Uncertainties in dose assessments

In the conduct of practices, rates of release of radionuclides are generally low and the possibilities for a detailed analysis of exposure might be limited ik for example, the external dose rate attributed to releases is of the same order as the fluctuations in the dose rate due to background radiation. In this case, the dose can be assessed as a value less than the dose estimated with the minimum detectable activity for the measurement used as input data. This dose assessment can be assigned an estimated uncertainty that takes into account the uncertainties in the parameters of the dosimetric models. 

Besides the uncertainties associated with monitoring procedures, an important source of uncertainty arises from the modelling and especially from people s habits. Often only nationwide average values, if any at all, for the relevant parameters are known, which may deviate considerably from the values for specific persons in specific areas. For the conduct of practices and for chronic (prolonged) exposure, activity levels in the environment may be 

Pathway of human exposure Quantity monitored Source of uncertainties in dose estimates 

External Gamma dose rate in air as Location and duration of stay of people in 

Ingestion Activity concentration in foods as fnnction of time and space Age dependent food intake origin of food seasonal variation of food intake 

The models that have been developed by the ICRP for describing the behaviour of radionuclides in the body, and hence for assessing intakes, provide the most up-to-date methods available for dose assessment. There are, however, a number of uncertainties that should be considered when interpreting monitoring data. 

Direct methods rely on the results of either whole or partial bocfy monitoring. The accuracy of atty measurements will depend principally on the level of activity, but also on the acciuacy of calibration of the monitoring equipment. Limits of detection for ai r particular radionuclide can be calculated from a knowledge of the sensitivity of the equipment and the background count in the region of interest. 

For indirect methods, the accuracy of measurements of levels of activity in physical or biological samples depends on similar considerations. It is, however, generally possible to define the counting geometry accurately, and counting times can be extended if necessary to obtain acceptable counting statistics for all samples except those with very low activity (or very short half-hves). 

From an assessment of the activity in the whole body or in samples of tissues or excreta, the models used to describe the behaviour of radionuclides in the body are then used to assess intake and dose. The reliability of the estimates of dose therefore depend upon the accuracy of the models, and any limitations on their application in particular circumstances. This will depend upon many factors. In particular, knowledge of the time of the intake(s) and of whether the intake was acute or chronic is essential for a reliable dose estimate. 

The Annex gives dose coefficients e g)j from the BSS [2] and DACs for selected radionuclides that are likely to be of concern in the woikplace. The DACs are calculated on the basis of an effective dose limit of 20 mSv in a year, a woiking time of 2000 h per year and a standard breathing rate of 1.2 m /h, and are given for AMADs of 1 and 5 pm. 

---

The uncertainties in dose assessments should be taken into account in the interpretation of data from radiation monitoring. 

Uncertainties in dose-response assessment. Several aspects of dose-response assessment result in significant uncertainties in the accuracy of the resulting relationship. 

Deficiencies in dose-response assessment. In addition to the sources of uncertainty in dose-response assessment described above, there are several important deficiencies in the way that the... 

Many sources of uncertainty in mixture assessment are comparable to those in the assessment of single substances. Examples are uncertainties about emission loads, the fate of the substances in the environment, exposure scenarios, variability in individual characteristics, extrapolation of toxicity data between species, and the shape of the dose-response curve. But there are also uncertainties that are typical for mixture assessment. An example is the composition of the mixture, which is often unknown or only partly known. This is irrelevant if the mixture of concern can be tested directly in the laboratory or the field, but it results in uncertainty if the assessment is based on sufficient similarity or on the known individual components. It is therefore important to explicitly state... 

Interspecies and intraspecies UFs have been used in the development of safe or threshold exposure levels for chronic, noncancer toxicity by health organizations throughout the world. Examples include the acceptable daily intake (ADI) (Lu 1988 Truhaut 1991 Lu and Sielken 1991), the tolerable daily intake (TDI) or tolerable concentration (TC) (Meek et al. 1994 IPCS 1994), the minimal risk level (MRL) (Pohl and Abadin 1995), the reference dose (RfD) (Barnes and Dourson 1988 Dourson 1996), and the reference concentration (RfC) (EPA 1994 Jarabek et al. 1990). The importance of using distribution-based analyses to assess the degree of variability and uncertainty in risk assessments has been emphasized in recent trends in risk analysis. This will enable risk managers to make more informed decisions and... 

Stayner L, Bailer AJ, Smith R, et al. 1999. Sources of uncertainty in dose-response modeling of epidemiological data for cancer risk assessment. Arm N Y Acad Sci 895 212-222. 

Uncertainties as to the origins of foods remain important contributors to the uncertainties in the assessment of ingestion doses. Although in rural areas a significant proportion of the diet may be produced locally, at least a part of the food consumed is produced elsewhere. Where there are usually no reliable data on this matter, it may be assumed that all the food consumed is produced locally, which gives a conservative bias. 

Properly calibrated methods of individual monitoring, with the inherent uncertainties taken into account, provide the most precise data for use in dose assessment. The results of individual monitoring should be used to specify models for dose assessment by means of the comparison of appropriate radiological quantities (i.e. external doses for particular periods and/or radionuclide activity for the whole body at the time of individual measurements). If systematic discrepancies are identified, appropriate correction factors should be introduced into the dose assessment models. 

In risk characterization, step four, the human exposure situation is compared to the toxicity data from animal studies, and often a safety -margin approach is utilized. The safety margin is based on a knowledge of uncertainties and individual variation in sensitivity of animals and humans to the effects of chemical compounds. Usually one assumes that humans are more sensitive than experimental animals to the effects of chemicals. For this reason, a safety margin is often used. This margin contains two factors, differences in biotransformation within a species (human), usually 10, and differences in the sensitivity between species (e.g., rat vs. human), usually also 10. The safety factor which takes into consideration interindividual differences within the human population predominately indicates differences in biotransformation, but sensitivity to effects of chemicals is also taken into consideration (e.g., safety faaor of 4 for biotransformation and 2.5 for sensitivity 4 x 2.5 = 10). For example, if the lowest dose that does not cause any toxicity to rodents, rats, or mice, i.e., the no-ob-servable-adverse-effect level (NOAEL) is 100 mg/kg, this dose is divided by the safety factor of 100. The safe dose level for humans would be then 1 mg/kg. Occasionally, a NOAEL is not found, and one has to use the lowest-observable-adverse-effect level (LOAEL) in safety assessment. In this situation, often an additional un-... 

Risk characterization is lire process of estimating llie incidence of a health effect under the various conditions of human or animal exposure described in lire exposure assessment. It is performed by combining the exposure (see Cliapter 12) and dose response (see Cluipter 11) assessments. The summary effects of the uncertainties in lire preceding steps should also be described in lliis step. 

Decision Analysis. An alternative to making assumptions that select single estimates and suppress uncertainties is to use decision analysis methods, which make the uncertainties explicit in risk assessment and risk evaluation. Judgmental probabilities can be used to characterize uncertainties in the dose response relationship, the extent of human exposure, and the economic costs associated with control policies. Decision analysis provides a conceptual framework to separate the questions of information, what will happen as a consequence of control policy choice, from value judgments on how much conservatism is appropriate in decisions involving human health. 

The principal application of PBPK models is in the prediction of the target tissue dose of the toxic parent chemical or its reactive metabolite. Use of the target tissue dose of the toxic moiety of a chemical in risk assessment calculations provides a better basis of relating to the observed toxic effects than the external or exposure concentration of the parent chemical. Because PBPK models facilitate the prediction of target tissue dose for various exposure scenarios, routes, doses, and species, they can help reduce the uncertainty associated with the conventional extrapolation approaches. Direct application of modeling includes... 

It is recognized that the NOAEL derived by using this traditional approach for dose-response assessment is not very accurate with respect to the degree to which it corresponds with the (unknown) tme NAEL. Furthermore, in this traditional approach, only the data obtained at one dose (NOAEL) are used in the hazard assessment rather than the complete dose-response data set. In case sufficient data are available, the shape of the dose-response curve should be taken into account in the hazard assessment. In the case of a steep dose-response curve, the derived NOAEL can be considered as more reliable because the greater the slope, the greater the reduction in response to reduced doses. In the case of a shallow dose-response curve, the uncertainty in the derived NOAEL may be higher and this has to be taken into account in the hazard assessment (see Section 5.7). If a LOAEL has to be used in the hazard assessment, then this value can only be considered reliable in the case of a very steep dose-response curve. 

Probabilistic methods can be applied in dose-response assessment when there is an understanding of the important parameters and their relationships, such as identification of the key determinants of human variation (e.g., metabolic polymorphisms, hormone levels, and cell replication rates), observation of the distributions of these variables, and valid models for combining these variables. With appropriate data and expert judgment, formal approaches to probabilistic risk assessment can be applied to provide insight into the overall extent and dominant sources of human variation and uncertainty. 

One of the most evident limitations in the NOAEL approach in the derivation of tolerable intakes is that it does not take into account the slope of the dose-response curve for the particular response of interest (Section 4.2.4). The NOAEL is by definition one of the doses tested, and apart from ensuring that the number and spacing of data points are adequate to provide a reasonable estimate of the NOAEL, all other data points are ignored. Although the NOAEL could be considered an estimate of the tme NAEL, the quality of the estimate cannot be assessed. For the dose-response relationship and precision in the NOAEL, consideration should therefore be given to the uncertainties in the NOAEL as the surrogate for the NAEL. 

In case a NOAEL cannot be set for the critical effect, a LOAEL is then set and extrapolated to a NOAEL. The extrapolation from a LOAEL to a NOAEL can be regarded as part of the dose-response analysis. Consideration should therefore also be given to the uncertainties in the extrapolation of the LOAEL to the NAEL in cases where only a LOAEL is available as the starting point for the assessment. 

In theory, the steeper the slope of the dose-response curve, the smaller the assessment factor and vice versa. However, there is no scientific basis for any value of a default factor to account for uncertainty in the NOAEL, nor any distribution. 

The T25 approach was discussed at a workshop organized by the European Centre for the Ecotoxicologicy and Toxicology of Chemicals (ECETOC) (Roberts et al. 2001, ECETOC 2002). It was concluded that the use of the T25 method in risk assessment is problematic due to uncertainties arising from the false assumption of both precision and linearity in the dose-response curves for tumor induction. 

---