Patient-Physician Messaging by Race, Ethnicity, Insurance Type, and Preferred Language.

Importance: Asynchronous patient-portal messaging has emerged as a central component of patient-physician communication, yet disparities in response patterns remain underexplored. Objective: To examine variations in asynchronous messaging responses in primary care by race and ethnicity, insurance type, and primary language. Design, Setting, and Participants: This cross-sectional study of messaging data included 795 170 adult patients receiving primary care at Mass General Brigham in 2021, 341 836 of whom sent an asynchronous patient portal message. Data were analyzed from April 17, 2023, to July 29, 2025. Exposures: Race and ethnicity, insurance status, and preferred language. Main Outcomes and Measures: Receipt of a response within 1 or 3 business days and time to first response from care team. Multivariable regression models adjusted for patient demographics, clinic, and time fixed effects. Results: The analytic sample consisted of 341 836 patients, who sent a total of 3 525 905 messages comprising 1 270 662 message threads to 1113 physicians. Among these patients (mean [SD] age, 52.2 [17.3] years), 18 442 were Asian (5.4%), 14 089 were Black or African American (4.1%), 9979 were Hispanic or Latino (2.9%), 285 919 were White (83.6%), 232 661 had commercial insurance (68.1%), and 332 004 primarily spoke English (97.1%). Characteristics of patient-threads that received a response from any care team member within 1 business day were as follows: 65.7% of threads from Black participants (32 165 of 48 983), 63.9% of threads from Hispanic or Latino (21 732 of 34 014), 68.5% of threads from White participants (743 161 of 1 085 517), 70.0% of threads from patients with commercial insurance (529 803 of 756 923), 60.9% of threads from patients with dual-eligibility (40 846 of 67 045), 68.4% of threads from patients preferring English (847 489 of 1 239 768), and 58.0% of threads from patients preferring Spanish (6898 of 11 903). Similar disparities were found in the response rate from primary care physicians. After adjustment, compared with White patients, the rate of response from any care team member within 1 day was lower for Black patients (1.1 [95% CI, 0.2-2.0] percentage points; P = .01) and Hispanic patients (1.1 [95% CI, 0.3-1.9] percentage points; P = .01). Compared with patients with commerical insurance, the response rate within 1 day was lower for dual-eligible patients (4.9 [95% CI, 4.2-5.5] percentage points; P < .001), and compared with patients who preferred English, the response rate within 1 day was lower for patients who preferred Spanish (4.1 [95% CI, 2.5-5.7] percentage points; P < .001). Adjustment for clinic and time fixed effects accounted for most of the observed disparities in response rates and timing. Conclusions and Relevance: In this cross-sectional study of primary care patients in a single health system, there were significant disparities in the responsiveness of primary care teams to asynchronous patient-portal messages by race and ethnicity, insurance type, and language. These were partially attributable to slower response times at practices that treat underserved patients.