Elsevier

Annals of Emergency Medicine

Volume 46, Issue 3, September 2005, Pages 263-271
Annals of Emergency Medicine

Original Research
A Risk Quantification Instrument for Acute Acetaminophen Overdose Patients Treated With N-Acetylcysteine

https://doi.org/10.1016/j.annemergmed.2005.04.004Get rights and content

Study objective

The risk of hepatotoxicity after acute acetaminophen overdose varies with timed serum acetaminophen concentration and delay to treatment. The ability to accurately predict hepatotoxicity is needed to reduce confusion about the optimal treatment regimen for individual patients and the effects of risk modifiers such as ethanol. We quantitatively estimate the risk of hepatotoxicity based on the degree and duration of pretreatment exposure to supratherapeutic concentrations of acetaminophen.

Methods

We examined all hospitalizations for acute acetaminophen overdose within a retrospective multicenter Canadian registry. We used a previously developed composite measure incorporating timed serum acetaminophen concentration and time to N-acetylcysteine treatment into a single parameter. We then modeled hepatotoxicity on this parameter, as well as age, sex, and ethanol use. Hepatotoxicity was defined as peak aminotransferase level of 1,000 IU/L or greater at 24 hours or longer.

Results

Of 1,270 admitted patients treated mostly with intravenous N-acetylcysteine for less than 24 hours, our model accurately identified the 94 patients who developed hepatotoxicity (discriminatory index 0.93). Hepatotoxicity occurred in none of the 313 patients (95% confidence interval [CI] 0% to 1.0%) above the traditional 150 μg/mL treatment line who were classified as low risk (<1%) using our instrument. After adjustment for severity of exposure, the risk of hepatotoxicity was considerably higher in the absence of coingested ethanol (median hepatotoxic dose 16.5 mmol/L × hour [95% CI 8.74 to 31.0 mmol/L × hour] versus 27.1 mmol/L × hour [95% CI 11.1 to 66.3 mmol/L × hour]), particularly among alcoholics (4.79 mmol/L × hour [95% CI 2.13 to 10.8 mmol/L × hour]).

Conclusion

Our risk prediction instrument identifies a large group of low-risk patients for whom 20-hour intravenous N-acetylcysteine therapy is sufficient. Our results suggest that acute and chronic ethanol use dramatically influences acetaminophen toxicity. This work may facilitate the evaluation of individualized treatment strategies for higher-risk patients.

Introduction

Acetaminophen is the most common pharmaceutical taken in overdose. In most Western countries, it is responsible for the greatest number of hospital admissions and deaths after overdose.1, 2, 3 Accurately predicting the risk of hepatotoxicity after acetaminophen overdose is essential for several reasons. The clinical and biochemical signs of end-organ toxicity are not detectable until about 24 hours after overdose, well past the optimum time for antidote therapy.4 Accurate risk stratification is necessary to minimize under- and overtreatment. Uncertainty currently exists about the optimal dose, duration, and route of N-acetylcysteine administration because imprecise risk prediction precludes direct comparison of the results of the large-scale effectiveness trials.5, 6, 7, 8, 9

More accurate risk estimation would also allow investigators to test controversial hypotheses about the effect of potential risk modifiers such as ethanol and would enhance the power of future clinical trials studying optimal therapy.7, 10 Ultimately, more accurate risk stratification might be used clinically to individualize therapy. High-risk patients identified early might be candidates for more intensive therapy, whereas low-risk patients could be targeted for abbreviated courses of N-acetylcysteine to shorten hospital stays and minimize the cost of therapy.

The Rumack-Matthew nomogram has proven to be a highly sensitive decision instrument to identify the need for N-acetylcysteine therapy after acute acetaminophen overdose.5, 6, 7, 11, 12 Once treatment is begun, it is no longer appropriate to use the nomogram for risk prediction because the timely administration of N-acetylcysteine substantially reduces the risk of hepatotoxicity.3, 6, 7, 13, 14 Moreover, the effects of postulated risk modifiers such as alcoholism and acute ethanol ingestion are poorly quantified.15, 16 Therefore, only a rough approximation of risk is possible for patients treated with N-acetylcysteine, despite the frequency of this occurrence.

To illustrate, it has been recognized for nearly 20 years that hepatotoxicity develops in about 6% of patients with a serum acetaminophen concentration above 200 μg/mL at 4 hours who are treated within 10 hours of overdose.7 Not all patients within this group, however, have the same degree of risk. Patients with higher serum acetaminophen concentrations, longer delays to N-acetylcysteine, and other risk factors (such as perhaps alcoholism) are at a higher individual risk. Determining which patients are at higher risk requires a more precise characterization of the dose-response curve of acetaminophen toxicity, allowing for varying delays to treatment with N-acetylcysteine.

We sought to construct a risk prediction instrument for hepatotoxicity after acute acetaminophen overdose treated with N-acetylcysteine, based on pretreatment duration and degree of supratherapeutic acetaminophen concentrations. By using data from Canadian hospitals in which a 20-hour intravenous N-acetylcysteine protocol has been the norm for 2 decades, we sought to quantify the effectiveness of this protocol. We also hoped to use the new prediction instrument to identify a subset of low-risk patients who might benefit from abbreviated therapy and to explore the influence of other factors, including ethanol, on the risk of hepatotoxicity.

Section snippets

Study Design and Setting

Using the Canadian Acetaminophen Overdose Study (CAOS), we identified subjects from tertiary care hospitals in 6 large Canadian cities from 1980 to 2002. CAOS is an ongoing, structured, retrospective review of patients hospitalized with acetaminophen overdose. CAOS identifies subjects by electronic search of discharge diagnoses, which include International Classification of Diseases, Ninth Revision 965.4 or International Classification of Diseases, 10th Revision T391. The primary objective of

Results

Of 3,520 admissions identified, 3,202 (91%) charts were available for review and entered into the CAOS registry. A total of 2,488 (77.7%) admissions resulted from single acute acetaminophen ingestion. Of these, we excluded 1,218 (49.0%). One hundred ninety-six were excluded because there was no serum acetaminophen level obtained within 24 hours of ingestion, 1,012 because no serum aminotransferase concentration was obtained more than 24 hours after ingestion, and 10 because of preexisting

Limitations

Our study represents the initial clinical application of a pharmacokinetically derived mathematical model. Our approach will benefit from independent validation using other large databases of acetaminophen overdose patients. Although our data provide a benchmark to compare the effectiveness of different treatment strategies, no such direct comparison was performed in this analysis.

Each of the underlying pharmacokinetic assumptions imperfectly models the real world of clinical toxicology.17 In

Discussion

The ability to accurately predict hepatotoxicity shortly after acetaminophen overdose is crucial because treatment decisions must be made before liver injury is detectable.4 Initiating antidote therapy based on the Rumack-Matthew nomogram has withstood the test of time, with only rare failures reported using the 150 μg/mL (1,000 μmol/L) treatment line.15, 25, 26 Which N-acetylcysteine protocol to initiate, however, is less clear. Without the ability to identify low-risk patients at

References (43)

  • E. Hazai et al.

    Reduction of toxic metabolite formation of acetaminophen

    Biochem Biophys Res Commun

    (2002)
  • J.A. Critchley et al.

    Is there a place for cimetidine or ethanol in the treatment of paracetamol poisoning?

    Lancet

    (1983)
  • K. Hawton et al.

    Effects of legislation restricting pack sizes of paracetamol and salicylate on self poisoning in the United Kingdom: before and after study

    BMJ

    (2001)
  • D.N. Juurlink

    The Clinical Epidemiology of Poisoning in Ontario

    (2003)
  • L.F. Prescott et al.

    Intravenous N-acetylcysteine: the treatment of choice for paracetamol poisoning

    BMJ

    (1979)
  • B.H. Rumack et al.

    Acetaminophen overdose: 662 cases with evaluation of oral acetylcysteine treatment

    Arch Intern Med

    (1981)
  • M.J. Smilkstein et al.

    Efficacy of oral N-acetylcysteine in the treatment of acetaminophen overdose: analysis of the national multicenter study (1976 to 1985)

    N Engl J Med

    (1988)
  • N.A. Buckley et al.

    Oral or intravenous N-acetylcysteine: which is the treatment of choice for acetaminophen (paracetamol) poisoning?

    Clin Toxicol

    (1999)
  • D. Parker et al.

    Safety of late acetylcysteine treatment in paracetamol poisoning

    Hum Exp Toxicol

    (1990)
  • B.H. Rumack et al.

    Acetaminophen poisoning and toxicity

    Pediatrics

    (1975)
  • C.I. Wallace et al.

    Paracetamol overdose: an evidence based flowchart to guide management

    Emerg Med J

    (2002)
  • Cited by (67)

    • Acetaminophen Poisoning

      2021, Critical Care Clinics
      Citation Excerpt :

      The acetaminophen ratio (the initial acetaminophen concentration/the nomogram line acetaminophen concentration), is currently used to guide increased acetylcysteine doses in some centers.19,48 Several modifications to these traditional strategies (acetaminophen-aminotransferase multiplication product; psi parameter; and a combination of initial acetaminophen ratio, ALT, and time to treatment) have been shown to predict outcome better in cohort studies (see Table 2)69–74 The advantage of these tools is that they could easily be integrated into current management protocols because they use available tests. Possible uses for these risk stratification tools would be to guide shorter courses or ongoing need for acetylcysteine in RSTI.65,71

    • Outcomes of Patients with Premature Discontinuation of the 21-h Intravenous N-Acetylcysteine Protocol after Acute Acetaminophen Overdose Presented at the North American Congress of Clinical Toxicology, October 2010, Denver, Colorado.

      2016, Journal of Emergency Medicine
      Citation Excerpt :

      We have also proposed using the area under the serum APAP vs. time curve before the start of treatment after acute overdose, as a single, continuous measure of exposure. This parameter, termed ψ, incorporates both dose (as estimated by the extrapolated 4-h APAP concentration) and the duration of pretreatment exposure (as estimated by the time from ingestion to the start of NAC), and has been validated in both Canadian and Thai patient populations (11,26,27). In the present study, we found that ψ values up to 5 mmol/L-h are associated with good outcomes in patients administered >1–2 h i.v. NAC, yet still receiving less than the complete 21-h protocol.

    View all citing articles on Scopus

    Supervising editor: G. Randall Bond, MD

    Author contributions: MLAS had the original idea for the study. MCY and DWJ planned and coordinated the Canadian Acetaminophen Overdose Study and oversaw data collection. DNJ assisted in the analysis and interpretation of the results. AMG piloted and helped refine the methodology. All authors were involved in the writing of the manuscript. MLAS takes responsibility for the paper as a whole.

    Funding and support: Supported by the physicians of Ontario through the P.S.I. Foundation (MLAS grant 02-38); the Canadian Association of Emergency Physicians (MCY, DWJ); the Centre for Advancement of Health, Foothills Hospital Calgary AB (MCY, DWJ); a Canadian Institute of Health Research New Investigator Award (DNJ) and Student Research Award (AMG); and an Alberta Heritage Foundation Studentship (DWJ).

    Presented at the North American Congress of Clinical Toxicology, September 2003, Chicago, IL.

    Reprints not available from the authors.

    View full text