Key words
thorax - CT-spiral - pediatrics
Abbreviations
CF:
Cystic Fibrosis
CI:
Confidence Interval
CT:
Computed Tomography
ECG:
Electrocardiography
HSCT:
Hematopoietic Stem Cell Transplantation
MRI:
Magnetic Resonance Imaging
Introduction
In the last decades, computed tomography (CT) has emerged as a basic diagnostic tool
in the evaluation of thoracic disease in many clinical subgroups of children. It plays
a significant role in the characterization of various pulmonary and mediastinal pathologies
and pleural and chest wall anomalies [1]
[2]
[3]. Therefore, there is no doubt about the efficacy of pediatric chest CT [3]
[4]. However, especially after the introduction of modern and fast CT scanners, there
has been an increased number of CT scans in pediatric patients [5].
Although the role of chest CT in childhood has been addressed in diagnostic guidelines
and regulations for different diseases, several studies have suggested overuse of
CT [6]
[7]
[8]. Hence, the efficiency in terms of radiation exposure is questionable. Some studies
concentrated on trends in pediatric emergency departments, including patients with
blunt chest trauma and brain CT [8]
[9]. Various reasons for unnecessary CT scans have been reported with the redundant
repetition of scans or the preference of CT over other imaging modalities with comparable
diagnostic performance (e. g., ultrasound and MRI) [8]
[10]. In solid tumors or in chronic lung disease (e. g., cystic fibrosis), some reports
have recommended omitting standard surveillance by chest CT in asymptomatic patients
or at least reducing routine follow-up imaging in cases in which the clinical impact
is limited [11]
[12]
[13]. Thus, to justify CT, referrers have to provide adequate medical information relevant
to the requested exposure. A detailed patient history allows a deeper understanding
of symptoms, and lab results may, for example, guide towards an infectious disease.
Also, the results of prior diagnostic procedures should be considered. Facing the
wide range of thoracic pathologies in children and the expanding diagnostic capabilities
of chest CT, careful patient preselection seems mandatory to avoid unnecessary examinations
and maintain the adequacy of the technique in the respective clinical scenario [14]. In this context, the interdependence between justifying CT by clinical question
and image interpretation can be understood as a critical factor for the clinical effectiveness
of chest CT.
As an essential factor, numerous studies have been published dealing with technical
aspects, radiation exposure, or specific chest pathologies in children [3]
[15]
[16]. However, no study has analyzed the effectiveness of pediatric chest CT considering
the relevance of findings with respect to the referring question to the best of our
knowledge. To better understand the utility of chest CT in our institution, the purpose
of this study was to estimate the effectiveness and efficiency of chest CT in children
based on the suspected diagnosis in relation to the number of positive, negative,
and inconclusive CT results.
Materials and Methods
This retrospective study was approved by the local ethics committee (project number
586/2019BO2), which waived the requirement for written informed consent. All study
procedures were conducted under the ethical standards as laid down in the 1964 Declaration
of Helsinki and its later amendments or comparable ethical standards.
Patient selection
A retrospective patient cohort was selected from our radiology information system
between January 2007 and December 2014. The inclusion criteria included age under
18 years and at least one CT examination, including the chest. In all cases, the justification
was authorized by a radiologist. The following patient-related data were collected
for later analysis: patient age, sex, use of sedation or anesthesia, information about
the referring department including the subdivision, admission to policlinic or hospital
ward, and outpatient or inpatient care (see also Analysis of CT findings).
CT examination
All CT examinations were performed at a single center with a board-certified division
of pediatric radiology. During the observation time of seven years, six different
CT devices made by the manufacturer Siemens (Siemens Healthineers AG, Forchheim, Germany)
were used. All acquisitions were performed with specific pediatric CT protocols determined
from the division of pediatric radiology. Scan parameters were adapted to patient
body weight. The CT settings followed the current standard operating procedures at
the time of the examination.
Patients younger than six years were examined breathing freely, whereas children older
than six years received a breathing command if they were able to follow the command.
A support cushion was used in newborns and infants to ensure a stable position on
the moving CT table. Board-certified pediatric radiologists carried out diagnostic
reading. For interpretation, further clinical information was accessible for the radiologist
via the in-house database, including lab results and information about the current
treatment.
When intravenous contrast agent was applied, Iomeprol (Cameron 400®) or Iopromid (Ultravist 370®; both Bayer Healthcare Deutschland, Leverkusen, Germany) was used. The injected volume
was adjusted to the patient’s body weight. The injection was performed either by a
dual-head power injection device (Medtronic, Saarbruecken, Germany) or via a manual
injection when the amount of contrast agent was under 20 ml (body weight < 10 kg).
Analysis of CT findings
We analyzed CT reports concerning technical features, including the application of
contrast agents. Clinical information was delivered with the request for an examination
justifying the indication for CT scan. Each request for a CT examination included
a primary question formulating the objective of the scan or the primary suspected
diagnosis based on the symptoms, blood parameters, and preceding diagnostics (patient
history). Clinical subgroups were formed according to the referring question. [Table 1] presents the inclusion criteria and settings for grouping in useful clinical categories.
Table 1
Clinical questions, additional criteria for inclusion, and examples for positive and
negative answers in the clinical subgroups.
|
subgroup
|
number
(n)
|
clinical question
|
additional condition
|
positive answers
(n; %)
|
negative answers
(n; %)
|
inconclusive
|
|
inflammatory process[*]
|
920
|
infection focus, pneumonia?
|
n/a
|
pneumonia
|
581
(63 %)
|
no infection
|
313
(34 %)
|
26
(3 %)
|
|
monitoring/response evaluation
|
268
|
disease activity
|
under therapy
|
new findings, stable disease
|
195
(73 %)
|
no primary finding
|
69
(26 %)
|
4
(1 %)
|
|
pulmonology
|
216
|
pulmonary/tracheobronchial malformation, interstitial lung disease, NEHI, CF?
|
n/a
|
malformation (CPAM, …), lung disease, bronchiectasis, mucus plugging, BPD
|
174
(81 %)
|
no malformation or lung disease
|
38
(17 %)
|
4
(2 %)
|
|
staging (tumor)
|
195
|
metastasis, classification?
|
initial diagnosis, prior to surgery
|
tumor findings, lymphoma, metastasis, classification
|
125
(64 %)
|
no metastasis
|
64
(33 %)
|
6
(3 %)
|
|
cardiovascular imaging
|
182
|
anatomy, vascular malformation, shunt/anastomosis?
|
congenital heart disease
|
visualized anatomy, shunt stenosis
|
149
(82 %)
|
no malformation
|
29
(16 %)
|
4
(2 %)
|
|
trauma management
|
144
|
injury, fractures, pneumothorax, hematoma?
|
recent trauma, trauma center
|
injuries, contusion, hematoma, pneumothorax, fractures
|
81
(56 %)
|
no trauma sequelae
|
61
(42 %)
|
2
(1 %)
|
|
follow-up
|
94
|
postoperative care,
non-oncological follow-up, relapse, treatment effects?
|
after therapy or after primary imaging
|
finding associated with primary diagnosis
|
71
(76 %)
|
no primary finding
|
21
(22 %)
|
2
(2 %)
|
BPD: Bronchopulmonary dysplasia; CPAM: Congenital pulmonary airway malformation; CF:
Cystic fibrosis; HSCT: Hematopoietic stem cell transplantation; NEHI: Neuroendocrine
hyperplasia of infancy.
* including HSCT patients.
The examination reports were stored electronically in the in-house database using
SAP NetWeaver (SAP, Walldorf, Germany). For each examination, we identified whether
the report included a useful result concerning the primary question (conclusive result)
– which may also contain the exclusion of a suspected pathology – or if no statement
was possible regarding the question. The latter means that neither a confirmation
nor an exclusion of the suspected pathology was possible (inconclusive result). In
the case of an inconclusive result, the reason for the limitation was taken from the
report. Subsequently, we figured out if the report provided a positive answer to the
primary clinical question – defined as a confirmation of the suspected diagnosis –
or a negative answer – defined as a rejection of the suspected diagnosis. [Table 1] presents examples for positive and negative answers for each subgroup. As target
values, a proportion of more than 50 % positive results for each subgroup and an overall
proportion of inconclusive results of fewer than 5 % were estimated.
Finally, we recorded whether additional findings were detected. We classified additional
findings as clinically important (major) or as a minor additional finding without
clinical relevance. Examples of clinically significant findings are pneumonia, previously
unknown tumors and pulmonary nodules, catheter-associated complications, like thrombosis
or an abscess, and other findings which required further workup. Other additional
findings were estimated as not relevant for the further clinical course of the patient,
for example, preexisting rib fractures and other known osseous pathologies (e. g.,
spondylolysis, synostosis), or small pleural effusions and soft tissue emphysema after
an intervention.
Statistical Analysis
Data were entered into Excel 2010 for Windows (Version 14.0; Microsoft Corporation,
Redmond, WA). For statistical analysis, SAS jmp (version 11.1.1 for Windows, SAS Institute
Inc., Cary, NC, USA) was used. The analysis involved descriptive statistics. Continuous
variable data were presented as means ± standard deviations. Data that did not follow
a normal distribution were presented as median with a 95 % confidence interval (CI).
The normality Shapiro-Wilk test was applied to study the distribution of patient data.
The Mann-Whitney U test was used to analyze the difference in age between male and
female patients. The Kruskal-Wallis H test was applied to study the correlation between
inconclusive CT results and scanner types and between inconclusive results and the
use of sedation. A p-value less than 0.05 was considered to indicate statistical significance.
Results
Study population
The final patient cohort consisted of 973 patients (mean age ± standard deviation,
10 ± 5.6 years; median and 95 % confidence level, 10.5 years [9.7; 10.2]) including
581 male patients (10.2 years ± 5.6; 10.7 years, [9.9; 10.6]) and 392 female patients
(9.5 years ± 5.7; 10.4 years, [9.1; 9.9]; difference between male and female, p = 0.003).
Finally, 2019 chest CT scans were included.
Technical results
62 % (1251/2019) of scans were non-enhanced. Intravenous contrast agents were used
in 768 CT examinations. The most prominent clinical subgroups of contrast-enhanced
scans were cardiovascular (n = 140) and trauma patients (n = 136). 65/2019 CT scans
(3.2 %) were performed with sedation or anesthesia, most of them in the age group
under one year (n = 37).
Clinical information
1457/2019 exams (72 %) were referred from in-house, which means that they were also
accommodated for inpatient treatment. In 562 cases, a polyclinic requested the chest
CT examination, which means that the scan was performed during an outpatient visit.
Most examinations (n = 987) were requested from the Department of Hematooncology,
including the polyclinic for hematopoietic stem cell transplantation (HSCT) patients.
[Fig. 1] illustrates the distribution of the subdivisions requesting the CT examinations.
The subgroup of 267 non-oncological and non-surgical cases included pediatric pulmonology
patients and other pathologies like pulmonary embolism and pneumothorax. There were
only 36 cases with CF in this subgroup.
Fig. 1 Distribution of the clinical partners (pediatric subdivisions) requesting CT examinations
(n). 1 HSCT: Hematopoietic stem cell transplantation. 90 examinations were excluded
from the figure for better visualization. These included non-pediatric subdivisions,
like orthopedics, trauma surgery, sports medicine, urology, gynecology and external
patients (n = 8).
Abb. 1 Verteilung der klinischen Zuweiser nach Anzahl der angeforderten CT-Untersuchungen
(n). 90 Untersuchungen wurden in der Abbildung für eine bessere Übersicht ausgeschlossen.
Diese beinhalten nicht Pädiatrie-spezifische Abteilungen wie Orthopädie, Unfallchirurgie,
Sportmedizin, Urologie, Gynäkologie und externe Zuweiser (n = 8). HSCT = Stammzelltransplantation.
[Table 1] summarizes the clinical questions for the CT scans and the number of patients in
each subgroup. In the subgroup of a suspected infection process (n = 920), 361 of
the cases (18 % of all scans) were performed to diagnose pneumonia after HSCT. Furthermore,
179 examinations (9 %) were carried out to exclude a pulmonary infection before HSCT.
Conclusive/inconclusive results
The proration of conclusive vs. inconclusive interpretation is listed in [Table 1]. In 97.6 % (1971/2019) of the CT scans, the primary question was answered sufficiently,
or the suspected diagnosis was ruled out. Only in 48 examinations (2.4 %), the interpretation
was inconclusive. Typical examples of inconclusive results were impaired vascular
contrast or breathing artifacts so that the area of interest (a pulmonary nodule or
cardiac anatomy) was not assessable. In one case, CT pointed towards fungal pneumonia,
while bronchoalveolar lavage remained negative. There was no correlation between the
use of different scanner types and the number of inconclusive results (p = 0.20).
One inconclusive CT in the subgroup of patients with sedation (1/66; 1.5 %; p = 0.65)
resulted from an unclear cystoid mass next to the esophagus without a visible connection
to the esophageal lumen proving to be an esophageal duplication cyst in endoscopy
(18-month-old girl).
CT findings in the subgroups
The reports in 595 examinations (30 %) rejected the suspected diagnosis. [Table 1] shows the number of positive and negative answers with clinical examples for each
subgroup. In 8 of 9 categories the proportion of positive interpretations was larger
than 50 % ([Fig. 2]). Three categories containing infection and pneumonia, lung imaging after HSCT,
and response assessment showed a proportion of positive results of 68 % to 76 %. Only
in the category “prior to HSCT”, 61 % of the examinations in this group had a negative
result and confirmed the suspected diagnosis “normal lung.”
Fig. 2 Proportion of positive and negative results of the CT examinations in the different
clinical subgroups. The width of the boxes corresponds to the number of patients in
the different clinical subgroups. The 50 % line is highlighted to emphasize differences
between the categories. Inconclusive results were excluded from this figure. 1 HSCT:
Hematopoietic stem cell transplantation.
Abb. 2 Anteil positiver und negativer Befunde der CT-Untersuchungen in den verschiedenen
klinischen Untergruppen. Die Breite der Kästen richtet sich nach der Anzahl der Patienten
in der jeweiligen Untergruppe. Die 50 %-Linie ist dargestellt, um die Unterschiede
zwischen den Untergruppen zu verdeutlichen. Inkonklusive Ergebnisse wurden für diese
Darstellung ausgeschlossen. HSCT = Stammzelltransplantation.
In the category “trauma management”, most of the thoracic findings were combined thoracic
injuries with more than one thoracic pathology related to chest trauma (n = 23). The
most frequent combination was lung contusion with pneumothorax and rib fractures (n = 12).
Other complex injuries included mediastinal hematoma (n = 4) and diaphragmatic rupture
(n = 2). Single findings referred to spinal fractures (n = 13), mediastinal hematoma
(n = 3), hemothorax (n = 2), and pulmonary laceration (n = 2). Further findings included
single lung contusions (n = 12), rib fractures (n = 7), and clavicle fractures (n = 6)
without other pathologies.
Additional findings
In 396 scans (20 % of all scans), an additional finding was detected. 222 of the additional
findings (56 %) were classified as clinically important. For example, non-enhanced
CT with the question for pulmonary metastasis in an immunocompromised patient showed
consolidations with peripheral ground-glass attenuation and additional centrilobular
nodules indicating fungal infection such as aspergillosis. Afterwards, oral antimycotic
therapy was started.
The remaining 174 findings (44 %) were classified as minor additional findings. As
an example, a small pleural fluid collection was detected after thoracoscopic wedge
resection of a pulmonary metastasis as a postoperative finding.
[Fig. 3] presents a correlation of the additional findings to the CT results regarding the
primary question. There were 79 examinations with a negative result concerning the
primary referring reason, but with a relevant additional finding. 24/79 in this subgroup
were examined for staging or tumor response assessment. In 406/2019 (20 %) examinations,
neither a primary nor an additional finding was reported, which means that thoracic
morphology was normal. In the subgroup of scans with a negative result and a minor
or no additional finding (n = 512), 204 examinations were performed before or after
an HSCT. 60 scans in this subgroup were carried out during trauma management.
Fig. 3 Correlation of the additional findings to the CT results regarding the primary question
(positive or negative).
Abb. 3 Korrelation zwischen den Nebenbefunden und den CT-Befunden mit Bezug zur primären
klinischen Fragestellung (positiv oder negativ).
Discussion
This study provides the most extensive analysis of chest CT reports in children concerning
clinical data and CT findings. In general, large-scale prospective studies to identify
influences of imaging on treatment regimens are challenging to establish in a pediatric
patient cohort due to ethical reasons (individual benefit vs. risks of radiation exposure).
The cohort’s heterogeneity was accepted to illuminate the full spectrum of chest CT
findings in a wide range of patients.
As expected, our analysis yielded useful results proving that chest CT in children
is a powerful imaging tool when considering the clinical situation and possible relevance
for therapy. Overall, 98 % of all chest CT examinations in our study provided conclusive
interpretation by either confirming or excluding the suspected pathology. Thus, our
presumed target value of < 5 % inclusive reports was achieved, supporting the effectiveness
of the method in our daily routine. As the pre-CT suspected diagnosis was predominantly
(> 50 %) confirmed in nearly all of the clinical categories, we assumed that the primary
indications for CT were adequately chosen ([Fig. 2]).
Most scans in our study were performed without a contrast agent (62 %). In line with
this, the most frequent indication for chest CT was the imaging of suspected pulmonary
inflammation (e. g., pneumonia; 46 %). In these subgroups, including examinations
in the context of HSCT, 63 % of the scans proved positive for an inflammatory process
([Table 1]). Since chest CT may also limit the differential diagnosis or suggest the underlying
pathology, e. g., a pattern suggesting viral pneumonia or identifying specific morphological
features of tuberculosis, it can be assumed that the positive results may encourage
further examinations or bronchoalveolar lavage or even enable optimization of an anti-infective
therapy [17]
[18]. Regarding the high proportion of this category in our cohort, it should be kept
in mind that CT should be used to distinguish between pneumonia and other pathologies,
like pulmonary hemorrhage or congenital anomalies. Besides, as immunocompromised children
suffer from pneumonia more frequently and the rate of severe complications is high,
the threshold for performing chest CT should be lower compared to immunocompetent
children [19]. In this context, our results point towards an interesting subgroup of 179 CT examinations
(9 % of all scans) before an HSCT to exclude a pulmonary infection. Children undergoing
HSCT are at increased risk of opportunistic infections, e. g., invasive fungal infections,
due to prolonged immunosuppression [20]. Nevertheless, the practice of chest CT screening to reduce the risk of post-HSCT
infections is not universal. Only one study from the St. Jude Children’s Research
Hospital evaluated the clinical utility of pre-HSCT chest CT screening [21]. Authors stated that the indication of pre-transplantation evaluation by chest CT
should consider the individual patient’s history, previous treatments, and the present
clinical risk for a pulmonary infection to avoid unnecessary medical evaluations or
a delay of transplant [21]. As our Department for Oncology follows the St. Jude Children’s Research Hospital’s
experiences for referring to CT before HSCT, we had a relatively large group of patients
in this investigation. Predominantly (61 %) in this subgroup, CT ruled out any evidence
of a silent inflammatory disease. This is in line with the results of the study of
Kasow et al. with normal chest CT in 56 % of the allo-recipients (n = 62). However,
the proportion of positive scans in our study (39 %) is rather high since unrecognized
infection may be fatal for these patients. Therefore, in this subgroup, CT may be
justified even more than in a non-HSCT setting. Although not surprisingly, more positive
results occurred after HSCT (68 % positive findings). In summary, more than 50 % of
the scans were found without relevant findings in the context of HSCT. Therefore,
to increase the efficiency of CT in this group, we support the recommendation of Kasow
et al. to carefully evaluate the indication for CT if the patient is asymptomatic
and to consider co-factors like medication and the present risk for a pulmonary infection
[21].
Apart from the established application such as in congenital anomalies, malignant
diseases, or interstitial lung disorders, chest CT is the imaging reference standard
also in patients with cystic fibrosis [22]
[23]. However, in this patient group, pulmonary MRI has now been accepted as a radiation-free
modality offering anatomical and functional details with impacts on therapy management
[24]
[25]
[26]
[27]. This may also explain the relatively low number of CF cases in our study (n = 36).
Rising opportunities in chest imaging require us, even more than before, to choose
the suitable imaging modality according to the clinical context and to consider each
patient’s radiation history when indicating repeated chest imaging [28]. In the context of improved outcomes and increasing life expectancy – not only concerning
CF, but also in pediatric hematooncology (e. g., lymphoma) – imaging surveillance
carried out for a longer period over a lifetime should be assessed critically [12]
[13]. The use of CT in an inappropriate clinical situation may result in false-positive
results (pseudo-disease) or incidentalomas that are not relevant for treatment or
may even hamper an immediate target-oriented therapy [8]. Risk-adapted strategies for imaging surveillance have been proposed for survivors
of childhood lymphoma [11].
Concerning the prevention of unnecessary CT in children, former reports have dealt
with the clinical benefit of chest CT in children with blunt trauma [29]. A chest radiograph may not be the superior imaging modality but may be sufficient
to answer the clinical question and exclude relevant pathologies in a trauma setting.
Therefore, CT in pediatric chest trauma patients is frequently used as a confirmatory
diagnostic tool [9]. This applies to our study only to a limited extent. Positive results were found
in 57 % of the subgroup of trauma patients ([Fig. 2]). As most of the findings described combined thoracic pathologies or other complex
injuries, like spinal fractures or diaphragmatic rupture, it can be estimated that
chest CT added clinical information in the trauma setting in most cases. The decision
to perform chest CT should be based on interdisciplinary clinical evaluation including
close communication between the radiologist and the clinical care team [30].
48 scans in our study delivered inconclusive results. As technical issues like reduced
vascular contrast or breathing artifacts were frequent reasons for impaired assessment,
these cases draw attention towards the necessity of individually adjusting scan protocols
and optimizing examination conditions in a pediatric setting. These aspects seem especially
important for the subgroup of cardiovascular patients. Concerning cardiovascular pathologies
in our study, indications were often indisputable and clinical usefulness was evident
in children with congenital heart disease for the visualization of anatomy (82 %)
([Table 1]). However, in this patient group, the scan protocols and technical details, like
the use of ECG or high-pitch scanning, are of particular concern to avoid a reduction
of image quality [31].
Our study demonstrated a small number of CT examinations (n = 79) which rejected the
primary suspected diagnosis, but an additional finding with clinical significance
was detected ([Fig. 3]). This means that chest CT may also identify relevant changes besides the reason
for the referral. Also, in cases with a low pretest probability for disorders, the
findings could substantially impact clinical decisions. However, this does not justify
random use of CT but can be helpful in complex situations. 30 % of the patients in
this subgroup (n = 24) were examined for staging or tumor response assessment in malignant
disorders in which complications (tumor-associated or therapy-related) are more likely
to occur.
Our study has several limitations that require discussion. The most critical limitation
is that our study was set up at a single center. Consequently, it depends on the local
standard of care, limiting the generalization of the study findings. Undoubtedly,
the classification of examinations into different clinical categories somehow simplifies
many real-life situations in clinically complex cases with more than one question
to be answered by the CT examination or different suspected diagnoses. Nevertheless,
the statistical relevance of simplification may be expected to be low in a study population
of more than 2000 patients. The CT findings were not compared to other imaging modalities.
In summary, our study confirms that pediatric chest CT in specialized centers has
a high effectiveness when the application is based on sufficient clinical data and
well-known patient history. The pre-CT evaluation appears crucial for preventing unnecessary
CT examinations, especially in the context of suspected pneumonia or trauma management.
Thus, the risk/benefit ratio (efficacy) should be looked at even more closely. Furthermore,
other imaging modalities, mainly MRI, should be considered if the individual clinical
situation is suitable.
-
Chest CT in children has high clinical value, also in clinically complex situations.
-
Knowledge of clinical parameters and patient history is essential for the indication
and evaluation of pediatric chest CT.
-
In an HSCT setting, CT indication and interpretation should consider the exact transplantation
phase.
