Severe Acute Liver Injury After Hepatotoxic Medication Initiation in Real-World Data

Key Points

Question  What are the most hepatotoxic medications based on real-world rates of severe acute liver injury and how do these rates compare with hepatotoxicity categorization based on published case reports?

Findings  This series of cohort studies among 7 899 888 persons without liver or biliary disease who initiated any of 194 suspected hepatotoxic medications in the outpatient setting from 2000 to 2021 found that 17 medications had rates of severe acute liver injury at 5.0 or more events per 10 000 person-years, representing the most potentially hepatotoxic medications; 11 medications (64%) were not included in the highest hepatotoxicity category by case reports.

Meaning  Because case reports of medication hepatotoxicity do not consider the number of persons exposed, safety signals derived from case reports should be investigated using epidemiologic data.

Importance  Current approaches to classify the hepatotoxic potential of medications are based on cumulative case reports of acute liver injury (ALI), which do not consider the size of the exposed population. There is little evidence from real-world data (data relating to patient health status and/or the delivery of health care routinely collected from sources outside of a research setting) on incidence rates of severe ALI after initiation of medications, accounting for duration of exposure.

Objective  To identify the most potentially hepatotoxic medications based on real-world incidence rates of severe ALI and to examine how these rates compare with categorization based on case reports.

Design, Setting, and Participants  This series of cohort studies obtained data from the US Department of Veterans Affairs on persons without preexisting liver or biliary disease who initiated a suspected hepatotoxic medication in the outpatient setting between October 1, 2000, and September 30, 2021. Data were analyzed from June 2020 to November 2023.

Exposures  Outpatient initiation of any one of 194 medications with 4 or more published reports of hepatotoxicity.

Main Outcomes and Measures  Hospitalization for severe ALI, defined by either inpatient: (1) alanine aminotransferase level greater than 120 U/L plus total bilirubin level greater than 2.0 mg/dL or (2) international normalized ratio of 1.5 or higher plus total bilirubin level greater than 2.0 mg/dL recorded within the first 2 days of admission. Acute or chronic liver or biliary disease diagnosis recorded during follow-up or as a discharge diagnosis of a hospitalization for severe ALI resulted in censoring. This study calculated age- and sex-adjusted incidence rates of severe ALI and compared observed rates with hepatotoxicity categories based on cumulative published case reports.

Results  The study included 7 899 888 patients across 194 medication cohorts (mean [SD] age, 64.4 [16.4] years, 7 305 558 males [92.5%], 4 354 136 individuals [55.1%] had polypharmacy). Incidence rates of severe ALI ranged from 0 events per 10 000 person-years (candesartan, minocycline) to 86.4 events per 10 000 person-years (stavudine). Seven medications (stavudine, erlotinib, lenalidomide or thalidomide, chlorpromazine, metronidazole, prochlorperazine, and isoniazid) exhibited rates of 10.0 or more events per 10 000 person-years, and 10 (moxifloxacin, azathioprine, levofloxacin, clarithromycin, ketoconazole, fluconazole, captopril, amoxicillin-clavulanate, sulfamethoxazole-trimethoprim, and ciprofloxacin) had rates between 5.0 and 9.9 events per 10 000 person-years. Of these 17 medications with the highest observed rates of severe ALI, 11 (64%) were not included in the highest hepatotoxicity category when based on case reports.

Conclusions and Relevance  In this study, incidence rates of severe ALI using real-world data identified the most potentially hepatotoxic medications and can serve as a tool to investigate hepatotoxicity safety signals obtained from case reports. Case report counts did not accurately reflect the observed rates of severe ALI after medication initiation.

Drug-induced acute liver injury (ALI) is the most common cause of acute liver failure in the US and Europe^1-3 and historically was a frequent reason for withdrawal of approved drugs from 1975 to 2007.⁴ Despite its clinical importance, no systematic approach to classify hepatotoxicity exists, to our knowledge. Researchers have used the number of published case reports of medication hepatotoxicity listed on the US National Institutes of Health LiverTox website⁵ to create categories of likelihood for medications to cause severe ALI (category A [well-known cause], ≥50 cases; category B [highly likely], 12-49 cases; category C [probable], 4-11 cases; and category D [possible], 1-3 cases).⁶ However, cases of drug-induced ALI are frequently underreported.^7,8 Moreover, categorizing hepatotoxic drugs using case reports does not consider the number of individuals exposed and may not accurately reflect incidence of severe ALI.

Assessing incidence rates of drug-induced ALI following medication initiation within real-world data (data relating to patient health status and/or the delivery of health care routinely collected from sources outside of a research setting) offers an independent means of categorizing the potential hepatotoxicity of medications. Because drug-induced ALI is challenging to confirm in clinical practice, remains a diagnosis of exclusion, and is susceptible to misclassification,⁷ researchers could evaluate severe ALI based on laboratory measures of substantial hepatic injury after systematically excluding non–drug-related causes. This approach provides a more objective screen for hepatotoxicity. Cohorts of initiators of suspected hepatotoxic drugs can be created, with follow-up censored at incident non–drug-related liver or biliary disease diagnoses to avoid capturing events unrelated to medication exposure. Rates of severe ALI could be measured in large cohorts, allowing investigation of hepatotoxicity safety signals from post-marketing adverse event reports. Yet, real-world evidence on rates of severe ALI following initiation of suspected hepatotoxic medications remains unknown.

This study evaluated rates of hospitalization for severe ALI after outpatient initiation of suspected hepatotoxic medications. Drugs with at least 4 published reports of hepatotoxicity were considered suspected hepatotoxins to focus analyses on the most commonly implicated products.⁶ To increase the likelihood that severe ALI events were medication related, rates were assessed after censoring for liver diseases, biliary diseases, or other conditions precipitating findings of ALI. Finally, we assessed whether observed rates of severe ALI reflected existing categories of hepatotoxicity based on the number of published case reports.

This study conducted cohort studies of initiators of suspected hepatotoxic medications within the US Department of Veterans Affairs (VA) using data between October 1, 1999, and September 30, 2021. The VA comprises more than 1200 points of care nationwide, including hospitals and outpatient clinics. The VA electronic health record data include demographic characteristics, outpatient and hospital diagnoses, laboratory results, and dispensed drugs.⁹ Race and ethnicity were not reported in this study because they should not alter the rate of severe ALI. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.¹⁰ This study was approved by the institutional review boards of the VA Connecticut Health System, Yale University, and the University of Pennsylvania and was granted a waiver of informed consent as authorized by 45 CRF §46.116(d).

Patients were eligible for inclusion in a medication initiator cohort if they had (1) a dispensed outpatient fill for any of 220 suspected hepatotoxic medications (defined by ≥4 published reports of hepatotoxicity⁶ [eTable 1 in Supplement 1]) between October 1, 2000, and September 30, 2021; and (2) 365 days or longer prior to the dispensed fill without a prior fill for that medication. We focused on outpatient medication initiation to increase the likelihood that severe ALI events were drug related; inclusion of medication initiators during hospitalization would increase the potential that severe ALI might be due to hospital-related events. A medication was not evaluated if it was (1) not dispensed in the VA during the study period; (2) administered via injection or intravenous route (excluding chemotherapy and hormone therapy); (3) used for alcohol use disorder or liver disease treatment; or (4) an anticoagulant, which would prevent identification of ALI-related coagulopathy. We did not evaluate topical, otic, ophthalmic, subdermal, rectal, or vaginal medications.

The index date was the date the medication was initially dispensed in the outpatient setting. For patients who were dispensed multiple courses, only the first was evaluated. Patients could be included in more than 1 cohort if dispensed multiple suspected hepatotoxic medications during the study period. The 365 days prior to the index date represented the baseline period. Patients were excluded if, during baseline, they had (1) hospitalization for severe ALI (to avoid including prevalent outcomes); (2) anticoagulant dispensed; (3) preexisting liver or biliary disease (to increase the likelihood that severe ALI events were medication related), defined by hepatitis B infection (positive hepatitis B surface antigen, e antigen, or DNA test result), hepatitis C infection (positive hepatitis C RNA or genotype test result), or liver or biliary disease diagnosis (eTable 2 in Supplement 1); or (4) a condition that might precipitate findings consistent with severe ALI (eTable 3 in Supplement 1). To minimize potential for inclusion of persons with undiagnosed liver or biliary disease, we further excluded those who, during baseline, had 2 alanine aminotransferase (ALT) results at 40 U/L or higher (to convert to μkat/L, multiply by 0.0167) at least 6 months apart (validated to identify persons with metabolic dysfunction–associated steatotic liver disease in the VA health care system¹¹), 1 ALT level greater than 100 U/L, or 1 alkaline phosphatase result with a level greater than 172 mg/dL (1.5 times the upper limit of normal [ULN], 115 mg/dL) (to convert to μkat/L, multiply by 0.0167).

Follow-up began on the index date and continued until study end point (defined in the first paragraph of the next section), medication discontinuation (no further fills within 30 days after last day’s supply), anticoagulant dispensing, incident liver or biliary disease or other condition precipitating findings of ALI, 12 months after index date (because drug-induced ALI typically develops within 12 months of initiation¹²), last VA contact, or September 30, 2021, whichever occurred first.

The primary outcome was hospitalization for severe ALI, defined by meeting either of the following definitions within the first 2 days of admission: (1) ALT level greater than 120 U/L (3 times ULN, 40 U/L) plus total bilirubin level greater than 2.0 mg/dL (to convert to mmol/L, multiply by 17.104, 2 times ULN, 1.0 mg/dL) (definition 1); or (2) international normalized ratio (INR) of 1.5 or higher plus total bilirubin level greater than 2.0 mg/dL (definition 2). Both definitions represent severe ALI and have been used by the US Food and Drug Administration’s Sentinel System to assess clinically significant drug-induced ALI in the postmarketing period.¹³ Definition 1 represents Hy’s Law biochemical criteria,¹⁴ which identifies hepatocellular injury severe enough to interfere with bilirubin excretion and predisposes a high mortality risk.^15,16 Definition 2 identifies hepatic dysfunction that might present in an advanced stage of acute liver failure, in which liver aminotransferases might not be high enough to meet definition 1.¹³ We evaluated severe ALI within the first 2 days of admission to avoid ascertaining outcomes that developed as a result of hospitalization. Outcomes were classified on the admission date. We assessed hospitalizations for severe ALI because such events represent ALI at the severe end of the spectrum, minimizing potential for misclassification bias. To increase the likelihood that severe ALI events were drug related, patients with a discharge diagnosis of liver or biliary disease or other condition that precipitates findings consistent with severe ALI (eTables 2 and 3 in Supplement 1) were censored as nonevents on the admission date.

To estimate the frequency that severe ALI events were medication related, we randomly sampled 75 patients who had an event. Two hepatologists independently reviewed the hospital records of these patients and assessed whether a medication possibly caused or contributed to the severe ALI. Among the 75 patients, 57 (76.0%; 95% CI, 64.7%-85.1%) had a hospitalization for severe ALI that was deemed medication related. Details are given in the eMethods in Supplement 1.

Baseline data included date and route of medication administration and days’ supply; age; sex; body mass index (BMI; calculated as weight in kilograms divided by height in meters squared); diabetes and hyperlipidemia diagnoses (eTable 4 in Supplement 1); medication count at initiation date; and INR, total bilirubin, ALT, and alkaline phosphatase results. Medication count was evaluated because of the observed linear association between the number of dispensed medications and drug interactions.¹⁷ Medication count was defined by the number of unique drugs dispensed 90 days or less prior to the index date; patients were classified as having polypharmacy if dispensed 5 or more medications.^18,19 For co-formulated medications, each component was counted separately. Data collected during follow-up included number of days’ supply of medications; hospital ALT, INR, and total bilirubin level; anticoagulant use; liver transplant (eTable 5 in Supplement 1); and all-cause mortality within 180 days of hospitalization for severe ALI.

We calculated age- and sex-adjusted rates of outcomes for each medication using Poisson regression analysis. The model included age (as a continuous variable), sex, and suspected hepatotoxic drug. A test for goodness of fit confirmed the data were not overdispersed. To ensure sufficient precision to estimate rates, we included medication cohorts in the primary analysis if either (1) the 95% CI width for the rate was less than 3 times the point estimate, or (2) 10 000 or more person-years of follow-up was present.

We organized medications into groups based on observed rate of severe ALI (≥10.0 [group 1], 5.0-9.9 [group 2], 3.0-4.9 [group 3], 1.0-2.9 [group 4], and <1.0 [group 5] events per 10 000 person-years). We classified groups 1 and 2 medications as the most potentially hepatotoxic. We identified the number (percentage) of groups 1 and 2 medications that were classified as category A, B, or C by case report–based categorization.⁶

We performed several sensitivity analyses to assess the robustness of the results. First, we repeated the analysis after additionally adjusting for diabetes, hyperlipidemia, and obesity (BMI > 30), since metabolic comorbidities may alter rates of severe ALI.²⁰ Second, we stratified results by sex. Third, we reran our analysis after limiting the cohorts to persons younger than 65 years, since veterans younger than 65 years are less likely to receive care outside the VA because they are typically ineligible for US Medicare.²¹ Fourth, we explored the effects of drug-drug interactions on the results. Since patients could be dispensed multiple hepatotoxic medications that might increase rates of severe ALI and misclassify assignment of medications to a group, we excluded individuals who were additionally dispensed any medication from groups 1 and 2 no more than 90 days prior to the index date and recalculated the rates. To explore whether our groupings were preserved with polypharmacy, we recalculated rates among initiators dispensed 5 or more unique medications within 90 days prior to the index date. Data were analyzed from June 2020 to November 2023 using SAS Enterprise Guide 8.2 (SAS Institute Inc).

The study included 7 899 888 patients who initiated treatment with a suspected hepatotoxic medication in the outpatient setting from 2000 to 2021 (mean [SD] age, 64.4 [16.4] years, 7 305 558 males [92.5%], and 4 354 136 individuals [55.1%] had polypharmacy).

Among the 220 medications with 4 or more published reports of hepatotoxicity, 26 could not be evaluated because they were (1) not dispensed in the VA (n = 8), (2) administered via injection or an intravenous route (n = 13), (3) used for alcohol use disorder or liver disease treatment (n = 2), or (4) anticoagulants (n = 3), leaving 194 medications for evaluation (eTable 1 in Supplement 1). For each of the 194 medication cohorts, eTable 6 in Supplement 1 provides the reasons for exclusion and final samples. The most common reason for exclusion was 365 or fewer days in the VA prior to the initial outpatient fill.

Among the 7 899 888 patients, we identified 1739 hospitalizations for severe ALI; 5 patients (0.3%) underwent liver transplant and 473 (27.2%) died within 180 days of an event. After calculating age- and sex-adjusted rates of severe ALI, 91 medications (46.9%) were not included in the primary analysis because either the 95% CI was more than 3 times the point estimate of severe ALI or there was too little person-time of follow-up (eTable 7 in Supplement 1). Three of these drugs (didanosine, nevirapine, and pyrazinamide) had rates of 10 events or more per 10 000 person-years, while 68 drugs had no observed events.

Figure 1 shows the 17 medications in groups 1 and 2 with the highest observed rates of severe ALI, their age- and sex-adjusted rates of severe ALI, and hepatotoxicity category based on the number of reported cases. Sample sizes and numbers of events are reported in the Table. Incidence rates of severe ALI ranged from 0 events per 10 000 person-years (candesartan, minocycline) to 86.4 events per 10 000 person-years (stavudine). Seven medications (stavudine, erlotinib, lenalidomide or thalidomide, chlorpromazine, metronidazole, prochlorperazine, and isoniazid) exhibited rates of 10.0 or more events per 10 000 person-years, and 10 (moxifloxacin, azathioprine, levofloxacin, clarithromycin, ketoconazole, fluconazole, captopril, amoxicillin-clavulanate, sulfamethoxazole-trimethoprim, and ciprofloxacin) had rates between 5.0 and 9.9 events per 10 000 person-years. Despite these medications having the highest rates of severe ALI, 11 medications (64%) were classified in the less hepatotoxic categories B or C based on case reports. Eleven medications (64%) in groups 1 and 2 were antimicrobials. Rates of severe ALI, sample sizes, and number of events for the medications in group 3 (n = 7), group 4 (n = 37), and group 5 (n = 42) are shown in Figure 2 and Figure 3 and in eTable 8 in Supplement 1. Among the group 3, 4, and 5 medications, 3 (43%), 5 (14%), and 8 (19%), respectively, were classified as category A (most hepatotoxic) by case reports.

After additionally adjusting for diabetes, hyperlipidemia, and obesity, rates of severe ALI remained similar (eFigures 1-3 in Supplement 1). Second, among medications with rates that could be estimated with sufficient precision, the groupings were similar in males and females (eTable 9 in Supplement 1). The rate of severe ALI with estrogen or progestin medications was higher in males than females (10.4 vs 0.0 events per 10 000 person-years). Third, after limiting each cohort to persons younger than 65 years, severe ALI rates were slightly lower in magnitude compared with the primary analysis. Rates of severe ALI for captopril and sulfamethoxazole decreased from 5.8 to 2.7 and 5.1 to 3.1 events per 10 000 person-years, respectively; however, the groupings of the remaining medications remained preserved (eFigures 4-6 in Supplement 1). Fourth, after excluding persons additionally dispensed groups 1 and 2 medications 90 or fewer days prior to suspected hepatotoxic drug initiation (eTable 10 in Supplement 1), rates of severe ALI for groups 1 and 2 medications were slightly lower than in the primary analysis (eFigures 7-9 in Supplement 1). Rates of outcomes for fluconazole and captopril decreased from 6.0 to 1.8 and from 5.8 to 2.6 events per 10 000 person-years, respectively, but groupings for other medications were generally preserved with the recalculated rates. When severe ALI in the presence of polypharmacy was examined, rates of outcomes were similar to the primary analysis and groupings were preserved (eFigures 10-12 in Supplement 1).

This series of cohort studies used real-world data to measure incidence rates of hospitalization for severe ALI following outpatient initiation of suspected hepatotoxic medications among persons without preexisting liver or biliary disease. The analyses identified 17 medications (groups 1 and 2) with the highest rates of severe ALI. These drugs remained the most potentially hepatotoxic after excluding concomitant users of other medications of groups 1 and 2 and with polypharmacy. Categorization of hepatotoxicity based on the number of published case reports did not accurately reflect observed rates of severe ALI.

Antimicrobial medications represented 64% of the medications from groups 1 and 2 with the highest rates of severe ALI. Antifungal and older antiretroviral medications have been particularly implicated as hepatotoxic.^22,23 This is consistent with our findings.

This study represents a systematic, reproducible approach to using real-world data to measure rates of severe ALI following medication initiation among patients without liver or biliary disease. Regulators and clinicians have relied on published case reports to assess the hepatic safety of drugs, but our results show that the number of reports inaccurately represents rates of severe ALI. We created cohorts of new initiators of outpatient medications, defined the period of exposure as the time receiving medication through 30 days after discontinuation (to capture events shortly after cessation) and evaluated a composite laboratory-based severe ALI outcome indicating significant hepatic injury. We reduced the likelihood of capturing non–drug-related severe ALI by censoring as nonevents patients with liver or biliary disease or other conditions precipitating ALI. We accounted for potential effects of drug-drug interactions by examining rates of outcomes after exclusion of concomitant users of groups 1 and 2 drugs and those with polypharmacy; results showed that medications from groups 1 and 2 remained the most hepatotoxic. These methods could allow measurement of rates of severe ALI for medications and enable evaluation of hepatotoxicity safety signals from spontaneous reports.

The study has important implications for clinical care. Patients initiating a medication with a high rate of severe ALI might require closer monitoring of liver-related laboratory tests to detect evolving hepatic dysfunction earlier, which might improve prognosis. Within electronic health record systems, automated messages could alert clinicians ordering a high-risk medication to the potential for severe ALI and to consider laboratory monitoring. The results also provide real-world evidence of the hepatic safety of certain medications. For example, although some statins were included in category A or B based on reported cases of hepatotoxicity, rates of severe ALI for these medications were low (<1.0 event per 10 000 person-years).

Our method of measuring rates of severe ALI after suspected hepatotoxic medication initiation could be applied within other electronic medical record data outside the VA system. The approach could be modified to identify new potentially hepatotoxic drugs by looking back from hospitalizations for severe ALI and examining associations with proximal medications. This method could also be expanded to examine other medication-related toxic effects on organs.

Our study has several limitations. First, we did not perform a causality assessment of all outcomes. Within a random sample of severe ALI events, 76% were deemed medication related by hepatologist review. Some severe ALI events might have been caused by herbal and dietary supplements, over-the-counter drugs such as acetaminophen, or other hepatotoxic medications that were not ascertained. Some events may have been due to non–drug-related conditions that evaded censoring because they were not recorded as discharge diagnoses. We also might have missed steatotic liver disease among individuals without baseline liver-related laboratory tests. However, there is no reason to believe that these products or conditions were differentially distributed across cohorts.

Second, because ascertainment of ALI relied on laboratory tests, surveillance bias may have contributed to the results, as clinicians might have performed more testing on persons dispensed medications with a greater number of hepatotoxicity reports. Third, our approach might have underestimated rates of outcomes, since veterans may present for emergent care outside the VA. Fourth, the 95% CIs around the severe ALI incidence rates could have been too narrow, as they did not account for the rate of outcome misclassification. Fifth, because we did not perform hypothesis testing or make formal statistical comparisons in rates between the medications of interest, we did not apply adjustment for multiple comparisons in our results.

Sixth, while we examined medication count as a proxy for drug-drug interactions, it was beyond the scope of our analyses to examine rates of severe ALI with specific combinations of medications. Seventh, we did not evaluate the influence of medication dose on severe ALI rates. Eighth, we did not examine rates of severe ALI after initiation of medications that had 1 to 3 published reports of hepatotoxicity. Instead, we focused on medications with the highest number of hepatotoxicity reports, since these are likely to be perceived as having the highest risk of severe ALI. Future studies should evaluate these other medications. Ninth, our study included predominantly male patients without liver or biliary disease. Results may not be generalizable to females or individuals with liver disease.

This series of cohort studies demonstrated that real-world data can be used to measure rates of hospitalization for severe ALI following outpatient initiation of suspected hepatotoxic medications among persons without preexisting liver or biliary disease. We identified medications with the highest rates of severe ALI after censoring for non–drug-related causes. Categorization of hepatotoxicity by case reports did not accurately reflect severe ALI rates using real-world data. This study provides a framework for investigating postmarketing hepatotoxicity safety signals. Future studies should evaluate rates of severe ALI among persons with chronic liver disease to provide evidence on the hepatic safety of medications in these patients.

Accepted for Publication: March 30, 2024.

Published Online: June 24, 2024. doi:10.1001/jamainternmed.2024.1836

Correction: This article was corrected on October 14, 2024.

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2024 Torgersen J et al. JAMA Internal Medicine.

Corresponding Author: Vincent Lo Re III, MD, MSCE, Center for Real-World Effectiveness and Safety of Therapeutics, Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, 423 Guardian Dr, 836 Blockley Hall, Philadelphia, PA 19104 (vincentl@pennmedicine.upenn.edu).

Author Contributions: Dr Lo Re had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Torgersen, Mezochow, Carbonari, Hennessy, Tate, Bräu, Roy, Justice, Lo Re.

Acquisition, analysis, or interpretation of data: Torgersen, Mezochow, Newcomb, Carbonari, Hennessy, Rentsch, Park, Tate, Bhattacharya, Lim, Mezzacappa, Njei, Taddei, Justice, Lo Re.

Drafting of the manuscript: Torgersen, Mezochow, Newcomb, Carbonari, Roy, Lo Re.

Critical review of the manuscript for important intellectual content: Torgersen, Mezochow, Carbonari, Hennessy, Rentsch, Park, Tate, Bräu, Bhattacharya, Lim, Mezzacappa, Njei, Taddei, Justice, Lo Re.

Statistical analysis: Newcomb, Rentsch, Lim, Roy, Justice, Lo Re.

Obtained funding: Justice, Lo Re.

Administrative, technical, or material support: Torgersen, Carbonari, Park, Tate, Njei, Justice.

Supervision: Torgersen, Tate, Lim, Justice, Lo Re.

Conflict of Interest Disclosures: Dr Torgersen reported receiving grants from the National Institute of Allergy and Infectious Diseases and the National Institute of Diabetes and Digestive and Kidney Diseases during the conduct of the study. Dr Hennessy reported receiving personal fees from Urovant Sciences, the Medullary Thyroid Carcinoma Registry Consortium (Novo Nordisk, AstraZeneca Pharmaceuticals LP, GlaxoSmithKline LLC, Lilly), i2o Therapeutics, Basilea, Provention Bio Inc, Ipsen Bioscience Inc, Covis Pharma GmbH, Biogen MA Inc, and Bluebird bio, Inc outside the submitted work. Dr Tate reported receiving grants from the National Institutes of Health during the conduct of the study. Dr Bhattacharya reported receiving grants from Gilead Sciences, Inc outside the submitted work. Dr Lim reported receiving grants from Gilead Sciences Inc, Intercept Pharmaceuticals, Inventiva Pharma, Novo Nordisk, Pfizer, and Viking Therapeutics outside the submitted work. Dr Lo Re reported receiving personal fees from Entasis Therapeutics and Urovant Sciences outside the submitted work. No other disclosures were reported.

Funding/Support: This study was funded, in part, by grant R01CA206465 (multiple principle investigators [MPIs], including Drs Justice and Lo Re) from the National Cancer Institute; grants P01AA029545 (PI, Dr Justice), U24AA020794 (PI, Dr Justice), U01AA020790 (PI, Dr Justice), U24AA022001 (PI, Dr Justice), and U01AA013566 (PI, Dr Justice) from the National Institute on Alcohol Abuse and Alcoholism; grant K08DK132977 (PI, Dr Torgersen) from the National Institute of Diabetes and Digestive and Kidney Diseases; and grant T32AI055435 (Dr Mezochow) from the National Institute of Allergy and Infectious Diseases.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The contents of this article do not represent the views of the US Department of Veterans Affairs (VA) or the US government. This work uses data provided by patients and collected by the VA as part of their care and support.

Meeting Presentation: Initial results of this study were presented at the virtual meeting of The Liver Meeting Digital Experience (oral abstract 119); November 15, 2020, and the International Conference on Pharmacoepidemiology 2023 (oral abstract 178); August 26, 2023; Nova Scotia, Canada.

Data Sharing Statement: See Supplement 2.