Article Text
PDF
Abstract
Background Prenatal alcohol exposure (PAE) is a community health problem with up to 50% of pregnant women drinking alcohol. The relationship between low or sporadic binge PAE and adverse child outcomes is not clear. This study examines the association between PAE in the general antenatal population and child neurodevelopment at 2 years, accounting for relevant contributing factors.
Methods This prospective population-based cohort recruited 1570 pregnant women, providing sociodemographic, psychological and lifestyle information and alcohol use for five time periods. PAE categories were ‘low’, ‘moderate/high’, ‘binge’, in trimester 1 or throughout pregnancy. Measures of cognitive, language and motor development (Bayley Scales of Infant and Toddler Development) were available for 554 children, while measures of sensory processing (Infant/Toddler Sensory Profile) and social–emotional development (Brief Infant Toddler Social Emotional Assessment) were available for 948.
Results A positive association in univariate analysis with low-level PAE throughout pregnancy and cognition (β=4.1, 95% CI −0.02 to 8.22, p=0.05) was attenuated by adjusting for environmental/social deprivation risk factors (β=3.06 (−1.19 to 7.30), p=0.16). Early binge drinking, plus continued PAE at lower levels, was associated with the child being more likely to score low in sensation avoidance (adjusted OR 1.88 (1.03 to 3.41), p=0.04).
Conclusion Early binge exposure, followed by lower-level PAE, demonstrated an increase in sensation-avoiding behaviour. There were, however, no significant associations between PAE and neurodevelopment following adjustment for important confounders and modifiers. Follow-up is paramount to investigate subtle or later onset problems.
- pregnancy
- cohort studies
- paediatrics
- alcohol
Statistics from Altmetric.com
Introduction
Prenatal alcohol exposure (PAE) is a community health problem. The prevalence of children with fetal alcohol spectrum disorder (FASD) in mainstream educational settings is between 2% and 5%, accounting for poor school functioning in many children.1 2 Some of these children will have been exposed in utero to chronic high levels of alcohol, the risks of which are well recognised.3 4 However, for others, alcohol exposure may have been moderate5 or at an occasional binge level,6 for which the risks are less clear. Data on the impact of maternal occasional binge drinking on the developing fetus are variable: the ALSPAC study in the UK reported developmental and learning problems (hyperactivity and inattention) associated with this exposure in 4-year-olds,7 while a Danish study found no impact on intelligence in 5-year olds.8 When it comes to low-level PAE, the evidence pertaining to its safety is even more conflicting and poorly understood.9 10
In this climate of uncertainty, pregnancy guidelines in more than 45 countries have used the precautionary principle and advised that abstinence from alcohol in pregnancy is the safest option.11 This recommendation is problematic for women who have unintended pregnancies, for example, 40% in Australia,12 and who may be drinking before pregnancy recognition.13 It is also recognised that many women choose to drink alcohol at low levels during pregnancy,14 15 possibly because of inconsistent or lack of advice from health professionals, who themselves may be confused by the uncertainty of the relationship between low PAE and FASD.16–19
Genetic and environmental variability make it challenging to assign an attributable risk to specific levels of PAE,16 20 but attempts should be made to produce more convincing evidence about the possible harm associated with commonplace patterns of drinking, such as low PAE or early pregnancy binge drinking. In doing so, researchers must confront major methodological imperatives. First, accurate measures of PAE are essential but inherently difficult to obtain, are often based on maternal self-report and retrospective and lack details about the dose, pattern and timing of drinking. In addition, the definition of exposure levels varies around the world, which adds to the difficulties in interpretation of results.6 Second, appropriately adjusted analyses of multiple different measures of neurodevelopmental outcomes, accounting for as many contributing factors as possible, are extremely complex and consequently vary from study to study.21–23
The Asking QUestions about Alcohol (AQUA) study comprises a unique cohort of mother/child dyads recruited from the general population in early pregnancy for longitudinal observation.24 As PAE is the primary focus of this study, its accuracy has been sought through detailed prospective data collection and classification.13 Extensive information on contextual factors that may help explain any apparent association between PAE and child neurodevelopment was also collected. The aim of this paper was to examine the relationship between common patterns of PAE (low to moderate PAE and occasional binge) and early childhood development, minimising residual confounding and modification.
Methods
Participants and data collection
All women with a singleton pregnancy, attending their first antenatal appointment before 19 weeks gestation between 25 July 2011 and 30 July 2012 and attending one of seven public hospital recruitment sites in metropolitan Melbourne, Australia, were eligible to participate. Being 16 years or older and being able to read and write English were prerequisites for participation. The methods are described in detail in the protocol publication.24 During pregnancy, women completed three questionnaires. Questionnaire 1 was administered at the time of recruitment and collected information about the 3 months prior to conception until 13 weeks gestation (n=2146). Questionnaire 2 covered the period from the 14th to the 26th week (n=1715, 79.9%), and questionnaire 3 included the remainder of the pregnancy (n=1571, 73.2%).24 Postnatal questionnaires were sent to women who had completed questionnaires 1–3. Questionnaire 4 was administered at 12 months after birth (n=1285, 81.8%) and questionnaire 5 at 24 months after birth (n=1038, 66.1%).
Assessment of exposure
Prior to recruitment and as part of the design of the study, we discussed draft questions about alcohol in pregnancy with pregnant women in focus groups.25 The final exposure measure used in the AQUA study was informed by their opinions and preferences. Data on frequency of drinking and amount and type of alcoholic drink(s) on each occasion were collected in all three pregnancy questionnaires. Trimester 1 (T1) data were separated into before pregnancy awareness (pre-aware) and after pregnancy awareness (post-aware). In total, there were detailed data on alcohol consumption for five stages of pregnancy: (1) 3 months before, (2) T1 pre-aware, (3) T1 post-aware, (4) second trimester and (5) third trimester. Levels of exposure were expressed as grams of absolute alcohol (AA) using algorithms previously described13 24 26: abstinent during pregnancy (but not lifetime abstainer), low (≤20 g AA/occasion and ≤70 g AA/week), moderate (21–49 g AA/occasion and ≤70 g AA/week), high (>70 g AA/week) and binge (≥50 g AA/occasion).24 In Australia, 10 g AA equals one standard drink.
The seven PAE groups used in the analysis were based on stage of pregnancy and exposure level.
Control group: abstinent throughout pregnancy (but not lifetime abstainers)
Low in T1, abstinent in T2 and T3
Moderate/high in T1, abstinent in T2 and T3
Binge pre-aware, abstinent in T2 and T3
Low in T1, low/moderate in T2 and/or T3
Moderate in T1, any level in T2 and/or T3
Binge pre-aware, low/moderate in T2 and/or T3
Assessment of neurodevelopment at 24 months
Clinical review
As their child approached 2 years of age, 1163 women who completed all pregnancy questionnaires and were not lifetime abstainers were invited to participate in a clinical review of their child. Three follow-up contacts were possible, asking mothers to bring their child in to the Royal Children’s Hospital, Melbourne, for a neurodevelopmental assessment using the Bayley Scales of Infant and Toddler Development–Third Edition (Bayley III).27 The Bayley III assesses cognitive, language and motor development, with age-standardised composite scores for each domain, with a mean of 100 and an SD of 15. A difference in mean score of 3 to 4 points in these scales is equivalent to an effect size of 0.2 to 0.3 SD and is considered clinically meaningful. The assessment was administered in a hospital setting by one of three trained psychologists blinded to the child’s PAE grouping. Inter-rater reliability was maximised through a thorough training protocol and quality control (cross-observation and scoring) throughout the data collection period.
Maternal report
Questionnaire 5, completed by the mothers (n=1038) when their child was 2 years old, included two validated screening tests of the child’s sensory processing and social–emotional development. The Infant/Toddler Sensory Profile is a 48-item screen for sensory processing behaviours in daily experiences (ie, auditory, visual, tactile, vestibular and oral sensory processing).28 Mothers rate the frequency of each behaviour (1 ‘almost always’ to 5 ‘almost never’), and scores are grouped to generate four quadrants: low registration (length/intensity of stimuli needed to respond), sensation seeking (level of interest and pleasure associated with sensory stimuli), sensory sensitivity (ability to notice sensory stimuli) and sensation avoiding (need to control amount/intensity of sensory stimuli). Lower scores suggest more of the behaviour representing that sensory processing quadrant.
Social–emotional problems and competencies were assessed with the Brief Infant Toddler Social Emotional Assessment (BITSEA).29 The BITSEA has 42 items, each scored as 0, 1 or 2 (never occurs, sometimes occurs and always occurs, respectively), yielding two subscales: the Problem scale and the Competence scale. The Problem scale comprises 31 items assessing elements of social–emotional problems such as externalising problems, internalising problems, dysregulation and maladaptive and atypical behaviours. The Competence scale comprises 11 items assessing social–emotional abilities such as compliance, attention, peer relations and empathy.
Maternal, infant and family explanatory factors (online supplementary table 1)
Supplementary file 1
The data pertaining to potential confounders and effect modifiers of neurodevelopmental outcomes are presented in detail in supplementary table 1.
Potential confounders measured in pregnancy, questionnaires 1–3
Factors determined from the literature as a priori confounders were collected during pregnancy using questionnaires 1–3.24 Univariable data are shown in online supplementary table 1. The reference category used in analyses is assigned an asterisk (*) here: maternal age at recruitment (<25, 25–29, 30–34* and ≥35); maternal educational attainment (secondary, trade/diploma and tertiary*); household income (up to $40 000, $40–70 000, >$70–100 000 and >$100 000*); ethnicity (Caucasian* and Asian/other); language spoken at home (English always*/sometimes/never); parity (0*, 1 and >1); prepregnancy body mass index (BMI: underweight, normal*, overweight and obese); folic acid supplementation in trimester 1 and trimester 2/3 (none, <0.4 mg, 0.4–0.5 mg*, >0.5–0.8 mg and >0.8 mg); smoking in pregnancy (no*, pre-aware period only and throughout); age started drinking regularly (<18 years and 18+ years*); paternal drinking 3 months before pregnancy (abstinent*, light and occasional, low moderate, moderate and frequent).
The Australian Index of Relative Socio-economic Disadvantage summarises and ranks a range of information about the economic and social conditions of people and households within a geographic area, that is, the mother’s postal code at recruitment, using the 2011 Census of Population, published by the Australian Bureau of Statistics.30 The socioeconomic index included was categorised into quartiles with the lowest comprising the most disadvantaged areas, and the highest being the reference category.
Dietary health was based on detailed assessment of information provided in questionnaire 2, using a modified Food Frequency Questionnaire.31 This measures usual dietary intake and has been validated for use in large-scale epidemiological studies32 and analysed in depth for a separate publication.33 The 74 food type variables in the food frequency questionnaire were used as input variables for extraction of dietary patterns using principle components analysis. A ‘Healthy’ pattern, characterised by high intake of fruits and vegetables, and an ‘Unhealthy’ pattern, characterised by high intake of meats, processed foods, sweets and pasta, were generated. Participants’ scores on each dietary pattern were standardised and divided into tertiles, the second tertile being the reference category.
Although not in the original protocol paper as an a priori variable, amount of exercise was also included, as it has been described as a potential confounder in more recent studies,34 at least a 30 min exercise in T1 (none, 1–3 days a week*, 4–7 days a week).
Potential effect modifiers measured after birth
Postnatal factors determined a priori as having the potential to modify a PAE effect on neurodevelopment were measured after birth.24 Details are also shown in supplementary table 1. Variables were: sex of baby (male*, female), birth weight and breast feeding (none, breast fed <6 months, breast fed 6–12 months, breast fed >12 months*). The caregiving environment at 24 months was accounted for using measures collected in questionnaire 5: maternal anxiety, stress and depression (DASS-21)35; the General Family Functioning subscale of the McMaster Family Assessment Device (normal*, abnormal)36; global rating of self-efficacy (lowest quartile, other quartiles*)37; Child Rearing Questionnaire (6/30 items rating parental warmth (lowest quartile, other quartiles*))37 38; and the Hostile Parenting Scale (lowest quartile, other quartiles*).39
Data analysis
Stata V.14 was used to undertake all analyses.40 Observations with missing data were excluded from the analysis due to their small numbers.
Bayley III at 24 months
The association of PAE with neurodevelopment was investigated using multivariable linear regression and a two-model stepwise approach. Model 1 was adjusted for covariates measured in pregnancy that were associated with PAE and the outcome, or the outcome alone, at a 10% level (p value ≤0.1) (online supplementary table 1). Likewise, Model 2 was adjusted for all infant, parental characteristics and family measures collected after birth associated with the outcome (p value ≤0.1) and retained all covariates from model 1 associated with the outcome (p value ≤0.1).
The results are reported as difference in mean score (regression coefficient) with 95% CI and p value. Significance was designated at the adjusted p value<0.01; however, results at p<0.05 were deemed suggestive of possible associations.
ITSP and BITSEA at 24 months
All scores were calculated according to the ITSP and BISTEA manuals. Age-corrected ITSP raw scores for each quadrant were standardised to a z-scale. The z scores were dichotomised, with the cut-off for an ‘atypical’ response set at 1 SD below the mean. The reference category was those without an atypical response. Age-corrected and sex-corrected BITSEA scores were calculated for each of the two subscales and dichotomised. The cut-off for the ‘Problem’ subscale was the ≥25th percentile and ≤15th percentile for ‘Competence’.29 The reference categories were <25th percentile and > 15th percentile, respectively. Both the ITSP and the BITSEA were analysed for an association with PAE using multivariable logistic regression. Covariates with a p value <0.1 in univariate regressions (data not shown) were included in a single multivariate model. Results are reported as ORs with 95% CIs and p values.
Results
The number and percent of children with a Bayley III assessment in each PAE category are shown in table 1. Almost 23% were controls without alcohol in pregnancy. The smallest group was exposed to binge levels before pregnancy awareness and then no more (9.4%), while the largest alcohol-exposed group comprised children of women who drank moderately in trimester 1 and continued to drink at any level in trimesters 2 and/or 3 (17.3%).
PAE and cognition, language and motor development at 24 months
Bayley III
Response rate
There were 564 2-year-old children (49%) who completed the assessment as 296 (25%) mothers declined, 254 (22%) did not respond and 49 (4%) cancelled or were not able to schedule an appointment time. Online supplementary table 2 shows that the largest attrition occurred in the control group (abstainers) and that the children who completed the assessment were evenly spread across the other PAE groups. In online supplementary table 3, BITSEA and ITSP summary data on approximately 240 children who did not complete the Bayley III assessment at 24 months are compared with those who did complete it. The percentage categorised as abnormal is not significantly different between the groups.
Ten children were excluded from the analysis. These included two children with developmental disorders of genetic aetiology (Down syndrome and dopa-responsive dystonia) and one child who was very shy and refused to complete the assessment. The remaining seven exclusions were one maternal lifetime abstainer and six children exposed to alcohol in only the second and/or third trimester. Table 1 shows the distribution of the 554 children across the seven PAE groups and the corresponding mean composite scores and SD in each of the three developmental domains of the Bayley III.
Table 2 presents the results of the stepwise approach to multivariable linear regression of the association between the PAE and the three Bayley III domains of cognitive, language and motor development.
Multivariate analysis of the association between PAE and child outcome at 24 months: Bayley III
When compared with children of abstainers, the group with low PAE in T1 and low/moderate in T2/3 scored on average 4.1 points higher on the cognitive scale (p=0.05) and 5.5 points higher on the language scale in the unadjusted models (p=0.06). These differences are reduced (3.0 and 5.1, respectively) following adjustment for important confounders, which included sociodemographic variables, exercise in pregnancy, parity, diet, smoking and folate supplementation. Those with low PAE, only in T1, had an adverse result in the adjusted model for their child’s motor development (−4.2). For the early binge drinking categories, no results approached significance in any of the three domains.
After adjusting for all other covariates, the ones that significantly affected the models are shown in table 2 as footnotes. By using a conservative p value of ≤0.01, only the strongest covariates were included. Being female had a positive effect on both cognition and language, and higher parity had a positive effect on motor development. Negative effects on cognition were observed with low maternal education, low household income and breastfeeding limited to 6–12 months, as compared with 12 months or more. A shorter duration of breastfeeding also had a negative effect on motor development. Negative effects on language were observed with non-Caucasian ethnicity and prepregnancy obesity (as determined by BMI). Data not shown in table 2 are other variables contributing to the model with p levels >0.01, but <0.05. These included a lower level of maternal education (a 3-point to 5-point negative score in all three models, p<0.02), smoking throughout pregnancy (−8.8, p=0.03, in the language domain) and low household income (−4.3, p=0.03, for motor development).
ITSP and BITSEA
Of the 1038 mothers who completed questionnaire 5, the maximum number available for analysis was 948 as 51 were lifetime abstainers and 39 did not fit into any of the six other PAE groupings (eg, drank only in trimester 2 or 3). The numbers available for calculation of the unadjusted ORs for each of the ITSP quadrants are shown in table 3, along with the adjusted ORs (AORs).
Multivariate analysis of the association between PAEand child outcome at 24 months: Infant/Toddler Sensory Profile
Children of abstainers during pregnancy were the control group. Remembering that low scores in these quadrants indicate a higher propensity to show abnormal behaviours, of borderline significance is the finding that mothers who drank at a low level, only in the first trimester, were less likely to have a child with low scores in sensation seeking (AOR 0.38, p=0.04). Also, children with early binge level exposure, followed by exposure at low, moderate or high levels later in pregnancy, were almost two times as likely to have a low score in the sensation avoiding quadrant as those without PAE (AOR 1.88, p=0.04).
BITSEA data are shown in table 4. There was no evidence of associations between PAE and Problem or Competence scores of the BITSEA.
Multivariate analysis of the association between PAE and child outcome at 24 months: Brief Infant–Toddler Social and Emotional Assessment
Variables used in the regressions and those that remained significant (p<0.01) in the adjusted models are shown as footnotes to tables 3 and 4. Maternal stress and aspects of parenting style were independently significant in more than one outcome analysis.
Discussion
Summary of findings
Using a general antenatal population, rather than one comprised of mainly chronic, high-level drinkers, we found no evidence of significant associations between PAE and neurodevelopment at 2 years. Although initial analysis suggested there may be a positive association between cognitive development and low-level drinking in T1 and continuing into T2/T3, the inclusion of many important prenatal and postnatal covariates in multivariable analyses reduced this difference.
In addition to direct assessment of cognitive, language and motor development using the Bayley III, we used maternally reported measures to investigate effects of different patterns of PAE on sensory modulation (ITSP) and social–emotional development (BITSEA) at age 2 years. When compared with children without PAE, several atypical responses (greater than −1 SD) in the ITSP were seen. First, children of women who binge drank before knowing about the pregnancy and then continued to drink at a lower level for the remainder of the pregnancy had an almost twofold increased likelihood of scoring low in sensation avoiding. Children with low scores for sensation avoiding are more likely to engage in disruptive behaviours as a coping strategy to reduce sensory input.28 Conduct problems such as these may accumulate in early childhood through exposure to stressful situations and possibly lead to problem behaviours later.41 Second, children of women who drank at low levels before pregnancy awareness had a lower likelihood to score low in sensation seeking, compared with those without PAE. Children who have low scores for sensation seeking have been proposed to be more active and continuously engaged in their environment searching for new sensory stimulation. Given their borderline significance, these findings should be interpreted with caution but can be reassessed as the children grow older.
Comparison with other studies
The finding in this study that low-level PAE is not associated with a positive neurodevelopmental outcome in 2-year-olds is important in that it adds considerable weight to the argument that any putative association between such an exposure and improved outcomes was due to methodological issues in previous studies.10 42–44 One example is the UK Millennium Cohort study, which found that 3 -year-old children of mothers who drank lightly during pregnancy had slightly fewer developmental problems than controls.45 As this study used a different measure of cognition, direct comparison with our findings is not possible, but we conjecture that inclusion of covariates such as prepregnancy BMI, folate supplementation throughout pregnancy and breastfeeding may have reduced this association, as it did in our study. In addition, the UK measure of PAE was collected 9 months after birth and therefore likely subject to recall bias. Likewise, a 2014 meta-analysis reported a small but significant positive association with cognition and mild/moderate PAE.46 The authors placed many caveats on the importance of this finding, such as the accuracy of their measures of PAE and the likelihood of residual confounding, especially as the association was reduced if studies not controlling for SES were excluded.
A covariate finding of particular interest was that breastfeeding beyond 12 months compared with 6–12 months had an independent positive influence on all three domains of neurodevelopment. This finding may be of importance to women who drank alcohol before pregnancy recognition and are anxious about outcomes for their baby. Although the role of this modifiable risk factor in improving cognitive development is controversial,47 a novel biological study of this topic provides strong evidence for a positive association.48 Using brain imaging in children from 2.2 to 4 years, a study has shown that breastfeeding beyond 12 months improves white matter microstructure in somatosensory regions of the brain. This in turn is positively associated with cognitive and behavioural performance.48 Our results are in accordance with this finding, keeping in mind that the variable we used was duration, not exclusivity of breastfeeding. Therefore, in line with recommendations in Australian and WHO guidelines, our data suggest that women should be encouraged to breastfeed for as long as possible,49 especially if they have been drinking any alcohol in pregnancy.
Strengths and weaknesses
Along with the prospective and detailed collection of PAE data,13 the extensive collection of confounder and modifier data for inclusion in the analyses was a strength of this study. Many of the sociodemographic, psychological and lifestyle determinants of health had strong associations with the neurodevelopmental outcomes in our study, independently of PAE. Non-modifiable risk factors typical of environmental/social deprivation, such as young maternal age and low household income, were associated with lower neurodevelopmental scores. Non-Caucasian ethnicity remained in the final model as an independent predictor of lower language scores, and we observed the influence of child sex on both cognition and language, with females doing better than males. The fact that these variables remained significant in the final models provides confidence in the rigour of our analyses. A limitation was the response rate for the Bayley III assessment. There was disproportionate attrition of the control group, who may have been less interested in the project as they were non-drinkers, with attrition evenly spread across the alcohol-exposed groups. Importantly, on examination of data available on the maternally reported neurodevelopment of approximately 240 children of women who completed questionnaire 5 but did not participate in the clinical review, there was no difference between children of participants and non-participants in the proportion with outcomes in the lower range (less than −1 SD).
This study was designed to determine whether it was possible to detect subtle, early signs of neurodevelopmental delay in association with exposure to socially prevalent patterns of alcohol consumption in pregnancy. Although these results are reassuring for parents whose children have been exposed to low-level PAE, there are limitations to the predictive capacity of neurodevelopmental assessment at age 2 years.50 51 We plan to follow-up these children longitudinally to investigate more subtle or later onset problems.4 52
Finally, cognition, behaviour and personality all have a genetic basis,53 54 and the contribution of PAE to such developmental outcomes will be influenced by the genetic profiles of mother and child. In addition, knowledge of genotypes related to alcohol metabolism is yielding important information about cognitive development in children following PAE,55 as is a growing understanding of epigenetic markers in the fetus,56 and in paternal sperm,57 that alter developmental programming. Research in these areas is in its infancy but likely to yield novel and important information in the near future as advances in genomic technology allow for more detailed investigations into the risk of having a child with FASD.
Conclusion
This study of 2-year-old children, using a prospective measurement of PAE and carefully controlling for important confounders and modifiers in the analyses, afforded an opportunity to unravel the effect that commonplace patterns of PAE have on early cognitive, language, motor, sensory and behavioural development. The only evidence of adverse associations between these developmental domains and PAE at 2 years of age was in sensory development, where binge drinking before pregnancy awareness plus continued lower-level drinking throughout pregnancy was associated with sensation-avoiding behaviours in children. We emphasise the importance of following these children as they develop.
What is already known on this subject
Children born after prenatal alcohol exposure (PAE) to low-level or sporadic drinking are at an unknown risk of adverse neurodevelopmental outcomes. Chronic, high-level drinking is strongly associated with fetal alcohol spectrum disorder.
Studies examining child outcomes after low-level exposure have produced conflicting results, summarised in a meta-analysis by Flak in 2014, some even suggesting there may be a positive association. Inaccurate measurement of PAE and lack of measurement of important cofactors contribute to uncertainty about the effect of all PAE on the unborn baby.
What this study adds
Using detailed measures of common patterns of PAE in the general population, and adjusting analyses for many maternal cofactors reflecting social deprivation, lifestyle, diet and psychological status, the children in our study had no adverse Bayley III results at 2 years of age.
Follow-up to 2 years may not be long enough to detect effects of low level and sporadic PAE; as these could manifest later in childhood development, follow-up is essential.
Acknowledgments
The authors are extremely grateful to all the women and their children who took part in this study. We thank the psychologists who undertook the Bayley III assessments: Michelle Livock, Claire Corbett and Siobhan James.
References
Footnotes
Contributors JH, EM, SL, EE, DA, CO’L, SD, DF, CN, JC and PA all contributed to the planning, conduct and reporting of this study. Analysis of data was conducted by JH who also drafted the manuscript. Further interpretation of data and intellectual input were provided by EM, PA and SL in particular. JH is the guarantor. JH, EM, SL, EE, DA, CO’L, SD, DF, CN, JC and PA all contributed to writing and reviewing the manuscript and approved the final version.
Funding This work was supported by the Australian National Health and Medical Research Council (grant no. 1011070; senior research fellowships nos. 1081288 (PA) and 1021252 (JH); practitioner fellowship no. 1021480 (EE)) and the Victorian state government’s operational Infrastructure support program.
Competing interests None declared.
Patient consent Detail has been removed from this case description/these case descriptions to ensure anonymity. The editors and reviewers have seen the detailed information available and are satisfied that the information backs up the case the authors are making.
Ethics approval Multiple human research ethics committees at participating sites in Victoria Australia.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No additional dare are available.