Individual symptoms and individual symptoms and signs

The evidence from prospective cohort studies demonstrates a number of individual symptoms (i.e. drowsiness, decreased activity, poor feeding, pale, reduced urine output, bile-stained vomiting) and signs (i.e. being drowsy, moderate/severe chest recession, respiratory rate > 60 breaths/minute, nasal flaring, grunting, crackles, lump > 2 cm, being pale, not looking well, bulging fontanelle) that are associated with serious illness in infants and young children. Most of the evidence is limited to data relating to infants less than 6 months in a secondary care setting. In isolation, none of these symptoms or signs are reliably associated with serious illness.

Scoring systems of combinations of symptoms and signs

Scoring more than 10 using the YOS scoring system after a history and examination may help identify other infants and children at high risk of serious illness.

A YOS of 6 with a well-appearing child makes the presence of a serious illness very unlikely. However, the development of features of serious illness including the symptoms listed on the YOS should prompt further evaluation.

In isolation, none of these symptoms are strongly associated with serious illness. A child identified as ‘ill’ when assessed by an experienced healthcare professional is likely to have an SBI. To ensure that children with serious illness are recognized early, many children without serious illness will need to be examined.

Health economics

The GDG did not identify any issues where cost-effectiveness issues were a priority for this clinical question.

Individual symptoms and individual symptoms and signs

Prospective cohort studies of children with fever have identified a number of symptoms and signs that are predictive of serious illness. Much of the most reliable data relates to infants up to the age of 6 months. The GDG decided that it was reasonable based on clinical experience to extrapolate the symptoms and signs to older children and use them as part of the assessment of older children with a feverish illness. The GDG is aware that there is currently a large prospective study being conducted in Australia on the predictive values of symptoms and signs in febrile children of all ages. In the UK, a project is in development on the recognition of acute illness in children (Dr R MacFaul, personal communication). It is hoped that the results of these studies will inform future guidance on the assessment of the risk of serious illness in children with feverish illness.

Scoring systems of combinations of symptoms and signs

The features used in the YOS associated with serious illness are validated and show good correlation with those children who go on to develop serious illness in children aged 3 months to 3 years. The GDG felt that these features can be extrapolated for use on children up to the age of 5 years, based on clinical experience and extrapolated to the UK population.

‘Traffic light’ system

The GDG attempted to summarise the results of risk stratification from the prospective cohort studies and scoring studies in a ‘traffic light’ system. From the scoring studies, those symptoms and signs that scored only 1 on the YOS were designated ‘green’. Those individual symptoms and signs that scored 5 in the YOS were designated ‘red’, as a child with only one ‘red’ symptom and all other ‘green’ symptoms (i.e. scoring 10 in the YOS) was at significant risk of serious illness. Those symptoms and signs that scored 3 in the YOS were designated ‘amber’, because while a child with a combination of ‘amber’ symptoms or signs was at significant risk of serious illness, a child with only one ‘amber’ feature was not at significant risk of serious illness.

From the prospective cohort studies, the GDG assigned ‘red’, ‘amber’ or ‘green’ status to additional symptoms and signs based on their associated risk of serious illness and on clinical experience.

Narrative summary

No evidence was found that provided ‘normal values’ for heart rate in the population of children under 5 years old. There is one EL 2+ study¹¹¹ that compared heart rate in children under 1 year with their body temperature. This study found that for every 1 °C rise in body temperature, the resting heart rate rose by 9.6 beats/minute (Figure 4.1). The GDG is aware that there is an ongoing UK study to determine normal values for resting heart rate in children with fever aged 3 months to 12 years.

Figure 4.1

Heart rate rise with rising temperature in children less than 1 year old; adapted with permission from Hanna and Greenes.

There are unvalidated tables of normal resting heart-rate values in young infants and children without fever which are widely taught (Figure 4.2).

Figure 4.2

Widely quoted values for paediatric heart rates at various ages (left diagonals; right diagonals) and the heart rates of children with minor blunt trauma at various ages (vertical lines).

Evidence summary

There is a lack of evidence regarding heart rate as a marker of serious illness. Despite this, the GDG felt that heart rate is a potentially important marker of serious illness. The Delphi panel was used to decide whether heart rate should be part of the routine assessment of a child with a fever, because a raised heart rate can be a sign of serious illness, particularly septic shock.

Delphi statement

‘Healthcare professionals examining children with fever must measure and record heart rate as part of their routine assessment.’

Seventy-five percent of the Delphi panel agreed with this statement in round 1 (consensus achieved).

‘Healthcare professionals should refer a child for specialist paediatric (children’s) care if the resting heart rate is above the expected range for a feverish child.’

This statement did not reach consensus despite adaptations made to the original statement after round 1.

GDG translation

Heart rate was not placed in the ‘traffic light’ system (see below) as the Delphi panel did not agree that heart rate per se should be used as a basis for referral to specialist care. The statement ‘healthcare professionals examining children with fever must measure and record heart rate as part of their routine assessment’ was adapted and combined with the statement about the physiological parameters that should be documented as part of the assessment (see the end of Section 4.5.2.4). The GDG felt it important to make healthcare professionals aware of the significance of a raised heart rate particularly in septic shock (see the recommendations at the end of Section 4.5.2.4).

The GDG felt that basic physiological parameters in children should be backed up by a better weight of evidence. The GDG is aware that one research project on the predictive value of heart rate and other vital signs in children with fever is currently in progress in the UK (Drs R MacFaul and M Thompson, personal communications) but it is likely that larger studies will be needed to produce definitive results. The GDG therefore recommends that studies are performed to confirm normal ranges for heart rate at various body temperatures and to determine whether children with heart rates outside these ranges are at higher risk of serious illness.

Narrative summary

Five studies were found investigating the prognostic value of the capillary refill time (CRT) with three EL 2+ prospective studies^113–115 and one EL 2− retrospective study¹¹⁶ which included children in ICU post-resuscitation, which was excluded owing to the lack of relevance. In addition, there is one EL 2+ SR¹¹⁷ for signs and symptoms of dehydration which included CRT. Overall, the studies were conducted in a range of settings varying from primary care to intensive care in the UK,¹¹³ the USA¹¹⁴ and Kenya¹¹⁵ with different baselines which made meta-analysing inappropriate.

The SR¹¹⁷ showed that prolonged CRT had sensitivity of 0.60 (95% CI 0.29 to 0.91) and specificity of 0.85 (95% CI 0.72 to 0.98) of detecting 5% dehydration, which made CRT the most specific sign of dehydration. The results from prospective cohort studies showed that there was no significant association of CRT of 3 seconds with meningococcal disease, other significant bacterial illness or white blood cell count (WBC) (statistics not provided).¹¹³ In one prospective cohort study, the receiver operating characteristic (ROC) curve showed that the best performance was obtained when a CRT of 3 seconds was taken to be ‘prolonged’; furthermore, a prolonged CRT (> 3 seconds) was associated with a more urgent triage category, the administration of fluid bolus and the length of hospital stay (all P < 0.05).¹¹³ Moreover, children with dehydration had prolonged CRT of 2 seconds, with a sensitivity of only 44% for predicting a fluid deficiency of < 5% or more of body weight (other statistics not provided).¹¹⁴ Overall agreement for CRT was moderate (k = 0.42), and was better for normal values (≤ 1 second) (k = 0.48) and clearly abnormal values (≥ 4 seconds) (k = 0.49).¹¹⁵

Furthermore, in a search of the specific signs and symptoms of meningococcal disease, CRT was found to be indicative (the OR of CRT > 3 seconds of having meningococcal disease is 29.4 (95% CI 9.4 to 92.6)¹¹⁸ in children with a petechial rash. In another SR¹¹⁷ that included four trials investigating the usefulness of prolonged CRT to indicated dehydration, the findings showed that the pooled sensitivity of prolonged CRT (defined differently in different studies) was 0.60 (95% CI 0.29 to 0.91), with a specificity of 0.85 (95% CI 0.72 to 0.98), for detecting 5% dehydration.

Evidence summary

The authors used different cut-offs of CRT and it appeared that CRT of 2 seconds was a weak predictor of dehydration and serious illness while a CRT ≥ 3 seconds is associated with dehydration and significant illness (e.g. meningococcal disease) in children.

GDG translation

The GDG noted that CRT is quick to carry out and exhibits moderate reproducibility. A statement about measuring CRT was combined with the statement about the physiological parameters which should be documented as part of the assessment (see the end of Section 4.5.2.4). The GDG considered that a CRT of ≥ 3 seconds was an ‘amber’ sign (see the recommendations at the end of Section 4.5.2.4).

Evidence summary

Blood pressure was not identified as an independent risk factor for serious illness in any of the prospective cohort studies and scoring systems. Low blood pressure was identified as one of several risk factors for adverse outcome in children with meningococcal disease.¹¹⁹

GDG translation

The GDG agreed with stakeholder comments that blood pressure should be measured in children with fever who are displaying features of possible serious illness. Blood pressure can be a helpful measurement to monitor children with possible sepsis although low blood pressure is a late feature of septic shock. Other markers such as raised heart rate and prolonged capillary refill time are present earlier and require no special equipment to measure. The GDG concluded that blood pressure should be measured when facilities exist to monitor blood pressure and other markers of inadequate organ perfusion (i.e. shock) are detected (see the recommendations at the end of Section 4.5.2.4).

Evidence summary

Refer to Sections 4.5.1 (General symptoms and signs of serious illness), 4.5.4 (Assessment of dehydration) and 4.6.6 (Pneumonia) for evidence relating to respiratory rate.

GDG translation

An abnormal respiratory rate has been shown to be a non-specific marker of serious illness, a specific feature of pneumonia and required for the assessment of dehydration. The GDG felt that respiratory rate is therefore an important physiological parameter which needs to be assessed by healthcare professionals. A statement about measuring respiratory rate was combined with the statement about the physiological parameters which should be documented as part of the assessment (see below).

Clinical question

Can the height of body temperature in a young child with fever be used to predict the risk of serious illness or mortality?

Narrative evidence

The literature search was restricted to prospective cohort studies because this would yield the highest quality evidence (EL 2). Twelve prospective cohort studies,^{93,95,98,99,120–127} of which three were EL 2–,^124,125,127 were found that reported on the relationship between height of fever and the outcome in terms of serious illness. The studies varied widely in terms of setting (e.g. hospital emergency department or paediatric assessment units in different countries such as Australia,⁹³ the UK¹²¹ or the USA, and Puerto Rico¹²⁰), ages of children included (e.g. < 28 days¹²⁷ to 3–36 months¹²⁸), definition of fever (e.g. rectal temperature ≥ 38 °C or rectal temperature ≥ 39 °C) and outcomes measured. There was also wide variation in the methods of analysis. For these reasons it was not possible or appropriate to pool the data.

Several large EL 2+ studies reported a higher relative risk of SBI with increasing body temperature, with body temperatures ≥ 39 °C in particular being associated with a higher risk. Other studies did not report this association. The sensitivity of a high body temperature to detect SBI is low. With one exception, the sensitivity of a temperature ≥ 39 °C to detect SBI was between 10% and 32%. In developed countries the sensitivity of a temperature ≥ 39 °C to detect SBI was between 10% and 14%. The PPV of a temperature ≥ 39 °C varied between 4% and 40% in developed countries.

Evidence summary

Twelve prospective cohort studies (nine EL 2+ and three EL 2−) that reported on the relationship between height of fever and the outcome in terms of serious illness were found.

Several large EL 2+ studies reported a higher relative risk of SBI with increasing body temperature, with body temperatures ≥ 39 °C in particular being associated with a higher risk. Other EL 2+ studies did not report this association.

Health economics

The GDG did not identify any issues that required a cost-effectiveness analysis for this clinical question.

GDG translation

The GDG noted that most large EL 2+ studies suggest that the risk of serious illness increases with height of fever in young children. Body temperatures ≥ 39 °C in particular were usually associated with a higher relative risk of SBI. The strongest associations were reported in studies involving children aged less than 6 months. However, the sensitivity and PPV of temperature ≥ 39 °C were low, which suggests that most cases of serious illness would be missed if height of body temperature was used in isolation to identify children with serious illness. Furthermore, the GDG noted that other features of a child with feverish illness, such as his or her age or an ‘ill appearance’ were often more predictive.

The GDG concluded that healthcare professionals should be aware that there is an association between height of body temperature and risk of SBI. However, this association is not sufficiently robust to recommend immediate action or referral based on body temperature alone. An exception was made for children aged under 6 months with body temperature ≥ 39 °C because the evidence was strongest for this age group.

In addition, the GDG noted that children aged under 3 months with fever are generally at a higher risk of serious illness (see Section 7.3). The incidence of serious illness in this group, for instance, is over ten times higher than that in older children. The clinical studies that provide the evidence for this age group used a body temperature ≥ 38 °C as the definition of fever. The GDG therefore decided that children aged under 3 months with a body temperature ≥ 38 °C should also be included in the recommendation about risk of serious illness.

Clinical question

Can the duration of fever in a febrile young child be used to predict the risk of serious illness or mortality?

Narrative evidence

Three EL 2+ prospective studies^126,129,130 that looked at the duration of fever as a risk factor for SBIs in general were found. One of them¹²⁹ reported that a duration of fever > 48 hours had an odds ratio of 3.85 (95% CI 1.11 to 13.3) for predicting serious illness. This relationship just reached statistical significance as an independent predictor of SBI. Another prospective cohort study¹²⁶ reported that duration of fever was longer in infants with SBIs (26.5 ± 41.5 hours) than those without (18.6 ± 21.7 hours) (P < 0.01). Furthermore, in comparison with < 24 hours, duration of fever > 48 hours had an odds ratio of 1.04 (95% CI 0.35 to 3.12) of having SBIs.¹³⁰ Of the other two EL 2 studies, one reported that children with SBI had statistically significant longer duration of fever while the other did not.

Two EL 2+ prospective studies^122,123 were also found that looked at the incidence of (predominantly occult) bacteraemia in relation to duration of fever in children with temperature ≥ 39 °C. Both studies reported a higher relative risk of bacteraemia with a shorter duration of fever (RR 1.5¹²² to 4.6¹²³). The PPVs of a short duration of fever were 4% and 10%.^122,123

Evidence summary

It was noted that there was a weak association between duration of fever and risk of serious illness from the three studies that looked at SBI in general. There was also an apparently converse association between duration of fever and risk of one particular SBI, namely bacteraemia.

Health economics

The GDG did not identify any issues that required a cost-effectiveness analysis for this clinical question.

GDG translation

The GDG noted a weak association between duration of fever and risk of serious illness from the five studies that looked at SBI in general. They also noted an apparently converse association between duration of fever and risk of one particular SBI, namely bacteraemia. The GDG concluded that the evidence was equivocal and relatively weak in both directions. They concluded that, on the basis of existing evidence, duration of fever could not usefully be included in the list of features that may be used to help predict serious illness.

The GDG was aware that longer durations of fever than those reported in the studies above may be associated with certain serious illnesses. In particular, the GDG noted that a fever lasting 5 days or more is one of the diagnostic criteria for Kawasaki disease. For this reason, it was decided to include a fever lasting 5 days or more as one of the ‘amber’ features in the traffic light system. A recommendation on the diagnosis of Kawasaki Disease is included in Section 4.6.9.