Clinical and sonographic features predict testicular torsion in children: a prospective study




  • To test the clinical and sonographic predictors of testicular torsion (TT) with the aim of reducing negative exploration rates.

Patients and Methods

  • We performed a prospective study of all boys treated for ‘acute scrotum’ at our institute between January 2001 and April 2012 and clinical findings were documented.
  • If available, ultrasonography (US) was added to the diagnostic evaluation.
  • A prediction of the diagnosis was based on clinical and sonographic features, and was followed by surgical exploration in all patients.


  • A total of 104 patients were included in the 16-month period of the study.
  • No single finding excluded TT. The clinical features (pain <24 h, nausea/vomiting, abnormal cremasteric reflex, high position of the testis) appeared predictive (100% sensitivity) and the clinical scoring system was proven to be reliable, reducing the negative exploration rate by >55%.
  • Ultrasound predictors alone were not able to identify all boys with TT.


  • It is safe to refrain from routine surgical exploration in every child with acute scrotum if the clinical score is applied, which results in a marked reduction of negative explorations.
  • A reliable diagnosis could not be obtained based on US alone. As scrotal US is unpleasant for the child, we propose to refrain from this if the clinical score is positive. Patients with a negative clinical score are suitable candidates for US to establish and secure diagnosis.

testicular torsion




torsion of the testicular appendix




odds ratio


positive predictive value


negative predictive value


Testicular torsion (TT), epididymo-orchitis (EO), and torsion of the testicular appendix (AT) are the three most common causes of ‘acute scrotum’ in children [1, 2]. The annual incidence of TT is 3.8% in males aged <18 years [3]. It has a bimodal distribution, with peaks in the perinatal period and in adolescence, which reflects the clinical distinction between extravaginal torsion in newborns and intravaginal torsion in older children. TT accounts for 5–25% of acute scrotum in children, and a delay in its diagnosis may result in loss of the testis [1, 2, 4], so typical history and clinical findings merit immediate intervention; however, when no scoring system is applied or no additional diagnostic measures are undertaken, the negative surgical exploration is very high (86%) [4].

In children and adolescents, a variety of clinical predictors such as a short pain duration, nausea or vomiting, high position of the testicle, abnormal ipsilateral cremasteric reflex and scrotal skin changes have been identified, mostly in retrospective studies, as being associated with an increased likelihood of TT [4-6]. Seasonal variations have even been implicated in contributing to TT, with the lower temperatures and humidity in winter and spring being associated with increased incidence rates [7, 8].

The imaging methods MRI and nuclear scintigraphy have very high accuracy rates, with a sensitivity and specificity of 93 and 100% and 100 and 59%, respectively [9, 10]; however, they are expensive, not universally available, scintigraphy involves exposure to ionizing radiation, and, more importantly, they are time-consuming and may thus significantly delay diagnosis. Hence, in addition to clinical examination, Doppler ultrasonography (US) is increasingly used in the management of patients with suspicion of TT. It gives excellent imaging of anatomical details and perfusion, with a reported sensitivity of 69–90% and specificity 98–100% [11, 12]. Because of its usual prompt availability, short duration and low costs, US has become the standard imaging method for acute scrotum [13]. Nevertheless, US has its limitations, such as a high operator dependency and the occurrence of false-negative results, leading to a non-surgical strategy potentially dangerous to the testis [14].

After compiling a retrospective inventory of the results from our institution, we identified the most sensitive clinical predictors of TT [4]. The aim of the present study was to test a clinical scoring system derived from these predictors and Doppler US prospectively, without changing our preoperative clinical evaluation strategy. Our ultimate goal was to change our future preoperative management only if the application of the scoring system was proven safe and if a reduction in surgical explorations would not have resulted in increased risk of missed TT.

Patients and Methods

All patients treated for acute scrotum at the Emergency Department of the Hamburg-Eppendorf University Medical Centre and the Altona Children's Hospital between January 2011 and April 2012 were included in the study. Institutional review board approval was obtained before beginning the investigation.

Data gathered for medical history included duration of symptoms, associated symptoms, history of trauma, previous episodes of pain, medications, other medical problems and sexual activity.

All participants were examined physically by a resident or the consultant of Paediatric Surgery. Aspects of the physical examination that were included were the involved testicle side, the presence of erythema, swelling, or tenderness of the testicle and/or epididymis, the position of the testicle in relation to the contralateral side, the blue dot sign, urethral discharge, and the presence of a normal cremasteric reflex. This reflex is elicited by stroking or pinching the medial thigh, causing contraction of the cremaster muscle, which elevates the testis. The cremasteric reflex was considered pathological if the testicle moved <0.5 cm on the ipsilateral side and was normal on the contralateral side.

Based on the retrospective analysis and identification of prognostic clinical features, a score was created that was completed for every patient presenting with acute scrotum. The diagnostic prediction was based on four symptoms: duration of pain (<24 h), nausea/vomiting, high testicular position and abnormal cremasteric reflex. TT was predicted if at least one of the four symptoms were present.

Urine analysis was performed in the Emergency Department using a urine dipstick (Combur; Roche Diagnostics GmbH, Mannheim, Germany). For the purpose of the analysis, the test was considered abnormal if leukocytes, erythrocytes or protein levels were at least twice the normal value.

Ultrasonography was performed during office hours by experienced paediatric radiologists using a high-resolution 12-5-MHz linear-array transducer (HDI 5000 SonoCT, Philips Medical Systems, Eindhoven, the Netherlands). US could not be performed during night hours. The radiologist was not informed of the outcome of the clinical prediction. Determination of the sonomorphology of the scrotum and epididymis, including echogenicity and echotexture, was performed. Two terms were used to describe echogenicity, as in the study by Chmelnik et al. [15], normal echogenicity (a homogeneous pattern) and diffuse or focal hyper- or hypoechogenicity (a heterogeneous and homogeneous pattern). The evaluation included measuring the bilateral testicular volume using a double ellipsoid method on two dimensions. Finally, central and peripheral perfusion of the testicle with colour Doppler US in comparison with the other side was assessed. An ultrasound diagnosis of TT was made if central perfusion was absent or uncertain. Based on the clinical prediction, and the Doppler US findings, the eventual prediction was made (TT or non-TT) before the surgical exploration took place.

According to local protocol, surgical exploration was performed in all patients. Under general anaesthesia the scrotum was opened through a short transverse mid-scrotal incision. The testis was exposed, and if torsion was present, it was detorqued and wrapped in warm saline sponges for 30 min. If no signs of perfusion were present, the testis was removed. If the testicle appeared viable it was secured in the correct position with two or three sutures. In the case of TT, contralateral orchidopexy was routinely performed in the same session. Postoperatively, patients received pain medication and, depending on the eventual diagnosis, antibiotic therapy.

Statistical analysis was performed using SAS 9.0 and SPSS 17.0. Data are presented as mean (sd) values. Differences between groups and predictive values were calculated using ANOVA, chi-squared test or CART for univariate or multivariate analysis, respectively, and are expressed by value with 95% CIs. A P value <0.05 was considered to indicate statistical significance.


During the study period 104 boys presented with acute scrotum. Of these, 12 (11.5%) had TT, 71 (68.3%) had AT and 17 (16.3%) had EO. These diagnoses were based on the results of the surgical exploration. Other pathologies included idiopathic scrotal oedema and haematoma. US was performed in 67 boys, which included all but one boy with TT.

Before surgery, TT was predicted by the clinical scoring system in 47 (45.2%) boys, which included all 12 boys who truly had TT, so no testis at risk would have been missed by using the score. Of the other 35 boys explored, 60% had AT, 13% had epididymitis, and one had idiopathic scrotal oedema.

In all the boys with TT, the score assigned to them was ≥1 (score of 1, n = 3; score of 2, n = 5; score of 3, n = 3 and score of 4, n = 1). As described in Table 2, 11 boys with TT had a short pain history (<24 h), four had nausea and/or vomiting, six had a pathological cremasteric reflex on the ipsilateral side and five had a testicle higher in position.

Application of the clinical score alone would have reduced our negative exploration rate by 55%. The clinical score showed a sensitivity of 100% (95% CI: 71.6–100%), a specificity of 63% (95% CI: 59.3–63.0%), and an accuracy of 66% (95% CI: 59.8–66.3%). If features detected by US alone had been applied, one boy with TT would have been missed. The sensitivity of US was 91% (95% CI: 61.2–99.5%), specificity 87% (95% CI: 81.3–89.0%), and accuracy was 90% (95% CI: 79.9–92.4%). If clinical score and Duplex US were combined the negative exploration rate would have been reduced by 51%; the sensitivity of this combination was 100% (95% CI: 69.5–100%), specificity 57% (95% CI: 53.4–57.0%), and accuracy 61% (95% CI: 55.1–61.5%). The highest reduction (59%) in negative exploration without risking to miss TT was achieved when combining short duration of symptoms with reduced central perfusion; the sensitivity was 100% (95% CI: 69.6.-100%), specificity 66% (95% CI: 62.0–65.6%) and accuracy 69% (95% CI: 62.8–69.2%).

The analysis of the positive predictive factors for TT is shown in Table 1. A fast onset of pain (pain <24 h, odds ratio [OR]: 22.7, 95% CI: 2.8–493), nausea and/or vomiting (OR: 22.5, 95% CI: 2.9–217), abnormal cremasteric reflex (OR: 45.0, 95% CI: 6.1–425) and high position of the testis (OR: 21.2, 95% CI: 3.4–149), as well as focal or segment testicular hyper-/hypoechogenity (OR: 34.7, 95% CI: 5.3–281) and reduced central perfusion (OR: 47.2, 95% CI: 6.4–453), were associated with TT.

Table 1. Variables with a positive and negative association with TT
FeaturesOR (95% CI)PPV, % (95% CI)NPV, % (95% CI)
Pain duration <24 h22.7 (2.8–493)27 (18–29)98 (93–99)
Nausea/vomiting22.5 (2.9–217)67 (25–94)92 (89–93)
Abdominal pain1.3 (0.1–13.1)14 (1–55)89 (88–92)
Erythema1.7 (0.3–11.9)13 (8–15)92 (78–99)
Swelling2.6 (0.3–51.3)13 (9–14)94 (74–99)
High position21.2 (3.4–149)63 (28–89)93 (90–95)
Pathological cremasteric reflex45.0 (6.1–425)75 (38–95)94 (91–95)
Blue dot sign0.0 (0.0–1.4)0 (0–15)85 (85–89)
Fever >38.5 °C0.0 (0.0–19.2)0 (0–68)88 (88–90)
Testicular hyper-/hypoechogenity34.7 (5.3–281)67 (40–84)95 (89–98)
Affected testicle larger1.1 (0.2–4.9)17 (9–23)84 (71–94)
Reduced central perfusion47.1 (6.4–453)73 (44–90)95 (89–98)

Additionally, factors were identified that made TT unlikely, which were a slow onset of symptoms (pain >24 h, negative predictive value [NPV] 98%, 95% CI: 93–99%), absence of swelling (NPV 94%, 95% CI: 74–99%), normal cremasteric reflex (NPV 94%, 95% CI: 91–95%), testicular hyper-/hypoechogenity (NPV 95%, 95% CI: 89–98%) and reduced central perfusion (NPV 95%, 95% CI: 89–98%). TT could have been excluded in all boys who had a combination of long pain duration (>24 h) and no nausea and vomiting or normal cremasteric reflex or normal echogenicity.

In boys with TT, US examination detected testicular hyper-/hypoechogenity more often than in boys without TT (8/11 vs 4/92, P < 0.001). Scrotal hyper-/hypoechogenity was more often found in boys without TT than those with TT (6/92 vs 0/11 patients, P > 0.05), whereas hyper-/hypoechogenity of the epididymis was more frequent in boys with EO (8/9) than AT (12/92) or TT (0/11, P < 0.001). The involved testicle was enlarged in the majority of the boys (65/104, P > 0.05). Doppler US showed reduced central perfusion of the affected testis (TT 8/12 vs non-TT 5/92, P < 0.001). Increased peripheral perfusion on the involved side was found more frequently in boys with EO (8/9) than in those with AT (16/92) or TT (1/11, P < 0.001).

The reported symptoms as well as clinical and US findings of the study population are shown in Table 2. The median (range) age of the study population was 10 (0–16) years. There was no significant difference in mean age between those with or without TT: the mean (sd) age of boys with TT was 9.6 (6.3) years (P > 0.05), with a trimodal peak ageof 0–1 year, 10–11 years and 15–16 years. Right-sided abdominal pain (6.7%) as well as fever (2.9%) was rare in all boys. Trauma was uncommon (n = 2 in the non-TT group). Only one boy had dysuria in the EO group and no boy reported being sexually active. A painful epididymis was mainly present in boys with EO (58.8%) and rarely in those with AT (16.9%) or TT (P < 0.001). In boys with AT (31.0%) a blue dot sign was more frequent than in those EO (n = 1, 5.9%) and no boy had a blue dot sign in the TT group (P = 0.001).

Table 2. Incidence of presenting symptoms and clinical and sonographic findings in boys with and without TT
FeaturesTT group, N = 12 n (%)Non-TT group, N = 92 n (%)P
Mean age, years9.69.7>0.05
Left side8 (66.7)49 (53.3)>0.05
Pain duration <24 h11 (91.7)30 (32.6)<0.001
Nausea/vomiting4 (33.3)2 (2.2)<0.001
Abdominal pain1 (8.3)6 (6.5)>0.05
Trauma0 (0.0)2 (2.2)>0.05
Erythema10 (83.3)69 (75.0)>0.05
Swelling11 (91.7)76 (82.6)>0.05
High position5 (41.7)3 (3.3)<0.001
Pathological cremasteric reflex6 (50.0)2 (2.2)<0.001
Painful epididymis0 (0.0)23 (25.0)0.05
Blue dot sign0 (0.0)23 (25.0)0.05
Fever >38.5 °C0 (0.0)3 (3.3)>0.05
Pathological urine analysis0 (0.0)5 (5.4)>0.05
Testicular hyper-/hypoechogenity8 (72.7)4 (7.1)<0.001
Affected testicle is enlarged7 (63.6)35 (62.5)>0.05
Reduced central perfusion8 (72.7)5 (8.9)<0.001

Urine analysis was normal in all the boys with TT and occasionally pathological in those with EO (11.8%) and AT (2.8%, P = 0.05). In 56.7% of all boys i.v. antibiotics were administered postoperatively (mean 3.3 days after surgery). Boys in the EO group (82.4%) received antibiotic treatment more often than those in the AT (46.5%) and TT groups (66.7%, P < 0.001). Testicular salvage was high (91.7%). Only in one newborn did the testicle not recover, as is common in antenatal torsion of the testis.


The overlooking of TT is a serious problem with major potential consequences, e.g. loss of testis with potential reduced fertility and sperm counts [16]. It is the third most common cause of malpractice lawsuits in adolescent males [17] so many peadiatric centres explore all cases of acute scrotum surgically, but this approach carries the risks of anaesthesia and surgery for many children that could have been treated conservatively.

The results of the present study show that our clinical scoring system would have identified all cases of TT and would have reduced our negative exploration rate by 55%. Unfortunately, US studies alone would have missed one boy with TT. Combining clinical and US findings may improve specificity; for instance the combination of a fast onset of symptoms and reduced central perfusion in colour Doppler US had the highest potential in safely reducing the negative exploration rate in children with TT; however, it could well be that the small number of examinations may have negatively influenced the results of the US.

The four clinical variables chosen for the score had been previously proven to be highly associated with TT [4]. As in most previous, predominantly retrospective, studies there was no single feature that could have excluded TT; however, the combination of two out of three features (fast onset of symptoms, high testicle position, and reduced testicular blood flow) were highly predictive of TT. By contrast, TT becomes unlikely if patients present with symptoms lasting >24 h and show no sonomorphological conspicuousness, e.g. hyper- or hypoechogenity or reduced central Doppler flow. TT may be excluded entirely in patients without a short pain history and nausea and vomiting or those with normal cremasteric reflex or normal echogenicity.

As reported previously, acute scrotal pain of short duration before first presentation is highly suggestive of TT [4, 5, 18]. Ischaemia after torsion results in excruciating pain and prompt presentation at an emergency department. Thus, very high NPVs were found in several studies ranging from 92 to 98% [4, 18]; however, rapid onset can be seen in AT or EO, and TT may start gradually [19].

A lack of the cremasteric reflex is another feature highly associated with TT. It was effective as a predictor in half of the cases of TT in the present study, with a very high OR of 45, a positive predictive value (PPV) of 75% and an NPV value of 94%. These results are similar to those of previous studies, with ORs of 27, PPVs of 43–83% and NPVs of 96–98% [5, 18, 20], but the presence of the cremasteric reflex does not preclude the possibility of torsion, as the reflex was intact in 8% of infants with TT. In previous studies a normal reflex was found in 29–40% of patients TT [5, 18].

The elevated and transverse location of a testis is considered to be important for the diagnosis of TT [4, 5, 18, 20]. High ORs (21–59) and NPVs (93–95%) make it a good tool [5, 18, 20], but massive swellings may make it impossible to judge testicular orientation. Yet again, vertical orientation of the testicle does not exclude TT. As in previous reports, not all patients with TT showed an atypical position of the affected testis (17–58%) [4, 18, 20].

The presence of nausea and vomiting is an important symptom of TT. Reflex stimulation of the coeliac ganglion induces nausea and vomiting in patients with TT, with ORs reported of 9 and 22 [4, 5]. Other features, e.g. fever, abdominal pain, erythema and testicular swelling, are not specific for TT, as mentioned in previous studies [1].

Reliance on various aspects of the history and physical examination alone may not be sufficient. We therefore invesitgated whether the addition of US to the diagnostic evaluation would achieve a higher diagnostic yield, i.e. would achieve a further reduction of surgical exploration with a maintained 100% sensitivity, without negatively affecting the specificity. Concurrently, although the accuracy of imaging is excellent, it is well documented as having a degree of error and inaccuracy [11-13, 17]. In particular, normal central arterial or venous perfusion is considered to be a reliable diagnostic variable for excluding TT [12]. In the present study, however, three boys with TT showed no abnormalities in colour Doppler imaging. False-positive findings showing a central perfusion are probably attributable to the special vascular supply of the testes. Even a necrotic testis after detorsion may demonstrate a certain amount of perfusion, mimicking vital tissue [17].

Volumetry was not helpful in the differential diagnosis of acute scrotum. Considering the large variability of testicular size, volumes ranged from normal to elevated in those with and without TT [13]; therefore colour Doppler US should be performed with reference to sonomorphological variables, including parenchymal echo texture as in the study by Chmelnik et al. [15]. Despite a very high OR, echogenicity would not have identified all patients with TT. When echogenicity and central perfusion status were combined, only one boy with TT was missed.

Even though US is a valuable diagnostic tool, correct diagnosis could not be ensured in every case and the method is highly dependent on the expertise and technique of the investigator. If a mismatch exists between the clinical suspicion of TT and the US findings, the patient should undergo surgical exploration.

In conclusion, based on the results of the present study, we concluded that it is safe to refrain from routine surgical exploration in every boy with acute scrotum. The application of the clinical score tested has shown that in >55% of boys in our study exploration could safely have been omitted without resulting in loss of a testis owing to undetected torsion. US is a very useful tool in experienced hands, but the current sensitivity results from our institution do not allow a reliable diagnosis based on US alone. As US can be quite unpleasant for the child and time-consuming to perform, we propose to refrain from US studies in cases with a positive clinical score, thus shortening the time until surgery and reducing costs. Patients with a negative score are good candidates for US to establish and secure diagnosis. Nevertheless, physicians not experienced in examining children should exercise extra caution as the diagnosis of a high-riding testicle and an abnormal cremasteric reflex can be a demanding task.

Conflict of Interest

None declared.