Interrater Reliability and Diagnostic Accuracy of Pelvic Girdle Pain Classification

Cook, Chad; Massa, Lisa; Harm-Ernandes, Ingrid; Segneri, Rachel; Adcock, Jennifer; Kennedy, Colleen; Figuers, Carol

doi:10.1016/j.jmpt.2007.03.008

« Previous Next »Journal of Manipulative and Physiological Therapeutics
Volume 30, Issue 4 , Pages 252-258, May 2007

Interrater Reliability and Diagnostic Accuracy of Pelvic Girdle Pain Classification

Chad Cook, PT, PhD, MBA
- Assistant Professor, Department of Community and Family Medicine, Duke University, Durham, NC
- Submit requests for reprints to: Chad Cook, PT, PhD, MBA, Assistant Professor, DUMC 3907, Duke University, Durham, NC 27278.
Lisa Massa, MPT
- Physical Therapist, Physical and Occupational Therapy, Duke University, Durham, NC
Ingrid Harm-Ernandes, PT
- Physical Therapist, Physical and Occupational Therapy, Duke University, Durham, NC
Rachel Segneri, SPT
- Graduate Student, Division of Physical Therapy, Department of Community and Family Medicine, Duke University, Durham, NC
Jennifer Adcock, SPT
- Graduate Student, Division of Physical Therapy, Department of Community and Family Medicine, Duke University, Durham, NC
Colleen Kennedy, DPT
- Physical Therapist, Physical and Occupational Therapy, Duke University, Durham, NC
Carol Figuers, PT, EdD
- Associate Professor, Department of Community and Family Medicine, Duke University, Durham, NC

Received 7 November 2006; received in revised form 1 January 2007; accepted 9 January 2007.

Abstract

Objective

The purpose of this study was to measure the reliability of a classification system for pelvic girdle pain (PGP) and diagnostic accuracy of selected examination and clinical special findings for diagnosis of PGP.

Methods

The design involved a prospective epidemiological study of pregnancy-related PGP. Consecutive subjects were recruited and classified using criteria defined by previous studies. Two clinicians examined the subjects and classified each patient into 1 of 5 classification groups. Clinical examination and clinical special tests were performed on the patients with PGP. Statistical analysis involved interobserver agreement using a κ statistic and sensitivity and specificity values for the examination and clinical special testing.

Results

Twenty-one subjects were included in the analyses. Aggregated percentage of agreement for the classification system was 84.6%. The Cohen κ was 0.78 (CI, 0.64-0.92; P < .0001), which indicated substantial agreement during selection of the classification system. Most clinical examination and clinical special-test findings demonstrated low sensitivity and high specificity, whereas clusters of findings including the lunge, manual muscle testing of the hip (lower extremities), and passive range of motion of the hip demonstrated the strongest diagnostic value.

Conclusion

Selected tests and measures are moderately discriminatory in diagnosing PGP. A classification system for diagnosing PGP demonstrates strong agreement and may be useful for clinicians.