The four participating hospitals (Hôtel-Dieu de Lévis, St-Georges, Montmagny, and Thetford Mines) are part of the CISSS-CA, a new integrated health organization under the leadership of a single CEO and created in 2015 by a legislative health reform in the province of Quebec, Canada (Act to Modify the Organization and Governance of the Health and Social Services Network) [72]. The CISSS-CA has 318 health facilities within its large mixed rural and urban territory, including the four acute care hospitals in this study. Although this new supraregional organization now has the administrative infrastructure to offer integrated care, it still lacks the knowledge management infrastructure to mobilize knowledge within its organization that has 10,000 employees serving a population of more than 415,000 spread across a large territory (15,079 km²) [73].

Our intervention is the delivery of a context-adapted CISSS-CA ACE program using a KT strategy based on our Wiki-suite. This intervention will be deployed in sequential phases.

An executive committee will oversee the entire study (Figure 1). This committee, which will be led by the first author (PA) and Director of Nursing, will meet every month during this 4-year study. The other members include a community-based geriatric nurse specialist, a home care coordinator, the ED director, the ED head nurse, a geriatrician, a database and measurement specialist, an information technology analyst, our research coordinator, and a patient representative. A local implementation team for each participating hospital, including an ED physician, a hospitalist, a family physician, a home care nurse, an inpatient unit manager, a research assistant, and a local patient or caregiver, will also meet every month at each participating hospital starting 9 months before the active implementation of the ACE program (Phase ID). This team will include locally identified champions to lead the local implementation. Regular meetings will also be organized with local hospital and community-based clinicians and decision makers to gather relevant feedback.

All local team members and eligible health professionals or decision makers will complete the Wiki101 web-based training to learn how to navigate and edit knowledge tools in WikiTrauma. After completing Wiki101, our executive committee in collaboration with ACE experts will then offer tailored support to each local team and the local champions to implement the ACE program. The training and support aim to inform and empower the local teams by providing information on the various ACE program interventions to implement locally, the rationale for each program intervention, and the sharing of existing knowledge tools and expertise developed in each center.

The context-adapted knowledge tools created at each site will be kept in WikiTrauma [74,75]. A local working group will be created in each center. All clinicians and decision makers will be encouraged to copy, edit, and update knowledge tools within WikiTrauma to create their own context-adapted tools. Our Wiki will track all changes automatically, and we will use Google Analytics to track the use of the tools. Any changes will be reviewed during our executive committee meetings and integrated after review to ensure reliability. Partners at Mount Sinai, CFHI, and Institut national d’excellence en santé et services sociaux (INESSS) will also have access to these web-based tools to provide expert oversight of our tool adaptation process. For any newly developed patient-centered knowledge tools (eg, self-care management guides), we will solicit in-depth feedback from a patient representative, who will lead a subcommittee of caregiver and patient representatives from the participating sites.

In the 9 months following training and local adaptation, we will implement the context-adapted CISSS-CA ACE program with the support of WikiTrauma and local implementation teams who will have the responsibility to roll out the different elements of our intervention within their respective hospitals. Our context-adapted CISSS-CA ACE program will include a series of systematic predischarge, postdischarge, and across transition period interventions for eligible patients as follows: (1) screening of patients in need of multidimensional evaluation of their loss of autonomy (with the Programme de Recherche sur l’Intégration des Services de Maintien de l’Autonomie 7 [PRISMA-7] tool) [76]; (2) a geriatric emergency management (GEM) nurse; (3) geriatric training [77]; (4) communication tools for transmitting information to nursing homes and other community-based stakeholders in the health system; (5) a fall prevention program; (6) systematic medication reconciliation; (7) elder-friendly ED environment adaptation; (8) access to clinician- and patient-centered Wiki-based KT tools [78,79] (eg, standing orders for geriatric patients admitted to the ED, geriatric analgesia prescription order set, and patient communication tools and decision aids); and (9) access to a community-based telemonitoring service. This service, which is offered by Télé-Surveillance Santé - Chaudière-Appalaches (TSS-CA) [80], will be offered for free to all eligible patients (ie, patients aged over 65 years who have at least seven ED visits in last 12 months, where over half of these visits are triaged to the ED observation unit and where two visits are not followed by hospitalization) transitioning from the hospital or ED to their home. This service currently offers remote monitoring of patients; nurses available 24 hours a day, 7 days a week, 365 days a year; and monthly phone check-ups. It also includes a customized emergency response intervention when patients are in need and connects patients with a network of community-based volunteer caregivers who are notified to visit patients when in need or simply to conduct a routine check-up.

In parallel with these interventions, we will also offer audit and feedback for health professionals and decision-makers in order to support organizational learning that will allow for real-time adjustment of interventions that are implemented. Feedback will take the form of monthly newsletters for the CISSS-CA covering patient-, clinician-, and hospital-level quality indicators. Moreover, our research team will be embedded within the CISSS-CA elder-friendly hospital committee, which plays an active role in supporting quality improvement initiatives and program implementation. This will ensure that results generated by our research team will benefit our population and ensure timely integrated knowledge translation.

Results from the centers will be compared with their own results in the preintervention period using data collected for the ITS study.

Our primary outcome will be a hospital-level composite outcome of 30-day hospital readmission and ED visit rate. Our secondary outcomes will be hospital-, clinical-, and patient- or caregiver-level outcomes (Table 1).

Hospital administrative databases (eg, Med-GPS, Logibec) will be used to calculate monthly hospital-level outcomes. Monthly data will then be analyzed to form points in time. Data will be extracted from the Régie de l’assurance maladie du Québec (RAMQ) physician billing database and Maintenance et exploitation des données pour l'étude de la clientèle hospitalière (Med-ECHO) database (containing data on hospitalizations and health professional consultations for all institutions) in addition to databases available at the INESSS in order to identify all public health services used prior to and after the implementation of the CISSS-CA ACE program.

Patient and caregiver baseline sociodemographic data will include age, sex, race, language, education level, family income, prehospital living situation (eg, home, intermediate nursing homes, etc), geography of residence (rural vs urban: as defined by Statistics Canada for Rural and Small Town [81,82]), and reason for hospital admission or consultation in the ED. Patient and caregiver outcomes are presented in Table 1. The living situation will be noted in the medical file when available at 30 days after discharge. Other measures that will be collected after discharge include the Care Transitions Measure-3 (CTM-3) [17,83], the Geriatric Anxiety Inventory-Short Form (GAI-SF) [84,85], and the Zarit Burden Interview (ZBI) [86]. The CTM-3 is a three-item questionnaire that measures the perceived quality of care transition on a 0-4 scale (0, fully disagree; 4, fully agree). The French version of the tool will be used [87]. The GAI-SF measures anxiety among seniors. The short version comprises five questions. The French-Canadian version of the tool has good psychometric properties [88]. The ZBI measures the burden of caregivers. The brief French version (12 questions) of the scale has good psychometric properties and is comparable to the original version [89,90].

In order to measure the change in care processes, we will measure process outcomes such as the proportion of patients seen by the GEM nurse, proportion of patients having a medication reconciliation list, and proportion of patients receiving telemonitoring services (Table 1).

After receiving ethics approval, we will start recruitment for our patient-level outcomes at least 3 months before the implementation of phase 1A at each site. Due to the pragmatic nature of our study, some sites will have longer periods of patient-level outcomes collected before phase 1A than others. Patient recruitment will continue throughout our study and up to a maximum of 24 months after the intervention. We will recruit consecutive eligible ED patients aged 65 years or older, as well as one caregiver whenever possible. We will recruit 38 patients or caregivers per month for each of the four hospitals based on a precision estimate of the CTM-3 measured for the study point of each month.

With authorization from the Director of Nursing and the Professional Services Director, a member of our embedded research team will contact by telephone a randomly selected daily sample of patients who visited the ED in the last 24 hours and up to a maximum of 7 days after their ED visit. This research assistant will administer the CTM-3 questionnaire to patients for quality improvement purposes. The research assistant will then ask the patients if they agree to be contacted a second time by a member of the research team to answer additional questions for the purpose of our study. Consenting patients will then be contacted within 7 days to obtain their verbal consent to participate in the study and consent to access their medical charts. We will ensure patients’ understanding by asking them to summarize in their own words the objectives of the study and what their participation involves using the Nova Scotia criteria [91]. The team member will then collect baseline sociodemographic data and administer the questionnaires for the study. The research team will then send a written consent form by mail after the interview. If a patient refuses to be contacted by our research team or to participate in this study, the CTM-3 data collected will only be available to the CISSS-CA. The research assistant will also ask permission to contact a caregiver. Identified caregivers will be contacted by telephone to complete the Zarit questionnaire. Our research assistant will first obtain the caregivers’ verbal consent to participate in the study and then ensure the caregivers’ understanding by asking them to summarize in their own words the objectives of the study and what their participation implies.

We will use segmented regression statistics to measure the changes in the level and slope in the postintervention period compared with the preintervention period in each center for each of our primary and secondary outcomes [92]. Thus, we will present a regression model with different intercept and slope coefficients for the pre- and postintervention time periods for each center. We will compare the changes in a composite primary outcome (total 30-day hospital readmission and ED visit rate) at our four intervention centers. We will use a Durbin-Watson test to verify the presence of autocorrelation and use an autoregressive error model to correct for this serial correlation. For an ITS, 10 measurement points before and 10 points after an intervention provide 80% power to detect a change in the level of 5 standard deviations (of the predata) only if the autocorrelation is greater than 0.4 (ie, extent to which data collected close together in time are correlated with each other) [93].

We have calculated the sample size for our patient cohort, which will be based on the smallest clinically significant difference (ie, 11%) that the CTM-3 can capture on a maximum score of 100. To detect a difference of 11% with a type I error of 5% and a type II error of 20%, 38 patients per month are needed to calculate a monthly CTM-3 estimate, based on a previous study in which CTM-3 was measured in 21 patients at Hôtel-Dieu Hospital in Lévis and showed an average of 75% and a standard deviation of 23%. Considering a call rejection rate of 70%, the number of people to be contacted in total will be 127 patients per month for each ED. We will therefore contact 5 patients per day in each of the four participating centers to obtain our targeted sample size of 38 patients at the end of the month. To ensure the creation of a random and representative sample of our population of elderly patients being discharged from the ED, a randomization table based on all patients discharged in the previous 24 hours from the ED will be provided every day to select which patients to call.