Clinical impact of medication reconciliation on physicians’ treatment policies at outpatient clinics: a cross-sectional study

Abstract

Background

Guidelines recommend implementing medication reconciliation (MR) at every healthcare transition. However, limited research has been conducted in outpatient clinics to explore whether MR affects the physicians’ decisions. This study aims to evaluate the clinical impact of MR in these settings, specifically investigating whether MR leads to adjustments in treatment policies. The study also aims to assess the severity of potential harm avoided by these adjustments.

Methods

A cross-sectional study was performed between 1 June 2019 to 18 April 2022 at Maastricht University Medical Center+, The Netherlands, involving 20 different outpatient clinics. All patients who had an appointment at one of the involved clinics were assessed for eligibility (n = 3,031). Pharmacy technicians obtained the best possible medication history (BPMH) through MR. The BPMH was shared with the physician after the patient’s visit. The physician could adjust his initial treatment policy based on the new information provided with the BPMH. The physician scored the severity of these policy adjustments, using the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) index.

Main outcomes

The primary outcome was the proportion of patients with a policy adjustment. Secondary outcomes included the proportion of patients with potential severe harm and the physician’s perceived value of the BMPH for possible future treatment. Descriptive data analysis was used.

Results

In total, 2,289 patients were included (mean age: 59, 47.6% male). MR in outpatient clinics resulted in treatment policy adjustments by physicians in 3.8% of patients, with physicians deeming 0.5% of all MRs necessary to prevent severe harm. Physicians considered the BPMH valuable for future treatment in 13.8% of patients.

Conclusion

Routine medication reconciliation for all patients at every outpatient visit is not justified by this study. Future research should focus on identifying high-risk patient who would benefit most from MR in outpatient clinics.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12913-026-15021-z.

Introduction

Inadequate knowledge of patients’ actual medication use poses a significant risk during healthcare transitions, particularly upon hospital admission [1–8]. A Cochrane review, based on 20 studies, highlighted that 559 out of 1,000 patients are vulnerable to experiencing one or more medication discrepancies during care transitions, defined as a difference between a patient’s actual medication use and the medication listed in the electronic health record (EHR) [9]. In addition, a systematic review reported that 11% to 59% of these discrepancies occurring at hospital admission could potentially result in harm [10].

To reduce potential harm due to medication discrepancies during care transitions, medication reconciliation (MR) has emerged as a widely accepted method. While numerous definitions of MR exist in the literature, for the purposes of this study we adopted the following definition: creating an accurate list of all medications a patient is taking and comparing it with those prescribed by the clinician (i.e., those listed in the patient’s chart or electronic health record), while also considering allergies .The ultimate goal of MR is to provide the patients with the right medication at all healthcare transitions [11]. The World Health Organization (WHO) emphasizes the importance of MR in its report “Medication Safety in Transitions of Care,” recommending its implementation at every healthcare transition to reduce medication errors, defined as errors in the prescribing, dispensing, administration, or documentation of a drug that could lead to harm, by 50% [12].

Multiple studies have demonstrated that implementing MR during hospitalization can significantly reduce medication discrepancies and prevent adverse events such as emergency department visits and hospital readmissions [1, 5, 13–18]. However, limited research has been conducted in outpatient clinics, i.e. the ambulatory setting. Existing studies in outpatient clinics primarily focus on identifying discrepancies [19–21]. These studies do not thoroughly assess to which extent these discrepancies result in patient harm or whether MR leads to changes in medical policies. In addition, implementing WHO’s advice of performing MR for every outpatient clinic patient poses a challenge due to the high number of patients attending these clinics within a short timeframe. Furthermore, MR is a time-consuming and therefore costly process while the healthcare system faces workforce shortages. Before implementing MR on a large scale for all outpatient clinic visits, it is crucial to determine the clinical impact of MR and the treatment adjustments hospital physicians might perform when having an accurate medication list.

Therefore, this study aims to evaluate the clinical impact of MR in outpatient clinics, defined as MR leading to treatment policy adjustments (e.g. change of medication or laboratory controls), and to assess the severity of potential harm avoided by these adjustments in relation to patients’ health.

Methods

Design and setting

A cross-sectional study was performed at Maastricht University Medical Center + in the Netherlands (MUMC+, 715 beds) between June 2019 and April 2022. In this study, 11 departments with 20 different outpatient clinics participated (the clinics are listed in additional file 1). Patient data were obtained and handled in accordance with privacy regulations and Dutch legislation. We obtained oral informed consent for MR, data collection and data analysis of all the patients included in this study. The informed consent was obtained immediately prior to the MR by phone.

Study population

The 20 outpatient clinics included both surgical and nonsurgical outpatient clinics. All ambulatory patients, both adults and children, who had an appointment at one of these clinics were eligible for inclusion. Patients who could not be reached by phone to perform a MR interview on two different days were excluded. Individuals who were unable to participate in the MR interview, e.g. due to language barriers, were also excluded. Furthermore, patients who revisited either the same outpatient clinic or a different one were excluded.

Study procedure

In the Netherlands, it is mandatory to perform MR prior to prescribing or dispensing medication and can be reimbursed, depending on the contracts of individual health insurers with the hospital. While the responsibility for MR ultimately lies with the prescriber, it may be delegated to a qualified professional. MR is often performed by trained pharmacy technicians (PT) with support from pharmacists. Previous studies have shown that MR conducted by PT are comparable to those performed by a pharmacist to obtain a best possible medication history (BPMH) [18, 22–28]. It has also been previously shown that MR via phone represents a reliable and time-efficient alternative to in-person MR [29]. The PTs involved in this study were all trained by a clinical pharmacist and experienced in conducting MR with patients. In addition, they were trained in the study procedure and data collection.

One week prior to the outpatient clinic visit, the community pharmacy records were retrieved by the PT through the Nationwide Medication Record System (NMRS) to obtain insight in the patient’s medication history. The NMRS is an electronic healthcare infrastructure in the Netherlands that contains nationwide pharmacy dispensing data, accessible only if the patient has provided informed consent [30]. The medication list from the NMRS, representing the patient’s medication history, was combined with the medication list present in the EHR, which reflects the physician’s records. Subsequently, the PT conducted a medication reconciliation interview by phone during which patients were asked about their current medication use and any additional medications listed in the NMRS or EHR were evaluated. The interview followed the guidelines outlined by Ebbens et al. [31]. Discrepancies between the combined medication list (NMRS plus EHR) and the patient’s actual medication use -including over-the-counter (OTC) medications- and reported allergies as stated by the patient were identified. The final medication overview was considered the BPMH.

At the outpatient clinic visit, the hospital physician established the treatment policy during the patient’s visit without having initial access to the BPMH. Following the clinic visit, the BPMH was shared with the hospital physician. The PT communicated any discrepancies and any pharmacokinetic or pharmacodynamic interactions between newly prescribed medications and the patient’s actual medication usage to the physician. Subsequently, the hospital physician assessed whether adjustments to the initially established treatment policy were necessary. These adjustments could encompass various actions, including initiating or discontinuing medication, changing the dosage of current medication, performing additional laboratory tests, or consulting with other healthcare providers. If adjustments were deemed necessary, the hospital physician utilized the National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP) index to determine the potential harm severity to the patient avoided by these adjustments [32]. The hospital physician then personally informed the patient via telephone about any adjustments to the treatment policy.

Finally, the hospital physician was asked if the BPMH was considered valuable for future treatment decision, such as providing insights into the patient’s specific medical condition or recognizing their vulnerability. The study procedure is illustrated in Fig. 1.

Fig. 1.

Flowchart of the study design. MR= medication reconciliation, NMRS= Nationwide Medication Record = System with dispensing records of community pharmacies, EHR= electronic health record, BPMH= best possible medication history, PT= pharmacy technicians, HP = hospital physician, NCC MERP= national coordinating council for medication error reporting and prevention

Data collection and classification

Three categories of parameters were collected (additional file 2 shows an overview of all parameters). Patient characteristics (e.g., age, sex) were collected from the EHR, the patient themselves and the hospital physician. Medication related characteristics (e.g., number and type of medicines) were collected during MR. The current medication was classified into high and low risk categories for causing harm according to literature [1, 32] (additional file 3 shows a list of medications identified as high-risk medication). Process-related characteristics (e.g., number of emergency room visits or hospital admissions in the past 12 months) were collected from the EHR, the patient themselves and the hospital physician.

Any BPMH resulting in a treatment policy adjustment was classified as a MR with clinical impact. The hospital physician used the NCC MERP index to score the treatment policy adjustment on potential harm [33].

Outcome measures

The primary outcome was the proportion of medication reconciliations with clinical impact.

The secondary outcomes were (1) the NCC MERP index score to assess the severity of potential harm avoided due to treatment policy adjustments based on MR and (2) the percentage of MRs for which the hospital physician indicated that the BPMH was considered valuable for future treatment.

Sample size

We used the formula provided in reference 37 to calculate the sample size. To calculate the sample size for a prevalence, the Z-statistic, expected prevalence, and precision are needed. However, the prevalence of policy adjustments due to medication reconciliation in outpatient clinics is unclear. Previous studies focussed on discrepancies and limited studies addressed the occurrence of adverse drug events (ADEs), i.e. harm occurring due to discrepancies (range 1.8% to 17.1%) [34–36]. We assumed that this harm due to discrepancies would require treatment policy adjustments and therefore used these studies to estimate the prevalence of policy adjustments. Since the population in our study consisted of patients of all ages attending 20 different outpatient clinics, we anticipated a conservative prevalence in our study of 4.4% [34, 35].

Therefore, by setting the expected prevalence at 4.4%, the precision at 1%, and the confidence level at 95%, we determined that a minimum of 1,616 MRs were required [37]. The calculations were performed using Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA).

Data analysis

The acquired data were documented in a Microsoft Access database and validated (in duplicate) by checking the data-entry by a second person. Descriptive data analysis was employed for both primary and secondary outcomes. Patient characteristics were presented as means and standard deviations for numeric variables and as numbers and percentages for categorical variables.

Results

A total of 3031 patients were approached to participate in the study. Of these 357 (11.8%) were excluded primarily because they either could not be reached (5.6%) or did not give consent for participation (4.6%). After exclusion, 2674 patients were eligible for inclusion. The primary reasons for loss to follow-up were appointment cancellations (7.1%) and the hospital physician lacking time (2.0%) to discuss MR results with the PT. The final study sample consisted of 2289 (85.6%) patients. (Fig. 2).

Fig. 2.

Flowchart of patient inclusion .MR= medication reconciliation

The study population had a mean age of 59 (SD 18.5) years and 47.6% were male (Table 1).

Table 1.

Demographic characteristics of study population (n = 2289)

		Study population (n = 2289)
Patient Characteristics
Sex	male n (%)	1089 (47.6)
Age (group)	years, mean, (sd)	59 (18.5)
	0–18 years, n (%)	71 (3.1)
	19–70 years, n (%)	1511 (66.0)
	71–100 years, n (%)	707 (30.9)
Housing	independent, n (%)	2048 (89.5)
	home care, n (%)	212 (9.3)
	institution, n (%)	29 (1.2)
Education	primary and below, n (%)	134 (5.8)
	lower secondary, n (%)	618 (27.0)
	upper secondary, n (%)	684 (29.9)
	advanced education, n (%)	735 (32.1)
	missing, n (%)	118 (5.2)
Health literacy	adequate, n (%)	1847 (80.7)
	sub-optimal, n (%)	340 (14.8)
	insufficient, n (%)	95 (4.2)
	missing, n (%)	7 (0.3)
Independent use of medication	independent, n (%)	1838 (80.3)
Patient known at visited outpatient clinic	yes, n (%) no, n (%	1693 (74.0) 596 (26.0)
Medication use characteristics
Use of any prescribed medication	yes, n (%)	2044 (89.2)
Number of medications, excluding OTCa	mean (SD) 0–4, n (%)	5.9 (4.7) 1036 (45.3)
	5–10, n (%)	856 (37.4)
	> 10, n (%)	397 (17.3)
Use of OTC medication	yes, n (%)	1409 (61.6)
Use of high risk medicationb	yes, n (%)	1691 (73.9)
Utilization of healthcare services in the previous 12 months
Visits to outpatient clinics	0, n (%) 1, n (%) 2–5, n (%)	393 (17.1) 331 (14.5) 803 (35.1)
	> 5, n (%)	762 (33.3)
Hospitalization	yes, n (%)	714 (31.2)
Emergency admission	yes, n (%)	762 (33.2)
Multiple specialties involved	yes, n (%)	1836 (80.2)

^a number of medications established during MR (without OTC = over-the-counter)^b additional file 3 displays medication classified as high risk medication

A total of 2044 (89.2%) of the included patients used prescribed medication with a mean of 5.9 prescribed medications (SD 4.7). In total, 73.9% used high-risk medication (see additional file 3).

Of the 2289 MRs included, the physicians adjusted the initial treatment policy in 88 (3.8%) patients after receiving the BPMH. Physicians classified 11 (13%) of the policy adjustments as necessary to prevent severe harm (MERP index score F to I) and 69 (78%) as resulting in no or minor harm (MERP index score C to E), see Table 2.

Table 2.

MR with clinical impact according to MERP index (n = 2289)

MERP Classification*	n (%)	Example treatment policy adjustment
C	40 (1.7%)	Discontinuation OTC medication: glucosamine/chondroitin Requesting lab work
D	14 (0.6%)	Increasing candesartan dosage as dose in the EHR was incorrect Ordering a DEXA scan
E	15 (0.7%)	Discontinuation of prescribed medication: apixaban, metoclopramide Adjustment of a medical procedure (using a smaller needle) due to clopidogrel usage
F	5 (0.2%)	Not prescribing carbamazepine based on interaction with apixaban (another anti-epileptic was chosen).
G	3 (0.1%)	Starting medication: bisphosphonate for a patient at high risk
H	2 (0.1%)	Replacing of oral miconazole with nystatin based on an interaction with a coumarin
I	1 (0.04%)	Starting insulin (The patient, contrary to what the physician believes, was not using insulin)
Not classified	8 (0.3%)
All	88 (3.8%)

*Options A and B do not occur due to the study design; by definition, the error has reached the patient because the physician had already established treatment policy at the time of MERP classification.

In eight (9%) patients the potential harm was not classified due to a human error. When assuming the eight unclassified MRs in patients as a worst case scenario, the maximum amount of policy adjustments to prevent severe harm would still be low (11 and 8 unclassified patients resulting in 19 policy adjustments, equalling 0.8% of all MRs in patients). Treatment policy adjustments avoiding severe harm were found in 0.4% of patients younger than 70 years old and in 0.6% of patients aged 70 years and over (6 patients versus 5 patients respectively). Additionally, these adjustments were observed in 0.2% of patients using less than 5 medications and in 0.7% of patients using 5 medications or more (2 patients versus 9 patients respectively).

Of all MRs, 2.8% involved medication adjustments and 1% were non-pharmacological. Medication interventions were diverse, most frequently affecting analgesics, proton pump inhibitors and antihypertensive drugs. Non-pharmacological interventions included, among others, monitoring of laboratory results and conducting consultations regarding medication adherence.

The physicians indicated the BPMH to be meaningful for future treatment in 316 patients (13.8%), including the 88 (3.8%) patients in which the treatment policy was adjusted. The BPMH was meaningful, for example, due to acquired knowledge regarding a patient’s vulnerability.

Discussion

In this study, we have demonstrated that medication reconciliation performed in outpatient clinics results in a change in treatment policy in 3.8% of patients. Furthermore, this study has shown that only in 0.5% of all MRs the adjustment of the treatment policy was deemed necessary by the physician to prevent severe harm. Finally, the study has revealed that, in addition to the 3.8% of patients in which the policy was adjusted, in another 10.0% of patients physicians perceived the BPMH to be meaningful for future treatment (thus, in total for 13.8% of the 2289 patients).

To our knowledge, this is the first study to address the impact of MR on treatment policy by employing a BPMH at outpatient clinics. Previous studies in outpatient clinics have focused on discrepancies between the patient’s actual medication use and the medication listed in the EHR [19, 38–40]. In these studies, the reported percentages of patients experiencing discrepancies varied from 47% to 83% [19, 40]. In three studies the discrepancies were scored on their ability to cause potential harm and this ranged from approximately 32% to 71% [38–40]. In our study much lower proportions of potential severe harm were found. This could be due to differences in patient population. In our study 11 departments with 20 different outpatient clinics participated, whereas other studies focused on a specific population. Furthermore, we focused on the changes that physicians actually assessed as necessary rather than potential harm. The results of our study are more in line with studies that assessed ADEs as these focused on actual harm that occurred (see sample size analysis).

A meta-analysis found that preventable medication-related harm occurs in 3% of patients across medical care settings and at least a quarter of preventable medication-related harm was severe or potentially life threatening [41]. In this study, we have illustrated that MR in outpatient clinics prevents potentially severe harm, as indicated by physicians, in only 0.5% of patients (this would be 0.8% if the missing cases were considered severe in a worst-case scenario). It is important to note that this percentage is expected to be even lower as our study did not measure actual harm and did not consider whether an error would have been identified and rectified further along the care continuum. For example, physicians considered policy adjustments prompted by drug–drug interactions necessary to prevent potentially severe harm. Yet, community pharmacists would also be alerted to these interactions during dispensing and could intervene accordingly. In light of the small proportion of patients where potentially severe harm could have been prevented and taking into account limited availability of resources both in terms of funding and personnel, the question arises as to whether it is justified to perform MR for all patients visiting an outpatient clinic [42]. On the other hand, given the large number of outpatient clinic visits, 0.5% still represents a significant number of patients for whom MR is of added value; for example, at MUMC+, the number of outpatient clinic visits was 361,007 in 2022 equalling 1805 patients for whom potentially severe harm could be avoided. Other potential benefits of implementing MR for all outpatient clinic visits have not been explored. For example, large scale MR might reduce physician workload in an outpatient clinic. However, in a previous study, pharmacy-led medication reconciliation did not show an impact on time savings for the physician. This could be due to the limited time available for MR [20, 43]. In addition, in some countries MR is conducted by pharmacists, alongside other interventions, such as providing education to physicians or conducting medication reviews at outpatient visits, which could result in policy changes not related to medication reconciliation. A multifaceted intervention strategy is anticipated to be more effective on reducing drug-related problems than single interventions, particularly when embedded within a model of seamless care [44]. Nevertheless, it should be evaluated which patient populations benefit most, and which specific interventions achieve the highest impact. Future research should aim at identifying the patients who would benefit the most from MR in outpatient clinics. This study has shown that physicians indicated in 13.8% of all MRs the BPMH to be meaningful for future treatment. The extent to which this information justifies the use of MR for all patients visiting an outpatient clinic needs further investigation. Recent evidence suggests that patients can conduct MR themselves in a patient portal, a development that could prove beneficial given the constraints of limited resources and the fact that the patient knows best which medication is used [31]. However, considering the findings of this study, a more rational approach may be to concentrate MR efforts on patients identified as being at a higher risk. It is for this reason that we will analyse the obtained data in a future study using machine learning to develop a model to predict which patients require MR to prevent harm. Decreasing medication reconciliation for low-risk patients might lead to greater implementation for high-risk patients.

Strengths and limitations of the study

A notable strength of this study lies in its substantial patient sample size and inclusion of many departments. Additionally, this study has examined the actual impact of MR on the treatment policies set by the physician. There are also a number of limitations to this study. Firstly, not all departments were included in this study and the study was conducted in one setting. For example, no data were collected in the psychiatry and pulmonology outpatient clinics. Nevertheless, there is a balanced representation of surgical- and non-surgical departments in the study. Given the many outpatient clinics included and the large sample size, we believe our study does give an insight in clinical impact of MR at outpatient clinics. Secondly, children and adolescents are present in the research population but are underrepresented. Further research is needed to determine if findings are applicable to these groups. Thirdly, in this study, patients who could not be interviewed for MR (e.g. due to a language barrier) were excluded. It is possible that the number of impactful MR in this group is higher. Therefore, a different approach must be used for this group of patients to retrieve the BPMH, e.g., involving a caregiver. A fourth limitation is that the NCC MERP index score was determined by the same physician who initially set the treatment policy. The physician might underestimate or overestimate the severity of the necessary policy adjustment based on the personal connection with the error. However, the treating physician knows the context of the patient best and, therefore, is expected to have additional information that a blinded assessor would not have. Importantly, Davis et al. stated that physicians can provide a reasonably accurate self-assessment in certain situations, but overall, self-assessment is unreliable [45]. Future studies, should also use an expert panel to determine the impact of the policy adjustment more objectively. Fifthly, during the assessment of the patient’s health literacy, the physician had limited information from the MR. This information could have influenced the evaluation of the patient’s health literacy. Finally, in 8 patients for whom the policy was adjusted, the severity was not registered using the MERP index due to a human error. Assuming that all these adjustments would fall into the category of severe harm, the total percentage of MRs with clinical impact on severe harm would add up to 0.8%.

Finally, in our study, the BPMH was provided after the clinic consultation. This design was deliberately chosen to measure the independent impact of medication reconciliation on treatment policies. Theoretically, provision of the BPMH real-time could result in a different treatment policy when the physician uses additional information in the preparation of the outpatient clinic visit, e.g. check of laboratory values or previous clinic notes. However, we expect this to be limited, as the time that physicians have for medication reconciliation is very limited [20, 43].”

Conclusion

Medication reconciliation in outpatient clinics results in policy adjustments by physicians in only 3.8% of patients. In 0.5% of all 2289 patients, MRs result in a policy adjustment that could potentially prevent severe harm. Physicians regarded the BMPH as valuable for future treatment in 13.8% of MRs. Routine medication reconciliation for all patients at every outpatient visit is debatable and may be better targeted to patients at high risk of harm. Further research is needed to identify the high-risk patients who would benefit the most from MR at outpatient clinics.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

FK was a major contributor in writing the manuscript. All authors have contributed to editing the manuscript and agreed to the manuscript’s contents and its submission.

Funding

No funding was received for this study.

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the local ethics committee (medisch-ethische toetsingscommissie (METC) azM/UM, reference METC 2019 − 1449). Informed consent to participate was obtained from all participants in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.