In your experience, how often do researchers conducting epidemiology studies using biomonitoring data for exposure assessments consider the variability (between- and within-person) in the exposure metric?

Answer Explanations

• Hardly ever Sometimes

  Expert 3
   In my research projects, my collaborators consistently consider both between- and within-person exposure variability. However, I recognize that this group is somewhat unique in this regard, as they were among the pioneers in developing these concepts (Kromhout, Vermeulen, Heederik, Brunekreef, Burstyn). Outside of my group, I have encountered only one research team from Finland that also accounted for between- and within-person variability. To date, I have personally not come across a U.S.-based health research group led by MDs that has incorporated this consideration. I have only reviewed one article that addressed between- and within-person variability, although this is emphasized in the leading textbook (Quantitative Exposure Assessment by Lawrence Kupper and Steven  Rappaport) on quantitative exposure assessment.  
• Sometimes

  Expert 4
  Depends on data availability, if multiple measurements are available most researchers will use all of them, and sometimes may even exclude participants with too few measurements. 
• Sometimes

  Expert 6
  This emphasis make this paper a useful contribution.
• Sometimes

  Expert 7
  There is an increasing recognition of the importance of considering both between- and within-person variability in epidemiology studies using biomonitoring data. However, cost considerations often require choosing between enrolling more subjects vs making more measurments.  Prior knowledge of between- and within-person variability is needed for best study design but is rarely available.
• Sometimes

  Expert 1
  There are obviously numerous factors to consider, these include sample size, timing of biomonitoring measurements (hourly, daily, weekly, etc...), assessment of confounding factors (e.g., diet), purpose of biomonitoring (e.g., athletes for drug assessment), and the type of testing method used including type of sample used (e.g., urine, sweat, tears, etc...). Studies sometimes assess both between and within variability, but this is not always the case. 
• Most of the time

  Expert 9
  Most researchers I know consider all likely sources of variability (both between- and within-person) in designing studies or critiquing other studies.  They might not do so explicitly in sample size calculations, but might consider it implicitly for example, by increasing overall sample size to account (partially) for within-person variation, or by other approaches to decrease within-person variation (such as taking 24-hour voids or first morning void for urine specimens).
• Sometimes

  Expert 8
  Often there are some measures of between- and within-person variability described, but not always. I would not say 'most of the time'. Between person is more often reported than within person. Also depends on the study design, e.g. whether longitudinal data are available. 
• Most of the time

  Expert 5
  Overall, I think that most epidemiologists don't often consider the "within" variation for individuals, but almost always consider the between variation as that measure is necessary to know (or estimate) for  basic sample size calculations.  
  
  Those epidemiologists who work primarily on studies using biomonitoring are well aware of the within-individual variation and the effects of this and repeated measurements on exposure measurement error, study power, and bias in risk estimates.  
  
  Generally, we don't have good estimates (from the field) of either within or between variation to assist with sample size/power estimations. 
• Most of the time

  Expert 2
  The consideration of variability (both between-person and within-person) in exposure metrics has become increasingly common in epidemiology studies using biomonitoring data. Over the past two decades, the importance of accounting for variability has been widely recognized, driven by the understanding that ignoring it can lead to exposure misclassification and biased results.
  
  

  Many studies now incorporate designs and statistical methods that address variability, such as repeated measures and mixed-effects models. This shows that researchers are mindful of variability in their assessments. Additionally, the trend towards following guidelines and best practices (such as WHO’s Biological Monitoring of Chemical Exposure in the Workplace, EFSA’s Human biomonitoring data collection from occupational exposure to pesticides) has led to more consistent reporting on how variability is handled, especially in studies published in reputable journals.
  
  

  However, there are still instances where variability might not be fully considered, particularly in studies with limited resources or less experienced researchers. In these cases, variability may be acknowledged but not thoroughly addressed, or it might be overlooked due to practical constraints. 
  
  

  The frequency of considering variability also varies depending on the study context, such as the population being studied and the nature of the exposure. Occupational studies, might be more rigorous in this regard compared to some environmental studies where repeated measures are harder to obtain.
  
  

  In summary, addressing variability is often practiced but not yet universally guaranteed. The field has made significant progress, but there is still room for improvement in ensuring that variability is consistently and thoroughly considered across all contexts.