Patient safety in marginalised groups: a narrative scoping review

Background Marginalised groups (‘populations outside of mainstream society’) experience severe health inequities, as well as increased risk of experiencing patient safety incidents. To date however no review exists to identify, map and analyse the literature in this area in order to understand 1) which marginalised groups have been studied in terms of patient safety research, 2) what the particular patient safety issues are for such groups and 3) what contributes to or is associated with these safety issues arising. Methods Scoping review. Systematic searches were performed across six electronic databases in September 2019. The time frame for searches of the respective databases was from the year 2000 until present day. Results The searches yielded 3346 articles, and 67 articles were included. Patient safety issues were identified for fourteen different marginalised patient groups across all studies, with 69% (n = 46) of the studies focused on four patient groups: ethnic minority groups, frail elderly populations, care home residents and low socio-economic status. Twelve separate patient safety issues were classified. Just over half of the studies focused on three issues represented in the patient safety literature, and in order of frequency were: medication safety, adverse outcomes and near misses. In total, 157 individual contributing or associated factors were identified and mapped to one of seven different factor types from the Framework of Contributory Factors Influencing Clinical Practice within the London Protocol. Patient safety issues were mostly multifactorial in origin including patient factors, health provider factors and health care system factors. Conclusions This review highlights that marginalised patient groups are vulnerable to experiencing a variety patient safety issues and points to a number of gaps. The findings indicate the need for further research to understand the intersectional nature of marginalisation and the multi-dimensional nature of patient safety issues, for groups that have been under-researched, including those with mental health problems, communication and cognitive impairments. Such understanding provides a basis for working collaboratively to co-design training, services and/or interventions designed to remove or at the very least minimise these increased risks. Trial registration Not applicable for a scoping review.


Background
Improving patient safety is at the forefront of healthcare policy and practice across the globe [1] but may be especially challenging for marginalised groups of patients [2][3][4]. The European Network for social inclusion and health defines marginalisation as the "position of individuals, groups or populations outside of 'mainstream society'") [5]. Marginalised patients experience severe health inequities which can result in poorer health status, higher premature morbidity and increased risk for patient safety incidents in comparison to the general population [6] [2][3][4]. There are several reasons underlying these poor health care outcomes among marginalised patients. At the macro-level for example, marginalised people may have no voice on healthcare policy planning and/or resource allocation because they are "systemically excluded from national or international policy making forums" [5] [7]. At the meso-level, poor or non-inclusive organisational service designs can lead to gaps in service provision for marginalised patients [3]. Finally, at the micro-level, marginalised people may experience barriers to communication regarding their health care needs and treatment due to impairment or personal context (e.g. language barriers or sensory, learning or age related disability) [8,9] or as a consequence of perceived [10] or actual stigma enacted (e.g. labelling of some homeless patients as 'difficult' leading to barriers in accessing care) [3,11].
Although published reviews have sought to capture the nature, causes and consequences of patient safety incidents in various settings [12,13], to our knowledge, none have specifically focused on marginalised populations. A scoping review is particularly suited to when the aim is to identify and map out the literature as opposed to a systematic review, which typically aims to responds to a very specific well defined research questions for a specific patient group [14]. We therefore chose the scoping review approach in order to determine the range of patient safety issues and in which types of marginalised patient groups. In order to be inclusive when mapping this potentially diverse literature, we also chose the broader definition of 'patient safety issues' [15] as opposed to a specific patient safety incident, to enable consideration of wider underlying circumstance and complexities for patients from marginalised groups as opposed to those from the general population.
This scoping review examines the range of patient safety issues for people considered to be marginalised. Our four main aims were: 1) to identify which marginalised patient groups have been studied in terms of patient safety research, 2) to understand what the particular patient safety issues are for these groups and 3) what contributes to the safety issues arising.

Methods
This scoping review was conducted in accordance with the guidance for conducting systematic scoping reviews [16].

Definitions
In the absence of an identifiable agreed definition within the literature, we chose as stated above, the European Network for social inclusion and health's definition of marginalisation, which simply states that marginalisation is the "position of individuals, groups or populations outside of 'mainstream society" [5]. The definition is broad and reflects the fact that marginalisation in an umbrella term. Marginalised people however can be grouped due to them sharing common features or outcomes (e.g. reduced access to health services) as a result of their marginalisation, but may have other differing attributes (e.g. ethnicity, disability etc.) which lead or have led to their marginalisation. We do hypothesise however that marginalised groups may experience negative consequences or disparities in patient safety as a result of their marginalisation. Consequently, we also included studies utilising the terms 'seldom heard,' 'hard to read' and vulnerable groups.' The inclusion of these terms reflects the fact that they have also been used in the literature to represent the same groups designated as marginalised elsewhere in the literature. Hard-to-reach, for example, is a term cited by National Health Service (NHS) reports in the UK [17]. These reports acknowledge that certain groups are marginalised from services and therefore 'harder to reach' for health services whose goal is to provide appropriate and equitable health care for all populations. 'Seldom heard' groups have been defined as groups who may experience barriers to accessing services or are under-represented in healthcare decision making [18,19]. Finally, vulnerability has been defined as "susceptibility to any kind of harm whether physical, moral or spiritual, at the hands of an agent or agency" [20], a factor which "… needs to be recognised and negotiated in health care transactions." [21] . The Organisation for Economic Co-operation and Development (OECD) report into integrating Social Services for Vulnerable Groups defines 'vulnerable populations' as "people or households who live in poverty, or who are confronted with life situations that increase the likelihood of extreme forms of poverty [22]. These populations often face multiple risks and may require a range of services, from low-cost interventions such as food parcels, to more costly interventions such as housing, or mental or physical health care." Vulnerability can be identified as occurring as a result of one or more social, structural, situational or other causes. Such definitions and causes clearly have significant overlap with the definitions for marginalised groups and have clear applications to patient safety within a healthcare context.

Patient and public involvement
We worked with our patient-research partners in one of our departmental patient and public involvement (PPI) groups in the design of the study. Specific suggestions were made by the PPI groups and added to the protocol such as additional terms (e.g. care leaver) for the search strategy.

Data sources and search strategy
Six electronic bibliographic databases were searched from January 2000 until September 2019: MEDLINE, Embase, PsycINFO, CINAHL, ASSIA and Sociological abstracts. We selected 2000 as the start date of our searches because it coincides with when the published patient safety research began to increase in volume after the publication of the landmark report To Err is Human: Building a Safer Health System in 1999 [23]. Our search strategy (see Additional file 1) included search combinations of two key blocks of terms: Patient Safety and Marginalised groups. We used the standardised search strategy for patient safety used in previous patient safety reviews published by our research centre [24]. For the second block of terms, we used a combination of terms derived from two prior reviews on marginalisation (conducted in other topic areas) to represent the concept of marginalisation as well as terms that represent specific groups previously cited as marginalised [7,25]. We also supplemented these terms with additional terms in order to be as comprehensive as possible. Specifically, the supplementary terms include 'hard to reach,' 'seldom heard' and 'vulnerable groups.'

Eligibility criteria
Studies were included if they met the following criteria:

Inclusion criteria
Types of studies: empirical studies and systematic /scoping reviews. Study designs were not restricted and included both quantitative and qualitative studies including case studies; Types of participants: Patients who are considered to belong to a marginalised group according to the definition provided above; Types of outcomes: data on types of patient safety issues experienced by marginalised people and what factors lead to or were associated with these issues. Language: only studies published in the English Language.

Exclusion criteria
Studies concerned with a very specific drug or medical procedure rather than broader categories of patient safety issues; Studies concerned with people with a single health condition (unless they also concern a marginalised group); Studies that are solely focused on healthcare professionals; Studies that are not concerned with health care related safety (e.g. safety in the home, quality of care).

Study selection
Search results were downloaded first into Endnote and then uploaded and the review process managed via the use of the review software Covidence [26]. All citations deemed relevant after title and abstract screening were retrieved for subsequent review of the full-text article. Studies were assessed for inclusion by two independent reviewers (SCS and GDW) with arbitration by a third reviewer (MP).
Charting the data A form was developed by the authors to confirm relevance and to extract key study characteristics such as: 1) publication year, 2) publication type, 3) country, 4) economic level (as classified by the World Bank), 4) study aim, 5) population, 6) key safety outcomes and 7) contributing/associated factors related to the patient safety issues. This form was reviewed by the research team and pretested by all reviewers (SCS, GDW, AP, SG, LR and MP) before implementation. Six independent reviewers were involved in the data extraction. In particular, upon independently reviewing a batch of 20 to 30 articles, the reviewers met to resolve any conflicts and to help ensure consistency between reviewers and with the research question and purpose [27].

Data synthesis
The data were compiled in a single spreadsheet and imported into Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA) for validation and coding. Studies were then coded and grouped by SCS and GDW (any disagreements were resolved via discussion) according to 1) marginalised group, 2) patient safety issues and 3) contributing or associated factors according to the 7 different factor types from the Framework of Contributory Factors Influencing Clinical Practice within the London Protocol [28] see Table 1. The London Protocol was chosen as it can be applied to all areas of healthcare reflecting the diversity in settings across included studies. Study quality appraisals were not conducted in accordance with standard practice for scoping reviews.

Search and selection of studies
The original searches yielded 3346 potentially relevant citations. After completion of deduplication and screening, 67 studies met the eligibility criteria and were included in the review. The flow of articles from identification to final inclusion is presented in Fig. 1.

Description of general characteristics of included studies
An overview of the included study characteristics is provided in Table 2. All included studies were published between 2002 and July 2019. We identified 8 reviews [29][30][31][32][33][34][35][36] and 59 empirical studies [8,. The vast majority of these studies were conducted in high income countries (82%), used a mixture of methods (predominantly quantitative (66%)) and were conducted across multiple settings, with the majority (49%) in secondary care settings. Table 3 (supplementary material) provides details of the individual included studies.

Description of contributory/associated factors
In total, 157 factors, 2 mapped to one of 7 different factor types (from the London Protocol), contributed to or were associated with patient safety issues (see Table 5).
In the vast majority of studies (52 or 78%) the identified feature(s) of marginalisation (e.g. a patient factor such as frailty) led to patients in that group experiencing negative implications for their patient safety thereby leading authors to conclude that the characteristic itself was a contributing/associated factor to the patient safety issue of interest. Four studies reported no discernible/neutral effect [36,59,84,92], two indicated a positive effect on patient safety [55,88] and one mixed effects as two outcomes were measured and had different directions [93].  1 This study defined culturally unsafe practice as any actions that diminish, demean, or disempower the cultural identity and well-being of an individual and applied this concept to health care. 2 Within studies, multiple factors were counted individually e.g. if ethnicity and the person's condition where both identified, then these would be counted separately within the patient domain. Additionally, where studies indicated multiple instances of the same individual factor, these were also counted separately.
In 7 studies, no factors were identified [31,34,50,58,65,78,81] and in two, it was unclear [44]. Most studies reporting factors, discussed multiple individual factors (range = 1-7, average = 2.3) across multiple domains (range = 1-4, average = 2.0). The single largest domain concerned patient factors with 95 counts followed by individual staff factors (n = 27) and institutional context (n = 16). A brief summary of examples in each factor type is presented beneath and ordered by frequency from highest to lowest count.

Patient factors (n = 95)
This was the largest factor type, with 61% of all individual factors being identified as belonging to this category. We classified any contributing or associated factors that were either intrinsic to the patient or as a result of their social/economic/cultural characteristics as belonging to this factor type. There was wide variation in the types of examples, but a patients' race/ethnicity, their condition (mental and/or physical e.g. frailty, disability), issues in communication capabilities (language, disability or literacy) and help-seeking behaviour (e.g. route of admission, cultural beliefs, how they perceived themselves to be treated by clinical staff) were the largest sub-categories within this factor type.

Individual (staff) factors (n = 27)
Communication skills (e.g. perceived behaviour/manner towards patients) issues as well as knowledge/cognition based errors (e.g. errors in prescribing) were most commonly identified amongst coded examples. A lack of policy adherence/enactment by clinicians was also identified. However, an example of how this factor can positively impact on patient safety was seen in one study which hypothesised the outcome to be due to recognition of patient vulnerability (arising from their intellectual disability) resulting in more considered/careful behaviour by clinicians.

Institutional context factors (n = 16)
Access to care was the largest example of this factor, particularly access being moderated by the requirement for patients to make (co-)payments in order to access care. Policies in terms of a lack of, or lack of enactment as well as issues in transitions of care (e.g. lack of consideration and responsiveness to patient factors) were also identified as leading to patient safety issues occurring.

Organisational & Management Factors (n = 8)
Organisational policies availability and their variation in implementation was the primary example of this factor. Organisational size, specialisation i.e. staff and patient type within the organisation and responsiveness were also identified as impacting on patient safety.

Work environmental factors (n = 7)
Staff workload, shortages and time pressures led to patients' perceptions of staff 'busyness' and in one case, perceived patient neglect formed the coded examples in this factor type.

Task and technology factors (n = 2)
Only two occurrences of this factor were identified (the availability of communication tools and personally held written health information) and both arose from the same study concerning patients with communication impairment/disability. Both studies were reviews which did not extract data to allow the identification of this characteristic      Germany High "To examine the association between region of origin and severe illness bringing a mother close to death (Near miss)." "We examined the association between maternal regional origin and four indicators of maternal Near miss, namely hysterectomy, haemorrhage, sepsis and eclampsia, first to describe the association between maternal regional origin and near miss complications and second to test whether higher risks among women with migrant background can be explained by the women's socioeconomic characteristics, health-related behaviour and pre-existing ma-

Team factors (n = 2)
Only two examples of this factor were identified, across two separate studies and both concerned team communication.

Discussion
This scoping review brings together the published academic literature regarding patient safety in marginalised groups and included 67 studies in total. Most studies were from high-income countries and were quantitative (observational) in nature, designed to ascertain whether or not there was a discernible impact on patient safety as a result of the marginalised groups characteristics(s) investigated. Results revealed that in most cases, multiple contributing factors and factor types linked to marginalisation, appeared to lead to negative implications for patient safety. Medication related safety issues and studies around ethnicity constituted the two largest areas with existing evidence. This coalescence however also leaves many gaps in knowledge in the literature allowing for new research agendas to be clearly identified. What is clear, is that there is a relative paucity of patient safety research conducted with respect to marginalised groups in general and that this aligns with a recent priority setting exercise that highlighted vulnerable patients as the top research priority for patient safety research in primary care [96]. Common to studies showing a negative impact on patient safety, was the finding that the studied attributes from the particular marginalised group of interest and their interaction with the health system, created spaces or 'safety vulnerabilities' for patient safety issues to occur (or to be more likely to occur). In mapping the studies reviewed to categories according to the London Protocol Framework, the results of this review point to patient factors being the primary area as to where these vulnerabilities occur. However, many of these patient factors are not transmutable and are necessarily tied to social and organisational context [97], therefore an attempts to improve patient safety for people from marginalised groups requires the system and those working within it to respond and change appropriately. On the basis of the current evidence identified in this review systems, organisations, and those working within it, were for the most part seemingly unable to compensate for or respond adequately to these patient factors and our review highlights that the reasons for this (e.g. work-environmental factors, team factors) have not been well studied.
Access to high quality, safe health care is a fundamental indicator societal and health equity. The findings of this review highlight the need for high quality research to understand the patient, health provider and systemic factors which explain the present inability of health care organisations to provide high and equitable standards of care and safety to marginalised patients. Given that most incident reporting systems are limited in scope [98], explicitly listing marginalised patient groups at high risk for patient safety incidents requires immediate attention by policymakers and practitioners.
An important research implication is the need to acquire a deeper understanding of the underlying vulnerabilities of patient safety in marginalised groups of patients and design improvement strategies. Such understanding and improvements will require researchers to study and address the multi-factorial nature of patient safety issues and their occurrences drawing from a range of disciplines in order to address the multiple factors and issues identified ranging from the micro-level patient-provider interaction to specific and innovative service design to address macro-level issues such as the reduced access to care experienced by people from marginalised groups. A number of possible avenues could be productive. Firstly, existing theoretical frameworks can support a critical consideration of the relationships between patient factors, clinical interactions and wider organisational context of systems within patient safety research. For example, the social model of disability makes an important distinction between bodily impairment and disability and associated disadvantage created by environmental and social exclusion [99]. Secondly,the distinction between medical and social models also resonates with previous qualitative research on patient safety highlighting the tendency of patients to highlight the importance of psychosocial aspects of safety such as trust, communication and continuity [100]. Such issues are likely to be even greater concerns for groups where there is little current evidence, such as those with mental health problems, communication and cognitive impairments, or in specific contexts such as homelessness. In addition, research focusing on such groups entails consideration of intersectionality where multiple social markers (e.g. age, gender, ethnicity, socioeconomic status) may synergistically influence the degree to which people are marginalised, vulnerable, excluded or disadvantaged within care systems [101]. Exploration of these issues (quantitatively and qualitatively) will promote further understanding of the overlaps and distinctions regarding marginalisation and vulnerability, as well as an understanding of amenable contributors to patient safety. The identification and understanding of amenable factors for patient safety provides a crucial base for generating solutions and draws attention to additional avenue for further research focused on marginalised groups and patient safety: the co-design and evaluation of appropriate interventions to improve the quality and safety of care. Whilst there has been a growing acknowledgement on the need for patient and public involvement and engagement to achieve such improvements there is limited evidence of such work, even in relation to black and minority ethnic groups representing the largest marginalised group focused on in the literature reviewed here [102,103]. Furthermore, the drive for increasing digitisation within care services in many high income countries [104,105] can potentially increase any existing      [P]: 1) Health care communication difficulties due to patient nondisclosure of condition, 2) passivity and vulnerability, and 3) frustration with family Negative inequalities [106] and indeed create new and as yet unknown patient safety issues for marginalised people(s) [107]. Conversely, there are potentially opportunities for digital technology to reduce inequalities e.g. service gap provision. What is clear however, is that the development of any technology designed to ameliorate patient safety issues for marginalised people(s), will have to first understand the specific issues as a basis for co-design. This entails a focus on multiple dimensions of experience as discussed above; for example physical impairment as well as the material and interactional contexts where technologies are deployed [108]. New research to improve knowledge and understanding of patient safety risks for marginalised groups would also allow policymakers access to information as to where patient safety vulnerabilities are occurring and enable more effective planning and system responsiveness as well as evidence-based policies of inclusion, particularly those that recognize inequities in resources [109]. What is clear from this review, is that the field of patient safety research for marginalised groups has much scope for research, with many areas of patient safety and groups being under-researched.

Strengths and limitations
This is the first attempt to identify and analyse the academic literature for patient safety within marginalised groups. The study provides a clear platform for further research by highlighting where the gaps in literature are. We conducted systematic searches and double screened all studies. Identifying studies and key words for marginalised groups however was challenging. Thus, there is a possibility that some relevant studies were not included despite thorough attempts to do so. In addition, our focus on studies of marginalised groups meant that we excluded studies where health professionals were the focus and their views may have been different to those of the patients within marginalised groups. Only including studies in the English Language will have also affected the range of possible included studies and consequently meant that majority of studies were from high income countries. Furthermore, defining marginalisation is difficult and often overlaps with other concepts such as vulnerability. We have tried to be inclusive and used search terms from prior published reviews (and appropriate inclusion and exclusion criteria) and we established interrater agreement whilst determining the eligibility of the studies but admittedly operationalising marginalisation involves some degree of subjectivity. Finally, although we found that in the majority of cases, the features of marginalisation in the included studies appeared to lead to negative implications for patient safety for marginalised groups, we cannot say what the strength of this relationship is as scoping reviews do not aim to produce a critically appraised and synthesised result.

Conclusions
Our review identified a range of patient safety issues for people in marginalised groups, whether these groups are defined by social, economic, demographic or by other means of stratification. The findings indicate the need for further research to understand the intersectional nature of marginalisation and the multi-dimensional nature of patient safety issues, for groups that have been underresearched, including those with mental health problems, communication and cognitive impairments. Understanding which groups in particular are most likely to experience safety issues, what these issues are and why they occur in turn provides a basis for working collaboratively to co-design training, services and/or interventions designed to remove or at the very least minimise these increased risks.