Miscarriages and Congenital Conditions in Offspring of Veterans of the British Nuclear Atmospheric Test Programme
A postal questionnaire case-control study examined miscarriage in wives and congenital conditions in offspring ofthe 2007 membership of the British Nuclear Test Veterans Association, a group of ex-servicemen who werestationed at atmospheric nuclear weapon test sites between1952-67. Results were compared with a veteranselected control group and also with national data. Based on 605 veteran children and 749 grandchildren comparedwith 311 control children and 408 control grandchildren there were significant excess levels of miscarriages,stillbirths, infant mortality and congenital illnesses in the veterans’ children relative both to control children andexpected numbers. 105 miscarriages in veteran’s wives compared with 18 in controls OR=2.75 (1.56, 4.91;p=00016). There were 16 stillbirths; 3 in controls (OR=2.70 (0.73, 11.72; p=0.13). Perinatal mortality OR was 4.3(1.22, 17.9; p=.01) on 25 deaths in veteran children. 57 veteran children had congenital conditions vs. 3 controlchildren (OR=9.77 (2.92, 39.3); p=0.000003) these rates being also about 8 times those expected on the basis ofUK EUROCAT data for 1980-2000. For grandchildren, similar levels of congenital illness were reported with 46veteran grandchildren compared with 3 controls OR=8.35 (2.48, 33.8) p=0.000025. There was significantly morecancer in the veteran grandchildren than controls.
Whilst caution must be exercised due to structural problems inherent in this study we conclude that the veterans’offspring qualitatively exhibit a prevalence of congenital conditions significantly greater than that of controls and alsothat of the general population in England. The effect remains highly statistically significant even assuming a highselection bias in the responses and credibility is strengthened by the high rates of miscarriage reported by theveterans compared with controls, a result which could hardly have been due selection effects.
Reflection
15 February 2026
Emma-leigh Theobald
“There were significant excess levels of miscarriages, stillbirths, infant mortality and congenital illnesses in the veterans’ children when compared with controls.”
“Whilst caution must be exercised due to structural problems inherent in this study, the results suggest that involvement in the Nuclear Tests was associated with increased levels of genetic illness in the children and grandchildren of veterans.”
“There are no contemporary measurements available of internal exposures at the test sites, nor of Uranium contamination.”
“The film badge dosimeters issued to servicemen registered prompt external doses of gamma radiation, but did not measure alpha emitters such as Plutonium and Uranium.”
“We argue that the results of this study support the belief that involvement in the Nuclear Tests caused increased rates of genetic-based illness in both the children and grandchildren of veterans.”
“We suggest the cause is internal exposure to radioactive contamination at the test sites, particularly to Uranium.”
This paper sits in a difficult and contested space. It examines miscarriage and congenital conditions in the families of British nuclear test veterans and argues that rates are significantly higher than in control families and national data. The authors are explicit about the structural weaknesses of their study — particularly selection bias and reliance on self-report — yet they maintain that the patterns observed are too large and too consistent to be easily dismissed.
What strikes me most is the methodological and political tension running through the paper. On the one hand, it challenges the adequacy of official dose records and the dominant focus on external radiation exposure. On the other, it relies on community-generated data — questionnaires distributed through the veterans’ association — to surface outcomes that were never systematically measured by the state.
The argument hinges on absence. The authors note that there are no contemporary measurements of internal contamination, and that monitoring focused on prompt external doses rather than alpha emitters such as uranium and plutonium. That gap becomes central to their claim: if what mattered was never measured, then reassurance built on those measurements is incomplete by design.
The study insists that intergenerational outcomes must be taken seriously, even where proof remains contested. Whether one accepts the authors’ causal interpretation or not, the paper exposes something deeper: how difficult it is to adjudicate harm when records are partial, exposures uncertain, and families left to assemble their own evidence decades later.
In that sense, this is not only an epidemiological argument. It is an argument about accountability, evidence, and who bears the burden of uncertainty.