Postgrad Med J 88:21-27 doi:10.1136/postgradmedj-2011-130217
  • Original article

Systematic review and meta-analysis of the effects of antipyretic medications on mortality in Streptococcus pneumoniae infections

  1. Richard Beasley1,2
  1. 1Medical Research Institute of New Zealand, Wellington, New Zealand
  2. 2Departments of Respiratory Medicine, Internal Medicine and Intensive Care, Capital & Coast District Health Board, Wellington, New Zealand
  3. 3Department of Medicine, University of Otago Wellington, Wellington, New Zealand
  1. Correspondence to Dr Sarah Jefferies, Medical Research Institute of New Zealand, Private Bag 7902, Wellington 6242, New Zealand; sarah.jefferies{at}
  1. Contributors All contributions to the production of this manuscript were from those named as authors.

  • Received 7 June 2011
  • Accepted 30 October 2011
  • Published Online First 25 November 2011


Aim To determine whether the use of antipyretic medications in the treatment of Streptococcus pneumoniae infection affects mortality in humans or animal models.

Design A systematic search of Medline, Embase, and The Cochrane Register of Controlled Trials was undertaken to identify in vivo animal experiments or randomised, controlled trials in humans of antipyretic medication in S pneumoniae infection which reported mortality data. Meta-analysis was by inverse variance weighted method for odds ratios.

Setting Antipyretics are recommended for the symptomatic treatment of various diseases caused by S pneumoniae. However, there is evidence that fever is a protective physiological response to infection, that treating fever secondary to infection may be harmful, and that some strains of S pneumoniae are temperature sensitive.

Main outcome measures Mortality associated with antipyretic use in S pneumoniae infection.

Results Four studies from two publications met the inclusion criteria and investigated the use of aspirin in animal models. The pooled estimate of mortality was an OR with aspirin treatment of 1.97 (95% CI 1.22 to 3.19). There were no suitable human studies identified.

Conclusions A twofold increased risk of mortality was found with aspirin treatment in animal models of S pneumoniae infection. No relevant human studies were identified. It is difficult to generalise from animal models to clinical medicine, but based on these findings and the prevalence and severity of S pneumoniae infections worldwide, future study of the effects of antipyretic therapy in S pneumoniae infection in humans is recommended.


In the 1880s, the discovery of Streptococcus pneumoniae (Diplococcus pneumoniae, pneumococcus) and its role in the aetiology of pneumonia was followed by research investigating the effects of temperature on S pneumoniae infection.1 Early in vivo animal experiments crudely indicated that raised ambient temperatures might provide survival benefits2–5 and potentiate humoral immunity.4 In 1909, Strouse described the native resistance of pigeons to S pneumoniae type III due to their higher normal core temperature (approximately 41°C) and found they succumbed to infection when cooled with ice or the antipyretic pyramidon.6 Further experiments in S pneumoniae infected rabbits demonstrated higher mortality rates with external cooling.7 8 In contrast, others demonstrated increased mortality in mice exposed to raised environmental temperatures9 and reduced mortality rates in mice subjected to drug induced hypothermia.10 11 However, in rabbits infected with S pneumoniae, a febrile response was associated with improved outcomes.7 12 In human patients, in an early 20th century case series, fever therapy (external radiation) in severe S pneumoniae type 3 meningitis appeared to reduce cerebrospinal fluid bacterial cell counts, but did not prevent mortality.13 In vitro experiments confirmed that elevated temperatures caused impaired replication and death of many S pneumoniae strains12 14 and also found an associated increase in in vitro antibiotic activity.15

More recently, evidence suggesting a beneficial effect of fever and detrimental effect of antipyretics in various infectious diseases has been described. In humans, observational studies have shown a positive correlation between febrile temperature during bacteraemia and survival,16 17 and hypothermia as a manifestation of sepsis is a negative predictor of outcome.18 In animals the suppression of fever with antipyretic drug treatment has been shown to increase mortality in viral,19 bacterial,20 and parasitic infections21 and in humans, antipyretic drugs have been shown to increase the duration of chickenpox illness22 and malarial parasitaemia,23 and augment rhinovirus shedding24 25 as well as inhibit antibody responses.24 26 Furthermore, a recent randomised controlled trial demonstrated a trend towards increased mortality in critically ill patients assigned to the ‘aggressive’ treatment of fever with paracetamol.27

S pneumoniae remains a major cause of global morbidity and mortality through a variety of diseases.28 29 It is the most common cause of lower respiratory tract infection which is the third most common cause of mortality worldwide.30 In the developed world, the use of antipyretic drugs in the treatment of pneumonia is routine, and recommended in international guidelines.28 In view of the evidence for S pneumonia temperature sensitivity, we hypothesised that the use of antipyretic medications in S pneumoniae infection may be associated with an increased risk of mortality. The purpose of this systematic review is to identify studies in animal models and humans of the effect of antipyretic drugs on mortality in the treatment of S pneumoniae infection, and by meta-analysis to investigate whether antipyretic drug use causes an increased risk of mortality.


Search strategy

Three databases were used to identify studies investigating the effect of antipyretics on mortality in S pneumoniae infection: Medline (1950 to present); Embase (1947 to present); and the Cochrane Central Register of Controlled Trials (1991 to present): search date 7 December 2010. Keywords were: ‘Streptococcus pneumoniae’ or ‘pneumococcal infections’ or ‘Diplococcus pneumoniae’ or ‘pneumococi*’; and ‘antipyre*’ or ‘paracetamol’ or ‘acetaminophen’ or ‘non-steroidal anti-inflammatory’ or ‘salicyl*’ or ‘aspirin’ or ‘ibuprofen’ or ‘diclofenac’. Potentially relevant studies which were not written in English were translated. Two people (SJ and RB) examined each paper's title and abstract and the full paper if necessary. The reference lists of all relevant papers were also reviewed and additional hand-searching was carried out.

Inclusion criteria

Studies were required to be in vivo animal experiments where treatment allocation did not have to be specifically stated as randomised, or randomised controlled trials in humans, which reported mortality outcomes in the investigation of antipyretic medications in S pneumoniae infection.

Exclusion criteria

  • All study arms manipulate temperature (ie, there is lack of a placebo or control group to enable comparison of normal thermal response to infection)

  • Antipyresis by non-pharmacological means

  • Antipyresis by steroid drug, due to potentially confounding immune effects

  • Potentially lethal dosing of antipyretics in the treatment group

  • Inactive bacterial products were used, for example, pneumococcal vaccines.

Data extraction

Data extraction was by reported counts or proportions of research subjects dying. Where data was provided as percentages, numbers were calculated and rounded to the nearest whole number.

Statistical method

The outcome variable was mortality. The categorical variables were pooled using the inverse variance weighting method for odds ratios.31 Homogeneity statistics and the I2 statistic were calculated for each analysis.32 Publication bias was examined through a funnel plot and formal tests of publication bias.


Figure 1 shows the results of the search strategy which identified four animal studies and no human randomised controlled trials. Review of 416 papers initially identified one human randomised controlled trial and 18 articles describing in vivo animal studies relating to antipyresis in S pneumoniae infection. Of these, two papers met the inclusion criteria,33 34 and 17 were excluded for the following reasons: one human trial and one animal study used non-living S pneumoniae products; seven animal studies induced hypothermia in the treatment group; one animal study exposed all study arms to temperature manipulation; and seven animal studies did not investigate or report mortality outcomes. The characteristics of the studies included in the meta-analysis are summarised in table 1 and further described below.

Figure 1

Flow diagram of search strategy.

Table 1

Characteristics of included studies

The two papers which were suitable for inclusion comprised four studies using animal models, three in mice and one in rabbits, and infection with type III S pneumoniae.33 34 All four studies used aspirin (acetylsalicylic acid, ASA) as their active treatments. Three studies used a placebo treatment, and one compared an antibiotic and aspirin with antibiotic alone but did not have a control substance. Three hundred and twenty-two animals were studied: 159 receiving antipyretic medication and 163 controls. Table 2 summarises the individual study results and pooled estimate together with the tests for heterogeneity. The pooled estimate of the risk of mortality with aspirin was an OR of 1.97 (95% CI 1.22 to 3.19) and the I2 statistic was 0 (95% CI 0 to 83.7). Figure 2 shows the forest plot for these estimates. A funnel plot and formal tests showed no evidence of publication bias (details not shown).

Table 2

Risk of mortality with aspirin

Figure 2

Forest plot of the relationship of aspirin treatment to mortality. This shows the individual trial estimates as well as pooled estimates with the size of the boxes on the forest plot inversely proportional to the size of the variance of the study estimates, so that more precise studies have larger boxes.

Included studies

Esposito et al33 carried out a series of studies on CD-1 adult mice to investigate the effect of aspirin on mortality and pulmonary antibacterial responses following the intratracheal infiltration of S pneumoniae type III. Three placebo controlled studies involving a total of 120 mice were suitable for inclusion (table 1). These experiments differed in the temporal commencement of the aspirin dosing regimen (subcutaneous 400 mg/kg/24 h) relative to S pneumoniae infection: (1) pretreatment—72 h before infection; (2) immediate treatment—10 min after infection; or (3) delayed treatment—6 h after infection. Survival was recorded at 96 h post-infection. The safety of the aspirin dosing regimen was pre-examined by a 2 week pilot in normal CD-1 mice which demonstrated no associated toxic effects. A higher mortality rate was demonstrated in the pretreatment and immediate treatment aspirin groups versus their corresponding controls (table 2). Additional experiments were also described whereby pairs of aspirin treated and control mice were sacrificed at different points post-infection for the evaluation of pulmonary bacterial and inflammatory responses and the exclusion of superinfection. Pulmonary bacterial clearance was found to be significantly impaired at 48 h post-infection in all aspirin-treated groups relative to controls.

Klastersky's paper,34 translated from French, primarily investigated the effects of corticosteroids on bacteraemia in rabbits, rodents, and humans. However, in one study it describes the effect of 250 mg aspirin administered enterally, in conjunction with antibiotic therapy, following intraperitoneal S pneumoniae type III infection (table 1). The control group received antibiotic treatment alone. Mortality rates were recorded at 24 h post-infection, with no significant difference found between groups. The author stated that there was no statistical difference between the ‘febrile index’ of the two groups, although specific temperature data was not provided. Of interest, Klastersky also described improved survival, but prolonged bacteraemia, in a group of S pneumoniae infected, antibiotic treated rabbits that were shaved of fur (for cooling effect) versus controls that remained unshaven. In one further study, when shaved rabbits were exposed to elevated ambient temperatures, mortality rates then equalled those of the unshaven controls.

Excluded studies of interest

Eight papers describing the effect of antipyretic medications in animal models of S pneumoniae infection were excluded because the control arm was also subjected to temperature manipulation,35 or they did not report mortality rates (table 3).36–42 In experimental models of meningitis39 and peritonitis40–42 antipyretic treatment was associated with significantly higher bacterial counts versus controls. In the study by Sande et al,35 all rabbits were anaesthetised using the antipyretic urethane, and for 12 h post-infection kept in a room temperature environment or warmed to maintain a core body temperature >40°C. The warmed group demonstrated lower S pneumoniae bacterial counts compared with those treated with antipyretic alone (p<0.001), until returned to a room temperature environment where this difference was no longer seen. In mice treated with sodium salicylate, Spanuolo et al40–42 noted a reduction in the median lethal inoculum (LD50)—bacterial dose required to kill 50%—as well as numbers of granulocytes found in peritoneal exudate compared with placebo treated controls. However, in contrast, two experimental models of S pneumoniae acute otitis media found fewer inflammatory changes37 and lower bacterial cell counts36 in the middle ears of chinchillas and gerbals treated with a combination of ibuprofen and antibiotic versus antibiotic therapy alone. One further experimental animal model of S pneumoniae acute otitis media investigated combinations of paracetamol and/or antibiotics and reported no significant difference in bacterial cell counts between the groups.38

Table 3

Summary of excluded studies of interest


This systematic review and meta-analysis demonstrated a twofold increased risk of mortality with aspirin treatment in animal models of S pneumoniae infection. There were no identified randomised controlled trials of antipyretic therapy in human S pneumoniae infection. In view of the prevalence and severity of S pneumoniae disease,28 29 and difficulties in extrapolating animal data to humans, we propose that it should be a priority to investigate whether the current practice of routine antipyretic therapy in S pneumoniae infection is associated with worse outcomes.

There are numerous mechanisms by which antipyretic medications could potentially influence outcomes in S pneumoniae infection. At febrile temperatures there are potential benefits to the host by the direct inhibition of heat sensitive S pneumoniae strains. Whereas S pneumoniae shows temperature-responsive adaptive genetic changes in the range of 21–37°C,43 many strains exhibit impaired replication and death at higher temperatures.14 Temperatures within the physiological febrile range also increase in vitro antibiotic activity against S pneumonia.15 Conversely, non-steroidal anti-inflammatory drugs have also been found to increase antibiotic activity against S pneumoniae in vitro, possibly by displacement of protein bound antibiotic44 and enhanced leucocyte bacterial uptake and killing.45 Other evidence suggests that antipyretics have the potential to impair immune responses which are promoted by fever.46 In a randomised controlled trial, Pyrmula et al26 found that paracetamol (acetaminophen) impaired humoral immune responses to vaccination in infants, although in post hoc analysis this effect was shown to be independent of a febrile response. Following primary vaccination, the paracetamol treated group had significantly lower antibody titres to antigens, including all 10 pneumococcal vaccine serotypes. This difference persisted after booster vaccination in all but one of the pneumococcal serotypes. Lower opsonophagocytic activity titres were also found in the paracetamol group.

There are a number of methodological issues relevant to the interpretation of the findings of this meta-analysis. The first is whether all relevant studies were identified. We are confident that our sensitive search strategy of three major databases, and the reference lists of relevant papers, identified all eligible studies published since 1947, including those not written in English. Furthermore, there was no evidence of publication bias in formal tests or funnel plots, although with the small number of included studies, these approaches to detecting publication bias may not be very sensitive.

Second, the outcome of this meta-analysis may be limited by the heterogeneity and quality of the studies included. Although the homogeneity statistic was not statistically significant, the I2 statistic was large with a wide confidence interval. Both papers did not use randomisation, and Klastersky's paper was not placebo controlled. Although all experiments used the same antipyretic drug, aspirin, different routes and regimens were used, and there was no description of safety profiling by Klastersky. Klastersky's paper did not clearly specify the doses of pneumococcus and penicillin used in the included experiment. However, the administration of antibiotic treatment in both groups did create a model which is better representative of clinical practice.

All four experiments had homogeneity in using the same serotype of pneumococcus, type III. This corresponds to serotype 3 of the current S pneumoniae classification systems47 which today is among the top 20 serotypes identified in humans with invasive pneumococcal disease globally.48 49 However, important differences in virulence can exist between the strains within serotypes, and with more than 90 serotypes now identified, these results cannot be generalised to all S pneumoniae infections.

Another important issue in this meta-analysis is that all identified studies were in animals, and no human trials were found. Generalisation of our results is therefore further limited by the physiological differences between rabbits, mice and man. Most notably, unlike most mammalian species, mice generally do not generate febrile responses to S pneumoniae infection9 and relative hypothermia may in fact be protective.10 11 Furthermore, while the primary outcome variable in all experiments was mortality (or survival) it was the anti-inflammatory effect of aspirin, not antipyretic actions, which was the purpose of the experiments by Esposito et al.33 Indeed, temperature was not reported by Esposito et al,33 and in the study by Klastersky, temperature was found not to be significantly altered.34 Thus the degree to which the increased aspirin associated mortality found in our meta-analysis is a result of antipyretic effects is uncertain. The limitations of generalising from experimental animal models to human clinical disease cannot be understated.

The results of this meta-analysis are consistent with a recent systematic review and meta-analysis of the effect of antipyretic medication on mortality in influenza infection.50 In animal models of influenza infection, antipyretic treatment was associated with an increased risk of mortality, with a fixed effects pooled OR of 1.34 (95% CI 1.04 to 1.73). This effect was seen with the use of different antipyretics including aspirin, paracetamol, and diclofenac, demonstrating a class effect of antipyretic medications.

Complex interactions exist between host microbial susceptibility and immune response to infection, fever, antipyretics, and antibiotics, and clinical outcomes may vary with specific circumstances. In view of the prevalence and severity of S pneumoniae infections,28 29 increasing pneumococcal antibiotic resistance,51 and the twofold increased mortality associated with aspirin therapy demonstrated in animal studies of S pneumoniae infection, we recommend that this issue requires definitive investigation by way of randomised placebo controlled trials of antipyretic drugs in specific S pneumoniae diseases in humans.

In conclusion, this systematic review and meta-analysis identified two papers describing four animal studies in which a twofold increased risk of mortality was associated with aspirin treatment in experimental S pneumoniae infection. No human studies were identified. Due to the limited generalisability of findings from animal models to clinical medicine and the prevalence and severity of S pneumoniae infection worldwide, future study of this relationship in humans is recommended as a priority.

Main messages

  • Streptococcus pneumoniae is the most common cause of lower respiratory tract infection, which is the third most common cause of mortality worldwide.

  • Antipyretic medication is recommended and commonly used in S pneumoniae infections, including pneumonia.

  • Many S pneumoniae strains are temperature sensitive and there is evidence that fever may be a protective response to infection.

  • There are currently no human studies investigating the effects of antipyretic medication on clinical outcomes, including mortality, in S pneumoniae infections.

  • A twofold increased risk of mortality has been found with aspirin treatment in animal models of S pneumoniae infection.

Current research questions

  • Is fever a beneficial response to infection?

  • Do antipyretic medications impair immune responses?

  • Is the administration of antipyretic medications in S pneumoniae infection a safe practice?

Key references

  • Rich AR, McKee CM. The mechanism of a hitherto unexplained form of native immunity to the type III pneumococcus. Bull Johns Hopkins Hosp 1936;59:171–207.

  • Esposito AL. Aspirin impairs antibacterial mechanisms in experimental pneumococcal pneumonia. Am Rev Respir Dis 1984;130:857–62.

  • Klastersky J. Etude experimentale et clinique des effets favorables et defavorables de la fievre et de l'administraion de corticoides au cours d'infections bacteriennes. Acta Clinica Belgica 1971;26:1–84.

  • Roberts NJ Jr. Impact of temperature elevation on immunologic defenses. Rev Infect Dis 1991;13:462–72.

  • Eyers S, Weatherall M, Shirtcliffe P, et al. The effect of mortality on antipyretics in the treatment of influenza infection: systematic review and meta-analysis. J R Soc Med 2010;103:403–11.


We thank Dr Maak Weinert (Wellington, New Zealand) and Dr Alain Marcuse (Wellington, New Zealand) for translating papers from German.


  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Responses to this article

Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of PMJ.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.