As highlighted by the Working Group on Medication Overload (1), problematic polypharmacy is an overwhelming problem for older adults. Nearly two thirds of community dwelling adults are taking five or more medications (2) and while most medications are prescribed to help people live longer, healthier lives, the more complex a medication regimen is, the more dangerous it becomes. With each added drug, the risk of an adverse drug event (ADE) increases (3), as well as the worsening of geriatric syndromes (4). A growing body of evidence suggests that patients and physicians may not recognize ADEs and additional new medications may be prescribed to treat an existing ADE (5). This so-called prescribing cascade further increases polypharmacy and the risk of ADEs. Aging itself is a risk factor for ADEs as changes in organ function, body weight, fat distribution, cognition, balance, function and drug receptor sensitivity all affect medication metabolism and side effects (6). Experts in geriatrics have published guidelines and lists of medications to be avoided or used cautiously in older adults (7–9). These medications are often referred to as potentially inappropriate medications or PIMs. PIMs are medications associated with a high risk of ADEs when administered to older adults, and may be of uncertain or no clinical benefit (10). Continued use of PIMs in later life, further increases the risk of ADEs associated with polypharmacy in older adults. Furthermore, as highlighted in a 2018 systematic review, there is a significant association between an increased number of medications and frailty, which is associated with an increased risk of adverse events in older adults (11).
Older adults living in nursing homes are at a particularly high risk of polypharmacy and ADEs as they are frail and often take ten or more medications (2). Some of these medications increase the risk of falls, fractures, kidney injury, bleeding, and death. Costs to the healthcare system from polypharmacy is very high and stem from emergency department visits, hospital admissions, and direct and indirect drug costs. As such, strategies to address polypharmacy are needed. One approach is the process of deprescribing, whereby a prescriber or pharmacist establishes the full list of medications a patient is taking and contextualizes them for the individual patient (12). Based on the medical history, a measure of frailty, and the patient and/or caregiver's values or preferences, safer classes of medications are selected, doses may be decreased, and medication may be discontinued (12). This process can be thought of as a "prescription check-up".
While deprescribing may be effective for stopping PIMs, manually reviewing all drugs and cross referencing them with patient conditions and lists of inappropriate drugs in older adults requires an expert command of the literature and is time-consuming. In many Canadian provinces nursing home medication reviews are mandatory, however there are no consistent guidelines for the reviews and therefore existing processes often result in medications being re-prescribed without documentation of a clear rationale or reassessment. Studies in acute care hospitals have demonstrated that electronic decision support can facilitate the process of deprescribing and augment the proportion of patients with one or more PIM (potentially inappropriate or problematic) medications stopped upon discharge from the hospital (13). More studies looking at electronic tools for deprescribing in nursing homes are needed to determine if an investment on the part of the government into a software that supports deprescribing is warranted.
An electronic decision support tool for deprescribing called MedSafer has been studied in the Canadian acute care context. MedSafer augmented the proportion of patients who were discharged from the hospital with one or more PIMs deprescribed as compared to usual care (13). New Brunswick (NB) has the highest proportion of older adults compared to other provinces in Canada. In NB, 20.8% of the population is over 65 years of age and this is expected to increase to 31.3% by the year 2038 (14). Strategies to promote regular prescription check-up are needed to reduce polypharmacy in the setting of LTC. For this study, MedSafer researchers partnered with the Centre for Innovation and Research in Aging (CIRA) at York Care Centre (YCC) to improve a previously developed web-based application (Polypharmacy App) that allowed prescribers in New Brunswick Nursing Homes (NBNHs) to visualize MedSafer deprescribing reports on a desktop computer or tablet. The polypharmacy App, called MedReviewRx, will be implemented in a number of NBNHs.
Research questions
-
Is the MedReviewRx App developed at YCC, in Fredericton NB, easy to use, does it provide useful information, and does it reduce the time needed to conduct a medication review in other NBNHs?
-
How often do nursing home health care providers log into the MedReviewRx App, and how often does this action result in a prescription check-up (review and deemed appropriate or review and medication change)?
-
What are the challenges to implementing the MedReviewRx App in other NBNHs and what improvements are required for widespread use of the MedReviewRx App?
-
What is the impact of the MedReviewRx App on deprescribing PIMs and associated adverse events in NBNHs?
Objectives (7)
To embed software enabled deprescribing in five NBNHs using MedReviewRx, an App containing MedSafer recommendations, in order to study the effectiveness at deprescribing PIMs in LTC, along with feasibility of use.
Trial design (8)
This is a hybrid type 2 effectiveness-implementation design for quality improvement research that will make use of mixed methods of evaluation. Implementation consists of using MedReviewRx. Effectiveness analysis consists of measuring the impact of the MedReviewRx on PIMs. Exploratory surveys (the mixed methods component of the study) which look at qualitative and quantitative user feedback and patient and caregiver attitudes about deprescribing will be used to explain study findings. Informal feedback, surveys and semi-structured interviews will be used to examine user experience with MedReviewRx. The Revised Patient Attitude Towards Deprescribing (rPATD) questionnaire “older adult” version for the patients and a “caregiver” version for family members/substitute decision makers will be used to describe patient and family member attitudes about deprescribing medications.
Intervention description {11a}
Estimated study duration is 18 months, and the effectiveness component of the study will use a deployment strategy which approximates a Stepped-Wedge Cluster-Randomized Trial (SW-CRT). This design has many advantages and is considered the most robust type of study design for pragmatic quality improvement interventions. Each cluster will contribute a control period. Subsequently, at each of 3 pre-defined time periods, one cluster will be randomly assigned to transition from control to intervention. In this way, all clusters act as an internal control (before and after) as well as a contemporaneous control for the other clusters (external control). The study is composed of three clusters and each cluster contains the following number of beds: Cluster A 218, Cluster B 314, and Cluster C 272.
Criteria for discontinuing or modifying allocated interventions {11b}
There is no planned interim analysis as the intervention is considered safe and best practice. The study will be analyzed according to the intention to treat principle.
Strategies to improve adherence to interventions {11c}
We will provide training for the software users, along with memos and email remainders about the study during the intervention phase. Ultimately the decision to use the software will up to the individual prescriber as this is a pragmatic intervention to study the real-world effectiveness of making the software available for use in the LTC homes.
Relevant concomitant care permitted or prohibited during the trial {11d}
All usual care received by the participant is allowed to continue during the trial.
Provisions for post-trial care {30}
Participants in the trial will continue to receive usual care in the LTC home post trial completion.
Sample size {14}
Based on the Canadian Institute for Health Information’s report, titled “Drug Use Among Seniors in Canada, 2016”, we expect that about 70% will be taking one or more PIMs. This yields an average cluster size of 175 for patients with at least 1 PIM. Uptake of deprescribing using the MedSafer report in acute care studies ranges from 8-25% (13) over usual care. Conservatively, we think we can increase the proportion of patients with 1 or more PIM deprescribed from a baseline rate of 10% to 20% (absolute increase of 10% or a number needed to treat of 10). Using the Shiny CRT calculator (15) we estimate that we will have at least 72% power to detect this difference under assumptions of cluster autocorrelation of 0.8 and a within period intra-cluster correlation (ICC) 0.05. This increases to 96% power if the ICC is 0.02.
Recruitment {15}
Nursing homes may experience 20 to 30% patient turnover each year due to deaths, therefore the total number of nursing home residents who have at least one interRAI LTCF assessment during the study period is approximately 1000. For the purposes of analysis, only the initial closed cohort will be included in the study. New residents will not be analyzed in the current study. All homes have agreed to participate in the study and an individual waiver of consent has been granted by the Horizon Health Network Research Ethics Board.
Sequence generation {16a}
Computer-generated random numbers were used in order to generate the allocation sequence of the clusters.
Concealment mechanism {16b}
Study sites are not aware of which clusters they are assigned to and will only be made aware that they are entering into intervention mode approximately four-weeks prior, to allow for training on the use of the software.
Implementation {16c}
The principal investigator (EGM) of the study generated the allocation sequence. Participants are automatically enrolled in the study and assigned to the intervention when the targeted nursing homes agree to take part in the study.
Who will be blinded {17a}
The study investigators and analyst will be blinded to the intervention status at the time of outcome adjudication. Prescribers cannot be blinded as the intervention is the use of a computer software.
Procedure for unblinding if needed {17b}
There is no procedure for unblinding (does not apply to the trial design).
Plans for assessment and collection of outcomes {18a}
Each study site will appoint a site contact to load information into MedReviewRx and to communicate with the study team. The nursing home site contact or designate(s) will be e-mailed a Nursing Home Site Assessment Questionnaire for completion prior to project implementation. The purpose of this questionnaire is to determine information about the nursing home including workflow, structure of medication reviews, physician, pharmacy and nursing support. This information will be used to inform implementation and determine where MedReviewRx fits into the workflow. Study sites will also receive an implementation toolkit containing a user manual, instructions on how to use the Momentum extract report, a research study PowerPoint presentation, template for Staff/Crew Meeting “huddles”, sample announcement of study frequently asked questions and a list of deprescribing resource websites. The CIRA research manager or designate will review the Nursing Home Site Assessment Questionnaire and Toolkit resources with the study site contact. If a nursing home site is not completing section N1 (List of all medications) of the interRAI LTCF assessment, a research assistant will be trained to enter this information for the study site.
The study will begin with a minimum three-month control phase. During the control phase, MedReviewRx will not be accessible to healthcare professionals at the nursing homes. This serves to obtain baseline deprescribing levels for each nursing home. Every three months thereafter, a cluster of nursing homes will enter intervention mode.
During the control phase, nursing home residents and their families or substitute decision makers will be recruited to complete a survey to explore patient attitudes about deprescribing in this NB study population. The rPATD questionnaire “older adult” version will be provided to nursing home residents and the “caregiver” version will be provided to family members and substitute decision makers. Survey questions will be entered in Lime survey to promote electronic completion whenever possible. A recruitment poster will be displayed at study sites and a recruitment communication will be sent to family members and substitute decision makers identified in the resident’s admission package. Paper copies will also be available from the study site contact. Residents and family members who ask nursing home staff about the survey will have their contact information forwarded to the CIRA research manager to answer questions and obtain verbal consent to have a survey e-mailed to them. If email is not available, a research assistant will contact the participant to administer the survey over the phone. Study site contacts may also distribute paper copies of the surveys if requested and permitted by the study site.
During the control phase, health care providers and study site staff will be recruited to register for MedReviewRx. A recruitment communication will be sent by e-mail and placed in study site mailboxes. Study recruitment will also be encouraged at interdisciplinary nursing home meetings such as Pharmacy and Therapeutics and Medical Advisory Committee meetings. User training will be provided by the study site contact or designate as determined by the site-specific implementation plan. Nursing home site contacts will participate in an interview to evaluate MedReviewRx implementation. Implementation interviews will be conducted within the first 3 months of the site intervention phase.
During the intervention phase, MedReviewRx will be made available to nursing homes with the understanding that it will be used to facilitate medication reviews and prescription check-ups. MedReviewRx provides clinicians with access to individualized and prioritized deprescribing information from MedSafer which:
A) Identifies PIMs.
B) Explains why the medication is potentially inappropriate.
C) Provides instructions on how to safely stop/taper the medication.
D) Links to patient informational brochures with non-pharmacologic interventions and rationale for deprescribing from the Canadian Deprescribing Network
Nursing homes will also have the option to print MedSafer deprescribing reports from MedReviewRx for clinicians who do not have computer access or who choose not to register to use MedReviewRx. If reports are printed, they will be placed in a binder for the prescriber to sign and date indicating that they have read the report. Prescribers and nursing home staff will be encouraged to write feedback directly on the report if they wish too. To assess the proportion of reports that were read, and feedback provided on the reports, signed reports will be kept in a locked cabinet for research assistants to collect. If visitor restrictions are in place, the study site contact will gather signed reports, place them in a sealed envelope(s) and send them to the CIRA research manager using a courier method that can be tracked and requires a sending signature and signature of receipt.
Clinicians will review deprescribing opportunities (electronically or via a paper report) and determine if medications can be tapered or stopped. Medication changes are discussed with the resident or substitute decision maker as part of usual care. This process will not change with the implementation of MedReviewRx. If the prescriber decides to alter a medication based on the deprescribing opportunities provided by MedSafer and their expert knowledge, the prescriber will do so in the same manner as they did prior to implementation of MedReviewRx, and discussion with the resident and/or substitute decision maker will occur according to the process in place at each nursing home.
During the intervention phase, an analysis of deprescribing opportunities will be conducted by MedSafer once every 3 months for each resident in the nursing home. Results will be stored in the MedReviewRx system to be accessed at any time.
Patients who die or are hospitalized will be censored from the study.
MedReviewRx has been designed to have Administrator accounts and Clinician accounts which restrict access according to user function. Study site contacts (or site designate) will load data into MedReviewRx using the Administrator function. Prescribers (physicians and nurse practitioners) and pharmacists will be provided with Clinician accounts. Clinicians will only access MedReviewRx records for patients under their care. MedReviewRx registration has been designed to restrict nursing home staff access to residents at their site only. Prescriber and pharmacist access will be restricted to study sites where they practice. MedReviewRx access will be audited by the CIRA research manager or designate.
User feedback on MedReviewRx and MedSafer deprescribing information will be solicited from prescribers, pharmacists and nursing home staff throughout the study using surveys. Surveys may be completed via Lime Survey, on paper or by telephone. Three surveys will be distributed by e-mail to registered MedReviewRx users as well as other study site prescribers, pharmacists and nurses who may have access to printed reports from MedReviewRx. Paper copies of the surveys will also be available at the study site nursing units. A MedReviewRx user feedback survey will be distributed at the end of the first and third quarter of the intervention and an acceptability and feasibility survey will be distributed at the end of the study. Paper copies of completed surveys will be collected by the study site contact, placed in a sealed envelope(s) and sent to the CIRA research manager using a courier method that can be tracked and requires a sending signature and signature of receipt. Courier costs will be paid for by CIRA. MedReviewRx users will be provided with a study e-mail which they may use to submit feedback and questions at any time throughout the study. Informal feedback received by email from users will be reviewed and actioned by the CIRA research manager or designate. E-mail feedback will be documented anonymously in an excel spreadsheet and reported to the study principal investigators within 1 week of receipt to determine if tasks need to be submitted to the technical team for software updates. Rapid cycle improvements in usability of the system and MedSafer output will be made based on survey results and informal feedback.
Informal feedback written on printed MedSafer reports will be documented in an excel spreadsheet and included in the analysis of qualitative feedback provided by survey and interview.
Frequency of MedReviewRx use will be measured electronically by counting the number of times a user accesses their account.
Overall experience with the research project will be evaluated using semi-structured interviews conducted at the end of the intervention period.
Sampling procedures
Convenience sampling was used to select nursing homes for the study. NBNH administrators and directors of nursing were contacted by email to determine interest in study participation.
Convenience sampling will also be used to recruit health care providers, older adults and family members of older adults from study nursing homes. A recruitment poster for resident and family member surveys will be posted in study nursing homes and a communication will be sent to families and substitute decision makers by regular mail or e-mail using the study site contact list. Health care providers and study site staff will be recruited to register for MedReviewRx by placing written recruitment notices in their nursing home site mailbox and/or by email. All pharmacists, nurses, nurse practitioners and physicians providing care at the participating nursing home will receive recruitment emails for MedReviewRx feedback surveys. Members of healthcare teams at the study sites will be recruited to participate in interviews to explore overall experience with the research project.
Variables under investigation include:
A) Patient demographics (date of admission to LTC, age, sex, gender, nursing home identifier, medical and psychiatric conditions documented in the interRAI LTCF assessment, and date of discharge from LTC if applicable).
B) Medication use data: absolute number of medications, number of PIMs, proportion of all medications that are PIMs, and type of deprescribing opportunities identified (PIMs that are high risk, intermediate risk or of little added value).
C) Safety data: death, fracture, fall, use of restraints, transfer to hospital, change in functional status based on interRAI activities of daily living (ADL) scores. Episodes of delirium based on responses to questions in Section C of the interRAI LTCF assessment form (periodic disordered thinking or awareness and acute change in mental status from a person's usual functioning).
D) Health care provider experience with MedReviewRx and MedSafer deprescribing information as described in the user feedback surveys and the acceptability and feasibility survey.
E) Nursing Home evaluation of the implementation toolkit.
F) Patient and caregiver attitudes about deprescribing.
Plans to promote participant retention and complete follow-up {18b}
Participants are automatically enrolled when targeted nursing homes agree to take part in the study.
Data management {19}
Patient demographics and safety data is entered in the interRAI LTCF assessment software by nursing home staff as part of usual care. Medication information will be obtained from nursing home MARs and entered in section N1 of the interRAI LTCF assessment software by nursing home staff or a trained research assistant. Research assistants who enter medication data will receive training on how to enter medication information into Momentum LTCF assessment software. Patient demographics, medication use, and safety data will be extracted from the LTCF interRAI assessment software and loaded into MedReviewRx by the study site contact (or designate) at each nursing home.
Confidentiality {27}
MedReviewRx will de-identify the data and transmit data sets to MedSafer. Multiple safeguards have been put in place to protect this data. Data sets do not contain patient specific information. Age is transmitted rather than date of birth, and a unique identifier is assigned to each patient data set. MedSafer is unable to identify patients however the unique identifier allows MedReviewRx and MedSafer applications to anonymously communicate about specific patients through an application programming interface (API). Data sets will be analyzed by the MedSafer software program. Files containing one or more targeted PIM and associated triggering condition(s) will have deprescribing opportunities flagged and returned to MedReviewRx. MedReviewRx will link the unique identifier to the correct nursing home patient and translate the MedSafer analysis into an output for review by their health care team. A message will also be displayed by MedReviewRx if there are no deprescribing opportunities identified for a data set. MedReviewRx transmits and stores data using a secure web-based system which was designed by Missing Links Technology (MLT) under the supervision of the New Brunswick Community College (NBCC). The MedReviewRx system safeguards personal health information (PHI) that it collects, processes and disseminates within trusted and authorized circles of care and ensures the confidentiality and privacy of the PHI. Each study site using MedReviewRx has site specific policies and procedures with respect to privacy and confidentiality of PHI.
Data from User Feedback Surveys and Acceptability and Feasibility Surveys and survey responses collected from the rPATD questionnaires completed via Lime Survey will be extracted into excel sheet format. Survey data collected from paper surveys or telephone surveys will be entered into the excel database by a research assistant. Research assistants will be provided with templates for excel data entry and trained on how to complete the template. A minimum of 5% of the research assistant data entry will be audited for accuracy and completeness by the CIRA research manager or designate. Interview data will be recorded and transcribed verbatim by research assistants. Interview recordings will be deleted after transcripts have been validated. Study data (PIMs, survey responses, interview transcripts) will be stored in password protected, members only secure cloud-based study sites hosted within Canada. Study data will only be accessible to the study research team and the CIRA research manager. On study closure, after final data analysis is completed, study data will be downloaded to a password protected study folder on the YCC secure network and the cloud-based study site will be permanently deleted. Study data will be stored for 7 years and then destroyed in accordance with YCC/CIRA policies and procedures.
Plans for collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis in this trial/future use {33}
No biological specimens were collected as part of this trial.
Data analysis plan
This mixed methods study will employ both qualitative and quantitative data analysis.
Statistical methods for primary and secondary outcomes {20a}
Analysis of the primary outcome will be conducted using the stepped wedge function in Stata v. 15 or a similar statistical program.
Interim analyses {21b}
There is no planned interim analysis.