Search

Researchers faced with incomplete data are encouraged to consider whether their data are ‘missing completely at random’ (MCAR), ‘missing at random’ (MAR) or ‘missing not at random’ (MNAR) when planning their analysis. However, there are two major problems with this classification as originally defined by Rubin in the 1970s. First, when there are missing data in multiple variables, the plausibility of the MAR assumption is difficult to assess using substantive knowledge and is more stringent than is generally appreciated. Second, although MCAR and MAR are sufficient conditions for consistent estimation with specific methods, they are not necessary conditions and therefore this categorization does not directly determine the best approach for handling the missing data in an analysis. How best to handle missing data depends on the assumed causal relationships between variables and their missingness, and what these relationships imply in terms of the ‘recoverability’ of the target estimand (the population parameter that encodes the answer to the underlying research question). Recoverability is defined as whether the estimand can be consistently estimated from the patterns and associations in the observed data without needing to invoke external information on the extent to which the distribution of missing values might differ from that of observed values. In this manuscript we outline an approach for deciding which method to use to handle multivariable missing data in an analysis, using directed acyclic graphs to depict missingness assumptions and determining the implications in terms of recoverability of the target estimand.

Three-level data structures arising from repeated measures on individuals clustered within larger units are common in health research studies. Missing data are prominent in such studies and are often handled via multiple imputation (MI). Although several MI approaches can be used to account for the three-level structure, including adaptations to single- and two-level approaches, when the substantive analysis model includes interactions or quadratic effects, these too need to be accommodated in the imputation model. In such analyses, substantive model compatible (SMC) MI has shown great promise in the context of single-level data. Although there have been recent developments in multilevel SMC MI, to date only one approach that explicitly handles incomplete three-level data is available. Alternatively, researchers can use pragmatic adaptations to single- and two-level MI approaches, or two-level SMC-MI approaches. We describe the available approaches and evaluate them via simulations in the context of three three-level random effects analysis models involving an interaction between the incomplete time-varying exposure and time, an interaction between the time-varying exposure and an incomplete time-fixed confounder, or a quadratic effect of the exposure. Results showed that all approaches considered performed well in terms of bias and precision when the target analysis involved an interaction with time, but the three-level SMC MI approach performed best when the target analysis involved an interaction between the time-varying exposure and an incomplete time-fixed confounder, or a quadratic effect of the exposure. We illustrate the methods using data from the Childhood to Adolescence Transition Study.

Aim To evaluate changes in in-hospital mortality rate following implementation of a comprehensive electronic medical record (EMR) system. Methods Before and after study of 355,709 hospital discharges, over an 8-year period, at a paediatric teaching hospital. The major outcome measures were crude number of in-hospital deaths, deaths per 1000 discharges, and standardised mortality ratio. Results Primary analysis of data from 2 years before and 2 years after EMR go-live showed a reduction in absolute mortality of 33 deaths, a reduction in the mortality rate of 0.48 per 1000 discharges (95% CI 0.09, 0.88 per 1000): and a relative 22% decrease (95% CI: 4%, 36%, P = 0.02) in deaths per 1000 discharges from 2.20 to 1.72. There was also a reduction in standardised mortality ratio of 47% (95% CI: 18%, 66%, P = 0.004). Post-hoc analysis of mortality rates for an additional 2-year pre-intervention period indicated that these changes in the mortality rate were not part of a pre-existing downward trend. Further analysis of an additional 20-month post-intervention period suggests that the reduced mortality rate has been sustained. Conclusion We documented evidence of a clinically important decrease in in-hospital mortality rate following the implementation of a modern comprehensive EMR system in an Australian paediatric teaching hospital. The study does not prove a causal relationship, and it is possible that other factors explain some, or all, of this difference, but no changes in the hospital population or other major interventions were identified as alternative explanations for this observed change.

OBJECTIVES: To describe characteristics and outcomes of children requiring intensive care therapy (ICT) within 12 hours following a medical emergency team (MET) event. DESIGN: Retrospective cohort study. SETTING: Quaternary paediatric hospital. PATIENTS: Children experiencing a MET event. MEASUREMENTS AND MAIN RESULTS: Between July 2017 and March 2019, 890 MET events occurred in 566 patients over 631 admissions. Admission to intensive care followed 183/890 (21%) MET events. 76/183 (42%) patients required ICT, defined as positive pressure ventilation or vasoactive support in intensive care, within 12 hours. Older children had a lower risk of requiring ICT than infants aged < 1 year (age 1–5 years [risk difference, -6.4%; 95% CI, -11% to -1.6%; P = 0.01] v age > 5 years [risk difference, -8.0%; 95% CI, -12% to -3.8%; P < 0.001]), while experiencing a critical event increased this risk (risk difference, 16%; 95% CI, 3.3–29%; P = 0.01). The duration of respiratory support and intensive care length of stay was approximately double in patients requiring ICT (ratio of geometric means, 2.0 [95% CI, 1.4–3.0] v 2.1 [95% CI, 1.5–2.8]; P < 0.001) and the intensive care mortality increased (risk difference, 9.6%; 95% CI, 2.4–17%; P = 0.01). Heart rate, oxygen saturation and respiratory rate were the most commonly measured vital signs in the 6 hours before the MET event. CONCLUSIONS: Approximately one-fifth of MET events resulted in intensive care admission and nearly half of these required ICT within 12 hours. This group had greater duration of respiratory support, intensive care and hospital length of stay, and higher mortality. Age < 1 year and a critical event increased the risk of ICT.

Medical research often involves using multi-item scales to assess individual characteristics, disease severity, and other health-related outcomes. It is common to observe missing data in the scale scores, due to missing data in one or more items that make up that score. Multiple imputation (MI) is a popular method for handling missing data. However, it is not clear how best to use MI in the context of scale scores, particularly when they are assessed at multiple waves of data collection resulting in large numbers of items. The aim of this article is to provide practical advice on how to impute missing values in a repeatedly measured multi-item scale using MI when inference on the scale score is of interest. We evaluated the performance of five MI strategies for imputing missing data at either the item or scale level using simulated data and a case study based on four waves of the Longitudinal Study of Australian Children (LSAC). MI was implemented using both multivariate normal imputation and fully conditional specification, with two rules for calculating the scale score. A complete case analysis was also performed for comparison. Based on our results, we caution against the use of a MI strategy that does not include the scale score in the imputation model(s) when the scale score is required for analysis. Refereed/Peer-reviewed

BackgroundThere is increasing evidence that patterns of pubertal maturation are associated with different patterns of health risk. This study aimed to explore the associations between anthropometric measures and salivary androgen concentrations in pre-adolescent children.MethodsWe analysed a stratified random sample (N=1151) of pupils aged 8–9 years old from 43 primary schools in Melbourne, Australia from the Childhood to Adolescence Transition Study. Saliva samples were assayed for dehydroepiandrosterone (DHEA), DHEA-sulfate and testosterone. Anthropometric measures included height, weight, body mass index (BMI) and waist circumference. Associations between (1) anthropometric measures and each androgen, and (2) hormone status with obesity and parental report of pubertal development were investigated using linear regression modelling with general estimating equations.ResultsGreater height, weight, BMI and waist circumference were positively associated with higher androgen concentrations, after adjusting for sex and socioeconomic status. Being overweight or obese was associated with higher testosterone and DHEA concentrations compared with the normal BMI category. Those who were obese were more likely (OR=2.7, 95% CI 1.61 to 4.43, pConclusionThis study provides clear evidence for an association between obesity and higher androgen levels in mid-childhood. The adrenal transition may be a critical time period for weight management intervention strategies in order to manage the risk for metabolic problems in later life for high-risk individuals.

Objectives To estimate SARS‐CoV‐2‐specific antibody seroprevalence after the first epidemic wave of coronavirus disease 2019 (COVID‐19) in Sydney. Setting, participants People of any age who had provided blood for testing at selected diagnostic pathology services (general pathology); pregnant women aged 20–39 years who had received routine antenatal screening; and Australian Red Cross Lifeblood plasmapheresis donors aged 20–69 years. Design Cross‐sectional study; testing of de‐identified residual blood specimens collected during 20 April – 2 June 2020. Main outcome measure Estimated proportions of people seropositive for anti‐SARS‐CoV‐2‐specific IgG, adjusted for test sensitivity and specificity. Results Thirty‐eight of 5339 specimens were IgG‐positive (general pathology, 19 of 3231; antenatal screening, 7 of 560; plasmapheresis donors, 12 of 1548); there were no clear patterns by age group, sex, or location of residence. Adjusted estimated seroprevalence among people who had had general pathology blood tests (all ages) was 0.15% (95% credible interval [CrI], 0.04–0.41%), and 0.29% (95% CrI, 0.04–0.75%) for plasmapheresis donors (20–69 years). Among 20–39‐year‐old people, the age group common to all three collection groups, adjusted estimated seroprevalence was 0.24% (95% CrI, 0.04–0.80%) for the general pathology group, 0.79% (95% CrI, 0.04–1.88%) for the antenatal screening group, and 0.69% (95% CrI, 0.04–1.59%) for plasmapheresis donors. Conclusions Estimated SARS‐CoV‐2 seroprevalence was below 1%, indicating that community transmission was low during the first COVID‐19 epidemic wave in Sydney. These findings suggest that early control of the spread of COVID‐19 was successful, but efforts to reduce further transmission remain important.

Introduction There is an emerging realisation that paediatric reference intervals (RIs) estimated using discrete age-groups may be misleading, especially close to age cut-off values. This limitation has been addressed by estimating RIs that vary continuously with age. This systematic review examines the range of statistical methods used over the past 25 years for estimation of age-specific RIs, and identifies trends in usage and reporting. Methods Literature searches were conducted using predefined search criteria for original publications between 1993 and 2018 on the MEDLINE and Embase databases. Data related to sample size, treatment of age (as categorical or continuous), and statistical methods were extracted from the selected publications. Results A total of 238 publications were reviewed. Not all publications reported the statistical methods used in different steps. Among the publications, 167 (70%) reported discrete age-group RIs, 54 (23%) reported continuous RIs and 17 (7%) reported both types of RIs. The nonparametric statistical method was commonly used for discrete age-group RIs (64%, n = 117), whereas a wide variety of curve-fitting approaches, including Cole's lambda-mu-sigma method (28%, n = 20), parametric curve-based methods (28%, n = 20), generalised additive model for location, scale and shape method (13%, n = 9) and quantile regression (11%, n = 8) were used for continuous RIs. Conclusions Improvement in the reporting of statistical methods used for estimating age-specific paediatric RIs is required. There has been insufficient uptake of methods for producing continuous RIs, especially for biomarkers that display strong age-dependence.

Aim This paper describes the use of the single patient therapy plan (SPTP). The SPTP has been designed to assess the efficacy at an individual level of a commercially available cannabinoid product, cannabidiol, in reducing seizure frequency in paediatric patients with intractable epilepsy. Methods The SPTP is a randomised, double-blind, placebo-controlled N-of-1 trial designed to assess the efficacy of treatment in a neurology outpatient setting. The primary objective of the SPTP is to assess the efficacy of cannabidiol in reducing seizure frequency in each patient with intractable epilepsy, with change in seizure frequency being the primary outcome of interest. The analysis adopts a Bayesian approach, which provides results in the form of posterior probabilities that various levels of benefit (based on the primary outcome measure, seizure frequency) have been achieved under active treatment compared to placebo, accompanied by decision rules that provide thresholds for deciding whether treatment has been successful in the individual patient. The SPTP arrangement is most accurately considered part of clinical practice rather than research, since it is aimed at making clinical treatment decisions for individual patients and is not testing a hypothesis or collecting aggregate data. Therefore, Human Research Ethics Committee approval was considered not to be required, although it is recommended that hospital Clinical Ethics Committees provide ethical oversight. Conclusion These SPTP resources are made available so that they may inform clinical practice in the treatment of severe epilepsy or adapted for use in other conditions.

Objectives: Gestational age at birth is declining, probably because more deliveries are being induced. Gestational age is an important modifiable risk factor for neonatal mortality and morbidity. We aimed to investigate the association between gestational age and mortality in hospital for term-born neonates (>= 37 wk') admitted to PICUs in Australia and New Zealand. Design: Observational multicenter cohort study. Setting: PICUs in Australia and New Zealand. Patients: Term-born neonates (>= 37 wk) admitted to PICUs. Interventions: None Measurements and Main Results: We studied 5,073 infants born with a gestational age greater than or equal to 37 weeks and were less than 28 days old when admitted to a PICU in Australia or New Zealand between 2007 and 2016. The association between gestational age and mortality was estimated using a multivariable logistic regression model, adjusting for age, sex, indigenous status, Pediatric Index of Mortality version 2, and site. The median gestational age was 39.1 weeks (interquartile range, 38.2-40 wk) and mortality in hospital was 6.6%. Risk of mortality declined log-linearly with gestational age. The adjusted analysis showed a 20% (95% CI, 11-28%) relative reduction in mortality for each extra week of gestation beyond 37 weeks. The effect of gestation was stronger among those who received extracorporeal life support: each extra week of gestation was associated with a 44% (95% CI, 25-57%) relative reduction in mortality. Longer gestation was also associated with reduced length of stay in hospital: each week increase in gestation, the average length of stay decreased by 4% (95% CI, 2-6%). Conclusions: Among neonates born at "term" who are admitted to a PICU, increasing gestational age at birth is associated with a substantial reduction in the risk of dying in hospital. The maturational influence on outcome was more strongly noted in the sickest neonates, such as those requiring extracorporeal life support. This information is important in view of the increasing proportion of planned births in both high- and low-/middle-income countries.

Objective: Poorer mental health in adulthood is associated with increased risk of cardiovascular disease and reduced life expectancy. However, little is known of the molecular pathways underpinning this relationship and how early in life adverse metabolite profiles relate to self-reported variation in mental health. We examined cross-sectional associations between mental health and serum metabolites indicative of cardiovascular health, in large Australian population-based cohorts at two stages of the life-course. Methods: We characterised cross-sectional serum nuclear magnetic resonance metabolite profiles of positively and negatively framed mental health in a large population-based sample of Australian 11- to 12-year-olds ( n = 1172; 51% girls) and mid-life adults (n = 1322; mean age 45 years; 87% women). We examined multiple standard self-report mental health scales, spanning psychosocial health, general well-being, life satisfaction, and health-related quality of life. Linear regression was used to investigate the cross-sectional association between mental health and each metabolite (n = 73) in children and adults separately, unadjusted and adjusted for age, sex, socioeconomic position and body mass index. Results: Better child and adult mental health were associated with lower levels of the inflammatory marker glycoprotein acetyls, and a favourable, less atherogenic lipid/lipoprotein profile. Patterns of association in children were generally weaker than in adults. Associations were generally modest and partially attenuated when adjusted for body mass index. Conclusions: In general, metabolite profiles associated with better child and adult mental health closely aligned with those predictive of better cardiovascular health in adults. Our findings support previous evidence for the likely bidirectional relationship between mental health and cardiovascular disease risk, by extending this evidence base to the molecular level and in children.

Introduction To facilitate best possible patient care, reference intervals (RIs) adopted by a laboratory must be appropriate for the population demographics and, where applicable, the analytical principle and/or the analytical instrument used. While guidelines from the Clinical and Laboratory Standard Institute (CLSI) recommend a validation process for discrete RIs, there are no current recommendations for the validation process for continuous RIs. This study aimed to validate recently published, HAPPI Kids continuous RIs, in a routine laboratory. Methods Initially, the difference in test results between the primary study laboratory that contributed to previous RIs development and a routine laboratory was assessed using specimens from 77 children tested in both laboratories using the Siemens ADVIA 1800 or Centaur/XP/XPT. Later, validation of the HAPPI Kids RIs was undertaken using 279 pediatric samples tested on the same analyzer type in the routine laboratory. The previously published RIs were validated if more than 90% of results in the routine laboratory were within the RIs. Results There was minimal evidence of clinically significant differences in test results between the primary and routine laboratories. The continuous RIs were validated after initial analysis for 16 of the 18 biochemistry analytes tested, and after secondary analysis for the remaining 2 analytes. Conclusion This study validates the HAPPI Kids RIs in a routine laboratory, satisfying the laboratory accreditation requirements for evaluation, implementation, and sourcing of RIs. In addition, this study presents a modification of the current CLSI method for validation of continuous RIs that will benefit routine laboratories in general.

Background: Blinding of treatment allocation from treating clinicians in neonatal randomised controlled trials can minimise performance bias, but its effectiveness is rarely assessed. Methods: To examine the effectiveness of blinding a procedural intervention from treating clinicians in a multicentre randomised controlled trial of minimally invasive surfactant therapy versus sham treatment in preterm infants of gestation 25–28 weeks with respiratory distress syndrome. The intervention (minimally invasive surfactant therapy or sham) was performed behind a screen within the first 6 h of life by a ‘study team’ uninvolved in clinical care including decision-making. Procedure duration and the study team’s words and actions during the sham treatment mimicked those of the minimally invasive surfactant therapy procedure. Post-intervention, three clinicians completed a questionnaire regarding perceived group allocation, with the responses matched against actual intervention and categorised as correct, incorrect, or unsure. Success of blinding was calculated using validated blinding indices applied to the data overall (James index, successful blinding defined as > 0.50), or to the two treatment allocation groups (Bang index, successful blinding: −0.30 to 0.30). Blinding success was measured within staff role, and the associations between blinding success and procedural duration and oxygenation improvement post-procedure were estimated. Results: From 1345 questionnaires in relation to a procedural intervention in 485 participants, responses were categorised as correct in 441 (33%), incorrect in 142 (11%), and unsure in 762 (57%), with similar proportions for each of the response categories in the two treatment arms. The James index indicated successful blinding overall 0.67 (95% confidence interval (CI) 0.65–0.70). The Bang index was 0.28 (95% CI 0.23–0.32) in the minimally invasive surfactant therapy group and 0.17 (95% CI 0.12–0.21) in the sham arm. Neonatologists more frequently guessed the correct intervention (47%) than bedside nurses (36%), neonatal trainees (31%), and other nurses (24%). For the minimally invasive surfactant therapy intervention, the Bang index was linearly related to procedural duration and oxygenation improvement post-procedure. No evidence of such relationships was seen in the sham arm. Conclusion: Blinding of a procedural intervention from clinicians is both achievable and measurable in neonatal randomised controlled trials.

ObjectivesTo (1) describe the dispensing of asthma preventers at hospital discharge and estimate its effect on hospital readmissions, and (2) estimate the effect of community asthma preventer dispensing on readmissions for the subgroup of children who were not prescribed an asthma preventer at discharge.DesignMultisite cohort study with linked administrative data.ParticipantsChildren aged 3–18 years admitted with asthma to a tertiary paediatric, mixed paediatric and adult, or regional hospital between 2017 and 2018.Main outcome measureHospital readmission for asthma within 12 months.ResultsOf the 767 participants, 201 (26.2%) were newly prescribed or requested to continue with asthma preventers. Of these, only 91 (45.3%) dispensed their discharge prescription within 3 days or had an active prescription. There was no evidence for a protective effect of discharge asthma preventer dispensing on asthma hospital readmissions within 12 months (OR 1.17, 95% CI 0.69 to 1.97, p=0.57). Of the 566 children who were not prescribed asthma preventers at discharge, 269 (47.5%) had one or more prescriptions dispensed in the community within 12 months. Participants who were in the protected period (asthma preventer dispensed) had reduced risk of an asthma hospital readmission (HR 0.61, 95% CI 0.36 to 1.02, p=0.06), including preschool children (HR 0.48, 95% CI 0.25, 0.93, p=0.03) on subgroup analysis.ConclusionsThere was a low rate for prescribing and dispensing of hospital discharge asthma preventers and no protective effect was found for its impact on readmissions. A protective effect on readmissions was found for community asthma preventer dispensing.

Background As of mid-2021, Australia’s only nationwide coronavirus disease 2019 (COVID-19) epidemic occurred in the first 6 months of the pandemic. Subsequently, there has been limited transmission in most states and territories. Understanding community spread during the first wave was hampered by initial limitations on testing and surveillance. To characterize the prevalence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-specific antibody seroprevalence generated during this time, we undertook Australia’s largest national SARS-CoV-2 serosurvey. Methods Between June 19 and August 6, 2020, residual specimens were sampled from people undergoing general pathology testing (all ages), women attending antenatal screening (20–39 years), and blood donors (20–69 years) based on the Australian population’s age and geographic distributions. Specimens were tested by Wantai total SARS-CoV-2-antibody assay. Seroprevalence estimates adjusted for test performance were produced. The SARS-CoV-2 antibody-positive specimens were characterized with microneutralization assays. Results Of 11 317 specimens (5132 general pathology; 2972 antenatal; 3213 blood-donors), 71 were positive for SARS-CoV-2-specific antibodies. Seroprevalence estimates were 0.47% (95% credible interval [CrI], 0.04%–0.89%), 0.25% (CrI, 0.03%–0.54%), and 0.23% (CrI, 0.04%–0.54%), respectively. No seropositive specimens had neutralizing antibodies. Conclusions Australia’s seroprevalence was extremely low (

IMPORTANCE The benefits of surfactant administration via a thin catheter (minimally invasive surfactant therapy [MIST]) in preterm infants with respiratory distress syndrome are uncertain.OBJECTIVE To examine the effect of selective application of MIST at a low fraction of inspired oxygen threshold on survival without bronchopulmonary dysplasia (BPD).DESIGN, SETTING, AND PARTICIPANTS Randomized clinical trial including 485 preterm infants with a gestational age of 25 to 28 weeks who were supported with continuous positive airway pressure (CPAP) and required a fraction of inspired oxygen of 0.30 or greater within 6 hours of birth. The trial was conducted at 33 tertiary-level neonatal intensive care units around the world, with blinding of the clinicians and outcome assessors. Enrollment took place between December 16, 2011, and March 26, 2020; follow-up was completed on December 2, 2020.INTERVENTIONS Infants were randomized to the MIST group (n = 241) and received exogenous surfactant (200 mg/kg of poractant alfa) via a thin catheter or to the control group (n = 244) and received a sham (control) treatment; CPAP was continued thereafter in both groups unless specified intubation criteria were met.MAIN OUTCOMES AND MEASURES The primary outcome was the composite of death or physiological BPD assessed at 36 weeks' postmenstrual age. The components of the primary outcome (death prior to 36 weeks' postmenstrual age and BPD at 36 weeks' postmenstrual age) also were considered separately.RESULTS Among the 485 infants randomized (median gestational age, 27.3 weeks; 241 [49.7%] female), all completed follow-up. Death or BPD occurred in 105 infants (43.6%) in the MIST group and 121 (49.6%) in the control group (risk difference [RD], -63% [95% CI, -14.2% to 1.6%); relative risk [RR], 0.87 [95% CI, 0.74 to 1.03]; P = .10). Incidence of death before 36 weeks' postmenstrual age did not differ significantly between groups (24 [10.0%) in MIST vs 19 [7.8%) in control; RD, 2.1% [95% CI, -3.6% to 7.8%); RR, 1.27 [95% CI, 0.63 to 2.57]; P = .51), but incidence of BPD in survivors to 36 weeks' postmenstrual age was lower in the MIST group (81/217 [37.3%] vs 102/225 [45.3%) in the control group; RD, -7.8% [95% CI, -14.9% to -0.7%); RR. 0.83 [95% CI, 0.70 to 0.98]; P = .03). Serious adverse events occurred in 10.3% of infants in the MIST group and 11.1% in the control group.CONCLUSIONS AND RELEVANCE Among preterm infants with respiratory distress syndrome supported with CPAP, minimally invasive surfactant therapy compared with sham (control) treatment did not significantly reduce the incidence of the composite outcome of death or bronchopulmonary dysplasia at 36 weeks' postmenstrual age. However, given the statistical uncertainty reflected in the 95% CI, a clinically important effect cannot be excluded. (C) 2021 American Medical Association. All rights reserved.

Background Dynamic treatment regimens (DTRs) formalise the multi-stage and dynamic decision problems that clinicians often face when treating chronic or progressive medical conditions. Compared to randomised controlled trials, using observational data to optimise DTRs may allow a wider range of treatments to be evaluated at a lower cost. This review aimed to provide an overview of how DTRs are optimised with observational data in practice. Methods Using the PubMed database, a scoping review of studies in which DTRs were optimised using observational data was performed in October 2020. Data extracted from eligible articles included target medical condition, source and type of data, statistical methods, and translational relevance of the included studies. Results From 209 PubMed abstracts, 37 full-text articles were identified, and a further 26 were screened from the reference lists, totalling 63 articles for inclusion in a narrative data synthesis. Observational DTR models are a recent development and their application has been concentrated in a few medical areas, primarily HIV/AIDS (27, 43%), followed by cancer (8, 13%), and diabetes (6, 10%). There was substantial variation in the scope, intent, complexity, and quality between the included studies. Statistical methods that were used included inverse-probability weighting (26, 41%), the parametric G-formula (16, 25%), Q-learning (10, 16%), G-estimation (4, 6%), targeted maximum likelihood/minimum loss-based estimation (4, 6%), regret regression (3, 5%), and other less common approaches (10, 16%). Notably, studies that were primarily intended to address real-world clinical questions (18, 29%) tended to use inverse-probability weighting and the parametric G-formula, relatively well-established methods, along with a large amount of data. Studies focused on methodological developments (45, 71%) tended to be more complicated and included a demonstrative real-world application only. Conclusions As chronic and progressive conditions become more common, the need will grow for personalised treatments and methods to estimate the effects of DTRs. Observational DTR studies will be necessary, but so far their use to inform clinical practice has been limited. Focusing on simple DTRs, collecting large and rich clinical datasets, and fostering tight partnerships between content experts and data analysts may result in more clinically relevant observational DTR studies.

Multiple imputation (MI) is increasingly popular for handling multivariate missing data. Two general approaches are available in standard computer packages: MI based on the posterior distribution of incomplete variables under a multivariate (joint) model, and fully conditional specification (FCS), which imputes missing values using univariate conditional distributions for each incomplete variable given all the others, cycling iteratively through the univariate imputation models. In the context of longitudinal or clustered data, it is not clear whether these approaches result in consistent estimates of regression coefficient and variance component parameters when the analysis model of interest is a linear mixed effects model (LMM) that includes both random intercepts and slopes with either covariates or both covariates and outcome contain missing information. In the current paper, we compared the performance of seven different MI methods for handling missing values in longitudinal and clustered data in the context of fitting LMMs with both random intercepts and slopes. We study the theoretical compatibility between specific imputation models fitted under each of these approaches and the LMM, and also conduct simulation studies in both the longitudinal and clustered data settings. Simulations were motivated by analyses of the association between body mass index (BMI) and quality of life (QoL) in the Longitudinal Study of Australian Children (LSAC). Our findings showed that the relative performance of MI methods vary according to whether the incomplete covariate has fixed or random effects and whether there is missingnesss in the outcome variable. We showed that compatible imputation and analysis models resulted in consistent estimation of both regression parameters and variance components via simulation. We illustrate our findings with the analysis of LSAC data.

BACKGROUNDAge-specific reference intervals (RIs) have been developed for biochemistry analytes in children. However, the ability to interpret results from multiple laboratories for 1 individual is limited. This study reports a head-to-head comparison of reference values and age-specific RIs for 30 biochemistry analytes for children across 5 analyzer types.METHODSBlood was collected from healthy newborns and children 30 days to RESULTSThe variations of age-specific mean reference values between analyzer types were within allowable total error (Westgard QC) for most analytes, and common age-specific reference limits were reported as functions of age and/or sex. Analyzer-specific reference limits for all analytes on 5 analyzer types are also reported as functions of age and/or sex.CONCLUSIONSThis study provides quantitative and qualitative measures of the extent to which results for individual children can or cannot be compared across analyzer types, and the feasibility of RI harmonization. The reported equations enable incorporation of age-specific RIs into laboratory information systems for improving evidence-based clinical decisions in children.

ObjectivesTo describe a well-established marker of cardiovascular risk, carotid intima–media thickness (IMT) and related measures (artery distensibility and elasticity) in children aged 11–12 years old and mid-life adults, and examine associations within parent–child dyads.DesignCross-sectional study (Child Health CheckPoint), nested within a prospective cohort study, the Longitudinal Study of Australian Children (LSAC).SettingAssessment centres in seven Australian major cities and eight selected regional towns, February 2015 to March 2016.ParticipantsOf all participating CheckPoint families (n=1874), 1489 children (50.0% girls) and 1476 parents (86.8% mothers) with carotid IMT data were included. Survey weights and methods were applied to account for LSAC’s complex sample design and clustering within postcodes and strata.Outcome measuresUltrasound of the right carotid artery was performed using standardised protocols. Primary outcomes were mean and maximum far-wall carotid IMT, quantified using semiautomated edge detection software. Secondary outcomes were carotid artery distensibility and elasticity. Pearson’s correlation coefficients and multivariable linear regression models were used to assess parent–child concordance. Random effects modelling on a subset of ultrasounds (with repeated measurements) was used to assess reliability of the child carotid IMT measure.ResultsThe average mean and maximum child carotid IMT were 0.50 mm (SD 0.06) and 0.58 mm (SD 0.05), respectively. In adults, average mean and maximum carotid IMT were 0.57 mm (SD 0.07) and 0.66 mm (SD 0.10), respectively. Mother–child correlations for mean and maximum carotid IMT were 0.12 (95% CI 0.05 to 0.23) and 0.10 (95% CI 0.03 to 0.21), respectively. For carotid artery distensibility and elasticity, mother–child correlations were 0.19 (95% CI 0.10 to 0.25) and 0.11 (95% CI 0.02 to 0.18), respectively. There was no strong evidence of father–child correlation in any measure.ConclusionsWe provide Australian values for carotid vascular measures and report a modest mother–child concordance. Both genetic and environmental exposures are likely to contribute to carotid IMT.

ObjectivesNuclear magnetic resonance (NMR) metabolomics is high throughput and cost-effective, with the potential to improve the understanding of disease and risk. We examine the circulating metabolic profile by quantitative NMR metabolomics of a sample of Australian 11–12 year olds children and their parents, describe differences by age and sex, and explore the correlation of metabolites in parent–child dyads.DesignThe population-based cross-sectional Child Health CheckPoint study nested within the Longitudinal Study of Australian Children.SettingBlood samples collected from CheckPoint participants at assessment centres in seven Australian cities and eight regional towns; February 2015–March 2016.Participants1180 children and 1325 parents provided a blood sample and had metabolomics data available. This included 1133 parent–child dyads (518 mother–daughter, 469 mother–son, 68 father–daughter and 78 father–son).Outcome measures228 metabolic measures were obtained for each participant. We focused on 74 biomarkers including amino acid species, lipoprotein subclass measures, lipids, fatty acids, measures related to fatty acid saturation, and composite markers of inflammation and energy homeostasis.ResultsWe identified differences in the concentration of specific metabolites between childhood and adulthood and in metabolic profiles in children and adults by sex. In general, metabolite concentrations were higher in adults than children and sex differences were larger in adults than in children. Positive correlations were observed for the majority of metabolites including isoleucine (CC 0.33, 95% CI 0.27 to 0.38), total cholesterol (CC 0.30, 95% CI 0.24 to 0.35) and omega 6 fatty acids (CC 0.28, 95% CI 0.23 to 0.34) in parent–child comparisons.ConclusionsWe describe the serum metabolite profiles from mid-childhood and adulthood in a population-based sample, together with a parent–child concordance. Differences in profiles by age and sex were observed. These data will be informative for investigation of the childhood origins of adult non-communicable diseases and for comparative studies in other populations.

There are now a growing number of applications of multilevel regression and poststratification (MRP) in population health and epidemiological studies. MRP uses multilevel regression to model individual survey responses as a function of demographic and geographic covariates. Estimated mean outcome values for each demographic-geographic respondent subtype are then weighted by the proportions of each subtype in the population to produce an overall population-level estimate. We recently reported an extensive case study of a large nationwide survey and found that MRP performed favorably compared to conventional survey sampling weights for the estimation of population descriptive quantities in a highly selected sample. In this study, we aimed to evaluate, by way of a simulation experiment, both the accuracy and precision of MRP versus survey sampling weights in the context of large population health studies. While much of the research into MRP has been focused on U.S. political and social science, we considered an alternative population structure of smaller size and with notably fewer geographic subsets. We explored the impact on MRP performance of sample size, model misspecification, interactions, and the addition of a geographic-level covariate. MRP was found to achieve generally superior performance in both accuracy and precision at both the national and state levels. Results were generally robust to model misspecification, and MRP performance was further improved by the inclusion of a geographic-level covariate. These findings offer further evidence that MRP provides a promising analytic approach for addressing participation bias in the estimation of population descriptive quantities from large-scale health surveys and cohort studies.

Focus of Presentation Multivariable regression models are widely used in epidemiological data analysis. Traditional teaching often focusses on technical aspects with insufficient attention paid to the purposes for which regression methods are used. Findings We have addressed these issues in a new short course that provides an introduction to regression analysis in the context of learning about causal inference, beginning from the standpoint that the majority of research questions in epidemiology are causal in nature. This approach leads naturally to using regression models in two different ways, firstly for direct estimation of a causal effect, under an assumption of constancy of the effect across strata of confounders, and secondly for prediction of outcomes, as a necessary step in the estimation of causal effects via g-computation. Conclusions/Implications Approaching the teaching of regression methods within a causal inference framework helps to dispel confusion created by traditional statistical approaches that imply the existence of “true models” and encourage the building of models in a way that is unclear about the purpose for which they will be used, for example seeking to identify “risk factors” in an exploratory manner. Key messages The teaching and practice of regression methods in epidemiology can be enhanced by emphasising the key differences between three distinct analytic purposes: description, prediction, causal. Regression models may play a role in all three but the way in which models are developed and interpreted differs between them.

Background Causal inference from cohort studies is central to epidemiological research. Targeted Maximum Likelihood Estimation (TMLE) is an appealing doubly robust method for causal effect estimation, but it is unclear how missing data should be handled when it is used in conjunction with machine learning approaches for the exposure and outcome models. This is problematic because missing data are ubiquitous and can result in biased estimates and loss of precision if handled inappropriately. Methods Based on a motivating example from the Victorian Adolescent Health Cohort Study, we conducted a simulation study to evaluate the performance of available approaches for handling missing data when using TMLE with machine learning. These included complete-case analysis; an extended TMLE approach incorporating an outcome missingness probability model; the missing indicator approach for missing covariate data (MCMI); and multiple imputation (MI) using standard parametric approaches or machine learning algorithms. We considered 11 missingness mechanisms typical in cohort studies, and a simple and a complex setting, in which exposure and outcome generation models included two-way and higher-order interactions. Results MI using regression with no interactions and MI with random forest yielded estimates with the highest bias. MI with regression including two-way interactions was the best performing method overall. Of the non-MI approaches, MCMI performed the worst Conclusions When using TMLE with machine learning to estimate the average causal effect, avoiding standard MI with no interactions and MCMI is recommended. Key messages We provide novel guidance for handling missing data for causal effect estimation using TMLE.

Background Outcome regression remains widely applied for estimating causal effects in observational studies, in which causal inference is conceptualised as emulating a randomized controlled trial (RCT). Multiple imputation (MI) is a commonly used method for handling missing data, but while in RCTs it has been shown that MI should be conducted by treatment group to reduce bias, whether imputation should be conducted by exposure group in observational studies has not been studied. Methods We conducted a simulation study to evaluate the performance of seven methods for handling missing data: Complete-case analysis (CCA), MI of main effect, MI with interactions (between exposure and: outcome, a strong confounder, outcome and a strong confounder, all incomplete), and MI conducted by exposure group. We simulated data based on an example from the Victorian Adolescent Health Cohort Study. Three exposure prevalences and seven outcome generation models were considered, the latter ranging from no interaction to strong-positive or negative exposure-confounder interaction. Various missingness scenarios were examined: with incomplete outcome only or also incomplete confounders, and three levels of complexity regarding the missingness mechanism. Results For all scenarios, MI by exposure led to the least bias, followed by MI approaches that included exposure-confounder interactions. Conclusions If MI is adopted in outcome regression, we recommend conducting MI by exposure group and, when not feasible, including exposure-confounder interactions in the imputation model. Key messages Similar to RCTs, MI should be conducted by exposure group when estimating average causal effects using outcome regression in observational studies.

Chronic respiratory diseases are often difficult to cure and are likely to originate early in life. Therefore, early identification of such diseases is of interest for early prevention.We explored the potential to predict these almost from birth; using data at 1 month of age, we attempted to predict disease occurrence 4 years later in life. Our data came from the Barwon Infant Study; after cleaning and processing, we had measurements on 41 variables from 401 participants.We considered three respiratory diseases: asthma, wheeze and hay fever. As predictors, we used a variety of information that would be available in a clinical setting. Of particular interest to our investigation was whether lung function measurements (newly available at such an early age) would helpfully improve predictive accuracy. We also investigated whether maternal smoking (previously associated with respiratory illnesses) is a helpful predictor.Our methods included logistic regression as the main model, multiple imputation to deal with missing values, stepwise selection and LASSO to select variables, and cross-validation to assess performance. We measured predictive performance using AUC (area under the receiver operating characteristic curve), sensitivity and specificity.Broadly, we found that the best models had only modest predictive power for each disease. For example, for asthma we achieved an AUC of 0.67, a sensitivity of 68% and a corresponding specificity of 63%. Performance for the other two diseases was similar.We also found that our lung function measurements didnotimprove predictive performance; some-what surprisingly, this was also true for maternal smoking. The most useful predictors included, among others, family history of these diseases and variables relating to the size of the infants.Given the modest performance of these models, our findings suggest that very early prediction of respiratory illnesses is still a challenging task.

For cluster randomized trials (CRTs) with a small number of clusters, the matched-pair (MP) design, where clusters are paired before randomizing one to each trial arm, is often recommended to minimize imbalance on known prognostic factors, add face-validity to the study, and increase efficiency, provided the analysis recognizes the matching. Little evidence exists to guide decisions on when to use matching. We used simulation to compare the efficiency of the MP design with the stratified and simple designs, based on the mean confidence interval width of the estimated intervention effect. Matched and unmatched analyses were used for the MP design; a stratified analysis was used for the stratified design; and analyses without and with post-stratification adjustment for factors that would otherwise have been used for restricted allocation were used for the simple design. Results showed the MP design was generally the most efficient for CRTs with 10 or more pairs when the correlation between cluster-level outcomes within pairs (matching correlation) was moderate to strong (0.3-0.5). There was little gain in efficiency for the MP or stratified designs compared to simple randomization when the matching correlation was weak (0.05-0.1). For trials with four pairs of clusters, the simple and stratified designs were more efficient than the MP design because greater degrees of freedom were available for the analysis, although an unmatched analysis of the MP design recovered precision for weak matching correlations. Practical guidance on choosing between the MP, stratified, and simple designs is provided.