About the Roussel Uclaf Causality Assessment Method (RUCAM) for Drug- and Herb-Induced Liver Injury
Before scoring anything you have to decide which of two tables you are on, because four of RUCAM's seven domains differ between them. The R ratio does that: R = (ALT ÷ ULN) ÷ (ALP ÷ ULN), with R of 5 or more hepatocellular, 2 or less cholestatic, and in between mixed — cholestatic and mixed share a table. The seven domains are then scored from −3 to +3 each: time to onset, course after stopping, risk factors, concomitant drugs, exclusion of other causes, prior knowledge of the drug's hepatotoxicity, and response to unintentional re-exposure. The total maps to excluded (0 or below), unlikely (1–2), possible (3–5), probable (6–8) or highly probable (9 or more).
Formula
R = (ALT ÷ ULN) ÷ (ALP ÷ ULN) → R ≥ 5 hepatocellular · R ≤ 2 cholestatic · 2 < R < 5 mixed
RUCAM = onset + course + risk factors + age + concomitant drugs + alternative causes + prior knowledge + re-exposure- Domains 1, 2, 3 and 7
- Differ between the hepatocellular and cholestatic/mixed tables in their options, their windows or both.
- Domains 4, 5 and 6
- Identical in both tables: concomitant drugs, search for alternative causes, and previous information on hepatotoxicity.
- Causality bands
- 0 or below excluded · 1–2 unlikely · 3–5 possible · 6–8 probable · 9 or above highly probable.
- Cholestatic and mixed injury share one table. There is no separate mixed table, which surprises people who expect three.
- The cholestatic course domain caps at +2 while the hepatocellular one reaches +3, so an identical clinical story scores one point lower on the cholestatic table. That is a real feature of the instrument, not a rounding artefact.
- Three domains can go negative — concomitant drugs to −3, alternative causes to −3, and re-exposure to −2 — which is how a case gets to 'excluded' rather than merely scoring low.
- The 2016 update added hepatitis E to Group I, making seven causes rather than six. Scoring against the older six-cause list inflates the alternative-causes domain.
- Re-exposure means unintentional re-exposure. Deliberately rechallenging a patient to gain 3 points is not acceptable practice and no guideline endorses it.
- RUCAM measures causality only. Severity and prognosis are assessed separately, with Hy's law — hepatocellular injury with jaundice — as the marker of a poor outcome.
Interpreting the result
Read the category, but read where the points came from first. A total of 6 built on chronology and dechallenge alone, with the alternative-cause work-up incomplete, is a weaker case than the same 6 with viral serology and imaging negative — and the second will hold up to challenge where the first will not. The possible/probable boundary at 5 to 6 is where most real cases sit and where most disagreement happens; if a case is sitting at 5, the highest-yield move is almost always to complete Group I rather than to re-argue the chronology. Note the structural biases before quoting a low score as reassurance: a drug with no published hepatotoxicity cannot score the +2 available in domain 6, and a patient with several plausible concomitant drugs is penalised for a situation that is common and does not by itself exonerate the index drug. Conversely, a high score in a patient on a single well-known hepatotoxin is close to unarguable. RUCAM says nothing about severity — a highly probable case may be trivial and an unlikely one may be in liver failure.
| Score | Band | What it means | Action |
|---|---|---|---|
| ≤ 0 | Excluded | The drug or herb is excluded as a cause of this liver injury | Look elsewhere. A negative total usually means an incompatible chronology or a highly probable alternative cause |
| 1–2 | Unlikely | The drug or herb is an unlikely cause | Continue the search for an alternative explanation; re-scoring rarely changes this band without new information |
| 3–5 | Possible | Causality is possible but not established. The band most real cases start in | Complete the Group I work-up and re-take the drug history including over-the-counter and herbal products — that is what moves the score |
| 6–8 | Probable | The drug or herb is a probable cause | Stop the agent, record it as suspected drug-induced liver injury, and report to pharmacovigilance |
| ≥ 9 | Highly probable | The drug or herb is a highly probable cause | Stop the agent, document lifelong avoidance, and report. Do not rechallenge to confirm |
What the RUCAM needs (9 inputs)
- R ratio inputs — ALT and ALP as multiples of the upper limit of normal
- Optional here, and used only to compute R and cross-check the table you select. R = (ALT ÷ ULN) ÷ (ALP ÷ ULN). Both values must be expressed as multiples of their own upper limit of normal, not as raw laboratory numbers.
- Pattern of injury
- Hepatocellular, or cholestatic/mixed. This selects the scoring table and is the single most consequential input — four of the seven domains have different options and different point values between the two.
- 1. Time to onset
- Scored from the start of the drug, or alternatively from its cessation. The windows differ by pattern: re-exposure onset of 1–15 days scores +2 in hepatocellular injury but the cholestatic window runs to 90 days, and the from-cessation window is 15 days versus 30.
- 2. Course after stopping
- Hepatocellular injury is scored on the fall in ALT, with +3 for a 50% fall within 8 days and +2 within 30. Cholestatic injury is scored on ALP or bilirubin over 180 days, because it resolves far more slowly — the maximum available is +2.
- 3. Risk factors
- Alcohol above 2 drinks daily in women or 3 in men scores +1, and age 55 or over scores +1. In the cholestatic table, alcohol and pregnancy are combined into a single +1 risk factor rather than being scored separately.
- 4. Concomitant drugs or herbs
- Penalties from 0 to −3. Only agents whose own time course fits the injury attract a penalty; a long-standing drug with an incompatible chronology scores 0. A known hepatotoxin with a compatible onset costs −2, and positive evidence for its role costs −3.
- 5. Search for alternative causes
- The domain that moves the score most. Group I holds seven causes — hepatitis A, B, C and E, hepatobiliary obstruction on imaging, alcoholism, and recent hypotension, shock or ischaemia. Group II covers complications of underlying disease plus CMV, EBV and HSV/VZV. Ruling out both groups scores +2; failing to exclude five or more of Group I costs −2.
- 6. Previous information on hepatotoxicity
- +2 if the reaction is labelled in the product characteristics, +1 if published but unlabelled, 0 if unknown. This is where a genuinely novel hepatotoxin is structurally penalised.
- 7. Response to unintentional re-exposure
- +3 for a doubling of ALT (or ALP) on re-exposure to the drug alone, +1 if given alongside the drugs present at the first reaction, −2 for a rise that stays within normal limits, and 0 for anything else including no re-exposure. Unintentional only — deliberate rechallenge to score this item is not acceptable practice.
What it returns
- RUCAM total
- Ranges from −9 to +14 in practice. Most real cases fall between 3 and 9, which is why the possible/probable boundary carries so much weight.
- Causality category
- Excluded, unlikely, possible, probable or highly probable.
- Table used, and the R ratio if supplied
- Reported explicitly, with a warning where the computed R disagrees with the selected table — the commonest way a RUCAM total goes wrong.
How it is calculated
The method came out of international consensus meetings convened to make drug causality assessment reproducible, and its structure reflects that origin: rather than modelling a probability, it enumerates the considerations experts actually use and assigns each a weight. Chronology carries the most positive weight because a compatible time course is the strongest single argument for causality and an incompatible one is close to disqualifying. Dechallenge — what happened after the drug stopped — is scored separately and generously, since improvement on withdrawal is the nearest thing to an experiment that ethical practice allows. The alternative-causes domain is the only one that can both add and subtract substantially, which encodes a specific judgement: a thorough negative work-up is positive evidence for the drug, and an absent work-up is evidence against, not neutral. The companion 1993 paper validated the method against cases with positive rechallenge, using them as an external reference for what a true positive looks like.
Facts & figures
| Domain | Hepatocellular | Cholestatic / mixed |
|---|---|---|
| 1. Time to onset (re-exposure window) | 1–15 days for +2 | 1–90 days for +2 |
| 1. From cessation | Within 15 days for +1 | Within 30 days for +1 |
| 2. Course after stopping | ALT: −50% within 8 days +3, within 30 days +2, max +3 | ALP or bilirubin: −50% within 180 days +2, less than 50% +1, max +2 |
| 3. Risk factors | Alcohol +1; age ≥ 55 +1 | Alcohol OR pregnancy +1 (single point); age ≥ 55 +1 |
| 7. Re-exposure | ALT doubling, ALT < 5×ULN before | ALP doubling, ALP < 2×ULN before |
| 4, 5, 6 | Identical in both tables | Identical in both tables |
Because the cholestatic course domain caps at +2, the same clinical story scores one point lower on that table — enough to move a case from probable to possible. Establishing the pattern from R before scoring is not a formality.
| Group | Causes |
|---|---|
| Group I (7) | Hepatitis A (anti-HAV IgM) · hepatitis B (HBsAg, anti-HBc IgM, HBV DNA) · hepatitis C (anti-HCV, HCV RNA) · hepatitis E (anti-HEV IgM/IgG, HEV RNA) · hepatobiliary obstruction on imaging · alcoholism (AST/ALT ≥ 2) · recent hypotension, shock or ischaemia |
| Group II | Complications of underlying disease · CMV · EBV · HSV or VZV infection |
Both groups ruled out scores +2; all seven Group I causes ruled out scores +1; six or five scores 0; fewer than five costs −2. Hepatitis E was added in the 2016 update and is the one most often omitted — testing for it is what separates a defensible +1 from an undeserved one.
Evidence
Derivation — international consensus meetings (CIOMS)
1993Developed from the conclusions of international consensus meetings convened to standardise causality assessment of adverse drug reactions affecting the liver, producing a domain-scored instrument rather than a fitted model.
Consensus-derived, so it has no C-statistic. Its claim is reproducibility and structure rather than measured discrimination against an objective reference standard, since no such standard exists short of rechallenge.
Companion validation — Danan & Benichou part II
1993The second of the paired 1993 papers, presenting a model for validating drug causality assessment methods using cases with positive rechallenge as the reference.
Used rechallenge-positive cases as an external standard for what a true positive looks like, which is the closest available approximation to a gold standard in this field.
Comparison with structured expert opinion — DILIN
2010Drug-Induced Liver Injury Network study comparing RUCAM with a structured expert opinion process in prospectively enrolled cases.
Found only moderate agreement between RUCAM and expert consensus, with RUCAM tending to assign lower causality categories. The disagreement is a documented limitation of the instrument rather than evidence that either approach is correct.
Update — Danan & Teschke 2016
2016Revision of the original method, clarifying the criteria and extending explicit application to herb-induced liver injury.
Added hepatitis E to the Group I alternative causes, tightened the definitions of several domains, and set out the method for herbal products, where concomitant-product and composition problems are more common than with licensed drugs.
Guideline adoption — ACG 2021 and EASL 2019
2021The ACG clinical guideline on idiosyncratic drug-induced liver injury and the EASL clinical practice guidelines on drug-induced liver injury.
Both identify RUCAM as the causality assessment instrument in routine use, while noting its limitations and that expert opinion remains the practical reference in complex cases.
How it compares
RUCAM vs Structured expert opinion (DILIN process)
Expert opinion remains the practical reference standard and RUCAM tends to score lower — the two agree only moderately, which is a documented limitation rather than a scoring error.
The Drug-Induced Liver Injury Network compared its structured expert opinion process with RUCAM in prospectively enrolled cases and found only moderate agreement, with RUCAM assigning lower causality categories. That matters practically: a case sitting at 'possible' by RUCAM may well be one a hepatology panel would call probable. RUCAM's advantages are that it is reproducible, requires no panel, and can be audited domain by domain, which is why guidelines and regulators use it. Expert opinion handles the situations RUCAM handles badly — novel agents with no published hepatotoxicity, several plausible concomitant drugs, and unusual latency. Where a case is consequential and RUCAM sits at the possible/probable boundary, specialist review is the right next step rather than re-scoring.
RUCAM vs R ratio alone
The R ratio classifies the pattern of injury; it says nothing about whether a drug caused it, and using it as though it did is a category error.
R sorts liver injury into hepatocellular, cholestatic and mixed, which determines the differential diagnosis, the expected time course of recovery, and — within RUCAM — which table applies. It carries no causal information at all: a hepatocellular pattern is equally compatible with viral hepatitis, ischaemia and drug injury. Its role here is as a prerequisite rather than an alternative, and its most common misuse is being computed from raw ALT and ALP values instead of multiples of the upper limit of normal, which changes the ratio and can select the wrong scoring table.
RUCAM vs Hy's law
Complementary and frequently confused — RUCAM asks whether the drug caused the injury, Hy's law asks whether the injury is likely to be dangerous.
Hy's law identifies hepatocellular injury accompanied by jaundice without significant biliary obstruction, a combination associated with a substantial risk of death or transplantation. It is a prognostic observation, not a causality assessment, and it applies whatever the RUCAM total is. The two are used together: RUCAM to establish attribution and support reporting, Hy's law to decide how closely the patient needs watching. A highly probable RUCAM case with normal bilirubin may be clinically trivial, while a case scoring 'possible' that meets Hy's law needs urgent attention.
RUCAM vs Naranjo adverse drug reaction probability scale
Naranjo is a general adverse-drug-reaction instrument; RUCAM is liver-specific and is what hepatology guidelines and regulators expect for suspected DILI.
The Naranjo scale scores any suspected adverse drug reaction across ten generic questions and is widely used in general pharmacovigilance. It has no concept of the hepatocellular versus cholestatic pattern, no liver-specific alternative-cause list, and no dechallenge criteria calibrated to how liver enzymes actually behave — all of which RUCAM encodes. For suspected liver injury, using Naranjo produces a number that is harder to defend and that neither the ACG nor the EASL guidelines reference for this purpose.
Pearls & pitfalls
- Establish the pattern from R before scoring anything. Four of the seven domains differ between the tables, and the cholestatic course domain caps one point lower — enough on its own to move a case from probable to possible.
- R uses multiples of the upper limit of normal, not raw values. Dividing a raw ALT by a raw ALP gives a meaningless ratio and will frequently select the wrong table.
- Cholestatic and mixed injury share one table. There is no third table, and looking for one wastes time.
- Hepatitis E is in Group I since the 2016 update. Omitting it and claiming all Group I causes were excluded overstates that domain by a point.
- 'No information' is not neutral in every domain. An incomplete alternative-cause work-up costs points rather than being ignored, which is deliberate — it prevents a high score from being built on an unexamined case.
- Only concomitant drugs with a compatible time course attract a penalty. A drug the patient has taken for ten years scores 0, not −1.
- Never rechallenge deliberately to score domain 7. It carries real risk, no guideline supports it, and the domain is written for unintentional re-exposure.
- A novel hepatotoxin is structurally penalised. Domain 6 awards nothing when the reaction is unpublished, so a genuinely new agent starts two points behind a well-characterised one.
- RUCAM and expert opinion agree only moderately, and RUCAM tends to score lower. A 'possible' result in a case experts would call probable is a known behaviour of the instrument, not necessarily an error in your scoring.
- Causality is not severity. Assess prognosis separately, and treat hepatocellular injury with jaundice — Hy's law — as the marker of a poor outcome regardless of the RUCAM total.
Critical actions
- Stop the suspected agent while the assessment is being done; do not wait for the score to justify the decision.
- Take a drug history that explicitly asks about over-the-counter medicines, herbal products, traditional remedies, supplements and bodybuilding products — patients do not volunteer these because they do not consider them drugs.
- Complete the Group I work-up in every case: hepatitis A, B, C and E serology, hepatobiliary imaging, an alcohol history, and a review for recent hypotension or ischaemia.
- Compute R from multiples of the upper limit of normal and record the pattern before scoring.
- Document the score domain by domain rather than as a total, so the assessment can be reviewed and re-scored when new results arrive.
- Re-score after the alternative-cause results return — a case scored on admission is nearly always provisional.
- Report probable and highly probable cases to the national pharmacovigilance scheme; RUCAM-assessed reports are what make future assessments possible.
- Record the agent as a suspected cause on the allergy or intolerance list so re-exposure is not prescribed inadvertently.
- Assess severity separately and monitor for Hy's law, which predicts outcome in a way RUCAM does not attempt to.
Why this score exists
The method exists because the pharmaceutical and regulatory worlds needed the same case to be assessed the same way twice. Adverse-reaction reports were being adjudicated by individual judgement, which meant a signal could appear or vanish depending on who read the file, and the consensus meetings that produced RUCAM were an attempt to make that process auditable. Two design decisions follow from that purpose rather than from clinical convenience. Scoring is deliberately conservative — several domains can subtract, and the absence of information is frequently scored as zero rather than being ignored, so an incomplete work-up cannot produce a high score. And the instrument is built around what can be documented rather than what a clinician believes: there is no domain for 'this looks like a drug reaction'. The 2016 update extended the same logic to herbal products, where the authors argued the problem is worse, because the agent's actual composition is often unknown and multiple constituents may be involved.
About the creator
First author, RUCAM derivation and the 2016 update
Led the development of the method from the international consensus meetings, and its subsequent revision.
Co-author, the paired 1993 derivation and validation papers
Co-authored both the original method and the companion paper validating causality assessment against rechallenge-positive cases.
Co-author of the 2016 update
Co-authored the update and much of the subsequent work applying RUCAM to herb-induced liver injury.
Limitations
- Consensus-derived with no measured discrimination against an objective reference standard, because none exists short of rechallenge — which cannot ethically be performed to test the instrument.
- Agrees only moderately with structured expert opinion, and tends to assign lower causality categories than expert panels.
- Structurally penalises novel hepatotoxins: a drug with no published hepatotoxicity cannot earn the +2 available in domain 6, so new agents start behind established ones.
- Penalises polypharmacy, which is common and does not by itself exonerate the index drug — a patient on several plausible agents may be scored down for a situation that is entirely ordinary.
- Depends heavily on the completeness of the alternative-cause work-up, so a busy service that does not test for hepatitis E will systematically under-score.
- The re-exposure domain is unavailable in most cases, since deliberate rechallenge is not acceptable, so 3 of the possible points are usually out of reach.
- Latency windows are fixed and do not accommodate agents with characteristically long or unusual time courses, such as some immune checkpoint inhibitors and certain herbal products.
- Not designed for paracetamol overdose, where the diagnosis rests on history and levels rather than causality scoring.
- Provides no information on severity, prognosis or management.
If you are the patient
RUCAM is a checklist doctors use to work out how likely it is that a medicine, herbal remedy or supplement caused liver damage. It is not a blood test — it is a structured way of weighing up the evidence so that different doctors reach the same conclusion about the same case. Seven things are scored: how long after starting the medicine the problem appeared, whether liver tests improved after stopping it, your age and alcohol intake, whether other medicines could be responsible, whether other causes such as viral hepatitis or a blocked bile duct have been ruled out, whether this medicine is already known to affect the liver, and what happened if you happened to take it again. The points are added up, and the total places the case somewhere between 'excluded' and 'highly probable'. The single most useful thing you can do is give a complete list of everything you take — including things bought without a prescription, herbal products, traditional remedies, vitamins, protein powders and bodybuilding supplements. People often leave these out because they do not think of them as medicines, but several are well-recognised causes of liver injury, and the assessment cannot be accurate without them. If a medicine is found to be the likely cause, it will be stopped and recorded so you are not given it again — and your doctors will not re-test by giving it back to you, because that carries real risk.
Frequently asked questions
What is the RUCAM score?#
A structured instrument for assessing whether a drug or herbal product caused liver injury. Seven domains are scored — time to onset, course after stopping, risk factors, concomitant drugs, exclusion of alternative causes, previous knowledge of hepatotoxicity, and response to unintentional re-exposure — and the total maps to excluded (0 or below), unlikely (1–2), possible (3–5), probable (6–8) or highly probable (9 or more).
How is the R ratio calculated and why does it matter?#
R = (ALT ÷ upper limit of normal) ÷ (ALP ÷ upper limit of normal). R of 5 or more is hepatocellular, 2 or less is cholestatic, and between 2 and 5 is mixed. It matters because it selects which of RUCAM's two scoring tables applies, and four of the seven domains differ between them. Computing R from raw laboratory values instead of multiples of the upper limit of normal is a common error that can select the wrong table.
Is there a separate RUCAM table for mixed liver injury?#
No. Cholestatic and mixed injury share a single table. Only hepatocellular injury has its own, so there are two tables in total rather than three.
What RUCAM score means a drug probably caused the injury?#
Six to eight is 'probable' and nine or above is 'highly probable'. At six or more the usual response is to stop the agent, record it as a suspected cause of drug-induced liver injury, and report it to the national pharmacovigilance scheme. Three to five is 'possible', which is where most real cases start.
Which alternative causes have to be excluded for RUCAM?#
Group I holds seven: hepatitis A, B, C and E, hepatobiliary obstruction on imaging, alcoholism, and recent hypotension, shock or ischaemia. Group II covers complications of the underlying disease plus CMV, EBV and HSV or VZV infection. Excluding both groups scores +2, all seven Group I causes scores +1, and failing to exclude five or more costs −2. Hepatitis E was added in the 2016 update and is the one most often forgotten.
Why does the same case score lower on the cholestatic table?#
Mainly because of the course domain. Hepatocellular injury can score +3 for a 50% fall in ALT within eight days, while the cholestatic table's best available score is +2 for a 50% fall in ALP or bilirubin within 180 days — reflecting that cholestatic injury genuinely resolves more slowly. The time-to-onset windows and the re-exposure criteria also differ. A one-point difference is enough to move a case between possible and probable.
Should you rechallenge a patient to complete the RUCAM score?#
No. Domain 7 is written for unintentional re-exposure — a patient who was inadvertently given the drug again. Deliberate rechallenge carries a real risk of severe recurrent injury, no guideline endorses it, and the three points it might yield do not justify it. In most cases that domain simply scores zero.
How does RUCAM compare with expert opinion?#
They agree only moderately, and RUCAM tends to assign lower causality categories than expert panels. The DILIN comparison documented this directly. RUCAM's advantages are reproducibility and auditability — it can be reviewed domain by domain and does not require a panel. Expert review is better in the situations RUCAM handles poorly: novel agents, several plausible concomitant drugs, and unusual latency.
Does RUCAM tell you how severe the liver injury is?#
No — it assesses causality only. A highly probable case may be biochemically trivial and a case scoring 'possible' may be in liver failure. Severity and prognosis are assessed separately, and Hy's law — hepatocellular injury with jaundice and no significant biliary obstruction — is the standard marker of a poor outcome.
References
Original / primary reference
- Danan G, Benichou C. Causality assessment of adverse reactions to drugs — I. A novel method based on the conclusions of international consensus meetings: application to drug-induced liver injuries. J Clin Epidemiol. 1993;46(11):1323-1330.
- Benichou C, Danan G, Flahault A. Causality assessment of adverse reactions to drugs — II. An original model for validation of drug causality assessment methods: case reports with positive rechallenge. J Clin Epidemiol. 1993;46(11):1331-1336.
- Danan G, Teschke R. RUCAM in drug and herb induced liver injury: the update. Int J Mol Sci. 2016;17(1):14 (source of the updated scoring tables).
Clinical practice guidelines
- Chalasani NP, Maddur H, Russo MW, Wong RJ, Reddy KR; Practice Parameters Committee of the American College of Gastroenterology. ACG Clinical Guideline: Diagnosis and Management of Idiosyncratic Drug-Induced Liver Injury. Am J Gastroenterol. 2021;116(5):878-898.
- European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Drug-induced liver injury. J Hepatol. 2019;70(6):1222-1261.