Hospital Pharmacy - December 2020 - 387

387

Popp et al

Figure 1. Sample selection and attrition criteria.

the rate at which they occurred (Table 3). As a group, cardiovascular agents had the highest rate of discharge medication
changes (26%) and gastrointestinal agents had the highest
rate of TI (65%). Psychotropic agents had the lowest rates of
both discharge medication changes (12%) and TI (7%).
Excluding sample sizes in the lowest quartile (n < 22), the
drug classes with the highest rates of discharge medication
changes were as follows: beta-blockers (33%), dihydropyridine calcium channel blockers (30%), angiotensin II receptor
blockers (ARBs, 29%), and angiotensin-converting enzyme
(ACE) inhibitors (24%). The drug classes with the highest
rates of TI were proton-pump inhibitors (PPIs, 68%), histamine-2 receptor antagonists (55%), bronchodilators (38%),
ARBs (28%), and statins (21%). Histamine-2 receptor antagonists were the most likely drug class to remain interchanged
following discharge (14%). Discharge medication changes
for other reasons (eg, dosing changes and switching to an
alternative non-interchange medication) generally occurred
at a low rate across drug classes.
For the entire patient cohort, the 30-, 60-, and 90-day
same-hospital readmission rates were 10.5%, 14.9%, and
16.7%, respectively. The 30-, 60-, and 90-day emergency
department visit rates were 6.4%, 9.5%, and 11.9%, respectively. A multiple regression model was performed to measure the predictive value of key variables on hospital
readmission and emergency department visits. The model
revealed that neither the presence of TI nor the number of
discharge medication changes had a significant effect on
either hospital readmissions or emergency room visits
(Table 4). Post hoc findings of the model revealed that a
higher total number of discharge medications predicted a

greater risk of hospital readmission or emergency department visit at all time points (30-, 60-, and 90 days post discharge), and that medication-specific discharge counseling
was predictive of a lower risk of hospital readmission at 90
days post discharge. Other variables including patient age,
hospital service, and insurance status were not predictive of
any outcome.

Discussion
This study was designed to expand upon the results of two
previous studies by Glaholt et al11 and Wang et al,12 each of
which examined the relationship between TI and discharge
medication changes. Our study included a broader array of
drug classes than the previous studies and was the first
study to examine clinical outcomes as a potential consequence of TI. Importantly, our rates of overall TI differ
from previous studies because our study was done at a different health care system with a different formulary and
patient population.
In this study, TI occurred in one-fifth of admission medications and was associated with a 70% increased risk for a
discharge medication change without an increase in 30-,
60-, or 90-day hospital readmission or emergency department visit rates. The greatest concern with unnecessary discharge medication changes is the risk of adverse drug
effects as a result of the change. As this was a retrospective
study, it was not feasible to directly measure the presence
and severity of adverse drug events following discharge
from the hospital. As a result, surrogate markers for adverse
events were utilized including hospital readmission or

Hospital Pharmacy - December 2020

Table of Contents for the Digital Edition of Hospital Pharmacy - December 2020