| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters and Responses |
We would like to comment on the article by Levangie entitled "Four Clinical Tests of Sacroiliac Joint Dysfunction: The Association of Test Results With Innominate Torsion Among Patients With and Without Low Back Pain" that appeared in the November 1999 issue of Physical Therapy.
We enjoyed the article and are happy to see that papers describing epidemiological data are being published by physical therapists. Levangie's conclusion that the "data do not support the value of these tests in identifying innominate torsion" appears to be correct, based on her calculations of the odds ratios. However, an alternative conclusion could also be obtained from her published performance characteristics (eg, sensitivity, specificity). In Table 4, Levangie reports the specificity of the Gillet test as 93% and of the sitting flexion test also as 93%. A high test result (90% or higher) effectively rules in the target disorder or conditions.1 Therefore, according to the data, the Gillet test and the sitting flexion test are effective tests to detect (rule in) innominate torsion. Thus, we believe that Levangie may have inadvertently left out this important interpretation of her data.
Jefferson County Rehab & Sports Clinic
430 S Truman Blvd
Crystal City, MO 63019
DPT Student
Creighton University
2500 California Plaza
Omaha, NE 68178
References
The literal interpretation of the values for specificity that I found for the Gillet and sitting flexion tests would be that 93% of the subjects determined not to have static pelvic asymmetry2 (4 mm) had a negative Gillet test or sitting flexion test. A highly specific test succeeds in identifying most individuals who do not have the target disorder. This may, but does not necessarily, occur by overdiagnosing absence of the disorder. A specific test may have numerous false negatives but should have relatively few false positives. The relatively few number of false positives and resulting relative utility of the positive test leads to Cibulka and Aslin's conclusion that the tests may be effective for ruling in the target condition or disorder.
Although Cibulka, Aslin, and I agree on the meaning of a test with high specificity, it can be misleading to interpret one test performance characteristic in isolation. Sensitivity and specificity do not predict the actual number of individuals who will be correctly categorized, because this is affected by the prevalence of the disease (pelvic asymmetry) in the target population. One must also consider the positive predictive value. The positive predictive value (PV+) answers the more clinically relevant question: "Given this [positive] test result, what is the likelihood that my patient has the disease [static pelvic asymmetry]?"2 In my study data, the PV+ values indicated that a positive Gillet test or a positive sitting flexion test was correct only 67% and 78% of the time, respectively.
The PV+ values in my study indicate a substantially weaker association of a positive test outcome with pelvic asymmetry than one might expect from examination of the specificity alone. The PV+ values are deflated compared with the specificity values because a large number of subjects with static pelvic asymmetry did not have a positive Gillet test or a positive sitting flexion test (false negative test results were obtained for 65% and 72% of the subjects, respectively). This discrepancy highlights the need to consider the prevalence of the disease (pelvic asymmetry) in the sample on whom the data are generated. In my study, both those with and without low back pain (LBP) were examined. We typically apply tests of sacroiliac joint dysfunction only to patients with LBP—a subset that may show a different prevalence. A separate calculation of test performance characteristics for subjects with LBP showed very consistent results. In fact, the prevalence of pelvic asymmetry was similar for the 2 groups.3 It is also useful to note that negative predictive values for the Gillet and sitting flexion tests (the likelihood that the patient does not have the condition, given that you obtain a negative test result) were 35% and 28%, respectively. The net effect of the findings on sensitivity, specificity, and predictive values reinforces the finding from the odds ratios that test outcome is not associated with static pelvic asymmetry in a meaningful way.
I appreciate that Cibulka and Aslin took the time to independently evaluate my findings and to initiate this dialogue. I would also like to thank Physical Therapy for providing a platform for this discussion.
Associate Professor
Physical Therapy Program
Sacred Heart University
Fairfield, CT 06432
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
