Protocol and registration
This systematic review was prospectively registered in the PROSPERO international prospective register of systematic reviews (reg CRD42020198438, September 11, 2020). This register (https://www.crd.york.ac.uk/prospero/) was searched for ongoing or previously published reviews with similar topics and criteria before registering our own review and initiating the search strategy; none were identified as overlapping.
The included study types were randomized control trials (RCTs) including randomized cross-over trials, cluster randomized trials, and within-subjects designs, as well as quasi-experimental studies including non-randomized control studies, before-and-after studies, and interrupted time series. Mixed-methods studies that included relevant quantitative data were included. Grey literature including reports from public agencies, theses, and conference abstracts with similar study designs as for the peer-reviewed literature were also included. Studies must have been published between January 2005 and August 2020, in English, French, or Scandinavian languages.
The included population was adults 50 years or older residing in the home or non-penal care institutions in OECD countries or countries of equivalent developmental level. Interventions must have been digital technologies, using audiovisual or sensor-based monitoring, specifically intended for nocturnal use.
Proof-of-concept, conceptual, observational, and qualitative studies that e.g. described attitudes or feasibility without providing any relevant quantitative outcome data were excluded from the review. Systematic, literature and scoping reviews, as well as other grey literature not reporting primary study results were excluded.
Hospitalized populations were excluded. Interventions intended for continuous monitoring of specific physiological conditions (i.e. apnea, seizure, heart rate etc.) were not included, nor were interventions used for daytime surveillance or outside the home or institutional setting.
The following databases were searched between July and October 2020 for peer-reviewed literature: PubMed (pubmed.ncbi.nlm.nih.gov), Cinahl Plus (EBSCOhost), Cochrane Library (www.cochranelibrary.com), Web of Science (www.webofknowledge.com), IEEE Xplore (ieeexplore.ieee.org), APA Psycinfo (EBSCOhost), Academic Search Elite (EBSCOhost), Applied Social Sciences Index and Abstracts-ASSIA (ProQuest), International Bibliography of the Social Sciences – IBSS (ProQuest), Scopus (www.scopus.com), and SocIndex (EBSCOhost).
Searches in the grey literature were conducted in September 2020 in the following international registers and databases of grey literature: OpenGrey (www.opengrey.eu), OAlster (oaister.worldcat.org), Bielefeld Academic Search Engine – BASE (www.base-search.net), WHO ICTRP (apps.who.int/trialsearch), ClinicalTrials.gov (clinicaltrials.gov/ct2/home), the International Health Technology Assessment Database (https://www.inahta.org/hta-database), DART-Europe (www.dart-europe.eu/basic-search.php), and Dissertations and Theses A&I (ProQuest). National sources of grey literature in the Nordic countries including publications and websites of public agencies, national databases, NGOs and interest groups were also searched in October 2020 using search strings in the respective Scandinavian languages. A complete list of the grey literature sources searched with links to websites is included in Supplement 1.
Searches in both types of information sources were followed up with searches in Google Scholar in English, Swedish, Norwegian, Danish, and Finnish languages.
The following search string was used in searches of the peer-reviewed literature databases:
(elderly OR "older adult*" OR "older person*" OR aged (MeSH only databases)) AND (nocturnal OR "night-time" OR "nighttime" OR "night time") AND (surveillance OR camera* OR "video monitor*" OR "in-home monitor*" OR "home monitor*" OR "safety monitor*" OR "digital monitor*" OR telemonitor* OR "remote monitor*" OR "digital camera" OR "digital sensor*" OR "monitoring system*")
The search string was adjusted in those databases that imposed limitations on search fields, operators and/or wildcards, so that several overlapping searches may have been required.
If the database allowed, the search string was identical to that used in the peer-reviewed literature searches. However, many databases did not allow advanced search strings, in particular websites of e.g. public agencies. The search string from the peer-reviewed literature was therefore divided into its components (delineated by the AND operator in that search string), or if necessary, into individual keywords. Screening of clearly irrelevant literature in these cases was conducted during the search.
An example of this was a grey literature database with a search maximum of 5 terms and without phrase searching. The resulting publications at each step in the search strategy were then opened, and the search terms re-combined in-document to determine inclusion in the study selection process:
Grey literature database search (maximum 5 terms), with results downloaded after each step
Search 1: elderly OR older adult OR older person OR aged
Search 2: nocturnal OR night-time OR nighttime OR night
Search 3: surveillance OR camera OR video monitor OR in-home monitor OR home monitor
Search 4: safety monitor OR digital monitor OR telemonitor OR remote monitor OR digital camera
Search 5: digital sensor OR monitoring system
Presence of any of terms in 1 AND any terms in 2 AND any terms in (3 or 4 or 5) in a publication: included in study selection process.
Study selection process
Altogether, 629 publications were identified in the literature search after removal of duplicates. All identified publications were entered into the Covidence systematic review software (Veritas Health Innovation Ltd, Melbourne, Australia, www.covidence.org) for the remainder of the review process. The PRISMA guidelines for reporting of systematic reviews(4) were followed and a summary of the study selection results can be found in Table 1.
Two researchers and one information sciences specialist (hereafter referred to as reviewers) conducted the review process, which had four steps:
Initial screening: The titles, keywords, and in some cases abstracts of the 629 publications were screened for relevance by two reviewers independently. Each reviewer voted on whether the publication was relevant for further review; consensus resulted in inclusion or exclusion at this step. Any remaining conflicts were resolved by the third reviewer.
Full-text screening: The full text for all publications proceeding to this step were obtained and read independently by two reviewers. Each reviewer voted on whether the publication was relevant for inclusion in the review and to proceed to data extraction. Consensus resulted in inclusion or exclusion at this step, and any conflicts remaining were resolved by the third reviewer.
Citation searching: We searched the cited references of seven review-type publications that proceeded past the screening stage but were then excluded as they were not original studies. We also searched Scopus for articles citing the same seven review-type publications. The cited references for, and citations of, any publications included after the full-text screening step were also searched in the same manner. These citation searches were conducted in August and September 2020. Both the cited and the citing articles followed the steps preceding this one in the review process.
Data extraction: Essential information regarding the study aim, design, conduct, population, intervention, and outcomes, as well as results data for relevant outcomes, was extracted from the publication by two reviewers independently. The template for this extraction can be found in Supplement 2. Consensus resulted in inclusion of the extracted data in the review’s summary of findings. Conflicts in extraction were discussed among the two reviewers, and any remaining conflicts remaining were resolved by the third reviewer.
Risk of bias assessment: The included individual publications were assessed for risk of bias at the study level by two researchers independently. The criteria assessed for randomized studies followed the Risk-of-bias tool 2.0(5) and included the method for random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, the completeness of outcome data, the possibility of selective outcome reporting, and other sources of bias. For non-randomized studies, the ROBINS-I tool(6) was used to assess bias due to confounding, missing data and selective reporting, and in selection of participants, classification of and deviations from interventions. Consensus about the risk of bias resulted in inclusion of the risk assessment in the review’s summary of findings. Conflicts in risk assessment were discussed among the two reviewers, and any conflicts remaining were resolved by the third reviewer.
The principle study measure was difference in means. In those cases where means were not the primary statistic in the study (e.g. cost-benefit analyses), the main statistic in the study was reported.