Abdominal Radiology

https://doi.org/10.1007/s00261-018-1858-8 (0123456789().,-volV)(0123456789().,-volV)

PELVIS

Structured versus narrative reporting of pelvic MRI in perianal

fistulizing disease: impact on clarity, completeness, and surgical
planning
Ozum Tuncyurek1 • Alejandro Garces-Descovich1 • Adrian Jaramillo-Cardoso1 • Elena Esteban Durán1 •

Thomas E. Cataldo2 • Vitaliy Y. Poylin2 • Said Fettane Gómez1 • Atenea Morcillo Cabrera1 •
Tarek Hegazi1 • Kevin Beker1 • Koenraad J. Mortele1

Ó Springer Science+Business Media, LLC, part of Springer Nature 2018

Abstract
Objective To evaluate clarity, completeness, and impact on surgical planning of MRI reporting of perianal fistulizing
disease using a structured disease-specific template versus narrative reporting for planning of disease treatment by col-
orectal surgeons.
Materials and methods In this HIPAA-compliant, IRB-approved study with waiver of informed consent, a structured
reporting template for perianal fistulizing disease MRIs was developed based on collaboration between colorectal surgeons
and abdominal radiologists. The study population included 45 consecutive patients who underwent pelvic MRI for perianal
fistulizing disease prior to implementation of structured reporting, and 60 consecutive patients who underwent pelvic MRI
for perianal fistulizing disease after implementation of structured reporting. Objective evaluation of the reports for the
presence of 12 key features was performed, as also subjective evaluation regarding the clarity and completeness of reports,
and impact on surgical planning.
Results Significantly more key features were absent in narrative reports [mean: 6.3 ± 1.8 (range 3–11)] than in structured
reports [mean: 0.3 ± 0.9 (range 1–5)] (p B 0.001). The use of structured reporting also increased the percentage of
completeness (72.5–88.3% for surgeon 1, and 61.2–81.3% for surgeon 2; p = 0.05 and 0.03, respectively), helpfulness in
surgical planning (7.1 ± 1.5–7.6 ± 1.5 for surgeon 1, and 5.8 ± 1.4–7.1 ± 1.1 for surgeon 2; p = 0.05 and p \ 0.001,
respectively), and clarity (7.6 ± 1.3–8.3 ± 1.1 for surgeon 1, and 5.2 ± 1.4–7.1 ± 1.3 for surgeon 2; p = 0.006 and
p \ 0.001, respectively) of the reports.
Conclusion Structured MRI reports in patients with perianal fistulizing disease miss fewer key features than narrative
reports. Moreover, structured reports were described as more complete and clear, and more helpful for treatment planning.

Keywords Structured reporting
Conventional reporting
Perianal fistulizing disease
Magnetic resonance imaging

Fistula treatment

2
Division of Colorectal Surgery/Department of Surgery, Beth
& Alejandro Garces-Descovich Israel Deaconess Medical Center, Harvard Medical School,
[email protected] 330 Brookline Avenue, Boston, MA 02115, USA
Ozum Tuncyurek
[email protected]
1
Division of Abdominal Imaging/Department of Radiology,
Beth Israel Deaconess Medical Center, Harvard Medical
School, 330 Brookline Avenue, Boston, MA 02115, USA

123
 Abdominal Radiology

Introduction Materials and methods

Radiology reports are the key source of information in Institutional review

medical imaging and are the main mode of contact between
radiologists and referring physicians [1]. Adequate report- This Health Insurance Portability and Accountability Act
ing delivers one of the cornerstones of integrative and (HIPAA)-compliant retrospective study was conducted
personalized medicine. Narrative reporting (NR) com- with the approval of our institutional review board (IRB).
monly lacks a proper structure, clarity, and universally The requirement for informed consent was waived due to
established lexicon; therefore, the quality and completeness the retrospective nature of this study.
is heavily dependent on the radiologists training and
experience [1, 2]. Recently, the use of structured reporting Subjects
(SR) has been shown to improve diagnostic precision,
reduce errors, and allow for a consistent use of terminology We conducted a search of our picturing archiving com-
[2]. To the best of our knowledge, however, no prior munication system (PACS) from December 2016 to
studies have assessed the value of disease-specific SR in September 2017 and identified 597 consecutive contrast-
the evaluation of patients with perianal fistulizing disease enhanced pelvic MRI studies. All reports were reviewed by
with MRI. one observer for the word ‘‘perianal fistula’’ in the indi-
Fistulas are defined as abnormal connections between cation for imaging tab. A total of 122 reports were iden-
two epithelial-lined surfaces. Perianal fistulas are typically tified; of these, 17 reports were excluded because they
identified between the anal canal and the perineal cuta- belonged to same patients at different time points of
neous tissue. They occur predominately in the male pop- evaluation. The final sample population was 105 reports in
ulation (2:1 ratio) and their estimated annual incidence is 105 individual patients. Implementation of a structured
one in 68.000–96.000 people in the United States [3, 4]. report (SR) took place in our institution on April 9, 2017 to
Perianal fistulas arise predominantly (90%) from impaired be used in patients undergoing MRI for evaluation of
drainage of anal crypts, and are possibly related to perianal fistulizing disease. Due to personal preference of
increased sphincter tone [5]. Infection of anal glands may some radiologists, some of the reports after implementation
result in abscess formation in the intersphincteric space and (AI) of the SR were narrative reports (NR). Subsequent
then possibly spreads along adjacent spaces. Other less classification was performed by type of report in 3 groups,
common causes (10%) of perianal fistulas include Crohn (1) narrative report before implementation (NRBI), (2)
disease, tuberculosis, sexually transmitted disease, and narrative report after implementation (NRAI), and (3)
trauma [6–9]. structured report after implementation (SRAI). The first
Historically, surgical or imaging-based classifications of group [NRBI] yielded 45 reports, the second group [NRAI]
perianal fistulas are proven helpful in planning for surgical yielded 17, and the third group [SRAI] yielded 43 reports
intervention. Because of its superior soft tissue resolution, (Fig. 1). For each of the 105 patients, we recorded their
the use of MR imaging is widely accepted as the imaging age, gender, and prior history of inflammatory bowel
modality of choice for diagnosis and classification disease.
[4, 10–14]. Moreover, certain key MRI features are
important when planning for surgical management; these Pelvic MRI protocol
include: the number and location of the fistulas, the loca-
tion of the internal anal opening, the fistula type, the Multiplanar, multisequence MRI data were acquired
presence of abscess, and the presence and location of through the pelvis using 1.5 and 3.0 Tesla scanners (Signa
secondary branches, etc. LX Excite and 750 W, GE Healthcare, Milwaukee, WI,
For this study, in collaboration with our institutional USA; Aera and Espree, Siemens Healthcare GmbH,
colorectal surgery group, we therefore developed a disease- Erlangen, Germany). The MRI protocol included, after
specific structured reporting template to be used in patients placement of gauze within the anal cleft, single shot fast
who underwent MRI for the assessment of their perianal spin echo (SSFSE) T2-weighted sagittal imaging (TE =
fistulas. The primary endpoint of this study was to evaluate 100 ms, TR = 4000 ms, slice thickness = 4.5 mm, pixel
objectively the clarity and completeness of NR versus SR spacing = 0.875/0.875 mm), single shot fast spin echo
of pelvic MRI in the assessment of perianal fistulas. (SSFSE) T2-weighted fat suppressed coronal oblique and
Secondarily, we also sought to evaluate the impact of SR axial oblique imaging (TE = 59 ms, TR = 4840 ms, slice
on the clarity, completeness, and impact on surgical thickness = 4.5 mm, pixel spacing = 0.875/0.875 mm),
planning. and axial two dimensional in phase and out of phase T1-

123
Abdominal Radiology

MRI structured reporting template

(N=597) On April 9th 2017, an MRI structured reporting template

12/2016 – 09/2017 for perianal fistula evaluation was implemented at our
Contrast enhanced pelvic institution. This template was designed using the existing
MRI reports literature [15–17] by an abdominal radiologist (KJM, with
19 years of post-fellowship experience in reading pelvic
MRI) and two colorectal surgeons (VYP and TEC, with 8
and 22 years of post-fellowship experience in treating
patients with perianal fistulas, respectively). The ultimate
Key word: goal was to increase report simplicity and usability as a
“Perianal fistula” surgical planning decision tool for perianal fistulas. The
template, shown in Fig. 2, was reviewed and approved by
Excluded: all other radiologists of the body MRI section. Twelve
475 reports features were deemed ‘‘key’’ by colorectal surgeons in the
planning of management of these patients (Fig. 3).
122 reports Although the use of the template was not mandated, it was
recommended for routine use during clinical readout.

Reports from the Evaluation of structured reports [SR]

same patients at versus narrative reports [NR]
different times
Objective evaluation for the existence of all 12 key features
Excluded: in the NRBI group (n = 45), NRAI group (n = 17) SRAI
17 reports group (n = 43) was performed by two radiologists (OT and
EED, with 17 and 14 years of abdominal radiology expe-
105 reports rience, respectively). In addition, the reports were assessed
for the number of fistulas present in each patient by the
same 2 radiologists.
BI AI
Subjective assessments of both the SR and the NR for
our entire sample (n = 105) were performed blinded and
independently by two colorectal surgeons (VYP and TEC
NR: 45 NR: 17 SR: 43 with 8 and 22 years of colorectal surgery experience,
reports reports reports respectively) for clarity, completeness, and clinical use-
fulness for surgical planning (Fig. 4). All MRI reports were
Fig. 1 Flow chart of inclusion criteria for study. SR structured report,
assigned in an anonymous random order. Randomization
NR narrative report, BI before implementation, AI after
implementation was performed sequentially with the aid of random number
generator provided by random.org. A list of four specific
weighted spoiled gradient echo (TE = 2.38 ms, TR = 180 questions that were to be completed by both colorectal
ms, slice thickness = 5 mm, pixel spacing = 0.546875/ surgeons was attached to each report. The first question
0.546875 mm) sequences. Gradient-echo, dynamic, high- assessed the clarity of the report using a Likert scale
resolution 3D T1-weighted fat suppressed axial oblique ranging from 1 (not at all clear) to 10 (completely clear).
VIBE/SPGR sequence (TE = 2.19 ms, TR = 4.7 ms, slice The second question (dichotomous) asked if, following
thickness = 3 mm, pixel spacing = 1.25/1.25 mm) were reading the report, the surgeon had enough information to
acquired before and after the administration of Gadobutrol then decide on the management approach (e.g., medical
contrast medium (0.1 mL/kg body weight at a 2 ml/s rate) treatment, fistulotomy, seton placement, abscess drainage,
followed by a saline flush. Postcontrast imaging (2 phases) etc.). The third question asked, using a Likert scale ranging
was obtained 45 s and 150 s following the contrast from 1 (not at all helpful) to 10 (completely helpful),
injection. whether the report was helpful for surgical planning. The
final question (dichotomous) asked if, following reading
the report, the surgeon still wanted to review the images.

123
 Abdominal Radiology

CHECK INTERNAL ANAL OPENING

Number of fistula openings
If mulple fistulas are found, please describe each one individually following No: ___
the same template
Fistula internal anal opening locaon
Please note posion accordingly to quadrant and clock face locaon (supine)
□ Anterior (midline) □ Posterior (midline)
□ Le anterolateral □ Right anterolateral
□ Le posterolateral □ Right posterolateral (1-12 o’clock)
Distance between internal anal opening and the anal verge
______ mm
Please note in millimeters

CHECK FISTULA TRACT

Maximum tract diameter
______ mm
Please note in millimeters
Fistula type
Please note according to Park’s classificaon for perianal fistulas
□Intersphincteric □Transphincteric □Extrasphincteric □Suprasphincteric □Superficial
Presence of secondary branches
Please note accordingly, If present, please describe where branch extends
□ None □ One □ Mulple □ Extension:______________________
Exit site locaon
Please note accordingly. Please circle the laterality as: le (L) or right (R).
□Gluteal □Scrotal □Vaginal □Labial □Urethral □Blind ending (sinus tract)
L-R L-R L-R L-R L-R L-R
Hyperintensity of the tract on T2 weighted sequence (Van Assche Classificaon)
Please note accordingly
□ Absent □ Mild □ Marked
Enhancement of the fistula tract
Please note accordingly
□ No central enhancement □ Central enhancement with □ Minimal progressive
(tract is fluid-filled) high signal intensity on T2 enhancement with low signal
(granulaon ssue) intensity on T2 (fibrosis)

CHECK OTHER FINDINGS

Presence of abscess □YES □NO
If “YES”, please describe if the abscess is separated from the fistula tract, locaon and size
□Separated □ Connected Locaon: ___________________ Size: ________ mm
Please note if there is any of these specific addional findings:
□ Anovaginal Fistula □ Rectal or sigmoid wall inflammaon □ Seton/drains/prior surgeries
Fig. 2 Structured reporting template for pelvic MRI to assess perianal fistulas

Statistical analysis and Dunn’s post hoc test were used to compare the three
reporting groups (NRBI, NRAI, and SRAI).
Student t test was used to compare the mean age and the Differences in responses of colorectal surgeons for NR
mean presence of key features in both cohorts (BI and AI). and SR groups were tested using Chi square for dichoto-
Analysis of presence of key features among the groups was mous answers and t test for Likert scale answers. Interob-
performed with Wilcoxon signed-rank test. Kruskal–Wallis server agreement was assessed using McNemar’s test for

123
Abdominal Radiology

Fig. 3 Key features of MRI MRI structured report key features and the rationale of including them in
structured report and the Figure 3.
perianal fistula evaluation template
rationale of including them in
perianal fistula evaluation Key feature Justification

An increased number of fistulas relate to a higher rate

Number of fistulas
of operative and postoperative complications
Location (quadrant or clock face) describes the point of
Internal anal opening
origin and direction of the fistula
Distance between internal anal opening As it facilitates surgical planning
and the anal verge
Relation between internal anal opening Guides surgical planning
and internal anal sphincter
Helps with surgical planning to minimize recurrence,
Fistula type based on pathogenesis and
postoperative incontinence and identify clinically
anatomy from the Park’s classification
unapparent disease.
Secondary tracks may be complicated by abscess
Presence of secondary branches formation and require specific surgery for adequate
drainage
Presence of abscess Same as above

Exit site location Same as above

According to the Van Assche classification, determines

T2 hyperintensity
activity

Contrast enhancement of the tract Crucial for assessment of fistula activity

Associated rectal and sigmoid colon To assess unresponsiveness to medical treatment for
inflammation inflammatory bowel disease (IBD)
Presence of setons, drainage catheters To assess rate of success of prior management
and features of prior surgeries

yes/no answers and paired samples t test for Likert scale Evaluation of structured reports [SR]
answers, and kappa values were calculated. A p value of versus narrative reports [NR]
less than 0.05 was considered statistically significant. All
statistical analyses were performed using SPSS version 25 Assessment of the number of fistulas/patient revealed that a
(IBM Corporation, US). single fistula was identified in 80% (n = 36) of the BI
group patients, and 75% (n = 45) in the AI group (p = 0.6).
In the BI group patients, multiple fistulas (2, 3 ,and 4) were
Results identified in 5 (11.1%), 3 (6.6%), and 1 (2.2%) of patients,
respectively. In the AI group, 2, 3, 4, and 5 fistulas were
Subjects noted in 5 (8.3%), 5 (8.3%), 4 (6.6%), and 1 (1.6%) of
cases, respectively (p [ 0.05) (Table 1).
The final study population consisted of 105 patients Regarding the presence or absence of the 12 key features
(Table 1). The BI group consisted of 45 patients: 31 males in both BI and AI groups, the following findings were
(68.8%) and 14 females (31.1%) with a median age of encountered (Table 2): (1) mentioning of the number of
43 years (IQ range 36–58.5 years). The AI group consisted fistulas was only present in 13 (28.8%) of the NRBI
of 60 patients: 29 males (48.3%) and 31 females (51.6%) reports, 0 (0%) of the NRAI reports and, 42 (98%) of SRAI
with a median age of 42.5 years (IQ range reports; (2) the internal opening location (quadrant and
30.2–54.8 years) (p value [ 0.05). A prior history of clock face) was present in 41 (91.1%) of the NRBI reports,
inflammatory bowel disease was found in 42.2% (n = 19) in 17 (100%) of the NRAI reports, and 42 (98%) of the
for the BI group, and 46.6% (n = 28) in the AI group SRAI reports; (3) reporting of the distance from the anal
(p value = 0.7). verge to the fistula opening was present in 25 (55.5%) of
the NRBI reports, in 9 (53%) of the NRAI reports and, 42
(98%) of the SRAI reports; (4) the relation of the fistula to
the internal sphincter was not mentioned in the NRBI

123
 Abdominal Radiology

Fig. 4 Evaluation chart of Randomized assigned #: ______

radiology reports by colorectal
surgeons Evaluaon of MRI Reports by Colorectal Surgeons

Please mark the correct numeric or Yes/No response to each queson.

1. How easy to understand is this MRI report?

Please mark one
Not at all Completely
clear clear
1 2 3 4 5 6 7 8 9 10

2. Do you have enough informaon to decide on the management approach?

(Seton placement, abscess drainage, fistulotomy, conservave approach).
Please mark one.

YES NO

3. How helpful is this MRI report for proper surgical planning?

Please mark one
Not at all Completely
helpful helpful
1 2 3 4 5 6 7 8 9 10

4. Based on this report do you sll want to review the images?

Please mark one

YES NO

reports while in 1 (6%) NRAI report and 42 (98%) of the presence/absence of an abscess was mentioned in 35
SRAI reports, the opening was addressed; (5) the type of (77.7%) of the NRBI reports versus 16 (94%) of NRAI and
the fistula (intersphincteric, transphincteric, extrasphinc- 41 (95%) of SRAI reports; (11) absent/associated rectal or
teric, suprasphincteric, and superficial) was specified in 28 sigmoid wall inflammation was commented upon in 15
(62.2%) of the NRBI reports, while 13 (76.4%) of NRAI (33.3%) of NRBI reports versus 3 (18%) of NRAI reports
reports and 43 (100%) of the SRAI reports was specified; and, 37 (86%) of the SRAI; and finally, (12) absence/
(6) presence of secondary branches were mentioned in 16 presence of a seton, stitch, drain or sequela from prior
(35.5%) of the NRBI reports versus 3 (17.6%) of the NRAI surgery was mentioned in 13 (28.8%) of NRBI reports, 6
reports and, 43 (100%) of the SRAI reports; (7) the fistula (35.2%) of NRAI and, 42 (98%) of SRAI reports.
exit site (gluteal, scrotal, vaginal, labial, urethral or blind An average of 6.3 ± 1.8 (range 3–11) key features were
ending) was reported in 36 (80%) of the NRBI reports, in 6 absent in NRBI reports (n = 45) and only 0.3 ± 0.9 (range
(35.2%) of the NRAI reports and, in 43 (100%) of the 1–5) in the SRAI group reports (n = 43) (p B 0.001). For
SRAI; (8) hyperintensity of the tract on T2-weighted completeness, an average of 7.1 ± 1.0 (range 5–9) key
images was reported in 9 (20%) of the NRBI reports while features were absent in the NRAI group reports (n = 17)
in 2 (12%) of the NRAI reports and, 43 (100) of the SRAI (p B 0.001). Statistical comparison analysis of the three
was mentioned; (9) assessment of contrast enhancement of groups by the MRI template as well as the subjective
the tract was done in 19 (42.2%) of NRBI reports, 6 assessment (colorectal surgeons) is depicted in Table 3.
(35.2%), and, in 43 (100%) of the SRAI reports; (10) the

123
Abdominal Radiology

Table 1 Demographics, clinical presentation, imaging features and fistula number distribution among patients with perianal fistula
Variable All patients (N = 105) Before implementation (n = 45) After implementation (n = 60) p value

Sex, no. (%)

Age, median (IQR), year 43 (32–56) 43 (36–58.5) 42.5 (30.2–54.7) 0.10
Female 45 (43) 14 (31.1) 31 (51.6) 0.01
Male 60 (57) 31 (68.8) 29 (48.3)
Clinical presentation, no. (%)
IBD history 47 (44.7) 19 (42.2) 28 (46.6) 0.69
Fistula number distribution, no. (%)
Total number of fistulas
1 81 (77.1) 36 (80) 45 (75) 0.641
2 10 (9.5) 5 (11.1) 5 (8.3) 0.741
3 8 (7.6) 3 (6.6) 5 (8.3) 0.99
4 5 (4.7) 1 (2.2) 4 (6.6) 0.389
5 1 (0.9) 0 (0) 1 (1.6) 0.99
IQR interquartile range, IBD inflammatory bowel disease

Table 2 Number of key features

Key features no. (%) NRBI (n = 45) NRAI (n = 17) SRAI (n = 43)
presence in all reports
Number 13 (28.8) 0 (0) 42 (98)
Internal opening 41 (91.1) 17 (100) 42 (98)
Distance from verge 25 (55.5) 9 (53) 42 (98)
Opening relation 0 (0) 1 (5.9) 42 (98)
Type 28 (62.2) 13 (76.4) 43 (100)
Branches 16 (35.5) 3 (17.6) 43 (100)
Exit site 36 (80) 6 (35.2) 43 (100)
T2 signal 9 (20) 2 (11.8) 43 (100)
Enhancement 19 (42.2) 6 (35.2) 43 (100)
Abscess 35 (77.7) 16 (94) 41 (95)
Wall inflammation 15 (33.3) 3 (18) 37 (86)
Past Sx history* 13 (28.8) 6 (35.2) 42 (98)
NRBI narrative report before implementation, NRAI narrative report after implementation, SRAI structured
report after implementation
*Seton, drainage catheter, prior surgery

Subjectively, 105 reports were evaluated by two sur- More importantly, however, when comparing the NR
geons who were asked to rate them using a Likert scale and group (n = 62) with the SR group (n = 43), the SR group
yes/no questions for completeness, helpfulness, and clarity reports significantly increased the percentage of com-
(Table 3). Overall, the AI group reports increased the pleteness (72.5–88.3% for surgeon 1, and 61.2–81.3% for
percentage of completeness (73.3–83.3% for surgeon 1, surgeon 2; p = 0.05 and 0.03, respectively), helpfulness in
and 57.8–78.3% for surgeon 2; p = 0.2 and 0.02, respec- surgical planning (7.1 ± 1.5–7.6 ± 1.5 for surgeon 1, and
tively), helpfulness in surgical planning 5.8 ± 1.4–7.1 ± 1.1 for surgeon 2; p = 0.05 and
(7.2 ± 1.5–7.3 ± 1.5 for surgeon 1, and p \ 0.001, respectively), and clarity (7.6 ± 1.3–8.3 ± 1.1
5.8 ± 1.3–6.8 ± 1.4 for surgeon 2; p = 0.84 and for surgeon 1, and 5.2 ± 1.4–7.1 ± 1.3 for surgeon 2;
p \ 0.001, respectively), and clarity (7.7 ± 1.3–8.0 ± 1.3 p = 0.006 and p \ 0.001, respectively) of the reports. The
for surgeon 1, and 5.1 ± 1.2–6.7 ± 1.5 for surgeon 2; interobserver agreement (kappa) between two surgeons NR
p = 0.33 and p \ 0.001, respectively) of the reports. and SR groups are presented in Table 4.

123
 Abdominal Radiology

Table 3 Statistical results of

Key features p value p value p value
pairwise comparison of the
NRBI versus NRAI NRBI versus SRAI NRAI versus SRAI
groups according to type of
report, time of implementation, Number 0.129 \ 0.001 \ 0.001
and surgeon’s evaluation of the
reports Internal opening 0.215 0.99 0.99
Distance from verge 0.99 \ 0.001 0.002
Opening relation 0.99 \ 0.001 \ 0.001
Type 0.639 \ 0.001 0.123
Branches 0.609 \ 0.001 \ 0.001
Exit site \ 0.001 0.052 \ 0.001
T2 signal 0.99 \ 0.001 \ 0.001
Enhancement 0.99 \ 0.001 \ 0.001
Abscess 0.249 0.038 0.99
Wall inflammation 0.817 \ 0.001 \ 0.001
Past Sx History* 0.99 \ 0.001 \ 0.001
Subjective assessment of all reports by colorectal surgeons
Evaluation of report p value p value p value
NRBI versus NRAI NRBI versus SRAI NRAI versus SRAI

SURGEON # 1
Completeness 0.146 – –
Need for re-review images 0.432 – –
Helpfulness 0.316 0.716 0.039
Clarity 0.295 0.175 0.007
SURGEON # 2
Completeness 0.057 – –
Need for re-review images 0.294 – –
Helpfulness 0.99 \ 0.001 0.005
Clarity 0.747 \ 0.001 0.004
NRBI narrative report before implementation, NRAI narrative report after implementation, NRBI narrative
report before implementation, SRAI structured report after implementation, Sx surgical
*Seton placement, drainage placement or prior surgery

Discussion Nörenberg et al. determined that referring physicians were

more satisfied with SR type of reports in staging of rectal
In this study, we evaluated the clarity, completeness, and cancer with MRI [19]. A similar conclusion was reached by
impact on surgical planning of MRI reporting of perianal Franconeri et al. who stated that MRI-SR for uterine fibroid
fistulizing disease using a structured disease-specific tem- disease helped determine the course of treatment better
plate (SR) versus narrative reporting (NR) for planning of than NR [20]. Gassenmaier et al. also recently claimed that
disease treatment by colorectal surgeons. SR type reporting SR type reports used to report shoulder MRI were more
system has many advantages. SR serves as a guide for readable and understandable [21]. Lastly, the Annotation
radiologists as it prescribes a standard methodology and and Image Markup (AIM) project of the National Institutes
systematic approach [18]. This checklist type reporting of Health Cancer Biomedical Informatics Grid has shown
style prevents significant omissions, errors due to spelling that reports of both oncologic and nononcologic imaging
and grammar, and improves standardization [2]. Numerous studies using SR type reporting are more easily integrated
prior studies have concluded that in a vast array of into the AIM system [2, 22].
abdominal diseases, SR type of reports have been extre- One of the most critical findings in our study was that
mely helpful to referring physicians [18–20]. For example, objective evaluation showed that the AI group reports
Brook et al. reported that SR-type reports were more contained significantly more key features than BI group
guiding to referring doctors with regards to tumor reports. The lack of description of the key features in BI
resectability in pancreatic cancers [18]. Similarly, group reports possibly relates to cognitive bias on the part

123
Abdominal Radiology

Table 4 Subjective assessment of all reports by colorectal surgeons

Narrative reports (n = 62) Structured reports (n = 43) p value J value
Reviewers S1 S2 S1 S2 S1 S2

Question, no. (%)

Completenessa 45 (72.5) 38 (61.2) 38 (88.3) 35 (81.3) 0.05 0.03 0.063
Need for re-review imagesb 59 (95.1) 48 (77.4) 39 (90.6) 28 (65.1) 0.37 0.17 0.190
Narrative reports (n = 62) Structured reports (n = 43) p value J value
S1 S2 S1 S2 S1 S2

Question, mean ± SD, (range)

Helpfulnessc 7.1 ± 1.5 (4–10) 5.8 ± 1.4 (2–9) 7.6 ± 1.5 (5–10) 7.1 ± 1.1 (5–9) 0.05 \ 0.001 \ 0.001
Clarityd 7.6 ± 1.3 (4–10) 5.2 ± 1.4 (3–9) 8.3 ± 1.1 (5–10) 7.1 ± 1.3 (4–9) 0.006 \ 0.001 \ 0.001

S1 surgeon 1, S2 surgeon 2
a
Graded on a 0–1 scale (0 = no, 1 = yes)
b
Graded on a 0–1 scale (0 = no, 1 = yes)
c
Graded on a scale of 0–10 (0 = not helpful, 10 = very helpful)
d
Graded on a scale of 0–10 (0 = unclear, 10 = very clear)

of the radiologist. In this type of bias, at some point of the may not be transferable to the surgical community at large.
report, the radiologist concludes that all relevant informa- However, one would assume that less experience would
tion has been communicated and fails to mention additional even benefit more from structured reporting. Thirdly, since
abnormal findings [23]. In our study, there was a statisti- we did not mandate the use of the SR after implementation,
cally different discrepancy between the two groups for the the adherence to it was only 70%. It is possible that radi-
inclusion of 10 out of 12 key features defining perianal ologists were ‘‘educated’’ on relevant points but still chose
fistulizing disease. Only for 2/12 key features, the location not to use the SR and were mentioning ‘‘more relevant’’
of the internal fistula opening and the location of the exit key features in the AI group. This highlights the need for
site, the two groups performed equally, most likely as both education on top of the report standardization. One could
features are among the most imperative in the description hypothesize that a higher adherence rate would even
of a perianal fistula. improve the results further. Finally, no comparison was
In addition, colorectal surgeons, when assessing the made between the findings in the original MRI reports and
reports subjectively, found that SR reports to be more the accuracy of the reported findings. However, the point of
complete, more helpful, and clearer than NR reports. In our the study was solely to assess the report content and clarity,
study, both surgeon 1 (highly experienced) and surgeon 2 as well as the colorectal surgeons’ satisfaction with the
(moderately experienced) found SR more clear, more report, and its impact on clinical decision making—not the
helpful, and more complete than narrative reports in ability of radiologists to predict correct perianal fistulizing
determining the treatment plan. This finding differs from disease diagnosis.
the results of Franconeri et al. [20] who found that, when In conclusion, structured MRI reports in patients with
assessing fibroids mapping before surgery, the highly perianal fistulizing disease miss fewer key features than
experienced surgeons did not significantly benefit from SR. narrative reports. Moreover, structured reports were
In contrast, when assessing pancreatic cancer staging CT described as more complete and clear, and more helpful for
reports, less-experienced surgeons did not find SR benefi- treatment planning. This potentially contributes to better
cial [18]. Of note, no significant change was seen in both surgical planning and patient outcomes.
surgeons’ opinions regarding the need to review the images
after reading the report, for both groups.
Our study has several limitations. First, although the
reports were presented anonymously, the retrospective References
design may have included inherent recall bias for the sur-
1. Bink A, Benner J, Reinhardt J, et al. Structured reporting in
geons. Secondly, only two expert surgeons in a tertiary care neuroradiology: Intracranial tumors. Front Neurol 2018; 32:1–10
center were included in the study, and therefore the results

123
 Abdominal Radiology

2. Ganeshan D, Duong PAT, Probyn L, et al. Structured reporting in 14. Sharma G, Khandige G, Mohan M. Magnetic resonance imaging
radiology. Acad Radiol 2018; 25:66–73 in perianal fistulas-a pictorial atlas. Indian J Gastroenterol 2016;
3. de Miguel Criado J, del Salto LG, Rivas PF, et al. MR imaging 35:337–342
evaluation of perianal fistulas: spectrum of imaging features. 15. Parks AG, Gordon PH, Hardcastle JD. A classification of fistula-
Radiographics 2012; 32:175–194 in-ano. Br J Surg 1976; 63:1–12
4. Abcarian H, Anal fistula principles and management, 1st ed. New 16. Van Assche G, Vanbeckevoort D, Bielen D, et al. Magnetic
York, NY: Springer, 2014:1–3 resonance imaging of the effects of infliximab on perianal fis-
5. Hamadani A, Haigh PI, Liu IA, Abbas MA. Who is at risk for tulizing Crohn’s disease. Am J Gastroenterol 2003: 98:332–339
developing chronic anal fistula or recurrent anal sepsis after ini- 17. Tutein Nolthenius CJ, Bipat S, Mearadji B, et al. MRI charac-
tial perianal abscess? Dis Colon Rectum 2009; 52:217–221 teristics of proctitis in Crohn’s disease on perianal MRI. Abdom
6. O’Malley RB, Al-Hawary MM, Kaza RK, Wasnik AP, Liu PS, Radiol 2016; 41:1918–1930
Hussain HK. Rectal imaging: part 2, perianal fistula evaluation on 18. Brook OR, Brook A, Vollmer CM, Kent TS, Sanchez N, Pedrosa
pelvic MRI - what the radiologist needs to know. AJR 2012; I. Structured reporting of multiphasic CT for pancreatic cancer:
199:43–53 potential effect on staging and surgical planning. Radiology
7. Gecse K, Khanna R, Stoker J, et al. Fistulizing crohn’s disease: 2015; 274:464–472
diagnosis and management. United Eur Gastroenterol J 2013; 19. Nörenberg D, Sommer WH, Thasler W, et al. Structured report-
1:206–213 ing of rectal magnetic resonance imaging in suspected primary
8. Beaugerie L, Carrat F, Nahon S, et al. High risk of anal and rectal rectal cancer: potential benefits for surgical planning and inter-
cancer in patients with anal and/or perianal Crohn’s disease. Clin disciplinary communication. Invest Radiol 2017; 52:232–239
Gastroenterol Hepatol 2018; 16:892–899 20. Franconeri A, Fang J, Carney B, et al. Structured vs narrative
9. Protyniak B, Shutt T, Farmer R. Squamous cell carcinoma orig- reporting of pelvic MRI for fibroids: clarity and impact on
inating from a Crohn’s enterocutaneous fistula. Case Rep Surg treatment planning. Eur Radiol 2017; 28:3009–3017
2017; 2017:1–4 21. Gassenmaier S, Armbruster M, Haasters F, et al. Structured
10. Ong EM, Ghazi LJ, Schwartz DA, Mortelé KJ. Guidelines for reporting of MRI of the shoulder – improvement of report qual-
imaging of Crohn’s perianal fistulizing disease. Inflamm Bowel ity? Eur Radiol 2017; 27:4110–4119
Dis 2015; 21:731–736 22. Zimmerman SL, Kim W, Boonn WW. Informatics in radiology:
11. Sahni VA, Ahmad R, Burling D. Which method is best for automated structured reporting o imaging findings using at AIM
imaging of perianal fistula? Abdom Imaging 2008; 33:26–30 standard and XML. Radiographics 2011; 31:881–887
12. Kim MJ. Transrectal ultrasonography of anorectal diseases: 23. Brady AP. Error and discrepancy in radiology: inevitable or
advantages and disadvantages. Ultrasonography 2015; 34:19-31 avoidable? Insights Imaging 2017; 8:171–182
13. Burdan F, Sudol-Szopinska I, Staroslawska E, et al. Magnetic
resonance imaging and endorectal ultrasound for diagnosis of
rectal lesions. Eur J Med Res 2015; 20:1–14

123