Background

Autism has historically, and incorrectly, been pathologised in research and discussed using deficit-based language in line with the medical model [1]. Autism is a normal part of cognitive diversity, resulting in differences to communication style and sensory processing which are often experienced as disabling impairments in a neurotypical world [1], as described within a neurodiversity paradigm [2, 3]. Autistic people have worse health compared to their neurotypical peers [4]. This includes significantly worse physical and mental health and a lower life expectancy [5]. This is associated with systematic stigmatisation [6]—including by healthcare professionals [7]—and healthcare inaccessibility for Autistic people [8]. Autistic adults’ experiences of health and healthcare have been severely under-explored in research [9].

Health passports (HP), also known as hospital passports, are a digital or physical source of information regarding patients’ care needs and preferences designed to aid information transfer between patients and healthcare staff [10]. It has been argued that the use of HPs can address the differences in communication that may cause misunderstandings between patients and healthcare professionals, improve patient safety, and promote person-centred care [11]. HPs are many and heterogenous, and can include: personal details, contact information, communication needs, care needs, preferences, [12] signs of pain and distress, a medical history, and information on consent and capacity [11]. There is no current standardised approach to the development or implementation of HPs [11, 13–15] and a realist review of their use for medically complex children has found that more attention needs to be paid to the needs of patients, staff and organisations which are intending to use HPs [16].

Within healthcare systems, patient-centred tools have previously been developed and implemented to reduce barriers to care. For example, Greenberg et al., [17] evaluated the impact of the Asthma Passport, a patient-centred tool, finding it both significantly increased the number of patients who completed an Asthma Action Plan (AAP) and Asthma Control Test (ACT) within a clinical setting, and increased satisfaction with the care itself. Similarly, "Kardio-Passes", a patient passport containing relevant diagnostic and treatment data for those patients who have experienced cardiovascular diseases (CVD), have been rated by patients as a helpful tool for documenting follow-up data in rehabilitation [18]. It may be that the potential for observable clinical deterioration, or even death, related to uncontrolled asthma and cardiovascular diseases have been responsible for these passport type tools leading to benefits in these clinical specialities and in primary care clinics. Another example of patient-centric tools are various birth plan templates which aim to increase child-birth related patient satisfaction. Collaborative birth plans have been associated with positive birth Outcomes, [19] however other research has shown a reduction in patient satisfaction and control [20] and insufficient evidence to support or refute that birth plans improve birth experiences or increase satisfaction [21].

Autism-specific HPs (AHPs) have been recommended within UK policy documents [22–24] and NICE clinical guidelines [25]. AHPs are recommended to support Autistic people to access more equitable healthcare. Some policies suggest the use of generic HP tools, for example, the Welsh Government Autism Code of Practice, [24] encourages NHS Health Boards and Trusts to use the Once for Wales Health Passport. In their guidance notes for health professionals, [26] it is recommended that staff familiarise themselves with the contents of the profile and inform their colleagues of important care-related details, as well as noting that any changes should be made by staff on the HP. Examples of specific AHPs for adults within Europe include tools developed by the UK National Autistic Society [27] and the Healthcare Passport for Children and Adults on the Autism Spectrum developed by Ireland’s National Autism charity AsIAm [28]. Title two of The Americans with Disabilities Act prohibits discrimination against disabled people in both public and private hospitals and health facilities [29] and requires modified policies and procedures, such as the use of auxiliary aids, to improve communication and remove barriers to accessing care. In North America, the Autism Healthcare Accommodations Tool (AHAT) was developed as part of the AASPIRE Healthcare toolkit [30] and has been referenced within the Australian Journal of General Practice, [31] alongside the NAS My Health Passport as examples of resources to aid in the provision of care for Autistic people. In addition, the My Autism Passport App has been introduced to Canada [32].

The potential of AHPs to reduce health inequalities is often stated without reference to evidence of their effectiveness in policy documents, [22] clinical guidance, [25] and guidelines accompanying AHPs within individual health services [33]. When the use of AHPs for adults is evaluated, evidence of limited usage and implementation challenges can be found, [34–36] including one survey which identified only 4% of respondents used an AHP, but 30% would use one if given by their primary care doctor [37]. The theoretical underpinnings of AHP interventions have not yet been thoroughly considered. Throughout the rest of this article, we refer to all health passports as HPs, even if they are Autism specific, because many have multiple intended populations.

Methodology

A systematic literature search and realist review was undertaken following guidance within the Preferred Reporting Guidelines for Systematic Reviews and Meta-Analyses (PRISMA), [43] and the RAMESES II reporting standards for realist reviews [44]. The protocol was prospectively registered with PROSPERO (registration ID: CRD42022304756).

Search

We identified search terms by undertaking test searches based on hand searching keywords in relevant papers. The search strategy, developed with the support of a specialist librarian, involved two main terms, relating to (i) Autistic people and (ii) HPs (see Table 1). Five electronic databases were searched to ensure wide coverage within bio-medical and social science disciplines (Medline via OVID, PsychINFO via Ebscohost, CINAHL via Ebscohost, Web of Science via Clarivate, and SCOPUS via SciVal). Databases were searched from 2010 to January 2022 and limited to humans. Review articles identified in database searching were unpicked to identify additional papers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Search terms.

https://doi.org/10.1371/journal.pone.0279214.t001

Study selection and eligibility criteria.

Papers were independently screened by two authors (RE and AG) at two stages: title and abstract; and full text. We did not exclude sources based on research design or methodology and commentaries were included. Due to limited resources for this study and the relative newness of AHPs, we included literature from the past 12 years. The following inclusion and exclusion criteria were utilised:

Inclusion criteria

• Population: Autistic adults (≥16 years) and those involved in supporting access to healthcare or providing healthcare to Autistic adults
• Context: any healthcare setting
• Phenomenon: HPs that were either (i) exclusively for the use of Autistic people, or (ii) were for a broader population of patients and reported on their use with Autistic patients.

Exclusion criteria

• Did not focus on HPs for Autistic adults (≥one paragraph of relevant text) e.g.:
  • No passport-type tool described
  • Passports focused exclusively on something other than health, such as social care or education
  • HPs for non-Autistic populations
  • HPs for children aged under 16 years
• Published pre-2010
• Full text not available
• Full text not available in English

Quality assessment, data extraction and realist synthesis methods.

Included sources were extracted independently by two authors (RE and AG) for core information relating to research design, if applicable (See Table 2), quality for realist review (See Table 3), and for contents relating to Context, Intervention, Mechanisms and Outcomes (See Table 4). Realist principles [45] were used to consider the impacts of Context on intervention components and intervention Mechanisms [42] in acknowledgement that interventions will not work in all Contexts, and that the ‘messy’ Context of healthcare can be particularly challenging [46] and thus impact on Outcomes. A summary document of the realist synthesis was presented to all authors to discuss and refine concepts and to identify areas of salience between sources. This was completed on two occasions, until consensus was reached.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of included papers.

https://doi.org/10.1371/journal.pone.0279214.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Quality of studies for realist evaluation.

https://doi.org/10.1371/journal.pone.0279214.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Context, mechanisms and outcomes of included papers.

https://doi.org/10.1371/journal.pone.0279214.t004

Results

A total of 162 unique records were identified through database searching, and 30 papers were sought for full-text review (see Fig 1). Two sources were unable to be retrieved via inter-library loan; 12 sources met the eligibility criteria. Ten reviews were identified and unpicked, identifying one additional paper, leading to a total of 13 items included in the review. Whilst the authors prefer identity-first language, and reject the notion that Autism is a disorder, we have used language from the original text when discussing the literature below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram study design, methodological quality, populations and settings.

https://doi.org/10.1371/journal.pone.0279214.g001

Most of the included sources originated from the USA (n = 5) and the UK (n = 4) (see Table 2). Some of the tools described were only for Autistic patients (n = 6); others were not Autism-specific but they reported on health passport use by Autistic people (n = 7), as well as people with Intellectual Disabilities (ID), or Intellectual and Developmental Disabilities (IDD), referred to as a Learning Disability within a UK Context. Content that was not relating to the use of health passports by Autistic people was excluded from our analysis. The most common target populations for the HPs included: Two of the tools were focused on the transition from paediatric to adult healthcare providers [47, 48]. Most interventions were focused on a general hospital setting (n = 4), whilst others were for specific environments such as emergency departments (EDs) (n = 2) [49, 50]. Most sources reported on a single HP tool (n = 8); others were part of a wider toolkit [50] or intervention [48]. Five sources discussed or evaluated tools which sought to improve communication [34, 48–51]. These included the testing and re-testing of a particular intervention [34] or piloting an intervention and seeking stakeholder feedback [34]. Four sources did not report a research study, and were descriptive in nature, including commentaries and blog posts. Other sources described research or quality improvement initiatives but did not include details regarding data collection and analysis [34] or described a tool without detailing its evaluation or creation [52].

Realist evaluation

Within this section, we present the contents of the included sources following a Context, Intervention, Mechanisms and Outcomes (CMO) format, which is summarised for each study in Table 4.

All sources included some contextual elements, but these varied in their depth and the population, for example considering the Context for patients with ID. Interventions ranged from a singular HP tool to multi-stage (development and evaluation) and multi-component Interventions. There were also variations in the contents of the HP tools. Mechanisms were more often implied by the researchers than explicitly noted by authors, and the vast majority of these did not reference a particular theory. Only 4 sources included quantitative Outcome measures, and those which did tended to report low usage of HPs. Only two sources contained sufficient detail to allow the development of full CMO configurations; the remaining sources lacked Outcomes measures that were directly associated with Mechanisms.

Context.

In this section we discuss the Context in terms of population, setting, staff and known issues. Within the 13 sources, the most frequently discussed intervention target population were Autistic people (n = 8), with other populations including individuals with ID/IDD (n = 5), Parents or Families (n = 2), children (n = 1), those with severe disabilities (n = 1), professionals (n = 1) and those with LDs (n = 1) (see Table 4). Among papers that described research participants in the evaluation of HPs (n = 5), the demographics, and sometimes even the number, of participants in these samples, were often poorly described. Four papers which detailed participants, listed the professionals involved, [30, 47–49] with one mentioning a multidisciplinary team but not specifying any further details [51]. Two papers mentioned age, [30, 47] sex [30, 47] and ethnicity [30, 48] of participants. Nicolaidis et al. described their participants the most, including race, participant education level, living arrangement and level of assistance needed [30].

Most papers (n = 10) focused on a hospital setting, including hospitals in general (n = 4) and emergency departments (n = 2). Staff members anticipated to be recipients of HPs included hospital staff, community staff, or health professionals in general.

There was considerable overlap in the known contextual factors affecting healthcare equity for Autistic adult patients, including a lack of staff training (n = 11) and corresponding knowledge of frequent co-occurring conditions (n = 10). Specific barriers identified as impacting Autistic healthcare experiences included communication differences (n = 6) and sensory difficulties (n = 2). It is important to note that these barriers are not caused by the Autistic qualities or Autistic people themselves, but by the interaction of these traits within poorly adapted services and environments [8]. Barriers within healthcare systems included staff shortages resulting in a lack of time (n = 5) and inaccessible or inappropriate hospital settings (n = 4). Contextual challenges were identified during the transitional period from paediatrics to adult care (n = 2) [47, 48]. Furthermore, patient dissatisfaction and a need for wider systematic change [52] were also described as known issues.

Interventions.

When reviewing interventions, we included both HP tools, and wider interventions around HP tools, where applicable (see Table 4). The majority of sources (n = 7) did not describe the development of their intervention. Others provided limited details, such as naming the team the HP was developed by, [51] but not necessarily how this was achieved. Three papers mentioned an element of co-production, such as the use of service users in the evaluation, [30, 51, 53] or development of the tools [30, 53]. Two papers mentioned the use of a pilot as part of the HP development [30, 51]. Four sources noted that the HP was an aspect of a wider intervention, including projects focused on: transition from paediatrics to adult primary care services, [48] improving quality of care [34, 55] or as part of a healthcare toolkit [30].

There was variation in the depth of HP descriptions, with some sources containing detailed information, including visual representations of the tool itself, [48, 50] whereas others only mentioned a few elements within the HP [56]. Table 5 compares the HPs contents against the elements identified by Northway et al. [11] in their review of HPs for people with learning disabilities. Erickson Warfield et al. [47] and Perkins and Vanzant [52] were excluded from Table 5 as they both included descriptions of more than one tool. The most frequently included elements in HPs were levels of communication (n = 7), name (n = 5) and date of birth (n = 5). We interpreted “maintaining comfort” as attempts to reduce distress, for example: "ways to help me avoid distress", [50] and "accommodations to help patient stay calm and comfortable" [30]. Some HP elements described by Northway et al. [11] were not included in any of the tools, including: advanced care planning (do not resuscitate orders), next of kin, contacts in relation to discharge, requirements for an interpreter, how I communicate when I feel well or unwell, the person completing the form, oral hygiene needs, risk assessments, reasonable adjustments required, best interests meeting and/or decision, sleeping, behaviour or support with toileting. However, in addition to the elements listed by Northway et al., [11] the following items were identified from our included sources: impairments, triggers, interests, strengths, personality, physical placements for passport and ethnicity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Elements included within HPs.

https://doi.org/10.1371/journal.pone.0279214.t005

Three sources specified that HPs were intended to be completed by patients and caregivers [51–53] and three sources by staff [55, 51, 57]. In terms of training to facilitate use of HPs, six sources specified training, six did not and one was unclear as to whether it was provided. Training was provided for staff (n = 5), family and caregivers (n = 4) and Autistic individuals (n = 1), with some projects offering training to two or more stakeholder groups. Other papers focused on a wider intervention, with less of a descriptive focus on the HP tools themselves. For example, Harris et al. [48] noted the use of a HP to aid with transitions to adult care but did not include a reminder to complete a HP within their own transition checklist aimed at clinicians. Two sources did not describe the contents of the HPs discussed [34, 57].

Mechanisms.

For two papers no information was provided as to the intended use of the intervention [48, 49]. Within our evaluation, we divided Mechanisms into those stated explicitly by authors and those implied by the research team. 11 papers reported at least one explicit Mechanism, however, Mechanisms were rarely clearly stated, and often lacked detail. Explicit Mechanisms that were based on robust theorisation were largely absent. An exception was Nicolaidis et al., [30] who utilised self-efficacy as part of their theory of change. Some additional explicit Mechanisms described within this study were the use of personalised patient information as an acknowledgement of heterogeneity within the Autistic population and identifying “Potential leverage points” (p.1181) from the research to explore in future practices. Also worthy of note, Lalive d’Epinay Raemy and Paignon [55] described multiple explicit Mechanisms (n = 8), including the need for establishing trust and improved communication between disabled individuals and professionals, reducing emergency admissions, and improving the accessibility of hospital buildings. Additionally, Heifetz and Lunsky [49] included several explicitly stated Mechanisms (n = 7), including the use of an implementation champion to aid embeddedness, and the suggestion that a co-ordinated approach and higher awareness of the tool would increase usage, improve embeddedness with staff and reduce the repetition of information by patients and carers [49]. An additional three papers described "some" theory behind the intervention.

Across the included sources, the most frequently described explicit Mechanisms included: improving knowledge for both patients and clinicians (n = 4), strengthening care and person-centred practices (n = 4), and embedding the tool within systems (n = 3). Other potential Mechanisms were poorly described and under theorised, for example, expecting that a HP would "increase quality of care", [50] "reduce anxiety for families" [54] or "improve communication" [55] without providing further details as to how this would be achieved. Our analysis of the sources resulted in a range of implicit Mechanisms being generated by the research team including: increasing accessibility (n = 2), humanising the Autistic patient (n = 2) and ensuring the tool is embedded within the system (n = 1). Other implicit Mechanisms mentioned include creating more holistic care [51], empowering staff and increasing quality of care, [53] changing staff behaviour [56] increasing acceptability through professional endorsements therefore increasing the acceptability of the tool, [52] easing transition between services [47] and updating the tool regularly to ensure relevance [51]. Implicit Mechanisms were not derived from five papers (Table 4).

Outcomes.

Of the 13 sources included, four contained at least one quantitative Outcome. Half of the papers did not include either quantitative or qualitative Outcome data (n = 7). Most quantitative Outcomes were focused on usage of HPs, the majority of which reported low usage. For example, Heifetz and Lunsky [49] demonstrated that from their small number of participants (n = 28), 82% did not have a chance to use their HP in the one-month follow-up period. Likewise, the transition checklist included in Harris et al. [48] intervention was shown to clinicians in 100% of appointments, aiming to prompt discussion about transition to adult healthcare services, however it did not include reference to their HP tool and was completed by clinicians in only 44% of appointments. In addition, Nicolaidis et al. [30] developed and tested an intervention which resulted in significant reductions in barriers to care, increased self-efficacy and higher a satisfaction of patient-practitioner communication. In this study, Autistic individuals said that the toolkit was useful (95%), and that they would recommend the toolkit to a friend (92%), or healthcare professional (95%) [30]. Two studies reported on usage without including numerical Outcomes. First, Lalive d’Epinay Raemy and Paignon’s evaluation [55] of the Disability Admission sheet stated that it had been completed for nearly all disabled individuals within Geneva. Second, by contrast, within Kelbrick [34] usage of the HP tool was described as "limited".

Seven sources included qualitative data focused on the views and experiences of those using the HP or wider intervention. Within these studies, a range of methods and participant groups were used, with some qualitative Outcomes reported by authors without obvious sources for their assertions (see Table 2). Qualitative findings included HPs facilitating better communication, [49] and increased understanding of the patient [57]. However, there were concerns regarding the practicalities of implementation surrounding the lack of staff awareness of HPs [49] and the inexperience of the professionals who were using them [57]. Several barriers to HP usage were noted by families and healthcare professionals such as the additional time needed to implement accommodations, Nico 2016 and reduced physician engagement [48].

CMO configurations.

CMO configurations are created to understand causality within realist evaluation and to identify if an intervention is working, in which Contexts, for whom, and to what extent [41]. This is achieved through consideration of the underlying Mechanisms that were or were not triggered in the evaluation Context [58]. Due to a lack of theories of change and/or Outcome measures, the majority (n = 11) of the literature reviewed was unsuitable for generating CMO configurations. In Box 1 we present CMO configurations for Nicolaidis et al. [30] and Harris et al. [48]. We do not feel that it would be appropriate to create an overarching CMO configuration for HPs, due to the weakness of the included evidence.

Box 1: CMO configurations

Nicolaidis et al.’s intervention [30] was conducted within the United States, was intensively co-designed, using community-based participatory research (CBPR), and research evidence from the team’s prior participatory studies. The low intensity online toolkit, which had separate versions for Autistic people and supporters, included a HP, health information for Autistic people and additional resources. The intervention was tested in a Context where Autistic patients had unmet and complex health needs, and primary care providers (PCPs) lacked knowledge, made assumptions, and were unwilling to make accommodations. Mechanisms that were triggered, resulting in desired Outcomes, included high levels of community acceptance of the toolkit, including ease of use and likelihood to recommend, which may have been achieved through the participatory design process. Additionally, the majority of participants agreed that their HP could be mailed to their PCP in advance of appointments, reducing barriers to use within consultations. Alongside acceptability to patients, the majority of PCPs found it useful as well. Accordingly, the toolkit increased Autistic patients’ self-efficacy and reduced barriers to accessing healthcare. Among a minority of Autistic patients who saw their PCP within one month, satisfaction with care quality was increased. Qualitative data suggested that self-efficacy was improved by Autistic patients being aided by the tool to prepare for visits and allowing for more effective self-advocacy within appointments. Longer term follow-up of patients, and a more comprehensive evaluation from PCPs would have aided the strength of these findings.

Harris et al.’s intervention, [48] focused on the transition from paediatric to adult healthcare for Autistic young people and was developed by a “transition team” (p.755) that included neither young nor Autistic people but included the parents of an adult with cerebral palsy. The intervention was tested in a Context where adult PCPs were unwilling and untrained to care for Autistic adults, and Autistic young people had high rates of co-occurring conditions, including mental ill health. Furthermore, there was inadequate support for both groups around transition. Mechanisms were sometimes, but not always, triggered, impacting on Outcomes. For example, electronic prompts displaying transition checklists were displayed during every appointment, although lack of training for and/or confidence in PCPs, alongside inadequate appointment length, meant that this Mechanism did not result in the completion of the checklist, let alone the HP tool, more than half of the time. In recognition of the time constraints, PCPs could refer patients to a social worker for a specific transition appointment, which resulted in a social worker completing a transition checklist for almost half of eligible patients. However, the electronic prompts for PCPs (and presumably social workers) did not include reference to the HP element of the intervention, likely reducing the number of times it would have been discussed by PCPs and social workers. We therefore theorise this absence would reduce the number of times the HP tool would have been used by Autistic patients, although there is no Outcome data to support this. Additional potential Mechanisms around supporting transition did not have corresponding Outcome data including the impact of a half day training event for families.

Discussion

This realist review included 13 sources focused on HPs for Autistic adults. The papers were of varying quality with only one meeting all eight criteria we included in our quality assessment for realist review. The majority of sources described contextual information and at least some intervention details, but most were lacking in Mechanisms. Only four sources included quantitative Outcomes. Nicolaidis et al., [30] was the most highly rated source in quality, and demonstrated increased measures of self-efficacy, reduced barriers to care and increased satisfaction following appointments amongst passport users. However, within this study, it is unclear as to how much of the impact was related to the HP tool, compared to the wider toolkit intervention.

The UK Medical Research Council guidance for the development and evaluation of complex interventions [59] provides a framework that researchers can follow to develop well-theorised interventions. This includes identifying problems, developing theoretically informed potential solutions, and evaluating those interventions, to ensure they have the best chance of succeeding in their current Context. Within our review, no paper other than Nicolaidis et al. [30] showed clear adherence to this iterative process. Other sources were identified as quality improvement initiatives, which busy health professionals juggled alongside their clinical duties, often relying on “common sense” and “ground-up” approaches due to high workloads impacting on time available to develop and evaluate interventions [60]. Furthermore, where interventions are moved to new Contexts, it is acknowledged that they may need to be re-designed and subjected to further evaluation, [61] for example, Heifetz and Lunsky [49] found that when three groups of patients were asked to update an existing tool, it varied significantly across the three Canadian study sites.

Both government policies [24] and much of the literature we reviewed suggest that HPs improve care for Autistic people and should be widely utilised. Although HPs might have potential, their use and efficacy is still in its infancy and the varying quality of current research impedes the replication and evaluation of these studies. The literature included within this review described barriers to accessing care, including communication, staff attitudes and training. Our realist evaluation has therefore shown that the recommendation to roll out HP tools to reduce health inequality for Autistic adults is currently unwarranted. Despite the current lack of widespread evidence of efficacy, our analysis identified some potential in interventions which included a HP alongside a wider toolkit. For example, those which included service users in intervention development could be an important mechanism to drive change [30, 55]. Furthermore, recurrent contextual barriers to high-quality healthcare for Autistic adults, and Mechanisms that were shared across multiple interventions show that there is some common ground on which tools can be developed.

However, any benefits have not been seen consistently in relation to asthma [17], and there is also conflicting evidence regarding the value and feasibility of birth plans [21]. A realist lens would suggest that even when there is a ‘life or death’ clinical implications of not using a tool that a sufficiently challenging context can reduce the feasibility of using tools. This has been noted by Nicolaidis et al: "…It is likely that an accommodation report may not be (…) sufficient to eliminate all constraints affecting PCPs’ ability to care for Autistic patients…"[30, p.1188]. We summarised the contextual barriers we extracted and inferred from included sources graphically in Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Graphical representation to contextual barriers to AHP successful implementation.

https://doi.org/10.1371/journal.pone.0279214.g002

Implementing a HP tool into healthcare systems will have limited impact if the wider systems are not also changed as care is "…a complex interplay between an individual’s Autistic characteristics, the healthcare provider’s knowledge and attitudes about Autism, and the healthcare system"[30, p.1181]. For HPs to achieve patient benefit, there is a need for systems that allow healthcare practitioners to engage with these tools in a useful and sensitive way, for example including longer consultation times and training to increase knowledge and confidence of how to support Autistic patients. If barriers to health professionals using HPs in consultations were removed, social factors such as the routine normalising of neurotypical behaviours and communication which result in the widespread stigmatisation of Autistic people would still be likely to influence the likelihood of Autistic individuals opting to self-disclose an Autism diagnosis. In a governmental review of ‘Think Autism’, [23] Autistic respondents felt that disclosure led to diagnostic overshadowing. Some Autistic people also have reported withholding their diagnosis in healthcare appointments as they felt it would negatively affect the treatment they received [62]. HPs cannot be embedded without a respect for Autistic communication. There is a need for a cultural change to reduce stigmatisation by healthcare staff [7] and provide a supportive sensory environment for Autistic people [63]. Accordingly, training and resources for clinicians must be developed and delivered by Autistic people, based on a neurodivergent-affirming model of Autism, as has been present in the Social Model of Disability since the 1960s. Working with Autistic clinicians will also increase the likelihood of acceptable interventions [64].

However, we acknowledge that this will not be easy to achieve; in one of the papers within this review a 5-day training course was initially envisaged but was replaced by 15 minute "on ward" sessions, [55] showing the significant pressures within the system and barriers to implementing training [65].

With regards to the hospital setting itself, there are significant sensory barriers for Autistic staff and patients [66]. Best practice guidance has been developed to help attend to sensory challenges within inpatient environments, [67] although this has not yet been routinely implemented. Part of making healthcare more accessible to Autistic people involves creating Autism friendly environments through cultural and systemic changes; for this to be more than tokenistic, Autistic voices must be centred within these efforts [68]. Organisations such as Autistic Doctors International [69] and the Maternity Autism Research Group [70] can play an important role in advancing change, due to members’ dual status as clinicians and Autistic people.

Conclusion

Our review highlights that there is currently insufficient evidence, and the evidence that exists is low quality. Accordingly, we conclude that HP tools do not improve the accessibility of healthcare for Autistic adults. There is a legal duty in the UK, USA, and many other countries for services to make reasonable adjustments in order to ensure disabled individuals can access the healthcare they need, and it is clear that interventions to facilitate this for Autistic adults are very much in their infancy. To date, there has been inadequate inclusion of Autistic researchers and lay Autistic co-researchers in many studies; this is based on an outdated deficit-based understanding of Autism. If researchers are serious about improving Autistic adults’ health, they must utilise strengths-based understandings of Autism which challenge neurotypical conventions and value the assets that Autistic people bring. When designing new tools to reduce the health inequalities Autistic adults face, researchers must take account of the barriers inherent in the intervention Context, ensure that appropriate theories are used in designing interventions, and that there is a clear map that shows the intended Mechanism of action. We are aware that Autism research currently tends to attract people from more privileged backgrounds. Reasonable adjustments are needed to ensure research conducted is accessible for Autistic individuals who wish to be involved in health intervention evaluation and development [78]. Interventions should be robustly evaluated in a way designed to assess how the intervention works, including showing if intended Mechanisms were triggered, and if any unintended consequences occurred, as well as measuring Outcomes related to health inequality for Autistic patients and knowledge and confidence for clinicians.

Supporting information

Acknowledgments

We gratefully acknowledge support in developing our search strategy from Stephen Storey, a specialist librarian at Swansea University.