reevaluating the sonographic criteria for acute appendicitis in children
TRANSCRIPT
Reevaluating the SonographicCriteria for Acute Appendicitis
in Children:
A Review of the Literature and a Retrospective Analysis of 246 Cases
Andrew T. Trout, MD, Ramon Sanchez, MD, Maria F. Ladino-Torres, MD
Ac
FrCCCCAd
ªht
13
Rationale and Objectives: There has been little rigorous evaluation of the sonographic criteria for acute appendicitis in children. Our clin-
ical experience has called the traditional diagnostic criteria into question. We set out to review the literature, evaluate the most commonly
applied diagnostic criteria for acute appendicitis, and identify those criteria that best predict the presence of disease.
Materials and Methods: A critical review of the literature concerning the sonographic diagnosis of acute appendicitis was performed.
Based on diagnostic criteria identified in that review, two independent, blinded pediatric radiologists retrospectively reviewed 246 right
lower quadrant ultrasound examinations in which the appendix was identified with attention to commonly described diagnostic criteriafor acute appendicitis. Multivariate and classification and regression tree analysis were performed to identify criteria that predict
appendicitis.
Results: In a multivariate analysis, inflammation of the periappendiceal fat is the only finding that statistically significantly predicts acute
appendicitis (OR = 68.93,P< .0001). Other criteria such as diameter, noncompressibility, hyperemia, the presence of an appendicolith, andloss of stratification of the appendiceal wall do not independently predict appendicitis.
Conclusion: Periappendiceal fat infiltration is themost important diagnostic criterion for acute appendicitis in children. Strict application of
other criteria such as diameter should be avoided.
Key Words: Appendix; ultrasound; diagnosis.
ªAUR, 2012
Sonographic evaluation of the appendix was originally
described in 1986 in a population of adults (1). Since
that time, the technique has been adopted in children
and due to the lack of ionizing radiation has largely become
the technique of choice in this population (2). In the 25 years
since the original article, there have been numerous analyses
of the performance of ultrasound in diagnosing acute appen-
dicitis in children. Interestingly, the diagnostic criteria that are
applied for acute appendicitis are largely based on the adult
population and these criteria, as they apply to children, have
received significantly less analytic attention over that same
25-year period. A few studies have evaluated the diagnostic
value of specific sonographic findings (3–7) and a recent
analysis has called into question the diagnostic criteria as
ad Radiol 2012; 19:1382–1394
om the Department of Radiology, Section of Pediatric Radiology, C.S. Motthildren’s Hospital, University of Michigan Medical Center, 1500 East Medicalenter Drive, 3-228, Ann Arbor, MI 48109-4252 (A.T.T., R.S., M.F.L.-T.);incinnati Children’s Hospital Medical Center, Department of Radiology,incinnati, OH (A.T.T.). Received February 3, 2012; accepted June 23, 2012.dress correspondence to: R.S. e-mail: [email protected]
AUR, 2012tp://dx.doi.org/10.1016/j.acra.2012.06.014
82
historically defined in the literature (8). This recent publica-
tion resonated with our clinical experience as we have anec-
dotally noted the limited value of some of the traditional
diagnostic criteria. Specifically, we hypothesized that appen-
diceal diameter >6 mm and lack of compressibility of the
appendix are poor diagnostic criteria for acute appendicitis
and that secondary findings such as periappendiceal fat infil-
tration better predict the presence of appendicitis.
Based on our clinical experience and this hypothesis, we set
out to critically review the literature and to evaluate the diag-
nostic performance of common sonographic criteria for acute
appendicitis in children.
MATERIALS AND METHODS
Institutional review board approval was obtained for this ret-
rospective analysis, which represents an analysis of a subset of
patients from a cohort previously reported (9) in the literature.
Sonographic examinations of the appendix between May
2005 and May 2010 were identified through a search of the
Radiology Informatics System. The following search criteria
were used:
TABLE 1. Diagnostic Criteria Assessed by Independent Reviewers and the Specific Definitions of these Criteria
Parameter Definition
Maximal appendiceal diameter Maximum anteroposterior dimension of the appendix measured in millimeters from
serosal to serosal surface while imaging the appendix in the transverse plane
Compressibility (Figure 6) Measurable, reproducible change in the caliber of the lumen of the appendix or
noticeable deformation of the adjacent or overlying soft tissues
1. Yes
2. No
3. Inadequate compression
Appendiceal wall signature (Figure 7) 1. Three concentric hyperechoic rings (normal (17)): Five distinct layers of bowel
wall that are alternately hyperechoic and hypoechoic (hyperechoic mucosal
surface, hypoechoic mucosa, hyperechoic submucosa, hypoechoic muscularis
and hyperechoic serosa)
2. Two concentric hyperechoic rings: Preservation of the outer two hyperechoic
rings (serosa and submucosa), with nonvisualization of the inner hyperechoic
ring (mucosa)
3. One hyperechoic ring: Preservation of the outer most hyperechoic ring (serosa)
with nonvisualization of the inner two hyperechoic rings (mucosa and submucosa)
Appendiceal vascularity (Figure 8) Assessed with color or power Doppler, relative to adjacent normal soft tissue
1. Normal
2. Increased
3. Decreased or absent
4. Not assessed
Periappendiceal fat
infiltration (Figures 3 and 4)
Increased echogenicity of the periappendiceal fat relative to normal intraabdominal
fat with associated thickening
1. Absent: Very minimal (noncircumferential) or no echogenic fat adjacent to
the appendix
2. Present: Echogenic periappendiceal fat with ill-defined margins that
circumferentially surrounds the appendix
Presence of appendicolith (Figure 9) Echogenic, well-defined focus within the appendix with posterior acoustic shadowing
1. Yes
2. No
Presence of fluid Simple or complex fluid without a defined wall
1. Yes
2. No
Fluid characteristics 1. Simple: Anechoic
2. Complex: Echogenic components, including septations
Presence of organized fluid collection Loculated fluid collection surrounded by a definable wall with associated mass effect
1. Yes
2. No
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
1. Patient demographics: <18 years of age
2. Referral source: outpatient or emergency room
3. Exam types: right upper quadrant, right lower quadrant,
abdominal ultrasound
4. Exam/clinical indications: abdominal pain, right lower
quadrant pain, pelvic pain, appendicitis, fever
Of the approximately 4000 records identified that met these
search criteria, only those in which the appendix was identified
by the initial interpreting radiologistwere included in the analysis.
Two pediatric radiologists with 11 and 4 years of clinical
experience (R.S., M.LT.) independently reviewed each ultra-
sound examination in which the appendix was identified.
These reviews were performed using Siemens KinetDx
WS3000 workstations (Siemens Medical Solutions USA
Inc., Mountain View, CA) and were independent of the
clinical interpretation of the ultrasound examinations that
had been previously provided by another radiologist. Research
reviewerswere blinded to patient outcomes andwere provided
with only the clinical history recorded by the technologist at
the time of the examination. Each reviewer recorded data
related to sonographic criteria for acute appendicitis previ-
ously reported in the literature (Table 1). These criteria were
derived from a review of the literature and included those cri-
teria that had been previously reported to have diagnostic
value. Diagnostic criteria were standardized between the two
reviewers before independent review of the data. In addition
to recording diagnostic criteria, each reviewer provided an
impression of ‘‘normal appendix,’’ ‘‘acute appendicitis,’’ or
‘‘perforated acute appendicitis’’ for each case.
The electronic medical records for each patient were
reviewed by a third study team member for demographic
1383
TABLE 2. Summary of Diagnostic Parameters as Recorded by Both Readers
Parameter n (%)
Pathology Conclusion
Normal Appendicitis Perforated Appendicitis
Mean maximal diameter
Reader 1: 5.9 � 1.8 mm 10.2 � 3.1 mm 10.1 � 4.1 mm
Reader 2: 5.7 � 1.6 mm 10 � 2.9 mm 10.3 � 3.1 mm
Compressibility
Yes
Reader 1: 40 (18.2%) 26 9 5
Reader 2: 42 (19%) 36 6 0
No
Reader 1: 17 (7.7%) 1 11 5
Reader 2: 58 (26.2%) 6 17 35
Inadequate compression
Reader 1: 163 (74.1%) 90 56 17
Reader 2: 121 (54.8%) 76 35 10
Appendiceal wall signature
3 rings (normal)
Reader 1: 155 (70.5%) 108 38 9
Reader 2: 154 (69.7%) 112 34 8
2 rings
Reader 1: 59 (26.8%) 9 34 16
Reader 2: 58 (26.2%) 6 36 16
1 ring
Reader 1: 6 (2.7%) 0 4 2
Reader 2: 9 (4.1%) 0 6 3
Appendiceal vascularity
Normal
Reader 1: 78 (53.4%) 59 14 5
Reader 2: 61 (41.8%) 49 9 3
Increased
Reader 1: 55 (37.7%) 7 38 10
Reader 2: 67 (45.9%) 12 43 12
Decreased/absent
Reader 1: 13 (8.9%) 0 10 3
Reader 2: 18 (12.3%) 2 12 4
Periappendiceal fat infiltration
Absent
Reader 1: 120 (52.4%) 110 10 0
Reader 2: 119 (51.5%) 110 9 0
Present
Reader 1: 109 (47.6%) 15 66 28
Reader 2: 112 (48.5%) 18 67 27
Presence of appendicolith
Yes
Reader 1: 38 (17.3%) 3 25 10
Reader 2: 32 (14.5%) 3 21 8
No
Reader 1: 182 (82.7%) 114 51 17
Reader 2: 189 (85.5%) 115 55 19
Presence of fluid
Yes
Reader 1: 135 (58.7%) 67 46 22
Reader 2: 104 (45.2%) 49 38 19
No
Reader 1: 95 (41.3%) 59 30 6
Reader 2: 124 (53.9%) 77 38 9
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
1384
TABLE 2. (continued) Summary of Diagnostic Parameters as Recorded by Both Readers
Parameter n (%)
Pathology Conclusion
Normal Appendicitis Perforated Appendicitis
Fluid characteristics
Simple
Reader 1: 105 (78.4%) 65 34 6
Reader 2: 80 (75.5%) 44 26 10
Complex
Reader 1: 29 (21.6%) 1 12 16
Reader 2: 26 (24.5%) 5 12 9
Presence of organized fluid collection
Yes
Reader 1: 8 (3.4%) 0 2 6
Reader 2: 5 (2.1%) 0 1 4
No
Reader 1: 225 (96.6%) 129 74 22
Reader 2: 228 (97.9%) 129 75 24
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
information (age, sex, and body mass index) and outcomes
(surgery, surgical pathology, and discharge diagnosis). Dis-
charge diagnosis and pathologic diagnosis were considered
the gold standard for diagnosis and outcomes.
Statistical Analysis
Assessment of interrater agreement and reader agreement
with outcomes was performed using the kappa (k) statistic
with 95% confidence interval as well as frequency of agree-
ment. Individual parameters were assessed for diagnostic value
for acute appendicitis using Fisher’s exact test and calculation
of odds ratios for discrete parameters, and logistic regression
with resulting odds ratios for continuous parameters. Multiple
explanatory variables were assessed using logistic regression
using stepwise elimination of parameters. P < .05 was consid-
ered significant for all analyses. To determine if combinations
of criteria better predict appendicitis, Classification and
Regression Tree (CART) analysis was performed with nodes
restricted to a minimum of 20 patients. This type of analysis
creates a decision tree in which variables with the greatest dis-
criminatory efficacy are placed earlier in the tree. By restrict-
ing node size to n = 20, diagnostic criteria that subdivide
patients into groups smaller than 20 are excluded. Accepting
smaller node sizes allows more accurate subdivision of the
patient population but also complicates the diagnostic tree
making it less clinically useful. All analyses were performed
for data from both readers. Statistical analysis was performed
with SAS version 9.2 (SAS Institute Inc., Cary, NC), except
for the CART analysis, which was performed using S-PLUS
version 7.0 (Insightful Corp., Seattle, WA).
RESULTS
The appendix had been reportedly identified by the initial inter-
preting radiologist in 246 examinations. In 13 of these examina-
tions, the reviewers felt that the appendixhadnot beenaccurately
identified. Of the 233 patients with a visualized appendix, out-
comes data were available in 220, with the other 13 lost to
follow-up. A total of 125 (53.6%) of the examinations were in
female patients and 108 (46.4%) were in males. The mean age
for the patient population was 10.7 � 4.1 years. Raw data
with findings for each of the variables are detailed in Table 2.
Based on pathologic and clinical conclusions, 76 patients
were diagnosed with nonperforated acute appendicitis and
27 were diagnosed with perforated (gross and microscopic
perforation) appendicitis. A total of 117 patients (53.2% of
220) did not have appendicitis. Readers 1 and 2 correctly cate-
gorized patients into the three groups (appendicitis, perfo-
rated appendicitis, and normal) 82.9% (k = 0.70, 95% CI:
0.62–0.79) and 79.2% (k = 0.64, 95% CI: 0.55–0.73) of the
time, respectively. When patients were subdivided based on
the presence of acute appendicitis alone (combining perfo-
rated and nonperforated cases), readers 1 and 2 correctly diag-
nosed appendicitis in 88.7% (k = 0.77, 95% CI: 0.69–0.86)
and 89.1% (k = 0.78, 95% CI: 0.78–0.87) of cases. Diagnostic
agreement between the two readers was present 87.3% of the
time (k = 0.77, 95% CI: 0.7–0.85) when three diagnostic cat-
egories were used (appendicitis, perforated appendicitis, and
normal) and 93.2% (k = 0.86, 95% CI: 0.8–0.93) of the
time when distinguishing appendicitis from normal.
Appendiceal Diameter
Mean appendiceal diameter for each of the diagnostic catego-
ries is detailed for both readers in Table 2. Table 3 details the
observed frequency of acute appendicitis for each millimeter
of appendiceal diameter in this patient population. This table
also lists diagnostic performance results including sensitivity,
specificity, positive predictive value, negative predictive value
and accuracy for a diagnostic cut-off of 6 mm as commonly
1385
TABLE 3. Frequency of Acute Appendicitis (Perforated and Nonperforated) Based on Maximal Appendiceal Diameter
Diameter n
Appendicitis
(Perforated, Nonperforated) Normal Sensitivity Specificity PPV NPV Accuracy
2–4 mm
R1: 22 (10%) 2 (1, 1) 20
R2: 25 (11.3%) 1 (0, 1) 24
5 mm
R1: 32 (14.5%) 3 (3, 0) 29 95.1% 41.9% 59% 90.7% 66.8%
R2: 36 (16.3%) 2 (1, 1) 34 97.1% 49.2% 62.5% 95.1% 71.5%
6 mm
R1: 45 (20.5%) 10 (3, 7) 35 85.4% 71.8% 72.7% 84.8% 78.2%
R2: 43 (19.5%) 7 (2, 5) 36 90.3% 79.7% 79.5% 90.4% 84.6%
7 mm
R1: 23 (10.5%) 11 (1, 10) 12 74.8% 82.1% 78.6% 78.7% 78.6%
R2: 21 (9.5%) 12 (2, 10) 9 78.6% 87.3% 84.4% 82.4% 83.3%
8 mm
R1: 16 (7.3%) 5 (2, 3) 11 69.9% 91.5% 87.8% 77.5% 81.4%
R2: 14 (6.3%) 7 (2, 5) 7 71.8% 93.2% 90.2% 79.1% 83.3%
$9 mm
R1: 82 (37.3%) 72 (17, 55) 10
R2: 82 (37.1%) 74 (20, 54) 8
NPV, negative predictive value; PPV, positive predictive value; R1, reader 1; R2, reader 2.
Sensitivity, specificity, PPV and NPV results are based on using that diameter as a diagnostic cutoff.
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
described in the literature and 7 mm as recently proposed by
Goldin et al (8). Receiver operating characteristic curves for
the range of appendiceal diameters as measured by each reader
are shown in Figures 1 and 2. The area under the curve is
0.874 and 0.915 for readers 1 and 2, respectively.
Other Diagnostic Parameters
Based on univariate analysis which looks at each variable
isolated from the others, the statistical significance of indi-
vidual diagnostic parameters as predictors of acute appendi-
citis with associated odds ratios are detailed in Table 4. As a
continuous variable, appendiceal diameter is not a signifi-
cant predictor of acute appendicitis (P = .1286 and P =
.0933 for readers 1 and 2, respectively). Specific cutoffs of
6 mm, 7 mm, and 8 mm, however, are significant predictors
(Table 4).
On multivariate regression which accounts for interactions
between variables, the only diagnostic parameter that statisti-
cally significantly predicted the presence of acute appendicitis
for both readers was the presence of infiltration of the periap-
pendiceal fat (P < .0001 for both; Figs 3, 4). The diagnostic
value of infiltration of the periappendiceal fat is as follows
(listed reader 1/reader 2): sensitivity 90.4%/91.3%, specificity
88%/85.9%, positive predictive value 86.2%/83.9%, negative
predictive value 91.7%/92.4%, accuracy 89.1%/88.3%. For
reader 2 only, bowel wall signature (specifically two vs. three
rings) was also a significant predictor or acute appendicitis
(P = .0132). For reader 1, bowel wall signature approached
but did not reach statistical significance (P = .0561). The
remainder of the diagnostic criteria were not statistically sig-
nificant predictors of appendicitis on multivariate analysis.
1386
Classification and regression tree analysis demonstrates that
combinations of diagnostic criteria do not change, to a great
extent, the classification of a given case beyond that assigned
by the presence or absence of infiltration of the periappendi-
ceal fat (Fig 5).
DISCUSSION
Despite 25 years of experience with the use of sonography in
the evaluation of acute appendicitis, there has been little
refinement of the diagnostic criteria in a pediatric population.
Most of the data concerning the use of this modality in chil-
dren focus on the overall diagnostic performance of sonogra-
phy as it relates to clinical management and as it compares to
computed tomography. Few articles have rigorously evaluated
the specific criteria for a sonographic diagnosis of acute
appendicitis in children.
Currently, the most commonly used diagnostic criteria are
a noncompressible appendix with an outer diameter >6 mm
(7,10–12). Other criteria that have been described include
hypervascularity, periappendiceal fat infiltration, loss of the
normal bowel signature in the appendiceal wall, and the
presence of an appendicolith (13–18).
Appendiceal Diameter
An appendiceal diameter of >6 mm is the most commonly
described diagnostic criterion for acute appendicitis. This cut-
off was first described in 1988 by Jeffrey et al who prospec-
tively evaluated 250 adult patients with suspected acute
appendicitis (19). In this population, 78/84 (93%) patients
with an appendix measuring >6 mm had appendicitis. Based
Figure 1. Receiver operating characteristic
curve for appendiceal diameter for reader 1.
Figure 2. Receiver operating characteristiccurve for appendiceal diameter for reader 2.
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
on their findings, the authors stated that ‘‘.maximal appen-
diceal diameter may be an additional useful parameter in guid-
ing clinical management’’ but qualified this by saying ‘‘.our
criteria regarding appendiceal diameter are not absolute and
are presented merely as guidelines’’ (19).
Although the 6 mm cutoff was originally described in an
adult population, descriptive studies have since defined the
diameter of both the normal and acutely inflamed appendix
in children. Results of these studies are detailed in Table 5.
Although there is a broad range of measurements for normal
appendices, studies have shown no association between
appendiceal diameter and patient specific factors such as
height, weight, or body mass index (6,20). Some of the
observed variation in appendiceal diameter may reflect
differences in measurement. In our review of the literature,
we have been unable to identify an agreed-on measurement
method, and most authors do not define how they are
measuring the appendix. Differences are introduced by
measuring anterior to posterior versus transversely as well as
measuring under different degrees of compression, particu-
larly if the appendix is compressible (6,20). Additionally,
human variation introduces measurement differences.
Between our two reviewers, the mean difference in diameter
measurement over all patients was 0.7 � 1.3 mm.
1387
TABLE 4. Statistical Significance of Individual Diagnostic Parameters for a Diagnosis of Acute Appendicitis Based upon UnivariateAnalysis of Findings Recorded by Readers 1 and 2
Parameter
Reader 1 Reader 2Interrater Agreement
(95% Confidence Interval)Odds Ratio P Value Odds Ratio P Value
Sex (male) 1.99 .0120 –
Age 1.03 .0914 –
Body mass index 2.43 .0262 –
Appendiceal diameter
Continuous variable 1.02 .1286 1.02 .0933 0.9*
6-mm cutoff 14.12 <.0001 32.22 <.0001
7-mm cutoff 14.93 <.0001 36.43 <.0001
8-mm cutoff 13.54 <.0001 25.28 <.0001
Compressibility (no) 0.03 <.0001 0.02 <.0001 0.675 (0.469–0.882)
Appendiceal wall signature 0.741 (0.645–0.836)
1 ring vs. 3 rings NA .0010 NA <.0001
2 rings vs. 3 rings 12.77 <.0001 23.09 <.0001
Vascularity 0.755 (0.653–0.857)
Decreased vs. normal NA <.0001 NA <.0001
Increased vs. normal 21.29 <.0001 18.72 <.0001
Periappendiceal fat infiltration (yes) 68.93 <.0001 63.83 <.0001 0.886 (0.825–0.946)
Appendicolith (yes) 19.56 <.0001 15.02 <.0001 0.898 (0.818–0.979)
Free fluid (yes) 1.66 .0801 1.91 .0173 0.674 (0.579–0.769)
Fluid characteristics (complex vs. anechoic) 45.5 <.0001 5.13 .0015 0.713 (0.552–0.873)
Organized fluid collection NA .0014 NA .0168 0.763 (0.497–1)
Interrater agreement for each parameter is also provided expressed as kappa statistics with Pearson correlation (*) reported for appendiceal
diameter as a continuous variable. Odds ratios indicated as ‘‘NA’’ are a result of inability to calculate the odds ratio in situations in which there is
a frequency of zero in any one of the diagnostic categories.
Figure 3. Periappendiceal fat infiltration.
(a) A 2 year old with abdominal pain. Ultra-
sound showsa normal appendixwith normalperiappendiceal fat. The fat is intermediate in
echogenicity and without mass effect. (b) A10yearoldwith right lower quadrantpain.Ul-trasound shows circumferential echogenic
fat (asterisks) around a distended, abnormal
appendix in a patient with acute appendici-
tis. Note the increased thickness and poordefinition of the fat in addition to the altered
echogenicity.
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
Despite the difficulties inherent in measurement, several
studies in both children and adults have looked at the diagnos-
tic performance of a diameter cutoff in acute appendicitis.
Those studies that evaluated diameter as an independent var-
iable are summarized in Table 6. Much of the attention has
focused on the 6 mm cutoff described by Jeffrey et al, which
has been shown to be variably predictive in both children
and adults. Although some results support the use of a
6 mm cutoff (4,5), there is evidence that this cutoff may not
be ideal. The diameter measurement of the appendix
depends on wall thickness and luminal content. Normal
physiologic processes such as lymphoid hyperplasia leading
1388
to wall thickening as well as fecal impaction resulting in
distention of the appendiceal lumen have been shown to
increase the diameter of the appendix to >6 mm (14,21). In
their recent pediatric study, Goldin et al found that using
the traditional 6 mm cutoff would have only correctly classi-
fied 84% of appendicitis cases and would have resulted in 26
unnecessary surgeries (8). Based on their findings, the authors
recommend a 7 mm cutoff (8).
Studies in adult patients have also called a 6 mm cutoff into
question as up to 23% of normal patients and 14%–32% of
patients with right lower quadrant pain but without acute
appendicitis have appendices >6 mm in diameter (22,23).
Figure 4. A 10 year old with right lower quadrant pain and periap-
pendiceal fat infiltration resulting from acute appendicitis. In this im-age, the appendix is oriented longitudinally. Note the echogenic,
thickened periappendiceal fat (asterisks) surrounding the abnormal
appendix (A). The signature of the appendiceal wall is abnormal
with only a single echogenic layer visible.
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
Based on a high sensitivity (100%) but low specificity (68%),
Rettenbacher et al concluded that a 6 mm cutoff was more
appropriately suited to exclude appendicitis than to diagnose
it (22).
In our population, appendiceal diameter as a continuous
variable was not statistically significantly associated with a
diagnosis of acute appendicitis. While cutoff diameters of
6 mm, 7 mm, and 8 mm were all significant predictors in
the univariate analysis, the accuracy of these cutoffs was
only moderate (Table 3) and was not as good as described
by Goldin et al (8). This is difficult to explain, but might be
attributed to population or measurement differences. Median
appendiceal diameters were a full millimeter larger in both
normals and patients with acute appendicitis in our popula-
tion than in the population described by Goldin et al (5 vs.
6 mm for normal; 9 vs. 10 mm for appendicitis).
In multivariate analyses, diameter cutoffs were no longer
statistically significant predictors of acute appendicitis in our
population. To date, the only other multivariate analysis of
diagnostic criteria for acute appendicitis in the literature is a
study of adult patients by van Randen et al that did find a
diameter of 6 mm as a statistically significant predictor of
appendicitis (OR = 9.4) (24). This discrepancy is difficult to
explain but could relate to study design or possibly differences
between adult and pediatric patients.
Although statistically our data do not support the use of a
diameter cutoff, if one were to be used, an 8 mm cutoff would
provide greater diagnostic accuracy than a 6 mm cutoff
(Figs 1, 2; Table 3).
Compressibility
In the original description of the use of ultrasound for the
diagnosis of acute appendicitis, Puylaert described the diag-
nostic criteria for acute appendicitis as a noncompressible
appendix with a faint or invisible echogenic layer (Fig 6)
(1). These criteria were simply stated by the authors and not
evaluated for their specific diagnostic performance. Noncom-
pressibility is one of the diagnostic criteria that is most com-
monly accepted but has received little evaluation in isolation
of other diagnostic parameters. The only studies that have
evaluated this criterion in isolation are in adults and the results
are conflicting. Kessler et al found that the identification of a
noncompressible appendix in a population of 125 adults had a
sensitivity and specificity of 96%with an accuracy of 96% (12).
van Randen et al however found that a noncompressible
appendix was not a significant predictor of appendicitis in a
population of 942 adults (OR = 0.8) (24).
This discrepancy may be explained by methodological dif-
ferences. To the best of our understanding, it appears that
Kessler et al evaluated their data in a univariate fashion
whereas van Randen et al performed a multivariate analysis.
The importance of this difference is apparent in our findings.
Based on a univariate analysis, identification of a noncompres-
sible appendix is significantly associated with the presence of
acute appendicitis. Under multivariate analysis, however, a
lack of compressibility does not predict the presence of appen-
dicitis, suggesting interplay of diagnostic variables.
It is our opinion that compressibility is one of the most
operator dependent factors assessed in a right lower quadrant
ultrasound. As we were designing this study, we struggled to
define ‘‘adequate compression.’’ Although the concept seems
simple, there are many hindrances to adequate compression.
These include:
� Depth of the appendix: it is difficult to exert compressive
force on an appendix that is at depth within the abdomen
� Overlying/adjacent tissues: a subhepatic, perirenal, or ret-
rocecal appendix can be difficult to compress
� Patient guarding: adequate compression cannot be
achieved in a patient who is tensing their abdominal
musculature
� Patient body habitus: adequate compression can be difficult
to achieve in obese patients because of intervening body fat
(subcutaneous or intraabdominal)
� The presence of fluid, particularly loculated: fluid is non-
compressible and thus may not allow compression of an
underlying appendix
� Decompressed appendix: if the appendix is collapsed it may
not compress further
� Mobile appendix: a normal appendix is mobile and may be
displaced rather than compressed
Based on our definition of adequate compression which
required a measurable change in the diameter of the
appendiceal lumen or clear deformation of the adjacent soft
tissues, our reviewers felt that compression was inadequate
1389
TABLE 5. Diameters of Normal and Acutely Inflamed Appendices Reported in the Pediatric Literature
Study Study Type n Mean Normal Diameter Mean Appendicitis Diameter
Goldin, 2011 Retrospective 204 5.0 mm (range 2.0–11.0)* 9.0 (range 1.3–20.0 mm)*
Hahn, 1998 Prospective 542 suspected appendicitis
54 healthy
3.9 (�1.2) mm 10.9 (�3.1) mm
Ozel, 2011 Prospective 142 4.2 (�0.9) mm –
Park, 2007 Prospective 51 – 9.7 mm (range 6.7–21.2 mm)
Wiersma, 2005 Prospective 118 3.9 mm (range 2.1–6.4 mm) –
Our data Retrospective 216 5.9 (�1.7) mm 10.5 (�3.1) mm
*Diameters reported are median diameters.
Figure 5. Classification and Regression Tree (CART) analysis based on reader 1. Each decision point in the tree is the result of application of
one of the diagnostic criteria. Criteria with the greatest discriminatory efficacy are placed earlier in the tree. Percentages indicated in parenthe-
ses are the rate of appendicitis in that subgroup. Note that periappendiceal fat strandingmost effectively classifies patients (highest percentage
of appendicitis in the positive group and lowest percentage in the negative group). Application of additional diagnostic criteria change thediagnostic probability as indicated but do not substantially change the classification of patients to ‘‘appendicitis’’ or ‘‘not appendicitis.’’
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
in 74% and 55% of the examinations they reviewed (reader 1,
reader 2).
Altered Appendiceal Wall Signature
The appendix has been described as having five distinct echo
layers including (Fig 7) (18,25):
Echogenic mucosal surface
Hypoechoic mucosa
Echogenic submucosa
Hypoechoic muscularis propria
Echogenic serosal fat
In pathologic specimens of acute appendicitis, the integrity
and stratification of these layers is lost (25). Similar changes
were noted in vivo in a multicenter trial in which alteration
of the normal mural stratification was the only ultrasound
finding that in isolation had adequate sensitivity for a diagnosis
1390
of acute appendicitis (26). Other than these studies, there are
few data in the literature to support the use of this diagnostic
criterion. Our data show that loss of the normal appendiceal
wall signature is a statistically significant predictor of acute
appendicitis in univariate analysis for both readers but only
for reader 2 in multivariate analysis. The odds ratio for this
parameter, however, is much lower than that for periappendi-
ceal fat infiltration and CART analysis shows that the most
important criteria to stratify patients is periappendiceal fat
infiltration (Fig 5).
Vascularity
Doppler flow within the appendiceal wall that is greater than
that in normal adjacent soft tissues (which have little to no
detectable flow) has been described as a finding diagnostic
of acute appendicitis (Fig 8) (16). Quillin et al showed that
in children, increased Doppler flow had a sensitivity of 87%,
specificity of 97%, and accuracy of 93% for acute appendicitis
Figure 7. Appendiceal wall signature. In this 4 year old with right lower quadrant pain, the normal appendix (a) has three concentric hyper-
echoic rings formed by the serosa, submucosa, and mucosa. Progressive inflammation results in loss of these concentric rings beginningwith the mucosa. (b) In this 15 year old with right lower quadrant pain, the appendix has a two-ring appearance with the echogenic submucosal
and serosa still visible. (c) In this 10 year old with acute appendicitis, ultrasound shows loss of all but the outer serosal echogenic ring.
Figure 6. Compression of a normal appen-dix. In this 15 year old with right lower quad-
rant pain, the normal appendix is indicated
by the arrows. (a) With gentle pressure theappendix is more round in configuration.
During compression (b), there is a notable
change in diameter of the appendix which
is now more ovoid in configuration.
TABLE 6. Performance of Specific Diameter Cutoffs in the Diagnosis of Acute Appendicitis as Reported in the Literature
Study Study Type Population Total n n Used for Analysis Diameter Cutoff Sens Spec Acc OR
Je, 2009 Retrospective Pediatric 160 160 >5.7 mm 89.6 93.2
Peletti, 2006 Prospective Pediatric 107 100 >6 mm 100 98
Goldin, 2011 Retrospective Pediatric 304 204 >7 mm 89.7 95.4 93.2
Kessler, 2004 Prospective Adult 125 104 >6 mm 98 98 97
Rettenbacher, 2001 Prospective Adult 710 518 >6 mm 100 68 79
van Randen, 2010 Prospective Adult 942 528 >6 mm 9.4
Worrel, 1990 Prospective Adult 200 200 >6 mm 66 93
Sens, sensitivity; spec, specificity; acc, accuracy; OR, odds ratio.
Although diameter is commonly used as a diagnostic criterion, these were among the few studies that looked at diameter as an independent
predictor in the sonographic evaluation of appendicitis.
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
(16). Results in adults were less convincing with Kessler et al
showing a sensitivity of 52%, specificity of 96%, and accuracy
of 73% for color Doppler and Incesu et al showing a sensitivity
of 74%, specificity of 93%, and accuracy of 80% for Power
Doppler (12,27). More recently, studies in both children
and adults have found that vascularity is not a significant
predictor of acute appendicitis (3,8,24). Our findings
support the latter conclusion. Although vascularity was a
significant univariate predictor of appendicitis, this effect did
not persist in the multivariate analysis.
Assessment of the vascularity of the appendix is inherently
limited because it is dependent on Doppler settings and user
technique. Additional difficulty is introduced by the fact that
the assessment of hyperemia is subjective and the development
of appendiceal necrosis has been shown to result in decreased
or absent vascularity, clouding the diagnostic picture (14).
1391
Figure 8. Hypervascularity in acute appendicitis. Normally, flow
greater than that of normal adjacent soft tissue should not be detect-able within thewall of the appendix. In this 10 year old with right lower
quadrant pain, there is increased blood flow in the wall (arrowheads)
and within the echogenic periappendiceal fat.
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
Periappendiceal Fat Infiltration
In our population, infiltration of the periappendiceal fat had
the highest odds ratio of all diagnostic criteria in the univariate
analysis and was the only diagnostic criterion that was statisti-
cally significantly associated with the presence of acute appen-
dicitis in multivariate analysis for both readers (Figs 3, 4). The
diagnostic performance statistics for periappendiceal fat infil-
tration are higher than the diagnostic statistics for any of the
evaluated appendiceal diameters (6, 7, or 8 mm) and higher
than the 70.5% sensitivity and 88.9% accuracy described by
Goldin et al, who found that fat infiltration was an insensitive
indicator of acute appendicitis (8).
In children, there have been no other studies that have stat-
istically evaluated the diagnostic value of periappendiceal fat
stranding but descriptive studies have shown an association
between fat infiltration and acute appendicitis. Wiersma
et al identified echogenic fat in 97.2% of the 71 patients
with appendicitis in their series (7). Park et al also showed
that the presence of periappendiceal fat infiltration can help
to distinguish appendicitis from a dilated, fecally impacted
appendix because this was present in 50/51 cases of appendi-
citis but none of the fecal impaction cases (21).
Findings in studies of adult appendicitis also support the use
of periappendiceal fat infiltration as a diagnostic criterion.
Kessler et al found that this had a sensitivity of 91%, specificity
of 76%, and accuracy of 83% (11). Fat infiltration was one of
the few variables that remained statistically significantly pre-
dictive of appendicitis in the multivariate analysis by van Ran-
den et al where it had an odds ratio of 3.2 (24). The degree of
fat infiltration has also been shown by Noguchi et al to signifi-
cantly correlate with the severity of acute appendicitis (28).
Periappendiceal fat infiltration is not a perfect diagnostic
criterion for acute appendicitis. Echogenic fat has been
reported to be present in rare cases in normal individuals
and was seen in a small number of normal patients in our pop-
ulation (6,20). Echogenic fat can also be the result of an
1392
adjacent inflammatory process such as infectious or
inflammatory terminal ileitis, omental infarction, epiploic
appendagitis, or Meckel diverticulitis. Generally these
processes will result in adjacent but not circumferential
periappendiceal fat infiltration. These diagnoses, however,
must be excluded before a diagnosis of acute appendicitis is
made. Finally, some would argue that the identification of
infiltration of the periappendiceal fat is subjective with
potential confusion arising from misidentification of normal
periappendiceal fat. It is for this reason that we strictly
defined this criterion as ‘‘circumferential echogenic
periappendiceal fat.’’ We contend that using a strict
definition will maximize the diagnostic value of this criterion.
Appendicolith
The association of an appendicolith with acute appendicitis
was initially described with plain radiography (Fig 9) (29).
Despite the fact that many authors use the presence of an
appendicolith as one of several diagnostic criteria for acute
appendicitis, this has not been well evaluated as an isolated cri-
terion (13,14,30,31). In children, Goldin et al found a
sensitivity of 33%, specificity of 96%, and accuracy of 72%
for appendicoliths (8). In our population, although rarely
seen in normal appendices, the presence of an appendicolith
was not a significant predictor of appendicitis on multivariate
analysis. Similar findings were described in adults by van Ran-
den et al (24).
Presence of Fluid
According to the literature, the diagnostic value of periappen-
diceal fluid is variable. Both Goldin et al and Franke et al
found periappendiceal fluid to be highly specific (99%) but
not sensitive for acute appendicitis (20% Goldin, 14% Franke)
and both authors described poor accuracy for free fluid as a
diagnostic criterion (<70% Goldin, 77% Franke) (8,26). In
their analysis, van Randen et al showed that periappendiceal
free fluid had one of the lowest odds ratios for acute
appendicitis (0.9) in univariate analysis and was not a
significant predictor in multivariate analysis (24). Our results
are similar in that the presence of fluid was a significant pre-
dictor in univariate analysis (OR < 2) but was not a predictor
in multivariate or CART analysis.
The literature describes several diagnostic criteria that pre-
dict the presence of acute appendicitis and yet most of these
show no statistically significant associationwith acute appendi-
citis in our population. One possible explanation for this is that
prior articles predominately represent univariate analyses of
diagnostic criteria that do not account for the interaction
between variables. In our population, many of these same var-
iables are statistically significant predictors of acute appendicitis
in univariate analyses but when multivariate modeling and
CARTanalysis are performed, periappendiceal fat infiltration
is shown to be themost valuable predictorof acute appendicitis.
Figure 9. Longitudinal (a) and transverse(b) images of appendicoliths. In this 11 year
old with right lower quadrant pain, appendi-
coliths appear as echogenic foci (arrows)
with associated posterior acoustic shadow-ing (arrowheads) within the appendiceal
lumen. Note the normal appendiceal wall in
(a) with three discrete echogenic layers.
Academic Radiology, Vol 19, No 11, November 2012 PEDIATRIC APPENDICITIS: ULTRASOUND CRITERIA
This study has several limitations. This is a retrospective
analysis, and the reviewers were limited to the images that
had been saved to the picture archiving and communications
system over the study period. These images may not have
adequately assessed all of the diagnostic parameters of interest
or provided sufficient imaging of an area of interest. Addition-
ally, this analysis concerns sonographically apparent criteria
and does not incorporate clinical data. The findings of this
analysis will need to be confirmed with prospective studies
and in clinical practice. These data are also limited by the
gold standard we used for a diagnosis of acute appendicitis.
In cases where surgery was not performed, appendicitis is
assumed to be absent. False-negative clinical diagnoses or
patients with mild appendicitis that resolved may have conta-
minated these data. Finally, these data are limited by sample
size. Larger samples might identify diagnostic criteria that
reach statistical significance that are not identified in this
analysis.
Approach to Ultrasound in Suspected AcuteAppendicitis
As a means of summarizing our results and review of the liter-
ature, we propose the following possible approach to inter-
preting right lower quadrant ultrasounds in acute appendicitis.
First evaluate for circumferentially thickened, echogenic
periappendiceal fat. If this is present, appendicitis should be
highly suspected (present in 87% of patients in our popula-
tion). Additional findings such as increased appendix diameter,
loss of normal appendiceal wall architecture, or the presence of
an appendicolith can further support the diagnosis. The
absence of any or all of these findings should not deter diagno-
sis unless another explanation for inflamed right lower quad-
rant fat is elicited. If, however, circumferentially thickened,
echogenic periappendiceal fat is absent, the likelihood of
appendicitis is low (present in 9% of patients in our popula-
tion). In this setting, the presence of a single diagnostic crite-
rion should not be considered diagnostic of acute appendicitis.
Appendiceal diameter $7 mm can be helpful to segregate
patients in this group, identifying those with a slightly higher
rate of appendicitis in the absence of periappendiceal fat
inflammation (20% rate of appendicitis in our population),
but even this criterion should not be applied in isolation.Mul-
tiple abnormal findings, however, may suggest the presence of
appendicitis even in the absence of abnormal periappendiceal
fat. Other criteria that should be given particular weight
include loss of normal mural stratification in the appendix
and the presence of an appendicolith. In all cases of suspected
acute appendicitis, we caution against overreliance on user-
dependent factors such as compressibility of the appendix or
appendiceal vascularity.
In conclusion, based on our findings and a review of the lit-
erature, periappendiceal fat infiltration appears to be the most
important single diagnostic criterion for acute appendicitis.
Strict application of other diagnostic criteria such as appendi-
ceal diameter, which is recently being called into question,
should be avoided.
ACKNOWLEDGMENT
The authors thankMike D. Smith, PhD, for statistical support.
REFERENCES
1. Puylaert JB. Acute appendicitis: US evaluation using graded compression.
Radiology 1986; 158:355–360.
2. Rosen MP, Ding A, Blake MA, et al. ACR Appropriateness Criteria� right
lower quadrant pain-suspected appendicitis. J Am Coll Radiol 2011; 8:
749–755.
3. Baldisserotto M, Peletti AB. Is colour Doppler sonography a good method
to differentiate normal and abnormal appendices in children? Clin Radiol
2007; 62:365–369.
4. Je BK, Kim SB, Lee SH, et al. Diagnostic value of maximal-outer-diameter
andmaximal-mural-thickness in use of ultrasound for acute appendicitis in
children. World J Gastroenterol 2009; 15:2900–2903.
5. Peletti AB, Baldisserotto M. Optimizing US examination to detect the nor-
mal and abnormal appendix in children. Pediatr Radiol 2006; 36:
1171–1176.
6. Wiersma F, Sramek A, Holscher HC. US features of the normal appendix
and surrounding area in children. Radiology 2005; 235:1018–1022.
7. Wiersma F, Toorenvliet BR, Bloem JL, et al. US examination of the appen-
dix in children with suspected appendicitis: the additional value of secon-
dary signs. Eur Radiol 2009; 19:455–461.
8. Goldin AB, Khanna P, Thapa M, et al. Revised ultrasound criteria for ap-
pendicitis in children improve diagnostic accuracy. Pediatr Radiol 2011;
41:993–999.
9. Trout AT, SanchezR, Ladino-TorresMF, et al. A critical evaluation of US for
the diagnosis of pediatric acute appendicitis in a real-life setting: how can
we improve the diagnostic value of sonography? Pediatr Radiol 2012; 42:
813–823.
1393
TROUT ET AL Academic Radiology, Vol 19, No 11, November 2012
10. Baldisserotto M, Marchiori E. Accuracy of noncompressive sonography of
children with appendicitis according to the potential positions of the
appendix. AJR Am J Roentgenol 2000; 175:1387–1392.
11. Kaiser S, Frenckner B, Jorulf HK. Suspected appendicitis in children: US
and CT—a prospective randomized study. Radiology 2002; 223:
633–638.
12. Kessler N, Cyteval C, Gallix B, et al. Appendicitis: evaluation of sensitivity,
specificity, and predictive values of US, Doppler US, and laboratory find-
ings. Radiology 2004; 230:472–478.
13. Ang A, Chong NK, Daneman A. Pediatric appendicitis in ‘‘real-time’’: the
value of sonography in diagnosis and treatment. Pediatr Emerg Care
2001; 17:334–340.
14. Hahn HB, Hoepner FU, Kalle T, et al. Sonography of acute appendicitis in
children: 7 years experience. Pediatr Radiol 1998; 28:147–151.
15. Patriquin HB, Garcier JM, Lafortune M, et al. Appendicitis in children and
young adults: Doppler sonographic-pathologic correlation. AJR Am J
Roentgenol 1996; 166:629–633.
16. Quillin SP, Siegel MJ. Appendicitis: efficacy of color Doppler sonography.
Radiology 1994; 191:557–560.
17. Sivit CJ. Diagnosis of acute appendicitis in children: spectrum of sono-
graphic findings. AJR Am J Roentgenol 1993; 161:147–152.
18. Berrocal T, delPozo G. Imaging in pediatric gastrointestinal emergen-
cies. In: Devos AS, Blickman JG, eds. Radiological Imaging of the
Digestive Tract in Infants and Children. Secaucus, NJ: Springer-Verlag,
2007; 1–79.
19. Jeffrey RB Jr, Laing FC, Townsend RR. Acute appendicitis: sonographic
criteria based on 250 cases. Radiology 1988; 167:327–329.
20. Ozel A, Orhan UP, Akdana B, et al. Sonographic appearance of the normal
appendix in children. Journal of clinical ultrasound. JCU 2011; 39:
183–186.
1394
21. Park NH, Park CS, Lee EJ, et al. Ultrasonographic findings identifying the
faecal-impacted appendix: differential findings with acute appendicitis. Br
J Radiol 2007; 80:872–877.
22. Rettenbacher T, Hollerweger A, Macheiner P, et al. Outer diameter of the
vermiform appendix as a sign of acute appendicitis: evaluation at US. Ra-
diology 2001; 218:757–762.
23. Simonovsky V. Sonographic detection of normal and abnormal appendix.
Clin Radiol 1999; 54:533–539.
24. van Randen A, Lameris W, van Es HW, et al. Profiles of US and CT imaging
features with a high probability of appendicitis. Eur Radiol 2010; 20:
1657–1666.
25. Spear R, Kimmey MB, Wang KY, et al. Appendiceal US scans: histologic
correlation. Radiology 1992; 183:831–834.
26. Franke C, Bohner H, Yang Q, et al. Ultrasonography for diagnosis of acute
appendicitis: results of a prospective multicenter trial. Acute Abdominal
Pain Study Group. World J Surg 1999; 23:141–146.
27. Incesu L, Yazicioglu AK, SelcukMB, et al. Contrast-enhanced power Dop-
pler US in the diagnosis of acute appendicitis. Eur J Radiol 2004; 50:
201–209.
28. Noguchi T, Yoshimitsu K, Yoshida M. Periappendiceal hyperechoic struc-
ture on sonography: a sign of severe appendicitis. J UltrasoundMed 2005;
24:323–327. quiz 328–330.
29. Gill B, Cudmore RE. Significance of faecoliths in the diagnosis of acute
appendicitis. Br J Surg 1975; 62:535–536.
30. Chang YJ, Kong MS, Hsia SH, et al. Usefulness of ultrasonography in
acute appendicitis in early childhood. J Pediatr Gastroenterol Nutr 2007;
44:592–595.
31. Ramachandran P, Sivit CJ, Newman KD, et al. Ultrasonography as an
adjunct in the diagnosis of acute appendicitis: a 4-year experience. J Pe-
diatr Surg 1996; 31:164–167. discussion 167–169.