Epidemiology 4
Presenter: Heiko Becher
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Modelling semicontinuous exposures: Simulation study and practical application to survival data
Risk factors in epidemiology are often semicontinuous such that a proportion of individuals have exposure zero, and a continuous distribution among those exposed. We call this a spike at zero (SAZ). Typical examples in cancer epidemiology are consumption of alcohol and tobacco, breastfeeding duration or others. To model non-linear functional relationships for continuous variables, the fractional polynomial procedure (FP)has been shown to be useful. For SAZ variables, an extension of the FP approach was proposed. To indicate whether or not a value is zero, a binary variable is added to the model. In a two-stage procedure, called FP-spike, it is assessed whether the binary variable and/or the continuous FP function for the positive part is required for a suitable fit. While the good performance of the procedure has been shown in linear or logistic regression, the applicability for survival data has not been investigated. In this presentation, we compared the performance of two approaches standard FP and FP-spike in the Cox model. We present the procedure and use survival data from a breast cancer study as an example. The performance will be demonstrated with a simulation study. We consiter a covariable X with a normal and a log-normal distribution and zero fractions ranging from 0.1 to 0.5. We assume an exponential distribution of the survival time with constant baseline hazard of ?0 = 0.05 resulting in a mean survival time of 1/0.05=20 for X=0. 1000 observations were generated per scenario. The constant hazard of censoring times ?c was chosen to achieve censoring rates of approximately 10, 30 and 50 percent. We show that the FP-spike procedure performes better in realistic settings. However, it is difficult to provide general guidelines since the performance depends simultaneously on effect size, fraction of zeros and distribution of the continuous part of the risk factor.
Epidemiology 4
Presenter: Kishore Das
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Study of the Occurrence of Seasonal Diseases: A Circular Statistical Approach
In the present study, we attempt to analyse the occurrence of seasonal diseases, both season-wise (unequal length of months) and month-wise with the aid of circular statistical tools. That the season-wise (unequal length of months) analysis has not been attempted before using circular statistical tools is the main motivation behind this study. The data has been taken from the project entitled "Statistical Modeling in Circular Statistics: An Application to Health Science" sponsored by the UGC, India. The study area have been chosen the Kamrup (rural) district of Assam, India. It is revealed that the occurrence of seasonal diseases is highest in the months of March or equivalently, during the Pre-monsoon season. The distribution of occurrence of seasonal diseases both month-wise and season-wise is found to be marginally positively skewed and platykurtic. The presence of seasonal effect is verified by the Rayleigh Uniformity test in both the cases. This study will aid the health officials to understand the scenario of month-wise and season-wise occurrence of seasonal diseases in the concerned part of Assam state so that adequate steps may be taken to curb them.
Epidemiology 4
Presenter: John Ferguson
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Extending Average Attributable Fractions
In epidemiology, the attributable fraction represents the proportional reduction in population disease prevalence that might be observed if a particular risk factor could be eliminated from the population. In some regards, it is a more relevant measure of disease association than odds ratios or relative risks as it hints at the potential impact of an intervention targeting the risk factor. Average attributable fractions1 are a related concept, more tailored to the situation where several risk factors are known to be associated with the disease, in which case they define a partition of cumulative disease burden into contributions from each risk factor. At first sight, average attributable fractions seem an extremely useful tool to quantify the portion of risk contributed by each risk factor; a pertinent calculation in describing chronic disease epidemiology. However, they are seldom used by practitioners in epidemiology. Perhaps the main reason for this may be the technical issues that researchers face in their application. In brief, some of these hurdles relate to: (a) computational difficulties when the number of risk factors is large (b) no proposed method to produce confidence intervals (c) a lack of flexible software to assist in their calculation. In this presentation, we describe these issues in more depth, and propose some solutions. We have developed an R-package, averisk, to implement our methods which can be downloaded from the CRAN repository. References Eide, Geir Egil, and Olaf Gefeller. "Sequential and Average Attributable Fractions as Aids in the Selection of Preventive Strategies." Journal of clinical epidemiology 48, no. 5 (1995): 645-55.
Epidemiology 4
Presenter: Biljana Jonoska Stojkova
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Estimating parameters of epidemiological model via Simulated Tempering Without Normalizing Constants
We consider a Susceptible-Infected-Recovered epidemiological model where disease spread dynamics is described by ordinary differential equations model. The mixture of discrete and continuous parameters in the model induces multi-modality in the likelihood surface, which makes sampling from the posterior distribution challenging. Standard MCMC methods fail to explore efficiently the multi-modal posterior surface because they could get easily trapped in some of the local modes. Simulated tempering algorithm is developed to sample efficiently from multi-modal distributions by augmenting the sampling space with an auxiliary temperature parameter. In order to ensure sampling of the temperature parameter, Simulated Tempering requires computation of the normalizing constants which contributes in large to its unpopularity in practice. The proposed Simulated Tempering algorithm does not require normalizing constants.
Epidemiology 4
Presenter: Theresa Keller
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Is There A Puberty-related Sex-switch Of Allergic Rhinitis Prevalence? Pooled Analyses Of Longitudinal Birth Cohorts.
Background/Aims. Allergic rhinitis is one of the most common chronic diseases. Cross sectional studies indicate that boys might suffer more often from allergic rhinitis than girls during childhood, while this pattern seems to change in adolescence towards a female predominance. Objective. To examine possible longitudinal sex-specific prevalence changes before and after puberty in allergic rhinitis in participants from existing European birth cohort studies. Methods. We included six population-based birth cohorts (PIAMA, BAMSE, LISAplus, GINIplus, DARC and MAS) from the EU Project MeDALL. Harmonized prospectively collected validated self/parent reported questions were used to assess puberty (Pubertal Development Scale PDS) and the clinical outcome (allergic rhinitis). For each birth cohort separately we used generalized estimating equations (GEE) to analyse the effects of age, sex, puberty, and possible confounders on the prevalence of the outcome, with the focus on the interaction of puberty (yes/no) and sex as an indicator of sex-specific changes in allergic rhinitis prevalence in participants before versus those in or after puberty. We performed a pooled analysis of individual participant data of the six birth cohorts. As a further sensitivity analysis, random-effect meta-analyses with the inverse-variance method were performed. I2 was calculated to quantify inconsistency across studies. Results. Data from 18,852 children (birth to age 20) were included. The interaction term sex*puberty indicated a sex-specific change in allergic rhinitis prevalence after the onset of puberty. Allergic rhinitis occurred less often in girls compared to boys both before and after the onset of puberty. However pooled analysis showed, the sex specific difference (female vs. male) decreased after puberty onset from adjusted Odds Ratio (OR) (0.60, 95%CI 0.52-0.68, up to 0.77, 95%CI 0.68-0.86 (p-value for interaction sex*puberty 0.001). In sensitivity analyses, the meta-analyses showed similar results: OR before onset of puberty 0.58, 95%CI 0.48-0.72 (I2 = 57%), OR after onset of puberty 0.75, 95%CI 0.67-0.84 (I2 = 3%). Conclusions. Combining data from six European birth cohorts, we found a decrease in the male predominance in allergic rhinitis prevalence after puberty onset.
Epidemiology 4
Presenter: Iris Pigeot
When: Thursday, July 14, 2016 Time: 2:30 PM - 4:00 PM
Room: Salon A Carson Hall (Level 2)
Session Synopsis:
Associations between early body mass index trajectories and later metabolic risk in European children: The IDEFICS study
Background: Adiposity is a known risk factor for several health outcomes. However, yet it is not clear whether it is mainly the actual weight status or (in addition) the trajectory of adiposity, especially changes throughout early life, that affects health. In this study, a 2-step procedure was developed to investigate the association between body mass index (BMI) trajectories during infancy/childhood and metabolic risk at a later age in order to identify periods of growth related to later metabolic health (Börnhorst et al. 2015). Methods: In a first step, BMI trajectories of 3,301 European children that participated in the multi-centre IDEFICS study were modelled using linear-spline mixed-effects models (step 1). The resulting random coefficients indicating initial subject-specific BMI and rates of change in BMI over time were used as exposure variables in the second step to assess associations between BMI growth and a metabolic risk score as well as with its single components measured in later childhood (mean age at outcome assessment: 8.5 yrs). The metabolic risk score was derived applying a z-score standardisation based on recently published reference values for young children to combine blood pressure, dyslipidaemia, waist circumference and hyperglycaemia into one continuous variable. Results: All exposures under investigation, i.e. BMI at birth, rates of BMI change during infancy (0 to <9 mths), early childhood (9 mths to <6 yrs) and later childhood (?6 yrs) as well as current BMI z-score were significantly associated with the later metabolic risk score. Associations were strongest for the rate of BMI change in early childhood (?=1.72; p<.0001) and for current BMI z-score (?=1.28; p<.0001) and less pronounced for BMI at birth (?=0.61; p<.0001). Results slightly differed with regard to the single metabolic risk factors. Conclusion: Rapid growth, especially in the time window of 9 mths to <6 yrs, is significantly related to later metabolic risk. Large parts of the associations of early BMI growth may be mediated through its effect on later BMI growth and current BMI. References: Börnhorst C et al. (2015). Associations between early body mass index trajectories and later metabolic risk in European children: The IDEFICS study. Eur J Epidemiol. 2015 Aug 22. [Epub ahead of print]
