Disagreement concerning atopic dermatitis subtypes between an English prospective cohort (ALSPAC) and linked electronic health records.

BACKGROUND: Subtypes of atopic dermatitis (AD) have been derived from the Avon Longitudinal Study of Parents and Children (ALSPAC) based on presence and severity of symptoms reported in questionnaires (Severe-Frequent, Moderate-Frequent, Moderate-Declining, Mild-Intermittent, Unaffected/Rare). Good agreement between ALSPAC and linked electronic health records (EHRs) would increase trust in the clinical validity of these subtypes and allow inferring subtypes from EHRs alone, which would enable their study in large primary care databases. OBJECTIVES: 1. Explore if presence and number of AD records in EHRs agrees with AD symptom and severity reports from ALSPAC; 2. Explore if EHRs agree with ALSPAC-derived AD subtypes; 3. Construct models to classify ALSPAC-derived AD subtype using EHRs. METHODS: We used data from the ALSPAC prospective cohort study from 11 timepoints until age 14 years (1991-2008), linked to local general practice EHRs. We assessed how far ALSPAC questionnaire responses and derived subtypes agreed with AD as established in EHRs using different AD definitions (e.g., diagnosis and/or prescription) and other AD-related records. We classified AD subtypes using EHRs, fitting multinomial logistic regression models tuning hyperparameters and evaluating performance in the testing set (ROC AUC, accuracy, sensitivity, and specificity). RESULTS: 8,828 individuals out of a total 13,898 had both been assigned an AD subtype and had linked EHRs. The number of AD-related codes in EHRs generally increased with severity of AD subtype, however not all with the Severe-Frequent subtypes had AD in EHRs, and many with the Unaffected/Rare subtype did have AD in EHRs. When predicting ALSPAC AD subtype using EHRs, the best tuned model had ROC AUC of 0.65, sensitivity of 0.29 and specificity of 0.83 (both macro averaged); when different sets of predictors were used, individuals with missing EHR coverage excluded, and subtypes combined, sensitivity was not considerably improved. CONCLUSIONS: ALSPAC and EHRs disagreed not just on AD subtypes, but also on whether children had AD or not. Researchers should be aware that individuals considered as having AD in one source may not be considered as having AD in another.

Non-linear associations of maternal pre-pregnancy body mass index with risk of stillbirth, infant, and neonatal mortality in over 28 million births in the USA: a retrospective cohort study.

Background: Higher maternal pre-pregnancy body mass index (BMI) has been associated with higher risk of stillbirth, infant and neonatal mortality. Studies exploring underweight have varied in their conclusions. Our aim was to examine the risk of stillbirth, infant and neonatal mortality across the BMI distribution and establish a likely healthy BMI range. Methods: In this retrospective cohort study, we used publicly available datasets (covering 1st January 2014 to 31st December 2021) from the US National Center for Health Statistics National Vital Statistics System. All births were eligible; analyses included those with non-missing data. Fractional polynomial multivariable logistic regression was used to examine the associations of maternal pre-pregnant BMI with stillbirth (birth with no signs of life at ≥24 weeks), infant mortality (death of a live born baby aged <365 days) and neonatal mortality (death of a live born baby aged <28 days). Findings: There were 77,896/28,310,154 (2.8 per 1000 births) stillbirths, 143,620/28,231,807 (5.1 per 1000 live births) infant deaths and 94,246/28,231,807 (3.3 per 1000 live births) neonatal deaths among complete cases. Mean (SD) BMI was 27.1 kg/m2 (6.7 kg/m2). We found non-linear associations between BMI and all three outcomes; risk was elevated at both low and high BMIs although, for stillbirth, the increased risk at low BMI was less marked than for infant/neonatal mortality. The lowest risk was at a BMI of 21 kg/m2 for infant and neonatal mortality and 19 kg/m2 for stillbirth. Interpretation: Public health messaging for preconception and postnatal care should focus on healthy weight to maximise maternal and child health, and not focus solely on maternal overweight or obesity. Funding: European Research Council, US National Institute of Health, UK Medical Research Council and National Institute for Health and Care Research.

Multiple imputation of missing data under missing at random: including a collider as an auxiliary variable in the imputation model can induce bias.

Epidemiological studies often have missing data, which are commonly handled by multiple imputation (MI). In MI, in addition to those required for the substantive analysis, imputation models often include other variables ("auxiliary variables"). Auxiliary variables that predict the partially observed variables can reduce the standard error (SE) of the MI estimator and, if they also predict the probability that data are missing, reduce bias due to data being missing not at random. However, guidance for choosing auxiliary variables is lacking. We examine the consequences of a poorly chosen auxiliary variable: if it shares a common cause with the partially observed variable and the probability that it is missing (i.e., it is a "collider"), its inclusion can induce bias in the MI estimator and may increase the SE. We quantify, both algebraically and by simulation, the magnitude of bias and SE when either the exposure or outcome is incomplete. When the substantive analysis outcome is partially observed, the bias can be substantial, relative to the magnitude of the exposure coefficient. In settings in which a complete records analysis is valid, the bias is smaller when the exposure is partially observed. However, bias can be larger if the outcome also causes missingness in the exposure. When using MI, it is important to examine, through a combination of data exploration and considering plausible casual diagrams and missingness mechanisms, whether potential auxiliary variables are colliders.

Multiple imputation of missing data under missing at random: compatible imputation models are not sufficient to avoid bias if they are mis-specified.

OBJECTIVES: Epidemiological studies often have missing data, which are commonly handled by multiple imputation (MI). Standard (default) MI procedures use simple linear covariate functions in the imputation model. We examine the bias that may be caused by acceptance of this default option and evaluate methods to identify problematic imputation models, providing practical guidance for researchers. STUDY DESIGN AND SETTING: Using simulation and real data analysis, we investigated how imputation model mis-specification affected MI performance, comparing results with complete records analysis (CRA). We considered scenarios in which imputation model mis-specification occurred because (i) the analysis model was mis-specified or (ii) the relationship between exposure and confounder was mis-specified. RESULTS: Mis-specification of the relationship between outcome and exposure, or between exposure and confounder, can cause biased CRA and MI estimates (in addition to any bias in the full-data estimate due to analysis model mis-specification). MI by predictive mean matching can mitigate model mis-specification. Methods for examining model mis-specification were effective in identifying mis-specified relationships. CONCLUSION: When using MI methods that assume data are MAR, compatibility between the analysis and imputation models is necessary, but not sufficient to avoid bias. We propose a step-by-step procedure for identifying and correcting mis-specification of imputation models.

Health and Well-Being in Surviving Congenital Heart Disease Patients: An Umbrella Review With Synthesis of Best Evidence.

Background: Advances in the management of congenital heart disease (CHD) patients have enabled improvement in long-term survival even for those with serious defects. Research priorities (for patients, families and clinicians) have shifted from a focus on how to improve survival to exploring long-term outcomes in patients with CHD. A comprehensive appraisal of available evidence could inform best practice to maximize health and well-being, and identify research gaps to direct further research toward patient and clinical need. We aimed to critically appraise all available published systematic reviews of health and well-being outcomes in adult patients with CHD. Methods: We conducted an umbrella review, including any systematic reviews that assessed the association of having vs. not having CHD with any long-term health (physical or mental), social (e.g., education, occupation) or well-being [e.g., quality of life (QoL)] outcome in adulthood (≥18-years). Results: Out of 1330 articles screened, we identified five systematic reviews of associations of CHD with adult outcomes. All but one (which studied QoL) explored health outcomes: one cardiovascular, two mental, and one mortality after transplant. CHD patients had a higher risk of stroke, coronary heart disease and heart failure, with the pooled relative risk (RR) for any outcome of 3.12 (95% CI: 3.01 to 3.24), with substantial heterogeneity (I2 = 99%) explained by the outcome being studied (stronger association for heart failure) and geography (stronger in Europe compared with other regions). CHD patients had a higher risk of anxiety (OR = 2.58 (1.45 to 4.59)], and higher mean scores for depression/anxiety symptoms (difference in means = -0.11 SD (-0.28 to 0.06), I2 = 94%)]. Compared with patients having a cardiac transplant for other (non-CHD) diseases, CHD patients had higher short-term mortality (RR at 30-days post-transplant = 2.18 [1.62 to 2.93)], with moderate heterogeneity (I2 = 41%) explained by previous surgery (higher mortality with prior Fontan/Glenn operation). All domains of QoL were lower in patients with Fontan's circulation than non-CHD adults. Conclusion: Adults with CHD have poorer cardiovascular, mental health and QoL outcomes, and higher short-term mortality after transplant. The paucity of systematic reviews, in particular for outcomes such as education, occupation and lifestyles, highlights the need for this to be made a priority by funders and researchers. Systematic Review Registration: [www.crd.york.ac.uk/prospero], identifier [CRD42020175034].

Effect of SARS-CoV-2 infection in neonates or in pregnancy on developmental outcomes at 21-24 months (SINEPOST): study protocol for a prospective cohort study.

INTRODUCTION: Exposure to SARS-CoV-2 during pregnancy or in the neonatal period may impact fetal or neonatal brain development either through direct central nervous system infection or indirectly through the adverse effects of viral infection-related inflammation in the mother or newborn infant. This study aims to determine whether there are early neurodevelopmental effects of SARS-CoV-2 infection. METHODS AND ANALYSIS: We will conduct a prospective national population-based cohort study of children aged 21-24 months who were born at term (≥37 weeks' gestation) between 1 March 2020 and 28 February 2021 and were either antenatally exposed, neonatally exposed or unexposed (comparison cohort) to SARS-CoV-2. Nationally, hospitals will identify and approach parents of children eligible for inclusion in the antenatally and neonatally exposed cohorts using information from the UK Obstetric Surveillance System (UKOSS) and British Paediatric Surveillance Unit (BPSU) national surveillance studies and will identify and approach eligible children for the comparison cohort through routine birth records. Parents will be asked to complete questionnaires to assess their child's development at 21-24 months of age. Outcome measures comprise the Ages and Stages Questionnaire, Third Edition (ASQ-3), Ages and Stages Questionnaire Social-Emotional, Second Edition (ASQ-SE-2), Liverpool respiratory symptoms questionnaire and questionnaire items to elicit information about healthcare usage. With parental consent, study data will be linked to routine health and education records for future follow-up. Regression models will compare ASQ-3 and ASQ-SE-2 scores and proportions, frequency of respiratory symptoms and healthcare usage between the exposed and comparison cohorts, adjusting for potential confounders. ETHICS AND DISSEMINATION: Ethics approval was obtained from the London-Westminster Research Ethics Committee. Findings will be disseminated in scientific conference presentations and peer-reviewed publications. ISRCTN REGISTRATION NUMBER: ISRCTN99910769.

Framework for the treatment and reporting of missing data in observational studies: The Treatment And Reporting of Missing data in Observational Studies framework.

Missing data are ubiquitous in medical research. Although there is increasing guidance on how to handle missing data, practice is changing slowly and misapprehensions abound, particularly in observational research. Importantly, the lack of transparency around methodological decisions is threatening the validity and reproducibility of modern research. We present a practical framework for handling and reporting the analysis of incomplete data in observational studies, which we illustrate using a case study from the Avon Longitudinal Study of Parents and Children. The framework consists of three steps: 1) Develop an analysis plan specifying the analysis model and how missing data are going to be addressed. An important consideration is whether a complete records' analysis is likely to be valid, whether multiple imputation or an alternative approach is likely to offer benefits and whether a sensitivity analysis regarding the missingness mechanism is required; 2) Examine the data, checking the methods outlined in the analysis plan are appropriate, and conduct the preplanned analysis; and 3) Report the results, including a description of the missing data, details on how the missing data were addressed, and the results from all analyses, interpreted in light of the missing data and the clinical relevance. This framework seeks to support researchers in thinking systematically about missing data and transparently reporting the potential effect on the study results, therefore increasing the confidence in and reproducibility of research findings.

Factors associated with participation over time in the Avon Longitudinal Study of Parents and Children: a study using linked education and primary care data.

BACKGROUND: In observational research, choosing an optimal analysis strategy when variables are incomplete requires an understanding of the factors associated with ongoing participation and non-response, but this cannot be fully examined with incomplete data. Linkage to external datasets provides additional information on those with incomplete data, allowing examination of factors related to missingness. METHODS: We examined the association between baseline sociodemographic factors and ongoing participation in the Avon Longitudinal Study of Parents and Children. We investigated whether child and adolescent outcomes measured in linked education and primary care data were associated with participation, after accounting for baseline factors. To demonstrate the potential for bias, we examined whether the association between maternal smoking and these outcomes differed in the subsample who completed the 19-year questionnaire. RESULTS: Lower levels of school attainment, lower general practitioner (GP) consultation and prescription rates, higher body mass index (BMI), special educational needs (SEN) status, not having an asthma diagnosis, depression and being a smoker were associated with lower participation after adjustment for baseline factors. For example, the adjusted odds ratio (OR) for participation comparing ever smokers (by 18 years) with non-smokers was: 0.65, 95% CI (0.56, 0.75). The associations with maternal smoking differed between the subsample of participants at 19 years and the entire sample, although differences were small and confidence intervals overlapped. For example: for SEN status, OR = 1.19 (1.06, 1.33) (all participants); OR = 1.03 (0.79, 1.45) (subsample). CONCLUSIONS: A range of health-related and educational factors are associated with ongoing participation in ALSPAC; this is likely to be the case in other cohort studies. Researchers need to be aware of this when planning their analysis. Cohort studies can use linkage to routine data to explore predictors of ongoing participation and conduct sensitivity analyses to assess potential bias.

The potential for linking cohort participants to official criminal records: a pilot study using the Avon Longitudinal Study of Parents and Children (ALSPAC).

Introduction: Linking longitudinal cohort resources with police-recorded records of criminal activity has the potential to inform public health style approaches and may reduce potential sources of bias from self-reported criminal data collected by cohort studies. A pilot linkage to police records in the Avon Longitudinal Study of Parents and Children (ALSPAC) allows us to consider the acceptability of this linkage, its utility as a data resource, differences in self-reported crime according to consent status for data linkage, and the appropriate governance mechanism to support such a linkage. Methods: We carried out a pilot study that linked data from the ALSPAC birth cohort to Ministry of Justice (MoJ) records on criminal cautions and convictions. This pilot was conducted on a fully anonymous basis, meaning we cannot link the identified records to any participant or the wider information within the dataset. Using ALSPAC data, we used summary statistics to investigate differences in self-reported criminal activity according to socio-economic background and consent status. We used MoJ records to identify the geographic and temporal concentration of criminality in the ALSPAC cohort. Results: We found that the linkage appears acceptable to participants (4% of the sample opted out), levels of criminality are high enough to support research and that the majority of crimes occurred in Avon & Somerset (the policing area local to ALSPAC). Both those who opted out of linkage or did not respond to consent requests had higher levels of self-reported criminal behaviour compared to participants who provided explicit consent. Conclusions: These findings suggest that data linkage in ALSPAC provides opportunities to study criminal behaviour and that linked individual-level records can provide robust research in the area. Our findings also suggest the potential for bias when only using samples that have explicitly consented to data linkage, highlighting the limitations of opt-in consent strategies.

Is screen time associated with anxiety or depression in young people? Results from a UK birth cohort.

BACKGROUND: There is limited and conflicting evidence for associations between use of screen-based technology and anxiety and depression in young people. We examined associations between screen time measured at 16 years and anxiety and depression at 18. METHODS: Participants (n = 14,665; complete cases n = 1869) were from the Avon Longitudinal Study of Parents and Children, a UK-based prospective cohort study. We assessed associations between various types of screen time (watching television, using a computer, and texting, all measured via questionnaire at 16y), both on weekdays and at weekends, and anxiety and depression (measured via the Revised Clinical Interview Schedule at 18y). Using ordinal logistic regression, we adjusted for multiple confounders, particularly focussing on activities that might have been replaced by screen time (for example exercising or playing outdoors). RESULTS: More time spent using a computer on weekdays was associated with a small increased risk of anxiety (OR for 1-2 h = 1.17, 95% CI: 1.01 to 1.35; OR for 3+ hours = 1.30, 95% CI: 1.10 to 1.55, both compared to < 1 h, p for linear trend = 0.003). We found a similar association between computer use at weekends and anxiety (OR for 1-2 h = 1.17, 95% CI: 0.94 to 1.46; OR for 3+ hours = 1.28, 95% CI: 1.03 to 1.48, p for linear trend = 0.03). Greater time spent using a computer on weekend days only was associated with a small increased risk in depression (OR for 1-2 h = 1.12, 95% CI: 0.93 to 1.35; OR for 3+ hours = 1.35, 95% CI: 1.10 to 1.65, p for linear trend = 0.003). Adjusting for time spent alone attenuated effects for anxiety but not depression. There was little evidence for associations with texting or watching television. CONCLUSIONS: We found associations between increased screen time, particularly computer use, and a small increased risk of anxiety and depression. Time spent alone was found to attenuate some associations, and further research should explore this.

Defining adolescent common mental disorders using electronic primary care data: a comparison with outcomes measured using the CIS-R.

OBJECTIVE: To compare the prevalence of common mental disorders (CMDs) derived from data held in primary care records with that measured using the revised Clinical Interview Schedule (CIS-R) in order to assess the potential robustness of findings based only on routinely collected data. DESIGN AND SETTING: Comparison study using linkage between the Avon Longitudinal Study of Parents and Children (ALSPAC) and electronic primary care records. PARTICIPANTS: We studied 1562 adolescents who had completed the CIS-R in ALSPAC at age 17-18 years and had linkage established to their primary care records. OUTCOME MEASURES: Outcome measures from ALSPAC were whether or not an individual met International Classification of Diseases-10 criteria for a diagnosis of (1) a CMD or, specifically, (2) depression. Lists of Read codes corresponding to diagnoses, symptoms and treatments were used to create 12 definitions of CMD and depression alone using the primary care data. We calculated sensitivities and specificities of these, using CIS-R definitions as the reference standard. RESULTS: Sensitivities ranged from 5.2% to 24.3% for depression and from 3.8% to 19.2% for CMD. The specificities of all definitions were above 98% for depression and above 96% for CMD.For both outcomes, the definition that included current diagnosis, treatment or symptoms identified the highest proportion of CIS-R cases. CONCLUSIONS: Most individuals meeting case definitions for CMD based on primary care data also met CIS-R case definitions. Conversely many individuals identified as cases using the CIS-R had no evidence of CMD in their clinical records. This suggests that clinical databases are likely to yield underestimates of the burden of CMD in the population. However, clinical records appear to yield valid diagnoses which may be useful for studying risk factors and consequences of CMD. The greatest epidemiological value may be obtained when information is available from survey and clinical records.

Using linked educational attainment data to reduce bias due to missing outcome data in estimates of the association between the duration of breastfeeding and IQ at 15 years.

BACKGROUND: Most epidemiological studies have missing information, leading to reduced power and potential bias. Estimates of exposure-outcome associations will generally be biased if the outcome variable is missing not at random (MNAR). Linkage to administrative data containing a proxy for the missing study outcome allows assessment of whether this outcome is MNAR and the evaluation of bias. We examined this in relation to the association between infant breastfeeding and IQ at 15 years, where a proxy for IQ was available through linkage to school attainment data. METHODS: Subjects were those who enrolled in the Avon Longitudinal Study of Parents and Children in 1990-91 (n = 13 795), of whom 5023 had IQ measured at age 15. For those with missing IQ, 7030 (79%) had information on educational attainment at age 16 obtained through linkage to the National Pupil Database. The association between duration of breastfeeding and IQ was estimated using a complete case analysis, multiple imputation and inverse probability-of-missingness weighting; these estimates were then compared with those derived from analyses informed by the linkage. RESULTS: IQ at 15 was MNAR-individuals with higher attainment were less likely to have missing IQ data, even after adjusting for socio-demographic factors. All the approaches underestimated the association between breastfeeding and IQ compared with analyses informed by linkage. CONCLUSIONS: Linkage to administrative data containing a proxy for the outcome variable allows the MNAR assumption to be tested and more efficient analyses to be performed. Under certain circumstances, this may produce unbiased results.

Socio-economic position and childhood multimorbidity: a study using linkage between the Avon Longitudinal Study of Parents and Children and the General Practice Research Database.

INTRODUCTION: In adults, multimorbidity is associated with social position. Socially disadvantaged adults typically experience more chronic illness at a younger age than comparable individuals who are more advantaged. The relation between social position and multimorbidity amongst children and adolescents has not been as widely studied and is less clear. METHODS: The NHS Information Centre (NHS IC) linked participants in the Avon Longitudinal Study of Parents and Children (ALSPAC) to the General Practice Research Database (GPRD). Multimorbidity was measured in three different ways: using a count of the number of drugs prescribed, a count of chronic diseases, and a person's predicted resource use score; the latter two measures were derived using the Johns Hopkins ACG system. A number of different socio-economic position variables measured as part of ALSPAC during pregnancy and early childhood were considered. Ordered logistic and negative binomial regression models were used to investigate associations between socio-economic variables and multimorbidity. RESULTS: After mutually adjusting for the different markers of socio-economic position, there was evidence, albeit weak, that chronic condition counts among children aged from 0 to 9 years were higher among those whose mothers were less well educated (OR = 0.44; 95% confidence interval 0.18-1.10; p = 0.08). Conversely, children whose mothers were better educated had higher rates of chronic illness between 10 and 18 years (OR = 1.94; 95% CI 1.14-3.30). However, living in a more deprived area, as indicated by the Townsend score, was associated with a higher odds of chronic illness between 10 and 18 years (OR for each increasing decile of Townsend score = 1.09; 95% CI 1.00-1.19; p = 0.06). CONCLUSIONS: We have found some evidence that, in younger children, multimorbidity may be higher amongst children whose parents are less well educated. In older children and adolescents this association is less clear. We have also demonstrated that linkage between prospective observational studies and electronic patient records can provide an effective way of obtaining objectively measured outcome variables.