BACKGROUND AND OBJECTIVES:

The pediatric inpatient discharge medication process is complicated, and caregivers have difficulty managing instructions. Authors of few studies evaluate systematic processes for ensuring quality in these care transitions. We aimed to improve caregiver medication management and understanding of discharge medications by standardizing the discharge medication process.

METHODS:

An interprofessional team at an urban, tertiary care children’s hospital trialed interventions to improve caregiver medication management and understanding. These included mnemonics to aid in complete medication counseling, electronic medical record enhancements to standardize medication documentation and simplify dose rounding, and housestaff education. The primary outcome measure was the proportion of discharge medication–related failures in each 4-week period. Failure was defined as an incorrect response on ≥1 survey questions. Statistical process control was used to analyze improvement over time. Process measures related to medication documentation and dose rounding were compared by using the χ2 test and process control.

RESULTS:

Special cause variation occurred in the mean discharge medication–related failure rate, which decreased from 70.1% to 36.1% and was sustained. There were significantly more complete after-visit summaries (21.0% vs 85.1%; P < .001) and more patients with simplified dosing (75.2% vs 95.6%; P < .001) in the intervention period. Special cause variation also occurred for these measures.

CONCLUSIONS:

A systematic approach to standardizing the discharge medication process led to improved caregiver medication management and understanding after pediatric inpatient discharge. These changes could be adapted by other hospitals to enhance the quality of this care transition.

More than 70 000 emergency department (ED) visits occur annually for pediatric medication overdoses, and up to 16% of children experience ambulatory adverse drug events, most often because of caregiver administration errors.1,2  One potential cause for >60 000 children experiencing out-of-hospital medication errors annually3  is poor transitions between inpatient and outpatient care. Because medication administration responsibility shifts from the hospital provider to the primary caregiver at discharge, children are at risk for errors. Effective approaches for these transitions are understudied.4  In general, there are no widely accepted standards to maximize quality of pediatric discharges, especially related to medications.5  However, any transition should be family centered, should be coordinated among all participants, and should use standardized documentation to improve communication and address health literacy.4  Previous research by this group suggests that more than two-thirds of caregivers have a discharge medication–related failure, and most instructions lack key medication information after inpatient discharge.6 

The pediatric discharge medication process is complicated, and caregivers have difficulty managing medication instructions, increasing the risk for errors.7  Recent literature is focused on ED discharges,813  with less focus on inpatient discharges.1416  A multicenter improvement collaborative decreased discharge-care failures using a bundled intervention; however, there was no specific focus on the medication process.14  Additionally, discharge instructions are prone to discrepancies, with 19% to 26% of pediatric charts containing conflicting information.17,18  Furthermore, liquid medications are often prescribed in complex, difficult-to-measure volumes, making administration challenging.19  Our group found that 70% of caregivers mismanage or misunderstand inpatient discharge medications (measured as discharge medication–related failures), particularly related to side effects, duration, and start time after discharge.6  Thus, systematic processes for ensuring accurate medication management and understanding by caregivers and consistent medication documentation are paramount to promoting safe inpatient discharges.20 

This project’s purpose was to improve caregiver medication management and understanding of medications by standardizing the discharge medication process as measured by responses on a telephone questionnaire. We aimed to reduce the proportion of discharge medication–related failures from 70% to 53% (25% relative decrease) in 12 months. We hypothesized that caregiver medication management and understanding would improve with systematic counseling, standardized documentation, and simplified dose rounding. In addition, we analyzed measures related to improvements in discharge medication documentation and dosing and the sustainability of these changes. A key driver diagram is used to describe the framework believed to contribute to the project’s aim (Fig 1).

FIGURE 1

Key driver diagram: key drivers hypothesized to contribute to the project’s specific aim. RN, registered nurse.

FIGURE 1

Key driver diagram: key drivers hypothesized to contribute to the project’s specific aim. RN, registered nurse.

Close modal

This project took place at a 132-bed, urban, tertiary care, academic children’s hospital serving a predominantly low-income racial and ethnic minority population. Standardizing medication counseling and documentation is especially important because one-third of adults in this population do not complete high school and therefore are at high risk for low health literacy.21,22  A 3-month baseline period (January 2017 to March 2017) was followed by a 12-month intervention period (April 2017 to March 2018) and then a 6-month sustainability period (April 2018 to September 2018). A hospital-based outpatient pharmacy opened in September 2016 to perform bedside prescription delivery and medication counseling. The hospital has a strong local quality improvement (QI) culture with history of multiple interprofessional collaboratives and a dedicated QI leadership group who runs a yearly QI course open to all staff. There is an anonymous incident reporting system used to promote a positive patient safety culture.

The study population included caregivers of patients admitted to pediatric hospital medicine teams, including medical patients younger than 13 years and complex care and surgical patients up to 21 years old. Teams are led by an attending physician and run by housestaff (physician assistants and residents) and students. The local electronic medical record (EMR) uses Epic Systems, a widely used health care software application.23  Before interventions, the discharge process included preparing a discharge summary, completing an after-visit summary (AVS), and delivering a discharge talk with medication counseling. The discharge summary is prepared by housestaff, is intended for the primary pediatrician, and contains a complete medication list. The AVS is a standardized document containing automatically imported discharge prescription information, optional free-text instructions, and a dosing calendar. The AVS is completed collaboratively by housestaff and nurses and given to caregivers at discharge. Discharge medication counseling is performed by housestaff, nurses, and/or a pharmacist. It is done at the bedside and/or in the pharmacy 24 hours before discharge. The approach to counseling and the content of additional written instructions was not standardized before this study.

A stakeholder team consisting of residents, physician assistants, nurses, pharmacists, attending physicians, and QI specialists was created in November 2016. The Model for Improvement and plan-do-study-act cycles were used to develop interventions that standardized discharge medication counseling and improved medication documentation.24  These included the MEDRITES acronym (medication name, engage family, dose, route, indications, timing, effects to watch for, storage and syringe), EMR enhancements, and housestaff education.

MEDRITES Acronym

In April 2017 an intervention bundle focused on standardizing discharge medication counseling and improving documentation was rolled out. Stakeholders determined important components of counseling on the basis of previous literature and decided on the outcome measure by group consensus.6  We then created the MEDRITES mnemonic and bundle to provide a teachable framework for key components of medication counseling (Fig 2). Care teams were educated on the mnemonic through information sessions and flyers, and the hospital-based pharmacy adjusted their counseling script to include all MEDRITES elements. After reviewing the medication name and engaging the family (to determine existing medication knowledge), providers discussed dose, route, indications, timing (including frequency, duration, and next dose due after discharge), and expected effects and side effects. Finally, caregivers were educated on proper storage and given a syringe (for liquid medications) for administration. The QI team provided monthly feedback on outcome and process data via e-mail.

FIGURE 2

MEDRITES mnemonic. The MEDRITES mnemonic provided a framework for key components of medication counseling throughout the project.

FIGURE 2

MEDRITES mnemonic. The MEDRITES mnemonic provided a framework for key components of medication counseling throughout the project.

Close modal

EMR Enhancements

EMR components were used to improve documentation and standardization of medication counseling. In our previous work, we found that 80% of AVSs were missing key medication administration information.6  To fix this, “smart phrases” for the most commonly prescribed medications were created, and housestaff added them to the AVS. These smart phrases contained prefilled text and prompts incorporating all MEDRITES elements (Supplemental Fig 5).

Next, because 24% of discharge liquid medication doses were difficult to measure (11.8 vs 12.0 mL),6  we simplified and standardized the EMR’s rounding algorithm. Safe rounding factors were implemented to appropriately dose the most commonly prescribed medications to the nearest 0.1, 0.5, or 1 mL.25 

Housestaff Education and Deliberate Practice

Two months after rollout, survey data revealed that caregivers most frequently lacked knowledge related to medication side effects, duration, and next dose due after discharge. We believed these elements were most commonly omitted from discharge counseling. The new academic year was identified as a critical opportunity to set expectations and provide a medication counseling framework. Therefore, in July 2017, we started deliberate practice, or one-on-one teaching, in which interns practiced medication counseling via role-play and a more experienced clinician provided real-time feedback on how to improve their technique by using MEDRITES.

Lastly, a health literacy lecture was given to housestaff in December 2017, which incorporated a custom-made video on MEDRITES counseling. Learners identified contextual factors for the project’s importance, discussed barriers and facilitators of performing medication counseling, and reviewed data.

Eligible study subjects were English-speaking caregivers of children <18 years old discharged from a pediatric hospital medicine service on at least 1 new oral prescription medication, including a change in dose of a previous medication. A 10-item questionnaire was administered via phone to caregivers 48 to 96 hours after discharge (Table 1). If >1 eligible medication was present, the questionnaire proceeded with the first medication listed by the caregiver. A quasi-random sample of caregivers confirmed to be primarily responsible for medication administration were called weekly. Each week, a random number generator identified a letter of the alphabet. Calls were made for patients whose last names started with that letter, continuing alphabetically until a predetermined number of surveys were completed (see Analysis section) or until the week ended. Calls were made during regular business hours, evening hours, and on weekends, as the primary study investigator’s (PI’s) (K.P.) schedule allowed. The number of calls made weekly in each period is described in the power analysis below.

TABLE 1

Failure Rates by Survey Question in the Baseline Versus Intervention Periods

QuestionCategory of FailureDefinition of Incorrect AnswerFailed in Baseline Period (n = 157), n (%)Failed in Intervention Period (n = 249), n (%)P
Did you go home on any medications? Unaware prescription exists Unaware child was to continue taking medication after discharge 0 (0) 0 (0) NA 
What medication was it? Unaware of name or class Unaware of or wrong medication name or class 3 (1.9) 2 (0.8) .38a 
Did you have any trouble getting it from the pharmacy? Pharmacy trouble Did not pick up medication or delay in obtaining medication 10 (6.4) 9 (3.6) .20 
When did you start the medication? Wrong start time Missed dose between discharge and when next dose was due 25 (15.9) 26 (10.4) .11 
How much medication are you giving? Wrong dose Unaware of dose or reported that dose differs by >10% of prescription26  12 (7.6) 7 (2.8) .03 
How many times a day are you giving the medication? Wrong interval Unaware of interval or reported interval does not match prescription 5 (3.2) 4 (1.6) .32a 
Was there ever a time you had to miss a dose? Missed doses <80% reported adherence to dosing regimen27  10 (6.4) 11 (4.4) .39 
How long will you give the medication for? Wrong duration Unaware of total duration, stopped medication early and so is at <80% adherence,27  or unaware that medication is chronic and has no end date 26 (16.6) 12 (4.8) <.001 
What are you watching for to know the medication is working? Unaware of desired effects Unaware of or never told why child is taking medication; unaware of or never told how to know child is improved 18 (11.5) 1 (0.40) <.001 
Are there any side effects of the medication you are looking for? Unaware of expected side effects Unaware of or never told about any side effects of medication 82 (52.2) 69 (27.7) <.001 
QuestionCategory of FailureDefinition of Incorrect AnswerFailed in Baseline Period (n = 157), n (%)Failed in Intervention Period (n = 249), n (%)P
Did you go home on any medications? Unaware prescription exists Unaware child was to continue taking medication after discharge 0 (0) 0 (0) NA 
What medication was it? Unaware of name or class Unaware of or wrong medication name or class 3 (1.9) 2 (0.8) .38a 
Did you have any trouble getting it from the pharmacy? Pharmacy trouble Did not pick up medication or delay in obtaining medication 10 (6.4) 9 (3.6) .20 
When did you start the medication? Wrong start time Missed dose between discharge and when next dose was due 25 (15.9) 26 (10.4) .11 
How much medication are you giving? Wrong dose Unaware of dose or reported that dose differs by >10% of prescription26  12 (7.6) 7 (2.8) .03 
How many times a day are you giving the medication? Wrong interval Unaware of interval or reported interval does not match prescription 5 (3.2) 4 (1.6) .32a 
Was there ever a time you had to miss a dose? Missed doses <80% reported adherence to dosing regimen27  10 (6.4) 11 (4.4) .39 
How long will you give the medication for? Wrong duration Unaware of total duration, stopped medication early and so is at <80% adherence,27  or unaware that medication is chronic and has no end date 26 (16.6) 12 (4.8) <.001 
What are you watching for to know the medication is working? Unaware of desired effects Unaware of or never told why child is taking medication; unaware of or never told how to know child is improved 18 (11.5) 1 (0.40) <.001 
Are there any side effects of the medication you are looking for? Unaware of expected side effects Unaware of or never told about any side effects of medication 82 (52.2) 69 (27.7) <.001 

NA, not applicable.

a

Fisher’s exact test was used.

Survey questions related to medication name, dose, frequency, duration, adherence, expected effects, and side effects were adapted from previous studies.813,28  Questions related to knowledge of new prescriptions, timing of the next dose after discharge, and difficulty obtaining the medication were added.6  The scripted questionnaire was pilot tested with 10 caregivers, and language was modified on the basis of feedback.

Additional data were collected, and medication information was confirmed in the EMR during the survey. These variables were related to medication documentation, dose rounding, use of the hospital-based pharmacy, number and class of eligible medications, discharge date, average daily census, and case-mix index. Hospital-specific software (Looking Glass Clinical Analytics, Streamline Health, Atlanta, GA) was used to collect length of stay (LOS), age, sex, race, ethnicity, and socioeconomic status (SES).29 

For consistency in questionnaire administration and data collection, all calls and chart reviews were done by the PI. Initial survey responses were discussed between the PI and senior author (M.L.R.) to develop correct response criteria and ensure survey administration and response recording consistency (Table 1).

Measures

The primary outcome measure was the weekly proportion of discharge medication–related failures, defined as an incorrect response on ≥1 survey questions. Failed questions were tracked individually and as a composite outcome. Process measures included the proportion of charts with (1) complete medication information on the AVS, (2) consistent information on the discharge summary and AVS, and (3) simplified dosing. Medication information on the AVS was defined as complete if all information typically present on an electronic prescription and the next dose due were present. The discharge summary and AVS were defined as consistent if prescription information was present and identical on both documents. Dosing was defined as simplified if rounded to 0.5 mL or a whole number. Because precise rounding (0.1 mL) is appropriate for some medications, the frequency of complex dosing was to be reduced, not eliminated.25  The weekly proportion of charts reviewed that used a smart phrase was an additional process measure (to ensure fidelity to the intervention), and LOS was the balancing measure (to assess increased work leading to discharge delays). Hospital-based pharmacy use was a key variable of interest because it aligned medication counseling with MEDRITES during interventions.

Analysis

Statistical process control was used to measure the interventions’ effect over time. Specifically, the percentage of weekly discharge medication–related failures was plotted on a p-chart, and initial trial limits and the center line (mean) were calculated from baseline data by using QI Charts 2.0.23 (Process Improvement Products, Austin, TX) for Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA). Baseline data were plotted weekly; intervention data, every 4 weeks; and sustainability data, monthly. Rules for detecting special cause variation were used to determine a system shift.30  The same analysis was used for process measures. To detect a 25% relative improvement in the baseline failure rate at 80% power, 14 questionnaires per week in the baseline period and 5 questionnaires per week in the intervention period were needed on average. To ensure sustainability, 5 questionnaires per week for 2 weeks/month continued for 6 months.

Data, including failure rate by survey question, in the baseline and intervention periods were compared by using the χ2 test or Fisher’s exact test for categorical variables and the Wilcoxon rank test for continuous variables. Unadjusted odds ratios (ORs) comparing those with and without a failure in the intervention period by process measures and use of the hospital-based pharmacy were calculated. If data related to survey answers or documentation were missing, a sensitivity analysis counting missing answers as (1) passed or present, (2) failed or absent, or (3) excluded was performed. For the primary analyses, conclusions did not change, and therefore missing data were set to “passed answer” or “present documentation” to bias results toward the null hypothesis. All other analyses were performed by using Stata 14.1 (Stata Corp, College Station, TX).

This study was approved by the Institutional Review Board at Albert Einstein College of Medicine. Verbal consent for data collection was obtained from caregivers.

There were 245 eligible patients contacted in the baseline period, of whom 157 (64%) agreed to participate. Of the 438 eligible patients contacted in the intervention period, 249 (57%) agreed to participate. In Table 2, we compare patients in the baseline and intervention periods. Patients did not differ significantly in age, sex, race and/or ethnicity, SES, or hospital-based pharmacy use. Patients with ≥2 prescriptions were less frequent in the intervention period (24.8% vs 16.1%; P = .03), and the distribution of eligible medications differed.

TABLE 2

Baseline and Intervention Period Cohort Characteristics

Total PatientsBaseline (n = 157)Intervention (n = 249)P
Female sex, n (%) 76 (48.4) 101 (40.6) .12 
Age, y, median (IQR) 3.0 (1.6 to 6.0) 4 (1.9 to 6.0) .31 
SES, median (IQR)a −3.0 (−6.6 to −1.2) −3.1 (−6.5 to −1.1) .93 
Race and/or ethnicity, n (%)   .09 
 Hispanic 61 (38.8) 98 (39.4)  
 Non-Hispanic Black 53 (33.8) 86 (34.5)  
 Non-Hispanic other 24 (15.3) 51 (20.5)  
 Unavailable or declined 19 (12.1) 14 (5.6)  
No. eligible medications, n (%)   .03 
 1 118 (75.2) 209 (83.9)  
 ≥2 39 (24.8) 40 (16.1)  
Type of eligible medication, n (%)   <.001 
 Corticosteroid 50 (31.9) 115 (46.2)  
 Antimicrobial 67 (42.7) 124 (49.8)  
 Antiviral 23 (14.6) 5 (2.0)  
 Other 17 (10.8) 5 (2.0)  
Used hospital-based pharmacy, n (%) 88 (56.1) 159 (63.9) .12 
Median monthly average daily census of the 2 participating units (IQR)b 54.6 (46.2 to 55) 47.9 (43.9 to 50.1) .11 
Average case-mix index of the 2 participating units, median (IQR)b 1.1 (1 to 1.3) 1.2 (1.1 to 1.2) .65 
Total PatientsBaseline (n = 157)Intervention (n = 249)P
Female sex, n (%) 76 (48.4) 101 (40.6) .12 
Age, y, median (IQR) 3.0 (1.6 to 6.0) 4 (1.9 to 6.0) .31 
SES, median (IQR)a −3.0 (−6.6 to −1.2) −3.1 (−6.5 to −1.1) .93 
Race and/or ethnicity, n (%)   .09 
 Hispanic 61 (38.8) 98 (39.4)  
 Non-Hispanic Black 53 (33.8) 86 (34.5)  
 Non-Hispanic other 24 (15.3) 51 (20.5)  
 Unavailable or declined 19 (12.1) 14 (5.6)  
No. eligible medications, n (%)   .03 
 1 118 (75.2) 209 (83.9)  
 ≥2 39 (24.8) 40 (16.1)  
Type of eligible medication, n (%)   <.001 
 Corticosteroid 50 (31.9) 115 (46.2)  
 Antimicrobial 67 (42.7) 124 (49.8)  
 Antiviral 23 (14.6) 5 (2.0)  
 Other 17 (10.8) 5 (2.0)  
Used hospital-based pharmacy, n (%) 88 (56.1) 159 (63.9) .12 
Median monthly average daily census of the 2 participating units (IQR)b 54.6 (46.2 to 55) 47.9 (43.9 to 50.1) .11 
Average case-mix index of the 2 participating units, median (IQR)b 1.1 (1 to 1.3) 1.2 (1.1 to 1.2) .65 

Continuous variables are presented as median (IQR); categorical variables are presented as n (%). IQR, interquartile range.

a

SES is calculated from small census tract data based on the child’s home address, and the deviation from the mean SES of the New York State population is reported as a z score29 ; sample size: n = 151 (baseline), n = 248 (intervention).

b

Based on all hospital admissions on those 2 units during the study period.

The mean weekly discharge medication–related failure rate decreased from 70.1% to 36.1% (48% relative decrease), exceeding our aim of a 25% relative decrease (Fig 3). In the intervention period, special cause variation occurred (at least 8 points below the previous center line) immediately. For sustainability, the original p-chart’s center line was extended for 6 months after the intervention period, and the decreased failure rate was sustained.

FIGURE 3

p-chart: discharge medication–related failures over time. The mean weekly discharge medication–related failure rate decreased from 70.1% to 36.1%; a shift occurred immediately, as special cause variation occurred (at least 8 points below the previous center line); the center line was extended for 6 months during sustainability while outcomes were monitored less frequently, and the decreased failure rate was sustained. LCL, lower control limit; UCL, upper control limit.

FIGURE 3

p-chart: discharge medication–related failures over time. The mean weekly discharge medication–related failure rate decreased from 70.1% to 36.1%; a shift occurred immediately, as special cause variation occurred (at least 8 points below the previous center line); the center line was extended for 6 months during sustainability while outcomes were monitored less frequently, and the decreased failure rate was sustained. LCL, lower control limit; UCL, upper control limit.

Close modal

Individual question answers were tracked, and the top 3 failure categories in both periods were the same: unaware of medication side effects, wrong duration, and did not start medication on time (Table 1). Given the predominance of failures related to side effects, analysis excluding this question revealed that a 61% relative decrease in the failure rate persisted in the intervention period (P < .001) (Table 3).

TABLE 3

Measures by Intervention Period

Baseline (n = 157)Intervention (n = 249)P
Outcome measures, n (%)    
 Discharge medication–related failures 110 (70.1) 90 (36.1) <.001 
 Discharge medication–related failures, excluding side effects 72 (45.9) 44 (17.7) <.001 
Process measures, n (%)    
 AVS information complete 33 (21.0) 212 (85.1) <.001 
 Consistent chart present 149 (94.9) 238 (95.6) .75 
 Simplified dosing present 118 (75.2) 238 (95.6) <.001 
 Used MEDRITES smart phrases NA 178 (71.5) NA 
Balancing measure, median (IQR)    
 LOS, h 48.0 (31.9 to 75.1) 39.1 (27.1 to 60.0) .002 
Baseline (n = 157)Intervention (n = 249)P
Outcome measures, n (%)    
 Discharge medication–related failures 110 (70.1) 90 (36.1) <.001 
 Discharge medication–related failures, excluding side effects 72 (45.9) 44 (17.7) <.001 
Process measures, n (%)    
 AVS information complete 33 (21.0) 212 (85.1) <.001 
 Consistent chart present 149 (94.9) 238 (95.6) .75 
 Simplified dosing present 118 (75.2) 238 (95.6) <.001 
 Used MEDRITES smart phrases NA 178 (71.5) NA 
Balancing measure, median (IQR)    
 LOS, h 48.0 (31.9 to 75.1) 39.1 (27.1 to 60.0) .002 

Categorical variables are presented as n (%) and analyzed with the χ2 test; continuous variables are presented as median (IQR) and analyzed with the Wilcoxon rank test. IQR, interquartile range; NA, not applicable.

p-charts for each process measure are shown in Fig 4. The mean weekly percentage of charts with complete AVS information increased from 20.3% to 84.8%, and the mean weekly percentage of charts with simplified dosing present increased from 75.7% to 94.9%. Both charts revealed special cause variation (Fig 4 A and C). When the overall rates were compared from the baseline to the intervention period, there were significantly more complete AVSs (21.0% vs 85.1%; P < .001) and more patients with simplified dosing (75.2% vs 95.6%; P < .001) in the intervention period (Table 3). The percentage of consistent charts remained high throughout the study, without special cause variation (Fig 4B). After introduction of MEDRITES smart phrases in the intervention period, their use averaged 72% per week and remained stable through the sustainability period, without special cause variation (Fig 4D).

FIGURE 4

p-charts of process measures: control charts for each process measure. A, Percentage of charts with AVS information complete over time. B, Percentage of consistent charts present over time. C, Percentage of charts with simplified dosing present over time. D, Percentage of charts that used MEDRITES smart phrases over time. LCL, lower control limit; UCL, upper control limit.

FIGURE 4

p-charts of process measures: control charts for each process measure. A, Percentage of charts with AVS information complete over time. B, Percentage of consistent charts present over time. C, Percentage of charts with simplified dosing present over time. D, Percentage of charts that used MEDRITES smart phrases over time. LCL, lower control limit; UCL, upper control limit.

Close modal

For a balancing measure of increased work around discharge, LOS was significantly shorter in the intervention period (48.0 vs 39.1 hours; P = .002) (Table 3). The use of the hospital-based pharmacy, an additional measure of interest, was not statistically different between periods (Table 2).

Caregivers in the intervention period who did and did not fail the survey were compared on process measures and use of the hospital-based pharmacy in an exploratory analysis of drivers of improved medication management and understanding. Those who used the pharmacy after counseling aligned with MEDRITES were approximately half as likely to fail (OR: 0.43; 95% confidence interval: 0.25–0.73). Similarly, caregivers who received an AVS with complete medication information were half as likely to fail (OR: 0.48; 95% confidence interval: 0.24–0.97).

An interprofessional QI team created a bundle of interventions to standardize the inpatient discharge medication process, which led to a sustained decrease in failures of medication management and increased understanding among caregivers of children discharged from the hospital. The mean weekly rate of discharge medication–related failures decreased from 70% to 36%, surpassing our goal. Additionally, complete medication documentation improved significantly. In our project, we identify specific categories in which caregivers lack understanding and provide a framework for approaching the entire discharge medication process. To our knowledge, this is one of the first improvement projects focused on the entire inpatient discharge medication process, specifically addressing categories in which caregivers lack understanding and standardizing documentation while testing improvement strategies.

Multiple ED studies7  suggest that caregivers have difficulty managing discharge medication plans, whereas few inpatient studies are focused on this topic.1416  In a multicenter collaborative, researchers also found lower rates of discharge-care failures using improvement strategies.14  Although sites reported working on obtaining discharge medications to reduce failures, interventions or outcomes related to improving medication management were not discussed.14  Both Mallory et al16  and Bhansali et al15  conducted studies assessing medication management and understanding after discharge; however, they did not analyze categories in which caregivers lacked understanding or analyze discharge documentation. Our study speaks to improved medication management by providing a framework for medication counseling, improving documentation, and measuring caregiver understanding.

In this study, we introduced a comprehensive method for systematic discharge medication counseling and a standardized approach to medication documentation (MEDRITES) across disciplines. This method resulted in a higher-quality discharge, as measured by caregiver’s knowledge and management of medications. The MEDRITES acronym created an approach to standardize written and verbal aspects of the discharge medication process. Housestaff, nurses, and pharmacists were guided to all counsel on medications similarly, and written instructions were augmented to support this approach. By changing our EMR’s medication rounding factors and creating templates for medication documentation, our system changed, supporting standardization in provider behaviors and practices. Subjectively, there was satisfaction with the tools to streamline this process, and LOS did not worsen during the intervention period, indicating that we did not overburden the system.

The context of our hospital is such that housestaff do most of the discharge counseling. Thus, we chose to target interventions toward this population, which may differ from other hospitals, where nurses, pharmacists, or clinical educators take on this primary role. It is unclear if the hospital-based pharmacy was crucial to the success of our interventions. We aligned with the pharmacy to deliver consistent medication information. However, we did not target use of the pharmacy as a process measure to achieve our aim. We deliberately assessed this predictor and found no significant difference in pharmacy use between study periods. We did, however, find significantly decreased odds of failure when pharmacy counseling aligned with housestaff counseling. Therefore, these results may be valuable to hospitals with a similar service or to those considering opening a hospital-based pharmacy.

This project took place at a single center with an outpatient hospital-based pharmacy and only included English-speaking caregivers. Additionally, our primary outcome was based on a nonvalidated questionnaire, although it was pilot tested with caregivers and modified before the study’s onset. The PI administered all surveys and, as is typical of QI studies, was not blinded to results. Standardization by using pre-scripted responses reduces but does not eliminate this bias. Questionnaire answers were based on caregiver self-report and recall, and we could not confirm if caregivers were actually administering the medication as reported, creating potential social desirability bias. Although we randomly assigned patient identification for calls, there is likely selection bias in both the baseline and intervention groups. When compared with the overall population of children discharged from the hospital, caregivers reached on the phone and willing to participate are likely a biased sample of the entire cohort exposed to these interventions. Because of resource constraints, baseline data are from 1 winter season, when hospital census and patient acuity is typically high. However, use of the hospital-based pharmacy, average daily census, and case-mix index did not differ significantly between periods, suggesting that the overall environment of the hospital was relatively stable. Additionally, improvements persisted through the winter season of the intervention period. Between the baseline and intervention periods, we found significant differences in the number and type of eligible medications and LOS. The higher number and distribution of eligible medications and shorter LOS seen in the baseline period may be a proxy for the medical complexity of the population studied. Shorter LOS may also be related to the lower average daily census. Although it was not statistically significant, it may contribute to improvements in available staffing and resources, which may improve discharge efficiency. Finally, there were several changes in nursing leadership during this time, which may affect team dynamics and culture, although the specific relationship to the discharge process is unclear.

This project demonstrates a successful approach to standardizing the discharge medication process to enhance caregiver medication management and understanding after inpatient discharge. The quality of this entire process improved with an interprofessional approach focused on systematic medication counseling, standardized medication documentation, and simplified dose rounding. Results were sustained, and next steps include spreading this work to other services and non–English-speaking caregivers.

We acknowledge Drs Patricia Hametz and Alexander Hogan for their thorough review of the article, as well as the pharmacists, housestaff, nurses, and attending physicians who participated in this project.

Dr Philips conceptualized and designed the study, designed the survey tool, collected data, conducted analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Zhou helped design the study and survey tool, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Rinke conceptualized and designed the study, supervised data collection and analyses, and critically reviewed and revised the manuscript; Drs Lee, Marrese, Nazif, Browne, and Sinnet, Mr Tuckman, and Ms Modi helped design the study and survey tool and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Supported by the National Institutes of Health National Center for Advancing Translational Science Clinical and Translational Science Award grants TL1TR001072 and UL1TR001073. The study’s contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health. Funded by the National Institutes of Health (NIH)

     
  • AVS

    after-visit summary

  •  
  • ED

    emergency department

  •  
  • EMR

    electronic medical record

  •  
  • LOS

    length of stay

  •  
  • MEDRITES

    medication name, engage family, dose, route, indications, timing, effects to watch for, storage and syringe

  •  
  • OR

    odds ratio

  •  
  • PI

    primary study investigator

  •  
  • QI

    quality improvement

  •  
  • SES

    socioeconomic status

1
Schillie
SF
,
Shehab
N
,
Thomas
KE
,
Budnitz
DS
.
Medication overdoses leading to emergency department visits among children
.
Am J Prev Med
.
2009
;
37
(
3
):
181
187
2
Kaushal
R
,
Goldmann
DA
,
Keohane
CA
, et al
.
Adverse drug events in pediatric outpatients
.
Ambul Pediatr
.
2007
;
7
(
5
):
383
389
3
Smith
MD
,
Spiller
HA
,
Casavant
MJ
,
Chounthirath
T
,
Brophy
TJ
,
Xiang
H
.
Out-of-hospital medication errors among young children in the United States, 2002–2012
.
Pediatrics
.
2014
;
134
(
5
):
867
876
4
Auger
KA
,
Simon
TD
,
Cooperberg
D
, et al
.
Summary of STARNet: seamless transitions and (Re)admissions network
.
Pediatrics
.
2015
;
135
(
1
):
164
175
5
Berry
JG
,
Blaine
K
,
Rogers
J
, et al
.
A framework of pediatric hospital discharge care informed by legislation, research, and practice
.
JAMA Pediatr
.
2014
;
168
(
10
):
955
962–966
6
Philips
K
,
Zhou
R
,
Lee
DS
, et al
.
Caregiver medication management and understanding after pediatric hospital discharge
.
Hosp Pediatr
.
2019
;
9
(
11
):
844
850
7
Glick
AF
,
Farkas
JS
,
Nicholson
J
, et al
.
Parental management of discharge instructions: a systematic review
.
Pediatrics
.
2017
;
140
(
2
):
e20164165
8
Yin
HS
,
Dreyer
BP
,
van Schaick
L
,
Foltin
GL
,
Dinglas
C
,
Mendelsohn
AL
.
Randomized controlled trial of a pictogram-based intervention to reduce liquid medication dosing errors and improve adherence among caregivers of young children
.
Arch Pediatr Adolesc Med
.
2008
;
162
(
9
):
814
822
9
Yin
HS
,
Dreyer
BP
,
Moreira
HA
, et al
.
Liquid medication dosing errors in children: role of provider counseling strategies
.
Acad Pediatr
.
2014
;
14
(
3
):
262
270
10
Al-Harthy
N
,
Sudersanadas
KM
,
Al-Mutairi
M
, et al
.
Efficacy of patient discharge instructions: a pointer toward caregiver friendly communication methods from pediatric emergency personnel
.
J Family Community Med
.
2016
;
23
(
3
):
155
160
11
Grover
G
,
Berkowitz
CD
,
Lewis
RJ
.
Parental recall after a visit to the emergency department
.
Clin Pediatr (Phila)
.
1994
;
33
(
4
):
194
201
12
Yin
HS
,
Parker
RM
,
Sanders
LM
, et al
.
Liquid medication errors and dosing tools: a randomized controlled experiment
.
Pediatrics
.
2016
;
138
(
4
):
e20160357
13
Kaestli
LZ
,
Noble
S
,
Combescure
C
, et al
.
Drug information leaflets improve parental knowledge of their child’s treatment at paediatric emergency department discharge
.
Eur J Hosp Pharm Sci Pract
.
2016
;
23
(
3
):
151
155
14
Wu
S
,
Tyler
A
,
Logsdon
T
, et al
.
A quality improvement collaborative to improve the discharge process for hospitalized children
.
Pediatrics
.
2016
;
138
(
2
):
e20143604
15
Bhansali
P
,
Washofsky
A
,
Romrell
E
,
Birch
S
,
Winer
JC
,
Hoffner
W
.
Parental understanding of hospital course and discharge plan
.
Hosp Pediatr
.
2016
;
6
(
8
):
449
455
16
Mallory
LA
,
Diminick
NP
,
Bourque
JP
, et al
.
Pediatric patient-centered transitions from hospital to home: improving the discharge medication process
.
Hosp Pediatr
.
2017
;
7
(
12
):
723
730
17
Gattari
TB
,
Krieger
LN
,
Hu
HM
,
Mychaliska
KP
.
Medication discrepancies at pediatric hospital discharge
.
Hosp Pediatr
.
2015
;
5
(
8
):
439
445
18
Johnson
KB
,
Butta
JK
,
Donohue
PK
,
Glenn
DJ
,
Holtzman
NA
.
Discharging patients with prescriptions instead of medications: sequelae in a teaching hospital
.
Pediatrics
.
1996
;
97
(
4
):
481
485
19
Jones
AN
,
Miller
JL
,
Neely
S
, et al
.
Prevalence of unrounded medication doses and associated factors among hospitalized pediatric patients
.
J Pediatr Pharmacol Ther
.
2017
;
22
(
4
):
286
292
20
Cua
YM
,
Kripalani
S
.
Medication use in the transition from hospital to home
.
Ann Acad Med Singapore
.
2008
;
37
(
2
):
136
141
21
US Census Bureau
. QuickFacts: Bronx County (Bronx Borough), New York. Available at: https://www.census.gov/quickfacts/fact/map/bronxcountybronxboroughnewyork/EDU635216#viewtop. Accessed July 18, 2018
22
Yin
HS
,
Johnson
M
,
Mendelsohn
AL
,
Abrams
MA
,
Sanders
LM
,
Dreyer
BP
.
The health literacy of parents in the United States: a nationally representative study
.
Pediatrics
.
2009
;
124
(
suppl 3
):
S289
S298
23
Glaze
J
. Epic Systems draws on literature greats for its next expansion. 2015. Available at: https://madison.com/news/local/govt-and-politics/epic-systems-draws-on-literature-greats-for-its-next-expansion/article_4d1cf67c-2abf-5cfd-8ce1-2da60ed84194.html. Accessed October 21, 2019
24
Langley
GJ
,
Nolan
KM
,
Nolan
TW
,
Norman
CL
,
Provost
LP
.
The Improvement Guide: A Practical Approach to Enhancing Organizational Performance
, 1st ed.
San Francisco, CA
:
Jossey-Bass
;
1996
25
Johnson
KB
,
Lee
CK
,
Spooner
SA
,
Davison
CL
,
Helmke
JS
,
Weinberg
ST
.
Automated dose-rounding recommendations for pediatric medications
.
Pediatrics
.
2011
;
128
(
2
). Available at: https://pediatrics.aappublications.org/content/128/2/peds.2011-0760d
26
Rinke
ML
,
Moon
M
,
Clark
JS
,
Mudd
S
,
Miller
MR
.
Prescribing errors in a pediatric emergency department
.
Pediatr Emerg Care
.
2008
;
24
(
1
):
1
8
27
Osterberg
L
,
Blaschke
T
.
Adherence to medication
.
N Engl J Med
.
2005
;
353
(
5
):
487
497
28
Isaacman
DJ
,
Purvis
K
,
Gyuro
J
,
Anderson
Y
,
Smith
D
.
Standardized instructions: do they improve communication of discharge information from the emergency department?
Pediatrics
.
1992
;
89
(
6 pt 2
):
1204
1208
29
Diez Roux
AV
,
Merkin
SS
,
Arnett
D
, et al
.
Neighborhood of residence and incidence of coronary heart disease
.
N Engl J Med
.
2001
;
345
(
2
):
99
106
30
Provost
LP
,
Murray
SK
.
The Health Care Data Guide: Learning from Data for Improvement
.
San Francisco, CA
:
Jossey-Bass
;
2011

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data