The utility of sonographic signs to diagnose simple and complicated appendicitis in children

Our tertiary paediatric surgical centre has a dedicated paediatric radiology department with rotating sonographers who perform USS, and paediatric radiologists (on-call 24-h) that report them. We utilise a Samsung RS85 ultrasound, with multiple convex (C3-10, C4-9) and linear (L3-12) transducers for each abdominal examination. When scanning the appendix, Puylaert’s graded compression technique along with various visualisation strategies are used [8]. These include left lateral decubitus positioning, posterior manual compression, micturition, and respiration; with a second sonographer requested where feasible. Scanning is performed using a standardised sonographer worksheet (Supplementary Material). Operators must indicate appendiceal visualisation and confirmation of blind-ending, as well as four primary sonographic signs: maximal diameter (mm), hyperaemia, compressibility, absent luminal gas; and three secondary signs: probe tenderness, peri-appendiceal fluid and peri-appendiceal echogenic fat.

All children presenting to the Emergency Department (ED) with a principal complaint of abdominal pain were retrospectively considered for inclusion between January 2017 and May 2019. Consecutive USS reports including the terms “appendix” or “appendicitis” were screened. To verify they were performed for suspected appendicitis, correlation with the electronic hospital records and the clinical indications were reviewed. USS were excluded if primarily investigating chronic non-specific abdominal pain, intussusception, or a known appendiceal mass. Children with previous appendicectomy were excluded.

USS were grouped into four operative outcomes: No Appendicectomy (NoA), Negative appendicectomy (NegA), SA and CA. NoA was confirmed via clinical follow-up for 12 months, NegA was confirmed on histology in the absence of acute inflammation. SA was defined as non-perforated, incorporating inflammatory changes or gangrenous appendices. CA was intraoperatively defined as macroscopic perforation, intraperitoneal appendicolith or four-quadrant pus [1]. CA also included clinically or radiologically detected appendiceal abscess or mass. If a child had undergone two USS before intervention, the former was classified as NoA, with the latter in the appropriate operative category; assuming this aided the decision for surgery. When an intraoperative result suggested SA, and histology demonstrated CA; intraoperative results were utilised. This reflects clinical management and acknowledges potential iatrogenic perforation during appendicectomy. Clinical follow-up was for 12 months.

HREC approval was granted before commencement (Local reference: RES-20-0000250Q-63608), with informed consent waived.

The median age was 10.7 years [8.0–13.4], with 54% being female. Children with CA were significantly younger than children with SA (p < 0.001); 9 [6‐13] and 12 [9‐14] years, respectively (Table 1). Females comprised a significant majority (92%) of the NegA group (p < 0.001).

Most USS did not result in a surgical intervention, with 61% having NoA (n = 571), 3% had a NegA (n = 24), 24% had SA (n = 226) and 12% had CA (n = 113). Appendix visualisation rate on USS was 61% (n = 569), with 62% (n = 580) having a conclusive report (including where other diagnoses were sonographically identified n = 137).

Sonographic signs which suggested appendicitis included an appendiceal diameter above 7 mm (AUC 0.92, [95% CI: 0.90–0.94] sensitivity 86%, specificity 86%) (Fig. 2), presence of an appendicolith (p = 0.003), hyperaemia (p = 0.001), non-compressibility (p = 0.03), and no luminal gas (p = 0.004) (Table 2). When these latter four significant variables were grouped, the regression model correctly classified 80% of cases with AUC 0.87 [95% CI: 0.83–0.90]. Secondary signs included probe tenderness (p < 0.001) and presence of peri-appendiceal echogenic fat (p < 0.001). When these two variables were grouped, the regression model correctly classified 82% of cases with AUC 0.85 [95% CI: 0.82–0.89]. Overall, we found USS had diagnostic accuracy of 89%, with sensitivity 84.7% [CI: 80.4–88.1] and specificity 92.1% [CI: 90.0–94.0].

Sonographic signs which suggest CA over SA comprised a diameter above 10.1 mm (AUC 0.63, [95% CI: 0.57–0.69] presence of an appendicolith (p = 0.003) and the secondary sign of peri-appendiceal fluid (p = 0.004) (Table 3). When the two primary signs were grouped, the regression model correctly classified 73% of cases with AUC 0.68 [95% CI 0.60–0.75]. To diagnose CA over SA, we found USS had diagnostic accuracy of 82%, with sensitivity 52.2% [CI: 42.1–62.1] and specificity 96.4% [CI: 92.8–98.3].

Our sonographic worksheet was utilised in 83% of USS, with significantly lower use in the NegA group (71%) compared to other operative outcomes (p = 0.006). Conclusive USS reports matched the operative outcomes over 90% of the time (Fig. 3). A fifth of children (18%) who had an inconclusive USS had surgery (8% SA, 5% CA, and 5% NegA). When the appendix was not viewed, the presence of secondary signs did not often lead to surgery (26%). If neither the appendix nor secondary signs were viewed (n = 215), appendicitis was the diagnosis in 9% of cases.

Visibility of sonographic signs is markedly reduced if the appendix itself is not visualised. Our study had an overall 62% conclusive scan rate, closely linked to our overall 61% appendix visualisation rate. Studies of diagnostic accuracy of USS for paediatric appendicitis report a wide range of visualisation, with two Australian studies reporting 41% and 92% [5, 7]. Cundy et al. attribute their superior visualisation rate to diligent paediatric sonographers who employ multiple techniques and use tightly curved transducers more suited to smaller patients [6]. Reddan et al. were able to improve visualisation rates to 69% in a subsequent study, following sonographer training and implementation of a worksheet [13]. Aside from implementation of a worksheet, other institutional variables such as sonographer experience, type of hospital (regarding paediatric volume), and time of day also contribute [22]. Visualisation is also dependent on patient factors, such as duration of symptoms and their clinical presentation pre-USS. Patients who have a high suspicion of appendicitis could have developed more sonographic signs than clinically equivocal patients [7, 12, 22, 24].

Our study demonstrated appendiceal visualisation rates of 92% in SA, with only 48% NoA viewed. One explanation is that non-inflamed appendices are more difficult to visualise due to their smaller size and absent secondary features. However, our study included a higher appendicitis prevalence (36%) than Cundy et al. (28%) implying a smaller patient volume in our study but also suggesting consistent visualisation is still possible. Our CA visualisation rate (72%) was lower than SA, which we attribute to the difficulty delineating an appendix amongst associated inflammatory changes.

With non-visualisation of the appendix, secondary signs were present in 16% of USS, well within the range of 5–23% reported in the literature [10, 24‐26]. In our cohort, a quarter of this group had appendicitis (26%), and 69% of these were CA. This relatively low rate of USS diagnosed appendicitis compared to Partain et al. (42%), is likely due to their stricter definition of a secondary sign including a ‘significant amount of fluid’ rather than our ‘peri-appendiceal fluid’, which is often physiological [12]. However, Held et al. report a 17% appendicitis rate, likely explained by the majority of their USS being in the non-visualised category (76%) [10]. When neither the appendix nor secondary signs were viewed, our appendicitis rate was within the range of previously reported values 2–9% [10, 12, 24]. It has been suggested that these children can be observed or discharged with increased confidence, although it remains case dependent.

Overall, our reported accuracy was inferior to previous meta-analyses reporting pooled sensitivity (88–89%) and specificity (94–97%) [27, 28]. However, this included heterogenous data from 12,926 children from different institutions. For example, some centres excluded non-diagnostic USS from analysis, which overestimate the true accuracy. For diagnosing CA over SA, our sensitivity was higher than previously reported values of 23–44%, with specificity similar at 93–100% [17‐19, 29].

Our sonographic features were analysed using an established institutional worksheet. This provides sonographers with a conventional framework to ensure a systematic and thorough examination and ensures a standardised outcome source which increases the quality of our retrospective study. The application of a standardised worksheet is lacking in existing published literature [9, 10, 12, 19‐21]. Therefore, their retrospective data are limited to sonographic signs that radiologists report, which means the relevant absence of signs may not be detailed.

The maximal appendiceal diameter is the most commonly studied sonographic sign. Our service uses 6 mm as a threshold, (sensitivity 97%, specificity 69%). However, our analysis suggests 7 mm is more accurate (sensitivity 86%, specificity 86%), particularly in the NegA group. This is similar to other studies, with some suggesting three categories [30‐33]. We suggest balancing the statistical superiority of a more specific 7 mm threshold with the clinical risk of false negatives.

An appendicolith was a significant predictor of appendicitis in our cohort with specificity 94% (adjusted OR 1.15, p = 0.003), similar to Partain et al. (adjusted OR 7.9 [95% CI 1.7–37.2]) and Telesmanich et al. (OR 15.8 (p = 0.03)), with Trout et al. finding no association [12, 14, 15]. An appendicolith has been generally associated with failure of non-operative management; however, none of the “appendicolith-positive” patients in our series (6%) developed appendicitis after 12 months follow-up; a longer follow-up will be needed to confirm this finding.

Hyperaemia on doppler was most frequently seen in SA (86%), and significantly associated with appendicitis. This is consistent with most previous studies [12, 15, 34, 35].

Non-compressibility on probe pressure significantly predicted appendicitis, and had increased prevalence in SA. Potentially CA may have been decompressed with a perforation at the time of USS, or the appendiceal structure could be distorted amongst heterogenous inflammatory echoes. This sign is uncommonly reported in the literature [15].

Absence of luminal gas was significantly associated with appendicitis and most frequent in SA. There is a paucity of reported data studying this sign, with our data suggesting it has utility to rule in appendicitis (specificity 92%), with its absence having little value (sensitivity 43%).

Focal pain with transducer pressure is uncommonly studied due to its similarity to palpation [11]. However, we found significant association with appendicitis, present in 97%, and probe tenderness carried the highest sensitivity (97%) and lowest specificity (44%) of all the sonographic signs. Its relative frequency in children who had a NegA (92%), could suggest clinicians value clinical right iliac fossa tenderness for diagnostic evaluation.

Peri-appendiceal fluid was not significantly associated with SA, consistent with the literature [9, 11, 13‐15, 35]. However, other studies reported significance of right lower quadrant fluid, with some additionally finding free fluid [10, 12, 35]. This emphasizes the importance of location, volume and character descriptors, as free fluid can be physiological in small amounts, whereas heterogenous fluid can suggest pathology.

The presence of peri-appendiceal echogenic fat was most associated with appendicitis in our cohort, which is consistent with the published literature [11, 14, 15, 35].

The statistically optimal threshold maximal appendix diameter was 10.1 mm for CA (sensitivity 54%, specificity 75%), which is larger than previously studied (9 mm) [21]. An appendicolith was a significant predictor of CA, which has been reported previously [16, 17, 20, 21, 23]. Visualisation of a blind-ending structure is included as a confirmatory variable on our worksheet to ensure viewing of the correct anatomy. It was negatively associated with CA which could be due to difficulty identifying the appendix tip amongst inflammatory tissue. Hyperaemia, non-compressibility, and absent luminal gas were not associated with CA in our study although some previous studies have found them to be helpful [17, 22].

Probe tenderness was negatively associated with CA, which was unexpected with unclear clinical significance. It has not been previously studied in the literature and reflects a subjective, non-specific yet sensitive sign of CA. Peri-appendiceal fluid was detected most frequently and significantly associated with CA, although it was also present in both SA and children without appendicitis. Heterogenous fluid has been reported as the most predictive sign for CA [17, 20, 22, 23]. We hypothesize this finding reflects perforation with leaked intra-luminal content or inflammatory exudate, and should be further characterised in a sonographic worksheet or report. Echogenic fat was not associated with CA as it was also common in SA, consistent with most previous studies [16, 17, 20‐23].

Strengths of our study include a large study population over a relative short period and the use of a standardised worksheet, reducing the variability in reporting, and ensuring accurate data collection. Limitations of our study stem from its retrospective design. Our literature review revealed other sonographic signs reported to be associated with appendicitis, such as appendiceal mural thickness and bowel signs which were not assessed in our study. Peri-appendiceal fluid is one of the signs found to be good differentiator between SA and CA; however, our study did not specifically address if it was simple or heterogeneous; we plan to add this information to our worksheet in the future. Investigating inter-observer reliability between sonographers or radiologists was another potential limitation. This would have allowed the operator-dependence of USS to be quantified and whether experience of sonographers improved their performance. Additional patient data such as body mass index, clinical presentation details, time from onset of symptoms to USS, and complication rates could enable a thorough investigation into factors affecting USS accuracy. It would also be useful to explore the clinician’s views about USS and appendicectomy, allowing us to gauge their perceived utility and how certain findings influence their management.