Fallout Memory Trajectories at Semipalatinsk 209
epidemiology is a scientific discipline that has been highly dependent on
state infrastructures for standardized, long-term collection of data. Once the
methods of counting are changed, the numbers are no longer comparable,
requiring complex validation studies, modeling and translation work to render data combinable – and the more of these transformations are done, the
more uncertainties are generated. Epidemiological studies ideally require
standardized and unchanged data collection over a long time that is independent of exposure, so that there is no systematic error introduced in the
data collection. This makes routine data recorded by the state, such causes of
death records important data sources. Conducting an epidemiological study
is a complex scientific memory practice, with data recording and measurement changing with improved diagnostics and therapeutics. Isolating the
effect of radiation exposure at aggregate levels implies many assumptions,
especially in the lower dose range and in retrospect when case ascertainment is different and often insecure. For instance, epidemiological studies of
fallout-related leukemia at Semipalatinsk have proved difficult to carry out,
with cases expected to have peaked 10 years after exposure – likely to be
undiagnosed in the rural areas during the 1950s and 1960s, let alone what
physicians outside the secret radiation research centers dared (not) to write
into the medical death certificates.
Establishing a kind of “nuclear census” is a prerequisite to reconstruct
exposures and retrieve disease to assess radiation risk. During the 1990s,
both local and international research efforts focused on the establishment of
registries, i.e., exposure and disease registries. Unless epidemiologists – as
done in some prevalence studies – carried out an entirely new data collection from scratch, all studies had to work with data collected in the past.
These epidemiological memory practices proved highly entangled with the
political and administrative systems as well as with the enactment of Soviet
modernity through public health. This was routine data, recorded by the
USSR administration – in rural places it were the obligatory birth and death
records as well as the kolkhoz books that functioned as registries for the
exposed rural populations. Especially for the period of atmospheric nuclear
testing 1949–1963, they were used as replacements for the lacking passport
system for the rural USSR population.
Following up on concerns by health authorities and the UN resolution on
Semipalatinsk of 1998, WHO commissioned a reproductive health study to
be conducted in collaboration with institutes in Semipalatinsk, UK epidemiologists, and WHO researchers. Called upon by WHO to produce “a scientifically sound” epidemiological study, western scientists looked into and
surveyed the possibilities and availability of records. Putting methods and
exclusion of possible bias first, they decided to work with those long-term
Soviet administrative data from vital statistics offices, for which completeness could be ascertained; the expectation was that they would be recorded
exactly the same way independent of whether the area was exposed or not,
thus avoiding bias in the data. This however restricted the outcomes that
epidemiology is a scientific discipline that has been highly dependent on
state infrastructures for standardized, long-term collection of data. Once the
methods of counting are changed, the numbers are no longer comparable,
requiring complex validation studies, modeling and translation work to render data combinable – and the more of these transformations are done, the
more uncertainties are generated. Epidemiological studies ideally require
standardized and unchanged data collection over a long time that is independent of exposure, so that there is no systematic error introduced in the
data collection. This makes routine data recorded by the state, such causes of
death records important data sources. Conducting an epidemiological study
is a complex scientific memory practice, with data recording and measurement changing with improved diagnostics and therapeutics. Isolating the
effect of radiation exposure at aggregate levels implies many assumptions,
especially in the lower dose range and in retrospect when case ascertainment is different and often insecure. For instance, epidemiological studies of
fallout-related leukemia at Semipalatinsk have proved difficult to carry out,
with cases expected to have peaked 10 years after exposure – likely to be
undiagnosed in the rural areas during the 1950s and 1960s, let alone what
physicians outside the secret radiation research centers dared (not) to write
into the medical death certificates.
Establishing a kind of “nuclear census” is a prerequisite to reconstruct
exposures and retrieve disease to assess radiation risk. During the 1990s,
both local and international research efforts focused on the establishment of
registries, i.e., exposure and disease registries. Unless epidemiologists – as
done in some prevalence studies – carried out an entirely new data collection from scratch, all studies had to work with data collected in the past.
These epidemiological memory practices proved highly entangled with the
political and administrative systems as well as with the enactment of Soviet
modernity through public health. This was routine data, recorded by the
USSR administration – in rural places it were the obligatory birth and death
records as well as the kolkhoz books that functioned as registries for the
exposed rural populations. Especially for the period of atmospheric nuclear
testing 1949–1963, they were used as replacements for the lacking passport
system for the rural USSR population.
Following up on concerns by health authorities and the UN resolution on
Semipalatinsk of 1998, WHO commissioned a reproductive health study to
be conducted in collaboration with institutes in Semipalatinsk, UK epidemiologists, and WHO researchers. Called upon by WHO to produce “a scientifically sound” epidemiological study, western scientists looked into and
surveyed the possibilities and availability of records. Putting methods and
exclusion of possible bias first, they decided to work with those long-term
Soviet administrative data from vital statistics offices, for which completeness could be ascertained; the expectation was that they would be recorded
exactly the same way independent of whether the area was exposed or not,
thus avoiding bias in the data. This however restricted the outcomes that