Giard R W
Medisch Centrum Rijnmond-Zuid, locatie St. Clara, afd. Klinische Pathologie, Postbus 9119, 3007 AC Rotterdam.
Ned Tijdschr Geneeskd. 2004 Feb 21;148(8):352-5.
Descriptive studies are important for the analysis of the effects of medical interventions. Especially in breast cancer screening, where the balance between gains and losses is often difficult to establish, detailed, empirical, long-term follow-up data on outcomes are of vital importance. Recent Dutch data comparing the prognosis, tumour stage at diagnosis and surgical treatment before and after the introduction of a population-based breast cancer screening programme reveal a better prognosis and a more favourable tumour stage after introduction of the programme. However, these data have their limitations. The interpretation of such figures is hampered by different types of bias (lead-time, length and over-diagnosis). The final assessment is always subjective and prone to confirmation bias.
描述性研究对于分析医学干预措施的效果很重要。特别是在乳腺癌筛查中，收益与损失之间的平衡往往难以确定，关于结局的详细、实证性、长期随访数据至关重要。荷兰最近的数据比较了基于人群的乳腺癌筛查计划实施前后的预后、诊断时的肿瘤分期和手术治疗情况，结果显示该计划实施后预后更好，肿瘤分期更有利。然而，这些数据有其局限性。此类数据的解读受到不同类型偏差（领先时间、病程和过度诊断）的阻碍。最终评估始终是主观的，且容易出现确认偏差。
