Ultrasound for the Diagnosis of Malrotation and Volvulus in Children and Adolescents: A Systematic Review and Meta-analysis

Abstract

Context:

Despite the advantages of ultrasound (US), upper gastrointestinal contrast series (UGI) remains the first line diagnostic modality in the diagnosis of midgut malrotation and volvulus in children.

Objective:

Evaluate the diagnostic accuracy of US in the diagnosis of malrotation with or without volvulus in children and adolescents aged 0–21 years, compared to the reference standard (diagnosis by surgery, UGI, CT, MRI, and clinical follow-up individually or as a composite).

Data sources:

We searched the electronic databases Ovid-MEDLINE, Embase, Scopus, CINAHL, and the Cochrane library in October 2019 and updated on August 18, 2020.

Study selection:

Studies evaluating the diagnostic performance of US for diagnosis of midgut malrotation with or without volvulus in children (0–21 years).

Data extraction and synthesis:

The data were extracted independently by 2 authors and a bivariate model was used for synthesis.

Results:

Meta-analysis of 17 cohort or cross-sectional studies and 2257 participants estimated a summary sensitivity of 94% (95% CI, 89%−97%) and summary specificity of 100% (95% CI, 97%−100%) (moderate certainty evidence) for the use of US for the diagnosis of malrotation with or without midgut volvulus compared to the reference standard. Subgroup analysis and meta-regression revealed better diagnostic accuracy in malrotation not complicated by volvulus, in the neonatal population and enteric fluid administration before US.

Conclusions:

Moderate certainty evidence suggests excellent diagnostic accuracy and coupled with the advantages, a strong case exists for the use of abdominal US as the first line diagnostic test for suspected midgut malrotation with or without volvulus in children and adolescents.

INTRODUCTION

Midgut malrotation is a congenital abnormality that, if complicated by volvulus, becomes a surgical emergency (1, 2). Although asymptomatic malrotation may remain undiscovered, when symptomatic, 75% will present before the age of 5 (3–5). Delayed diagnosis of midgut volvulus can lead to necrosis of the midgut, serious morbidity and mortality.

In the late 1960s, contrast upper gastrointestinal series (UGI) emerged as the preferred modality for evaluation of midgut malrotation and volvulus (6, 7). UGI requires contrast by mouth or via enteric tube and requires exposure to ionizing radiation. Diagnosis of malrotation on UGI is made by evaluating the position of the duodeno-jejunal junction (DJJ), which is a proxy for the ligament of Treitz, and looking for signs of obstruction and/or volvulus. UGI is performed by a radiologist or radiology practitioner assistant. Results of UGI can be difficult to interpret, since this three dimensional problem is being interpreted with two dimensional images (8, 9).

Since the first report of ultrasound to diagnose malrotation in 1987 (10) and the first description of the “whirlpool sign” to diagnose midgut volvulus in 1992 (11), ultrasound has been increasingly used for evaluation of midgut malrotation, especially when complicated with volvulus (11). Midgut malrotation can be diagnosed on US if the third portion of the duodenum (D3) is intraperitoneal as opposed to retroperitoneal between the SMA (12) and aorta and/or by reversal of the SMA and SMV relationship (13–25). Malrotation with superimposed midgut volvulus occurs when the midgut twists on the narrow mesenteric pedicle of the malrotated gut, and can be diagnosed by a whirlpool sign on US, which is direct visualization of the SMV and proximal small bowel swirling around the SMA (11, 26). Increasingly, studies have been reported in the last two decades on the diagnostic accuracy of US (13–25, 27–30). The advantages of US in the pediatric population include portability to the bedside, absence of ionizing radiation and faster to perform compared with the necessity for oral contrast and enteric tubes in UGI (31). However, US is operator dependent and limited by air interface, which may interfere with the visibility of the intraabdominal structures. The use of ultrasound as the primary imaging modality to diagnose malrotation without or with midgut volvulus remains debatable within the pediatric radiology and surgery communities (32, 33). Studies have reported that reversal of SMA/SMV position is not always sensitive or specific for diagnosis of malrotation (15, 34–36).

We performed a systematic review and meta-analysis to address this clinical conundrum. Our primary objective was to assess the diagnostic accuracy of abdominal US for the evaluation of midgut malrotation and volvulus in children and adolescents aged 0–21 years compared to the reference standard (diagnosis by surgery, UGI, computed tomography (CT), magnetic resonance (MR), clinical follow-up or any combination). Our secondary objective was to explore heterogeneity among studies by subgroup analyses and meta-regression.

METHODS

The systematic review was performed and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy (PRISMA-DTA) guidelines (37). The protocol was registered on PROSPERO (doi: CRD42019128332) (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=128332).

Literature Search

We systematically searched in Ovid-MEDLINE, Embase, Scopus, CINAHL, and Cochrane library up to October 30, 2019 and updated the search on Aug 18, 2020, with the assistance of a librarian (AS). The search strategies for the databases searched are reported in eAppendix 1. There were no language restrictions and one article in Chinese was evaluated and included.

Inclusion Criteria

Studies were included if they satisfied all of the following criteria: (1) patients (0–21 years) who underwent US for suspected midgut malrotation and/or volvulus, (2) included any or a combination of the following reference standards: surgery, UGI contrast series, CT or MR which visualize the third portion of D3, and/or clinical follow-up and (3) reported results for construction of 2 × 2 tables.

Exclusion Criteria

Studies were excluded if : (1) case reports, case series or case-control studies where false positives or false negatives were not available, (2) studies with a target condition other than midgut malrotation and/or volvulus, (3) studies where data for age group 0–21 years cannot be derived and (5) letters, reviews, opinions, meta-analyses, consensus statements, and guidelines.

Data Extraction and Quality Assessment

Data were extracted and managed using REDCap (Vanderbilt University, Nashville, TN) electronic data capture tool(38) hosted at Texas Children’s Hospital, Houston, USA. The details of study data extracted are reported in the supplementary file. The methodological quality of each study was assessed using the Quality Assessment of Diagnostic-Accuracy Studies (QUADAS-2) tool(39) by two authors independently (H.N., M.K.). The independent assessment was managed using REDCap (38). The certainty of evidence was assessed according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) for diagnostic test studies (40–43). GradPro software was used to generate summary of findings (SoF) table (44).

Data Synthesis and Analysis

The results of the index test were dichotomous and the bivariate random-effects approach was used to calculate the summary estimates of sensitivity and specificity with 95% confidence intervals (CI) (45). The bivariate model deals with variation beyond chance in sensitivity and specificity between studies and any correlation that may exist between sensitivity and specificity. Publication bias was assessed using the Deek’s funnel plot and Deeks’ asymmetry test (46).

We performed separate meta-analyses of the two US signs, (1) inversion of the SMA and SMV for malrotation, defined as the SMV to the left of the SMA and (2) whirlpool sign for midgut volvulus, identified as clockwise wrapping of the SMV and mesentery around the SMA, since they were reported separately in some studies. We performed meta-regression with the following covariates: (1) study design (prospective vs. retrospective), (2) clinical follow-up (included vs. not included in the reference standard), (3) prevalence (higher vs. lower than the summary prevalence of the include studies, 17%), (4) use of enteric fluid, (5) age group (infants vs. all children, and (6) number of US operators and/or interpreters (single vs. multiple). Exploratory analysis using the Moses-Littenberg SROC method (45), by fitting a separate receiver operating characteristic (ROC) curve, which provides visual qualitative inference of associations of covariates but not statistical significance, was used when meta-regression was not possible due to fewer studies or sparse data. P values <.05 were considered as having statistical significance. We calculated summary estimates of sensitivity and specificity using ‘xtmelogit’ and publication bias (funnel plot and Deek’s test) using „metandi‟ and MIDAS packages (47–49) for Stata (Stata Corp, College Station, TX). We generated forest plots with 95% confidence intervals (CIs) for sensitivity and specificity in Stata and SROC plots in Review Manager 5 (50). We calculated likelihood ratios for evaluating post-test probabilities using the Fagan nomogram. We performed meta-analysis of studies reporting UGI as the index test for comparison.

RESULTS

Literature search

The process of inclusion of studies starting from 3,536 publications to 17 included studies (13–25, 27–30) are reported in Figure 1. Fifteen articles were excluded after reviewing full text (etable1).

Figure 1. PRISMA flow diagram.

Study selection process for this systematic review and meta-analysis.

Characteristics of Included Studies

The detailed patient and ultrasound characteristics are reported in Table 1 and Table 2, respectively. The study size ranged from 23 to 427 patients. The methodological assessment of the included studies using QUADAS-2 is summarized in Figures 2A and 2B. Unclear risk of bias was seen in less than 40% of the studies (13, 14, 16, 18, 19, 23, 24, 27, 28, 30). High and unclear risk of applicability concerns were seen in less than 20% of the studies (13, 19, 23, 24). One study had a high risk of applicability concern because it evaluated the relationship of the SMA and SMV and midgut malrotation in neonates being investigated for pyloric stenosis (13). The GRADE certainty of evidence was moderate (downgraded for risk of bias and inconsistency, upgraded for strong association or high diagnostic accuracy) (Table 3).

Table 1.

Characteristics of the Included Studies

Study id/Country	Participants	Clinical setting	Study Design	Target Condition	Index Test	Sample size/M:F ratio	Reference Standard
Wong et al.32 2020 USA	Children (0 – 20 yr) referred for MGV by US	Presenting to a CH for imaging	R	MGV	US – WP sign	195 N/A	UGI, S, F
Karaman et al.28 2018 Turkey	Children (days – 15.6 yr) referred for malro by US	Presenting to a CH for imaging	R	Malro	US – SMA/V & WP	82 M49:F33	UGI, S
Taghavi et al.29 2018 Iran	Children (2 days – 15 yr) referred for malro by US-Some asymptomatic cases with abnormal US or symptomatic cases with normal US also had an UGI	Presenting to a CH for imaging	P, cross-sectional study	Malro	US – SMA/V	67 M35:F32	UGI, S
Kumar et al.26 2017 India	Children (median age of 12 months) referred for malro by US	Presenting to a CH for imaging	R	Malro	US – SMA/V& WP	52 M42:F10	S
Zhang et al.25 2017 China	Neonates referred for suspected malro and volvulus by US	Presenting to a teaching hospital for imaging	P	Malro, MGV	US – SMA/V & WP	83 M45:F38	UGI, O, S
Dao et al.33 2015 USA	Children (1 day – 16 yr) referred for MGV by US	Presenting to a CH for imaging	R	MGV	US - WP	245 US/226 pts N/A	UGI, S, F-Some have UGI, no F for negative US cases
Zhou et al.30 2015 China	Children (2 days – 13 yr) referred for malro by US and UGI	Presenting to a CH for imaging	P	Malro	US – SMA/V, D3, & WP	70 M44:F26	UGI, O, S, F-F at least 1 year to confirm true negative
Esposito et al.34 2014 Italy	Children (1 day – 12 yr) with proven malro by S with preceding US	Cases from S	R	MGV	US - WP	34 M10:24F	S
Hennessey et al.27 2014 Australia	Children (median age 47 days) referred for malro and volvulus by US first, then UGI or S	Presenting to a CH for imaging	R	Malro, MGV	US – SMA/V & D3	139 M83:F56	UGI, S, F-F time mean 575 days
Alehossein et al.22 2012 Iran	Neonates (1 – 27 days) presenting with bilious vomiting with US first, then UGI or S	Presenting to a CH for imaging	P	Malro	US – SMA/V, D3, WP	23 M13:F10	UGI, S, F
Yang et al.21 2011 China	Children (1 day – 15 yr) referred for malro by US first, then UGI or S	Presenting to a CH for imaging	R	MGV	US –WP	138 M87:F51	UGI, S
Orzech et al.19 2006 Canada	Children (newborn – 17 yr) referred for malro by US first then UGI or S	Presenting to a CH for imaging	R	Malro	US –SMA/V, WP	211 M110:F101	UGI, S
Chao et al.23 2000 Taiwan	Neonates referred for suspected malro and MGV by US	Presenting to a CH for imaging	P	Malro, MGV	US – SMA/V, proximal D appearance, WP, ascites, dilatation of distal SMV	31 M11:F9	S
Shimanuki et al.35 1996 Japan	Children (0 day – 14 yr) referred for MGV by US first then UGI or S	Presenting to a CH for imaging	R	MGV	US –WP	236 M160:F76	UGI, S
Dufour et al.20 1992 Belgium	Children (95% infants) w/recurrent abdominal pain and/or vomiting presenting for UGI had preceding US looking at SMA and SMV relationship	Presenting to a CH for imaging	P	Malro	US – SMA/V	427 N/A	UGI, S
Weinberger et al.18 1992 USA	Infants (2–218 days) referred for pyloric stenosis; SMA/SMV relationship on US and correlated with malrotation by UGI or S	Presenting to a CH for imaging	R	Malro	US – SMA/V	337 M239:F98	UGI, S
Cohen et al.24 1987 USA	Infants referred for vomiting and suspected malrotation with US first then UGI or S	Presenting to a CH for imaging	R	Malro, MGV	US – proximal duodenum appearance, D3	261 N/A	UGI, S

Characteristics of included studies. Abnormal relationship of SMA and SMV was used to determine malrotation and the whirlpool sign was used to determine MGV on US unless specified otherwise. Abbreviations: USA= United States of America; R= retrospective; P= prospective; SMA= superior mesenteric artery; SMV= superior mesenteric vein; D= duodenum; D3= third portion of the duodenum; US = ultrasound; UGI= upper gastrointestinal contrast series; Malro=midgut malrotation; MGV= midgut volvulus; M= male; F= female; S= surgery; F= clinical follow-up; O= Other imaging studies which may include computed tomography, and magnetic resonance imaging; N/A= none available; pts=patients; CH=Children’s hospital; peds rad= pediatric radiologist. Funding sources were reported by only one study.

Table 2.

Ultrasound Characteristics and Design of the Included Studies

Study	Vendor (Model) and probe frequency	Neonates & Infants group vs. All ages	Enteric fluid (Yes/No)	Multiple interpreters (Yes/No/Unsure)
Wong et al.32	N/A	All ages	No	No – retrospective review by 1 blinded pediatric radiologist
Karaman et al.28	Toshiba (Aplio 500) 7.5 MHz linear	All ages	No	No – performed by 1 pediatric radiologist with 10 yr experience
Taghavi et al.29	N/A 3.5 MHz curved & 7.5 – 12 MHz linear	All ages	No	Yes – performed by 2 pediatric radiologists
Kumar et al.26	GE (Logiq E9) Linear for children & curvilinear for older children	All ages	No	Unsure
Zhang et al.25	Philips (iU22) C5-1 MHz & L12-5 MHz	Neonates/Infants	Yes, water w/ NGT	No – performed by 1 pediatric sonographer
Dao et al.33	N/A	All ages	No	Unsure
Zhou et al.30	Toshiba & Hitachi Linear & curved	All ages	No	No – performed by 1 pediatric radiologist with 3 yr of experience
Esposito et al.34	Esaote (MyLab Twice) 6 to 10MHz microconvex , 7.5 to 10 MHz linear	All ages	No	Unsure
Hennessey et al.27	N/A	All ages	Yes	Yes – peds sonographers (15–30 yr experience) & radiologists (5–30 yr experience)
Alehossein et al.22	Siemens (G50) 7.5 convex and 10MHz linear	Neonates/Infants	Yes, water w/ NGT	No – performed by 1 pediatric radiologist
Yang et al.21	GE (Vivid 7), GE (Logiq S6) and Siemens (Sequoia 512) Convex & linear array 5 – 8 M Hz	All ages	No	Unsure
Orzech et al.19	N/A	All ages	No	Yes
Chao et al.23	Siemens (Acuson 128) 5 MHz curved for children; 7 MHz curved for newborns	Neonates/Infants	Yes, water w/ NGT	No – performed by 1 gastroenterologist
Shimanuki et al.35	Siemens (Quantum 2000) 5 MHz convex, 7.5 MHz linear	All ages	No	Unsure
Dufour et al.20	Siemens (Acuson 128) 5 MHz linear probe	All ages	No	Unsure
Weinberger et al.18	N/A	Neonates/Infants	No	Unsure
Cohen et al.24	ATL (MK) 7.5 MHz, occasionally 5 MHz	Neonates/Infants	Yes, water w/ NGT	Unsure

Ultrasound Characteristics and Design of the Included Studies. N/A = none available, MHz = megahertz; GE = General Electric; ATL = Advanced Technology Laboratories; NGT = nasogastric tube

Figures 2a and 2b: Methodological assessment of the included studies by QUADAS-2.

2a risk of bias and applicability concerns summary: review authors’ judgements about each domain for the included study and 2b risk of bias and applicability concerns graph: review authors’ judgements about each domain presented as percentages across included studies.

TABLE 3-.

SUMMARY OF FINDINGS TABLE

Explanations

^a.Clinical applicability concerns in one study and unclear risk of bias in 40% of studies

^b.Variability in study design, clinical pathway and heterogeneity

True positives: patients with malrotation and midgut volvulus, False negatives: patients incorrectly classified as not having malrotation and midgut volvulus, True negatives: patients without malrotation and midgut volvulus, False positives: patients incorrectly classified as having malrotation and midgut volvulus. We present true positives, false negatives, true negatives and false positives at the summary prevalence from included studies of 17% and two lower prevalence of 10 and 5 as prevalence lower than our summary prevalence have been reported in literature.

^cCertainty of evidence assessed by GRADE was moderate, downgraded for risk of bias and inconsistency, upgraded for strong association or high diagnostic accuracy)upgraded for strong association or high diagnostic accuracy.

Diagnostic Performance of Ultrasound

The sensitivity and specificity of the studies for midgut malrotation with or without midgut volvulus are depicted in Figure 3 and the summary sensitivity was 94% (95% CI, 89%−97%) and the summary specificity 100% (95% CI, 97%−100%) by meta-analysis. The summary prevalence for midgut malrotation with or without volvulus was 17% (range 2–83%). The studies plotted in the ROC space with summary curve and summary points and 95% confidence and prediction regions are depicted in Figure 4. The positive likelihood ratio was 317 and negative likelihood ratio was 0.06, meaning a positive test on US increases the probability of having malrotation without or with volvulus to 98% and a negative US test decreases the probability to 1% (efigure 1).

Figure 3: Forest plot.

Sensitivity and specificity for the individual included studies for ultrasound for midgut malrotation with or without volvulus are shown in separate left and right panels. The summary estimate is shown by the diamond and the vertical dotted line passes through the middle of the diamond outlining the summary estimate for sensitivity and specificity. Forest plots allow for visual assessment of variability among studies and heterogeneity statistics, Cochrane Q statistic is reported separately for sensitivity and specificity.

Figure 4: Receiver operator characteristics summary curve.

Included studies are plotted in the Receiver operator characteristics (ROC) space. A summary curve (black uninterrupted curve), summary point (black dot) from the meta-analytic estimates are depicted. The smaller ellipse with closely dotted line (95% confidence region), and the larger ellipse with spaced dotted line (prediction region for future study estimates) are also depicted in the figure.

Variability (heterogeneity) is to be expected in diagnostic accuracy studies and often due to differences in clinical study design, patient spectrum, prevalence of the target condition, study quality or unexplained (45, 51). There was considerable heterogeneity with Higgins I² statistic of 92 (84–100). The likelihood of publication bias, assessed by the Deek’s funnel plot and test, was not statistically significant (p=.86) (efigure 2).

Subgroup Analyses and meta-regression

We performed separate meta-analyses for studies reporting US signs for malrotation without volvulus and malrotation complicated by volvulus (Figure 5). The summary sensitivity was 95% (95% CI, 89%−98%) and summary specificity 100% (95% CI, 96%−100%) for US detection of inversed SMA and SMV relationship for malrotation (13 studies, 1568 participants). US detection of the whirlpool sign as an indicator of midgut volvulus had a summary sensitivity of 92% (95% CI, 82%−97%) and summary specificity of 99% (95% CI, 96%−100%) (8 studies, 876 participants). There was considerable heterogeneity with US for malrotation with Higgins I² statistic of 88 (76–100) and 0 (0–100), respectively. The likelihood of publication bias, assessed by the Deek’s funnel plot and test, was not statistically significant for US for diagnosis of malrotation (p=.68) or for malrotation with volvulus (p=.07).

Figure 5:

Comparative ROC curves for ultrasound (USS) used for diagnosis of malrotation without volvulus and US for the diagnosis of malrotation complicated by volvulus. a) Summary curve for ultrasound detection of malrotation with or without volvulus is black and the studies represented by black open oval shapes adjusted for sample size and b) Summary curve for ultrasound detection of midgut volvulus (MGV) is red and the studies represented by red diamonds adjusted for sample size.

The covariates evaluated by multiple univariate meta-regression are reported in efigure 3. The covariates, administration of enteric fluid (efluid) and clinical follow-up had better diagnostic accuracy. In the joint model (indicating both sensitivity and specificity in the meta-regression), only efluid showed better accuracy and those with clinical follow up showed trends with difference (p=0.07) but not statistically significant. Studies with prospective study design or prevalence above or below the median prevalence (17%) did not contribute to better accuracy of the diagnostic accuracy of US. The ROC curves showed better diagnostic accuracy in the neonate and infants group compared to children of all ages and in studies with a single US interpreter (compared to multiple interpreters) (efigures 4A and 4B).

Meta-analysis of UGI as the index test for the diagnosis of malrotation with or without volvulus (11 studies, 843 participants), estimated a summary sensitivity of 0.91(95% CI, 84%−96%) and summary specificity of 0.94 (95% CI, 72%−99%) (efigure 5). No significant difference between the sensitivities of US and UGI (p=0.533, Fisher’s exact test), but US had a significantly higher specificity (p=0.037).

DISCUSSION

We report the first systematic review and meta-analysis of studies evaluating the diagnostic accuracy of abdominal US for midgut malrotation with or without volvulus in children. Meta-analysis for US detection of malrotation with or without volvulus estimated the summary sensitivity to be 94% (95% CI, 89%−97%), summary specificity to be 100% (95% CI, 97%−100%) (17 studies, 2257 participants, moderate certainty of evidence).

There was considerable heterogeneity, I² Higgins static of 92% (95% CI, 84–99) which was explored by subgroup analysis and meta-regression. The diagnostic accuracy of ultrasound for malrotation with or without volvulus was better in the groups: inversion of SMA and SMV relationship, use of enteric fluid, and in infants including neonates. We hypothesize that visualizing the SMA and SMV relationship, which is at the level of the pancreas, is more feasible as it is not typically obscured by overlying bowel gas. Additionally, the SMV relationship relative to the SMA-aorta axis may be a better assessment of the SMA/SMV relationship alone, however, published studies did not evaluate this specifically (52). Prior studies have reported mixed sensitivity and specificity for the SMV and SMA relationship. Zerin and DiPetro showed the inversed relationship of the SMV and SMA on US was seen in six out of nine malrotation cases (sensitivity of 67%, small sample size) (35). Taylor evaluated the SMV and SMA relationship on CT in a case control study and demonstrated a sensitivity and specificity of 71% and 79%, respectively for malrotation (36). The whirlpool sign, which is the SMV, and duodenum wrapping around the SMA, occurs below the level of the pancreas and can be obscured by air in the transverse colon. The use of enteric fluid, typically water via a nasogastric tube in six studies, can be used at bedside as an adjunct to help improve the visualization of the duodenum and to increase the accuracy in the diagnosis of midgut malrotation. Proficiency may be improved through continual sonographer and radiologist education, standardization of ultrasound protocols, and structured and clear reporting (53). Other covariates, which may be important include additional signs on US for dilatation of the duodenum, ascites, and dilatation of the SMV and types of ultrasound machines used could not be confirmed due to poor reporting in the included studies.

We found that the risk for bias and applicability concerns were low, in less than 40% and 20% of the total studies, respectively and the GRADE certainty of evidence moderate giving us excellent confidence in our estimation of diagnostic accuracy. The common issues identified included inconsistent use of a reference standard. Many studies (13–16, 18–20, 23, 24, 29, 30) did not report clinical follow-up to ensure that the negative US patients who did not have surgery were truly negative. UGI was often used as the reference standard when surgical confirmation was not reported. This reflects the currently accepted practice of UGI as the accepted standard for diagnosis of malrotation. One study had high risk for applicability concern because they evaluated malrotation in infants referred for pyloric stenosis (13). In addition to the diagnostic accuracy shown in our data, US has the additional advantages of portability of the ultrasound machine to the patient’s bedside, decreases the necessity of oral contrast in a vomiting, sick child, no ionizing radiation, and faster results and decreased cost. Using Time-Driven Activity-Based Costing methodology, Nguyen et al. showed that US was on average 35 minutes faster and cost on average $127 less per patient compared to UGI (31). Based on Bossuyt et al.’s tool for assessing new tests against existing diagnostic pathways (54), US can be strongly considered to replace UGI as the first line test for children suspected of midgut malrotation with or without volvulus.

Clinical Applicability

In a hypothetical cohort of 1000 infants where US is used to diagnose malrotation with or without volvulus, with a prevalence of 17%, applying the summary sensitivity of 0.94 and summary specificity of 1.00 from this review, 10 patients with the disease will be missed by US (Table 3). There will be no false positives or patients wrongly diagnosed with malrotation because of the high specificity of 1.00. Reported false negative rate (FNR) and false positive rate (FPR) ranges for malrotation with or without volvulus by UGI from single published studies are 1.4–8.3% and 1–18%, respectively(55–59). When our meta-analysis of studies reporting UGI as index test was compared with US, there was not a significant difference between the two sensitivities (p=0.533), but US had a significantly higher specificity (p=0.037).

Strengths and Limitations

The strengths of the review include following standard recommended guidelines for performing and reporting diagnostic accuracy studies (PRISMA-DTA and the Cochrane Collaboration)(37, 60). Limitations include substantial heterogeneity among the included studies some of which were explored in subgroup analyses and meta-regression.

CONCLUSION

Moderate certainty evidence suggests that abdominal ultrasound has excellent diagnostic accuracy for midgut malrotation with and without volvulus. Substantial heterogeneity among the included studies and inadequate number of studies for meta-regression analysis of some of the sonographic signs urges for multicenter and prospective data for further evaluation of the specific sonographic techniques and signs.

What is already known on this topic

• Ultrasound is portable, decreases the necessity for oral contrast, absence of ionizing radiation, and faster to perform in children and has distinct advantages over UGI.

• Debate exists within the pediatric radiology and surgery communities as to whether US is a sufficiently sensitive modality to be able to screen at-risk populations for malrotation with or without midgut volvulus.

• The diagnostic accuracy data for US in malrotation has not been systematically reviewed.

What this study adds

• First systematic review of diagnostic accuracy of abdominal ultrasound in the diagnosis of midgut malrotation with or without volvulus in children.

• Summary sensitivities and specificities were 94% and 100% for US detection of malrotation with or without volvulus (17 studies, 2257 participants, moderate certainty evidence).

• Ultrasound has excellent diagnostic accuracy to be used as first line diagnostic test in the diagnosis of malrotation with or without volvulus in children.

Abbreviations

AUROC

area under receiver operating characteristic curve

CI

Confidence intervals

CT

computed tomography

DJJ

Duodenojejunal junction

DTA

Diagnostic test accuracy

GRADE

Grading of Recommendations, Assessment, Development and Evaluations

MR

Magnetic resonance

PRISMA-DTA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Diagnostic Test Accuracy Studies

QUADAS-2

Quality Assessment of Diagnostic Accuracy Studies (version 2)

ROC

Receiver-operating characteristic

SMA

Superior mesenteric artery

SMV

Superior mesenteric vein

US

ultrasound

UGI

Upper gastrointestinal contrast series