Professional Documents
Culture Documents
Author's Accepted Manuscript: 10.1016/j.juro.2016.01.101
Author's Accepted Manuscript: 10.1016/j.juro.2016.01.101
PII: S0022-5347(16)00224-X
DOI: 10.1016/j.juro.2016.01.101
Reference: JURO 13294
Please cite this article as: Sheth KR, Keays M, Grimsby GM, Granberg CF, Menon VS, DaJusta DG,
Ostrov L, Hill M, Sanchez E, Kuppermann D, Harrison CB, Jacobs MA, Huang R, Burgu B, Hennes H,
Schlomer BJ, Baker LA, Diagnosing Testicular Torsion before Urologic Consultation and Imaging: A
Validation of the TWIST Score, The Journal of Urology® (2016), doi: 10.1016/j.juro.2016.01.101.
DISCLAIMER: This is a PDF file of an unedited manuscript that has been accepted for publication. As a
service to our subscribers we are providing this early version of the article. The paper will be copy edited
and typeset, and proof will be reviewed before it is published in its final form. Please note that during the
production process errors may be discovered which could affect the content, and all legal disclaimers
that apply to The Journal pertain.
Embargo Policy
All article content is under embargo until uncorrected proof of the article becomes available
online.
We will provide journalists and editors with full-text copies of the articles in question prior to the embargo
date so that stories can be adequately researched and written. The standard embargo time is
12:01 AM ET on that date. Questions regarding embargo should be directed to jumedia@elsevier.com.
ACCEPTED MANUSCRIPT
PT
1
Affiliations: University of Texas Southwestern Medical Center, Dallas, TX
2
Children’s Hospital of East Ontario, Ottawa, Ontario
3
RI
Phoenix Children’s Hospital, Phoenix, AZ
4
Mayo Clinic, Rochester, MN
5
Nationwide Children’s Hospital, Columbus, OH
6
SC
Harvard Medical School, Boston, MA
7
Ankara Üniversitesi Tıp Fakültesi, Ankara, Turkey
8
Children’s Health, Dallas, TX
Descriptive Runninghead (41 characters, max 50): Diagnosing Testicular Torsion with
TWIST score
EP
Funding: This study was supported by an NIH grant, R21DK092654 (PI: Baker, LA).
AC
1
ACCEPTED MANUSCRIPT
Abstract
PURPOSE: TWIST (Testicular Workup for Ischemia and Suspected Torsion) score uses
urologic history and physical exam to assess risk of testis torsion. The parameters include
testis swelling (2 points), hard testis (2), absent cremasteric reflex (1), nausea/vomiting
(1), and high-riding testis (1). While TWIST has been validated when scored by urologists,
PT
its diagnostic accuracy amongst non-urologic providers is unknown. We assessed the
utility of the TWIST score when collected by non-urologic non-physician providers,
mirroring the ER evaluation of acute scrotal pain.
RI
MATERIALS AND METHODS: Pediatric patients with unilateral acute scrotum were
prospectively enrolled in a NIH clinical trial. After undergoing basic history and physical exam
SC
training, EMTs calculated TWIST score and determined Tanner stage per a pictorial diagram.
Clinical torsion was confirmed by surgical exploration. All data were captured into RedCap, and
receiver operating characteristic (ROC) curves evaluated the diagnostic utility of TWIST.
U
RESULTS: Of 128 patients (mean age 11.3), 44 (mean age 13.0) had torsion. TWIST score
cutoff values of 0 and 6 derived from ROC analysis identified 31 high, 57 intermediate, and 40
AN
low risk patients (positive predictive value 93.5%; negative predictive value 100%).
CONCLUSIONS: TWIST score assessed by non-urologists, such as EMTs, is accurate. Low risk
patients do not require ultrasound to rule out torsion. High risk patients can proceed directly to
M
surgery, avoiding the ultrasound in >50% of patients. In the future, EMTs and/or ER triage
personnel could calculate TWIST score to guide radiologic workup and immediate surgical
intervention at initial assessment long before urologic consultation.
D
TE
C EP
AC
2
ACCEPTED MANUSCRIPT
Introduction
Testicular torsion is one of the few pediatric urologic emergencies, accounting for only
10-15% of acute scrotum presentations1, 2 with an annual incidence of 3.8 per 100,000 pediatric
patients.3 Intervention within 4-8 hours is critical to prevent permanent testicular loss or atrophy
PT
from compromised testicular arterial flow.4, 5 Thereafter, the testicle is often not salvageable
RI
resulting in orchiectomy rates of 32-41%.3, 6 While delayed ER presentation cannot be
controlled, prompt and accurate diagnosis upon arrival is essential to identify patients requiring
SC
surgical detorsion.
U
The classic presentation for testicular torsion in pubertal males is acute onset unilateral
AN
testicular pain with nausea and vomiting. An absent cremasteric reflex has been considered
specific for testicular torsion,7 although there are reports of torsion with present cremasteric
M
reflex.8 Testicular swelling, tenderness and high lie are nonspecific, often making diagnosis
solely on physical exam difficult.9 Therefore, testicular ultrasound with doppler is heavily relied
D
upon for confirmation.10, 11 Since awaiting any imaging constitutes a time delay, risk scoring
TE
systems based on signs and symptoms only, such as the TWIST (Testicular Workup for Ischemia
Barbosa et al. devised and validated the TWIST scoring system, which assigns a summed
C
TWIST score (range 0-7 points) based on the absence (0 points) or presence of the following 5
AC
variables: testicular swelling (2 points), hard testicle (2), absent cremasteric reflex (1),
nausea/vomiting (1) and high riding testis (1). Their ROC analysis yielded all binary variables.
Thus, for the categories of testicular swelling and hard testis, patients could only receive a score
of 0 or 2 points (no option for 1 point). Per their initial validation, patients at high risk for
3
ACCEPTED MANUSCRIPT
torsion (TWIST ≥ 5) could proceed straight to OR without imaging as the PPV was 100%.
Patients with intermediate risk (TWIST = 3-4) required ultrasound to evaluate for torsion, and
patients at low risk for torsion (TWIST ≤ 2) did not require scrotal ultrasound as the NPV was
100%.12 However, the TWIST score does not account well for physiologic differences in
PT
children, control for inter-observer variability, or substitute for medico-legal need of ultrasound
RI
documentation. In Barbosa et al. data were collected by urologists, but in practice, this would
SC
evaluate the utility of the TWIST score when measured by trained non-physician, non-urologic
personnel, specifically EMTs, who are often the first medical providers to encounter patients in
an emergency setting.
U
AN
Materials and Methods
M
The study population was drawn from an NIH-funded prospective study evaluating the
TE
use of Near Infrared Spectroscopy (NIRS) in the diagnosis of torsion.13 Between March 2013
and March 2015, we prospectively evaluated males aged one month to 21 presenting to a tertiary
EP
care pediatric ER with acute scrotum, defined as painful scrotum or testis, abdominal pain,
C
and/or waddling gait ("Cowboy shuffle") from painful scrotum. Patients with synchronous
AC
bilateral testicular torsion or a previously known testicular or scrotal pathology were excluded in
the study protocol due to inability to use the contralateral testis as an internal control for NIRS
vascular problems that could affect total body tissue oxygenation levels, and thus NIRS
measurements.
4
ACCEPTED MANUSCRIPT
Study Design
Per the NIH study protocol, the ER was instructed to page the on-call EMT research personnel
upon arrival of any patients with acute scrotum. Potentially eligible study subjects were
PT
approached by the on-call EMT and screened for inclusion and exclusion criteria. Informed
consent and required study data were collected while patients were receiving care, ultrasound
RI
and/or surgery, avoiding any delay in care. Scrotal ultrasound was used as the gold standard for
SC
diagnosis of testicular torsion and was intended to be performed for all patients included in the
study, unless the physician's clinical suspicion was high enough to forego ultrasound and
U
proceed to the OR for urgent detorsion. Patients with no evidence of torsion on ultrasound were
AN
given a urology follow-up appointment in 2 weeks and ER warning signs.
For all enrolled study patients, the research EMTs evaluated the patient, assigning binary
M
components of the TWIST score and Tanner stage using a descriptive and pictorial table (see
D
Figure 1). The EMT received no training specific to the scrotal exam or TWIST scoring, but
TE
rather relied on basic history and physical exam training they learned during EMT certification.
Patients without complete TWIST score components were excluded from this study.
EP
Data Analysis
C
All collected data were entered into a RedCap database (CTSA NIH Grant
AC
UL1TR001105)14 and extracted as needed for analysis. T-tests, Fischer’s exact test and Wilcoxan
rank sum tests were used for comparisons. ROC curves were constructed to analyze and compare
the performance of the TWIST score as a diagnostic test for torsion.15 Optimal cutoff values for
low, intermediate, and high risk groups were chosen to maximize performance of the test, taking
into account clinically meaningful results to optimize NPV and PPV while limiting false
5
ACCEPTED MANUSCRIPT
negatives and false positives. All statistical analysis was performed with Stata 12 (College
Station, TX).
PT
RI
U SC
AN
M
D
TE
C EP
AC
6
ACCEPTED MANUSCRIPT
Results
Among the 316 patients assessed for eligibility, 115 did not meet study criteria and 47
declined to participate. One patient was taken straight to the OR for high suspicion of torsion
PT
without ultrasound and 2 patients without torsion did not receive an ultrasound. One of these
patients refused ultrasound and the other had low clinical suspicion and ultrasound was deferred
RI
per the ED. Both patients had Urology follow-up with no evidence of torsion at that time.
SC
128/154 enrolled patients had all TWIST data available (Figure 2). 44/128 (34.4%) patients were
diagnosed with torsion and surgically confirmed. Amongst those not diagnosed with torsion in
U
the ER, no patients presented with missed torsion, although only 45% returned for clinical
AN
follow-up. Demographic characteristics of all patients with and without torsion are demonstrated
in Table 1. Patients with torsion were older (13.0 vs 10.4 yrs, p = 0.001), were more likely to be
M
white or black (p = 0.003), had higher Tanner stage (p < 0.001), and had lower median hours of
The TWIST score components and total score (range 0-7) distribution is shown in Table
2. The median TWIST score for torsion patients was 6 and the median TWIST score of non-
EP
torsion patients was 1 (p < 0.001). The ROC curve using all points in the TWIST score had an
AUC = 0.95 (95% CI 0.91-0.98) (Figure 3a). Clinically meaningful TWIST score cutoff values
C
of 0 and 6 were used to categorize patients into low risk (TWIST = 0), intermediate risk (TWIST
AC
= 1-5), and high risk (TWIST ≥ 6) with an optimized AUC of 0.90 (95% CI 0.85-0.94) (Figure
3b). There were no patients with a TWIST score of 0 that had torsion, giving a negative
predictive value (NPV) of 100% and specificity of 47.6%. Of those with TWIST ≥ 6, 29/31 had
torsion for a positive predictive value (PPV) of 93.5% and a sensitivity of 65.9%. The two
7
ACCEPTED MANUSCRIPT
torsion patients with a TWIST score of 1 for nausea/vomiting were manually detorsed at and
outside facility and operative room findings remained consistent with this.
For Tanner stage 3-5 patients, a high risk TWIST score had a PPV of 100% and
PT
sensitivity of 65.6% (Figure 4a). In contrast, for Tanner stage 1-2 patients, a high risk TWIST
score had a PPV of 77.8% and sensitivity of 70.0% (Figure 4b). The two patients with high risk
RI
TWIST score without testicular torsion were a Tanner 1 and Tanner 2 patient with torsion of the
SC
appendix testis.
U
AN
M
D
TE
C EP
AC
8
ACCEPTED MANUSCRIPT
Discussion
In this study, we assessed the TWIST score obtained by trained non-physician, non-
urologic providers. In our population, the TWIST score performed well as a diagnostic test for
PT
torsion, although not as well as previously reported and with different optimal cutoff values.12
Based on our results we have devised an algorithm to evaluate patients who present to the ER
RI
with an acute scrotum (Figure 5).
SC
In our population, no torsions were missed on follow-up and all patients taken to the
operating room had torsion or testicular ischemia in the absence of torsion, indicating recent
U
detorsion (spontaneous, manual by ER, or with anesthesia). In current practice, ultrasound is
AN
increasingly used to guide the diagnosis of testicular torsion,10, 11 with reported 100% sensitivity,
97.9% specificity and 98.1% diagnostic accuracy. 2 Thus, ultrasonography served as the gold
M
standard in our study population although constituting a 30-60 minute time delay in diagnosis.
D
Thus, there is a growing effort to return to traditional history and physical exam findings
TE
to diagnose torsion, decreasing the reliance on imaging, minimizing cost, and facilitating rapid
surgical intervention.16-19 The TWIST score is easy to calculate with a simple patient evaluation.
EP
In our study, 29/44 (65.9%) testicular torsions were detectable by a high risk TWIST score (≥ 6)
C
and only 2/31 (6.5%) in the high risk group without testicular torsion would undergo a negative
AC
unnecessary surgical exploration. Per our ROC analysis the high risk score cutoff was 6 rather
than 5, which was reported previously.12 In our population 4/12 (33.3%) patients with a TWIST
score of 5 did not have testicular torsion, which would lead to an unacceptably high negative
exploration rate. Alternatively, a high risk cutoff of 7 would yield a 100% PPV, but lower
sensitivity to 34.1%, leading to an optimal cutoff of 6. In addition, our low risk group cutoff of 0
9
ACCEPTED MANUSCRIPT
was different than previously published. 40/128 (31.3%) comprised the low risk group with no
cases of testicular torsion (100% NPV). If the goal of a low risk TWIST score is to avoid use of
ultrasound could be avoided in all low (31.3%) and high (24.2%) risk patients, comprising over
PT
50% of patients. In comparison, Barbosa et al. found ultrasound unnecessary in ~80% of
RI
patients.12
SC
Torsion has a bimodal age distribution with the first peak in the neonatal period and the
second peak around puberty.20 While post-pubertal children usually present with severe testicular
U
pain, identifying typical torsion symptoms and performing sonography to appropriately diagnose
AN
torsion is more challenging in prepubescent children.6, 21 In our study, the two patients without
torsion who were in the high risk group were Tanner stage 1 or 2 with a diagnosis of torsion of
M
In Tanner stage 3-5 patients, the high risk TWIST score had a PPV of 100%, signifying
TE
that the TWIST score performs better for peri-pubertal or post-pubertal children. TWIST also
performs well in Tanner 1-2 patients, but some of these children with torsion of appendix testis
EP
will be categorized as high risk for torsion. Therefore, ultrasound should be considered for these
One key difference and asset in this validation of the TWIST score that may explain the
AC
differences in group stratification is that all the involved components were collected by non-
available in the ER and the initial evaluation and decision to obtain an ultrasound or not is quite
often done by non-urologists. Thus, we suggest the cutoff values seen in our study are more valid
10
ACCEPTED MANUSCRIPT
for use in the ER. Implementation and evaluation of the TWIST score by ER providers and triage
nurses following our diagnostic algorithm will be the next step to potentially expedite urologic
consultation and minimize time to the OR. Taking this one step further, validating EMT-
generated TWIST scoring opens the door to early risk stratification in the field or during transit
PT
to the hospital, analogous to Glasgow Coma Scale use. Such rapid triage could allow for
RI
expedited care on arrival for patients with TWIST ≥ 6, bypassing the ER with direct transport to
the OR. Of course, this time-saving approach would require further investigation and evaluation
SC
prior to routine implementation.
U
Limitations
AN
The NIH study was powered for NIRS performance to diagnose testicular torsion. Thus, this
secondary outcome sub-analysis evaluating TWIST score performance is limited by the small
M
number of patients. Similarly, patients without torsion in the ER had poor follow-up, which
D
could lead to misclassification of non-torsion patients. Furthermore, the TWIST score does not
TE
account for time since initial symptom onset. Generally, as torsion progresses, more signs and
symptoms associated with the TWIST score will be present. Our tertiary care center is often a
EP
referral center for pediatric patients with testicular pain, and our torsion rate in this study of
34.4% is much higher than reported in the literature (10-15%).1, 2 Many patients are transferred
C
from outside hospitals, prolonging their duration of symptoms. Thus, our study cohort may
AC
represent a biased group of patients with prolonged torsion, enabling the TWIST score to be
more diagnostic. Furthermore, the TWIST score does not incorporate severity of pain in risk
stratifying patients. While this may be hard to quantify for patients, clinical suspicion tends to be
higher when a patient presents with sudden onset severe pain. Lastly, the results are only
11
ACCEPTED MANUSCRIPT
applicable to the group of patients that were not excluded in our study due to prior testicular
pathology or other medical co-morbidities. However, strengths of this study include the
prospective internally controlled study design and the patient evaluation with TWIST scoring by
PT
Conclusion
RI
The TWIST score was highly predictive in our population when evaluated by EMTs, especially
SC
in Tanner 3-5 patients where both the PPV and NPV were 100%. Therefore, our proposed
algorithm can potentially guide emergency room physicians and staff to triage patients
U
presenting with acute scrotum. Due to difficulty in definitive torsion diagnosis in Tanner 1 and 2
AN
patients, we recommend obtaining an ultrasound even in high risk TWIST patients. Since low
risk patients do not require ultrasound to rule out torsion and high risk patients Tanner 3-5 can
M
12
ACCEPTED MANUSCRIPT
References
PT
in children presenting to the pediatric department with acute scrotum. AJR
American journal of roentgenology 2013, 200(5):W444-449.
3. Zhao LC, Lautz TB, Meeks JJ et al: Pediatric testicular torsion epidemiology using a
RI
national database: incidence, risk of orchiectomy and possible measures toward
improving the quality of care. The Journal of urology 2011, 186(5):2009-2013.
4. Visser AJ, Heyns CF: Testicular function after torsion of the spermatic cord. BJU
SC
International 2003, 92(3):200-203.
5. Kapoor S: Testicular torsion: a race against time. International journal of clinical
practice 2008, 62(5):821-827.
6. Cost NG, Bush NC, Barber TD et al: Pediatric testicular torsion: demographics of
U
national orchiopexy versus orchiectomy rates. The Journal of urology 2011, 185(6
AN
Suppl):2459-2463.
7. Rabinowitz R: The importance of the cremasteric reflex in acute scrotal swelling in
children. The Journal of urology 1984, 132(1):89-90.
8. Nelson CP, Williams JF, Bloom DA: The cremasteric reflex: a useful but imperfect
M
1):73-76.
10. Liguori G, Bucci S, Zordani A et al: Role of US in acute scrotal pain. World journal of
TE
14. Harris PA, Taylor R, Thielke R et al: Research electronic data capture (REDCap)--a
metadata-driven methodology and workflow process for providing translational
research informatics support. Journal of biomedical informatics 2009, 42(2):377-
381.
15. DeLong ER, DeLong DM, Clarke-Pearson DL: Comparing the areas under two or
more correlated receiver operating characteristic curves: a nonparametric
approach. Biometrics 1988, 44(3):837-845.
16. Boettcher M, Bergholz R, Krebs TF et al: Clinical predictors of testicular torsion in
children. Urology 2012, 79(3):670-674.
13
ACCEPTED MANUSCRIPT
17. Boettcher M, Krebs T, Bergholz R et al: Clinical and sonographic features predict
testicular torsion in children: a prospective study. BJU Int 2013, 112(8):1201-1206.
18. Srinivasan A, Cinman N, Feber KM et al: History and physical examination findings
predictive of testicular torsion: an attempt to promote clinical diagnosis by house
staff. Journal of pediatric urology 2011, 7(4):470-474.
19. Beni-Israel T, Goldman M, Bar Chaim S et al: Clinical predictors for testicular torsion
as seen in the pediatric ED. The American journal of emergency medicine 2010,
PT
28(7):786-789.
20. Sharp VJ, Kieran K, Arlen AM: Testicular torsion: diagnosis, evaluation, and
management. American family physician 2013, 88(12):835-840.
RI
21. Patriquin HB, Yazbeck S, Trinh B et al: Testicular torsion in infants and children:
diagnosis with Doppler sonography. Radiology 1993, 188(3):781-785.
U SC
AN
M
D
TE
C EP
AC
14
ACCEPTED MANUSCRIPT
Race
PT
White 11 (25.0%) 16 (19.1%) 0.0032
Hispanic 18 (40.9%) 58 (69.1%)
Black 14 (31.8%) 9 (10.7%)
RI
Asian 1 (2.3%) 0
Other 0 1 (1.2%)
SC
1 (6.3 yrs) 6 (14.3%) 28 (33.3%) <0.0014
2 (10.8 yrs) 4 (9.5%) 31 (36.9%)
3 (13.7 yrs) 12 (28.6%) 11 (13.1%)
U
4 (15.3 yrs) 17 (40.5%) 10 (11.9%)
5 (15.0 yrs) 3 (7.1%) 4 (4.8%)
AN
Median hours of pain prior to 17.3 (0.6-129.3) 29.2 (0.9-346.1) 0.024
arrival (range)
1
Two-tailed t-test
M
2
Fischer exact test used
3
Tanner stage missing in two patients with torsion
4
Wilcoxan rank sum test used due to non-normal distribution
D
Table 1: Patient characteristics: This table describes patient demographics and characteristics in the
TE
torsion and non-torsion groups. The patients with torsion were found to be significantly older, more
likely to be black or white, had a higher Tanner stage and shorter duration of pain.
C EP
AC
ACCEPTED MANUSCRIPT
PT
2 0 9 (10.7%)
3 2 (4.6%) 9 (10.7%)
4 2 (4.6%) 4 (4.8%)
RI
5 8 (18.2%) 4 (4.8%)
6 14 (31.8%) 2 (2.4%)
7 15 (34.1%) 0
SC
TWIST Score Risk Category
Low (0) 0 40 (47.6%) <0.001
Intermediate (1-5) 15 (34.1%) 42 (50.0%)
U
High (6-7) 29 (65.9%) 2 (2.4%)
Wilcoxan rank sum test used to compare groups due to non-normal distribution
AN
Table 2: TWIST score results: The individual components of the TWIST score are identified for all
patients with and without torsion, and the break-up of patients in the low (0), intermediate (1-5), and
M
high risk groups (6-7) are shown. There was no significant difference in the break-down of TWIST score
components for patients with torsion when sub-stratified by Tanner stage.
D
TE
C EP
AC
ACCEPTED MANUSCRIPT
PT
RI
U SC
AN
M
D
TE
C EP
AC
Figure 1:Tanner stage chart used by EMTs to identify patient Tanner stage at time of evaluation
ACCEPTED MANUSCRIPT
Excluded (n=162)
PT
Assessed for eligibility (n = 316) ♦ Not meeting inclusion criteria (n= 115)
• Consent not properly obtained (n=16)
• Comorbid conditions (n=23)
RI
• Obvious bug bites (n=11)
• Bilateral scrotal pain (n=10)
• Age less than 1 month (n=1)
SC
• No scrotal pain on assessment (n=21)
• Did not follow study protocol (n=20)
• Previous scrotal surgery (n=6)
U
• Prior or chronic testis pain (n=7)
AN
Enrolled (n= 154) ♦ Declined to participate (n= 47)
♦ Other reasons (n = 0)
M
TWIST data unavailable (n = 26)
D
Analysed for primary endpoint (n= 128)
TE
EP
Underwent scrotal ultrasound (n = 125) No scrotal ultrasound (n = 3)
C
AC
PT
RI
U SC
7
AN
M
D
TE
Figure 3: ROC curve for TWIST score demonstrates an AUC of 0.95 (95% CI 0.91-0.98) (A). ROC curve for TWIST score risk categories
demonstrates an AUC of 0.90 (95% CI 0.85-0.94) (B).
C EP
AC
ACCEPTED MANUSCRIPT
1 1
PT
RI
SC
7
7
U
AN
M
D
TE
Figure 4: ROC curve for TWIST score in Tanner 2-5 patients demonstrates an AUC of 0.95 (95% CI 0.91-0.99) (A). The ROC curve for TWIST score
in Tanner 1 patients demonstrates an AUC of 0.96 (95% CI 0.91-1.00) (B). The cutoff risk category cutoff points (0 and 6) are circled in both ROC
curves.
C EP
AC
ACCEPTED MANUSCRIPT
PT
RI
Calculate TWIST Score (0-7) and Tanner Stage (1-5)
U SC
AN
Low Risk Intermediate Risk High Risk
(TWIST score = 0) (TWIST score = 1-5) (TWIST score ≥ 6)
M
D
TE
No Testicular Torsion Obtain Scrotal US to evaluate Tanner 3-5 Tanner 1-2
(No US necessary) for testicular torsion
C EP
List of abbreviations:
ER Emergency Room
PT
ROC Receiver Operating Characteristics
RI
NPV Negative Predictive Value
SC
NIRS Near Infrared Spectroscopy
OR Operating Room
U
MRI Magnetic Resonance Imaging AN
M
D
TE
C EP
AC