John B Carlin

Murdoch Childrens Research Institute, Melbourne, Victoria, Australia

Are you John B Carlin?

Claim your profile

Publications (408)2000.21 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Multiple imputation (MI), a two-stage process whereby missing data are imputed multiple times and the resulting estimates of the parameter(s) of interest are combined across the completed datasets, is becoming increasingly popular for handling missing data. However, MI can result in biased inference if not carried out appropriately or if the underlying assumptions are not justifiable. Despite this, there remains a scarcity of guidelines for carrying out MI. In this paper we provide a tutorial on the main issues involved in employing MI, as well as highlighting some common pitfalls and misconceptions, and areas requiring further development. When contemplating using MI we must first consider whether it is likely to offer gains (reduced bias or increased precision) over alternative methods of analysis. Once it has been decided to use MI, there are a number of decisions that must be made during the imputation process; we discuss the extent to which these decisions can be guided by the current literature. Finally we highlight the importance of checking the fit of the imputation model. This process is illustrated using a case study in which we impute missing outcome data in a five-wave longitudinal study that compared extremely preterm individuals with term-born controls.
    No preview · Article · Feb 2016 · International Journal of Social Research Methodology
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background and objectives: Infant body composition and postnatal weight gain have been implicated in the development of adult obesity and cardiovascular disease, but there are limited prospective data regarding the association between infant adiposity, postnatal growth, and early cardiovascular parameters. Increased aortic intima-media thickness is an intermediate phenotype of early atherosclerosis. The aim of this study was to investigate the relationship between weight and adiposity at birth, postnatal growth, and aortic intima-media thickness. Methods: The Barwon Infant Study (n=1074 mother-infant pairs) is a population-derived birth cohort. Infant weight and other anthropometry were measured at birth and 6 weeks of age. Aortic intima-media thickness was measured by trans-abdominal ultrasound at 6 weeks of age (n=835). Results: After adjustment for aortic size and other factors, markers of adiposity including increased birth weight (β=19.9μm/kg, 95%CI 11.1, 28.6 p<0.001) and birth skinfold thickness (β=6.9μm/mm, 95%CI 3.3, 10.5 p<0.001) were associated with aortic intima-media thickness at 6 weeks. The association between birth skin-fold thickness and aortic IMT was independent of birth weight. In addition, greater postnatal weight gain was associated with increased aortic intima-media thickness, independent of birth weight and age at time of scan (β=11.3μm/per kg increase, 95%CI 2.2, 20.3 p=0.01). Conclusion: Increased infant weight and adiposity at birth, as well as increased early weight gain, were positively associated with aortic intima-media thickness. Excessive accumulation of adiposity during gestation and early infancy may have adverse effects on cardiovascular risk.
    No preview · Article · Dec 2015 · Clinical Science
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Multiple imputation (MI) is a well-recognised statistical technique for handling missing data. As usually implemented in standard statistical software, MI assumes that data are ‘Missing at random’ (MAR); an assumption that in many settings is implausible. It is not possible to distinguish whether data are MAR or ‘Missing not at random’ (MNAR) using the observed data, so it is desirable to discover the impact of departures from the MAR assumption on the MI results by conducting sensitivity analyses. A weighting approach based on a selection model has been proposed for performing MNAR analyses to assess the robustness of results obtained under standard MI to departures from MAR. Methods In this article, we use simulation to evaluate the weighting approach as a method for exploring possible departures from MAR, with missingness in a single variable, where the parameters of interest are the marginal mean (and probability) of a partially observed outcome variable and a measure of association between the outcome and a fully observed exposure. The simulation studies compare the weighting-based MNAR estimates for various numbers of imputations in small and large samples, for moderate to large magnitudes of departure from MAR, where the degree of departure from MAR was assumed known. Further, we evaluated a proposed graphical method, which uses the dataset with missing data, for obtaining a plausible range of values for the parameter that quantifies the magnitude of departure from MAR. Results Our simulation studies confirm that the weighting approach outperformed the MAR approach, but it still suffered from bias. In particular, our findings demonstrate that the weighting approach provides biased parameter estimates, even when a large number of imputations is performed. In the examples presented, the graphical approach for selecting a range of values for the possible departures from MAR did not capture the true parameter value of departure used in generating the data. Conclusions Overall, the weighting approach is not recommended for sensitivity analyses following MI, and further research is required to develop more appropriate methods to perform such sensitivity analyses.
    Preview · Article · Dec 2015 · BMC Medical Research Methodology
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose Mental and behavioral disorders increase in prevalence with the passage through puberty. Yet the first symptoms for many children emerge between seven and 11 years, before the pubertal rise in gonadal hormones. A possibility that symptom onset may be linked to the adrenarchal rise in androgens has been little explored. Methods The Childhood to Adolescence Transition Study recruited a stratified random sample of 1,239 eight-nine year olds from primary schools in Melbourne, Australia. Saliva samples were assayed for dehydroepiandrosterone, dehydroepiandrosterone-sulphate (DHEA-S), and testosterone. Emotional and behavioral problems were assessed through parental report on the Strengths and Difficulties Questionnaire. Results In males, high levels of all androgens were associated with greater total difficulties and peer problems. Higher dehydroepiandrosterone and testosterone were associated with emotional symptoms and DHEA-S with conduct problems. In females, DHEA-S was associated with peer problems. Conclusions In late childhood, androgens are associated with emotional and behavioral problems in males, raising a possibility that the adrenarchal transition plays a contributing role. If so, the late primary school years may prove to be an important phase for preventing the onset of mental health and behavioral problems in boys.
    No preview · Article · Dec 2015 · Journal of Adolescent Health
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: There are limited longitudinal data on the associations between different social contexts of alcohol use and risky adolescent drinking. Methods: Australian prospective longitudinal cohort of 1943 adolescents with 6 assessment waves at ages 14-17 years. Drinkers were asked where and how frequently they drank. Contexts were: at home with family, at home alone, at a party with friends, in a park/car, or at a bar/nightclub. The outcomes were prevalence and incidence of risky drinking (≥5 standard drinks (10g alcohol) on a day, past week) and very risky drinking (>20 standard drinks for males and >11 for females) in early (waves 1-2) and late (waves 3-6) adolescence. Results: Forty-four percent (95 % CI: 41-46 %) reported past-week risky drinking on at least one wave during adolescence (waves 1-6). Drinking at a party was the most common repeated drinking context in early adolescence (28 %, 95 % CI 26-30 %); 15 % reported drinking repeatedly (3+ times) with their family in early adolescence (95 % CI: 14-17 %). For all contexts (including drinking with family), drinking 3+ times in a given context was associated with increased the risk of risky drinking in later adolescence. These effects remained apparent after adjustment for potential confounders (e.g. for drinking with family, adjusted RR 1.9; 95 % CI: 1.5-2.4). Similar patterns were observed for very risky drinking. Conclusions: Our results suggest that consumption with family does not protect against risky drinking. Furthermore, parents who wish to minimise high risk drinking by their adolescent children might also limit their children's opportunities to consume alcohol in unsupervised settings.
    Full-text · Article · Nov 2015 · BMC Public Health
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background.—Distinguishing pathological from physiological relationships between vessel size and aortic intima-media thickness (aIMT) is an important challenge, especially in growing children. We examined the relationship between childhood vessel diameter and aIMT and assessed common analytic approaches used to address this relationship.Methods.—We analyzed aIMT in two population-derived cohorts; 6-week-old infants and 19-year-olds. We simulated datasets in which we assumed a simple physiological relationship between vessel diameter and aIMT, and then superimposed possible pathological effects on aIMT; (a) intrauterine growth retardation, (b) macrosomia and (c) both intrauterine growth retardation and macrosomia. Using simulated datasets and cohorts, we evaluated analytic strategies including those in which the relationship between vessel diameter and aIMT was (a) ignored, (b) adjusted for by dividing aIMT by weight, or (c) adjusted for using varying regression techniques.Results.—aIMT was found to increase in proportion to vessel diameter in both cohorts (138 μm/mm at 6 weeks and 52 μm/mm at 19 years of age). Simply dividing aIMT by weight produced negative associations with weight across all datasets. By contrast, adjusting for vessel diameter as a covariate enabled accurate distinction of the direction of the association between aIMT and weight in all simulated datasets. These results were replicated in the cohort studies for both aIMT and carotid intima-media thickness.Conclusion.—There is a physiological relationship between vessel diameter and aIMT. Simply dividing aIMT by weight may lead to incorrect assumptions regarding the relationship between weight and aIMT. However, the physiological relationship is appropriately estimated by including vessel diameter as a covariate in regression.
    No preview · Article · Sep 2015

  • No preview · Article · Aug 2015 · The Lancet
  • [Show abstract] [Hide abstract]
    ABSTRACT: Despite the success of rotavirus vaccines, suboptimal vaccine efficacy in regions with a high burden of disease continues to present a challenge to worldwide implementation. A birth dose strategy with a vaccine developed from an asymptomatic neonatal rotavirus strain has the potential to address this challenge and provide protection from severe rotavirus disease from birth. This phase 2a randomised, double-blind, three-arm, placebo-controlled safety and immunogenicity trial was undertaken at a single centre in New Zealand between Jan 13, 2012, and April 17, 2014. Healthy, full-term (≥36 weeks gestation) babies, who weighed at least 2500 g, and were 0-5 days old at the time of randomisation were randomly assigned (1:1:1; computer-generated; telephone central allocation) according to a concealed block randomisation schedule to oral RV3-BB vaccine with the first dose given at 0-5 days after birth (neonatal schedule), to vaccine with the first dose given at about 8 weeks after birth (infant schedule), or to placebo. The primary endpoint was cumulative vaccine take (serum immune response or stool shedding of vaccine virus after any dose) after three doses. The immunogenicity analysis included all randomised participants with available outcome data. This trial is registered with the Australian New Zealand Clinical Trials Registry, ACTRN12611001212943. 95 eligible participants were randomised, of whom 89 were included in the primary analysis. A cumulative vaccine take was detected in 27 (90%) of 30 participants in the neonatal schedule group after three doses of RV3-BB vaccine compared with four (13%) of 32 participants in the placebo group (difference in proportions 0·78, 95% CI 0·55-0·88; p<0·0001). 25 (93%) of 27 participants in the infant schedule group had a cumulative vaccine take after three doses compared with eight (25%) of 32 participants in the placebo group (difference in proportions 0·68, 0·44-0·81; p<0·0001). A serum IgA response was detected in 19 (63%) of 30 participants and 20 (74%) of 27 participants, and stool shedding of RV3-BB was detected in 21 (70%) of 30 participants and 21 (78%) of 27 participants in the neonatal and infant schedule groups, respectively. The frequency of solicited and unsolicited adverse events was similar across the treatment groups. RV3-BB vaccine was not associated with an increased frequency of fever or gastrointestinal symptoms compared with placebo. RV3-BB vaccine was immunogenic and well tolerated when given as a three-dose neonatal or infant schedule. A birth dose strategy of RV3-BB vaccine has the potential to improve the effectiveness and implementation of rotavirus vaccines. Australian National Health and Medical Research Council, the New Zealand Health Research Council, and the Murdoch Childrens Research Institute. Copyright © 2015 Elsevier Ltd. All rights reserved.
    No preview · Article · Aug 2015 · The Lancet Infectious Diseases
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In cost-effectiveness analyses of drugs or health technologies, estimates of life years saved or quality-adjusted life years saved are required. Randomised controlled trials can provide an estimate of the average treatment effect; for survival data, the treatment effect is the difference in mean survival. However, typically not all patients will have reached the endpoint of interest at the close-out of a trial, making it difficult to estimate the difference in mean survival. In this situation, it is common to report the more readily estimable difference in median survival. Alternative approaches to estimating the mean have also been proposed. We conducted a simulation study to investigate the bias and precision of the three most commonly used sample measures of absolute survival gain - difference in median, restricted mean and extended mean survival - when used as estimates of the true mean difference, under different censoring proportions, while assuming a range of survival patterns, represented by Weibull survival distributions with constant, increasing and decreasing hazards. Our study showed that the three commonly used methods tended to underestimate the true treatment effect; consequently, the incremental cost-effectiveness ratio (ICER) would be overestimated. Of the three methods, the least biased is the extended mean survival, which perhaps should be used as the point estimate of the treatment effect to be inputted into the ICER, while the other two approaches could be used in sensitivity analyses. More work on the trade-offs between simple extrapolation using the exponential distribution and more complicated extrapolation using other methods would be valuable. Copyright © 2015 John Wiley & Sons, Ltd. Copyright © 2015 John Wiley & Sons, Ltd.
    Full-text · Article · Jul 2015 · Pharmaceutical Statistics
  • [Show abstract] [Hide abstract]
    ABSTRACT: Perinatal depression is a neglected global health priority, affecting 10–15% of women in high-income countries and a greater proportion in low-income countries. Outcomes for children include cognitive, behavioural, and emotional difficulties and, in low-income settings, perinatal depression is associated with stunting and physical illness. In the Victorian Intergenerational Health Cohort Study (VIHCS), we aimed to assess the extent to which women with perinatal depressive symptoms had a history of mental health problems before conception.
    No preview · Article · Jun 2015 · The Lancet
  • [Show abstract] [Hide abstract]
    ABSTRACT: Children with cardiomyopathy (CM) are at risk of sudden cardiac death (SCD), but the incidence and risk factors for this outcome are not clear. This study sought to determine the incidence and risk factors for SCD in children with varying CM phenotypes from a long-term population-based study of childhood CM. The NACCS (National Australian Childhood Cardiomyopathy Study) is an ongoing longitudinal cohort study including all children in Australia with primary CM who were diagnosed between January 1, 1987, and December 31, 1996, and were <10 years of age. The cumulative incidence and risk factors for SCD within individual CM phenotypes were explored using survival analysis. Of 289 eligible patients, 16 (5.5%) experienced SCD over a median follow-up of 11.9 years (interquartile range: 1.7 to 15.4). The risk of SCD varied according to CM phenotype (p = 0.007). The cumulative incidence of SCD at 15 years was 5% for dilated cardiomyopathy (DCM), 6% for hypertrophic cardiomyopathy (HCM), 12% for restrictive cardiomyopathy, and 23% for left ventricular (LV) noncompaction. Older age at diagnosis, positive family history of CM, and severity of LV dysfunction were related to increased risk of SCD in patients with DCM, and a higher posterior wall thickness Z-score was the sole risk factor identified for patients with HCM. Predictors of SCD include CM phenotype, family history of CM (DCM), severity of systolic dysfunction (DCM), and extent of LV hypertrophy (HCM). Continuing follow-up of this cohort into adulthood is likely to reveal an ongoing risk of SCD. Copyright © 2015 American College of Cardiology Foundation. Published by Elsevier Inc. All rights reserved.
    No preview · Article · Jun 2015 · Journal of the American College of Cardiology
  • Cattram D. Nguyen · Katherine J. Lee · John B. Carlin
    [Show abstract] [Hide abstract]
    ABSTRACT: Multiple imputation is gaining popularity as a strategy for handling missing data, but there is a scarcity of tools for checking imputation models, a critical step in model fitting. Posterior predictive checking (PPC) has been recommended as an imputation diagnostic. PPC involves simulating "replicated" data from the posterior predictive distribution of the model under scrutiny. Model fit is assessed by examining whether the analysis from the observed data appears typical of results obtained from the replicates produced by the model. A proposed diagnostic measure is the posterior predictive "p-value", an extreme value of which (i.e., a value close to 0 or 1) suggests a misfit between the model and the data. The aim of this study was to evaluate the performance of the posterior predictive p-value as an imputation diagnostic. Using simulation methods, we deliberately misspecified imputation models to determine whether posterior predictive p-values were effective in identifying these problems. When estimating the regression parameter of interest, we found that more extreme p-values were associated with poorer imputation model performance, although the results highlighted that traditional thresholds for classical p-values do not apply in this context. A shortcoming of the PPC method was its reduced ability to detect misspecified models with increasing amounts of missing data. Despite the limitations of posterior predictive p-values, they appear to have a valuable place in the imputer's toolkit. In addition to automated checking using p-values, we recommend imputers perform graphical checks and examine other summaries of the test quantity distribution. © 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.
    No preview · Article · May 2015 · Biometrical Journal
  • Jock Lawrie · John B. Carlin · Andrew B. Forbes
    [Show abstract] [Hide abstract]
    ABSTRACT: Stepped wedge randomized trials are increasingly popular. Here we derive the optimal design for a fixed number of periods; this does not allocate an equal number of cluster units to each treatment sequence as might otherwise have been expected.
    No preview · Article · Apr 2015 · Statistics [?] Probability Letters
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: There is substantial interest in studying lung function in infants, to better understand the early life origins of chronic lung diseases such as asthma. Multiple breath washout (MBW) is a technique for measuring lung function that has been adapted for use in infants. Respiratory sighs occur frequently in young infants during natural sleep, and in accordance with current MBW guidelines, result in exclusion of data from a substantial proportion of testing cycles. We assessed how sighs during MBW influenced the measurements obtained using data from 767 tests conducted on 246 infants (50% male; mean age 43 days) as part of a large cohort study. Sighs occurred in 119 (15%) tests. Sighs during the main part of the wash-in phase (before the last 5 breaths) were not associated with differences in standard MBW measurements compared with tests without sighs. In contrast, sighs that occurred during the washout were associated with a small but discernible increase in magnitude and variability. For example, the mean lung clearance index increased by 0.36 (95% CI: 0.11-0.62) and variance increased by a multiplicative factor of 2 (95% CI: 1.6-2.5). The results suggest it is reasonable to include MBW data from testing cycles where a sigh occurs during the wash-in phase, but not during washout, of MBW. By recovering data that would otherwise have been excluded, we estimate a boost of about 10% to the final number of acceptable tests and 6% to the number of individuals successfully tested. © 2015 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of the American Physiological Society and The Physiological Society.
    Full-text · Article · Apr 2015
  • Kim Mulholland · John Carlin · Trevor Duke · Martin Weber

    No preview · Article · Mar 2015 · The Lancet Respiratory Medicine
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The modern environment is associated with an increasing burden of non-communicable diseases (NCDs). Mounting evidence implicates environmental exposures, experienced early in life (including in utero), in the aetiology of many NCDs, though the cellular/molecular mechanism(s) underlying this elevated risk across the life course remain unclear. Epigenetic variation has emerged as a candidate mediator of such effects. The Barwon Infant Study (BIS) is a population-derived birth cohort study (n = 1074 infants) with antenatal recruitment, conducted in the south-east of Australia (Victoria). BIS has been designed to facilitate a detailed mechanistic investigation of development within an epidemiological framework. The broad objectives are to investigate the role of specific environmental factors, gut microbiota and epigenetic variation in early-life development, and subsequent immune, allergic, cardiovascular, respiratory and neurodevelopmental outcomes. Participants have been reviewed at birth and at 1, 6, 9 and 12 months, with 2- and 4-year reviews under way. Biological samples and measures include: maternal blood, faeces and urine during pregnancy; infant urine, faeces and blood at regular intervals during the first 4 years; lung function at 1 month and 4 years; cardiovascular assessment at 1 month and 4 years; skin-prick allergy testing and food challenge at 1 year; and neurodevelopmental assessment at 9 months, 2 and 4 years. Data access enquiries can be made at [www.barwoninfantstudy.org.au] or via [peter.vuillermin@deakin.edu.au]. © The Author 2015; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association.
    Full-text · Article · Mar 2015 · International Journal of Epidemiology
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: As there is limited knowledge regarding the longitudinal development and early ontogeny of naïve and regulatory CD4(+) T-cell subsets during the first postnatal year, we sought to evaluate the changes in proportion of naïve (thymic and central) and regulatory (resting and activated) CD4(+) T-cell populations during the first postnatal year. Blood samples were collected and analyzed at birth, 6 and 12 months of age from a population-derived sample of 130 infants. The proportion of naïve and regulatory CD4(+) T-cell populations was determined by flow cytometry, and the thymic and central naïve populations were sorted and their phenotype confirmed by relative expression of T cell-receptor excision circle DNA (TREC). At birth, the majority (94%) of CD4(+) T cells were naïve (CD45RA(+)), and of these, ~80% had a thymic naïve phenotype (CD31(+) and high TREC), with the remainder already central naïve cells (CD31(-) and low TREC). During the first year of life, the naïve CD4(+) T cells retained an overall thymic phenotype but decreased steadily. From birth to 6 months of age, the proportion of both resting naïve T regulatory cells (rTreg; CD4(+)CD45RA(+)FoxP3(+)) and activated Treg (aTreg, CD4(+)CD45RA(-)FoxP3(high)) increased markedly. The ratio of thymic to central naïve CD4(+) T cells was lower in males throughout the first postnatal year indicating early sexual dimorphism in immune development. This longitudinal study defines proportions of CD4(+) T-cell populations during the first year of postnatal life that provide a better understanding of normal immune development.
    Full-text · Article · Mar 2015

  • No preview · Article · Jan 2015 · Pediatrics
  • [Show abstract] [Hide abstract]
    ABSTRACT: We describe Pseudomonas aeruginosa acquisitions in children with cystic fibrosis (CF) aged ≤5-years, eradication treatment efficacy, and genotypic relationships between upper and lower airway isolates and strains from non-CF sources. Of 168 CF children aged ≤5-years in a bronchoalveolar lavage (BAL)-directed therapy trial, 155 had detailed microbiological results. Overall, 201/271 (74%) P. aeruginosa isolates from BAL and oropharyngeal cultures were available for genotyping, including those collected before and after eradication therapy. Eighty-two (53%) subjects acquired P. aeruginosa, of which most were unique strains. Initial eradication success rate was 90%, but 36 (44%) reacquired P. aeruginosa, with genotypic substitutions more common in BAL (12/14) than oropharyngeal (3/11) cultures. Moreover, oropharyngeal cultures did not predict BAL genotypes reliably. CF children acquire environmental P. aeruginosa strains frequently. However, discordance between BAL and oropharyngeal strains raises questions over upper airway reservoirs and how to best determine eradication in non-expectorating children. Copyright © 2014 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.
    No preview · Article · Jan 2015 · Journal of cystic fibrosis: official journal of the European Cystic Fibrosis Society
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: As there is limited knowledge regarding the longitudinal development and early ontogeny of naı ̈ve and regulatory CD4+ T-cell subsets during the first postnatal year, we sought to evaluate the changes in proportion of naı ̈ve (thymic and central) and regulatory (resting and activated) CD4+ T-cell populations during the first postnatal year. Blood samples were collected and analyzed at birth, 6 and 12 months of age from a population-derived sample of 130 infants. The proportion of naı ̈ ve and regulatory CD4+ T-cell populations was determined by flow cytometry, and the thymic and central naı ̈ve populations were sorted and their phenotype confirmed by relative expression of T cell-receptor excision circle DNA (TREC). At birth, the majority (94%) of CD4+ T cells were naı ̈ ve (CD45RA+), and of these, ~ 80% had a thymic naı ̈ ve phenotype (CD31+ and high TREC), with the remainder already central naı ̈ve cells (CD31− and low TREC). During the first year of life, the naı ̈ve CD4+ T cells retained an overall thymic phenotype but decreased steadily. From birth to 6 months of age, the proportion of both resting naı ̈ve T regulatory cells (rTreg; CD4+CD45RA+FoxP3+) and activated Treg (aTreg, CD4+CD45RA−FoxP3high) increased markedly. The ratio of thymic to central naı ̈ve CD4+ T cells was lower in males throughout the first postnatal year indicating early sexual dimorphism in immune development. This longitudinal study defines proportions of CD4+ T-cell populations during the first year of postnatal life that provide a better understanding of normal immune development.
    Full-text · Article · Jan 2015

Publication Stats

25k Citations
2,000.21 Total Impact Points

Institutions

  • 2001-2015
    • Murdoch Childrens Research Institute
      • • Clinical Epidemiology & Biostatistics (CEBU)
      • • Research Group for Healthy Mothers Healthy Families
      Melbourne, Victoria, Australia
  • 1991-2015
    • The Royal Children's Hospital
      • • Department of Allergy and Immunology
      • • Clinical Epidemiology and Biostatistics Unit (CEBU)
      • • Centre for Adolescent Health
      • • Department of Respiratory Medicine
      Melbourne, Victoria, Australia
    • University of Melbourne
      • • Department of Paediatrics
      • • Department of Psychiatry
      Melbourne, Victoria, Australia
  • 2010
    • University of Bristol
      • Faculty of Medicine and Dentistry
      Bristol, England, United Kingdom
  • 2006
    • Royal Hospital for Women
      Sydney, New South Wales, Australia
    • Alfred Hospital
      • Department of Allergy, Immunology & Respiratory Medicine (AIRmed)
      Melbourne, Victoria, Australia
    • Cancer Council Victoria
      • Centre for Behavioural Research in Cancer
      Melbourne, Victoria, Australia
  • 2005
    • Harvard University
      Cambridge, Massachusetts, United States
  • 2000-2003
    • Royal Melbourne Hospital
      Melbourne, Victoria, Australia
  • 1999
    • Papua New Guinea Institute of Medical Research
      New Garoka, Eastern Highlands, Papua New Guinea
  • 1998
    • Victoria University Melbourne
      Melbourne, Victoria, Australia