This project is a VA HSR&D-funded Quality Enhancement Research Initiative (QUERI) project to translate into practice evidence about clinical management of primary hypertension. The project aims to contribute to quality improvement of care for patients with primary hypertension. The project implemented a clinical decision support (CDS) system for primary care clinicians and evaluated the implementation by studying the following: impact on the clinicians' prescribing and their patients' blood pressures; the clinician satisfaction with the CDS; and organizational factors in the implementation.

Background / Rationale:

Despite the existence of evidence-based guidelines, there is a gap between evidence-based recommended best medical practice and actual practice. In previous work, we used hypertension as a model to study the translation of research regarding clinical blood pressure targets and drug choice into primary care practice through computer-based implementation of clinical practice guidelines. In collaboration with experts in guideline-automation at Stanford Center for Biomedical Informatics Research (BMIR) (formerly known as Stanford Medical Informatics, SMI), we developed an innovative clinical decision support system: ATHENA-CDS-HTN (hypertension) also known as ATHENA-HTN. ATHENA-HTN is a knowledge-based CDS that uses knowledge bases (KBs) of clinical knowledge of hypertension encoded into computable formats. Clinical data from an electronic health records system is processed with the knowledge in the KB by an execution engine (also known as a guideline interpreter) that generates patient-specific outputs with conclusions about the current state of the patient with respect to the guidelines as shown in the patient's electronic health record data and also generates detailed recommendations for next steps in clinical management of the patient's hypertension. The system is designed to take account of multiple data elements in the patients' clinical data including co-morbid diagnoses that are relevant to hypertension and its treatment, other medications currently prescribed, history of adverse drug reactions/events, and selected laboratory values relevant to hypertension and its management. In previous work, we had demonstrated that deploying this CDS system in VA primary care clinics was feasible and that clinicians found the system usable and useful, as shown by their actual extensive use of the system and their response to a questionnaire survey. The current project was funded by the Department of Veterans Affairs Health Services Research and Development (HSR&D), under a special funding initiative for Quality Enhancement Initiative (QUERI) projects designed to improve care for hypertension by collaboration with a Veterans Integrated Service Network (VISN) to implement programs to encourage use of established best-practices for managing hypertension according to evidence-based guidelines available at the time of the study. These studies were known as "VISN Collaboratives". The studies did not involve any new drugs or new uses of established drugs, but rather encouraging use of established drugs recommended in standard evidence-based guidelines for care. The focus of the study was implementation of existing known best practices in managing hypertension, using the CDS as a vehicle to bring information to the point-of-care, with detailed individualized recommendations about patients presented to the primary care provider at the time the provider sees patients in scheduled outpatient primary care clinic visits.

Objective(s):

Our objectives in this VISN Implementation Collaborative included: (1) implement evidence-based guidelines for hypertension in a CDS system by upgrading the ATHENA-HTN knowledge-base (KB) to the most recent guidelines; (2) deploy the CDS system in 5 medical centers within VISN 1 to generate individualized recommendations to primary care clinicians caring for patients in outpatient clinics; (3) evaluate the implementation including the organizational aspects.

Methods:

In Phase 1 we plan to update the KB and conduct offline testing; to revise the M (formerly known as MUMPS) program that extracts patient data daily from VistA to extract additional data elements; to streamline the system architecture to make it easier to implement in multiple sites; to work with the site-PIs to obtain IRB approval at 5 implementation sites in VISN 1 plus the coordinating site at VA Palo Alto; to improve the user interface design; to identify and resolve issues in implementing new information technology in multiple different VA medical centers (VAMC) with different Computerized Patient Record System (CPRS) implementations and different methods of distributing computer access to clinicians; to train the site PIs in use of the system; and perform baseline data analysis to inform the randomization. In Phase 2 we plan to recruit primary care providers from the participating 5 medical centers in VISN 1; randomize clinics to ATHENA-HTN intervention or usual care; deploy the system for intervention providers; train intervention providers in use of the system; and conduct a 12-month clinical trial of the system.

Status:

The project has been completed. We completed Institutional Review Board (IRB) approvals at all five medical centers. We updated the knowledge base of computable knowledge regarding hypertension and conducted offline testing. Our system was deployed to all five medical centers. While we were preparing for deployment of the CDS, one site changed from a rich-client environment to a thin-client environment for the medical center computers; we developed new code to run the system in a thin client environment. The previous programming computed recommendations of the CDS based on clinical data available the night before the clinic visit (in order to pre-compute advisories for faster display during clinic); in this project we developed software code to obtain automatic blood pressure updates for the day of visit if these were entered by clinical staff at the time of the visit. VA Palo Alto staff worked with staff at each VAMC to validate clinical data extraction at each site. We developed code for a clinician to select, optionally, for blood pressure write-back to VA VistA so that if clinicians entered to our CDS a blood pressure measurement that they had just taken so they could view updated recommendations based on the repeat blood pressure they would not also have to enter that blood pressure separately to electronic health record. We recruited primary care providers (clinicians) at each of 5 medical centers to participate in the project. One of the 5 medical centers completed the installation of the ATHENA-HTN system, but did not continue with the intervention period, so the intervention included 4 rather than 5 medical centers. Our site-principal investigators and project coordinator trained intervention providers on use of the system by offering short phone calls for training or in-person introduction, and by making the system available for a short period of time for familiarity. Records of patients seen by the clinicians in this training period were excluded from analyses. We completed a 6-8 month intervention period for the clinical trial at the 4 participating medical centers. Our random allocation was at clinic level, where a clinic is a substation of a medical center (station). Before initiating data analyses, we prepared a detailed data analysis plan.

Inclusion Criteria:

• Must be primary care clinicians (for example, physician or nurse practitioner) at one of the participating VA medical centers. This study is NOT recruiting patients.
• The primary care clinician must have a panel of patients for whom he or she provides direct care.

Exclusion Criteria:

• Anyone who does not meet inclusion criteria.

Baseline characteristics of primary care providers who started intervention period

Up to one year.

The study did not provide any intervention to patients. Patients were receiving their usual care from their primary care providers (clinicians.) The providers in the study were the study participants. The providers were theoretically at risk that the study might identify or disclose adverse information about their clinical management of patients. Another theoretical risk was that a provider might find the clinical decision support intrusive and problematic for clinical care.

The study has originally planned to include 5 large medical centers. One of the medical centers dropped out before the intervention started. This lowered the number of clinic sites and of participants (clinicians).