SSHB logosshb-logo-white

Our Blog

18th Apr 2019

The Children with HIV early antiretroviral trial (CHER) was aimed at investigating ART initiation strategies in infants and children in a resource limited setting in South Africa. The resulting cohort of infants followed up for a 36 months provided data on growth in anthropometric measurements, including head circumference that serves as a proxy measurement for brain size, especially before the age of two which is the period when almost all brain growth occurs. The aim of this analysis was to look at the impact of HIV infection and ART treatment on head circumference measurements, as a proxy for brain development, to assess the appropriateness of the WHO reference ranges for our cohort of healthy children, and to derive South African specific reference ranges using this cohort.

We adopted a broken-stick growth model for the head circumference profiles by including piecewise linear splines in a mixed effect model for the impact of the ART strategies on head circumference profiles over age. The LMS method was used to calculate z-scores using the WHO references ranges and the groups were compared with reference to these z-scores. Cohort specific percentiles and z-scores were calculated based on the Box-Cox Power distribution.

The model illustrated the positive effect of ART treatment with children on early ART catching up faster to the HIV- children than children on delayed ART. The latter group started catching up as the children start using ART at later ages.

Head circumference measurements for children included in the HIV-uninfected control groups were larger than expected based on WHO standards. Cohort specific derived reference ranges thus illustrated an upward adjustment of the percentile bands.


Introduction: Lower childhood cognitive ability, commonly measured by the Intelligence Quotient (IQ), is associated with worse later life social, economic, and health outcomes. This study aimed to assess inequalities in childhood IQ trajectories in a large sample of Guatemalan school children, according to birth year, school, and height-for-age Z-scores (HAZ).
Methods: A multilevel model was developed to describe 57 244 IQ observations (level 1) in 22 646 children (level 2), aged 6-15 years born between 1955 and 1993, in five schools (level 3) from Guatemala City, Guatemala. Briefly, IQ trajectories were modelled as a quadratic polynomial age function (with random effects at level 2 & 3), birth year was added as a linear term and interacted with age, and HAZ was included as a two-term linear spline (with a random effect at level 3 for the first term).
Results: IQ trajectories differed across schools across ages, the inequality in average IQ at age 11 between the students of a high fee private institution and a no-fee public school was 28.5 points. Later born individuals had higher IQ compared to earlier born individuals (B=0.16, 95% CI 0.15 – 0.18), although this inequality narrowed in adolescence. A higher HAZ score was associated with higher IQ only in individuals with HAZ scores below 0 (1.4, 0.8 – 2.05), this effect was stronger in public compared to private school students.
Conclusions: We found large inequalities in the IQ of Guatemalan children. Lower HAZ was associated with lower IQ more strongly in children who attended disadvantaged schools, possibly reflecting the damaging effects of poor early life environments both for linear growth and cognitive development.


17th Apr 2019

Introduction: Gestational diabetes mellitus (GDM) management consists of advice on diet and exercise, followed by pharmacotherapy if hyperglycaemia persists. Previous studies have highlighted maternal differences between treatment groups however only a few of them involved metformin-treated patients, in a UK clinical setting. This UK-based study aimed to identify relationships between maternal characteristics and GDM treatment modalities. Methods: Maternal records from Born in Bradford cohort participants receiving treatment for GDM during their singleton pregnancies were studied (N=727). Treatment groups consisted of lifestyle modifications (diet and/or exercise), pharmacotherapy (insulin and/or metformin) and combined treatment (lifestyle modifications and pharmacotherapy). Differences between groups were evaluated using Pearson’s χ2 and Fisher’s exact tests for categorical variables and Kruskal-Wallis test for continuous variables. Multinomial logistic regression examined maternal predictors of GDM treatment. Results: Mothers receiving lifestyle modifications (N=196) and combined therapy (N=209) were younger than mothers receiving pharmacotherapy (N=322). 57.4% of women treated with pharmacotherapy were obese compared to 17.7% and 24.9% of women in lifestyle modifications and combined therapy groups, respectively. Pakistani women were less likely to be treated with pharmacotherapy than lifestyle modifications (RRR 0.6(0.3-1.3)). Higher fasting glucose levels at diagnosis increased the risk of combined treatment compared to lifestyle modifications (RRR 1.9(1.3-2.6)). Conclusions: Being older and having a less healthy clinical profile increased the risk of treatment involving insulin and/or metformin for women with GDM.


15th Apr 2019

Introduction: The Saskatchewan Pediatric Bone Mineral Accrual Study (PBMAS) was initiated in 1991, when 248 boys and girls, aged 8-15 years, were recruited. Using serial measures (1991-1997, 2002-2007, 2009-2011 and 2016-2017) the study aimed to identify the development of body composition and its relationship to risk of adult disease. Previously, a relationship between adolescent trunk fat mass (TFM) and cardiometabolic risk at 26 years was identified. The present study examines developmental trajectories of TFM, during both adolescence and emerging adulthood (EA), of individuals categorized as either low or high metabolic risk (MRS) at 36 years of age. Methods: Fifty-five individuals were assessed from adolescence (11.5 ± 1.8 years), through EA (26.2 ± 2.2) into young adulthood (35.6 ± 2.2 years) (median of 11 visits) for anthropometrics, blood pressure, blood metabolites, DXA, diet and physical activity. MRS groups were created using sex-specific median splits of continuous standardized risk scores (blood pressure and serum markers) at 36 years of age. TFM trajectories were analyzed using multilevel random effects models. Results: The high MRS group had significantly steeper trajectories of TFM development in both adolescence and EA, 0.65±0.11 and 0.44±0.11 log g (p0.05) to TFM accrual, however physical activity was (−0.04±0.02; p<0.05). Conclusions: Young adults with high cardiometabolic risk at 36 years of age had greater trunk fat mass accrual during both adolescence and emerging adulthood. These results support the need for intervention at both these critical periods of fat accrual.


12th Apr 2019

Data linkage is an important tool for increasing the value of existing longitudinal studies. Linkage to multiple data sources can provide a low-cost, efficient means of collecting extensive and detailed data on cross-sectoral services, society, and the environment, as well as augmenting direct data collection through linkage with biological samples, social media and other digital sources. These data can be used to supplement traditional cohort studies, or to create population-level electronic cohorts generated from administrative data. Such administrative data cohorts offer the ability to answer questions that require large sample sizes or detailed data on heard-to-reach populations, and to generate evidence with a high level of external validity and applicability for policy-making. There is increasing interest in using these two models of data collection in conjunction, combining population-level administrative data with detailed attribute data collected directly from participants, in order to provide a deeper insight into what determines our health.

Lack of access to unique or accurate identifiers means that linkage is not always straightforward. Errors occurring during linkage (false-matches and missed-matches) can lead to substantial bias in results based on linked data. This issue is compounded by difficulties in evaluating linkage quality or determining the potential impact of errors on results due to the separation of linkage from analyses of inked data. This talk will give an overview of the opportunities, challenges and methods for using data linkage in cohort studies.


02nd Apr 2019

Introduction
Diverse tooth development staging techniques use a prediction of the mature tooth root length as a reference to classify the observed tooth development. Proportions of the predicted mature root length are used as tresholds between the root stages. Longitudinal data permit to collect information of a specific tooth position while in development and while mature. This information allows to establish the exact tooth development stage thresholds for the considered subjects and can be used to validate observers’ stage allocation performances.
Methods
Longitudinal tooth development data were extracted from 119 series of retrospectively collected digital dental panoramic radiographs. Each series included at least two radiographs from the same subject registered at different moments. The youngest radiograph contained mature and the older maturing second molars. All second molars were staged by six observers according to the technique of Köhler et al.. The ratio between the second molar root length measured in the last and each previously recorded radiograph was calculated for each subject. The calculated range of second molar root length ratios that confirmed correct Köhler staging were as follows: range=0.25 to <0.50, Stage 5; range=0.5 to <0.75, Stage 6; range=0.75 to <1, Stage 7; ratio=1 (i.e., no range), Stage 8-10. The registered Köhler stages and the calculated ratios were independently verified for each second molar position and for each of the six observers.
Results
Verification of the calculated ratios and registered Köhler stages revealed that all observers generally classified the developing tooth in a more advanced stage than the correct stage, except for Stage 5.
Conclusion
Significant discrepa


01st Apr 2019

Despite the ubiquity of the modern ‘obesogenic’ environment, we have not uniformly developed obesity. On the contrary, there is considerable variation in weight gain, which is observable from early infancy. In fact, it is not uncommon for siblings to vary considerably in their weight, even when they live in the same household. Obesity risk is about far more than just the environment we are exposed to. Genetic susceptibility to the environment is thought to explain some of the variation in adiposity. Nearly 100 twin studies have established that weight is a highly heritable trait (50-90%), and ~1000 common genetic variants (single nucleotide polymorphisms) have been discovered. Individual differences in appetite have been implicated as one of the mechanisms through which genes influence adiposity, so-called ‘Behavioural Susceptibility Theory’ (BST). BST hypothesises that individuals who inherit a set of genes that confer greater responsiveness to food cues (wanting to eat in response to the sight, smell or taste of palatable food), and lower sensitivity to satiety (fullness), are more vulnerable to overeating in response to the modern food environment, and therefore to developing obesity. At the same time, our environment is not simply an ‘exposure’ – we select and shape our environment to suit our preferences, many of which have some genetic basis (called gene-environment correlation). This talk summarises the interplay between genes, appetite and the home family environment in early weight gain, using data from Gemini and TEDS – two large, population-based British twin birth cohorts set up to study genetic and environmental influence on health and development.



06th Feb 2019

There has been interest in finding separate effects of age, year and birth cohort for decades, in both the biological and social sciences. However, the exact collinearity between these three (Age = Year – Birth Year) lead to difficulty in estimating these effects. Because of this, it is impossible to estimate near-linear effects (or linear components of these effects) without making strong assumptions about at least one of these. This is a problem for anyone interested in any of age, period and/or cohort patterns in a particular outcome.

There have been many attempts to ‘solve’ this identification problem without having to make strong assumptions – however in each case, it turns out such models are, in fact, making hidden assumptions that were not intended by the user, as I show with simulations. I then consider what researchers should do, drawing on literature from across the social, biological and health sciences. This includes consideration of non-linearities around linear APC effects (with both statistical and graphical techniques), strong and explicit assumptions based on theory (for example assuming there are no linear period effects), including constraints on certain parameters to estimate ranges within which other parameters must fall. I provide an example focusing on mortality in the twentieth century. In each case, these methods acknowledge that there is a ‘line of solutions’ of possible combinations of APC effects, and not a single answer that can be estimated empirically. None of these methods represent a solution to the identification problem – rather they are an honest acknowledgement of the problem, with an awareness that the methods are limited by their assumptions.



We use cookies to improve your website experience. To learn about our use of cookies and how you can manage your cookie settings, please see our Cookie Policy. By closing this message, you are consenting to our use of cookies.