Study of coverage of confidence intervals for the standardized mortality ratio in studies with missing death certificates.

This paper assesses the coverage probability of commonly used confidence intervals for the standardized mortality ratio (SMR) when death certificates are missing. It also proposes alternative confidence interval approaches with coverage probabilities close to .95. In epidemiology, the SMR is an important measure of risk of disease mortality (or incidence) to compare a specific group to a reference population. The appropriate confidence interval for the SMR is crucial, especially when the SMR is close to 1.0 and the statistical significance of the risk needs to be determined. There are several ways to calculate confidence intervals, depending on a study characteristics (ie, studies with small number of deaths, studies with small counts, aggregate SMRs based on several countries or time periods, and studies with missing death certificates). This paper summarizes the most commonly used confidence intervals and newly applies several existing approaches not previously used for SMR confidence intervals. The coverage probability and length of the different confidence intervals are assessed using a simulation study and different scenarios. The performance of the confidence intervals for the lung cancer SMR and all other cancer SMR is also assessed using the dataset of French and Czech uranium miners. Finally, the most appropriate confidence intervals to use under different study scenarios are recommended.

Bagged neural network model for prediction of the mean indoor radon concentration in the municipalities in Czech Republic.

The purpose of the study is to determine radon-prone areas in the Czech Republic based on the measurements of indoor radon concentration and independent predictors (rock type and permeability of the bedrock, gamma dose rate, GPS coordinates and the average age of family houses). The relationship between the mean observed indoor radon concentrations in monitored areas (∼22% municipalities) and the independent predictors was modelled using a bagged neural network. Levels of mean indoor radon concentration in the unmonitored areas were predicted using the bagged neural network model fitted for the monitored areas. The propensity to increased indoor radon was determined by estimated probability of exceeding the action level of 300Bq/m3.

Radiocaesium levels in game in the Czech Republic.

The paper summarizes results of monitoring of (137)Cs activities in game species roaming in the woods over the territory of the Czech Republic for the time interval of 1986-2012. Geometric means and other statistical characteristics were estimated from the data sets on the assumption of log-normal distribution of the data from the time interval 2004-2012 where the character of data distribution had displayed no significant change. Geometric means (in Bq/kg) in meat were: wild boar 5.1, red deer 1.9, roe deer 0.77 and feathered game 0.14. The mean value in the less frequent game amounted to 0.36 Bq/kg. The geometrical standard deviation (GSD) widely varied from 1.6 to 21 for the studied species. Based on mass activity dependence on time, we assessed the effective and environmental half-lives of activity decline. For red deer and roe deer, the effective (137)Cs half-life was 2.9 and 3.2 years, and environmental half-life 3.2 and 3.6 years respectively. The effective half-life of (137)Cs in wild boar of 38 years was determined with large uncertainty and it shows constant influx of (137)Cs activity to the digestive tract of wild boars. A statistically significant season-based (137)Cs level was found in red deer and wild boar. Higher winter and spring activities of (137)Cs in wild boar are linked with decreasing access to naturally occurring food with lower (137)Cs content (chestnuts, acorns, and beech nuts), making boar grub around for ground-deposited food (often for mushrooms with higher activity). Higher winter activities of (137)Cs in red deer meat, most probably, are due to lower access to green diet in winter. The average annual committed effective dose for Czech population based on estimates of game species meat consumption between 2004 and 2012 was insignificant, only 0.03 µSv.