Results of search
The initial search yielded 11481 studies of which on removing duplicates amounted to 10812 studies. Furthermore, we found 12 studies by searching the references in the included studies. We extracted studies using a reference manager system (46). Finally, 17 studies were included in the analysis:
2 RCTs (47, 48),
5 prospective observational studies (49–53),
2 retrospective observational studies (54, 55),
1 case series (56) and
7 case reports (57–63).
The data from these studies was extracted to personalised data extraction forms (38) of the Cochrane Effective Practice and Organisation of Care (EPOC) (supplementary material). Full-text screened studies, when excluded, were done so with explanation in these forms. Table 1 provides the characteristics of included studies including the study design, number of participants, interventions, outcomes assessed, and the risk of bias.
Totally, there were 25,722 participants from 17 studies.
The interventions in these studies were acetaminophen and its derivatives, and NSAIDs, compared between themselves, external cooling, with placebo or not compared at all.
A meta-analysis was not possible as the data could not be pooled. This was because, i) the study designs were heterogenous ii) the outcomes were measured at different time points, using various techniques and were reported in heterogenous ways and iii) some studies used comparison between interventions, and others did not.
Therefore, a narrative synthesis was conducted, following the guidelines of the Cochrane Consumers and Communications Group (44) (Supplementary material) and the ESRC methods programme (45). The results were synthesised accordingly, and the review reported, in keeping with the PRISMA (64) and the SWiM protocol extension for narrative synthesis (65) (Supplementary material). To assess the certainty of evidence, the guidelines adapting the GRADE assessment for narrative synthesis (43) was followed. The final findings were represented in a summary of findings table.
Grouping by study types
The studies were grouped according to the study design as their natural similarities rendered well for comparing effects. Below is a summary of all included studies according to the study type.
RCTs
Both the RCTs compared the antipyretic drugs among themselves and with placebo for efficacy of fever reduction. They also reported the adverse events. Cunietti’s study included participants with simple self-limiting upper respiratory tract bacterial/viral infections.
Observational studies
Four of the five prospective observational studies focused on the cardiovascular effects along fever reduction from intravenous antipyretics. Furthermore, Krajcova’s study investigated the mechanism of hypotension by observing cardiac output related measurements. Cantais’ and Lee’s studies investigated mortality from antipyretics. Lee further sought to compare the effect of antipyretic therapy between sepsis and non-sepsis groups. They also incorporated fever temperature ranges and investigated for association of mortality and antipyretic therapy.
Both the included retrospective observational studies investigated mortality from antipyretics and external cooling, comparing those who developed sepsis with those who did not.
All the included observational studies were conducted on ICU patients.
Case series and reports
The case series reported Steven Johnson syndrome and toxic epidermal necrolysis from antipyretics during viral infections. The case reports reported on adverse events from antipyretics taken for common conditions such as flu and upper respiratory tract infections. The adverse evets reported were pneumonitis, hypotension, eosinophilia, fixed drug eruption, SJS/TEN and aseptic meningitis.
Risk of bias (ROB)
The two RCTs had unclear ROB, due to unavailability of the full text and original publication and all the other studies had high ROB due to their study designs and some due to confounding and missing data. (Table 1)
Table 1
Characteristics of included studies
Study ID
|
Country
|
Design
|
Setting
|
Population
|
Intervention
|
Comparator
|
Outcomes
|
Observation period
|
N
|
ROB
|
Cunietti
1993
|
Not mentioned
|
RCT
|
Not mentioned
|
≥ 65 Y
viral/ bacterial infections
URTI/LRTI
Fever > 38oC
informed consent
|
Nimesulide
|
Paracetamol
|
1. Fever reduction
2. Progress to complications, hemodynamic
3. Laboratory markers of morbidity
|
3 days
|
39
|
U
|
Reiner
1985
|
Not mentioned
|
RCT
|
Not mentioned
|
18–90 Y
with fever
|
Nimesulide
|
Diclofenac, Placebo
|
Fever reduction
|
6 hours
|
81
|
U
|
Cantais
2016
|
France
|
Obs - P
|
ICU
|
Adult patients requiring IV acetaminophen infusion according
to the attending physician’s judgment and having arterial
pressure monitored via an arterial catheter
|
Paracetamol
|
Nil
|
1. Fever reduction
2. Progress to complications, hemodynamic
3. Mortality
|
3 hours
|
160
|
H
|
Hersch 2008
|
Israel
|
Obs – P
|
ICU
|
critically ill
patients in the ICU who were febrile (body
temperature ≥ 38°C), ventilated, sedated, and
experiencing sepsis
|
Propacetamol
|
Nil
|
1. Fever reduction
2. Progress to complications, hemodynamic
|
2 hours
|
14
|
H
|
Krajcova 2012
|
Czech Republic
|
Obs - P
|
ICU
|
> 18Y, artificially ventilated and administered paracetamol, monitored by PiCCO, with sinus rhythm
|
Paracetamol
|
Ranitidine
|
1. Fever reduction
2. Progress to complications, hemodynamic
|
Till resolution of fever occurred
|
6
|
H
|
Lee 2012
|
Korea and Japan
|
Obs - P
|
ICU
|
All adult patients requiring ICU > 48 hours
|
NSAIDs, paracetamol
|
External cooling
|
1. Fever reduction
2. Mortality
|
28 days
|
1425
|
H
|
Poblete 1997
|
Switzerland
|
Obs - P
|
ICU
|
Patients undergoing mechanical ventilation in ICU with rectal temperature > 38.5C and in whom the attending physician wished to reduce fever (Patients did not have inspired O2 fraction of > 0.6; did not consume caloric intake exceeding energy expenditure calculated at baseline)
|
Propacetamol, metamizole
|
External cooling
|
1. Fever reduction
2. Progress to complications, hemodynamic
3. Mortality
|
Till metabolically stable state was acheived
|
20
|
H
|
Ye 2017
|
USA
|
Obs - R
|
ICU
|
All patients meeting criteria for sepsis, receiving mechanical ventilation; with fever and hypothermia
|
NSAIDs, paracetamol
|
External cooling
|
Mortality
|
Not specified
|
8711
|
H
|
Zhang 2015
|
USA
|
Obs - R
|
ICU
|
All patients meeting criteria for sepsis
|
NSAIDs, paracetamol
|
External cooling
|
Mortality
|
Not specified
|
15268
|
H
|
Ban 2016
|
Korea
|
CS
|
Allergy clinic
|
Patients presenting with fever and blistering lesions with a history of acetaminophen exposure preceding onset of symptoms
|
Paracetamol
|
Nil
|
1. AdEv
2. LOS
3. Laboratory marker of morbidity
|
Case 1: 67 days
Case 2: 36 days
|
2
|
H
|
Akashi 1997
|
Japan
|
CR
|
Hospital
|
64- and 70-year-old with Acetaminophen induced pneumonitis
|
Paracetamol
|
Nil
|
1. AdEv
2. Laboratory marker of morbidity
|
Case 1: 4 months
Case 2: 3 days
|
2
|
H
|
Ayonrinde 2000
|
Australia
|
CR
|
Hospital
|
65-year-old with anaphylactoid reaction
|
Paracetamol
|
Nil
|
1. AdEv
2. Laboratory marker of morbidity
|
3 days
|
1
|
H
|
Danguy 2010
|
France
|
CR
|
Hospital
|
71-year-old with Acetaminophen induced hypotension
|
Paracetamol
|
Nil
|
Progress to complications, hemodynamic
|
5 days
|
1
|
H
|
Gonzalo - Garijo 2006
|
Spain
|
CR
|
Hospital
|
68-year-old with Ibuprofen induced fever
|
Ibuprofen
|
Nil
|
1. AdEv
2. LOS
3. Laboratory marker of morbidity
|
3 months
|
1
|
H
|
Kim 2014
|
Korea
|
CR
|
Hospital
|
60-year-old with TEN
|
Paracetamol
|
Nil
|
1. AdEv
2. LOS
3. Laboratory marker of morbidity
|
30 days
|
1
|
H
|
Kondo 1993
|
Japan
|
CR
|
Hospital
|
63-year-old with Acetaminophen induced eosinophilia
|
Paracetamol
|
Nil
|
1. AdEv
2. Laboratory marker of morbidity
|
7 days
|
1
|
H
|
Prabhu 2005
|
Nepal
|
CR
|
Hospital
|
65-year-old with fixed drug eruption
|
Paracetamol
|
Nil
|
1. AdEv
2. Laboratory marker of morbidity
|
48 hours
|
1
|
H
|
(RCT: randomised controlled trials; Obs P: prospective observational; Obs R: retrospective observational; CS: case series; CR: Case reports; U: unclear; H: high; ICU: intensive care unit; NSAIDs: non-steroidal anti-inflammatory drugs; LOS: length of stay; AdEv: adverse events; TEN: toxic epidermal necrolysis) |
Narrative Synthesis
No study investigated the primary outcome, onset/worsening of chronic inflammatory disease from antipyretic use. We could find no study that investigated patient satisfaction from use of these drugs either.
Fever reduction:
Antipyretics were beneficial in reducing fever as seen in the included RCTs [Mean 1.6\(\pm\)0.42oC] (47, 48). This did not, however, appear to extend to the sepsis scenario or where the patients were critically ill from infections [fever reduction ranged from 0.21±1.15 to 0.69±0.4oC], as seen in observational studies (50–53). The effect of external cooling was similar [fever reduction ranged from 0.2±0.1 to 2±1.44oC] in all the studies investigating this therapy (50, 53). The case series did not contribute to this outcome and only one case report mentioned that “fever reduced” after paracetamol (60).
Table 2
Outcomes – Fever reduction
Study ID
|
Temperature reduction Intervention
|
Temperature reduction Comparator
|
Time points reported
|
Cunietti 1993
|
1.6 ± 0.42
|
1.6 ± 0.42
|
6, 7, 8, 9 and 10 am and at 2, 6 and 10 pm on the first day and at 6 and 10 am and at 2, 6 and 10 pm on the second and third days
|
Reiner 1985
|
Significant
|
|
Before, 30, 60, 90, 120, 240 and 360 minutes after drug administration
|
Cantais 2016
|
0.69 ± 0.4
p < 0.0001
|
Nil
|
Mean drop at 30 mins post infusion
|
Hersch 2008
|
0.3 ± 0.75
|
Nil
|
At 30 mins post infusion.
Measures reported at before and 15, 30, 45, 60, 90, 120 minutes after propacetamol administration;
|
Krajcova 2012
|
0.35 ± 0.13
p = 0.02
|
Nil
|
69th minute
|
Lee 2012
Antipyretic drugs and external cooling in sepsis patients
|
0.39 ± 0.14
|
0.2 ± 0.1
|
From application of antipyretic to next temperature monitoring
|
Lee 2012
Antipyretic drugs and external cooling in non-sepsis patients
|
0.38 ± 0.16
|
0.1 ± 0.05
|
From application of antipyretic to next temperature monitoring
|
Poblete 1997
Antipyretic drugs and external cooling
|
0.21 ± 1.15
|
2 ± 1.44
p < 0.0001
|
On achievement of metabolically stable state
|
Prabhu 2005
|
Fever reduced
|
Nil
|
Not mentioned
|
Temperature changes in oC |
Progress to haemodynamic complications:
Seven of the 17 studies reported development of hypotension, although the magnitude differed from study to study [SBP drop of 3 to 4% in one RCT (47); MAP drop ranging from 4.8 ± 12.28 to 12.59 ± 8.1 mmHg in prospective observational studies, (49, 51, 52) except in one study, that showed no change in MAP (50) and drop of varying degrees in BP in the case reports (58, 61)].
Heart rate was also measured in four studies and showed a trend towards reduction after the administration of antipyretics, although the degree of reduction did not reach statistical significance (47, 49, 50, 58). One study found a non-statistical significant trend towards reduction in peripheral resistance and cardiac output from antipyretic drugs (49).
Table 3
Outcomes - Progress to hemodynamic complications
Study ID
|
Mean blood pressure drop Intervention
|
Mean blood pressure drop Comparator
|
Time points reported
|
Cunietti 1993
|
0 events
|
0 events
|
6, 7, 8, 9 and 10 am and at 2, 6 and 10 pm on the first day and at 6 and 10 am and at 2, 6 and 10 pm on the second and third days
|
Cantais 2016
(≥ 15% MAP drop group)
|
11.5 ± 25.44
p < 0.0001
(83/160 events)
|
overall MAP drop: 5.96 ± 4.91
p < 0.0001
|
30 minutes after infusion of acetaminophen
|
Hersch 2008
|
13.48 ± 11
p < 0.05
(24/72 events)
|
Nil
|
At 30 mins; measures also reported at before and 15, 30, 45, 60,
90, and 120 minutes after propacetamol administration
|
Krajcova 2012
Febrile cycles
|
12.59 ± 8.1
|
Mean rise
3.28 ± 37.69
p = 0.001
|
19th minute and post intervention
|
Krajcova 2012 afebrile cycles
|
6.59 ± 9
|
Mean rise
1.36 ± 18.72
|
19th minute and post intervention
|
Poblete 1997
Antipyretic drugs and external cooling
|
0 ± 22.27
|
Mean rise
2 ± 25.5
|
End of intervention
|
Ayonrinde 2000
|
SBP: 85
DBP: 50
|
Nil
|
30 mins after ingestion of drug
|
Danguy 2010
|
SBP: 72
DBP: 40
|
Nil
|
Day 4 post intervention
|
blood pressure changes in mmHg |
Mortality: (Fig 2, 3 & 4)
There were no events of mortality in studies involving common self-limiting viral/bacterial disease, as reported in the RCTs and case reports. However, in sepsis and ICU patients (reported in observational studies), mortality was positively associated with antipyretic therapy in the highest fever temperature range [OR (95%CI): 2.61 (1.11, 6.11) and 2.05 (1.19, 3.55) for NSAIDs and acetaminophen respectively] (Table 3), when compared to external cooling [OR (95%CI): 1.2 (0.70, 2.05)] (55). Mortality was not affected by antipyretic drugs in other temperature ranges and did not differ from non-sepsis group (54, 55).
When all the studies comparing overall mortality between antipyretic drugs and external cooling in sepsis patients were considered, the OR and 95%CI was 0.75 (0.64, 0.87) (p = 0.0003)(26, 53, 55). Mortality in antipyretic drugs group compared to external cooling in the non-sepsis group, however, was not different and yielded an OR and 95%CI of 0.30 (0.07, 1.30) p = 0.11. Mortality in antipyretic drugs group compared to non-treatment in sepsis patients was higher but statistically non-significant OR and 95%CI: 1.11 (0.96, 1.28) p = 0.15
Table 4
Study ID
|
No of events
|
Mortality Intervention
OR and 95%CI
|
Time points reported
|
Cantais 2016
(For development of
≥ 15% MAP drop)
N = 160
|
42/83
|
1.6 (0.86, 3)
P = 0.14
|
In hospital and ICU
|
Lee 2012
Antipyretic drugs and external cooling in sepsis patients
|
53/147:64/307
|
2.14 (1.38, 3.43)
P = 0.0006
|
28 days from admission
|
Lee 2012
Antipyretic drugs and external cooling in non-sepsis patients
|
2/131:18/364
|
0.3 (0.68, 1.32)
P = 0.11
|
28 days from admission
|
Poblete 1997
|
1/8
|
Cannot calculate as not associated with antipyretic
|
In hospital/ICU
|
Ye 2017
Antipyretic drugs and external cooling (all sepsis patients)
|
121/652:206/892
|
0.22 (0.18, 0.27)
P < 0.0001
|
In ICU
|
Ye 2017
Antipyretic drugs and non-therapy (all sepsis patients)
|
121/652:1196/7167
|
1.14 (0.93, 1,5)
P = 0.22
|
In ICU
|
Zhang 2015
Antipyretic drugs and external cooling (all sepsis patients)
|
129/1027:212/1006
|
0.54 (0.42, 0.68)
P < 0.0001
|
In ICU
|
Zhang2015
Antipyretic drugs and non-therapy (all sepsis patients)
|
129/1027:1389/12208
|
0.54 (0.42, 0.68)
P < 0.0001
|
In ICU
|
Length of stay (ICU/hospital):
A mean 51.5 days resulted from the antipyretic treatment adverse effects in the case series (56). Case reports yielded a mean 4 days from the adverse effect of these drugs (57–63). The observational study that reported the length of stay did not assess the effect of treatment on it (53).
Table 5
Outcomes - Length of stay
Study ID
|
length of stay Intervention
|
Length of stay Comparator
|
Time point measured
|
Lee 2012
(No assessment of antipyretic drugs on LOS)
|
With sepsis: Mean: 8 days (5, 14)
Without sepsis: Mean: 5 days (4, 7)
|
Nil
|
At discharge
|
Ban 2016
|
Case 2: 67 days
Case 6: 36 days
|
Nil
|
At discharge
|
Gonzalo - Garijo 2006
|
2.5 days
|
Nil
|
At discharge
|
Kim 2014
|
17 days
|
Nil
|
At discharge
|
Morbidity indicators (laboratory and radiological findings) (supplementary material)
The details of laboratory markers influenced by antipyretic drugs was provided in the case reports of adverse events (56–63). Detailed laboratory readings from these studies are provided in the supplementary material
Adverse events
Adverse events were reported in the RCTs, the case series and all the case reports. Altogether, 18 episodes of adverse effects were reported in 10 studies (47, 48, 56–63), including mild transitory effects, hypotension, Steven Johnson syndrome/toxic epidermal necrolysis, aseptic meningitis, pneumonia, and fixed drug eruption. All the case reports confirmed causality of the adverse events.
Table 6
Outcomes - Adverse events
Study ID
|
Adverse events Intervention
|
Adverse events Comparator
|
Time point reported
|
Cunietti 1993
|
1/39 (skin rashes)
|
Nil
|
End of trial
|
Reiner 1985
(Both were antipyretics)
|
3/30
(Transient adverse effects)
|
4/24
(transient adverse effects)
|
End of trial
|
Ban 2016
|
SJS/TEN
|
Nil
|
7days after acetaminophen
|
Akashi 1997
|
Acetaminophen induced pneumonitis
|
Nil
|
18 and 5 days after ingestion of acetaminophen
|
Ayonrinde 2000
|
Anaphylactoid reaction
|
Nil
|
30 mins after ingestion of acetaminophen
|
Gonzalo - Garijo 2006
|
Aseptic meningitis (clinically suspected)
|
Nil
|
15 minutes after ibuprofen
|
Kim 2014
|
SJS/TEN
|
Nil
|
14 days after administration of paracetamol
|
Kondo 1993
|
Acetaminophen induced eosinophilia
|
Nil
|
5 days after ingestion of Kinuya-chinetsu
|
Prabhu 2005
|
Fixed drug eruption
|
Nil
|
After paracetamol
|
Certainty of evidence
The certainty of evidence from included studies turned out to be low for fever reduction and mortality and very low for other outcomes.
Table 7
GRADE Summary of Findings table
Certainty assessment
|
Impact
|
Certainty
|
Importance
|
№ of studies
|
Study design
|
Risk of bias
|
Inconsistency
|
Indirectness
|
Imprecision
|
Other considerations
|
Fever reduction (assessed with: temperature)
|
8
|
observational studies
|
serious a
|
serious b
|
not serious
|
not serious
|
publication bias strongly suspected
strong association c
|
Quality of evidence does not generate confidence in the effect
|
⨁⨁◯◯
LOW
|
CRITICAL
|
Mortality (assessed with: death or alive)
|
5
|
observational studies
|
serious a
|
serious b
|
not serious
|
not serious
|
none
|
Quality of evidence generates moderate level of confidence in the effect
|
⨁⨁◯◯
LOW
|
CRITICAL
|
Hemodynamic changes (assessed with: BP)
|
6
|
observational studies
|
serious a
|
not serious b
|
not serious
|
serious d
|
publication bias strongly suspected c
|
Quality of evidence does not generate confidence in the effect
|
⨁◯◯◯
VERY LOW
|
CRITICAL
|
Morbidity indicators (assessed with: laboratory values)
|
7
|
observational studies
|
serious a
|
not serious
|
not serious
|
serious d
|
publication bias strongly suspected c
|
Quality of evidence does not generate confidence in the effect
|
⨁◯◯◯
VERY LOW
|
IMPORTANT
|
Length of stay (assessed with: days)
|
8
|
observational studies
|
serious a
|
not serious
|
not serious
|
serious d
|
publication bias strongly suspected
all plausible residual confounding would reduce the demonstrated effect c
|
Quality of evidence does not generate confidence in the effect
|
⨁◯◯◯
VERY LOW
|
NOT IMPORTANT
|
Adverse events (assessed with: development of adverse events)
|
8
|
observational studies
|
not serious
|
serious b
|
not serious
|
serious d
|
publication bias strongly suspected c
|
Quality of evidence does not generate confidence in the effect
|
⨁◯◯◯
VERY LOW
|
IMPORTANT
|
(a. observational studies; wide variance of point estimates/minimal or no overlap of CIs/all studies show low p values; c. small studies with significant changes; d. small sample size) |