PRACTICE REPORTS Practice model change
PRACTICE REPORTS
ar Layar
Effects of a hospitalwide pharmacy practice model change on readmission and return to emergency department rates Sammuel V. Anderegg, Samaneh T. Wilkinson, Rick J. Couldry, Dennis W. Grauer, and Eric Howser
Scan to access AJHP Voices
H
An audio interview that supplements the information in this article is available on AJHP’s website at www.ajhp.org/ site/misc/podcasts.xhtml.
ospital readmissions are a significant clinical and economic problem in the United States. Nearly one in every five hospital discharges results in readmission.1 Historically, the American healthcare system has been driven by a prospective payment model that encourages minimizing patient length of stay and maximizing turnover.2 In October 2012, the Centers for Medicare and Medicaid Services (CMS) began reducing payments to hospitals reporting excess readmissions in high-risk patient populations.3 Recent studies have shown that 23.4% of readmissions in elderly patients and 4.5% of all-cause readmissions are associated with adverse drug reactions
Purpose. The impact of an innovative medication reconciliation and discharge education program on 30-day readmissions and emergency department (ED) visits was evaluated. Methods. An observational pre–post analysis was conducted at an academic medical center to compare rates of hospital readmissions and return to ED visits during three-month periods before and after implementation of a restructured pharmacy practice model including (1) medication reconciliation at transitions of care for every patient and discharge education for a high-risk subgroup, (2) new or expanded services in the preanesthesia testing clinic and ED, (3) a medication reconciliation technician team, and (4) pharmacist-topatient ratios of 1:30 on acute care floors and 1:18 on critical care units. The primary outcome was the composite of rates of readmissions and return to ED visits within 30 days of discharge. Results. A total of 3,316 patients were included in the study. Pharmacy teams
(ADRs).4,5 Pharmacists are uniquely positioned to prevent postdischarge
Sammuel V. Anderegg, Pharm.D., M.S., BCPS, is Pharmacy Manager, Oncology Service Line, Georgia Regents Medical Center, Augusta, GA; at the time of writing, he was Postgraduate Year 2 Health-System Pharmacy Administration Resident, The University of Kansas Hospital (UKH), Kansas City, and University of Kansas School of Pharmacy, Lawrence. Samaneh T. Wilkinson, Pharm.D., M.S., is Assistant Director of Pharmacy, Clinical Services; and Rick J. Couldry, B.S.Pharm., M.S., FASHP, is Director of Pharmacy, UKH. Dennis W. Grauer,
completed medication reconciliation in 95.8% of cases at admission and 69.7% of cases at discharge. Discharge education was provided to 73.5% of high-risk patients (defined as those receiving anticoagulation therapy or treatment for acute myocardial infarction, chronic obstructive pulmonary disease, congestive heart failure, or pneumonia). No significant difference was observed between the preimplementation and postimplementation groups with regard to the primary outcome. In the high-risk subgroup, there was a significant reduction in the 30-day rate of hospital readmissions, which declined from 17.8% to 12.3% (p = 0.042); cost projections indicated that this reduction in readmissions could yield annual direct cost savings of more than $780,000. Conclusion. Implementation of a teambased pharmacy practice model resulted in a significant decrease in the rate of 30-day readmissions for high-risk patients. Am J Health-Syst Pharm. 2014; 71:146979
ADRs and promote appropriate use of medication therapy by reconcil-
M.S., Ph.D., is Associate Professor and Graduate Program Director, University of Kansas School of Pharmacy. Eric Howser is Decision Support Systems Specialist, UKH. Address correspondence to Dr. Anderegg ([email protected]). The authors have declared no potential conflicts of interest. Copyright © 2014, American Society of Health-System Pharmacists, Inc. All rights reserved. 1079-2082/14/0901-1469$06.00. DOI 10.2146/ajhp130686
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
1469
PRACTICE REPORTS Practice model change
ing medication lists and providing patient education.6,7 Recent studies demonstrate conflicting evidence regarding pharmacists’ ability to influence outcomes and reduce readmissions.8,9 Previously evaluated models have focused mainly on specific diseases, patient populations, and inpatient medical services or care units. There remains a gap in information to evaluate the overall effect of comprehensive pharmacy-led transition-of-care programs. The purpose of the study described here was to evaluate the impact of a hospitalwide medication reconciliation program and discharge education for high-risk patients on 30-day readmission and return to emergency department (ED) rates at an academic medical center. Methods This study was structured around an expansion of pharmacy services that took place in October 2012 at the University of Kansas Hospital (UKH) in Kansas City. The UKH pharmacy practice model is an integrated, teambased, patient-centered care model. Pharmacists are divided into five core clinical divisions: internal medicine, cardiology, pediatrics, hematology– oncology, and critical care. Pharmacists practice in at least two specialties within each core division (e.g., neonatal critical care and pediatric critical care, pediatric critical care and general pediatrics), and at least two pharmacists practice in each specialty. A clinical coordinator who is responsible for administrative, care coordination, and clinical advancement duties supervises each division. Pharmacy teams consist of attending pharmacists,10 pharmacy residents, and pharmacy students on clerkship rotations. Thirteen teams serve medical rounding teams and additional off-service patients on weekdays and provide daily comprehensive medication reviews; therapy interventions based on identified 1470
drug-related problems; vancomycin, aminoglycoside, warfarin, phenytoin, and additional medication dosing management per pharmacyto-dose protocols; and medication reconciliation as needed. On evenings, anywhere from two (pediatrics and cardiology) to six (internal medicine) pharmacy teams are replaced by one pharmacist covering each division, who is responsible for the same clinical responsibilities pharmacy teams have on weekdays. On weekends, one to three pharmacists cover each clinical division based on average patient volume. Attending pharmacists and residents rotate to staff the three core operational shifts in the central pharmacy, which support the pharmacy teams seven days per week. Attending pharmacists rotate through these shifts at least once per month to stay competent in central pharmacy operations. Pharmacy residents work with pharmacy teams during the week, staff clinical shifts on evenings and weekends, and may rotate through operational shifts depending on departmental needs. Operational shifts focus on order verification, i.v. room throughput, phone calls, total parenteral nutrition, and chemotherapy production. The distribution model is primarily centralized, with medication dispensing cabinets on each patient care unit stocked with emergent and as-needed medications. The transformation of the UKH pharmacy practice model was focused on accountability. Pharmacy teams were charged with completing 100% of medication reconciliations at admission, level-of-care transfers, and discharges, as well as discharge education for a high-risk patient population (Figure 1). Additional resources were allocated to the pharmacy department to make the practice model change possible. Pharmacists were added to inpatient rounding teams (increasing the total number of pharmacy teams to 20)
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
and the preanesthesia testing clinic (PTC). Pharmacy services were reestablished in the ED (these services had been discontinued in 2010), and a six-member pharmacy technician team focused on medication reconciliation was created. ED and PTC pharmacy teams were strategically placed to complete medication reconciliations before the admission of patients to the hospital. At UKH, around one third of patients are admitted to inpatient services through the ED, and another one third are postsurgical patients who are evaluated before the scheduled procedure in the PTC. To facilitate the practice model change, inpatient pharmacy services were restructured to accomplish ratios of attending pharmacists to patients of 1:30 for acute care services and 1:18 for critical care services. Medication reconciliation technicians were hired to work unit-based shifts and cover the ED and areas of high patient turnover where inpatient pharmacy teams would need additional help to fulfill medication reconciliation responsibilities. This study was a quasiexperimental, observational, pre–post study of patients discharged from UKH between July and December 2012; eligible patients were those 18 years of age or older and discharged from internal medicine, family medicine, cardiology, or orthopedic surgery medical services during the study period. Patients were excluded from the analysis if they were admitted for a mental illness diagnosis or alcohol or drug use; died during their first admission; were discharged to a rehabilitation unit, skilled nursing facility, long-term care facility, or hospice; or were readmitted for chemotherapy, radiation therapy, rehabilitation therapy, or dialysis. These groups were excluded because (1) it was thought that the effect of discharge counseling may not be adequately captured by the study methodology and (2) the diagnoses
PRACTICE REPORTS Practice model change
and reasons for readmission listed above are commonly associated with planned readmissions. Patients discharged between July and September 2012 who met the inclusion criteria were considered the preimplementation, or control, group. The practice model change was implemented on October 1, 2012, after extensive planning, preparation, and training. Patients discharged between October and December 2012 who met the inclusion criteria served as the postimplementation, or intervention group. Patients were stratified into
two readmission risk groups (highrisk and non-high-risk) and evaluated for changes in readmission and return to ED rates. The high-risk group included patients who were hospitalized with acute myocardial infarction (AMI), congestive heart failure (CHF), or pneumonia (per CMS core measures criteria); had chronic obstructive pulmonary disease (COPD) as the primary reason for admission; or were discharged on oral anticoagulation (warfarin, dabigatran, or rivaroxaban) therapy. High-risk patients on oral anticoagulation were included
in an additional high-risk category if they met criteria for that admission. The non-high-risk group included all other patients meeting the inclusion criteria who were not considered to be at high risk. Pharmacy team members manually identified high-risk patients during profile review and documented the need for discharge education in the pharmacy notes area of the electronic medical record (EMR) used for “pass-off ” communication between pharmacists. In the postimplementation group, admission medication his-
Figure 1. Medication reconciliation (Med Rec) and patient education activities before and after implementation of a pharmacy practice model aimed at reducing 30-day rates of readmissions and return to ED visits at University of Kansas Hospital.
Preimplementation All Patients
Postimplementation NonHigh-Risk Patients
+ Med Rec
High-Risk Patients
Admission Med Rec
Admission
Admission
Discharge Med Rec
Discharge Med Rec
Patient Education
Medication Calendar
Discharge
Discharge
Discharge
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
1471
PRACTICE REPORTS Practice model change
tories were performed by pharmacy technicians, pharmacy students, pharmacy residents, and attending pharmacists. Pharmacy technicians were the primary team members conducting medication reconciliations. EMR tools were used to identify newly admitted patients. Columns added to patient lists displayed the admission date and indicated whether a pharmacy admission history had been performed. Similar indicators were used to quickly identify discharge medication reconciliation status and discharge education status. After each medication history was completed, a pharmacist worked with the medical team to reconcile the admission medication list with medications the patient was receiving as an inpatient. Attending pharmacists and pharmacy residents performed discharge medication reconciliation by “pending” discharge orders, a task that was achieved by leveraging EMR functionality. Pharmacists were granted access to the discharge navigator, which is typically utilized by providers and case managers. Within the navigator, the discharge medication plan can be proactively queued for action prior to order execution (e.g., stress ulcer prophylaxis and i.v. antibiotics can be discontinued and oral antibiotics ordered, home medications can be restarted and refills initiated as pending orders); such pharmacist-pended orders serve as electronic recommendations ready for provider review prior to the signing of discharge orders. As part of the required discharge workflow within the EMR, providers open the discharge navigator and view all pended orders generated by a pharmacist. Providers have the ability to modify or remove recommendations as needed and can add orders when appropriate. Intensive care unit pharmacists also used this functionality to reconcile medications when patients were transferred from critical to acute care services. 1472
Discharge education was provided to high-risk patients by attending pharmacists, pharmacy residents, and pharmacy students under the supervision of a pharmacist. Takehome medication calendars listing active medications, dosages, routes, and frequencies were provided to supplement education sessions and serve as a postdischarge medication reference. These lists were generated using an EMR macro within a pharmacy discharge progress note. After a provider signed the discharge orders, the macro pulled the finalized medication list into the discharge note. Medication calendars were printed and delivered to patients during counseling. Discharge education was provided to non-high-risk patients if requested by the patient, caregiver, or medical team. In some cases, pharmacy teams identified patients who would benefit from education due to health literacy deficiencies. Patients were identified for the study analysis using reports generated from the hospital EMR. Demographics, readmission, and return to ED data were collected before and after implementation of the transitionsof-care program and supplied by the UKH Department of Organizational Improvement. Healthcare utilization data did not include readmissions or return to ED visits outside of the UKH health system. Data points indicating the completion status of medication reconciliation (at admission and discharge), as well as discharge education and medication calendar distribution, were documented in the EMR by pharmacists and compiled using a report designed by the department of pharmacy. The primary outcome was the composite rate of 30-day readmission and/or return to the ED. This composite outcome included patients who experienced a readmission or a return to ED visit, or both, within 30 days of discharge. In addition, each outcome was evaluated
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
separately. Postdischarge healthcare utilization rates at baseline and after program implementation in both high-risk and non-high-risk patients were compared. The composite outcome was further broken down by discharge medical service and highrisk criteria to compare differences in healthcare utilization before and after program implementation. The secondary outcome was health system cost savings associated with differences in readmission and return to ED rates before and after project implementation. Cost savings estimates were calculated using the average direct and total costs of patient readmissions and ED encounters at UKH during the study period, as reflected by data obtained from the University HealthSystem Consortium Clinical Data Base/ Resource Manager (UHC, Chicago, IL). The database calculates direct cost based on the estimated cost for medication, laboratory testing, imaging, and other resource charges from all encounters; in calculating total cost, the database attempts to factor in overhead expenses such as facility costs and labor. Since our study period spanned six months of discharges, the total number of highrisk patients in the preimplementation and postimplementation groups was multiplied by two in order to estimate the total number of high-risk patients discharged in one year. This study was approved by the UKH institutional review board. Analysis Descriptive and bivariate statistics were performed to analyze the study results. Student’s t tests were used for continuous data (the age and length of stay data points and number of medications prescribed), and the chisquare method was used for discrete data. The a priori level of significance was set at 0.05. The study population effectively comprised a census of all patients meeting the inclusion criteria within
PRACTICE REPORTS Practice model change
the defined time period. Assuming that a 5% reduction in the composite outcome could be achieved, with a 5% probability of type I error in the conclusion and the power set at 80%, it was determined that a population of 90 patients per group would be required; the groups compared in the study met this sample size requirement. Results Overall, 1664 and 1652 patients were included in the preimplementation and postimplementation groups, respectively. Evaluation of baseline characteristics indicated statistically significant differences
between the groups in the number of medications prescribed at admission, in discharge service classification, and in patient classification into high-risk categories; these differences were due to the large sample size, and it was determined that they were not clinically relevant to the study. No significant change was observed in the mean length of stay between the preimplementation and postimplementation groups (4.1 days versus 4.2 days, p = 0.477), as shown in Table 1. In the postimplementation group, pharmacy teams completed medication reconciliation in 95.8% of patients at admission and 69.7% of
patients at discharge (Table 2). For high-risk patients, medication reconciliation was completed at higher rates at both admission (98.9%) and discharge (77.9%) relative to the completion rates for non-highrisk patients (95.0% and 67.4%, respectively). A total of 389 patients (23.4%) were readmitted to inpatient services or returned to the ED (or both) within 30 days of discharge in the preimplementation group, compared with 373 (22.6%) in the postimplementation group (p = 0.585) (Table 3); the corresponding 30-day readmission rates were 16.2% and 15.6% (p = 0.632). The 30-day
Table 1.
Baseline Characteristics of Preimplementation and Postimplementation Groupsa Variable Male, no. (%) Mean ± S.D. age, yr Race, no. (%) Caucasian African-American Asian Other, unknown or declined Marital status, no. (%) Married or life partner Single Divorced, separated, or widowed Unknown Discharge service, no. (%) Internal medicine Cardiology Orthopedic surgery Family medicine Mean ± S.D. no. medications prescribed On admission At discharge New at discharge High-risk patients, no. (%) Anticoagulation CHF COPD AMI Pneumonia Mean ± S.D. length of stay, days
Preimplementation (n = 1664)
Postimplementation (n = 1652)
878 (52.8) 54.2 ± 17.1
832 (50.4) 54.2 ± 16.4
0.167b 0.086c
1174 (70.6) 273 (16.4) 21 (1.3) 196 (11.8)
1146 (69.4) 311 (18.8) 13 (0.8) 182 (11.0)
0.160b 0.160b 0.160b 0.160b
799 (48.0) 512 (30.8) 350 (21.0) 3 (0.2)
778 (47.1) 523 (31.7) 341 (20.6) 10 (0.6)
0.237b 0.237b 0.237b 0.237b
831 (49.9) 391 (23.5) 273 (16.4) 169 (10.2)
931 (56.4) 365 (22.1) 231 (14.0) 125 (7.6)
0.001b 0.001b 0.001b 0.001b
11.2 ± 7.8 12.2 ± 7.2 3.4 ± 2.8 325 (19.5) 226 (13.6) 45 (2.7) 31 (1.9) 30 (1.8) 20 (1.2) 4.1 ± 3.9
11.8 ± 8.0 12.4 ± 7.0 3.8 ± 3.1 358 (21.7) 188 (11.4) 51 (3.1) 62 (3.8) 41 (2.5) 41 (2.5) 4.2 ± 4.2
PRACTICE REPORTS
ar Layar
Effects of a hospitalwide pharmacy practice model change on readmission and return to emergency department rates Sammuel V. Anderegg, Samaneh T. Wilkinson, Rick J. Couldry, Dennis W. Grauer, and Eric Howser
Scan to access AJHP Voices
H
An audio interview that supplements the information in this article is available on AJHP’s website at www.ajhp.org/ site/misc/podcasts.xhtml.
ospital readmissions are a significant clinical and economic problem in the United States. Nearly one in every five hospital discharges results in readmission.1 Historically, the American healthcare system has been driven by a prospective payment model that encourages minimizing patient length of stay and maximizing turnover.2 In October 2012, the Centers for Medicare and Medicaid Services (CMS) began reducing payments to hospitals reporting excess readmissions in high-risk patient populations.3 Recent studies have shown that 23.4% of readmissions in elderly patients and 4.5% of all-cause readmissions are associated with adverse drug reactions
Purpose. The impact of an innovative medication reconciliation and discharge education program on 30-day readmissions and emergency department (ED) visits was evaluated. Methods. An observational pre–post analysis was conducted at an academic medical center to compare rates of hospital readmissions and return to ED visits during three-month periods before and after implementation of a restructured pharmacy practice model including (1) medication reconciliation at transitions of care for every patient and discharge education for a high-risk subgroup, (2) new or expanded services in the preanesthesia testing clinic and ED, (3) a medication reconciliation technician team, and (4) pharmacist-topatient ratios of 1:30 on acute care floors and 1:18 on critical care units. The primary outcome was the composite of rates of readmissions and return to ED visits within 30 days of discharge. Results. A total of 3,316 patients were included in the study. Pharmacy teams
(ADRs).4,5 Pharmacists are uniquely positioned to prevent postdischarge
Sammuel V. Anderegg, Pharm.D., M.S., BCPS, is Pharmacy Manager, Oncology Service Line, Georgia Regents Medical Center, Augusta, GA; at the time of writing, he was Postgraduate Year 2 Health-System Pharmacy Administration Resident, The University of Kansas Hospital (UKH), Kansas City, and University of Kansas School of Pharmacy, Lawrence. Samaneh T. Wilkinson, Pharm.D., M.S., is Assistant Director of Pharmacy, Clinical Services; and Rick J. Couldry, B.S.Pharm., M.S., FASHP, is Director of Pharmacy, UKH. Dennis W. Grauer,
completed medication reconciliation in 95.8% of cases at admission and 69.7% of cases at discharge. Discharge education was provided to 73.5% of high-risk patients (defined as those receiving anticoagulation therapy or treatment for acute myocardial infarction, chronic obstructive pulmonary disease, congestive heart failure, or pneumonia). No significant difference was observed between the preimplementation and postimplementation groups with regard to the primary outcome. In the high-risk subgroup, there was a significant reduction in the 30-day rate of hospital readmissions, which declined from 17.8% to 12.3% (p = 0.042); cost projections indicated that this reduction in readmissions could yield annual direct cost savings of more than $780,000. Conclusion. Implementation of a teambased pharmacy practice model resulted in a significant decrease in the rate of 30-day readmissions for high-risk patients. Am J Health-Syst Pharm. 2014; 71:146979
ADRs and promote appropriate use of medication therapy by reconcil-
M.S., Ph.D., is Associate Professor and Graduate Program Director, University of Kansas School of Pharmacy. Eric Howser is Decision Support Systems Specialist, UKH. Address correspondence to Dr. Anderegg ([email protected]). The authors have declared no potential conflicts of interest. Copyright © 2014, American Society of Health-System Pharmacists, Inc. All rights reserved. 1079-2082/14/0901-1469$06.00. DOI 10.2146/ajhp130686
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
1469
PRACTICE REPORTS Practice model change
ing medication lists and providing patient education.6,7 Recent studies demonstrate conflicting evidence regarding pharmacists’ ability to influence outcomes and reduce readmissions.8,9 Previously evaluated models have focused mainly on specific diseases, patient populations, and inpatient medical services or care units. There remains a gap in information to evaluate the overall effect of comprehensive pharmacy-led transition-of-care programs. The purpose of the study described here was to evaluate the impact of a hospitalwide medication reconciliation program and discharge education for high-risk patients on 30-day readmission and return to emergency department (ED) rates at an academic medical center. Methods This study was structured around an expansion of pharmacy services that took place in October 2012 at the University of Kansas Hospital (UKH) in Kansas City. The UKH pharmacy practice model is an integrated, teambased, patient-centered care model. Pharmacists are divided into five core clinical divisions: internal medicine, cardiology, pediatrics, hematology– oncology, and critical care. Pharmacists practice in at least two specialties within each core division (e.g., neonatal critical care and pediatric critical care, pediatric critical care and general pediatrics), and at least two pharmacists practice in each specialty. A clinical coordinator who is responsible for administrative, care coordination, and clinical advancement duties supervises each division. Pharmacy teams consist of attending pharmacists,10 pharmacy residents, and pharmacy students on clerkship rotations. Thirteen teams serve medical rounding teams and additional off-service patients on weekdays and provide daily comprehensive medication reviews; therapy interventions based on identified 1470
drug-related problems; vancomycin, aminoglycoside, warfarin, phenytoin, and additional medication dosing management per pharmacyto-dose protocols; and medication reconciliation as needed. On evenings, anywhere from two (pediatrics and cardiology) to six (internal medicine) pharmacy teams are replaced by one pharmacist covering each division, who is responsible for the same clinical responsibilities pharmacy teams have on weekdays. On weekends, one to three pharmacists cover each clinical division based on average patient volume. Attending pharmacists and residents rotate to staff the three core operational shifts in the central pharmacy, which support the pharmacy teams seven days per week. Attending pharmacists rotate through these shifts at least once per month to stay competent in central pharmacy operations. Pharmacy residents work with pharmacy teams during the week, staff clinical shifts on evenings and weekends, and may rotate through operational shifts depending on departmental needs. Operational shifts focus on order verification, i.v. room throughput, phone calls, total parenteral nutrition, and chemotherapy production. The distribution model is primarily centralized, with medication dispensing cabinets on each patient care unit stocked with emergent and as-needed medications. The transformation of the UKH pharmacy practice model was focused on accountability. Pharmacy teams were charged with completing 100% of medication reconciliations at admission, level-of-care transfers, and discharges, as well as discharge education for a high-risk patient population (Figure 1). Additional resources were allocated to the pharmacy department to make the practice model change possible. Pharmacists were added to inpatient rounding teams (increasing the total number of pharmacy teams to 20)
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
and the preanesthesia testing clinic (PTC). Pharmacy services were reestablished in the ED (these services had been discontinued in 2010), and a six-member pharmacy technician team focused on medication reconciliation was created. ED and PTC pharmacy teams were strategically placed to complete medication reconciliations before the admission of patients to the hospital. At UKH, around one third of patients are admitted to inpatient services through the ED, and another one third are postsurgical patients who are evaluated before the scheduled procedure in the PTC. To facilitate the practice model change, inpatient pharmacy services were restructured to accomplish ratios of attending pharmacists to patients of 1:30 for acute care services and 1:18 for critical care services. Medication reconciliation technicians were hired to work unit-based shifts and cover the ED and areas of high patient turnover where inpatient pharmacy teams would need additional help to fulfill medication reconciliation responsibilities. This study was a quasiexperimental, observational, pre–post study of patients discharged from UKH between July and December 2012; eligible patients were those 18 years of age or older and discharged from internal medicine, family medicine, cardiology, or orthopedic surgery medical services during the study period. Patients were excluded from the analysis if they were admitted for a mental illness diagnosis or alcohol or drug use; died during their first admission; were discharged to a rehabilitation unit, skilled nursing facility, long-term care facility, or hospice; or were readmitted for chemotherapy, radiation therapy, rehabilitation therapy, or dialysis. These groups were excluded because (1) it was thought that the effect of discharge counseling may not be adequately captured by the study methodology and (2) the diagnoses
PRACTICE REPORTS Practice model change
and reasons for readmission listed above are commonly associated with planned readmissions. Patients discharged between July and September 2012 who met the inclusion criteria were considered the preimplementation, or control, group. The practice model change was implemented on October 1, 2012, after extensive planning, preparation, and training. Patients discharged between October and December 2012 who met the inclusion criteria served as the postimplementation, or intervention group. Patients were stratified into
two readmission risk groups (highrisk and non-high-risk) and evaluated for changes in readmission and return to ED rates. The high-risk group included patients who were hospitalized with acute myocardial infarction (AMI), congestive heart failure (CHF), or pneumonia (per CMS core measures criteria); had chronic obstructive pulmonary disease (COPD) as the primary reason for admission; or were discharged on oral anticoagulation (warfarin, dabigatran, or rivaroxaban) therapy. High-risk patients on oral anticoagulation were included
in an additional high-risk category if they met criteria for that admission. The non-high-risk group included all other patients meeting the inclusion criteria who were not considered to be at high risk. Pharmacy team members manually identified high-risk patients during profile review and documented the need for discharge education in the pharmacy notes area of the electronic medical record (EMR) used for “pass-off ” communication between pharmacists. In the postimplementation group, admission medication his-
Figure 1. Medication reconciliation (Med Rec) and patient education activities before and after implementation of a pharmacy practice model aimed at reducing 30-day rates of readmissions and return to ED visits at University of Kansas Hospital.
Preimplementation All Patients
Postimplementation NonHigh-Risk Patients
+ Med Rec
High-Risk Patients
Admission Med Rec
Admission
Admission
Discharge Med Rec
Discharge Med Rec
Patient Education
Medication Calendar
Discharge
Discharge
Discharge
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
1471
PRACTICE REPORTS Practice model change
tories were performed by pharmacy technicians, pharmacy students, pharmacy residents, and attending pharmacists. Pharmacy technicians were the primary team members conducting medication reconciliations. EMR tools were used to identify newly admitted patients. Columns added to patient lists displayed the admission date and indicated whether a pharmacy admission history had been performed. Similar indicators were used to quickly identify discharge medication reconciliation status and discharge education status. After each medication history was completed, a pharmacist worked with the medical team to reconcile the admission medication list with medications the patient was receiving as an inpatient. Attending pharmacists and pharmacy residents performed discharge medication reconciliation by “pending” discharge orders, a task that was achieved by leveraging EMR functionality. Pharmacists were granted access to the discharge navigator, which is typically utilized by providers and case managers. Within the navigator, the discharge medication plan can be proactively queued for action prior to order execution (e.g., stress ulcer prophylaxis and i.v. antibiotics can be discontinued and oral antibiotics ordered, home medications can be restarted and refills initiated as pending orders); such pharmacist-pended orders serve as electronic recommendations ready for provider review prior to the signing of discharge orders. As part of the required discharge workflow within the EMR, providers open the discharge navigator and view all pended orders generated by a pharmacist. Providers have the ability to modify or remove recommendations as needed and can add orders when appropriate. Intensive care unit pharmacists also used this functionality to reconcile medications when patients were transferred from critical to acute care services. 1472
Discharge education was provided to high-risk patients by attending pharmacists, pharmacy residents, and pharmacy students under the supervision of a pharmacist. Takehome medication calendars listing active medications, dosages, routes, and frequencies were provided to supplement education sessions and serve as a postdischarge medication reference. These lists were generated using an EMR macro within a pharmacy discharge progress note. After a provider signed the discharge orders, the macro pulled the finalized medication list into the discharge note. Medication calendars were printed and delivered to patients during counseling. Discharge education was provided to non-high-risk patients if requested by the patient, caregiver, or medical team. In some cases, pharmacy teams identified patients who would benefit from education due to health literacy deficiencies. Patients were identified for the study analysis using reports generated from the hospital EMR. Demographics, readmission, and return to ED data were collected before and after implementation of the transitionsof-care program and supplied by the UKH Department of Organizational Improvement. Healthcare utilization data did not include readmissions or return to ED visits outside of the UKH health system. Data points indicating the completion status of medication reconciliation (at admission and discharge), as well as discharge education and medication calendar distribution, were documented in the EMR by pharmacists and compiled using a report designed by the department of pharmacy. The primary outcome was the composite rate of 30-day readmission and/or return to the ED. This composite outcome included patients who experienced a readmission or a return to ED visit, or both, within 30 days of discharge. In addition, each outcome was evaluated
Am J Health-Syst Pharm—Vol 71 Sep 1, 2014
separately. Postdischarge healthcare utilization rates at baseline and after program implementation in both high-risk and non-high-risk patients were compared. The composite outcome was further broken down by discharge medical service and highrisk criteria to compare differences in healthcare utilization before and after program implementation. The secondary outcome was health system cost savings associated with differences in readmission and return to ED rates before and after project implementation. Cost savings estimates were calculated using the average direct and total costs of patient readmissions and ED encounters at UKH during the study period, as reflected by data obtained from the University HealthSystem Consortium Clinical Data Base/ Resource Manager (UHC, Chicago, IL). The database calculates direct cost based on the estimated cost for medication, laboratory testing, imaging, and other resource charges from all encounters; in calculating total cost, the database attempts to factor in overhead expenses such as facility costs and labor. Since our study period spanned six months of discharges, the total number of highrisk patients in the preimplementation and postimplementation groups was multiplied by two in order to estimate the total number of high-risk patients discharged in one year. This study was approved by the UKH institutional review board. Analysis Descriptive and bivariate statistics were performed to analyze the study results. Student’s t tests were used for continuous data (the age and length of stay data points and number of medications prescribed), and the chisquare method was used for discrete data. The a priori level of significance was set at 0.05. The study population effectively comprised a census of all patients meeting the inclusion criteria within
PRACTICE REPORTS Practice model change
the defined time period. Assuming that a 5% reduction in the composite outcome could be achieved, with a 5% probability of type I error in the conclusion and the power set at 80%, it was determined that a population of 90 patients per group would be required; the groups compared in the study met this sample size requirement. Results Overall, 1664 and 1652 patients were included in the preimplementation and postimplementation groups, respectively. Evaluation of baseline characteristics indicated statistically significant differences
between the groups in the number of medications prescribed at admission, in discharge service classification, and in patient classification into high-risk categories; these differences were due to the large sample size, and it was determined that they were not clinically relevant to the study. No significant change was observed in the mean length of stay between the preimplementation and postimplementation groups (4.1 days versus 4.2 days, p = 0.477), as shown in Table 1. In the postimplementation group, pharmacy teams completed medication reconciliation in 95.8% of patients at admission and 69.7% of
patients at discharge (Table 2). For high-risk patients, medication reconciliation was completed at higher rates at both admission (98.9%) and discharge (77.9%) relative to the completion rates for non-highrisk patients (95.0% and 67.4%, respectively). A total of 389 patients (23.4%) were readmitted to inpatient services or returned to the ED (or both) within 30 days of discharge in the preimplementation group, compared with 373 (22.6%) in the postimplementation group (p = 0.585) (Table 3); the corresponding 30-day readmission rates were 16.2% and 15.6% (p = 0.632). The 30-day
Table 1.
Baseline Characteristics of Preimplementation and Postimplementation Groupsa Variable Male, no. (%) Mean ± S.D. age, yr Race, no. (%) Caucasian African-American Asian Other, unknown or declined Marital status, no. (%) Married or life partner Single Divorced, separated, or widowed Unknown Discharge service, no. (%) Internal medicine Cardiology Orthopedic surgery Family medicine Mean ± S.D. no. medications prescribed On admission At discharge New at discharge High-risk patients, no. (%) Anticoagulation CHF COPD AMI Pneumonia Mean ± S.D. length of stay, days
Preimplementation (n = 1664)
Postimplementation (n = 1652)
878 (52.8) 54.2 ± 17.1
832 (50.4) 54.2 ± 16.4
0.167b 0.086c
1174 (70.6) 273 (16.4) 21 (1.3) 196 (11.8)
1146 (69.4) 311 (18.8) 13 (0.8) 182 (11.0)
0.160b 0.160b 0.160b 0.160b
799 (48.0) 512 (30.8) 350 (21.0) 3 (0.2)
778 (47.1) 523 (31.7) 341 (20.6) 10 (0.6)
0.237b 0.237b 0.237b 0.237b
831 (49.9) 391 (23.5) 273 (16.4) 169 (10.2)
931 (56.4) 365 (22.1) 231 (14.0) 125 (7.6)
0.001b 0.001b 0.001b 0.001b
11.2 ± 7.8 12.2 ± 7.2 3.4 ± 2.8 325 (19.5) 226 (13.6) 45 (2.7) 31 (1.9) 30 (1.8) 20 (1.2) 4.1 ± 3.9
11.8 ± 8.0 12.4 ± 7.0 3.8 ± 3.1 358 (21.7) 188 (11.4) 51 (3.1) 62 (3.8) 41 (2.5) 41 (2.5) 4.2 ± 4.2
