is limited, despite the use of eDTCA by all pharmaceutical companies [6]. Some studies of eDTCA have investigated location of risk information on prescription drug websites [7], ‘fair balance’ of risk and benefit information [8] and use of persuasive appeals [9]. Despite this, more studies are needed that address message design of prescription drug websites. The informational content and design elements of eDTCA are important to evaluate because many people use the Internet to seek health information (72% of Internet users) [10] and many people may not have a sufficient level of health literacy, or specifically ehealth literacy, to understand the information they encounter [11]. We know that DTCA influences patient requests for prescription drugs [12], but to what extent are these requests based on adequate information about those drugs? By analyzing visual and textual design components of eDTCA, we can determine to what extent the strategies displayed on pharmaceutical websites educate consumers. Supporters of DTCA believe that it serves an important role in health education and encourages patients to take charge of their health and talk to their doctors, whereas opponents argue that it misinforms consumers and promotes over-prescribing [13]. Many previous studies of eDTCA are dated [9], focus on one specific type of pharmaceutical [8, 14] or are limited to one category of illness (stigmatized illnesses) [15]. Whether pharmaceutical websites are providing fair and balanced information that is readable, as well as the ways in which they market to consumers, has important implications for health educators, clinicians and the consumers/patients themselves. The current study examined textual and visual message design elements of prescription drug websites to expand the current understanding of eDTCA as a specific form of online health information. Although the pharmaceutical industry’s advertising expenditure by medium demonstrates the rise of eDTCA [16], few recent studies have utilized content analysis to analyze message design elements of websites promoting pharmaceutical products. The current content analysis investigates visual and textual content and design features of prescription drug websites (across drug/illness categories) of the most prescribed medications in the United States. As evidence suggests that pharmaceutical advertising leads to improved sales [3, 17], the current study uncovers insights about visual and textual content and design strategies in eDTCA by content analyzing a stratified random sample from among a list of the most-prescribed medications. The purpose of this analysis is not only to describe message content and features but also to demonstrate how pharmaceutical websites’ message design may differ according to the nature of the medication and its ranking (according to the number of prescriptions sold). Readability of eDTCA Literacy is important to consider when examining message design in eDTCA because the text of prescription drug websites typically conveys important information about medications, including risks and benefits to audiences with different levels of education [2]. A previous content analysis of pharmaceutical websites indicated that 100% of the websites included some textual content [9], so it is reasonable to presume that medication information is presented to some extent in textual content of eDTCA. As it relates to health literacy and eDTCA, our focus in the current study is on the readability of medication information presented in eDTCA, which is conceptually defined as the language complexity of textual information. There is reason to believe that textual content in eDTCA may not be accessible to consumers with lower (health) literacy. For example, online health promotion materials for patients often exceed the average reading level of the US population [11]. Furthermore, the Flesch–Kincaid grade level of textual content for contraceptive websites is above the standard of 7–8th grade reading level [14], which indicates that contraceptive eDTCA (and perhaps eDTCA more generally) is not readable for the average US consumer. The differential readability of benefit and risk information on prescription drug websites is also important to consider because the US Food and Drug Administration’s (FDA) regulatory requirements necessitate ‘fair and balanced’ information of risks and benefits in DTCA [7, 18]. The FDA defines a benefit as ‘help provided by a drug for the person who is taking it’ that must be substantiated by clinical evidence for the purposes of advertising regulation, whereas a risk describes various possibilities of harmful effects of the medication [19]. Benefits and risks often appear on prescription drug websites, but their font sizes and locations often differ. Huh and Cude [7] found that approximately half of the pharmaceutical websites they examined employed larger font size for benefits than risks. In addition to the font size, Ledford [14] found that, for contraceptive prescription websites, benefits were placed within easy view on the webpage whereas viewing risk information required scrolling down the page. Similarly, Charbonneau [8] found that risk information on websites for menopausal hormone therapies was not available on the main page and required subsequent clicking. In addition to analyzing readability, the current study also addressed placement of benefit and risk information, but with a sample that extends beyond contraceptive and menopause websites. Information processing models, such as the elaboration likelihood model (ELM) [20] and heuristic-systematic model [21], would suggest that ability is a critical determinant of how a message is processed. If a person lacks the ability to centrally/systematically process a message, they are left to rely on heuristic cues to evaluate the message. In the case of eDTCA, if risk information is too difficult to understand, consumers will rely on benefit information and/or the visual information. This could be problematic to the extent that visual information is congruent with benefit information rather than risk information. RQ1: How readable is textual benefit and risk information in eDTCA? Visual message design of eDTCA Visual design can inform and persuade audiences in many ways, such as capturing the audience’s attention, clarifying messages and conveying messages optically [22]. In examining how visuals influence audiences’ health behavior, analyses often focus on the images’ description and source and reasons for their selection [23]. Previous research has found that aesthetically appealing images can elicit greater engagement by users [24] and that reducing visual clutter and increasing accessibility of information can improve health practices [25]. In print and television DTCA, bar charts can enhance recall of drug efficacy information [26]. Other visual design elements, such as emotional appeals and color scheme, can also affect message perceptions [27–29]. Importantly, visuals can be designed to persuade audiences to behave in ways that affect their health and the health of others. There is very limited research focused on the types of visual message design features used on prescription drug websites. Thus, the current study included visual information on prescription drug websites, which was conceptualized as images that communicate meaning or emotion independent of textual content [22]. Visuals can be incorporated into health message design to help manage the perception of risk associated with a drug [30]. Those who perceive the risk to be higher spend more time accessing information, have more time to think about the benefits and are more likely to be persuaded by visual elements of the website [30]. Macias and Lewis [9] found that 89% of the websites in their sample included at least one photograph, but that only 20% of the sample used video. The placement of visuals on pharmaceutical websites is also important. Risk information is sometimes placed in a location that requires users to scroll down [9] or click to subsequent pages [8], whereas visual content, which tends to emphasize benefits, may be prominent. Dual coding theory provides a framework for our examination of the visual content of eDTCA. Although it is focused on information processing, similar to the models mentioned previously (ELM and heuristic-systematic model) [20, 21], it provides insights for evaluating the content of eDTCA. Dual coding theory suggests that visual information has an advantage over textual information when they are contained within the same message [31]. As visual information is encoded twice, it is easier to recall than textual information and is given priority when processing a multi-modal message. This is certainly true for mediums in which visual and textual information is presented simultaneously, such as televised DTCA, and may be detrimental to consumers’ recall of risk information [32]. To the extent that visual information in eDTCA relates to benefits, rather than risks, it presents an advantage for benefit information over risk information because visuals are relied on more heavily than textual information when they are inconsistent. The current study focuses in particular on the amount and types of visuals that are used on the homepages of prescription medication websites. RQ2: What is the nature of the visual content used in eDTCA? User-centric components of eDTCA User-centric characteristics are conceptualized as textual and/or visual communication that targets particular characteristics of audiences using available data. Specifically, we focus on the demographic characteristics of people pictured on the website as well as financial incentives offered, languages available and geographic scope indicated. Prior content analyses of DTCA in magazines demonstrated trends in the demographics of people in visuals, including depictions of solely women in 49.1% of the ads [33]. Furthermore, advertising in magazines targeting African Americans and women more than doubled after the FDA relaxed requirements pertaining to simplifying information in DTCA ads [33]. In today’s globally connected marketplace, scholars have warned that pharmaceutical companies could disseminate social media ads globally despite the fact that DTCA is only legal in two nations [6, 16]. Thus, the digital era allows more precise targeting of audiences and utilizes a wealth of data to do so—including race, geographic location and gender. The demographics of the people pictured in the ads are important considering that targeted visuals that promote identification can increase behavioral intentions of the audience [34]. Financial information, particularly financial incentives, are relevant as they may prompt consumers to ask for the advertised brand, whether or not it is the best choice for them among alternatives, including generic versions [9]. Geographic scope and languages provided speak to the target audience of the ads and geographic scope also relates to federal regulations. Although DTCA is only legal in the United States and New Zealand, consumers in other countries may still gain access to eDTCA from the United States and New Zealand during online health information searches [35]. The current study considers various user-centric variables [33] and design-focused variables, as explained [28]. User-centric characteristics, consisting of a mix of demographic and psychographic factors [2], examined in the current study include financial status [36], age [37], race [37], gender [38], language [39] and geographic location [16] as well as the health condition and treatment type [7]. RQ3: What are the types of user-centric characteristics communicated in the content of eDTCA? Social media in eDTCA Research related to social media features of eDTCA is scant. Social media may enhance the ability of pharmaceutical companies to target consumer groups [38, 40]. Liang and Mackey [6] found that the top 10 global pharmaceutical corporations had Facebook pages. Of the top 10 global firms they analyzed, Liang and Mackey [6] found that nine of them had content on Friendster and Twitter and eight posted eDTCA on YouTube. According to Guidry et al. [24], social media use for health marketing is both unregulated and under-researched. RQ4: What are the types of social media linked to eDTCA? eDTCA differences across drugs Message targeting research clearly expresses that demographic variables are insufficient for targeting specific audiences because the nature of an illness may not relate to a geographic location or financial status [1]. Thus, a distinction is often made between types of illnesses [7, 15] when it comes to message targeting. Although acquiring prescription medications shown in eDTCA requires a gatekeeper, patients’ requests for some types of medications are typically honored and lead to higher prescribing volume [41]. Huh and Cude [7] analyzed eDTCA by categorizing medications that treat short- and long-term health conditions to investigate differences in presentation of risk information. Approximately a third of pharmaceutical products did not list their risks on the product’s homepage, and these medications were more likely to be for long-term health conditions than short-term health conditions [7]. In the current study, Huh and Cude’s [7] classification for type of medication is employed, with short-term drugs being those used on a repeat, periodic basis and long-term drugs being those used for ongoing maintenance. RQ5: What are the differences in content and design elements of eDTCA according to the longevity of the prescription drug treatment (short-term versus long-term use)? We also sought to determine to what extent the top-prescribed drugs were homogenous or heterogeneous in the types of content and message design strategies that they used. Specifically, we wondered whether the medications with greater sales were using any content or design strategies differently from the medications with lower sales. Previous studies have found relationships between DTCA volume and prescription sales [42] but have not examined how the content and design of DTCA influence sales. More important than the mere volume of DTCA is what those messages convey to consumers. Therefore, our study seeks to address which content and design elements are the greatest predictors of medication sales. RQ6: Which content and design elements in eDTCA are the strongest predictors of product ranking (quartiles according to number of prescriptions sold)? Materials and methods Content analysis is an appropriate method for examining eDTCA as our interest is in evaluating the content and design aspects of the ads themselves. The current study has high external validity by employing real-world correlates, sales numbers, to examine the websites of pharmaceutical companies that are successful. The current study examines visual, textual, user-centric and social media features on prescription medication websites. Sample A study conducted by the IMS Institute for Healthcare Informatics [43] determined the best-selling prescription drugs in the United States dispensed through retail pharmacies. The 100 top-selling pharmaceutical products were listed by the amount of sales generated and the amount of prescriptions filled, and they were reported to the online health news websites [44]. The ranking of the most prescribed medications, which differed from the ranking of the most revenue-generating medications, was used as a more accurate gauge of the influence of eDTCA. The list was split into four quartiles of 25 prescription drugs each and a stratified random sample was conducted to ensure even sampling within quartiles. This technique was used to allow us to analyze differences in content and design of eDTCA according to product ranking (by quartile). Five prescription drugs were randomly selected from each quartile, resulting in a sample size of 20, shown in Table I. The patient version of each website in the sample was located and coded. Coders were trained to code for several categories related to visual, textual and social media components on the homepages of websites. Huh and Cude [7] emphasized treating the homepage of pharmaceutical products as separate than the entire website due to its prominence. Similarly, Ledford [14] demonstrated that benefit information does not require users to scroll, demonstrating prominence denied to some risk information. This measure of prominence has been used as a rationale for the study the homepages of health websites [28]. Table I. Sample of prescription drug websites (listed in decreasing order by number of prescriptions sold) Name of medication . Medical condition/ purpose of medication . Short-term versus long-term . Website . Lantus Solostar Diabetes Long-term https://www.lantus.com/ Vyvanse ADHD and binge eating disorder Long-term http://www.vyvanse.com/ Suboxone Opioid dependence Short-term https://www.suboxone.com/ Xarelto Atrial fibrillation Long-term https://www.xarelto-us.com/ Bystolic High blood pressure Long-term http://www.bystolic.com/ Diovan High blood pressure Long-term http://www.diovan.com/ Benicar High blood pressure Long-term http://benicar.com/ Humalog Diabetes Long-term https://www.humalog.com/index.aspx Pataday Contact lens solution Short-term http://www.pataday.com/ Toprol-XL High blood pressure Long-term https://www.toprol-xl.com/about-toprol-xl.html Invokana Diabetes Long-term https://www.invokana.com/ Levemir Flexpen Diabetes Long-term https://www.levemir.com/ Zostavax Shingles vaccine Short-term https://www.zostavax.com/ Combigan Hypertension Long-term https://www.combigan.com/patient/Default.aspx Onglyza Diabetes Long-term https://www.onglyza.com/ Effient Acute coronary syndrome Long-term https://www.effient.com/ Norvir HIV Long-term http://norvir.com/ Advair HFA Asthma Long-term https://www.advair.com/ Epiduo Acne Long-term https://www.epiduoforte.com/ Novolog Flexpen Mix 70/30 Diabetes Long-term https://www.novolog.com/ Name of medication . Medical condition/ purpose of medication . Short-term versus long-term . Website . Lantus Solostar Diabetes Long-term https://www.lantus.com/ Vyvanse ADHD and binge eating disorder Long-term http://www.vyvanse.com/ Suboxone Opioid dependence Short-term https://www.suboxone.com/ Xarelto Atrial fibrillation Long-term https://www.xarelto-us.com/ Bystolic High blood pressure Long-term http://www.bystolic.com/ Diovan High blood pressure Long-term http://www.diovan.com/ Benicar High blood pressure Long-term http://benicar.com/ Humalog Diabetes Long-term https://www.humalog.com/index.aspx Pataday Contact lens solution Short-term http://www.pataday.com/ Toprol-XL High blood pressure Long-term https://www.toprol-xl.com/about-toprol-xl.html Invokana Diabetes Long-term https://www.invokana.com/ Levemir Flexpen Diabetes Long-term https://www.levemir.com/ Zostavax Shingles vaccine Short-term https://www.zostavax.com/ Combigan Hypertension Long-term https://www.combigan.com/patient/Default.aspx Onglyza Diabetes Long-term https://www.onglyza.com/ Effient Acute coronary syndrome Long-term https://www.effient.com/ Norvir HIV Long-term http://norvir.com/ Advair HFA Asthma Long-term https://www.advair.com/ Epiduo Acne Long-term https://www.epiduoforte.com/ Novolog Flexpen Mix 70/30 Diabetes Long-term https://www.novolog.com/ Open in new tab Table I. Sample of prescription drug websites (listed in decreasing order by number of prescriptions sold) Name of medication . Medical condition/ purpose of medication . Short-term versus long-term . Website . Lantus Solostar Diabetes Long-term https://www.lantus.com/ Vyvanse ADHD and binge eating disorder Long-term http://www.vyvanse.com/ Suboxone Opioid dependence Short-term https://www.suboxone.com/ Xarelto Atrial fibrillation Long-term https://www.xarelto-us.com/ Bystolic High blood pressure Long-term http://www.bystolic.com/ Diovan High blood pressure Long-term http://www.diovan.com/ Benicar High blood pressure Long-term http://benicar.com/ Humalog Diabetes Long-term https://www.humalog.com/index.aspx Pataday Contact lens solution Short-term http://www.pataday.com/ Toprol-XL High blood pressure Long-term https://www.toprol-xl.com/about-toprol-xl.html Invokana Diabetes Long-term https://www.invokana.com/ Levemir Flexpen Diabetes Long-term https://www.levemir.com/ Zostavax Shingles vaccine Short-term https://www.zostavax.com/ Combigan Hypertension Long-term https://www.combigan.com/patient/Default.aspx Onglyza Diabetes Long-term https://www.onglyza.com/ Effient Acute coronary syndrome Long-term https://www.effient.com/ Norvir HIV Long-term http://norvir.com/ Advair HFA Asthma Long-term https://www.advair.com/ Epiduo Acne Long-term https://www.epiduoforte.com/ Novolog Flexpen Mix 70/30 Diabetes Long-term https://www.novolog.com/ Name of medication . Medical condition/ purpose of medication . Short-term versus long-term . Website . Lantus Solostar Diabetes Long-term https://www.lantus.com/ Vyvanse ADHD and binge eating disorder Long-term http://www.vyvanse.com/ Suboxone Opioid dependence Short-term https://www.suboxone.com/ Xarelto Atrial fibrillation Long-term https://www.xarelto-us.com/ Bystolic High blood pressure Long-term http://www.bystolic.com/ Diovan High blood pressure Long-term http://www.diovan.com/ Benicar High blood pressure Long-term http://benicar.com/ Humalog Diabetes Long-term https://www.humalog.com/index.aspx Pataday Contact lens solution Short-term http://www.pataday.com/ Toprol-XL High blood pressure Long-term https://www.toprol-xl.com/about-toprol-xl.html Invokana Diabetes Long-term https://www.invokana.com/ Levemir Flexpen Diabetes Long-term https://www.levemir.com/ Zostavax Shingles vaccine Short-term https://www.zostavax.com/ Combigan Hypertension Long-term https://www.combigan.com/patient/Default.aspx Onglyza Diabetes Long-term https://www.onglyza.com/ Effient Acute coronary syndrome Long-term https://www.effient.com/ Norvir HIV Long-term http://norvir.com/ Advair HFA Asthma Long-term https://www.advair.com/ Epiduo Acne Long-term https://www.epiduoforte.com/ Novolog Flexpen Mix 70/30 Diabetes Long-term https://www.novolog.com/ Open in new tab Coding categories Based on the literature review, several coding categories were developed to analyze the visual, textual and social media elements of pharmaceutical websites as well as aspects of the medication/condition. We address each of them, in turn, below. Visual elements Visual coding categories included presence of a video, pictures of people, product image, number of visuals, emotional tone and color. Presence of a video from or on the homepage was coded as yes/no. If a video was mentioned with a link to it opening in a new page, it was counted as ‘yes’ for presence of a video. Pictures of people were coded at the ratio level (i.e. number of pictures). The total number of people shown was counted, excluding repetitive depictions of the same people. For example, if a homepage had a single picture with four people clearly visible in it, the code entered was 4. Product image was coded at the ratio level (number of times a product is visually depicted), regardless of the form (i.e. a pill or a package). Total number of visuals was coded at the ratio level and included photographs, motion graphics and informational graphics. This category excluded company logos or design features that do not communicate a message independent of text (i.e. colored circles containing text). Emotional tone of homepages was coded to assess the affective information on the website. Websites without people pictured were coded as neutral. Websites with visuals of people were coded as positive or negative, according to the facial expressions and body language of the people shown. Using the logic presented by Nagel et al. [45], facial expressions and body language that are typically associated with positive emotions were coded as positive. For example, when people were pictured smiling, laughing, hugging, etc. then the emotional tone was coded as positive. Conversely, facial expressions and body language that are typically associated with negative emotions were coded as negative. When people were pictured frowning, crying, etc. then the emotional tone was coded as negative. Colors are highly influential visual cues, previously categorized into warm and cool colors in an analysis of non-surgical cosmetic procedure websites [21]. The dominant colors of the website were coded according to the previous study mentioned, and any number of colors were counted when white was the most dominant color. The coded colors were then categorized into warm, cool or combined warm and cool colors for analysis. We used Grumbein and Goodman’s [28] definitions of cool colors as blues, purples and greens and warm colors as reds, oranges and yellows. When two colors were used that fit in different categories (cool versus warm), we coded the dominant colors as both. Textual elements Readability statistics were calculated for risks and benefits using Microsoft Word, including measures of passive sentences, Flesch reading ease, and Flesch–Kincaid grade level. This strategy has been used in previous studies [11, 46]. The placement of the risk and benefit information was coded separately for each type of information as before the scroll (or click), after the scroll (or click), or not on the homepage. If seeing the full information required clicking on a box or scrolling down the page to reveal the full textual information, it was coded as after the scroll. This applied to both risk and benefit information. Finally, the total number of products advertised on the website was also coded. We coded the total number of products advertised based on the number of product names located on the homepage. Social media Homepages were examined for links to social media platforms. The total number of social media links was coded at the ratio level, where the presence of a link(s) for each type of social media platform was coded as present (1) or absent (2) on the homepage. Facebook, Twitter and YouTube were included in the coding rubric, but additional categories for other social media links could be added, as needed. We did not locate links to any other social media platforms on the homepages in the sample. User-centric content The user-centric design of webpages was assessed by visuals/text that target consumers in a particular market segment (e.g. information appealing to low-socioeconomic-status individuals, such as financial incentives). User-centric coding categories included financial information, languages, gender and race of people in visuals, and geographic information. The placement of financial information on the homepage was coded as appearing before the scroll or after the scroll. [Jewett and DiPasquale [47] explain that placement of text in a place that does not require the user to scroll is analogous to placing an item on the front of the newspaper ‘above the fold’ (p. 10). Thus, presenting visuals before the scroll on the homepage of prescription drug websites is akin to giving them the prominence of front page news.] The number of language options on the homepage was coded at the ratio level. If the website was provided in English with no additional language options, it was coded as 1 for this variable. The gender and race of people shown on the homepage were recorded because demographics often help identify target audiences in health communication [23]. Following the guidelines of Heuer et al.’s[48] visual content analysis examining obesity in the news, sex was categorized as male/female and race was divided into four potential categories: Caucasian, African American, Latinx and Asian. The people pictured were coded as female or male based on aspects of appearance typically associated with females and males. Race was similarly coded based on the skin tone and physical appearance of people as is typically associated with a given race. Finally, the existence and placement of information relating to the geographical scope of eDTCA on the homepages was coded (i.e. stating that a website is intended for customers located in the United States). We coded any indication located on the homepage, regardless of where it appeared on the homepage, as either present or absent. The indication did not need to use the exact wording but needed to convey the same general meaning. Nature of health condition Medications were coded according to whether they were intended for long-term or short-term use. Thus, a medication for a chronic illness would be considered long-term, whereas a medication for an acute condition would be considered short-term. Table I lists the medications and shows the nature of the condition for which they are used. Inter-coder reliability Two coders were trained to analyze the data according to the conceptual coding categories. Following communication research norms [49], 20% of the data (n = 4) were coded by both coders to ensure consistency. Krippendorff’s alpha was utilized to calculate the degree of inter-coder reliability using Hayes and Krippendorff’s [50] macro for SPSS. Some categories exhibited low reliability. To stay within the acceptable range for communication research, the coders were retrained and two categories were discarded. Krippendorff’s alpha for the categories that were retained ranged between 0.97 and 1.0, with a mean of α = 0.98 (see Table II). Table II. Inter-coder reliability for coding categories Coding category . Coding guidance . Coding values . Krippendorff’s alpha value . Video Whether or not a video is located on the homepage 1 = Yes 2 = No 1.00 Social media: Facebook Whether or not there is an icon/link to a Facebook account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: Twitter Whether or not there is an icon/link to a Twitter account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: YouTube Whether or not there is an icon/link to a YouTube account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Nature of health condition(medication use) A health condition that does not require continuous use of medication was coded as short-term. A health condition that does require continuous use of medication was coded as long-term 1 = Short-term 2 = Long-term 1.00 Risk information Whether all risk information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Complete information not provided on homepage 1.00 Benefit information Whether all benefit information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Financial information Whether information related to the price of the drug or financial promotion for the drug was provided on the homepage 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Picture of person The number of visuals of people (not including fuzzy/ blurry background people without clearly visible faces) Number (ratio) 0.97 Race of person in picture Race was determined based on skin color and facial features of people pictured 1 = White 2 = Black 3 = Black and White 1.00 Females in picture Females were defined based on facial and bodily features typically associated with women Number (ratio) 1.00 Males in picture Men were defined based on facial and bodily features typically associated with men Number (ratio) 0.97 Picture of product The number of visuals that showed the drug, whether in its packaging or in its original form (capsule, drop, etc.) Number (ratio) 1.00 Visuals The number of visual elements that convey information, including people, animals and drugs. Excluded were branding visuals (e.g. company logos) and purely decorative visual elements Number (ratio) 1.00 Emotional tone Emotional tone was determined based on the visuals on the homepage. Visuals of the drug or neutral facial expressions were coded as neutral. Visuals of people smiling, laughing or embracing were coded as positive. Visuals of people frowning or looking sad were coded as negative 1 = Positive 2 = Neutral 3 = Negative 1.00 Dominant color(s) Cool colors were defined as blues, purples and greens. Warm colors were defined as reds, oranges and yellows. If both warm and cool colors were dominant, the dominant color was coded as 3 (both) 1 = Warm 2 = Cool 3 = Both 1.00 Products advertised The number of products advertised was determined based on logos/names of drugs located on the homepage Number (ratio) 1.00 Languages available The number of languages was determined based on a full examination of the homepage. If the website was in English without any other language options, the code was 1 Number (ratio) 1.00 Intended for US customers Whether there was any indication on the homepage that the website was intended only for US consumers. We coded for the meaning rather than the exact phrase, so this indication could have been included using various language to be counted as present 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Coding category . Coding guidance . Coding values . Krippendorff’s alpha value . Video Whether or not a video is located on the homepage 1 = Yes 2 = No 1.00 Social media: Facebook Whether or not there is an icon/link to a Facebook account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: Twitter Whether or not there is an icon/link to a Twitter account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: YouTube Whether or not there is an icon/link to a YouTube account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Nature of health condition(medication use) A health condition that does not require continuous use of medication was coded as short-term. A health condition that does require continuous use of medication was coded as long-term 1 = Short-term 2 = Long-term 1.00 Risk information Whether all risk information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Complete information not provided on homepage 1.00 Benefit information Whether all benefit information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Financial information Whether information related to the price of the drug or financial promotion for the drug was provided on the homepage 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Picture of person The number of visuals of people (not including fuzzy/ blurry background people without clearly visible faces) Number (ratio) 0.97 Race of person in picture Race was determined based on skin color and facial features of people pictured 1 = White 2 = Black 3 = Black and White 1.00 Females in picture Females were defined based on facial and bodily features typically associated with women Number (ratio) 1.00 Males in picture Men were defined based on facial and bodily features typically associated with men Number (ratio) 0.97 Picture of product The number of visuals that showed the drug, whether in its packaging or in its original form (capsule, drop, etc.) Number (ratio) 1.00 Visuals The number of visual elements that convey information, including people, animals and drugs. Excluded were branding visuals (e.g. company logos) and purely decorative visual elements Number (ratio) 1.00 Emotional tone Emotional tone was determined based on the visuals on the homepage. Visuals of the drug or neutral facial expressions were coded as neutral. Visuals of people smiling, laughing or embracing were coded as positive. Visuals of people frowning or looking sad were coded as negative 1 = Positive 2 = Neutral 3 = Negative 1.00 Dominant color(s) Cool colors were defined as blues, purples and greens. Warm colors were defined as reds, oranges and yellows. If both warm and cool colors were dominant, the dominant color was coded as 3 (both) 1 = Warm 2 = Cool 3 = Both 1.00 Products advertised The number of products advertised was determined based on logos/names of drugs located on the homepage Number (ratio) 1.00 Languages available The number of languages was determined based on a full examination of the homepage. If the website was in English without any other language options, the code was 1 Number (ratio) 1.00 Intended for US customers Whether there was any indication on the homepage that the website was intended only for US consumers. We coded for the meaning rather than the exact phrase, so this indication could have been included using various language to be counted as present 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Open in new tab Table II. Inter-coder reliability for coding categories Coding category . Coding guidance . Coding values . Krippendorff’s alpha value . Video Whether or not a video is located on the homepage 1 = Yes 2 = No 1.00 Social media: Facebook Whether or not there is an icon/link to a Facebook account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: Twitter Whether or not there is an icon/link to a Twitter account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: YouTube Whether or not there is an icon/link to a YouTube account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Nature of health condition(medication use) A health condition that does not require continuous use of medication was coded as short-term. A health condition that does require continuous use of medication was coded as long-term 1 = Short-term 2 = Long-term 1.00 Risk information Whether all risk information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Complete information not provided on homepage 1.00 Benefit information Whether all benefit information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Financial information Whether information related to the price of the drug or financial promotion for the drug was provided on the homepage 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Picture of person The number of visuals of people (not including fuzzy/ blurry background people without clearly visible faces) Number (ratio) 0.97 Race of person in picture Race was determined based on skin color and facial features of people pictured 1 = White 2 = Black 3 = Black and White 1.00 Females in picture Females were defined based on facial and bodily features typically associated with women Number (ratio) 1.00 Males in picture Men were defined based on facial and bodily features typically associated with men Number (ratio) 0.97 Picture of product The number of visuals that showed the drug, whether in its packaging or in its original form (capsule, drop, etc.) Number (ratio) 1.00 Visuals The number of visual elements that convey information, including people, animals and drugs. Excluded were branding visuals (e.g. company logos) and purely decorative visual elements Number (ratio) 1.00 Emotional tone Emotional tone was determined based on the visuals on the homepage. Visuals of the drug or neutral facial expressions were coded as neutral. Visuals of people smiling, laughing or embracing were coded as positive. Visuals of people frowning or looking sad were coded as negative 1 = Positive 2 = Neutral 3 = Negative 1.00 Dominant color(s) Cool colors were defined as blues, purples and greens. Warm colors were defined as reds, oranges and yellows. If both warm and cool colors were dominant, the dominant color was coded as 3 (both) 1 = Warm 2 = Cool 3 = Both 1.00 Products advertised The number of products advertised was determined based on logos/names of drugs located on the homepage Number (ratio) 1.00 Languages available The number of languages was determined based on a full examination of the homepage. If the website was in English without any other language options, the code was 1 Number (ratio) 1.00 Intended for US customers Whether there was any indication on the homepage that the website was intended only for US consumers. We coded for the meaning rather than the exact phrase, so this indication could have been included using various language to be counted as present 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Coding category . Coding guidance . Coding values . Krippendorff’s alpha value . Video Whether or not a video is located on the homepage 1 = Yes 2 = No 1.00 Social media: Facebook Whether or not there is an icon/link to a Facebook account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: Twitter Whether or not there is an icon/link to a Twitter account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Social media: YouTube Whether or not there is an icon/link to a YouTube account for the drug (company) on the homepage 1 = Present 2 = Absent Undefined Nature of health condition(medication use) A health condition that does not require continuous use of medication was coded as short-term. A health condition that does require continuous use of medication was coded as long-term 1 = Short-term 2 = Long-term 1.00 Risk information Whether all risk information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Complete information not provided on homepage 1.00 Benefit information Whether all benefit information could be viewed with or without scrolling or clicking 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Financial information Whether information related to the price of the drug or financial promotion for the drug was provided on the homepage 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Picture of person The number of visuals of people (not including fuzzy/ blurry background people without clearly visible faces) Number (ratio) 0.97 Race of person in picture Race was determined based on skin color and facial features of people pictured 1 = White 2 = Black 3 = Black and White 1.00 Females in picture Females were defined based on facial and bodily features typically associated with women Number (ratio) 1.00 Males in picture Men were defined based on facial and bodily features typically associated with men Number (ratio) 0.97 Picture of product The number of visuals that showed the drug, whether in its packaging or in its original form (capsule, drop, etc.) Number (ratio) 1.00 Visuals The number of visual elements that convey information, including people, animals and drugs. Excluded were branding visuals (e.g. company logos) and purely decorative visual elements Number (ratio) 1.00 Emotional tone Emotional tone was determined based on the visuals on the homepage. Visuals of the drug or neutral facial expressions were coded as neutral. Visuals of people smiling, laughing or embracing were coded as positive. Visuals of people frowning or looking sad were coded as negative 1 = Positive 2 = Neutral 3 = Negative 1.00 Dominant color(s) Cool colors were defined as blues, purples and greens. Warm colors were defined as reds, oranges and yellows. If both warm and cool colors were dominant, the dominant color was coded as 3 (both) 1 = Warm 2 = Cool 3 = Both 1.00 Products advertised The number of products advertised was determined based on logos/names of drugs located on the homepage Number (ratio) 1.00 Languages available The number of languages was determined based on a full examination of the homepage. If the website was in English without any other language options, the code was 1 Number (ratio) 1.00 Intended for US customers Whether there was any indication on the homepage that the website was intended only for US consumers. We coded for the meaning rather than the exact phrase, so this indication could have been included using various language to be counted as present 1 = No need to scroll 2 = Need to scroll 3 = Information not provided on homepage 1.00 Open in new tab Results Types of textual content The first research question asked about the types of textual content used in eDTCA will specific focus on the readability of the text. Textual content included benefit and risk information (before and after the scroll) and the readability of both types of information. Over half of the websites (n = 11; 55%) included benefit information before the scroll. Five websites (25%) included benefit information after the scroll and four websites (20%) did not include any benefit information on the homepage. The mean proportion of passive sentences in benefit information was 27.50% (SD = 35.88, range: 0–100). The average Flesch reading ease score for benefit information was 35.91 (SD = 21.49, range: 0–79.3) and the Flesch–Kincaid grade level score was 10.94 (SD = 4.84, range: 0–19.6). Only one website (5%) included full risk information before the scroll. The mean proportion of passive sentences in risk information was 14.55% (SD = 12.45, range: 1.1–54.5). The average Flesch reading ease score for risk information was 48.00 (SD = 11.70, range: 18.9–59.9) and the Flesch–Kincaid grade level score was 10.49 (SD = 2.19, range: 7.8–15.7). There were no significant differences in the number of passive sentences or Flesch–Kincaid grade level scores between benefit information and risk information. However, there was a significant difference in the Flesch reading ease score between benefit information and risk information (t = 3.43, df = 19, P < 0.01). Benefit information had a lower overall reading ease score (M = 35.91, SD = 21.49) than risk information (M = 48.00, SD = 11.70) (Table III). Types of visual content The second research question asked about the types of visual content used in eDTCA. Visual content included total number of visuals, videos, pictures of people, pictures of products, colors and emotional tone. The average number of visuals on a website was 2.75 (SD = 1.94), with a minimum of 0 and a maximum of 7 (see Table IV). The majority of websites did not include videos (n = 13; 65%). The average number of people pictured before the scroll was 2.0 (SD = 2.03), with a minimum of 0 and a maximum of 7. Sixteen (80%) of the 20 websites included pictures of people. Pictures of products on the websites were fairly rare. The average number of product pictures before the scroll was 0.40 (SD = 0.82), with a minimum of 0 and a maximum of 3. The majority of websites (n = 14; 70%) was dominated by cool colors. The number of websites with warm colors (n = 2; 10%) and a combination of warm and cool colors (n = 4; 20%) were much lower. The emotional tone of websites was mostly positive (n = 12; 60%), as determined by visual components (particularly pictures of people). The second most common emotional tone was neutral (n = 7; 35%) and only one website (5%) had a negative emotional tone. Table IV. Results for types of visual content Coding category . Mean (SD) . Number (%) . Total number of visuals 2.75 (1.94) — Videos — 13 (65) Pictures of people 2.0 (2.03) 16 (80) Pictures of products 0.4 (0.82) 5 (25) Colors  Cool — 14 (70)  Warm — 2 (10)  Combined — 4 (20) Emotional tone  Positive — 12 (60)  Neutral — 7 (35)  Negative — 1 (5) Coding category . Mean (SD) . Number (%) . Total number of visuals 2.75 (1.94) — Videos — 13 (65) Pictures of people 2.0 (2.03) 16 (80) Pictures of products 0.4 (0.82) 5 (25) Colors  Cool — 14 (70)  Warm — 2 (10)  Combined — 4 (20) Emotional tone  Positive — 12 (60)  Neutral — 7 (35)  Negative — 1 (5) Open in new tab Table IV. Results for types of visual content Coding category . Mean (SD) . Number (%) . Total number of visuals 2.75 (1.94) — Videos — 13 (65) Pictures of people 2.0 (2.03) 16 (80) Pictures of products 0.4 (0.82) 5 (25) Colors  Cool — 14 (70)  Warm — 2 (10)  Combined — 4 (20) Emotional tone  Positive — 12 (60)  Neutral — 7 (35)  Negative — 1 (5) Coding category . Mean (SD) . Number (%) . Total number of visuals 2.75 (1.94) — Videos — 13 (65) Pictures of people 2.0 (2.03) 16 (80) Pictures of products 0.4 (0.82) 5 (25) Colors  Cool — 14 (70)  Warm — 2 (10)  Combined — 4 (20) Emotional tone  Positive — 12 (60)  Neutral — 7 (35)  Negative — 1 (5) Open in new tab Table III. Results for types of textual content . Before the scroll n (%) . After the scroll n (%) . Not included n (%) . Flesch–Kincaid grade level Mean (SD) . Flesch reading easea Mean (SD) . Benefit information 11 (55) 5 (25) 4 (20) 10.94 (4.84) 35.91 (21.49) Risk information 1 (5) 19 (95) 0 (0) 10.49 (2.19) 48.00 (11.70) . Before the scroll n (%) . After the scroll n (%) . Not included n (%) . Flesch–Kincaid grade level Mean (SD) . Flesch reading easea Mean (SD) . Benefit information 11 (55) 5 (25) 4 (20) 10.94 (4.84) 35.91 (21.49) Risk information 1 (5) 19 (95) 0 (0) 10.49 (2.19) 48.00 (11.70) a Flesch reading ease score (t = 3.43, df = 19, P < 0.01). Open in new tab Table III. Results for types of textual content . Before the scroll n (%) . After the scroll n (%) . Not included n (%) . Flesch–Kincaid grade level Mean (SD) . Flesch reading easea Mean (SD) . Benefit information 11 (55) 5 (25) 4 (20) 10.94 (4.84) 35.91 (21.49) Risk information 1 (5) 19 (95) 0 (0) 10.49 (2.19) 48.00 (11.70) . Before the scroll n (%) . After the scroll n (%) . Not included n (%) . Flesch–Kincaid grade level Mean (SD) . Flesch reading easea Mean (SD) . Benefit information 11 (55) 5 (25) 4 (20) 10.94 (4.84) 35.91 (21.49) Risk information 1 (5) 19 (95) 0 (0) 10.49 (2.19) 48.00 (11.70) a Flesch reading ease score (t = 3.43, df = 19, P < 0.01). Open in new tab Table V. Results for user-centric content Coding category . n (%) . Race of pictured people  Caucasian 10 (62.5)  African American 1 (6.3)  Both 5 (31.3) Sex of pictured people  Female 2 (12.5)  Male 6 (37.5)  Both 8 (50.0) Financial information  Before the scroll 10 (50)  After the scroll 8 (40)  Not included 2 (10) Languages available  English 17 (85)  English and Spanish 3 (15) Geographical scope indicated (for US audiences only)  Before the scroll 6 (30)  After the scroll 9 (45)  Not included 5 (15) Link to information for non-US audiences  Yes 19 (95)  No 1 (5) Coding category . n (%) . Race of pictured people  Caucasian 10 (62.5)  African American 1 (6.3)  Both 5 (31.3) Sex of pictured people  Female 2 (12.5)  Male 6 (37.5)  Both 8 (50.0) Financial information  Before the scroll 10 (50)  After the scroll 8 (40)  Not included 2 (10) Languages available  English 17 (85)  English and Spanish 3 (15) Geographical scope indicated (for US audiences only)  Before the scroll 6 (30)  After the scroll 9 (45)  Not included 5 (15) Link to information for non-US audiences  Yes 19 (95)  No 1 (5) Open in new tab Table V. Results for user-centric content Coding category . n (%) . Race of pictured people  Caucasian 10 (62.5)  African American 1 (6.3)  Both 5 (31.3) Sex of pictured people  Female 2 (12.5)  Male 6 (37.5)  Both 8 (50.0) Financial information  Before the scroll 10 (50)  After the scroll 8 (40)  Not included 2 (10) Languages available  English 17 (85)  English and Spanish 3 (15) Geographical scope indicated (for US audiences only)  Before the scroll 6 (30)  After the scroll 9 (45)  Not included 5 (15) Link to information for non-US audiences  Yes 19 (95)  No 1 (5) Coding category . n (%) . Race of pictured people  Caucasian 10 (62.5)  African American 1 (6.3)  Both 5 (31.3) Sex of pictured people  Female 2 (12.5)  Male 6 (37.5)  Both 8 (50.0) Financial information  Before the scroll 10 (50)  After the scroll 8 (40)  Not included 2 (10) Languages available  English 17 (85)  English and Spanish 3 (15) Geographical scope indicated (for US audiences only)  Before the scroll 6 (30)  After the scroll 9 (45)  Not included 5 (15) Link to information for non-US audiences  Yes 19 (95)  No 1 (5) Open in new tab User-centric content/design The third research question asked about the types of user-centric content and design included in eDTCA. User-centric content and design included financial information, race and sex of people in pictures, languages, geographical scope (intended for United States only indication) and link for citizens outside the United States. Complete financial information was included before the scroll on half of the websites, after the scroll on eight (40%) of the websites and not shown on the homepage of two (10%) of the websites. The majority (n = 10; 62.5%) of the websites with pictures of people included only Caucasians. Some (n = 5; 31.3%) included both Caucasian and African American people in pictures. One website (6.3%) pictured only an African American. Caucasians and African Americans were the only races represented on the websites (Table V). Overall, females and males appeared in similar numbers across websites that included pictures of people; however, the number of websites with males only (n = 6) was higher than the number of websites with females only (n = 2). The average number of females pictured was 0.90 (1.02) and the average number of males pictured was 1.15 (1.09). Only three (15%) websites included languages other than English, which in each case was Spanish. Most websites (n = 15; 75%) included a disclaimer that the information on the site is intended for US audiences only; however, only six (30%) included the information before the scroll. One website (5%) included a link to another website with information for audiences outside of the United States. Types of social media content The fourth research question asked about the types of social media content used in eDTCA. Very few websites (n = 2; 10%) included social media links. The two websites that included social media links, Epiduo and Xarelto, featured links to Facebook and Twitter. Differences in content and design elements by treatment length The fifth research question asked how content and design elements differed across eDTCA according to the longevity of the prescription medication treatment (short-term versus long-term). There was a significant difference in emotional tone for short-term versus long-term prescription drug websites. There were significantly more long-term prescription drug websites with positive emotional tones than short-term prescription drug websites (Χ2 = 8.80, P < 0.05). Fisher’s exact test also indicated a significant difference (7.00, P < 0.05). None of the other coding categories was significant (see Table VI). Table VI. Results for differences in content and design elements by treatment longevity Coding category . Χ2-statistic . P-value . F-statistic . P-value . Total number of visuals — — 1.10 0.31 Videos 1.52 0.23 — — Pictures of people — — 0.00 1.00 Race of pictured people 0.46 0.80 — — Sex of pictured people 3.43 0.18 — — Pictures of product — — 0.83 0.37 Colors 1.51 0.47 — — Emotional tone 8.80 0.02* — — Benefit information 2.89 0.24 — — Risk information 0.19 0.67 — — Financial information 2.16 0.34 — — Languages available — — 0.58 0.46 Geographical scope indicated 1.53 0.47 — — Link to information for non-US audiences 1.18 0.56 — — Coding category . Χ2-statistic . P-value . F-statistic . P-value . Total number of visuals — — 1.10 0.31 Videos 1.52 0.23 — — Pictures of people — — 0.00 1.00 Race of pictured people 0.46 0.80 — — Sex of pictured people 3.43 0.18 — — Pictures of product — — 0.83 0.37 Colors 1.51 0.47 — — Emotional tone 8.80 0.02* — — Benefit information 2.89 0.24 — — Risk information 0.19 0.67 — — Financial information 2.16 0.34 — — Languages available — — 0.58 0.46 Geographical scope indicated 1.53 0.47 — — Link to information for non-US audiences 1.18 0.56 — — * Significant at P < 0.05. Open in new tab Table VI. Results for differences in content and design elements by treatment longevity Coding category . Χ2-statistic . P-value . F-statistic . P-value . Total number of visuals — — 1.10 0.31 Videos 1.52 0.23 — — Pictures of people — — 0.00 1.00 Race of pictured people 0.46 0.80 — — Sex of pictured people 3.43 0.18 — — Pictures of product — — 0.83 0.37 Colors 1.51 0.47 — — Emotional tone 8.80 0.02* — — Benefit information 2.89 0.24 — — Risk information 0.19 0.67 — — Financial information 2.16 0.34 — — Languages available — — 0.58 0.46 Geographical scope indicated 1.53 0.47 — — Link to information for non-US audiences 1.18 0.56 — — Coding category . Χ2-statistic . P-value . F-statistic . P-value . Total number of visuals — — 1.10 0.31 Videos 1.52 0.23 — — Pictures of people — — 0.00 1.00 Race of pictured people 0.46 0.80 — — Sex of pictured people 3.43 0.18 — — Pictures of product — — 0.83 0.37 Colors 1.51 0.47 — — Emotional tone 8.80 0.02* — — Benefit information 2.89 0.24 — — Risk information 0.19 0.67 — — Financial information 2.16 0.34 — — Languages available — — 0.58 0.46 Geographical scope indicated 1.53 0.47 — — Link to information for non-US audiences 1.18 0.56 — — * Significant at P < 0.05. Open in new tab Content and design predictors of product ranking/prescriptions The sixth research question asked about the strongest predictors of product ranking (quartiles according to number of prescriptions). There were two significant predictors of product ranking, which were the number of females pictured on the homepage (t = −3.85, P = 0.001) and the presence of a link to information for people outside of the United States (t = 4.04, P = 0.001). The regression model including those two variables explained 62.4% of the variance in product ranking. The relationship was such that the greater the number of females, the higher the quartile ranking of the prescription medication. (t = 3.18, P = 0.007), explaining 37.8% of the variance in prescriptions sold. The relationship was such that the greater the number of females, the higher the number of prescriptions sold. All other content and design variables were excluded using stepwise regression. Discussion The first research question asked about the types of textual content used in eDTCA, with a focus on readability of the text. Results found a statistically significant difference in reading ease scores, which were lower for benefits than risks. A higher Flesch reading ease score implies that a reader requires less education to understand the text [51]. Therefore, our findings show that risk information was easier to understand than benefit information. This finding is counterintuitive and inconsistent with previous research, which has found that risk information in DTCA is written at a higher level of difficulty than benefit information [14]. The second research question asked about the types of visual content used in eDTCA. The study found that most websites included pictures of people displaying positive emotions. Nabi and Green [29] have suggested that emotional elements of messages affect the persuasiveness of narratives relating to health. Furthermore, there is documented evidence of emotional appeals influencing health information consumers’ attitudes, knowledge-seeking and overall behavior [29, 52]. Considering that negative emotions may motivate online health information seeking to begin with (e.g. health-related fear, worry) [51], it is not surprising that eDTCA portrays positive emotions. Such positive emotions may serve to alleviate negative emotions by visually presenting prescription drugs as the solution to negative emotions. Further, our study found that prescription drug websites were mostly dominated by cool colors. This is consistent with Grumbein and Goodman’s [28] finding that websites of surgical procedures were mostly neutral or cool colors, which they attributed to a relaxing and non-threatening effect in comparison to warmer colors. It is also worth noting that the infusion of colors into ads has a long-documented persuasive effect [27]. The dominance of cool colors for this sample of top-selling prescription drugs is consistent with research findings that blue has a greater effect than other colors on purchase intentions [53]. Visual elements are particularly important to consider as it relates to individual searches for health information about a personally relevant risk, where distractions resulting from message design elements could help manage the perception of risk [23]. Rather than focusing on the message itself, defensiveness—among other states—may encourage processing of unrelated cues, such as visual appeal and speaker credibility, rather than the message itself. Thus, message design in the digital age must include an evaluation of the persuasive appeal of visual content of eDTCA. Dual coding theory suggests that visual content has an advantage over textual content because it is encoded in two different ways. Considering this, the findings of the current study point to an advantage of benefit information over risk information. The visuals were overwhelming of people displaying positive emotions, whereas risk information was generally only shown within textual content. Similarly, websites were dominated by cool colors, which are non-threatening and relaxing, as mentioned above. This provides yet another example of the dominance of positive information in the visual content of eDTCA, which is dually encoded by viewers. The third research question examined the types of user-centric content of eDTCA. Results showed that Caucasians were depicted most commonly, followed by African Americans. Lee and Begley [54] showed that African Americans are less likely to be exposed to DTCA than Whites, which may partially explain why they are shown less in eDTCA. Photos elicit more positive responses when the audience perceives the people pictured as similar to them; thus, images used in eDTCA seem targeted to, and are likely to have a more favorable response from, Caucasians [34]. Furthermore, results also showed no significant difference in the number of depictions of females and males. Further, the study also showed that financial information was usually available on the homepage of the website, particularly in the form of coupons or rebates. This strategy seems to be a useful one considering that Pho [55] indicates that patients tend to use these coupons to avoid the purchase of generic alternatives. Additionally, consumers prefer ads with coupons to those without and brands that provide coupons [12]. Even high-earning patients are more likely to ask doctors about medications when their ads are accompanied by coupons [12]. Thus, it seems that socioeconomic factors are often considered when user-centric content is developed for prescription drug websites. Patients may not be best served, though, by obtaining these brand name prescription medications over generic alternatives. Incorporating such knowledge into health literacy campaigns is essential in the digital age. Moreover, the study also found that the majority of websites was in English and the only second language option was Spanish. Most websites state that the website is only for audiences located in the United States. However, one website had a link for audiences outside the United States. Southwell and Rupert [56] explained that individuals across the world nevertheless access these websites, which raises questions about online drug promotion globally, international media effects research and ethical practices. As it relates to pharmaceutical companies’ eDTCA as a whole, they seem to be compliant with global regulations (DTCA bans in other countries) by providing a disclaimer that the website is intended for US consumers and providing the information in languages that are used widely in the United States. The fourth research question explored the types of social media links on eDTCA websites. The study found that few websites of the most prescribed pharmaceuticals included links to social media sites, despite findings that a majority of the highest grossing drugs have social media presences [25]. It may be that pharmaceutical companies are using eDTCA 2.0 (social media) to present information that does not meet the FDA’s fair balance regulations, which may be why they do not link from their website to the social media sites, as websites may be under greater scrutiny than the newer technology of social media. Mobile and social media pharmaceutical product claims incorporate risks to a lesser extent than benefits and require more effort to locate [57, 58]. The fifth research question investigated differences in content and design elements of eDTCA according to length of medication treatment (long-term versus short-term). Medications for long-term health conditions were more likely to be depicted using a positive emotional tone than those for short-term health conditions. Previous research showed that risks more often appeared on the main website for pharmaceuticals treating short-term than long-term conditions [7]. It may be that positive emotional tone is used to manage perceived risk of long-term more so than short-term conditions due to potentially increased risk of longer medication exposure. Emotional valence can act as a frame, which ‘eventually shapes the way in which one interprets’ information being processed [59]. The sixth research question asked which content and design elements in eDTCA are the strongest predictors of product ranking according to number of prescriptions sold. The findings demonstrated that the number of females depicted on the website positively correlated with more prescriptions sold. The images of women do not provide context for understanding health conditions; hence, their correlation with highly prescribed medications cannot be explained only from an information-processing perspective. Rather, the cultural relevance of these images may explain their inclusion [60]. In the United States, informal caregivers tend to be women [61, 62]. Women have long been found to influence healthcare-seeking for men [63]. They also tend to seek health care sooner and more often than men [64, 65]. These cultural tendencies are reflected in the message design of pharmaceutical medication websites. Images of females may resonate with audiences in a comforting manner, as speaker credibility is among the heuristic cues that consumers of health information may rely on when they perceive a health risk is personally relevant [23]. For people with low literacy/health literacy, photos of women could serve as peripheral cues that allow them to evaluate the message without fully understanding the benefit and risk information; this would be consistent with propositions of the ELM [20]. Implications This study offers several contributions to existing knowledge of eDTCA message design. First, our study not only investigates textual and visual design elements to explore message design strategies in eDTCA but also examines readability and use of social media in eDTCA. Previous studies of eDTCA have addressed message design strategies and fair balance of benefit and risk information, but there is a lack of recent research addressing readability of textual content. Our findings indicate that the most-prescribed pharmaceuticals provide benefit and risk information that is far above the average American’s reading level. This presents a substantial concern as limited literacy generally, and health literacy specifically, can prevent consumers from adequately processing important benefit and risk information. The US Office of Disease Prevention and Health Promotion [66] indicates that Internet users with low literacy may skip entire sections of text that seem difficult to read or read each word carefully, but still struggle to understand the content. This makes eDTCA message design extremely important, as 72% of Internet users search for health information online [10]. The current findings have implications for health educators, who may need to help consumers understand the risks and benefits that they encounter in eDTCA and drive consumers toward other reliable (and readable) information sources related to particular health conditions and available medications, as well as alternatives that may help their conditions, such as lifestyle changes and therapy. Healthcare providers may choose to provide patients with supplementary and more accessible information than pharmaceutical websites. McInnes and Haglund [67] found that domains that end with .com are more likely to require a lower reading level than a .edu domain but a higher reading level than a .gov or .nhs domain. Thus, healthcare providers can guide their patients to find information from governmental organizations that provide information for a non-specialized audience. This may allow consumers with low literacy/health literacy to better understand important benefit and risk information as it relates to various courses of treatment. In interpersonal discussions with patients, healthcare professionals and health educators may want to emphasize treatment options and lifestyle changes that can be combined with or used as an alternative to prescription medications, when warranted. Second, this study explores the websites of the highest selling pharmaceutical products. Health communication studies previously examining visuals in eDTCA often focused on a specific illness or health condition [38]. By developing generalizable results on pharmaceutical websites more broadly, our study contributes to the literature on visual message design strategies used in eDTCA. Our findings indicate that visuals typically feature people conveying a positive emotional tone. The visuals in this context seem to serve as illustrations of the benefits, or expected outcomes, of taking the medication. Considering that such images are easy to interpret and that visual information is encoded more quickly and is easier to recall than textual information, our findings and dual coding theory [31] lead us to suggest that benefit information in eDTCA may be over-represented compared with risk information. Future experimental studies could examine the differential effects of the information conveyed via visuals by comparing emotionally positive pictures of people with emotionally neutral images of people and/or the medication itself. Although limited research has addressed this as it relates to visuals and audio in the context of televised DTCA [68], it has yet to be addressed for visuals and written text in eDTCA. Findings of the current study indicate that textual benefit and risk information is balanced and actually presents risk information that is easier to read; however, whether the visual elements add to the case for benefits over risks and distract from risk information processing is something that should be addressed by experimental studies to inform FDA regulations of DTCA. It is clear from the current study that links to social media and videos on the homepages of eDTCA are infrequent. What is unclear, however, is why eDTCA is lacking in videos, considering that they have great potential for helping to educate consumers about prescription drugs and promote the brand. Videos could be used to demonstrate how the drug works as well as model how one can talk to their doctor about the prescription medication. When consistent information is presented in dual modalities as it typically is in videos (through visual and audio modalities) it is easier to recall, as suggested by dual coding theory [31]. It is worth noting that we only coded information on the homepage of the websites, so some websites may have had videos that were not located on the homepage. Limitations and future directions All scientific studies have their limitations, and this study is no exception. The sampled websites consist of only the highest-selling medications, which limits its generalizability to lower-selling medication websites. This sample also included medications for a range of health issues, some which are more prevalent than others. Studies with more homogenous samples (e.g. focused on a single disease or type of medication) may reveal which visual and textual strategies are especially prominent for particular types of medications/diseases and examine their effects on consumers. Qualitative research may also help to enhance the depth of analysis related to visual and textual components of DTCA across mediums. Future researchers should conduct similar examinations for over-the-counter medications to see how the strategies they use differ from those for prescription medications. Content and design elements that lead to high sales for prescription medications may be different from those that enhance sales of over-the-counter medications, which do not require access through a gatekeeper. Further, future researchers should conduct content analyses of mobile content [45] produced by pharmaceutical corporations to promote their products, with a particular emphasis on how pharmaceutical companies use social media and how/whether they drive traffic to their social media sites. More studies that address the eDTCA 2.0 (social media) may shed some light on the reason(s) that pharmaceutical companies are not promoting their social media accounts on their websites. Another important avenue for future research is to combine content analysis with experiments or surveys to determine how specific message design elements affect consumers’ cognitive processing, attitudes and behaviors related to eDTCA [68]. Conclusion Using content analysis, we analyzed the types of visual and textual information displayed in eDTCA (prescription medication websites) as well as the user-centric characteristics addressed through visual and textual elements. Among the key findings of our study is that eDTCA tends to use positive portrayals of people, which may serve to enhance recall of benefit information over risk information. Additionally, eDTCA targets mainly Caucasians and females through the visuals of people that they display. Textual information that conveys risks and benefits is above the average reading level of Americans, although the risk information is significantly easier to read than benefit information. In light of this study’s findings, dual coding theory can be applicable to pharmaceutical websites for prescription drugs. Such websites have competing information in terms of risks and benefits, which are differentially presented in textual and visual content. Results have important implications for health educators and healthcare practitioners as they may need to help patients access information that is readable and provides an unbiased view of prescription medications and alternative treatments. Future research should address how visual information in eDTCA may enhance consumer understanding and recall of benefits and risks as well as address whether visuals in eDTCA provide an advantage for benefit or risk information. Further, future eDTCA studies may also consider studying social media pages of the most prescribed medications in addition to their websites. The current study provides a foundation for which these future studies can be built. Acknowledgements The authors would like to acknowledge the fellowship for faculty at Gulf University for Science & Technology and the University of Missouri-St. Louis as it helped facilitate the initial conceptualization of this study in the summer of 2016. Conflict of interest statement None declared. References 1 Schmid KL , Rivers SE , Latimer AE et al. Targeting or tailoring? Maximizing resources to create effective health communications . Mark Health Serv 2008 ; 28 : 32 – 7 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 2 Slater MD , Flora JA. Health lifestyles: audience segmentation analysis for public health interventions . Health Educ Q 1991 ; 18 : 221 – 33 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Babar Z , Siraj AM , Curley L. A review of DTCA techniques: appraising their success and potential impact on medication users . Res Social Adm Pharm 2018 ; 14 : 218 – 27 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Lucas T , Manning M , Hayman LW et al. Targeting and tailoring message-framing: the moderating effect of racial identity on receptivity to colorectal cancer screening among African–Americans . J Behav Med 2018 ; 41 : 747 – 56 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Willis E. Visual elements in direct-to-consumer advertising: messages communicated to patients with arthritis . Health Mark Q 2017 ; 34 : 1 – 17 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Liang BA , Mackey TK. Prevalence and global health implications of social media in direct-to-consumer drug advertising . J Med Internet Res 2011 ; 13 : e64 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Huh J , Cude BJ. Is the information ‘fair and balanced’ in direct-to-consumer prescription drug websites? J Health Commun 2004 ; 9 : 529 – 40 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Charbonneau DH. An analysis of benefits and risk information on pharmaceutical web sites for the treatment of menopause . Health Info Libr J 2013 ; 30 : 212 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Macias W , Lewis LS. A content analysis of direct-to-consumer (DTC) prescription drug web sites . J Advert 2003 ; 32 : 43 – 56 . Google Scholar Crossref Search ADS WorldCat 10 Pew Research Center. Health Fact Sheet. 2016 . Available at: http://www.pewinternet.org/fact-sheet/. Accessed: 9 November 2018 11 Eltorai AE , Sharma P , Wang J et al. Most American Academy of Orthopaedic Surgeons’ online patient education material exceeds average patient reading level . Clin Orthop Relat Res 2015 ; 473 : 1181 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Bhutada NS , Cook CL , Perri M III . Consumers responses to coupons in direct-to-consumer advertising of prescription drugs . Health Mark Q 2009 ; 26 : 333 – 46 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Ventola CL. Direct-to-consumer pharmaceutical advertising: therapeutic or toxic? PT 2011 ; 36 : 669 – 74 . OpenURL Placeholder Text WorldCat 14 Ledford CJ. Content analysis of Internet marketing strategies: how pharmaceutical companies communicate about contraceptives with consumers online . Social Mark Q 2009 ; 15 : 55 – 71 . Google Scholar Crossref Search ADS WorldCat 15 Kang H , An S. How direct-to-consumer drug websites convey disease information: analysis of stigma-reducing components . J Health Commun 2013 ; 18 : 1477 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Mackey TK , Cuomo RE , Liang BA. The rise of digital direct-to-consumer advertising? Comparison of direct-to-consumer advertising expenditure trends from publicly available data sources and global policy implications . BMC Health Serv Res 2015 ; 15 : 236 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Arney J , Menjivar C. Disease mongering in direct‐to‐consumer advertising and the expansion of the antidepressant market . Sociol Inq 2014 ; 84 : 519 – 44 . OpenURL Placeholder Text WorldCat 18 Huh J , Suzuki-Lambrecht Y , Lueck J et al. Presentation matters: comparison of cognitive effects of DTC prescription drug advergames, websites, and print ads . J Advert 2015 ; 44 : 360 – 74 . Google Scholar Crossref Search ADS WorldCat 19 Center for Drug Evaluation and Research at Food and Drug Administration. Prescription Drug Advertising - Drug Advertising: A Glossary of Terms. 2015 . Available at: https://www.fda.gov/drugs/resourcesforyou/consumers/prescriptiondrugadvertising/ucm072025.htm. Accessed: 24 February 2020. 20 Petty RE , Cacioppo JT. Communication and Persuasion . New York, NY : Springer , 1986 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC 21 Chaiken S. Heuristic versus systematic information processing and the use of source versus message cues in persuasion . J Pers Soc Psychol 1980 ; 39 : 752 – 66 . Google Scholar Crossref Search ADS WorldCat 22 King AJ , Jensen JD , Davis LA et al. Perceived visual informativeness (PVI): construct and scale development to assess visual information in printed materials . J Health Commun 2014 ; 19 : 1099 – 115 . Google Scholar Crossref Search ADS PubMed WorldCat 23 McWhirter JE , Hoffman-Goetz L. A systematic review of visual image theory, assessment, and use in skin cancer and tanning research . J Health Commun 2014 ; 19 : 738 – 57 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Guidry J , Jin Y , Haddad L et al. How health risks are pinpointed (or not) on social media: the portrayal of waterpipe smoking on Pinterest . Health Commun 2016 ; 31 : 659 – 67 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Mackert M , Guadagno M , Lazard A et al. Improving gestational weight gain and breastfeeding promotion: visual communication to overcome health literacy barriers . J Commun Healthc 2016 ; 9 : 90 – 7 . Google Scholar Crossref Search ADS WorldCat 26 Sullivan HW , O’Donoghue AC , Aikin KJ et al. Visual presentations of efficacy data in direct-to-consumer prescription drug print and television advertisements: a randomized study . Patient Educ Couns 2016 ; 99 : 790 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Fahmy S , Bock MA , Wanta W. Visual Communication Theory and Research: A Mass Communication Perspective . New York, NY : Palgrave Macmillan , 2014 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC 28 Grumbein A , Goodman JR. Pretty as a website: examining aesthetics on nonsurgical cosmetic procedure websites . Vis Commun 2015 ; 14 : 485 – 523 . Google Scholar Crossref Search ADS WorldCat 29 Nabi RL , Green MC. The role of a narrative’s emotional flow in promoting persuasive outcomes . Media Psychol 2015 ; 18 : 137 – 62 . Google Scholar Crossref Search ADS WorldCat 30 Klein WM , Stefanek ME. Cancer risk elicitation and communication: lessons from the psychology of risk perception . CA Cancer J Clin 2007 ; 57 : 147 – 67 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Paivio A. Mental Representations: A Dual Coding Approach , vol. 9. New York, NY: Oxford University Press , 1990 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC 32 Dan V. Audiences in the dark: deception in pharmaceutical advertising through verbal–visual mismatches. In Docan-Morgan T (ed.). The Palgrave Handbook of Deceptive Communication . Basingstoke, UK: Palgrave Macmillan , 2019 , 839 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC 33 Mastin T , Andsager JL , Choi J et al. Health disparities and direct-to-consumer prescription drug advertising: a content analysis of targeted magazine genres, 1992–2002 . Health Commun 2007 ; 22 : 49 – 58 . Google Scholar Crossref Search ADS PubMed WorldCat 34 Buller MK , Bettinghaus EP , Fluharty L et al. Improving health communication with photographic images that increase identification in three minority populations . Health Educ Res 2019 ; 34 : 145 – 58 . Google Scholar Crossref Search ADS PubMed WorldCat 35 Shir-Raz Y , Avraham E. Under the regulation radar: PR strategies of pharmaceutical companies in countries where direct advertising of prescription drugs is banned—the Israeli case . Public Relat Rev 2017 ; 43 : 382 – 91 . Google Scholar Crossref Search ADS WorldCat 36 Mackey TK , Liang BA. Global reach of direct-to-consumer advertising using social media for illicit online drug sales . J Med Internet Res 2013 ; 15 : e105 . Google Scholar Crossref Search ADS PubMed WorldCat 37 Netemeyer RG , Burton S , Andrews JC et al. Graphic health warnings on cigarette packages: the role of emotions in affecting adolescent smoking consideration and secondhand smoke beliefs . J Public Policy Mark 2016 ; 35 : 124 – 43 . Google Scholar Crossref Search ADS WorldCat 38 Hallahan K. Improving public relations web sites through usability research . Public Relat Rev 2001 ; 27 : 223 – 39 . Google Scholar Crossref Search ADS WorldCat 39 Kang J , Lin CA. Effects of message framing and visual-fear appeals on smoker responses to antismoking ads . J Health Commun 2015 ; 20 : 647 – 55 . Google Scholar Crossref Search ADS PubMed WorldCat 40 Martinez LS , Lewis N. The role of direct-to-consumer advertising in shaping public opinion surrounding prescription drug use to treat depression or anxiety in youth . J Health Commun 2009 ; 14 : 246 – 61 . Google Scholar Crossref Search ADS PubMed WorldCat 41 Becker SJ , Midoun MM. Effects of direct-to-consumer advertising on patient prescription requests and physician prescribing: a systematic review of psychiatry-relevant studies . J Clin Psychiatry 2016 ; 77 : e1293 – 300 . Google Scholar Crossref Search ADS PubMed WorldCat 42 Chang H-Y , Murimi I , Daubresse M et al. Effect of direct-to-consumer advertising on statin use in the United States . Med Care 2017 ; 55 : 759 – 64 . Google Scholar Crossref Search ADS PubMed WorldCat 43 Aitken M , Kleinrock M , Pennente K et al. Medicines Use and Spending Shifts: A Review of the Use of Medicines in the US in 2014. IMS Institute for Healthcare Informatics, 2015 . Available at: https://www.iqvia.com/-/media/iqvia/pdfs/institute-reports/medicines-use-and-spending-shifts-in-the-us-in-2014. Accessed: 24 February 2020. 44 Brown T. The 10 Most-Prescribed and Top-Selling Medications. WebMD, 2015 . Available at: https://www.webmd.com/drug-medication/news/20150508/most-prescribed-top-selling-drugs. Accessed: 24 February 2020. 45 Nagel F , Maurer M , Reinemann C. Is there a visual dominance in political communication? How verbal, visual, and vocal communication shape viewers’ impressions of political candidates . J Commun 2012 ; 62 : 833 – 50 . Google Scholar Crossref Search ADS WorldCat 46 Mailloux SL , Johnson ME , Fisher DG et al. How reliable is computerized assessment of readability? Comput Nurs 1995 ; 13 : 221 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 47 Jewett A , DiPasquale D. What’s black and blue and read online: an analysis of newspaper website aesthetics and the influence of circulation size. In: AEJMC August 2013 , Washington DC. 48 Heuer CA , McClure KJ , Puhl RM. Obesity stigma in online news: a visual content analysis . J Health Commun 2011 ; 16 : 976 – 87 . Google Scholar Crossref Search ADS PubMed WorldCat 49 Riffe D , Lacy S , Fico F. Analyzing Media Messages: Using Quantitative Content Analysis in Research , 2nd edn. Mahwah : Lawrence Erlbaum Associates , 2014 . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC 50 Hayes AF , Krippendorff K. Answering the call for a standard reliability measure for coding data . Commun Methods Meas 2007 ; 1 : 77 – 89 . Google Scholar Crossref Search ADS WorldCat 51 Van Stee SK , Yang Q. Online cancer information seeking: applying and extending the comprehensive model of information seeking . Health Commun 2018 ; 33 : 1583 – 92 . Google Scholar Crossref Search ADS PubMed WorldCat 52 Murphy ST , Frank LB , Moran MB et al. Involved, transported, or emotional? Exploring the determinants of change in knowledge, attitudes, and behavior in entertainment‐education . J Commun 2011 ; 61 : 407 – 31 . Google Scholar Crossref Search ADS WorldCat 53 Hall RH , Hanna P. The impact of web page text-background colour combinations on readability, retention, aesthetics and behavioural intention . Behav Inf Technol 2004 ; 23 : 183 – 95 . Google Scholar Crossref Search ADS WorldCat 54 Lee D , Begley CE. Racial and ethnic disparities in response to direct-to-consumer advertising . Am J Health Syst Pharm 2010 ; 67 : 1185 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat 55 Pho K. Discount prescription drug coupons no bargain. USA Today 24 October 2012 . 56 Southwell BG , Rupert DJ. Future challenges and opportunities in online prescription drug promotion research: comment on “Trouble spots in online direct-to-consumer prescription drug promotion: a content analysis of FDA warning letters” . Int J Health Policy Manag 2016 ; 5 : 211 –1 3 . Google Scholar Crossref Search ADS PubMed WorldCat 57 Aikin KJ , Sullivan HW , Dolina S et al. Direct-to-consumer promotion of prescription drugs on mobile devices: content analysis . J Med Internet Res 2017 ; 19 : e225 . Google Scholar Crossref Search ADS PubMed WorldCat 58 Tyrawski J , DeAndrea DC. Pharmaceutical companies and their drugs on social media: a content analysis of drug information on popular social media sites . J Med Internet Res 2015 ; 17 : e130 . Google Scholar Crossref Search ADS PubMed WorldCat 59 Nabi RL. Exploring the framing effects of emotion: do discrete emotions differentially influence information accessibility, information seeking, and policy preference? Commun Res 2003 ; 30 : 224 – 47 . Google Scholar Crossref Search ADS WorldCat 60 Houts PS , Doak CC , Doak LG et al. The role of pictures in improving health communication: a review of research on attention, comprehension, recall, and adherence . Patient Educ Couns 2006 ; 61 : 173 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat 61 Anderson LA , Edwards VJ , Pearson WS et al. Adult caregivers in the United States: characteristics and differences in well-being, by caregiver age and caregiving status . Prev Chronic Dis 2013 ; 10 : e135 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 62 Trivedi R , Beaver K , Bouldin ED et al. Characteristics and well-being of informal caregivers: results from a nationally-representative US survey . Chronic Illn 2014 ; 10 : 167 – 79 . Google Scholar Crossref Search ADS PubMed WorldCat 63 Norcross WA , Ramirez C , Palinkas LA. The influence of women on the health care-seeking of behavior of men . J Fam Pract 1996 ; 43 : 475 – 80 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 64 Galdas PM , Cheater F , Marshall P. Men and health help-seeking behaviour: literature review . J Adv Nurs 2005 ; 49 : 616 – 23 . Google Scholar Crossref Search ADS PubMed WorldCat 65 Thompson AE , Anisimowicz Y , Miedema B et al. The influence of gender and other patient characteristics on health care-seeking behavior: a QUALICOPC study . BMC Fam Pract 2016 ; 17 . OpenURL Placeholder Text WorldCat 66 Office of Disease Prevention and Health Promotion. Health Literacy Online: A Guide to Simplifying the User Experience. 2015 . Available at: https://health.gov/healthliteracyonline/. Accessed: 24 February 2020. 67 McInnes N , Haglund BJ. Readability of online health information: implications for health literacy . Inform Health Soc Care 2011 ; 36 : 173 – 89 . Google Scholar Crossref Search ADS PubMed WorldCat 68 Russell CA , Swasy JL , Russell DW et al. Eye-tracking evidence that happy faces impair verbal message comprehension: the case of health warnings in direct-to-consumer pharmaceutical television commercials . Int J Advert 2017 ; 36 : 82 – 106 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) 2020. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Electronic direct-to-consumer advertising of pharmaceuticals: an assessment of textual and visual content of websites
JO - Health Education Research
DO - 10.1093/her/cyaa004
DA - 2020-04-01
UR - https://www.deepdyve.com/lp/oxford-university-press/electronic-direct-to-consumer-advertising-of-pharmaceuticals-an-QHC5SWltXk
SP - 134
VL - 35
IS - 2
DP - DeepDyve
ER -