This systematic review protocol follows the PRISMA-P reporting guideline [25]. The PRISMA-P Checklist can be found in the supplementary material. This review is registered on PROSPERO (ref: CRD42021221041).
Eligibility criteria
We will include studies that are written in English and fulfill the following criteria.
Topic of interest
The topic of interest will focus on the general health check, defined as “a service that aims to look at the general risk factors of different diseases using several screening tests to assess the general health” [11]. Using this definition, disease-specific screening such as cancer screening of any kind will be excluded.
Population
We will include studies with a target population that is inclusive of the population aged 50 years or above. We will exclude healthcare providers because this review aims to understand what interventions have been implemented on the service-users side to increase the uptake rate. We will also exclude people with diagnosed diseases, as the health-seeking behaviour of this population is likely to be cue by symptoms, which does not align with the definition of prevention, where the individual is asymptomatic.
Intervention
Intervention that is designed to improve the uptake of the general health check will be included. The intervention must be targeted at changing behaviour (e.g. general health check attendance behaviour). Intervention studies that compare the effectiveness of different screening modalities will be excluded unless screening modalities are adopted as a way to increase the uptake rate (as a way to change the targeted behaviour).
Outcome
The primary outcome will be the actual uptake rate. An actual uptake/attendance is defined as the full completion of a general health check service. We will exclude studies with a primary outcome measure of intention rather than action, for instance, intention to attend general health check attendance. These outcome variables will be excluded as these do not reflect an actual uptake behaviour.
Type of studies
We will consider empirical experimental studies that consisted of a control group and a comparison group, e.g. randomized controlled trials. We will exclude studies that were not published as empirical research articles, e.g. systematic review and research protocol.
Information sources
A literature search will be developed using the key terms related to “general health check”, “intervention”, “uptake”. These key terms will be combined with their alternative terms and together a search strategy will be developed. This search strategy will be entered into four electronic databases (e.g. PubMed, PsychINFO, EMBASE, and Web of Science) to obtain a list of potentially relevant literature. We chose these databases because of their relevance to social science, psychology and health-care related topics.
Search
The search strategy was developed and was discussed with a university librarian who has expertise in developing search strategies. The search strategy was discussed and moderated by the research team (WL, JL, MY). An example of the search strategy entered into PubMed is as follows:
((("general" AND "health check") OR "health check?" OR "health check" OR "preventive health check?" OR "medical checkup?" OR "medical check?" OR "comprehensive health check?")
AND
("uptake" OR attend* OR participat* OR utili?ation OR adher* OR appointment?)) AND
("intervention" OR "strategy" OR "strategies" OR "method?" OR "technique?")
The search strategy was entered into PubMed, PsychINFO, EMBASE and Web of Science on 18 August 2020. At this stage, no language and study design limits were imposed on the search strategy.
Study records
Data management
The retrieved records from the four databases will be imported to EndNote, where duplicates of records will be discarded. After this, a list of retrieved records will be generated and exported to Excel for titles and abstracts screening. The reason for any exclusions will be documented.
Selection process
Two reviewers (WL, AL) will independently screen the title and abstract of the retrieved articles by following the eligibility criteria. Any articles with uncertainty in their eligibility will be included in the full-text screening. The full-text screening will be conducted independently. When a source is not accessible on the publication website, additional help will be sought from the university librarian to get a copy. The reason for any exclusion will be recorded. Any discrepancies in the selection process will be discussed between the two reviewers. The final selection process will be illustrated in a PRISMA flow diagram [25].
Data collection process
A standard table will be used to collect the study characteristics of the included articles [26]. One reviewer will initially extract the study characteristics independently and the results will be checked by another reviewer to ensure accuracy. Other than extracting the generic information of the intervention study, we will also identify the behaviour change techniques used in each intervention.
Coding of behaviour change techniques
Behaviour change technique is the “observable, replicable and irreducible” component of an intervention [18]. Behaviour Change Technique Taxonomy (BCTTv1) is a 93-item hieratically structured system that helps to characterise intervention components. These 93 BCTs are grouped into 16 behaviour change groups [18].
Two reviewers will receive online BCT coding training before conducting the content analysis [27]. The coding process will adopt the following sequences:1) the reviewers will read the intervention description line by line until they feel familiar with the content; 2) they will follow the description and identify the ‘active ingredient’ and label the relevant BCTs.
A content analysis using the BCTTv1 will be conducted to code the BCTs implemented in each intervention. Content analysis is chosen because there is an established taxonomy, i.e., BCTTv1, used to identify the intervention components. By knowing the frequency of BCT identified, we could explore the possible relationship between BCTs in individual studies and across different studies.
Data items
We will extract the following study characteristics:
- Study design (control/ comparison group; sample size)
- Participant characteristics (such as age, gender, sample size)
- Setting (e.g. general practice setting)
- Intervention characteristics (intervention type, mode of delivery)
- Primary outcome measures
- Results (e.g. uptake rate for each intervention)
Outcomes and prioritization
The primary outcome is the general health check uptake rate. If a study has measured uptake rate at different time points, these variables will be reported in ascending order, e.g. 3-month period, 6-month period and 12-month period. This review will not specify any secondary outcome measures at this stage. If the eligible studies have included any secondary measures, we will also present them in the Results section.
Risk of bias in individual studies and across studies
We will use the Cochrane Risk of Bias tool for quality assessment [28]. This tool collects the following information to determine the risk of bias: 1) random sequence generation; 2) allocation concealment; 3) blinding of participants and personnel; 4) blinding of outcome assessment; 5) incomplete outcome data; 6) selective reporting. RevMan 5 will be used to document verbatim quotations from the eligible studies as evidence for the quality assessment. Two reviewers will assess the study quality independently based on the Cochrane Risk of Bias criteria [15]. They will label the study as ‘high risk’, ‘low risk’ or ‘unclear risk’ accordingly. Any discrepancies will be resolved through discussion between the two reviewers. The risk of bias graph of individual studies and across studies will be generated using RevMan 5.
Data analysis
Data analysis will include two sections to address the objectives. Objectives 1,3 and 4 will be addressed using narrative synthesis. Objective 2 will be addressed, if possible, by a meta-analysis.
Identification of intervention functions
Objective 3 aims to identify the intervention used to increase general health check uptake. A description of each intervention condition of the study will be presented narratively and which of the nine intervention functions in the BCW each intervention serves [17].
Identification of behaviour change technique
Objective 4 aims to identify the behaviour change techniques used to increase the general health check uptake rate. Content analysis will be used to identify the BCT and the result will be presented in the form of narrative synthesis. The result of the content analysis will be entered into the NVivo software. The inter-rater reliability of the result will be compared using Cohen’s Kappa [29], with an acceptable agreement level defined as a coefficient 0.6 or higher [30]. Any discrepancies will be discussed between the two reviewers. For any unresolved discrepancies, a third reviewer will be engaged in the final discussion. The final result will be presented in a table with illustrative quotations to support data synthesis. The narrative synthesis will address from three perspectives in the following sequence: the frequency of BCT identified, BCT reported effectiveness, and the possible BCT combination identified across studies.
Statistical analysis
The primary outcome will be compared using the absolute difference in the uptake rate between the randomized groups.
If there are sufficient data, a meta-analysis will be conducted to address Objective 2 to provide quantitative analysis for the overall intervention effectiveness. The heterogeneity between studies will be firstly assessed by visual inspection of forest plots. Then a formal statistical test of heterogeneity by using the chi-square test and the I2statistic will be used. The I2statistic implies the proportion of variation between studies not due to chance, varying from 0% to 100%.
The random-effect model which assumes the effect size could vary between studies will be used to estimate the pooled difference in the uptake rate across studies. In the case of multiple interventions within a trial, the groups will be combined to create a single pairwise comparison [31-33].