Cohort study of hepatotoxicity associated with nimesulide and other non-steroidal anti-inflammatory drugs
BMJ 2003; 327 doi: https://doi.org/10.1136/bmj.327.7405.18 (Published 03 July 2003) Cite this as: BMJ 2003;327:18All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Dear Editor,
We are writing in response to the article by Traversa et al.
assessing the hepatic damage associated with nimesulide and other non-
steroidal anti-inflammatory drugs (NSAIDs) recently published in your
journal (1); the article presents a study in which the risks were
estimated in a retrospective cohort of users of such drugs in the region
of Umbria (Italy). The main stated conclusion of this study is that the
risk of liver injury in patients taking nimesulide and other NSAIDs is
small. This may be so but the results also confirm that the risk is higher
for nimesulide. We have recently published some estimates with data from
the Spanish Pharmacovigilance System reaching a conclusion that clearly
differs from that of these authors (2) and would like to specifically
comment on some aspects of the study as well as discuss how small risk
ratios may have strong public health impact:
1. It is at least arguable that with a rate ratio of 1.9 (95% CI, 1.1
-3.8) for the more severe hepatic injuries –when compared to the rest of
NSAIDs– nimesulide is still regarded as having a small risk. Furthermore,
since the drug is in widespread used in Italy, the population attributable
risk could account for a considerable number of cases that could be
avoided: 25 millions of prescriptions of nimesulide were issued in 2001,
corresponding to 18 DDDs per 1000 inhabitants per day (3). With the rate
figures from this study (more severe liver injury: nimesulide, 29.0 per
100 000 person-years; other non-steroidal non-antiinflammatory drugs, 15.6
per 100 000 person-years) and the above figures of consumption, around 140
cases of severe liver injury could be expected to appear in excess with
nimesulide in one year in Italy.
2. Based on the numbers of packets per prescription, it is assumed
that the characteristics of users of the different non-steroidal anti-
inflammatory drugs were similar. If, rather than the number of packets,
the number of DDDs per person had been taken –a more reliable approach–,
the conclusion would have been different since the mean number of DDDs per
user of nimesulide and NSAIDs were in fact 50.9 and 73.2, respectively.
This does mean that these drugs are being used in a different manner,
nimesulide being used for shorter periods or, unlikely, in smaller doses.
Also, the number of prescriptions per user was different: nimesulide
accounted for 2.9 and NSAIDs for 4.1 (not 2.2 as presented in table 1;
also the number of prescriptions per user when considering “any NSAID” is
not 2.4 as stated but 4.9).
3. An acute pattern of use of nimesulide when compared to the other
NSAIDs has been consistently observed in Italy. It was described for the
first time by the same authors with information coming from the same
database although in the period 1993-1994 (4); then it was concluded that
nimesulide was “used mainly in the symptomatic treatment of milder
conditions (e.g., fever and upper respiratory infections) and the duration
of use is shorter”. Recently, with data from the same database and for the
period June 2000-July 2001, a similar pattern was found: most of the
patients received nimesulide for shorter periods or in small amounts:
92.7% of nimesulide versus 54.1% of other NSAIDs users had less than 15
DDDs (3).
4. Time of exposure is an important feature for some types of
hepatotoxicity to appear. A substantial proportion of nimesulide-induced
hepatotoxicity occurred after relatively long time of exposure. In Spain,
5 out of the 11 most severe cases of hepatotoxicity related to nimesulide
appeared following more than 1 month of treatment; the average duration
until occurrence in a series of 13 published cases of severe
hepatotoxicity was 62 days (5). As the use in Spain for nimesulide was
mostly as an anti-inflammatory for chronic conditions and longer periods,
hepatotoxicity was more likely to appear. Noteworthy, the study by
Traversa et al (1) based on a particular pattern of use, may have missed
the most severe cases appearing after longer treatment periods.
In conclusión, the epidemiological study by Traversa et al (1) have
shown that nimesulide has an increased risk of acute liver injury, in
particular for the most severe cases, as compared to other NSAIDs,
confirming the signal raised by the spontaneous reporting systems of Spain
and Finland. Probably, the way the drug is being used in Italy –in shorter
periods of time– precludes to some extent the occurrence of a number of
cases –also the publicity given to this particular association has
refrained many doctors from using the drug in those prone to develop
hepatic damage. Notwithstanding, widespread use could account for an
excess of this type of reactions. On the other hand, nimesulide does not
appear to have any demonstrated advantage over the most widely used NSAIDs
either in efficacy or safety. Regarding upper gastrointestinal
complications, it has been shown by García-Rodríguez et al in a case-
control study performed in Italy (nimesulide OR=4.4; 95% CI: 2.5-7.7;
ibuprofen, OR=2.1; 0.6-7.1; diclofenac, OR=2.7, 1.5-4.8) (6). Bearing all
this in mind, our opinion is that the comparative benefit-risk balance of
nimesulide is very questionable and we would discourage its generalized
use.
Alfonso Carvajal
Professor of Pharmacology
Institute of Pharmacoepidemiology,
University of Valladolid (Spain)
Miguel Angel Maciá
Pharmacovigilance Centre of Castilla-La Mancha,
Toledo (Spain)
Javier García del Pozo*
Francisco de Abajo*
Dolores Montero*
Fernando de Andrés-Trelles*
* Spanish Medicines Agency,
Madrid (Spain)
References
1. Traversa G, Bianchi C, Da Cas R, Abraha I, Minneti-Ippolito F,
Venegoni M. Cohort study of hepatotoxicity associated with nimesulide and
other non -steroidal anti-inflammatory drugs. BMJ 2003;327:18-22.
2. Macia MA, Carvajal A, Garcia del Pozo J, Vera E, del Pino A.
Hepatotoxicity associated with nimesulide: data from the Spanish
Pharmacovigilance System. Clin Pharmacol Ther 2002; 72:596-7.
3.http://www.ministerosalute.it/medicinali/resources/documenti/note_inform...
4. Menniti-Ippolito F, Maggini M, Raschetti, Da Cas R, Traversa G,
Walker AM. Ketorolac use in outpatients and gastrointestinal
hospitalization: a comparison with other non-steroidal anti-inflammatory
drugs in Italy. Eur J Clin Pharmacol 1998; 54: 393-7.
5. Merlani G, Fox M, Oehen HP, Cathomas G, Renner EL, Fattinger K,
Schneemann M, Kullak-Ublick GA. Fatal hepatoxicity secondary to nimesulide
Eur J Clin Pharmacol 2001; 57: 321-6.
6. García Rodríguez LA, Cattaruzz C, Troncon MG, Agostinis L. Risk of
hospitalization for upper gastrointestinal tract bleeding associated with
ketorolac, other nonsteroidal anti-inflammatory drugs, calcium
antagonists, and other antihypertensive drugs. Arch Intern Med 1998;158:
33-39.
Competing interests:
None declared
Competing interests: No competing interests
Editor - The article by Traversa et al. provided an interesting
estimation of the risk of acute hepatotoxicity associated with nimesulide in
comparison with other non-steroidal anti-inflammatory drugs (NSAIDs).1
Overall, the risk for all hepatopathies in patients exposed to nimesulide was
small and not significantly higher than that of other NSAIDs, with the
exception relating to more severe liver injuries.
We are currently performing a similar cohort study on the entire
population listed in the Local Health Unit (LHU) of Ravenna by cross-linking
information stored in different databases on a patient-by-patient basis. The
analysis of all prescriptions for nimesulide or other NSAIDs registered in the
pharmaceutical database in the period between January 1st 1997 and
August 31th 2002 (with an observation period of 5.7 years) generated
three different cohorts: patients exposed to nimesulide (n=45,884), those
exposed to other NSAIDs (n=113,316), and the control group of those never
exposed to any NSAID (n=239,513). In each patient of each cohort, on the basis
of the personal health number, from the nosocomial database have been retrieved
all hospital admissions for liver injury through the following ICD-9 codes:
570, 573.3, 573.8, 573.9. A total number of 587 patients hospitalised for liver
injury has been classified as eligible for analyses. The frequency of
hospitalisation in each cohort is reported in table 1.
Pharmacoutilisation pattern |
Nimesulide ‰ (cases/total) |
Other NSAIDs ‰ (cases/total) |
Population not exposed to NSAIDs ‰ (cases/total) |
p-value* |
Single |
1.80 (51/28,392) |
1.57 (80/50,901) |
- |
0.228 |
Multiple |
1.77 (31/17,492) |
1.28 (80/62,415) |
- |
0.062 |
Total |
1.79 (82/45,884) |
1.41 (160/113,316) |
- |
0.040 |
|
1.79 (82/45,884) |
- |
1.44 (345/239,513) |
0.039 |
*z-test
A significantly higher frequency of liver injury has been observed among
patients exposed to nimesulide either compared to that of patients treated with
NSAIDs or that of population who never received any NSAID. Interestingly, no
significantly difference was found between patients exposed to other NSAIDs and
population never exposed. Based on the registered frequencies, two
epidemiological measures of impact relating to nimesulide were calculated: the
population attributable risk (PAR) and the attributable risk percentage (ARP).2
The PAR, that considers the proportion of cases of liver injury in the entire
population which can be attributed to the exposure to nimesulide, corresponded
to 4%. Furthermore the ARP, that considers in the exposed group the proportion
of cases which can be attributed to the exposure itself accounted for 19.5%.
Contrary to the study by Traversa et al.,1 the chosen
approach to cross-link different databases on a patient-by-patient basis gave
the advantage to measure the actual epidemiological impact concerning liver
injuries attributable to an exposure to nimesulide. Data we evaluated provide
an indication suggesting that liver injury might be not an infrequent
experience and raise some important questions about the real risk of
hepatotoxicity associated with an exposure to nimesulide. This adverse event
surely deserves more attention, also considering the mean cost of 3,5 thousand
euros per patient per event, incurred by the LHU of Ravenna, which was
calculated on a fixed value (the DRG’s tariff) including large part of
hospitalisation expenses exclusively.
References
1 – Traversa G, Bianchi C, Da Cas R, Abraha I, Menniti-Ippolito F,
Venegoni M. Cohort study of hepatotoxicity associated with nimesulide and other
non-steroidal anti-inflammatory drugs. BMJ
2003;327:18-22
2 – Szklo M, Nieto FJ. Epidemiology: beyond the basics. Gaithersburg,
Maryland: AN Aspen Publication, 2000
Competing interests:
None declared
Competing interests:
Sir,
In a retrospective cohort and nested case-control study, Traversa et
al(1) compared the incidence of acute hepatotoxicity associated with
nimesulide and certain other non-steroidal anti-inflammatory drugs
(NSAIDs). They concluded that the risk of liver injury in patients taking
nimesulide was small, albeit slightly higher than for other NSAIDs. In
fact, the risk of all hepatopathies among current users of nimesulide was
comparable to that for other NSAIDs (rate ratio 1.3, 95% confidence
interval 0.7 to 2.3). However, this rate ratio was 1.9 (1.1 to 3.8) when
only more severe diagnoses of liver injury were included in the
analysis(1). In other words, the severity of acute hepatic injury rather
than its frequency appeared to be increased in patients receiving
nimesulide in comparison with those receiving other NSAIDs. Based on
spontaneous reports to the Spanish Pharmacovigilance System (2) and the
World Health Organization (3), nimesulide was also found to induce a high
proportion of severe hepatic adverse reactions compared with other NSAIDs.
This may explain why nimesulide was removed from the market in some
countries.
Since no fulminant hepatitis was observed in their study, the authors
speculated that some of the spontaneously reported cases of fulminant
hepatitis attributed to NSAIDs were linked to concomitant conditions that
would lead to exclusions in epidemiological studies (1). In some cases,
however, a causal relationship between the occurrence of fulminant
hepatitis and nimesulide use was very likely (3,4).
Finally, the authors underlined the need to evaluate the overall risk
profile of NSAIDs, particularly the risk of serious gastrointestinal
complications (1). In this respect, nimesulide does not offer any
advantage over conventional NSAIDs such as diclofenac, ibuprofen or
naproxen (5). In summary, nimesulide should not be considered as an NSAID
of first choice.
1 Traversa G,Bianchi C,Da Cas R,Abraha I,Minneti-Ippolito F,Venegoni
M. Cohort study of hepatotoxicity associated with nimesulide and other non
-steroidal anti-inflammatory drugs. BMJ 2003;327:18-22.
2 Macia MA,Carvajal A,Garcia del Pozo J,Vera E,del Pino A. Hepatotoxicity
associated with nimesulide: data from the Spanish Pharmacovigilance
System. Clin Pharmacol Ther 2002; 72:596-7.
3 Merlani G,Fox M,Oehen HP,Cathomas G,Renner EL,Fattinger K, et al. Fatal
hepatotoxicity to nimesulide. Eur J Clin Pharmacol 2001;57:321-6.
4 Dumortier J,Borel I,Delafosse B,Vial T,Scoazec JY, Boillot O.
Transplantation hépatique pour hépatite subfulminante après prise de
nimésulide. Gastroenterol Clin Biol 2002;26:415-6.
5 Garcia Rodriguez LA,Cattaruzzi C,Troncon MG,Agostinis L. Risk of
hospitalization for upper gastrointestinal tract bleeding associated with
ketorolac, other nonsteroidal anti-inflammatory drugs, calcium
antagonists, and other antihypertensive drugs. Arch Intern Med 1998;158:33
-9.
Competing interests:
None declared
Competing interests: No competing interests
Dear Sir
I can not retrieve which NSAID's was implicated in the study. I would like
to how many out of all prescription were Diclofenac. This drug was
involved in many case of hepatitis, in large number of publication.
Competing interests:
None declared
Competing interests: No competing interests
Authors reply
In the conclusion of his letter, prof. Bannwarth affirms that
“nimesulide should not be considered as an NSAID of first choice” and
that, with regard to serious gastrointestinal complications, it “does not
offer any advantage over conventional NSAIDs such as diclofenac,
ibuprofen, or naproxen” (1). We entirely share the widely accepted view
that ibuprofen, diclofenac, and naproxen, should be considered as first
choice NSAIDs: these substances have been studied in a large number of
efficacy trials, and represent the reference NSAIDs for the study of
gastroduodenal toxicity, the main adverse reaction associated with NSAID
use.
When we planned our study, at the end of the summer 2002, the
discussion did not concern the selection of first choice NSAIDs. We were
facing the signal of a greater hepatotoxicity associated to nimesulide
use, which had brought to suspending the marketing of the drug in Finland
and later in Spain. Despite the fact that the signal was not confirmed in
Italy, it was imperative to conduct further analyses since nimesulide was
the most frequently prescribed NSAID in Italy. The overall conclusions of
our study suggested that the increase in the risk of liver damage
associated to NSAID use was low (rate ratio, RR, 1.4; 95% confidence
intervals 0.9 to 2.1) (2). In comparison with other NSAIDs the increase
associated to the use of nimesulide ranged from 1.3 (0.7 to 2.3) when all
hepatopathies were included in the analysis, to 1.7 (0.9 to 3.3) when the
analysis was restricted to liver injuries, to 1.9 (1.1 to 3.8) when
applying a further restriction to more serious forms of liver injury.
Dr. Degli Esposti and colleagues presented some preliminary findings
of a cohort study which is currently performed in Ravenna (Italy) (3). It
is suggested that their data “raise some important questions about the
real risk of hepatotoxicity associated with an exposure to nimesulide”.
For the time being, from the table presented we can derive a relative risk
of liver injury of 1.3 (for nimesulide in comparison with other NSAIDs),
which coincides with the lowest estimate provided in our study. The
publication of the study will allow to understand in more detail both
methods and findings, and may provide a further contribution to the
current discussion on NSAIDs hepatotoxicity.
Most of the comments by prof. Carvajal and colleagues concern the
interpretation of our findings (4). A main objection relates to a
supposedly shorter duration of use of nimesulide in comparison with other
NSAIDs. In Table 1 of our article was indicated that the number of
packages per prescription was similar for nimesulide and for other NSAIDs
(1.2 and 1.3 respectively). Prof. Carvajal et al. recalculated the number
of DDDs and of prescriptions per user. We are sorry that, due to
insufficient information reported in Table 1, we may have induced an
incorrect calculation. To clarify the discrepancy (and to answer dr.
Gottehrer’s question (5) regarding diclofenac) we provide some
prescription details relevant to the three most prescribed NSAIDs (other
than nimesulide), which represent 47.7% of the overall prescriptions.
During the 5 year period, users of diclofenac (464,979 prescriptions and
6,487,494 DDDs), ketoprofen (249,030 prescriptions and 3,445,572 DDDs),
and piroxicam (216,177 prescriptions and 4,801,941 DDDs) received a number
of prescriptions per user (2.6, 2.0, and 2.1, respectively) and DDDs per
user (36.0, 28.3, and 46.8, respectively) which were even lower than for
nimesulide (6). In Table 1 of our article (2), the number of users
referred to individuals who received one or more NSAIDs (if a user
received different NSAIDs he/she contributed only once to the number of
users), whereas in the calculation of prescriptions per user he/she
contributed as many times as the number of different NSAIDs. As a
consequence, a user of any NSAID received 4.9 prescriptions during the
study period, though the average number of prescriptions per user of each
single NSAID was 2.4 (2.2 for the category “other NSAIDs”; Table 1,
reference 2).
Similarly, no discrepancy is present in a pilot analysis we conducted
on the use of nimesulide (7), where in one of the nine tables is indicated
that 92.7% of incident users appear to have received prescriptions of
nimesulide for up to 15 days (versus 54.1 for the other NSAIDs). This is
not surprising when considering that basically all the packages available
in Italy have the same duration, namely 15 days (Appendix ref. 7). Had we
put the cut off at 14 days, we would have found that no user of
nimesulide, and 48% for the other NSAISDs, had received prescriptions for
up to two weeks.
It is obviously a matter of speculation whether a greater
hepatotoxicity would have been estimated, had the actual pattern of use of
nimesulide been consistently longer than the one observed. We would also
discard the possibility that “the publicity given to this particular
association has refrained many doctors from using the drug [nimesulide] in
those prone to develop hepatic damage” (4). On the one hand, our exclusion
criteria should have already removed the population of patients at higher
risk of liver damage. On the other hand, a crude indication of the
concomitant clinical conditions which affected users of different NSAIDs
can be derived from the proportion of concomitant prescriptions (on the
same day of the NSAID prescription) of any other drug (different from
NSAIDs and gastroprotective agents). During the 5 year period the
prescriptions of other drugs associated to NSAIDs were greater for
nimesulide (46.6% of the prescriptions) than for other NSAIDs (40.2% of
the prescriptions) (6).
Prof. Carvajal and colleagues express their doubts that the highest
risk estimate we provided might be regarded as a small increase in the
risk. We understand their concern. What we propose for the present and
future discussions is to adopt a coherent view for safety assessment of
all NSAIDs. It is well known that the main adverse events associated to
NSAID use is represented by gastroduodenal lesions. Even limiting the
analysis to the most severe ones (haemorrhages and perforations), which
carry a distinct risk of death, it has been estimated that around 400
events per 100,000 persons per year are expected among NSAID users
(assuming a relative risk of 4), with around 300 extra cases per 100,000
persons per year (8).
Is there any reason to accept for severe gastroduodenal lesions,
whose incidence among users of NSAIDs may even be 10 times higher than
liver damages, differences in the relative risk far greater than the ones
described in our study? To answer this question let us focus, as an
example, on the study mentioned by prof. Carvajal and colleagues, which
concerns the risk of gastrointestinal bleeding associated with different
NSAIDs. We deem of interest to mention, in addition to the odds ratios
already reported (for ibuprofen, diclofenac, and nimesulide), that a level
of risk similar to nimesulide was observed for naproxen (OR 4.3; 1.6 to
11.2) and tenoxicam (OR 4.3; 1.9 to 9.7), whereas higher risks were
relevant to piroxicam (OR 9.5; 6.5 to 13.8) and ketorolac (OR 24.7; 9.6 to
63.5) (9).
Similar levels of variability have been observed in many other
studies concerning gastroduodenal adverse events associated to NSAID use.
In comparison with these wide ranges of estimates, we believe that the
differences between nimesulide and other NSAIDs in the risk of hepatic
events can be regarded as limited. It is also limited the difference among
individual NSAIDs: in comparison with non use, we estimated a rate ratio
of liver injury of 2.2 (1.3 to 3.9) among users of nimesulide and 1.5 (0.7
to 3.2) among users of diclofenac, the second most prescribed NSAIDs.
We are aware of a risk: the use of highly gastrotoxic NSAIDs does not
justify the prescription of substances whose safety profile may be
considered poorer for other organs. Nonetheless, it would not be wise to
support regulatory actions based on rarer adverse events combined with
lower risk differences. Until an overall revision of NSAIDs as a class is
carried out, though, other interventions are possible. The overall
available evidence already provides a large amount of information to
encourage a cautious use of NSAIDs, and to promote the prescription of
safer NSAIDs in everyday practice.
G. Traversa(1), C. Bianchi(1), R. Da Cas(1), I. Abraha(2), F. Menniti
-Ippolito(1), and M. Venegoni(3)
(1) Dept. of Epidemiology, Istituto Superiore di Sanità, Rome, Italy.
(2) Regional Health Authority of the Umbria Region, Perugia, Italy. (3)
Dept. of Internal Medicine, Fatebenefratelli Hospital, Milano, Italy.
References
1. Bannwarth BG. Overall safety profile of nimesulide. BMJ 327:18-22,
rapid response.
2. Traversa G, Bianchi C, Da Cas R, Abraha I, Menniti-Ippolito F,
Venegoni M. Cohort study of hepatotoxicity associated with nimesulide and
other non-steroidal anti-inflammatory drugs. BMJ 2003;327:18-22.
3. Degli Esposti E, Liverani S, Capone A, Russo P. Liver injury
associated to nimesulide: is it a truly negligible event? BMJ 327:18-22,
rapid response.
4. Carvajal A, Maciá MA, García del Pozo J, de Abajo F, Montero D, de
Andrés-Trelles F. How small risk ratios may have strong public health
impact. BMJ 327:18-22, rapid response.
5. Gottehrer NP. NSAIDs including diclofenac? BMJ 327:18-22, rapid
response.
6. Traversa G, Bianchi C, Da Cas R, Abraha I, Menniti-Ippolito F,
Venegoni M. Nimesulide-associated hepatotoxicity. Epidemiological study in
the Umbria region. Report to the CPMP, May 2003.
7. Traversa G, Bianchi C, Da Cas R, Maggini M. Rapporto sull’uso
della nimesulide in Italia.
http://www.ministerosalute.it/medicinali/resources/documenti/note_inform...
8. Hernandez-Diaz S, García-Rodríguez LA. Epidemiologic assessment of
the safety of conventional nonsteroidal anti-inflammatory drugs. Am J Med
2001; 110: 20S-27S.
9. García-Rodríguez LA, Cattaruzzi C, Troncon MG, Agostinis L. Risk
of hospitalisation for upper gastrointestinal tract bleeding associated
with ketorolac, other non-steroidal antinflammatory drugs, calcium
antagonists, and other antihypertensive drugs. Arch Intern Med 1998;158:33
-39.
Competing interests:
None declared
Competing interests: No competing interests