- Email: [email protected]
Radiologic Reporting and Interpretation of Occult Inguinal Hernia
ORIGINAL SCIENTIFIC ARTICLES
Radiologic Reporting and Interpretation of Occult Inguinal Hernia Joseph Miller,
MD, MS,
Aylish Tregarthen,
BSc,
Rola Saouaf,
MD,
Shirin Towfigh,
MD, FACS
Inguinal hernias are common entities. Occult inguinal hernias are difficult to diagnose on examination and can cause groin and pelvic pain. Imaging is heavily relied on to help diagnose these hernias; as such, correct interpretation of imaging studies can prevent delay in treatment for a patient with pain. We evaluated the accuracy and reliability of radiologic reports for detection of occult inguinal hernias in patients with groin and pelvic pain. STUDY DESIGN: All CT and MRI studies ordered for groin or pelvic pain during a 5-year period were analyzed. Studies were included if the original radiologic reports were available for review, and if the patient underwent operative exploration. A blinded radiologist was asked to “overread” the images. Operative findings were considered the gold standard with which radiologic reports were compared. RESULTS: Of 322 CT and MRI studies, 125 groins met criteria. Original radiologic reports were 35% accurate, with 97% positive predictive value (PPV) and 13% negative predictive value (NPV). Over-read radiologist reports were significantly different (p < 0.0001), with 79% accuracy, 97% PPV, and 30% NPV. CONCLUSIONS: Most radiologic reports issued for CT and MRI studies were incorrect for evaluation of occult inguinal hernia. Over-read radiologist reports were more than twice as accurate when evaluating the same images. The physician who is relying on radiologic reports to determine plan of care for a patient with groin or pelvic pain should inquire further into any negative study, especially if there is strong clinical suspicion for inguinal hernia. (J Am Coll Surg 2018;227:489e495. 2018 by the American College of Surgeons. Published by Elsevier Inc. All rights reserved.)
BACKGROUND:
Inguinal hernia is a well-known cause of groin pain that can be cured with operative repair.1 In the majority of patients, physical examination alone is sufficient to diagnose inguinal hernia.2 Symptomatic hernias that cannot be palpated on examination are referred to as occult inguinal hernias, a clinical entity first described in 1975.3 These hernias are confirmed at the time of operation.3
A few publications have examined the role for imaging in the identification of occult inguinal hernias,4-6 however, the standards proposed are difficult to interpret and can be contradictory. This is likely due in part to differing methodologies, in particular an inconsistent reliance on gold standard comparisons, such as operative findings. Additionally, image interpretation itself is a complex cognitive process prone to multiple sources of error7; yet many studies examine the utility of imaging modalities without inclusion of radiologists as study members, and without a study design that includes expert or dedicated review of imaging studies by a radiologist. In our surgical practice, we have observed that radiologic reports of cross-sectional imaging often did not accurately identify inguinal hernias, and patients were denied necessary hernia repair (Fig. 1). We have learned to overcome this problem by reviewing the images ourselves or with our own radiologist. Our concern, and the impetus for this research, is that it is not the common practice of most physiciansdsurgeons and non-surgeons alike. That is, physicians tend to rely purely on the
CME questions for this article available at http://jacscme.facs.org Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Received June 24, 2018; Revised July 23, 2018; Accepted August 6, 2018. From the Department of Imaging, Cedars-Sinai Medical Center, Los Angeles (Miller, Saouaf), Beverly Hills Hernia Center, Beverly Hills (Towfigh), CA, and University of Exeter Medical School, Exeter, England, United Kingdom (Tregarthen). Correspondence address: Shirin Towfigh, MD, FACS, Beverly Hills Hernia Center, 450 N Roxbury Dr #224, Beverly Hills, CA 90210. email: [email protected]
ª 2018 by the American College of Surgeons. Published by Elsevier Inc. All rights reserved.
489
https://doi.org/10.1016/j.jamcollsurg.2018.08.003 ISSN 1072-7515/18
490
Miller et al
Radiologic Report in Occult Inguinal Hernia
J Am Coll Surg
Figure 1. (A) Representative axial and (B) coronal CT images of a patient ultimately found to have bilateral direct inguinal hernias at time of surgical exploration. Imaging study was acquired with original indication of “chronic left groin pain.” Original CT report included mention of incidental renal cyst and hepatic hemangioma, but “no suspicious CT findings to explain chronic left groin pain.” Over-read report identified bilateral direct inguinal hernias.
radiologic reports. This is based on the feedback that our patients provided us about their earlier consultations. This practice can result in underdiagnosis or misdiagnosis in a large proportion of patients experiencing pain due to inguinal hernias. The aim of this study is to quantify the value of reliance on radiologic reports alone, and to determine the reliability of radiology reports in identifying inguinal hernias in patients with groin or pelvic pain. We hypothesize that radiologic reports of CT and MR pelvis commonly misreport inguinal hernias, and they are typically falsely negative.
METHODS We performed a retrospective review of records for all patients referred to a hernia-centric specialty surgical practice during a 5-year period. Inclusion criteria included all patients with examination findings suggestive, but not necessarily diagnostic, for inguinal hernia, for example, groin tenderness without palpable hernia defect or bulge, preoperative cross-sectional imaging (CT or MR) of the pelvis for the specific indication of “groin pain” or “pelvic pain” with available original radiologic report, and an operation to address the groin or pelvic pain as likely due to inguinal hernia. History of inguinal hernia repair was an exclusion criterion. All preoperative imaging studies were analyzed independently by radiologist study staff (RS). Dr Saouaf is fellowship-trained in body imaging with more than 20 years of MRI experience and a specific interest in
hernia-related pathology. She has worked side by side with our expert hernia surgeon (ST) learning the subtle findings of clinically suspicious inguinal hernias on cross-sectional imaging. For this study, she was blinded to the patient’s plan of care and their operative findings. She was tasked with identifying all pathologies she found pertinent, mirroring her standard practice of interpretation. Comparisons were made between findings in the original radiologic report and those made by blinded over-read radiologist. Definition of inguinal hernia was determined to be a direct or indirect inguinal defect with content. By this definition, discovery of preperitoneal fat extending through the fascial defect would be considered an inguinal hernia; however, an isolated cord lipoma within the scrotum only and without visible extension from the preperitoneal fat would not be considered an inguinal hernia. Informed consent for operative exploration was obtained from each patient. Preoperative discussion included the plan to repair any hernia noted, to address any undetermined pathology that might explain the patient’s chronic groin or pelvic pain (eg endometriosis), and the possibility that no hernia repair or other intervention can be performed if associated pathology was not found. The various options for hernia repair and their associated risks and benefits were reviewed in detail. All operations were performed by a single surgeon (ST). The decision to operate was based on her clinical evaluation of the patient, as well as her own review of the preoperative images. The primary preoperative diagnosis
Miller et al
Vol. 227, No. 5, November 2018
was occult inguinal hernia. If laparoscopic exploration was performed, this included peritoneal dissection and complete evaluation of the myopectineal orifice for any content, including preperitoneal fat. Operative findings were considered the gold standard. Accuracy, sensitivity, specificity, and positive and negative predictive values for diagnosis of inguinal hernia were calculated for each category of reader (original radiologic report vs over-read report). Two-tailed McNemar’s analysis was performed for comparison between reader categories. Significance was set at p ¼ 0.05. This retrospective study was approved by our IRB with a waiver of informed consent. A post-hoc qualitative review of a representative sample of the original radiology reports was also performed in an effort to identify and qualify the potential deficiencies in radiologic reports when addressing groin and pelvic pain. The sample consisted of all available reports from the first 6 months of the study period.
RESULTS A total of 322 patient records across the study period were reviewed based on CT and/or MRI pelvis performed for “groin pain” or “pelvic pain.” Radiologic reports were provided from a varied mixture of academic and community radiologists from all across the nation. Two hundred and twenty-four (70%) patients underwent operative evaluation of at least 1 groin. One hundred and twenty-five individual groins met study criteria, as they had original
Table 1.
Operative Findings (n ¼ 125)
Variable
n
%
Direct inguinal Direct only Direct þ indirect Direct þ femoral Direct þ indirect þ femoral Indirect inguinal Indirect only Indirect þ direct Indirect þ femoral Indirect þ direct þ femoral Femoral Femoral only Femoral þ direct Femoral þ indirect Femoral þ direct þ indirect No inguinal or femoral pathology
27 12 12 2 1 93 70 12 10 1 18 5 2 10 1 13
22 10 10 2 1 74 56 10 8 1 14 4 2 8 1 10
Total is >100% because more than 1 type of hernia was noted in many patients.
Radiologic Report in Occult Inguinal Hernia
491
images with original radiologic reports available and they underwent operative evaluation for occult inguinal hernia as the likely cause of their groin or pelvic pain. Inguinal hernia was confirmed at time of operation in 112 (90%), with the majority representing indirect inguinal hernia or some permutation thereof (74%) (Table 1). The true-positive reading from original radiologic reports was 26% and from the over-read radiologist was 71% (Table 2). False-negative rate was 64% and 18%, respectively, and false-positive rate was 1% and 2%, respectively. Relative performance of original radiologic report vs over-read radiologic report was similar for both CT and MR (Table 2). The original radiologic reports demonstrated an overall accuracy of 35% and over-read radiologic reports an accuracy of 79% in classifying normal and abnormal groins in comparison with operative gold standard (Table 3). This resulted in sensitivity of 0.29 and 0.79, respectively, and specificity of 0.92 and 0.77, respectively. Relative performance of original radiologic report vs over-read radiologic report was similar for both CT and MR (Table 3). The original radiologic report and the over-read both classified the same imaging study correctly 42 times (34%) (Table 4). The over-read classified a study correctly that the original report classified incorrectly 57 times (46%). McNemar’s test for correlated proportions demonstrates a significant difference between the original radiologic report and the over-read radiologic report (p < 0.0001). This significant difference persisted when limiting evaluation to either CT (p < 0.0001) or MR (p ¼ 0.0018) (Table 4). As part of a post-hoc analysis, 41 original radiology reports from the first 6 months of the studied period were qualitatively reviewed for deficiencies. Of these reports, only 18 (44%) made any reference to an inguinal canal or to the abdominal wall in any regard.
DISCUSSION Although the majority of patients with symptomatic inguinal hernia have a palpable mass or hernia defect on physical examination, a subset of patients will have occult findings. As our practice is a referral center focusing exclusively on hernia-related conditions, we see a disproportionately higher number of patients with occult inguinal hernias. These patients have symptoms highly suggestive of inguinal hernia with tenderness, but no palpable bulge or defect on examination. Imaging can help identify these occult hernias, allowing the surgeon to repair them and cure the patient of their chronic groin and pelvic pain.
492
Miller et al
J Am Coll Surg
Radiologic Report in Occult Inguinal Hernia
Table 2. Comparison of Reliability Original Radiologic Report with Blinded Radiologist Over-Read of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias Variable
True positive n %
False positive n %
False negative n %
True negative n %
Original report (n ¼ 125) Over-read report Original CT report (n ¼ 88) Over-read CT report Original MRI report (n ¼ 37) Over-read MRI report
32 89 21 65 11 24
1 3 0 2 1 1
80 23 64 20 16 3
12 10 3 1 9 9
26 71 24 74 30 65
Classic papers on the imaging findings of abdominal wall hernia most often focus on identification of the hernia sac itself, although this definition is likely to underdiagnose abdominal wall defects solely containing preperitoneal fat.8 In some situations, these might be smaller hernias with occult findings on physical examination. More subtle pathologies are likely to be recognized by looking for “abnormal ballooning of the anteroposterior diameter of the inguinal canal and/or simultaneous protrusion of fat and/or bowel within the inguinal canal (either originating from the posterior wall of the inguinal canal or through the internal inguinal ring).”9 Any perceptible defect in the abdominal wall fascia about the myopectineal orifice should be identified as representing a hernia, as should associated signs, such as fat stranding within the inguinal canal, the “lateral crescent sign” described in direct inguinal hernia,10 or the venous compression seen in femoral hernia.11 In our experience, inguinal hernias are poorly reported on CT and MRI, and occult inguinal hernias are most likely to be missed (Fig. 1).4 Patients are commonly referred to us for groin or pelvic pain of unknown etiology, as negative findings have been reported for their CT and/or MR pelvis imaging studies. These patients have often undergone multiple cross-sectional imaging studies over a span of many years, resulting in a delay in their diagnosis and treatment, as surgical repair was denied to them. As these patients come to us from across the
1 2 0 2 3 3
64 18 73 23 43 8
10 8 3 1 24 24
nation with imaging studies already having been performed, we did not control specific aspects of imaging acquisition, such as use of oral/IV contrast, or nonstandard protocols, such as dynamic or Valsalva imaging. The majority of CT and MRI studies were not ordered as dynamic or with Valsalva. We have learned to read these images ourselves and rely heavily on consultation with our own radiologist to help accurately evaluate the images. The impetus of our study was to quantify our experience to educate other physicians, with the goal of reducing the delay in diagnosis and treatment of patients with hernia as the cause of their groin or pelvic pain. We analyzed our patients with chronic groin and pelvic pain that was clinically most likely due to occult inguinal hernia, and on which we operated for primary diagnosis of inguinal hernia. We compared their original radiologic reports from CT and MRI pelvis studies against our blinded radiologist interpretation of those same studies using the surgeon’s decision to operate and her operative findings as the gold standard. It is important to note that complete surgical evaluation for inguinal hernia must include peritoneal dissection of the myopectineal orifice and evaluation of any preperitoneal fat within it. We found significant deficiencies in the original radiologic reports, with nearly 2 of 3 reports incorrectly identifying groin pathology. Only 1 of every 4 reports was truly positive; only 1 of every 10 reports was truly negative.
Table 3. Comparison of Value of Original Radiologic Report vs Blinded Radiologist Over-Read of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias Variable
Original report (n ¼ 125) Over-read report Original CT report (n ¼ 88) Over-read CT report Original MRI report (n ¼ 37) Over-read MRI report
Sensitivity
Specificity
PPV
NPV
Accuracy
0.29 0.79 0.25 0.76 0.41 0.89
0.92 0.77 1 0.33 0.90 0.90
0.97 0.97 1 0.97 0.92 0.96
0.13 0.30 0.04 0.05 0.36 0.75
0.35 0.79 0.27 0.75 0.54 0.89
NPV, negative predictive value; PPV, positive predictive value.
Miller et al
Vol. 227, No. 5, November 2018
Radiologic Report in Occult Inguinal Hernia
493
Table 4. Correlations Between Original Radiologic Report and Blinded Over-Read Radiologist Report of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias
Variable
Over-read report Correct Incorrect n % n %
Over-read CT report Correct Incorrect n % n %
Over-read MRI report Correct Incorrect n % n %
Original report (n ¼ 125) Correct Incorrect Original CT report (n ¼ 88) Correct Incorrect Original MRI report (n ¼ 37) Correct Incorrect
d 42 57 d d d d d d
d d d d 22 44 d d d
d d d d d d d 20 13
d 34 46 d d d d d d
d 2 24 d d d d d d
d 2 19 d d d d d d
These findings were similar regardless of whether CT or MRI reports were evaluated, although MRI reports were more accurate than CT reports for both over-read and original reports; this is consistent with prior studies comparing cross-sectional imaging modalities for evaluation of groin pathology.4 A positive finding of occult inguinal hernia issued by the original interpreting radiologist was predictive 97% of the time, however, the negative predictive value was only 13%. The radiologist over-read was twice as accurate (79% vs 35%, p < 0.0001), but still not as high as the surgeon’s clinical diagnosis (90%), which was made after history, examination, and her own review of the images. Negative predictive value of imaging read by over-read radiologist remained low, at 30%. The immediate implication of this study is that reliance on a radiologic report alone is inadequate in the evaluation of patients with chronic groin and pelvic pain that is due to inguinal hernia. There is an aphorism well known among radiologists: “The question is not what you look at, but what you see.”12 It is important for physicians to realize that even if an imaging study is acquired with technical perfection, we rely on interpretation by imperfect humans. We show this to be glaringly true for occult inguinal hernias. The radiologic literature has recently begun to grapple with the sources of error implicit in image interpretation, which can be broadly classified as perceptual (60% to 80%) and cognitive (20% to 40%).13 Although all included studies were ordered with specific indication of “groin pain” or “pelvic pain,” it is possible that the original interpreting radiologists simply did not perceive the underlying pathology when present. On our post-hoc review of the body of original reports, only about half (44%) mentioned the inguinal canals or abdominal wall in any way; even among these reports, radiologists would often detail a finding unilaterally
d d d d 25 50 d d d
d d d d 2 20 d d d
d d d d 2 23 d d d
d d d d d d d 54 35
d d d d d d d 1 3
d d d d d d d 3 8
p Value
<0.0001 d d <0.0001 d d 0.0018 d d
(especially if the study indication was explicitly for lateralized pain), but make no mention of the contralateral side. Our surgeon had the privilege of capturing a relevant history and physical examination, resulting in 90% accuracy in her diagnosis. Reliance on radiologic reports alone would result in only 35% of patients being offered inguinal hernia repair. This is significant, as the common practice by physicians (both surgeons and non-surgeons) is to rely on reports when determining diagnosis and plan of care, instead of independently evaluating the images themselves or performing in-person review with a radiologist. The major problem with interpretation of imaging for occult inguinal hernias is that most are missed, that is, a positive finding on imaging should be believed, but a negative finding should be questioned. We urge physicians to question any “negative” results for a CT or MRI pelvis, especially when there is high clinical suspicion for occult inguinal hernia as a cause of a patient’s groin or pelvic pain. This will help reduce delay in treatment of inguinal hernia as the cause of groin or pelvic pain. One method is to review the images oneself. We also highly encourage in-person communication with the radiologist to re-review the imaging, while the physician can share relevant clinical history or examination findings with the radiologist. This should lower the very high (64%) false-negative rate when the radiologist reviews the imaging alone. An implicit bias of our study is the attention our radiologist paid to the inguinal region in reinterpreting the images. Although blinded to the operative findings, she was aware that the patient underwent imaging for groin or pelvic pain and that hernia was being ruled out. However, the original radiologists were also reading these images for express evaluation of groin or pelvic pain,
494
Miller et al
Radiologic Report in Occult Inguinal Hernia
although hernia was not necessarily noted in the request by the ordering physician. This might explain the higher accuracy of her readings. Although the true reason for the discrepancies in interpretation between the original radiologists and our own radiologist is unknown, we know that our radiologist has gained experience in interpreting cross-sectional imaging for inguinal hernias, in part based on years of close collaboration with our surgeon. She has been able to correlate imaging findings, however subtle, with real patient symptoms and presentations. It is consistent with our own experiences that unless a radiologist is used to thinking of subtle hernias as a real pathologydas opposed to as an incidental findingdcapable of generating severe and chronic pain, they might be more likely to omit inguinal hernias from their perceptual and cognitive processes. The reduction of such omissions is an implicit goal of Radiological Society of North America’s “Radiology Reporting Initiative,” which provides best-practice templates for use in structured reporting.14 It has been shown that structured reporting can improve the content and clarity of radiologic interpretations.15 That being said, it is important to note that just because a standard template includes a comment on the groin, does not guarantee that the radiologist has carefully considered it. One interpretation of these findings suggests that radiologists might not be aware of occult inguinal hernia as a cause of groin pain. If there is any doubt in a physician’s mind about the accuracy of an imaging report, we recommend in-person review with the issuing radiologist. Collaboration of this manner can help to mitigate the risks of a perceptual error resulting in a false-negative interpretation. A secondary implication of this study concerns the status of imaging studies performed for occult hernias as examined in the literature. We show significant deficiencies in the quality of the original radiologic reports issued for cross-sectional imaging studies (ie CT and MR) performed for an indication of groin/pelvic pain. As such, any research study relying on a review of radiologic reports alone might be suspect, with results being inferior to those studies that include direct and focused re-review of the images by a study radiologist. We posit that this might be a factor that can explain some of the variability seen in the literature about the utility of cross-sectional imaging in diagnosis of hernias.
CONCLUSIONS Two of 3 radiologic reports of CT or MRI pelvis misreport occult inguinal hernias (35% accuracy). We show that a positive study is typically correct (97%), however,
J Am Coll Surg
only 1 in 10 negative studies is truly negative. Because of this high false-negative rate, if there remains clinical suspicion for an inguinal hernia as the cause of the patient’s groin or pelvic pain, we recommend that surgeons review images themselves or with their radiologist to confirm any negative findings. This will allow patients to be correctly diagnosed in a timely fashion, with the chance to undergo a potentially curative hernia operation for their groin or pelvic pain. Author Contributions Study conception and design: Miller, Towfigh Acquisition of data: Miller, Saouaf Analysis and interpretation of data: Miller, Tregarthen Drafting of manuscript: Miller, Tregarthen Critical revision: Miller, Saouaf, Towfigh Acknowledgment: Thank you to the efforts of Ali Khader, who helped in writing portions of the manuscript and with the background research. He is a medical student at Royal College of Surgeons in Ireland Medical University of Bahrain. REFERENCES 1. Towfigh S, Neumayer L. Inguinal hernia. In: Cameron JL, Cameron AM, eds. Current Surgical Therapy. 11th ed. Philadelphia: Saunders; 2013:531e535. 2. HerniaSurge Group. International guidelines for groin hernia management. Hernia 2018;22:1e165. 3. Herrington JK. Occult inguinal hernia in the female. Ann Surg 1975;181:481e483. 4. Miller J, Cho J, Michael MJ, et al. Role of imaging in the diagnosis of occult hernias. JAMA Surg 2014;149:1077e1980. 5. Garvey JF. Computed tomography scan diagnosis of occult groin hernia. Hernia 2012;16:307e314. 6. Robinson A, Light D, Kasim A, Nice C. A systematic review and meta-analysis of the role of radiology in the diagnosis of occult inguinal hernia. Surg Endosc 2013;27:11e18. 7. Busby LP, Courtier JL, Glastonbury CM. Bias in radiology: the how and why of misses and misinterpretations. Radiographics 2018;38:236e247. 8. Zarvan NP, Lee FT Jr, Yandow DR, Unger JS. Abdominal hernias: CT findings. AJR Am J Roentgenol 1995;164: 1391e1395. 9. van den Berg JC, de Valois JC, Go PM, Rosenbusch G. Detection of groin hernia with physical examination, ultrasound, and MRI compared with laparoscopic findings. Invest Radiol 1999;34:739e743. 10. Burkhardt JH, Arshanskiy Y, Munson JL, Scholz FJ. Diagnosis of inguinal region hernias with axial CT: the lateral crescent sign and other key findings. Radiographics 2011; 31:E1eE12. 11. Suzuki S, Furui S, Okinaga K, et al. Differentiation of femoral versus inguinal hernia: CT findings. AJR Am J Roentgenol 2007;189:W78eW83.
Vol. 227, No. 5, November 2018
Miller et al
12. Torrey B, ed. The Writings of Henry David Thoreau, Journal II, 1850eSeptember 15 1851. Boston: Houghton Mifflin; 1906:373. 13. Bruno MA, Walker EA, Abujudeh HH. Understanding and confronting our mistakes: the epidemiology of error in radiology and strategies for error reduction. Radiographics 2015; 35:1668e1676.
Radiologic Report in Occult Inguinal Hernia
495
14. Radiological Society of North America. RSNA Reporting Initiative. Available at: https://www.rsna.org/Reporting_ Initiative.aspx. Accessed February 8, 2018. 15. Schwartz LH, Panicek DM, Berk AR, et al. Improving communication of diagnostic radiology findings through structured reporting. Radiology 2011;260:174e181.
Radiologic Reporting and Interpretation of Occult Inguinal Hernia Joseph Miller,
MD, MS,
Aylish Tregarthen,
BSc,
Rola Saouaf,
MD,
Shirin Towfigh,
MD, FACS
Inguinal hernias are common entities. Occult inguinal hernias are difficult to diagnose on examination and can cause groin and pelvic pain. Imaging is heavily relied on to help diagnose these hernias; as such, correct interpretation of imaging studies can prevent delay in treatment for a patient with pain. We evaluated the accuracy and reliability of radiologic reports for detection of occult inguinal hernias in patients with groin and pelvic pain. STUDY DESIGN: All CT and MRI studies ordered for groin or pelvic pain during a 5-year period were analyzed. Studies were included if the original radiologic reports were available for review, and if the patient underwent operative exploration. A blinded radiologist was asked to “overread” the images. Operative findings were considered the gold standard with which radiologic reports were compared. RESULTS: Of 322 CT and MRI studies, 125 groins met criteria. Original radiologic reports were 35% accurate, with 97% positive predictive value (PPV) and 13% negative predictive value (NPV). Over-read radiologist reports were significantly different (p < 0.0001), with 79% accuracy, 97% PPV, and 30% NPV. CONCLUSIONS: Most radiologic reports issued for CT and MRI studies were incorrect for evaluation of occult inguinal hernia. Over-read radiologist reports were more than twice as accurate when evaluating the same images. The physician who is relying on radiologic reports to determine plan of care for a patient with groin or pelvic pain should inquire further into any negative study, especially if there is strong clinical suspicion for inguinal hernia. (J Am Coll Surg 2018;227:489e495. 2018 by the American College of Surgeons. Published by Elsevier Inc. All rights reserved.)
BACKGROUND:
Inguinal hernia is a well-known cause of groin pain that can be cured with operative repair.1 In the majority of patients, physical examination alone is sufficient to diagnose inguinal hernia.2 Symptomatic hernias that cannot be palpated on examination are referred to as occult inguinal hernias, a clinical entity first described in 1975.3 These hernias are confirmed at the time of operation.3
A few publications have examined the role for imaging in the identification of occult inguinal hernias,4-6 however, the standards proposed are difficult to interpret and can be contradictory. This is likely due in part to differing methodologies, in particular an inconsistent reliance on gold standard comparisons, such as operative findings. Additionally, image interpretation itself is a complex cognitive process prone to multiple sources of error7; yet many studies examine the utility of imaging modalities without inclusion of radiologists as study members, and without a study design that includes expert or dedicated review of imaging studies by a radiologist. In our surgical practice, we have observed that radiologic reports of cross-sectional imaging often did not accurately identify inguinal hernias, and patients were denied necessary hernia repair (Fig. 1). We have learned to overcome this problem by reviewing the images ourselves or with our own radiologist. Our concern, and the impetus for this research, is that it is not the common practice of most physiciansdsurgeons and non-surgeons alike. That is, physicians tend to rely purely on the
CME questions for this article available at http://jacscme.facs.org Disclosure Information: Authors have nothing to disclose. Timothy J Eberlein, Editor-in-Chief, has nothing to disclose. Received June 24, 2018; Revised July 23, 2018; Accepted August 6, 2018. From the Department of Imaging, Cedars-Sinai Medical Center, Los Angeles (Miller, Saouaf), Beverly Hills Hernia Center, Beverly Hills (Towfigh), CA, and University of Exeter Medical School, Exeter, England, United Kingdom (Tregarthen). Correspondence address: Shirin Towfigh, MD, FACS, Beverly Hills Hernia Center, 450 N Roxbury Dr #224, Beverly Hills, CA 90210. email: [email protected]
ª 2018 by the American College of Surgeons. Published by Elsevier Inc. All rights reserved.
489
https://doi.org/10.1016/j.jamcollsurg.2018.08.003 ISSN 1072-7515/18
490
Miller et al
Radiologic Report in Occult Inguinal Hernia
J Am Coll Surg
Figure 1. (A) Representative axial and (B) coronal CT images of a patient ultimately found to have bilateral direct inguinal hernias at time of surgical exploration. Imaging study was acquired with original indication of “chronic left groin pain.” Original CT report included mention of incidental renal cyst and hepatic hemangioma, but “no suspicious CT findings to explain chronic left groin pain.” Over-read report identified bilateral direct inguinal hernias.
radiologic reports. This is based on the feedback that our patients provided us about their earlier consultations. This practice can result in underdiagnosis or misdiagnosis in a large proportion of patients experiencing pain due to inguinal hernias. The aim of this study is to quantify the value of reliance on radiologic reports alone, and to determine the reliability of radiology reports in identifying inguinal hernias in patients with groin or pelvic pain. We hypothesize that radiologic reports of CT and MR pelvis commonly misreport inguinal hernias, and they are typically falsely negative.
METHODS We performed a retrospective review of records for all patients referred to a hernia-centric specialty surgical practice during a 5-year period. Inclusion criteria included all patients with examination findings suggestive, but not necessarily diagnostic, for inguinal hernia, for example, groin tenderness without palpable hernia defect or bulge, preoperative cross-sectional imaging (CT or MR) of the pelvis for the specific indication of “groin pain” or “pelvic pain” with available original radiologic report, and an operation to address the groin or pelvic pain as likely due to inguinal hernia. History of inguinal hernia repair was an exclusion criterion. All preoperative imaging studies were analyzed independently by radiologist study staff (RS). Dr Saouaf is fellowship-trained in body imaging with more than 20 years of MRI experience and a specific interest in
hernia-related pathology. She has worked side by side with our expert hernia surgeon (ST) learning the subtle findings of clinically suspicious inguinal hernias on cross-sectional imaging. For this study, she was blinded to the patient’s plan of care and their operative findings. She was tasked with identifying all pathologies she found pertinent, mirroring her standard practice of interpretation. Comparisons were made between findings in the original radiologic report and those made by blinded over-read radiologist. Definition of inguinal hernia was determined to be a direct or indirect inguinal defect with content. By this definition, discovery of preperitoneal fat extending through the fascial defect would be considered an inguinal hernia; however, an isolated cord lipoma within the scrotum only and without visible extension from the preperitoneal fat would not be considered an inguinal hernia. Informed consent for operative exploration was obtained from each patient. Preoperative discussion included the plan to repair any hernia noted, to address any undetermined pathology that might explain the patient’s chronic groin or pelvic pain (eg endometriosis), and the possibility that no hernia repair or other intervention can be performed if associated pathology was not found. The various options for hernia repair and their associated risks and benefits were reviewed in detail. All operations were performed by a single surgeon (ST). The decision to operate was based on her clinical evaluation of the patient, as well as her own review of the preoperative images. The primary preoperative diagnosis
Miller et al
Vol. 227, No. 5, November 2018
was occult inguinal hernia. If laparoscopic exploration was performed, this included peritoneal dissection and complete evaluation of the myopectineal orifice for any content, including preperitoneal fat. Operative findings were considered the gold standard. Accuracy, sensitivity, specificity, and positive and negative predictive values for diagnosis of inguinal hernia were calculated for each category of reader (original radiologic report vs over-read report). Two-tailed McNemar’s analysis was performed for comparison between reader categories. Significance was set at p ¼ 0.05. This retrospective study was approved by our IRB with a waiver of informed consent. A post-hoc qualitative review of a representative sample of the original radiology reports was also performed in an effort to identify and qualify the potential deficiencies in radiologic reports when addressing groin and pelvic pain. The sample consisted of all available reports from the first 6 months of the study period.
RESULTS A total of 322 patient records across the study period were reviewed based on CT and/or MRI pelvis performed for “groin pain” or “pelvic pain.” Radiologic reports were provided from a varied mixture of academic and community radiologists from all across the nation. Two hundred and twenty-four (70%) patients underwent operative evaluation of at least 1 groin. One hundred and twenty-five individual groins met study criteria, as they had original
Table 1.
Operative Findings (n ¼ 125)
Variable
n
%
Direct inguinal Direct only Direct þ indirect Direct þ femoral Direct þ indirect þ femoral Indirect inguinal Indirect only Indirect þ direct Indirect þ femoral Indirect þ direct þ femoral Femoral Femoral only Femoral þ direct Femoral þ indirect Femoral þ direct þ indirect No inguinal or femoral pathology
27 12 12 2 1 93 70 12 10 1 18 5 2 10 1 13
22 10 10 2 1 74 56 10 8 1 14 4 2 8 1 10
Total is >100% because more than 1 type of hernia was noted in many patients.
Radiologic Report in Occult Inguinal Hernia
491
images with original radiologic reports available and they underwent operative evaluation for occult inguinal hernia as the likely cause of their groin or pelvic pain. Inguinal hernia was confirmed at time of operation in 112 (90%), with the majority representing indirect inguinal hernia or some permutation thereof (74%) (Table 1). The true-positive reading from original radiologic reports was 26% and from the over-read radiologist was 71% (Table 2). False-negative rate was 64% and 18%, respectively, and false-positive rate was 1% and 2%, respectively. Relative performance of original radiologic report vs over-read radiologic report was similar for both CT and MR (Table 2). The original radiologic reports demonstrated an overall accuracy of 35% and over-read radiologic reports an accuracy of 79% in classifying normal and abnormal groins in comparison with operative gold standard (Table 3). This resulted in sensitivity of 0.29 and 0.79, respectively, and specificity of 0.92 and 0.77, respectively. Relative performance of original radiologic report vs over-read radiologic report was similar for both CT and MR (Table 3). The original radiologic report and the over-read both classified the same imaging study correctly 42 times (34%) (Table 4). The over-read classified a study correctly that the original report classified incorrectly 57 times (46%). McNemar’s test for correlated proportions demonstrates a significant difference between the original radiologic report and the over-read radiologic report (p < 0.0001). This significant difference persisted when limiting evaluation to either CT (p < 0.0001) or MR (p ¼ 0.0018) (Table 4). As part of a post-hoc analysis, 41 original radiology reports from the first 6 months of the studied period were qualitatively reviewed for deficiencies. Of these reports, only 18 (44%) made any reference to an inguinal canal or to the abdominal wall in any regard.
DISCUSSION Although the majority of patients with symptomatic inguinal hernia have a palpable mass or hernia defect on physical examination, a subset of patients will have occult findings. As our practice is a referral center focusing exclusively on hernia-related conditions, we see a disproportionately higher number of patients with occult inguinal hernias. These patients have symptoms highly suggestive of inguinal hernia with tenderness, but no palpable bulge or defect on examination. Imaging can help identify these occult hernias, allowing the surgeon to repair them and cure the patient of their chronic groin and pelvic pain.
492
Miller et al
J Am Coll Surg
Radiologic Report in Occult Inguinal Hernia
Table 2. Comparison of Reliability Original Radiologic Report with Blinded Radiologist Over-Read of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias Variable
True positive n %
False positive n %
False negative n %
True negative n %
Original report (n ¼ 125) Over-read report Original CT report (n ¼ 88) Over-read CT report Original MRI report (n ¼ 37) Over-read MRI report
32 89 21 65 11 24
1 3 0 2 1 1
80 23 64 20 16 3
12 10 3 1 9 9
26 71 24 74 30 65
Classic papers on the imaging findings of abdominal wall hernia most often focus on identification of the hernia sac itself, although this definition is likely to underdiagnose abdominal wall defects solely containing preperitoneal fat.8 In some situations, these might be smaller hernias with occult findings on physical examination. More subtle pathologies are likely to be recognized by looking for “abnormal ballooning of the anteroposterior diameter of the inguinal canal and/or simultaneous protrusion of fat and/or bowel within the inguinal canal (either originating from the posterior wall of the inguinal canal or through the internal inguinal ring).”9 Any perceptible defect in the abdominal wall fascia about the myopectineal orifice should be identified as representing a hernia, as should associated signs, such as fat stranding within the inguinal canal, the “lateral crescent sign” described in direct inguinal hernia,10 or the venous compression seen in femoral hernia.11 In our experience, inguinal hernias are poorly reported on CT and MRI, and occult inguinal hernias are most likely to be missed (Fig. 1).4 Patients are commonly referred to us for groin or pelvic pain of unknown etiology, as negative findings have been reported for their CT and/or MR pelvis imaging studies. These patients have often undergone multiple cross-sectional imaging studies over a span of many years, resulting in a delay in their diagnosis and treatment, as surgical repair was denied to them. As these patients come to us from across the
1 2 0 2 3 3
64 18 73 23 43 8
10 8 3 1 24 24
nation with imaging studies already having been performed, we did not control specific aspects of imaging acquisition, such as use of oral/IV contrast, or nonstandard protocols, such as dynamic or Valsalva imaging. The majority of CT and MRI studies were not ordered as dynamic or with Valsalva. We have learned to read these images ourselves and rely heavily on consultation with our own radiologist to help accurately evaluate the images. The impetus of our study was to quantify our experience to educate other physicians, with the goal of reducing the delay in diagnosis and treatment of patients with hernia as the cause of their groin or pelvic pain. We analyzed our patients with chronic groin and pelvic pain that was clinically most likely due to occult inguinal hernia, and on which we operated for primary diagnosis of inguinal hernia. We compared their original radiologic reports from CT and MRI pelvis studies against our blinded radiologist interpretation of those same studies using the surgeon’s decision to operate and her operative findings as the gold standard. It is important to note that complete surgical evaluation for inguinal hernia must include peritoneal dissection of the myopectineal orifice and evaluation of any preperitoneal fat within it. We found significant deficiencies in the original radiologic reports, with nearly 2 of 3 reports incorrectly identifying groin pathology. Only 1 of every 4 reports was truly positive; only 1 of every 10 reports was truly negative.
Table 3. Comparison of Value of Original Radiologic Report vs Blinded Radiologist Over-Read of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias Variable
Original report (n ¼ 125) Over-read report Original CT report (n ¼ 88) Over-read CT report Original MRI report (n ¼ 37) Over-read MRI report
Sensitivity
Specificity
PPV
NPV
Accuracy
0.29 0.79 0.25 0.76 0.41 0.89
0.92 0.77 1 0.33 0.90 0.90
0.97 0.97 1 0.97 0.92 0.96
0.13 0.30 0.04 0.05 0.36 0.75
0.35 0.79 0.27 0.75 0.54 0.89
NPV, negative predictive value; PPV, positive predictive value.
Miller et al
Vol. 227, No. 5, November 2018
Radiologic Report in Occult Inguinal Hernia
493
Table 4. Correlations Between Original Radiologic Report and Blinded Over-Read Radiologist Report of Cross-Sectional Imaging for Diagnosis of Occult Inguinal Hernias
Variable
Over-read report Correct Incorrect n % n %
Over-read CT report Correct Incorrect n % n %
Over-read MRI report Correct Incorrect n % n %
Original report (n ¼ 125) Correct Incorrect Original CT report (n ¼ 88) Correct Incorrect Original MRI report (n ¼ 37) Correct Incorrect
d 42 57 d d d d d d
d d d d 22 44 d d d
d d d d d d d 20 13
d 34 46 d d d d d d
d 2 24 d d d d d d
d 2 19 d d d d d d
These findings were similar regardless of whether CT or MRI reports were evaluated, although MRI reports were more accurate than CT reports for both over-read and original reports; this is consistent with prior studies comparing cross-sectional imaging modalities for evaluation of groin pathology.4 A positive finding of occult inguinal hernia issued by the original interpreting radiologist was predictive 97% of the time, however, the negative predictive value was only 13%. The radiologist over-read was twice as accurate (79% vs 35%, p < 0.0001), but still not as high as the surgeon’s clinical diagnosis (90%), which was made after history, examination, and her own review of the images. Negative predictive value of imaging read by over-read radiologist remained low, at 30%. The immediate implication of this study is that reliance on a radiologic report alone is inadequate in the evaluation of patients with chronic groin and pelvic pain that is due to inguinal hernia. There is an aphorism well known among radiologists: “The question is not what you look at, but what you see.”12 It is important for physicians to realize that even if an imaging study is acquired with technical perfection, we rely on interpretation by imperfect humans. We show this to be glaringly true for occult inguinal hernias. The radiologic literature has recently begun to grapple with the sources of error implicit in image interpretation, which can be broadly classified as perceptual (60% to 80%) and cognitive (20% to 40%).13 Although all included studies were ordered with specific indication of “groin pain” or “pelvic pain,” it is possible that the original interpreting radiologists simply did not perceive the underlying pathology when present. On our post-hoc review of the body of original reports, only about half (44%) mentioned the inguinal canals or abdominal wall in any way; even among these reports, radiologists would often detail a finding unilaterally
d d d d 25 50 d d d
d d d d 2 20 d d d
d d d d 2 23 d d d
d d d d d d d 54 35
d d d d d d d 1 3
d d d d d d d 3 8
p Value
<0.0001 d d <0.0001 d d 0.0018 d d
(especially if the study indication was explicitly for lateralized pain), but make no mention of the contralateral side. Our surgeon had the privilege of capturing a relevant history and physical examination, resulting in 90% accuracy in her diagnosis. Reliance on radiologic reports alone would result in only 35% of patients being offered inguinal hernia repair. This is significant, as the common practice by physicians (both surgeons and non-surgeons) is to rely on reports when determining diagnosis and plan of care, instead of independently evaluating the images themselves or performing in-person review with a radiologist. The major problem with interpretation of imaging for occult inguinal hernias is that most are missed, that is, a positive finding on imaging should be believed, but a negative finding should be questioned. We urge physicians to question any “negative” results for a CT or MRI pelvis, especially when there is high clinical suspicion for occult inguinal hernia as a cause of a patient’s groin or pelvic pain. This will help reduce delay in treatment of inguinal hernia as the cause of groin or pelvic pain. One method is to review the images oneself. We also highly encourage in-person communication with the radiologist to re-review the imaging, while the physician can share relevant clinical history or examination findings with the radiologist. This should lower the very high (64%) false-negative rate when the radiologist reviews the imaging alone. An implicit bias of our study is the attention our radiologist paid to the inguinal region in reinterpreting the images. Although blinded to the operative findings, she was aware that the patient underwent imaging for groin or pelvic pain and that hernia was being ruled out. However, the original radiologists were also reading these images for express evaluation of groin or pelvic pain,
494
Miller et al
Radiologic Report in Occult Inguinal Hernia
although hernia was not necessarily noted in the request by the ordering physician. This might explain the higher accuracy of her readings. Although the true reason for the discrepancies in interpretation between the original radiologists and our own radiologist is unknown, we know that our radiologist has gained experience in interpreting cross-sectional imaging for inguinal hernias, in part based on years of close collaboration with our surgeon. She has been able to correlate imaging findings, however subtle, with real patient symptoms and presentations. It is consistent with our own experiences that unless a radiologist is used to thinking of subtle hernias as a real pathologydas opposed to as an incidental findingdcapable of generating severe and chronic pain, they might be more likely to omit inguinal hernias from their perceptual and cognitive processes. The reduction of such omissions is an implicit goal of Radiological Society of North America’s “Radiology Reporting Initiative,” which provides best-practice templates for use in structured reporting.14 It has been shown that structured reporting can improve the content and clarity of radiologic interpretations.15 That being said, it is important to note that just because a standard template includes a comment on the groin, does not guarantee that the radiologist has carefully considered it. One interpretation of these findings suggests that radiologists might not be aware of occult inguinal hernia as a cause of groin pain. If there is any doubt in a physician’s mind about the accuracy of an imaging report, we recommend in-person review with the issuing radiologist. Collaboration of this manner can help to mitigate the risks of a perceptual error resulting in a false-negative interpretation. A secondary implication of this study concerns the status of imaging studies performed for occult hernias as examined in the literature. We show significant deficiencies in the quality of the original radiologic reports issued for cross-sectional imaging studies (ie CT and MR) performed for an indication of groin/pelvic pain. As such, any research study relying on a review of radiologic reports alone might be suspect, with results being inferior to those studies that include direct and focused re-review of the images by a study radiologist. We posit that this might be a factor that can explain some of the variability seen in the literature about the utility of cross-sectional imaging in diagnosis of hernias.
CONCLUSIONS Two of 3 radiologic reports of CT or MRI pelvis misreport occult inguinal hernias (35% accuracy). We show that a positive study is typically correct (97%), however,
J Am Coll Surg
only 1 in 10 negative studies is truly negative. Because of this high false-negative rate, if there remains clinical suspicion for an inguinal hernia as the cause of the patient’s groin or pelvic pain, we recommend that surgeons review images themselves or with their radiologist to confirm any negative findings. This will allow patients to be correctly diagnosed in a timely fashion, with the chance to undergo a potentially curative hernia operation for their groin or pelvic pain. Author Contributions Study conception and design: Miller, Towfigh Acquisition of data: Miller, Saouaf Analysis and interpretation of data: Miller, Tregarthen Drafting of manuscript: Miller, Tregarthen Critical revision: Miller, Saouaf, Towfigh Acknowledgment: Thank you to the efforts of Ali Khader, who helped in writing portions of the manuscript and with the background research. He is a medical student at Royal College of Surgeons in Ireland Medical University of Bahrain. REFERENCES 1. Towfigh S, Neumayer L. Inguinal hernia. In: Cameron JL, Cameron AM, eds. Current Surgical Therapy. 11th ed. Philadelphia: Saunders; 2013:531e535. 2. HerniaSurge Group. International guidelines for groin hernia management. Hernia 2018;22:1e165. 3. Herrington JK. Occult inguinal hernia in the female. Ann Surg 1975;181:481e483. 4. Miller J, Cho J, Michael MJ, et al. Role of imaging in the diagnosis of occult hernias. JAMA Surg 2014;149:1077e1980. 5. Garvey JF. Computed tomography scan diagnosis of occult groin hernia. Hernia 2012;16:307e314. 6. Robinson A, Light D, Kasim A, Nice C. A systematic review and meta-analysis of the role of radiology in the diagnosis of occult inguinal hernia. Surg Endosc 2013;27:11e18. 7. Busby LP, Courtier JL, Glastonbury CM. Bias in radiology: the how and why of misses and misinterpretations. Radiographics 2018;38:236e247. 8. Zarvan NP, Lee FT Jr, Yandow DR, Unger JS. Abdominal hernias: CT findings. AJR Am J Roentgenol 1995;164: 1391e1395. 9. van den Berg JC, de Valois JC, Go PM, Rosenbusch G. Detection of groin hernia with physical examination, ultrasound, and MRI compared with laparoscopic findings. Invest Radiol 1999;34:739e743. 10. Burkhardt JH, Arshanskiy Y, Munson JL, Scholz FJ. Diagnosis of inguinal region hernias with axial CT: the lateral crescent sign and other key findings. Radiographics 2011; 31:E1eE12. 11. Suzuki S, Furui S, Okinaga K, et al. Differentiation of femoral versus inguinal hernia: CT findings. AJR Am J Roentgenol 2007;189:W78eW83.
Vol. 227, No. 5, November 2018
Miller et al
12. Torrey B, ed. The Writings of Henry David Thoreau, Journal II, 1850eSeptember 15 1851. Boston: Houghton Mifflin; 1906:373. 13. Bruno MA, Walker EA, Abujudeh HH. Understanding and confronting our mistakes: the epidemiology of error in radiology and strategies for error reduction. Radiographics 2015; 35:1668e1676.
Radiologic Report in Occult Inguinal Hernia
495
14. Radiological Society of North America. RSNA Reporting Initiative. Available at: https://www.rsna.org/Reporting_ Initiative.aspx. Accessed February 8, 2018. 15. Schwartz LH, Panicek DM, Berk AR, et al. Improving communication of diagnostic radiology findings through structured reporting. Radiology 2011;260:174e181.