Section of Urology, University of Arizona Health Sciences Center, Tucson, Arizona 85724-5077, USA.
Urology. 2010 Feb;75(2):266-70. doi: 10.1016/j.urology.2009.09.020. Epub 2009 Nov 13.
To report a weakness in the April 2006 release of the Surveillance, Epidemiology, and End Results (SEER) dataset, in which the primary tumor size of small (< 1.8 cm) metastatic renal cancers was often incorrectly coded into the dataset from the measurement as listed in the patient's chart.
In the SEER dataset, 167 patients with tumor size < or = 2.5 cm had metastatic disease at presentation in 1998-2003. Each patient's chart was individually re-examined by SEER registries to determine the correct primary tumor size. This confirmed data were compared with the coded tumor size in the SEER dataset.
Of the 167 re-examined cases, 2 had incorrect histology and 6 could not be verified. Of the remaining 159 cases, 87 (55%) were correctly coded for primary tumor size while 72 (45%) were incorrect. The error rate decreased with increasing size; for tumors < or = 1 cm, > 1-2 cm, and > 2-2.5 cm, error rates were 88%, 53%, and 6.8%, respectively (P < .001). A breakpoint in error rate occurred between tumor sizes < 1.8 cm (78%) and > or = 1.8 cm (10%) (P < .001). Most errors (72%) were miscoded by a factor of 10. Analysis of the latest April 2009 release suggests that most corrections have been incorporated into the public access dataset.
Coded primary tumor sizes in the April 2006 release SEER dataset for metastatic renal tumors < 1.8 cm from 1998 to 2003 were often inaccurate. Verification of tumor size in this subset was essential to insure data accuracy and quality of research. Researchers should recognize potential limitations of population-based cancer registries.
报告 2006 年 4 月发布的监测、流行病学和最终结果（SEER）数据集的一个缺陷，在该数据集中，小（<1.8cm）转移性肾肿瘤的原发肿瘤大小经常被错误地编码为患者图表中列出的测量值。
在 SEER 数据集中，1998-2003 年间有 167 名肿瘤大小<或=2.5cm 的患者在初次就诊时就出现了转移性疾病。SEER 登记处单独重新检查了每位患者的图表，以确定正确的原发肿瘤大小。将这些确认数据与 SEER 数据集中编码的肿瘤大小进行比较。
在 167 例重新检查的病例中，有 2 例存在错误的组织学，有 6 例无法核实。在其余 159 例中，87 例（55%）原发肿瘤大小编码正确，而 72 例（45%）编码错误。错误率随肿瘤大小的增加而降低；肿瘤大小<或=1cm、>1-2cm 和>2-2.5cm 的错误率分别为 88%、53%和 6.8%（P<0.001）。在肿瘤大小<1.8cm（78%）和>或=1.8cm（10%）之间，错误率出现了一个转折点（P<0.001）。大多数错误（72%）的编码错误为 10 倍。对 2009 年 4 月最新发布的分析表明，大多数错误已被纳入公共访问数据集。
1998 年至 2003 年间，转移性肾肿瘤<1.8cm 的 2006 年 4 月 SEER 数据集的编码原发肿瘤大小往往不准确。对这一小部分肿瘤大小进行验证对于确保数据的准确性和研究质量至关重要。研究人员应认识到人群癌症登记处的潜在局限性。
