Abstract
Fever is a common clinical complaint in adults and children with a variety of infectious illnesses, as well as a frequently reported adverse event following immunization. Although the level of measured temperature indicative of a “fever” was first defined in 1868, it remains unclear what role fever has as a physiologic reaction to invading substances, how best to measure body temperature and compare measurements from different body sites, and, consequently, how to interpret fever data derived from vaccine safety trials or immunization safety surveillance. However, even with many aspects of the societal, medical, economic, and epidemiologic meanings of fever as an adverse event following immunization (AEFI) still elusive, it is a generally benign—albeit common—clinical sign. By standardizing the definition and means of assessment of fever in vaccine safety studies, thereby permitting comparability of data, we hope to arrive at an improved understanding of its importance as an AEFI.
Fever is an increase in body temperature to more than the normal range, usually resulting from a complex physiologic reaction to the presence of substances that induce the production of endogenous cytokines [1]. These cytokines, in turn, alter the activity of hypothalamic neurons, raising the hypothalamic set point [1]. Fever is most often caused by viral, bacterial, or rickettsial infections [2], unlike hyperthermia, which typically results from insufficient heat dissipation in the face of excessive heat production (e.g., strenuous exercise), excessive environmental heat exposure, heat shock, or defective hypothalamic thermoregulation [3, 4]. In 1868, Carl Wunderlich determined that “normal body temperature” is actually a range of values rather than a specific temperature, and 38°C (100.4°F) was established as the upper limit of normal [5, 6]. Although this definition of “normal temperature” has since been questioned [1], it is clear that elevated body temperature due solely to endogenous causes (e.g., fever and pyrexia) rarely exceeds 41°C (105.8°F). Hyperthermia with an increase in core body temperature to >40°C (>104.0°F) can result in cellular damage and multiple-organ dysfunction (heat stroke), which may include encephalopathy, circulatory failure, acute respiratory distress syndrome, intestinal ischemia and endotoxemia, and disseminated intravascular coagulation [3, 7]. Such excessive temperature elevations are generally the result of excessive heat exposure coupled with thermoregulatory failure, and they rarely, if ever, occur as a result of fever alone.
How Is Fever Determined?
The “physiologic gold standard” of true core body temperature can only be obtained during surgery or under experimental conditions permitting measurement of the temperature of the blood bathing the thermoregulatory centers in the hypothalamus [8, 9]. Temperatures recorded by sensors placed in the pulmonary artery or within the esophagus have been described as good comparators [8, 10, 11]. Although rectal temperature is viewed by many as the closest practical correlate of core temperature, measurement at this site involves several problems, including a slow response to changes in core temperature ("rectal lag"), inappropriate placement of the thermometer, inadequate measurement time, real or imagined discomfort, concerns over safety, and lack of cultural acceptance. None of the external sites for obtaining body temperature (i.e., oral, rectal, axillary, umbilical, tympanic, temporal artery, or cutaneous sites) have been shown to be consistently superior, nor has it yet been possible to reliably predict temperature differences among these anatomic sites. For example, a recent review showed that differences between infrared ear thermometry and rectal temperature measurements are substantial in both directions [12]. Although rectal temperature tends to be higher than oral temperature, and oral temperature tends to be higher than axillary temperature, these relationships are inconstant and often misleading [10, 13–15].
Misconceptions About Fever
Despite the recognition that fever is generally a physiologic reaction to an underlying disease process and not a disease in itself, the presence of an elevated temperature may cause undue concern and is often tainted by misconceptions on the part of parents, patients, and healthcare providers [2, 16–20]. These misconceptions can give rise to an unwarranted fear of severe side effects, such as permanent brain damage, leading to inappropriate or excessive temperature measurement, as well as unnecessary clinic visits, laboratory testing, and antipyretic and antimicrobial therapy. Concern over a febrile seizure may precipitate hospital admission and performance of lumbar punctures or other costly tests attempting to define the cause of the fever. When seen in this context, it is not surprising that “fever phobia,” with or without accompanying signs or symptoms of illnesses (such as upper respiratory tract infection or ear infection), is one of the most common reasons that parents seek medical attention for their children [21–23]. Thus, in addition to monitoring for potential physical consequences, a multitude of often unfounded psychological and societal concerns need to be considered when assessing the significance of fever as an adverse event following immunization (AEFI).
Despite these fears, only a small proportion of febrile infants and children with elevated temperature have a serious bacterial infection [24, 25]. Conversely, invasive life-threatening and other infections, such as acute otitis media, can occur in the absence of fever (e.g., up to 45% of children with acute otitis media do not present with fever) or even with hypothermia [26–31]. Thus, the presence of fever is only one of many clinical observations that can be used to assess the nature and severity of an illness; it must always be taken in context with a thorough clinical evaluation integrating historical data, physical findings, behavior, age (special caution needs to be given to different approaches for different age groups) [26], and responsiveness [2, 26, 29, 32]. When there is suspicion of significant infection, serial clinical and laboratory examinations, as well as repeated inquiries about medical history, may be necessary to achieve a diagnosis and to determine an appropriate course of management. For example, Baraff et al. [26] report that the probability of serious bacterial infection in febrile (measured rectal temperature, ⩾38°C) infants ⩽90 days of age ranges from 1.4% (95% CI, 0.4%–2.7%) to 17.3% (95% CI, 8.0%–30.0%). Therefore, it is important that, when fever is being considered as an AEFI, it be evaluated not only in the context of its temporal association with immunization, but also in conjunction with other historical and clinical observations that may identify a coincident and unrelated cause, necessitating according management.
Aspects of Fever as an Aefi
Frequency of fever as an AEFI. The passive surveillance system for AEFI in the United States, known as the Vaccine Adverse Event Reporting System, or VAERS, has found fever to be the most frequently reported “serious” and “nonserious” AEFI for all age groups, with the exception of persons aged 18–64 years, in whom a wide range of local reactions are more frequently reported as nonserious events (e.g. swelling, pruritis, pain and/or redness in any combination coded as “hypersensitivity at the injection site”). (Code of Federal Regulations 600.80 criteria define “serious adverse experience”: as any adverse experience occurring at any dose that results in any of the following outcomes: death, a life-threatening adverse experience, inpatient hospitalization or prolongation of existing hospitalization, a persistent or significant disability/incapacity, or a congenital anomaly/birth defect. Important medical events that do not result in death, are not life-threatening, and do not require hospitalization may be considered serious adverse experiences when, on the basis of appropriate medical judgment, it has been determined that they may jeopardize the patient or subject and may require medical or surgical intervention to prevent one of the outcomes listed in this definition.) There are, however, fundamental problems with the assessment of these data. First, reported fever may only be one sign among a multitude of signs and symptoms which, if given equal and appropriate significance, would often lead to an assignment of causality to some factor other than vaccination. For example, fever in conjunction with a sore throat and cough after immunization would far more likely be caused by a viral common cold than by a vaccine. Second, lacking denominator data and a control group, it is sometimes impossible to infer the rate or relative risk of fever after immunization, compared with a nonvaccinated population. For example, in a Canadian surveillance system, the baseline population incidence of adverse events was calculated through the use of a computerized registry linking hospital and immunization records and data from all physician billing claims. The study showed that the incidence of hospitalization for pyrexia and convulsions increased significantly within 2 and 7 days, respectively, after receipt of diphtheria-tetanus—whole-cell pertussis (DTwP) immunization. However, the paucity of observations on unimmunized infants prevented calculation of any population-attributable risks, despite the fact that pyrexia and nonepileptic convulsions were the most frequently designated reasons for hospitalization in the first year of life, irrespective of immunization status [33].
Similar difficulties occur in trying to calculate vaccine-attributable risks. Clinical trials of routine childhood vaccines have shown that fever typically occurs in 1%–10% of vaccinees [34], but it can be as frequent as 30% to >70% among vaccinees receiving multiple vaccines or DTwP vaccine [34, 35]. The temperature of subjects in these studies was seldom >39°C (>102.2°F), and fever was typically self-limited. The studies, however, did not provide any valid comparison with the rate of fever among unvaccinated children of similar age by which to determine the “background rate” of elevated temperature in these populations. Calculation of vaccine-attributable risk was, therefore, impossible.
Economic implications. Whether caused by intercurrent infection or immunization, fever after childhood immunization has important economic implications, with a 2-fold increase in utilization of medical services and a 3-fold increase in work loss for at least 1 parent [35]. Not surprisingly, parents reported a willingness to pay a median of US$50 (range, US$0 to >US$9999; mean, US$261) to avoid fever after immunization.
Febrile seizures. Although fever itself is usually transient and without long-lasting effects, an elevated risk of febrile seizure following some immunizations has been found. Administration of acetaminophen at the time of primary immunization has been shown to significantly reduce or prevent the appearance of fever with an inactivated component vaccine (e.g., DTwP-polio) and has found wide acceptance [36, 37]. Ibuprofen and acetaminophen have equally been recommended for administration at the time of diphtheria-tetanus—acelluar pertussis (DTaP) immunization and every 4 h for 24 h thereafter for children with a history of febrile seizures to reduce the possibility of postvaccination fever [38].
In addition, prophylactic antipyretics, such as acetaminophen (with aspirin avoided, particularly after receipt of live viral vaccines) after immunization, objective temperature measurement, and use of a less reactogenic vaccine may also have value in reducing the likelihood of a febrile seizure, which occurs most frequently during the first day of fever [39]. Although such episodes are a cause of great concern to parents and give impetus to the concept of a relationship between immunization and “brain damage,” it needs to be kept in mind that (1) febrile seizures are common in childhood, affecting 3%–5% of children <5 years of age, (2) children who experience febrile seizures after immunization do not appear to be at higher risk for subsequent seizures or neurodevelopmental disability [40–43], and (3) although they are frequently reported AEFIs, febrile seizures are, nevertheless, rare events with an attributable risk for DTwP and measles-mumps-rubella (MMR) vaccines of 6–9 and 25–35 cases per 100,000 immunized children, respectively [40]. In comparison, measles disease itself results in 1 in 1000 infected children developing encephalitis, and 1 in 50 and 1 in 250 children with pertussis disease experience convulsions and encephalopathy, respectively.
Therefore, despite the occurrence of this transient complication from fever after immunization, it needs to be emphasized that vaccination against measles and pertussis [44–47], as well as vaccination against invasive pneumococcus [48] and Haemophilus influenzae type b disease [49–51], has significantly reduced not only the overall incidence of neurologic disorders and fever-causing bacteremia associated with the diseases themselves, but it has also reduced that of the associated serious and often permanent neurodevelopmental disabilities due to these conditions [52].
Inference of causality. In general, although increased temperature is a frequently reported AEFI, it is primarily concurrent infectious processes rather than immunizations that represent the most common causes of such fever [53]. Because children experience an average of 2–6 acute febrile episodes in the first 2 years of life, with medical care sought for two-thirds of cases [2, 22, 23, 26], any meaningful assessment of fever as an AEFI should include measurement of the background rate. Most randomized, controlled vaccine trials compare the rate of adverse events among study vaccine recipients with that seen among a similar population given a licensed vaccine rather than a placebo. To our knowledge, only 1 vaccine safety study compared fever in vaccinated and unvaccinated children using the “gold standard” of clinical trials—that is, a randomized, double-blind, placebo-controlled trial of MMR vaccine in 581 twin pairs [54, 55]. In this trial, the authors reported that, although temperatures of ⩾38.5°C (⩾103.1°F) within 3 weeks of immunization were seen significantly more often in 14–18-month-old vaccinated children than in placebo recipients (25% vs. 6%; OR, 3.28; 95% CI, 1.23–4.82; P < .001), it is notable that fever occurred in as many as 6% of the placebo recipients in that age group. Among children 6 years of age, 3% of both vaccine and placebo recipients developed fever, and 88% of fevers with a temperature of ⩾38.5°C at peak period of occurrence were unrelated to vaccination. Had there been no placebo groups under observation, any elevations in temperature might easily have been attributed to the vaccine rather than to coincident infectious conditions. A double-blind, placebo-controlled efficacy trial of live, attenuated varicella vaccine and of inactivated hepatitis A vaccine each also failed to find any significant difference in the incidence of fever (defined as an oral temperature of ⩾38.0°C or ⩾38.3°C, respectively) between the vaccinated and the placebo groups [56, 57]. Finally, a placebo-controlled efficacy trial of intranasal influenza vaccine in children found a significant 21% reduction in the number of febrile illnesses overall and a 30% reduction in the incidence of febrile otitis media in vaccinated children during a 6–9 month follow-up period [58]. A significant increase in the rate of fever in vaccinees, compared with nonvaccinees, was present only on day 2 after immunization, and the increase was of low grade (mean temperature, 38.2°C [100.7°F]) and of short duration (mean duration, 1.4 days).
There are, therefore, a multitude of reasons precluding definitive conclusions regarding the frequency and significance of fever as an AEFI, including limited knowledge of the variable background rates of fever in various age groups and a lack of sufficient comparisons of fever rates in vaccinated and unvaccinated populations in a controlled setting.
Improving Current Understanding of Fever as an Aefi
Defining the true importance of fever as an AEFI has been hindered by the limited comparability of data derived from different vaccine safety surveillance systems and by the use in different vaccine safety trials of different temperature limits, diverse sites for measuring temperature, and a wide variety of devices for identifying fever. The lack of a standardized definition is not unique to the evaluation of fever; it extends to virtually all other AEFIs as well. For example, no standardized definitions exist for well-established postimmunization events, such as “persistent crying,” “anaphylactic reaction,” or even common injection site reactions. To overcome this difficulty, an international task force was recently assembled, consisting of professionals from patient care, public health, and scientific, pharmaceutical, regulatory, and professional organizations in both developed and developing countries [59]. This group—the Brighton Collaboration, named after the town in Great Britain where it was conceptualized in 1999—has as its goal the development of globally accepted standardized case definitions and guidelines for data collection, analysis, and presentation of AEFIs.
Towards that end, a working group was formed within the Brighton Collaboration in 2001 to standardize the definition of fever as an AEFI. In a broad survey of the literature on vaccine studies and fever, the working group found widely varying lower limits for fever (temperature, ⩾37.1°C to ⩾38.5°C), 9 different sites for temperature measurement (i.e., rectum, oral sites, axillary sites, tympanic membrane, umbilical sites, inguinal sites, great-toe, forehead, and abdominal skin), and the use of 4 different devices for measurement (i.e., mercury-in-glass, electronic, infrared, and thermophototropic liquid crystal thermometers). Any meaningful comparison of fever rates using such disparate criteria and methodology—even rates attributable to the same vaccine evaluated at different times at different study sites—would be difficult at best and, in most cases, impossible.
Clearly, standardization of the definition and method of diagnosis and analysis of fever in surveillance systems and vaccine trials would substantially improve our understanding of this common event. To this end, the working group developed a definition of fever as an AEFI as a temperature elevation of ⩾38°C (100.4°F), measured at any site, utilizing any validated device (selected guidelines of temperature measurement to facilitate comparability of data are show in table 1). Although it was recognized that this value is, to some extent, arbitrary, it is based on a conservative interpretation of definitions proposed and used over the years by clinicians, investigators, and the public at large. Because it would be practically and scientifically inappropriate to impose a global standard for devices and route of measurement, the working group also developed a set of guidelines for the standardized collection, analysis, and presentation of data on fever as an AEFI to enable the highest degree of comparability. Despite the wide leniency in choice of measurement site and device, establishing consistency within and between trials was deemed essential for comparability of data. The working group document was made available on the Internet for public comment for 6 months and appears to have garnered wide acceptance. It is hoped that this definition document, together with others currently under evaluation by the Brighton Collaboration, will facilitate standardized recording and presentation of AEFI data acceptable for global use.
Selected guidelines for temperature measurement for standardized understanding of fever data in immunization safety.
Selected guidelines for temperature measurement for standardized understanding of fever data in immunization safety.
Conclusions
In conclusion, although many aspects of the societal, medical, economic, and epidemiologic meaning of fever as an AEFI are still elusive, it is a common, generally benign, clinical sign. We consider the cost of a potentially unnecessary clinical/diagnostic evaluation of a child with fever after immunization to be clearly off-set by the risk of complications from disease that would result had we no vaccination programs. In addition, a globally standardized assessment and reporting of the event, as proposed by the Brighton Collaboration, is a step toward a more scientifically rigorous understanding of its incidence and true significance.
Acknowledgments
Financial support. The review was conducted as part of the Brighton Collaboration, which receives its funding from the Centers for Disease Control and Prevention, the World Health Organization, and the European Research Grant for Improved Vaccine Safety Surveillance.
Conflict of interest. M.B. is a full time employee of Wyeth. All other authors: No conflict.
Comments