Integrating Contextual Factors Into Clinical Decision Support

The term patient context refers to the myriad contextual factors in patients' lives that complicate the application of research evidence to patient care. For instance, the inability of a patient to afford a medication for a particular condition is a contextual factor. Contextual factors can be addressed when correctly identified. Substituting a low cost generic for a high cost brand name medication may enable a patient to afford a medication. Addressing contextual factors in a care plan is termed contextualizing care. Conversely, the failure to address a contextual factor when it is feasible to so is a contextual error, because it results in an inappropriate plan of care. In sum, contextual errors are medical errors caused by inattention to patient context. They are common and linked to both diminished health care outcomes and an increase in health care costs related to overuse and misuse of medical services. These findings were determined using a validated method for coding audio recorded data called Content Coding for Contextualization of Care ("4C") collected during the encounters by both real patients, and by unannounced standardized patients (USPs) employing checklists.

Preventing contextual errors requires heightening clinician responsiveness to clues that there are contextual factors during the clinical encounter, in real time. These clues, termed contextual red flags are evident in two sources: the medical record and from patients directly. An effective intervention would prompt clinicians to determine whether there are underlying contextual factors that could be addressed in the care plan, averting contextual error. This desirable process is termed contextual probing.

While clinical decision support (CDS) has been used to provide physicians with timely biomedical information at the point of care to prevent errors and promote appropriate care, this technology also affords an opportunity to draw physician attention to both contextual red flags and contextual factors in order to avert contextual errors. This study assesses the potential of "contextualized CDS" to improve contextualization of care through a randomized controlled intervention trial, with assessment measures of both patient health care outcomes and averted costs associated with overuse and misuse of medical services. The three hypotheses are that CDS:

1. Reduces contextual error: CDS tools that inform clinicians of contextual factors and prompt them to explore contextual red flags should result in a reduction in contextual error.
2. Improve health care outcomes: Contextualized CDS predicts improved health care outcomes defined as a partial or full resolution of the contextual red flag (e.g. elevated HgB A1c) after the index visit.
3. Reduces avoidable health care costs: Contextualized CDS is associated with a reduction in misuse and overuse of inappropriate or unnecessary medical services.

To test the hypotheses, patients who consent to participate will be randomized to usual care or care enhanced with contextualized CDS. Participants will audio record their visits, and the data will be coded using 4C. They will be followed several months after the index visit for assessment of outcomes by blinded assessors using an established tracking method. In addition, USPs presenting with cases containing complicating contextual factors that if overlooked result in overuse and misuse of medical services, will be employed to assess the third hypothesis, and to supplement the data obtained by observing the effects of contextual alerts on the care of real patients for the first hypothesis.

• Contextual clinical decision support (Other): Incorporation of contextual data into EHR clinical decision support alerts
• Contextual survey (Behavioral): Patients complete a survey asking about red flags that could signal contextual factors relevant to their care

Primary outcomes

• Resolution of Contextual Red Flags [6-9 months following index visit]: Proportion of red flags noted at index visit that have resolved

Secondary outcomes

• Probing of Contextual Red Flags [At index visit]: Proportion of red flags which the examining physician probes
• Planning for Contextual Factors [At index visit]: Proportion of contextual factors identified during visit that are incorporated into care plan

Sex: All, ages: 18 Years to any

Inclusion Criteria:

* English-speaking adult patients presenting to outpatient primary care clinics for scheduled appointments who can be contacted in advance of their appointment and the clinicians (physicians or nurse practitioners) seeing those patients at those visits.

Exclusion Criteria:

* • Patients with emergent or unscheduled visits or who do not speak English.

Participant flow

Outcome results

• Resolution of Contextual Red Flags (Primary), Count of units, Red flags: Contextual Survey + Contextual CDS 116; Contextual Survey Only 240 (p .91)
• Probing of Contextual Red Flags (Secondary), Count of units, Red flags: Contextual Survey + Contextual CDS 215; Contextual Survey Only 271 (p 0.02)
• Planning for Contextual Factors (Secondary), Count of units, Contextual factors: Contextual Survey + Contextual CDS 221; Contextual Survey Only 255 (p .006)

Adverse events

Time frame: At index visit for each patient during the study period.

• Results publication: Weiner SJ, Schwartz A, Weaver F, Galanter W, Olender S, Kochendorfer K, Binns-Calvey A, Saini R, Iqbal S, Diaz M, Michelfelder A, Varkey A. Effect of Electronic Health Record Clinical Decision Support on Contextualization of Care: A Randomized Clinical Trial. JAMA Netw Open. 2022 Oct 3;5(10):e2238231. doi: 10.1001/jamanetworkopen.2022.38231. PMID 36279133

• Study Protocol and Statistical Analysis Plan (PDF, 2017-06-08)

• Saul J Weiner, MD, Principal investigator, University of Illinois at Chicago