Article Text

Download PDFPDF

Comparison of peripheral and central capillary refill time in febrile children presenting to a paediatric emergency department and its utility in identifying children with serious bacterial infection
  1. Evelien de Vos-Kerkhof1,
  2. Tarik Krecinic1,
  3. Yvonne Vergouwe2,
  4. Henriëtte A Moll1,
  5. Ruud G Nijman1,
  6. Rianne Oostenbrink1
  1. 1Department of General Paediatrics, Erasmus MC-Sophia Children's Hospital, Rotterdam, The Netherlands
  2. 2Department of Public Health, Centre for Medical Decision Making, Erasmus MC, Rotterdam, The Netherlands
  1. Correspondence to Dr Rianne Oostenbrink, Erasmus MC-Sophia Children's Hospital, Wytemaweg 80, 3015 CN Rotterdam, Department of General Paediatrics, room Sp-1549, P.O. Box 2060, Rotterdam 3000 CB, The Netherlands; r.oostenbrink{at}


Objective To determine the agreement between peripheral and central capillary refill time (pCRT/cCRT) and their diagnostic values for detecting serious bacterial infection (SBI) in febrile children attending the paediatric emergency department (ED).

Design Prospective observational study.

Setting Paediatric ED, Erasmus Medium Care-Sophia Children's hospital, the Netherlands.

Patients 1193 consecutively included, previously healthy, febrile children (1 month–16 years) with both pCRT measurements and cCRT measurements available. SBI diagnosis was based on abnormal radiographic findings and/or positive cultures from normally sterile locations in addition to clinical criteria.

Main outcome measures Agreement between pCRT and cCRT (Cohen's κ), overall and stratified for age and body temperature. The diagnostic value of pCRT and cCRT for SBI was assessed with logistic regression.

Results Overall agreement was 0.35 (95% CI 0.27 to 0.43; considered ‘fair’). Although not significant, agreement was lower in children aged 1–<5 years (κ: 0.15 (95% CI 0.04 to 0.27)) and decreased with higher body temperatures with κ ranging from 0.55 (95% CI 0.32 to 0.79) for temperature <37.5°C to 0.21 (95% CI 0.07 to 0.34) for temperature >39.5°C. Abnormal pCRT (>2 s) was observed in 153 (12.8%; 95% CI 10.9% to 14.7%) and abnormal cCRT in 55 (4.6%; 95% CI 3.4% to 5.8%) children. The OR of abnormal pCRT (>2 s) for predicting SBI was 1.10 (95% CI 0.65 to 1.84). For abnormal cCRT (>2 s), the OR was 0.43 (95% CI 0.13 to 1.39).

Conclusions The pCRT and cCRT values showed only fair agreement in a general population of febrile children at the ED, and no significant association with age or body temperature was found. Only a small part of febrile children at risk for serious infections at the ED show abnormal CRT values. Both abnormal pCRT and cCRT (defined as >2 s) performed poorly and were non-significant in this study detecting SBI in a general population of febrile children.

  • Accident & Emergency
  • Infectious Diseases
  • Measurement
  • Circulatory
  • General Paediatrics
View Full Text

Statistics from

What is already known on this topic?

  • Abnormal capillary refill time (CRT) is considered a warning sign for serious infection in febrile children.

  • CRT differs by body site, peripheral (fingertip) CRT differs from central (sternum) CRT.

  • Studies on the association between abnormal CRT and serious infection in febrile children are conflicting.

What this study adds?

  • Dichotomised peripheral (fingertip) and central (sternum) capillary refill times (CRTs) differ, as assessed with chance-corrected agreement and this difference is not significantly associated with categorised age and body temperature.

  • Only a small part of febrile children at risk for serious infections at the emergency department (ED) show abnormal CRT values.

  • Both abnormal peripheral CRT or central CRT (defined as >2 s) have low diagnostic value for serious infections in febrile children at the ED.


Assessing febrile children presenting to a paediatric emergency department (ED) is challenging. Primarily, we aim to establish the risk of having a serious bacterial infection (SBI) with a potential lethal outcome or serious sequelae.1 ,2 Several clinical determinants that could serve as warning signs for SBI have been identified. Capillary refill time (CRT) has been shown to be associated with an increased risk of SBI.1 ,3 ,4 It has also been included in National Institute for Health and Care Excellence (NICE) guidelines as a marker of severe infection and of dehydration in gastroenteritis.5–7 The NICE guideline on the evaluation of feverish illness in young children includes CRT assessment at initial evaluation, but labels prolonged CRT an amber (not red) warning sign in the ‘traffic light system’, reflecting some ambiguity regarding the predictive value for SBI.5 This may be supported by studies including children within broader age ranges, which showed a modest predictive value of CRT for serious infection.3 ,8

From a pathophysiological perspective CRT is considered to indicate peripheral skin perfusion.9 In states of reduced effective circulating volume, for example, shock or dehydration, the CRT is presumed to be prolonged as a result of a vasoconstrictive response in peripheral vascular beds.10 This mechanism aims at retaining circulating volume for vital organs (eg, central nervous system) while reducing flow to non-vital tissues (eg, gastrointestinal tract and extremities).9 Different distances to the heart could lead to a discrepancy between peripheral CRT (pCRT), as commonly measured at the fingertip, and central CRT (cCRT) as commonly measured at the sternum. Several studies confirmed that CRT differs at different body sites11 ,12 although based on selected populations of, for example, healthy children, children with minor injury or illness, neonates, or infants. Uncertainty remains as we do not know the potential effect of these differences in CRT on its predictive value for SBI in febrile children. We investigated both the potential agreement between pCRT and cCRT, and the role of both as a clinical marker of SBI in a general population of febrile children presenting to a paediatric ED.


Study design and setting

The current study was undertaken as part of a larger prospective observational diagnostic study at the ED of the Erasmus Medium Care-Sophia Children's Hospital in Rotterdam, The Netherlands.13 We determined the intermethod agreement between pCRT and cCRT, and the diagnostic value for the detection of SBI for either pCRT or cCRT.

Study population

We consecutively enrolled febrile children aged 1 month to 16 years at the ED from February 2009 to May 2012. Children were included if fever (≥38.5°C) was observed at home in the last 24 hours, measured at the ED or if the child was triaged as febrile. All children were triaged according to the Manchester Triage System (MTS).14 Children with significant underlying comorbidity (eg, immune deficiency or malignancy) were excluded. Only data from initial visits of a febrile episode were used.13 We evaluated data from children with both pCRT and cCRT measurements available. Ethical approval was obtained from the institutional review board and informed consent was obtained.13


CRT was assessed by measuring the time from release until return of skin colour after pressing on either the sternum for cCRT, or fingertip/nail bed for pCRT for 5 s, following the Advanced Paediatric Life Support (APLS) guidelines.15 Following how the CRT is used in clinical practice, time values were categorised as normal (≤2 s), prolonged (>2–≤4 s) and severely prolonged (>4 s). SBI (eg, pneumonia, meningitis, urinary tract infection) were defined present based on abnormal radiographic findings and/or positive cultures from normally sterile locations or based on consensus diagnosis using clinical signs and symptoms and laboratory tests.13 For this purpose data were retrieved from the structured ED-specific record and thorough review of the electronic patient record. The radiographs were assessed by the local team of radiologists, non-blinded, similar to clinical practice. In a previous study of our research group two blinded radiologists made independent assessments of chest X-rays with a κ of 0.64 (95% CI 0.52 to 0.75).16 ,17 If final classification of the diagnosis could not be made clearly, a clinical consensus diagnosis was made by the investigators (RGN and RO).13 Standardised follow-up by telephone was conducted 72 hours after the visit in order to reduce the possibility of missed diagnoses of SBI.13

Data collection

Data concerning signs and symptoms were routinely recorded in a structured electronic record form that was integrated in the patient's digital file and vital signs were routinely measured at triage assessment, including heart and respiratory rate, peripheral oxygen saturation, temperature, pCRT and/or cCRT. Heart and respiratory rate were either measured by count or by standard monitoring devices. Temperature was measured rectally in children younger than 6 months and with auricular devices in older children. CRT was measured by estimate of seconds and classified in the above-mentioned three categories. Staff members were all experienced paediatric emergency nurses, attending resident paediatricians or senior paediatricians. All data on vital signs were collected by the nursing staff who all had received training in the APLS method. Additional diagnostic tests were conducted at the discretion of the treating physician.

Statistical analysis

The intermethod agreement between pCRT and cCRT was determined using Cohen's κ statistic to quantify the proportion of chance-corrected agreement between two measurement methods.18 Interpretation was based on a scale (poor: κ<0.00; slight: 0.00≤κ≤0.20; fair: 0.21≤κ≤0.40; moderate: 0.41≤κ≤0.60; substantial: 0.61≤κ≤0.80; almost perfect: κ>0.80), with values above 0.67 generally regarded as acceptable.19 ,20

Subgroup analyses for factors potentially associated with CRT agreement, that is, age, body temperature and the presence of SBI were performed. Strata for age were <1 years, 1–<5 years and ≥5 years and for body temperature <37.5°C, 37.5–38.4°C, 38.5–39.4°C and >39.5°C. As we had a low number of patients in the severely prolonged (>4 s) CRT group (table 1), further analyses were conducted with pCRT and cCRT dichotomised as normal (≤2 s) and abnormal (>2 s). The potential diagnostic value of dichotomous pCRT and cCRT for the detection of SBI was evaluated using logistic regression analysis.21 All analyses were conducted using SPSS PASW software (20.0; SPS, Chicago, Illinois, USA) and based on a two-sided level of α: 0.05.

Table 1

General characteristics of the included children with both pCRT and cCRT measured (CRT group) and excluded children missing either or both measures


Population characteristics

Of the entire cohort of 1993 children we included 1193 (59.9%) children with both measurements of pCRT and cCRT (CRT group). Patient characteristics for this group in comparison to the 800 children excluded for missing one or both CRT measurements were similar (table 1). Minor differences were observed in age, MTS urgency category, temperature and cCRT between the two groups and were considered clinically irrelevant. SBIs were present in 138 (11.8%): 112 (81%) with positive bacteriological cultures or infiltrate on chest radiograph, and 26 based on a combination of clinical signs and laboratory tests (consensus) (see online supplementary appendix).

Supplementary appendix

SUPPLEMENTAL TABLE 1. Specification of all SBI diagnoses with data for all of the relevant reference criteria. Values are numbers and (percentages) unless stated otherwise.

The intermethod agreement between pCRT and cCRT

Within the CRT group, pCRT was significantly more frequently and severely prolonged (χ2 test p<0.0001) than cCRT (table 2). A discrepancy was observed in 112 (9.4%) children with an abnormal pCRT (>2 s) together with a normal cCRT (≤2 s) with 16 (14.3%) of these having an SBI. Fourteen (1.2%) children had a normal pCRT together with an abnormal cCRT; however, none had an SBI.

Table 2

Cross table comparing peripheral capillary refill time (pCRT) and central capillary refill time (cCRT)

The overall agreement was κ: 0.35 (95% CI 0.27 to 0.43) (table 3), considered fair.19 Agreement stratified for age did not differ from the overall agreement. Between age groups, the agreement tended to be lower in children aged 1–<5 years (κ: 0.15; 95% CI 0.04 to 0.27) compared with younger children (<1 year, κ: 0.46; 95% CI 0.33 to 0.58) and older children (≥5 years, κ: 0.44; 95% CI 0.20 to 0.67). Stratified for body temperature, the agreement of pCRT and cCRT in the various temperature bands did not differ from the overall agreement. With higher body temperatures, the agreement tended to decrease consistently. Agreement between cCRT and pCRT in children with SBI (κ: 0.24; 95% CI 0.02 to 0.47) was similar to the agreement in all children.

Table 3

The intermethod agreement between dichotomous (normal/abnormal) peripheral and central CRT in the total study cohort and in subgroups

Diagnostic performance of pCRT/cCRT

Both pCRT and cCRT had no diagnostic value for the detection of SBI. For pCRT the OR for SBI was 1.10 (95% CI 0.65 to 1.84). cCRT showed an OR for an SBI of 0.43 (95% CI 0.13 to 1.39). In children admitted to the paediatric intensive care unit (ICU), (n=7) we observed an abnormal pCRT in 3 (42.9%) and an abnormal cCRT in 2 (28.6%) cases. In children diagnosed with sepsis, (n=5) abnormal pCRT and abnormal cCRT were observed in 1 (16.7%) child.


Main findings

Only a small part of febrile children at risk for serious infections at the ED show abnormal CRT values. The overall agreement between abnormal (>2 s) pCRT and abnormal (>2 s) cCRT assessed in a general population of febrile children was fair and neither significantly associated with categorisations of body temperature nor with age. The diagnostic values of either abnormal pCRT or cCRT for detecting the presence of SBI in this large febrile population were poor.

Comparison with existing literature

Previous studies have discussed the considerable variability of CRT assessment at the site of the fingertip or sternum, but had been performed in (small) specific populations or in children without significant illness.11 ,12 Our data confirmed this variability in febrile children as well. Studies determining the association of body temperature with CRT showed inconsistent results, with some confirming an association in neonates and adults. For example, one study among adults showing a decrease in CRT by 5% for every degree increase in body temperature.10 We could not prove our hypothesis that agreement between CRT at sternum and fingertip would be better in febrile children with higher body temperature.

Given the low agreement between dichotomised pCRT and cCRT observed, the question arises whether this results in different diagnostic values for SBI. Our study showed poor diagnostic performance for both dichotomised pCRT and cCRT to predict presence of SBI in a general cohort of febrile children. This is in contrast with a systematic review of 2010 reporting poor peripheral perfusion as a warning sign for serious infection. Four out of the six studies described in this review applied dichotomous measurement of CRT1 Two additional studies in children presenting with acute illness or fever to the ED showed only a weak OR of CRT for the presence of SBI (CRT measured categorically)3 and no significant association between CRT and meningococcal disease (CRT counted as integer seconds without stopwatch).8 Alternatively, a large cohort study in febrile children under 5 years did show a significant, though modest predictive value for SBI.4 In this study CRT was also measured categorically. We also observed low prevalence for abnormal CRT in a general population of febrile children: abnormal pCRT was observed in 13.8% and abnormal cCRT in 2.2% of cases (table 2). This observation is confirmed by a systematic review of 21 studies on the validity of the CRT measurement;12 ,22 17 of the 21 included studies reported CRT as a continuous measurement by stopwatch.12 As CRT represents poor perfusion, it may be expected late in the natural history of SBI, that is, when shock is developing. This may explain limited diagnostic value of CRT as a marker of SBI in a large general population of febrile children with a very low prevalence of shock/serious illness. Although low numbers hinder reliable conclusions from our own data in ICU-admitted patients or patients with sepsis, we indeed observed the same trend of pCRT being more frequently abnormal than cCRT. The hypothesis that CRT has stronger diagnostic value in more selected populations with a higher risk of SBI, such as more severely ill children or children suspected of sepsis, is acknowledged by others. In neonates suspected of sepsis, for example, a prolonged CRT had a higher sensitivity of 50% and specificity of 90% for late-onset bacterial infection.22

Strengths and limitations

One of the main strengths of this study was the large, prospective cohort of febrile children enrolled consecutively with a relatively wide range of ages that is representative of the general ED population. This contrasts with most studies evaluating CRT, which were mostly conducted in relatively selected populations.1 Data were collected in a standardised fashion by experienced ED nursing staff members, all of whom were trained in APLS methods and additionally trained according to the study protocol. One limitation was that pCRT and cCRT were not reported in all children, and hence could not be assessed for the entire cohort. However, a substantial cohort of children at risk for SBI remained for our study purposes. Comparison of the patient characteristics of included and excluded children in this study showed only minor differences, related to more frequent measurement of pCRT and cCRT in younger children and in more severely ill children. Higher frequency of measuring CRT in more severely ill patients is a common phenomenon and complies with clinical guidelines for management of febrile children.5 Interestingly, we did not observe a difference in SBI prevalence in both groups. Moreover, if more severely ill children were selectively included, this likely would have resulted in an overestimation of the diagnostic value of CRT. Second, CRT measurement was neither conducted in a strictly standardised way nor in a controlled environment. From previous research we know that extrinsic factors have been shown to affect CRT (eg, ambient/body temperature, pressing time).10 ,12 ,22 It was not feasible to control for these factors in a large cohort study integrated in routine clinical care. In addition, we did not observe significant influence of body temperature to the agreement of cCRT and pCRT. From a statistical point of view one might expect better results by evaluating continuous values of age and temperature. Under the assumption of linear associations between age or temperature and CRT agreement, we also analysed the impact of continuous age or temperature on the association between cCRT and pCRT by logistic regression analysis, but did not come to different conclusions.23

However, our categories for age are based on previous association studies between age and SBI.3 Given the large numbers in this study cohort, we believe that any significant association between body temperature, age and agreement that would merit changes in the method of measurement or interpretation would be shown, despite suboptimal modelling, that is, categorisation, of predictor variables. Similarly, measuring CRT categorically, rather than as a continuous value assessed by stopwatch, may have limited the model performance due to loss of information.24 However, the method used and the circumstances of CRT measurement in this study reflect clinical practice and thus show the clinical diagnostic value of CRT as it is commonly applied. The observed categorised CRT values in our study include the normal value range;11 observed abnormal CRT values were mostly within the ‘prolonged’ (2–≤4 s) category, and only in less than 1% of the population above 4 s (n=9). With the low variability of CRT with only 5–10% in abnormal ranges in a general population of febrile children we do not believe that continuous data on CRT would improve the diagnostic value. Our results showed no relationship between the two measures of CRT and SBI. Last, crucial to our findings, is the manner in which we coded SBI, more specifically the diagnosis of pneumonia. There is no good reference standard for this diagnosis, and it has been agreed upon that radiographic focal abnormalities in combination with clinical signs and symptoms are the best justifiable alternative for the diagnosis of definite bacterial infection.25–27 We do not think the absence of an association of CRT with SBI is due to our definition of SBI. Our definition coheres with previous studies3 ,28 conducted at this centre, as well as other studies.29 In addition, we strived to minimise potential misclassification by applying the reference criteria with both objective criteria, signs of systemic illness, consensus diagnosis by experienced clinicians and we included a follow-up period in our studies as a valid proxy for missing reference tests, as is recommended in literature.30 ,31


We observed only a fair agreement between dichotomised pCRT and cCRT in a large cohort of febrile children attending the ED. Categorisations of age and temperature were not associated with the agreement between pCRT and cCRT. Neither measure was found to be preferable over the other in the context of predicting the risk of SBI as an outcome in febrile children at the ED. Only a small part of febrile children at risk for serious infections at the ED show abnormal CRT values. The diagnostic performance of abnormal (>2 s) pCRT and abnormal (>2 s) cCRT for the presence of SBI in febrile children attending the ED was poor, with low sensitivity for SBI in particular.


The authors thank all of the emergency department staff members of the Erasmus MC-Sophia Children's Hospital for their participation and careful collection of the required data. The authors also thank paediatricians and researchers Mrs Dorien Geurts and Ms Joany Zachariasse for their contribution to the discussions and the revision of the manuscript.


View Abstract


  • EdV-K and TK contributed equally.

  • Twitter Follow Ruud Nijman at @rgnijman

  • Contributors EdV-K: supervised data collection, participated in recoding and data checking, participated in the data analysis, interpretation of the data, the discussion of the results, wrote the first draft of the manuscript and approved the final manuscript as submitted. TK: was involved in writing the protocol, in the data collection, conducted the data analysis and interpretation, the discussion of the results, contributed equally in writing the first draft of the manuscript and approved the final manuscript as submitted. YV: co-led the study design, supervised the data analysis at each successive step and contributed in the critical appraisal of the manuscript. She also approved the final manuscript as submitted. RO: was one of the initiators of the cohort study, led the study design, the development of the protocol, supervised data analysis at each step, participated in the discussion of the results, and co-wrote the manuscript. She also approved the final manuscript as submitted. RGN: substantially contributed to the conception and design of the study, supervised data collection, participated in recoding and data checking, and reviewed and revised the manuscript. He also approved the final manuscript as submitted. HAM: was one of the initiators of the cohort study, substantially contributed to the conception and design of the study, participated in the interpretation of the data, and reviewed and revised the manuscript. She participated in the discussion of the results and analyses and approved the final manuscript as submitted.

  • Funding EdV-K is supported by ZonMW, a Dutch organisation for health research and development.

  • Competing interests None declared.

  • Ethics approval The Institutional Review Board of Erasmus Medical Centre.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles