Patient E-Message Character Limit Impact on Primary Care Clinician Burnout

Introduction

Clinician burnout is prevalent in healthcare,¹ especially among primary care (PC) clinicians.² One driver is asynchronous work, an important component of which is electronic messages (e-messages) between patients and clinicians in the electronic medical record (EMR).³ Studies have described mixed results regarding the impact of patient e-messages on clinician burnout and patient care,^4–6 though to our knowledge, no studies have described the impact on PC clinicians of limiting e-message character counts.

As our institution implemented a character count limit on e-messages sent by patients to clinicians, we hypothesized this change would decrease burnout rates and/or e-message burden among PC clinicians, presumably by decreasing patient e-message length. This project aimed to evaluate the impact of this character count limit on PC clinician burnout and e-message burden.

Methods

A patient-initiated e-message length limitation (1000 characters) was implemented in June 2023 across a single academic health system. While this character limit was selected and put in place by the institution, the study team determined to assess its impact on PC clinician burnout and e-message-related burden using a pre-/post-survey design. Voluntary electronic (REDCap) questionnaires were distributed to all of the family medicine and general pediatrics attending physicians and advanced practice clinicians (i.e. nurse practitioners and physician assistants) in their respective academic departments at the institution approximately 1 month prior to and 8 months after the character limit implementation. Responses pre-/post- were unmatched, and non-missing responses to each item were analyzed. The questionnaires included standard sociodemographic questions (e.g. sex, race/ethnicity, age range) and Likert scale questions assessing PC clinician burnout and e-message-related burden, based on the Mini Z burnout inventory and other similar surveys.^1,4,7,8 Authors who were part of the study population reviewed the questionnaires for validity.

During data analysis, sociodemographic characteristics pre-/post-implementation were compared using chi-squared and Fisher’s exact tests. Mean Likert scale responses pre-/post- were compared using unadjusted linear regression with robust standard errors to account for lack of independence between pre-/post-respondents. In sensitivity analyses, adjusted analyses were conducted, and Likert scales were treated as ordinal variables, where similar results were found. Two-sided tests were computed with statistical significance p<0.05 using R version 4.4.1.

This project was determined to be not human subjects research by the institutional review board.

Results

Approximately 150 clinicians met inclusion criteria and were invited to participate in the study. A total of 81 eligible responses were received for the pre-intervention survey and 72 for the post-intervention survey (approximate response rates 54% and 48%, respectively). There were no significant sociodemographic differences between eligible respondents in the pre-/post-intervention surveys (Table 1). There were also no significant differences in Mini Z scores or subjective measures of e-message burden before and after the intervention (Table 2). Overall missing data was 3.4% with no more than 4.6% missing on any individual item.

Discussion

Recently, patient use of EMR-based messaging has expanded in volume and complexity. Such e-messages are associated with increased resource utilization and burnout, and addressing them is often unreimbursed and completed after hours.^3,5,6 Hence, this must remain a focus given high PC clinician burnout and the role of asynchronous, electronic communication in workload and fatigue.^3,4,7 While character count limits may be helpful, implementing such a limit on patient-to-clinician e-messages was insufficient to significantly improve burnout or e-message-related burden for PC clinicians at our institution, consistent with a cross-sectional study that showed no association between word count and clinician burnout when clinicians from multiple specialties were surveyed.⁹ The impact of the intervention may have been reduced by patients working around the limit, such as by sending multiple messages or attaching documents, suggesting that varying the size of the character count limit may not change the results. Notably, differences in messaging workflows (e.g. nurses triaging messages prior to forwarding them to clinicians) between clinics and institutions could influence the extent to which a character limit would impact clinician burnout and e-message burden at other institutions. The nature of this study taking place at a single academic center with unlinked pre-/post-surveys and suboptimal response rate may limit some of its generalizability, yet the results highlight that PC clinician burnout is a complex issue, and a single institutional change may be insufficient to make an impact. This complexity is supported in the literature, such as by other work on billing for patient e-messages that demonstrated only modest benefit in reducing e-messages¹⁰ or findings that patient e-messaging was felt to improve efficiency while also increasing the workload of clinicians.¹¹ In such a context, our work highlights the need for other studies to evaluate additional approaches to address patient messages and their impact on PC clinicians.