Electronic Health Record–Based Messages to Primary Care Providers: Valuable Information or Just Noise?

Communication between clinicians is critical to coordination of care and prevention of adverse outcomes in the outpatient setting. Increasing the adoption of electronic health records (EHRs) and medical home-based care models will greatly increase electronic communication between different members of the health care team.^1-3 One method of clinician-to-clinician communication is note-based messaging through the EHR, where the recipient is requested to provide their “additional signature” to a message to attest that it was received. We recently found that primary care providers (PCPs) receive a large number of EHR-based additional signature request (ASR) alerts and spend considerable time processing them.⁴ Large numbers of messages^5,6 might also cause PCPs to miss certain higher-priority notifications.^7,8 Whether ASR alerts and other types of electronic messaging (called “routing” in some systems) are relevant to patient care or just a medium for distributing legal risk is unclear.⁹

To determine the value of clinician-to-clinician messaging in the EHR, we developed and tested a new method to evaluate the content of electronic messages and determined whether they were essential to clinical care.

We conducted the study in the outpatient clinics of a large tertiary care Department of Veterans Affairs (VA) facility. In the VA, clinician-to-clinician electronic communication occurs through an asynchronous alert notification inbox within the EHR, much like e-mail, where the sender and recipient need not be simultaneously engaged. We defined an ASR alert as any note-based message that required an electronic signature to complete the alert. Messaging systems with capabilities to track routing are now available in many commercially available EHRs.

We queried a centralized alert tracking file containing details of all ASR alerts. Using methods developed in prior work,⁴ we extracted 160 days of ASR alerts transmitted to any full-time PCP (physician, physician assistant, or nurse practitioner) beginning May 27, 2009. A 1% sample was randomly selected for further analysis, based on the feasibility of medical chart review.

Because each ASR alert originated from a specific note within the EHR, medical chart reviews were focused on the content of the parent note. Two PCP reviewers (H.K. and K.H.) identified each parent note and rated alerts on 3 “value” attributes: (1) urgency with which follow-up action was needed to avoid patient harm, (2) level of patient harm that might occur if the PCP missed the alert, and (3) subjective importance of the alert to PCP's care. To identify alerts that most affected clinical care, we defined “high-value” alerts as those which both reviewers rated as urgent, potentially harmful if missed, and important.

Reviewers also determined the alert sender's role, reason for the alert, and whether the information transmitted would be received through other means of communication regardless of the alert. To determine the proportion of pertinent note content, reviewers collected word counts for both the parent note and the section directly relevant to the recipient.

Of 420 927 total alerts collected during the study period, 53 606 (12.7%) were ASR alerts, of which 536 (1.0%) were reviewed. For analysis, 525 ASR alert–parent note pairs were usable.

Additional signature request alerts were most commonly (38.7%) transmitted by other members of the primary care clinic, including medical assistants, technicians, nurses, and less commonly, mid-level health care providers and other PCPs. Twenty-six percent originated from the telephone triage service, which provides after-hours telephone support, while the remaining were transmitted by specialists, pharmacists, and other support services. In almost all alerts (99.2%), the PCP would not typically receive the information outside the alert notification system. Parent notes contained a median of 142 words, of which 28 (19.7%) were considered relevant to the PCP receiving the note.

Reviewers identified 15 unique reasons for alert transmission (Table), the most frequent of which was to inform PCPs about patients' medication refill requests (40.0%). In addition, 18.9% relayed new or persistent symptoms reported by patients.

Overall, 282 alerts (53.7%) met high-value criteria. Most refill requests (89.0%) and reports of new or persistent symptoms (64.6%) were deemed high value. Conversely, alerting about patient home events, order status updates, inpatient visits, and progress note completion (residents to supervising attending) were infrequently (<15%) of high value, even though many were rated as “important.”

Approximately half of EHR-based alerts for clinician-to-clinician messaging were of high value, but inefficiencies in information transfer required PCPs to read through large amounts of extraneous text to find relevant information. Few alerts were deemed nonessential, but in conjunction with the low percentage of relevant text in parent notes, they likely lead to a perception of information overload from ASR alerts.

Strategies to improve efficiency of electronic clinician-to-clinician messaging should be pursued to avoid burdening busy frontline health care providers. Other members of a medical home team could efficiently follow up many of the alerts we encountered. Alert systems could also allow the sender to highlight relevant text within the note. Because clinician-to-clinician messaging is likely to increase as systems become more integrated, our study might be useful to others as they explore interventions to improve outpatient communication.

Correspondence: Dr Murphy, Houston VA HSR&D Center of Excellence (152), Michael E. DeBakey Veterans Affairs Medical Center (MEDVAMC), 2002 Holcombe Blvd, Houston, TX 77030 (drmurphy@bcm.edu).

Author Contributions: All authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Murphy, Reis, Kadiyala, Hirani, Sittig, and Singh. Acquisition of data: Murphy, Reis, Kadiyala, Hirani, and Singh. Analysis and interpretation of data: Murphy, Reis, Sittig, Khan, and Singh. Drafting of the manuscript: Murphy, Reis, Kadiyala, Hirani, and Khan. Critical revision of the manuscript for important intellectual content: Murphy, Kadiyala, Hirani, Sittig, and Singh. Statistical analysis: Murphy and Khan. Obtained funding: Singh. Administrative, technical, and material support: Murphy, Kadiyala, and Hirani. Study supervision: Murphy, Sittig, and Singh.

Funding/Support: This work was funded by the Veterans Affairs National Center for Patient Safety, Baylor College of Medicine Department of Family & Community Medicine Post Doctoral Fellowship program, a SHARP contract (ONC #10510592) from the Office of the National Coordinator for Health Information Technology, and in part by the Houston VA HSR&D Center of Excellence (grant HFP90-020).

Role of the Sponsors: These sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.