Methods

This systematic review followed the standard guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). The review protocol was registered with the PROSPERO international prospective register of systematic reviews (registration number: CRD42016049061)

Eligibility Criteria

Study design : all quantitative studies (randomized controlled trials, observational or quasi-experimental studies, etc…).
Population : Older adults (aged 65 and over) from all countries, being discharged from acute care hospital in-patient stays to a LTCF. For the purpose of the study, LTC will be defined as health services provided for people with complex health needs/moderate to extensive functional deficits or chronic conditions, who are unable to remain at home or in a supportive living environment, and involving nursing care and personal care.23 Acute care hospital in-patient stays could be for any health condition (i.e. frailty, geriatric syndrome, hip fracture, stroke, Alzheimer/dementia, multimorbidity, chronic disease exacerbation, oncology, infection…), planned or unplanned.
Interventions : Any TCi, defined as “a set of actions designed to ensure the coordination and continuity of healthcare as patients transfer between different locations or different levels of care within the same location”.24 In this study, TCis were included if they targeted an acute care hospital in-patient discharge back to a LTCF. TCis could include care/discharge planning in conjunction with the patients/caregivers/nursing home personnel, systematic medication reconciliation by a pharmacist, telecare/telemedicine/telemonitoring, formal discharge summary, structured follow-up and coordination among the different healthcare professionals.
Outcomes : Any quality of care, patient-related and healthcare services use reported outcomes.
A fully detailed list of inclusion/exclusion criteria is available in Supplementary Table 1.

Data Sources and Searches

Systematic searches were performed in Medline, CINHAL, EMBASE, Cochrane Central and Social Work Abstracts combining the concepts of LTC, hospital, older population and transitional care. We used MeSH terms and related and free key words (see Supplementary Table 2 for a more detailed search strategy). Reference lists of included studies were screened manually and companion papers were searched. Articles published between January 1st, 1995 (first TCis) and October, 2016 were considered for inclusion.

Study Selection

Based on inclusion criteria, two reviewers (MLB, AS) independently examined and selected the titles and abstracts obtained from the database searches. Full texts of the selected references were then retrieved and independently examined and selected by the same reviewers. At each step, any disagreement was resolved by consensus and discussion with a third reviewer (MW or IV).

Data Extraction and Quality Assessment

Data on study characteristics (authors, publication date, title, journal, study design), settings and participants (country, mean age, proportion of female, sample size, percentage going to a LTCF, reason of hospitalization, description of the intervention, healthcare professionals involved in the intervention, any coordination measures with the LTCF) were extracted from each study by two reviewers working independently (MLB, AS) and reconciled. Outcomes on quality of care (i.e. medication errors), patient-related (i.e. mortality, health-related outcomes) and healthcare services use (readmission, ED visits and total readmission days) were similarly extracted from the studies for all reported time points.
Study quality was assessed independently by two reviewers (MLB, AS) using the latest version Mixed Methods Appraisal Tool (MMAT), updated in 2018.25

Data Synthesis and Analysis

The wide variety of the reported outcomes and the discrepancy of the various cutoffs used (i.e. “Patients with at least 1 medication error”, “Patients with at least 3 medication errors” and “Patients with at least 5 medication errors”) precluded any meta-analysis pooling. Thus, we conducted a narrative synthesis26 by organizing the included studies into homogeneous groups according to the outcome of interest.
We thus reported the results, regrouped by the main key study outcomes identified: “Quality of care” (medication problems, and advance directives), “Clinical outcomes” (mortality, mobility and function and confusion/behavioral symptoms), “Services use” (hospital readmissions and ED visits and hospital days) and “Satisfaction” with TCi (from either healthcare professionals or patients/caregivers). We then looked for patterns within and across groups and compared similarities and differences.
A sensitivity analysis was performed by exploring the results with and without the studies with at least one MMAT item with a negative response in each identified study outcome.25