Pediatric Appendiceal Ultrasound

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281081662

Pediatric appendiceal ultrasound: accuracy, determinacy and clinical


outcomes

Article in Pediatric Radiology · August 2015


DOI: 10.1007/s00247-015-3432-7 · Source: PubMed

CITATIONS READS

84 244

8 authors, including:

Larry A Binkovitz Kyle Unsdorfer


Mayo Foundation for Medical Education and Research Northeast Ohio Medical University
66 PUBLICATIONS 1,704 CITATIONS 6 PUBLICATIONS 109 CITATIONS

SEE PROFILE SEE PROFILE

Prabin Thapa
Mayo Foundation for Medical Education and Research
218 PUBLICATIONS 6,064 CITATIONS

SEE PROFILE

All content following this page was uploaded by Prabin Thapa on 13 October 2017.

The user has requested enhancement of the downloaded file.


Pediatr Radiol (2015) 45:1934–1944
DOI 10.1007/s00247-015-3432-7

ORIGINAL ARTICLE

Pediatric appendiceal ultrasound: accuracy, determinacy


and clinical outcomes
Larry A. Binkovitz 1 & Kyle M. L. Unsdorfer 2 & Prabin Thapa 3 & Amy B. Kolbe 1 &
Nathan C. Hull 1 & Shannon N. Zingula 1 & Kristen B. Thomas 1 & James L. Homme 4

Received: 14 February 2015 / Revised: 8 May 2015 / Accepted: 7 July 2015 / Published online: 18 August 2015
# Springer-Verlag Berlin Heidelberg 2015

Abstract appendicitis was 18.5% (146/790). There were 109 true-


Background Ultrasonography is considered the most appro- positive, 440 true-negative, 17 false-positive, 6 false-neg-
priate initial imaging study in the evaluation of acute appen- ative, 218 equivocal and 41 technically inadequate US
dicitis in children but has recently come under criticism with studies. A definitive interpretation was made in 72% of
reports of low specificity and high indeterminate study rates, the studies, with an accuracy, sensitivity and specificity of
particularly when used in obese patients and patients early in 0.960, 0.948 and 0.963, respectively. No patient or system
the course of their disease, or when performed by factors significantly affected US accuracy. Indeterminate
sonographers with limited experience. studies (28%) had significantly higher CT utilization
Objective To (1) assess the impact of patient factors (gen- (46% vs. 11%) and normal appendectomy rates (6.9%
der, age, body mass index, and symptom duration) and vs. 3.5%).
system factors (call status or year of exam) on pediatric Conclusion US should be the initial imaging study of
appendiceal US accuracy and indeterminate study rate, choice for pediatric appendicitis. When a definitive inter-
(2) assess the impact of indeterminate study results on pretation was given, the accuracy was 96%, was indepen-
follow-up CT and negative laparotomy rates and (3) pres- dent of patient and system factors and resulted in reduced
ent strategies to reduce the rate of indeterminate US studies follow-up CTs and negative laparotomies. Accuracy can be
and improve accuracy. increased by requiring the presence of periappendiceal in-
Materials and methods We retrospectively reviewed all US flammatory changes prior to interpreting a mildly
reports performed for the assessment of acute appendicitis in distended appendix as positive for acute appendicitis. The
children <18 years old at Mayo Clinic Rochester from January indeterminate study rate can be reduced by not requiring
2010 to June 2014. visualization of the normal appendix for the exclusion of
Results A total of 790 US examinations were performed acute appendicitis.
in 452 girls (57%) and 338 boys (43%). The prevalence of
Keywords Appendicitis . Appendix . Children . Diagnostic
accuracy . Ultrasound
* Larry A. Binkovitz
[email protected]
Introduction
1
Department of Diagnostic Radiology, Mayo Clinic,
200 First St. SW, Rochester, MN 55905, USA Acute appendicitis is the most common cause of abdominal
2
NEOMED, pain requiring emergency surgery in children [1]. The
Rootstown, OH, USA signs and symptoms of acute appendicitis include anorex-
3
Department of Biostatistics, Mayo Clinic, ia, nausea, fever, leukocytosis and periumbilical pain mi-
Rochester, MN, USA grating to the right lower quadrant (RLQ). However the
4
Department of Emergency Medicine, Mayo Clinic, clinical presentation can vary considerably, making prompt
Rochester, MN, USA diagnosis challenging and resulting in prolonged
Pediatr Radiol (2015) 45:1934–1944 1935

emergency department evaluations, delayed or unneces- US, presence of right-lower-quadrant (RLQ) pain, wheth-
sary surgeries and increased perforation rates [2–4]. Imag- er sonography was performed on-call or off-call, US re-
ing can play an important role in evaluation of acute ap- sults, pathological findings and clinical follow-up for chil-
pendicitis, especially when classic signs and symptoms are dren who did not undergo surgery.
absent [5]. Imaging with US, CT and more recently MRI
have been used [6, 7], and there continues to be a debate Imaging protocol
over which modality is best [8–10]. US evaluation for
acute appendicitis was first described by Puylaert [11] The US examinations were performed supine with graded
and was quickly applied to pediatric patients [12]. The compression of the RLQ using the most current GE Logiq
American College of Radiology considers US to be the platform scanners, presently E-9 (GE Healthcare, Fair-
most appropriate imaging modality for the initial evalua- field, CT) and high-frequency linear and sector probes.
tion of acute appendicitis because of its lack of ionizing Static transverse and longitudinal images as well as mul-
radiation [13], high sensitivity [14] and low cost [15]. tiple cine clips throughout the RLQ with and without
However CT has demonstrated higher specificity (94% graded compression are routinely obtained. The imaging
vs. 88%) with similar sensitivity (95% vs. 94%) compared protocol did not change during the course of the study.
to US [14]. MRI for acute appendicitis has also shown Studies obtained during routine hours (off-call, Monday–
excellent sensitivity and specificity [6, 7]. There is a wide Friday 7 am–5 pm) were performed by any one of more
range of sensitivities and specificities reported with US than 50 general sonographers. These studies are reviewed
[14, 16], and this is thought to be a result of operator by one of several pediatric radiologists with a Certificate
dependence and patient and system factors. It has been of Added Qualification, prior to patient discharge. If the
suggested that CT might be a better first-line imaging normal appendix was not visualized, the radiologist would
modality in obese patients and those with short symptom typically rescan the child. The reviewing radiologist may
duration [17, 18]. The diagnostic yield of appendiceal US also choose to rescan patients to clarify imaging findings.
has been found to be poor when performed by technolo- Studies conducted outside of the routine hours (on-call)
gists with little pediatric experience [16, 19], at institu- were performed by one of four on-call sonographers with
tions where it is not the initial imaging study and when an average of more than 20 years of sonographic
the appendiceal visualization rate is low [16]. Thus in experience.
some practices and for some patients, CT may be chosen
over US for the initial assessment of acute appendicitis. US report categorization
The purpose of this report is to assess the impact of patient
and system factors on the accuracy and indeterminate study We categorized the US reports in the patients’ record at
rate on pediatric appendiceal US and to investigate how inde- the time of the decision to operate or discharge. These
terminate results impact patient outcomes. include six revised reports of the 581 on-call studies ini-
tially interpreted by radiology residents and subsequently
reviewed by a staff pediatric radiologist. The US reports
Materials and methods were retrospectively categorized (by K.M.L.U. and
L.A.B.) as follows:
Patient population
& US-0, no appendicitis: Normal appendix visualized from
Our institutional review board approved this retrospective base to tip. Rarely, a small portion of the appendix might
review of all pediatric (<18 years old) abdominal US re- have been obscured by cecal gas but the radiologist’s im-
ports performed at Mayo Clinic Rochester between Janu- pression was no appendicitis. There was no recommenda-
ary 2010 and June 2014. We included all studies ordered tion for CT or clinical follow-up.
for the evaluation of acute abdominal symptoms. Children & US-1, no appendicitis: Appendix not visualized. No sec-
were identified for inclusion based on a search of all ab- ondary signs of appendicitis, including periappendiceal fat
dominal US reports that contained the words “appendix” thickening, increased echogenicity or hyperemia. The
or “appendicitis.” Children were excluded if they previ- presence of pain with transducer pressure was evaluated
ously had an appendectomy or a CT scan for acute appen- but not incorporated into the US categorization because of
dicitis prior to the US, if the US was performed at an the variability in patients’ ability to report pain and the
outside institution, and if the child did not undergo an highly subjective nature of this finding. There was no
appendectomy and had no clinical follow-up after dis- recommendation for CT or clinical follow-up.
charge. We recorded patient age, gender, height, weight, & US-2, appendicitis: Abnormal appendix visualized and
body mass index (BMI), symptom duration at the time of interpreted as consistent with acute appendicitis. There
1936 Pediatr Radiol (2015) 45:1934–1944

was no recommendation for CT or clinical follow-up. for evaluation of continuous variables such as BMI-FAP;
Standard signs for appendicitis were used and included: the chi square and Fisher exact tests were used for com-
appendiceal diameter greater than 6 mm and non-com- parison of discrete variables, such as accuracy off-call vs.
pressible, loss of appendiceal wall architecture, hyperemia on-call. Statistical significance was set at P < 0.05. The
and presence of an appendicolith as well as presence of the following definitions were used for assessment of diag-
secondary signs listed in US-1. nostic performance:
& US-3, equivocal: Appendix visualized but the appearance
was thought to be indeterminate. Includes cases where the & True positive (TP): US-2 and P-1, P-2 or P-3
appendiceal diameter was >6 mm or non-compressible or & True negative (TN): US-0 or US-1 and P-0 or P-X
there was RLQ fluid but the studies were not thought to be & False positive (FP): US-2 and P-0 or P-X
diagnostic of acute appendicitis. CT or clinical follow-up & False negative (FN): US-0 or US-1 and P-1, P-2 or P-3
was sometimes recommended. & Indeterminate-positive (IP): US-3, US-4 or US-5 and
& US-4, equivocal: Appendix not visualized and acute P-1, P-2 or P-3
appendicitis not ruled out. These reports differ from & Indeterminate-negative (IN): US-3, US-4 or US-5 and
US-1 in that the initial radiologist’s report did not P-0 or P-X
specifically describe the absence of periappendiceal
findings. CT or clinical follow-up was sometimes Determination of the diagnostic yield of US for acute
recommended. appendicitis was performed using the intention-to-
& US-5, technically inadequate: The RLQ was insufficiently diagnose calculations [20], including all 790 US studies,
evaluated because of bowel gas. as follows: accuracy=(TP+TN)/(TP+TN+FP+FN+IP+IN),
sensitivity=TP/(TP+FN+IP) and specificity=TN/(TN+FP+
IN). A second set of calculations was performed exclud-
Pathology report categorization ing the equivocal (US-3 and US-4) or technically inade-
quate (US-5) studies. This was done because these inde-
Pathology reports were categorized as follows based on the terminate studies do not contribute to the decision to
initial reading: operate or discharge in our practice [21]. The diagnostic
performance of US was then calculated as follows:
& P-0, negative: Normal appendix. sensitivity=TP/(TP+FN), specificity=TN/(TN+FP) and
& P-1, minimal/early appendicitis: The terms minimal or accuracy=(TP+TN)/(TP+TN+FP+FN).
early were used in the report. This typically indicates neu-
trophilic infiltration, predominately of the appendiceal
epithelium.
& P-2, appendicitis: Abnormal with detailed description of Results
microscopic findings of acute appendicitis with neutro-
philic infiltration of the appendiceal epithelial and muscu- During the study period 849 US reports included the
lar layers. terms appendix or appendicitis. Studies were excluded
& P-3, perforated appendix: Acute appendicitis with gross as follows: no clinical follow-up (n=20), duplicate or-
perforation of the appendiceal wall. ders (n=17), not performed at Mayo Clinic Rochester
& P-X, patients who did not undergo surgery and did not (n=11), performed to clarify CT findings (n=6) or per-
have acute appendicitis on clinical follow-up. formed in children with a prior appendectomy (n=5). A
total of 790 examinations met inclusion criteria and in-
cluded 452 girls (57%) and 338 boys (43%). The mean
Statistical analysis age was 10.4+/−4.5 years. BMI was available in 622
(79%) children and averaged 19.8+/−4.8. BMI-FAP is
Using SAS© (version 9.2; SAS Institute, Cary, NC) for defined for children older than 2 years and was avail-
analysis, group statistics are presented as mean/standard able in 571 (72%) children, distributed as follows: 3%
deviation (SD) for continuous variables and n (%) for underweight (<5th percentile) and 23% overweight
categorical variables. The Cochran-Armitage trend test (>85th percentile) indicating a slightly overweight bias
was used to detect trends in accuracy, appendiceal visual- in our study group [22]. Most studies (581, or 73%)
ization and determinacy across years of study as well as were performed on-call; 209 (27%) studies were per-
accuracy among groups categorized by age, BMI for age formed off-call. There were 146 cases of acute appen-
percentile (BMI-FAP), symptom duration, pathology dicitis for an overall prevalence of 18.5%, which was
grade or symptom duration. The Student’s t-test was used similar for off-call and on-call studies, with off-call
Pediatr Radiol (2015) 45:1934–1944 1937

Table 1 Group 1 (definitive US) vs. group 2 (indeterminate US) Table 3 Diagnostic yield of US

Group 1 Group 2 Total P-value Definitive studies All studies groups 1


(US=0, 1, 2) (US=3, 4, 5) group 1 n=572 and 2 n=790

n 572 218 790 Sensitivity 0.948 0.747


Age in years 0.2 Specificity 0.963 0.683
Mean (SD) 10.3 (4.4) 10.7 (4.7) 10.4 (4.5) Accuracy 0.960 0.747
Median 11 11 11 Positive predictive value 0.865
Range 0.6–17.0 0.6–17 0.6–17 Negative predictive value 0.986
Gender 0.97
Male 245 (43%) 93 (43%) 338 (43%)
Female 327 (57%) 125 (57%) 452 (57%)
BMI 0.15 (US-3 or US-4, 22.4%) or technically inadequate (US-5,
n 449 173 622 5.2%). The prevalence of acute appendicitis in groups 1
Mean (SD) 19.5 (4.5) 20.4 (5.7) 19.8 (4.8) and 2 was similar, 20% vs. 14%, respectively, or P=0.9.
Range 12.0–37.0 12.0–52.0 12.0–52.0 The demographic data of groups 1 and 2 are summarized
Call status 0.03 in Table 1 and the US score distribution and outcomes are
Off-call 139 (24%) 70 (32%) 209 (26%) listed in Table 2. The diagnostic performance of US for
On-call 433 (76%) 148 (68%) 581 (74%) group 1 and for all patients is compared in Table 3. There
Path grade 0.59 were no statistically significant differences between the
1 8 (7%) 3 (11%) 11 (7%) groups for any patient factor, including age, gender,
2 88 (76%) 21 (78%) 109 (75%) BMI, symptom duration or pathological grade of appen-
3 20 (17%) 3 (11%) 26 (18%) dicitis (Figs. 1, 2, 3, and 4). Specifically, neither young
age nor short symptom duration resulted in decreased US
BMI body mass index, Path pathological, SD standard deviation accuracy. Accuracy was similar between off-call and on-
call studies (Fig. 5). However, on-call studies were more
likely to yield a definitive diagnosis, P=0.03.
prevalence of 17.7% (37/209) and on-call prevalence of Of the 218 indeterminate studies, the US report specif-
18.7% (109/581), P=0.74. ically indicated that a follow-up CT should be obtained if
clinically indicated in only 46 patients (21%), reflecting
Accuracy and determinacy variability in follow-up recommendations among the
interpreting pediatric radiologists. A follow-up CT was
For purposes of analysis, we divided US reports into two four times less likely to be obtained when a definitive
groups. Group 1 reports gave a definitive diagnosis and US diagnosis was given, (group 1, 62/572, 11%) than
included 572 (72.4%) studies with either no acute appen- when an indeterminate diagnosis was given (group 2,
dicitis (US-0 or US-1) or acute appendicitis (US-2); group 101/218, 46%), P<0.0001. Normal appendectomies were
2 reports were indeterminate studies, and this group in- half as frequent in group 1, 20/572 (3.5%) as in group 2,
cluded 218 (27.6%) studies that were either equivocal 15/218 (6.9%), P=0.04.

Table 2 US score distribution and final diagnosis


a
Group 1 (Definitive) n True positive True negative False positive False negative Acute appendicitis
US-0 205 204 1
US-1 241 236 5
US-2 126 109 17
Total 572 109a 440 17 6a 20% (115/572)
Group 2 (Indeterminate) n Indeterminate positive Indeterminate negative
US-3 37 9 28
US-4 140 18 122
US-5 41 4 37
Total 218 31a 187 14% (31/218)
a
Total acute appendicitis (TP+FP+IP)=146, prevalence=18.5% (146/790)
1938 Pediatr Radiol (2015) 45:1934–1944

Fig. 3 US accuracy and symptom duration. Graph shows that US


accuracy is independent of symptom duration; similar accuracies were
found among all groups from symptom durations of less than 12 h to
greater than 48 h

Appendiceal visualization

The data were also analyzed based on whether the appendix


was visualized (US-0, US-2, US-3, n=368, 49%) or not (US-1
and US-4, n=381, 51%) (Table 4). There were no statistically
significant differences in the patient factors (age, gender and
BMI) between children in whom the appendix was visualized
and those in whom it was not. Visualization was more likely in
children with acute appendicitis (109/146, 75%) than without
(204/603, 34%), P<0.0001, and more likely in studies per-
formed on-call (288/551, 52%) than off-call (80/198, 40%),
Fig. 1 US accuracy according to age group and gender. a, b Graphs
show that US accuracy is independent of patient age (a) and gender (b)
P=0.004.
Studies without appendiceal visualization and without sec-
ondary signs of acute appendicitis (US-1, n = 241) were

Fig. 2 US accuracy and body mass index. Graph shows that US accuracy Fig. 4 US accuracy and pathological grade of appendicitis. Graph shows
is independent of body mass index for age percentile (BMI-FAP). The that US accuracy is independent of pathological grade. The trend of
study group was skewed slightly toward overweight, with 22% of the higher accuracy with increasing severity of appendicitis did not reach
children above the 85th percentile statistical significance
Pediatr Radiol (2015) 45:1934–1944 1939

and was stool-filled but the pathological impression was acute


appendicitis without specific mention of inflammatory chang-
es in the appendiceal wall. There were 17 false-positive cases
where the appendix was thought to be abnormal (US-2) but
was determined to be normal based on pathological (n=10) or
clinical (n=7) follow-up. Fifteen (88%) of these 17 false-
positive US studies were interpreted as abnormal based solely
on appendiceal diameter greater than 6 mm and lack of com-
pressibility but without secondary signs of acute appendicitis.
The remaining two US studies were normal aside from mildly
increased appendiceal diameter and vascularity. Pathological
analysis of the 10 false-positive cases that underwent appen-
dectomy revealed lymphoid hyperplasia in three, one of which
had serosal adhesions. The remaining seven showed no path-
Fig. 5 US accuracy according to call status. Graph shows that US
ological abnormality.
accuracy is independent of call status (on-call or not on-call) of the
examination. Note that most studies were performed on-call
Discussion

considered negative [23, 24]. In this group, there were five US accuracy and determinacy
false-negative studies, yielding a 98% negative predictive val-
ue. The decisions to operate in the five false-negative children Despite decades of clinical experience with US for the evalu-
were based on CT in one and on clinical findings in four. ation of acute appendicitis, controversy remains. Its sensitivity
There was a single false-negative case of acute appendicitis and specificity have been reported to be approximately 90%
where the appendix was visualized and thought to be normal [14], but certain patient and system factors have been reported
(US-0). The resected appendix measured 5 mm in diameter to reduce its accuracy [17–19]. Its diagnostic yield has been

Table 4 Appendiceal
visualization Visualized (US 0, 2, 3) Not visualized (US 1, 4) Totala P-value

Age (years)
n 368 381 749 0.27
Mean (SD) 10.6 (4.3) 10.2 (4.6)
Range 0.0–17.0 0.0–17.0
Gender
Male 158 160 318 (42%) 0.58
Female 208 223 431 (58%)
Call status
On-call 288 263 551 (74%) 0.004
Off-call 80 118 198 (26%)
BMI
n 290 301 591 0.23
Mean (SD) 19.4 (4.2) 20.1 (5.3) 19.7 (4.8)
Range 12.0–36.0 12.0–52.0 12.0–52.0
BMI-FAP
n 283 287 570 0.08
Mean (SD) 61.1 (26.5) 64.1 (28.5) 62.6 (27.6)
Median 64 71 66.5
Range 3–99 3–99 3–99

Statistically significant P-values (<0.05) are in bold


BMI-FAP body mass index for age percentile, SD standard deviation
a
excludes 41 technically inadequate studies (US-5)
1940 Pediatr Radiol (2015) 45:1934–1944

reported to be well below reported accuracies because of rel- confidently interpreted as negative for appendicitis. Had all
atively high rates of indeterminate results [21]. Although such studies been re-categorized, our negative predictive val-
appendiceal US exams may be indeterminate in clinical prac- ue would have decreased from 0.979 to 0.940, indicating a
tice, the definition of what makes a study indeterminate is not slight increase in false-negative reports. This would substan-
standard. Lack of complete visualization of the appendix has tially reduce the indeterminate rate but could also result in a
been considered to be indeterminate for acute appendicitis, small increase in the number of false-negative studies. In these
and with this definition Krishnamoorthi et al. [25] reported a patients our false-negative rate was 2%, similar to other au-
47% indeterminate study rate. Prior to determining the accu- thors [23, 26], and was not significantly different from the
racy and diagnostic yield of US in acute appendicitis, the false-negative rate when a normal appendix was visualized,
decision of how to categorize indeterminate studies must be 0.5%. It is important to note that none of the five children with
made because the manner in which these studies are included false-negative US studies was discharged from the emergency
in statistical analyses has a great effect on the diagnostic yield department with unrecognized acute appendicitis; all were
of US. Authors may choose a best-case scenario (indetermi- further assessed and underwent an appendectomy where acute
nate studies included as true positive and true negative) or a appendicitis was confirmed. With these findings, we have
worst-case scenario (studies included as false positive and since implemented a US report template that includes docu-
false negative) or both [14, 16]. Trout et al. [19] included mentation of the presence or absence of secondary signs of
exams that were indeterminate as “positive” because at their acute appendicitis, and in the absence of such findings the
institution, sonography that did not exclude the diagnosis of study is considered negative for acute appendicitis. An addi-
acute appendicitis could result in additional imaging or appen- tional strategy to decrease indeterminacy is the adoption of a
dectomy. This led to an increase in the number of false- binary reporting template that requires the interpreting radiol-
positive studies and decreased accuracy [19]. ogist to categorize the study as definitively positive or nega-
We analyzed the data two ways. In the first analysis, inde- tive for acute appendicitis. Using this strategy, Nielsen et al.
terminate studies were excluded and US was shown to have an [27] found their non-diagnostic rate to decrease from 48% to
accuracy of 96%. This analysis reflects the fact that in our 0.1% and found an increase in both sensitivity (67% to 92%)
practice indeterminate studies are not viewed as positive or and negative predictive value (89% to 98%).
negative and do not directly result in patient discharge, admis-
sion or operation [20]. The second analysis used the intention- Increasing accuracy
to-treat method and included indeterminate studies and
assessed the diagnostic yield of appendiceal US [21]. Our Although our accuracy rate of 96% is high when a definitive
indeterminate study rate of 28% is similar to other reports diagnosis is made, increases might be achieved with changes
[17, 18]. Including indeterminate studies in the statistical anal- to the standard US technique of supine RLQ graded compres-
ysis conveys to clinicians the proportion of patients with sion that increase appendiceal visualization [28, 29]. Increased
suspected acute appendicitis whose US study has not yielded visualization might allow re-categorization of US-4 cases to
definitive information and who need additional clinical or im- US-2 and increase the number of true-positive studies. How-
aging follow-up. ever, increased appendiceal visualization might not result in
improved accuracy [16]. Our appendiceal visualization rate
Decreasing indeterminacy increased steadily throughout the 5-year study period, but
US accuracy and the proportion of definitive US diagnoses
From a clinician’s standpoint, the need to reduce the number did not (Fig. 6). These observations suggest that the relation-
of indeterminate US studies is significant. If a relatively high ship between appendiceal visualization and accuracy is com-
proportion of US evaluations for acute appendicitis is indeter- plex. This may be because of the acute appendicitis cases that
minate, CT might be chosen as the initial imaging study or are always non-visualized (deep in the pelvis) or have minimal
patients might undergo unnecessary laparotomies. A diagnos- histological changes are still deemed acute appendicitis by the
tic test, no matter how accurate when a definitive diagnosis is pathologist. It is important to note that a reduction in the num-
given, is not clinically useful if the indeterminate rate is too ber of false-positive exams could be achieved if the US criteria
high. We have identified several strategies that can be used to for diagnosing acute appendicitis required the demonstration
reduce indeterminacy. The biggest reduction in the number of of thickening and increased echogenicity of the
indeterminate results with an increase in true-negative studies periappendiceal fat and did not rely solely on an increased
might be achievable by increasing the emphasis placed on the appendiceal diameter [24]. The US diagnosis of acute appen-
significance of the presence or absence of secondary signs of dicitis in 15 of our 17 false-positive studies relied solely on a
acute appendicitis. When a thorough RLQ US examination minimal increase of appendiceal diameter above 6 mm with-
has been performed, the absence of appendiceal visualization out any periappendiceal inflammatory changes. Appendiceal
and any secondary findings for acute appendicitis can be diameter in isolation is a poor predictor of acute appendicitis,
Pediatr Radiol (2015) 45:1934–1944 1941

with significant overlap between normal and abnormal appen-


dices [24].

US accuracy: patient factors

Age

The utility of US for acute appendicitis in children younger


than 3 years was evaluated by Chang et al. [30], who reported
a sensitivity of 95% and an accuracy of 93%. We also found
that US was highly accurate regardless of age throughout
childhood, ranging from 1.000 in the subgroup of children
younger than 2 years to 0.970 for children 2–6 years of age.

Body habitus

Several authors have examined the impact of body habitus on


appendiceal US accuracy. Schuh et al. [17] found that over-
weight children were twice as likely to have inaccurate US
results as normal-weight children. Sulowski et al. [31] found
that obese children were three times more likely to require a
follow-up CT after the initial US than normal-weight children
despite similar incidence and severity of acute appendicitis
and clinical outcomes. Abo et al. [32] found that as BMI
increases in children, the sensitivity of US decreases. Howev-
er in a study of 122 children with surgically proven acute
appendicitis, Yigiter et al. [33] demonstrated no statistical dif-
ference in US accuracy among underweight, normal-weight
and overweight patients. They concluded that body habitus
did not affect US accuracy in the evaluation of the inflamed
appendix, but their study did not include normal appendices.
In a study of 963 children with an average weight percentile of
93%, Trout et al. [19] concluded that body habitus was the
only patient factor that contributed to their low appendiceal
visualization rate. In contrast, we found no statistical differ-
ence in accuracy across BMI-FAP. The weight distribution of
our patients was closer to normal than that reported by Trout
et al. [19], with an average BMI-FAP of 67 and with only 22%
of our patients being overweight, and this may contribute to
our differing results.

Symptom duration

Assessment of the impact of symptom duration on the diag-


nostic yield of appendiceal US in children has received rela-
Fig. 6 US accuracy and appendix visualization. a, b Note that US tively little attention. In a multicenter study, Bachur et al. [18]
accuracy was independent of year performed (a), but appendix found that as the duration of abdominal pain increased, US
visualization increased over the years of the study (b). c The rate of
definitive diagnoses was stable throughout the years of study. These sensitivity increased; sensitivity was 81% at <8 h and 96%
findings suggest a complex relationship between appendiceal >48 h, P<0.001. The authors recommended that “clinicians
visualization and US accuracy and determinacy should not rely on ultrasonography early in the course of
1942 Pediatr Radiol (2015) 45:1934–1944

illness.” Our US accuracy was unaffected by symptom dura- frequently than general sonographers [19]. We found no dif-
tion, indicating that additional imaging confirmation of nega- ference in the accuracy of off-call and on-call US examina-
tive US findings in patients with symptoms less than 24 h tions, at 99% and 95% respectively, P=0.08. However, we
should not be routinely obtained. found that on-call studies had higher rates of appendiceal vi-
sualization (52% vs. 40%, P=0.0042) and determinacy (75%
Pathological grade vs. 67%, P=0.03). This likely reflects the increased experi-
ence of the on-call sonographers at our institution.
The relationship between the diagnostic yield of
appendiceal US and the pathological grade of acute ap- The negative clinical impact of indeterminate studies
pendicitis might be complicated by the relatively subjec-
tive nature of pathological grading. Bachur et al. [18] One of the major reasons for imaging children who are
found that US sensitivity was greater for perforated than suspected of having acute appendicitis is to reduce the
non-perforated acute appendicitis. We had similar sensi- need for CT imaging and the negative appendectomy rate.
tivities for early/mild appendicitis (88%), appendicitis The negative appendectomy rate based on the clinical eval-
(94%) and perforated (100%) appendicitis, P=0.37. uation alone can be reduced with the addition of US.
Karakas et al. [35] found the rate to decrease from 13%
US accuracy: system factors and appendiceal visualization to 8% with the addition of US. When using a standardized
and validated US reporting template that required the
Trout et al. [19] noted increasing US accuracy and rates of interpreting radiologist to make a definitive diagnosis,
appendiceal visualization in the later years of their 5-year ret- Nielsen et al. [27] found the rate of negative appendecto-
rospective review, suggesting that the progressively improv- mies to be reduced four-fold and the follow-up CT rate
ing performance might be a result of increasing technical and reduced five-fold compared to studies that were reported
interpretive expertise attained during the study. Specifically, without following the template and could yield indetermi-
appendiceal visualization rates increased from 6% in the first nate results. Our results confirm these findings; we found a
year of their review to 29% in the fifth year. We also found an two-fold reduction in the negative appendectomy rate and
increasing appendiceal visualization rate over time, 32% visu- a four-fold reduction in the rate of follow-up CT studies
alization in the first year and 51% in the fifth, P=0.01. A wide when definitive US interpretations were made.
range of pediatric appendiceal US visualization rates have
been reported. Lee et al. [34] noted a 98% visualization rate Limitations
in patients of all ages with the use of posterior manual
compression in addition to standard supine compression, There are several limitations to our study. First, the US scores
but this high level has not been repeated in North America. were based on the interpretations of the US studies by several
Rompel et al. [29] increased their appendiceal visualization pediatric radiologists at the time care was given and were
rate from 86% to 93% with the addition of harmonic im- reviewed and retrospectively categorized. The vast majority
aging to the standard protocol. Mittal et al. [16] found the of the initial interpretations clearly fell into one of the US
visualization rate to be lower at hospitals where US is used categories but approximately 40 studies did not; this occurred
less frequently or is less available (25%) than at hospitals almost entirely when the US interpretation indicated that there
where it is available at all times and is the primary imaging were no secondary signs of acute appendicitis but there was a
test for acute appendicitis (56%). Chang et al. [28] in- lack of visualization of the normal appendix. These studies
creased the appendiceal visualization rate from 35%, when were initially interpreted as indeterminate but were catego-
they used supine graded compression US, to 59% when left rized in this study as negative for acute appendicitis, US-1,
posterior oblique imaging of the right flank was added to rather than indeterminate, US-4.
supine imaging. A second potential limitation is the lack of dedicated
Trout et al. [19] found that sensitivity and accuracy of follow-up in children who did not have surgery for acute ap-
appendiceal US was greater during off-call hours (when stud- pendicitis. If their clinical charts indicated any form of follow-
ies were performed by dedicated pediatric sonographers and up after the initial acute appendicitis evaluation, with no indi-
staffed by pediatric radiologists) than on-call hours (when cation of an interval appendectomy, the children were consid-
studies were performed by general sonographers and staffed ered negative for acute appendicitis. The lack of dedicated
by non-pediatric radiologists). They also found that the dedi- follow-up, such as phone calls at 24 h after discharge, could
cated pediatric sonographers identified the appendix more result in under-reporting of missed cases of acute appendicitis.
Pediatr Radiol (2015) 45:1934–1944 1943

This potential limitation is mitigated by the nature of our prac- Conclusion


tice. Ninety-six percent of the study patients were local, 63%
were county residents and 33% lived in adjacent counties. In our study US accuracy was not influenced by patient or
Mayo Clinic Rochester is the only pediatric care center for system factors (age, gender, BMI, symptom duration, severity
the entire tri-state region of southern Minnesota, northern Io- of acute appendicitis, or call status of the study). With experi-
wa and southwest Wisconsin and has an extensive regional enced sonographers and radiologists, pediatric appendiceal
practice for pediatric referrals. It is unlikely that local patients US can definitively rule in or rule out acute appendicitis in
would have undergone an appendectomy for missed appendi- approximately 70% of children, with an accuracy of 96%.
citis elsewhere. Additionally, an internal audit of our pediatric Interpretation guidelines that emphasize the presence or ab-
emergency department found no cases of missed appendicitis sence of periappendiceal inflammatory changes and utilize
between July 2008 and June 2009 [36]. This suggests that it is specific report templates can increase US accuracy and de-
unlikely that any cases of acute appendicitis were missed in crease indeterminacy. Additionally, when a definitive US di-
our study patients with clinical follow-up. agnosis is given, CT utilization and negative laparotomy rates
A third limitation, inherent in all studies of the efficacy decrease.
of US in the evaluation of acute appendicitis, is the reli-
ance on pathology as the gold standard for acute appen- Acknowledgments The authors thank Sonia Watson, PhD, for assis-
dicitis. It is well accepted that some cases of early or mild tance in editing the manuscript.
acute appendicitis resolve spontaneously [37], with 6.5%
of operated cases showing acute on chronic pathological Conflicts of interest None
changes [38]. Cases of spontaneously resolving acute ap-
pendicitis would be scored as false positive in our series.
Additionally, there might be a confirmation bias for acute References
appendicitis wherein the pathologist might be influenced
by clinical or surgical findings and label subtle changes 1. Sivit CJ (2004) Imaging the child with right lower quadrant pain
minimal acute appendicitis when these same findings and suspected appendicitis: current concepts. Pediatr Radiol 34:
447–453
would otherwise be passed as within normal limits. This
2. Flum DR, Koepsell T (2002) The clinical and economic correlates
might have been a factor in our single false-negative case of misdiagnosed appendicitis: nationwide analysis. Arch Surg 137:
when the appendix was visualized (US-2). 799–804
A fourth limitation of our study is the potential lack of 3. Morrow SE, Newman KD (2007) Current management of appen-
generalizability of our results to other practices. The skill of dicitis. Semin Pediatr Surg 16:34–40
4. Williams RF, Blakely ML, Fischer PE et al (2009) Diagnosing
the person performing the US study is of paramount impor-
ruptured appendicitis preoperatively in pediatric patients. J Am
tance in achieving highly accurate and determinate results. Coll Surg 208:819–825
Our highly experienced sonographers contribute greatly to 5. Taylor GA, Wesson DE (2015) Acute appendicitis in children: di-
these results. Additionally, the role of clinicians in the diagno- agnostic imaging. UpToDate. http://www.uptodate.com/contents/
sis or exclusion of appendicitis cannot be understated. The US acute-appendicitis-in-children-diagnostic-imaging. Accessed 23
April 2015
report is only a single piece of information among several that 6. Koning JL, Naheedy JH, Kruk PG (2014) Diagnostic performance
are used to determine whether patients go to surgery, are of contrast-enhanced MR for acute appendicitis and alternative
dismissed, or have further imaging or clinical follow-up. None causes of abdominal pain in children. Pediatr Radiol 44:948–955
of the false-negative US studies resulted in a child being 7. Orth RC, Guillerman RP, Zhang W et al (2014) Prospective com-
dismissed from the Emergency Department with missed ap- parison of MR imaging and US for the diagnosis of pediatric ap-
pendicitis. Radiology 272:233–240
pendicitis because of the judgment of the clinicians caring for 8. Frush DP, Frush KS, Oldham KT (2009) Imaging of acute appen-
these patients. dicitis in children: EU versus U.S. … or US versus CT? A North
Last, the prevalence of acute appendicitis in our study American perspective. Pediatr Radiol 39:500–505
group, 18.5%, was lower than in many other reported series 9. Strouse PJ (2010) Pediatric appendicitis: an argument for US.
Radiology 255:8–13
investigating appendiceal US accuracy in children. Doria et al.
10. Hernanz-Schulman M (2010) CT and US in the diagnosis of appen-
[14] reported an average appendicitis prevalence of 31% in dicitis: an argument for CT. Radiology 255:3–7
their meta-analysis that included 21 pediatric US and CT stud- 11. Puylaert JB (1986) Acute appendicitis: US evaluation using graded
ies. However they found that inclusion of data from the studies compression. Radiology 158:355–360
with prevalence rates less than 25% did not affect their results. 12. Okoji GO, Cameron BH (1991) Appendicitis presenting with dys-
Specifically, they noted that the pooled specificity and sensi- uria in a 2-year-old: ultrasound-aided diagnosis — a case report.
Ann Trop Paediatr 11:389–390
tivity values were not confounded by disease prevalence. 13. Rosen MP, Ding A, Blake MA et al (2011) ACR appropriateness
Thus our relatively low prevalence rate likely would not criteria® right lower quadrant pain — suspected appendicitis. J Am
weaken our conclusions. Coll Radiol 8:749–755
1944 Pediatr Radiol (2015) 45:1934–1944

14. Doria AS, Moineddin R, Kellenberger CJ et al (2006) US or CT for 26. Wiersma F, Toorenvliet BR, Bloem JL et al (2009) US examination
diagnosis of appendicitis in children and adults? A meta-analysis. of the appendix in children with suspected appendicitis: the addi-
Radiology 241:83–94 tional value of secondary signs. Eur Radiol 19:455–461
15. Pena BM, Taylor GA, Fishman SJ et al (2000) Costs and 27. Nielsen JW, Boomer L, Kurtovic K et al (2015) Reducing comput-
effectiveness of ultrasonography and limited computed to- ed tomography scans for appendicitis by introduction of a standard-
mography for diagnosing appendicitis in children. Pediatrics ized and validated ultrasonography report template. J Pediatr Surg
106:672–676 50:144–148
16. Mittal MK, Dayan PS, Macias CG et al (2013) Performance of 28. Chang ST, Jeffrey RB, Olcott EW (2014) Three-step sequential
ultrasound in the diagnosis of appendicitis in children in a multi- positioning algorithm during sonographic evaluation for appendici-
center cohort. Acad Emerg Med 20:697–702 tis increases appendiceal visualization rate and reduces CT use.
17. Schuh S, Man C, Cheng A et al (2011) Predictors of non-diagnostic AJR Am J Roentgenol 203:1006–1012
ultrasound scanning in children with suspected appendicitis. J 29. Rompel O, Huelsse B, Bodenschatz K et al (2006) Harmonic US
Pediatr 158:112–118 imaging of appendicitis in children. Pediatr Radiol 36:1257–1264
18. Bachur RG, Dayan PS, Bajaj L et al (2012) The effect of abdominal 30. Chang YJ, Kong MS, Hsia SH et al (2007) Usefulness of ultraso-
pain duration on the accuracy of diagnostic imaging for pediatric nography in acute appendicitis in early childhood. J Pediatr
appendicitis. Ann Emerg Med 60:582–590, e583 Gastroenterol Nutr 44:592–595
31. Sulowski C, Doria AS, Langer JC et al (2011) Clinical outcomes in
19. Trout AT, Sanchez R, Ladino-Torres MF et al (2012) A critical
obese and normal-weight children undergoing ultrasound for
evaluation of US for the diagnosis of pediatric acute appendicitis
suspected appendicitis. Acad Emerg Med 18:167–173
in a real-life setting: how can we improve the diagnostic value of
32. Abo A, Shannon M, Taylor G et al (2011) The influence of body
sonography? Pediatr Radiol 42:813–823
mass index on the accuracy of ultrasound and computed tomogra-
20. Polites SF, Mohamed MI, Habermann EB et al (2014) A simple
phy in diagnosing appendicitis in children. Pediatr Emerg Care 27:
algorithm reduces computed tomography use in the diagnosis of
731–736
appendicitis in children. Surgery 156:448–454
33. Yigiter M, Kantarci M, Yalcin O et al (2011) Does obesity limit the
21. Fedko M, Bellamkonda VR, Bellolio MF et al (2014) Ultrasound sonographic diagnosis of appendicitis in children? J Clin
evaluation of appendicitis: importance of the 3×2 table for outcome Ultrasound 39:187–190
reporting. Am J Emerg Med 32:346–348 34. Lee JH, Jeong YK, Park KB et al (2005) Operator-dependent tech-
22. (2002) 2000 CDC growth charts for the United States: Methods and niques for graded compression sonography to detect the appendix
Development. http://www.cdc.gov/nchs/data/series/sr_11/sr11_ and diagnose acute appendicitis. AJR Am J Roentgenol 184:91–97
246.pdf. Accessed 22 April 2015 35. Karakas SP, Guelfguat M, Leonidas JC et al (2000) Acute appen-
23. Estey A, Poonai N, Lim R (2013) Appendix not seen: the predictive dicitis in children: comparison of clinical diagnosis with ultrasound
value of secondary inflammatory sonographic signs. Pediatr Emerg and CT imaging. Pediatr Radiol 30:94–98
Care 29:435–439 36. Spahr C, Homme J, Mannenbach M et al (2010) Management of
24. Trout AT, Sanchez R, Ladino-Torres MF (2012) Reevaluating the the elusive inconclusive ultrasound for pediatric appendicitis.
sonographic criteria for acute appendicitis in children: a review of Pediatric Academic Societies Meeting, Vancouver
the literature and a retrospective analysis of 246 cases. Acad Radiol 37. Andersson RE (2007) The natural history and traditional manage-
19:1382–1394 ment of appendicitis revisited: spontaneous resolution and predom-
25. Krishnamoorthi R, Ramarajan N, Wang NE et al (2011) inance of prehospital perforations imply that a correct diagnosis is
Effectiveness of a staged US and CT protocol for the diagnosis of more important than an early diagnosis. World J Surg 31:86–92
pediatric appendicitis: reducing radiation exposure in the age of 38. Barber MD, McLaren J, Rainey JB (1997) Recurrent appendicitis.
ALARA. Radiology 259:231–239 Br J Surg 84:110–112

View publication stats

You might also like