Keywords acute appendicitis - appendectomy - pediatric surgery - plasma sodium concentration
- diagnostic accuracy
Introduction
Acute appendicitis (AA) is a common surgical emergency in children, and acute appendectomy
is the most common emergency operation on this population. The diagnostic method to
differentiate AA from other causes of abdominal pain has evolved over the years. Initially
AA was a clinical diagnosis, later aided by biochemical markers showing signs of inflammation.
In 1985, Gale et al presented computed tomography (CT) as a diagnostic tool for AA,[1 ] followed by Puylaert's description of ultrasound diagnosis[2 ] in 1986 and Incesu et al's introduction of magnetic resonance imaging in 1997.[3 ] Simultaneously several clinical scores have been presented, giving the clinician
support in making the diagnosis of AA or to differentiate between perforated and nonperforated
AA.[4 ] Finally, novel biochemical markers have been tested with the ambition of increasing
the accuracy of the diagnosis.[5 ]
[6 ]
At the same time, the treatment has changed from laparotomy on wide indications, via
minimally invasive procedures to recent advances in nonoperative treatment of AA.
The historical dogma that AA always progresses to gangrene and perforation was challenged[7 ] and has now been rejected.[8 ] Hence, there may be an increasing benefit in the ability of differentiating perforated
from nonperforated AA in modern treatment algorithms.
Plasma sodium concentration at admission to the pediatric emergency department was
chosen as a marker to differentiate between perforated and nonperforated AA. The reason
to test plasma sodium concentration originates from an earlier postoperative study,
where only patients with perforated AA were planned for inclusion.[9 ] One of the inclusion criteria was normonatremia at the end of surgery. The study
could not be conducted due to a very high prevalence of hyponatremia. Based on these
pilot data, we hypothesized that plasma sodium concentration might act as a marker
to differentiate between perforated and nonperforated AA. Arginine vasopressin (AVP)
concentration is often increased in inflammatory states such as acute pediatric surgery.[10 ] To investigate a possible mechanism of hyponatremia development, AVP was also measured
on admission.
The primary objectives of this study were to investigate the association between plasma
sodium concentration on hospital admission and perforation status in children with
AA and to evaluate the characteristics of plasma sodium concentration on admission
as a clinical marker for perforation status in children with AA.
Materials and Methods
Study Design and Participants
This prospective diagnostic accuracy study was conducted at the Department of Pediatric
Surgery at Astrid Lindgren Children's Hospital, Karolinska University Hospital in
Stockholm, Sweden, a tertiary referral center for Pediatric surgery. Patients were
included between May 2016 and July 2017, with an interruption for 3 months during
the relocation to a new hospital building, providing a total inclusion time of 10
months. The intended study population and the final inclusion criteria were previously
healthy children, 1 to 14 years of age, with a histopathologic diagnosis of AA according
to Carr.[11 ] Histopathologic perforation was defined as a macroscopic or microscopic perforation
with loss of appendix wall integrity.
Previously known metabolic or endocrine diseases were exclusion criteria. The study
protocol was approved by the Regional Ethics Review Board in Stockholm (reference
No. 2016/181–31/2). The study protocol was registered at the Australian New Zealand
Clinical Trial Registry (ACTRN 12617000047392).
Procedures
Children presenting at the pediatric emergency department with symptoms of suspected
appendicitis were invited to participate, and their parents were presented formalized
oral and written study information. All patients underwent imaging diagnostics, in
78 cases ultrasound and in 2 cases CT, which depending on clinical status was performed
before or after inclusion. For study participants, written consent was obtained prior
to inclusion and blood sampling, which was performed before any intravenous rehydration
and/or drug treatment was given. Children who were eventually not diagnosed with AA
were not included in the study.
Blood sampling for the study included bedside blood gas analysis (ABL 90 Flex Plus
[ABL], Radiometer Medical ApS, Denmark) where plasma sodium concentration, plasma
glucose (P-glucose), and base excess (BE) were obtained. The latter two variables
were included since they theoretically may show an association with perforated appendicitis
due to possible increased insulin resistance and starvation, respectively, in these
patients. In addition, white blood cell count (WBC) and C-reactive protein (CRP) were
determined as part of routine testing in this patient group. These variables were
analyzed at the Karolinska University Hospital Laboratory according to standard procedures.
Plasma-AVP was obtained in 52 patients (10 patients with and 42 without perforated
appendicitis) and was determined by radioimmunoassay as previously described,[12 ] using a specific AVP antibody (AB3096).[13 ] AVP was extracted from plasma using Sep-Pak Plus C18 extraction cartridges (Waters
Corporation, Milford, Massachusetts, United States). The detection limit was 0.10
pg/mL plasma and the interassay coefficient of variation was 8%.
The sodium concentration analysis was based on potentiometric measuring principles
in plasma. The variation in plasma sodium concentration has been investigated with
a resulting coefficient of variation of 0.2 to 0.3% using this method.[14 ]
Patient age (years) and sex, symptom duration (days), and body temperature (°C) were
registered at admission to the pediatric emergency department.
Statistical Analysis
Pilot observations suggested that a difference of 3 mmol/L in plasma sodium concentration
and a standard deviation 2 mmol/L was applicable to the present study in similar patients,
corresponding to a standardized effect size of 1.5 (effect size = difference/standard
deviation). Retrospective data suggested a rate of perforated appendicitis of ∼20%
among acute appendectomies at our hospital. With a total study size of n = 80 and at least 10 of these in the perforated group, there is a power of 80% to
find a standardized effect size of 1.0 with a two-sided t -test and a significance level of 5%, corresponding to a difference of equal size
as the standard deviation.
D'Agostino & Pearson omnibus normality test was used to assess normality. Normally
distributed data are presented as mean (95% confidence interval [CI]) and nonparametric
data are presented as median (interquartile range [IQR]).
Student's t -test or Wilcoxon matched pair test was used as applicable, Mann–Whitney U-test was
used to compare groups with nonparametric data, and two-sided Fisher's exact test
was used to compare dichotomized variables. Receiver operating characteristic (ROC)
curve analysis was performed to assess the best cutoff for the prediction of perforated
AA and values for area under the curve (AUC), sensitivity, and specificity are given
as mean (95% CI of the mean).
GraphPad Prism 6 was used for the statistical analyses (GraphPad Software, Inc. La
Jolla, California, United States), except the power analysis that was performed by
Dell Statistica 13.2 (Dell Software, Inc. Round Rock, Texas, United States).
Predictors of perforated AA were analyzed by univariate logistic regression followed
by step-wise forward multiple logistic regression using the NCSS software (NCSS11,
Kaysville, Utah, United States). The variables included in the univariate analysis
were sodium concentration, glucose concentration, BE, symptom duration, CRP concentration,
WBC, temperature, and age. Only factors with a univariate p < 0.2 were used in the multiple logistic regression analysis, and a decrease in deviance
by 3.84 (p = 0.05) was necessary for an item to be included in the final predictive model.
The study was conducted according to the Standards for Reporting Diagnosis Accuracy
Studies.[15 ]
Role of the Funding Source
The funders of the study had no role in the study design, data collection, data analysis,
data interpretation, or writing of the report. The corresponding author had full access
to all the data in the study and had final responsibility for the decision to submit
the article.
Results
Included and excluded patients are presented in [Fig. 1 ]. During the active inclusion periods, 228 children with suspected AA were admitted
to the pediatric emergency department; 212 children were diagnosed with AA and underwent
subsequent laparoscopic appendectomy. Nine children were randomized to nonoperative
treatment in a randomized controlled trial, the Appendectomy Versus Non-Operative
Treatment For Acute Nonperforated Appendicitis in Children (APPY)-trial.[16 ] Another seven children accepted participation in the study and had blood samples
taken on admission, but were later excluded, as they were not diagnosed with AA, nor
had an appendectomy. Out of the 212 patients later having an appendectomy, seven declined
participation in the study and 125 were never asked to participate. Eighty children
with AA, that was later confirmed by histopathology, were included in the study and
final analyses.
Fig. 1 STARD flow-chart. *Denotes patients that were randomized to not having appendectomy
within the APPY trial.[16 ] AA, acute appendicitis.
Patient characteristics of all eligible patients and the 80 children included in the
study are presented in [Table 1 ].
Table 1
Patient characteristics on admission to the pediatric emergency department
Included
Not included
p -Value
Perforated appendicitis
Nonperforated appendicitis
p– Value
n = 80
n = 132
n = 15
n = 65
Age (years)[a ]
9.2 (7.3–11.1)
8.9 (6.6–11.2)
0.511
7.5 (6.3–9.5)
9.2 (8.0–11.1)
0.167
Male, n (%)
53 (66)
69 (52)
0.062
12 (80)
41 (63)
0.245
CRP (mg/L)
32 (13–66)
44 (16–88)
0.134
79 (44–119)
24 (9–54)
<0.001
WBC (x109 /L)
14.1 (10.3–17.7)
14.9 (11.8–17.8)
0.353
17.1 (14.0–18.9)
13.5 (10.3–17.4)
0.026
Symptom duration (days)
2 (1–3)
2 (1–3)
0.057
3 (2–4)
2 (1–2.5)
<0.001
Perforation (%)
19
23
0.493
100
0
Plasma sodium (mmol/L)[a ]
138 (136–140)
134 (132–136)
139 (137–140)
<0.001
Body temperature (°C)
37.5 (37.1–38.0)
37.7(37.2–38.2)
0.236
37.8 (37.4–38.8)
37.5(37.1–37.9)
0.052
Plasma glucose (mmol/L)
5.1 (4.7–5.9)
5.4 (4.9–7.3)
5.1 (4.7–5.8)
0.020
Base excess (mmol/L)
−0.4 (−2–0.5)
−2 (−5–1)
−0.2 (−1.6–0.6)
0.003
AVP (pg/mL) n = 52
4.2 (2.8–7.9)
8.6 (5–14.6) n = 10
3.4 (2.5–6.6) n = 42
0.004
Abbreviations: AVP, arginine vasopressin; CRP, C-reactive protein; WBC, white blood
cell count.
Note: Data are expressed as mean (95% confidence interval).
a Data are expressed as median (interquartile range).
Age, sex, CRP, WBC, duration of symptoms, and proportion of perforated AA were comparable
between included and not included patients.
On admission to the pediatric emergency department, the median plasma sodium concentration
in patients with perforated AA was significantly lower as compared with patients with
nonperforated AA, 134 (IQR 132–136) mmol/L and 139 (IQR 137–140) mmol/L, respectively
([Fig. 2 ]). The patients who were sampled at the pediatric emergency department but later
excluded as they did not have appendicitis (n = 7) had similar median plasma sodium concentration, 139 (IQR 138–141), as compared
with patients with nonperforated AA.
Fig. 2 Plasma sodium concentrations at admission to the emergency department were compared
in patients with acute appendicitis verified by histopathology, with and without perforation,
respectively. Lines denote median values, whereas dashed line corresponds to the chosen
value for dichotomization of data. The p -value was obtained by Mann–Whitney U -test.
The ROC curve of plasma sodium concentration at admission identifying patients with
perforated AA showed an AUC of 0.93 (95% CI, 0.87–0.99). A cutoff value of plasma
sodium concentration of ≤136 mmol/L was shown to give the best possible sensitivity
and specificity, 0.82 (95% CI, 0.70–0.90) and 0.87 (95% CI, 0.60–0.98), respectively
(p < 0.001) ([Fig. 3 ]). By Fisher's exact test on dichotomized data, plasma sodium concentration ≤136
mmol/L was associated with an odds ratio (OR) of 31.9 (95% CI, 6.3–161.9), or a 15-fold
increase in relative risk (RR) (RR = 15, 95% CI, 3.7–62) for perforation compared
with values >136 mmol/L, with similar sensitivity and specificity as compared with
when continuous data was used ([Fig. 4 ]).
Fig. 3 Receiver operating characteristic curve for plasma sodium concentration as a predictor
of perforated AA. AUC is area under the curve; dashed line is the line of no predictive
value, that is, AUC = 0.5.
Fig. 4 Plasma sodium concentration dichotomized at ≤136 mmol/L and >136 mmol/L, respectively.
Dark bars denote perforated AA, gray bars denote nonperforated AA. Odds ratio was
31.9 (95% CI, 6.3–161.9) by Fisher's exact test, sensitivity 0.87 (95% CI, 0.60–0.98),
and specificity 0.83 (95% CI, 0.72–0.91). AA, acute appendicitis; CI, confidence interval.
Measures of possible and commonly used predictors of perforated AA are given in [Table 2 ]. From these univariate data, a forward stepwise multiple logistic regression analysis
was performed. This analysis confirmed the association between low plasma sodium concentration
and perforated AA. Starting with plasma sodium concentration as the strongest predictor,
adding any of the other proposed factors failed to statistically improve the predictive
model.
Table 2
Predictors of perforated appendicitis by univariate logistic regression (n = 80)
Variable
OR (95% CI)
p -Value
Correctly classified (%)
Sodium (per mmol/L)
2.3 (1.5–3.4)
< 0.001
81.1
Sodium (dichotomized)[a ]
25.5 (5.0–128.0)
< 0.001
81.1
Glucose (per mmol/L)
1.6 (1.0–2.6)
0.032
63.5
Base excess (per mmol/L)
1.6 (1.2–2.1)
<0.001
71.6
Symptom duration (days)[b ]
0.73 (0.54–0.99)
0.041
75.7
CRP (per mmol/L)
0.98 (0.97–0.99)
0.009
78.4
WBC (per 109 /L)
0.87 (0.78–0.96)
0.005
71.6
Temperature (per C°)
0.51 (0.23–1.12)
0.054
66.2
Age (per year)
1.2 (0.93–1.49)
0.159
62.2
Abbreviations: CI, confidence interval; CRP, C-reactive protein; OR, odds ratio; WBC,
white blood cell count.
Note: The investigated predictors are treated as continuous parameters.
a Cutoff at ≤136 mmol/L.
b Per 1 day increase.
On admission, median plasma AVP concentration was higher in patients with perforated
as compared with nonperforated appendicitis, 8.6 (IQR, 5.0–14.6) and 3.4 (IQR, 2.5–6.6)
pg/mL, respectively (p = 0.004) ([Table 1 ]).
As a post-hoc analysis not included in the original protocol, the sensitivity and
specificity for ultrasound imaging to predict perforation were calculated. Ultrasound
imaging was performed in 78 of the patients and indicated perforation in 14 patients,
whereof perforation was later confirmed histopathologically in six patients. Ultrasound
ruled out perforation in 64 patients, whereof perforation was later confirmed in seven
patients. Hence, the sensitivity and specificity for ultrasound in predicting perforation
later confirmed or ruled out by histopathology were 0.46 and 0.88, respectively.
Discussion
The main finding of this prospective diagnostic accuracy study was a strong correlation
between low plasma sodium concentration on hospital admission and perforation in children
with AA. Children with perforated AA had significantly lower plasma sodium concentration
as compared with children with nonperforated AA. In addition, low plasma sodium concentration
had a high predictive value discriminating between perforated and nonperforated AA.
In children with AA and plasma sodium concentration of ≤136 mmol/L at emergency department
admission, there was a 32-fold increase in OR of finding a perforated AA. In the multivariate
analysis, none of the commonly used variables predicting the severity of AA (CRP,
WBC, temperature, duration of symptoms, and age) improved the prediction obtained
from plasma sodium concentration alone. BE and P-glucose (included in the blood gas
analyses) were also found to be associated with perforated AA, but the contribution
to overall accuracy (plasma sodium concentration + BE + P-glucose) was minor as compared
with plasma sodium concentration alone. However, with only 15 cases, multiple significant
predictors are unlikely to be found in a multivariate analysis, and their absence
must be interpreted with caution.
Possible limitations of the current study are inherent to the sample size and the
single center study design. Further drawbacks are the interruption of the inclusion
period and that the patients were not consecutively included. Nonetheless, the validity
of the study is enhanced by the prospective design. Furthermore, our results remain
to be reproduced in larger, prospective multicenter trials before being incorporated
into standardized pathways for diagnosing AA in children.
The findings of this study are consistent with our pilot data observations and also
with the sparse previous retrospective data on plasma sodium concentration as a marker
in children[17 ] and adults[18 ]
[19 ] in this context. A recent meta-analysis found the pooled sensitivity and specificity
to differentiate complicated (perforated or gangrenous appendicitis) from noncomplicated
AA to be within the range of 0.14 to 0.59 and 0.74 to 1, respectively, when nine different
CT features informative for complicated AA were studied.[20 ] Compared with this, our findings with sensitivity and specificity of 0.82 and 0.87,
respectively ([Fig. 4 ]) combined with a ROC curve AUC of 0.93 are notable. The post-hoc finding of a sensitivity
and specificity of ultrasound to predict perforation of 0.46 and 0.88, respectively,
also indicates that this imaging modality would benefit from other indices of perforation
such as plasma sodium concentration, in our setting.
Differentiation between uncomplicated and complicated appendicitis is not possible
before treatment as there is no proper modality to differentiate between phlegmonous
and gangrenous appendicitis except histopathology. Multiple randomized controlled
trials, both presented[21 ] and ongoing, use suspected perforated appendicitis as an exclusion criterion. For
these reasons, the differentiation of perforated appendicitis will become increasingly
important as differentiated treatment modalities are developed for the subtypes of
nonperforated and perforated AA. There is evolving evidence that medical treatment
of nonperforated AA is safe, feasible, and noninferior to appendectomy, in both children
and adults.[21 ]
[22 ]
[23 ] Children with abdominal comorbidity such as previous abdominal surgery or children
with respiratory compromise such as ongoing infections or cystic fibrosis may benefit
the most from nonoperative treatment of nonperforated AA. Whether determination of
plasma sodium concentration will be useful to select patients for nonoperative treatment
remains to be proven.
Hyponatremia is a common finding in various pediatric patient groups, for example,
in children with pneumonia, bronchiolitis, meningitis, and encephalitis and might
be regarded as a surrogate marker of disease severity.[24 ]
[25 ] Different patient groups may have different pathogenesis for hyponatremia. However,
one important factor they have in common is increased AVP release, although other
mechanisms have been proposed.[26 ] Children with bronchiolitis in the pediatric intensive care unit (PICU) setting
may be used as an adequate illustration, since patients with hyponatremia at admission
have been shown to have a higher mortality, increased ventilator time, and longer
duration of stay in the PICU as compared with children with normal sodium values.[27 ] The concentration of AVP was also shown to be high in the most severe cases of respiratory
syncytial virus-bronchiolitis.[28 ]
Known nonosmotic stimuli for AVP release associated with AA, and aggravated in perforated
AA, are pain, fever, nausea, vomiting, physiological stress, and reduced intravascular
volume. Furthermore, perforation of the appendix will initiate peritonitis, with an
aggressive inflammatory as well as acute neuroendocrine stress response, stimulating
AVP release that in turn may cause low plasma sodium concentration.[29 ]
[30 ] Therefore, we hypothesized that AVP concentration might be increased in these patients.
Indeed, patients with perforated AA had significantly higher AVP concentration as
compared with patients with no perforation. In our opinion, it is likely that this
is a major factor contributing to the lower sodium concentration seen among patients
with perforated AA
Plasma sodium concentration is not only highly predictive for distinguishing perforated
from nonperforated AA; it is also easily performed at a low cost, and readily available
in the pediatric emergency department. In settings where radiographic techniques are
not accessible, for example, in resource scarce environments, a strong marker for
perforation could possibly impact the care of patients with AA, although this remains
to be investigated.
Conclusion
This represents the first prospective study to identify plasma sodium concentration
on hospital admission as a predictive marker discriminating perforation from nonperforation
in children with AA. In our view, plasma sodium concentration may be used as a supporting
tool to supplement clinical examination, blood sampling, and imaging techniques in
this setting.
