Fever is frequently observed in secondary care, occurring in up to 36% of patients in general medical wards, and up to 50% in critical care (Niven et al, 2012; Kiekkas et al, 2013). Fever is a symptom, not an illness in itself, and it is defined as the ‘controlled’ rise in body temperature, triggered by infectious or non-infectious mechanisms (Suzuki et al, 2015).
It is a complex, physiological, adaptive response to infection, which gives hosts survival advantages during illness (Carey, 2010). Fever inhibits microbial reproduction and viral replication, as well as accelerating the rate of phagocytosis (Dai et al, 2015). This is achieved by regulated phagocytic responses to pathogenic presence and the stimulation of cytokines, which react with arachidonic acid, altering the hypothalamus' thermoregulation system (Drewry et al, 2017).
Fever also involves an increase in metabolic rate and oxygen consumption (Young et al, 2012). The increased physiological demands during pyresis are identified as potentially detrimental to health (Niven et al, 2013). It is argued that any potential benefit of fever is undermined by the additional physiological demands on the body (Niven et al, 2013; Young and Saxena, 2014). Thus, an antipyretic drug may be administered to reduce fever-related complications (Niven et al, 2012). Paracetamol is reportedly the most common antipyretic used in current practice, although many forms of non-steroidal anti-inflammatory drugs (NSAIDS) are also used, including ibuprofen (Cannon, 2013; Suzuki et al, 2015). It is reported within the literature that fever is detrimental to patients with non-infectious, neurological pathologies, for example, after stroke, thus antipyretics may be administered to prevent worsening outcomes (Young et al, 2012; Zhang, 2015).
By contrast, there is no clear recommendation or guidance regarding antipyretic drugs and febrile patients with infection. The use of antipyretics during infection remains a controversial topic. Indeed, many authors have argued that suppressing fever interferes with the body's natural defence mechanisms, and may worsen patient outcomes (Carey, 2010; Young et al, 2012; Lee et al, 2012). In support, several authors have reported that infection, in the absence of fever, is highly correlated with poor prognosis (Lee et al, 2012; Young et al, 2012; Kiekkas et al, 2013). There is currently one guideline available, specific to children, from the National Institute for Health and Care Excellence (NICE) (2017), which clearly states that antipyretics must not be administered with the sole aim of suppressing fever, unless the child is distressed. Despite the publication of this guideline, several authors have suggested that health professionals hold an ongoing belief that fever is detrimental and requires suppression (Carey, 2010; Kiekkas et al, 2014). Subsequently, aggressive, routine fever-suppression methods have been frequently demonstrated in practice (Carey, 2010).
The aim of this literature review is to explore, analyse and appraise current research regarding antipyretic drugs in adult patients with fever and infection in secondary care, including intensive care units (ICU).
Methods
A literature search was undertaken in July 2017 using both the Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Medline databases (2010-2017). Keywords were searched, using truncation, including: fever, pyre*, infect*, and antipyr*. The Boolean logical operator ‘OR’ was used, in addition to Boolean operator ‘AND’ between ‘fever’ and ‘infection’ because both search terms needed to be present for a study to be considered for inclusion.
A critical, narrative approach was used to review current literature on the topic. Current literature was initially considered based on publication within the previous 5 years (2012-2017); however, due to a lack of literature eligible for inclusion, the relative definition of ‘current’ includes all relevant publications between 2010 and 2017. The primary author reviewed all search citations and abstracts, and each of the studies deemed eligible for inclusion at full text review were examined for quality and relevance using the tool developed by the Critical Appraisal Skills Programme (CASP) (2018) in Oxford.
The target population for this review was hospitalised adult patients with fever and infection. Studies included in the review had to meet one or both of the defined inclusion criteria:
Included studies varied in the outcomes measured, such as patient mortality/morbidity, patient experiences and perceptions of fever/antipyretics, and professionals' attitudes towards fever/antipyretics.
Findings
Outcomes of the literature search
The process of selecting articles for inclusion is outlined in Figure 1. Collectively, the database searches identified 1523 research papers. Each study was considered for inclusion based on the title and abstract, whereby 1501 were excluded, and 22 articles were selected for full text review. Reasons for exclusion at title and abstract review included: duplicates, the main focus was not infection, fever or antipyretics, not relevant, paediatric-specific or neurologically focused. After full-text evaluation, 13 articles were finally included in the review. Reasons for exclusion at this stage included: not relevant, inaccessible, and one study that failed to state the reason for paracetamol administration (analgesic versus antipyretic).
Overview of the included studies
The methodologies of the studies within this review include: randomised controlled trials (RCT) (n=3; one pilot, two comprehensive), cross-sectional survey/questionnaires (n=2), qualitative interviews (n=1), prospective observational studies (n=2), and retrospective observational studies (n=5). Table 1 details the characteristics of each included study. Each of the 13 included studies were examined against CASP quality checklists, and the overall methodological quality of the studies was deemed satisfactory. Despite being considered satisfactory for inclusion, the quality of these studies could have been improved further by increased blinding, data collection, and adopting alternative methods to convenience sampling. Two key themes were identified within the included studies: ‘antipyretics, fever and patient outcomes’ and ‘professionals’ and patients’ experiences and perceptions of antipyretics and fever’.
| Author (year) | Study type | Setting/country | Population | Sample size | Investigation | Results |
|---|---|---|---|---|---|---|
| Ames et al (2013) | Qualitative Interviews | Oncology/transplant ward, USA | Febrile inpatients | 28 | Patient experience of fever |
|
| Brick et al (2017) | Cross sectional Electronic questionnaire | Paediatric ICUs and transport teams, UK and Ireland | Doctors and nurses | 462 | Attitudes of nurses and medical staff towards fever and antipyretics |
|
| Dai et al (2015) | Prospective Observational | Tertiary teaching hospital, Taiwan | Febrile inpatients with hospital-acquired bacteraemia | 502 | Association between antipyretics and mortality |
|
| Janz et al (2015) | RCT | ICU, tertiary hospital, USA | Febrile inpatients with severe sepsis | 40 | Association between antipyretics and oxidative stress |
|
| Kiekkas et al (2014) | Cross sectional Prospective survey | Wards and ICUs, 9 hospitals, Greece | Adult nurses and healthcare assistants | 458 | Attitudes towards fever and antipyretics |
|
| Lee et al (2012) | Multicentred Perspective Observational | 25 ICUs, Japan and Korea | Febrile inpatients with and without sepsis | 1425 | Association between antipyretics and mortality |
|
| Mohr et al (2011) | Single centre Retrospective | Academic medical centre, USA | Febrile adult inpatients with Gram-negative sepsis | 241 | Demographic and clinical factors associated with antipyretics |
|
| Mohr et al (2012) | Retrospective | Academic medical centre, USA | Febrile adult inpatients with Gram-negative sepsis | 278 | Impact of antipyretics on 28-day in-hospital mortality |
|
| Morris et al (2010) | RCT | Wards and ICUs, USA, Thailand, Australia | Febrile, critically ill and non-critically ill adult patients | 30 | Association between IV ibuprofen and adverse events |
|
| Niven et al (2013) | RCT (pilot) | 2 ICUs, Canada | Febrile inpatients with and without infection | 26 | Association between antipyretics and mortality |
|
| Suzuki et al (2015) | Multicentred Retrospective Observational | 4 ICUs, Australia | Febrile inpatients with and without infection | 15 818 | Association between paracetamol and mortality |
|
| Weinkove et al (2015) | Retrospective Observational | 157 ICUs, Australia and New Zealand | Febrile inpatients with neutropenic and non-neutropenic sepsis | 118 067 | Association between peak temperature in first 24 hours and mortality |
|
| Young et al (2012) | Retrospective Observational | ICU, Australia and New Zealand | Febrile inpatients with and without infection | 636 051 | Association between peak temperature in first 24 hours and mortality |
|
ICU = intensive care unit; IV = intravenous; NSAIDS = non-steroidal anti-inflammatory drugs; RCT = randomised controlled trial
Antipyretics, fever and patient outcomes
One study found early antipyretic exposure (paracetamol and/or ibuprofen) in Gram-negative sepsis/septic shock reduced mortality risk (22% treatment group versus 35% control, P=0.01) (Mohr et al, 2012). Another study based on paracetamol (in both afebrile and febrile patients) reported significant decreases in mortalities within the treatment group (paracetamol: 996 deaths, versus placebo: 1168 deaths; P<0.0001), although, in medical patients with infection, paracetamol was not associated with increased survival but instead prolonged time to death (Suzuki et al, 2015).
An RCT demonstrated that paracetamol in septic ICU patients may reduce oxidative injury and improve renal function; however, mortality rates were not significantly different between treatment and control groups (paracetamol 5.6% versus control 18.2%, P=0.355) (Janz et al, 2015). Similarly, an RCT by Niven et al (2013) reported no differences in mortalities within the aggressively treated group (higher doses of paracetamol and physical cooling) compared with the permissive treatment group (21% aggressive versus 17% permissive, P=1.0). Additionally, Morris et al's (2010) RCT specific to ibuprofen demonstrated no differences in mortality risk in those receiving ibuprofen versus placebo.
Contrasting results were demonstrated by a prospective study by Lee et al (2012), which found that NSAIDS increased mortality in septic patients (P=0.028) and non-significantly decreased mortality in non-septic patients (P=0.15). Additionally, paracetamol was correlated with increased mortalities in septic patients (P=0.01), and the opposite for non-septic patients, although this was not found to be statistically significant (P=0.63). In non-septic patients peak temperatures above 39.5°C were associated with increased mortality (P=0.01). However, in septic patients, peak temperatures between 37.5°C and 38.4°C were associated with reduced mortality (P=0.014), and mortality risk remained reduced at peak temperatures above 39.5°C.
Perhaps also supporting fever's protective role during infection, Young et al (2012) reported that the lowest mortality risk was found in infectious patients with peak temperatures above 39.0°C. However, in non-infectious patients (including neurologically injured patients), peak temperatures of 39°C and above were correlated with higher mortality rates. Similarly, Weinkove et al (2015) reported that patients with the lowest peak temperatures were more at risk of imminent mortality (P<0.0001). Peak temperatures between 37.5°C and 39.4°C, and above 39.4°C, were associated with reduced mortality in non-neutropenic septic patients, and non-significantly increased in neutropenic septic patients.
One study found that patients with hospital-acquired bacterial infections exhibited higher mortality rates if their maximum temperature was lower than 37.9°C (Dai et al, 2015). Temperatures of 39°-39.9°C were associated with fewer mortalities, and no mortalities were observed in temperatures equal to or higher than 40°C. Each 1-degree rise in body temperature was correlated with a 28% decrease in mortality risk (P=0.023). Additionally, patients were at higher risk of mortality if they suffered infection, but were afebrile (Dai et al, 2015). The latter finding is consistent with two other included studies (Lee et al, 2012; Young et al, 2012).
Professionals' and patients' experiences and perceptions of antipyretics and fever
A cross-sectional descriptive study found that lower fever knowledge demonstrated by health professionals was positively correlated independently with antipyretic administration, and negative attitudes towards fever (P<0.05) (Kiekkas et al, 2014). Additionally, longer professional experience was also positively correlated independently with antipyretic drug administration (P=0.002). Brick et al (2017) investigated both doctors' and nurses' attitudes towards fever. The mean temperature that professionals reported acceptable for initiating antipyretic treatments was 38°-38.5°C. The reported temperature threshold was higher among doctors than nurses (38.5°C versus 38.0°C, P<0.001) and also higher in senior staff compared with junior staff (38.5°C versus 38.0°C, P<0.001). They also reported that only 29.5% of participants felt it was likely or very likely that the unit would comply with NICE guidance on fever management. Another study by Mohr et al (2011) highlighted inconsistencies in practice, including a general preference among nurses to administer paracetamol over ibuprofen, and that only 29% of patients in the study received antipyretic medication for initial temperature spikes of over 40°C.
Ames et al (2013) found in their qualitative interviews that patients most commonly reported feeling cold, ‘shivery’, warm, weak and ‘sweaty’ during infection, which were all generally perceived as ‘symptoms of fever’.
Discussion
The routine practice of administering antipyretic drugs remains controversial. Two studies demonstrated trends towards improved patient outcomes following antipyretic administration (Mohr et al, 2012; Suzuki et al, 2015) and in contrast four studies demonstrated trends towards increased mortality risk or demonstrated fever's benefits during infection (Lee et al, 2012; Young et al, 2012; Dai et al, 2015; Weinkove et al, 2015).
A common finding within the included studies is that the absence of fever during infection was significantly correlated with increased mortality (Lee et al, 2012; Young et al, 2012). It is argued that this may be due to blunted immunological responses and reduced cytokine production (Cannon, 2013; Dai et al, 2015). Another potentially confounding factor is that antibiotic efficacy is reportedly increased in the presence of fever (Cannon, 2013).
Within the literature there is ongoing controversy regarding the severity of illness that warrants antipyretic administration. Presumably, patients suffering severe sepsis would benefit from a reduction in metabolic and cardiac workload through the suppression of fever. Considering this, NSAIDS may, theoretically, benefit septic patients given their anti-inflammatory properties; however, the existing evidence does not currently support this theory. One of the studies demonstrated increased mortality risk following NSAID administration in septic patients (Lee et al, 2012) and, while another study demonstrated no adverse effects, ibuprofen was not associated with reduced mortality risk (Morris, 2010).
Perhaps the protective role of heat-shock proteins explains why fever remains beneficial even during severe illness. In the presence of fever, heat-shock proteins prevent thermal damage to cells by inhibiting proinflammatory-signalling pathways (Kiekkas et al, 2013; Young and Saxena, 2014). Therefore, by administering antipyretics, the function of heat-shock proteins may be affected, and their ability to protect, as part of the immunological defence process, is hindered.
Another important consideration for nursing practice are patients with comorbidities, who may be less likely to cope with increased physiological demands during fever (Carey, 2010; Launey et al, 2011). This could also be said for the elderly, who are typically vulnerable during severe infection due to blunted febrile responses (Hammond and Boyle, 2011). This is an important issue and one that has implications for practice, because identifying infection in the elderly may be more difficult without the presence of fever as a diagnostic aid (Cannon, 2013).
Similarly, ‘clinically unstable’ patients, typically in ICU, generally have limited cardiopulmonary reserves, and are often unable to compensate during times of increased metabolic demands. Thus, antipyretic drugs may reduce the risk of haemodynamic instability, as well as hypoxic tissue injury (Kiekkas et al, 2013). However, contrasting results were reported in the included studies, and subsequently this remains an ongoing controversy. Additionally, three studies reported that antipyretics failed to consistently achieve normothermia (Morris 2010; Mohr et al, 2012; Niven et al, 2013). If antipyretics fail to lead to fever abating, the aforementioned benefits are likely to be minimal or non-existent. Finally, another important result is Suzuki's (2015) finding that paracetamol prolonged time to death, rather than preventing it. One possible explanation for this is that illness duration is extended by fever suppression, which presents both ethical and economic considerations for practice, associated with prolonged hospital admissions.
Risks associated with antipyretic medications have been highlighted within the literature, including side effects and patient safety issues. If antipyretics frequently fail to lower body temperature, patients may be unnecessarily exposed to medication and potential side effects.
An important finding within Niven et al's (2013) RCT were the mild troponin rises experienced within the aggressively treated group. The reason for this finding remains unknown, however the relationship between aggressive paracetamol administration and cardiotoxicity is a factor worth consideration (Ralapanawa et al, 2016).
In addition to physiological risks associated with antipyretic administration, clinical implications were highlighted. Several authors suggested that antipyretic administration is associated with delayed diagnosis and initiation of treatment (Launey et al, 2011; Niven et al, 2012; Mohr et al, 2012; Dai et al, 2015). Fever has been described as a key diagnostic sign in clinical practice, indicating the presence of pathology (Mohr et al, 2012). Therefore, masking fever may lead to incorrect assumptions that a patient is recovering (Williams and Bellamy 2008). Furthermore, delayed initiation of antimicrobial therapy is associated with increased mortality (Niven et al, 2012).
The decision to administer an antipyretic is influenced by several factors, including beliefs, attitudes and knowledge. Despite the wide recognition of fever's benefits within the literature, the same cannot be said for clinical practice. Current practice favours aggressive antipyretic administration, which has been described as a ‘ritualistic’ and ‘persistent’ nursing intervention (Young et al, 2012; Lee et al, 2012). Perpetuating this cycle of routinely suppressing fever is the lack of clear guidance available to professionals working with adults.
The phenomenon termed ‘fever-phobia’ is widely used to describe significant misconceptions and negative attitudes towards fever, demonstrated by both health professionals and patients. Highlighting the extent of fever-phobia is the observation that nurses would waken a sleeping patient to administer an antipyretic (Demir and Sekreter, 2012), and that fever is treated more promptly than pain (Dvorkin et al, 2014). Edwards et al (2007) suggested that the heavy focus on reducing fever in children is based on the misconception that antipyretics prevent febrile convulsions. However, several studies including RCTs have confirmed that antipyretics do not prevent initial or further febrile convulsions (Leung and Robson, 2007; Strengell et al, 2009; Sullivan and Farrar, 2011). One explanation for the current practices demonstrated by nurses is the perceived need to ‘do something’ and ‘cure’ patients. Carey (2010) pointed out that pressure on nurses to ‘intervene’ in relation to fever may also stem from patients' and relatives' expectations. Websites aimed at the general public, such as NHS Inform in Scotland, while providing reassurance it is not always necessary to seek help for a fever, still point to the use of antipyretics ‘to help the uncomfortable feelings associated with a fever’ (NHS Inform, 2019).
The literature suggests that professionals' awareness and understanding of fever's benefits are lacking, which is concerning, given that current research cannot support routine antipyretic administration. An important point raised by Ames et al's (2013) study is that patients experience a range of symptoms based on their underlying illness. Therefore, rather than focusing solely on suppressing fever, practitioners should aim to relieve symptoms of the underlying disease.
Implications for practice
In the absence of clear, adult-specific guidance, antipyretic therapies remain a grey area, which raises both safety and practical concerns. Antipyretics should be considered based on each individual patient, their underlying illness and comorbidities. This approach promotes selective use of antipyretic drugs based on evidence, rather than as part of routine practice. In light of this, Carey (2010) advocated a protocol-based approach to assist decision-making in practice. Based on Carey's protocol, a theoretical protocol is shown in Figure 2, which incorporates the most recent available evidence. The protocol promotes selective use of antipyretics; however, it also accounts for physiologically vulnerable patients, who may benefit from fever suppression.
Strengths and limitations
This literature review has important strengths arising from rigorous selection criteria. The review is also based on diverse study types, and includes recently published RCTs and large observational studies. However, the overall statistical power of this review is limited, based on the heterogeneity of the included studies, including variable methods of temperature measurement and antipyretic drug selected, as well as the dose, frequency, duration and route. This review is also subject to limitations including the general lack of published literature on the topic, in particular, patients' experiences and perceptions of fever and antipyretics. For the purposes of obtaining an adequate sample size for review, some inconsistencies are consequently present within the included studies.
Conclusion
Fever is a beneficial, adaptive response to infection. However, it also involves an increase in metabolic demand and oxygen consumption, which may be deleterious for physiologically vulnerable patients. The literature suggests that health professionals remain ‘fever-phobic’ and continue to administer antipyretics inappropriately. If patients exhibit comorbidities, including coronary disease or significant respiratory disease, and are typically vulnerable at febrile temperatures, antipyretics may be beneficial. In view of this, the decision to administer antipyretic drugs must be based on clear, evidence-based rationales, and based on patients' individual needs, as part of a holistic assessment.
Based on the available evidence, routine, aggressive antipyretic administration cannot be advocated in secondary care. A large RCT is urgently required to close the gap between research and practice, and to respond to ongoing clinical uncertainty towards antipyretics. Considering that antipyretics influence patient outcomes, further research would facilitate evidence-based decision-making in practice, which will ultimately improve patient outcomes and patient safety.