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

Is a normal CRP sensitive enough to rule out septic arthritis

Three Part Question

In [adults presenting to the ED with an acute hot joint] is a [normal CRP] sensitive enough [to rule out septic arthritis]

Clinical Scenario

A 48 year old with a presents with a one day history of red, swollen right knee. On examination she is febrile, has decreased range of movement and a left knee effusion. You order bloods and the CRP comes back within normal range. You wonder what benefit CRP gives in ruling out SA?

Search Strategy

Multi-file search using OVID – Medline (1950-2008), Embase (1980-2008), CINHAL (1982-2008), Cochrane ({[exp Infective Arthritis OR hot OR septic OR septic] AND [ exp Blood Sedimentation OR exp C-Reactive Protein OR exp laboratory techniques and procedures OR OR OR inflammatory]}) Limited to Humans, English Language and Adults
The following number of articles were identified from each of the databases: Medline 118 citations, Embase 125 citations, CINHAL 6 citations, Cochrane 2 citations.

Search Outcome

Duplicates and irrelevant titles removed from 251 articles leaving a total of three relevant articles.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Gupta et al,
75 Patients with adult-onset SA (Newman A classification),* 46 with primary joint disease and 26 with RAProspective cohort study (level 2)CRP (g/l)Elevated in 98% only one patient had a normal CRP all patients with SA (75) median (range) 185 (6–440) IQR 101–256. RA patients with SA (n = 25) median (range) 184 (16–422) IQR 137–233. Non-RA patients with SA (n = 50) 191 (6–440) IQR 94–281Imperfect gold standard.* Small sample. No sample size calculation. Notable differences between compared groups
Soderquist, et al,
The Netherlands
54 Patients with culture-verified SA of which 34 patients had verified crystal-associated arthritisRetrospective case–control series (level 2)CRP (mg/l)SA patients (n = 50) median (range) 182 (<7–405) crystal-associated arthritis (n = 32) median (range) 101 (22–353). SA patients showed a significantly higher CRP than patients with crystal-associated arthritis (p = 0.002) (Mann–Whitney U test)Small sample. No sample size calculation. Larger proportion of SA patients than crystal-associated arthritis. Collection bias. No record of the time at which ESR was taken
Margaretten et al,
Fourteen studies involving 6242 patients, of whom 653 met gold standard for diagnosis of SA and met all respective inclusion criteriaSystematic review (level 1)FindingsRetrospective case–control series CRP >100 mg. Sensitivity 77 specificity 53 LR(+) 1.6 (1.1–2.5) LR(–) 0.44 (0.24–0.82)1Imperfect gold standard*
Quality of study/level of evidenceLow according to the study’s predetermined criteria for the quality of evidence in primary studies
Studies found8


The evidence is limited. Most of the literature comes from a paediatric population and is most often retrospective. The best evidence, a prospective cohort of 75 patients, in which 98% have a raised CRP, shows promise for the value of CRP in potentially ruling out septic arthritis (Gupta). In contrast, a lower quality study of 54 patients with culture verified septic arthritis, 34 of which had verified crystal-associated arthritis, had the sensitivity and specificity of CRP >100 mg in septic arthritis determined in a recent systematic review that were not sufficient to rule out septic arthritis (Soderquist). Although the serum CRP level usually increases 100 to 1000-fold during an acute bacterial infection, the patient group examined in this study and the absence of a value of CRP given the prospective study make it difficult to weigh up the evidence homogenously (Margaretten). In conclusion, CRP as with other inflammatory indices should be regarded as an adjunct in the diagnosis of the acute hot joint and synovial fluid and blood culture analysis should be performed in the case of a high suspicion of septic arthritis.

*Imperfect gold standard—Gram stain (sensitivity 29–50%) and culture (sensitivity 82%).2 CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; IQR, interquartile range; RA, rheumatoid arthritis; SA, septic arthritis.

Clinical Bottom Line

Although CRP shows promise, there is insufficient evidence for its sensitivity to be high enough to rule out septic arthritis.


  1. Gupta MN, Sturrock RD, Field M. Prospective comparative study of patients with culture proven and high suspicion of adult onset septic arthritis. Ann Rheumatol Dis 2003;62:327–31.
  2. Soderquist B, Jones I, Fredlund H, Vikerfors T. Bacterial or crystal-associated arthritis? Discriminating ability of serum inflammatory markers. Scand J Infect Dis 1998;30:591–6.
  3. Margaretten ME, Kohlwes J, Moore D, Bent S. Does this adult patient have septic arthritis? JAMA 2007;297:1478–88.