This review adheres to published guidelines for conducting and reporting systematic reviews [18]. The review is not registered with PROSPERO.
Search strategy
Four electronic bibliographic databases will be searched: MEDLINE, Embase, PsycInfo and CINAHL. We will also identify eligible studies by checking the reference lists of those studies identified in the search that meet our inclusion criteria.
Our search strategy will include search combinations of three key blocks of terms: System/contributory Factors, Patient Safety and Primary Care, similar to those used in the previous review [14], except the context is primary care. An example of our search strategy and the terms used (in MEDLINE) is listed in Figure 2.
Eligibility criteria
Studies will be excluded if they fail to meet (a “NO” choice) any of the 3 criteria. Studies will be eligible for full-text screening if they fully (a “YES” choice to each criterion) or partly (one or more “UNSURE” choice) meet criteria A1, A2 and A3.
-
A.
For any study type (including review articles and opinion pieces):
-
1)
Is it an empirical research?
-
YES, NO, UNSURE
-
Is it worth continuing?
-
2)
Does it make reference to contributory factors to patient safety incidents?
-
YES, NO, UNSURE
-
Is it worth continuing?
-
3)
The research has been conducted in primary care?
-
YES, NO, UNSURE
-
Is it worth continuing?
We will include:
-
Types of studies: We will include empirical studies which provide data on factors that contribute to patient safety incidents in primary care. Study designs will not be restricted and will include both quantitative designs (that is, randomised controlled trials, quasi-experimental studies, cohort studies, cross-sectional studies) and qualitative studies including case studies. We will also include grey literature reports.
-
Types of participants: patients in primary care. We will not exclude participants on the basis of age or diagnosis.
-
Phenomena of interest: contributory factors of active failures or threats to patient safety. On the basis of the findings, an existing systematic review which examined contributory factors to patient safety incidents in secondary care settings, we anticipate that such contributory factors may include healthcare system/organisational factors (for example, communication failures between different professionals or patient-professional communication failures, staff workload, training/education, supervision/leadership, availability/use of equipment and supplies; policy issues; characteristics of the physical environment), health professional factors (inexperience, stress, personality attitudes) and patient factors (for example, language problems, personality characteristics, multimorbidity).
-
Setting/context: We will focus on identifying studies conducted in primary care. We will also include studies conducted in the interface of primary and secondary care. We defined primary care as ‘the medical care involving first contact and on-going care to patients, regardless of the patient’s age, gender or presenting problem’ [19]. We will not restrict our search in specific geographical areas or date of publication.
We will exclude:
-
Articles not published in English (for pragmatic reasons such as translation difficulties)
-
Non-empirical studies (primarily because we aim to build our conceptual framework based on empirical evidence rather than theoretical hypotheses and views that have not been empirically tested. Additionally, we do not expect to find any relevant systematic reviews of empirical studies given the lack of systematic evidence in this research area)
-
Studies that report only patient safety incidents without providing information on factors that may account for these incidents
-
Studies relating to home care (considered to be contextually different from general primary care settings).
Management of search outcomes and study eligibility screening
The results of the searches of each database will be exported to an Endnote reference management database (version X4) and merged to identify and delete duplicates.
We will use the taxonomy proposed by intra-EuropeanLINNEAUS Euro-PC collaboration for abstract and full-text screening. According to this taxonomy, patient safety incidents fall into three categories, access of health care (that is, incident related to availability, accessibility, accommodation, affordability and acceptability of health care), clinical task (that is, incident related to history/examination/problem identification, diagnosis, treatment, delivery, rehabilitation, prevention) and organisational task (that is, incident related to administration, supervision/management, maintenance, payment).
Using PRISMA guidelines [18], screening will be completed in two stages (see Figure 3). Initially, the titles and abstracts of the identified studies will be screened for eligibility (see ‘Eligibility criteria’ section). A proportion of titles and abstracts (50%) will be screened by two researchers independently to assess reliability using the kappa statistic. Assuming reliability is confirmed, screening of the remaining titles and abstracts will be completed by one reviewer.
Eligibility criteria
Next, the full texts of studies initially assessed as ‘relevant’ for the review will be retrieved and checked against our inclusion/exclusion criteria. Full-text screening will be completed by two researchers independently, with disagreements resolved by discussion.
Methodological quality of the studies
Quantitative studies: We expect the main body of the research included in this review to be observational studies (cross-sectional or prospective). Thus, for the main bulk of anticipated studies (observational studies), we decided to assess the methodological quality using criteria adapted from guidance on the assessment of observational studies [20]. The quality review will include assessment of the design, conduct and analysis of each study and will be used as a framework for the narrative synthesis of the results. Three key criteria will be used to conduct the quality review, and each study will be awarded one point for each criterion met These key criteria are:
-
1)
A response rate of 70% or greater at baseline
-
2)
Adequate control for confounding factors in analysis
-
3)
A follow-up rate of greater than 70% (in prospective studies).
These criteria allow rapid evaluation and have been previously used by members of our research group to assess the methodological quality of observational studies [21]. The criteria will not be used to exclude papers prior to the synthesis; rather, it will be used to provide a context for the interpretation of the findings. The methodological appraisal of any randomised controlled trials (RCTs) identified in this review will be assessed using the Cochrane risk of bias assessment tool [20]. However, if a range of different types of experimental studies are identified including RCTs, non-randomised trials, controlled before after studies and time series studies, the methodological quality of those studies will be assessed using nine standardised criteria developed by the Effective Practice and Organisation of Care (EPOC) [17].
Qualitative studies: The methodological appraisal of qualitative studies indentified in this review will be assessed using the Critical Appraisals Skills Checklist (CASP) for qualitative studies [22].
Each paper will be independently appraised by two reviewers, and discrepancies will be resolved by discussion.
Data extraction
A data extraction sheet will be devised that will include the following:
-
Study characteristics - year, objective, research method, setting, country.
-
Participant characteristics - number, age, gender, diagnoses.
-
Patient safety incident type and characteristics
-
Practice characteristics
-
Main outcomes
-
I.
Contributory factors to patient safety in primary care
-
Results of the study quality appraisal.
Data extraction will be completed by two researchers. Disagreement will be resolved by discussion until consensus is reached.
Data synthesis
The outcomes of the systematic review will be organized and presented descriptively. The heterogeneity of the research designs and outcomes of this review are unlikely to allow the use of formal meta-analytic procedures.
On the basis of a previous paradigm in hospital settings [14], a contributory factor framework will be developed. Studies will be categorised by types/domains of contributory factors to patient safety (that is, communication failures, equipment and supply, active failures and so on) in primary care. Coding contributory factors into different domains will be conducted by two authors independently. In addition, expert advice (RL) will be sought about the contributory factor coding process. Depending upon the consistency in reporting methods and the number of studies retrieved, we aim to further group studies according to study design and methodological quality. This synthesis will be performed by the lead author (SG) and then reviewed independently by co-authors. Disagreements will be resolved by consensus.