The NUDGE trial pragmatic trial to enhance cardiovascular medication adherence: study protocol for a randomized controlled trial

Facilitating family communication of familial hypercholesterolemia genetic risk: Assessing engagement with innovative chatbot technology from the IMPACT-FH study

Objective

To assess use of two web-based conversational agents, the Family Sharing Chatbot (FSC) and One Month Chatbot (OMC), by individuals with familial hypercholesterolemia (FH).

Methods

FSC and OMC were sent using an opt-out methodology to a cohort of individuals receiving a FH genetic result. Data from 7/1/2021 through 5/12/2022 was obtained from the electronic health record and the chatbots' HIPAA-secure web portal.

Results

Of 175 subjects, 21 (12%) opted out of the chatbots. Older individuals were more likely to opt out. Most (91/154, 59%) preferred receiving chatbots via the patient EHR portal. Seventy-five individuals (49%) clicked the FSC link, 62 (40%) interacted, and 36 (23%) shared a chatbot about their FH result with at least one relative. Ninety-two of the subjects received OMC, 22 (23%) clicked the link and 20 (21%) interacted. Individuals who shared were majority female and younger on average than the overall cohort. Reminders tended to increase engagement.

Conclusion

Results demonstrate characteristics relevant to chatbot engagement. Individuals may be more inclined to receive chatbots if integrated within the patient EHR portal. Frequent reminders can potentially improve chatbot utilization.

Innovation

FSC and OMC employ innovative digital health technology that can facilitate family communication about hereditary conditions.

Description of patient questions received by clinical pharmacists in the Nudge Study

PURPOSE

The Nudge Study is a patient level-randomized trial testing different text message medication refill reminders sent to patients assigned to 4 arms: (1) usual care, (2) generic text, (3) optimized text, and (4) optimized text plus chatbot. This report describes the frequency and types of patient questions sent to clinical pharmacists (CPs) following text reminders.

METHODS

Patients were enrolled from Denver Health and Hospital Authority (DHHA) and Veterans Affairs Eastern Colorado Health Care System (VA ECHCS) from October 1, 2019, through May 30, 2021. Included patients responded to at least 1 text or interactive voice response (IVR) message. Patients were dichotomized as those who posed at least 1 question to a CP and those who posed no questions.

RESULTS

Of the 6,325 patients enrolled in an intervention arm, 3,323 (52.5%) responded to at least 1 text or IVR message, and among those responding, 305 (9.2%) responded with a pharmacist question. Patient factors associated with submitting a CP question included age (45-74 years), enrollment from DHHA, and receipt of the optimized text or optimized text plus chatbot message versus the generic text. Questions to CP were in the following categories: medication related (48.2%), refill logistics (38.4%), cost (9.2%), and other (17.7%).

CONCLUSION

In a text messaging intervention focused on medication refills, there were few questions directed to the CP. Patients assigned to receive optimized texts were more likely to have questions. We hypothesize that this may suggest greater patient engagement regarding their condition, resulting in more questions.

Potential bias and lack of generalizability in electronic health record data: reflections on health equity from the National Institutes of Health Pragmatic Trials Collaboratory

Embedded pragmatic clinical trials (ePCTs) play a vital role in addressing current population health problems, and their use of electronic health record (EHR) systems promises efficiencies that will increase the speed and volume of relevant and generalizable research. However, as the number of ePCTs using EHR-derived data grows, so does the risk that research will become more vulnerable to biases due to differences in data capture and access to care for different subsets of the population, thereby propagating inequities in health and the healthcare system. We identify 3 challenges-incomplete and variable capture of data on social determinants of health, lack of representation of vulnerable populations that do not access or receive treatment, and data loss due to variable use of technology-that exacerbate bias when working with EHR data and offer recommendations and examples of ways to actively mitigate bias.

Secondary analysis of electronic opt-out consent in pragmatic research: A study design method to diversify clinical trials?

We conducted a multi-center pragmatic trial of a low-risk intervention focused on medication adherence using an opt-out consent approach, where patients could opt out by letter and then electronically. We focus on the cohort after opt-out by mail. Here, we describe that 8% of patients opted out electronically, resulting in a 92% participation rate. Patients who self-identify as Black or Hispanic were less likely to opt out in the study, and half the study cohort was female. This demographic data is useful for planning future trials employing this approach.

General improvements versus interruptive or non-interruptive alerts in the blood order set: study protocol for a randomized control trial to improve packed red blood cell utilization

BACKGROUND

Blood transfusions can serve as a life-saving treatment, but inappropriate blood product transfusions can result in patient harm and excess costs for health systems. Despite published evidence supporting restricted packed red blood cell (pRBC) usage, many providers transfuse outside of guidelines. Here, we report a novel prospective, randomized control trial to increase guideline-concordant pRBC transfusions comparing three variations of clinical decision support (CDS) in the electronic health record (EHR).

METHODS

All inpatient providers at University of Colorado Hospital (UCH) who order blood transfusions were randomized in a 1:1:1 fashion to the three arms of the study: (1) general order set improvements, (2) general order set improvements plus non-interruptive in-line help text alert, and (3) general order set improvements plus interruptive alert. Transfusing providers received the same randomized order set changes for 18 months. The primary outcome of this study is the guideline-concordant rate of pRBC transfusions. The primary objective of this study is to compare the group using the new interface (arm 1) versus the two groups using the new interface with interruptive or non-interruptive alerts (arms 2 and 3, combined). The secondary objectives compare guideline-concordant transfusion rates between arm 2 and arm 3 as well as comparing all of arms of the study in aggregate to historical controls. This trial concluded after 12 months on April 5, 2022.

DISCUSSION

CDS tools can increase guideline-concordant behavior. This trial will examine three different CDS tools to determine which type is most effective at increasing guideline-concordant blood transfusions.

TRIAL REGISTRATION

Registered on ClinicalTrials.gov 3/20/21, NCT04823273 . Approved by University of Colorado Institutional Review Board (19-0918), protocol version 1 4/19/2019, approved 4/30/2019.

User-Centered Design to Reduce Inappropriate Blood Transfusion Orders

BACKGROUND

To improve blood transfusion practices, we applied user-centered design (UCD) to evaluate potential changes to blood transfusion orders.

OBJECTIVES

The aim of the study is to build effective transfusion orders with different designs to improve guideline adherence.

METHODS

We developed three different versions of transfusion orders that varied how information was presented to clinicians ordering blood transfusions. We engaged 14 clinicians (residents, advanced practice providers [APPs], and attending physicians) from different specialties. We used the think aloud technique and rapid qualitative analysis to generate themes to incorporate into our modified orders.

RESULTS

Most end-users who participated in the semi-structured interviews preferred the interruptive alert design plus behavioral nudges (n = 8/14, 57%). The predominant rationale was that the in-line alert was not visually effective in capturing the end-user's attention, while the interruptive alert forced a brief stop in the workflow to consider the guidelines. All users supported the general improvements, though for different reasons, and as a result, the general improvements remained in the designs for the forthcoming trial.

CONCLUSION

The user experience uncovered through the think aloud approach produced a clear and rich understanding of potentially confounding factors in the initial design of different intervention versions. Input from end-users guided the creation of all three designs so each was addressing human factors with parity, which ensured that the results of our study reflected differences in interruptive properties of the alerts and not differences in design.

Comparison of Pharmacy Refill Data With Chemical Adherence Testing in Assessing Medication Nonadherence in a Safety Net Hospital Setting

Background Pharmacy fill data are a practical tool for assessing medication nonadherence. However, previous studies have not compared the accuracy of pharmacy fill data to measurement of plasma drug levels, or chemical adherence testing (CAT). Methods and Results We performed a cross-sectional study in patients with uncontrolled hypertension in outpatient clinics in a safety net health system. Plasma samples were obtained for measurement of common cardiovascular drugs, including calcium channel blockers, thiazide diuretics, beta blockers, angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, and statins, using liquid chromatography mass spectrometry. Proportion of days covered (PDC), a method for tracking pharmacy fill data, was calculated via linkages with Surescripts, and its diagnostic test characteristics were compared with CAT. Among 77 patients with uncontrolled hypertension, 13 (17%) were nonadherent to at least 1 antihypertensive drug and 23 (37%) were nonadherent to statins by CAT. PDC was significantly lower in the nonadherent versus the adherent group by CAT only among patients prescribed an angiotensin-converting enzyme inhibitor/angiotensin receptor blocker or statin (all P<0.05) but not in patients prescribed other drug classes. The sensitivity and specificity of PDC in detecting nonadherence to angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and statin drugs by CAT were 75% to 82% and 56% to 79%, respectively. The positive predictive value of PDC in detecting nonadherence was only 11% to 27% for antihypertensive drugs and 45% for statins. Conclusions PDC is useful in detecting nonadherence to angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and statins but has limited usefulness in detecting nonadherence to calcium channel blockers, beta blockers, or thiazide diuretics and has a low positive predictive value for all drug classes.

OUP accepted manuscript

The increasing volume and richness of healthcare data collected during routine clinical practice have not yet translated into significant numbers of actionable insights that have systematically improved patient outcomes. An evidence-practice gap continues to exist in healthcare. We contest that this gap can be reduced by assessing the use of nudge theory as part of clinical decision support systems (CDSS). Deploying nudges to modify clinician behaviour and improve adherence to guideline-directed therapy represents an underused tool in bridging the evidence-practice gap. In conjunction with electronic health records (EHRs) and newer devices including artificial intelligence algorithms that are increasingly integrated within learning health systems, nudges such as CDSS alerts should be iteratively tested for all stakeholders involved in health decision-making: clinicians, researchers, and patients alike. Not only could they improve the implementation of known evidence, but the true value of nudging could lie in areas where traditional randomized controlled trials are lacking, and where clinical equipoise and variation dominate. The opportunity to test CDSS nudge alerts and their ability to standardize behaviour in the face of uncertainty may generate novel insights and improve patient outcomes in areas of clinical practice currently without a robust evidence base.

Leave me out: Patients' characteristics and reasons for opting out of a pragmatic clinical trial involving medication adherence

Opt-out procedures are sometimes used instead of standard consent practices to enable patients to exercise their autonomous preferences regarding research participation while reducing patient and researcher burden. However, little is known about the characteristics of patients who opt-out of research and their reasons for doing so. We gathered such information in a large pragmatic clinical trial (PCT) evaluating the effect of theory informed text messages on medication adherence.Eligible patients, identified through electronic health records, were sent information about the study and provided with an opportunity to opt-out. Those opting out were asked to complete a voluntary survey regarding their reasons for doing so. Demographic data were compared among patients opting-out vs those included in the study using chi-squared tests and a log binomial regression model.Of 9046 patients receiving study packets, 906 (10.0%) patients returned opt-out forms. Of those, 451 (49.8%) returned the opt-out survey. Patients who opted out were more likely to be older, white, and nonHispanic than those who were included in the PCT. Survey respondents expressed high levels of trust in their health care providers, research, and system. Nearly half (46.6%) reported concerns about time as a reason to opt-out.In this PCT, 10% of patients receiving packets opted out, with significant differences in age, race, gender, and ethnicity compared to those included. Future trials should further investigate representativeness and reasons patients choose to opt-out of participating in research.