|Diagnosis and management of febrile infants (0-3 months)
|Hui C, Neto G, Tsertsvadze A, Yazdi F, Tricco AC, Tsouros S, Skidmore B, Daniel R
This review primarily assessed the accuracy of tests for diagnosing serious bacterial infections in children aged three months or younger with a fever. The main conclusion was that combined clinical and laboratory criteria had fairly high sensitivity and hence reliability in not missing cases of infection. This conclusion seems reliable, but the low specificity of these criteria should be considered.
The objectives that met the inclusion criteria for DARE were to review the evidence for the diagnostic accuracy of screening for serious bacterial illness and invasive herpes simplex virus in febrile infants, aged three months or younger, and to evaluate the benefits and harms of various management strategies.
PubMed, CINAHL, EMBASE, PsycINFO, EBM Reviews, Cochrane Central Register of Controlled Trials (CENTRAL) and Cochrane Database of Systematic Reviews were searched to September 2010. Search terms were reported. Websites of relevant organisations were searched for unpublished studies and experts in febrile infections in infants were contacted. Only studies reported in English were included.
Studies reporting on the diagnosis or management of febrile infants aged zero to three months, with no history of major disease that could predispose to fever, serious bacterial infection or herpes simplex virus infection, were eligible. Fever was defined as rectal temperature of 38°C or higher. Eligible settings were hospital emergency departments, out-patient office practices or acute care walk-in clinics. Studies conducted in North America, Australia, New Zealand, Western Europe, Scandinavia, Israel and four Asian countries (listed) were eligible. Eligible study designs were randomised or non-randomised controlled trials, case series, and cohort, case-control or cross-sectional studies.
The included studies evaluated the diagnostic accuracy of a wide range of clinical features, laboratory tests and scoring instruments (alone or in combination) for identifying serious bacterial illness (including bacterial meningitis, bacteraemia and urinary tract infection) in infants with a fever. Various tests were used as the reference standard. The management strategies that were evaluated were immediate antibacterial or antiviral therapy for infants at high risk, compared with delayed treatment until the diagnostic workup was complete.
Two reviewers independently selected studies for inclusion. Disagreements were resolved by consensus.
Assessment of study quality
The quality of the included studies was assessed using the 14-item Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool. This assessment was performed by two independent reviewers.
Data were extracted from the diagnostic accuracy studies to calculate the sensitivity, specificity and positive and negative predictive values, and their associated 95% confidence intervals. These data were extracted by two independent reviewers and checked by a third reviewer.
Methods of synthesis
The diagnostic accuracy parameters were pooled, using a DerSimonian and Laird random-effects model, where the studies applied the same criteria, in similar populations of infants, for predicting a specific type of infection. Heterogeneity across studies was examined graphically by plotting the values of sensitivity and specificity, and statistically using Ι² and Χ². Potential sources of heterogeneity that were specified in advance were age (zero to 28 days versus over 28 days), prevalence of infection, different index tests and different thresholds for each test.
Results of the review
Across the whole review, 84 studies (92 publications) were included. Participant numbers were reported for individual studies. The mean QUADAS score was nine (range three to 14).
Diagnostic accuracy: Sixty-two studies were included. Combined clinical and laboratory criteria (Rochester, Philadelphia, Boston and Milwaukee) had a similar overall accuracy in identifying patients with versus without serious bacterial infections (sensitivity 84.4 to 100%; specificity 26.6 to 69.0%; negative predictive value 93.7 to 100%; and positive predictive value 3.3 to 48.6%). The pooled sensitivity for the Rochester criteria was 94% (95% CI 91 to 96; Ι²=24.5%; seven studies) and specificity 49% (95% CI 47 to 51; Ι²=95.9%; seven studies). The pooled sensitivity for the Philadelphia protocol was 93% (95% CI 89 to 95; Ι²=75.5%; six studies) and specificity 46% (95% CI 44 to 48; Ι²=96.8%; six studies).The Boston, Philadelphia and Milwaukee criteria showed better overall accuracy in older infants than in neonates. The Rochester criteria were more accurate in neonates.
Clinical history alone (15 studies) showed higher sensitivity (94.0 to 95.4%), but lower specificity (11.3 to 33.2%) than criteria based on age, gender and degree of fever. Among laboratory criteria alone (27 studies), the overall accuracy (based on the area under the receiver operating characteristic curve) of C-reactive protein was greater than that of absolute neutrophil count, absolute band counts, serum white blood cells and calcitonin. The results for specific types of bacterial infection and those of studies using other diagnostic criteria were reported.
Four studies examined diagnostic accuracy for herpes simplex virus, but they provided insufficient data for any firm conclusions.
Management: Ten studies reported the benefits and harms of immediate antibiotic or antiviral therapy for infants at high risk of infection. There was no evidence that directly compared the outcomes for immediate versus delayed treatment.
Combined clinical and laboratory criteria showed fairly high sensitivity and hence reliability in not missing possible cases of serious bacterial infection. Clinical criteria or laboratory tests alone may have limited ability to rule out serious infection.
This complex review addressed a number of research questions. The inclusion criteria for diagnostic accuracy and management were broad, but generally clear. The search was thorough and included attempts to locate unpublished studies. Restricting the review to studies in English means that some relevant studies could have been missed. Appropriate methods were used to minimise errors and bias during the review process.
The quality of the diagnostic accuracy studies was assessed, using a standard tool. Relevant details of the included studies were reported. The synthesis was mainly narrative, which was appropriate in view of the wide range of tests included. The synthesis was complicated by a lack of standard definitions, limited follow-up data, heterogeneity and few studies for inclusion in meta-analyses.
The authors' conclusion about the sensitivity of combined clinical and laboratory criteria reflects the evidence presented and appears to be reliable, but the relatively low specificity of these criteria should be considered when interpreting the results.
Implications of the review for practice and research
Practice: The authors did not state any implications for practice.
Research: The authors stated that there was a need for research on the risks of testing and the management options for infants with fever. These studies should report separately on infants aged zero to 28 days, one to two months, and two to three months. They should examine the subtypes of serious bacterial infection and report fully on the harms of diagnostic and observational protocols.
Funded by the Agency for Healthcare Research and Quality, USA.
Hui C, Neto G, Tsertsvadze A, Yazdi F, Tricco AC, Tsouros S, Skidmore B, Daniel R. Diagnosis and management of febrile infants (0-3 months) Rockville, MD, USA: Agency for Healthcare Research and Quality. Evidence Report/Technology Assessment; 205. 2012
Subject indexing assigned by CRD
Bacteremia; Drug Therapy; Fever; Herpes Simplex; Humans; Infant; Mass Screening; Sensitivity and Specificity
Date bibliographic record published
Date abstract record published
This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.