Background: Over 8 million Americans with diabetes require insulin for disease management. Insulin is a high-risk medication with immense potential for harm. Transitions of care from the hospital to ambulatory settings are particularly prone to error. Studies have suggested a substantial number of insulin prescriptions at hospital discharge contain errors which can result in increased morbidity and mortality for patients. Studies have also highlighted the increased risk of harm with insulin initiation, particularly in older adults, and suggested overall clinical inertia in intensifying insulin when indicated. In most large hospital systems, hospitalists manage insulin at discharge. We aimed to review current insulin prescribing practices among the hospital medicine providers at a large urban medical center.
Methods: Hospital Medicine (HM) providers and Internal Medicine (IM) residents at a single academic institution were surveyed to determine their approach to prescribing insulin at discharge. Subjective survey data was then compared to a retrospective medical record review of 282 patients with diabetes discharged on insulin by HM or IM providers over a 3-month period. A multi-disciplinary team was formed including hospitalists, endocrinologists, pharmacists, a social worker, and nurse educators to identify system-based interventions to reduce error and improve the quality of insulin prescribing.
Results: Survey data revealed only 23% of respondents felt “very confident” in prescribing the correct type and dose of insulin at discharge, while 57% felt “mostly confident” and 20% felt “somewhat” or “not at all” confident. Regardless of confidence level or provider role, most respondents (71%) based discharge insulin regimens primarily on in-hospital insulin requirements rather than HbA1c as recommended by current American Diabetes Association (ADA) guidelines. Less than half of providers (44%) always or most of the time included prescriptions for insulin supplies at discharge. Less than half of providers (45%) included diabetes specific insulin discharge instructions. Only 24% of providers checked insurance coverage when prescribing new insulin. Retrospective review of medical records revealed less than half (45%) of patients with a HbA1c >9% had their insulin increased at discharge, as advised by ADA guidelines. Non-insulin diabetes medications were increased or added in 2% of patients with type 2 diabetes and HbA1c between 7-9%. Additionally, 73% of patients discharged on insulin had none or only some of the necessary insulin supplies prescribed. Inpatient consult to diabetes education was associated with increased likelihood of patients being discharged with all supplies. Discharge instructions communicating insulin regimens to patients were completed in less than half (44%) of discharges and 27% of those completed were inaccurate. For the 89 patients with outpatient data for follow up within 30 days of discharge, insulin remained unchanged in 73%.
Conclusions: Survey and retrospective chart review data indicated providers discharging patients on insulin could benefit from system-based interventions with clinical decision-making aids and education to improve provider confidence and reduce insulin prescription errors. Proposed interventions include development of a diabetes discharge order set with embedded clinical decision-making support, easy reference educational materials for providers, and restructuring of current diabetes discharge instructions.