Source 出处 · Shanghan Lun · Doses are not given here: they are set by a licensed practitioner for the individual.
Safety. The indication is three characters of symptom - 身必黄, the body will turn yellow - and four characters of mechanism, 瘀热在里. Shanghan Lun gives no more than that. A modern reader has one obligation before anything else: jaundice is a sign, and the first question about it is not which pattern it belongs to but which of three compartments it comes from.
The split is made by a blood test that nearly every clinic can run. Unconjugated or indirect hyperbilirubinaemia with otherwise normal liver enzymes points at haemolysis or at Gilbert syndrome: haemolysis is checked with reticulocytes, LDH, haptoglobin and a direct antiglobulin test, and Gilbert is a benign, common condition in which the bilirubin rises with fasting, illness, exertion or poor sleep and falls again. Conjugated hyperbilirubinaemia with a raised alkaline phosphatase and GGT, dark urine, pale stool and itching, points at obstruction: common bile duct stone, pancreatic head cancer, cholangiocarcinoma, ampullary tumour, primary biliary cholangitis, primary sclerosing cholangitis, postoperative or inflammatory stricture, Mirizzi syndrome, and parasitic infestation. Between them sits hepatocellular injury with a transaminase-predominant picture: viral hepatitis A through E, alcoholic liver disease, drug-induced liver injury, autoimmune hepatitis, Wilson disease, haemochromatosis, alpha-1 antitrypsin deficiency, ischaemic hepatitis from sepsis or hypoperfusion, and the pregnancy-specific conditions.
Three features change the timing, and they are worth stating in plain words. Jaundice with fever and rigors and right upper quadrant pain is acute cholangitis - Charcot's triad - and with hypotension or altered mental state it becomes Reynolds' pentad; it is drained and covered with antibiotics, and antibiotics do not substitute for drainage. Painless jaundice is not the mild version; it is the classic presentation of malignant obstruction and is the reason painless is a word that should worry rather than reassure. Jaundice in a pregnancy is always obstetric business, and pre-eclampsia with liver involvement and HELLP syndrome are in that list. In a newborn, jaundice appearing in the first twenty-four hours, or any direct hyperbilirubinaemia, or jaundice persisting beyond two weeks, is biliary atresia until it is excluded, and the window for the Kasai procedure is measured in weeks - that single item is the most time-critical thing in this entire entry.
Drug-induced liver injury deserves its own sentence here, because herbal medicines and dietary supplements are a recognised and frequently under-reported cause of it, including in people who take them precisely because they believe them to be gentle. Any jaundice that appears during the taking of any medicine or supplement is a reason to stop and to be assessed, not a reason to continue.
On the materials. 赤小豆 is the adzuki bean, Vigna angularis or V. umbellata, a food legume. It is not 相思子, Abrus precatorius, the rosary pea or jequirity bean, whose seeds contain abrin, a toxin of extreme potency, and which is sold worldwide as beads for bracelets and rosaries; confusion between the two, and childhood ingestion of the beads, are real poisoning pathways. 杏仁, apricot kernel, contains amygdalin, a cyanogenic glycoside that releases hydrogen cyanide on hydrolysis; bitter apricot kernel and sweet apricot kernel are not the same material, and children are the group at risk. 连轺 is the root of Forsythia; later practice generally uses 连翘, the fruit, and this library records the classical ingredient without adjudicating between them. 生梓白皮 is the fresh root bark of Catalpa; it is largely unavailable and is substituted in practice, and this library declines to recommend a substitute. Ephedra carries the sympathomimetic cautions recorded under 越婢汤, licorice the glycyrrhizin cautions, and jujube is a sugar load. There are no adequate safety data for pregnancy or breastfeeding and pregnancy is an absolute exclusion.
Chinese clinical reports on this formula exist, chiefly in jaundice and dermatological contexts, small in scale and limited in method. This library records that the research exists, not that benefit has been established - and it records that jaundice is a laboratory question before it is a pattern question.
主治里症状只有三个字 —— 身必黄，病机四个字 —— 瘀热在里。《伤寒论》没有多给。今天的读者在其它一切之前有一件事必须做：黄疸是一个体征，关于它的第一个问题不是它属于哪个证，而是它来自三个隔间中的哪一个。
这个区分靠一项几乎每家诊所都能做的血检。以间接胆红素（未结合）升高为主、肝酶其余正常，指向溶血或 Gilbert 综合征：溶血查网织红细胞、LDH、结合珠蛋白、直接抗人球蛋白试验；Gilbert 是良性且常见的状态，胆红素在饥饿、生病、劳累、睡不好时升高，之后自行回落。以直接胆红素（结合）升高为主、伴碱性磷酸酶与 GGT 升高、尿色深、大便色浅、皮肤瘙痒，指向梗阻：胆总管结石、胰头癌、胆管癌、壶腹周围癌、原发性胆汁性胆管炎、原发性硬化性胆管炎、术后或炎症性狭窄、Mirizzi 综合征、寄生虫感染。两者之间是肝细胞性损伤、以转氨酶升高为主的一类：甲至戊型肝炎病毒、酒精性肝病、药物性肝损伤、自身免疫性肝炎、Wilson 病、血色病、α1 抗胰蛋白酶缺乏、脓毒症或低灌注造成的缺血性肝炎，以及妊娠特有的那几种。
有三个特征会改变时间，值得直说：**黄疸＋发热寒战＋右上腹痛＝急性胆管炎**（Charcot 三联），若再加上低血压或意识改变就是 Reynolds 五联 —— 它要靠引流并用抗生素覆盖，**抗生素不替代引流**；**无痛性黄疸不是轻的那一种**，它是恶性梗阻的经典表现，「无痛」这个词应当让人警觉而不是放心；**妊娠期的黄疸一律是产科的事**，子痫前期肝受累与 HELLP 综合征都在清单里。而在新生儿身上：出生 24 小时内出现的黄疸、任何直接胆红素升高、或持续超过两周的黄疸，在排除之前是胆道闭锁，而 Kasai 手术的时间窗以周计 —— 这一条是整篇里时间最紧的一项。
药物性肝损伤值得单写一句：草药与膳食补充剂是它公认且常被漏报的一类原因，其中不少恰恰是「因为觉得温和」才服用的人。服用任何药物或补充剂期间出现的黄疸，是停下并去就诊的理由，不是继续服用的理由。
药物本身：赤小豆是豆科豇豆属的食用豆（Vigna angularis／umbellata）。它**不是相思子**（鸡母珠、Abrus precatorius）—— 后者种子含鸡母珠毒蛋白 abrin，毒性极强，全球作为手串与念珠的珠子出售；**把这两样搞混，以及儿童误食珠子，是真实的中毒路径**。杏仁含苦杏仁苷，属氰苷，水解后释放氰化氢；苦杏仁与甜杏仁不是同一物料，儿童是危险人群。连轺是连翘的根，后世多通用连翘（果实）；本库记古典药味，不裁决二者异同。生梓白皮是梓树（Catalpa）的鲜根皮，今多缺用、临床常代；本库不推荐代用品。麻黄带越婢汤条目里写明的拟交感胺各项禁忌，甘草带甘草酸禁忌，大枣是糖负荷。孕期与哺乳期没有充分的人体安全性数据，孕期为绝对排除项。
中文文献里这方多用于黄疸与皮肤科场景，规模小、方法学质量有限；本库如实记「研究存在」，不记「疗效已被确立」—— 并且记下：黄疸在成为辨证问题之前，先是一项化验问题。