Activity
Not responding normally to social cues or no response to social cues (included in 2007 traffic light table)
‘Activity’ was included in the 2007 traffic light as part of the YOS.
The 2013 review found evidence to support assessing activity level at presentation. The reported evidence showed that children with decreased social interaction were not more likely to have a urinary tract infection than children with normal social interaction (not a strong positive likelihood ratio). Children without a urinary tract infection often showed decreased social interaction (low specificity). Children with a urinary tract infection did not usually show decreased social interaction (low sensitivity). The evidence was of low quality.
The GDG acknowledged that it would be helpful to define ‘social cues’ for parents, caregivers or less experienced healthcare professionals. The glossary of the guideline has been updated to outline that this can include the parents’ perception of a baby behaving differently, response to their name, smiling and/or giggling.
The GDG chose to keep decreased activity in the ‘amber’ column as the evidence did not support movement into the ‘red’ category based on definitions used in the study. If the decreased activity is severe, healthcare professionals may use their clinical judgement of ‘appears ill to a healthcare professional’ to manage the child appropriately. Therefore, no change was made to the traffic light table.
Appears ill to a healthcare professional (included in 2007 traffic light table) and parents/carers
‘Appears ill to a healthcare professional’ was included in the 2007 traffic light table as part of the YOS.
The 2013 review supported the results of the 2007 review. The results of the studies were mixed, with some studies showing that children who appeared unwell were not more likely to have a serious illness than those who appeared well (not a strong positive likelihood ratio), and other studies showing that children who appeared unwell were more likely to have a serious illness (convincing positive likelihood ratio).
The GDG members acknowledged that being ‘very ill’ was more predictive than ‘appears ill’. However, they were aware that it is hard to distinguish between the two terms. The majority of studies reporting this sign did not define ‘appears ill’, and those that did used a combination of symptoms and signs that are presented elsewhere in the traffic light table. As there was no separate data available on parent/carer reports of ‘appears ill’, the GDG decided that that no recommendation could be made specifically on parent/carer reports of ‘appears ill’. However, the GDG highlighted that parent/carer reports of fever and other specific symptoms were covered by other recommendations in the guideline.
Based on their expert opinion, the GDG members noted that ‘appears ill to a healthcare professional’ can be subjective and difficult to define. Therefore, the GDG concluded that for this sign, the definition of ‘healthcare professional’ should be restricted to those who are trained in assessing children, for example GPs, specialist nurses and paediatricians. The GDG members concluded, based on their clinical experience, that there was not a strain on resources for children who are inappropriately referred because of this sign.
Given the mixed quality of the evidence, the GDG did not change ‘appears ill to a healthcare professional’ in the traffic light table.
Wakes only with prolonged stimulation or does not wake, or if roused, does not stay awake (included in 2007 traffic light table)
‘Wakes only with prolonged stimulation or does not wake, or if roused, does not stay awake’ was included in the 2007 traffic light table as part of the YOS.
The evidence was mixed, with some studies showing that children who were difficult to rouse were more likely to have a serious illness than those who were not difficult to rouse (convincing and strong positive likelihood ratio), and some studies showing that children who were difficult to rouse were not more likely to have a serious illness than those who were not difficult to rouse (not a strong positive likelihood ratio). The evidence was of low and very low quality, and most of the studies focused on detecting bacterial meningitis rather than serious illness in general.
The GDG did not believe the evidence was strong enough to move or remove this from the traffic light table, and therefore no changes were made to the traffic light table for this sign.
Decreased activity (included in 2007 traffic light table)
‘Decreased activity’ was included in the 2007 traffic light table as part of the YOS.
Some studies showed that children with decreased activity were more likely to have a serious illness than children with normal levels of activity (strong positive likelihood ratio); however, other studies showed that children with decreased activity were not more likely to have a serious illness than children with normal levels of activity (not strong positive likelihood ratio). The evidence was mainly of low to very low quality.
The 2007 recommendation referred to decreased activity by parental report, but the 2013 review shows that decreased activity at presentation to a healthcare professional was also a useful symptom or sign of serious illness. The GDG acknowledged that decreased activity was difficult to define, and that it was difficult to distinguish between ‘moderate’ and ‘severe’ impairment, as reported in one of the included studies.
Given the varied definitions and quality of the evidence the GDG decided to keep decreased activity in the ‘amber’ column, and so no changes were made to the traffic light table.
No smile (included in 2007 traffic light table)
The GDG stated this was included in the 2007 traffic light table as part of the YOS.
No new evidence was identified in the 2013 review. Therefore, the GDG agreed that this sign would not be changed or removed.
Weak, high-pitched or continuous cry (included in 2007 traffic light table)
This feature was included in the 2007 traffic light table as part of the YOS.
The evidence from the 2013 review was low in quality, and only one study reported diagnostic data or data that allowed diagnostic data to be calculated. The study showed that children with an abnormal cry were not more likely to have serious illness than children without an abnormal cry (not a strong positive likelihood ratio).
The GDG therefore stated that the 2013 data was not strong enough to change or remove ‘weak, high-pitched or continuous cry’ from the traffic light table.
Irritability (identified in 2013 review)
The evidence showed that children who were irritable were not more likely to have a serious illness than children who were not irritable (not a strong positive likelihood ratio).
The sign ‘content/smiles’ is already included in the ‘green’ column of the traffic light table. The GDG believed that this is in line with the evidence that shows children without irritability usually do not have a serious illness. The GDG believed there was a general consensus in clinical practice that irritability can be defined as when an infant or child is uncomfortable when picked up or moved; however, none of the studies adequately defined irritability.
As ‘content/smiles’ is already included in the ‘green’ column of the table, the GDG did not add irritability to the traffic light table.
Decreased consciousness and/or coma (identified in 2013 review)
The evidence implied that children with decreased consciousness were not more likely to have a serious illness than children with a normal level of consciousness (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious bacterial infection usually did not have decreased consciousness (high specificity). However, children with a serious bacterial infection did not usually present with decreased consciousness (low sensitivity).
The reviewed evidence was based on a population outside the intended guideline population; that is, children older than 5 years or those with febrile convulsions. Furthermore, the GDG believed that this sign was already included in the traffic light table as ‘does not wake, or if roused, does not stay awake’.
Based on the quality of the available evidence and its discussion, the GDG decided that no changes relating to decreased consciousness and/or coma were needed to the traffic light table.
Restlessness (identified in 2013 review)
The evidence regarding restlessness was reported in one study. Children who were restless were not more likely to have a serious illness than children who were not restless (not a strong positive likelihood ratio). Children with a serious illness were often restless (moderate sensitivity); however, children without a serious illness were also often restless (low specificity).
Based on the limited evidence, the GDG did not believe restlessness was a useful symptom to detect serious illness. Therefore, restlessness was not added to the traffic light table.
Respiratory
The majority of respiratory symptoms were originally included in the traffic light table as indicators of pneumonia.
Nasal flaring and grunting (included in 2007 traffic light table)
No new evidence was found for nasal flaring or grunting in the 2013 review.
The GDG emphasised that clinical judgment should be used to distinguish between nasal flaring (amber symptom/sign) and grunting (red symptom/sign).
Based on the available evidence and its discussion, the GDG decided that no changes relating to nasal flaring and grunting were needed to the traffic light table.
Tachypnoea (included in 2007 traffic light table)
Abnormal respiratory rate was included in the 2007 traffic light table as a non-specific marker of serious illness, a specific feature of pneumonia and required for the assessment of dehydration. A statement about measuring respiratory rate was combined with the statement about the physiological parameters which should be documented as part of the assessment.
The 2013 review of the evidence showed that children who had tachypnoea were not more likely to have a serious illness than children who did not have tachypnoea (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious illness often did not have tachypnoea (moderate specificity). However, the evidence showed that children with a serious illness also did not usually have tachypnoea (low sensitivity). The available evidence was of low quality.
The cut-offs proposed by Fleming et al. (2011) and Nijman et al. (2012) were reviewed, but there was no significantly clear evidence on specific rates to alter the categories.
The GDG members concluded from their experience that respiratory rate is an important physiological parameter which needs to be assessed by healthcare professionals.
Given the low quality of the evidence, the GDG did not believe the evidence was strong enough to change or remove an existing recommendation. Therefore, no changes relating to tachypnoea were made to the traffic light table.
Oxygen saturation (included in 2007 traffic light table)
Oxygen saturation was included in the original traffic light table as a specific sign of pneumonia.
The current review did not find any evidence regarding oxygen saturation for detecting serious illness.
However, the GDG members were aware that the measurement of oxygen saturation is becoming more common amongst GPs and non-paediatric accident and emergency departments. Using their expert opinions, the GDG members believed that oxygen saturation should be retained in the traffic light table.
Based on the available evidence and its discussion, the GDG decided that no changes relating to oxygen saturation were needed to the traffic light table.
Moderate or severe chest indrawing (included in 2007 traffic light table)
Chest indrawing was included in the original traffic light table as a specific sign of pneumonia.
The current review did not find any further evidence regarding chest indrawing for detecting serious illness.
The GDG decided it should be retained in the traffic light table. A definition of chest indrawing is provided in the glossary.
Crackles (included in 2007 traffic light table)
The evidence relating to crackles in the 2013 review was of low and very low quality. The evidence showed that children with crackles were not more likely to have a serious illness than children who did not have crackles (not a strong positive likelihood ratio). In addition, the evidence suggested children without a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, usually did not have crackles (high specificity). However, children with a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, also did not usually have crackles (low sensitivity). The evidence was of low to very low quality.
Given the quality of the evidence, the GDG did not believe the evidence was strong enough to change or remove an existing recommendation. Therefore, no changes relating to crackles were made to the traffic light table.
Respiratory symptoms (identified in 2013 review)
The 2013 review highlighted that the evidence supports existing symptoms and signs in the original traffic light table.
The GDG believed that the new evidence was not defined well enough to add anything further to the assessment of respiratory symptoms.
Therefore, no changes relating to respiratory symptoms were made to the traffic light table.
Nasal symptoms (identified in 2013 review)
The evidence shows that serious illness is not ruled out by a lack of nasal symptoms (low sensitivity).
The GDG members were aware from their clinical experience that less serious complaints, such as upper respiratory tract infections, are often used to rule out the presence of a serious illness. However, the GDG stated that nasal symptoms were too common to be of practical use.
The GDG, therefore, did not add nasal symptoms to the ‘green’ column of the traffic light table.
Wheeze (or stridor) (identified in 2013 review)
The evidence shows that children who had wheeze were not more likely to have a serious illness than children who did not have wheeze (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, usually did not have wheeze (high specificity). However, children with a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, also did not usually have wheeze (low sensitivity).
The GDG agreed that wheeze was too common a symptom to be moved into the ‘amber’ or ‘red’ columns. Therefore, no changes relating to wheeze were made to the traffic light table.
Chest findings/abnormal chest sounds (identified in 2013 review)
The evidence showed that children who had abnormal chest sounds were not more likely to have a serious illness than children who did not have abnormal chest sounds (not a strong positive likelihood ratio). In addition, the evidence showed that children without pneumonia, a urinary tract infection or bacteraemia usually did not have abnormal chest sounds (high specificity). However, it also showed that children with pneumonia, a urinary tract infection or bacteraemia also did not usually have abnormal chest sounds (low sensitivity).
The GDG highlighted that ‘crackles’ was already included in the traffic light table, which was a better defined sign than ‘chest findings’ or ‘abnormal chest sounds’. One of the studies included in the review was for detecting urinary tract infection, and the GDG was unsure how relevant chest findings or abnormal chest sounds would be to this diagnosis.
Given the quality of the evidence and the fact that an item already covering this feature was already included in the traffic light table, the GDG decided not make any changes relating to chest findgins/abnormal chest sounds to the traffic light table.
Cough (identified in 2013 review)
The available evidence showed that children who had a cough were not more likely to have a serious illness than children who did not have a cough (not a strong positive likelihood ratio). There was some evidence that children without a urinary tract infection usually did not have a cough (high specificity), but other evidence showed that children without a urinary tract infection, pneumonia, bacteraemia or meningococcal disease often had a cough (low specificity). In addition, children with a urinary tract infection, pneumonia, bacteraemia or meningococcal disease did not usually have a cough (low sensitivity).
The evidence suggests cough was not a useful predictor of serious illness, although the GDG highlighted that two of the studies were on detecting urinary tract infection and it was not clear how relevant cough was to this diagnosis. There was not enough evidence for the GDG to determine that cough was a useful symptom or sign in the detection of serious illness. Furthermore, the GDG stated ‘cough’ was too common to be of practical use.
Based on the available evidence and the results of its discussion, the GDG decided not make any changes relating to cough to the traffic light table.
Circulation and hydration
In the 2007 guideline the GDG recognised that dehydration was a marker of serious illness but there was a lack of evidence to determine the difference between mild, moderate and severe dehydration. The most specific symptoms and signs of dehydration have been highlighted for healthcare professionals to assess in order to ensure a low false positive rate and are included in the guideline Diarrhoea and vomiting in children under 5 (NICE, 2009). As evidence was found relating to the use of heart rate in the diagnosis of serious illness, the ‘hydration’ category was changed to ‘circulation and hydration’ for greater clarity.
Dry mucous membranes and reduced skin turgor (included in 2007 traffic light table)
The GDG acknowledged that dry mucous membranes and reduced skin turgor were included in the 2007 traffic light table based on a study that reviewed signs and symptoms of dehydration, rather than a study of serious illness associated with fever. However, the GDG members stated that, in their experience, dehydration was a marker for serious illness and therefore should be included in the traffic light table.
No new evidence was found for dry mucous membranes and/or reduced skin turgor in the 2013 review.
The GDG acknowledged that the recommendations regarding signs of dehydration in the 2007 Fever guideline were intended for use primarily in children who had been sent home after seeing a healthcare professional. Since the publication of the 2007 Feverish Illness in Children guideline, a clinical guideline on diarrhoea and vomiting has been published (Diarrhoea and vomiting in children under 5, NICE 2009). The Diarrhoea and vomiting guideline concluded that looking at physical signs of dehydration was an inaccurate way of determining whether a child was moderately or severely ill, as it is difficult to distinguish between different severities of dehydration. However, the two guidelines consider different populations, and if a child exhibits diarrhoea and/or vomiting they are treated in accordance with that guideline rather than the Fever guideline. The GDG also emphasised that the purpose of the traffic light table is to raise awareness rather than to make clear definitive diagnosis.
In the absence of evidence to challenge the 2007 recommendation, the GDG did not change it.
Poor feeding (included in 2007 traffic light table)
The 2013 review did not find clear evidence relating poor feeding to an increased risk of serious illness. Children who showed poor feeding were not more likely to have a serious illness than children who showed normal feeding (not a strong positive likelihood ratio). The evidence was of low to very low quality.
However, the GDG members stated that, in their clinical experience, poor feeding was a key reason that parents or caregivers bring their child to a healthcare professional. In recognition that poor feeding was a worrying feature, but not an immediate alarm feature, its position was in the amber column in the 2007 traffic light table. The GDG acknowledged that it was hard to define poor feeding. Depending on the age of the child, it can be difficult to assess how much the child is feeding, for example if the child is being breastfed. Furthermore, the GDG also acknowledged that the Nademi et al. (2001) study includes children up to age 16 years, who have more control over their own feeding habits, and therefore the data may not be applicable to the population covered by this guideline who are under aged 5 years. In addition, the Newman et al. (2002) study investigates urinary tract infection, which is not relevant to this sign.
The GDG’s decision was that the new data was not strong enough to support changing the 2007 recommendation, and so no changes were made to it.
Capillary refill time of 3 seconds or more (included in 2007 traffic light table)
In the 2007 guideline the GDG noted that capillary refill time is quick to carry out and exhibits moderate reproducibility. A statement about measuring capillary refill time was combined with the statement about the physiological parameters which should be documented as part of the assessment (see the end of Respiratory rate section). The GDG considered that a capillary refill time of 3 seconds or more was an ‘amber’ sign (see the recommendations at the end of Respiratory rate section).
For the 2013 review the evidence showed that children with a capillary refill time of more than 3 seconds were more likely to have a serious illness than children with a capillary refill time of 3 seconds or less (strong positive likelihood ratio). In addition, evidence showed that children without a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, usually did not have an increased capillary refill time (high specificity). However, children with a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, did not usually have an increased capillary refill time either (low sensitivity). The evidence was of low to very low quality.
The GDG acknowledged that in the cut-offs reported in the Craig et al. (2010) study the capillary refill time is measured in whole seconds, and so greater than 3 seconds would be 4 seconds or more. However, the other cut-off reported in the study is 2 to 3 seconds. It was not clear whether the data for children with a capillary refill time of 3 seconds were included in the results.
In the 2013 review the GDG acknowledged that there is a difference in central and peripheral capillary refill time. The GDG was aware that peripheral capillary refill time can be affected without indicating a serious illness, and that taking peripheral measurements can be inaccurate and lead to false positives. The GDG emphasised that it is not a sign that should be used in isolation. For further details, please refer to the guideline Bacterial meningitis and meningococcal septicaemia (NICE, 2010).
The GDG stated that the data identified in the 2013 review was of limited quality and not strong enough to change the 2007 recommendations. Therefore, no changes relating to capillary refill time of 3 seconds or more were made to the 2007 recommendations.
Reduced urine output (included in 2007 traffic light table)
The evidence in the 2013 review showed that children with a reduced urine output were not more likely to have a serious illness than children with a normal urine output (not a strong positive likelihood ratio). In addition, the evidence showed children without a serious bacterial infection or a urinary tract infection often did not have reduced urine output (moderate specificity). However, children with a serious bacterial infection or a urinary tract infection also did not usually have reduced urine output (low sensitivity). The evidence was of low to very low quality.
The GDG members stated that in their experience reduced urine output is commonly reported by parents and caregivers as a marker of dehydration and its position in the amber column reflected its relevance.
Based on the quality of the evidence and its discussion, the GDG decided to keep the existing recommendation.
Other
Fever for 5 days or more (included in 2007 traffic light table)
This sign was included in the 2007 guideline as it was indicative of Kawasaki disease; however, the new review found only two studies that reported on Kawasaki disease, neither of which reported on duration of fever. There was evidence that those with a serious bacterial illness had had fever for longer than children without serious illness (significant P values), and children who had had fever for three days or more were significantly more likely to have a urinary tract infection than those who had not (significant relative risk). No evidence was reported that examined fever duration of longer than 5 days.
Based on their clinical experience, the GDG members argued that most non-serious illnesses will resolve themselves after 5 days, and therefore a fever of more than 5 days duration is a good indicator of serious illness. The GDG acknowledged that in the evidence there is a weak correlation between duration of fever and severity of illness. However, it believed this may be in part to relying on parental/caregiver recall of when the fever started. Also, the evidence was limited as many studies excluded children who had had fever for 5 days or longer and none of the studies used 5 days as a cut-off.
The GDG concluded that the evidence in the current review was not strong enough to change the 2007 recommendations and therefore no such changes were made.
Temperature of 38°C or more in children age under 3 months, temperature of 39°C or more in children age 3–6 months (included in 2007 traffic light table)
In the 2007 guideline the GDG concluded that healthcare professionals should be aware that there is an association between height of body temperature and risk of serious bacterial illness. However, this association was not sufficiently robust to recommend immediate action or referral based on body temperature alone. An exception was made for children aged less than 6 months with a body temperature of 39°C or higher because the evidence was strongest for this age group.
In the 2013 review, the GDG acknowledged the ambiguity of the age groups in the 2007 recommendation regarding height of fever, and altered the text of the recommendation to reflect the intended meaning of less than 3 months for one group, and age 3 to 6 months (inclusive) for the other group. No studies were identified for the 3 to 6 month age group specifically, although most studies included this age group in their sample. The studies often did not report how the temperature was measured, and the studies tended to look at one or two cut-offs rather than a range of temperatures, making it hard to compare data from different temperature cut-offs. Despite these limitations in the data, the GDG highlighted that there is a correlation between high temperature and serious bacterial infection in general, but that, on an individual basis, high temperature was not useful for detecting serious illness. The current review suggests that there is a plateau in positive predictive values, negative predictive values and likelihood ratios around 39°C and 40°C, suggesting that a temperature above this does not provide a better indication of serious illness. The GDG therefore decided to move the recommendation regarding height of fever in the 3 to 6 month age group from the red column to the amber column. The GDG acknowledged that any fever in a child under 3 months is a risk factor for serious illness in itself, and so the recommendation for this age group remained in the red column.
The GDG made it clear that use of height of fever alone should not be used to diagnosis a serious illness. In addition, the GDG noted that children aged less than 3 months with fever are generally at a higher risk of serious illness (see Section 8.2). The incidence of serious illness in this group, for instance, was over ten times higher than that in older children. The clinical studies that provide the evidence for this age group used a body temperature of 38°C or higher as the definition of fever.
The GDG was also aware that infants in England and Wales have their first immunisations at age 2 months and that most of these infants experience post-immunisation fever. There was a discussion about what impact a recommendation on height of fever in this age group would have on health services, with a potential for health services to be overwhelmed. However, it was highlighted that parents and carers were routinely advised to expect their child to have a fever within 48 hours of immunisation and that there was no evidence of an increase in consultations due to this.
The GDG therefore decided that children aged less than 3 months with a body temperature of 38°C or higher should be included in the recommendation about risk of serious illness.
Non-blanching rash, bulging fontanelle and neck stiffness (included in 2007 traffic light table)
In the 2007 traffic light table there were several symptoms and signs that were included because they are indicative of meningococcal septicaemia or bacterial meningitis, including non-blanching rash, bulging fontanelle and neck stiffness. The evidence was of low to very low quality.
The 2013 review reported that there was some evidence that children with a non-blanching rash were more likely to have a serious illness than children who did not have a non-blanching rash (convincing positive likelihood ratio); however, there was also evidence that children with a non-blanching rash were not more likely to have a serious illness than children who did not have a non-blanching rash (not a strong positive likelihood ratio). In addition, children without a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, usually did not have a non-blanching rash (high specificity). Children with a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, also did not usually have a non-blanching rash (low sensitivity).
The evidence for the 2013 review showed that children with a bulging fontanelle were not more likely to have a serious illness than children without a bulging fontanelle (not a strong positive likelihood ratio). Children without a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, usually did not have bulging fontanelle (high specificity). Children with a serious illness, such as pneumonia, a urinary tract infection or bacteraemia, also did not usually have a bulging fontanelle (low sensitivity).
The 2013 review reported that there was some evidence that children with neck stiffness were more likely to have meningococcal disease than children who did not have neck stiffness (convincing positive likelihood ratio). In addition, children without a serious illness, such as meningitis, pneumonia, a urinary tract infection or bacteraemia, usually did not have neck stiffness (high specificity). Children with a serious illness, such as meningitis, pneumonia, a urinary tract infection or bacteraemia, also did not usually have neck stiffness (low sensitivity).
The 2013 review also found that there was evidence that children with focal seizures were more likely to have bacterial meningitis than children who did not have focal seizures (strong positive likelihood ratio). In addition, children without a serious illness, such as meningitis, pneumonia, a urinary tract infection or bacteraemia, usually did not have focal seizures (high specificity). However, children with a serious illness, such as meningitis, pneumonia, a urinary tract infection or bacteraemia, also did not usually have focal seizures (low sensitivity).
Since the 2007 Fever guideline, a guideline on bacterial meningitis in children and young people has been published. The guideline Bacterial meningitis and meningococcal septicaemia (NICE, 2010) includes a comprehensive list of symptoms and signs of bacterial meningitis and meningococcal septicaemia. However, it is worth noting that the bacterial meningitis guideline is relevant when bacterial meningitis or meningococcal septicaemia is suspected, whereas the Fever guideline is relevant for children that do not have a known source of fever. The GDG stated that the most relevant symptoms and signs of bacterial meningitis and meningococcal septicaemia were included in the 2007 traffic light table, and the 2013 review found no strong evidence to move or remove these from the traffic light table. The GDG was aware that the symptoms of cold hands and feet and limb pain are included in the list of clinical features found in meningococcal disease and meningitis in the 2010 guideline.
Although it was of low quality, the available evidence supported the existing recommendation and matched the opinion of the GDG. Therefore, it was decided that the traffic light table did not need to be changed.
Status epilepticus (included in 2007 traffic light table)
No evidence was identified in the 2013 review for status epilepticus.
Based on their clinical experience, the GDG members stated that status epilepticus should remain in the ‘red’ column, as it is a serious condition and a child with status epilepticus needs urgent referral. Therefore, no changes were made to the recommendation on status epilepticus.
Focal neurological signs and focal seizures (included in 2007 traffic light table)
The GDG highlighted that focal neurological and focal seizures were included in the traffic light table as they may be indicative of Herpes simplex encephalitis.
There was no evidence identified in the 2013 review that reported on neurological signs or focal seizures for identifying serious illness.
Based on their clinical experience, the GDG members did not know of any clinical reason to move these signs from the ‘red’ column of the traffic light table and therefore no changes were made.
Swelling of a limb or joint, and non-weight bearing limb/not using an extremity (included in 2007 traffic light table)
The GDG highlighted that both swelling of a limb or joint and non-weight bearing limb/not using an extremity were included in the 2007 traffic light table as they are indicative of septic arthritis.
No evidence was identified in the 2013 review regarding swelling of a limb or joint and/or non-weight bearing limb for detecting serious illness. The GDG acknowledged that the consequences of missing the diagnosis of septic arthritis in a child are serious. However, it was also aware that this is not a common illness. The GDG also acknowledged that many children with swelling and/or non-weight bearing will recover from these symptoms in a few days, and so they do not require immediate referral.
Based on the available evidence and its discussion, the GDG decided that no changes were needed and these two symptoms should remained in the amber category of the traffic light table.
A new lump greater than 2 cm (included in 2007 traffic light table)
There was no evidence in the 2013 review to support including ‘new lump greater than 2 cm’ in the traffic light table. The study on which the 2007 recommendation was based was excluded as it included non-febrile surgical patients.
The GDG highlighted that ‘new lump greater than 2 cm’ was originally included in the traffic light table based on one study that was excluded from the update as the population included a high proportion of children without fever. A significant number of children in this study were diagnosed with hernias and other surgical conditions. Moreover, in a subset analysis of children with fever from this study, a new lump larger than 2 cm did not feature in a set of risk factors for serious illness. The GDG stated that a new lump larger than 2 cm most likely indicated a hernia or an abscess requiring surgical intervention, and was not associated with fever.
The GDG therefore decided to remove the existing recommendation, and so removed ‘new lump greater than 2 cm’ from the traffic light table.
Bile-stained vomiting (included in 2007 traffic light table)
There was no evidence in the 2013 review to support including ‘bile-stained vomiting’ in the traffic light table. The study on which the 2007 recommendation was based was excluded as it included non- febrile surgical patients.
The GDG was aware that bile-stained vomiting is more likely to indicate a surgical problem, rather than a serious bacterial illness. It was included in 2007 based on one study that was excluded from the updated review, as it included a high proportion of children without fever. A significant number of children in this study were diagnosed with hernias and other surgical conditions. Moreover, in a subset analysis of children with fever from this study, bile-stained vomiting did not feature in a set of risk factors for serious illness.
The GDG therefore decided to remove the existing recommendation, and hence removed ‘bile-stained vomiting’ from the traffic light table.
Diarrhoea (identified in 2013 review)
The evidence relating to diarrhoea was mixed, with some studies showing that children without a serious bacterial infection, a urinary tract infection or a bacterial illness usually did not have diarrhoea (high and moderate specificity) and some showing that children without serious bacterial infection often had diarrhoea (low specificity). However, children with a serious bacterial infection, a urinary tract infection or a bacterial illness did not usually have diarrhoea (low sensitivity). Children with diarrhoea were not more likely to have a serious illness than children without diarrhoea (not strong positive likelihood ratio).
The GDG stated that the evidence was not consistent enough to add diarrhoea to the traffic light table. The GDG highlighted that dehydration was already included in the traffic light table. The GDG also highlighted that a child presenting with diarrhoea and/or vomiting should be managed as outlined in the guideline Diarrhoea and vomiting in children under 5 (NICE, 2009).
Based on the available evidence and its discussion, the GDG decided that no changes relating to diarrhoea were needed to the traffic light table.
Vomiting (identified in 2013 review)
The evidence showed children with vomiting were not more likely to have a serious illness than children without vomiting (not a strong positive likelihood ratio). In addition, some studies showed that children without a serious bacterial infection, a urinary tract infection or a bacterial illness usually did not have vomiting and some showed that the children without bacterial meningitis or urinary tract infection often had vomiting (moderate to high specificity). However, children with a serious bacterial infection, a urinary tract infection or a bacterial illness did not usually have vomiting (low sensitivity). The evidence was of low to very low quality.
The GDG stated that the evidence was not consistent enough to add vomiting to the traffic light table and highlighted that dehydration was already included in the traffic light table. The GDG also highlighted that a child presenting with diarrhoea and/or vomiting should be managed as outlined in the guideline Diarrhoea and vomiting in children under 5 (NICE, 2009).
Based on the available evidence and its discussion, the GDG decided that no changes relating to vomiting were needed to the traffic light table.
Abdominal pain (identified in 2013 review)
The evidence showed that children with abdominal pain were not more likely to have a serious illness than children without abdominal pain (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious illness usually did not have abdominal pain (high specificity). However, children with a serious illness also did not usually have abdominal pain (low sensitivity). The evidence was of low to very low quality.
The GDG stated that the evidence was of low and very low quality, and evidence on diagnostic accuracy was limited to that of one study. The other included study did not report diagnostic data or data that would allow diagnostic data to be calculated. It is worth noting that the temperature used as an inclusion criterion for this study was lower than other studies in the review.
The non-diagnostic accuracy evidence stated that abdominal pain is not predictive of urinary tract infection. Therefore, the GDG concluded that abdominal pain should not be added to the traffic light table.
Crying on micturition/dysuria (identified in 2013 review)
The evidence showed that children who cried on micturition were not more likely to have a urinary tract infection than children who did not cry on micturition (not a strong positive likelihood ratio). In addition, the evidence showed that children without a urinary tract infection often did not cry on micturition (moderate specificity); however, children with a urinary tract infection also did not usually cry on micturition (low sensitivity).
The GDG highlighted that the evidence was of low quality and limited to that of one study. Furthermore, the GDG stated that a child presenting with crying during micturition or dysuria would clearly be indicative of a urinary tract infection and should be managed as outlined in the guideline Urinary tract infection in children (NICE, 2007).
Based on the available evidence and its discussion, the GDG decided that no changes relating to crying on micturition/dysuria were needed to the traffic light table and this symptom was not added.
Headache (identified in 2013 review)
The evidence showed that children with a headache were more likely to have bacterial meningitis than children without a headache (convincing positive likelihood ratio). Evidence also showed that children without bacterial meningitis usually did not have a headache (high specificity) and that children with bacterial meningitis also did not usually have headache (low sensitivity).
The evidence for headache was of very low quality and limited to that of one study. The study included children from 6 months to 5 years, and it was not clear to the GDG how pre-verbal children would communicate that they had a headache. The GDG concluded that the evidence was not strong enough to add headache to the traffic light table.
Based on the quality of the available evidence and its discussion, the GDG decided that no changes relating to headache were needed to the traffic light table.
Conjunctivitis (identified in 2013 review)
The evidence showed that children with conjunctivitis were not more likely to have a urinary tract infection than children without conjunctivitis (not a strong positive likelihood ratio). In addition, the evidence showed that children without a urinary tract infection usually did not have conjunctivitis (high specificity). However, children with a urinary tract infection also did not usually have conjunctivitis (low sensitivity).
The evidence for conjunctivitis was in relation to detecting urinary tract infection, and the GDG was not convinced of a clinical link between the two conditions. Therefore, the GDG did not add conjunctivitis to the traffic light table.
Poor peripheral circulation (identified in 2013 review)
The evidence showed that children with poor peripheral circulation were not more likely to have a serious illness than children with normal peripheral circulation (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious bacterial infection often had normal peripheral circulation (moderate specificity); however, children with a serious bacterial infection also usually had normal peripheral circulation (low sensitivity).
The GDG highlighted that capillary refill time, which acts as an indicator of poor peripheral circulation with a recognised definition, is already included in the traffic light table. Furthermore, the evidence was of very low quality and was limited to that of one study. In addition, poor peripheral circulation was not defined in the study, and the evidence shows that it was not a good detector of serious illness.
Based on the available evidence and its discussion, the GDG decided that no changes relating to poor peripheral circulation were needed to the traffic light table.
Bulging abdomen (identified in 2013 review)
The evidence showed that children with a bulging abdomen were not more likely to have a serious illness than children without a bulging abdomen (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious bacterial infection often did not have a bulging abdomen (moderate specificity); however, children with a serious bacterial infection also usually did not have a bulging abdomen (low sensitivity).
Evidence was of very low quality and was limited to that of one study. The GDG found that the evidence that bulging abdomen was a useful predictor of serious illness was not convincing. Therefore, no changes relating to bulging abdomen were made to the traffic light table.
Paresis or paralysis (identified in 2013 review)
The evidence showed that children with paresis or paralysis were not more likely to have bacterial meningitis than children without paresis or paralysis (not a strong positive likelihood ratio). In addition, the evidence showed that children without bacterial meningitis usually did not have paresis or paralysis (high specificity). However, children with bacterial meningitis also did not usually have paresis or paralysis (low sensitivity).
The evidence for paresis or paralysis for detecting serious illness was of very low quality and was limited to that of one study. The included children had all had a febrile convulsion prior to inclusion in the studies. The GDG stated that a child with paresis or paralysis is likely to be identified using the traffic light table under ‘appears ill to a healthcare professional’ and ‘focal neurological signs’. The evidence was not convincing to add paresis or paralysis as an additional symptom or sign.
The GDG decided that paresis or paralysis should not be added to the traffic light table.
Abnormal neurological findings (identified in 2013 review)
The GDG stated that ‘abnormal neurological findings’ is already covered in the traffic light table under ‘focal neurological signs’ and ‘appears ill to a healthcare professional’. The new evidence was not strong enough to add abnormal neurological findings to the traffic light table as a separate symptom or sign. All of the included studies used abnormal neurological findings to detect bacterial meningitis, and a child presenting with bacterial meningitis should be managed as outlined in the guideline Bacterial meningitis and meningococcal septicaemia (NICE, 2010).Therefore, no changes relating to abnormal neurological findings were made to the traffic light table.
Impression of tone (identified in 2013 review)
The evidence for impression of tone was limited to one study, which did not report diagnostic accuracy data or data that would allow diagnostic accuracy data to be calculated. The evidence stated that tone was not significantly associated with bacteraemia.
Therefore, the GDD decided that impression of tone should not be added to the traffic light table.
Tenderness on examination (identified in 2013 review)
The review results showed that children who showed tenderness on examination were not more likely to have a urinary tract infection than children who did not show tenderness on examination (not a strong positive likelihood ratio). In addition, the evidence showed that children without a urinary tract infection usually did not have tenderness on examination (high specificity). However, children with a urinary tract infection also did not usually have tenderness on examination (low sensitivity).
The GDG stated that tenderness on examination was not described in enough detail in the study to be used, although the GDG acknowledged that it was likely to refer to abdominal tenderness, as the study reports on urinary tract infection. In addition, the evidence was not strong enough for it to be added to the traffic light table.
Therefore, the GDG decided that tenderness on examination should not be added to the traffic light table.
Urinary symptoms (identified in 2013 review)
The evidence showed that children with urinary symptoms were not more likely to have a serious bacterial infection than children without urinary symptoms (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious bacterial infection usually did not have urinary symptoms (high specificity). However, children with a serious bacterial infection also did not usually have urinary symptoms (low sensitivity). There was some evidence that children with urinary symptoms were more likely to have a serious bacterial infection than children without urinary symptoms (strong positive likelihood ratio).
‘Urinary symptoms’ was not defined in the studies, although the GDG acknowledged that the term is likely to refer to symptoms and signs of urinary tract infection. This suggests a definite source cause of fever, and was not a helpful symptom or sign to add to the traffic light table. A child presenting with urinary symptoms should be managed as outlined in the guideline Urinary tract infection in children (NICE, 2007).
The GDG stated that two of the symptoms described in the 2007 guideline – offensive urine and haematuria – were rare, and if present would refer to a urinary condition. Therefore, these were removed from the recommendation. This did not result in any changes to the traffic light table.
Abnormal ear, nose and throat signs (identified in 2013 review)
The evidence was mixed for ‘abnormal ear, nose and throat signs’. One study showed that children with abnormal ear, nose and throat signs were not more likely to have a serious illness than children with no signs (not a strong positive likelihood ratio). In addition, the evidence showed that children without a serious bacterial infection often had abnormal ear, nose and throat signs (low specificity), while another study showed that children without a serious bacterial infection usually did not have ear problems (high specificity). Both studies showed that children with a serious bacterial infection did not usually have abnormal ear, nose and throat signs or ear problems (low sensitivity).
The GDG highlighted that the evidence was of low and very low quality, and symptoms were too common to add ‘abnormal ear, nose and throat signs’ to the traffic light table.
The GDG therefore did not add ‘abnormal ear, nose and throat signs’ to the traffic light table.
Rigor and/or chills (identified in 2013 review)
The evidence suggested that children with rigors were not more likely to have a bacterial illness than children who did not have rigors (not strong positive likelihood ratio). The evidence showed that children without bacterial illness often did not have rigors (moderate specificity); however, children with a bacterial illness also usually did not have rigors (low sensitivity). The evidence was of very low quality.
The GDG highlighted that rigors are caused by a high body temperature, and are therefore associated with high temperatures in children. The GDG acknowledged that there was evidence of a link between higher temperatures in children and serious illness, and therefore rigors could be an indicator of serious illness. The GDG was aware that rigors are an uncommon symptom/sign in children under 5 years, but there was insufficient evidence that rigors alone signal the need for urgent attention.
The GDG stated that the quality of the evidence and positive likelihood ratio meant that rigors could not be added to the red column of the traffic light table. However, the GDG did feel it was an important feature and the decision was therefore made to add rigors to the amber category of the traffic light table.
Cold hands and feet (identified in 2013 review)
No evidence regarding cold hands and feet was reported in the 2013 review.
The GDG noted clinical overlap with poor peripheral circulation but that the NICE Bacterial meningitis and meningococcal septicaemia guideline (NICE, 2010) had identified cold hands and feet as a relevant sign when considering a diagnosis of meningitis.
The GDG was aware that the symptoms of cold hands and feet are included in the list of clinical features found in meningococcal disease and meningitis in the NICE 2010 meningitis guideline. However these symptoms were taken from uncontrolled studies and did not therefore fulfil the inclusion criteria of the updated Feverish illness guideline. Moreover, a study of these symptoms and signs in children with self-limiting viral illness found that cold hands and feet were reported in 20% to 24% of young children. The specificity of this symptom for detecting meningococcal disease would therefore be low. The GDG emphasised that in isolation, for undifferentiated children with fever, other features of the traffic light table were sufficient to identify high risk children and therefore did not add this symptom or sign to the traffic light table.
A child presenting with cold hands and feet should be diagnosed as outlined in the guideline Bacterial meningitis and meningococcal septicaemia (NICE, 2010).
Based on the available evidence and its discussion, the GDG decided that no changes relating to cold hands and feet were needed to the traffic light table.
Yale Observation Score
The evidence suggests that the Yale Observation Score was good at identifying children who do not have a serious illness. However, it was less good at identifying children who do have a serious illness. This was in line with the evidence found for the 2007 review that the YOS alone was not a good detector of serious illness. As highlighted in the 2007 review, the GDG acknowledged that the usefulness of the YOS was increased when it was used in combination with a history taken by a physician and examination.