Analyses of the risk posed by asbestos draw on data concerning occupationally exposed cohorts, gathered in the context of studies that have been described and discussed in the published literature since 1960. The studies in question are not all equally suitable for use as basis for risk analysis. In some of the studies, the characterisation of exposure in particular is neither complete nor optimal. 4.1 Measurement of asbestos exposure in epidemiological studies
Over the years, the various cohort studies have made use of a variety of measure- ment techniques for fibres. In the first of these studies, carried out before the Sec- ond World War, exposure to asbestos was measured by capturing fibres in wash bottles (impingers) filled with a liquid (usually alcohol) and counting the parti- cles. The measuring period was generally shorter than half an hour. Fibres were not distinguished from particles, and more detailed identification of the fibres was technically impossible. The concentration was expressed in millions of parti- cles per cubic foot (mppcf). Filter methods were also used in order to establish substance concentrations gravimetrically (by weighing).
Shortly thereafter, methods were introduced which made use of microscopic counting techniques. Phase contrast microscopy (PCM) has long been the most
widely used microscope technology in such counting methods*; it allows the
measurement of fibres thicker than approximately 0.25 µm. With transmission electron microscopy (TEM) or scanning electron microscopy (SEM)**, it is pos-
sible to also count fibres thinner than those observable using PCM; these tech- niques allow for the detection of fibres with a diameter of as little as 0.01 µm. However, the use of electron microscopy is more expensive than the use of PCM and requires the involvement of specially trained personnel. Initially, measure- ments were based on the use of static equipment. Later on, personal sampling became more widely applied.
4.2 Use of conversion factors for the comparison of research results Because the numerous studies of asbestos risk have made use of a variety of measurement techniques, their results are expressed in different units of expo- sure. Conversion factors are therefore needed if the results are to be compared and expressed in one unit of exposure.
A report of the US National Research Council provides an overview of widely used conversion factors.35 Table 8 (adopted from35) shows, for instance, that
where measurement by impinger would yield a value of 1 particle per m3, a
phase contrast microscope would have detected on average 6 fibres/m3. The use
of standard conversion factors results in a simplification of reality, however, because an accurate conversion factor is strongly dependent on the specifics of the environment, and should ideally be determined on a case-by-case basis for each environment. In practice, true conversion factors have been found to vary considerably***.
* Workplace measurements using PCM usually provide a good insight into the prevailing atmospheric asbestos con- centration. However, PCM does not allow for asbestos fibres to be distinguished from other fibres, such as cotton, paper, mineral wool and glass fibres. When indoor measurements are made in a setting where non-asbestos fibres are likely to be present, there is a significant risk of overestimation; yet the inability to detect fibres with a diameter of < about 0.25 µm is liable to lead to underestimation. Consequently, if non-asbestos fibres are likely to be
present, environmental concentrations are often measured by means of transmission electron microscopy (TEM) or scanning electron microscopy (SEM), possibly in combination with detection techniques such as XRMA or SAED, which allow for distinction to be made between different fibre types.
** In Europe, most laboratories (except those in France) use SEM, rather than TEM. The current generation of SEM equipment is at least as good as – and in some ways superior to – modern TEM equipment. SEM is better in terms of minimum detection limit, measurement uncertainty and the likelihood of contamination.
*** Comparative research has previously shown that the factor for the conversion of ‘light-microscope to electron- microscope chrysotile measurements’ varies from 19 to 76 for all airborne asbestos fibres, depending on the type of working environment (61, in ATSDR, 20017).
Nature and quality of the epidemiological research 49 Fibres shorter than 5 µm are considered to make little contribution to the carcino-
genic potency of asbestos (see subsection 2.2.1). These small fibres cannot be measured using PCM in a way that complies with the applicable counting con- vention. For regulatory purposes, the most important conversion factors are those used to convert PCM-based fibre concentrations into TEM figures for fibres longer than 5 µm. Verma and Clark’s comparative research established a PCM to TEM conversion factor for fibres longer than 5 µm (with a diameter greater than 0.3 µm) of between 1.2 and 10.4, but usually between 1.4 and 3.2.61 A report
published by the Health Effects Institute suggested that a conversion factor range from one order of magnitude below to one order of magnitude above would probably cover all workplace settings.17 The EPA concluded that the PCM-to-
TEM conversion factor (for fibres more than 5 µm long with a diameter greater than 0.4 µm) was between 2 and 4 (EPA, 1986).24
4.3 Lack of detail concerning the quantification of exposure in the occu- pationally exposed cohorts
The usefulness of earlier studies is limited not only by differences in measure- ment methods but also by the differences in the nature of the available data; for instance, data are incomplete, or have not been collected in accordance with cur- rent standards.
In some studies, for example, the measurement strategy used to characterise the exposure – i.e. to assign exposure levels to samples of workers and in time based on measurements of airborne asbestos or external data – was not consistent with modern principles, resulting in the misclassification of exposure and poten- tial attenuation or more generally bias of the exposure-response relationship. For some cohort studies, the researchers did not know exactly how long workplace
Table 8 Factors for the conversion of atmospheric fibre concentrations obtained by means of the various measurement methods used to measure workplace asbestos levels (adopted from 35). Bracketing of a conversion factor indicates that the
quoted figure is an estimate obtained by the combination of other conversion factors. Equivalent value for alternative method Original measurement obtained using: Impinger (particles) PCM (fibres/ml, fibres >5µm) TEM (fibres/ml) Gravimetrical (mass) mg/m3 Impinger (particles) 1 6 360 0.2 PCM (fibres/ml, fibres >5µm) 0.17 1 60 0.03 TEM (fibres/ml) 0.0028 0.017 1 0.0005
Gravimetric (mass) measurement mg/m3
exposure had lasted and they therefore simply made crude estimates regarding duration of exposure.
In many of the studies, subjects were exposed to various types of asbestos. Moreover, the distribution of fibre lengths differed very likely from one study to the next or was not known; the asbestos that mine workers are exposed to will not exhibit the same fibre length distribution as the asbestos that people working in an asbestos textile factory are exposed to. Yet fibre length distribution data are available only for a few of the studied cohorts.
There are also gaps in the information available regarding the asbestos-can- cer exposure-response relationship. The relationship between fibre length (and diameter) and the carcinogenic potency of asbestos is not clear from the pub- lished research. Some data are available from animal research, but not readily transferrable to humans. Moreover, while recent epidemiological analyses sug- gest that longer and thinner fibres may play a more important role, the precise relationship between fibre dimensions and health effects cannot reliably be deter- mined, because most of the studies provide too little information about fibre length and diameter distributions.
Another shortcoming is that information about smoking patterns is available only for a few of the cohorts, meaning that the association between asbestos exposure and lung cancer often could not be adjusted for smoking.
Finally, in several studies the information provided on the cause of death is not completely accurate (particularly with regard to diagnosis of mesothelioma cases in the decades before and immediately after the Second World War) The shortcomings in the nature and quality of the epidemiological research described above are expected to lead to misclassification and bias the apparent association between asbestos exposure and the occurrence of lung cancer or mes- othelioma. The degree of bias is difficult to estimate in most cases, being possi- ble only where validation studies were performed. Any interpretation of the entire body of evidence regarding health effects resulting from asbestos exposure data needs to take account of the considerations set out above.