Housing with support for older people: a mixed-methods systematic review protocol

Study design

This mixed-methods review will employ convergent parallel design. We chose this approach rather than a sequential approach because convergent parallel design is considered more appropriate for reviews addressing separate qualitative and quantitative research questions⁵. As such, in accordance with convergent parallel design qualitative and quantitative data will be analysed separately, and the findings will then be integrated⁵. By considering qualitative and quantitative studies in isolation prior to synthesising them, we will be able to do justice to each of the two overarching research paradigms before integrating our results.

Search strategy

The search strategy that was designed for this review is based around three overarching concepts – older people, models of housing with support, and mixed-methods research. One comprehensive search of the published, peer-reviewed research on housing with support for older people will be used to answer both research questions. Preliminary scoping searches indicated that research articles informing both research questions were to be found across a range of sources, including medical and sociological sources. The search strategy was initially developed for the MEDLINE (Ovid) database and was subsequently translated for use in the CINAHL, SocINDEX (both on the EBSCO platform), and Social Policy & Practice (Ovid) databases (see extended data for the full search strategy⁶). These databases cover a range of subject areas, professions, and geographical areas, ultimately providing a wide scope of research sources.

No language limits were applied to the search strategy. Following trial searches of MEDLINE and the examination of references from previously published reviews, the decision was taken to limit the search to articles published from the year 2000 to present. For the purposes of this review, older people have been defined as adults aged 50 years and over, so where possible, limits to include only research on older people were applied.

A separate grey literature search will also be undertaken to inform each of the two research questions, based on key terms used for the database search and using the same inclusion and exclusion criteria that will be used for the database search.

The search strategy was developed by an information specialist (LF) in consultation with the research team. It was reviewed and modified following consultation with another information specialist on the team (CL).

Supplementary search strategies

The following supplementary search strategies will be used:

These strategies will be used for each included study, including studies found through the supplementary search strategies and through the grey literature search. In addition, any systematic reviews that are identified at any stage in the screening process will be targeted and their reference sections will be checked for relevant studies.

Screening

All database search results will be imported into EPPI-Reviewer 4 for title and abstract screening for each research question. Therefore, the search results will be screened in full twice. This will ensure that screeners are able to focus on identifying relevant studies for each research question in isolation. EPPI-Reviewer’s priority screening function will be used to improve the efficiency of title and abstract screening⁷.

Study selection

For the first research question, focusing on older people’s perceptions and experiences of housing with support, studies will be selected according to the criteria outlined in Table 1. Only studies that used one of five widely recognised qualitative methodologies – case study research, ethnography, grounded theory research, narrative analysis, and phenomenology – as outlined by Creswell & Poth (2017)⁸ will be considered for inclusion. Qualitative methodologies are often ill-defined, and we feel that it is important to limit our included studies to those that specify a methodology and, moreover, use a widely recognised methodology.

For the second research question, focusing on the impact of housing with support for older people, studies will be selected according to the criteria outlined in Table 2. Quantitative studies will not be excluded based on outcome or comparison group, because we did not want to risk excluding studies with outcomes that we may not have considered.

Disagreements regarding inclusion for both research questions will be resolved through discussion until consensus is reached.

Quality assessment and confidence in evidence

For the first research question, focusing on older people’s perceptions and experiences of housing with support, Joanna Briggs Institute’s Critical Appraisal Checklist for Qualitative Research will be used⁹. Additionally, the CERQual (Confidence in the Evidence from Reviews of Qualitative Research) approach will be used to assess confidence in qualitative findings. The CERQual method enables reviewers to transparently assess and describe the extent to which a review finding is a reasonable representation of the phenomenon of interest, such that the phenomenon of interest is unlikely to be substantially different from the research finding¹⁰. It includes four elements: 1) methodological limitations; 2) relevance to the review question; 3) coherence; and 4) adequacy of data¹⁰. These elements are used to assess overall confidence in qualitative findings. There are four levels of confidence: high, moderate, low, or very low¹⁰.

For the second research question, focusing on the impact of housing with support, quality assessment will be carried out using a tool for observational cohort studies from the National Heart, Lung, and Blood Institute of the National Institutes of Health in the United States of America¹¹. This tool uses 12 items to assess quality. We will also use British Medical Journal guidelines to assign levels of evidence to each quantitative study¹², and the GRADE certainty of evidence tool¹³ to write our strength of evidence recommendation for quantitative findings. In the GRADE tool, the levels of evidence range from one to four. The certainty of evidence can be high, moderate, low, or very low. The quality of evidence drives the strength of recommendation, which is one of the last translational steps of research and is most proximal to patient care.

For both research questions, each included study will be independently quality assessed by two reviewers, with any disagreements being resolved by consensus. Quality assessment results will not be used to exclude studies.

Data extraction

Data will be extracted by a single reviewer into a bespoke extraction sheet for each research question (see extended data⁶). For the research question focusing on older people’s perceptions and experiences of housing with support, the following data will be extracted: first author and year of publication, study title, country, housing model, study design, data collection method, dates of data collection, population age, number of participants, proportion of males and females, and findings. For the quantitative research question focusing on the impact of housing with support, the following data will be extracted: first author and year of publication, study title, country, housing model, study design, timepoints, participants at baseline, participants at follow-up, loss to follow-up, population age, proportion of males and females, and findings.

Extracted data will be verified independently by a second reviewer. Journal websites for the included articles will be checked for supplementary data and errata, and authors will be contacted if data required for quality appraisal and analysis are missing.

Data analysis

Thematic synthesis will be used to integrate the results of the included qualitative studies. Thematic synthesis has three stages: 1) line-by-line coding of text; 2) the development of descriptive themes; 3) and the generation of analytical themes¹⁴. We will use Dedoose (2020) software to code text and to develop our descriptive themes, which we will draw inductively from the included studies. The generation of analytical themes represents the stage of synthesis whereby reviewers integrate the primary studies and generate novel interpretations of findings¹⁴. We will develop analytical themes by revisiting the descriptive themes in light of the qualitative research question – what are older people’s perceptions and experiences of housing with support?

For the quantitative research question focusing on the impact of housing with support, we will conduct a feasibility assessment to identify the level of heterogeneity among the included quantitative studies, which will determine whether or not meta-analysis is suitable. A core assumption underpinning meta-analysis is that the studies being pooled are homogeneous; therefore, all sources of heterogeneity and variation must be assessed before conducting a meta-analysis. Our feasibility analysis will consider population, comparator, intervention, measurement scale, and length of time to follow-up. If meta-analysis is feasible, R software (2020) will be used, and we will conduct a meta-analyses for main outcomes as well as sensitivity and subgroup analysis by age, gender, and socio-economic groups. If meta-analysis is not feasible, we will present a narrative synthesis of the included quantitative studies using their summary statistics and vote counting.

Dissemination of information

Findings will be disseminated as peer-reviewed publications.

Study status

The database search was conducted in November 2019. Full-text screening was completed in May 2020. It is anticipated that the review will be completed in November 2020.

Data availability

Underlying data. No data are associated with this article

Extended data. Open Science Framework: Housing with support for older people: a mixed-methods systematic review protocol. https://doi.org/10.17605/OSF.IO/YU7S6⁶

This project contains the following extended data:

Reporting guidelines. Open Science Framework: PRISMA-P checklist for ‘Housing with support for older people: a mixed-methods systematic review protocol’ https://doi.org/10.17605/OSF.IO/YU7S6⁶

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).