When evaluating environmental exposures, residential exposures are often most relevant. In most countries, it is impossible to establish full residential histories. In recent publications, childhood leukaemia and background radiation have been studied with and without full residential histories. This paper investigates the consequences of lacking such full data.

Data from a nationwide Finnish Case-Control study of Childhood Leukaemia and gamma rays were analysed. This included 1093 children diagnosed with leukaemia in Finland in 1990–2011. Each case was matched by gender and year of birth to three controls. Full residential histories were available. The dose estimates were based on outdoor background radiation measurements. The indoor dose rates were obtained with a dwelling type specific conversion coefficient and the individual time-weighted mean red bone marrow dose rates were calculated using age-specific indoor occupancy and the age and gender of the child. Radiation from Chernobyl fallout was included and a 2-year latency period assumed.

The median separation between successive dwellings was 3.4 km and median difference in red bone marrow dose 2.9 nSv/h. The Pearson correlation between the indoor red bone marrow dose rates of successive dwellings was 0.62 (95% CI 0.60, 0.64). The odds ratio for a 10 nSv/h increase in dose rate with full residential histories was 1.01 (95% CI 0.97, 1.05). Similar odds ratios were calculated with dose rates based on only the first dwelling (1.02, 95% CI 0.99, 1.05) and only the last dwelling (1.00, 95% CI 0.98, 1.03) and for subjects who had lived only in a single dwelling (1.05, 95% CI 0.98, 1.10).

Knowledge of full residential histories would always be the option of choice. However, due to the strong correlation between exposure estimates in successive dwellings and the uncertainty about the most relevant exposure period, estimation of overall exposure level from a single address is also informative. Error in dose estimation is likely to cause some degree of classical measurement error resulting in bias towards the null.

在评估环境暴露时，住宅暴露通常最相关。在大多数国家，建立完整的居民历史是不可能的。在最近的出版物中，无论有没有完整的居住史，都已经研究了儿童白血病和背景辐射。本文研究了缺少此类完整数据的后果。

分析了一项关于芬兰儿童白血病和伽玛射线的全国性病例对照研究的数据。其中包括1990-2011年在芬兰被诊断出患有白血病的1093名儿童。每个病例均按性别和出生年份与三个对照相匹配。完整的住宅历史可用。剂量估算基于室外本底辐射测量。获得具有居住类型特定转换系数的室内剂量率，并使用特定于年龄的室内占用率以及孩子的年龄和性别来计算各个时间加权平均红色骨髓剂量率。切尔诺贝利辐射的辐射已包括在内，并假设了2年的潜伏期。

连续住所之间的中位间隔为3.4 km，红色骨髓剂量的中位差为2.9 nSv / h。连续住所的室内红色骨髓剂量率之间的皮尔森相关性是0.62（95％CI 0.60，0.64）。在完整的居住历史中，剂量率增加10 nSv / h的优势比为1.01（95％CI 0.97，1.05）。对于仅基于第一个住所（1.02，95％CI 0.99，1.05）和仅最后一个住所（1.00，95％CI 0.98，1.03）的剂量率和仅居住于单个住所的受试者的剂量率计算出相似的比值比（1.05，95％CI 0.98，1.10）。

完整的居民历史知识将始终是您的选择。但是，由于连续住宅中的暴露估计与最相关的暴露时间之间的不确定性之间具有很强的相关性，因此从一个地址估算总体暴露水平也是有益的。剂量估算中的误差可能会导致某种程度的经典测量误差，从而导致对零值的偏倚。