Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Identifying Autism in the Female Population
Terry (Traolach) Brugha University of Leicester
Leicestershire Partnership NHS Trust (Assisted by a cast of thousands...)
Are we missing autism even more so in women than in men?
• What questions should we be asking women and men and what observations could we make to help answer this question?
• For example, might part of the answer be to ask people more probing questions about their possible repetitive behaviours and sensory difficulties or differences?
• What does our knowledge of the epidemiology of autism in adulthood and from growing referrals of women to an NHS clinic teach us already?
False positives and false negatives
• We can get this wrong both by under counting (women) and by over counting (men)
• Are we coy about asking gender specific questions because somehow that might be criticised as biasing a measure either way?
• And do we fail to do so in a balance or unbiased way because we only ask male gender specific and less so female gender specific questions?
• Could we be coy about asking questions about issues that are of greater interest to women?
• (Somehow this seems implausible given that most of our widely used measures were developed by women).
We don’t ask the right questions?
• Autism assessments rest largely on informant information on early development and on direct observation / a glimpse (i.e. ADOS) – Some ADOS items on understanding of
relationships, personal responsibilities, on descriptions of emotions are loaded against males?
• In adulthood we have probably failed to exploit the person’s ability to self report – The quest for new, better questions...
If so why?
• False positive diagnosis of autism in men that includes some men who are poorly socialised?
• Men are less concerned about how they present, appear, and are possibly less verbally competent?
• Are women more biologically protected?
• Greater concealment because a diagnosis could compromise freedom to choose to be a mother?
• It is possible (we also don’t know) that women on the spectrum are more likely to be in a sexual relationship; this could provide greater opportunities to learn to be socially responsive and responsible and seem ‘normal’?
Repetitive behaviours
• Could it be that there is a greater range of routine, repetitive behaviours, which women engage in, which go unnoticed, appear normal?
• Domestic roles, employment roles, outside interests (caring for animals, for example)
• Could it be that we don’t ask about interests that are commoner in women – Is it more ok to ask about male gender specific interests
than those that might be commoner in women?
– Food issues, dress, make up, celebrity adulation ...?
Special interests
• Draft one:
– Do you have special interests for example involving, history, technology?
• Revised draft:
– Do you have special interests for example involving, history, technology, food, animals, fashion, body adornment (jewellery, make up, tattoos, symbols), people who are famous (celebrities)?
Sensory differences
• Do you react differently to most people to things you experience – the feeling of things on your skin; smells; things you see (patterns)?
• Are you particularly sensitive to noise? Are you different to most people in the way you respond to it being too hot or too cold for you? Are you particularly sensitive to pain?
• Do you have any noticeable likes/dislikes? The feel of things – clothes, material, surfaces, what food feels like in your mouth? How do you find getting your hair cut (by someone different) or being hugged?
• Things you see – colours, reflections, appearances, movement/busyness – are you bothered by neon lights?
• What about smells – do you have a good sense of smell? Do any smells alarm you?
• How sensitive is your hearing? Are there noises that you can’t stand?
• How sensitive to pain are you? Are you more aware (or less aware) than most people of the temperature – when the weather is hot or cold? – Acknowledgement Tom Berney and SCAN 3.0
The hard evidence on gender?
• The epidemiology of autism in adults
• How are surveys carried out?
• What did we find?
• Let’s focus on gender
Background
• All previous epidemiological studies of autism (ASD) have been child based
• Most adult autism research samples have been clinic based and therefore are potentially not representative
• Recognition of autism in adulthood is largely confined to intellectual disability services
• No previous adult epidemiological methods developed or tested
Alerting Characteristics
Council Report CR136, Royal College of Psychiatrists, April 2006.
1) Difficulties with social relationships – social isolation
• few/no sustained relationships - those that there are, are likely to be either distant or intense.
• a persistent aloofness or an awkward interaction with peers (which sometimes may be unduly compliant or passive).
• unusually egocentric with little concern for others or awareness of their viewpoint and little empathy or sensitivity
• a lack of awareness of social rules: prone to social blunders
2) Problems in communication • an odd voice, monotonous and perhaps at an unusual volume
• talking ‘at’ (rather than ‘to’) you with little awareness of your response
• language superficially good but too formal/stilted/pedantic and with difficulty in catching any meaning other than the literal.
• impassive appearance with few gestures and a rather odd, poorly co-ordinated gaze that may either avoid looking at you or else look through you (misinterpreted as furtive or aggressive respectively) – i.e. limited nonverbal communicative behaviour.
• an awkward or odd posture and body language
3) Absorbing and narrow
interests
• obsessively pursued interests
• unusually circumscribed interests that contribute little to a wider life – e.g collecting facts and objects that have little practical/social value
• a set approach to everyday life that may include unusual routines or rituals; change is often upsetting
4) A disorder that, although its presentation
may change and moderate with age, has a childhood onset and is life-long
• Therefore if possible the assessment should include a developmental history
• Because the sufferer may not be able to recognise and describe their own problems and because key features emerge in the first 3 to 4 years of life it is advisable (if possible) to obtain collateral information on early development and current behaviour
How common was and is it (1)?
• In children the median rate for 16 surveys published in the period 1966–1991 was 4.4/10,000, whereas that for the 16 surveys published in the period 1992–2001 was 12.7/10,000.
– Fombonne E. Epidemiology of autistic disorder and other pervasive developmental disorders. J Clin Psychiatry 2005;66 Suppl 10:3-8.
– Newschaffer CJ, Croen LA, Daniels J et al. The Epidemiology of Autism Spectrum Disorders. Annu Rev Public Health 2007 January 2;28:235-8.
• Three large scale child epidemiological studies in GB suggest rates in childhood of 10 per 1,000 within the past decade
– Baird G, et al. Prevalence of disorders of the autism spectrum in a population cohort of children in South Thames: the Special Needs and Autism Project (SNAP). The Lancet 2006 July 15;368(9531):210-5.
– Green H, McGinnity A, Meltzer H, Ford T, Goodman R. Mental Health of Children and Young People in Great Britain, 2004. Hampshire: Palgrave McMillan; 2005 Sep.
– Baron-Cohen S, et al. Prevalence of autism-spectrum conditions: UK school-based population study. Br J Psychiatry 2009 June;194(6):500-9.
• Studies in children show male to female ratios of at least 4:1
• About 1 to 2 in 4 ASD children also have learning difficulties (depending on the definition of LD)
• Recognition rates: typically 3-5% (NAO, 2009)
• The rate in adult secondary psychiatric care based on screening was about 30 / 1,000
• (Nylander and Gillberg)
• Rates of severe complex forms lower and unknown
• General population rates in adulthood were unknown until now
And how would you study it?
• No adult epidemiological studies to go by
• Adults with the condition unlikely to know they have it and often are poorly aware of what makes them different
• Epidemiological methods to date dependent on collecting information on childhood development from parents / similar observers such as teachers
• Clinician diagnoses are highly unreliable...
Our study aims were to:
• Conduct an efficient two-phase survey of ASD
– as well as psychosis and personality disorder within a third Adult Psychiatric Morbidity Survey (APMS 2007)
• Maximise the exclusion of low scoring cases
– for accurate prevalence estimation
• Maximise the overall number of cases detected
– for further sub-group analysis, for example to study associations with ASD in an unbiased sample
In
Adult Household 1993
Institutional Residents 1994
Adult Household 2007
1993 2000 2007
Adult Household 2000
Homeless Persons 1994
Prisoners 1997
Adult Household Follow-up
2002
Children and Adolescent Follow-up
2002
Children and Adolescents 1998
Children and Adolescents 2004
Carers 2001
Looked After Children 2001
The British Psychiatric Morbidity Survey Programme
Household population of
England aged 16+
Random sample who consent to and complete an initial interview including AQ-20
7461
Non responders
Household population of
England aged 16+
Random sample who consent to and complete
an initial lay interview including
AQ-20 7461
Selection for second stage based on initial responses on
AQ, PSQ, PDQ N=849
Non responders
Household population of
England aged 16+
Random sample who consent to and complete an initial lay interview including AQ-
20 7461
Selection for second stage based on initial responses on
AQ, PSQ, PDQ 849
Agree to and complete second stage
clinical interviews with SCAN SCID-II ADOS
618
Uncooperative or untraceable
Non responders
Household population of
England aged 16+
Random sample who consent to and complete an initial lay interview including AQ-
20 7461
Selection for second stage based on initial responses on
AQ, PSQ, PDQ 849
Agree to and complete second stage
clinical interviews with SCAN SCID-II ADOS
618
Follow-up with DISCO developmental assessments
via informants N=60
Uncooperative or untraceable
Non responders
Phase one ASD screening tool
• Review of existing measures
• 50 item Autism-Spectrum Quotient (Simon Baron-Cohen, Sally Wheelwright, et al)
• Traits
• Validated only in specific populations (i.e. Clinics)
• Self-report (essential in adult surveys)
• AQ-20 subset of Autism-Spectrum Quotient?
• Threshold for best sensitivity/specificity?
Methods
• Range of sampling probabilities for phase two
• Prevalence by strata
• Selection probability increased with AQ-20 score
• Low threshold for eligibility to phase two
Phase two selection probability by phase one AQ score
0
0.2
0.4
0.6
0.8
1
0 2 4 6 8 10 12 14 16 18 20
AQ score at phase one
Probability of selection to phase two
Phase two diagnostic assessment in a field work context
• Requirement for a direct ‘face to face’ assessment of autism in field work interviews:
–Autism Diagnostic Observation Schedule (Lord et al) Module 4 (ADOS-4)
• Adapting the Autism diagnostic observation schedule (ADOS -4) for the ‘normal’ adult world – training a team to assess > 600 adults
Calibration of ADOS-4
• In phase three challenge of finding parents or carers to complete the informant based Diagnostic Interview for Social and Communication Disorders (DISCO) and ADI-R to assess childhood and current development
1. ADI-R and DISCO used to calibrate primary diagnostic assessment (ADOS-4)
2. Clinical case vignette rating by experienced professional diagnosticians (Psychol Med, 2011)
• ADOS-4 10+ threshold confirmed for definite ASD
Phase two assessments
• 618 ADOS assessments conducted at phase two
• 19 cases of ASD were diagnosed
• 5 of these had an AQ-20 score of less than 10
• Weighting corrects for sampling and non-response
• ~72 estimated cases overall (would have been found if we
could have conducted ADOS diagnostic assessments on all
phase one respondents – i.e. totally impractical)
Performance of AQ-20 screen and ADOS assessment
Best combination of sensitivity and specificity (%)
• 10+ AQ-20 cut off (to predict ADOS 10+)
• Sensitivity = 73.7 %
• Specificity = 62.0 %
• Sum = 135.7 %
• McNamee argues for minimum sum of the percentages of 160+ %
Findings: suitable for a two-phase design
• Further methodological development work has just now been funded in order to test a better phase one screening questionnaire.
• However, in lieu of a better lay instrument...
• A two-phase survey design with
- a low AQ score threshold for eligibility to phase two
- multiple sampling fractions
- inclusion of everyone with a high score
was appropriate.
Findings
• Distribution of traits and of the disorder in the adult population
• Prevalence according to a range of thresholds
• Associations with ASD in the community
Distribution of phase AQ-20 scores in ~ 7,400 adults
Distribution of ADOS Scores. Diagnostic threshold 10 or more
ADOS cut-off Unweighted
Number
Base = 2828
Weighted
Number
Base = 7333
Weighted estimate
(95% CI)
7+ 32 108 14.7 (7.0, 22.5)
8+ 26 88 12.0 (4.9, 19.1)
9+ 20 75 10.2 (3.4, 17.0)
*10+ 19 72 9.8 (3.0, 16.5)
11+ 16 65 8.9 (2.2, 15.5)
12+ 12 47 6.4 (0.6, 12.3)
13+ 10 44 6.0 (0.2, 11.8)
Unweighted and weighted estimates numbers of respondents per 1000 population for ADOS cut-offs from 7+ to 12+ including estimated prevalence of ASD at the recommended threshold of ADOS 10+
P-value for age as a continuous predictor of ASD (p=0.55) using the recommend threshold of ADOS 10+.
Prevalence by Gender
Prevalence by Educational Achievement
Key associations found
• Each analysis using the 10+ ADOS threshold was repeated with the 7+ cut-off: the statistically significant associations for gender and tenancy for ADOS 10+ were also found for the 7+ cut-off; those for other associations were also in the same direction but no longer statistically significant.
• As most (15/19) of the discovered phase two ADOS 10+ cases were males, the weighted univariable logistic model (males and females combined) was repeated in males only. The findings were unaltered.
Service contact and use
• Although there were no clearly significant findings ASD cases appeared to be less likely to be using health services for mental health reasons compared with other adults with mental and behavioural disorders
• (note that cases were less likely to know how to answer questions about receipt of welfare benefits)
Combined rates with the intellectually disabled
Repeat as far as possible the APMS 2007 in adults with intellectual disabilities: Living in communal care establishments Living in private households
Combine the results with data from the APMS 2007 to derive an overall prevalence for autistic spectrum conditions in adults in England.
Use same instruments used in the APMS 2007 where possible
Combined prevalence of autism
Prevalence of autism among adults living in private households
7274 adults aged 18 years or over who
took part in the APMS 2007
Prevalence of autism among adults living in communal care
establishments
Adults with intellectual
disability living in private
households who could not have taken part in APMS 2007
Adults living in communal care establishments sampled from case registers
Sensitivity analysis and other
adjustments to take account of adults
living in communal care establishments
who do not have intellectual disability
Calculation of autism prevalence
Sensitivity analysis
Effect on combined prevalence of autism
Prevalence of intellectual disability in communal care establishments: 100%-110%
Prevalence of severe intellectual disability in private households:
70%-150%
Variation in intellectual disability prevalence
Defence establishments: 70%-110% of APMS 2007 estimate
Variation in autism prevalence
Educational establishments: 70%-110% of APMS 2007 estimate
Prison establishments: 500%-4000% of APMS 2007 estimate
Study sample for study to extend the APMS 2007
Adult intellectual disability case registers in England
• Leicestershire • Sheffield • Lambeth
Methods: Screening/Diagnostics
APMS 2007
Extension
ADOS Module 41
(with participants) ADOS Module 11
(with participants & carer)
DISCO2 (with informant)
ADI-R3 (with informant)
ADI-R3 (with informant)
DISCO2 (with informant)
1. Lord et al. (2002). The autism diagnostic observations schedule – generic: a standard measure of social and communication deficits associated with the spectrum of autism. JADD 30:205-23
2. Lord et al. (1994). Autism Diagnostic Interview – Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. JADD 24:659-85
3. Wing et al. (2002). The diagnostic interview for social and communication disorders: Background, inter-rater reliability and clinical use. Journal of Child Psychology and Psychiatry 43:307-25
Results 1: overall prevalence of autism
7274 adults without intellectual disability were included in the APMS 2007.
276 adults with intellectual disability were included in the extension study, comprising: • 79 adults from private households; • 197 adults from communal care establishments.
The combined prevalence of autism among all adults was 1.1% (0.3% to 1.9%).
Results 2: autism by sex, age and ethnic group
Combined prevalence of autism among men was 2.0% and 0.3% among women.
Combined prevalence of autism was similar by age group (18-44 years: 1.3%; 45-74 years: 1.1%; 75+ years: 0.6%)
Combined prevalence of autism was similar in white and south Asian population (1.2% and 0.8% respectively) [numbers were too small to investigate other ethnic groups]
Results 3: autism by verbal IQ / severity of intellectual disability
So why is autism commoner in women – evidence from epidemiology?
• Not explained by age (such as women surviving longer)
• Not explained by differences in survey cooperation (young men cooperate less)
• Not explained by differential service contact activity (the gender difference is also found in the undiagnosed)
• Could possibly be explained by ability level – suggestion that autism is decreased in higher functioning / ability women and not in moderate to profoundly intellectually disabled women
The ‘so what’ question
• If it were the case that higher functioning women with autism are missed more often, partly because they are managing better, would it matter that they are missed more – after all most people in adulthood with autism are missed anyway?
• Where do we draw the line – some say having autism is not what matters and they would not want to be different – so what does matter?
• Exclusion and prejudice, restricted functioning, lower income, greater stress, poorer life quality?
My own conclusions
• I am increasingly concerned that we are less good at identifying autism in women than in men, which puts women at a disadvantage.
• I want to see if we can measure autism better.
• And then I want to try that out in future population research.
• Solving that might even lead to the world being a better place for all of us to live in.
Research Team
University of Leicester Traolach (Terry) Brugha Professor of Psychiatry / Principal Investigator Howard Meltzer Professor of Mental Health & Disability Jane Smith Fieldwork Manager Nicky Spiers Medical Statistician Freya Tyrer Research Manager
Interviewers Andrew Leaver, Ann Loughnane, Caroline Lovett, Emma Peters, Karen Ricci, Darren Sharpe
Leicestershire Partnership NHS Trust Sabyasachi Bhaumik Medical Director Reza Kiani Consultant Psychiatrist in Intellectual Disability
University of Glasgow Sally-Ann Cooper Professor of Intellectual Disabilities
Autism Research Centre, University of Cambridge Fiona Scott Consultant Chartered Psychologist
National Centre for Social Research, London Sally McManus Research Director
Susan Purdon Independent Survey Specialist