中国肝脏病杂志(电子版)

位置：首页 >> 期刊文章

RECAM 量表和 RUCAM 量表在药物性肝损伤诊断中的应用

作者： style="font-size: 12px ">李勤荣海珊徐智媛杨晋辉

单位：昆明医科大学第二附属医院消化内科云南昆明 650033

关键词：药物性肝损伤中草药诊断因果关系评估电子因果关系评估方法鲁塞尔-尤克拉夫因果关系评估法

分类号：

出版年,卷(期):页码：2026,18(1):16-24

摘要：

摘要：目的探讨电子因果关系评估方法（revised electronic causality assessment method，

RECAM）与鲁塞尔-尤克拉夫因果关系评估法（Roussel-Uclaf causality assessment

method，RUCAM）在药物性肝损伤（drug-induced liver injury，DILI）因果关系评估及

诊断性能中的一致性与差异。方法通过医院电子病历系统，以“药物性肝炎”、“药物

性肝损害”、“药物性肝衰竭”及“药物性肝炎伴胆汁淤积”等为关键词，检索 2021 年

1 月 1 日至 2023 年 12 月 31 日在昆明医科大学第二附属医院消化内科住院的所有患者。

研究初始共检索出 466 例带有 DILI 相关诊断的患者，随后依据纳入与排除标准进行筛

选，最终纳入 244 例确诊 DILI 患者。根据靶细胞受损类型将 DILI 患者分为肝细胞损

伤型（130 例）、胆汁淤积型和混合型（114 例）；根据患者发病前的可疑肝毒性药物将

患者分为中草药及中成药组（116 例）和其他组（128 例）。对患者分别采用 RECAM

与 RUCAM 量表进行评估。比较两量表评估结果（极可能、很可能、可能、不可能）

的差异以及与实际诊断的一致性（符合、疑似符合、不符合）。结果 ①所有 DILI 患

者的量表评估结果显示，“高度可能”（极可能+很可能）占比最高，其中 RECAM 量

表“极可能”占比最高（49.6%），RUCAM 量表“很可能”占比最高（52.5%），两量

表评估结果分布存在显著差异（z = -12.002，P ＜ 0.001）。RECAM 量表评估结果与实

际诊断符合、疑似符合、不符合的例数分别为 206 例、35 例、3 例，RUCAM 量表分

别为 142 例、87 例、15 例，两种量表评估结果与实际诊断的符合情况分布存在统计学

差异（χ2 = 76.605，P ＜ 0.001）。两量表诊断 DILI 一致性的加权 Kappa 系数为 0.333

（95%CI：0.236～0.429，P ＜ 0.001）。②对于肝细胞损伤型 DILI 患者，RECAM 量表

评估结果中“高度可能”比例高于 RUCAM 量表（91.5% 比 69.3%），而 RUCAM 量

表评估结果中“可能”比例高于 RECAM 量表（25.4% 比 7.0%），两种量表评估结果

分布的差异有统计学意义（z = -8.416，P ＜ 0.001）。RECAM 量表评估结果与实际诊

断符合、疑似符合、不符合的例数分别为 119 例、9 例、2 例，RUCAM 量表分别为

94 例、33 例、3 例，两种量表评估结果与实际诊断的符合情况分布存在统计学差异

（χ2 = 26.635，P ＜ 0.001）。两量表诊断 DILI 一致性的加权 Kappa 系数为 0.302 （95%CI：

0.108～0.495，P ＜ 0.001）。③对于胆汁淤积型和混合型 DILI 患者，RECAM 量表评

估结果中“高度可能”的比例高于 RUCAM 量表（76.3% 比 42.1%），两种量表评估

结果分布的差异有统计学意义（z = -8.591，P ＜ 0.001）。RECAM 量表评估结果与实

际诊断符合、疑似符合、不符合的例数分别为 87 例、26 例、1 例，RUCAM 量表分

别为 48 例、54 例、12 例，两种量表评估结果与实际诊断的符合情况分布存在统计学

差异（χ2 = 43.845，P ＜ 0.001）。两量表诊断 DILI 一致性的加权 Kappa 系数为 0.302

（95%CI：0.195～0.410，P ＜ 0.001）。④对于中草药及中成药所致 DILI 患者，RECAM

量表评估结果中“高度可能”比例高于 RUCAM 量表（83.6% 比 44.0%），两种量表评

估结果分布的差异有统计学意义（z = -8.686，P ＜ 0.001）。RECAM 量表评估结果与

实际诊断符合、疑似符合、不符合的例数分别为 97 例、18 例、1 例，RUCAM 量表分别为 51 例、57 例、8 例，两种量表评估结果与实际诊断的符合情况分布存在统计学

差异（χ2 = 37.745，P ＜ 0.001）。两量表诊断 DILI 一致性的加权 Kappa 系数为 0.259

（95%CI：0.153～0.366，P ＜ 0.001）。 ⑤ 对于其他药物所致 DILI 患者，RECAM 量

表评估结果中“极可能”分类结果的比例高于 RUCAM 量表（59.4% 比 10.9%），而

RUCAM 量表评估结果中“很可能”占比高于 RECAM 量表（60.2% 比 25.8%）。两种

量表评估结果分布的差异有统计学意义（z = -8.273，P ＜ 0.001）。RECAM 量表评估

结果与实际诊断符合、疑似符合、不符合的例数分别为 109 例、17 例、2 例，RUCAM

量表分别为 91 例、30 例、7 例，两种量表评估结果与实际诊断的符合情况分布存在统

计学差异（χ2 = 41.632，P ＜ 0.001）。两量表诊断 DILI 一致性的加权 Kappa 系数为 0.424

（95%CI：0.264～0.584，P ＜ 0.001）。结论 RECAM 量表与实际诊断的符合程度更高。

对于中草药及中成药制剂所致 DILI 患者，RECAM 量表的诊断效力优于 RUCAM 量表，

RECAM 量表可能为此类患者提供更高的诊断精确度。

Abstract: Objective To investigate the consistency and differences between the revised

electronic causality assessment method (RECAM) and the Roussel-Uclaf causality assessment

method (RUCAM) in causality assessment and diagnostic performance for drug-induced liver

injury (DILI). Methods Through the hospital electronic medical record system, a retrieval

was conducted for all patients admitted to the Department of Gastroenterology of the Second

Affiliated Hospital of Kunming Medical University from January 1st, 2021 to December

31st, 2023, with keywords including drug-induced hepatitis, drug-induced liver injury, druginduced liver failure and drug-induced hepatitis with cholestasis. A total of 466 patients

with DILI-related diagnoses were initially retrieved. Screening was subsequently performed

in accordance with pre-specified inclusion and exclusion criteria, and 244 patients with

confirmed DILI were finally enrolled in this study. DILI patients were classified into three

categories according to the type of target cell injury: hepatocellular injury type (130 cases),

cholestatic type and mixed type (114 cases). Patients were further divided into Chinese herbal

medicine and Chinese patent medicine group (116 cases) and other drug group (128 cases)

based on suspected hepatotoxic drug exposure before disease onset. All enrolled patients were

assessed using the RECAM scale and RUCAM scale respectively. Differences in assessment

results (definite, highly probable, probable, unlikely) between the two scales, as well as their

consistency with the actual clinical diagnosis (consistent, suspected consistent, inconsistent)

were compared. Results ① The scale assessment results of all DILI patients showed that the

proportion of the “highly likelihood” (sum of “probable” and “highly probable” grades) was

the highest. Among them, the “highly probable” grade accounted for the largest proportion

(49.6%) in RECAM scale assessment results, while the “probable” grade accounted for the

largest proportion (52.5%) in RUCAM scale assessment results. A significant difference

was observed in the distribution of assessment results between the two scales (z = -12.002,

P ＜ 0.001). For consistency with the actual clinical diagnosis, the number of cases with

consistent, suspected consistent and inconsistent results was 206, 35 and 3 respectively for the

RECAM scale, and 142, 87 and 15 respectively for the RUCAM scale. A statistically significant difference was detected in the diagnostic consistency distribution between the two scales

(χ2

= 76.605, P ＜ 0.001). The weighted Kappa coefficient for DILI diagnostic consistency

between the two scales was 0.333 (95%CI: 0.236～0.429, P ＜ 0.001). ② For patients with

hepatocellular injury-type DILI, the proportion of “highly likelihood” cases in the RECAM

scale assessment was higher than that in the RUCAM scale (91.5% vs. 69.3%), whereas

the proportion of “probable” cases in RUCAM scale was higher than that in the RECAM

scale (25.4% vs. 7.0%). The difference in the distribution of assessment results between the

two scales was statistically significant (z = -8.416, P ＜ 0.001). The number of cases with

RECAM assessment results were consistent, suspected to be consistent and inconsistent with

the actual diagnosis were 119, 9 and 2, respectively; while the corresponding numbers for the

RUCAM scale were 94, 33 and 3, respectively. The distribution of consistency between the

assessment results of the two scales and the actual diagnosis showed a statistically significant

difference (χ2

= 26.635, P ＜ 0.001). The weighted Kappa coefficient for the consistency

of DILI diagnosis between the two scales was 0.302 (95%CI: 0.108～0.495, P ＜ 0.001).

③ For patients with cholestatic and mixed type DILI, the proportion of “highly likelihood”

in RECAM scale assessment was higher than that in RUCAM scale (76.3% vs. 42.1%). The

difference in the distribution of assessment results between the two scales was statistically

significant (z = -8.591, P ＜ 0.001). The number of cases with RECAM assessment results

were consistent, suspected to be consistent and inconsistent with the actual diagnosis were

87, 26 and 1, respectively, while the corresponding numbers for the RUCAM scale were 48,

54 and 12, respectively. The distribution of consistency between the assessment results of the

two scales and the actual diagnosis showed a statistically significant difference (χ2

= 43.845,

P ＜ 0.001). The weighted Kappa coefficient for the consistency of DILI diagnosis between

the two scales was 0.302 (95%CI: 0.195～0.410, P ＜ 0.001). ④ For patients with DILI

induced by traditional Chinese herbs and proprietary Chinese medicines, the proportion of

“highly likelihood” cases in RECAM scale assessment was higher than that in the RUCAM

scale (83.6% vs. 44.0%). The difference in the distribution of assessment results between the

two scales was statistically significant (z = -8.686, P ＜ 0.001). The number of cases with

RECAM assessment results were consistent, suspected to be consistent and inconsistent with

the actual diagnosis were 97, 18 and 1, respectively, while the corresponding numbers for the

RUCAM scale were 51, 57 and 8, respectively. The distribution of consistency between the

assessment results of the two scales and the actual diagnosis showed a statistically significant

difference (χ2

= 37.745, P ＜ 0.001). The weighted Kappa coefficient for the consistency of

DILI diagnosis between the two scales was 0.259 (95%CI: 0.153～0.366, P ＜ 0.001). ⑤ For

patients with DILI induced by other drugs, the proportion of “highly probable” classification

results in RECAM scale assessment was higher than that in RUCAM scale (59.4% vs. 10.9%),

whereas the proportion of “probable” cases in RUCAM scale was higher than that in RECAM

scale (60.2% vs. 25.8%). The difference in the distribution of assessment results between the

two scales was statistically significant (z = -8.273, P ＜ 0.001). The number of cases with

RECAM assessment results were consistent, suspected to be consistent and inconsistent with

the actual diagnosis were 109, 17 and 2, respectively, while the corresponding numbers for the

RUCAM scale were 91, 30 and 7, respectively. The distribution of consistency between the

assessment results of the two scales and the actual diagnosis showed a statistically significant

difference (χ2

= 41.632, P ＜ 0.001). The weighted Kappa coefficient for the consistency

of DILI diagnosis between the two scales was 0.424 (95%CI: 0.264～0.584, P ＜ 0.001). Conclusions The RECAM scale had a higher degree of conformity in practical diagnosis. For

patients with liver injury caused by Chinese herbal and proprietary medicines, the diagnostic

efficacy of RECAM scale was better than that of RUCAM scale. The RECAM scale might

provide higher diagnostic accuracy for such patients.

基金项目：

作者简介：

参考文献：

服务与反馈：

【文章下载】【加入收藏】