3193 — Patient-Centered Rather than Disease-Centered CDS for Health Professionals: Steps Toward Coordinating Multiple Guidelines for the Same Patient

Goldstein MK, GRECC, VA Palo Alto Health Care System; Tu SW, Stanford University and VA Palo Alto Health Care System; Oshiro C, VA Palo Alto Health Care System; Martins SB, VA Palo Alto Health Care System; Wang D, VA Palo Alto Health Care System; Furman A, VISN 21 and VA Palo Alto Health Care System; Ashcraft M, VA Palo Alto Health Care System; Mendoza J, VA Palo Alto Health Care System;

Objectives:
Co-existing multiple chronic conditions (MCCs) are highly prevalent among Veteran patients; yet, most clinical decision support (CDS) is single-disease-focused. For Veteran-centered rather than disease-centered care, there is a need for CDS methods to consider multiple conditions that are present within the same patient. This study aims to provide information about choices faced in integrating multiple guidelines into an existing system.

Methods:
We extracted recommendations from relevant guidelines and encoded computable knowledge bases (KBs) for 5 common chronic conditions: hypertension (HTN), diabetes (DM), hyperlipidemia (HL), heart failure (HF), and chronic kidney disease (CKD). We designed an architecture for integrating this CDS into the existing VISN21 Clinical Dashboard for Patient-Aligned Care Teams (PACTs). Previous work had resolved many issues of comorbidity within a KB, eg, comorbid conditions that are contraindications to using a medication that would otherwise be preferred for the guideline condition. To explore further adjustments needed for multiple guidelines, we categorized new decisions made by the project team over an 8 month period.

Results:
We found the following example issues in encoding KBs: (a) When comorbid conditions appeared in two guidelines, we had to decide whether to encode the information in one or the other or both KBs, with attendant challenges of maintaining completeness and consistency; (b) Several guidelines included recommendations regarding the same comorbidity, eg CKD, but they defined the comorbidity differently; (c) Different guidelines sometimes used slightly different contraindications for the same drug, requiring guideline-specific lists of drug contraindications. We developed decision rules to resolve these issues.

Implications:
The process of operationalizing guidelines for CDS can bring to light inherent inconsistencies across guidelines that are intended to be applied, by clinicians, to the same patient.

Impacts:
This study will help future CDS developers to plan for guideline inconsistencies. It also illustrates the importance of encouraging guideline authors to consider commonly-occurring comorbid conditions and to harmonize definitions of conditions, targets, age ranges, and drug contraindications.