Patient-initiated electronic health record amendment requests

TL;DR

This study qualitatively analyzed all patient-initiated EHR amendment requests received at one academic health system over a 7-year period, finding that most requests sought correction of incorrect information and about half were approved. The authors conclude that patient access may improve record accuracy, but low request rates suggest patients need better information about how to view and amend records.

Key findings

The dataset included 205 amendment request forms from 181 distinct patients, containing 818 distinct requests; 77.8% (n=636) were made to rectify incorrect information, and about half of all requests were ultimately approved (n=406, 49.6%; abstract reports 49.7%). Requests to remove valid information were uncommon (6.6% of requests) and less likely to be approved (27.8%); among 14 requests to remove valid information about drug-seeking behavior, 13 were denied (92.9%). Outpatient clinic notes were the most common document type contested (n=308, 37.7%), followed by inpatient discharge summaries (n=84, 10.3%) and emergency department notes (n=83, 10.1%). Only about 0.2% of patients who requested a copy of their chart submitted an amendment request; among 36 requests based on information viewed through the patient portal or after-visit summary, all but one were approved.

Methods

Qualitative content analysis of all patient-initiated amendment requests received by the University of Michigan Health System from March 2006 through December 2012; the final dataset comprised 205 request forms from 181 patients, yielding 818 distinct requests after separating sub-requests requiring separate review. Coding followed a grounded theory approach with iterative theme identification across three analytic dimensions: clinical/documentation area, patient motivation, and outcome of the request; ambiguous cases were reviewed with Health Information Management staff, and clinician comments on denied requests were also analyzed. Statistical analyses used R V.2.15.3, with test of proportions for racial proportions and unpaired t tests for mean ages.

Limitations

The paper states that the work was conducted at a single institution, data were originally collected for operational tracking rather than prospectively for research, pre-2008 requests were not systematically archived (likely underestimating earlier request counts), unofficial clinician–patient amendment requests were not captured, and the study did not determine clinical significance/urgency or whether clinicians perceived requested changes as useful. The paper also notes it could not determine the proportion of patients who found errors but did not seek an amendment. Evident from the text: the retrospective, single-site design and reliance on archived operational records limit generalizability and may miss amendment activity outside the formal DHIM process.