The diagnostic accuracy of the clinical examination of the hip
MetadataShow full metadata
Hip related pain places a massive economic burden on society and can have a significant effect on the individual sufferer in terms of pain, ability to participate in activity and financial costs. A delay in the accurate identification of the cause of hip pain will lead to a delay in the initiation of appropriate management, prolonging the period of suffering and possibly allowing time for further deterioration of pathology. Currently, the differential diagnosis of hip pain is based on information collected from the patient interview, a physical examination and commonly from findings identified via medical imaging. However, there is lack of good quality evidence to support the use of information from the clinical examination for either identifying or ruling out a specific cause of hip joint pain. Furthermore, there is increasing evidence that pathology identified by medical imaging is not necessarily symptomatic. The aims of this thesis were to determine the diagnostic accuracy of the clinical examination of the hip, and of MRA, for the identification of intra-articular pathology. The thesis first explored the reliability of information gathered from physical tests of the hip. The prevalence of positive tests results and the reliability of patient reports of the reproduction of pain and of ratings of pain intensity were examined in study one. Standardised versions of physical tests were applied in a predetermined random order to both the symptomatic and asymptomatic hips of patients with unilateral hip pain. Tests were repeated one hour and 2-7 days later. The prevalence of positive findings in symptomatic hips ranged from zero to 80%, with resisted tests being the least likely to cause pain and tests that incorporated adduction or internal rotation in flexion being the most provocative. Several of these tests were also provocative in asymptomatic hips, although at a much lower prevalence. The majority of tests demonstrated ‘moderate’ to ‘almost perfect’ within-session (60 minutes apart) and between-session (2-7 days apart) reliability for both pain reproduction and ratings of pain intensity. However, the intensity of pain experienced during testing influenced reliability, with poor reliability being observed with tests that created low intensity pain. These findings indicate that clinicians can be confident that patients will reliably report both the reproduction and intensity of pain, provided the intensity is greater than 2 points on the numeric pain rating scale. Measures of strength (peak force) obtained with a hand held force dynamometer and range of movement (ROM), obtained with a gravity dependent inclinometer, were xiv examined in study two. Data was collected concurrently with that of study one. Within and between-session reliability was determined and strength and ROM values between the symptomatic and asymptomatic hips were compared. Despite the presence of pain and pathology, excellent levels of reliability were observed for both peak force and ROM measures. The percent standard error of measurement for most strength tests was close to 10% and around 3% for ROM. No statistically significant differences were seen in strength between sides. However, a statistically significant reduction in range of movement was observed with the bent knee fall out test. Study three investigated the association between information collected from a clinical examination and a positive response to a fluoroscopy-guided anaesthetic injection (FGAI). Consecutive patients with unilateral hip pain, referred for a magnetic resonance image arthrogram by selected sports or orthopaedic specialists, were recruited. Participants completed standardised questionnaires and were examined by an experienced physiotherapist, both before and immediately after the FGAI. The physiotherapist was blinded to previously obtained clinical information and the radiologist performing the FGAI was blinded to the findings of the clinical examination. A positive anaesthetic response (PAR) was defined as a reduction in pain intensity of at least 80%, calculated by comparing the mean pain intensity score from the three most provocative tests performed prior to the FGAI, to the score from the same three tests being reapplied after this procedure. A PAR was considered to indicate the presence of symptomatic intra-articular pathology. Two variables collected from the history demonstrated sufficient accuracy to indicate that they should be included in the diagnostic examination of the hip. Dominant pain in the groin had a negative likelihood ratio (LR) of 0.18 (95% CI 0.06, 0.58) and sensitivity of 0.91 (95% CI 0.77, 0.97). These values indicate that the absence of groin pain has utility as a screening test for intra-articular pathology of the hip. In contrast, the presence of crepitus had a high specificity [(0.91 (95% CI 0.77, 0.97)] and a moderate positive LR [3.67 (95% CI 1.12 to 11.9)], indicating utility for identifying intra-articular pathology. A number of physical tests demonstrated high sensitivity. However, only one of these, the quadrant test, had a negative LR [0.14 (95% CI 0.02, 1.10)] that indicated that a negative test significantly lowers the probability of a PAR. No physical test demonstrated sufficient specificity or high enough positive LR to indicate that it would be useful for identifying intra-articular pathology as stand-alone test. xv In study four, a clinical prediction rule was derived from the data collected in the previous study. Logistic regression analysis was used to conduct an in-depth, systematic exploration of all possible combinations of two or more key variables using a corrected version of the Akaike information criterion (AICc) to measure model adequacy. The best overall model determined by this analysis included six variables: dominant pain in the groin; age ≥ 39 years; the presence of crepitus; internal ROM <41º; self-reported limited ROM and positive quadrant test. The model demonstrated an overall accuracy of 81%, sensitivity of 91%, specificity of 70%, positive and negative likelihood ratios of 3.1 and 0.12 respectively. Hence, the model has diagnostic utility to both rule in and rule out a PAR. A screening score that appropriately weights the contribution of each test finding was developed to provide a simplified means of interpreting findings for an individual patient clinically. The screening score was employed to assess the accuracy of the commonly recommended practice of ruling a specific condition in or out on the basis of the number of positive or negative tests. This demonstrated that the accuracy of this approach to decision-making is dependent on there being relatively equal weightings in terms of the contribution each test makes to the likelihood of a PAR. Where this is not the case, decisions made on the basis of the number of positive tests may be inaccurate. Study five investigated the diagnostic utility of magnetic resonance imaging arthrogram (MRA) for identifying symptomatic intra-articular pathology of the hip, using pain response to FGAI as the reference standard. Data was collected concurrently with that obtained for study three. All participants underwent a 3 Tesla MRA immediately after the FGAI employed in study three. MR images were reported, following a standardised protocol, by a musculoskeletal radiologist with 30 years experience who was blinded to the results of the FGAI and clinical details of the participant. Despite a high prevalence of structural abnormalities identified by MRA, only the presence of subchondral bone oedema demonstrated a statistically significant association with a positive anaesthetic response. No individual structural abnormality demonstrated sufficient accuracy to indicate that it has diagnostic utility as a stand-alone finding. These observations suggest that the presence of abnormalities identified by MRA should not be considered as evidence that an intra-articular source of hip pain has been established. This thesis provides important new information that contributes significantly to current understandings regarding the diagnostic accuracy of findings from the patient history, physical examination and MRA imaging.