Measuring vital signs in children with fever at the emergency department: an observational study on adherence to the NICE recommendations in Europe

We performed a prospective observational study in 28 EDs in 11 European countries, including patients under the age of 16 and with a fever as their presenting complaint. Children were excluded if they presented to the ED repeatedly for the same problem within 7 days, if they were treated with antibiotics in the 7 days before the ED visit, or if they had a documented allergy to antibiotics. For the current study, children with comorbidities were also excluded, as disease-specific characteristics may influence their management. In the whole population, we evaluated the measurement of vital signs. In children under five, we assessed the adherence to the recommendations to measure four distinct vital signs from the NICE guideline ‘Fever in under 5s: assessment and initial management’ [13].

Data collection took place between October 2014 and February 2016 within the network of Research in European Pediatric Emergency Medicine (REPEM). Detailed methods have been published earlier [19]. In short, all participating 28 EDs recorded medical information for all attending children with fever for one random day each month. We recorded general characteristics of patients (age, sex, weight, height, comorbidities), vital signs (heart rate, respiratory rate, temperature, oxygen saturation, capillary refill time) and information regarding diagnosis and management. Data were extracted from routine patient records, and filled in on an electronic study case report form (CRF) by the local investigator after the sampling day (Electronic Supplementary Material 1). Comorbidities and diagnoses were recorded according to pre-specified categories. We neither used ICD-codes for the recording of diagnoses nor had we access to data after the ED visit. Consequently, ‘diagnosis’ in this manuscript refers to a presumed diagnosis at ED discharge. All items in the CRF were mandatory to fill in, with the option to choose ‘unknown’. Unknown values on vital signs were seen as ‘not measured’, and were therefore considered to be outcomes rather than omissions. Local investigators were aware of the sampling days and the general scope of the study as a registry of febrile children, but vital sign measurement was not known as a specific point of interest. Hospital information was collected using a survey, including questions on guideline use. We collected data on hospital setting (inner city/rural/mixed), hospital type (academic/teaching/non-teaching), triage system and number of annual paediatric ED visits, similar to other studies on the organization of care (Electronic Supplementary Material 2) [20]. Setting reflects the population in the catchment area of the hospital. Academic hospitals are connected to a university, teaching hospitals non-university hospitals that provide training for paediatrics residents; non-teaching hospitals do not provide training of residents.

Not every study hospital used the same triage system, but they all classified children according to a five-point scale, ranging from ‘non-urgent’ to ‘immediate’, making comparisons possible. Owing to the small number of cases, patients in the ‘immediate’ and ‘very urgent’ categories were grouped together. Tachycardia and tachypnoea were defined according to the advanced pediatric life support (APLS) guideline [21]. Fever was defined as temperature ≥ 38 °C, hypoxia as peripheral oxygen saturation level of ≤ 94%. Crowding of the ED was defined for each hospital according to their number of total paediatric ED visits on the sampling day (less than usual/as usual/more than usual). We defined a usual number of total visits as the interquartile range of the number of total visits across the different sampling days per hospital. If on a sampling day the number of total visits was lower than the 25th percentile for that hospital, crowding was less than usual, if the number was higher than the 75th percentile, the ED was more crowded than usual.

Adherence to the NICE guideline was based on the following indicator: ‘Measure and record temperature, heart rate, respiratory rate and capillary refill time as part of the routine assessment of a child with fever.’ [13]. Adherence to the NICE guideline was defined as the complete measurements of those four vital signs in children under 5 years old.

We used descriptive analyses to evaluate the frequency of measurement for all of the available vital signs in the study population. We examined practice variations between countries, age groups, triage levels and diagnoses, visualizing the measurement of vital signs by heat maps. We used diagnosis in these analyses as a proxy of presenting complaint (next to the fever) and suspicion of severity, assuming that children with RTIs would present with respiratory symptoms, enteric infections with vomiting or diarrhoea and that children with fever without source, urinary tract infections and sepsis/meningitis mostly present without specific symptoms but with a higher suspicion of invasive infections. We compared the frequency of detecting abnormal vital signs between countries that frequently measured vital signs and countries that measured them less often.

In assessing adherence to the NICE guideline, we measured the frequency of complete measurements in children under five from all hospitals that used the NICE recommendations. We tested the influence of age, triage level, diagnosis and crowding of the ED on adherence using a multilevel logistic regression model that included hospital as a random variable. Analyses were performed using SPSS (IBM, version 24) and R (version 3.5.2).

In total, 5255 children were included in the complete cohort, all presenting with fever and without prior antibiotic treatment or repeated ED visits. In the current study, we included 4560 children. Exclusion was mostly because of comorbidities (Fig. 1). Of the included children, 53.8% were male and the median age was 2.4 years (interquartile range 1.1–4.7). Table 1 shows their baseline characteristics and provides information regarding patients’ way of referral and follow-up. Baseline characteristics of children with comorbidities have been published earlier [19]. In general, these children were more ill and older than children without comorbidities. Of the 28 participating hospitals, 17 were academic hospitals, 10 teaching hospitals and one non-teaching hospital (Table 2). They varied from inner city hospitals (n = 17) to regional (n = 2) and mixed hospitals (n = 9), and their number of annual paediatric ED visits ranged from 2700 to 88,000. Most hospitals used a local triage system (n = 8) or the Manchester triage system (n = 7, Table 2). All except the Spanish hospitals used the recommendation to routinely measure vital signs as mentioned in the NICE guideline.

The measurement of vital signs occurred in varying degrees, both when comparing the different vital signs with each other and across participating countries. Overall, temperature was measured most frequently (97%, 4435/4560, 95% confidence interval 97–98%), ranging between countries from 70% (78/111) in Hungary to 100% in Denmark and England (n = 24 and n = 145 respectively). Capillary refill was next (86%, 85–89%), followed by heart rate (73%, 72–75%), saturation (56%, 55–58%) and respiratory rate (51%, 50–53%), although the latter two had much wider ranges between countries. Figure 2 a contains a heat map visualizing the frequency of vital sign measurements in participating countries. Variability between countries is apparent throughout all of the different vital signs and is most striking for respiratory rates. Temperature was the most consistent, as it was measured in more than 90% of cases in all countries but one.

In the hospital in Turkey, no routine triage was performed. In the remaining hospitals, 99% (3825/3852) of children were triaged upon their arrival to the hospital. Children requiring ‘very urgent’ or ‘immediate’ care had their vital signs measured most frequently (Fig. 2b). Differences across triage levels were greatest for heart rate, saturation and respiratory rate and amounted to about 30% points between the ‘very urgent/immediate’ and ‘standard’ categories (heart rate: 93% vs. 64%, saturation: 90% vs. 59%, respiratory rate: 65% vs. 37%).

Differences in measurement across age groups were smaller (Fig. 2c). Only heart rate and saturation were more frequently measured in infants than in children > 5 years of age (heart rate: 83% vs. 71%; saturation 78% vs. 47%).

We observed an association between the measurement of vital signs and diagnosis (Fig. 2d). Most children (3307/4461, 74%) had respiratory tract infections (RTIs); only 15 children had sepsis or meningitis. Temperature and capillary refill were measured quite consistently across the different infectious foci (97% and 86% of cases respectively), but the remaining vital signs exhibited a considerable variability. Saturation was measured substantially more often in lower RTIs and in patients with sepsis/meningitis than in other cases. Heart rate was recorded in all patients with sepsis/meningitis (15/15) and in 86% (416/486) of those with lower RTIs. For fevers of unknown origin, on the other hand, heart rate measurements were included in the work-up of only 61% (174/284) of cases. Respiratory rates were measured in less than half of patients for four out of seven infectious foci and were done most frequently in patients with lower RTIs, amounting to 64% (310/486) of cases.

The incidence of abnormal vital signs when measured was generally low. Of all patients with a measured temperature, 2403 (54.2%) had a fever at the time of evaluation in the ED (Table 1). Out of these children, 889 (37%) had a temperature of 39 °C or more. Other than that, heart rate was most often abnormal, in 34.1% of cases. Twenty-nine percent of children were found to be tachypnoeic, hypoxia was found in 3.3% of cases and prolonged capillary refill in 1.5%.

We observed no correlation between the frequency of measurement of a vital sign per country and the proportion of abnormal values (out of all values measured in that country). So, less frequent measurement of a vital sign was not related to a higher proportion of abnormal values detected.

From all hospitals using the NICE recommendations, 1450/3014 (48%) of children under five underwent a complete measurement of these vital signs (95% CI 46 to 50%). A complete measurement was most frequent in children with lower RTIs and sepsis, although at a moderate compliance of 55% and 46% respectively (193/350 for lower RTIs and 5/11 for sepsis; Table 3). Multivariable analysis showed that children with RTIs had complete measurements significantly more often than children with fever without focus (odds ratio for upper RTI 1.75 (1.10–2.77), for lower RTI 3.75 (2.21–6.37); Table 3). Also, younger children were more likely to have all recommended vital signs measured than children over 1 year of age. Last, children with high triage urgency had full measurements slightly more often than non-urgent children (immediate/very urgent OR 1.62 (0.95–2.76), urgent level OR 1.36 (0.96–1.95)). Crowding of the ED had no significant effect on the frequency of complete measurement of vital signs. After adjusting for diagnosis, age and triage urgency, a substantial variability between hospitals remained (data not shown).

Fever was an inclusion criteria for our study, which explains the high frequency of completed temperature in our database and the high proportion of abnormal temperatures. The relatively high proportion with abnormal heart rate can be explained by the physiological relationship between temperature and heart rate [22, 23]. Respiratory rate was least frequently measured and with large variation across subgroups. Other studies have suggested reasons for such variability, like crying or distress of a child, or limitations in the counting technique [11, 24, 25]. We had no information on the child’s well-being or the devices used for measurement of respiratory rate, but these factors may have contributed to the observed low frequency of measurement of this vital sign. Although ED crowding has been associated with decreased quality of care [26], we found no association between ED crowding and adherence to the vital signs measurement recommendation in our study.

We observed an overall adherence of 48% to the NICE recommendation to measure four vital signs in all children under five, in our study in 2014–2016. This is lower than reported by a previous audit study in primary care in the UK (62%) after educational sessions and introduction of a template to record vital signs in the electronic health record [15]. An audit among paediatric EDs in the UK found that temperature was similarly measured as in our study (94%), but reports lower numbers for capillary refill time (53%) and higher rates for heart rate (94%) and respiratory rate (89%) measurements [16]. It may be striking that full measurement of vital signs children under five was most frequently done in children suspected of RTIs, rather than in those with suspected urinary tract infections and fever without focus. Even though the discharge diagnosis is often unknown at the moment of vital sign measurement, it is likely to assume that children with these last two diagnoses might present without specific symptoms. These children may have more diagnostic uncertainty and be at higher risk of complicated disease. Less than half of the children with suspected sepsis—although represented by a small number in our study—received the full set of vital sign measurements needed for compliance with the NICE guidelines.

Patient characteristics can only partly explain the observed practice variations. Professional adherence to guideline recommendations can also be influenced by local policy or professional experience. Even though most participating centres mentioned that their guidelines were based on the NICE guideline, in the process of translation from the UK to another setting, the evidence probably is weighed according to the local setting and practice. This may induce further practice variation across centres [18].

This study had the advantage of a sizeable, prospectively generated database containing large amounts of high quality patient information from 28 hospitals of various sizes and hospital types, from 11 different countries in Europe. Compared to the available literature in European paediatric emergency medicine, this number of included hospitals and countries is large, supporting the generalizability of our findings. However, some countries and hospitals included more patients than others, which might have influenced results. Furthermore, countries were represented by different numbers of hospitals (some countries only by one hospital), which adds uncertainty to whether measurements are a reflection of national or local policies.

The study was performed in hospitals of the REPEM research collaboration, ensuring high-quality data [27]. Their interest in research indicates that they are likely to uphold a high standard of care. The staff of participating hospitals were only aware of the general study design as a registry of febrile children, so a special focus on vital sign measurement during the study period is unlikely. Lastly, because this research treated missing variables as decisions not to perform certain measurements, some room remains for human error in data collection. However, all items in the data collection form were mandatory, with the option to fill in ‘missing’. During the preparation of this manuscript, the local investigators confirmed that ‘missing’ values were indeed ‘not recorded’.

Our numbers on compliance to the NICE recommendation obtained from 28 European EDs calls for better recording of vital signs in children. Not measuring vital signs may pose children at risk of underestimating the severity of their illness or delaying necessary treatments [14]. Even though almost all included centres had adopted the NICE recommendation to measure vital signs in all febrile children, compliance in less than half of cases is striking. Even in children with sepsis, fever without source or urinary tract infections in less than 50% of cases the full set was measured. Therefore, special attention should be given to children presenting with fever without specific symptoms, since vital sign measurements may contribute most to the identification of severe infections in this patient group. Although measurement is influenced by age and triage, it might be questioned whether triage appropriately selects children with severe disease [28].

Future research should focus on identifying reasons for non-compliance, including cultural and healthcare factors at the individual, organizational and national level [18]. Qualitative research could provide more in-depth information on the reasons for the observed discrepancies in vital sign measurements across Europe. At the same time, more evidence is needed on the diagnostic value of vital signs in different settings and patient groups and their impact on health outcomes. Such research could provide evidence for targeted measuring of vital signs in children that benefit most from complete measurements.