Information Transfer and the Hospital Discharge Summary: National Primary Care Provider Perspectives of Challenges and Opportunities

Discussion

The results of this national provider survey reveal that PCPs who have access to hospital data are more likely to have a HDS available to them at the time of posthospital followup. On a superficial level this finding seems self confirmatory; that is, if a PCP has access to hospital records he or she is more likely to have an available hospital summary. The known logistic barriers of correct PCP identification and use of mail, fax, or e-mail⁸ are removed (our finding that solo practitioners have lower HDS availability than other practice types may indicate that these issues persist for some). However, even for PCPs with direct access to inpatient data >80% of the time, only 72.4% had an available HDS >80% of the time. Hospital requirements regarding the timing of HDS preparation vary widely; presumably this gap largely relates to noncompletion of the summary by the time of the posthospital follow-up visit, but further study of other potential barriers could provide additional insight. Critically, only a quarter of respondents reported that the HDS contained all information needed more than 80% of the time, and a quarter 40% of the time or less. Technological solutions such as shared electronic medical records may remedy some of the challenges of HDS availability, but issues of timely preparation and inclusion of relevant content remain.^{4,9,16⇓–18}

Previous studies have shown that many physician trainees feel unprepared to write discharge summaries¹³, and that academic medical centers may trail other practice settings in the timely provision of complete summaries.^1,16 The relatively small number of responses within the academic practice subgroup (n = 40) limits our ability to draw strong conclusions; however, the finding that PCPs in academic settings tend to be slightly less likely to have the HDS available at posthospital followup (when adjusted for direct access to inpatient data) offers some validation that this issue lingers. Although our survey did not address potential causative factors, differential utilization of electronic medical records in academic centers, conflicting priorities for learners and teachers (ie, not viewing documentation preparation as a learning experience) and time pressure (resulting from duty hour limitations and other educational commitments) may contribute to this result. Future study focusing on the academic practice setting could help clarify this question and strengthen the validity of conclusions. In addition, Figure 1 summarizes the perceived importance of discharge summary components by PCPs; >80% reported “list of diagnosis and problems,” “treatment provided,” and “medication list with changes” as “very important.” All 11 components were felt to be “somewhat important” or “very important” by >80% of PCPs. Further study of training and workflow optimization for learners, particularly as they relate to preparation of discharge summaries that include this high-yield content, could offer considerable benefit for patients and caregivers at academic centers as well as those in communities served by graduates.

In a model adjusted for access to inpatient data, community size, and practice setting the HDS “information score” (a reflection of whether specific desired information was contained in the summary) was found to be higher with more direct access to inpatient data. This may reflect automated HDS templates active within integrated systems or information recall errors in which PCPs access the desired information outside the HDS but do not recall this distinction at survey completion. Hospital-owned clinic practices tended to have lower information scores, possibly due to reliance on automated systems or hospitalist assumptions that PCPs use information sources outside the HDS.

Our study’s findings are strengthened by engagement of providers at the national level and a favorable response rate. Table 1 demonstrates broad representation by providers at various levels of experience across multiple practice and community settings, which will enhance the generalizability of these results. In addition, given that responses to the single question asking how often the discharge summary contains all information needed for follow-up care are associated with the frequency of specific pieces of information being contained in the summary we can infer validation of adequate performance for this survey question.

Limitations of our study include sampling bias; the use of a society-specific database potentially restricts our ability to generalize conclusions beyond the demographic profiles and practice preferences of AAFP members. However, the influence of this bias should be minimized by the broad membership of AAFP (approximately 125,000 members in the United States). In addition, nonresponse bias could favor the perspectives of providers who are more able and willing to complete surveys. PCPs who encounter barriers to HDS availability and instead obtain pertinent information directly from the patient could be disproportionately represented as well. Our favorable response rate, as well as the wide demographic profile of our respondents, should mitigate these effects. This survey was administered in 2015; continued implementation and refinement of electronic medical record systems could conceivably have improved HDS content and availability in the interim, but the deficits identified in our results are unlikely to have been wholly remedied since that time. This survey was limited to physicians. Inpatient and outpatient care teams increasingly incorporate Nurse Practitioners and Physician Assistants as core members; these groups, with unique perspectives stemming from alternate training and scope of practice models, may offer additional insights. Finally, the survey did not address specific barriers to availability of inpatient data and the HDS (such as conservative interpretation of protected health information sharing guidelines by hospital systems). Further study of contributing factors may aid practice redesign and enhance information access.

The HDS serves as the main conduit for communication between inpatient and outpatient care teams. More broadly, previous study has shown that increased communication between PCPs and specialists, using health information technology tools such as electronic health records, reduces hospitalizations.¹⁹ Focusing on the posthospital care transition, complete and available summaries have been shown to reduce hospital readmission risk.¹⁰ Despite this recognition, our survey demonstrates that PCPs continue to identify deficits in HDS content and availability; although having access to inpatient medical data increases the likelihood that a clinician will have the HDS available to them, a high level of deficiency in desired content remains. In this age of practice redesign and electronic medical record deployment, our findings suggest that initial efforts should include ensuring availability of a complete summary (containing items deemed important by PCPs including diagnosis list, treatment provided, and medication list) at the posthospital follow-up visit. Improved outcomes during and after this critical care transition will be a result welcomed by both patients and their care providers.