Diagnosing Ovarian Torsion
Amidst an increasingly busy winter flu season, you encounter a 25 year old female with acute onset abdominal pain. She has no past medical or surgical history (including no STD’s) and she has never been pregnant. She is monogamous with her heterosexual partner. Her last normal menstrual period ended two-weeks ago. On review of symptoms she denies any associated fever, nausea, vomiting, dysuria, back pain, diarrhea, constipation, or vaginal discharge. Her abdominal exam reveals moderate left lower quadrant and suprapubic discomfort to deep palpation without appreciable masses or peritoneal signs. The remainder of the physical exam is unremarkable and non-diagnostic. While performing her pelvic exam you do not appreciate any vaginal discharge or bleeding or adnexal mass, but she is more tender to palpation of her left adnexa than in the midline or on the right. You are concerned about ovarian torsion. As you thoughtfully contemplate your next diagnostic step in Bayesian fashion your mind wanders (and wonders) to the diagnostic accuracy of history, physical exam, sonography, and computed tomography for ovarian torsion.
Type of Study
“Retrospective Studies”[Mesh] OR “Prospective Studies”[Mesh] OR “Comparative Study “[Publication Type]
“Ovarian Diseases/diagnosis”[Mesh] AND “Torsion Abnormality”[Mesh]
“Gynecological Examination”[Mesh] OR “Physical Examination”[Mesh:NoExp]) OR “Ultrasonography”[Mesh]) OR “ultrasonography “[Subheading]) OR “Tomography, X-Ray Computed”[Mesh]
Sensitivity, specificity for individual tests
(sensitiv*[ Title/Abstract] OR sensitivity and specificity[MeSH Terms] OR “sensitivity and specificity”[All Fields] OR “sensitivity and specificity/standards”[All Fields]) OR “specificity”[All Fields]) OR “screening”[All Fields]) OR “false positive”[All Fields]) OR “false negative”[ All Fields]) OR “accuracy”[All Fields]) OR (“predictive value”[All Fields] OR “predictive value of tests”[All Fields]) OR “predictive value of tests/standards”[All Fields]) OR “predictive values”[All Fields]) OR “predictive values of tests”[All Fields]) OR (“reference value”[All Fields] OR “reference values”[All Fields]) OR”reference values/ standards”[All Fields]) OR (“roc”[All Fields] OR “roc analyses”[All Fields]) OR “roc analysis”[All Fields]) OR “roc and”[All Fields]) OR “roc area”[All Fields]) OR “roc auc”[All Fields]) OR “roc characteristics”[All Fields]) OR “roc curve”[All Fields]) OR “roc curve method”[All Fields]) OR “roc curves”[All Fields]) OR “roc estimated”[ All Fields]) OR “roc evaluation”[All Fields]) OR “likelihood ratio”[All Fields])
Population: Adult female emergency department patients with acute abdominal pain.
Intervention: Diagnostic tests (history, physical exam including pelvic exam, sonography, CT) for ovarian torsion
Outcome: Diagnostic accuracy (sensitivity, specificity, likelihood ratio) for bedside physical exam, EM physician ultrasound, and CT.
Article 1: Ovarian Torsion: A Fifteen-Year Review, Ann Emerg Med 2001; 38: 156-159
Article 2: Ovarian torsion: 10-year perspective, Emergency Medicine Australasia 2005; 17: 231–237
Article 3: Emergency laparoscopy for suspected ovarian torsion: are we too hasty to operate? Fertility and Sterility 2010; 93: 2012-2015
Diagnosing Ovarian Torsion
Article 4: Preoperative Sonographic andClinical Characteristics as Predictors of Ovarian Torsion, J Ultrasound Med 2008; 27:7–13
Diagnosing Ovarian Torsion
Ovarian torsions are reported to represent 3% of gynecological emergencies, but the true incidence is largely unknown since many women are likely never diagnosed. For example, the four case series examined at this Journal Club each required many years’ worth of patient data to be reviewed in order to identify a handful of ovarian torsion cases. Similarly, at Barnes Jewish Hospital (> 90,000 adult ED visits/year) HMED indicates that 6 and 9 cases of ovarian torsion were diagnosed in 2009 and 2010, respectively. The differential diagnosis of female abdominal pain includes appendicitis, ovarian cysts, ectopic pregnancy, renal colic, urinary tract infection, pelvic inflammatory disease, and diverticulitis. Ovarian torsion results from the twisting of the adnexal vascular pedicle and 50-80% of cases are associated with a (usually benign) ovarian tumor. Ovarian torsion is slightly more likely to occur on the right side, possible due to a stabilizing effect of the sigmoid colon on the left.
Because of the ovary’s dual blood supply from both the uterine and ovarian arteries, complete arterial obstruction is rare so attempts at salvage are warranted even if the diagnosis is made later. Cohen has demonstrated preserved ovarian function in 93% of laparoscopically detorsed ovarian torsion cases. Fortunately, Lass has previously demonstrated normal fertility with one ovary should detorsion not be successful. In pediatric (and probably adult) patients the best chance for salvage occurs in those taken to the operating room within 8-hours of diagnosis. Those delayed over 24-hours have near zero salvage rates.
The most important factoid to take away from this data is that all of the studies are case series. In other words, only disease-positive (confirmed ovarian torsion) cases were assessed. Therefore, critical readers can only estimate the sensitivity for the history, physical exam, laboratory, and sonographic findings that are reported in most of these trials. One exception was the PGY III paper demonstrating that Color Doppler sonography (in the hands of GYN or Radiology?) may increase the post-test probability of ovarian torsion if abnormal (however “abnormal” was not defined by the investigators) with a positive likelihood ratio of 5.3, but the absence of an abnormal Doppler does not decrease (negative likelihood ratio 0.61) the probability of ovarian torsion.
The remainder of the studies do not report on a cohort of disease negative patients so 2×2 tables cannot be reconstructed and estimates of specificity, positive likelihood ratio, and negative likelihood ratio cannot be computed for findings such as pelvic pain, palpable adenexal mass, and vaginal bleeding. In other words, we have no idea whether the presence or absence of these findings increase, decrease, or do not change the post-test probability of ovarian torsion. Even more disheartening, Kohn has suggested that diagnostic studies which only evaluate disease-positive patients overestimate sensitivity point-estimate. In fact, in contradiction to this fact, the authors of the PGY IV manuscript grossly misinterpret and overstate the relevance of their findings:
“Therefore, whereas the presence of lower quadrant pain plus 1 or more of those symptoms should raise clinical suspicion for torsion, the absence of these additional symptoms does not preclude the diagnosis”.
For this statement to be true the specificity would need to be acceptably high and the positive LR > 10 (Brown 1999). In reality, after reviewing all of this research (which represents the best evidence on the topic) we have no idea what the specificity or LR’s are because no disease-negative subjects were included. Let’s explore this concept.
Since LR+ = sen / (1 – spec) and LR- = (1- sen) /spec
we can manufacture a range of specificities to test what the LR+ and LR- would become.
Sample LR’s (not actual data) for loss of Doppler venous flow within 1-day of symptom-onset
|Reported Sensitivity||Invented Specificity||LR+||LR-|
So we can see that the specificity would need to be quite high to provide meaningful (i.e. LR+ > 10, LR- < 0.10) likelihood ratios that actually alter post-test probabilities significantly. And this is for the sonographic diagnostic test with the highest sensitivity!
Another limitation of these diagnostic trials is that they only focus upon accuracy. Future trials should use the STARD criteria and assess patient-important outcomes in addition to fully describing diagnostic accuracy (sensitivity and specificity, likelihood ratios) and the reproducibility (kappa) for each element of history, physical exam, lab testing, and sonography. Based upon these results, EP’s should maintain ovarian torsion in the differential diagnosis of women of all ages with abdominal pain (as long as they still have their ovaries) and maintain a low threshold for further imaging (US with Doppler, CT, or MRI) and GYN consult while recognizing that these modalities are also imperfect.The most sensitive sonographic finding is the loss of venous blood flow (85% sensitive when ultrasound performed within one-day of symptom onset) while loss of arterial blood flow has 54% sensitivity within one-day of symptom onset. On the other hand, CT is 34% sensitive for the diagnosis of ovarian torsion (again specificity unknown) and only agrees with ultrasound results 50% of the time. MRI may be considered in pregnant patients, but the diagnostic test characteristics for MRI have not been described.
Diagnostic certainty can only be reached via laparoscopic evaluation by an experienced Gynecologist.