The effects of a ‘discharge time-out’ on the quality of hospital discharge summaries
- Namita Mohta1,2,3,4,
- Prashant Vaishnava1,4,
- Cathy Liang2,
- Kye Ye5,
- Matt Vitale4,5,
- Anuj Dalal1,4,
- Jeff Schnipper1,2,4
- 1BWH Academic Hospitalist Service, Boston, Massachusetts, USA
- 2Division of General Medicine and Primary Care, Boston, Massachusetts, USA
- 3Brigham and Women's Hospital, Partners Healthcare, Boston, Massachusetts, USA
- 4Harvard Medical School, Boston, Massachusetts, USA
- 5Internal Medicine Residency Program, Boston, Massachusetts, USA
- Correspondence to Dr Prashant Vaishnava, Mount Sinai Medical Center, 1468 Madison Avenue, New York, NY 10029, USA;
Contributors NSM, PV, SY and JLS: conception of design, acquisition of data, analysis of data, generation of the article and critical revision; CLL, MV and AD: conception of design, acquisition of data, analysis of data and generation of the article.
- Accepted 27 March 2012
- Published Online First 5 May 2012
Background High-quality discharge summaries are a key component of a safe transition in care. The purpose of this study was to determine the effects of standardised feedback and a ‘discharge time-out’ (DTO) on the quality of discharge summaries.
Methods During 2006–2007, the authors trained hospitalists to provide two interventions at their discretion: (1) feedback on one discharge summary to each intern using a standardised form and (2) a DTO, modelled after the surgical time-out, in which key questions about the patient's hospital course and discharge plan are answered verbally by the intern during rounds on the day of discharge. To evaluate these interventions, trained clinicians, blinded to group assignment, performed an explicit review of two discharge summaries before and after intervention implementation. The authors used a mixed linear model to evaluate relative improvement over time.
Results The authors compared 14 interns who only received a 1-h lecture and a small-group resident-led training session with 13 interns who also received feedback and 12 interns who received feedback and a DTO. Save greater improvement in the documentation of tasks to be completed after discharge (39% vs 8% absolute improvement, p=0.05) by interns receiving an intervention, most domains were unaffected by having received a DTO and/or feedback.
Conclusion These results suggest that standardised feedback and a DTO integrated into attending rounds have limited potential to improve discharge summaries as currently designed. This study stresses the need for developing and refining interventions that can improve the narrative flow of discharge summaries.
- Transitions in care
- graduate medical education
- quality improvement
- patient safety
- healthcare quality improvement
Quality transitions of care across the care continuum are critical, specifically at the time of hospital discharge. Patient recovery from diseases requiring hospitalisation is frequently incomplete and requires ongoing evaluation and management in the outpatient or postacute setting. Appropriate postdischarge care requires complete and timely communication between inpatient and postdischarge providers.1–3 When effective communication is not realised, patients are at a higher risk of having an adverse event during the transition from hospital to home.4 5 Conversely, efforts to improve timely communication have been shown to improve compliance with follow-up tests and may decrease readmission rates.6–8
The hospital discharge summary is a core component of effective communication. It contains data elements important for postdischarge care. A recent systematic review showed that discharge summaries often lacked important information such as diagnostic test results (missing from 33% to 63% of cases), treatment or hospital course (7%–22%), discharge medications (2%–40%), test results pending at discharge (65%), patient or family counselling (90%–92%) and follow-up plans (2%–43%).3
Several studies have looked at interventions to improve the inclusion of core data elements in discharge documentation.7 9 However, even a ‘defect-free’ discharge summary may be suboptimal in terms of postdischarge provider satisfaction and patient continuity of care. Equally important, but less commonly studied than inclusion of key data elements, are the qualitative components of a discharge summary, that is, whether it is well written, organised, clear, succinct and ‘tells a story’ such that postdischarge care providers are able to understand what happened to the patient in the hospital and what potentially needs to happen next.
The best ways to improve both the quantitative and qualitative aspects of discharge documentation remain unknown. Moreover, it is unknown how best to incorporate resident training in discharge summary composition into clinical practice. This is important because at academic medical centres, residents compose the majority of discharge summaries, time for education and training are increasingly scarce and, more broadly, lessons regarding physician training in discharge summary composition may be applicable to many settings.
The aims of this study were to determine the effects of a ‘discharge time-out’ (DTO) (a method of real-time communication modelled after the surgical time-out) and standardised feedback on the quality of discharge summaries written by first year medical residents (interns).
Settings and patients
The study was conducted on the hospitalist medicine service at the Brigham and Women's Hospital, a tertiary academic medical centre in Boston, Massachusetts, USA. Each medical team consisted of a hospitalist attending (from a pool of approximately 25), one junior or senior resident, two interns, and one or two medical students. Patients typically presented with a wide variety of medical conditions, a high degree of medical complexity and multiple comorbidities. The study period was academic year 2006–2007 (ie, July 2006 through June 2007). The Partners Institutional Review Board approved the study.
From November 2006 through June 2007 (the intervention period), prior to the start of their 2-week rotations, hospitalist attendings were given training and tools to deliver two interventions: (1) feedback using a standardised form to each intern on the team and (2) a DTO during attending rounds. Training of attendings included a 1-h session on the above interventions by one of the study authors (JLS). Attendings were encouraged to provide both feedback and the DTO, but the choice to deliver either intervention was at the discretion of each attending. All interns at baseline received a lecture and a small-group resident-to-intern training session at the beginning of the academic year regarding the importance of discharge documentation, what data elements to include and how best to organise them (standard curriculum).
For standardised feedback, the study coordinator randomly selected one discharge summary per intern for evaluation during each 2-week attending rotation. The attending then reviewed and completed a form on the quality of the discharge summary (table 1). This feedback form was based on literature review,10 11 regulatory requirements such as The Joint Commission,12 expert opinion (as part of a process to develop metrics for the quality of discharge summaries across all Partners acute care hospitals)13 and local consensus. It included questions about whether key data elements regarding the hospitalisation were documented, such as relevant findings on physical exam and studies, hospital course by problem, discharge physical exam, and specific follow-up appointments and tasks. The intern and attending then reviewed the summary and feedback form together. Forms were returned to the study coordinator on a weekly basis.
We implemented a DTO, a term coined by Andrew Modest, a hospitalist at Mount Auburn Hospital, Cambridge, Massachusetts, USA (personal communication) as an analogy to the ‘surgical time-out,’ now considered standard of care prior to surgery.14 The DTO was structured as a brief verbal exchange during hospitalist attending rounds. When discussing a patient ready for discharge, the DTO addresses a series of questions to be briefly asked by the attending and answered by the intern caring for the patient (figure 1). The DTO includes key questions about the patient's initial presentation, clinical findings, final diagnosis, treatment rendered and response to treatment, condition on discharge, the discharge plan (such as details regarding medication adjustments) and the follow-up plan, including outpatient appointments and tasks.
Discharge summary composition
The discharge ordering process at the Brigham and Women's Hospital was electronic and contained a template for several discrete fields, including follow-up appointments, diet, activity, principal and secondary diagnoses, operations and procedures, discharge condition, and tasks to complete at follow-up. A tool for medication reconciliation prompted the intern to continue, discontinue or hold current medications at discharge. The entire history of present illness and hospital course section was free text. For patients whose length of stay was <1 week, the discharge paperwork produced by this electronic ordering process served as the discharge summary. For patients whose stay was longer than 1 week, the discharge summary had to be dictated using a pocket dictation card as a guide.
The primary outcome was improvement in the quality of discharge summaries based on explicit review. The review was conducted of two randomly selected discharge summaries written by each intern prior to the receipt of either intervention (usually during a prior rotation) and compared with two randomly selected discharge summaries by the same intern after receipt of feedback or at least one DTO (ie, during the same rotation in which they received the intervention(s)). We identified which interns received DTOs based on self-report, solicited from all medical teams each week.
Trained physician reviewers blinded to group assignment conducted the explicit reviews, evaluating performance against 31 key metrics in the following four areas: (1) clinical information, (2) discharge medications, (3) follow-up information and (4) global assessment. As with the feedback form, the review tool was developed based on literature review, expert opinion and local (Partners-wide) consensus. Evaluators were trained to ensure consistent and standardised evaluation across all summaries. Approximately 60% of the summaries were evaluated by two reviewers. For these summaries, the average κ score for all components of the summary was 0.44, with a range of 0.13 for hospital course to 0.96 for documentation of warfarin information.
Limited by the capacity of and funding for physician reviewers, we analysed data on a sample of 25 randomly selected interns who rotated on hospitalist medicine services during both the preintervention and intervention periods and who received at least one intervention. To control for temporal trends, assuming quality of discharge summaries would improve as the academic year progressed, we matched these interns to 14 interns who did not receive any interventions but who rotated on hospitalist medicine services during the same months of the year. For each intern, we calculated the average score for each data element in two discharge summaries written during the preintervention period. For Likert-type questions (eg, 1–5 strongly disagree to strongly agree), answers were dichotomised (agree and strongly agree vs all other responses). Then, we calculated the average score for each data element in the two summaries written in the rotation during the intervention period.
We used a mixed linear model (PROC MIXED in SAS) to compare improvement over time among the interns who received the interventions with interns who did not receive either intervention. We compared results among three groups of interns: (1) interns who received feedback and DTO; (2) interns who received feedback only; and (3) interns who received neither (ie, received standard curriculum). We also conducted a second analysis combining the two intervention groups compared with usual care. Two-sided p values <0.05 were considered significant. We used SAS V.9.0 for all analyses. Because this was a preliminary study, we did not perform power or sample size calculations.
One hundred interns rotated through the hospitalist general medicine services during the academic year 2006–2007. Twenty-five interns only rotated on these services prior to intervention training and so were ineligible for the study, while 75 interns rotated after November 2006 and therefore were eligible to receive at least one of the interventions. Of these, 51 interns received at least one intervention (68%) and 24 received only the standard curriculum (figure 2). The study population consisted of 39 interns: 25 interns (randomly selected from 51) who received any intervention and 14 usual care interns (selected from 24) matched to the intervention interns by their general medicine rotation schedule.
When combining the intervention groups together and comparing them with those interns who only received a 1-h resident-led training session, most domains were unaffected by the interventions over and above temporal trends, including documentation of the chief complaint, discharge diagnosis, and discharge lab results (table 2). Domains with improvement included documentation of clearly stated, specific tasks to follow-up (39.1% improvement vs 8.2% improvement, p<0.05). There were non-significant but directional trends towards improvement on a few measures, such as documentation of a discharge physical exam, and on several global measures of quality, such as whether the discharge summary provided the information the primary care provider would need to care for patient after discharge (27% vs 12% absolute improvement in the percentage of cases where the reviewers said ‘agree’ or ‘strongly agree’) and the overall quality of the discharge summary (29% vs 15% absolute improvement; table 2). The average intern in either arm wrote discharge summaries that contained 54% of applicable required data elements prior to the intervention; this improved to 65% postintervention among usual care interns and to 70% among interns who received the interventions (p=0.44 for comparison).
When comparing the three study groups (DTO and feedback combined, feedback only, and standard curriculum), there were some areas in which the combination of DTO and feedback demonstrated improvement over the other two groups. For example, clear documentation of follow-up tasks improved more in the combination group (54% improvement) compared with feedback alone (25%) or usual care (8%; table 3). Some global assessment metrics had non-significant trends towards improvement in the combination group compared with the other two groups. These included whether the summary provided the information the primary care provider would need to care for patient after discharge (38% improvement vs 18% vs 12%, respectively; table 3).
Overall, feedback alone or with a DTO did not significantly improve documentation of key elements of the discharge summary versus temporal trends alone, with the exception of documentation of follow-up tasks. In comparing feedback alone versus feedback combined with DTO, the combination group was no better on inclusion of most data elements, again with the exception of documentation of follow-up tasks. But even this finding is questionable given the issue of multiple testing.
Our negative findings may have been due to the relatively small sample size and limited statistical power to make comparisons. It is also possible that in at least some cases, the discharge summary may have already been prepared, at least in draft form, by interns before they received the DTO. Conversely, at least for the 10% of patients who required a dictated discharge summary, the dictation could have been done weeks later, again limiting the impact of the intervention. A DTO designed to improve ‘narrative flow’ (ie, organisation and clarity) of a discharge summary may not be capable of improving the inclusion of specific data elements: its impact may be more subtle. Last, the impact of the intervention may have been impacted by usability issues. On the one hand, real-time interventions, conducted on current inpatients during rounds, have the potential to significantly change intern behaviour. But issues of time pressure may limit the use of interventions such as these.
A few other studies have investigated how best to improve discharge summary training. One study conducted in the late 1980s at a university-affiliated Veterans Affairs hospital implemented a brief educational intervention consisting of guidelines and a specified format for dictation of discharge summaries.15 The intervention was associated with improvements in discharge summary clarity, brevity and inclusion of several data elements such as condition at discharge, discharge instructions and follow-up plans. The study was limited to 11 residents and did not have a control group. More recently, a study at the University of Pennsylvania implemented a discharge summary curriculum with or without hospitalist feedback to interns.16 Significant between group improvements were seen in use of headings and documentation of procedures when comparing usual care, curriculum alone and curriculum plus feedback. Non-significant trends in most other measures favoured the curriculum plus feedback group over curriculum alone over usual care.
Currently, much attention is being paid to the quality of discharge documentation, both from a regulatory standpoint and as a possible way of preventing unnecessary readmissions. To date, the focus has been on the inclusion of certain data elements: this is likely best handled with the use of information technology that allows for auto-importation of data and templates to solicit information.9 However, we would argue that it is at least as important to communicate a ‘story’ to the next provider. This is less well addressed by tools and technology and requires provider training. As currently designed, the DTO and feedback are not ready for widespread implementation. However, it may be that refinements in these tools may achieve the desired goals of this study. For example, during rounds, the focus of the DTO could be on the discharge and follow-up plan (ie, those elements that require specific discharge orders). For particularly complex patients, the first eight questions of the DTO could be addressed outside of rounds with that particular intern, with enough advanced notice to influence the composition of the discharge summary.
Limitations of the study include generalisability as the interventions were studied on one academic medicine service in one hospital. The study does not quantify the long-term impact of the intervention, as the postintervention evaluations occurred during the same rotation in which the intervention occurred. We also cannot rule out the possibility of contamination among interns, for example, those who received the interventions discussed it with those that did not (or that changes in culture affected all interns). There was potential confounding by the quality of attending; the choice of intervention was left to the discretion of the attending.
This study suggests that standardised feedback plus a DTO have limited potential to improve the quality of discharge documentation as currently designed and implemented. Future research should focus on how best to improve the quality of discharge summaries among interns and residents, both quantitatively and qualitatively. There is a need for novel and efficient techniques to teach these skills. Though the present study was largely negative, the potential value of real-time, patient-focused education should not be discounted.
Future studies are also necessary to establish better surrogate measures of discharge summary quality that correlate with more downstream outcomes such as postdischarge provider satisfaction. Such surrogate measures may serve as targets of interventions in their own right.
Finally, recognising that many house-staff attitudes and approaches to medical care are initially formed in medical school, there is an opportunity to improve early training of medical students regarding the importance of transitions of care generally, and the quality of discharge summaries specifically. Students spend much of their preclinical and clinical years honing their admission and progress note writing skills. We propose that equal effort be spent on honing their discharge summary composition skills so that quality can be improved during transitions in care.
Competing interests None.
Ethics approval The ethics approval was provided by the Partners Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data available on request from the corresponding author.