|Analgesics; Non-steroidal anti-inflammatory agents;
Ibuprofen; Naproxen sodium; Acetaminophen; Acute liver failure;
Acute kidney injury; Gastrointestinal hemorrhage
|A considerable proportion of the U.S. population uses nonprescription
analgesics on a regular basis . Four oral medications
are currently approved in the U.S. for non-prescription analgesia:
acetaminophen, aspirin, ibuprofen, and naproxen sodium. As with any
pharmaceuticals, these agents carry some risk of adverse events that
must be weighed against their therapeutic benefits.
|Some cases of acute liver failure have been attributed to
nontherapeutic use or unintentional overdose of acetaminophen
[2-5]. Acute liver failure is also theoretically possible for those with
severely damaged livers who take acetaminophen at recommended
doses, although this phenomenon has not been established.
Recent recommendations designed to reduce excessive dosing
with acetaminophen may also discourage some appropriate use of
acetaminophen, diverting patients to use alternative analgesics .
These alternatives to acetaminophen carry their own risks and benefits.
With nonspecific nonsteroidal anti-inflammatory drugs (NSAIDs),
there may be dose-related effects on the gastrointestinal lining and
on cardiovascular disease risk . Some of these effects are welcome;
for example, aspirin in low doses is recommended to lower risk of
myocardial infarction in high risk populations. Other effects, such as
acute renal failure and gastrointestinal bleeding, are undesirable.
|In this paper, we assess the net effect on mortality risk related to each
of the non-aspirin over-the-counter analgesics approved for use in the
U.S. There are four major fatal endpoints that are thought to be affected
by therapeutic analgesic use. These endpoints are upper gastrointestinal
hemorrhage, liver failure, renal failure and a sudden cardiovascular
event (either stroke or myocardial infarction). The first three categories of risk convey much smaller risks than cardiovascular events. As
a result, any attempt to weigh risks among alternative analgesics
would be dominated by the cardiovascular effects of the analgesics.
Although acetaminophen does not appear to have any association with
cardiovascular mortality, alternative analgesics have effects that may
be either protective or causative. Aspirin is used for chemoprevention
of cardiac outcomes, although optimal cardiopreventive dosing is too
low to offer effective analgesic relief. In contrast, other NSAIDs have
been reported to increase cardiovascular mortality [8-10]; two of
them, rofecoxib and valdecoxib, were withdrawn in 2004 and 2005
respectively because of an association with cardiovascular deaths
[11,12]. Because cardiovascular deaths far outnumber deaths from
gastrointestinal hemorrhage, liver failure or renal failure, even modest
variations in relative risk for cardiovascular deaths would dominate any
effects related to deaths from other endpoints.
|Nevertheless, for someone who was already getting optimal
chemoprevention against cardiovascular death, there remains the
question of what the attendant risks might be for someone weighing
the use of acetaminophen or an alternative analgesic. Thus, a patient
taking 81 mg/day of aspirin to lower cardiovascular risk might still face the question of what non-prescription analgesic to use for pain. That
hypothetical patient could benefit from understanding the comparative
risks among the choices available. In this paper we compare the risks
for fatal outcomes caused by gastrointestinal hemorrhage, acute liver
failure or acute renal failure. We estimated risks for each of these three
outcomes for each of the non-aspirin non-prescription analgesics
approved in the U.S., in order to estimate the risk of death from these
causes related to use of each of these agents.
|We studied the three oral non-aspirin drugs that are approved for
nonprescription analgesia in the US: acetaminophen, ibuprofen and
naproxen sodium. For each drug we obtained estimates of the recent
general-population point prevalence of use from the most recently
available data in the Slone surveys [13,14], which were conducted
during the period 2004 to 2007. The Slone Survey inquired about
analgesic use among adults age 18+ during the week before the survey.
For simplicity we assumed that users reporting that they used one
analgesic were nonusers of the others during the reporting period.
|We obtained the number of deaths in the U.S. from acute liver
injury, acute renal failure, and gastrointestinal hemorrhage for the years
2006, 2007, and 2008 from the National Center for Health Statistics
Compressed Mortality File (CMF) . The CMF uses counts of deaths
by cause from vital statistics registries and U.S. population estimates
from the Bureau of the Census. We restricted the mortality data to
age 20 years or older. Types of underlying cause of death were defined
by International Classification of Diseases, Tenth Revision (ICD-10)
codes . Upper gastrointestinal hemorrhage was defined by codes
for gastrointestinal ulcer with hemorrhage or perforation (K25 through
K29 range, K92.2). Acute liver failure deaths were defined as codes
K72.0 (acute and subacute hepatic failure) and K76.7 (hepatorenal
syndrome). For acute renal failure we used code N17, which included
acute renal failure with tubular necrosis, or acute cortical necrosis,
or other or unspecified pathology. We converted the three years of
death data to one-year risks by dividing the average annual deaths in
each outcome category over the three years by the average U.S. adult
population age 20+ for those years (Table 1).
|The total population one-year risk for each endpoint is a weighted
average of the risks among nonusers of analgesics and users of each of the
three analgesics. Most of the available literature on the risks associated
with non-prescription analgesic use compares the risk among users with
that of nonusers of that analgesic, and reports results as relative risks.
We searched the literature to obtain estimates of the relative risks for
each drug-endpoint combination. When more than one study reported
findings for a particular drug-endpoint combination, we combined the
results in a fixed-effects meta-analysis. For many combinations, there
was only one study reporting on the relation. We relied on populationbased
or general practice studies whenever available, rather than
clinical trials and studies with comparison groups based on either no NSAID exposure or past exposure. We used three estimates for each
of the 9 relative risks that we sought to estimate: we chose a central
estimate for our primary evaluation, and we also used a low estimate
and a high estimate for sensitivity analyses. Our estimates were based
on use of each product according to the label. We did not attempt to
incorporate risks stemming from accidental or deliberate overdoses,
although when low-dose or high-dose estimates were reported, we
used the high-dose result if it was within therapeutic guidelines. The
central estimate was either the only reported value or the result of a
meta-analysis of reported values. For the low and high estimates we
used the lower and upper confidence limits of the reported estimate
that we used for the central estimate, or the lower and upper confidence
limits of the meta-analysis, unless the lower bound was less than unity,
in which case we used unity instead.
|For each relative risk, the nonusers could be defined either as
nonusers of that product or as nonusers of any analgesic. For our
calculations we considered nonusers to be those not using any analgesic.
If users of other analgesics are part of the referent group for the relative
risks, the estimates for the relative risk obtained may differ from what
would have been obtained with just nonusers of analgesics as the referent
group. Generally the inclusion of users of other products in the referent
group for a specific product would lead to slight underestimates of the
relative risk for that product compared with nonusers of any analgesic,
presuming that the analgesics elevate risks above the levels in nonusers.
|From the relative risk values taken from the literature, we calculated
estimates of the absolute risk of each of the fatal endpoints for users
of each type of analgesic. We started from the overall risk of each
endpoint based on the deaths obtained from the National Center for
Health Statistics. That total risk for an endpoint is a weighted average
of the risks experienced by each category of analgesic user and the risk
among those unexposed to any of the analgesics. From the relative risks
and total risk, we back-calculated the risk among unexposed, from
which we could obtain the absolute risk for each analgesic category.
This calculation assumes that confounding between estimates is
|The prevalence of use of the three analgesics of interest among
people 18 years or older was 18% for acetaminophen , 17% for
ibuprofen, and 4.7% for naproxen sodium . The one-year overall
risks of death are given in table 1 for the three endpoints. Risks are
low for all three categories, but about one order of magnitude lower
for acute liver failure than for acute renal failure or gastrointestinal
|Table 2 gives the range of risk ratio estimates that we assembled
from the available literature. The first row indicates what we considered
the central estimate for each measure, based on either a fixed effects
meta-analysis or the use of a single study where only one was available.
The next two rows indicate lower and upper values that we took as the
plausible range. These values were either taken from the confidence
limits from the single study or meta-analysis, unless additional
information on dose or duration of use was available. The underlying
literature and reported values are summarized in appendix table 1,
which reports values for any use of the NSAID, for low-dose use, and
for high-dose use. These categories are sometimes not available and,
when they were, not consistently defined across studies. In table 3, the
number of deaths predicted in a population experience of 1,000,000
person-years is shown under three different scenarios, using the best
estimates for the risk ratios, the low estimates, and the high estimates.
|Under the best estimate scenario, use of any of the nonprescription
analgesics is related to some increase in the overall risk of death from
one of the three considered endpoints, compared with nonexposed.
Acetaminophen use was estimated to carry the smallest increase in
risk, about 35 deaths per million person-years of experience, most of
the increase being attributed to an increase in risk of acute renal failure.
Increased risk from using ibuprofen was estimated to be nearly twice as
great as the increase for acetaminophen, at about 64 deaths per million
person-years of experience, with a slightly larger increase in risk than
acetaminophen for acute renal failure and a much greater risk for upper
gastrointestinal hemorrhage. For naproxen sodium, the increase in risk
for acute renal failure is smaller than the increase for either of the other
two analgesics, but it is overshadowed by a much greater increase in risk
for fatal upper gastrointestinal hemorrhage than the alternative drugs.
The overall estimated increase in deaths related to naproxen sodium is
118 per million person-years. From the perspective of switching from
acetaminophen to one of the alternatives, there would be an increase
in the risk of a fatal adverse event in one of the three categories equal
to about 29 deaths per million person-years if the switch is from
acetaminophen to ibuprofen and 83 deaths per million person-years if
the switch is from acetaminophen to naproxen sodium.
|For the scenario based on using all the low estimates, switching from
acetaminophen to ibuprofen would result in an estimated increase of 37
deaths per million person-years; for the high estimates, we estimated
nearly the same number of deaths from using acetaminophen and
ibuprofen. For naproxen sodium, low estimates resulted in an estimate
of an increase of 86 deaths per million person-years for those switching
from acetaminophen to naproxen, and 53 deaths per million personyears
for the high estimates [17-27].
|The risks for acute liver failure are so low from any of these drugs
that this endpoint barely influences the calculations. Thus, if instead of a RR of 1.0 between acetaminophen and fatal acute liver failure, we had used a RR of 50, the risk of death among acetaminophen users from any of the three endpoints would have been 107 per million personyears
instead of 93 per million person-years. The component due to
acute liver failure increases from 2.7 to 16.8 deaths per million personyears,
with the risk among unexposed changing from 2.7 deaths per
million person-years to 0.34 deaths per million person-years. Under
this scenario, the risks related to use of ibuprofen and naproxen sodium
would be only slightly reduced from the values of 122 per million
person-years and 177 per million person-years respectively, and there
would be only a small change in the overall risk difference.
|The other two endpoints, acute renal failure and upper
gastrointestinal hemorrhage, are approximately equally common,
but the estimates from the literature of increased risk for each of the
analgesics are more narrowly spread for acute renal failure than for
upper gastrointestinal hemorrhage. Thus, the assumptions about the
magnitude of effects for upper gastrointestinal hemorrhage dominate
the risk calculations and projected differences in risk among these
|This analysis is premised on an array of reported findings from the
literature on analgesics, some of which are subject to considerable error.
We also made a number of assumptions for ease of comparison. We
assumed that all analgesics would be taken according to label directions.
Thus, our estimate that acetaminophen likely does not increase risk of
fatal liver failure is intended to describe the relation when users follow
the instructions on the label. According to the FDA, “The risk of liver
injury primarily occurs when patients take multiple products containing
acetaminophen at one time and exceed the current maximum dose of
4000 mg within a 24-hour period” . Nevertheless, even when the
risk ratio for liver injury from acetaminophen was assumed to be 50
rather than 1, the overall risk of death from any of these three outcomes
studied changed very little, because deaths from liver injury were so
|We also assumed that effects of analgesics were uniform over
age, sex and other variables, and that there was no uncontrolled
confounding that would have influenced the interpretation. Another
assumption was that each person uses only one analgesic at a given
time, although in reality there could be combination therapy. We
ignored non-fatal adverse effects and limited attention among fatal
outcomes to three endpoints, liver disease, renal disease, and upper
gastrointestinal hemorrhage. Doing so involved what was perhaps the
most important assumption, namely that the cardiovascular effects
were constant across all the analgesics, and that all users were receiving
optimal cardioprotective doses of aspirin. For someone who was not
taking low-dose aspirin, the higher-dose of aspirin taken for analgesia
might lower cardiovascular risk as a secondary effect and thus offset
some of the increase in risk from upper gastrointestinal bleeding. But if
the same person already takes low-dose aspirin, thus reducing the risk
of cardiovascular death but not achieving analgesia, acetaminophen
appears likely to be associated with a lower risk of death than ibuprofen
or naproxen sodium, the other two non-aspirin non-prescription
alternatives on the U.S. market.
|This work was supported by a contract between McNeil Consumer Healthcare
and RTI-Health Solutions. RTI Health Solutions is an independent, non-profit
research organization that does work for government agencies and pharmaceutical
companies. No ethical review was needed for this work.
|This work was supported by a contract between Mcneil Consumer Healthcare and Rti-Health Solutions. Rti health solutions is an independent, non-profit
research organization that does work for government agencies and pharmaceutical
- Kaufman DW, Kelly JP, Rosenberg L, Anderson TE, Mitchell AA (2002) Recent patterns of medication use in the ambulatory adult population of the United States: the Slone survey. JAMA 287: 337-344.
- Beer C, Pakravan N, Hudson M, Smith LT, Simpson K, et al. (2007) Liver unit admission following paracetamol overdose with concentrations below current UK treatment thresholds. QJM 100: 93-96.
- Dart RC, Bailey E (2007) Does therapeutic use of acetaminophen cause acute liver failure? Pharmacotherapy 27: 1219-1230.
- Larson AM, Polson J, Fontana RJ, Davern TJ, Lalani E, et al. (2005) Acetaminophen-induced acute liver failure: results of a United States multicenter, prospective study. Hepatology 42: 1364-1372.
- Norris W, Paredes AH, Lewis JH (2008) Drug-induced liver injury in 2007. Curr Opin Gastroenterol 24: 287-297.
- Food and Drug Administration (2009) Department of Health and Human Services: Organ-specific warnings; internal analgesic, antipyretic and antirheumatic drug products for over-the-counter human use; final monograph 74:19385–19409.
- Conaghan PG (2012) A turbulent decade for NSAIDs: update on current concepts of classification, epidemiology, comparative efficacy, and toxicity. Rheumatol Int 32: 1491-1502.
- García Rodríguez LA, Varas-Lorenzo C, Maguire A, González-Pérez A (2004) Nonsteroidal antiinflammatory drugs and the risk of myocardial infarction in the general population. Circulation 109: 3000-3006.
- Ray WA, Stein CM, Hall K, Daugherty JR, Griffin MR (2002) Non-steroidal anti-inflammatory drugs and risk of serious coronary heart disease: an observational cohort study. Lancet 359: 118-123.
- Varas-Lorenzo C, Castellsague J, Stang MR, Perez-Gutthann S, Aguado J, et al. (2009) The use of selective cyclooxygenase-2 inhibitors and the risk of acute myocardial infarction in Saskatchewan, Canada. Pharmacoepidemiol Drug Saf 18: 1016-1025.
- McDaid C, Maund E, Rice S, Wright K, Jenkins B, et al. (2010) Paracetamol and selective and non-selective non-steroidal anti-inflammatory drugs (NSAIDs) for the reduction of morphine-related side effects after major surgery: a systematic review. Health Technol Assess 14: 1-153, iii-iv.
- Slone Epidemiology Unit, Boston University School of Public Health: Acetaminophen use in the adult population of the United States in 2004-2007. Results of a population-based telephone survey (unpublished report for McNeil). May, 2009.
- (2006) Slone Epidemiology Center at Boston University: Patterns of medication use in the United States. A report from the Slone Survey.
- Centers for Disease Control and Prevention, National Center for Health Statistics.
- World Health Organization. International Statistical Classification of Diseases and Related Health Problems. 10th Revision. Version for 2006.
- de Abajo FJ, Montero D, Madurga M, Garcia Rodriguez LA (2004) Acute and clinically relevant drug-induced liver injury: a population-based case-control study. Br J Clin Pharmacol 58: 71-80.
- Perneger TV, Whelton PK, Klag MJ (1994) Risk of kidney failure associated with the use of acetaminophen, aspirin, and nonsteroidal antiinflammatory drugs. N Engl J Med 331: 1675-1679.
- García Rodríguez LA, Hernández-Díaz S (2001) Relative risk of upper gastrointestinal complications among users of acetaminophen and nonsteroidal anti-inflammatory drugs. Epidemiology 12: 570-576.
- Lee CH, Wang JD, Chen PC (2010) Increased risk of hospitalization for acute hepatitis in patients with previous exposure to NSAIDs. Pharmacoepidemiol Drug Saf 19: 708-714.
- García Rodríguez LA, Pérez Gutthann S, Walker AM, Lueck L (1992) The role of non-steroidal anti-inflammatory drugs in acute liver injury. BMJ 305: 865-868.
- Huerta C, Castellsague J, Varas-Lorenzo C, García Rodríguez LA (2005) Nonsteroidal anti-inflammatory drugs and risk of ARF in the general population. Am J Kidney Dis 45: 531-539.
- Lafrance JP, Miller DR (2009) Selective and non-selective non-steroidal anti-inflammatory drugs and the risk of acute kidney injury. Pharmacoepidemiol Drug Saf 18: 923-931.
- Castellsague J, Pisa F, Rosolen V, Riera-Guiardia N, Giangreco M, et al. (2011) Use of nimesulide and other NSAIDs and the risk of upper gastrointestinal complications in Firuli Venezia Giulia, Italy. Analysis of cases with codes with high positive predictive value. Abstract 409. PDS 20: S178.
- Riera-Guardia N, Castellsague J, Calingaert B, Varas-Lorenzo C, Fourrier-Réglat A, et al. (2010) Nonsteroidal anti-inflammatory drugs and upper gastrointestinal complications. Meta-analysis of epidemiological studies. Poster. International Conference on Pharmacoepidemiology.
- Traversa G, Bianchi C, Da Cas R, Abraha I, Menniti-Ippolito F, et al. (2003) Cohort study of hepatotoxicity associated with nimesulide and other non-steroidal anti-inflammatory drugs. BMJ 327: 18-22.
- Schneider V, Lévesque LE, Zhang B, Hutchinson T, Brophy JM (2006) Association of selective and conventional nonsteroidal antiinflammatory drugs with acute renal failure: A population-based, nested case-control analysis. Am J Epidemiol 164: 881-889.
- Food and Drug Administration. Department of Health and Human Services. Questions and Answers about Oral Prescription Acetaminophen Products to be Limited to 325 mg Per Dosage Unit.