MINI-REVIEW
Treating fever in
children: paracetamol
or ibuprofen?
Edward Purssell
arents worry greatly about the effects of high
P
Edward Purssell is a
Lecturer, Florence temperatures on their children (Schmitt,
Nightingale School of
Nursing and Midwifery,
1980; Kramer et al, 1985; Crocetti et al,
King’s College London, 2001), and the treatment of febrile illnesses in chil-
London dren with antipyretic drugs such as paracetamol
(acetaminophen) and ibuprofen is widespread. The
rationale for this is to relieve any discomfort associ-
ated with high temperatures (Temple, 1983;
Hopkins et al, 1991). However, the therapeutic ben-
efits of antipyretics are often overestimated; though
sometimes advocated as a means of reducing the
incidence of febrile convulsions (Temple, 1983)
there is no evidence that they are effective in this
(Purssell, 2000a).
Since much time is spent treating high tempera-
ture in children, and in view of parental concerns,
identification of the most effective method of
antipyresis is an important subject. It is particular-
ly important for nurses and health visitors to con-
sider this subject because they are the health pro-
fessionals primarily responsible for the treatment
of febrile children in hospital and community set-
tings, and advising parents and carers. This mini-
Series Editor: review (Griffiths, 2002) builds on previous reviews
Peter Griffiths looking at the use of antipyretics as prophylaxis
ABSTRACT
Community health practitioners frequently prescribe or advise parents on
antipyretic medications for children with fevers. This mini-review sets out to
examine the evidence for the relative effectiveness of two of the most widely
available and commonly used over the counter medicines - paracetamol and
ibuprofen. A systematic literature search was undertaken to identify al stud-
ies comparing the effects of the two drugs. The Medline, Embase, Cinahl
and RCN databases were searched. Eight randomised controlled trials that
reported temperature differences at time-points between 1 and 6 hours
after administration were identified. Statistical meta-analysis showed no
clear benefit for one drug over another 1 hour after administration. However,
by 6 hours after administration ibuprofen was clearly superior resulting in a
mean temperature 0.58°C lower than paracetamol. Both drugs appeared
well tolerated and no evidence of difference in short-term adverse effects
was observed. Both drugs are effective antipyretics but the longer action of
ibuprofen may make it preferable in some circumstances.
316
ish Journal of Community Nursing.Downloaded from magonlinelibrary.com by 130.216.129.208 on June 29, 2015. For personal use only. No other uses without permission. . All rights reserv
against febrile convulsions and tepid sponging as a
treatment of high temperature (Purssell, 2000a,b).
Method
The hypothesis tested was that ibuprofen is a more
effective than paracetamol in reducing temperature
in children with pyrexia. Additionally, any reported
side-effects of either treatment were noted.
Medline, Embase, CINAHL and the Royal College
of Nursing database were searched for all years
from 1970 using the following search terms in the
combinations shown in Table 1: children, infants,
pediatric, paediatric, fever, febrile, pyrexia, temper-
ature, paracetamol, acetaminophen, ibuprofen.
Textbooks and reference lists were also searched.
Criteria for inclusion were that each study should
be primary research comparing oral paracetamol and
ibuprofen as treatments for fever in children, and
should include sufficient statistics to extract mean
temperature and effect size at either or all of 0, 1, 2,
4 and 6 hours. The search yielded 13 studies com-
paring paracetamol and ibuprofen, five of which
were excluded (these are shown in Box 1). Data
extraction was done by one person (the author), and
so the potential for bias or error in extraction and
interpretation exists.
One problem when comparing these studies is the
lack of uniformity about the dosage of drugs used
(treatment doses of paracetamol ranged from
8 mg/kg to 15 mg/kg, and those for ibuprofen from
5 mg/kg to 10 mg/kg). As the recommended dosages
of both paracetamol and ibuprofen exist in a range,
however, the differing doses used in the studies may
reflect the way in which these drugs are used in clin-
ical practice. Some recommended and maximum
dosages have recently been changed, that for parac-
etamol now being up to 20 mg/kg every 6 hours and
ibuprofen 4–10 mg/kg every 4–6 hours (Guy’s, St
Thomas’ and Lewisham Hospitals, 2001).
Effect sizes (a measurement of the size of the rela-
tionship between two variables) were calculated
using META Meta-Analysis Programme 5.3
(Schwarzer, 1991), a free meta-analysis software
package that also calculates the homogeneity or
British Journal of Community Nursing, 2002, Vol 7, No 6
 heterogeneity of the studies. The test of homogeneity
seeks to establish whether the observed variance
within the studies analysed is greater than that which
would be expected if the overall population shared a
common effect size (Shadish and Haddock, 1997). In
other words, a high degree of heterogeneity suggests
that the studies are different in some respect.
Effect sizes were calculated from mean tempera-
tures and standard deviations where these were
given; if these were not, other statistics were used
where possible. These must be used with caution
however, as use of different methods for this calcu-
lation may bias the resultant effect size and so reduce
the reliability of the analysis (Ray and Shadish,
1996). A further drawback is that not all studies gave
statistics for all time points.
Results
Eight studies were analysed (Table 2). Five of them
clearly used random allocation to treatment groups.
However, even when they used randomisation it was
not always clear how it had been done, and in the
three remaining studies it was not clear if they had
randomised at all. Because of the different formula-
tions of the drugs, staff and subjects were not always
blinded to the treatment that they were receiving, with
only the studies by Autret et al (1994), Kauffman et al
(1992), Van Esch et al (1995) and Walson et al (1989)
being double-blinded. Six were based on a single clin-
ic or centre, although Autret et al (1994) and Joshi et
al (1990) used multiple centres. All studies except
that by Simila et al (1976) provided full inclusion and
exclusion criteria, and all gave the full treatment reg-
imen and outcome measures. Although studies used
different methods of recording temperatures, this is
not considered to be significant because the aim of
this analysis is to look at the differences between the
treatments within each study, rather than comparing
studies with each other directly.
All of the children studied were enrolled because
they had a high temperature as the result of one of a
variety of conditions. In most (though not all) of the
studies, the conditions were stated, and the mix of
conditions reflects those that cause fever in the pae-
diatric population, such as infections of the urinary
tract, upper and lower respiratory tracts, and unclas-
sified viral infections. Most children were other-
wise well, although one study (Van Esch et al, 1995)
looked at children who had a febrile convulsion and
then had a rectal temperature of over 38.5°C. The
subject of antipyretics as prophylaxis or treatment
for febrile convulsions has been the subject of a
recent review elsewhere (Purssell, 2000a) and so is
not considered any further here. The ages of the
children studied varied somewhat, from 4 months to
British Journal of Community Nursing, 2002, Vol 7, No 6
ish Journal of Community Nursing.Downloaded from magonlinelibrary.com by 130.216.129.208 on June 29, 2015. For personal use only. No other uses without permission. . All rights reserv
Table 1. Search strategy used
children A fever A paracetamol A ibuprofen
OR N OR N OR N
infants D febrile D acetaminophen D
OR OR
pediatric pyrexia
OR OR
paediatric temperature
13 years. The effect of age or condition on the
antipyretic efficacy is not considered in detail here.
The results of the analysis are shown in Table 3.
At 1 hour, the studies were evenly split between
those showing greater effect from paracetamol and
ibuprofen, resulting in some heterogeneity in the
results (although this was not statistically signifi-
cant). As the doses in the studies favouring both
treatments cover the range of recommended doses, it
is unlikely that dosages alone account for the hetero-
geneity seen. One possible explanation is the small
mean changes in temperature – use of paracetamol
resulted in a mean fall in temperature of just 0.24°C,
while ibuprofen reduced the mean temperature by
0.325°C, a difference of less than 0.1°C.
At 4 hours all studies reported a greater effect
from ibuprofen – the mean difference in temperature
between those treated with paracetamol and ibupro-
fen was 0.63°C. At 6 hours there was also a high
degree of homogeneity in the results, showing a
greater mean fall in those receiving ibuprofen of
0.58°C. The difference in the effect size between the
treatments was statistically significant (Table 3).
The differences that exist appear to be unrelated to
dosages of the drugs. This is further supported by the
high degree of homogeneity across drug doses at 4
and 6 hours. Overall it appears that ibuprofen is more
effective than paracetamol, particularly at 4 and 6
hours where data is both homogeneous and the differ-
ences between treatments are statistically significant.
Discussion
Analysis of drug levels in the blood suggests that
ibuprofen has a longer effect than paracetamol.
Box 1. Studies that were excluded from the analysis
Walson et al (1992), a multi dose study with first statistics at 6 hours.
Seth et al (1980) which provided inadequate statistics for all but 12 hours.
Wilson et al (1984) which used the rectal route for the paracetamol.
McIntyre and Hull (1996) which provided insufficient statistics.
Wilson et al (1991) which reported statistics only for the whole 6-hour
period, rather than for each hour.
317
 MINI-REVIEW

were wearing. As a result of this it is impossible to

Table 3. Differences in temperature between ibuprofen say what effect environment plays in the results
and paracetamol treatments, meta-analysis (positive reported, although the fact that studies were ran-
mean difference indicates greater efficacy for ibuprofen)
domised means that these factors should balance out
between groups and would not alter the conclusions
Mean Number of
difference in studies and about differences.
Time temperature (°C) 95% CI:°C subjects p Age may be a factor in the febrile response to
infection. In young infants under the age of 1 month,
1 hour -0.01 -0.04:0.02 5 studies 0.22
n = 448 the degree of fever may correlate with the rate of
serious bacterial infection (Bonadio et al, 1990). The
4 hours 0.63 0.59:0.67 6 studies 0.00003 studies discussed here all looked at broad age
n = 423
ranges. Walson et al (1989) found no difference in
6 hours 0.58 0.52:0.64 5 studies 0.005 response when they subdivided their sample into
n = 267 2–4, 5–8 and 9–11-year-old children. The study by
Wilson et al (1991) suggested that children over the
Kelley et al (1992) studied drug levels after the age of 6 years had a decreased response to the
administration of paracetamol 12mg/kg and ibupro- 5mg/kg dose of ibuprofen. Ibuprofen is not licensed
fen 6mg/kg, finding that the paracetamol reached for use in children under the age of 6 months in
peak levels at 27 minutes, although the maximum Britain (Guy’s, St Thomas’ and Lewisham
antipyretic effect was not reached until 133 minutes. Hospitals, 2001), although Wilson et al (1991) and
For ibuprofen the results were 54 and 183 minutes Simila et al (1976) used the drug with children as
respectively. From this it appears that antipyretic young as 3 months of age with no adverse effects.
effect is not related to blood concentration alone. Initial temperature also appears to have an effect on
Most studies comparing paracetamol and ibupro- the efficacy of antipyretic drug therapy. Wilson et al
fen give little information about the environment in (1991) found that children with high initial tempera-
which the children were treated or what clothing they tures (38.8°C or higher) had significantly less
response to the drugs than those with lower tempera-
Table 2. Details of studies included in analysis tures. This may be linked to their finding of less
antipyretic response among those children with posi-
Treatment tive blood cultures. Bonadio et al (1990) suggested
dose that there is a relationship between level of pyrexia
Study (date) n Randomisation Follow-up (mg/kg) and serious bacterial infection. However, while high
Aksoylar et al 101 Random, not 1, 2 hours P 15 initial temperature is proportional to the rate of seri-
(1997) blind I 8 ous bacterial infection, as a diagnostic tool it lacks
Amdekar and 38 Randomisation 1, 2, 4, 6 P 8 specificity (Bonadio et al, 1991).
Desai (1985) not clear, hours I 7 Most of the studies used accident and emergency
blinding not departments, children on inpatient wards, or groups
clear of both to recruit their samples. One (Walson et al,
Autret et al 154 Random, 4 hours P 10 1989) used paid volunteers. Consequently the stud-
(1994) double blind I 7.5 ies in this analysis cover children with a large variety
Joshi et al 175 Randomisaton 1, 2 hours P 8 of underlying conditions, particularly upper and
(1990) not clear, I 7 lower respiratory tract infections, otitis media, gas-
blinding not trointestinal infections and general systemic viral
clear infections. Insufficient data is given to make any
Kauffman et al 28 Random, 2, 4, 6 P 10 predictions about how the different diagnoses might
(1992) double blind hours I 7.5 affect the response to the drugs, so the results pre-
Simila et al 38 Randomisation 1, 2, 4, 6 P 12.5 sented here can only be generalized to the group of
(1976) not clear, not hours I 0.5 & 6 children covered by the samples included in the
blind studies. This does, however, cover the kinds of con-
Van Esch et al 72 Random, 2, 4, 6 P 10 ditions that would be seen in general paediatric prac-
(1995) double blind hours I 5 tice and allows us to see what effects the treatments
Walson et al 93 Random, 1, 2, 4, 6 P 10 may have on children seen in such settings.
(1989) double blind hours I 5 & 10 The method of recording temperature is another
variable that differs between studies. However, as
P = paracetamol; I = ibuprofen
temperatures are not being compared between

318 British Journal of Community Nursing, 2002, Vol 7, No 6

ish Journal of Community Nursing.Downloaded from magonlinelibrary.com by 130.216.129.208 on June 29, 2015. For personal use only. No other uses without permission. . All rights reserv
 studies, this is less important; the primary measure
here was detecting differences within each study.
Therefore, as long as the temperature measure is reli-
able within each study, the exact method by which it
is done is less important. One study showed wide
variation in the time that thermometers were left to
record temperature – 1–5 minutes for rectal tempera-
tures, and 3–10 minutes for axilla temperatures (Eoff
and Joyce, 1981). The same study found that greatest
temperature change occurred within 3 minutes for
rectal temperatures and 5 minutes for axilla tempera-
tures using glass thermometers. There is therefore the
possibility of reduced reliability caused by operator
error if such variations were to occur, or if there were
other errors such as instrumentation differences. It is
impossible to assess what part these might play in the
findings here. Nevertheless, where there is a high
degree of homogeneity between studies, it would
suggest that random error is not significant.
Tolerability and side-effects.
Antipyretic drugs such as paracetamol and ibupro-
fen have the potential to cause toxicity and side-
effects (Plaisance, 2000), so studies were examined
for signs of any such complications. The quality of
the reporting of side-effects varied. For example,
while Aksöylar et al (1997), Kauffman (1992) and
Joshi et al (1990) report no side-effects, other stud-
ies reported more symptoms. Even then, a number
of the symptoms were not related to the drugs, and
even among those that may have been due to the
drugs it is impossible to say for sure that the drugs
were directly responsible. Indeed Walson et al
(1989) reported that some symptoms, such as respi-
ratory complaints, were more common among the
placebo group than either of the treatment groups.
More reliable, perhaps, are the findings from the
same study using laboratory data which showed no
significant difference in haematological or bio-
chemical findings that could not be explained
through dehydration or disease. The same was true
for vital signs and physical examination (Walson et
al, 1989). Furthermore, Autret et al (1994) reported
that the acceptability of both drugs was good. A
study comparing paracetamol with an ibuprofen-
paracetamol combination showed that even combin-
ing the drugs did not appear to have any deleterious
effects (Lal et al, 2000).
Although no great differences were seen in the tol-
erability of the treatments, it must be remembered
that exclusions within the study samples prevent this
from being generalized to the entire paediatric popu-
lation (Walson et al, 1992). For example, most stud-
ies specifically excluded those with haematological
disorders and malignancies, and so it would be incor-
British Journal of Community Nursing, 2002, Vol 7, No 6
ish Journal of Community Nursing.Downloaded from magonlinelibrary.com by 130.216.129.208 on June 29, 2015. For personal use only. No other uses without permission. . All rights reserv
rect to assume that both drugs would be equally well
tolerated in these groups.
Conclusion
Both paracetamol and ibuprofen act as antipyretic
drugs. The aim of this mini-review was to identify
and quantify any difference in efficacy and side-
effects between them. The analysis suggests that
ibuprofen is a more potent antipyretic, both in terms
of maximum temperature fall and the length of
antipyretic action. The clinical usefulness of these
differences is harder to quantify – in many cases the
treatments are of equal value in terms of antipyretic
outcome. However individual practitioners may find
situations where such differences are of value,
although the value and advisability of antipyresis has
yet to be proven (Mackowiak, 2000).
Febrile illness are common among children, and
use of antipyretic agents is common. Treatments KEY POINTS
should be evidence-based wherever possible, yet there
● Treatment of fever
is evidence that with respect to antipyresis, this is still
in children is of
not the case (Mayoral et al, 2000). In particular it
great concern to
appears that many parents may be giving incorrect
parents.
doses of antipyretic medications, and that despite 20
years of parental education by nurses and other health- ● There is no
care professionals, relatively little progress has been evidence that
made in reducing ‘fever phobia’ (Crocetti et al, 2001). treatment of fever
Nurse prescribing makes it important that nurses is of benefit other
understand the drugs that they are prescribing and than for comfort of
advocating, and play a part in the evaluation of the the child and
efficacy and side-effects of these drugs. In the future reassurance of the
the scope of nurse prescribing will be extended to parent.
other groups, making knowledge of the most effec- ● Both paracetamol
tive drugs for conditions such as this more important and ibuprofen are
for larger numbers of nurses. Even those not pre- effective
scribing may be responsible for choosing, adminis- antipyretics, and
tering or recommending antipyretic drugs, and so both are well
studies such as this do have a wider value than those tolerated.
prescribing alone. ● Ibuprofen appears
The findings of this study suggest that there is lit- to have a longer
tle to recommend one drug over the other for their action and is more
immediate effect, although ibuprofen has a longer effective than
lasting effect which could reduce the number of paracetamol
doses required and reduce parental fears. ■ between 4 and 6
hours after
Aksöylar S, Akşit, S, Çağlayan S, Yaprak I, Bakiler R, Çetin administration.
F (1997) Evaluation of sponging and antipyretic medica-
tion to reduce body temperature in febrile children. Acta ● Ibuprofen may be
Peadiatr Jpn 39: 215–7 preferred in some
Amdekar YK, Desai RZ (1985) Antipyretic activity of
ibuprofen and paracetamol in children with pyrexia. Br J circumstances
Clin Pract 39(4): 140–3 because of its
Autret E, Breart G, Jonville AP, Courcier S, Lassale C,
Goehrs JM (1994) Comparative efficacy and tolerance of longer action. The
ibuprofen syrup and acetaminophen syrup in children clinical significance
with pyrexia associated with infectious diseases and
treated with antibiotics. Eur J Clin Pharmacol 46: of this is unclear.
197–201
Bonadio WA, Romine K, Gyuro J (1990) Relationship of
319
 MINI-REVIEW
fever magnitude to rate of serious bacterial infection
in neonates. J Pediatr 116(5): 733–5
Bonadio WA, McElroy K, Jacoby PL, Smith D (1991)
Relationship of fever magnitude to rate of serious bac-
terial infection in infants aged 4-8 weeks. Clin Pediatr
30(8): 478–80
Crocetti M, Moghbeli N, Serwint J (2001) Fever phobia
revisited: have parental misconceptions about fever
changed in 20 years? Pediatrics 107: 1241–6
Eoff MJF, Joyce B (1981) Temperature measurements in
chidren. Am J Nurs 81(5): 1010–1
Griffiths P (2002) Evidence informing practice: intro-
ducing the mini-review. Br J Community Nurs 7(1):
38–40
Guy’s, St Thomas’ and Lewisham Hospitals (2000)
Paediatric Formulary. Guy’s, St Thomas’ and
Lewisham Hospitals, London
Hopkins A, for the Joint Working Group of the Research
Unit of the Royal College of Physicians and the
British Paediatric Association (1991) Guidelines for
the management of convulsions with fever. BMJ 303:
634–6
Joshi YM, Sovani VB, Joshi V et al (1990) Comparative
evaluation of the antipyretic efficacy of ibuprofen and
paracetamol. Indian Pediatr 27: 803–6
Kauffman RE, Sawyer LA, Scheinbaum ML (1992)
Antipyretic efficacy of ibuprofen vs acetaminophen.
Am J Dis Child 146: 622–5
Kelley MT. Walson PD. Edge JH. Cox S. Mortensen ME
(1992) Pharmacokinetics and pharmacodynamics of
ibuprofen isomers and acetaminophen in febrile chil-
dren. Clin Pharmacol Ther 52(2): 181–9
Kramer M, Naimark L, Leduc DG (1985) Parental fever
phobia and its correlates. Pediatrics 75(6): 1110–3
Lal A, Gomber S, Talukdar B (2000) Antipyretic effects
of nimesulide, paracetamol and ibuprofen-paraceta-
mol. Indian J Pediatr 67(12): 865–70
Mackowiak PA (2000) Physiological rationale for sup-
pression of fever. Clin Infect Dis 31: S185–S189
Mayoral CE, Marino RV, Rosenfeld W, Greensher J
(2000) Alternating antipyretics: is this an alternative?
Pediatrics 105(5): 1009–12
McIntyre J, Hull D (1996) Comparing efficacy and tol-
erability of ibuprofen and paracetamol in fever. Arch
Dis Child 74: 164–7
Plaisance KL (2000) Toxicities of drugs used in the man-
agement of fever. Clin Infect Dis 31: S219–S223
ish Journal of Community Nursing.Downloaded from magonlinelibrary.com by 130.216.129.208 on June 29, 2015. For personal use only. No other uses without permission. . All rights reserv
Purssell E (2000a) The use of antipyretic medications in
the prevention of febrile convulsions in children. J
Clin Nurs 9(4): 473–80
Purssell E (2000b) Physical treatment of fever. Arch Dis
Child 82: 238–9
Ray JW, Shadish WR (1996) How Interchangeable are
different estimators of effect size? Consulting and
Clinical Psychology 64(6): 1316–25
Schmitt BD (1980) Fever phobia. Am J Dis Child 134:
176–81
Schwarzer R (1991) META meta-analysis programmes v
5.3. Freie Universitat, Berlin. http://www.yorku.ca/
faculty/academic/schwarze/meta_e.htm (accessed
20/05/02)
Seth UK, Gupta K, Paul T, Pispati PK (1980)
Measurement of antipyretic activity of ibuprofen and
paracetamol in children. J Clin Pharmacol 20(11-12):
672–5
Shadish WR, Haddock CK (1997) Combining estimates
of effect size. In: Cooper H, Ledges LV eds. The
Handbook of Research Synthesis. Russell Sage
Foundation, New York: 261–81
Simila S, Kouvalainin K, Keinanen S (1976) Oral
antipyretic therapy. Scand J Rheumatol 5: 81–3
Temple AR (1983) Review of comparative antipyretic
activity in children. Am J Med 75: 38–46
Van Esch A, Van Steensel-Mol HA, Steyerberg EW,
Offringa M, Habbema DF, Derksen-Lubsen G (1995)
Antipyretic efficacy of ibuprofen and acetaminophen
in children with febrile seizures. Arch Pediatr Adolesc
Med 149: 632–7
Walson PD, Galletta G, Braden NJ, Alexander L (1989)
Ibuprofen, acetaminophen, and placebo treatment of
febrile children. Clin Pharmacol Ther 46(1): 9–17
Walson PD, Galletta G, Chomilo F, Braden NJ, Sawyer
LA, Scheinbaum ML (1992) Comparison of multi-
dose ibuprofen and acetaminophen therapy in febrile
children. Am J Dis Child 146: 626–32
Wilson G, Guerra AJMS, Santos NT (1984)
Comparative study of the antipyretic effect of ibupro-
fen (oral suspension) and paracetamol (suppositories)
in paediatrics. J Int Med Res 12(4): 250–4
Wilson JT, Brown D, Kearns GL et al (1991) Single-
dose, placebo-controlled comparative study of ibupro-
fen and acetaminophen antipyresis in children. J
Pediatr 119(5): 803–11