Roudgari Hassan, Masson Lindsey F, Haites Neva E
Department of Medicine & Therapeutics, College of Life Sciences & Medicine, Polwarth Building, Foresterhill, Aberdeen, UK.
Fam Cancer. 2007;6(4):415-43. doi: 10.1007/s10689-007-9140-1. Epub 2007 May 23.
Drawing an informative pedigree is fundamental in genetic counselling. It is very common for some parts of pedigrees to remain ambiguous because of the proband's inability to recall the past history of her/his family. Current age, date of birth, date of death and age of diagnosis are the commonest missing information in pedigrees.
The Scottish Social Statistics website, National Statistics website and English language literature were used to model extrapolations. About 172 Grampian families and three high-risk Grampian families with complete information were chosen to evaluate the influence of extrapolations on models' performance. Differences between original data and extrapolated data were assessed by independent samples t-test.
Changes made by extrapolations in age- and cancer-related information were not statistically significant (P > 0.05) in comparison with original data, except for average age of diagnosis of breast cancer (P = 0.03). The differences made by extrapolations in estimated probabilities generated by probability assessment models were small and ignorable except that for Tyrer-Cuzick model for Grampian family 3.
Extrapolations based on National Health Statistics can scientifically cover missing information in a defined population with minimum effect on performance of probability assessment models.
绘制一份信息丰富的家系图是遗传咨询的基础。由于先证者无法回忆起其家族的既往史，家系图的某些部分仍不明确的情况很常见。当前年龄、出生日期、死亡日期和诊断年龄是家系图中最常见的缺失信息。
利用苏格兰社会统计网站、国家统计网站和英文文献进行模型推断。选择了约172个格兰扁地区的家庭以及三个信息完整的格兰扁地区高危家庭，以评估推断对模型性能的影响。通过独立样本t检验评估原始数据和推断数据之间的差异。
与原始数据相比，推断在年龄和癌症相关信息方面所做的更改无统计学意义（P>0.05），乳腺癌诊断平均年龄除外（P=0.03）。除格兰扁地区家庭3的泰勒-库齐克模型外，推断在概率评估模型生成的估计概率方面所产生的差异较小且可忽略不计。
基于国家卫生统计数据的推断能够科学地填补特定人群中的缺失信息，且对概率评估模型性能的影响最小。
