Møller Marie, Bressendorff Iain, Borg Rikke, Dieperink Hans, Gregersen Jon W, Hansen Helle, Hommel Kristine, Hornum Mads, Ivarsen Per, Jensen Karina H, Jørgensen Morten B, Kristensen Tilde, Krustrup Dorrit, Mose Frank H, Rossing Peter, Otte Kjeld E, Persson Frederik, Schandorff Kristine D, Hansen Ditte
Department of Nephrology, Copenhagen University Hospital Herlev-Gentofte, Herlev, Denmark.
Department of Medicine, Zealand University Hospital, Roskilde, Denmark.
Clin Kidney J. 2024 Jul 9;17(8):sfae203. doi: 10.1093/ckj/sfae203. eCollection 2024 Aug.
This study validates the application of Systematized Nomenclature of Medicine second edition (SNOMED II) codes used to describe medical kidney biopsies in Denmark in encoded form, aiming to support robust epidemiological research on the causes, treatments and prognosis of kidney diseases.
Kidney biopsy reports from 1 January 1998 to 31 December 2018 were randomly extracted from the Danish National Patobank, using SNOMED codes. A 5% sample was selected, and nephrologists assessed the corresponding medical records, assigning each case the applied clinical diagnoses. Sensitivity, specificity, positive predictive values (PPV), negative predictive values and Cohen's kappa coefficient for the retrieved SNOMED codes were calculated.
A total of 613 kidney biopsies were included. The primary clinical disease groups were glomerular disease ( = 368), tubulointerstitial disease ( = 67), renal vascular disease ( = 51), diabetic nephropathy ( = 51) and various renal disorders ( = 40). Several SNOMED codes were used to describe each clinical disease group and PPV for the combined SNOMED codes were high for glomerular disease (94%), diabetic nephropathy (85%) and systemic diseases affecting the kidney (96%). Conversely, tubulointerstitial disease (62%), renal vascular disease (60%) and other renal disorders (17%) showed lower PPV.
SNOMED codes have a high PPV for glomerular diseases, diabetic nephropathy and systemic diseases affecting the kidney, in which they could be applied for future epidemiological research.
本研究验证了以编码形式用于描述丹麦肾脏活检的医学系统命名法第二版（SNOMED II）代码的应用，旨在支持对肾脏疾病的病因、治疗和预后进行有力的流行病学研究。
使用SNOMED代码从丹麦国家病理库中随机提取1998年1月1日至2018年12月31日的肾脏活检报告。选取5%的样本，肾脏病专家评估相应的病历，为每个病例指定应用的临床诊断。计算检索到的SNOMED代码的敏感性、特异性、阳性预测值（PPV）、阴性预测值和科恩kappa系数。
共纳入613例肾脏活检。主要临床疾病组为肾小球疾病（n = 368）、肾小管间质疾病（n = 67）、肾血管疾病（n = 51）、糖尿病肾病（n = 51）和各种肾脏疾病（n = 40）。每个临床疾病组使用了多个SNOMED代码，肾小球疾病（94%）、糖尿病肾病（85%）和影响肾脏的全身性疾病（96%）的联合SNOMED代码的PPV较高。相反，肾小管间质疾病（62%）、肾血管疾病（60%）和其他肾脏疾病（17%）的PPV较低。
SNOMED代码对肾小球疾病、糖尿病肾病和影响肾脏的全身性疾病具有较高的PPV，可用于未来的流行病学研究。
