Professional Documents
Culture Documents
Background: Medication reconciliation at transitions tion patients there were 170 PADEs (1.05 per patient)
in care is a national patient safety goal, but its effects on (adjusted relative risk [ARR], 0.72; 95% confidence in-
important patient outcomes require further evaluation. terval [CI], 0.52-0.99). A significant benefit was found
We sought to measure the impact of an information tech- at hospital 1 (ARR, 0.60; 95% CI, 0.38-0.97) but not at
nology–based medication reconciliation intervention on hospital 2 (ARR, 0.87; 95% CI, 0.57-1.32) (P=.32 for test
medication discrepancies with potential for harm (po- of effect modification). Hospitals differed in the extent
tential adverse drug events [PADEs]). of integration of the medication reconciliation tool into
computerized provider order entry applications at
Methods: We performed a controlled trial, random- discharge.
ized by medical team, on general medical inpatient units
at 2 academic hospitals from May to June 2006. We en- Conclusions: A computerized medication reconcilia-
rolled 322 patients admitted to 14 medical teams, for tion tool and process redesign were associated with a de-
whom a medication history could be obtained before dis-
crease in unintentional medication discrepancies with po-
charge. The intervention was a computerized medica-
tential for patient harm. Software integration issues are
tion reconciliation tool and process redesign involving
physicians, nurses, and pharmacists. The main out- likely important for successful implementation of com-
come was unintentional discrepancies between pread- puterized medication reconciliation tools.
mission medications and admission or discharge medi-
cations that had potential for harm (PADEs). Trial Registration: clinicaltrials.gov Identifier:
NCT00296426
Results: Among 160 control patients, there were 230
PADEs (1.44 per patient), while among 162 interven- Arch Intern Med. 2009;169(8):771-780
E
FFORTS TO IMPROVE THE care Organizations designated medica-
quality and safety of health tion reconciliation as a “National Patient
care include attention to un- Safety Goal” in 2005.7 Medication recon-
intentional medication dis- ciliation is “a process of identifying the
crepancies, defined as un- most accurate list of all medications a pa-
explained differences among documented tient is taking . . . and using this list to pro-
regimens across different sites of care (eg, vide correct medications for patients any-
prior to admission compared with hospi- where within the health care system.”8
tal admitting orders).1,2 Discrepancies are Hospitals have undertaken diverse ap-
highly prevalent; up to 67% of inpatients proaches to comply with The Joint Com-
have at least 1 unexplained discrepancy in mission’s mandate. Some studies support
their prescription medication history at medication reconciliation as a means to re-
admission.3 duce medication discrepancies or ADEs9-15;
Because medication discrepancies are most are either pre-post studies or uncon-
an important contributor to adverse drug trolled evaluations of the types of errors
events (ADEs) among hospitalized and re- intercepted by the process. Few rigorous
Author Affiliations are listed at cently discharged patients,4-6 The Joint studies demonstrate that medication rec-
the end of this article. Commission on Accreditation of Health- onciliation efforts improve important pa-
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
771
Gathers additional sources of Gathers additional sources of PAML Gathers additional sources of
PAML information, helps resolve information, resolves uncertainties PAML information, helps resolve
uncertainties or discrepancies in or discrepancies in PAML with help uncertainties or discrepancies in
PAML, notifies responsible physician of nurse and/or pharmacist PAML, notifies responsible physician
During
hospitalization
Figure 1. Redesigned medication reconciliation workflow. Adapted from Poon et al17 with permission. PAML indicates preadmission medication list.
tient outcomes or define the best ways to implement these and 1 or 2 medical students. Patients were enrolled if study phar-
processes or identify the patients most likely to benefit. macists (generally 1 pharmacist per weekday per hospital) had
We sought to determine the effects of a redesigned pro- time to obtain a medication history prior to discharge. Pa-
cess for medication reconciliation, supported by infor- tients admitted to 1 of 7 randomly chosen medical teams and
floors were assigned to the intervention, while patients admit-
mation technology (IT), on potential ADEs (PADEs). We
ted to the other teams and on different floors received usual
hypothesized that our intervention would decrease PADEs care. Thus, patients in the 2 arms were cared for by different
in all patients and that patients at high risk for medica- physicians and nurses. Randomization was stratified by
tion discrepancies would benefit most. study hospital and assigned by the principal investigator
( J.L.S.) using random number generation in Microsoft Excel
METHODS (Microsoft Corp, Redmond, Washington). Patients dis-
charged from a nonstudy team or floor, patients transferred
between a control team or floor and an intervention team or
STUDY DESIGN, SETTING, AND PARTICIPANTS floor, and patients discharged after June 20, 2006, were
excluded. The study was approved by the institutional
The Partners Medication Reconciliation Study was a cluster- review board of Partners HealthCare, Boston; patient con-
randomized controlled trial conducted from May 1 through June sent was deemed unnecessary.
20, 2006, at 2 large academic hospitals in Boston, Massachu-
setts. The design of this study has been described.16 Eligible pa-
tients were admitted to one of several general medicine teams INTERVENTION
and floors of each hospital, according to a rotating call cycle.
Each team (6 at hospital 1 and 8 at hospital 2) consisted of 1 The intervention consisted of an IT application designed to fa-
attending physician, 1 junior or senior resident, 2 to 4 interns, cilitate medication reconciliation, integrated into the inter-
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
772
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
773
resolved by discussion and by a third adjudicator if necessary cluded gout medications, muscle relaxants, hyperlipidemic medi-
(needed in 86 of 4700 adjudicated medication discrepancies cations, antidepressants, and respiratory medications.16 Health-
[1.8%]). care utilization outcomes were compared using logistic regression
Weekly meetings were conducted to ensure consistency be- with general estimating equations clustered by admitting
tween sites and among study pharmacists and physician adju- physician.
dicators. Reliability of the gold standard preadmission medi- Subgroup analyses explored possible effect modification; pre-
cation histories and physician adjudication of outcomes have specified subgroups included study hospital, PADE risk score,
been shown to be moderate to high.16 Prespecified secondary number of preadmission medications, transfers from outside
outcomes, based on hospital administrative data, included emer- institutions, level of training of admitting physician, and pa-
gency department visits and hospital readmissions within 30 tient understanding of medications. The PADE risk score, de-
days of discharge. rived from the control group, was calculated by assigning 2
points for patients younger than 85 years and 1 point each for
STATISTICAL ANALYSIS low or medium patient understanding of medications, having
16 or more preadmission medications, 4 or more high-risk pread-
Patient characteristics and study results were calculated using pro- mission medications (as defined in the previous paragraph),
portions, means with standard deviations, and medians with in- 13 or more outpatient visits in the previous year, and having a
terquartile ranges. Poisson log-linear regression was used to de- family member or caregiver as a source of preadmission medi-
termine associations between the number of PADEs per patient cation information.16 Interaction terms (ie, subgroup⫻study
and study arm. To adjust for potential confounding, a weighted arm) were entered into secondary models to evaluate the sta-
propensity score approach was used in which the data for each tistical significance of any effect modification.
patient were weighted by the inverse of the probability of being Generalized estimating equations, using a robust covariance
in the treatment arm given each of the potential confounding co- matrix, were applied to adjust for clustering of results by the ad-
variates.25 Covariates, chosen a priori on clinical grounds, in- mitting physician. Model fit for the propensity score model of the
cluded patient age, number of outpatient visits within the previ- primary outcome was assessed based on aggregates of residu-
ous year, inpatient admissions to index hospital within the previous als26 using the ASSESS statement in SAS statistical software (SAS
month, number of preadmission medications, number of high- Institute Inc, Cary, North Carolina), with a P value computed based
risk preadmission medications, admission source, primary care on 10 000 simulated paths (P=.60, suggesting good model fit).
physician (PCP) from within the hospital network, whether the Analyses were intention to treat. P⬍.05 (2 sided) was consid-
PCP was the discharging physician, family or caregiver as a source ered significant. Analyses were implemented with SAS statistical
of preadmission medication information, level of training of ad- software, version 9.1 (SAS Institute Inc).
mitting physician, and patients’ understanding of their medica- Our target sample comprised 460 patients, which was es-
tions (subjectively categorized by the study pharmacist as low, timated to provide a 90% power to detect a 60% relative de-
medium, or high). High-risk medication classes, based on their crease (based on studies of paper-based medication reconcili-
risk for causing PADEs in the control group when prescribed, in- ation9,10) in the presence of any PADE (from 27%5,6,11 to 11%
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
774
(continued)
of patients), assuming an ␣ level of .05, 5 patients per admit- A PAML was “completed” (planned actions on admis-
ting physician, and an intraclass correlation coefficient by phy- sion assigned for all medications and the list signed off
sician of 0.10 (we did not adjust for additional correlation at as “ready for review” by a pharmacist) within 24 hours
the team level). The study was not powered to detect a differ- of admission for 75 patients (46%), including 42 of 84
ence in health care utilization.
patients (50%) at hospital 1 and 33 of 78 patients (42%)
at hospital 2 (P=.35 for comparison). A PAML was com-
RESULTS pleted prior to discharge in 121 patients (75%). The pri-
mary outcome could be evaluated in all 322 patients.
DESCRIPTION OF STUDY SAMPLE Figure 2 shows the study flow. Patient characteristics
in the 2 study arms were similar, except for a signifi-
Of 801 patients originally identified as potentially eli- cantly lower proportion of PCPs within the Partners
gible for the study, 322 were enrolled and were cared for HealthCare network and a higher proportion of non-
by each of the 14 randomized medical teams and by 117 medicine interns in the intervention group (Table 1).
different admitting physicians. The most common rea- Some differences were also noted by study hospital (a table
son for patient exclusion was lack of time of the study of the site differences in baseline patient characteristics
pharmacist to obtain a medication history before dis- is available from the corresponding author).
charge. Compared with enrolled patients, unenrolled pa-
tients had shorter lengths of stay (a table of the charac- EFFECT OF THE INTERVENTION ON PADEs
teristics of the study sample and the excluded subjects
is available from the corresponding author). Among the 162 patients assigned to the intervention, there
Of 162 patients assigned to the intervention arm, the were 170 unintentional medication discrepancies with
PAML Builder application was used in 160 patients (99%). potential for patient harm (1.05 PADEs per patient) vs
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
775
Abbreviations: DRG, diagnosis-related group; ED, emergency department; IQR, interquartile range; Ob-Gyn, obstetrics/gynecology; PCP, primary care physician.
a P⬍.05 for comparison between intervention and control groups (total population).
b P⬍.05 for comparison between intervention and control groups at hospital 2.
c Based on Medicare 2006 weights (version 23).27
d Based on administrative billing codes.28
e Excluding “as needed” medications and topical agents.
f In 5 medication classes most likely to cause potential adverse drug events when prescribed: gout medications, muscle relaxants, hyperlipidemic agents,
antidepressants, and respiratory medications.16
g P⬍.05 for comparison between intervention and control groups at hospital 1.
h Based on pharmacist assessment.
230 (1.44 PADEs per patient) among the 160 patients lower (adjusted and clustered relative risk, 1.09; 95% CI,
assigned to usual care, an adjusted 28% relative risk re- 0.49-2.44) (P value for interaction, .02). The interven-
duction (Table 2). Ninety-eight PADEs were consid- tion was associated with a significant reduction in PADEs
ered serious, ie, to have potential to cause serious harm at hospital 1 but not at hospital 2 (P value for interac-
such as rehospitalization or persistent alteration in health tion, .32) (Table 4). Differences between hospitals were
function, including 43 PADEs in the intervention arm qualitatively similar for the adjusted effect of the inter-
(0.27 per patient) and 55 PADEs in those assigned to usual vention on PADEs due to history errors, those due to rec-
care (0.34 per patient). The intervention was associated onciliation errors, PADEs at admission, and those at dis-
with a significant reduction in PADEs at discharge but charge (Table 4). No other subgroup differences were
not at admission (Table 2). Table 3 gives examples of found in the effect of the intervention.
different types of PADEs identified during the study.
No significant differences were found in health care uti-
COMMENT
lization. The rate of hospital readmission or emergency de-
partment visit within 30 days was 20% in the intervention
arm and 24% in the usual care arm (clustered odds ratio, In this 2-hospital cluster-randomized controlled trial, we
0.76; 95% confidence interval [CI], 0.43-1.35). found that a medication reconciliation intervention con-
In subgroup analyses, we found effect modification by sisting of novel IT and process redesign involving phy-
PADE risk score and a suggestion of an effect modifica- sicians, nurses, and pharmacists was associated with a
tion by hospital. The effect of the intervention was greater 28% relative risk reduction in unintentional medication
in the 167 patients with a PADE risk score of 4 or higher discrepancies with potential for harm, a type of PADE.
(adjusted and clustered relative risk, 0.62; 95% CI, 0.41- The absolute risk reduction between the 2 arms was 0.39
0.93) than in the 155 patients with a risk score of 3 or PADE per patient or a number needed to treat 2.6 pa-
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
776
tients to prevent 1 PADE. The intervention was more suc- A significant reduction in PADEs with the interven-
cessful in patients at high risk for medication discrep- tion at hospital 1 but not at hospital 2 could have been
ancies and possibly more successful at hospital 1 than at due to chance, since the study was not powered to de-
hospital 2. To our knowledge, this is the first random- tect a difference in PADEs at each hospital individually.
ized controlled trial of an IT-based medication recon- Alternatively, part of the answer may be the phased hos-
ciliation intervention. pitalwide rollout of the intervention at hospital 1, which
We believe our intervention was successful because led to more publicity about medication reconciliation,
it combined effective process redesign with IT. The new possibly resulting in greater compliance with the new pro-
reconciliation process encouraged interdisciplinary com- cesses. Hospital differences related to reducing “medi-
munication and cross-checks. The PAML Builder cation history errors” may be due in part to greater in-
application facilitated accurate medication histories volvement of nurses at admission at hospital 1. Based on
by presenting several sources of available medication in- informal, unblinded interviews with clinical personnel
formation, and it displayed the PAML with current in- conducted after study completion, nurses at hospital 1
patient medications during the discharge ordering more consistently confirmed PAML accuracy at admis-
process. sion compared with hospital 2. Hospital differences in
However, our intervention was far from perfect in reducing “reconciliation errors” were likely due to dif-
eliminating potentially harmful medication discrepan- ferences in discharge medication computer screens de-
cies (1.05 PADEs per patient in the intervention arm). veloped at the 2 hospitals—a hypothesis we generated a
Possible reasons include incomplete and inaccurate elec- priori. At hospital 1, it was easier to view inpatient and
tronic sources of ambulatory medications, lack of pa- preadmission medications simultaneously and to order
tient and caregiver knowledge of preadmission medica- medications from either list at discharge.
tion regimens, lack of clinician adherence with the Regarding other studies of medication reconcilia-
reconciliation process, and software usability issues (such tion, 2 recent pre-post studies of paper-based interven-
as ease of adding new medications to the PAML and lack tions evaluated effects on actual ADEs based on retro-
of integration with the admission ordering process). spective review of a random selection of medical records,
Our intervention was more successful in patients at demonstrating a 43% to 84% relative risk reduction.9,10
high risk for medication discrepancies, based on a risk A recent randomized controlled trial of a pharmacist-
score derived from the control group. If prospectively vali- run “medication reconciliation and seamless care ser-
dated in other populations, this score could prove use- vice” found that 56.3% of 119 control patients had a drug
ful in prioritizing those most in need of intensive recon- therapy inconsistency or omission at discharge on ret-
ciliation efforts, ie, above the minimum Joint Commission rospective medical record review compared with 3.6%
standard.7 of intervention patients (based on an independent re-
The intervention reduced PADEs at discharge but not view of 17% of intervention patients’ medical records by
at admission. This result was likely because of the fol- a second pharmacist).29 Mostly descriptive studies of IT-
lowing circumstances: (1) PADEs in general were more based medication reconciliation tools are also begin-
common at discharge than at admission16; (2) most er- ning to be published,30-34 and commercial vendors are
rors of reconciliation occurred at discharge, and the PAML starting to provide medication reconciliation modules in
Builder may have been particularly effective at reducing their applications.30,35
these kinds of errors; and (3) delays in completing a PAML We believe IT-based medication reconciliation inter-
would also have attenuated the effectiveness of the in- ventions have several advantages over paper-based so-
tervention at admission but not at discharge. lutions, including the ability to use existing electronic
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
777
History Timing of Error Type of Error Reason for Error Potential Severity
A patient with multiple cardiac risk factors was admitted for Admission Omission Reconciliation error Serious
atypical chest pain. At home he was taking atenolol,
100 mg by mouth daily. The medical team accurately
documented this medication in the preadmission
medication list but did not prescribe it (or any other
-blocker) on admission or any time during the
hospitalization. No medical reason could be found for
withholding it, and notes document a plan to continue
optimal medical management of possible coronary
disease. The patient was discharged on his home dose of
atenolol.
A patient with a seizure disorder was admitted for peripheral Admission Frequency History error Serious
edema and COPD exacerbation. At home the patient was
taking carbamazepine, 100 mg by mouth twice daily, but
the medical team incorrectly documented 100 mg by
mouth 3 times a day in the preadmission medication list
and then prescribed that frequency on admission. No
blood levels of the medication were drawn during the
hospitalization.
A patient with a history of asthma was admitted for an Discharge Omission Reconciliation error Serious
asthma exacerbation. At home, the patient was using an
albuterol inhaler, 2 puffs every 4 hours as needed for
shortness of breath. The team accurately documented the
medication in the preadmission medication list. During the
hospitalization the patient was given albuterol by
nebulizer. At discharge, no albuterol (or any other
short-acting bronchodilator) was prescribed.
A patient with gastroesophageal reflux disease was admitted Discharge Omission History error Significant
with nausea and vomiting and was found to have
hyponatremia. At home, the patient was taking
omeprazole 20 mg by mouth daily, but the medical team
did not document this medication in the preadmission
medication list. During hospitalization, the patient was
prescribed esomeprazole for stress ulcer prophylaxis. At
discharge, no proton pump inhibitor or other antireflux
medication was prescribed.
A patient with a history of myocardial infarction and Discharge Dose History error Serious
hypertension was admitted for a gastrointestinal bleed. At
home the patient was taking lisinopril, 10 mg by mouth
daily, but the medical team incorrectly documented 25 mg
by mouth daily in the preadmission medication list.
During the hospitalization the patient was hypertensive,
and the team appropriately ordered captopril instead of
lisinopril to manage his blood pressure. At discharge, with
a normal blood pressure, the team ordered what they
thought was his home dose of lisinopril, 25 mg by mouth
daily.
sources of ambulatory medication information, better in- taking fewer medications. Second, the study measured
tegration into workflow in hospitals with CPOE, easier potential and not actual ADEs, and while we analyzed
sharing of reconciliation information across providers, health care utilization, the study was not powered to de-
automatic production of documentation for discharge tect a reduction in this outcome. Third, full use of the
summaries, comparisons of medication lists to facilitate PAML Builder was not achieved: only 46% of patients had
reconciliation and patient education, provision of alerts a completed PAML within 24 hours of admission (al-
and reminders to ensure compliance, and ability to track though 75% were complete by discharge), thus limiting
compliance to inform further process improvement. the ability of the intervention to benefit patients. Based
This study has several limitations. First, because it was on an analysis of clinician attitudes and patterns of use
conducted on general medical services at academic hos- of the application, we attribute this nonadherence to the
pitals, results may not be generalizable to other settings. lack of integration of the application with CPOE at ad-
Patients with very short lengths of stay may have been mission.36 This problem has since been remedied; cur-
disproportionately discharged before enrollment, lead- rently 80% to 90% of PAMLs are complete within 24 hours
ing to selection of a sicker patient population; the re- of admission. Fourth, we cannot exclude the possibility
sults may not be generalizable to less sick patients or those of unmeasured provider characteristics confounding the
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
778
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
779
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 8), APR 27, 2009 WWW.ARCHINTERNMED.COM
780