diagnosing joint infections: synovial fluid differential is more sensitive than white blood cell...
TRANSCRIPT
ORIGINAL ARTICLE
Diagnosing joint infections: synovial fluid differential is moresensitive than white blood cell count
Sean Baran • Connie Price • David J. Hak
Received: 10 July 2013 / Accepted: 29 September 2013
� Springer-Verlag France 2013
Abstract In order to identify the predictive value of
synovial fluid white blood cell (WBC) count and differ-
ential white blood cell count in identifying nonprosthetic
joint infection in immunocompetent and immunosup-
pressed populations, we retrospectively reviewed 96 adult
patients who underwent hip or knee aspiration because of
symptoms suggesting a possible nonprosthetic joint infec-
tion. Medical history, including immunosuppressive dis-
ease or drugs, was recorded, and synovial fluid cell count,
differential, and culture results were compared. There were
44 patients with positive synovial cultures. Of 36 patients
who had a synovial WBC C50,000/mm3, 89 % had posi-
tive cultures. The sensitivity to synovial WBC C50,000/
mm3 was 0.727 (95 % CI 0.570–0.845), and specificity was
0.923 (95 % CI 0.806–0.975). There were 12 patients with
a synovial WBC \50,000/mm3 that had positive cultures.
The sensitivity of percentage polymorphonuclear cells
(%PMNs) to predict positive cultures when the %PMNs
were at least 80, 85, and 90 % was 0.932, 0.886, and 0.818,
respectively. The specificity when the %PMNs was at least
80, 85, and 90 % was 0.598, 0.577, and 0.673, respectively.
Among the 29 % of immunocompromised patients, the
sensitivity to synovial WBC C50,000/mm3 was 0.714
(95 % CI 0.420–0.904), and specificity was 1.000 (95 % CI
0.732–1.000). Twenty-nine percent of patients with a
synovial WBC \50,000/mm3 had positive cultures. The
sensitivity of %PMNs to predict positive cultures when the
%PMNs was at least 80, 85, and 90 % was 1.000, 0.929,
and 0.786, respectively. The specificity when the %PMNs
were at least 80, 85, and 90 % was 0.500, 0.643, and 0.714,
respectively. We found that the synovial WBC differential
(percentage synovial fluid PMNs) is a more sensitive pre-
dictor for nonprosthetic adult joint infection than the
synovial absolute WBC count. This was true in both the
general population and the immunosuppressed population.
Keywords Septic arthritis � Joint infection �Diagnosis
Introduction
Septic arthritis is a potentially devastating condition,
associated with significant morbidity and even mortality,
which requires emergent diagnosis and treatment [1].
Prompt diagnosis is imperative for initiating appropriate
treatment. Synovial fluid aspiration and analysis has long
been recognized and recommended for its usefulness in the
evaluation of the acutely tender, red, warm, and swollen
joint. The sophistication of individual assays in synovial
fluid analysis has grown immensely since Bauer and
Our local IRB reviewed and approved this study.
This study was performed at Denver Health Medical Center, Denver,
Colorado.
S. Baran
Department of Orthopaedic Surgery, Mayo Clinic,
200 First Street SW, Rochester, MN 55905, USA
e-mail: [email protected]
C. Price
Department of Internal Medicine, Denver Health Medical
Center, University of Colorado Denver, School of Medicine,
777 Bannock St, MC0188, Denver, CO 80204, USA
e-mail: [email protected]
D. J. Hak (&)
Department of Orthopaedic Surgery, Denver Health Medical
Center, University of Colorado Denver, School of Medicine,
777 Bannock St, MC0188, Denver, CO 80204, USA
e-mail: [email protected]
123
Eur J Orthop Surg Traumatol
DOI 10.1007/s00590-013-1331-x
Ropes’ early reports that variation in cell content and gross
appearance may be indicative of different arthritic diseases
[2]. Despite this growing sophistication, no single synovial
fluid test has been proven adequately discriminatory in the
diagnosis of the septic joint [3]. Even the synovial fluid
white blood cell (WBC) count, which has traditionally been
held as a useful marker of joint infection, displays a tre-
mendous amount of overlap with the other inflammatory
arthritides [4, 5]. Multiple studies have concluded that
synovial fluid WBC count is insufficient to rule in or out
septic arthritis [3–7]. Some research suggests that the
synovial fluid white blood cell differential (dWBC) count
may be a better marker of joint infection, but the amount of
evidence is scant, and the conclusions are mixed [5, 8–12].
Unfortunately, the recognized ‘‘gold standard’’ for the
diagnosis of septic arthritis remains the level of clinical
suspicion of a physician experienced in the diagnosis of
joint pathology [3, 4].
This situation presents a serious problem for the physi-
cian lacking sufficient experience in the diagnosis of joint
pathology and, most importantly, for the patient with a
potentially septic joint. The diagnosis is even further con-
founded in the immunosuppressed population, as their
synovial fluid WBC values have been shown to have an
even greater degree of overlap with the other inflammatory
arthritides [8, 12, 13].
The purpose of this study was to examine the predictive
value of the synovial fluid WBC and dWBC in the diag-
nosis of an adult septic joint in patients who underwent
synovial fluid aspiration at a major tertiary care center. Our
hypothesis was that dWBC would be a more sensitive, and
possibly more specific, marker of a septic joint than the
absolute synovial WBC count.
Patients and methods
This retrospective case–control study was conducted to
determine the sensitivity and specificity of synovial WBC
count and dWBC count to diagnose septic arthritis. The
institutional review board at our institution approved this
study.
The study population includes all patients who under-
went synovial fluid aspiration and analysis with cell count
of the hip or knee between May 2003 and May 2009 at an
academic, urban, tertiary care center which has greater than
900,000 patient contacts annually.
The hospital database for infectious diseases was que-
ried for all synovial fluid aspiration results between May
2003 and May 2009 and initially sorted based on the
synovial fluid culture results. The electronic medical
records associated with each culture-positive aspiration
were then reviewed and abstracted for demographics,
medical history, including immunosuppressive medication
or disease, joint sampled, synovial fluid culture results, and
synovial fluid analysis results. Only those cases in which
the patient had undergone aspiration of the hip or knee
were included. Cases were excluded if the patient was less
than 18 years of age, had been an inmate in the county jail
at the time of aspiration, or if the synovial fluid analysis
had been done at an outside facility. For those patients who
had undergone multiple aspirations during a single hospital
course or within 1 month of an initial aspiration, only the
initial synovial fluid analysis was included. Culture results
were reviewed to evaluate for possible contaminants, and
those that were listed as contaminants (i.e., broth-only
growth) were excluded. Finally, aspirations from prosthetic
joints were excluded because the number of infected total
joint arthroplasties was too low to analyze independently.
A cohort of 44 consecutive cases in which the patient
had culture-confirmed septic arthritis (i.e., growth of a
pathologic organism) were identified, and subsequently, a
cohort of 52 consecutive patients with negative culture
results during the same time period were established. The
same information that was abstracted for the culture-posi-
tive aspirations was abstracted from associated medical
records of the culture-negative cases. The same inclusion/
exclusion criteria were applied to these cases. In addition,
the case was excluded if the joint was treated as septic
based on clinical criteria, or if the negative culture was
obtained after repeat aspiration during the course of the
treatment of a previously diagnosed septic joint.
Statistical analysis
Sensitivity and specificity were reported with 95 % confi-
dence intervals (CI).
Results
The search of the database identified 44 consecutive
patients with culture-positive synovial fluid aspirates in the
6-year study. From the same time period, 52 consecutive
patients with culture-negative synovial fluid aspirates were
identified for comparison, yielding a total of 96 cases for
study. Patient demographics are summarized in Table 1.
The median age was 47 years with a range of 21–85 years.
Women comprised 39 % of the study population.
The knee (96 %) was more commonly involved than the
hip (Table 1). The most common organism (Table 2)
identified was Staphylococcus aureus (59 %) followed by
beta-hemolytic streptococcus (25 %).
In 36 of the 96 patients, the synovial WBC was
C50,000/mm3. Of these 36 patients, 32 (89 %) had positive
culture results. Of 60 patients with a synovial WBC
Eur J Orthop Surg Traumatol
123
\50,000/mm3, 12 (20 %) had positive culture results. The
sensitivity to synovial WBC C50,000/mm3 was 0.727
(95 % CI 0.570–0.845), and specificity was 0.923 (95 % CI
0.806–0.975) (Fig. 1).
Forty-one (93 %) of 44 patients with culture-positive
aspirates had a dWBC count with a %PMNs of at least
80 %. The sensitivity of %PMNs to predict positive cul-
tures when the %PMNs were at least 80, 85, and 90 % was
0.932 (95 % CI 0.803–0.982), 0.886 (95 % CI
0.746–0.957), and 0.818 (95 % CI 0.668–0.913), respec-
tively. Of 56 patients with culture-negative aspirates, 28
(85 %) had a dWBC count with %PMNs less than 80 %.
The specificity when the percentage PMNs was at least 80,
85, and 90 % was 0.538 (95 % CI 0.396–0.675), 0.577
(95 % CI 0.433–0.710), and 0.673 (95 % CI 0.528–0.793),
respectively.
Patients with immunosuppression
Immunosuppression (disease and/or medication) was pres-
ent in 28 patients (Tables 3, 4). Seventeen (61 %) of 28
patients had a known immunocompromising disease, and
nine additional patients were taking an immunosuppressive
medication. The remaining two patients had a known diag-
nosis of an immunocompromising disease and were also
taking an immunosuppressive medication. Seven patients
0
0.2
0.4
0.6
0.8
1
WBC > 50k 80% PMNs 85% PMNs 90% PMNs
Sensitivity Specificity
Fig. 1 Solid bars represent sensitivity values, and white bars
represent specificity values for WBC C50,000/mm3 (WBC), PMNs
C80 % (80), PMNs C85 % (85), and PMNs C90 % (90) for native
joints. Positive and negative error bars indicate 95 % CI
Table 3 Immunosuppressive diseases observed
Disease Number of patients affected
Type 1 diabetes mellitus 3
Type 2 diabetes mellitus 13
Human immunodeficiency virus 2
End-stage renal disease 4 (3 had comorbid DM)
Table 4 Immunosuppressive medications observed
Medication Number of patients taking
Prednisone 7
Hydroxychloroquine 7
Methotrexate 3
Sulfasalazine 3
Loxapine 1
Table 5 Isolated pathogens in immunosuppressed population
Organism Number of
patients
Percentage of patients
(%)
Staphylococcus aureus 9 64
MRSA 2 14
MSSA 7 50
Beta-hemolytic
streptococcus
3 21
Escherichia coli 1 7
Haemophilus influenzae 1 7
Table 1 Patient demographics
Median age (years) (range) 47 (21–85)
Sex (female) 39 %
Joint involved
Knee 96 %
Hip 4 %
Immunosuppressiona 30 %
a Includes patients with an immunosuppressive disease and/or on an
immunosuppressive medication
Table 2 Isolated pathogens
Organism Number of
patients
Percentage of
patients (%)
Staphylococcus aureus 26 59
MRSA 9 20
MSSA 17 39
Beta-hemolytic streptococcus 11 25
Streptococcus pneumoniae 1 2
Streptococcus milleri 1 2
Haemophilus influenzae 1 2
Bacteroides fragilis 1 2
Escherichia coli 1 2
Candida spp. 2 5
Eur J Orthop Surg Traumatol
123
were on greater than one immunosuppressive medication,
and the most common immunosuppressive medication was
prednisone. The most common immunocompromising dis-
ease was type 2 diabetes mellitus. Culture-positive aspirates
were present in 14 (50 %) of the 28 immunosuppressed
patients. The most common organism was S. aureus (64 %),
followed by beta-hemolytic strep (21 %) (Table 5).
There were nine immunosuppressed patients with a
synovial WBC C50,000/mm3, and 100 % of these patients
had culture-positive aspirates. Of 19 patients with a synovial
WBC\50,000/mm3, five (26 %) had positive culture results.
The sensitivity in immunosuppressed patients to synovial
WBC C50,000/mm3 was 0.714 (95 % CI 0.420–0.904), and
specificity was 1.000 (95 % CI 0.732–1.000) (Fig. 2).
Immunosuppressed patients with culture-positive aspi-
rates had a dWBC count with %PMNs of at least 80 % in
every case. The sensitivity of %PMNs to predict positive
cultures when the %PMNs were at least 80, 85, and 90 % was
1.000 (95 % CI 0.732–1.000), 0.929 (95 % CI 0.642–0.996),
and 0.786 (95 % CI 0.488–0.943), respectively. Seven of 14
immunosuppressed patients with culture-negative aspirates
had a dWBC count with %PMNs less than 80 %. The
specificity when the %PMNs were at least 80, 85, and 90 %
was 0.500 (95 % CI 0.240–0.760), 0.642 (95 % CI
0.356–0.860), and 0.714 (95 % CI 0.420–0.904), respec-
tively. These data are summarized in Table 2.
Discussion
Prompt diagnosis and treatment of intra-articular infections
is imperative to limit adverse outcomes [1]. Following
septic arthritis, permanently reduced joint function is
observed in around 40 % [1]. Synovial fluid white blood
cell (WBC) count has traditionally been used to diagnose
joint infection. The American Rheumatism Association has
classified WBC into three categories: \2,000 WBC/mm3
(noninflammatory), 2,000–50,000 WBC/mm3 (inflamma-
tory), and [50,000 WBC/mm3 (infectious) [8]. Impor-
tantly, these cutoffs are merely guidelines, since there is a
significant amount of overlap in the observed synovial
WBC values between infectious arthritis and various
inflammatory arthritides [4, 5]. Multiple studies have
examined the synovial WBC value as a predictive criterion
for diagnosing septic arthritis. While various sensitivities
and specificities for synovial absolute WBC values have
been reported, each of these studies has concluded that
synovial WBC alone is insufficient to diagnose or rule out
joint infection [3–7].
We found sensitivity and specificity values for the WBC
C50,000/mm3 of 0.727 and 0.923, respectively. These
values are consistent with previously reported values in the
literature. Importantly, using a 50,000/mm3 cutoff for
infectious arthritis may have resulted in delayed diagnosis
of 20 % of the septic joints observed in this study. These
results further support the need for a better marker of joint
infection.
We found that the dWBC, specifically the %PMNs, is a
more sensitive marker of joint infection than the synovial
fluid WBC. In our series, a %PMNs value of at least 80 %
had a sensitivity to predicting joint infection of 0.932,
which was a 20 % increase in sensitivity compared with
the synovial WBC value. In contrast to using the standard
cutoff of WBC C50,000/mm3, which may have resulted in
delayed diagnosis in up to 20 % of the observed cases of
joint infection, using a cutoff of %PMNs greater than 80 %
would have resulted in only 7 % of cases experiencing
delayed diagnosis. We found that the %PMNs represent a
more sensitive marker for joint infection than synovial
absolute WBC count and is therefore a more useful clinical
value in ruling out joint infection. When the %PMNs are
less than 80 %, it is unlikely that the etiology of the joint
inflammation is due to infection. However, in our series,
we found that the specificity of %PMNs remained low, and
therefore, it is not useful as a marker to rule in the diag-
nosis of a septic joint.
Evidence regarding the use of the dWBC in identifying
an infectious etiology of arthritis is limited, and the con-
clusions in the literature are mixed. McCutchan and Fisher
reported that the dWBC is of minimal value as it is not
consistently different in aspirates from infectious arthritis
versus inflammatory arthritis [8]. Coutlakis et al. [5]
commented in a 2002 study that the percentage of neu-
trophils tends to be higher in fluids with higher total white
blood cell counts. However, the article contains no further
0
0.2
0.4
0.6
0.8
1
WBC > 50k 80% PMNs 85% PMNs 90% PMNs
Sensitivity Specificity
Fig. 2 Solid bars represent sensitivity values, and white bars
represent specificity values for WBC C50,000/mm3 (WBC), PMNs
C80 % (80), PMNs C85 % (85), and PMNs C90 % (90) for native
joints in immunosuppressed patients. Positive and negative error bars
indicate 95 % CI
Eur J Orthop Surg Traumatol
123
analysis or discussion regarding this topic. Three older
studies propose that increased PMNs above 90 % sug-
gested septic arthritis, and in a 2007 review of the litera-
ture, Margaretten et al. [7, 9–11] reported that when the
percentage of PMNs is at least 90 %, there is an increased
likelihood ratio of the diagnosis.
Further confounding the diagnosis of septic arthritis is the
presence of immunosuppression. Patients with septic
arthritis who are taking immunosuppressive medications
tend to have WBC values in the inflammatory range (i.e.,
2,000–50,000 WBC/mm3) instead of the infectious range
[8]. McGillicuddy et al. [7] reported that there was no sta-
tistically significant relationship between WBC levels and
immunosuppression in culture-positive synovial fluid aspi-
rates. The literature regarding septic arthritis among patients
with an immunosuppressive disease is largely based upon
those with HIV infection, as HIV has been recognized to
significantly increase the probability of septic arthritis [12].
However, most of the evidence supporting this claim has
come from small studies and case reports. Zalavras et al. [13]
did find that patients with HIV infection in whom the serum
CD4 count was less than 200/mm3 tended to have WBC
values in the inflammatory range.
Using WBC C50,000/mm3 as a clinical marker for joint
infection, we found a sensitivity of 0.714 in the immuno-
suppressed population compared with a sensitivity of 0.727
in the overall population. The proportion of joint infections
in which the diagnosis would have been delayed using the
WBC cutoff rises to an unacceptable 36 % in the immu-
nosuppressed population, underscoring the necessity of a
more sensitive clinical marker. Our study demonstrates a
dramatically improved sensitivity for predicting joint
infection when using the %PMNs as a clinical marker. In
this study, a cutoff of 80 % PMNs resulted in early
detection of joint infection in 100 % of the observed
immunosuppressed patients. Interestingly, the specificity of
WBC C50,000/mm3 is 100 % in this study among immu-
nosuppressed patients. While this finding is potentially
useful to rule in the diagnosis, the more clinically useful
finding for early diagnosis is the increased sensitivity of the
dWBC.
Based on the results of this study, we believe clinicians
should use the dWBC value to increase early suspicion for
the potentially septic joint, especially in the immunosup-
pressed patient. Importantly, we do not advocate singular
reliance upon the dWBC. Rather, the clinician should
continue to rely upon the entire clinical picture to arrive at
the proper diagnosis. However, with the observed increase
in sensitivity of the dWBC over the WBC, it is likely that
joint infection will represent a greater possibility in the
early differential diagnosis of red, warm, tender, swollen
joint. Further, by maintaining this increased suspicion in
the face of a ‘‘noninfectious’’ WBC value, the physician
will potentially miss fewer cases of joint infection early in
their course.
Staphylococcus aureus is the most common etiologic
organism in septic arthritis. McGillicuddy et al. found that
the organism accounted for 55 % of the cases in their
49-patient study [8]. Li et al. [14] found an even higher
incidence, with 76 % of cases caused by S. aureus, in a
study of 55 patients. Similarly, Kaandorp et al. [15] found
that in a population of 157 patients with septic arthritis, S.
aureus was causative in 44 % of cases. Our results showing
64 % of positive cultures growing S. aureus are consistent
with these authors’ previous findings. This proportion
remained relatively constant regardless of the presence of
immunosuppression.
Study limitations
This study has several limitations. First, it is a hospital-
based retrospective, case–control study, and therefore,
the data utilized were originally recorded for the pur-
poses other than research, and the accuracy of these
recordings cannot be guaranteed. The study population is
limited to patients seen at a single major, urban, tertiary
care center. However, the records collected were not
limited only to hospital admissions, but rather included
all patients (both inpatient and outpatient) who under-
went synovial fluid analysis, which eliminates Berkson’s
bias from our study. Second, the population size is rel-
atively small. Third, the data collected only included
hips and knees, which may possibly limit the general-
izability of our findings to other joints. However, we did
choose to include only hips and knees because these are
the only two joints from which an adequate amount of
synovial fluid can routinely be obtained for differential
cell count analysis. Finally, our study does not address
infection in prosthetic joints, a subset of the population
known to experience higher rates of joint infection than
the general population.
Conflict of interest The authors report no conflicts of interest
regarding this manuscript. None of the authors, nor their institutions,
received any payment or services from a third party for any aspect of
the submitted work.
References
1. Shirtliff ME, Mader JT (2002) Acute septic arthritis. Clin
Microbiol Rev 15:527–544
2. Swan A, Amer H, Dieppe P (2002) The value of synovial fluid
assays in the diagnosis of joint disease: a literature survey. Ann
Rheum Dis 61:493–498
3. Mathews CJ, Kinglsey G, Field M, Jones A, Weston VC, Phillips
M, Walker D, Coakley G (2008) Management of septic arthritis:
a systematic review. Postgrad Med J 84:265–270
Eur J Orthop Surg Traumatol
123
4. Mathews CJ, Coakley G (2008) Septic arthritis: current diagnostic
and therapeutic algorithm. Curr Opin Rheumatol 20:457–462
5. Coutlakis PJ, Roberts WN, Wise CM (2002) Another look at
synovial fluid leukocytosis and infection. J Clin Rheumtol 8:67–71
6. Li SF, Cassidy C, Chang C, Gharib S, Torres J (2007) Diagnostic
utility of laboratory tests in septic arthritis. Emerg Med J 24:75–77
7. McGillicuddy DC, Shah KH, Friedberg RP, Nathanson LA, Ed-
low JA (2007) How sensitive is the synovial fluid white blood cell
count in diagnosing septic arthritis? Am J Emerg Med
25:749–752
8. McCutchan HJ, Fisher RC (1990) Synovial leukocytosis in
infectious arthritis. Clin Orthop Relat Res 257:226–230
9. Krey PR, Bailen DA (1979) Synovial fluid leukocytosis: a study
of extremes. Am J Med 67:436–442
10. Shmerling RH, Delbanco TL, Tosteson AN, Trentham DE (1990)
Synovial fluid tests: what should be ordered? JAMA 264:1009–1014
11. Kortekangas P, Aro HT, Tuominen J, Toivanen A (1992) Syno-
vial fluid leukocytosis in bacterial arthritis vs. reactive arthritis
and rheumatoid arthritis in the adult knee. Scand J Rheumatol
21:283–288
12. Margaretten ME, Kohlwes J, Moore D, Bent S (2007) Does this
adult patient have septic arthritis? JAMA 297:1478–1488
13. Zalavras CG, Dellamaggiora R, Patzakis MJ, Bava E, Holtom PD
(2006) Septic arthritis in patients with human immunodeficiency
virus. Clin Orthop Relat Res 451:46–49
14. Li SF, Henderson J, Dickman E, Darzynkiewicz R (2004) Lab-
oratory tests in adults with monoarticular arthritis: can they rule
out a septic joint? Acad Emerg Med 11:276–280
15. Kaandorp CJ, Dinant HJ, van de Laar MA, Moens HJ, Prins AP,
Dijkmans BA (1997) Incidence and sources of native and pros-
thetic joint infection: a community based prospective survey. Ann
Rheum Dis 56:470–475
Eur J Orthop Surg Traumatol
123