We used a cross-sectional survey to examine participants’ perceptions of PROME. A detailed description of the survey procedure is given in a doctoral dissertation [12].We scheduled visits to conduct the survey with the coordinators of 7 seniors’ centers in a metropolitan area of a southeastern state in the United States. On each visit, 2 research assistants recruited study participants into a reserved private room and explained the purpose of the study, along with the rights of the participants. Those who completed the survey received a US $20 grocery gift card as an appreciation for their participation. Data collection was started in August 2013 and was continued until we reached our goal of 300 completed surveys in December 2013. Before beginning the study, the University of Tennessee Health Science Center Institutional Review Board determined the study to be exempt from their oversight.

The survey instrument used for this study contained a chart from a hypothetical PROME Web portal that compared PROME for 5 common antihypertensive medications (Figure 1). The chart used arbitrary 5-star ratings and included the number of people who supposedly gave reviews for each medication. We asked for the following sociodemographic information from the participants: sex, age, years of education, race, family member(s) they live with, and income. We asked participants 17 questions in total: 6 demographic questions and 11 questions about the PROME Web portal.

Referring to the PROME chart, we surveyed participants’ perceptions of its potential applicability to health care using 7 questions: (1) 1 question on overall usefulness, (2) 4 questions on patient-physician communication, and (3) 2 final questions: 1 on the perceived credibility of the information provided by the PROME Web portal, and 1 on the willingness of the survey participants to provide their own medication experiences to a PROME Web portal. The 4 questions on patient-physician communication concerned the likelihood for patients to mention PROME to their physician, the likelihood for PROME to facilitate the communication, the perceived likelihood for physicians to be upset by the PROME mention, and the likelihood for patients to ask their physicians to prescribe the PROME-recommended medication. All the questions were answered on a 5-point Likert scale with the following choices: definitely, very probably, probably, probably not, and definitely not. Previous surveys on the effect of DTCA on patient-physician communication indicated that there would be more positive than negative evaluations [5]. We used the unbalanced scale to provide more discrimination between positive evaluations [13]

The survey also had 2 questions on the preferred reporting items, that is, what information the participants wanted to see in the PROME Web portal. The first question asked participants to indicate which reporting items (star ratings, number of reviews, and individual comments) they believed valuable. The second question asked participants to indicate any medication attributes (effectiveness, side effect, food interaction, convenience, and cost) they believed the PROME Web portal should report.

Lastly, the survey had 2 additional questions on PROME Web portal sponsorship, that is, what entity study participants believed should sponsor the PROME Web portal. The first question asked participants to rate the importance of PROME Web portal sponsorship on a 5-point Likert scale. The second question asked participants to indicate any types of sponsors (academic institutions, nonprofit foundations, chain pharmacies, health information companies, and drug plans) they would trust.

For the participants’ perceptions of PROME’s potential applicability to health care, we reported the percentages selected for each of the 5 response choices as a bar graph. Further, we dichotomized the 5 response choices into having a positive (definitely, very probably, and probably) or a negative perception (probably not or definitely not), and then estimated the probability of having a positive perception, along with 95% CIs. We constructed pie charts to describe the frequency distribution of medication attributes that study participants picked as important elements of PROME coverage and to generate the frequency distribution of PROME Web portal sponsors that study participants picked as trusted. We used chi-square tests to determine whether participants’ perceptions varied with their sociodemographic characteristics at a significance level of 5%. Analyses were conducted using SAS software, version 9.3 (SAS Institute Inc).

We contacted older adults (≥50 years) who were members of seniors’ centers in a metropolitan area in a southeastern US state to participate in our survey. Table 1 lists the characteristics of the 300 older adults who completed our survey. They had a mean age of 71.95 years (SD 8.65), were mostly non-Hispanic white (164/300, 54.7%) and female (231/299, 77.3%), with at least some high school education (287/297, 96.6%) and annual incomes of at least US $10,000 (239/262, 92.3%).

As Figure 2 shows, an overwhelming majority of study participants (n=243, 81.0%) were positive about the overall usefulness of PROME; however, the percentage positive decreased to 62.3% (187/300) when excluding probably as the response. As for patient-physician communication, 245 of the 300 participants (81.7%) said that they would mention PROME to their physician (166/300, 55.3% for the responses definitely and very probably), and 248 participants (82.9%) said that PROME would facilitate patient-physician communication (138/299, 46.2% for the responses definitely and very probably). However, 133 participants (44.3%) said that their physician would get upset if they mentioned PROME (42/299, 14.1% for the responses definitely and very probably). Further, 209 (69.9%) of study participants were positive that they would ask their physician to prescribe a PROME-recommended medication. When we further broke down the positive responses to the question on whether physicians would be upset by such a request into probably, very probably, and definitely, we found that more participants chose probably (n=110, 36.8%) than very probably and definitely combined (n=99, 33.1%). In other words, study participants had some reservations about asking their physician to prescribe a PROME-recommended medication.

As for information credibility, most of our study participants (n=257, 85.7%) were positive that PROME information is trustworthy. Further, two-thirds (n=216, 72.0%) were willing to participate in PROME Web portals by providing their own medication experiences. However, for information credibility, more participants responded probably (n=136, 45.3%) than very probably and definitely combined (n=121, 40.3%) (Figure 2).

When asked about how PROME should be displayed in a Web portal, of the 308 answers given, participants most frequently picked the number of reviews (n=105, 34.1%), followed by star ratings (n=97, 31.5%) and individual comments (n=96, 31.2%) (Figure 3). However, the difference was negligible. Study participants were also asked to indicate medication attributes (such as effectiveness, side effects, ease of use, costs, and interaction with food) that PROME should cover. Study participants gave 809 answers and picked side effects most often (n=224, 27.7%) and ease of use least often (n=110, 13%). Cost (n=168, 20.8%), effectiveness (n=153, 18.9%), and interaction with food (n=151, 18.7%) were picked almost equally (Figure 3). When asked to indicate the importance of the PROME Web portal’s sponsorship, 263 of 298 participants (88.2%) said that sponsorship is important (Figure 4). Academic institutions such as the University of Tennessee were viewed as the most trusted sponsors (120/400, 30.0%), followed by nonprofit foundations such as the American Heart Association (97/400, 24.3%) (Figure 4).

Across all sociodemographic characteristics, study participants were positive that PROME provide useful and credible information to facilitate patient-physician communication (Table 2). Further, their willingness to participate in PROME Web portals remained high across all sociodemographics except for income. Study participants in the highest income bracket (over US $70,000 per year) were least willing to participate in PROME Web portals (P=.02) (Figure 5). However, the perceived likelihood that physicians would be upset was significantly different across several demographic variables. With increased education, study participants were more likely to believe that physicians would get upset by a mention of PROME (3/10, 30% for people with middle school education or less and 61/92, 66% for people with some college) (Figure 6). This trend was also present with income levels: the higher the income, the more likely the participant was to believe physicians would get upset by a mention of PROME (Figure 5). Likewise, non-Hispanic whites were more likely than non-Hispanic blacks to believe that physicians would get upset by a mention of PROME.

Our findings should be interpreted carefully because of three potential study limitations. First, we have measured participants’ perceptions based on convenience sampling in one metropolitan area. Participants’ perceptions of the study sample may well have been different from those of the US population as a whole. However, the characteristics of our study participants were similar to the population who were over 50 years of age and lived in a metropolitan area of a southeastern state.

Second, study participants may not have been willing to disclose some information, especially when the information was sensitive because of privacy concerns. For instance, many of the respondents refused to disclose their estimated annual income even though we had guaranteed the anonymity and confidentiality of their responses. While this limitation is common to all survey studies, the study sample could have been more reluctant to release the sensitive information.

Third, this study used an unbalanced 5-point Likert scale (3 positive and 2 negative responses) to elicit study participants’ perceptions of PROME with regard to patient-physician communication. Although this scale provides more discrimination between positive responses [13,30], some neutral responses could have been forced to positive ones. When this happens, dichotomizing the responses into positive and negative responses could inflate the occurrence of positive evaluations. However, the likelihood is minimal considering survey participants are known to choose a response based on a label rather than the position on the 5-point scale [31,32]

This study found that older participants across most demographics considered PROME to provide useful and credible information to facilitate patient-physician communication, and thus were willing to participate in PROME Web portals by sharing their own medication experiences. These participants also believed that an academic institution should sponsor PROME Web portals. Overall, this study found that there is a need for developing a trustworthy Web portal to systematically compile PROME for older patients to communicate well with their physicians.