Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

The Negative Predictive Value of Absence of Fever in Excluding Septic Arthritis in Children

Three Part Question

In [an afebrile child with an acutely painful joint not caused by trauma] Is [the absence of fever sufficient evidence] to [safely exclude the diagnosis of septic arthritis]

Clinical Scenario

A 4 year old child present to the childrens area of the accident and emergency department with a painful tender hip joint. There is no history of trauma. His observations are normal and he has a temperature of 36.8C He has a normal xray and has has an ultrasound of the hip which shows a small effusion. He is awaiting blood tests. The mother feels blood tests are unnecessary. She asks you how necessary the blood tests are.

Search Strategy

Medline search via PubMed
Systematic process of search and review.
Medline search terms accepted for title review if <200 matched items.
Title review;
removed duplicate items and eliminated irrelevant titles and those not in English
Abstract Review;
eliminated publications with the wrong outcome, patient group, and those reporting single cases or series of fewer than 10 cases
Full Text Review;
Attempt to extract outcome data
Septic Arthritis 11808
"Septic Arthritis" 3017
"Septic Arthritis" children 782
"Septic Arthritis" children temperature 25 Included
"Septic Arthritis" paediatric 26 Included
"Septic Arthritis" pediatric 123 Included
pyarthrosis children 1858
pyarthrosis children temperature 30 Included

Systematic reviewing
Medline Search 204
Title review 18
Abstract Review 12
Paper review 8

Search Outcome

Systematic searching
Medline Search 204
Title review 18
Abstract Review 12
Paper review 8

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Sung Taek Jung et al
May 2003
Children with a diagnosis of either Transient Synovitis or Septic ArthritisRetrospective case control study comparing findings and results between 97 patients with transient synovitis and 27 with septic arthritisNumber of cases with Septic Arthritis and fever (temp <37.0C)21 out of 27 NPV no fever = 77.8%Single centre Retrospective not prospective Case control therefore selection bias No subgroup analysis looking separately at neonates
Michelle S. Caird et al
June 2006
Children who underwent hip aspiration because of a suspicion of septic arthritisProspective case conrtol study collected data on 53 children of which 34 had septic arthritisNumber of patients with septic arthritis who had a fever (defined in this study as temperature >38.5)15 out of 34 NPV for temp <38.5 = 44%Differing definition of fever to most studies Unclear as to criteria for diagnosis SA. States half confirmed half presumed. Only those who had hip aspiration included Small series Single centre
Ilkka Kunnamo et al
January 1987
Children seen with a presenting complaint of arthritis in a childrens hospital having been first seen and referred by primary care doctors in an 18 month period starting in May 1982Uni and Multivatiate analysis was performed on various clinical and laboritory findings for a total of 7 diagnostic groups including Septic Arthritis Number of Patients with Septic Arthritis who had a fever (defined in this study as temerature > 38.5C14 out of 18 NPV No fever = 77.8%Small number of cases with Septic Arthritis Dated Retrospective
G. F. Eich
November 1999
Children presenting with irritable hip to Major University Childrens Hospital between 1988 and 1992Retrospective case control study of 89 cases of irritable hip using uni and multivariate analysis to differentiate between cases with Transient Synovitis, Septic Arthritis and Perthes.Number of Patients with Septic Arthritis who had fever7 out of 8 NPV no fever 87.5%Small number of patients with septic arthritis No definition of fever given Retrospective Case control suffers from selection bias Single centre
David M. Klein; et al
Children aged 0 - 6 with confirmed diagnoses of septic arthritisRetrospective case control study of 26 children with septic arthritisNumber of patients with Septic Arthritis who had a temperature >37.021 out of 26 NPV normal temperature = 80.8%Small series n = 26 Retrospective Case control studies limited by selection bias Single centre
Mininder S. Kocher
August 2004
Children presenting to a Childrens' hospital with acutely irritable hip between 1997 and 2002Prospective case control study of 51 children with Septic Arthritis and 103 children with Transient Synovitis. The aim was to test the validity of a clinical prediction rule developed in the earlier paper by the same author.Patients with Septic Arthritis with a 23 out of 24 NPV no fever = 95.8%"History of Fever" not defined clearly Same investigators validating their own scoring system invites accusation of bias
Mininder S Kocher
December 1999
All children with acutely irritable hips seen at a major tertiary care children's hospital between 1979 and 1996Retrospective case control study. 282 cases of which 82 had a diagnosis of Septic Arthritis. 86 had Transient synovitis and 114 cases were excluded as being patients from atypical groups. The SA and TS groups were comparent on univariate and multivariate analysis with the purpose of drawing up a system for scoring the likelyhood of a diagnosis of SA Patients with Septic Arthritis with a fever67 out of 82 NPV no feverNo clear definition of term "history of fever" Varying criteria used to diagnose SA Large number of patients excluded from analysis Retrospective Case control risks selection bias Single centre


Quite a wide variety of results between investigators. Some studies differed in their definition of fever making direct comparison difficult. The largest study had the highest number and proportion of patients with septic arthritis without fever. This study however defined fever at a quite high temperature (38.5C) and may have included patients with unproven Septic Arthritis in the Disease group. Ignoring those issues the raw total number of patients with Septic Arthritis included in all the reviewed studies was 219 of which 168 had fever of some kind. This gives an overal negative predictive value of absence of fever of 76.7% If you exclude the outlying study this rises to 82.7% All the studies reviewed concluded that a combination of clinical findings, imaging and blood inflamatory markers were necessary to exclude septic arthritis.

Editor Comment


Clinical Bottom Line

Using lack of fever alone to exclude septic arthritis in children is by my calculations at best around 80% safe. Given the serious risks associated with the diagnosis of septic arthritis a panel of tests including inflamatory markers is necessary to avoid missing cases with disastrous results.

Level of Evidence

Level 2 - Studies considered were neither 1 or 3.


  1. Sung Taek Jung Significance of Laboritory and Radiologic Findings for Differentiating Between Septic Arthritis and Transient Synovitis of the Hip Lippincott Williams and Wilkins; Journal of Pediatric Orthopaedics May-June 2003;23(3):368-72
  2. Michelle S. Caird Factors Distinguishing Septic Arthritis From Transient Synovitis of the Hip in Children a Prospective Study The Journal of Bone anfd Joint Surgery (America) June 2006; 88(6):1251-7
  3. Ilkka Kunnamo et al Clinical Signs and Laboritory Tests in the Differential diagnosis of Arthritis in Children American Journal Of Diseases Of Children January 1987;141(1):34-40
  4. G F Eich et al The painful hip: evaluation of criteria for clinical decision-making Springer-Verlag; European Journal of Pediatrics November 1999 158: 923-928
  5. David M. Klein Sensetivity of Objective Parameters in Diagnosis of Pediatric Septic Hips Lippincott-Raven Publishers 1997 May;(338):153-9
  6. Mininder S Kocher Validation of a Clinical Prediction Rule for the Differentiation Between Septic Arthritis and Transient Synovitis of the hip in Children Journal of Bone and Joint Surgery August 2004; 86(8):1629-35
  7. Mininder S Kocher Differentiating Between Septic Arthritis and Transient Synovitis of the Hip in Children: An Evidence-Based Clinical Prediction Algorithm Jounal of Bone and Joint Surgery December 1999; 81(12):1662-70