Madhuri Sudan

Epidemiology of childhood leukemia in the presence and absence of Down syndrome

Down syndrome (DS) is a common congenital anomaly, and children with DS have a substantially higher risk of leukemia. Although understanding of genetic and epigenetic changes of childhood leukemia has improved, the causes of childhood leukemia and the potential role of environmental exposures in leukemogenesis remain largely unknown. Although many epidemiologic studies have examined a variety of environmental exposures, ionizing radiation remains the only generally accepted environmental risk factor for childhood leukemia. Among suspected risk factors, infections, exposure to pesticides, and extremely low frequency magnetic fields are notable. While there are well-defined differences between leukemia in children with and without DS, studies of risk factors for leukemia among DS children are generally consistent with trends seen among non-DS (NDS) children. We provide background on DS epidemiology and review the similarities and differences in biological and epidemiologic features of leukemia in children with and without DS. We propose that both acute lymphoblastic and acute myeloblastic leukemia among DS children can serve as an informative model for development of childhood leukemia. Further, the high rates of leukemia among DS children make it possible to study this disease using a cohort approach, a powerful method that is unfeasible in the general population due to the rarity of childhood leukemia.

Cell phone exposures and hearing loss in children in the Danish National Birth Cohort

BACKGROUND Children today are exposed to cell phones early in life, and may be the most vulnerable if exposure is harmful to health. We investigated the association between cell phone use and hearing loss in children. METHODS The Danish National Birth Cohort (DNBC) enrolled pregnant women between 1996 and 2002. Detailed interviews were conducted during gestation, and when the children were 6 months, 18 months and 7 years of age. We used multivariable-adjusted logistic regression, marginal structural models (MSM) with inverse-probability weighting, and doubly robust estimation (DRE) to relate hearing loss at age 18 months to cell phone use at age 7 years, and to investigate cell phone use reported at age 7 in relation to hearing loss at age 7. RESULTS Our analyses included data from 52 680 children. We observed weak associations between cell phone use and hearing loss at age 7, with odds ratios and 95% confidence intervals from the traditional logistic regression, MSM and DRE models being 1.21 [95% confidence interval [CI] 0.99, 1.46], 1.23 [95% CI 1.01, 1.49] and 1.22 [95% CI 1.00, 1.49], respectively. CONCLUSIONS Our findings could have been affected by various biases and are not sufficient to conclude that cell phone exposures have an effect on hearing. This is the first large-scale epidemiologic study to investigate this potentially important association among children, and replication of these findings is needed.

Marginal structural models, doubly robust estimation, and bias analysis in perinatal and paediatric epidemiology.

We thank Ahrens and Schisterman (henceforth, A&S) for their commentary1 on our article.2 Although it was not our original intention, we are grateful for the invited discussion on the place of causal inference in perinatal and pediatric epidemiology. In response, we briefly offer some clarifications and extensions. A&S claim that we did not adjust for reduced hearing at age 18 months (Y1) in our analysis of the impact of postnatal cellphone exposure (X2) on hearing loss at age 7 years (Y2). In fact, we adjusted for Y1 and other variables listed in the footnotes of Tables 3 and 4 in our article.2 In applying directed acyclic graphs (DAGs),3,4 A&S rightly caution against grouping variables into one node (e.g., B) in the DAG (Figure below or in our article2). This grouping would imply that all arrows pointing into and out of B apply to every variable in B. We used the grouping to avoid clutter and were always mindful of it. A&S raised the issue of a variable in a grouping also being a collider. This is applicable to every collider on a backdoor path and which is selected for confounding control (e.g., A is simultaneously a confounder and a collider with respect to X2→Y2 in the DAG below). To eliminate the collider bias introduced by conditioning on A for confounding control or on X2 (which by being a consequence of the collider A induces conditioning on A),4 we need to have measured variable(s) that can be used to close the open bidirected path between A and Y2 or A and X2. This issue leads us to an important but overlooked result that should be part of the existing causal assumptions: there should be no uncontrolled collider bias before or following confounding control. That is, one must control for any collider bias that arises from using a collider on an open backdoor path to close that backdoor or when the exposure under study is caused by a collider that also lies on an open backdoor path. Figure Directed acyclic graph modified from Sudan et al2 to incorporate uncontrolled confounding between X2 and Y2, unmeasured common causes of A and X2, and of A and Y2, and non-differential independent misclassification of Y2. We agree with A&S on the need for multiple bias modeling.5 We expect to see more of it in the literature as probabilistic bias analysis is increasingly accepted by journals, large data become more available, and investigators routinely use bias formulas6,7 and simulation techniques. We disagree with A&S that bias analysis must be preceded by “placement of the unmeasured confounder in the DAG”1 and that such placement can reveal when “the potential bias is no longer a concern” 1. A known but unmeasured variable should be part of the working DAG from the outset, and not left out until bias analysis. Adding a dashed bidirected arc between the exposure X2 and outcome Y2 in our DAG at the bias analysis stage implies the suspicion of, at least, an unmeasured, possibly unknown, common cause of X2 and outcome Y2. In our article, we triangulated our effect estimate using conventional logistic regression, inverse-probability-weighted (IPW) fitting of marginal structure models (MSM) and doubly robust estimation despite the differences in the qualitative meaning of their effect estimates. This is useful because finding conflicting quantitative results such as reversed effect directions can send warning signals. Importantly, in our article, the different estimates were in the same direction and of similar magnitude. A&S claim that this similarity in magnitude was simply due to minimal confounding in our study. Using hypothetical data with more confounding, we show this claim to be incorrect (see the first three models in the second column of the Table below). A&S then claim that we did not specify how we implemented DRE. Please see the text and Tables 3 and 4 of our article.2 Table Odds ratios (95% confidence limits) for the effect of X2 on Y2 obtained from conventional outcome regression, inverse probability weighted fitting of marginal structural models, doubly robust estimation, and union models using hypothetical data generated ... Given journal space limitations, commentaries can sometimes confusingly oversimplify complex issues. First, AS the others are IPW fitting of MSM, and g-estimation of structural nested models.8,9 Risking oversimplification, we conclude that causal analysis involves estimating well-defined causal effects using (i) (possibly untestable, qualitative) causal assumptions (e.g., no uncontrolled confounding), and (ii) appropriate statistical estimation techniques (e.g. doubly or multiply robust estimation) to remove existing bias without introducing new bias (e.g. handling time-varying confounding, mediation or effect modification without introducing collider bias).3,4,8–10 Even the most sophisticated estimation technique, sans causal assumptions, cannot endow an estimate with causal meaning. Conversely, the simplest conventional regression model coupled with appropriate causal assumptions can be used for causal effect estimation. For details, we defer to our and A&S’s references. Clear and defined research questions guided our analysis and presentation of the results of the difficult, yet important, pursuit of the role of environmental exposures in the health of children.

Pilot test of a peer-led small-group video intervention to promote mammography screening among Chinese American immigrants

This study evaluated the feasibility, acceptability, and potential effect of a small-group video intervention led by trained Chinese American lay educators who recruited Chinese American women not up to date on mammography screening. Nine lay educators conducted 14 Breast Health Tea Time Workshops in community settings and private homes that started with watching a culturally tailored video promoting screening followed by a question-and-answer session and distribution of print materials. Many group attendees did not have health insurance or a regular doctor, had low levels of income, and were not proficient in English. Forty-four percent of the attendees reported receipt of a mammogram within 6 months after the small-group session, with higher odds of screening among women who had lived in the United States less than 10% of their lifetime. Four of the educators were very interested in conducting another group session in the next 6 months.

Prospective cohort analysis of cellphone use and emotional and behavioural difficulties in children

Background We previously reported associations between cellphone exposure and emotional and behavioural difficulties in children in the Danish National Birth Cohort using cross-sectional data. To overcome the limitations of cross-sectional analysis, we re-examined these associations with prospectively collected data. Methods Based on maternal reports, prenatal and postnatal cellphone exposures were assessed at age 7 years, and emotional and behavioural difficulties were assessed at 7 and 11 years with the Strengths and Difficulties Questionnaire. We used multivariable-adjusted logistic regression models to estimate ORs and 95% CIs relating prenatal exposure and age-7 cellphone use to emotional and behavioural difficulties at age 11 years. Results Children without emotional and behavioural difficulties at age 7 years, but who had cellphone exposures, had increased odds of emotional and behavioural difficulties at age 11 years, with an OR of 1.58 (95% CI 1.34 to 1.86) for children with both prenatal and age-7 cellphone exposures, 1.41 (95% CI 1.20 to 1.66) for prenatal exposure only, and 1.36 (95% CI 1.14 to 1.63) for age-7 use only. These results did not materially change when early adopters were excluded, or when children with emotional and behavioural difficulties at age 7 years were included in the analysis. Conclusions Our findings are consistent with patterns seen in earlier studies, and suggest that both prenatal and postnatal exposures may be associated with increased risks of emotional and behavioural difficulties in children.

Prenatal and Postnatal Cell Phone Exposures and Headaches in Children

OBJECTIVE Children today are exposed to cell phones early in life, and may be at the greatest risk if exposure is harmful to health. We investigated associations between cell phone exposures and headaches in children. STUDY DESIGN The Danish National Birth Cohort enrolled pregnant women between 1996 and 2002. When their children reached age seven years, mothers completed a questionnaire regarding the childs health, behaviors, and exposures. We used multivariable adjusted models to relate prenatal only, postnatal only, or both prenatal and postnatal cell phone exposure to whether the child had migraines and headache-related symptoms. RESULTS Our analyses included data from 52,680 children. Children with cell phone exposure had higher odds of migraines and headache-related symptoms than children with no exposure. The odds ratio for migraines was 1.30 (95% confidence interval: 1.01-1.68) and for headache-related symptoms was 1.32 (95% confidence interval: 1.23-1.40) for children with both prenatal and postnatal exposure. CONCLUSIONS In this study, cell phone exposures were associated with headaches in children, but the associations may not be causal given the potential for uncontrolled confounding and misclassification in observational studies such as this. However, given the widespread use of cell phones, if a causal effect exists it would have great public health impact.

On the association of cell phone exposure with childhood behaviour

We were pleased to read the paper by Guxens et al,1 which examined the association of prenatal cell phone and cordless phone use with behavioural problems in children. We are encouraged to see new research on this topic, but would like to point out several limitations in that study. The authors concluded that they did not find an association between prenatal cell phone use and behavioural problems in children, and that their findings differed from associations in the Danish National Birth Cohort (DNBC) as reported by Divan et al .2 ,3 These conclusions overlook Guxens et al s small sample size of 2529 and much smaller number of exposed cases. The …

Maternal cell phone use during pregnancy and child behavioral problems in five birth cohorts

INTRODUCTION Previous studies have reported associations between prenatal cell phone use and child behavioral problems, but findings have been inconsistent and based on retrospective assessment of cell phone use. This study aimed to assess this association in a multi-national analysis, using data from three cohorts with prospective data on prenatal cell phone use, together with previously published data from two cohorts with retrospectively collected cell phone use data. METHODS We used individual participant data from 83,884 mother-child pairs in the five cohorts from Denmark (1996-2002), Korea (2006-2011), the Netherlands (2003-2004), Norway (2004-2008), and Spain (2003-2008). We categorized cell phone use into none, low, medium, and high, based on frequency of calls during pregnancy reported by the mothers. Child behavioral problems (reported by mothers using the Strengths and Difficulties Questionnaire or Child Behavior Checklist) were classified in the borderline/clinical and clinical ranges using validated cut-offs in children aged 5-7years. Cohort specific risk estimates were meta-analyzed. RESULTS Overall, 38.8% of mothers, mostly from the Danish cohort, reported no cell phone use during pregnancy and these mothers were less likely to have a child with overall behavioral, hyperactivity/inattention or emotional problems. Evidence for a trend of increasing risk of child behavioral problems through the maternal cell phone use categories was observed for hyperactivity/inattention problems (OR for problems in the clinical range: 1.11, 95%CI 1.01, 1.22; 1.28, 95%CI 1.12, 1.48, among children of medium and high users, respectively). This association was fairly consistent across cohorts and between cohorts with retrospectively and prospectively collected cell phone use data. CONCLUSIONS Maternal cell phone use during pregnancy may be associated with an increased risk for behavioral problems, particularly hyperactivity/inattention problems, in the offspring. The interpretation of these results is unclear as uncontrolled confounding may influence both maternal cell phone use and child behavioral problems.

Characterization of extremely low frequency magnetic fields from diesel, gasoline and hybrid cars under controlled conditions.

This study characterizes extremely low frequency (ELF) magnetic field (MF) levels in 10 car models. Extensive measurements were conducted in three diesel, four gasoline, and three hybrid cars, under similar controlled conditions and negligible background fields. Averaged over all four seats under various driving scenarios the fields were lowest in diesel cars (0.02 μT), higher for gasoline (0.04–0.05 μT) and highest in hybrids (0.06–0.09 μT), but all were in-line with daily exposures from other sources. Hybrid cars had the highest mean and 95th percentile MF levels, and an especially large percentage of measurements above 0.2 μT. These parameters were also higher for moving conditions compared to standing while idling or revving at 2500 RPM and higher still at 80 km/h compared to 40 km/h. Fields in non-hybrid cars were higher at the front seats, while in hybrid cars they were higher at the back seats, particularly the back right seat where 16%–69% of measurements were greater than 0.2 μT. As our results do not include low frequency fields (below 30 Hz) that might be generated by tire rotation, we suggest that net currents flowing through the cars’ metallic chassis may be a possible source of MF. Larger surveys in standardized and well-described settings should be conducted with different types of vehicles and with spectral analysis of fields including lower frequencies due to magnetization of tires.

Complexities of sibling analysis when exposures and outcomes change with time and birth order.

In this study, we demonstrate the complexities of performing a sibling analysis with a re-examination of associations between cell phone exposures and behavioral problems observed previously in the Danish National Birth Cohort. Children (52,680; including 5441 siblings) followed up to age 7 were included. We examined differences in exposures and behavioral problems between siblings and non-siblings and by birth order and birth year. We estimated associations between cell phone exposures and behavioral problems while accounting for the random family effect among siblings. The association of behavioral problems with both prenatal and postnatal exposure differed between siblings (odds ratio (OR): 1.07; 95% confidence interval (CI): 0.69–1.66) and non-siblings (OR: 1.54; 95% CI: 1.36–1.74) and within siblings by birth order; the association was strongest for first-born siblings (OR: 1.72; 95% CI: 0.86–3.42) and negative for later-born siblings (OR: 0.63; 95% CI: 0.31–1.25), which may be because of increases in cell phone use with later birth year. Sibling analysis can be a powerful tool for (partially) accounting for confounding by invariant unmeasured within-family factors, but it cannot account for uncontrolled confounding by varying family-level factors, such as those that vary with time and birth order.