Abstract
Background:
Insulin, a high-risk medication, is prone to prescribing errors. Patients with diabetes experience higher hospitalization rates and extended hospital stays. Prescription errors, such as missing orders, inappropriate insulin type, missing instructions, and lack of appropriate intensification of insulin regimens are common issues. This project explored the use of system-based interventions and educational tools to minimize errors and improve the quality of insulin discharge regimens.
Methods:
A needs assessment and baseline chart review were conducted before adapting a diabetes order set obtained from the University of California, San Diego. Subsequent beta testing and broader implementation were followed by repeat chart reviews to assess the impact.
Results:
Providers strongly desired an insulin discharge order set, with 98% of those surveyed expressing this preference. Those who were high utilizers of the order set showed increased rates of ordering all supplies (55%), compared with pre-intervention rates (27%). However, no change was observed in the practice of intensifying insulin regimens in patients with uncontrolled diabetes upon discharge.
Discussion:
Insulin prescribing is prone to error. A diabetes discharge order set may improve the percentage of patients who receive necessary insulin supplies at discharge and provide educational resources to encourage appropriate insulin regimens at hospital discharge.
Patients with diabetes face higher hospitalization rates and longer hospital stays than patients without diabetes.1-3 Diabetes affects 11.3% (37.3 million) of the U.S. population, 4 with nearly 25% using insulin. 5 Despite being a high-risk medication, 6 errors in insulin prescribing are common, especially during transitions of care such as hospital discharge.7-9 The American Diabetes Association (ADA) recommends adding or increasing insulin posthospital discharge if diabetes is uncontrolled, and especially with hemoglobin A1c (HbA1c) level >9%. 10 Unfortunately, clinical inertia in advancing diabetes medication regimens persists despite well-documented clinical risks of having HbA1c levels above goal. 11
Hospital discharge for patients on insulin is associated with increased risks, including missing prescriptions, inappropriate insulin type, lack of administration instructions, inaccurate dosages, and lack of supplies.7-9 The electronic medical record (EMR) has tools available to potentially improve outcomes for these patients. One group demonstrated that the implementation of a diabetes-specific order set increased the number of patients being discharged with appropriate supplies as well as the number being discharged with bolus insulin. 12 Another project utilized an EMR alert for insulin prescriptions without associated pen needle order at discharge and were able to show a decrease in 90-day hospital readmissions and a 30% reduction in incomplete insulin prescriptions. 13
At our institution, hospital medicine (HM) providers discharge approximately 60 to 70 patients on insulin every month. This quality improvement project investigated the discharge process for patients on insulin and implemented system-based interventions, leveraging the EMR, to reduce errors in ordering insulin supplies and optimize insulin regimens at discharge. The primary outcome of this study was to determine whether implementation of a diabetes discharge order set would improve ordering of all supplies in patients discharged on insulin. A secondary outcome was whether implementation of the order set with real-time access to ADA guidelines would result in more frequent insulin intensification for patients with uncontrolled diabetes (HbA1c >9%).
Methods
The institutional review board determined that our project did not require approval as it was not considered human subjects research. We conducted a retrospective interventional study at a large Midwest academic medical center using Epic’s EMR (Epic Systems Corporation. Verona, WI, USA). Our HM division includes physicians and advanced practice providers (APPs), with physicians involved in intern and resident training. Our retrospective analysis included adult (aged >19 years) patients with diabetes admitted to our HM service under either observation or inpatient status and discharged from the hospital with insulin. Patients with insulin pumps, those who were discharged to hospice care, those who left against medical advice, or those who were transferred to other hospitals were excluded. Pregnant patients were not specifically excluded but are not typically admitted to HM at our institution and were not likely to be included. No patients evaluated had a diagnosis of gestational diabetes. Patients discharged to a skilled nursing facility or acute rehabilitation facility were included as these patients are often discharged home within two to four weeks 14 and providing insulin supply prescriptions with discharge orders was felt to be important in smoothly managing transitions of care.
Study Design
We assessed the discharge process used by HM providers for patients with diabetes requiring insulin, through an anonymous survey (Supplemental Table S1). Pre-intervention and post-intervention chart reviews were then conducted to evaluate insulin prescription completeness and quality prior to and following implementation of a diabetes discharge order set. The pre-intervention chart reviews evaluated 282 patients with diabetes discharged on insulin by the HM service from July 1, 2021 to September 30, 2021. Post-intervention chart reviews evaluated 312 patients discharged from April 17, 2023 to July 31, 2023 and employed the same chart review method and personnel for consistency. Insulin supplies ordered at discharge were categorized into “All,” “Some,” or “None” for analysis. “All” was defined as blood glucose strips + lancets + pen needles/syringes (depending on whether pens or vials were prescribed) or a continuous glucose monitor + pen needles/syringes.
Using tools within Epic, we were able to identify those HM providers who ranked among the top 50 of all providers to use the order set. Forty HM providers were identified as “high utilizers” and were compared with “low utilizers” who used the order set less frequently or not at all.
The ADA guidelines for transitions of care recommend insulin intensification at hospital discharge for patients with an HbA1c of 7% to 9% and a greater degree of intensification for patients with an HbA1c of >9%. 15 To evaluate quality of insulin prescriptions, we compared HbA1c obtained within 3 months of admission and looked to see whether providers were intensifying insulin regimens in the most uncontrolled patients, those with HbA1c >9%. We compared the total daily dose of insulin for each patient preadmission with the total daily dose prescribed at discharge. Any increase in total daily dose was considered intensification of the patient’s insulin regimen.
Order Set Development
A multidisciplinary team, including hospitalists, endocrinologists, pharmacists, a social worker, nurse educators, and information technologists collaborated to develop a diabetes discharge order set to reduce error and encourage guideline-directed insulin prescription practices. We partnered with endocrine providers at the University of California, San Diego, who shared their standardized diabetes discharge order set. 16 Using this as a template, with leadership support from HM and endocrine departments, we collaborated with informatics to develop our own order set. Hard stops were incorporated to prevent missing supplies, and insulin ordering was standardized for fixed and mealtime dosing with correction scales. The order set contained embedded educational guidance for ordering providers, including indications and contraindications for various noninsulin, antidiabetic medications as well as ADA recommendations on insulin regimen intensification based on admission HbA1c to encourage evidence-based prescribing practices (Supplemental Figures S1a-f).15,17 The order set was designed for hospital discharge and outpatient clinic functionality. The initial version, excluding combination medications and insulin pumps, was created in September 2022, piloted on March 1, 2023, and implemented across the institution on April 17, 2023 (Supplemental Figure S2).
Anticipated challenges included order set accessibility, provider education and buy-in, limitations of the Epic discharge navigator, and the evolving pharmacology landscape.
Provider Education
Comprehensive provider education was conducted to encourage utilization of our new order set. Initially, key stakeholders in endocrinology and HM, as well as internal medicine (IM) resident teams, received an in-person presentation on the project background and aims, along with a walkthrough of the order set. Following this, a concise video covering the same content was sent to leaders in all relevant hospital departments for distribution to their staff. Some departments opted for an in-person presentation in addition to the video (Family Medicine, General Surgery, and outpatient General IM). We also created a poster (Supplemental Figure S3) with pre-intervention statistics, targeted improvement approaches, and suggestions for utilizing the order set to enhance insulin safety on discharge. This poster was displayed in HM and IM provider workrooms.
Statistical Analysis
Using PC SAS version 9.4 (SAS Institute, Cary, NC, USA), we analyzed data: categorical variables with frequency and percentages, and continuous variables with median and quartiles. Nonparametric continuous variables were analyzed, using either the Mann-Whitney test (when two groups were compared) or the Kruskal-Wallis test (three or more groups were compared), with adjusted pairwise comparisons using the Dwass, Steel, Critchlow-Fligner method. Categorical variables were analyzed using χ2 tests, or Fisher exact tests when expected cell counts were too small.
Results
Survey Results
The anonymous survey, distributed to HM providers and IM residents, achieved an overall response rate of 37% (n=66), with a 50% response rate among HM providers (Supplemental Table S1). Providers overwhelmingly (98%) expressed a desire for an insulin discharge order set, citing its potential value in simplifying insulin ordering. Confidence in prescribing insulin varied, with 33% of attending physicians feeling “very confident,” 57% “mostly confident,” and 10% “somewhat confident.” The APPs and residents reported lower confidence levels. Questions on discharge practices revealed that most prescribers based discharge insulin dosing primarily on hospital insulin requirements rather than HbA1c and 30% would discharge patients on correction insulin only. Providers reported consistently including prescriptions for diabetes supplies only 44% of the time. Written insulin discharge instructions, a templated form within Epic’s discharge navigator used to clarify insulin regimens, blood glucose goals, and providing anticipatory guidance, were included by only 45% of providers.
Chart Review
Demographic differences (Table 1) between pre-data and post-data included a significant increase in type 1 diabetes patients in the post-intervention group (15.6% vs 9.8%, P = .04). Primary admission problems and 90-day readmission rates also significantly differed between pre-intervention and post-intervention populations (P < .01 and P = .04, respectively).
Demographic Comparison for Pre-Data/Post-Data.
Abbreviation: GI, gastrointestinal; IQR, interquartile range.
Wilcoxon rank sum P value.
Chi-square P value.
Fisher exact P value.
Significant differences at
The initial three-month chart review at University of Nebraska Medical Center (UNMC) exposed deficiencies in the discharge process, with 73% of patients discharged with incomplete insulin supplies. Insulin was intensified in only 45% of patients with HbA1c >9% (Table 2).
Outcomes.
Significant differences at <.05.
After order set implementation, patients discharged with all supplies improved from 27% to 42%, P < .01 (Figure 1). When patients being discharged to a facility were removed from the analysis, all supplies improved from 32% to 49%, P value < .01 (Table 2). Diabetes education was consulted on 34% of patients and significantly increased the likelihood of patients being discharged with all supplies (52% vs 14%, P < .01). This remained significant in post-intervention data, with 57% of those with a diabetes education consultation discharging with all supplies compared with 33% of those without, P < .01. For patients newly started on insulin, defined as those patients not on insulin prior to hospital admission, ordering all supplies improved from 57% to 67%, P = .34. High utilizers of the order set were more likely to send all supplies compared with low utilizers (55% vs 40%, P = .17). Similar results were observed when “All” and “Some” groups were combined as those who were high utilizers of the order set sent all or some supplies 80% of the time, compared with 70% of low utilizers (P = .19). In the case of patients newly initiated on insulin at discharge, high utilizers provided all supplies at discharge in 7 of 8 (87%) compared with 18 of 29 (62%), P = .57.
Percentage of charts with insulin supplies at discharge relative to use order set.
Intensification of insulin regimens at discharge for patients with an HbA1c of >9% occurred in 45% of discharges pre-intervention, but only in 41% of discharges post-intervention, P = .52 (Supplemental Table S2). Insulin was newly started in an overall similar number of patients: 13% versus 12%, P = .64.
Noninsulin diabetes medications were added at discharge more frequently post-intervention, with 7% of patients being discharged with a new noninsulin medication compared with 2% pre-intervention, P < .01 (Table 2). High utilizers were more likely to prescribe noninsulin diabetes medications although the increase was not statistically significant due to the small sample size (15% vs 6%, P = .25; Table 2).
Unexpectedly, the percentage of patients who received accurate and complete discharge instructions for insulin decreased after implementation of the order set from 32% to 18% (P < 0.01). Consultation to diabetes education was significantly associated with increased likelihood of discharge instructions being filled both pre-intervention (58/95 = 61%, P < .01) and post-intervention (43/107 = 40%, P < .01).
Discussion
Our needs assessment identified insulin prescribing errors at discharge, consistent with literature findings. Most patients lacked necessary insulin supplies, a concern highlighted in a June 2022 Federal Drug Administration alert, citing missed doses and needle reuse due to missing or incorrect pen needle prescriptions. 6 Order sets have been known to reduce errors and enhance care. 18 The HM providers, accustomed to using order sets, sought standardized diabetes management, particularly when prescribing insulin.
Our findings show that an order set with hard stops can improve the frequency with which patients receive all insulin supplies. High utilizers of the order set were more likely to order complete supplies for patients newly initiated on insulin and those using insulin prior to admission. The improvement seen in those newly initiated on insulin did not reach significance, likely due to the small number of patients who met these criteria. Persistent issues with ordering supplies may be due to inconsistent use of the order set by providers and discharging patients to facilities without an immediate need for supplies. In addition, some patients may already have a portion of supplies at home and may not require new supplies at discharge. Interestingly, we found that the ordering of supplies was increased in the post–order set period even among low utilizers of the order set (Figure 1). This may be due to improved awareness of our institution’s deficiencies by all providers with our education interventions and some order set use by low utilizers.
A primary challenge in implementing the diabetes discharge order set was its accessibility. Epic’s discharge navigator allows adjustment in preadmission medications, meaning many patients already on insulin could have their regimens adjusted without opening our order set. Given the high usage of general discharge order sets at our institution, our next step was to integrate the diabetes order set directly into these discharge order sets, which occurred in August 2023. This is expected to further improve utilization. The order set will be regularly assessed and updated as new medications and prescribing updates become available. Future updates will include combination medications, continuous glucose monitors, and insulin pumps.
Previous studies have indicated clinical inertia in managing diabetes both inpatient and outpatient, despite the well-documented risks of uncontrolled diabetes. 11
Transitions of care guidelines recommend intensification of diabetes regimens at hospital discharge, especially when HbA1c is >9%.15,17 Unfortunately, we did not see improvement in increasing total daily insulin dose for patients with HbA1c >9% after order set implementation. This is likely due to the multiple known barriers to insulin intensification already reported, 11 including concern for hypoglycemia, changes in clinical condition, deference to outpatient primary care or endocrine providers, or patient nonadherence with prescribed regimens prior to hospitalization. Further studies are needed to determine the optimal rate of insulin intensification for those patients with uncontrolled diabetes and how to overcome barriers to intensification.
Furthermore, patient education challenges persisted with only 32% receiving accurate written insulin instructions at discharge, a figure that further decreased post-discharge. This may be due to the order set’s provision of standardized insulin ordering options, allowing correction scales to be included in the prescription signature, whereas prior to the order set, correction scales were written in separate instructions. In addition, higher patient volumes were experienced during the post-intervention period, which may have made extra steps in the discharge process, such as filling out insulin-specific instructions, less likely to be completed. As unclear instructions can impede post-discharge diabetes management, future initiatives aim to auto-extract data from discharge orders for accurate patient instructions.
This study has several additional limitations. Its retrospective nature studying nonconsecutive time periods raises the potential for our results to be explained by other changes occurring within our institution although no major insulin prescribing changes occurred during the study. We were only able to assess order set use by comparing high utilizers with low utilizers, not use of the order set in individual patient discharges. In addition, due to the small number of providers included in the “high utilizer” category, lack of statistical power may have precluded our ability to uncover significant differences between groups post-intervention. Focusing only on patients discharged with insulin limited insight into noninsulin medication trends although an increase in noninsulin prescriptions were observed. The order set’s introduction coincided with national shifts toward sodium-glucose cotransporter-2 inhibitors for heart failure and glucagon-like peptide 1 receptor agonists, potentially influencing our findings of increased noninsulin medication usage and perhaps the decrease in insulin intensification post-intervention. In addition, institution-wide provider education on the order set spanned months post-release, suggesting that initial data might underestimate the order set’s impact.
Strengths
High engagement and leadership support were critical to our success. Stakeholders in an interprofessional team with a diversity of skills and training were engaged early and often throughout development. Multiple departments were approached and open to receiving education about the intervention. Post-implementation feedback was highly positive on a follow-up HM provider survey although a small response rate prevented the survey from providing additional meaningful comparison.
Conclusions
In conclusion, we found that implementation of a diabetes discharge order set can improve the percentage of patients who receive necessary insulin supplies at hospital discharge. Such an order set can also provide real-time educational tools to support alignment of provider prescribing practices with evidence-based care. These findings emphasize the benefit of a systems-based intervention to standardize the approach to discharging patients on insulin.
Supplemental Material
sj-docx-10-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-docx-10-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-docx-7-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-docx-7-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-docx-9-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-docx-9-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-1-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-1-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-2-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-2-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-3-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-3-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-4-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-4-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-5-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-5-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-6-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-6-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Supplemental Material
sj-png-8-dst-10.1177_19322968241239621 – Supplemental material for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin
Supplemental material, sj-png-8-dst-10.1177_19322968241239621 for Developing a Diabetes Discharge Order Set for Patients With Diabetes on Insulin by Elizabeth Miles, Melissa McKnight, Claire C. Schmitz, Chelsea R. McElroy, Jana L. Wardian, Valerie Shostrom and Preethi Polavarapu in Journal of Diabetes Science and Technology
Footnotes
Acknowledgements
AcknowledgmentsThe authors would like to acknowledge the contributions of many other team members who used their expertise to create and evaluate the diabetes order set. Dr. Brady Bulian was instrumental in data collection both pre-data and post-data. Jocelyn Pearson, BSN, RN provided informatics support in the translation of our ideas into a cohesive order set. In addition, pharmacists Jon Knezevich, PharmD and Aaron Beck, PharmD, and UNMC nurse educators Shelby Hoskins, MS, RD, LMNT, CDCES, Dorotha Rohlfsen, BSN, RN, CDCES, CPT, Casey Vlach, BSN, RN, CDCES, CPT, and Shelly Kelley, BSN, RN, CDCES helped to improve the order set in numerous ways. We are grateful to our University of California, San Diego collaborators Kristen Kulasa, MD and Andrea Stallings, PA-C for sharing their order set and allowing us to adapt it for our particular needs.
Abbreviations
ADA, American Diabetes Association; APPs, advanced practice providers; EMR, electronic medical record; GI, gastrointestinal; HbA1c, hemoglobin A1c; HM, hospital medicine; IM, internal medicine; MAR, medical administration record; UNMC, University of Nebraska Medical Center.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
