Understanding Source-Related Influences on Physicians Translating Comparative Effectiveness Research into Patient Care: Results from a Study of Cardiologists

Comparative effectiveness research (CER), which refers to an evaluation of the clinical effectiveness of two or more medical interventions that are used to treat the same condition, has the potential to inform decision-making in both policy circles and physicians’ exam rooms. The ability of stakeholders to translate that research into practice has important implications for health outcomes, but the impact of information sources on physicians in translating CER remains understudied. This project examines the source-related influences on and motivations of cardiologists with respect to willingness to make changes in their practice based on emerging CER results. The results from this survey of cardiologists (N = 42) indicate that the sources of information (including perceived credibility of those sources) matter greatly to cardiologists when deciding whether to make a change in practice. These findings suggest data-based implications for researchers and practitioners that are engaged in closing the CER translation gap.

. The general conclusion is that translation in health refers to the long and complicated process of converting information from basic science to improvements in clinical and, ultimately, public health outcomes (Drolet & Lorenzi, 2011;Marincola, 2003;Pober et al., 2001;Rubio et al., 2010;Sussman et al., 2006). In a widely cited review, Balas and Boren claim that it takes 17 years to translate only 14% of original health/medical research into practice, and that a majority of original research is never translated into practice (Balas & Boren, 2000). Whether this long of a lag involved in converting research to practice exists is not in question, and countless researchers, advocates, and even government agencies have taken on the task of bridging this gap (Balas & Boren, 2000;Berwick, 2003;Denis et al., 2002;Green, 2008;Green et al., 2009).
Comparative effectiveness research (CER) in health care has had the attention of U.S.-based policymakers and politicians over the past three decades (Chalkidou et al., 2009;Denis et al., 2002;Tunis et al., 2010). The first stand-alone piece of legislation to address CER was introduced in 2002, and the American Recovery and Reinvestment Act (ARRA) of 2009 appropriated $1.1 billion to funding CER. This set the stage for CER support in the Patient Protection and Affordable Care Act (ACA) of 2010, which established the nongovernmental Patient-Centered Outcomes Research Institute (PCORI) to focus on the delivery of evidence-based health care (Allen, 2018).
CER generally refers to an evaluation of clinical effectiveness of two or more medical interventions (sometimes including a "wait and see" approach) that are used to treat the same condition, the purpose of which is to inform decision-makingstarting at high policy levels and ultimately funneling down to conversations about treatment between doctors and patients (Chalkidou et al., 2009;Dreyer et al., 2010;Weissman et al., 2015). The value of CER, then, lies in the ability of researchers and practitioners to translate it into clinical care (Shah et al., 2010). That said, there is an incorrect assumption among some that, as soon as CER is published that holds the potential to improve efficiency, quality, and cost-effectiveness of care, those findings will be immediately integrated by clinicians (Avorn & Fischer, 2010;Macintyre, 2012). Some recommendations for closing the translation gap between CER and practice have been decision aids (Shah et al., 2010), physician report cards (Avorn & Fischer, 2010), and clinical guidelines and continuing medical education (CME) (Avorn & Fischer, 2010;Shah et al., 2010). The Institute of Medicine has also advocated for the inclusion and meaningful participation of stakeholder involvement in many aspects of CER, including setting priorities and disseminating results (Devine et al., 2013;Sox & Greenfield, 2009). This is echoed by scholars who recognize that stakeholders in health care are likely to play different roles (fueled by different motivations) in the process of translating evidence-based research (e.g., translational research has its origins in the pharmaceutical industry) (Garland et al., 2021;Lean et al., 2008;Schumock & Pickard, 2018). The relationships between these stakeholders are ever-changing, representing feedback loops between basic and applied research, public opinion and the media, political platforms, and public health initiatives, to name a few (Macintyre, 2012;Sussman et al., 2006). Examining the CER translation process through a stakeholder-based framework has some potential to shorten the extensive gap by promoting more www.scholink.org/ojs/index.php/sssr Studies in Social Science Research Vol. 2, No. 4, 2021 68 Published by SCHOLINK INC. efficient uptake of findings and determining the feasibility of CER results in a clinical setting (Arterburn et al., 2020).
Of the potential stakeholders involved in translating CER into practice, physicians bear the responsibility of directly interacting with patients, and their professional impact depends on their ability to provide effective and efficient care. Naturally, there is an abundance of literature examining the factors that contribute to physician practice patterns, though researchers note it is difficult to isolate any one factor as a primary cause of behavior change (Cunningham et al., 2019;Keating, 2017). One category of factors includes personal characteristics, such as personality traits (Green et al., 2002) and personal biases and habits (Ubel & Asch, 2015). A second category of factors can be classified as economic and institutional dynamics, including reimbursement mechanisms (e.g., fee-for-service) (Campbell et al., 2007;Hillman et al., 1998;Khullar et al., 2015) and feelings of burnout (Avorn & Fischer, 2010;Dobrow et al., 2004;Sussman et al., 2006;Whippen & Canellos, 1991).
Finally, information processes, including concerns about accessing up-to-date information (Schumock & Pickard, 2018), beliefs about what constitutes evidence (Gupta et al., 2017;Ubel & Asch, 2015), and the source or combination of sources from which the physician receives the evidence (Gupta et al., 2017;Pollack et al., 2017) are likely to impact physician behavior. In one review of the literature on physician behavior change, researchers found that medical decisions are disproportionately influenced by the physicians' mentors (Cunningham et al., 2019). In general, the role that specific information sources and their characteristics (including credibility) play in physician behavior and engagement in the translation process remains understudied (Keating, 2017) and is the focus of this paper. In order to explore the role of information sources on physicians' translation of CER, the following two research questions were developed: RQ1: Do sources of information, given a hypothetical scenario, differentially predict patterns of change in physician practice?
RQ2: What characteristic(s) of information sources might motivate a behavioral response to that source?
Given the exploratory nature of this topic, the present study concentrates on one case prompt and one physician specialty, both of which are described below.

Method
This project involves an examination of influences on and motivations of cardiologists with respect to ability and willingness to make changes in their practice based on emerging CER results. Cardiologists were selected for this case for a few reasons. First, cardiac care and intervention continue to be a priority; the leading cause of death in the United States every year since 1921 has been heart disease (Centers for Disease Control and Prevention National Center for Health Statistics, various years).

Case Prompt
The specific case referring to the DECREASE family of trials was presented at the conclusion of this survey. In collaboration with a practicing physician, the following case prompt was developed:

Clinical guidelines recommend the prescription of beta-blockers in several classes of patients, including those who will be undergoing non-cardiac surgery. You learn that the series of trials that provide justification to prescribe perioperative beta-blockers in these classes of patients is insecure; much of the data had been lost, and that which remained was found to contain serious flaws, including, in one case, complete fabrication of a dataset. The European Society of Cardiology has stopped recommending the use of beta-blockers in patients undergoing non-cardiac surgery, but the American
Heart Association guidelines based on these trials have not been retracted.

Measures
After being presented with the case prompt, participants were asked, "Which of the following would best describe your reaction had you received the information about the insecurity of the clinical trials from:" and provided with the following nine sources: professional medical journal, non-profit patient advocacy organization, medical-related website (e.g., WebMD), colleague, CME class, news broadcast, industry representative (e.g., pharmaceutical, medical device), professional association (e.g.,

International Association of Cardiologists, U.S. Preventive Services Task Force), and online resources
for clinical guidelines (e.g., National Guideline Clearinghouse). Respondents were asked to select one of five reactions on a Likert scale, anchored by 1 (Not at all likely to change practice) and 5 (Very likely to change practice).
A factor analysis suggests that source items load on two factors: professional sources (professional medical journal, colleague, CME class, industry representative, professional association, and online resource for clinical guidelines) (α = .836), and non-professional sources (non-profit patient advocacy organization, medical-related website, and news broadcast) (α = .793). However, the two factors were highly correlated (r = .581, p < .001). Additionally, the nine items combined had higher reliability (α = .866), and reliability analysis suggested that removing any of the items would not improve this score.
Given that the sample for the factor analysis was small and the reliability of the overall measure high compared to the separate measures, physicians' average "likeliness to change" score was computed across all nine items. Source-specific "likeliness to change" scores were analyzed separately when appropriate.
A number of items from Harvard's Medical Professionalism Survey were also adapted to capture information processes, including engagement in the translation process (such as by serving as a reviewer for scholarly journals, working with community-based organizations to promote health research, and participating in the development of formal clinical practice guidelines, to name a few) (Campbell et al., 2007). A variety of source credibility items were developed for this study.
Respondents were asked to respond to a 5-point Likert scale anchored by 1 (Not at all credible) and 5 (Very credible) to reflect their perceived credibility of the same sources that followed the case prompt.

Analysis
Due to the sample size and the exploratory nature of this study, analyses were conducted using primarily descriptive and inferential statistics. In general and unless otherwise noted, all correlations used, as a dependent variable, a computation of the mean of all nine "likeliness to change" items, as all nine of them loaded on one factor (α = .866). Pearson r correlation coefficients or Kendall tau-b coefficients, depending on the nature of the variables and whether or not they are normally distributed, were used to correlate information processes with likeliness to change items.

Result
The results in testing the impact of information processes on physicians' translation of CER suggest a significant difference in changing practice based on source of information [F(1,8) = 22.35, p <.001].
The estimated marginal means of the likeliness of changing practice given a particular source are reported in Table 1. The estimated marginal mean for likeliness to change based on information from a medical journal (the source most likely to prompt change) is significantly higher (p < .05) than that for colleagues, medical-related websites, industry representatives, non-profit organizations, and news broadcasters as a source. The estimated marginal mean for the source least likely to prompt changea news broadcastis significantly different from every source except for industry representatives and non-profits. This information suggests that source of information does matter in a physician's decision-making process. Interestingly, independent samples t-tests did not suggest that physician engagement in the translation process itself (to name a few activities: whether a respondent serves as a reviewer for a professional journal, participates in the development of clinical guidelines, or volunteers with community-based health promotion organizations) impacts their overall likeliness to change practice, nor their likeliness to change practice for each individual source. This leads us to the second research question: what is it about a source that may predict a physician's response to that source?
To address this question, I examined whether there is a relationship between a respondent's likeliness to www.scholink.org/ojs/index.php/sssr Studies in Social Science Research Vol. 2, No. 4, 2021 72 Published by SCHOLINK INC.
change score for a source and perceived credibility of that source (Table 2). In the case of every source except medical journals and non-profit organizations (the latter of which approaches significance), the relationship between likeliness to change for a source and perceived credibility of that source is statistically significant (p <.05). It should be noted that an online resource for clinical guidelines (e.g., National Guideline Clearinghouse) was not used in this part of the analysis, as credibility is not in question because the guidelines represent fact.

Discussion
The first finding suggested by this research is simple: information sources matter. Cardiologists were significantly more likely to make changes in practice when provided with CER information in a medical journal or via a professional association; industry representatives, non-profit organizations, and news broadcasts were not compelling enough on their own to move a change in practice. In fact, likeliness to change practice with news broadcasts as a source fell closer to "unlikely to change" than to a neutral response. This supports previous research that suggests that translation models for CER should be inclusive of different stakeholders, taking into account the reciprocal influence those stakeholders have.
The second finding indicated by these data is that perceived source credibility strongly impacts cardiologist decision-making. With the exception of only two sourcesjournals and non-profit organizations (which reflect both extremes of the "likeliness to change" spectrum)source credibility and likeliness to change practice given a source were significantly correlated. There is a lesson to be taken from non-physician stakeholders in CER translation: when working with cardiologists (and, presumably, physicians representing other specialties), it is extremely important not only to understand physicians' information needs, but also how to promote the individual/organizational credibility of non-physician stakeholders.
The primary limitation of this study relates to the small sample size. A number of analyses that could have gauged the moderating and mediating influences of different variables on source-related predictors of physician behavior were simply not possible with this small sample and inadequate power for such analyses. Further, the lack of diversity both with respect to gender and to racial/ethnic identity compromises the generalizability of these findings. Future examinations into stakeholders in the CER translation process should aim to expand the sample in order to generalize findings. Finally, self-reported data, particularly as it involves a theoretical case (and, therefore, actions cannot be measured), may not necessarily capture behavioral tendencies.
Prior research on the role of physicians in translating CER indicates that there are likely a variety of factors that influence decision-making. These can include broad systemic and institutional factors, or could be as personal as a physician's history with the topic. The aim of this article is to report data for only one potential factor motivating physician behavior change. That said, though the decision-making framework is likely multivariate and complicated, there is strong evidence that information source and source credibility could impact physician behavior. This is noteworthy because it is changeable; while interventions cannot impact reimbursement systems or alter a physician's career trajectory, they can focus on perceived credibility of, and messaging delivery through, different sources. These findings suggest data-based implications for researchers and practitioners that are engaged in closing the CER translation gap. As we navigate what scientists are calling a "pandemic era," (The Lancet Planetary Health, 2021) with patients presenting with known and unknown disease and comorbidity, the value of CER to compare vaccinations, and behavioral and pharmacological interventions, will be more important than ever. Understanding nuances involved in bridging the communication gap from "bench to bedside" has the potential to enhance patient safety, improve cost effectiveness, and save lives.