Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Research in Social and Administrative Pharmacy

journal homepage: www.elsevier.com/locate/rsap

The effectiveness of medication reconciliation to prevent medication error:

A systematic review and meta-analysis
Daranee Chiewchantanakita,∗∗, Anupong Meakchaia, Natdanai Pituchaturonta,
Piyameth Dilokthornsakula,b, Teerapon Dhippayoma,∗
a
Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Naresuan University, Phitsanulok, Thailand
b
Center of Pharmaceutical Outcomes Research, Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Naresuan University, Phitsanulok, Thailand

ARTICLE INFO ABSTRACT

Keywords: Background: The impact of medication reconciliation (MR) in low-middle-income countries, including Thailand,
Medication reconciliation may differ from other developed countries.
Medication error Objective: To evaluate the effect of medication reconciliation (MR) on the reduction of medication error in
Hospital care Thailand.
Thailand
Methods: A systematic search was conducted in the following databases: PubMed, CENTRAL, CINAHL, Scopus,
Systematic review
Thai Journals Online, Thai index Medicus, Thai Medical Index, and Health Science Journal in Thailand from
inception to January 2018. Studies that evaluated the effect of MR compared to usual care within hospitals in
Thailand and reported the occurrence of medication error were included. Meta-analyses were performed using
random-effects model.
Results: Of the 107 articles retrieved, 7 articles involving 1581 patients were included in quantitative synthesis.
Three of the included studies were randomized controlled trials (RCT). Overall, the risk of medication error in
patients who received MR in all transitions of care was 75% lower than those receiving usual care (RR 0.25;
95%CI 0.15–0.43). The effect on the reduction of medication error appeared higher when MR was provided to
ambulatory patients (RR 0.17 [95%CI 0.04–0.80] compared with hospitalized patients during admission (RR
0.37 [95%CI 0.20–0.65]) and discharge (RR 0.27 [95%CI 0.17–0.43]). Effects on reducing medication error was
greater when MR was provided in secondary care hospitals compared with primary care hospitals both during
admission (RR 0.49 [95%CI, 0.34–0.69] vs RR 0.25 [95%CI, 0.05–1.26]), and discharge transition (RR 0.19
[95%CI, 0.09–0.39] vs RR 0.30 [95%CI, 0.12–0.79]).
Conclusion: Overall, current evidence indicates that the provision of MR in Thailand is effective in reducing
medication errors in all transitions of care. However, to promote patient safety, appropriate strategies should be
developed to support MR in specific transition of care and hospital setting so patients can benefit most from this
service.

Introduction histories is therefore crucial to prevent medication error associated

Medication error is a preventable event that may cause or lead to
patient harm.1 According to a previous systematic review, at least one
medication error was observed in up to 67% of patents at hospital
admission and nearly 60% of these errors were clinically important.2
Adverse drug event associated with medication error can prolong hos-
pital stays, lead to emergency visits and hospital readmissions, and
increase use of healthcare resources.3 Medication error can occur as a
result of medication discrepancy due to poor medication history taking
between transitions of care.4 Obtaining accurate patients’ medication
adverse health outcomes and improve patient safety.
Medication reconciliation (MR) is a formal process for creating the
most complete and accurate list possible of a patient's current medi-
cations and comparing the list to those in patient records or medication
orders.5 The aims of MR are to avoid errors of omission, duplication,
incorrect doses or timing, and adverse drug-drug or drug-disease in-
teractions. One of the well-recognized MR practices suggested four
steps as follows: 1) verify (collect a current medication list); 2) clarify
(make sure the medications and doses are appropriate); 3) reconcile
(compare new medications with the list and document changes in the

∗
Corresponding author. Naresuan University, Phitsanulok, 65000, Thailand.
∗∗
Corresponding author. Naresuan University, Phitsanulok, 65000, Thailand.
E-mail addresses: daraneec@nu.ac.th (D. Chiewchantanakit), teerapond@nu.ac.th (T. Dhippayom).

https://doi.org/10.1016/j.sapharm.2019.10.004
Received 7 March 2019; Received in revised form 20 September 2019; Accepted 6 October 2019
1551-7411/ © 2019 Published by Elsevier Inc.

Please cite this article as: Daranee Chiewchantanakit, et al., Research in Social and Administrative Pharmacy,
https://doi.org/10.1016/j.sapharm.2019.10.004
D. Chiewchantanakit, et al.
orders), and 4) transmit (communicate the updated and verified list to
the appropriate caregivers).6 A number of patient safety organizations
has adopted and promoted MR as a medication safety strategy in their
guidelines and recommendations.7
Several studies have showed beneficial effects of MR on a wide
range of outcomes. Specifically, findings from a recent systematic re-
view and meta-analysis indicated that MR reduced the risk of patients
with medication discrepancies by 42%.8 In addition, reduction of 67%,
28% and 19% in adverse drug event-related hospital revisits, emer-
gency department (ED) visits, and hospital readmissions were observed
in patients who received MR service.9 However, the existing evidence
was mainly derived from developed countries where medication in-
formation sharing are feasible or electronic medication records are
accessible with an aid of information technology.10
Like many other low-middle-income countries (LMIC), access to
electronic medication records and shared medication information
especially across healthcare settings in Thailand is limited. Therefore,
findings from other developed countries might not be applicable to Thai
context. In addition, a variation of MR effects on medication error was
observed in relevant Thai literature.11–13 These warrant the need to
have a conclusive evidence to support the implementation of MR in
Thailand, which could be generalized to other LMIC whose contexts are
similar to Thailand. The aim of this study is therefore to systematically
review and evaluate the effect of MR on medication error in Thailand.
Methods
Study search
This systematic review and meta-analysis was aligned with the
Preferred Reporting Items for Systemic Review and Meta-Analyses
(PRISMA) guideline.14 A systematic search was performed in electronic
databases including PubMed, CENTRAL, CINAHL, Scopus, Thai Jour-
nals Online, Thai Index Medicus, Thai Medical Index, and Health Sci-
ence Journal in Thailand from the inception of database to January
2018. Grey literature was also searched via the following databases:
Thai Thesis Database, Thai Library Integrated System (ThaiLIS), Thai-
Journal Citation Index Centre (TCI). Keywords for searching relevant
research included three domains: “Medication Reconciliation” AND
[“Medication Discrepancy” OR “Medication Discrepancy” OR Re-
hospitalization OR Readmission OR Hospitalization OR Admission OR
“Unplan visit”] AND Thailand.
Study selection
Full text studies of randomized controlled trials (RCT), non-RCT,
and controlled before-after studies (CBA) were included if they met the
following inclusion criteria: 1) assessed the effect of MR, either in-
patient or outpatient care, in Thai hospitals; 2) compared MR service
with usual care; and 3) reported the occurrence of medication error.
Both inpatient and outpatient studies were included in order to capture
a comprehensive evidence of MR in Thailand. Studies conducted using
MR as part of other services were excluded. Study selection was done
independently by two reviewers (AM and NP). Any disagreements were
discussed and resolved by the third reviewer (DC).
Data extraction
Data from the included studies were extracted based on the re-
commendation of the modified Cochrane Effective Practice and
Organization of Care Group (EPOC) guideline15 using a modified data
record form. Extracted data included study design; study centers and
location; length of follow-up; randomization process; number of pa-
tients; age of patients; gender; inclusion and exclusion criteria; number
of drug items; components of MR process; characteristics of usual care;
outcomes of interest such as scope of medication error, how it was
2
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
measured and its findings. Data extraction was performed in-
dependently by two reviewers (AM and NP) and verified by DC and TD.
Quality assessment
The EPOC tool15 was used to assess the risk of bias of each included
study. The EPOC consists of 9 domains including 1) sequence genera-
tion, 2) allocation concealment, 3) similar baseline outcome measure-
ments, 4) similar baseline characteristics, 5) incomplete outcome data,
6) prevent knowledge allocation, 7) protect against contaminating, 8)
free from selective outcome reporting, and 9) free from other risks of
bias. Studies that reported differences in qualification/competency of
pharmacist between study groups were considered as having high risk
for the domain of ‘other risks of bias’. Each topic was classified into
three levels of potential risk of bias: high, low and unclear risk.
The following risk of bias domains were justified as key domains for
the summary assessment of the risk of bias within a study: “similar
baseline characteristics” and “free from other risks of bias”. Since EPOC
can be used to assess risk of bias for RCT, non-RCT, and controlled
before-after studies. This means studies other than RCT would score
high risk of bias for sequence generation and allocation concealment
domains by default. However, the main purpose of these processes were
to avoid bias in allocating subjects into different study groups so that
both groups would have similar characteristics and less confounders.
Therefore, to overcome the problem of high risk of bias in non-RCT
studies during allocation process, “similar baseline characteristics” was
chosen as one of the key domains to determine the overall risk of bias of
individual study. According to the nature of service intervention study,
it is unlikely to “blind” or prevent knowledge of allocation for service
providers and subjects. This domain could be scored as having low risk
of bias if the outcome of interest is measured objectively or there is a
method to assure that the assessor could reliably measure the outcome.
To offset the potential risk of not be able to prevent knowledge allo-
cation, the qualification/competency of pharmacist were assessed if
they were capable to measure the outcome in a reliable manner. This
domain was added into “other risk of bias” domain and made one of the
key domains to evaluate the overall risk of bias in the present study.
The risk of bias summary for each study was classified as low risk
(low risk of bias for all key domains), high risk (high risk of bias for one
or more key domains), or unclear risk (unclear risk of bias for one or
more key domains). Quality assessment was undertaken by two re-
viewers (AM and NP) and verified by the third reviewer (PD).
Outcomes and statistical analysis
The study outcome was the number of patients with medication
error. Meta-analyses were performed under a random-effects model to
determine the effect of MR on medication error compared to usual care
and stratified by care transitions. The relative risk (RR) and its corre-
sponding 95% confidence interval (CI) were reported. Heterogeneity
across studies was assessed with I2- statistic which was expressed as %
of the variance of the overall analysis. Thresholds of I2 were interpreted
in accordance with the magnitude and direction of effects and strength
of evidence of heterogeneity (i.e. p-value) as follows: might not be
important (0%–40%); moderate heterogeneity (30%–60%); substantial
heterogeneity (50%–90%); and considerable heterogeneity
(75%–100%).16 A sensitivity analysis by including only RCT was per-
formed to explore the robustness of main findings. Subgroup analyses,
stratified by different transitions of care, were also performed to ex-
plore the effect modifiers of different baseline, which included hospital
types and number of drug items. Hospitals in Thailand are classified
into four groups: 1) sub-district health promoting hospitals, which are
capable of providing primary care with outpatient service in village and
sub-district level, and generally have no doctor on duty for the entire
time; 2) primary care hospitals, also known as community hospitals,
which are located in districts and are usually limited to providing
D. Chiewchantanakit, et al.
Fig. 1. PRISMA flow diagram of selected articles.
primary care treatment with a capacity of 10–120 beds; 3) secondary
care hospitals, also known as general hospitals, which are located in
province capitals or major districts, and are capable of providing sec-
ondary care treatment with a capacity of 120–500 beds; and 4) tertiary
care hospitals, also known as regional hospitals, which are located in
province centers or major cities, and are capable of tertiary care pro-
vided by a comprehensive set of specialists with a capacity of at least
500 beds. All hospital types, except sub-district health promoting hos-
pitals, have facilities to provide emergency, outpatient, and inpatient
services. All meta-analyses were conducted using STATA® version 15.0
3
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
(STATA Corp, College Station, TX, USA).
Results
Identification and selection of studies
The search of published and grey literature in various databases
yielded 107 articles after duplicates were removed (Fig. 1). The re-
maining articles were screened through titles and abstracts, of which 70
were removed because of the irrelevance to MR in Thailand. This
 D. Chiewchantanakit, et al.

Table 1
Characteristics of included studies.
Study Study design Setting Sample size Age (years)a Patients' conditions according to the No. of drug Comparators Study Scope of medication error
inclusion criteria items period
Int Ctrl

Kitjaorapin Quasi- Internal medicine 100 100 62.9 ± 15.7 Admitted and discharged at internal 4.6 ± 2.4 (Int), Non-specified HC provider 1 mo Allergy to drug ordered; commission error;
(2009)11 experiment ward medicine wards with underlying 4.7 ± 2.2 (Ctrl) collected medication duplication; omission error; wrong dose or
disease and required continuous record frequency; wrong drug; and wrong time
medication

Mensin (2010)12 RCT Male surgery ward 45 45 64.19 ± 14.1 Admitted in surgical ward and had 13.1 ± 5.7 Nurse copied previous 6 mo Allergy to drug ordered; illegible writing;
chronic diseases that required medication record to missing information; omission error;
continuous medication medical chart/informed potential drug interaction; wrong dose;
doctor wrong drug; and wrong time

Mungmee (2010)13 Quasi- OPD 150 150 69.3 ± 9.7 Had at least one chronic disease (but 7.7 ± 2.8 Similar to intervention 5 mo Allergy to drug ordered; drug-drug
experiment not psychiatric disease) that admitted group without OMR form interaction; duplication; omission error;
to the medical or surgical ward attached to patient record wrong dose or frequency; and wrong drug

Siwichi (2008)17 RCT IPD 239 235 61.5 ± 14.5 Had at least one chronic disease and 11.2 ± 4.5 Similar to intervention 12 mo Allergy to drug ordered; omission error;
used ≥1 continuous medication group without MR form wrong dose or frequency; wrong drug;

4
attached to patient record wrong route; and wrong time

Srisupanwitaya Quasi- Female surgery ward 45 45 62.7 ± 0.7 Had chronic disease (but not 6.5 ± 3.1 Similar to intervention 9 mo Allergy to drug ordered; illegible writing;
(2010)18 experiment and OPD psychiatric disease) that required group without DMR form omission error; potential drug interaction;
continuous medication attached to patient record wrong dose; wrong drug; wrong frequency;
and wrong time

Suetrong (2005)19 RCT Female internal 117 117 66b Aged ≥20 years old with no 6.5 Similar to intervention 4 mo Allergy to drug ordered; omission error;
medicine, surgery, psychiatric disease group without MR form wrong dose or frequency; wrong drug; and
and emergency attached to patient record wrong time
medicine wards

Thorchoo (2016)20 Quasi- OPD 97 96 60.2 ± 13.1 Aged ≥20 years old with one of the 5b Non-specified HC provider 3 mo Allergy to drug ordered; commission error;
experiment following medical conditions (with no collected medication duplication; omission error; potential drug
psychiatric disease): diabetes, HTN, record but not attached to interaction; unintentional changed of
asthma and COPD and required patient record medication; wrong dose or frequency;
continuous medication wrong drug; and wrong time

DMR: Discharge medication reconcile; HC: Health care; IPD: Inpatient Department; ME: medication error; OMR: Outpatient medication reconcile; OPD: Outpatient Department; pt: patient; RCT: Randomized controlled
trial.
a
Presented as mean ± SD.
b
Presented as median.
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
D. Chiewchantanakit, et al.
resulted in 37 articles being full text reviewed for eligibility. Of those,
30 articles were excluded for the reasons stated in Fig. 1. Finally, a total
of 7 studies11–13,17–20 was included in this systematic review for qua-
litative synthesis.
Characteristics of included studies
There were a total of 1581 patients included in the 7 studies with
the mean age ranging from 60.2 to 69.3 years (Table 1). Three stu-
dies12,17,19 were RCTs and the other four studies11,13,18,20 were non-
RCTs with quasi experimental design. A wide range of study period was
observed, ranging from one to twelve months. Four studies11,12,17,19
explored the effects of MR in hospitalized patients, whilst two stu-
dies13,20 delivered MR to patients at outpatient department (OPD) and
the remaining study18 expanded MR service from inpatient department
(IPD) to OPD follow-up visit. The study setting can be classified into
two groups based on hospital types, i.e. primary care hospital (3 stu-
dies)11,17,20 and secondary care hospitals (4 studies).12,13,18,19 Almost
all (6 out of 7) studies11–13,17,18,20 recruited patients with at least one
chronic medical condition or required continuous medication. The
average number of drug items incurred by participants among the in-
cluded studies ranged from 4.6 to 13.1. The following criteria were used
in all included studies to justify medication error: allergy to drug or-
dered, omission error, wrong dose and wrong drug.11–13,17–20 Drug in-
teraction and duplication were used to denoted medication error in
four12,13,18,20 and three studies,11,13,20 respectively. Only two studies
reported that commission error11,20 and illegible writing12,18 were
classified as medication error.
Detail of MR service
Pharmacist was the main healthcare provider that delivered MR
service in all included studies, with an exception of two studies11,18 that
had doctor and/or nurse took part in providing this intervention
(Table 2). In verification step, almost all studies collect a current
medication list from hospital medical record and medicines brought to
hospital by participating patients as well as interview patients or carers.
Clarification step was reported in five11,13,17–19 out of seven studies. All
five studies11,12,17–19 that were conducted in hospitalized patients had
MR provided in multi-transitions of both admission and discharge.
Moreover, two studies11,19 also reported additional reconciliation
during hospitalization. During MR transmission step, all included stu-
dies attached MR form with patient record. Patients in the usual care
either have their medication record collected by healthcare provider
(three studies) or received service similar to the intervention group
without having MR form attached to patient record (four studies)
(Table 1).
Quality of included studies
All three RCTs12,17,19 reported appropriate sequence generation, but
did not give enough information on allocation concealment process,
hence scored unclear risk of bias in this domain (Table 3). The re-
maining four non-RCTs11,13,18,20 were justified as having high risk of
bias in sequence generation and allocation concealments domains. Bias
for baseline outcome measurements could not be evaluated as none of
the seven studies reported baseline outcome measured. All four non-
RCTs11,13,18,20 were deemed as having low risk from contamination as it
is unlikely that the control groups received the intervention because
they were patients who received usual care before the introduction of
MR intervention in the study settings. No studies reported the compe-
tency of healthcare practitioners who delivered MR services or provided
detail of any given training course to ensure the quality and consistency
of service intervention. Therefore, all seven studies were scored unclear
risk in “other risk of bias” domain. Since this domain was set as one of
the key domains to estimate the summary risk of bias within each study,
5
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
scoring unclear risk of bias in this domain had turned the summary risk
of bias to unclear risk for all included studies.
Overall effect
The overall effect was a pooled estimate of relative risk from studies
that reported the total number of patients who were presented with
medication error following MR intervention at the end of study period
from all care transitions. Of the seven included studies, one study11 only
presented the number of patients who were classified as having medi-
cation error during MR at admission and discharge. The total number of
patient at the end of study period that can be used to justify the overall
effect of MR during hospitalization was not reported. This study was
therefore not included in the main analysis (Fig. 2). Overall, pooled
estimate of findings from six studies (1381 patients) showed that those
who received MR were less likely to encounter medication error com-
pared to those who had usual care (RR 0.25; 95%CI, 0.15–0.43) with
substantial heterogeneity (I2 = 71.9%, p = 0.003). A sensitivity ana-
lysis that included only three RCT studies (798 patients)12,17,19 showed
similar findings with the main analysis as the RR of having medication
error in patients receiving MR was 0.30 (95%CI, 0.16–0.56) compared
with usual care groups.
Effects on hospitalized patients
Findings from 5 studies11,12,17–19 involving 1088 hospitalized pa-
tients showed that MR at IPD admission reduced the risk of medication
error by 63% (RR 0.37, 95%CI, 0.20–0.65) of usual care service with
substantial heterogeneity (I2 = 78.8%, p = 0.001) (Table 4). Similarly,
patients receiving MR at IPD discharge were less likely to have medi-
cation error (RR 0.27, 95%CI, 0.17–0.43) compared to those who re-
ceived usual care with low heterogeneity (I2 = 13.0%, p = 0.331).
Effect on medication error was greater when MR was provided in
secondary care hospitals compared with primary care hospitals both
during admission (RR 0.49 [95%CI, 0.34–0.69] vs RR 0.25 [95%CI,
0.05–1.26]), and discharge transition (RR 0.19 [95%CI, 0.09–0.39] vs
RR 0.30 [95%CI, 0.12–0.79]). However, inconsistent findings were
presented at different transitions when studies were classified based on
number of drug items. Specifically, the RR of medication error fol-
lowing MR during admission among studies on patients who possessed
≥10 drug items (0.23 [95%CI 0.06–0.95]) was lower than the RR from
studies on patients with less than 10 drug items (0.52 [95%CI
0.38–0.71]). On the other hand, when MR was provided during dis-
charge, studies on patients who were prescribed with ≥10 drug items
showed higher RR compared with their counterparts, i.e. 0.33 [95%CI,
0.15–0.73] vs 0.18 [95%CI, 0.09–0.37].
Effects on ambulatory patients
The benefit of MR in reducing medication error by 83% has been
observed when it was provided to ambulatory patients (RR 0.17
[95%CI, 0.04–0.80]), which was accompanied by moderate hetero-
geneity (I2 = 59.2%, p = 0.086). Unlike the effects found in hospita-
lized patients, the reduction in medication error following MR to am-
bulatory patients was only profound in primary care hospitals (RR 0.06
[95%CI, 0.01–0.24]) whereas this effect was not significant in sec-
ondary care hospitals (RR 0.34 [95%CI, 0.07–1.58]). It was not possible
to perform a subgroup analysis based on number of drug items for
studies in OPD setting as no studies was conducted in patients who used
≥10 drug items.
Discussion
Results from this study reiterate findings from others,8–10,21 which
were mainly derived from high income countries, and strengthen the
beneficial effects of MR in reducing the risk of medication error. It
 Table 2
Characteristic of medication reconciliation among the included studies.
D. Chiewchantanakit, et al.

Study Verification – information source Clarification Reconciliation Transmission

HC provider Hospital Medicines Interview pt Others Admission/OPD Hospitalization DC Hospitalization/OPD DC

medical brought to or carer
recorda hospital

Kitjaorapin Doctor, nurse Yes Yes Yes Referring form; Clarified by Compared MR Compared MR form Two times MR form (in pink MR form and drug
(2009)11 and pharmacist patients' medical pharmacist (no form with order with doctor order (before doctor paper) attached on the profile given to
booklet detail given) on admission ordered and front page of patient patient when
before record discharge
dispensed)c

Mensin (2010)12 Pharmacistb NR NR NR NR NR Two times (after No Two times (after AMR form attached to DMR form attached to
doctor ordered doctor ordered patient record patient record
and before and before
dispensed)d dispensed)d

Mungmee Pharmacistb Yes Yes Yes NR Clarified by Compared OMR NA NA OMR form attached on NA
(2010)13 pharmacistb (no form with the front page of
detail given) prescribing order patient record

Siwichi (2008)17 Pharmacist Yes Yes Yes HC practitioner Clarified by Compared AMR No Compared DMR AMR form attached to DMR form attached to
from other HC pharmacist (no form with order form with DC patient record patient record

6
settings; patients' detail given) on admission order
medical booklet

Srisupanwitaya Nurse Yes Yes Yes NR Pharmacistb Compared AMR No Compared DMR AMR form attached to DMR form attached to
(2010)18 (admission) and clarified AMR form with order form with DC patient record patient record and DC
pharmacist taken by nurse at on admission order medication list form
(before DC) admission inserted in OPD
patient record

Suetrong (2005)19 Pharmacist Yes Yes Yes NR Clarified with pt Compared AMR Compared TMR form Compared DMR AMR and TMR form DMR form attached to
and/or their form with order with order in the form with DC attached to the front patient record
relatives on admission transferred ward order page of IPD patient
record

Thorchoo (2016)20 Pharmacist Yes NR Yes NR NR Physician NA NA MR form attached to NA

reviewed MR patient record
form before
prescribing

AMR: Admission medication reconcile; DC: Discharge; DMR: Discharge medication reconcile; IPD: Inpatient department; MR: Medication reconcile; NA: Not available; NR: Not report; OMR: Outpatient medication
reconcile; OPD: Outpatient department; TMR: Transferred medication reconcile.
a
Including hospital computer database and records from pharmacy department.
b
The investigator act as a pharmacist to provided MR intervention.
c
Two times by doctor before prescribed and pharmacist at pharmacy room before dispensed.
d
Two times by the pharmacist investigator after doctor ordered and pharmacist at pharmacy room before dispensed.
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
D. Chiewchantanakit, et al. Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx

Table 3
Risk of bias of included studies.

*Highlighted columns represent key domains; †High risk if missing data between intervention and control group > 5%; ‡Other sources of bias included qualification/
competency of pharmacist or training before conducting intervention; H=High risk of bias; L = Low risk of bias; U=Unclear risk of bias.

Fig. 2. Overall effects of medication reconcilation on reported medication errors.

should be noted, however, that the pooled estimate of overall effects of
MR from all care transitions should be interpreted with caution as it
was presented with a substantial heterogeneity from studies with un-
clear risk of bias. Nonetheless, what this study adds to the current lit-
erature is that the benefit of MR is well established in setting like
Thailand where access to shared patient information is limited. As
hospital care practice varies across countries, whether the effects of MR
in reducing medication error are prominent in other LMIC requires
further exploration. In addition, when stratified by transition of care,
this study revealed that the highest magnitude of risk reduction was
achieved when MR was performed to ambulatory patients. Subgroup
analyses also showed that the effect of MR in reducing risk of ME in
hospitalized patients was prominent in secondary care hospitals rather
than primary care hospitals. On the other hand, the effect on ME in
ambulatory patients was only evident in primary care hospitals, not
secondary care hospitals. However, the effect of MR on patients who
used 10 or more drug items is still inconclusive.
According to a previous systematic review on MR process,7 it is
difficult to identify best practice for MR as current evidence show
heterogeneity between MR interventions and how they were evaluated.
However, it is suggested that MR should encompass four main steps:
verify, clarify, reconcile and transmit.6 These four steps have been
covered by almost all included studies, indicating that MR in Thailand
is well aligned with international concept and practice of MR.
Medication discrepancies, deemed to be one type of medication
error, has been used interchangeably with medication error in several
studies.7 As there are limited number of meta-analysis study on the
effect of MR on medication error, evidence on medication discrepancies
is used to compare with findings from the present study. The positive
effect of MR on the reduction of medication error coincided with
findings from previous meta-analysis studies. However, when compare
with usual care, the risk of having mediation error among patients in
MR group showed in this study (RR 0.25) is lower than those reported
in previous meta-analysis studies (RR 0.3421 and 0.588). One possible

7
D. Chiewchantanakit, et al. Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx

Table 4
Effects of medication reconcilation on medication errors in subgroup meta-analysis stratified by different transitions of care.
Outcomes Pooled RR (95%CI) Heterogeneity, I2%, p-value No. of studies No. of studied population

11,12,17–19
Hospitalized – Admission 0.37 (0.20–0.65) 78.8%, p = 0.001 5 1088

Hospitals types
Primary care hospitals 0.25 (0.05–1.26) 93.6%, p = 0.000 211,17 674
Secondary care hospitalsa 0.49 (0.34–0.69) 0.0%, p = 0.801 312,18,19 414
Number of drug items
Drug items ≥10 0.23 (0.06–0.95) 89.7%, p = 0.002 212,17 564
Drug items < 10 0.52 (0.38–0.71) 0.0%, p = 0.806 311,18,19 524

Hospitalized – Discharge 0.27 (0.17–0.43) 13.0%, p = 0.331 511,12,17–19 1088

Hospitals types
Primary care hospitals 0.30 (0.12–0.79) 61.9%, p = 0.105 211,17 674
Secondary care hospitalsa 0.19 (0.09–0.39) 0.0%, p = 0.863 312,18,19 414
Number of drug items
Drug items ≥10 0.33 (0.15–0.73) 47.4%, p = 0.168 212,17 564
Drug items < 10 0.18 (0.09–0.37) 0.0%, p = 0.861 311,18,19 524

Ambulatory 0.17 (0.04–0.80) 59.2%, 0.086 313,18,20 583

Hospitals types
Primary care hospitals 0.06 (0.01–0.24) N/A 120 193
Secondary care hospitalsa 0.34 (0.07–1.58) 22.8%, p = 0.255 213,18 390
Number of drug items
Drug items ≥10 N/A N/A N/A N/A
Drug items < 10 0.17 (0.04–0.80) 59.2%, 0.086 313,18,20 583

a
Included general and regional hospitals.

explanation for the higher impact of MR showed in this review is that
the provision of MR in all fours studies in hospitalized patients was at
multi-transitions, i.e. admission and discharge. Findings from Me-
konnen et al.,21 on the other hand, was derived from studies on single
transitions whereas Cheema et al. study8 did not clearly define whether
the estimate effect was pooled from studies on MR at single or multiple
hospital transitions. In addition, the MR interventions in all of the in-
cluded studies in the present review have transmitted the verified
medication list to healthcare providers by attaching a dedicated unique
MR form on patient records. This may attract attention from clinicians
and prevent medication errors in the next hospital transition such as at
discharge.
The effect of MR in reducing medication error among studies in
ambulatory patients was slightly higher than those studies in hospita-
lized patients both at admission and discharge transitions. This is
probably due to the scope of medication error measured by studies in
ambulatory patients which was wider than the scope used among stu-
dies in hospitalized patients. Specifically, all three studies in ambula-
tory patients13,18,20 identified drug interaction as medication error,
whereas only one12 out of five studies in IPD reported drug interaction
in the scope of their medication error measures. In addition, duplication
was counted as medication error in two13,20 out of three studies in
ambulatory patients compared to one11 out of five studies in hospita-
lized patients. It should be noted, however, that the scope and defini-
tion of medication error is still inconclusive.22 A goal standard criteria
and definition of medication error is needed to better compare effects
on medication error among studies.
In general, patients admitted in secondary care settings are char-
acterized with more complex medical conditions than those admitted in
primary care hospitals. Accordingly, it is expected that the risk of
medication error is higher among patients admitted in secondary care
setting which indicates more room of improvement than primary care
premises. This could explain the reasons why the effect of MR was
prominent in secondary care rather than primary care hospitals as
shown in this study. Conversely, findings in ambulatory patients
showed that MR in primary care hospitals could lessen the risk of
medication error than in secondary care hospitals. Perhaps, pharmacists
would have more time to conduct effective MR in primary care
hospitals as they generally have smaller number of visiting patients
compared to secondary care hospitals. This conjecture, however, re-
quires further investigation.
One of the strengths of this study is that it stratified findings based
on hospital transitions, i.e. IPD-admission, IPD-discharge, and OPD to
better illustrate the effects of MR at different transition care. Subgroup
analysis was also conducted which could help decision makers to target
specific transition and hospital setting where patients could get the
most benefit from MR. Although a test for publication bias could not be
conducted due to limited number of included studies, a comprehensive
search to cover relevant grey literature was performed to alleviate the
concern over the risk of publication bias.
There are some limitations that should be mentioned for this study.
First, a substantial heterogeneity was witnessed in the main analysis
which may be due to differences in study design, setting, level of hos-
pital, and scope of medication errors used among the included studies.
However, the magnitude of effect size was similar among main and
sensitivity analysis, which confirmed that the primary finding was valid
and credible. Second, the generalizability of findings from this study
should be limited to patients with chronic diseases who visit primary or
secondary care hospitals. This is because the majority of studies re-
cruited patients with at least one chronic medical condition; hence,
findings in this study could not be generalized to patients without
chronic diseases. Studies conducted in tertiary care hospitals, which
may respond to MR with different magnitude, were also not included.
Third, the evidence generated came from studies with unclear risk of
bias since the competency of service providers could not be justified.
Lastly, although there are strong associations between medication error
and detrimental clinical outcomes,2,7 the impact of MR on clinical in-
dicators could not be determined in the present study due to limited
availability of relevant literature. Future studies should overcome these
limitations by exploring the effect of MR at tertiary care hospitals using
high standard approach with RCT design, providing a comprehensive
description of the competency of service providers, and measuring
clinical outcome such as mortality, length of hospital stay, and read-
mission.

8
D. Chiewchantanakit, et al.
Conclusion
The current evidence indicates that MR reduces the number of pa-
tients experiencing medication error in both outpatient and inpatient
settings in Thailand. Appropriate strategies should be developed to
support the provision of MR with priority given to ambulatory patients.
The uptake of MR to hospitalized patients should also be promoted,
especially in secondary care hospitals.
Funding
This work was supported by Naresuan University's Faculty of
Pharmaceutical Sciences Research Fund.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.sapharm.2019.10.004.
References
1. Knez L, Suskovic S, Rezonja R, Laaksonen R, Mrhar A. The need for medication re-
conciliation: a cross-sectional observational study in adult patients. Respir Med.
2011;105:S60–S66.
2. Tam VC, Knowles SR, Cornish PL, Fine N, Marchesano R, Etchells EE. Frequency, type
and clinical importance of medication history errors at admission to hospital: a
systematic review. CMAJ (Can Med Assoc J). 2005;173:510–515.
3. Mueller SK, Sponsler KC, Kripalani S, Schnipper JL. Hospital-based medication re-
conciliation practices: a systematic review. Arch Intern Med. 2012;172:1057–1069.
4. Vira T, Colquhoun M, Etchells E. Reconcilable differences: correcting medication
errors at hospital admission and discharge. Qual Saf Health Care. 2006;15:122–126.
5. Barnsteiner JH. Medication reconciliation. In: Hughes RG, ed. Patient Safety and
Quality: An Evidence-Based Handbook for Nurses [Internet]. Rockville (MD): Agency for
Healthcare Research and Quality (US); 2008. [cited 2018 Jun 6]. Available from:.
https://www.ncbi.nlm.nih.gov/books/NBK2648/.
6. Ptasinski C. Develop a medication reconciliation process. Nurs Manag. 2007;38:18.
7. Almanasreh E, Moles R, Chen TF. The medication reconciliation process and classi-
fication of discrepancies: a systematic review. Br J Clin Pharmacol. 2016;82:645–658.
Research in Social and Administrative Pharmacy xxx (xxxx) xxx–xxx
8. Cheema E, Alhomoud FK, Kinsara ASA, Alsiddik J, Barnawi MH, Al-Muwallad MA,
et al. The impact of pharmacists-led medicines reconciliation on healthcare outcomes
in secondary care: a systematic review and meta-analysis of randomized controlled
trials. PLoS One. 2018;13:e0193510.
9. Mekonnen AB, McLachlan AJ, Jo-anne EB. Effectiveness of pharmacist-led medica-
tion reconciliation programmes on clinical outcomes at hospital transitions: a sys-
tematic review and meta-analysis. BMJ Open. 2016;6:e010003.
10. Mekonnen AB, Abebe TB, McLachlan AJ, Brien JA. Impact of electronic medication
reconciliation interventions on medication discrepancies at hospital transitions: a
systematic review and meta-analysis. BMC Med Inf Decis Mak. 2016;16:112.
11. Kitjaorapin B. Outcome of Medication Reconciliation at Medicine Wards in Kabinburi
Hospital master's thesis Bangkok: Chulalongkorn University; 2009 111 p.
12. Mensin P. Cost-Effectiveness of Pharmacist-Participation in Medication Reconciliation
Process in Surgical Inpatient at Samuprakarn Hospital master's thesis Bangkok:
Chulalongkorn University; 2010 113 p.
13. Mungmee S. Medication Reconciliation in Outpatient Visiting Medicine and Surgery
Clinics at H.M. Queen Sirikit Hospital master's thesis Bangkok: Chulalongkorn
University; 2010 100 p.
14. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. Ann Intern Med.
2009;151:264–269.
15. Staresinic AG, Sorkness CA, Goodman BM, Pigarelli DW. Comparison of outcomes
using 2 delivery models of anticoagulation care. Arch Intern Med.
2006;166:997–1002.
16. Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions
Version 5.1.0 [updated March 2011]: The Cochrane Collaboration. 2011; 2011. [cited
2019 Mar 15]. Available from: http://handbook-5-1.cochrane.org/.
17. Siwichi W. Outcomes of Medication Reconciliation on Medicatuon Error in the Inpatient
Setting at Maetha Hospital, Maetha District, Lamphun Province master's thesis Chiang
Mai: Chiang Mai University; 2008 116 p.
18. Srisupanwitaya S. Outcomes of Medication Reconciliation Complete Process in Surgical
Chronic Disease Patients at Samutprakarn Hospitalt Samutprakarn Hospital master's
thesis Bangkok: Chulalongkorn University; 2010 95 p.
19. Sue-trong C. Developing and Implementing Medication Reconciliation Process Following
Transition Points in Inpatient Medication System. Bangkok: Chulalongkorn University;
2005 136 p.
20. Thorchoo S, Upakdee N. Clinical and economic outcomes of medication reconcilia-
tion at outcome department, Nongki hospital, Burirum. FDA Journal. 2016;28:34–44.
21. Mekonnen AB, McLachlan AJ, Brien JA. Pharmacy-led medication reconciliation
programmes at hospital transitions: a systematic review and meta-analysis. J Clin
Pharm Ther. 2016;41:128–144.
22. Ferner RE, Aronson JK. Clarification of terminology in medication errors: definitions
and classification. Drug Saf. 2006;29:1011–1022.