Effect of Electronic Health Record Clinical Decision Support on Contextualization of Care

Key Points Question Can customized clinical decision support (CDS) tools that provide clinicians with information about patient life challenges improve patient outcomes by helping clinicians contextualize their care? Findings In this randomized clinical trial that included 452 audio-recorded patient encounters, customized CDS tools did not improve patients’ outcomes but significantly improved contextualization of their care. Meaning This study found that CDS improved contextualization of care, suggesting that the technology has the potential to improve patient outcomes.

Outcomes in Ambulatory Care through Health Information Technology." We will assess the 110 potential of "contextualized CDS" to improve contextualization of care through a randomized 111 controlled intervention trial, with assessment measures of both patient health care outcomes 112 and averted costs associated with overuse and misuse of medical services. In addition to 113 pursuing the aforementioned aim, the study design will adopt best practices for CDS design. We 114 propose to implement highly personalized, concise, actionable contextual CDS strategies. The 115 proposed study will pursue these aims by testing three hypotheses about contextualized CDS, 116 and adhering to one design principle. The three hypotheses are that CDS:

124
The design principle, referred to as "Five Right" 17 is to provide the right information to the right 125 people through the right channels in the right format at the right point in care delivery.

126
To test the hypotheses, patients who consent to participate will be randomized to usual 127 care or care enhanced with contextualized CDS. Participants will audio record their visits, and ____ Integrating Contextual Factors into Clinical Decision Support to Reduce Contextual Error and Improve Outcomes in Ambulatory Care Version 2 Page 5 of 18 6/8/2017 the data will be coded using 4C. They will be followed for 4-6 months following the index visit for 129 assessment of outcomes using an established tracking method. 6

188
The approach would embrace the "Five Rights" framework already widely adopted in CDS 189 design. 17 CDS interventions must provide the right information, to the right people, through the 190 right channels, in the right intervention formats, at the right points in workflow.

191
In the following section, we outline a plan for incorporating and rigorously assessing  collection will require listening in on the visit and that we will be inviting patients to audio 250 record their visits. They'll also learn that this is a randomized study so that some of the 251 time they'll see contextualized CDS information and other times they won't. They'll learn 252 that they are not a unit of study, and we will be collecting no data about their individual 253 performance. We'll also inform them that a decision not to participate will not impact their 254 employment in any regard as we are a research team not connected to management.

255
Those indicating they would like to participate will be contacted by an RA to complete 256 the informed consent process 257 258 • Patients: The proposed protocol is that patients of participating physicians will be 259 contacted about 2 weeks prior to a scheduled appointment to the adult primary care 260 clinic at either of the two sites. Initial contact will be via mail with an opt out for a follow 261 up phone call. If they don't opt out, the research assistant will call them. They will be 262 informed that they are invited to participate in a study to determine whether providing 263 their health care team with additional information in the electronic medical record about 264 challenges or life circumstances they are facing that impact their care could improve the 265 quality of their care, including their health outcomes. They will be informed that if they 266 participate they will be asked and, assisted if needed, with completing a brief 267 questionnaire for their medical record about challenges they are having that might 268 impact their care. They'll also learn that when they arrive for their appointment they will 269 receive a small digital audio record to carry into the visit.

531
We believe this study is minimal risk. However, to ensure the safety of research participants 532 and to comply with NIH policies, a DSMB will be formed in early months of the project and 533 given responsibility to review and approve study methods and analysis plan for the 534 research. The DSMB will be organized by Dr. Weiner and will consist of senior, experienced 535 clinicians and health services researchers. If deemed necessary by the IRB, the Principal 536 Investigators will not be on the Data Safety and Monitoring Committee, thereby ensuring 537 some level of independent review. When necessary, we will bring in experts from outside 538 the project to serve on the committee. The DSMB will review interim data mid-way through 539 the study using a predetermined stopping rule to determine whether the intervention group 540 is being significantly benefitted (or harmed) over the control group and whether early 541 stopping is necessary. In the rare event that an adverse event attributable to the CDS 542 intervention is found, we will contact the patient's provider and document in the chart the 543 potential error that was found.

554
• Patients: The proposed protocol is that patients will be contacted about 2 weeks prior 555 to a scheduled appointment to the adult primary care clinic at either of the two sites.
556 Initial contact will be via mail with an opt out for a follow up phone call. If they don't 557 opt out, the research assistant will call them. They will be informed that they are 558 invited to participate in a study to determine whether providing their health care team 559 with additional information in the electronic medical record about challenges or life 560 circumstances they are facing that impact their care could improve the quality of their 561 care, including their health outcomes. They will be informed that if they participate 562 they will be asked and, assisted if needed, with completing a brief questionnaire for 563 their medical record about challenges they are having that might impact their care.

564
They'll also learn that when they arrive for their appointment they will receive a small 565 digital audio record to carry into the visit. They'll be told that it is preferable to 566 conceal the audio in their pocket or bag, but that they can take it out if they like.

580
• Clinicians: Clinicians will be informed of the study at their standing staff meetings.

581
They'll be told that the purpose of the project is to assess whether enhanced clinical 582 decision support, that provides both passive and actively delivered information 583 provided by patients and extracted from their medication record about life 584 challenges, or "contextual factors" that may be impacting their health care, can 585 improve clinical decision making and health care outcomes and costs. They'll be 586 informed that if they participate data collection will require listening in on the visit and 587 that we will be inviting patients to audio record their visits. They'll also learn that this 588 is a randomized study so that some of the time they'll see contextualized CDS 589 information and other times they won't. They'll learn that they are not a unit of study, 590 and we will be collecting no data about their individual performance. We'll also inform 591 them that a decision not to participate will not impact their employment in any regard 592 as we are a research team not connected to management. Those indicating they 593 would like to participate will be contacted by an RA to complete the informed consent 594 process.

624
• Clinicians: Encounters rather than clinicians are the unit of interest for this study.

625
There will not be sufficient data collection from any individual clinician to draw 626 inferences about his or her performance. In fact, clinicians are not randomized in this 627 study. The same clinician will see patients in both the intervention and control 628 groups, with and without contextualized CDS. He or she will also see USPs with and 629 without clinical decision support. Hence, we plan only to collect aggregate data on 630 the participating clinicians, including age range, years in practice, gender, and 631 whether they are trained in internal medicine, family medicine or as advance 632 practices nurses. As described directly above, the audio recordings by patients of 633 their visit with their doctor or APN will be encrypted and stored on a secure server 634 space.