17th International Congress for Radiation Research, Montréal, Canada, August 27-30, 2023
The USTUR director, Sergey Tolmachev, two USTUR faculty members, Stacey McComish and Xirui Liu, and the USTUR’s Department of Energy program manager, Joey Zhou, attended the 17th International Congress for Radiation Research in Montréal, Canada. Their work to explore the misclassification of diseases among USTUR Registrants was presented as two posters. The first poster by Ms. McComish quantified the rate of death certificate misclassification observed in the USTUR population, and the second poster by Ms. Liu explored the potential impact of those misclassification errors on radiation epidemiological studies. Additionally, Dr. Tolmachev and Ms. McComish were co-authors on a slide presentation by Sara Dumit of Los Alamos National Laboratory.
Death certificates vs. autopsy reports: misclassification of causes of death among USTUR Registrants
Stacey L. McComish (USTUR), Xirui Liu (USTUR), Florencio T. Martinez (USTUR), Joey Y. Zhou (DOE), Sergey Y. Tolmachev (USTUR)
The U.S. Transuranium and Uranium Registries performs autopsies on each of its Registrants as a part of its mission to follow up occupationally-exposed individuals. This provides a unique opportunity to explore death certificate misclassification errors, and the factors that influence them, among this small population of former nuclear workers. Underlying causes of death (UCOD) from death certificates and autopsy reports were coded using the 10th revision of the International Classification of Diseases (ICD-10). These codes were then used to quantify misclassification rates among 275 individuals for whom both death certificates and autopsy reports were available. The ICD-10 categorizes diseases using 22 chapters. Death certificates incorrectly identified the UCOD ICD-10 disease chapter in 25.5% of cases. The misclassification rates for the most common disease chapters were: 9.9% neoplasms, 16.4% circulatory, 37.5% nervous system, 59.3% respiratory, and 18.7% external causes. A logistic regression revealed that both clinical history and the use of autopsy findings have a statistically significant influence on the match rate. Calculating the odds ratio for clinical history indicates that the odds of a match were 2.7 times higher when clinical history was mentioned on the autopsy report than when it was not. Similarly, when cases in the unknown autopsy influence group were excluded, the odds of a match were 4.0 times higher when death certificates were completed using autopsy findings than when autopsy findings were not used. [USTUR-0646-23A]
Impact of death certificate misclassifications on radiation cancer risk estimates
Xirui Liu (USTUR), Stacey L. McComish (USTUR), Sergey Y. Tolmachev (USTUR), Joey Y. Zhou (DOE)
A prior US Transuranium and Uranium Registries (USTUR) study explored misclassification of underlying causes of death (UCODs) on death certificates by comparing UCODs on death certificates to those on autopsy reports. It found a false-positive (over-classification) rate of 4.2%and a false-negative (under-classification) rate of 14%. The current study evaluates the impact of misclassification errors such as these on radiation cancer risk estimates. Autopsy reports, death certificates, and external radiation doses were available for 243 USTUR Registrants. The average cumulative external radiation dose was used to divide cases into high and low dose groups. The true odds ratio for cancer deaths – 0.679 (p=0.185 > 0.05) – was calculated from the UCODs on the autopsy reports while the odds ratio according to the death certificates was 0.828 (p=0.517 > 0.05). This represented a shift in the odds ratio toward one due to death certificate misclassifications; however, neither of the odds ratios was statistically significant. This scenario represents the impact of one possible combination of over- and under-misclassification errors on the odds ratio. Of the 243 cases in this study, 81 died of cancer and 162 died of non-cancer. The over-misclassification rate of 4.2% identified in the previous study was used to evaluate how false positives on death certificates could impact the statistical significance of the odds ratio. Seven non-cancer cases (162×4.2%) were selected at random and changed to cancer cases. The odds ratio and p-value for the resulting dataset were calculated and saved. When this process was repeated 20,000 times, 3.1% of the p-values were less than 0.05 (two-sided), meaning that there was a 3.1% probability that over-misclassification errors would result in a statistically significantly odds ratio, even though the true odds ratio was not significant. Similarly, when the under-misclassification rate of 14% was applied, 12 cases (81×14%) were randomly misclassified from non-cancer to cancer, resulting in a 2.5% probability that under-misclassification errors would lead to the erroneous conclusion that the odds ratio was statistically significant. When the over-classification rate of 4.2% and the under-classification rate of 14% were simulated simultaneously, the p-value was significant 8.6% of the time. [USTUR-0645-23A]
*This is the abstract as it was published in the book of abstracts. The poster that was presented focused on more recent simulation results.
Modeling of a unique USTUR dataset: female nuclear worker treated with chelation therapy after plutonium inhalation
Sara Dumit (LANL), Maia Avtandilashvili (USTUR), Stacey L. McComish (USTUR), Guthrie Miller (retired LANL), Jasen Swanson (US Army), Sergei Y. Tolmachev (USTUR)
This study shows the modeling of a unique dataset of bioassay measurements from a female former nuclear worker available at the US Transuranium and Uranium Registries (USTUR). The worker was internally exposed to a plutonium-americium mixture via acute inhalation at a nuclear weapons facility. She was medically treated with injections of 1 g Ca-DTPA on days 0, 4, and 14 after the intake. A total of 13 fecal and 24 urine samples was collected and analyzed for plutonium (Pu) and americium (Am) immediately after the intake, from day 0 to day 20. Consequently, she was followed-up for bioassay monitoring over 14 years, with 13 additional post-treatment urine samples collected and analyzed for Pu. The uniqueness of this dataset is due to the availability of: (i) both early and long-term bioassay data from a female with Pu intake; (ii) data on chelation therapy for a female; and (iii) fecal measurement results. Chelation therapy with DTPA is known to aid in reducing the internal radiation dose by enhancing the excretion rate of Pu from the body. Such enhancement affects the normal biokinetics of Pu in vivo, posing a challenge to the internal dose assessment. The current dose assessment practice is to exclude the data affected by DTPA from the analysis. Using this standard approach, the worksite’s Radiation Protection staff estimated the Pu intake to be 73 Bq, with a Committed Effective Dose (CED) to the whole body of 16 mSv and a Committed Equivalent Dose (CEqD) to the bone surfaces of 340 mSv. The present analysis is the first one to explicitly model the combined biokinetics of Pu and DTPA by using a newly developed chelation model. The Markov Chain Monte Carlo method was used to investigate the model parameter uncertainty, given the bioassay data and assumed prior probability distributions. This work is groundbreaking because the modeling includes the bioassay data collected before, during, and after the DTPA administrations. Preliminary results of this study show that the worker’s Pu intake was 22 Bq, with a CED to the whole-body of 1.74 mSv and a CEqD to the bone surfaces of 42.8 mSv, which differ from the original worksite’s calculations. The difference in results is expected because this analysis incudes pre- and post-treatment bioassay data and uses a novel model that accounts for the effect of chelation therapy in removing Pu from the body. [USTUR-0644-23A]