How paediatricians investigate early developmental impairment in the UK: a qualitative descriptive study

Early developmental impairment (EDI) is a group of conditions manifesting when a child’s developmental abilities lie two or more standard deviations below the population mean in at least two developmental domains [1, 2]. EDI affects 2–3% of young children [1, 3]. The Office of National Statistics mid-year population statistics for 2019 estimate there are 3,857,263 children under 4 years of age in the UK; [4] therefore EDI will affect around 77,000–116,000 children and between 20,000–25,000 will present each year. EDI can be divided into two broad groups: isolated EDI (EDI-) is where there are no other clinical features in the phenotype other than the developmental issues, whilst EDI + is where there are additional features present, such as macrocephaly, microcephaly, dysmorphia, organomegaly, abnormalities of movement or muscle tone, eye abnormalities, seizures, consanguinity, or a relevant family history [5]. Paediatricians are more likely to investigate EDI + than EDI- [6]. There is variation in paediatricians’ views on the usefulness of specific investigations and whether or not children are investigated appropriately [6]. A number of guidelines exist for investigating EDI, but they are rarely evidence based, are often based on consensus, and the recommended investigations vary significantly [1, 3, 7‐21]. In 2017, we reviewed the frequency of diagnostic investigations for EDI in our centre and recommended rationalisation of investigations, with the consideration for cost-savings one potential benefit of this approach [5, 22]. These recommendations were controversial, with other groups arguing in favour of additional testing [23].

Whilst there is large variation in what investigations paediatricians request in children with EDI, there is no published evidence on how paediatricians choose aetiological investigations for EDI. Understanding why paediatricians choose the tests they do would identify internal and external driving factors, and could aid professional groups in reaching consensus about how to investigate EDI, particularly in the age in which greater genetic testing is being introduced. For example, it is tempting to assume that genetic testing through exome or genome testing will become the only first line test in EDI, but similar hopes about microarray have not come to fruition, [5] and understanding motivations for testing are important if investigations are to be rationalised in the future.

This study took place in the UK and the context of the health service is important. The UK has a nationalised health service, in which clinicians have freedom to choose investigations and treatments according to clinical need, other than for highly expensive and specialised conditions, without having to consider costs to the hospital or patients / family or involve insurance companies. On occasions, there may be pressures from hospital managers or the government to reduce costs and improve efficiency, but the degree to which clinicians feel this pressure will vary, and the ultimate decision on what a patient needs remains with the clinician. For children with EDI, referrals may be made to several specialists: general paediatricians, community paediatricians, paediatricians with expertise in neurodisability, or paediatric neurologists. Although a proportion of paediatric neurologists only see children with EDI via referral from other paediatricians following initial assessment and investigation, other neurologists take referrals direct from primary care. The preliminary training for all UK paediatricians is identical, with at least 5 years of general paediatric training, followed by additional training in sub-specialisms or general paediatrics to reach consultant level. Therefore, the initial training and experience of investigating EDI is the same across sub-specialisms, and there is likely to be significant concordance in views.

This qualitative descriptive study aimed to describe the thought processes and factors that affect UK paediatrician’s decisions on investigation choice in children with EDI, including the cost of testing.

Fourteen health care professionals were interviewed (Table 1). Two interviews were terminated early, one because of a meeting and another because of technical difficulties with the video link. Both were completed within 7 days. The duration of interviews ranged from 48 to 97 min (median 76 min).

Two major themes emerged during analysis (Fig. 1):

“The value of an aetiological diagnosis’”

“Managing risk and probability when investigating EDI”, from which the following subthemes were found:

The value of an aetiological diagnosis—illustrative quotations are shown in Table 2.

Consultants recognised the value to parents of an aetiological diagnosis for EDI, including: ‘understanding why’; the difficulties coping without an explanation; to allow for easier access to support in educational environments where the diagnosis of EDI was not accepted; access to support groups for specific conditions; prognostication; and recurrence risk. One participant highlighted a number of disadvantages to an aetiological diagnosis, particularly for ethnic minorities, including the inability to “take back” a genetic diagnosis, the negative effect on health and travel insurance, and how accurate prognostication could not be guaranteed because of phenotypical variation of genetic conditions.

Managing risk and probability when investigating EDI—illustrative quotations are shown in Table 3.

This theme, exploring the factors influencing choice of aetiological investigations, contained a spectrum of attitudes, at the end of which were two distinct approaches: ‘circumspection’ and ‘accepting appropriate risk’. Most participants were “in both camps”, and had changed the degree to which they ascribed to each attitude as they gained clinical experience, which formed our third subtheme, “Transition between styles of practice”.

Clinical phenotyping based on history and examination was highly valued. Participants preferred to adopt a stepwise approach to investigation, performing the tests most likely to reveal an aetiology first, possibly alongside a “general health screen”, then moving to second or third line testing where initial tests were non-diagnostic. There were two exceptions: children with an autistic spectrum disorder, where venesection would be distressing or impossible, and where a child required an MRI under general anaesthetic. The invasiveness of the test and the inconvenience and cost to the family of attending hospital for the investigations were considered important. Where initial investigations were non-diagnostic, clinical ‘watching and waiting’ helped to identify new ‘clues’ as the child grew older and enable access to new investigations when they became available. Parents influenced the choice of investigations if there was anxiety about specific conditions or investigations that affected their mental health.

False positive results or insufficient samples leading to repeated tests were an inconvenience that created additional, unnecessary work, and increased the child and families’ inconvenience. A low yield for investigations was considered “over-investigation”. Previously missed cases, either by the participant or someone that trained them, led to a focus on that particular investigation in the future.

The speciality of the consultant played a role: community and neurodisability paediatricians thought neurologists ordered more tests, which paediatric neurologists accepted. Neurologists reported a strong focus on the diagnosis of rare conditions and were more likely to request metabolic investigations, but noted they saw more complex children and felt there was an expectation from the families and referrers to perform some additional investigations to justify the referral. Phenotyping remained important to neurologists.

There was a complex relationship with the cost of investigations. It was suggested phenotyping led to rationalisation of investigations, and was justified as ‘cost efficient”. However, only one participant reported pressure from managers to reduce expenditure, participants did not know the cost of aetiological investigations, and all reported they would order a highly expensive test if they had a strong suspicion of the condition. The degree a condition was treatable was important, but was balanced against the incidence of the condition, rather than cost. For example, most participants would not order serum Cavorite if the incidence of the condition was extremely low, even if the cost was small. Neurologists were more likely to request Cavorite, but two changed their view when the cost was over £1000 a test, and suggested it should be on newborn screening instead. They also noted the health economic savings of improving a small number of children’s outcome may be greater than the cost of testing a large number of children. Although all participants agreed cost was important, they frequently deferred decisions on a population level to national governmental bodies. Participants agreed in principle that it would be good to redirect money saved on unnecessary investigations to fund improved access to therapeutic interventions they needed in their unit (Table 4), but noted this was unrealistic within current health service structures.

Paediatricians recognise the value of an aetiological diagnosis in EDI for families, and cited many of the same reasons that parents do: validation, information, procuring services or therapies in the education setting, early intervention, support, need to know, and prenatal diagnosis [28]. Many published guidelines recommend clinical phenotyping to choose which investigations are first and second line [1, 3, 5, 7‐21]. Previous qualitative data collected from 107 UK clinicians and published in 2017, showed 61% had departmental guidelines for the investigation of EDI and 62% of responders thought children with EDI + were investigated the right amount, compared to 38% children with EDI- [6]. Amongst our participants, 79% had departmental guidelines, and 50% thought guidelines investigated children the right amount, suggesting our data is valid and reflects the range of views on EDI.

We have previously suggested that reducing the burden of unnecessary tests could produce cost savings to the health service, and found the yield of metabolic investigations was low, particularly in EDI- [22]. This agrees with other groups’ work [1, 5, 14, 29‐31]. However, other publications extol the importance of diagnosing metabolic disorders, [20, 21, 23, 32‐34] particularly where they are treatable. Such divergence in views was also seen amongst our participants. Community / neurodisability and general paediatricians found the yield of metabolic investigations was low and recommendations for large numbers of routine metabolic tests were described as “bananas”! These interviewees reported problems with routine screening of metabolic conditions, including the large volumes of blood needed and the inconvenience to children and families of repeat testing when samples were insufficient, falsely positive, or of uncertain significance. Over time, many of our consultants dropped routine metabolic investigations because the yield was so low. Cost efficiency was highlighted in our previous work as a potential advantage of reducing investigations requested in EDI, [5, 22] and interviewees themselves cited this as one reason for reducing the number of metabolic tests they ordered. However, it was hard to establish cost as a behavioural modifier from our data, despite what was said: participants did not know the cost of investigations, rarely faced cost pressures from managers as a negative reinforcement, and did not experience a positive reinforcement of having saved money redirected to interventional services. There was also a suspicion money would redirected to other acute hospital services instead, perpetuating what one participant called the “Cinderella status” of community paediatrics. What is more likely is that consultants change their behaviour because of either the positive punishment factor of the additional work created by spurious results, or negative reinforcement from never finding a positive result.

In comparison, neurologists and more inexperienced paediatricians were more likely to investigate rare treatable metabolic conditions because they were treatable, but acknowledged many of these would be better investigated on newborn screening to allow for early diagnosis and treatment before developmental impairment occurred. Despite this divergence in views, all participants retained a belief that phenotyping was essential to aid with the choice of appropriate investigations. The differences between interviewees holding these two views related only to the degree of risk for missing a treatable condition considered acceptable. With phenotyping seen as important, guidelines were noted to be unable to cover every clinical situation, but were useful in making trainees and inexperienced consultants feel safe, particularly around rare conditions.

The criticism we faced when recommending reducing amounts of investigations in EDI guidelines [23] was valid in many ways, including the observation that not every child in our cohort received the same investigations. In reality, we think the actual difference in opinions is not so large: our data suggested little evidence for a large battery of tests in EDI-, which was where rationalisation of investigations was recommended, but we encouraged further testing in EDI + based on phenotype [5]. Given only 20% of our cohort had EDI-, this was the minority of children. Therefore, any disagreement primarily relates to how children are stratified based on clinical assessment and the acceptable level of risk for missing diagnoses. There is no good evidence to help with such stratification, and our participants all wanted a high-quality, evidence-based guideline according to clinical phenotype. Participant 13 suggested a large cohort study in which children with EDI were rigorously phenotyped and offered the same investigations, allowing evidence to be collected on which investigations are useful in different phenotypes. Such studies are both difficult to design and expensive, and we have been unsuccessful on a number of occasions to obtain funding for similar research projects.

In the near future, the availability of exome studies may also be a driver to change practice. A recent meta-analysis showed 31% of children with EDI- and 53% with EDI + had genetic abnormalities on exome studies, [35] so suggestions that exome studies become the only first line investigation may arise, with additional investigations reserved to determine the biochemical significance of variants of uncertain significance or for acquired, non-genetic causes. However, a number of observations suggest this may not be how paediatricians will practice in reality. Firstly, microarray was supposed to be introduced in our unit as the only first line investigation, with subsequent tests performed in children with negative results. Instead, microarray became yet another investigation in a long list of investigations. This is the recommendation in published guidelines, [3, 19‐21] suggesting our local experience is not unique. Secondly, our data suggests interviewees may perform additional tests, either because they are performing venesection to obtain DNA and wish to reduce trauma to the child and inconvenience to the family, or because they wish to arrange a general health screen for other modifiable health factors. Therefore, we should not assume that, even if this is how geneticists think exome studies should be used, paediatricians will behave in this way.

The strength of our data includes collecting the views of all types of consultants who see children with EDI, the open nature of our questions, and discussing the cost of investigating rare diseases by using a fictional condition, which removed preconceived ideas about specific diseases. The use of qualitative interviews allowed much richer exploration of this issue than we had managed previously via qualitative questionnaires [6]. There are also limitations to our data. As with any qualitative data, the findings may not be generalisable, but we purposefully chose clinicians from different specialities and geographical regions to obtain a full range of views. Our data is also in concordance with our previous qualitative data from a national survey [6]. We reached data saturation so our sample size was appropriate. The COVID-19 pandemic meant interviews had to be performed virtually midway through recruitment, and we do not know the impact of this on our data. The interviewer was junior to the participants interviewed, and this disparity may have played a role, alongside any previous working relationship he had with interviewees. He may also have preconceived opinions prior to the study, but we mitigated these factors in study design and training. Finally, the attitudes of clinicians in a nationalised health care system are likely to be different to other health care systems, so this data may not be applicable outside the UK.

In conclusion, paediatricians acknowledge the importance for families of finding an aetiology for EDI, and this is not restricted to treatable conditions: the “need to know why” is recognised as important. When choosing investigations, paediatricians fall into two broad groups: the first is where a large battery of tests is applied to all children investigated with no consideration of clinical phenotyping or likely differential diagnosis. Standardisation of practice and guidelines are highly valued. These paediatricians tend to be the least experienced and eschew risk taking. The second are those whose choice of investigations are based on the clinical phenotype and an acceptable level of risk is acknowledged. These paediatricians tend to be more senior and have transitioned from a low risk, standardised approach towards accepting some risk as they have gained clinical experience. The cost of tests played only a small part in the decision-making process, mainly because any savings in expenditure were unlikely to be reinvested in other services, such as therapy or psychology, so there was little motivation to change behaviour. Decisions on cost tended to be deferred to national or governmental bodies.

What this study adds