Engaging patients in medication reconciliation via a patient portal following hospital discharge

Abstract

Few ambulatory medication reconciliation tools exist. Transitions between inpatient and outpatient care can result in medication discrepancies. An interdisciplinary team designed a new ‘Secure Messaging for Medication Reconciliation Tool’ (SMMRT) within a patient web portal and piloted it among 60 patients at a Veterans Affairs hospital, an integrated system with a shared electronic health record. Recently discharged patients used SMMRT to view their medications in a secure email message and replied using SMMRT's interactive form, verifying their medication regimens and clarifying any inaccuracies. In total, 108 medication discrepancies and 23 potential adverse drug events (ADEs) were seen. Nearly 50% of the potential ADEs were classified as serious. Overall, participants were enthusiastic about SMMRT; 90% said they would use SMMRT again. Enabling patients to conduct medication reconciliation through a web portal is feasible in the transition from inpatient to outpatient care and may improve medication safety.

Introduction

Adverse drug events (ADEs) are the most common healthcare-associated adverse events.¹ Serious preventable medication discrepancies, which may predispose to ADEs, occur in 3.3 million outpatient visits each year in the USA, costing US$4.2 billion and resulting in 7000 deaths.^1–3 Transitions between the inpatient and ambulatory setting result in discontinuities in care and may lead to ADEs, which affect up to 20% of patients.^4–6 Medication discrepancies, defined as unintentional differences found between patients’ medical records and patients’ reports of the medication they are taking,⁷ are important contributors to adverse outcomes, particularly among hospitalized and recently discharged patients.^8–11 The Joint Commission has called for a process to ‘accurately and completely reconcile medications across the continuum of care’ and has designated medication reconciliation as a National Patient Safety Goal.^{8 10 12–15}

Prior efforts have targeted medication reconciliation upon hospital admission, before discharge home or during the ambulatory visit.^16–19 Randomized trials have shown that pharmacist-led medication reconciliation at discharge can decrease preventable ADEs,^{8 20} but few studies have demonstrated a reduction in hospital readmissions.¹⁴ Incorporating health information technology into medication reconciliation at hospital discharge can reduce potential ADEs and medication discrepancies.^{20 21} However, little is known about medication reconciliation strategies in the post-discharge period, when patients may not have prompt access to primary care and are at high risk for ADEs.⁵

Studies examining patient web portals have largely focused on management of chronic disease in outpatients. One randomized study found that pharmacist-assisted blood pressure control was better among patients whose medications were titrated using communication via a web portal,²² while another study failed to show improvement in glycemic control among patients with diabetes.^{22 23} We hypothesized that medication reconciliation could be performed asynchronously and remotely over the web with patient participation. Such a system could be valuable during vulnerable transitions, such as the post-discharge period, or among patients with difficulty attending in person. We therefore designed a pilot study and tested a medication reconciliation tool delivered via a web portal to improve medication safety among patients recently discharged from hospital.

Methods

Setting and participants

We recruited patients hospitalized at the Veterans Affairs (VA) Boston Healthcare System from June to December 2012 for pilot testing of the Secure Messaging for Medication Reconciliation Tool (SMMRT). Eligibility criteria included age ≥18 years, capacity to make medical decisions, primary care recipient at any VA facility, planned discharge home, computer access, and knowledge of the internet. After enrollment, patients were instructed on the use of My HealtheVet, the VA's patient web portal, and its secure messaging function. A US$50 cash incentive was offered for participation.

Development of the Secure Messaging for Medication Reconciliation Tool

A multidisciplinary team, including physicians (LH, SRS), a pharmacist (AP) and a nurse informaticist (TM) developed an interactive medication reconciliation document that displayed the patient's list of medications (generic name, dosage strength, formulation (eg, tablet or inhaler)) and directions for administration (eg, take one-half tablet each morning and each evening). The document included an interactive component so that patients could perform medication reconciliation independently at home by checking ‘yes’ or ‘no’ to each listed medication and add free-text questions or comments directed to the research team (figure 1). This document was embedded into the body of a secure message—similar to a typical email message—and delivered to and from patients using the secure messaging feature within My HealtheVet. This system of medication reconciliation was named ‘Secure Messaging for Medication Reconciliation Tool’, or SMMRT. Before testing SMMRT in this pilot study, we conducted three focus group discussions with veterans to optimize the usability of the SMMRT.

Figure 1.

Example of an initial outbound Secure Messaging for Medication Reconciliation Tool (SMMRT), pasted into the body of a secure message. All secure messages were addressed to the patient by name and specified the patient's date of birth, discharge date and diagnosis as well as allergies. Medications were grouped as ‘Current’, ‘Non-VA, Herbal, Over-the-counter’, ‘Stopped’, ‘Expired’ and ‘Other’ to emulate the display of the medication list provided by the inpatient team at discharge. EHR, electronic health record

Pilot testing of SMMRT

For each enrolled patient, the ultimate goal was to produce a fully reconciled medication list for the medical record, reflecting as accurately as possible the medications that the patient was taking after re-establishing his or her home routine after leaving hospital. To achieve this goal, the study pharmacist initiated a four-step process in consultation with a physician (LH). First, within 2 days of hospital discharge, the study pharmacist reviewed the patient's discharge summary and reconciled the listed medications with those dispensed from the VA pharmacy; any discrepancies were highlighted for the patient to consider and respond with clarification. Second, the pharmacist incorporated this reconciled list of medications into the SMMRT and sent it to the patient via secure message in My HealtheVet, within 72 h of the patient's discharge from the hospital (figure 2). Third, the patient responded by logging into My HealtheVet, selecting the secure messaging feature and replying to the message containing the SMMRT and clarifying any questions or discrepancies. Returned SMMRTs were received by the pharmacist, sometimes leading to several back-and-forth messages and occasional telephone calls to clarify medication regimens. Fourth, the pharmacist posted a summary of this process, including a final reconciled medication list, in the patient's electronic health record (EHR).

Figure 2.

Overview of the Secure Messaging for Medication Reconciliation Tool (SMMRT) pilot study. SMMRT was initiated within 72 h of discharge, establishing two-way communication with the patient. At home, the patient completed the interactive SMMRT and sent it back to the clinical pharmacist. Updated medication lists were posted to the electronic health record (EHR) to be viewed by providers.

Post-pilot in-depth interviews

We randomly selected 10 participants for a debriefing interview by telephone. Participants responded to questions about their experiences with completing medication reconciliation using the SMMRT tool, including perceived utility and barriers experienced during the pilot (Ssee online supplementary appendix).

Measurements and analysis

The main outcomes were clinically important medication discrepancies and potential ADEs. A physician (LH) and pharmacist (AP) identified clinically important medication discrepancies and potential ADEs by comparing (1) medications listed on the discharge summaries, (2) pharmacist-reconciled medication lists in the initial SMMRT (see ‘Pilot testing of SMMRT’), and (3) then returned the SMMRT, reflecting patients’ confirmed medication lists. Drug classes associated with potential ADEs were noted. One physician (SRS) classified the severity of each medication discrepancy and potential ADE using a previously validated system.²⁴ We used bivariate analysis to determine patient characteristics associated with medication discrepancies.

The VA Boston institutional review board approved this study. A p value ≤0.05 was considered significant. All analyses were conducted in SAS V9.3 (Cary, North Carolina, USA).

Results

We enrolled 60 patients, of whom 56 (93%) were male. Average age was 61 years and 58 (97%) had a primary care provider. Most (70%) had private health insurance or Medicare, and more than three-quarters (78%) had more than five medications on their discharge medication list. All pilot participants had at least one chronic condition, defined as diabetes, hypertension, prior myocardial infarction or stroke, hyperlipidemia, or heart disease. Demographic information was collected from 50 additional patients who were screened for enrollment but were determined not to be eligible; the most common reason for ineligibility was lack of computer use (78%). Ineligible patients were older than pilot participants (mean age 65 years) and did not have computer access or knowledge of the internet, but were otherwise similar with respect to gender (100% male), insurance status (80% insured) and affiliation with a primary care provider (96%).

Among the 60 enrolled patients, we sent 51 SMMRT messages within 72 h of discharge, and received 34 reply messages (response rate 67%). Nine patients became ineligible after enrollment owing to rehospitalization within 72 h (n=2), ambulatory follow-up visit before initial SMMRT (n=1), discharge to psychiatric facility (n=1), discharge to rehabilitation (n=4), and death (n=1).

Clinically important medication discrepancies and potential ADEs

We identified at least one medication discrepancy between the discharge summary and the medications filled at the pharmacy among 40 of the 51 patients (78%) to whom we sent an initial SMMRT message; in total, we found 108 clinically important medication discrepancies among these 40 patients. More than two-thirds of these discrepancies (68%) were medication omissions—that is, a medication the patient was apparently taking but which was not documented in either the physician's electronic prescription or the patient's discharge summary. Twenty medication discrepancies (19%) were identified as medication duplication (two or more concurrent orders for one medication), while 15 medication discrepancies (14%) were a result of gross inaccuracies in the medical record of the dose of the medication to be taken.

In the 34 returned SMMRTs, in which patients had independently carried out medication reconciliation after discharge home from hospital, we identified 26 additional clinically important medication discrepancies among 17 patients. Among these 26 medication discrepancies, 50% were due to medication omission, 38% were due to medication duplication, and 12% were the result of a dosing error. Ten patients had at least one medication discrepancy on both the initial and returned SMMRT (table 1).

Table 1.

Characteristics of the 108 clinically important medication discrepancies and the 23 potential adverse drug events (ADEs) among medication lists sent (n=51) and returned (n=34)

Outcome	Events (n) on initial SMMRT*	Events (n) on returned SMMRT†
Median medication discrepancies per patient	2	2
Clinically important medication discrepancies	108	26
Source of discrepancy
Medication missing from discharge summary/absence of electronic prescription for medication	73	13
Duplicate medication order	20	10
Dosing discrepancy	15	3
Medication type involved in discrepancy
VA medications	38	8
Non-VA medications	13	17
Remote VA medications§	46	1
Remote non-VA medications	11	0
Potential adverse drug event‡	23	0
Source of potential adverse drug event
Medication missing from discharge summary/absence of electronic prescription for medication	15
Duplicate medication order	2
Dosing discrepancy	6
Severity of potential adverse drug event
Serious error	13
Significant error	8
Life threatening error	2
Medication type involved in potential adverse drug event
VA medications	11
Non-VA medications	1
Remote VA medications	11

*Discrepancies identified when medications on the patient's discharge summary were compared with electronic medication prescription orders and refills at the time of discharge.

†Discrepancies identified when patient-reported medication use was compared with the initial SMMRT and the electronic medication prescription orders and refills at the time of discharge.

‡The most common medication class implicated in potential ADEs was psychiatric medications (35%), followed by hypoglycemic agents (17%), cardiovascular and anticoagulants medications (13% respectively). Opioid medications accounted for 9% of the potential ADEs, and there was one error (4%) involving an anti-epileptic drug.

§Other VA facilities.

SMMRT, Secure Messaging for Medication Reconciliation Tool; VA, Veterans Affairs.

In the 51 SMMRT messages sent, 23 potential ADEs were identified (table 1). Medication omissions were responsible for the majority (65%) of potential ADEs. Almost half (43%) of potential ADEs seen were classified as being a significant error or of life-threatening severity. Over half of the potential ADEs (52%) were noted to be in medications prescribed at other VA facilities (‘remote’ VA).

In bivariate analysis, we examined the relationship between patient characteristics and observed medication discrepancies (table 2). Although none of the patient characteristics was found to be associated with medication discrepancies, there was a trend suggesting that patients with a greater number of medications and those who had medications changed at hospital discharge (compared with admission medications) were more likely to have a medication discrepancy identified (table 2).

Table 2.

Correlates of the 34 patients who experienced one or more of the 108 clinically significant medication discrepancies observed

Characteristics	Patients with ≥ 1 medication discrepancy, N (%)	p Value
Age, years (mean)		0.16
<61 (n=23)	13 (57)
≥ 61 (n=28)	21 (75)
Insurance status		0.71
Insured (n=41)	28 (68)
Uninsured (n=10)	6 (60)
Length of stay, days		0.55
<4 (n=24)	15 (63)
≥4 (n=27)	19 (70)
Number of discharge medications		0.11
<14 (n=25)	14 (56)
≥14 (n=26)	20 (77)
Medication changes at discharge		0.10
≤2 (n=37)	22 (59)
>2 (n=14)	12 (86)
Expired medications at discharge		0.14
≤2 (n=41)	25 (61)
>2 (n=10)	9 (90)

Patients’ experience with medication reconciliation via patient portal

All 10 veterans completing post-intervention interviews found SMMRT easy to use. However, 30% required time to troubleshoot initial technical difficulties; 50% indicated some challenges navigating the web portal; and 40% reported difficulty with the cut-and-paste function. Most respondents (60%) appreciated the ability to communicate directly with a healthcare provider following hospital discharge to clarify questions about their medication and to request refills. A majority (90%) of participants said they would use SMMRT again, frequently emphasizing SMMRT's ability to provide rapid access to their healthcare provider (figure 3).

Figure 3.

Responses of 10 pilot participants to domains of experience following the Secure Messaging for Medication Reconciliation Tool (SMMRT) pilot. We conducted 10 in-depth interviews among participants who completed the pilot to evaluate their experience and seek feedback on SMMRT.

Discussion

This pilot study demonstrated that patient-mediated medication reconciliation via a web portal after hospital discharge is feasible and readily accepted. We found that over two-thirds of participants had at least one medication discrepancy at discharge, and nearly one-third had at least one potential ADE. Medication omissions comprised over two-thirds of discrepancies examined. These discrepancies were particularly prevalent among veterans who received primary care at a VA location different from the facility in which they were hospitalized. This finding demonstrates the challenges to ensuring medication safety even within an integrated care system using a shared EHR.²⁵

Our findings underscore the importance of conducting medication reconciliation soon after hospital discharge to reduce the risk of ADEs.⁸ The SMMRT enables patients to participate in medication reconciliation from home as they await primary care follow-up, a period during which medication discrepancies often occur.^{4 26}

Prior studies have largely focused on using existing technology, such as telephone outreach or in-person visits, for follow-up after discharge.^{12 20} Surveys have consistently found that patients are enthusiastic about electronic health information exchange with their primary care providers.^27–29 This study, with patients using an interactive tool within a web portal to reconcile their medications after hospital discharge, showed that patients are able and willing to use this technology to manage their care.

Our study had several limitations. First, our participants were all older, hospitalized veterans, limiting the generalizability of our findings and user perceptions. Second, patients performed medication reconciliation themselves without face-to-face verification, limiting our ability to confirm the accuracy of their medication lists. Third, although we limited this pilot study to patients receiving care primarily at a VA facility, medications filled outside the VA or purchased over-the-counter could not be reviewed by our study team.

In conclusion, this study demonstrates the potential for patients to carry out medication reconciliation via a web portal in the immediate post-hospital discharge period. Tools such as SMMRT, designed to conduct ‘virtual’ medication reconciliation following hospital discharge, stand to improve medication safety in the transition from inpatient to outpatient care. Future studies with prospective, controlled designs should examine the ability of this approach to reduce medication discrepancies and prevent ADEs, and to improve medication adherence.