A Structured Workshop to Improve the Quality of Resident Discharge Summaries Open Access

Poor communication at hospital discharge can increase the risk of adverse events, medication errors, and hospital readmission.1–4 As hospital stays become shorter and increasingly complicated, and as primary care physicians become less involved in the inpatient management of their patients,5 the potential for adverse events related to poor communication of discharge plans at care transition may increase.1 

The hospital discharge summary is a tool commonly used to document the details of the inpatient stay for the posthospital care team. The Joint Commission6 and the Society of Hospital Medicine7 stress the importance of the discharge summary to enhance collaboration during patient handoffs, and the Accreditation Council for Graduate Medical Education (ACGME) requires residency programs to assess resident performance on record keeping.8 However, in a study,3 only 16% of internal medicine residency programs reported having a discharge planning curriculum in place. Residency programs rely on the use of templates to guide trainees in creating discharge summaries,9,10 but trainees lack formal instruction11 and feedback3 related to that skill, and deficiencies in discharge summaries have been widely reported.12 

We designed an educational program to teach residents how to effectively create a hospital discharge summary. We hypothesized that implementation of the educational program would result in improvements in the quality of hospital discharge summaries.

The intervention took place at a community hospital internal medicine residency program in Waterbury, Connecticut. There were 10 residents in each of the second- and third-year classes, and a pool of 16 faculty members rotated as general medicine ward attending physicians. There were 4 inpatient ward teams during each 4-week block, each led by a second- or third-year resident who was responsible for dictating discharge summaries for all patients admitted to his or her team. Hospital policy on summaries reflects the Joint Commission requirements.5 All residents spent 9 to 11 months of each academic year onsite.

Before the introduction of our curricular innovation, unstructured interviews with program leadership revealed that resident notes in both the inpatient and outpatient settings required considerable editing before the attending physician's co-signature. Residents were expected to dictate discharge summaries using templates provided by the medical records department. The only formal training for the residents was a 1-hour chart documentation skills session during orientation. Feedback occurred on an ad hoc basis during the course of routine clinical care or, more rarely, as a chart stimulated recall exercise.

In addition, we anonymously surveyed upper-level residents, asking them to rate their experiences and attitudes regarding chart documentation. Seventy-nine percent reported having received feedback about a discharge summary (survey provided as online supplemental material). In the free-form comments section, residents noted that discharge summaries were a useful tool from which to gather data on a subsequent hospitalization.

In August 2007, we launched a monthly, hour-long, structured workshop series entitled “Reviewing Effective and Accurate Documentation (READ),” during which, residents critiqued each other's chart notes.13 Five types of notes (ie, hospital admission, inpatient progress, hospital discharge summary, medical consultation, and outpatient visit) were reviewed throughout the year, but only one type of note was reviewed during each session. Discharge summaries were reviewed during 3 sessions each academic year. The week before each workshop, 2 notes were selected from separate patient charts. Note selection was random, except for ensuring that no single resident had multiple notes reviewed through the course of the year. Notes were de-identified for both patient and author information.

All onsite residents attended the workshop through all 3 years of training. Residents worked in small groups to read and analyze the first note, using instruments developed for this purpose to guide them. After 10 minutes of small-group discussion, the entire group of attendees engaged in a 15-minute open conversation moderated by a faculty member. We repeated this process for the second note.

All curricular materials including a detailed instructor's guide are available online at www.stmh.org/read and were peer reviewed by MedEd PORTAL.13 

We requested dictation logs for each of the 4 residents rotating on the inpatient medicine wards during May and June of 3 consecutive years (2007 [pre-READ], 2008, and 2009). Up to 3 summaries dictated by each resident during the specified rotation were randomly selected for review. Summaries for patients who died in the hospital or were transferred to another inpatient facility were excluded. We were unable to locate the dictation log for 1 resident. In all, 63 discharge summaries (21 from each year) were selected for review. Each summary was de-identified to author, patient, and date by an administrator not involved in the study, and then reviewed by 2 of 4 faculty members using a 14-item evaluation tool that was designed based on the Society of Hospital Medicine's hospital discharge checklist,7 the Joint Commission's requirements,6 and the Discharge Summary Evaluation Tool from Myers and colleagues.11 Items were organized into 3 prespecified categories based on question content and format: presence of essential elements (yes/no), quality of essential elements (Likert scale), and overall quality of summary (Likert scale). Faculty reviewers received training on use of the evaluation tool and received all summaries simultaneously in August 2009. Information regarding transmission of summaries to outpatient providers and characteristics of patients and their discharge summaries were collected from the medical records department.

The Saint Mary's Hospital Institutional Review Board granted all phases of the evaluation process exemption from review and informed consent.

Data were compiled and analyzed using SPSS, version 17.0 (SPSS Inc., Chicago, IL). Interrater reliability was determined by Krippendorff α. Comparisons between the groups were performed by 1-way analysis of variance and Fisher exact probability test, with significance defined at P < .05.

There were no differences in demographic characteristics of the authors of discharge summaries or of the patients during the 3 years of the study (table 1). Interrater reliability was α  =  0.65 for the 4 yes/no items related to the presence of essential elements, 0.44 for the 7 Likert scale items indicating the quality of essential elements, and 0.65 for the 3 Likert-scale items measuring overall quality of the summary.

Mean scores for 10 of the 14 individual items (71%) improved in a stepwise manner during the 3 years of the study, as did the combined scores for 2 of the 3 (67%) of the prespecified item subtypes (table 2). Most discharge summaries contained the suggested components in 2007, before program implementation, and the mean score for the item subtype related to presence of essential elements did not show statistical improvement (P  =  .10). However, residents were significantly more likely to include all of the essential elements (P  =  .02) and use headings to separate sections of the note (P  =  .048) after program implementation (table 2).

Documentation of communication with the outpatient physician occurred in few of the summaries and did not improve after the educational intervention. However, query of the medical records department revealed that the percentage of summaries actually carbon copied to the responsible outpatient physician increased from 47% to 71% to 86% during the 3 years of the study (P  =  .03).

We found that the quality of hospital discharge summaries improved following the implementation of a structured program to teach chart documentation skills. The workshop resulted in improved discharge summary quality because we emphasized the importance of communication at the transition of care for hospitalized patients, which has been reported to have an effect. Forster and colleagues1 demonstrated that 59% of preventable or ameliorable peridischarge adverse events were directly related to lapses in communication. Other authors have shown the effect of hospital transition–related medical errors on readmission,4 and the potential effect that communication14 and discharge summary availability15 can have on that outcome. Residents' exposure to the literature on care transitions may have motivated them to improve their chart documentation and to be more deliberate about sending a carbon copy of summaries to outpatient physicians.

Residents benefited from receiving feedback on their notes. Other authors have demonstrated that chart review and feedback can positively affect resident documentation.16,17 Because individual feedback to each resident was minimal, with only 6 discharge summaries reviewed during the course of each academic year, we conclude that education was enhanced by the novel structure of the workshop. Residents were engaged in active learning through peer assessment, a well-described classroom strategy for adult learners18 that has been shown to be effective for both recipients19 and providers20 of feedback.

The READ workshop provided a forum for education and assessment regarding resident medical record keeping, while addressing the ACGME competencies7 of interpersonal and communication skills, systems-based practice, and practice-based learning and improvement with minimal trainee or faculty time commitment.

Despite our high baseline rate of inclusion of essential elements and no significant improvement in that metric, we found improvement in the overall quality of the notes, suggesting that meeting basic requirements on a dictation template does not translate to a high-quality discharge summary. Residents should be instructed to create discharge summaries that exceed the Joint Commission's requirements by including specific information that will be relevant to outpatient providers (eg, rationale for decisions made, pending results, contingency plans). The Society of Hospital Medicine discharge checklist7 may serve as a more useful training tool.

Our study had several limitations including single-site intervention, lack of a control group, and small sample size, which translated into suboptimal interrater agreement scores. The improvement we detected in carbon copy rates to outpatient providers was unrelated to the evaluation tool and, therefore, not affected by interrater agreement. Finally, it is unclear whether the improvements we detected during training will persist once trainees are in practice.

Our findings show that through peer assessment and faculty emphasis on the importance of the discharge summary, residency programs can equip residents with skills to improve their discharge summaries beyond typical template use.