Article, Pediatrics

A new clinical score to identify children at low risk for appendicitis

a b s t r a c t

Background: Besides clinical signs and imaging, in recent years, biomarkers have proven to be a viable diagnostic resource for Acute appendicitis .

Objective: The objective of this study was to develop a clinical score including clinical signs and a combination of biomarkers to identify children with abdominal pain at low risk of AA.

Design/methods: We prospectively included children 2 to 14 years of age with abdominal pain suggestive of AA who presented to the emergency department between July 2016 and September 2017. A new score, the Pediatric appendicitis Laboratory Score (PALabS) including clinical signs, leucocyte (WBC) and neutrophil (ANC) counts and plasma C-reactive protein (CRP) and calprotectin (CP) levels was developed and validated through second- ary analyses of two distinct cohorts The validation sample included visits to a single pediatric emergency depart- ment from 2012 to 2013 and 2016 to 2017.

Results: The derivation sample included 278 children, 35.9% of whom had AA and the validation sample included 255 children, 49% of whom had AA. Using logistic regression, we created a 6-part score that consisted of nausea (3 points), history of focal Right lower quadrant pain (4 points), ANC of >=7500/uL (7 points), WBC of >=10,000/uL (4 points), CRP >= 10.0 mg/L (2 points) and CP >= 0.50 >= ng/mL (3 points). This score exhibited a high Discriminatory power (area under the curve: 0.88; 95% confidence interval: 0.84 to 0.92) and outperformed the PAS and Kharbanda scores (area under the curve: 0.76; 95% confidence interval: 0.71 to 0.82 and 0.82; 95% confidence in- terval: 0.77 to 0.87, respectively). A PALabS <=6 had a sensitivity of 99.2% (95% confidence interval [CI]: 95.6-99.9), negative predictive value of 97.6% (95% CI: 87.7-99.6), and negative likelihood ratio of 0.03 (95% CI: 0.00-0.18) in the validation set.

Conclusion: In our validation cohort of patients with acute abdominal pain, the new score can accurately predict which children are at low risk of appendicitis and could be safely managed with close observation.

(C) 2019

Introduction

acute appendicitis in children is the primary cause of Urgent surgery in pediatric patients [1,2]. Diagnosis of AA continues to be a challenge, especially in the youngest children, who often present with abdominal pain accompanied by nonspecific signs. Physicians often face a clinical dilemma in deciding the timing of surgical intervention, since delayed diagnosis of appendicitis is associated with increased morbidity, mortality, and health care costs [3,4].

Today, Imaging techniques constitute the basis of diagnosis in the majority of cases, especially Abdominal ultrasound as it is innocuous to the patient. Nonetheless, that technology is not always available,

? All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

* Corresponding author at: Pediatric Emergency Department, Department of Pediatrics, Cruces University Hospital, Plaza de Cruces s/n, E-48903 Barakaldo, Bizkaia, Spain.

E-mail address: [email protected] (J. Benito).

and its diagnostic performance depends on the experience of the pro- fessional using it, with a sensitivity as low as 80% for some operators [5-7]. Abdominal computed tomography (CT) scans and magnetic reso- nance imaging (MRI) may also be used for improving the diagnostic precision but at the expense of exposure to significant ionizing radiation in the case of CT [8-11] or a lack of availability in the emergency depart- ment (ED) in the case of MRI [11].

Clinical scores based on clinical signs and white blood cell count or absolute neutrophil count (ANC) have been developed to standardize care and limit imaging for patients with possible appendici- tis [12-14]. These scores appeared promising in derivation samples, but most of them have revealed variable accuracy and limited clinical use- fulness on external validation [15]. In fact, only the Kharbanda score has been validated in pediatric patients [14].

On the other hand, various markers have been proposed in the scien- tific literature as possible markers of acute appendicitis [16,17]. Recent studies have identified a panel of biomarkers (the APPY1 Test), includ- ing WBC and C-reactive protein (CRP) and calprotectin (CP) plasma

https://doi.org/10.1016/j.ajem.2019.05.050

0735-6757/(C) 2019

derivation cohort“>levels, that has the potential to identify, with great accuracy, children with abdominal pain who are at low risk of AA [18,19]. Our hypothesis was that including new biomarkers (CRP and CP) would be helpful for designing a more accurate decision support tool to identify children suspected of having AA who are at low risk of the disease.

Patients and methods

Patient population

Derivation cohort

We prospectively reviewed the clinical records of all patients aged 2 to 14 years, between July 1, 2016, and August 31, 2017, with suspected appendicitis, who were admitted to Cruces University Hospital (CUH). All patients were recruited by registrar physicians in the ED. The deriva- tion cohort was built with data from a randomly chosen set of patients extracted from this cohort. For the randomization process we specified the number of cases needed and we estimate the percentage of cases. We used the dialog box of IBM SPSS Statistics for Windows to select a random sample. Sampling is performed without replacement; so, the same case cannot be selected more than once.

Validation cohort

For the validation, we used a temporal validation, which can be con- sidered external in time and thus intermediate between internal valida- tion and external validation [20]. The validation cohort was built with data from two independent prospectively analyzed cohorts of patients 2 to 14 years old with visits to CUH ED from 2012 to 2013 and from 2016 to 2017. Being consistent with the recommendations of Altman, these cohorts were chosen as the validation sample because there were many clear similarities between the two sets of patients and be- tween the clinical and laboratory techniques used for assessing them. All the children with suspected appendicitis recruited were identified by trained clinical researchers in the ED.

Data collection and consent

This study was approved by the local ethics committee (Ethics Com- mittee of CUH). Informed consent was obtained from all parents, and as- sent was obtained from children who were older than 7 years of age. Clinical researchers collected demographic, clinical and management data: age, sex, personal history, any treatment administered before ar- riving at the ED, symptoms, physical examination findings, test results, diagnosis, treatment administered, and course of the illness. All of these data were recorded in the hospital’s computer system and later retrieved and entered into a database by one of the investigators, with all the Security measures necessary to comply with European data pro- tection law. Recruitment was carried out 24 h a day, 7 days a week.

Outcome measures and definitions

In both derivation and validation cohorts, patients were defined as having appendicitis by the final diagnosis documented in the pathology and surgical reports (ICD-10: K35-K37). For patients who were discharged without a diagnosis of AA, a telephone follow-up was per- formed 15 days after the emergency consultation. In cases in which we failed to contact the family, searches were made through the hospital’s computer system and the electronic registries of the Basque Health System to determine whether the patient returned to the hospi- tal during the monitoring period.

Abdominal ultrasound and basic blood tests were performed, includ- ing measurement of WBC, ANC, and CRP and CP levels. After evaluation by the attending physician of the laboratory tests, diagnostic imaging, and/or Surgical consultations undertaken, the children whose condition was classified as “suggestive of appendicitis” were admitted for surgical intervention or further tests. The clinical diagnosis of suspected AA was based fundamentally on the physical exam (localized pain in the right iliac fossa and/or the presence of signs of peritoneal irritation) and

ultrasound findings. Ultrasound findings were considered to be sugges- tive of AA if was any of the following were observed: an appendix diam- eter of N6 mm, Inflammatory changes in the periappendicular tissues, free liquid in the peritoneal cavity, and/or appendicoliths.

Children who met any of the following criteria were excluded: symptoms lasting N5 days, prior appendectomy, diagnosis of urinary tract infection, cancer, or inflammatory intestinal disease, or treatment with systemic corticoids, as well as refusal to sign the informed consent form.

Laboratory procedures

The inflammatory markers (WBC, ANC, CRP, and CP) were measured immediately in the hospital lab. In addition, a blood sample was col- lected in ethylenediaminetetraAcetic acid tubes, centrifuged, and frozen at -40 ?C until analysis. The following systems were used for marker measurements: turbidimetric immunoassays (cobas c 501 module, Roche Diagnostics) for CRP, enzyme immunoassay (DS2 system with Matrix software, version 1.17, ALERE Healthcare) for CP, and a hematol- ogy analyzer (Beckman Coulter) for WBC. The following ranges were considered normal: WBC b 10,000 k/mL, ANC b 7500 k/mL, CRP b 2.0 mg/dl, and CP b 0.5 ngr/ml.

Data analysis

Score development

univariate and multivariate analyses were conducted using IBM SPSS Statistics for Windows Version 23.0 (released 2013; IBM Corp, Armonk, NY). Predictors considered for inclusion in the score were coded as dichotomized variables for the effective discrimination be- tween appendicitis and non-appendicitis diagnoses. threshold values (cutoffs) for continuous variables (WBC, ANC, CP and CRP) were ob- tained using the Youden index, which is represented graphically as the height above the chance line and is also equivalent to the area under the curve of a single operating point [21].

For consideration in the score, we included only predictors with b10% missing data. First, we entered the candidate variables (as de- scribed above) into a forward stepwise multivariate logistic regression analysis. Variables independently associated (P b .05) with appendicitis in the age- and sex-adjusted model were then ranked according to the magnitude of the ?-coefficient. Since the new score was based on more laboratory variables than previously proposed scores, it was called the Pediatric Appendicitis Laboratory Score (PALabS).

For the derivation cohort, we compared the discriminatory perfor- mance of PALabS versus the previously published Pediatric Appendicitis Score (PAS) [13] and Kharbanda Score (KS) [14]. The overall diagnostic performance of each score was analyzed using the area under the re- ceiver operating characteristic curve (AUC). Significance was set at the 5% level and 95% confidence intervals (CIs) were calculated.

Validation

The clinical decision rule derived from the logistic regression was then applied to the validation cohort. Only patients with complete infor- mation on all candidate variables were considered in the multivariate analyses. If a laboratory test result was not available, the value corre- sponding to the mean in our derivation cohort was used. We evaluated the discriminatory performance using the AUC. To establish the diag- nostic accuracy of the score, sensitivity, negative predictive value (NPV), and negative likelihood ratio were determined.

Results

Study population

Between July 2016 and August 2017, 59,473 episodes were regis- tered in the ED, of which 2689 corresponded to patients who presented

with abdominal pain. In 607 (22.5%), complementary tests were per- formed in order to rule out AA, with AA being confirmed in 217 cases (35.7%). A total of 204 patients were missed or not approached to partic- ipate and 17 met exclusion criteria, leaving 386 patients eligible for en- rollment. Twenty-five patients were excluded for having incomplete data and 361 patients were enrolled, with AA being confirmed in 140 cases (38.8%) (Fig. 1). Thirty-five (25%) of the 140 patients with appen- dicitis had a perforated appendix. The clinical characteristics of the pa- tients enrolled are summarized in Table 1.

The mean age of the children included in the analysis was 9.89 (+-2.8) years, and 60.4% were boys. The mean age (9.31 +- 3.2 years), sex (56.4% boys) and appendicitis rate (35.7%) in the group of 607 children who could have been, but were not, included in the study did not differ signif- icantly from those of the sample analyzed. We failed to contact seven pa- tients in the group without AA (2.7%) at the telephone follow-up. None of them returned to the hospital during the monitoring period.

Clinical course

An abdominal ultrasound scan was performed on all patients. Eleven patients with abdominal ultrasound consistent with AA were not in the end found to have AA. Just one patient with negative ultrasound find- ings was positive for AA, and of the 47 doubtful cases, the disease was ruled out in 39 children and confirmed in the other 8. After evaluation, 165 (45.7%) children were discharged from hospital, 138 (38.2%) went directly to the operating room, and 58 (16.1%) were admitted for obser- vation. Three (0.8%) patients taken to the operating room were found to have a normal appendix.

A randomly chosen set of 278 patients extracted from this cohort, 35.9% of whom had appendicitis, were included in the derivation

sample. The validation sample included 255 children from two indepen- dent cohorts of children, one of them composed of the other 83 patients from the aforementioned cohort and the other based on an existing co- hort of 172 children seen in our pediatric ED with suspected appendici- tis between 2013 and 2014 [19]. The rate of appendicitis in the validation sample was 49%. The derivation and validation groups did not differ significantly in age, sex, clinical characteristics or Laboratory test results. Characteristics of the derivation and validation samples are presented in Table 2.

PALabS development

Threshold values obtained with the Youden index for the continuous variables were: 10,000/uL for WBC, 7500/uL for ANC, 0.50 ng/ml for CP and 10.0 mg/L for CRP. All predictors evaluated in the derivation sample for inclusion in the score were significantly associated with AA risk (P value b .05), adjusted for age and sex (Table 3) Using logistic regression, we created a 6-part score that consisted of nausea or vomiting (3 points), maximal pain in the right lower quadrant (RLQ) (4 points), ANC of >=7500/uL (7 points), WBC of >=10,000/uL (4 points), CRP

>= 10.0 mg/L (2 points) and CP >= 0.50 ng/mL (3 points) (Table 3).

The PALabS exhibited a high discriminatory power (AUC: 0.88; 95% CI: 0.84 to 0.92) and outperformed the PAS and Kharbanda score (AUC: 0.76; 95% CI: 0.71 to 0.82 and 0.82; 95% CI: 0.77 to 0.87 respectively). (Fig. 2).

PALabS validation

The PALabS was applied to the 255 patients in the validation sample. A score <= 6 identified 42 (16.4%) patients, of which, 41 did not have ap- pendicitis. This cutoff point had a sensitivity of 99.2% (95% CI:

Fig. 1. Flow of patients through the study.

Table 1

Characteristics of patients prospectively enrolled between July 2016 and August 2017.

Variables

No appendicitis n=178 (64.0)

Appendicitis n = 100 (36.0)

P

Sex, n (%)

95 (53.4)

68 (68.0)

0.012

Age (years)

Median (IQR)

10.0 (8.0-12.0)

9.0 (8.0-11.7)

n.s.

Duration of pain (hours), n (%)

<=12 hours

70 (39.3)

43 (43.0)

13-24 hours

61 (34.3)

30 (30.0)

n.s.

25-48 hours

19 (10.7)

18 (18.0)

N48 hours

28 (15.7)

9 (9.0)

Pain score (0-10)

Median (IQR)

5.0 (5.0-7.0)

6.0 (5.0-8.0)

Temperature before ED arrival, ?C Mean (SD)

37.6 (1.1)

37.5 (1.0)

n.s.

Maximum temperature, ?C Mean (SD)

36.8 (0.9)

36.8 (0.9)

migration of pain to RLQ, n (%)

84 (48.6)

55 (55.0)

n.s.

History of focal pain in RLQ, n (%)

42 (23.6)

39 (39.0)

0.005

Symptoms

Anorexia, n (%)

104 (58.8)

76 (76.0)

0.002

Nauseas, n (%)

83 (46.6)

73 (73.0)

b 0.001

Physical exam

Maximal pain in RLQ, n (%)

91 (51.1)

83 (83.0)

b 0.001

Blumberg positive, n (%)

103 (58.2)

70 (70.0)

0.015

Psoas sign positive, n (%)

73 (41.0)

42 (42.0)

n.s.

Pain with hopping, n (%)

82 (46.1)

61 (61.0)

0.045.

Pain with walking, n (%)

76 (46.6)

64 (64.0)

0.002

Guarding, n (%)

50 (28.1)

43 (43.0)

0.036

Laboratory

CRP, mg/L

Median (IQR)

3.0 (0.2-20.6)

18.0 (4.0-40.3)

b 0.001

CP, ng/dL

Median (IQR)

0.5 (0.3-0.9)

0.8 (0.5-1.2)

b 0.001

ANC, k/mL

Median (IQR)

5425.0 (3700.0-9425.0)

12200.0 (8675.0-16225.0)

b 0.001

WBC, k/mL

Median (IQR)

8550.0 (6975.0-11800.0)

14700.0 (11600.0-18525.0)

b 0.001

SD: standard deviation; n.s.: no significant; IQR: interquartile range; RLQ: right lower quadrant.

95.6-99.9), NPV of 97.6% (95% CI: 87.7-99.6), and negative likelihood ratio of 0.03 (95% CI: 0.00-0.18). A score >= 18 identified 121 (47.4%) pa- tients, of which, 90 had appendicitis. (Fig. 3).

Table 2

Main characteristics of Derivation and Validation Sample.

PALabS and ultrasound

In Table 4, we present the relationship between PALabS and ultra- sound findings. Pooling patients from the derivation and validation samples, 60 (11.2%) had an inconclusive ultrasound result and the rate of appendicitis in this group was 18.3%. Eleven (3.1%) patients who

had positive ultrasound findings were not in the end diagnosed with ap-

Variables Derivation

Sample, n = 278

Validation Sample, n= 255

pendicitis and four of them had PALabS <=6. The positive predictive value of ultrasound was 61.2% for patients with PALabS <=6 versus >=93% for pa-

Sex, n (%) 163 male (58.6) 169 male (66.3) Age (years)

Median (IQR) 10.0 (8.0-12.0) 10.0 (8.0-12.0)

Duration of pain (hours), n (%)

<=12 hours

113 (40.6)

83 (32.8)

13-24 hours

91 (32.7)

83 (32.8)

25-48 hours

37 (13,3)

37 (14.6)

N48 hours

37 (13,3)

50 (19.8)

Pain score (0-10)

Median (IQR)

5.0 (5.0-8.0)

6.0 (5.0-7.0)

Nausea/Emesis, n (%)

156 (56.1) 141 (55.3)

Maximal pain in RLQ, n (%) 174 (62.6) 187 (73.3) Laboratory

CRP, mg/L

Median (IQR) 8.0 (0.4-30.2) 6.0 (1.0-20.5)

CP, ng/dL

Median (IQR) 0.6 (0.3-1.0) 0.8 (0.5-1.4)

Neutrophil count, k/mL

tients with PALabS N 6.

Patients missed by PALabS

Seven patients with appendicitis were misclassified as low risk by the PALabS, six in the derivation sample and one in the validation sam- ple. None of these patients were found to have complicated appendicitis in the operating room. Table 5 lists the clinical characteristics of these patients.

Discussion

In the present study, we derived and validated a new clinical score, PALabS, to identify children with abdominal pain who are at low risk of acute appendicitis. These children could be managed safely with

Median (RIQ) 7700.0

(4500.0-12300.0)

Leukocyte count, k/mL

Median (RIQ) 10600.0

9900.0

(5500.0-13806.0)

close observation, without other Diagnostic tests. This is the first clinical score for AA in which laboratory variables outweigh clinical ones and this characteristic makes it potentially more reliable than previously

(7600.0-15000.0) 12600.0

(8900.0-16200.0)

Appendicitis confirmed, n (%) 100 (35.9) 125 (49.0) IQR: interquartile range; RQL: right lower quadrant.

published appendicitis scores.

Other authors have developed clinical scores to predict appendicitis in adults and children, although few have been subsequently validated in pediatric patients [12-14]. Most of these scores are based mainly on

Table 3

Univariate and multivariate analysis of predictors of appendicitis in the derivation sample (n = 278). Diagnostic value of the independent variables identified in the multivariate analysis.

Variables

Univariate analysis

Multivariate analysis

Weight

? Coefficient

95% CI

Significance

? Coefficient

95% CI

Significance

Duration of pain (Ref. b24 h)

24-48 h

1.15

0.67-1.96

0.500

N 48 h

0.56

1.27-2.64

0.007

Migration of Pain

1.32

0.80-2.17

0.267

History of pain in RLQ

2.07

1.21-3.51

0.007

Temperature before ED arrival

0.87

0.63-1.19

0.398

Maximum temperature

0.98

0.74-1.30

0.917

Anorexia

2.31

1.33-4.04

0.003

Nausea

3.09

1.82-5.26

b 0.001

3.03

1.54-5.96

0.001

3

Maximal pain in RLQ

4.66

2.56-8.49

b 0.001

3.71

1.79-7.69

b 0.001

4

Positive Blumberg sign

1.86

1.09-3.18

0.023

Positive Psoas sign

1.01

0.68-1.66

0.981

Pain with hopping

2.00

1.14-3.50

0.015

Pain with walking

2.22

1.31-3.73

0.003

Guarding

1.90

1.38-3.19

0.014

WBC >= 10,000 k/mL

16.03

7.79-32.98

b 0.001

4.00

1.11-14.35

0.033

4

ANC >= 7500 k/mL

18.09

8.96-36.52

b 0.001

6.67

1.81-24.58

0.004

7

CRP >= 1.0 mg/dl

3.29

1.97-5.50

b 0.001

2.25

1.10-4.62

0.026

2

CP >= 0.50 ngr/ml

2.62

1.52-4.49

b 0.001

3.16

1.18-8.45

0.022

3

clinical parameters, with laboratory data playing a secondary role. This could partially explain the poor performance of these clinical tools when seeking to validate them in an external population, since some of the score components may be influenced by observer experience and subjectivity. On the other hand, many of the clinical variables that appear as components in such scores are used by clinicians to select which patients with abdominal pain should undergo laboratory or im- aging tests to rule out appendicitis. That is, most patients selected to rule out appendicitis have these symptoms and findings on physical ex- amination, regardless of whether they subsequently have AA. In this context, among the classic score components, laboratory results might have a greater discriminatory power to rule out AA. Nonetheless, to date, only the Kharbanda score [14] has shown that a laboratory param- eter, ANC N 6750 k/mL has greater diagnostic value than clinical param- eters to identify children with AA. This is also the case of our clinical score, in which not only ANC but also other laboratory parameters, namely, WBC, CRP and CP, may help to diagnose AA more accurately.

Samuel [13] published the PAS in 2002 and reported a sensitivity of 100% and NPV of 99%. The score was not validated, however, and nu- merous studies seeking to validate it have not found such good perfor- mance [15]. Further, a more recent study by Kharbanda et al. [22]

Fig. 2. Comparison of the new score and PAS and Kharbanda Score in the derivation sample.

found a significantly poorer overall performance of the PAS compared to another clinical rule for appendicitis. These investigators derived and validated, first a clinical score and more recently a risk calculator, both with greater discriminatory power for appendicitis than those pre- viously published. We applied the PAS and Kharbanda score to our der- ivation cohort of patients and found that all scores, PAS, Kharbanda and PALabS, predicted appendicitis with high accuracy. Nonetheless, the PALabS performed better than the Kharbanda Score and markedly bet- ter than the PAS. It seems that the greater the weight of the clinical var- iables in the design of the scores, the lower their discriminatory power. As already mentioned in the previous paragraph, all these scores have been developed and validated in patients with suspected appendicitis. In this population, history and physical examination are already used to de- cide who should undergo laboratory and imaging tests to rule out appen- dicitis. In addition, several studies [23-25] have shown that not all symptoms and clinical signs have the same discriminatory power for the diagnosis of appendicitis. A single study [26] carried out in patients with undifferentiated abdominal pain showed that pain migration to the RLQ and cough/hop pain were associated with a higher probability of appendicitis. On the other hand, in studies performed in patients with suspected appendicitis, the absence of nausea and of maximal pain in the RLQ made the diagnosis of appendicitis unlikely. In one recent study focusing solely on the interexaminer reliability of history and phys- ical examination findings in pediatric patients with abdominal pain with and without AA [27], only vomiting showed high interexaminer reliabil- ity. In our study, only nausea/vomiting and maximal pain in RLQ were

shown to be independent predictors of AA.

Notably, WBC was the most commonly reported laboratory test (10 studies) followed by ANC or percentage of neutrophils [28-30] (seven studies). The most commonly reported cutoff for WBC is 10,000 cells/mm3 (in nine studies), while the cutoff for ANC varied somewhat among studies: 7500 cells/mm3 in three studies [19,23,26] and 6750 cells/mm3 in one [31]. Our study confirms these cutoff points and the strong association between low ANC and the absence of appen- dicitis among patients with signs and symptoms suggestive of appendi- citis found in previous studies [32].

In recent years, given the complexity of the diagnosis of AA, numer- ous publications have appeared studying the relationship between AA and various biomarkers such as CRP, CP [32-36] and more recently proadrenomedullin [37]. Although relationships have been found indi- vidually between these markers and AA, they have not demonstrated sufficient reliability in ruling out the disease with certainty [12-16]. On the other hand, the combination of WBC and ANC with biomarkers such as CRP and CP has yielded satisfactory results with a high level of

Fig. 3. Score performance with the logistic-regression model.

sensitivity for acute appendicitis, suggesting that they could be useful in combination to identify children with abdominal pain who are at low risk of this disease. In a study published by Huckins et al. [18], 503 pa- tients with abdominal pain suggestive of AA, the APPY1 test, a panel of biomarkers that includes WBC, PCR and CP, yielded a sensitivity of 96.5% (95% CI: 92-99) and NPV of 96.9% (95% CI: 93-99) with a specific- ity of 43.2% (95% CI: 38-48). In addition, another more recent study [19] carried out in our pediatric ED showed that adding ANC b 7500 k/mL to the APPY1 test panel, the sensitivity and NPV to rule out AA increased to 100%. Nevertheless, neither of these studies combined laboratory re- sults with clinical data to create a reliable tool for the diagnosis of ap- pendicitis. Taking into account the clinical data, it was possible to identify some of the small number of patients identified as low risk in these two studies who were eventually found to have AA. The current study finds that the combination of biomarkers with clinical parameters not only helps identify children with a low risk of appendicitis but also to select more precisely those at very high risk of having this disease. It is important to note that the PALabS was able to identify 60% of patients as at >=75% risk or b5% risk for AA in our validation cohort, which could facilitate decisions about whether to perform imaging studies, consult the surgeon or indicate clinical observation.

At present, the diagnosis of AA is supported more by CT scans and ul-

trasound imaging than by laboratory tests [31,38-40]. At our hospital and in our care setting, ultrasound is the preferred imaging test, and has a high success rate in expert hands, with sensitivity and specificity above 90% [41]. Accordingly, ultrasound performed very well as a diag- nostic tool in the present study with an NPV of 99.5% and positive pre- dictive value of 95.2% for the diagnosis of appendicitis. The results were uncertain in 60 cases (11.2%), however, and it was necessary to ex- tend the stay in hospital under observation or carry out further tests, this resulting in a final diagnosis of AA in 11 (18.3%) of those cases. Twelve of the patients (9.9%) with an uncertain ultrasound diagnosis had PALabS <=6 and none of them had AA. In addition, 34 patients with an uncertain ultrasound diagnosis had PALabS <= 18 and only 5 (14.7%) had AA. Consistent with this, a recent study combining WBC cutoffs with inconclusive Ultrasound studies also identified low risk patients [42]. These patients with a low risk of appendicitis could be managed

with careful observation without more imaging studies. According to our study, both the PALabS and ultrasound, either alone or in combina- tion, could be used for initial screening in children with suspected AA, depending on the level of clinical suspicion and the diagnostic resources existing in each healthcare setting.

Strengths

Our score has two important strengths. First, it is simple to apply in the clinical setting, with few variables. In addition, we believe that the clinical variables selected are probably the most objective and therefore the least dependent on the interpretation of the clinician. On the other hand, the score has been validated which means that it is likely to be more reproducible in clinical environments similar to ours.

Limitations

Several limitations should, however, be noted, the primary one being that it is a single-center study conducted at a tertiary hospital where physicians in charge of treating the patient have access to radiol- ogy services and consultations with pediatric surgeons 24 h a day. Dif- ferences in the prevalence of AA and factors related to complementary tests, such as, whether they are available (e.g., ultrasound may not be) and the criteria for performing them, might make our results less appli- cable in other settings. Second, the PALabS was derived and validated using data from different time periods from a single children’s hospital. As such, our results require validation in new populations before wide- spread application. Similarly, our derivation and validation cohorts had appendicitis at rates of 35.9% and 49%, respectively. The discriminatory power of PALabS might be lower if applied in populations with higher or lower appendicitis rates. Third, the size and selection of the sample were subject to the presence of the investigators participating in the study and, therefore, may not accurately reflect the entire population of children with suspected AA. We believe, however, that this does not significantly affect the results of the analysis, in which cases and con- trols are compared in a balanced manner. Four, we failed to contact by telephone 10 of the 93 patients discharged, although a check was

Table 4

Evaluation of the ultrasound in relation to the values of the new score in 531 patients (278 from the derivation sample and 254 from the validation sample).

New score

Appendicitis/n? patients (%)

Inconclusive ultrasound, n (%)

Inconclusive ultrasound with appendicitis, n (%)

NPV of ultrasounda, %

PPV of ultrasounda,%

<=6

7/121 (5.7)

12 (9.9)

0 (0)

99.9

61.2

7-18

83/254 (32.6)

34 (13.3)

5 (14.7)

99.1

93.8

>=19

135/156 (86.5)

14 (8.9)

6 (42.8)

95.4

99.1

Overall

225/531 (42.3)

60 (11.2)

11 (18.3)

99.5

95.2

a NPV and PPV of ultrasound excluding inconclusive cases.

Table 5

Clinical characteristics of patients with a final diagnosis of appendicitis who were classified as low risk (new score <= 6) in the derivation and validation samples.

Sample

Age/sex

Nausea/vomiting

Maximal tenderness in RLQ

WBC/ul

ANC/ul

CP ng/mL

CRP mg/L

Score

Ultrasound positive for appendicitis

Derivation

8 y/F

+

9900

6900

0.43

5.35

4

+

Derivation

9 y/M

11,000

6800

0.35

0.10

4

+

Derivation

12 y/F

+

9800

6400

0.07

0.46

4

+

Derivation

12 y/M

+

8500

5400

0.22

1.20

3

+

Derivation

13 y/M

9900

5900

0.88

5.88

3

+

Derivation

13 y/F

+

7200

3300

0.42

2.40

4

+

Validation

9 y/M

9500

6393

1.15

20.40

5

+

made subsequently in the computer system to see whether there were follow-up consultations; however, if they had gone to a private provider or only returned to our hospital after the check was performed in the computer system, a false negative could occur, and this might have a significant effect on the results of the study. Last, our score was not per- fect: 6 patients were missed in the derivation sample, and 1 patient was missed in the validation sample. Before PALabS is applied clinically, it must be validated in other clinical settings.

In conclusion, the PALabS is useful in identifying children with ab- dominal pain suggestive of AA who are at low risk of this disease. In pa- tients in whom PALabS <=6 Clinical monitoring could be utilized without the need for more complementary tests, such as ultrasound, or extend- ing the hospital stay. Moreover, PALabS might represent an aid in cases in which ultrasound is inconclusive or in centers in which ultrasound is not available, making it possible to postpone radiological tests such as CT, referrals to other centers, or even interventions, waiting instead to observe the patients’ progression. Next steps include validation in other clinical settings and evaluation of how PALabS may influence the care delivered.

Funding source

No external funding was secured for this study.

Financial disclosure

The authors have no financial relationships relevant to this article to disclose.

Contributors’ statement page

Javier Benito: Dr. Benito conceptualized and designed the study, su- pervised data collection, analyzed the data, wrote and critically revised the initial draft of the manuscript, and approved the final manuscript as submitted.

Santiago Fernandez: Dr. Fernandez collaborated in the study design, supervised data collection, wrote and critically revised the initial draft of the manuscript, and approved the final manuscript as submitted.

Miriam Gendive: Dr. Gendive collaborated in the design of the study,

participated in data collection and critically revised the manuscript.

Paula Santiago: Dr. Santiago collaborated in the design of the study, participated in data collection and critically revised the manuscript.

Raquel Perez-Garay: Dr. Perez-Garay collaborated in the design of the study, participated in data collection and critically revised the manuscript.

Eunate Arana-Arri: Dr. Arana-Arri collaborated in the design of the study, analyzed the data and critically revised the manuscript.

Santiago Mintegi: Dr. Mintegi collaborated in the design of the study, wrote and critically revised the initial draft and approved the final man- uscript as submitted.

Declaration of Competing Interest

The authors have no conflicts of interest to disclose.

References

  1. Hall MJ, DeFrances CJ, Williams SN, Golosinskiy A, Schwartzman A. National Hospital Discharge Survey: 2007 summary. Natl Health Stat Rep 2010;24(29):1-20.
  2. Addiss DG, Shaffer N, Fowler BS, Tauxe RV. The epidemiology of appendicitis and ap- pendectomy in the United States. Am J Epidemiol 1990;132:910-25.
  3. Bonadio W, Peloquin P, Brazg J, Scheinbach I, Saunders J, Okpalaji C, et al. Appendi- citis in preschool aged children: regression analysis of factors associated with perfo- ration outcome. J Pediatr Surg 2015;50:1569-73.
  4. Narsule CK, Kahle EJ, Kim DS, Anderson AC, Luks FI. Effect of delay in presentation on rate of perforation in children with appendicitis. Am J Emerg Med 2011;29:890-3.
  5. Doria AS, Moineddin R, Kellenberger CJ, Epelman M, Beyene J, Schuh S, et al. US or CT for diagnosis of appendicitis in children and adults? A meta-analysis. Radiology 2006;241(1):83-94.
  6. Lowe Lh, Penney MW, Stein SM, Heller RM, Neblett WW, Shyr Y, et al. Unenhanced limited CT of the abdomen in the diagnosis of appendicitis in children: comparison with sonography. AJR 2001;176(1):31-5.
  7. Trout AT, Sanchez R, Ladino-Torres MF, Pai DR, Strouse PJ. A critical evaluation of US for the diagnosis of pediatric acute appendicitis in a real-life setting: how can we im- prove the diagnostic value of sonography? Pediatr Radiol 2012;42:813-23.
  8. Hernanz-Schulman M. CT and US in the diagnosis of appendicitis: an argument for CT. Radiology 2010;255(1):3-7.
  9. Brenner DJ, Hall EJ. Computed tomography-an increasing source of radiation expo- sure. NEJM 2007;357(22):2277-84.
  10. Bachur GR, Levy JA, Callahan MJ, Rangel SJ, Monuteaux MC. Effect of reduction in the use of computed tomography on clinical outcomes of appendicitis. JAMA Pediatr 2015. https://doi.org/10.1001/jamapediatrics.2015.0479 Published online June 22.
  11. Aspelund G, Fingeret A, Gross E, Kessler D, Keung C, Thirumoorthi A, et al. Ultraso- nography/MRI versus CT for diagnosing appendicitis. Pediatrics 2014;133(4): 586-93 Apr.
  12. Alvarado A. A practical score for the early diagnosis of acute appendicitis. Ann Emerg

    Med 1986;15:557-64.

    Samuel M. Pediatric appendicitis score. J Pediatr Surg 2002;37:877-81.

  13. Kharbanda AB, Taylor GA, Fishman SJ, Bachur RG. A clinical decision rule to identify children at low risk for appendicitis. Pediatrics 2005;116(3):709-16.
  14. Ebell MH, Shinholser J. What are the Most clinically useful cutoffs for the Alvarado and pediatric appendicitis scores? A systematic review. Ann Emerg Med 2014: 365-372.e2.
  15. Kwan KY, Nager AL. Diagnosing pediatric appendicitis: usefulness of laboratory markers. Am J Emerg Med 2010;28:1009-15.
  16. Allister L, Bachur R, Glickman J, Horwitz B. serum markers in acute appendicitis. J Surg Res 2011;168(1):70-5.
  17. Huckins DS, Simon HK, Copeland K, Spiro DM, Gogain J, Wandell M. A novel bio- marker panel to rule out acute appendicitis in pediatric patients with abdominal pain. Am J Emerg Med 2013;31(9):1368-75.
  18. Benito J, Acedo Y, Medrano L, Barcena E, Garay RP, Arri EA. Usefulness of new and traditional serum biomarkers in children with suspected appendicitis. Am J Emerg Med 2016;34(5):871-6 May.
  19. Altman DG, Vergouwe Y, Royston P, Moons KG. Prognosis and prognostic research: validating a prognostic model. BMJ 2009;338:b605.
  20. Schisterman EF, Perkins NJ, Liu A, Bondell H. Optimal cut-point and its correspond- ing Youden Index to discriminate individuals using pooled blood samples. Epidemi- ology 2005;16:73-81.
  21. Kharbanda AB, Vazquez-Benitez G, Ballard DW, Vinson DR, Chettipally UK, Kene MV, et al. Development and validation of a novel pediatric appendicitis risk calculator (pARC). Pediatrics 2018;141(4). https://doi.org/10.1542/peds.2017-2699 pii: e20172699. Apr. [Epub 2018 Mar 13].
  22. Zuniga RV, Arribas JL, Montes SP, Fernandez MN, Abad CG, Matin LG, et al. Applica- tion of pediatric appendicitis score on the emergency department of a secondary level hospital. Pediatr Emerg Care 2012;28(6):489-92.
  23. Bachur RG, Callahan MJ, Monuteaux MC, Rangel SJ. Integration of ultrasound find- ings and a clinical score in the diagnostic evaluation of pediatric appendicitis. J Pediatr 2015;166:1134-9.
  24. Schneider C, Kharbanda A, Bachur R. Evaluating appendicitis scoring systems using a prospective pediatric cohort. Ann Emerg Med 2007;49:778-84 (784.e1).
  25. Goldman RD, Carter S, Stephens D, Antoon R, Mounstephen W, Langer JC. Prospec- tive validation of the pediatric appendicitis score. J Pediatr 2008;153:278-82.
  26. Kharbanda AB, Fishman SJ, Bachur RG. Comparison of pediatric emergency physi- cians’ and surgeons’ evaluation and diagnosis of appendicitis. Acad Emerg Med 2008;15:119-25.
  27. Bachur RG, Dayan PS, Dudley NC, Bajaj L, Stevenson MD, Macias CG, et al. The influ- ence of age on the diagnostic performance of white blood cell count and absolute neutrophil count in suspected pediatric appendicitis. Acad Emerg Med 2016;23: 1235-42.
  28. Mandeville K, Pottker T, Bulloch B, Liu J. Using appendicitis scores in the pediatric ED. Am J Emerg Med 2011;29:972-7.
  29. Cayrol J, Miguez MC, Guerrero G, Tomatis C, Simal I, Maranon R. Diagnostic accuracy and prognostic utility of D dimer in acute appendicitis in children. Eur J Pediatr 2016;175:313-20.
  30. Sivitz AB, Cohen SG, Tejani C. Evaluation of acute appendicitis by pediatric emer- gency physician sonography. Ann Emerg Med 2014;64(358-64):e4.
  31. Andersson RE. Meta-analysis of the clinical and laboratory diagnosis of appendicitis. Br J Surg 2004;91:28-37.
  32. Kharbanda AB, Rai AJ, Yohaimi C, Liu K, Dayan PS. Novel serum and urine markers for pediatric appendicitis. Acad Emerg Med 2012;19:56-62.
  33. Bealer JF, Colgin M. S100A8/A9: a potential new diagnostic aid for acute appendicitis. Acad Emerg Med 2010;17:333-6.
  34. Mills AM, Huckins DS, Kwok H, Baumann BM, Ruddy RM, Rothman RE, et al. Diag- nostic characteristics of S100A8/A9 in a multicenter study of patients with acute Right lower quadrant abdominal pain. Acad Emerg Med 2012;19(1):48-55.
  35. Gavela T, Cabeza B, Serrano A, Casado Flores J. C-Reactive protein and procalcitonin are predictors of the severity of acute appendicitis in children. Pediatr Emerg Care 2012;28(5):416-9.
  36. Oikonomopoulou N, Miguez-Navarro Rivas-Garcia A, Garcia Gamiz M, Lopez-Lopez R, Oliver-Saez P, Riano-Mendez B, et al. Assessment of proadrenomedullin as diag- nostic or Prognostic biomarker of acute appendicitis in children with acute abdom- inal pain. Am J Emerg Med 2018;18:30793-9. https://doi.org/10.1016/j.ajem.2018.

    09.038 Sep 26. pii: S0735-6757. [Epub ahead of print].

    Saito JM, Yan Y, Evashwick TW, Warner BW, Tarr PI. Use and accuracy of diagnostic imaging by hospital type in pediatric appendicitis. Pediatrics 2013;131(1):e37-44.

  37. Hennelly KE, Bachur R. Appendicitis update. Curr Opin Pediatr 2011;23(3):281-5.
  38. Benabbas R, Hanna M, Shah J, Sinert R. Diagnostic accuracy of history, physical ex- amination, laboratory tests, and point-of-care ultrasound for pediatric acute appen- dicitis in the emergency department: a systematic review and meta-analysis. Acad Emerg Med 2017;24(5):523-51. https://doi.org/10.1111/acem.13181 May.
  39. Vazquez MA, Morteruel E, Garcia E, Mintegi S, Capape S, Benito J. diagnostic yield of Abdominal ultrasonography in acute appendicitis. An Pediatr (Barc) 2003;58(6): 556-61.
  40. Kennedy TM, Thompson AD, Choudhary AK, Caplan RJ, Schenker KE, DePiero AD. Utility of applying white blood cell cutoffs to non-diagnostic MRI and ultrasound studies for suspected pediatric appendicitis. J Emerg Med 2018. https://doi.org/10. 1016/j.ajem.2018.12.029 pii: S0735-6757 (18)30989-6. [Epub ahead of print].