American Journal of Health-System Pharmacy

Louden,, Les;Mirtallo, Jay, M.;Worley,, Marcia;Naseman,, Ryan;Hafford,, Amanda;Brown, Nicole, V.

doi: 10.2146/ajhp160877pmid: 29167144

Abstract Purpose Operational efficiency improvements for pharmacy workflow processes were evaluated using a barcode-enabled and integrated medication-tracking system for medications dispensed from the pharmacy to the emergency department (ED). Methods The preimplementation study period (Period 1) was defined as November 17–December 16, 2015; system implementation and training, were defined as December 17, 2015–January 18, 2016; and postimplementation (Period 2) was defined as January 19–February 17, 2016. Periods 1 and 2 were compared to (1) quantify the number and type of phone calls received related to medication inquiries, (2) evaluate the percentage of redispensed doses per total dispensed doses, and (3) assess the rate of medication administration record (MAR) messages received per total dispensed doses. Results A reduction in the total number of phone calls by 77% was observed (from 125 to 29 calls). A 0.7% difference was detected for re-dispensed doses as well as MAR messages (0.009% difference in rate) between the 2 study periods. This difference was observed despite an increase in the total amount of dispensed doses that occurred for both redispensed doses (936) and MAR messages (920) during Period 2. Conclusion A barcode-enabled and integrated medication-tracking system was successfully implemented into the medication distribution process in the ED. The process change increased operational pharmacy efficiencies by decreasing medication status phone calls, redispensed doses, and MAR messages. barcode, efficiency analysis, integrated medication-tracking system, missing medication A variety of factors exist in the hospital setting that prevent the timely delivery and administration of medications. Examples include missing medications (e.g., a nurse cannot locate a medication for administration), interruptions and distractions impeding pharmacy workflow (e.g., checking on orders), multiple delivery methods (e.g., pneumatic tube, pick up from pharmacy, delivery to medication bins) and patient transfers. Green et al.1 found the most common cause for medications not being administered (38% of omissions) was directly related to medications not being available on the patient care unit. Technological advancements such as barcoding, automated dispensing cabinets (ADCs), electronic healthcare records (EHRs), and Web-based medication-tracking systems have been developed to enhance pharmacy medication dispensing processes.2 However, inefficient methods still exist with these technologies that limit the ability to locate medications in the dispensing process. Without having an effective means to trace medications after leaving the pharmacy, there is a lack of transparency and accountability within the system resulting in rework for pharmacy and nursing staff. Consequently, nurses and other staff view only the time a pharmacist verifies medication but do not know where the medication is in the distribution process. Previous medication-tracking systems have historically been limited to using various Web-based tracking system vendors (e.g., Medboard, MedEx). These medication-tracking systems have been shown to reduce medication turnaround time (i.e., the time from physician order entry to final medication delivery).3 However, they require extra time and duplicate work from hospital staff to operate multiple electronic systems simultaneously that do not integrate with the EHR. In addition to the limited visibility for all staff (e.g., pharmacy, nursing, medical) to view real-time medication-tracking status, these systems require purchasing additional hardware (e.g., TV monitors, Bluetooth-enabled scanners, vendor-supplied 2-D barcodes) along with paying service fees for the software, which result in expensive ongoing costs. Epic Systems Corporation (Madison, WI) is a member of the Electronic Health Record Association, an organization that aims to improve patient care, increase efficiencies, enhance patient safety, and provide better outcomes for patients.4 Epic’s 2014 release of the Willow Inpatient Pharmacy System provides the capability to track individual patient medications within the EHR, which is now feasible via the dispense tracking system. Pairing this medication-tracking functionality within the EHR provides an integrated medication-tracking system combined with the use of barcode-scanning technology. Using an integrated medication-tracking system offers the ability for all medical staff to view real-time tracking updates and medication delivery status. Further, this tracking system provides several advantages over other tracking vendors, such as limited equipment (i.e., only required purchase of wired barcode scanners), no additional hardware or software purchases (i.e., wireless handheld computers, TV monitors for pharmacy and nursing units, and vendor-supplied 2-D barcodes), and no ongoing service fees (feature included within the Epic software). Open in new tabDownload slide Les Louden, Pharm.D., M.S., BCPS, is currently the pharmacy manager for St. Joseph’s Hospital with BayCare Health System in Tampa, Florida. He received his bachelor of science degree, cum laude, in chemistry from Appalachian State University in 2009 and his doctor of pharmacy degree, magna cum laude, from East Tennessee State University in 2014. Dr. Louden completed a postgraduate year 1/postgraduate year 2/M.S. in health-system pharmacy administration (residency) at The Ohio State University Wexner Medical Center in June 2016. He obtained his board-certified pharmacotherapy specialist certification in December 2015. His current interests include pharmacy automation technology, finance, and medication safety. Open in new tabDownload slide Les Louden, Pharm.D., M.S., BCPS, is currently the pharmacy manager for St. Joseph’s Hospital with BayCare Health System in Tampa, Florida. He received his bachelor of science degree, cum laude, in chemistry from Appalachian State University in 2009 and his doctor of pharmacy degree, magna cum laude, from East Tennessee State University in 2014. Dr. Louden completed a postgraduate year 1/postgraduate year 2/M.S. in health-system pharmacy administration (residency) at The Ohio State University Wexner Medical Center in June 2016. He obtained his board-certified pharmacotherapy specialist certification in December 2015. His current interests include pharmacy automation technology, finance, and medication safety. Implementation of a barcode-enabled and integrated medication-tracking system has the potential to improve operational efficiencies for our pharmacy and staff by delivering medications in a more timely and transparent manner. Operational efficiency is defined as the ability to deliver products and services in an efficient manner without sacrificing quality.5 Industries such as aviation have used operational efficiency to evaluate different input measures.6 The objective of this study was to perform an operational efficiency analysis of a barcode-enabled and integrated EHR medication-tracking system for medications dispensed from the pharmacy to the emergency department (ED) at a large academic medical center. This pilot unit was chosen because of the high volume of patient-specific doses, which require delivery by pharmacy technicians or pneumatic tube. Methods Site description. The Ohio State University Wexner Medical Center (OSUWMC) is a 1,311-bed academic medical center located in Columbus, Ohio. The ED at OSUWMC is a designated Level 1 Trauma Center with 105 beds and approximately 71,000 visits annually. Medication distribution to the ED is primarily decentralized with approximately (78–81%) of all dispensed doses coming from an ADC. In addition to dispensed doses from ADCs, the inpatient pharmacy dispenses over 70,000 first doses and 15,000 cart fill medications to the ED annually. The pharmacy utilizes pneumatic tubes and pharmacy technicians to deliver medications to the ED. Barcode-enabled and integrated medication-tracking system. In early 2015, after a request for proposal analysis, the OSUWMC department of pharmacy decided to select Epic’s barcode-enabled and integrated medication-tracking system over other available third-party medication-tracking systems. This decision was based on promoting the use of 1 interoperable system along with the ability to track medications directly from the electronic medication administration record (MAR). Other considerations included (1) barcoding—utilization of the same Epic system’s barcodes on either the product or production labels, (2) startup costs—limited equipment costs necessary (2-D USB cable barcode scanners at ~$300 each) for implementation, and (3) future costs—no additional ongoing fees for the barcode-enabled and integrated medication-tracking system. Study design. A single group before-and-after test study was developed to evaluate the success of a barcode-enabled and integrated medication-tracking system for the ED. The preimplementation period, Period 1, was 29 days long from November 17 to December 16, 2015 and allowed for the collection of initial data points. The system training and implementation period was 32 days long from December 17, 2015 (system go-live date) to January 18, 2016. The system training and implementation period served as a process standardization period to allow adequate staff training and understanding of operational changes. Postimplementation, Period 2, was 29 days long from January 19 to February 17, 2016, and provided data results after system implementation. This study was determined by The Ohio State University’s institutional review board (IRB) to not be human research and did not require a formal IRB review. System training and implementation. The 32-day implementation period was utilized to allow time for pharmacy and nursing staff to adapt to new process changes. Specific educational tutorials were developed for pharmacists, pharmacy technicians, and nursing staff. Each tutorial covered process changes as well as how to effectively navigate and operate the integrated medication-tracking system. Pharmacy and technician managers provided additional troubleshooting support. Further education and instructions were delivered to the phone triage technicians to support accurate documentation on the phone call log. Regular follow-up was provided to emphasize study objectives and increase compliance with consistent documentation of phone calls. A new pharmacy workflow process was developed to facilitate additional barcode-scanning steps required for the barcode-enabled and integrated EHR medication-tracking system. Medication-tracking barcode labels were added to cart fill medication production labels to support the barcode requirements necessary for medication tracking. Pharmacy technicians were responsible for barcode-scanning medications to utilize the barcode-enabled and integrated medication-tracking system for sending and receiving medications. Pharmacy technicians sent doses by 3 different routes including “tubed,” “pick up at pharmacy,” and “out for delivery.” Tubed medications were considered received upon sending via the pneumatic tube system. Any medications sent via “pick up at pharmacy” or “out for delivery” were subsequently scanned as received upon nurse pick up or after a completed delivery. Medications delivered to the ED were hand delivered to patient-specific bins within medication rooms. Hand-delivered medications included those unable to be sent by pneumatic tube or were part of the twice-daily cart fill delivery. Pharmacy technicians were able make additional notations within the embedded tracking system to each medication upon sending or receiving by adding notes (e.g., “tubed to ED-CDU #201,” “placed in refrigerator,” “left in medication bin for room #32,” “picked up by nurse aide Suzie Baker”) necessary to help assist in clarifying medication location. Nursing education explained how to view medication status updates directly from the MAR. Education focused on the convenience of using pharmacy messages, and specific pharmacy messages within the tracking system allowed for viewing all tracking event updates for all patient medications (Figure 1). Alternatively, nurses also checked individual medication status updates while simultaneously sending an MAR message to pharmacy staff (Figure 2). Individual tabs for tracking events for the last 12, 24, or 48 hours were available from the MAR message view. Follow-up education occurred by communication from nurse educators using e-mail notifications, afternoon huddles, and weekly news updates. Figure 1 Open in new tabDownload slide Tracking events for 12, 24, and 48 hours. Tracking events show actions that have occurred for tubed, picked up, or hand-delivered medications. Copyright Epic Systems Corporation. Used with permission. Figure 1 Open in new tabDownload slide Tracking events for 12, 24, and 48 hours. Tracking events show actions that have occurred for tubed, picked up, or hand-delivered medications. Copyright Epic Systems Corporation. Used with permission. Figure 2 Open in new tabDownload slide Medication administration record message with tracking events. Updated tracking event actions can be viewed before submitting a message to the pharmacy. Copyright Epic Systems Corporation. Used with permission. Figure 2 Open in new tabDownload slide Medication administration record message with tracking events. Updated tracking event actions can be viewed before submitting a message to the pharmacy. Copyright Epic Systems Corporation. Used with permission. Data collection. Data collected during the study included (1) phone calls received related to missing medications, (2) redispensed medications per total number of dispensed doses, and (3) MAR messages received per total dispensed doses. Only medications dispensed from University Hospital’s (UH) central pharmacy were included in this study. Excluded medications were those dispensed from ADCs. Further inclusion and exclusion criteria were defined within each efficiency measure subcategory. Data from Period 1 and Period 2 were collected using Epic reports. Data were described using descriptive statistics (means and percentages) using Statistical Analysis System 9.3 (SAS, Cary, NC). Phone calls. Phone triage technicians utilized a data collection sheet, or phone call log, to manually document the number and type of phone calls received relating to all ED medication inquiries. Only phone calls received by the phone triage technician located within UH’s main pharmacy were included. Medication inquiry types included missing medications, medication status, and other (e.g., medication no longer needed). Redispensed medications. Data reports of dispensed medications were utilized to track specific medication order ID numbers to identify the percentage of redispensed medications relative to the total dispensed doses during each study period. MAR messages. Data reports of dispensed medications along with a customized MAR message report were used to match the medication order ID number to identify the percentage of MAR messages received relative to the total dispensed doses during each study period. Dispense types included first dose, scheduled doses, and cart fill medications. Redispensed doses were excluded as MAR messages are placed only for existing medication orders. Results Phone calls A total of 125 phone calls were received during Period 1 compared to 29 phone calls during Period 2. In Period 1, 35 (28%) and 70 (56%) phone calls were related to missing medications and medication status versus 16 (55%) and 9 (31%) during Period 2, respectively. Although the number of phone calls was greatly reduced during the 2 study periods, the reduction in phone calls related to medication status should be noted. Medication status phone calls were reduced from 70 (56%) during Period 1 to 9 (31%) during Period 2. Higher percentages were noted for missing medications in Period 2 (55%) versus Period 1 (28%). Redispensed doses A total of 3,468 dispensed doses occurred during Period 1 versus 4,404 dispensed doses in Period 2. Despite the increase in overall dispensed doses from Period 1 to Period 2, we realized a decrease in the percentage of redispensed doses between these periods with 8% (267) in Period 1 compared to 7% (308) in Period 2. MAR messages Because of the difference in the inclusion and exclusion criteria, a slight difference in the total number of medications dispensed was observed regarding MAR messages. A total of 3,385 dispensed doses occurred during Period 1 versus 4,305 dispensed doses in Period 2. For Period 1, 826 MAR messages were received resulting in 0.244 MAR messages per dispensed dose or 1 MAR message per 4.10 dispensed doses. In Period 2, 1,012 MAR messages were received resulting in 0.235 MAR messages per dispensed dose or 1 MAR message per 4.25 dispensed doses. A slight decrease was observed for MAR messages received despite witnessing an increase in total dispensed doses during Period 2. Discussion Implementing a barcode-enabled and integrated medication-tracking system provided operational improvements to the medication delivery process. Nursing staff was provided with tracking information, which reduced the total number of phone calls related to ED medication inquiries to the pharmacy by 77% (from 125 to 29 total phone calls). When evaluating the reduction in the medication inquiries related to medication status, the number of calls decreased from 70 to 9 between the 2 periods. Increased tracking visibility to nursing staff regarding medication-tracking status reduced phone calls related to medication inquiries. Further, we saw a slight reduction of 0.009 MAR messages per dispensed dose (1 MAR message per every 111 dispensed doses) from Period 1 to Period 2, which was encouraging given that 920 more dispensed doses occurred during Period 2. Additionally, from a pharmacy operations standpoint, we preferred using a MAR message over a phone call to help facilitate medication inquires related to missing medications. This was a pilot study, and one of the limitations was the short study period. The results may have been affected by fluctuations in patient volumes within the ED, and the short study period limited the ability to effectively account for data fluctuation. Future studies should focus on lengthening Period 1 and Period 2 study periods to collect a larger data set, which would allow for the ability to account for greater fluctuations in patient volumes. This pilot study evaluated only 1 105-bed hospital unit. The pilot study was useful to organize operational changes and obtain feedback from pharmacy and nursing staff in considering next steps for expansion. Data derived from small-scale implementation could be utilized to identify expansion methods to additional hospital units. Manual documentation of phone calls was necessary to document the changes in phone call volumes. Pharmacy technicians were asked to perform this activity over the course of a 90-day period during all 3 study periods, which led to concerns that “burnout” could have played a role in relation to the drastic decrease in documentation of phone calls. However, technicians were highly engaged in the process of medication tracking as this feature allowed for the ability to defend their actions (i.e., prove if a medication was tubed or sent on delivery). As an additional limitation, we were not able to collect data on the total number of calls placed to the pharmacy, regardless of relatedness to missing medications. Although the total number of calls related to missing medications appears to have decreased over time, we were unable to comment on how this related to all calls placed to the pharmacy. Previous methods were unable to capture data and track missing medications. The results obtained from this study could be used to provide valuable baseline data to assess our pharmacy’s operational efficiencies (i.e., phone calls, redispensed doses). Additionally, new data points now exist for sent and delivered doses because of additional barcoding steps. This current system provides the opportunity to specifically capture data for the comparison of medication delivery times. Based on discussions with pharmacy and nursing staff, the system was a valuable addition to our medication distribution process. Study results demonstrated a decrease in phone call volume and redispensed doses, which helped prevent distractions and additional work performed by pharmacy technicians. In addition, our phone triage technician experienced enhanced communication with nursing staff because of the increased transparency while speaking with nursing staff (e.g., by confirming if medications were tubed, delivered, or picked up from the pharmacy). These benefits went beyond the data captured and provided additional value and justification for expanding the system to additional hospital units. Future studies at our institution could use these baseline data for comparison to track and trend these metrics in other units. Conclusion A barcode-enabled and integrated medication-tracking system was successfully implemented into the medication distribution process in the ED. The process change increased operational pharmacy efficiencies by decreasing medication status phone calls, redispensed doses, and MAR messages. Disclosures The authors have declared no potential conflicts of interest. References 1 Green CJ Du-Pre P Elahi N . Omission after admission: failure in prescribed medications being given to inpatients . Clin Med . 2009 ; 9 : 515 – 8 . Google Scholar Crossref Search ADS WorldCat 2 Ward MJ Boyd JS Harger NJ . An automated dispensing system for improving medication timing in the emergency department . World J Emerg Med . 2012 ; 3 : 102 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Calabrese SV Williams JP . Implementation of a web-based medication tracking system in a large academic medical center . Am J Health-Syst Pharm . 2012 ; 69 : 1651 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 4 HIMSS Electronic Health Record Association . EHR developer code of conduct . www.ehra.org/ASP/codeof-conduct.asp (accessed 2016 Mar 15). 5 Adedeji BB Lee CY Johnson AL . Operational efficiency . In: Badiru AB , ed. Handbook of industrial and systems engineering . 2nd ed. Boca Raton, FL : CRC Press ; 2013 : 17 . Google Preview WorldCat COPAC 6 Sarkis J . An analysis of the operational efficiency of major airports in the United States . JOM . 2000 ; 18 : 335 – 51 . Google Scholar Crossref Search ADS WorldCat Copyright © 2017 by the American Society of Health-System Pharmacists, Inc. All rights reserved.

Cotugno,, Stephanie;Morrow,, Gina;Cooper,, Chandra;Cabie,, Maribeth;Cohn,, Sara

doi: 10.2146/ajhp160755pmid: 29167145

Abstract Purpose Results of an initiative to improve assessment and documentation of the influenza vaccination status of adult psychiatric inpatients are reported. Methods A prospective quality-improvement study was conducted at a large, tertiary care academic medical center with the aim of improving compliance with the Influenza Immunization (IMM-2) quality measure, which was added to the Inpatient Psychiatric Facility Quality Reporting (IPFQR) program in 2015 and requires assessment and documentation of influenza vaccination status in specified groups of psychiatric inpatients. The primary objective was to improve the IMM-2 IPFQR compliance rate to 100% during the 2015–16 influenza season from a rate of 55% during the 2014–15 influenza season through pharmacist interventions; secondary objectives included analysis of types of pharmacist interventions, rates of influenza vaccination status assessment and ordering, and rates of vaccine refusal by psychiatric disease state. Results With pharmacist interventions, the IMM-2 IPFQR compliance rate was increased to 99% during the 2015–16 influenza season. Of the 1,413 patients included in the study population, 45% (n = 646) were targeted for pharmacist intervention. Influenza vaccine was ordered for 61% of the study population (n = 867 patients), with an overall refusal rate of 74% (n = 642). Differences in refusal rates by psychiatric diagnosis were not significant. Conclusion Pharmacist-conducted education of nurses and interventions to ensure completion of influenza vaccine assessments and documentation led to an improved IMM-2 IPFQR compliance rate at the study site. compliance, influenza, influenza vaccines, nursing, pharmacy, psychiatry Every year, approximately 1 in 5 people are infected with the influenza virus in the United States.1 Influenza is an acute, highly contagious viral infection of the nose, throat, and lungs that typically occurs in the late fall, winter, or early spring. Common symptoms include fever (a temperature of 38–39 °C), myalgia, malaise, sore throat, and nasal congestion. With about 226,000 people hospitalized for influenza complications each year, the disease is the fifth leading cause of death in the United States.1 Studies have shown influenza vaccine administration to be the most effective method of preventing influenza and its potentially severe complications. Vaccine use decreases influenza-associated hospitalizations by 71.1% among all adults and 76.8% among adults 50 years of age or older.2 In 2015, the Centers for Medicare and Medicaid Services (CMS) added the Influenza Immunization (IMM-2) prevention quality measure to the Inpatient Psychiatric Facility Quality Reporting (IPFQR) program. This quality measure (IMM-2 IPFQR) aims to ensure screening for seasonal influenza immunization status and vaccine administration prior to discharge, if indicated, among psychiatric inpatients 6 months of age or older admitted to acute care hospitals IMM-2 IPFQR compliance is defined as the completion of an influenza vaccination status assessment and documentation of vaccine administration or refusal or documentation of a vaccine contraindication or receipt of the vaccine prior to admission.3 Compliance with this quality measure requires only completion of assessment and documentation; it does not require influenza vaccine administration. Rates of adherence to medication regimens have been observed to be lower among inpatients with psychiatric disorders than among patients with only physical disorders.4 Against that backdrop, a quality-improvement study to prospectively evaluate seasonal influenza vaccine assessments and provide pharmacist interventions for acute care hospitalized psychiatric inpatients at a large, tertiary care academic medical center was conducted. The overall goals were to assess and improve compliance with the IMM-2 IPFQR measure, as hospitalization is an opportunity to assess vaccination status and potentially reduce influenza-associated hospitalizations. Methods Study design. The study was a prospective quality-improvement study conducted in a 1,541-bed tertiary care academic medical center with 102 adult psychiatric beds. An exemption was granted by the institutional review board, as the evaluated intervention is part of the institution’s standard of practice per the “Nurse/Pharmacist Standing Order Protocol for Influenza Vaccine.” The quality-improvement study was based on this protocol. The study was designed and implemented through collaboration by an interprofessional team of hospitalists, nurses, nursing managers, and pharmacists. View largeDownload slide Stephanie Cotugno, Pharm.D., BCPS, is an emergency department clinical pharmacist at UPMC Shadyside Hospital in Pittsburgh, Pennsylvania. She received her doctor of pharmacy degree from the University of Pittsburgh in 2015 and completed an ASHP-accredited residency in hospital pharmacy at Yale-New Haven Hospital in 2016. She is an active member of the American College of Clinical Pharmacy. Dr. Cotugno’s current professional interests include emergency medicine, critical care, and psychiatric pharmacy. View largeDownload slide Stephanie Cotugno, Pharm.D., BCPS, is an emergency department clinical pharmacist at UPMC Shadyside Hospital in Pittsburgh, Pennsylvania. She received her doctor of pharmacy degree from the University of Pittsburgh in 2015 and completed an ASHP-accredited residency in hospital pharmacy at Yale-New Haven Hospital in 2016. She is an active member of the American College of Clinical Pharmacy. Dr. Cotugno’s current professional interests include emergency medicine, critical care, and psychiatric pharmacy. Patient selection. The study included all adult psychiatric patients admitted to the institution from October 2015 through March 2016. All acute care hospitalized psychiatric inpatients older than 6 months of age and included in the IMM-2 IPFQR target population were identified. Patients who were less than 18 years of age, had a length of stay greater than 120 days, left against medical advice, were transferred or discharged to another acute care hospital, or were admitted to the psychiatric observational unit were excluded from the review. Patients less than 18 years of age were excluded because this patient population is admitted to the institution’s 15-bed children’s psychiatric inpatient service, which is under the jurisdiction of our children’s hospital and requires informed written or spoken consent prior to the administration of influenza vaccine. Study timeline. The study consisted of 3 phases. In phase 1, a retrospective chart review to determine the IMM-2 IPFQR compliance rate during the 2014–15 influenza season was completed to obtain baseline data and identify opportunities for improved compliance. In phase 2, the IMM-2 IPFQR quality measure was introduced to the psychiatric leadership, and a designated pharmacist provided education to all psychiatric nurses on each psychiatric unit. In phase 3, a daily report identifying all newly admitted psychiatric inpatients was developed. Once the patients were identified, a pharmacist began the intervention workflow. Interventions included a daily review of newly admitted patients meeting the inclusion criteria, review (and, as appropriate, completion) of influenza vaccine assessments, ordering of influenza vaccine if indicated, review of the medication administration record after 24 hours of vaccine ordering, and provision of feedback to nursing staff as necessary. Interventions. A pharmacist provided education on influenza vaccine assessment completion to the psychiatric nursing teams on each psychiatric inpatient unit prior to the beginning of influenza season. The in-service sessions introduced the IMM-2 IPFQR quality measure and compliance requirements. The pharmacist also reviewed the health system’s influenza vaccine ordering protocol, nurses’ scope of practice in Connecticut, and nurses’ existing workflow for completion of influenza vaccine assessment, vaccine ordering, and medication administration record documentation. The pharmacist also reviewed Centers for Disease Control and Prevention influenza vaccination information1 and created an influenza vaccine tip sheet and a frequently-asked-questions document. The pharmacist also encouraged the completion of computer-based influenza training and assessment questions created by the institution. The completion of computer-based training and assessment questions was considered mandatory by the institution; nurse attendance at the in-service conducted by the pharmacist was strongly encouraged but not mandatory. The pharmacist reviewed nursing staff influenza vaccine assessments documented in the electronic medical record and identified patients with incomplete assessments. The pharmacist then completed influenza vaccine assessments as necessary and ordered vaccine if indicated. The pharmacist also reviewed the medication administration record in the electronic medical record for appropriate documentation of vaccine administration or refusal. In addition to the daily pharmacist intervention process, the pharmacist provided regular feedback to psychiatric nursing staff leadership and, if indicated, education of psychiatric nurses on appropriate influenza vaccine assessment completion. Data collection. During the influenza season, a daily report generated from the electronic medical record was used to identify all acute care hospitalized psychiatric inpatients to be assessed for the influenza vaccine. An influenza vaccination history was obtained and evaluated prior to the review of each patient’s electronic medical record. Patient demographic information obtained included each patient’s medical record number, admission date, unit, sex, age, and axis I diagnostic code. Information obtained from the electronic medical record regarding the influenza vaccine assessment and influenza vaccine ordering included assessment completion, vaccination status prior to admission, documented vaccine contraindications, the presence of a vaccine order, administration documentation, patient vaccine refusal, and the presence of documentation in the medication administration record. Pharmacist interventions, including the completion of the influenza vaccine assessment and ordering of the influenza vaccine, as well as the predicted IMM-2 IPFQR compliance rate, were monitored. Endpoints. The primary endpoint was the IMM-2 IPFQR compliance rate; the target rate was 100%. Secondary endpoints included the number and types of pharmacist interventions, including rates of influenza vaccine assessment completion and influenza vaccine ordering, and vaccine refusal rates categorized by psychiatric disease state. Statistical analysis. Descriptive statistics were used to analyze the collected data and compare the IMM-2 IPFQR compliance rates for the 2014–15 and 2015–16 influenza seasons. Results Prior to the 2015–16 influenza season, a pilot study involving retrospective review of the medical records of 78 psychiatric patients admitted during the period October 2014–March 2015 was completed; the study found an IMM-2 IPFQR compliance rate of 55%. During the 2015–16 influenza season, a total of 1,413 electronic medical records were examined, with no psychiatric inpatients excluded from data collection. Baseline characteristics of the psychiatric patient population were collected (Table 1). Table 1 Baseline Patient Characteristics (n = 1,413) Characteristic Value Sex, no. (%)  Male 785 (55.5)  Female 628 (44.5) Age, yr  Mean ± S.D. 40 ± 17.4  Median (range) 38 (15–94) Psychiatric diagnosis, no. (%)  Major depressive disorder 343 (24)  Mood disorder 228 (16)  Bipolar disorder 204 (15)  Schizoaffective disorder 196 (14)  Psychosis 156 (11)  Schizophrenia 142 (10)  Polysubstance abuse disorder 60 (4)  Othera 84 (6) Characteristic Value Sex, no. (%)  Male 785 (55.5)  Female 628 (44.5) Age, yr  Mean ± S.D. 40 ± 17.4  Median (range) 38 (15–94) Psychiatric diagnosis, no. (%)  Major depressive disorder 343 (24)  Mood disorder 228 (16)  Bipolar disorder 204 (15)  Schizoaffective disorder 196 (14)  Psychosis 156 (11)  Schizophrenia 142 (10)  Polysubstance abuse disorder 60 (4)  Othera 84 (6) a Includes posttraumatic stress disorder, anxiety disorder, intermittent explosive disorder, obsessive compulsive disorder, impulse control disorder, borderline personality disorder, attention-deficit/hyperactivity disorder, neurocognitive disorder, and adjustment disorder. Open in new tab Table 1 Baseline Patient Characteristics (n = 1,413) Characteristic Value Sex, no. (%)  Male 785 (55.5)  Female 628 (44.5) Age, yr  Mean ± S.D. 40 ± 17.4  Median (range) 38 (15–94) Psychiatric diagnosis, no. (%)  Major depressive disorder 343 (24)  Mood disorder 228 (16)  Bipolar disorder 204 (15)  Schizoaffective disorder 196 (14)  Psychosis 156 (11)  Schizophrenia 142 (10)  Polysubstance abuse disorder 60 (4)  Othera 84 (6) Characteristic Value Sex, no. (%)  Male 785 (55.5)  Female 628 (44.5) Age, yr  Mean ± S.D. 40 ± 17.4  Median (range) 38 (15–94) Psychiatric diagnosis, no. (%)  Major depressive disorder 343 (24)  Mood disorder 228 (16)  Bipolar disorder 204 (15)  Schizoaffective disorder 196 (14)  Psychosis 156 (11)  Schizophrenia 142 (10)  Polysubstance abuse disorder 60 (4)  Othera 84 (6) a Includes posttraumatic stress disorder, anxiety disorder, intermittent explosive disorder, obsessive compulsive disorder, impulse control disorder, borderline personality disorder, attention-deficit/hyperactivity disorder, neurocognitive disorder, and adjustment disorder. Open in new tab The overall IMM-2 IPFQR compliance rate during the 2015–16 influenza season was 99% (1,393 of 1,413 patients), which was significantly higher than the IMM-2 IPFQR compliance rate during the 2014–15 influenza season (99% versus 55%, p = 0.0001). The monthly compliance rates for CMS-audited cases during the 2015–16 influenza season are provided in Table 2. Table 2 Monthly Audit-Confirmed IMM-2 IPFQR Compliance Rates at Study Site During 2015–16 Influenza Seasona,b Month Fraction (%) of Patients October 52/53 (98) November 53/54 (98) December 55/55 (100) January 50/50 (100) February 49/49 (100) March 49/50 (98) Month Fraction (%) of Patients October 52/53 (98) November 53/54 (98) December 55/55 (100) January 50/50 (100) February 49/49 (100) March 49/50 (98) a IMM-2 IPFQR = Influenza Immunization (IMM-2) prevention quality measure for the Inpatient Psychiatric Facility Quality Reporting (IPFQR) program. b Compliance was verified by Centers for Medicare and Medicaid Services audit of selected patient records. Open in new tab Table 2 Monthly Audit-Confirmed IMM-2 IPFQR Compliance Rates at Study Site During 2015–16 Influenza Seasona,b Month Fraction (%) of Patients October 52/53 (98) November 53/54 (98) December 55/55 (100) January 50/50 (100) February 49/49 (100) March 49/50 (98) Month Fraction (%) of Patients October 52/53 (98) November 53/54 (98) December 55/55 (100) January 50/50 (100) February 49/49 (100) March 49/50 (98) a IMM-2 IPFQR = Influenza Immunization (IMM-2) prevention quality measure for the Inpatient Psychiatric Facility Quality Reporting (IPFQR) program. b Compliance was verified by Centers for Medicare and Medicaid Services audit of selected patient records. Open in new tab Of the 1,413 psychiatric inpatients in the study population, 646 (45%) were identified as requiring pharmacist intervention. Of the pharmacist interventions, 376 (58%) involved both influenza vaccine assessment completion and vaccine ordering, 144 (22%) involved only influenza vaccine ordering, and 126 (20%) required only assessment completion. IMM-2 IPFQR compliance was documented via nursing staff workflow for 767 (55%) of patients. Based on influenza vaccine assessment data, it was determined that 391 patients (28%) had received the vaccine prior to admission, and 143 (10%) had refused the vaccine on admission. Influenza vaccine was ordered for 867 patients (61%). Assessment was not provided for 5 patients, an order was not provided for 5 patients, and allergy to influenza vaccine was documented for 2 patients. Among the 867 patients for whom influenza vaccine was ordered, 642 (74%) refused vaccine administration, and 225 (26%) received vaccine doses. Analysis of refusal rates by diagnosis (Table 3) found that patients with bipolar disorder had the highest refusal rate (67%), followed by patients with psychosis (62%), mood disorder (60%), schizophrenia (57%), or schizoaffective disorder (51%). Table 3 Influenza Vaccine Refusal Rates in Study Population, by Psychiatric Diagnosis Diagnosis n Vaccine Indicated and Ordered but Refused, No. (%) Bipolar disorder 204 137 (67) Psychosis 156 97 (62) Mood disorder 228 138 (60) Schizophrenia 142 81 (57) Schizoaffective disorder 196 100 (51) Major depressive disorder 343 162 (47) Polysubstance abuse disorder 60 24 (40)  Othera 84 46 (55) Diagnosis n Vaccine Indicated and Ordered but Refused, No. (%) Bipolar disorder 204 137 (67) Psychosis 156 97 (62) Mood disorder 228 138 (60) Schizophrenia 142 81 (57) Schizoaffective disorder 196 100 (51) Major depressive disorder 343 162 (47) Polysubstance abuse disorder 60 24 (40)  Othera 84 46 (55) a Includes posttraumatic stress disorder, anxiety disorder, intermittent explosive disorder, obsessive compulsive disorder, impulse control disorder, borderline personality disorder, attention-deficit/hyperactivity disorder, neurocognitive disorder, and adjustment disorder. Open in new tab Table 3 Influenza Vaccine Refusal Rates in Study Population, by Psychiatric Diagnosis Diagnosis n Vaccine Indicated and Ordered but Refused, No. (%) Bipolar disorder 204 137 (67) Psychosis 156 97 (62) Mood disorder 228 138 (60) Schizophrenia 142 81 (57) Schizoaffective disorder 196 100 (51) Major depressive disorder 343 162 (47) Polysubstance abuse disorder 60 24 (40)  Othera 84 46 (55) Diagnosis n Vaccine Indicated and Ordered but Refused, No. (%) Bipolar disorder 204 137 (67) Psychosis 156 97 (62) Mood disorder 228 138 (60) Schizophrenia 142 81 (57) Schizoaffective disorder 196 100 (51) Major depressive disorder 343 162 (47) Polysubstance abuse disorder 60 24 (40)  Othera 84 46 (55) a Includes posttraumatic stress disorder, anxiety disorder, intermittent explosive disorder, obsessive compulsive disorder, impulse control disorder, borderline personality disorder, attention-deficit/hyperactivity disorder, neurocognitive disorder, and adjustment disorder. Open in new tab Discussion Administration of influenza vaccine has been shown to be the most effective method of preventing influenza and potentially severe complications leading to hospitalization.2 CMS implements quality measures to monitor outcomes to ensure effective and safe patient-centered care. CMS has implemented a prevention quality measure (IMM-2 IPFQR) to improve influenza vaccine compliance, specifically in the psychiatric patient population, to enhance the quality, safety, and performance of inpatient psychiatric services.3 The results of the study described here demonstrated that pharmacist interventions, including influenza vaccine education, assessment, and ordering, can significantly improve the IMM-2 IPFQR compliance rate. From the 2014–15 influenza season (during which no influenza vaccine– specific pharmacist interventions were performed) to the 2015–16 influenza season (during which such interventions were performed), the IMM-2 IPFQR compliance rate at the study site increased by 44 percentage points. Among the 1,413 psychiatric inpatients hospitalized during the study period, 45% were targeted for pharmacist intervention, resulting in a total of 1,022 pharmacist interventions. The majority of interventions involved both completion of the influenza vaccine assessment and influenza vaccine ordering. With pharmacist intervention, the influenza vaccine was ordered for the majority of psychiatric patients admitted during the study period, giving them the opportunity to receive the vaccine and be protected from infection with influenza virus and associated complications. However, the majority of ordered influenza vaccine doses were refused. Refusal rates did not differ significantly by psychiatric diagnosis. Limitations of this quality-improvement study included the acute psychiatric condition of patients at admission—the time when the vaccine was initially offered. Depressive, manic, or psychotic symptoms may lead patients to refuse the influenza vaccine at the time of admission. It is possible that patients might accept the influenza vaccine at discharge once appropriate treatment for the underlying psychiatric condition has been initiated. A potential limitation regarding the completion of the influenza vaccine assessment and the proper documentation of influenza vaccine refusal or administration was a lack of knowledge of proper assessment and documentation procedures among some nurses due to the seasonal change in nursing staff workflow during the influenza season; that lack of knowledge may have been exacerbated if attendance at the pharmacist-conducted nursing staff in-service was not achievable. Additional limitations included difficulty in providing pharmacist-to-patient influenza vaccine education due to pharmacist time constraints. Lastly, the refrigerator containing the influenza vaccines was not linked to the institution’s medication dispensing system, which may have resulted in delayed vaccine administration or missed opportunities for administration prior to patient discharge. Future directions and opportunities for improvement include increasing the influenza vaccine administration rate by providing influenza vaccine education directly to patients. Although analysis of refusal rates by psychiatric diagnosis did not indicate statistically significant differences, the results may be useful in identifying patients who may benefit from vaccine education. Additional opportunities to improve influenza vaccine administration rates may include consideration of alternative influenza vaccine formulations, such as intradermal vaccine. Lastly, continuing annual nursing staff and pharmacist influenza vaccine education and providing feedback to nurses as indicated will be essential in maintaining high IMM-2 IPFQR compliance rates. Conclusion Pharmacist-provided education of nurses and interventions to help ensure completion of influenza vaccine assessments and documentation led to an improved IMM-2 IPFQR compliance rate at the study site. Disclosures The authors have declared no potential conflicts of interest. Previous affiliations At the time of the study, Dr. Cotugno was affiliated with Yale-New Haven Hospital. References 1 Centers for Disease Control and Prevention . Key facts about influenza and the influenza vaccine ( 2015 ). www.cdc.gov/flu/protect/keyfacts.htm (accessed 2016 Apr 11). 2 Talbot HF Zhu Y Chen Q . Effectiveness of influenza vaccine for preventing laboratory-confirmed influenza hospitalization in adults . Clin Infect Dis . 2013 ; 56 : 1774 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Specifications manual for National Hospital Inpatient Quality Measures, discharges 01-01-14 (1Q14) through 09-30-14 (3Q14) . IMM-2 IPFQR . Oak Brook, IL : Joint Commission ; 2013 . WorldCat COPAC 4 Cramer JA Rosenheck R . Compliance with medication regimens for mental and physical disorders . Psychiatr Serv . 1998 ; 49 : 196 – 201 . Google Scholar Crossref Search ADS PubMed WorldCat Copyright © 2017 by the American Society of Health-System Pharmacists, Inc. All rights reserved.

Awker, Amy, L.;Bell, Melissa, A.;McGraw,, Megan;Klein, Mark, A.

doi: 10.2146/ajhp160726pmid: 29167146

Abstract Purpose The effect of education regarding new evidence in periprocedural anticoagulation, with a focus on reducing use in patients at only moderate thromboembolic risk, is presented. Methods A retrospective cohort analysis and quasiexperimental design were used. The initial review identified the current state of practice regarding bridging anticoagulation. Education was then provided to primary care providers and pharmacists on recent evidence. A subsequent review was completed to assess the impact of this education on clinical decision-making. Inclusion criteria were adults taking warfarin for an indication of mechanical heart valve, atrial fibrillation (AF), or history of venous thromboembolism (VTE) and required interruption of warfarin therapy to undergo a planned procedure. Patients were excluded if their anticoagulation was managed outside of the Minneapolis Veterans Affairs Anticoagulation Clinic. Results The overall rate of bridging decreased from 38.8% to 24.8% (14% decrease; 95% confidence interval [CI], 2–26%; p = 0.028) confidence interval [CI], 0.02–0.26; p = 0.026) after educational intervention. This decrease occurred in the moderate thromboembolic risk group, in which the bridging rate decreased from 63.8% to 30.2% (33.6% decrease; 95% CI, 14–53%; p = 0.001). Bleeding complications occurred more frequently in patients who received bridging. There were no thromboembolic complications. Conclusion The majority of patients at moderate thromboembolic risk were previously receiving bridging until new evidence was released indicating that the risks may outweigh any benefits. The provision of education to primary care physicians and pharmacy staff regarding this new evidence in the area of periprocedural anticoagulation management significantly reduced the amount of bridging used for patients on warfarin for AF or a history of VTE who were at moderate thromboembolic risk. anticoagulants, atrial fibrillation, clinical decision-making, pharmacists, venous thromboembolism, warfarin Although the number of warfarin visits have declined from 2009 to 2014 with the release of the direct-acting oral anticoagulants (DOACs), a substantial number of patients remain on warfarin, and it is estimated that about 250,000 patients per year undergo procedures for which warfarin must be stopped during the periprocedural period.1,2 Many of these patients undergo bridging anticoagulation, which refers to the administration of a short-acting anticoagulant, such as a low molecular weight heparin (LMWH), for the 10–12-day period during which warfarin is held and the International Normalized Ratio is below the target range.2 Intuitively, the belief has been that the shorter the amount of time without therapeutic anticoagulation, the lower the rate of thromboembolism. However, this rationale remains theoretical and has not been proven in well-designed clinical studies. Current guidelines from the American College of Chest Physicians (CHEST) suggest the use of bridging anticoagulation in patients at high risk of thromboembolism and avoidance of its use in low-risk patients. There is a lack of data to support a specific recommendation in patients at moderate risk of thromboembolism, and guidelines suggest this decision should be made based on individual patient and surgery-related factors.2 The recommendations in these guidelines, however, are weak, as they are based on low-quality evidence. Since the release of the most recent CHEST guidelines in 2012, a number of studies have been published in this area. In the Orbit-AF study, Steinberg et al.3 performed a retrospective review of patients with atrial fibrillation (AF) undergoing warfarin interruption for an elective procedure. Patients who received bridging were compared with those who did not receive bridging. Average CHADS2 scores were 2.34 in the patients who did not receive bridging and 2.53 in those who received bridging. The rates of thromboembolism were low and were not statistically different between the 2 groups, but the rate of bleeding was higher in the patients who received bridging. In the Kaiser venous thromboembolism (VTE) study, Clark et al.4 performed a similar retrospective review in patients interrupting warfarin with a history of a prior VTE event. Results were similar, showing that the rates of recurrent VTE were not different between the 2 groups, but there was more bleeding in the patients who received bridging. Finally, the BRIDGE trial was the first prospective, randomized, multi-center, placebo-controlled trial in which patients with AF were randomized to receive bridging with a LMWH versus placebo.5 Patients were required to have a minimum CHADS2 score of ≥1 for inclusion. The average CHADS2 score for the patients who did not receive bridging compared with those who received bridging was 2.3 versus 2.4, respectively. The placebo group was statistically noninferior to the bridging group for rates of thromboembolism, but the bridging group had 3 times the rate of major bleeding complications and almost double the rate of minor bleeding complications.5 The majority of patients in all 3 of these studies were either considered to be at low or moderate risk of thromboembolism. These recent studies provided the first high-quality evidence favoring a strategy of foregoing bridging anticoagulation in patients at low or moderate risk of thromboembolism. Open in new tabDownload slide Amy L. Awker, Pharm.D., earned her bachelor of arts degree in chemistry in 2011 from the University of Minnesota and her doctor of pharmacy degree from the University of Minnesota College of Pharmacy in 2015. Dr. Awker completed a postgraduate year 1 pharmacy practice residency at the Minneapolis Veterans Affairs (VA) Health Care System, where she gained clinical skills in a variety of pharmacy practice settings, including cardiology, internal medicine, psychiatry, and ambulatory care. She currently works at the Minneapolis VA as a Clinical Rotational Pharmacist. Open in new tabDownload slide Amy L. Awker, Pharm.D., earned her bachelor of arts degree in chemistry in 2011 from the University of Minnesota and her doctor of pharmacy degree from the University of Minnesota College of Pharmacy in 2015. Dr. Awker completed a postgraduate year 1 pharmacy practice residency at the Minneapolis Veterans Affairs (VA) Health Care System, where she gained clinical skills in a variety of pharmacy practice settings, including cardiology, internal medicine, psychiatry, and ambulatory care. She currently works at the Minneapolis VA as a Clinical Rotational Pharmacist. In the Minneapolis Veterans Affairs Health Care System (MVAHCS), anticoagulation is managed by a centralized, pharmacist-run anticoagulation clinic. When patients on warfarin must interrupt their therapy to undergo a procedure, the service performing the procedure submits a periprocedural consult to the anticoagulation clinic. The pharmacist then performs an assessment for bridging that includes determining a patient’s individual level of thromboembolic and bleeding risk. Based on this risk level, a protocol allows the pharmacist to determine the treatment plan for patients at low or high thromboembolic risk. However, for patients at moderate risk, the pharmacist must consult the patient’s primary care physician (PCP) for agreement with the treatment plan. It was previously unknown how often and in which circumstances PCPs recommended bridging. The primary purpose of this study was to assess the impact of providing education to PCPs and pharmacists regarding updated literature on the implementation of a new evidence-based approach to periprocedural anticoagulation targeting a reduction in bridging for patients at moderate thromboembolic risk. Secondary outcomes included evaluations of 30-day postprocedure thromboembolic and bleeding complications. Methods We first conducted a retrospective cohort study to determine our practice patterns in the management of periprocedural anticoagulation. Based on the results of the initial study, we conducted a quasiexperimental study to identify a strategy to better implement evidence-based management of periprocedural anticoagulation. The electronic medical record was searched to identify all patients who received a consultation at the anticoagulation clinic for periprocedural management of anticoagulation between the two time periods of April 1– August 31, 2015, and January 1–March 31, 2016. Patients were included if they were age 18 or older; were actively on warfarin therapy for an indication of mechanical heart valve, AF, or VTE; needed to undergo interruption of warfarin therapy for an elective procedure; and the consult was completed during the study time periods. Patients were excluded if they were on a DOAC, they were on warfarin for an alternative indication, or their warfarin was not managed by the MVAHCS. Patients who underwent more than 1 procedure during the study time periods were only included once with their first procedure. The flow of patients included in the study is represented in Figure 1. Charts were reviewed for basic demographic information, indication for warfarin, applicable risk factors and level of thromboembolic risk, applicable risk factors and level of bleeding risk, type of procedure, bridging decision, and 30-day postprocedure bleeding and thromboembolic complications. Any bleeding symptom reported by the patient was recorded; however, major bleeding was classified according to the International Society on Thrombosis and Haemostasis definition as fatal bleeding, bleeding into a critical area (i.e., intracranial, intraspinal, intraocular, retroperitoneal, intraarticular or pericardial, or intramuscular with compartment syndrome), drop of hemoglobin of 2 g/dL or more, and/or leading to transfusion of 2 or more units of whole blood or red cells.6 Clinically relevant nonmajor bleeding was defined as overt bleeding not meeting the criteria for major bleeding but associated with a hospitalization or emergency department visit.6 Thromboembolic risk categories (high, moderate, low) were defined according to the criteria outlined in the 2012 CHEST guidelines.2 These guidelines categorize patients according to the indication for anticoagulation and then risk stratify based on presence of other risk factors. For patients with mechanical valves, the risk depends on whether it is a mitral or aortic valve plus the presence of other factors. For patients with AF, the risk depends on their CHADS2 score. Lastly, for patients with prior VTE, the risk depends on how recently the event occurred and the presence of other factors, such as thrombophilias or cancer. Figure 1 Open in new tabDownload slide Patient selection. DOAC = direct acting oral anticoagulants, MVAHCS = Minneapolis Veterans Affairs Health Care System. Figure 1 Open in new tabDownload slide Patient selection. DOAC = direct acting oral anticoagulants, MVAHCS = Minneapolis Veterans Affairs Health Care System. Patients who received bridging were bridged according to the MVAHCS standard bridging protocol. According to this protocol, warfarin is stopped 5 days prior to the procedure, and enoxaparin is initiated 3 days prior to the procedure in the morning. For patients with an estimated creatinine clearance (CLcr) greater than 30 mL/min, a dose of enoxaparin 1 mg/kg every 12 hours is used, and for those with an estimated CLcr of ≤30 mL/min, a dose of enoxaparin 1 mg/kg every 24 hours is used. Enoxaparin is stopped after the morning dose the day before the procedure (about 24–36 hours pre-procedure). Warfarin is resumed with the patient’s previous regimen the evening of the procedure day according to the surgeon or provider performing the procedure. Enoxaparin is resumed in the morning on the day after the procedure (post-op day 1). Following the initial study period, formal education was provided to the department of general internal medicine physicians at a monthly staff meeting covering the results of the Orbit-AF, Kaiser VTE, and BRIDGE trials as well as the results from the initial study period. Additionally, a formal continuing-education presentation was provided to pharmacy staff. The results from the posteducation period were then compared to the results from the preeducation study period. The overall rates of bridging and of bridging specifically in the moderate-risk patient population were compared between the preeducation and the posteducation study groups using a 2-sample test of proportions and a significance level set at 0.05. This study was approved by the Minneapolis Veterans Affairs institutional review board (IRB) as an IRB-exempt quality-improvement study. Results Patient characteristics. A combined total of 320 patients were identified as receiving a consult of perioperative anticoagulation management. Thirty percent of the patients were excluded in the preintervention group and 35% of the patients were excluded in the postintervention group (Figure 1). Thirteen patients in the postintervention group were having another procedure and were excluded because of previous inclusion in either the preintervention or the postintervention group. Other reasons for exclusion included consults completed outside of the study time frame, consults placed in error for patients who were not receiving anticoagulation, and consults placed in error for patients who did not require interruption of their anticoagulation therapy. Patient characteristics are listed in Table 1. The mean age of patients was 70.7 years and the median age was 70 years. Approximately 97% were males in this veteran population. The most common indication for warfarin was AF in both groups, and the majority of patients were at low or moderate thromboembolic risk. The mean CHADS2 score of patients in the preintervention group was 2.19, and the mean CHADS2 score of patients in the postintervention group was 2.14. The types of procedures performed are listed in Table 2. In both groups, the most common type of procedure was gastrointestinal, followed by urologic and orthopedic in the preintervention and postintervention groups, respectively. Table 1 Patient Characteristics Characteristic Study Period Preintervention (n = 116) Postintervention (n = 101) Mean ± S.D. age, yr 71.8 ± 8.7 69.4 ± 7.9 Sex, no. (%)  Male 112 (96.6) 99 (98)  Female 4 (3.4) 2 (2) Indication for warfarin, no. (%)  Mechanical valve 13 (11) 10 (10)  Atrial fibrillation 72 (62) 57 (57)  Venous thromboembolism 31 (27) 34 (34) Thromboembolic risk, no. (%)  High 13 (11) 12 (12)  Moderate 47 (41) 42 (42)  Low 56 (48) 46 (46) Characteristic Study Period Preintervention (n = 116) Postintervention (n = 101) Mean ± S.D. age, yr 71.8 ± 8.7 69.4 ± 7.9 Sex, no. (%)  Male 112 (96.6) 99 (98)  Female 4 (3.4) 2 (2) Indication for warfarin, no. (%)  Mechanical valve 13 (11) 10 (10)  Atrial fibrillation 72 (62) 57 (57)  Venous thromboembolism 31 (27) 34 (34) Thromboembolic risk, no. (%)  High 13 (11) 12 (12)  Moderate 47 (41) 42 (42)  Low 56 (48) 46 (46) Open in new tab Table 1 Patient Characteristics Characteristic Study Period Preintervention (n = 116) Postintervention (n = 101) Mean ± S.D. age, yr 71.8 ± 8.7 69.4 ± 7.9 Sex, no. (%)  Male 112 (96.6) 99 (98)  Female 4 (3.4) 2 (2) Indication for warfarin, no. (%)  Mechanical valve 13 (11) 10 (10)  Atrial fibrillation 72 (62) 57 (57)  Venous thromboembolism 31 (27) 34 (34) Thromboembolic risk, no. (%)  High 13 (11) 12 (12)  Moderate 47 (41) 42 (42)  Low 56 (48) 46 (46) Characteristic Study Period Preintervention (n = 116) Postintervention (n = 101) Mean ± S.D. age, yr 71.8 ± 8.7 69.4 ± 7.9 Sex, no. (%)  Male 112 (96.6) 99 (98)  Female 4 (3.4) 2 (2) Indication for warfarin, no. (%)  Mechanical valve 13 (11) 10 (10)  Atrial fibrillation 72 (62) 57 (57)  Venous thromboembolism 31 (27) 34 (34) Thromboembolic risk, no. (%)  High 13 (11) 12 (12)  Moderate 47 (41) 42 (42)  Low 56 (48) 46 (46) Open in new tab Table 2 Procedure Type in Study Patientsa Procedure No. (%) by Study Period Preintervention (n = 116) Postintervention (n = 101) Gastrointestinal endoscopy 41 (35) 48 (48) Orthopedic surgery 10 (9) 12 (12) Cardiothoracic surgery 8 (7) 4 (4) Urologic surgery 15 (13) 9 (9) ENT surgery 2 (2) 5 (5) Vascular surgery 10 (9) 4 (4) Spinal injection 3 (2) 9 (9) Intraabdominal surgery 9 (8) 3 (3) Other surgery/procedure 18 (15) 7 (7) Procedure No. (%) by Study Period Preintervention (n = 116) Postintervention (n = 101) Gastrointestinal endoscopy 41 (35) 48 (48) Orthopedic surgery 10 (9) 12 (12) Cardiothoracic surgery 8 (7) 4 (4) Urologic surgery 15 (13) 9 (9) ENT surgery 2 (2) 5 (5) Vascular surgery 10 (9) 4 (4) Spinal injection 3 (2) 9 (9) Intraabdominal surgery 9 (8) 3 (3) Other surgery/procedure 18 (15) 7 (7) a ENT = ear, nose, and throat. Open in new tab Table 2 Procedure Type in Study Patientsa Procedure No. (%) by Study Period Preintervention (n = 116) Postintervention (n = 101) Gastrointestinal endoscopy 41 (35) 48 (48) Orthopedic surgery 10 (9) 12 (12) Cardiothoracic surgery 8 (7) 4 (4) Urologic surgery 15 (13) 9 (9) ENT surgery 2 (2) 5 (5) Vascular surgery 10 (9) 4 (4) Spinal injection 3 (2) 9 (9) Intraabdominal surgery 9 (8) 3 (3) Other surgery/procedure 18 (15) 7 (7) Procedure No. (%) by Study Period Preintervention (n = 116) Postintervention (n = 101) Gastrointestinal endoscopy 41 (35) 48 (48) Orthopedic surgery 10 (9) 12 (12) Cardiothoracic surgery 8 (7) 4 (4) Urologic surgery 15 (13) 9 (9) ENT surgery 2 (2) 5 (5) Vascular surgery 10 (9) 4 (4) Spinal injection 3 (2) 9 (9) Intraabdominal surgery 9 (8) 3 (3) Other surgery/procedure 18 (15) 7 (7) a ENT = ear, nose, and throat. Open in new tab Bridging. Comparisons of the rates of bridging between the 2 groups are shown in Table 3. The overall rates of bridging were 38.8% in the preintervention group and 24.8% in the postintervention group. The rate of bridging in the low-risk group decreased from 5.4% to 0% in the preintervention and postintervention groups, and in the high-risk group the rate of bridging increased from 92.3% to 100% between the preintervention and postintervention groups, respectively. These between group differences were not significant. The rates of bridging for moderate-risk patients, however, were 63.8% in the preintervention group and 30.2% in the postintervention group. For patients at moderate-risk, the rates of bridging based on indication for warfarin are shown for each group in Figure 2. The rates of bridging did not decrease for moderate risk patients on warfarin for an indication of mechanical valve, whereas they decreased for those patients with AF or prior VTE history. Table 3 Comparison of Bridging by Study Perioda Variable Study Period % Decrease (95% CI) P Preintervention (n = 116) Postintervention (n = 101) Overall percent bridged 38.8 24.8 14.0 (2 to 26) 0.028 Percent of moderate risk bridged 63.8 30.2 33.6 (14 to 53) 0.001 Variable Study Period % Decrease (95% CI) P Preintervention (n = 116) Postintervention (n = 101) Overall percent bridged 38.8 24.8 14.0 (2 to 26) 0.028 Percent of moderate risk bridged 63.8 30.2 33.6 (14 to 53) 0.001 a CI = confidence interval. Open in new tab Table 3 Comparison of Bridging by Study Perioda Variable Study Period % Decrease (95% CI) P Preintervention (n = 116) Postintervention (n = 101) Overall percent bridged 38.8 24.8 14.0 (2 to 26) 0.028 Percent of moderate risk bridged 63.8 30.2 33.6 (14 to 53) 0.001 Variable Study Period % Decrease (95% CI) P Preintervention (n = 116) Postintervention (n = 101) Overall percent bridged 38.8 24.8 14.0 (2 to 26) 0.028 Percent of moderate risk bridged 63.8 30.2 33.6 (14 to 53) 0.001 a CI = confidence interval. Open in new tab Figure 2 Open in new tabDownload slide Percent bridged in moderate-risk group by indication, AF = atrial fibrillation, VTE = venous thromboembolism. Figure 2 Open in new tabDownload slide Percent bridged in moderate-risk group by indication, AF = atrial fibrillation, VTE = venous thromboembolism. Complications. Due to the nature of the study time frame, 30-day postprocedure complications were only able to be reviewed in 71 out of the 101 patients in the postintervention group. Some consults were completed far in advance, and therefore the 30-day postprocedure date did not fall within the study time frame. Other reasons included procedures getting rescheduled for future dates outside the study time frame or canceled procedures. Of the 71 patients reviewed, 56 were not bridged and 15 were bridged. Overall, 12 patients reported bleeding of any kind. This corresponded to a 12.5% bleeding rate in those who were not bridged (7 out of 56) and a 33.3% bleeding rate in those who were (5 out of 15). However, of these 12 reports, there was only 1 case meeting criteria for major bleeding plus 1 case meeting criteria for clinically relevant nonmajor bleeding. Both of these cases occurred in patients who received bridging. The patient with the major bleeding episode developed a preperitoneal actively bleeding hematoma requiring emergent surgical intervention, reversal of anticoagulation, and an admission to the intensive care unit. The clinically relevant nonmajor bleeding episode involved excessive bleeding from an external fistula site that required an emergency department visit and brief admission for observation. Minor bruising was the most common complaint for patients who were not bridged. There were no thromboembolic events in either group. Discussion In this study there was a significant decrease in the amount of bridging used for patients at moderate thromboembolic risk in response to education regarding recent evidence. Although new guidelines have not yet been introduced in this area, current guidelines are based on low-quality evidence and have not made strong recommendations.2 Therefore, with the release of some larger, well-designed studies, providers appeared willing to change their practice to fit the new evidence. In this study, the decrease in bridging was seen specifically in the moderate-risk patients, who were on warfarin for either AF or history of VTE, which follows the new evidence. The Orbit-AF study and the Bridge trial were both based on patients with AF, but there was a very small sample of patients at high thromboembolic risk.3,5 Additionally, the Kaiser VTE study involved patients with prior VTE and also had a small number of patients at high risk.4 These studies demonstrated that there was not an increased risk to patients at low and moderate risk to forego bridging. However, to date there are no high-quality studies done in patients with mechanical valves, so consequently providers were hesitant to forego bridging in these patients. Interestingly, the rate of bridging for patients at high thromboembolic risk, as well as for those at moderate risk with mechanical valves, actually increased from the preintervention to the postintervention group. Perhaps this may have been related to the fact that the education delivered to providers emphasized that there is still a lack of quality evidence regarding the risk versus benefit of bridging in these particular patient populations. Apart from recent trial evidence, the education delivered to providers also included the results from the initial study period, which highlighted the fact that the majority of moderate-risk patients had been receiving bridging. This fact combined with the recent trial evidence likely influenced providers to give additional attention to this patient population and to consider changing their clinical decision. There remain many questions to be answered in this clinical area. For example, the current risk stratification of patients with atrial fibrillation is based on a tool that is validated to predict the risk of stroke over a 1-year period and not necessarily in the periprocedural setting. Additionally, the current version of the CHEST guidelines, as well as the recent literature, uses the CHADS2 scoring system to assess thromboembolic risk rather than the CHADS2VaSc score, which has become the standard of care in clinical practice. Additionally, the risk categorization for patients with prior VTE does not currently take into account whether the event was provoked or unprovoked. Perhaps there are other factors that may be even more important in predicting thromboembolic risk in this setting. For example, certain procedural factors, such as the inherent bleed risk of the procedure itself, may play a larger role. Recent studies suggest that bridging may not provide additional benefit. However, there may still be certain patient circumstances in which bridging would be beneficial that were not as well represented in the clinical studies (e.g., patients who have multiple risk factors in a particular category, patients who have had a thromboembolic complication during interruption of warfarin therapy in the past). In this study, it appeared that prior to the release of new evidence, the default clinical decision was to bridge patients at moderate risk. After the education, however, providers did not default to bridging; when they made the decision to bridge, it was based on specific clinical rationale. In many cases, patients had mechanical aortic valves with multiple risk factors, had a history of experiencing a stroke during warfarin therapy interruption, or had a history of extensive venous thromboembolism. On 2 occasions, the patient was presented with the potential risks and elected to be bridged. While the sample size of this study was much smaller, the trend in bleeding complications was similar to those reported in the recent literature.3–5 Bleeding complications were more frequent than thromboembolic complications, and bleeding occurred more frequently in patients who received bridging. The bleeding symptoms reported in those patients who were not bridged were not considered clinically significant, whereas the major bleeding complication and the clinically relevant nonmajor bleeding complication that occurred were both in patients who received bridging. The patient who experienced a major bleeding complication also had a history of previously experiencing a stroke when not receiving warfarin therapy, so there was reasonable clinical rationale to provide bridging to this patient. There were some significant limitations of this study, including its retrospective, single-center, quasiexperimental design. Due to the small sample size and low frequency of complications, generalizable conclusions cannot be drawn regarding between group differences in the rates of complications. Additionally, this study only applied to patients who received a consult for periprocedural anticoagulation. This means that if a patient underwent a procedure and the consult was not placed appropriately, the patient would not have been included. A previous review of this process identified that this would only apply to a small number of patients. Finally, this review did not include information about the bridging decisions for patients who underwent unplanned procedures. Conclusion The majority of patients at moderate thromboembolic risk were previously receiving bridging until new evidence was released indicating that the risks may outweigh any benefits. The provision of education to PCPs and pharmacy staff regarding this new evidence in the area of periprocedural anticoagulation management significantly reduced the amount of bridging used for patients on warfarin for AF or a history of VTE who were at moderate thromboembolic risk. Disclosures The authors have declared no potential conflicts of interest. Acknowledgments The authors acknowledge Fatima Khan, M.D., Neil Patel, M.D., and Gerhard Johnson, M.D., for their continual support of the anticoagulation clinic, Thomas Rector, Pharm.D., for his support with statistical analysis, and Anders Westanmo, Pharm.D., for his assistance with collection of data. References 1 Barnes GD Lucas E Alexander GC Goldberger ZD . National trends in ambulatory oral anticoagulant use . Am J Med . 2015 ; 128 : 1300 – 5.e2 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Douketis JD Spyropoulous AC Spencer FA . Perioperative management of antithrombotic therapy—antithrombotic therapy and prevention of thrombosis, 9th ed. American College of Chest Physicians evidence-based clinical practice guidelines . CHEST . 2012 ; 141 ( suppl ): e326S – e350S . Google Scholar Crossref Search ADS PubMed WorldCat 3 Steinberg BA Peterson ED Kim S . Outcomes Registry for Better Informed Treatment of Atrial Fibrillation Investigators and Patients. Use and outcomes associated with bridging during anticoagulation interruptions in patients with atrial fibrillation: findings from the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF) . Circulation . 2015 ; 131 : 488 – 94 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Clark NP Witt DM Davies LE . Bleeding, recurrent venous thromboembolism, and mortality risks during warfarin interruption for invasive procedures . JAMA Intern Med . 2015 ; 175 : 1163 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Douketis JD Spyropoulos AC Kaatz S . Bridging anticoagulation in patients with atrial fibrillation . N Engl J Med . 2015 ; 373 : 823 – 33 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Schulman S Angeras U Bergqvist D . Definition of major bleeding in clinical investigations of antihemostatic medicinal products in surgical patients . Thromb Haemost . 2010 ; 8 : 202 – 4 . Google Scholar Crossref Search ADS WorldCat Copyright © 2017 by the American Society of Health-System Pharmacists, Inc. All rights reserved.

McCarthy, Bryan, C.;Tuiskula, Kristin, A.;Driscoll, Tara, P.;Davis, Andrew, M.

doi: 10.2146/ajhp160848pmid: 29167147

Abstract Purpose Results of an analysis of the economic impact of adverse drug events (ADEs) resulting in patient harm on hospitalization costs and length of stay (LOS) are reported. Methods In a retrospective single-site study, medication errors among patients admitted to an academic medical center during the period April 2014–May 2015 were identified using voluntary event reporting system data and diagnosis codes. Hospitalization cases involving documented ADEs resulting in harm, as defined on a widely used medication error classification index, were matched with control cases by admission period, diagnosis-related group, and patient age and sex. Total hospitalization costs and LOS in the study groups were analyzed using an independent 2-sample Mann–Whitney U test. Results Among 416 hospitalization cases evaluated for inclusion in the study, 242 were matched with 3,279 control cases for analysis. The primary drug classes implicated in the evaluated medication errors included chemotherapy agents (38%), corticosteroids (14%), and opioids (11%). Total hospitalization costs differed significantly (p = 0.044) between patients who experienced ADEs resulting in harm (median, $19,444; interquartile range [IQR], $13,481–$40,580) and those who did not (median, $17,173; IQR, $12,500–$27,125); the former group also had a significantly (p = 0.005) longer median LOS. Conclusion Chargemaster data for an academic medical center revealed that the median total hospitalization cost and LOS were significantly greater for hospitalizations during which a harm-causing medication error was recorded versus hospitalizations during which harm-causing medication errors were not recorded. chargemaster, cost of hospitalization, length of stay, medication error, medication safety Nearly 15 years ago, the Institute of Medicine published the “To Err Is Human” report, which exposed the substantial impact of medical errors in the U.S. healthcare system and called for dramatic systemwide changes, including an improved understanding of those errors.1 Since then, research on patient safety has accelerated, as evidenced by significant increases in relevant published literature.2 An adverse drug event (ADE) is “an injury resulting from the use of a drug. Under this definition, the term ADE includes harm caused by the drug (adverse drug reactions and overdoses) and harm from the use of the drug (including dose reductions and discontinuations of drug therapy).”3 ADEs are a major source of medical errors and, as such, are the focus of much of the aforementioned patient safety research. However, the current economic impact of medication errors remains an opportunity for investigation, according to the Department of Health and Human Services.4 The cost of hospitalization and hospital length of stay (LOS) for inpatients experiencing an ADE were determined in the late 1990s and early 2000s to be significantly higher than for those who did not experience an ADE. According to inclusion criteria unique to each study, medication errors identified on scales of mild to severe or significant to life-threatening were included for analysis.5–7 Of note, more recent related publications citing this research have applied healthcare inflation rate factors to costs to promote external validity in application of these conclusions.8,9 The purpose of the study described here was to determine the specific economic impact of medication errors resulting in harm on patients’ hospitalization costs and LOS using present-day cost and LOS data. Methods The study was approved by the University of Chicago investigational review board. The retrospective study was conducted at University of Chicago Medicine, which is an academic medical center with 463 adult acute care beds, including 6 adult intensive care units (ICUs), and 155 pediatric acute care beds, including 3 pediatric ICUs. Medication errors were identified in patients admitted between April 2014 and May 2015 using 2 internal databases, including a voluntary event reporting system and a hospital-based cost accounting system. The hospital-based cost accounting system was queried for patients whose records contained an International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code indicative of an ADE (see appendix). Each ADE was verified by 2 study investigators through electronic medical record review and then categorized using the National Coordinating Council for Medication Error Reporting and Prevention (NCC-MERP) Index for Categorizing Medication Errors.10 Patient hospitalizations that involved a medication error resulting in harm, defined as error classification at category E or higher on the NCC-MERP index (denoting the occurrence of an error that may have contributed to or resulted in temporary harm to the patient and required intervention), were included for analysis. Open in new tabDownload slide Tara P. Driscoll, Pharm.D., BCPS, received her doctor of pharmacy degree from Purdue University. Dr. Driscoll completed an ASHP-accredited postgraduate year 1 pharmacy residency at University of Chicago Medicine and a postgraduate year 2 ambulatory care pharmacy residency at Indiana University. Her research interests are in the areas of ambulatory care, primary care, and specialty pharmacy. Open in new tabDownload slide Tara P. Driscoll, Pharm.D., BCPS, received her doctor of pharmacy degree from Purdue University. Dr. Driscoll completed an ASHP-accredited postgraduate year 1 pharmacy residency at University of Chicago Medicine and a postgraduate year 2 ambulatory care pharmacy residency at Indiana University. Her research interests are in the areas of ambulatory care, primary care, and specialty pharmacy. Each case patient hospitalization (CPH; a hospitalization during which a medication error resulted in harm) was matched with at least 1 control (i.e., a hospitalization during which the patient did not experience a medication error resulting in harm). Controls were matched to each CPH based on the following criteria: (1) patient classification by Medicare Severity Diagnosis Related Group, (2) hospitalization during the study period, (3) patient age within 2 years, and (4) patient sex. For each CPH, the number of controls varied depending on how many fulfilled the above criteria. CPHs were excluded if the patient experienced the medication error prior to admission, if multiple medication errors occurred during the admission, if the medication error was due to an unspecified medication, or if no identifiable control hospitalizations were available. Cost of hospitalization and LOS for CPHs and controls were determined using the hospital-based accounting system, which sources data from the hospital chargemaster. For CPHs with more than 1 control match, the means of the controls’ total hospitalization costs and LOS values were compared with those of the CPH. As the data set was not normally distributed due to outlier data points, an independent 2-sample Mann–Whitney U test was performed to determine the differences in median total hospitalization cost and LOS values between CPHs and controls. Secondary analyses were performed in various subgroups: (1) hospitalizations involving the 3 therapeutic classifications of medications most commonly implicated in medication errors in the CPHs, (2) hospitalizations involving patients who were less than 65 years of age, and (3) hospitalizations involving patients who were 65 years of age or older. All analyses were performed in Stata SE 12 (StataCorp LLC, College Station, TX). The a priori level of significance was 0.05. Results Approximately 1,600 medication error reports submitted through the voluntary event reporting system and a total of 371 random patient hospitalizations in the hospital-based cost accounting system with a documented ICD-9-CM code indicative of a medication error during the study period were selected for evaluation for inclusion in the study; 120 and 296 of medication errors identified through those systems, respectively, were categorized by study investigators as category E or higher on the NCC-MERP index. Collectively, 65 hospitalizations were excluded due to a medication error occurring prior to admission; 17, due to multiple medication errors in a single admission; 30, due to a medication error from an unspecified medication; and 54, due to a lack of identifiable controls. In addition, 8 hospitalizations were removed as duplicate data from each system. A total of 242 CPHs (39 from the voluntary event reporting system and 203 from the hospital-based accounting system) were included and matched with 3,279 controls for analysis. Patient demographics are summarized in Table 1. The majority of the patients were male, and almost half were 45–64 years old. Table 2 shows the frequency with which medications of various therapeutic classifications were implicated in medication errors in the CPHs. The most commonly implicated classifications were antineoplastic agents, corticosteroids, and opioids. Table 1 Demographic Characteristics of Study and Control Groupsa Patient Characteristic Case Group (n = 242) Control Group (n = 3,279) No. (%) in age group  <15 yr 24 (9.9) 566 (17.3)  15–29 yr 20 (8.3) 194 (5.9)  30–44 yr 27 (11.2) 154 (4.7)  45–64 yr 108 (44.6) 1,523 (46.4)b  ≥65 yr 63 (26.0) 842 (25.7) No. (%) male 129 (53.3) 2,263 (69.0)b Patient Characteristic Case Group (n = 242) Control Group (n = 3,279) No. (%) in age group  <15 yr 24 (9.9) 566 (17.3)  15–29 yr 20 (8.3) 194 (5.9)  30–44 yr 27 (11.2) 154 (4.7)  45–64 yr 108 (44.6) 1,523 (46.4)b  ≥65 yr 63 (26.0) 842 (25.7) No. (%) male 129 (53.3) 2,263 (69.0)b a Unless otherwise indicated, differences between groups are not significant. b Significantly different (p < 0.05) from value for case group. Open in new tab Table 1 Demographic Characteristics of Study and Control Groupsa Patient Characteristic Case Group (n = 242) Control Group (n = 3,279) No. (%) in age group  <15 yr 24 (9.9) 566 (17.3)  15–29 yr 20 (8.3) 194 (5.9)  30–44 yr 27 (11.2) 154 (4.7)  45–64 yr 108 (44.6) 1,523 (46.4)b  ≥65 yr 63 (26.0) 842 (25.7) No. (%) male 129 (53.3) 2,263 (69.0)b Patient Characteristic Case Group (n = 242) Control Group (n = 3,279) No. (%) in age group  <15 yr 24 (9.9) 566 (17.3)  15–29 yr 20 (8.3) 194 (5.9)  30–44 yr 27 (11.2) 154 (4.7)  45–64 yr 108 (44.6) 1,523 (46.4)b  ≥65 yr 63 (26.0) 842 (25.7) No. (%) male 129 (53.3) 2,263 (69.0)b a Unless otherwise indicated, differences between groups are not significant. b Significantly different (p < 0.05) from value for case group. Open in new tab Table 2 Therapeutic Classification of Medications Implicated in Errors That Resulted in Harm (n = 242) Therapeutic Classification No. (%) Cases in Which Implicated  Antianemia drugs 2 (0.8)  Antibiotics 22 (9.1)  Anticoagulants 6 (2.5)  Anticonvulsants 4 (1.7)  Antidepressants 1 (0.4)  Antidiabetic agents 7 (2.9)  Antidiarrhea agents 1 (0.4)  Antiemetics 3 (1.2)  Antihistamines 1 (0.4)  Antiinflammatory agents 1 (0.4)  Antilipemic agents 1 (0.4)  Antimanic agents 3 (1.2)  Antineoplastic agents 82 (33.9)  Antiprotozoals 1 (0.4)  Antituberculars 1 (0.4)  Benzodiazepines 5 (2.1)  Bronchodilators 8 (3.3)  Calcium antagonists 1 (0.4)  Cardiac drugs 4 (1.7)  Corticosteroids 34 (14.0)  Diagnostic agents 4 (1.7)  Electrolytes 1 (0.4)  Hypotensive agents 2 (0.8)  Immunosuppressive agents 7 (2.9)  Local anesthetics 3 (1.2)  Opioid antagonists 1 (0.4)  Opioids 28 (11.6)  Parathyroid hormones 1 (0.4)  Sedatives and hypnotics 1 (0.4)  Skeletal muscle relaxants 1 (0.4)  Sympathomimetic agents 1 (0.4)  Vasodilating agents 3 (1.2)  Vitamin K 1 (0.4) Therapeutic Classification No. (%) Cases in Which Implicated  Antianemia drugs 2 (0.8)  Antibiotics 22 (9.1)  Anticoagulants 6 (2.5)  Anticonvulsants 4 (1.7)  Antidepressants 1 (0.4)  Antidiabetic agents 7 (2.9)  Antidiarrhea agents 1 (0.4)  Antiemetics 3 (1.2)  Antihistamines 1 (0.4)  Antiinflammatory agents 1 (0.4)  Antilipemic agents 1 (0.4)  Antimanic agents 3 (1.2)  Antineoplastic agents 82 (33.9)  Antiprotozoals 1 (0.4)  Antituberculars 1 (0.4)  Benzodiazepines 5 (2.1)  Bronchodilators 8 (3.3)  Calcium antagonists 1 (0.4)  Cardiac drugs 4 (1.7)  Corticosteroids 34 (14.0)  Diagnostic agents 4 (1.7)  Electrolytes 1 (0.4)  Hypotensive agents 2 (0.8)  Immunosuppressive agents 7 (2.9)  Local anesthetics 3 (1.2)  Opioid antagonists 1 (0.4)  Opioids 28 (11.6)  Parathyroid hormones 1 (0.4)  Sedatives and hypnotics 1 (0.4)  Skeletal muscle relaxants 1 (0.4)  Sympathomimetic agents 1 (0.4)  Vasodilating agents 3 (1.2)  Vitamin K 1 (0.4) Open in new tab Table 2 Therapeutic Classification of Medications Implicated in Errors That Resulted in Harm (n = 242) Therapeutic Classification No. (%) Cases in Which Implicated  Antianemia drugs 2 (0.8)  Antibiotics 22 (9.1)  Anticoagulants 6 (2.5)  Anticonvulsants 4 (1.7)  Antidepressants 1 (0.4)  Antidiabetic agents 7 (2.9)  Antidiarrhea agents 1 (0.4)  Antiemetics 3 (1.2)  Antihistamines 1 (0.4)  Antiinflammatory agents 1 (0.4)  Antilipemic agents 1 (0.4)  Antimanic agents 3 (1.2)  Antineoplastic agents 82 (33.9)  Antiprotozoals 1 (0.4)  Antituberculars 1 (0.4)  Benzodiazepines 5 (2.1)  Bronchodilators 8 (3.3)  Calcium antagonists 1 (0.4)  Cardiac drugs 4 (1.7)  Corticosteroids 34 (14.0)  Diagnostic agents 4 (1.7)  Electrolytes 1 (0.4)  Hypotensive agents 2 (0.8)  Immunosuppressive agents 7 (2.9)  Local anesthetics 3 (1.2)  Opioid antagonists 1 (0.4)  Opioids 28 (11.6)  Parathyroid hormones 1 (0.4)  Sedatives and hypnotics 1 (0.4)  Skeletal muscle relaxants 1 (0.4)  Sympathomimetic agents 1 (0.4)  Vasodilating agents 3 (1.2)  Vitamin K 1 (0.4) Therapeutic Classification No. (%) Cases in Which Implicated  Antianemia drugs 2 (0.8)  Antibiotics 22 (9.1)  Anticoagulants 6 (2.5)  Anticonvulsants 4 (1.7)  Antidepressants 1 (0.4)  Antidiabetic agents 7 (2.9)  Antidiarrhea agents 1 (0.4)  Antiemetics 3 (1.2)  Antihistamines 1 (0.4)  Antiinflammatory agents 1 (0.4)  Antilipemic agents 1 (0.4)  Antimanic agents 3 (1.2)  Antineoplastic agents 82 (33.9)  Antiprotozoals 1 (0.4)  Antituberculars 1 (0.4)  Benzodiazepines 5 (2.1)  Bronchodilators 8 (3.3)  Calcium antagonists 1 (0.4)  Cardiac drugs 4 (1.7)  Corticosteroids 34 (14.0)  Diagnostic agents 4 (1.7)  Electrolytes 1 (0.4)  Hypotensive agents 2 (0.8)  Immunosuppressive agents 7 (2.9)  Local anesthetics 3 (1.2)  Opioid antagonists 1 (0.4)  Opioids 28 (11.6)  Parathyroid hormones 1 (0.4)  Sedatives and hypnotics 1 (0.4)  Skeletal muscle relaxants 1 (0.4)  Sympathomimetic agents 1 (0.4)  Vasodilating agents 3 (1.2)  Vitamin K 1 (0.4) Open in new tab Values for total cost of hospitalization and LOS differed significantly between CPHs and controls when the entire cohorts were compared (Table 3). In the secondary analyses, the only significant differences detected between CPHs and controls were in the total cost of hospitalization and LOS for patients younger than 65 years. Despite equal median LOS (5.0 days) in the CPHs and controls, the independent 2-sample Mann–Whitney U test detected statistically significant differences in the spread of the data, as depicted by interquartile range values. Table 3 Comparison of Outcomes Between Patients Who Did (Case Group) and Did Not (Control Group) Have a Medication Error Resulting in Harma Subgroup Analyzed Case Group Control Group pb All Patients  n  242  3,279  Median total cost of hospitalization ($, IQR) 19,444 (13,481–40,580) 17,173 (12,500–27,125) 0.044  Median LOS (days, IQR)  5.0 (5.0–11.0)  5.0 (4.0–7.0) 0.005 Patients Receiving Antineoplastics  n  82  1,611  Median total cost of hospitalization ($, IQR) 18,616 (14,896–29,546) 17,827 (17,021–27,703) 0.536  Median LOS (days, IQR)  5.0 (5.0–8.8)  5.0 (4.9–7.7) 0.893 Patients Receiving Corticosteroids  n  34  306  Median total cost of hospitalization ($, IQR) 16,822 (10,967–38,621) 14,244 (11,207–19,008) 0.339  Median LOS (days, IQR)  5.0 (4.0–7.8)  4.6 (3.7–5.3) 0.235 Patients Receiving Opiates  n  28  381  Median total cost of hospitalization ($, IQR) 23,415 (14,283–56,356) 18,004 (12,370–28,787) 0.251  Median LOS (days, IQR)  7.0 (4.8–16.0)  5.2 (4.2–7.2) 0.251 Patient Age ≥ 65 yr  n  63  842  Median total cost of hospitalization ($, IQR) 16,485 (13,575–24,664) 17,692 (13,382–19,572) 0.878  Median LOS (days, IQR)  5.0 (4.0–7.5)  5.0 (4.2–5.9) 0.806 Patient age < 65 yr  n  179  2,423  Median total cost of hospitalization ($, IQR) 20,095 (13,327–43,994) 17,101 (12,050–27,603) 0.027  Median LOS (days, IQR)  5.0 (5.0–12.0)  5.0 (4.0–7.4) 0.002 Subgroup Analyzed Case Group Control Group pb All Patients  n  242  3,279  Median total cost of hospitalization ($, IQR) 19,444 (13,481–40,580) 17,173 (12,500–27,125) 0.044  Median LOS (days, IQR)  5.0 (5.0–11.0)  5.0 (4.0–7.0) 0.005 Patients Receiving Antineoplastics  n  82  1,611  Median total cost of hospitalization ($, IQR) 18,616 (14,896–29,546) 17,827 (17,021–27,703) 0.536  Median LOS (days, IQR)  5.0 (5.0–8.8)  5.0 (4.9–7.7) 0.893 Patients Receiving Corticosteroids  n  34  306  Median total cost of hospitalization ($, IQR) 16,822 (10,967–38,621) 14,244 (11,207–19,008) 0.339  Median LOS (days, IQR)  5.0 (4.0–7.8)  4.6 (3.7–5.3) 0.235 Patients Receiving Opiates  n  28  381  Median total cost of hospitalization ($, IQR) 23,415 (14,283–56,356) 18,004 (12,370–28,787) 0.251  Median LOS (days, IQR)  7.0 (4.8–16.0)  5.2 (4.2–7.2) 0.251 Patient Age ≥ 65 yr  n  63  842  Median total cost of hospitalization ($, IQR) 16,485 (13,575–24,664) 17,692 (13,382–19,572) 0.878  Median LOS (days, IQR)  5.0 (4.0–7.5)  5.0 (4.2–5.9) 0.806 Patient age < 65 yr  n  179  2,423  Median total cost of hospitalization ($, IQR) 20,095 (13,327–43,994) 17,101 (12,050–27,603) 0.027  Median LOS (days, IQR)  5.0 (5.0–12.0)  5.0 (4.0–7.4) 0.002 a IQR = interquartile range, LOS = length of stay. b Mann–Whitney U test. Open in new tab Table 3 Comparison of Outcomes Between Patients Who Did (Case Group) and Did Not (Control Group) Have a Medication Error Resulting in Harma Subgroup Analyzed Case Group Control Group pb All Patients  n  242  3,279  Median total cost of hospitalization ($, IQR) 19,444 (13,481–40,580) 17,173 (12,500–27,125) 0.044  Median LOS (days, IQR)  5.0 (5.0–11.0)  5.0 (4.0–7.0) 0.005 Patients Receiving Antineoplastics  n  82  1,611  Median total cost of hospitalization ($, IQR) 18,616 (14,896–29,546) 17,827 (17,021–27,703) 0.536  Median LOS (days, IQR)  5.0 (5.0–8.8)  5.0 (4.9–7.7) 0.893 Patients Receiving Corticosteroids  n  34  306  Median total cost of hospitalization ($, IQR) 16,822 (10,967–38,621) 14,244 (11,207–19,008) 0.339  Median LOS (days, IQR)  5.0 (4.0–7.8)  4.6 (3.7–5.3) 0.235 Patients Receiving Opiates  n  28  381  Median total cost of hospitalization ($, IQR) 23,415 (14,283–56,356) 18,004 (12,370–28,787) 0.251  Median LOS (days, IQR)  7.0 (4.8–16.0)  5.2 (4.2–7.2) 0.251 Patient Age ≥ 65 yr  n  63  842  Median total cost of hospitalization ($, IQR) 16,485 (13,575–24,664) 17,692 (13,382–19,572) 0.878  Median LOS (days, IQR)  5.0 (4.0–7.5)  5.0 (4.2–5.9) 0.806 Patient age < 65 yr  n  179  2,423  Median total cost of hospitalization ($, IQR) 20,095 (13,327–43,994) 17,101 (12,050–27,603) 0.027  Median LOS (days, IQR)  5.0 (5.0–12.0)  5.0 (4.0–7.4) 0.002 Subgroup Analyzed Case Group Control Group pb All Patients  n  242  3,279  Median total cost of hospitalization ($, IQR) 19,444 (13,481–40,580) 17,173 (12,500–27,125) 0.044  Median LOS (days, IQR)  5.0 (5.0–11.0)  5.0 (4.0–7.0) 0.005 Patients Receiving Antineoplastics  n  82  1,611  Median total cost of hospitalization ($, IQR) 18,616 (14,896–29,546) 17,827 (17,021–27,703) 0.536  Median LOS (days, IQR)  5.0 (5.0–8.8)  5.0 (4.9–7.7) 0.893 Patients Receiving Corticosteroids  n  34  306  Median total cost of hospitalization ($, IQR) 16,822 (10,967–38,621) 14,244 (11,207–19,008) 0.339  Median LOS (days, IQR)  5.0 (4.0–7.8)  4.6 (3.7–5.3) 0.235 Patients Receiving Opiates  n  28  381  Median total cost of hospitalization ($, IQR) 23,415 (14,283–56,356) 18,004 (12,370–28,787) 0.251  Median LOS (days, IQR)  7.0 (4.8–16.0)  5.2 (4.2–7.2) 0.251 Patient Age ≥ 65 yr  n  63  842  Median total cost of hospitalization ($, IQR) 16,485 (13,575–24,664) 17,692 (13,382–19,572) 0.878  Median LOS (days, IQR)  5.0 (4.0–7.5)  5.0 (4.2–5.9) 0.806 Patient age < 65 yr  n  179  2,423  Median total cost of hospitalization ($, IQR) 20,095 (13,327–43,994) 17,101 (12,050–27,603) 0.027  Median LOS (days, IQR)  5.0 (5.0–12.0)  5.0 (4.0–7.4) 0.002 a IQR = interquartile range, LOS = length of stay. b Mann–Whitney U test. Open in new tab Discussion The study demonstrated significant differences in both total cost of hospitalization and LOS between patients who experienced a medication error resulting in harm during hospitalization and those who did not. Although these results are congruent with data from previous studies using similar methodologies, the use of hospital chargemaster data for the period 2014–15 in the study methodology provided increased external validity for the benefit of modern-day researchers or healthcare administrators evaluating the economic impact of medication errors. We found that the drug classes most frequently associated with medication errors were similar to those previously reported in the literature. Suh et al.5 reported that antiinfective, cardiovascular, antineoplastic, and analgesic/antiinflammatory agents were the leading causes of adverse drug reactions. More recently, Sakuma et al.11 cited antibiotics, sedatives, laxatives, and antihypertensives as the agents most commonly associated with medication errors. The finding in our study that antineoplastic agents, corticosteroids, and opioids were most commonly associated with ADEs was not surprising given the wide scope of cancer specialty services at University of Chicago Medicine. Similar to the results of Suh et al.,5 our findings indicated that the total cost of hospitalization was significantly lower and LOS was significantly shorter among patients younger than 65 years of age when those who were harmed by a medication error were compared with those who were not harmed; we did not detect a significant difference in the same comparison among patients who were at least 65 years old. Suh et al. suggested that in their study population such differences may have been related to the increased likelihood of elderly patients being treated for a larger number of and more complex medical problems. Other studies have shown that elderly patients are more likely to experience medication errors in the community setting, potentially causing hospital admission, but the frequency of medication errors in elderly inpatients is less clear.12,13 Evaluating the preventability of medication errors was not within the scope of our research. In a previous study by Bates et al.,14 medication errors classified as preventable were found to be associated with a higher cost than events classified as nonpreventable, while a study by Hug et al.15 found no such difference. Although we did not classify preventability, the results can still provide a global impetus to continue exploring ways to prevent medication errors. While we studied only the impacts of medication errors on cost and LOS, there may be additional benefits to mitigation of medication errors. By decreasing morbidity and LOS, health systems may see increases in patient satisfaction scores. There may also be improvement in the culture of safety and willingness of staff to report errors as we gain a better understanding of adverse events. This study had several limitations. Two study investigators were responsible for verification and categorization of medication errors using the NCC-MERP index, and no assessment of interrater reliability was performed. Interpretation of exclusion criteria was limited to medication errors documented in the voluntary reporting system or identified using ICD-9-CM coding. It is possible that undocumented or unidentified medication errors occurred in some CPHs or control hospitalizations. Our criteria for matching were based on admission diagnosis, age, and sex. However, additional variables may influence the total cost of hospitalization and LOS. With respect to the latter, the literature suggests that hospital discharge may be delayed by the unavailability of test results or complications regarding patient placement or insurance status.16–18 We did not control for severity of illness, a variable that must be considered given that expenditures and LOS may increase for patients who are more ill. Conclusion Chargemaster data for an academic medical center revealed that the median total hospitalization cost and LOS were significantly greater for hospitalizations during which a harm-causing medication error was recorded versus hospitalizations during which harm-causing medication errors were not recorded. Disclosures The authors have declared no potential conflicts of interest. Previous affiliations At time of writing, Dr. Tuiskula and Dr. Driscoll were affiliated with University of Chicago Medicine, Chicago, IL. Appendix ICD-9-CM codes indicative of an adverse drug eventa E930.0, E930.1, E930.2, E930.3, E930.4, E930.5, E930.6, E930.7, E930.8, E930.9, E931.0, E931.1, E931.2, E931.3, E931.4, E931.5, E931.6, E931.7, E931.8, E931.9, E932.0, E932.1, E932.2, E932.3, E932.4, E932.5, E932.6, E932.7, E932.8, E932.9, E933.0, E933.1, E933.2, E933.3, E933.4, E933.5, E933.6, E933.7, E933.8, E933.9, E934.0, E934.1, E934.2, E934.3, E934.4, E934.5, E934.6, E934.7, E934.8, E934.9, E935.0, E935.1, E935.2, E935.3, E935.4, E935.5, E935.6, E935.7, E935.8, E935.9, E936.0, E936.1, E936.2, E936.3, E936.4, E937.0, E937.1, E937.2, E937.3, E937.4, E937.5, E937.6, E937.8, E937.9, E938.0, E938.1, E938.2, E938.3, E938.4, E938.5, E938.6, E938.7, E938.9, E939.0, E939.1, E939.2, E939.3, E939.4, E939.5, E939.6, E939.7, E939.8, E939.9, E940.0, E940.1, E940.8, E940.9, E941.0, E941.1, E941.2, E941.3, E941.9, E942.0, E942.1, E942.2, E942.3, E942.4, E942.5, E942.6, E942.7, E942.8, E942.9, E943.0, E943.1, E943.2, E943.3, E943.4, E943.5, E943.6, E943.8, E943.9, E944.0, E944.1, E944.2, E944.3, E944.4, E944.5, E944.6, E944.7, E945.0, E945.1, E945.2, E945.3, E945.4, E945.5, E945.6, E945.7, E945.8, E946.0, E946.1, E946.2, E946.3, E946.4, E946.5, E946.6, E946.7, E946.8, E946.9, E947.0, E947.1, E947.2, E947.3, E947.4, E947.8, E947.9, E948.0, E948.1, E948.2, E948.3, E948.4, E948.5, E948.6, E948.8, E948.9, E949.0, E949.1, E949.2, E949.3, E949.4, E949.5, E949.6, E949.7, E949.9 E930.0, E930.1, E930.2, E930.3, E930.4, E930.5, E930.6, E930.7, E930.8, E930.9, E931.0, E931.1, E931.2, E931.3, E931.4, E931.5, E931.6, E931.7, E931.8, E931.9, E932.0, E932.1, E932.2, E932.3, E932.4, E932.5, E932.6, E932.7, E932.8, E932.9, E933.0, E933.1, E933.2, E933.3, E933.4, E933.5, E933.6, E933.7, E933.8, E933.9, E934.0, E934.1, E934.2, E934.3, E934.4, E934.5, E934.6, E934.7, E934.8, E934.9, E935.0, E935.1, E935.2, E935.3, E935.4, E935.5, E935.6, E935.7, E935.8, E935.9, E936.0, E936.1, E936.2, E936.3, E936.4, E937.0, E937.1, E937.2, E937.3, E937.4, E937.5, E937.6, E937.8, E937.9, E938.0, E938.1, E938.2, E938.3, E938.4, E938.5, E938.6, E938.7, E938.9, E939.0, E939.1, E939.2, E939.3, E939.4, E939.5, E939.6, E939.7, E939.8, E939.9, E940.0, E940.1, E940.8, E940.9, E941.0, E941.1, E941.2, E941.3, E941.9, E942.0, E942.1, E942.2, E942.3, E942.4, E942.5, E942.6, E942.7, E942.8, E942.9, E943.0, E943.1, E943.2, E943.3, E943.4, E943.5, E943.6, E943.8, E943.9, E944.0, E944.1, E944.2, E944.3, E944.4, E944.5, E944.6, E944.7, E945.0, E945.1, E945.2, E945.3, E945.4, E945.5, E945.6, E945.7, E945.8, E946.0, E946.1, E946.2, E946.3, E946.4, E946.5, E946.6, E946.7, E946.8, E946.9, E947.0, E947.1, E947.2, E947.3, E947.4, E947.8, E947.9, E948.0, E948.1, E948.2, E948.3, E948.4, E948.5, E948.6, E948.8, E948.9, E949.0, E949.1, E949.2, E949.3, E949.4, E949.5, E949.6, E949.7, E949.9 Open in new tab E930.0, E930.1, E930.2, E930.3, E930.4, E930.5, E930.6, E930.7, E930.8, E930.9, E931.0, E931.1, E931.2, E931.3, E931.4, E931.5, E931.6, E931.7, E931.8, E931.9, E932.0, E932.1, E932.2, E932.3, E932.4, E932.5, E932.6, E932.7, E932.8, E932.9, E933.0, E933.1, E933.2, E933.3, E933.4, E933.5, E933.6, E933.7, E933.8, E933.9, E934.0, E934.1, E934.2, E934.3, E934.4, E934.5, E934.6, E934.7, E934.8, E934.9, E935.0, E935.1, E935.2, E935.3, E935.4, E935.5, E935.6, E935.7, E935.8, E935.9, E936.0, E936.1, E936.2, E936.3, E936.4, E937.0, E937.1, E937.2, E937.3, E937.4, E937.5, E937.6, E937.8, E937.9, E938.0, E938.1, E938.2, E938.3, E938.4, E938.5, E938.6, E938.7, E938.9, E939.0, E939.1, E939.2, E939.3, E939.4, E939.5, E939.6, E939.7, E939.8, E939.9, E940.0, E940.1, E940.8, E940.9, E941.0, E941.1, E941.2, E941.3, E941.9, E942.0, E942.1, E942.2, E942.3, E942.4, E942.5, E942.6, E942.7, E942.8, E942.9, E943.0, E943.1, E943.2, E943.3, E943.4, E943.5, E943.6, E943.8, E943.9, E944.0, E944.1, E944.2, E944.3, E944.4, E944.5, E944.6, E944.7, E945.0, E945.1, E945.2, E945.3, E945.4, E945.5, E945.6, E945.7, E945.8, E946.0, E946.1, E946.2, E946.3, E946.4, E946.5, E946.6, E946.7, E946.8, E946.9, E947.0, E947.1, E947.2, E947.3, E947.4, E947.8, E947.9, E948.0, E948.1, E948.2, E948.3, E948.4, E948.5, E948.6, E948.8, E948.9, E949.0, E949.1, E949.2, E949.3, E949.4, E949.5, E949.6, E949.7, E949.9 E930.0, E930.1, E930.2, E930.3, E930.4, E930.5, E930.6, E930.7, E930.8, E930.9, E931.0, E931.1, E931.2, E931.3, E931.4, E931.5, E931.6, E931.7, E931.8, E931.9, E932.0, E932.1, E932.2, E932.3, E932.4, E932.5, E932.6, E932.7, E932.8, E932.9, E933.0, E933.1, E933.2, E933.3, E933.4, E933.5, E933.6, E933.7, E933.8, E933.9, E934.0, E934.1, E934.2, E934.3, E934.4, E934.5, E934.6, E934.7, E934.8, E934.9, E935.0, E935.1, E935.2, E935.3, E935.4, E935.5, E935.6, E935.7, E935.8, E935.9, E936.0, E936.1, E936.2, E936.3, E936.4, E937.0, E937.1, E937.2, E937.3, E937.4, E937.5, E937.6, E937.8, E937.9, E938.0, E938.1, E938.2, E938.3, E938.4, E938.5, E938.6, E938.7, E938.9, E939.0, E939.1, E939.2, E939.3, E939.4, E939.5, E939.6, E939.7, E939.8, E939.9, E940.0, E940.1, E940.8, E940.9, E941.0, E941.1, E941.2, E941.3, E941.9, E942.0, E942.1, E942.2, E942.3, E942.4, E942.5, E942.6, E942.7, E942.8, E942.9, E943.0, E943.1, E943.2, E943.3, E943.4, E943.5, E943.6, E943.8, E943.9, E944.0, E944.1, E944.2, E944.3, E944.4, E944.5, E944.6, E944.7, E945.0, E945.1, E945.2, E945.3, E945.4, E945.5, E945.6, E945.7, E945.8, E946.0, E946.1, E946.2, E946.3, E946.4, E946.5, E946.6, E946.7, E946.8, E946.9, E947.0, E947.1, E947.2, E947.3, E947.4, E947.8, E947.9, E948.0, E948.1, E948.2, E948.3, E948.4, E948.5, E948.6, E948.8, E948.9, E949.0, E949.1, E949.2, E949.3, E949.4, E949.5, E949.6, E949.7, E949.9 Open in new tab Footnotes a ICD-9-CM = International Classification of Diseases, 9th Revision, Clinical Modification. References 1 Kohn LT Corrigan JM Donaldson MS , eds. To err is human: building a safer health system . Washington, DC : National Academies Press ; 1999 . Google Preview WorldCat COPAC 2 Stelfox HT Palmisani S Scurlock C . The “To Err is Human” report and the patient safety literature . Qual Saf Health Care . 2006 ; 15 : 174 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Nebeker JR Barach P Samore MH . Clarifying adverse drug events: a clinician’s guide to terminology, documentation, and reporting . Ann Intern Med . 2004 ; 140 : 795 – 801 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Office of Disease Prevention and Health Promotion, Department of Health and Human Services . National action plan for adverse drug event prevention ( 2014 ). https://health.gov/hcq/pdfs/ADE-Action-Plan-508c.pdf (accessed 2017 Aug 29). 5 Suh DC Woodall BS Shin S Hermes-De Santis ER . Clinical and economic impact of adverse drug events in hospitalized patients . Ann Pharmacother . 2000 ; 34 : 1373 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Classen DC Pestotnik SL Evans RS . Adverse drug events in hospitalized patients . JAMA . 1997 ; 277 : 301 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Bates DW Spell N Cullen DJ . The costs of adverse drug events in hospitalized patients . JAMA . 1997 ; 277 : 307 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Hamblin S Rumbaugh K Miller R . Prevention of adverse drug events and cost savings associated with PharmD interventions in an academic level 1 trauma center: an evidence-based approach . J Trauma Acute Care Surg . 2012 ; 73 : 1484 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Gardner KN Diefenderfer LA Nelson LA . Cost-avoidance and qualitative analysis of clinical pharmacy interventions by psychiatric pharmacy residents at state psychiatric facilities . Am J Health-Syst Pharm . 2016 ; 73 : e46 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat 10 National Coordinating Council for Medication Error Reporting and Prevention . Types of medication errors . www.nccmerp.org/types-medication-errors (accessed 2016 Dec 20). 11 Sakuma M Kanemoto Y Furuse A . Frequency and severity of adverse drug events by medication classes: the JADE Study . J Patient Saf . Epub ahead of print . 2015 Aug 13 . WorldCat 12 Routledge PA O’Mahony MS Wood-house KW . Adverse drug reactions in elderly patients . Br J Clin Pharmacol . 2004 ; 57 : 121 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Kongkaew C Noyce PR Ashcroft DM . Hospital admissions associated with adverse reactions: a systematic review of prospective observational studies . Ann Pharmacother . 2008 ; 42 : 1017 – 25 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Bates DW Cullen DJ Laird N , for the ADE Prevention Study Group . Incidence of adverse drug events and potential adverse drug events. Implications for prevention . JAMA . 1995 ; 274 : 29 – 34 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Hug BL Keohane C Seger DL . The cost of adverse drug events in community hospitals . Jt Comm J Qual Patient Saf . 2012 ; 38 ( 3 ): 120 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Silva SA Valácio RA Botelho FC Amaral CF . Reasons for discharge delays in teaching hospitals . Rev Saude Publica . 2014 ; 48 : 314 – 21 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Costa AP Poss JW Peirce T Hirdes JP . Acute care inpatients with long-term delayed-discharge: evidence from a Canadian health region . BMC Health Serv Res . 2012 ; 12 : 172 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Brasel KJ Rasmussen J Cauley C Weigelt JA . Reasons for delayed discharge of trauma patients . J Surg Res . 2002 ; 107 : 223 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Copyright © 2017 by the American Society of Health-System Pharmacists, Inc. All rights reserved.