Clinical practice guidelines for diagnosis of autism spectrum disorder in adults and children in the UK: a narrative review

One guideline suggested that clinical practice varies greatly [43] and we found this to be mirrored in CPGs with a wide range of potential assessment processes included. DSM-5 recommended that a diagnostic assessment should include gathering multiple sources of information from clinician’s observations, caregiver history and self-report (where possible). National guidelines, although providing far greater detail, tended to include these areas and additionally suggested various other detailed assessments such as gathering wider functional/assessment information [10]; using documentary evidence, assessing risks, and assessment of challenging behaviour [9]; assessing for co-conditions [9, 39]; physical examination [39]; comprehensive educational assessment [44]; assessment of communication, neuropsychological functioning, motor and sensory skills, and adaptive functioning [10]. Professional guidelines added other factors such as comprehensive cognitive assessment [40] and impact of individual’s mental health [41], accounts of relationships in different settings [11] and observation in school or another setting [43]. Journal articles tended to reflect national guidelines and varied in the level of detail outlined for assessment factors. Two articles gave little detail of assessment processes but one referred readers directly to NICE guidelines for further detail [45] and the other was aimed at community practitioners who would be more likely to be involved in referral than diagnosis [46]. Articles also included assessment of co-occuring conditions (e.g. [47‐52]) and a physical or medical examination (e.g. [47, 50]). Additional assessment areas included assessment of specific domains such as family stressors and coping abilities [47]. In one guideline [48] it was suggested that some clinicians bypass ICD/DSM criteria and instead undertake:

‘…testing for specific underlying difficulties such as lack of theory of mind or lack of central coherence and then using these to decide the presence of the behavioural criteria’ [48].

The RCSLT guideline [41] differed from most by suggesting consideration of theories relating to the triad of social impairments, such as executive functioning deficits, motivation, memory and central coherence, as well as social interaction and communication. However, some (e.g. [40]) suggested cognitive or neuropsychological testing whilst SIGN guidelines stated that such assessments are ‘useful for individual profiling but are not diagnostic instruments’ [10]. This anomaly may reflect the specialist role of SLTs in the diagnostic process.

Overall, we would concur with a reflection in one guideline, which noted how the HCP may be faced with ‘possible uncertainty as to where to go next in their investigation framework as this could be potentially enormous’ [53].

Recommendations about the use of diagnostic tools were mixed. One third of the guidelines (n = 7) did not specify any particular tool for diagnostic assessment. Other CPGs tended to suggest the consideration of a range of tools without specifically recommending any particular instrument(s), although regular references were made to ADOS (n = 13), ADI-R (n = 11), DISCO (n = 9) and 3di (n = 6). The NICE guideline for children and young people emphasised use of DSM/ICD criteria rather than tools; the NICE guideline for adults did the opposite [9, 39]. Overall, findings concurred with Penner et al. in that guidelines varied substantially in their recommendations for use of diagnostic tools [29].

There were differences in the way guidelines described the relationship between, or referred to, diagnosis, assessment, profiling, needs assessment and wider formulation. All guidelines encompassed the concept of a wider (needs related) assessment but few explicitly separated out these processes or discussed how this related to a diagnostic assessment. One exception to this was the Regional Autistic Spectrum Disorder Network (RASDN) children’s guideline, which separated the diagnostic from the formulation process, describing the latter as including examination of the person’s wider environment:

‘The outcome of the formulation should be to understand an individual in a more global holistic way rather than merely in terms of signs and symptoms, as in the case of diagnosis’ [44].

The RCPsych guideline suggested that diagnosis is only one component of the wider multidisciplinary exercise [11]. Some guidelines did not mention formulation but suggested a profile of strengths, abilities and weaknesses should be carried out alongside a diagnostic assessment (e.g. [10, 39]). Adult guidelines from RASDN separated out a diagnostic assessment from a full needs assessment [54]; NICE guidelines for adults considered comprehensive assessment to include diagnostic, needs and risk assessment [9]; whilst the full children’s guidelines similarly brought together the diagnostic and needs elements under ‘autism diagnostic assessment’, explaining that:

‘..the label of autism does not constitute a complete diagnostic assessment and a profile of the child or young person’s strengths and weaknesses is also essential. This requires a multidisciplinary team which has the skills to undertake the assessments necessary for profiling’ [55].

Operationally, therefore, there were contradictions between guidelines about what constitutes the diagnostic process, how it should be structured and which diagnostic tools should be used.

Where specified, all guidelines advocated for diagnosis to take place within a multidisciplinary setting with various guidelines suggesting this was ‘necessary’ [44], the ‘optimum approach’ [10] or ‘ideal’ [43] (See Table 4). Some suggested (n = 4) that an appropriately trained and experienced single professional is sufficient to diagnose in particular cases, but to be alert for indications for a more specialist assessment [11] and with access to multidisciplinary support if required [48].

Despite this almost universal recommendation, the extended version of NICE children’s guidelines (and cited by SIGN [10] and Carpenter [48]) questioned the evidence base for multidisciplinary assessment reporting a study [56] that showed moderate agreement between an individual HCP and an MDT in making a diagnosis, but stating that it was a low quality study [55]. These guidelines also suggested in practice that a diagnosis can be made by a single experienced HCP but that a comprehensive profile of the patient requires a multidisciplinary approach [55]. SIGN guidelines also cited research [57] which demonstrates that parents value a multidisciplinary assessment [10].

None of the guidelines in this review dealt with how HCPs come to a consensus within a multi-disciplinary context, although Northern Ireland guidelines recommended that training should include the promotion of collaborative and innovative working [44] and that clinicians must understand the profession specific roles and responsibilities of the overall team [44, 54].

Therefore, most guidelines referred to MDTs as best practice, but lacked recommendations about how roles within MDTs are negotiated, how disagreement is resolved (other than second opinion outside the team); or how teams should work together, a factor that is acknowledged by NICE adults guideline [9].

Many guidelines (n = 9) outlined the importance of keeping the person/family informed and involved throughout the process or recommended a person-centred approach. Some described the relationship with the person coming for diagnosis and their family as a partnership (e.g. NICE adult guideline [9]) or as person-centred (e.g. RASDN adult guideline [54]). Some guidelines (n = 6) acknowledged that the person or family may disagree with or be reluctant to accept a diagnosis or, alternatively desire one [46] and be determined on a particular outcome, which can lead to misleading results [11]. Carpenter asserted that some people may begin to see diagnosis as a desirable outcome and pre-prepare answers based on structured interviews published on the internet [48]. The potential for disagreement or desire for diagnosis, therefore, may impact on the interaction with the person or their family. So, although the relationship with the patient/family is considered within CPGs, there is little guidance as to how HCPs might deal with patient/family desire or disagreement.

All national guidelines (n = 5) outlined the requirement to consider the needs, preferences and values of the individual and their family and/or support them to communicate their needs and concerns. Most guidelines (n = 17) described elements of diagnosis that relate to either family environment, family needs and concerns, circumstances, relationships, functioning, experiences in different settings, contextual information or level of support needs. Many guidelines reflected the need to consider assessment of support required. Enquiries should be made about how symptoms impact on function within the family, at home, school or work [9, 39, 47, 54, 58]. Overall, therefore, there was a focus not only on the assessment of symptoms, and the way in which these affect the daily life of the person and their family, but the wider environmental and social context of the person and the way in which they are supported, or not, by that context.

Some guidelines (n = 6) reported the difficulties of diagnosing autism when compensation strategies may ‘mask’ difficulties in some contexts, particularly as an adult [33], and in girls [50] where autism may go unrecognised. Some suggested that individuals may come forward for diagnosis when their circumstances change and/or stressors increase (e.g. [10, 45, 54]). Some guidelines (n = 5) noted that cultural differences will exist in norms for social interaction or that cultural variations can deliver misleading signs and symptoms. DSM-5 suggested that the boundaries between normality and pathology differ between cultures and the level at which experience may become problematic may differ [33].

SIGN suggested that those with autism may not have met ‘normal’ adult milestones in work, relationships or independence and contained extensive information on how females can present with a different symptom profile [10]. Others warned that behaviours might be the result of disruptive home experiences, carer illness [39] or complex psychosocial or child protection backgrounds [53].

Despite research showing links between diagnostic rates and SES, there was very little mention of the impact of SES in CPGs. DSM-5 stated that cultural and socioeconomic factors may affect age at recognition or diagnosis [33] but generally guidelines failed to consider how this might be considered in practice, other than to be aware that ‘cultural variations can deliver misleading signs and symptoms’ [45] or that autism is ‘not restricted to particular ethnic or economic backgrounds’ [40]. RCSLT guidelines considered assessment of bilingual individuals [41] and some suggested that ethnicity may delay engagement in the diagnostic process [11] or may increase difficulty in accessing services [54].

Overall, guidelines suggested it was the responsibility of the HCP to make a judgement about which behaviours appear to be ‘normal’ in complex social and family circumstances, as well as against norms for behaviour.

Overall the general focus of guidelines was to outline a framework to find the best way to decide whether autism is present or not around a threshold of symptom severity. However, many guidelines problematized this, for example, one guideline discussed how definitions of autism have changed with DMS-5 [11] and others suggested that social factors, such as an upbringing characterized by lack of boundaries [45] or symptoms amplified by distress [11] may cause diagnostic difficulties.

All national guidelines considered uncertainties around diagnosis, particularly with very young children or those with co-existing disorders [39]; when there may be disagreement within the diagnosing team or between the team and the patient or family, or when there is a lack of local expertise [9]. Many warned of diagnostic difficulties, or ‘obscuring’ [11, 53] that can take place if there is an intellectual disability or other complex coexisting condition and several considered the difficulties of overlapping diagnostic criteria [33, 50, 51, 53]. Further uncertainty was outlined when individuals may not reach the diagnostic threshold [39] or when children with autism score below the cut-off as determined by the diagnostic instrument [43].

Despite this uncertainty, CPGs generally proposed a systematic approach to diagnosis and, in some cases, asserted that progress has been achieved in establishing consensus around a behavioural definition and established systematic clinical assessments (e.g. [50]) even whilst recognizing that the ‘boundaries between disorders are more porous than originally perceived’ [33].

Eight guidelines stressed the key role of clinical judgement in the diagnostic process. DSM-5 outlined that the use of diagnostic criteria should be informed by clinical judgement [33] and ICD-10 suggested that guidelines should be used flexibly in clinical work [32]. The full version of NICE children’s guideline recommended: ‘Use information from all sources, together with clinical judgement, to diagnose autism based on ICD-10 or DSM-IV criteria’ [55]. One guideline suggested that clinicians may depend on the ‘feel’ of the interaction with the patient for diagnosis [48]. The RCPsych guideline stated that:’…much will depend on the extent of the clinician’s experience, their rigour in applying standard criteria and their ability to recognise alternative diagnoses’ [11]. Uncertainty, clinical judgement and clinician experience, therefore, were all identified as important factors in the diagnostic process.

‘Diagnosis and the assessment of needs …can open doors to support and services…all of these can improve the lives of the child or young person and their family’ [39].

However, NICE guidelines for adults acknowledged that adults who are diagnosed may receive no support due to lack of services [9] and Pilling et al. stated that whilst care for children and young people is generally well coordinated, this is not always the case for adult services [58].

Although some guidelines acknowledged that people may not want a diagnosis and the label it brings with it (e.g. [44, 54]) or that it can be stigmatising or damaging to career plans [11], generally guidelines described the benefits of a diagnosis primarily as relating to improved quality of life, creating an opportunity to have needs met, greater understanding and reassurance about one’s own situation and access to interventions and services. Some guidelines considered that diagnosis can provide relief, understanding or an opportunity to move on with increased support [11, 39, 45].

Many guidelines stressed the importance of early diagnosis as this enables early intervention which leads to improved health outcomes (e.g. [41, 44, 47, 50]). However, the BMJ guideline asserted that the, ‘…efficacy of early intervention varies from child to child’ [43], and that ‘consideration of the direct financial costs, indirect costs… and the impact on relationships within the family… must be balanced against likely and possible improvements in outcome for the person with ASD’ [43], bringing uncertainty into the benefits of diagnosis. Furthermore, O’Hare asserted that it is difficult to prove that earlier intervention is more effective [53].

Overall, guidelines reflected a concern about the potential impact or benefits on the child or adult receiving a diagnosis and considered positive factors such as access to support and intervention, increased understanding or relief; as well as potential negative impacts such as stigma. Carpenter, however, questioned the relationship between need and diagnosis, by asking whether diagnosis is influenced by what intervention the person needs or ‘…explicitly determined by the person’s need to have the label to access a service… rather than their fitting strict diagnostic criteria?’ [48].

To conclude, whilst CPGs appeared to frame a methodical and clinical diagnostic process, they also rehearsed a number of subjective dilemmas that HCPs have to negotiate along the way. Some CPGs themselves drew attention to social issues that muddle the process: the difficulties of establishing a clear threshold in a condition where symptoms are impacted by the stressors of environment [11]; the problem of relying on mechanistic assessments or algorithms [11, 48]; the crucial role of clinical judgement [54]; the possibility of diagnostic uncertainty through disagreement, lack of local expertise or when a complex coexisting condition is present [9]; the complexity caused by interaction with co-occurring conditions; masking of autism by comorbid conditions in secondary care [58]; the impact of good (or poor) social support and coping strategies on how symptoms present [33], to name a few.