Between July 1996 and September 1997 all consecutive patients referred to the radiology department for painful hips or limping (n = 112) were examined with sonography in a prospective study. The patients were referred from outpatient clinics (orthopedic, pediatric, and emergency departments) and general practitioners. The patients included one male infant, 80 boys, and 31 girls, (age range, 9 months-15 years; mean age, 5 years 11 months). Complaints were unilateral in 108 patients and bilateral in four. In patients with unilateral disease, the contralateral asymptomatic hip was used as a normal reference.

In 11 hips with transient synovitis (nine patients with unilateral disease and one patient with bilateral disease), Doppler sonography was repeated 4-6 weeks later.

Sonography was performed, by the same investigator, on an Ultramark 9 HDI unit (Advanced Technology Laboratories, Bothell, WA) with a high-frequency broadband 7- to 10-MHz linear array transducer or a 128 XP10 unit (Acuson, Mountain View, CA) with a 7-MHz linear array transducer. Patients were examined in the supine position, with their legs extended and slightly externally rotated. Both hips were examined with sonography in a plane parallel to that of the femoral neck because the anterior capsule of the hip is best viewed in that plane.

In patients with unilateral symptoms, a pathologic amount of joint effusion was diagnosed if sonography revealed a thickening of the anterior joint capsule of more than 2 mm compared with that in the asymptomatic hip [8, 19,20,21], in which the thickness of the joint capsule was defined as the distance between the femoral cortex and the fascia of the iliopsoas muscle.

In patients with bilateral symptoms, the thickness of the effusion between the anterior and posterior layers of the capsule was measured [22] and considered abnormal if it exceeded 2 mm.

The posterior layer of the anterior joint capsule was subsequently examined with Doppler sonography to identify the anterior ascending cervical arteries (Fig. 1B). A low-pulse repetition frequency (1.25 kHz) and a wall filter of 50 kHz were used. The color gain was set at a level just below the disappearance of color noise. When no vessels were found with color Doppler sonography, another attempt to identify the vessels was made with power Doppler. Next, a pulsed Doppler examination was performed, placing the Doppler sample over the vessel. The pulse-repetition frequency was then adapted to the amplitude of the waveform. The resistive index (RI), defined as the difference between the peak systolic and end-diastolic flow velocities divided by the peak systolic velocity, was measured at least twice in each vessel.

The Doppler examination was limited to 5 min for each hip. In an attempt to estimate the overall vascularization of the femoral head, the number of vessels that could be identified within the limited time frame was assessed. None of the hips with an effusion was aspirated.

Afebrile patients with an irritable hip had transient synovitis if sonography revealed an abnormal effusion and the symptoms subsided within 4 weeks without specific therapy and remained symptom-free for at least 6 months thereafter.

Perthes' disease was diagnosed with radiography if the typical changes of avascular necrosis appeared on serial radiographs (i.e., condensation, subchondral fracture, collapse, and eventual fragmentation).

Patients had nonspecific synovitis when effusion was present in the hip joint without the classical presentation of transient synovitis [23] and without clinical or serologic signs of rheumatoid arthritis. The discharge diagnosis in this group of patients was reactive or postinfectious synovitis.

The final clinical diagnosis was established 6 months after presentation by reviewing medical and radiologic records.

For statistical analysis, the Statistical Package for the Social Sciences, version 7.5 (SPSS, Chicago, IL) was used. The Kruskal-Wallis test was used to analyze the differences between patient groups. Because the differences between the symptomatic and asymptomatic hips in each patient group were roughly normally distributed, a comparison of both hips was performed using a paired t test; for each patient group, outcomes did not differ significantly, regardless of whether the measurement on the contralateral side was available.

The 97.5 percentile was calculated to determine the upper limit of the RI in asymptomatic hips.

For hyperemia, the Wilcoxon's signed rank test was used to compare both hips in each disease group.

The correlation between the RI and age and between the RI and effusion was determined by calculating Spearman's rank correlation coefficient.

In this study, we studied only the anterior ascending cervical arteries because the joint capsule can be viewed to the best advantage with sonography using the anterior approach. The lateral, medial, and posterior ascending vessels could not be reliably viewed because of unfavorable geometric factors (large distance and overprojection of the greater trochanter) (Figs. 1A and 1B). However, it seems justified to extrapolate the findings in the anterior ascending cervical arteries to the latter vessels because all ascending cervical arteries lie in the same anatomic (intraarticular) compartment, share the same origin (femoral artery), and form a subsynovial anastomotic ring [1].

It was more difficult to obtain Doppler signals in older children than in younger children. This difficulty can be attributed to geometric factors, such as the increasing distance between the transducer and the femoral neck in older children and the progressive reduction of the number of vessels with age [1].

A recent sonographic study on the vascularity of the femoral head [18] examined the intrinsic vessels within the cartilaginous femoral head in neonates, but this method fails in older children because of the progressive ossification of the cartilaginous structures. The method used in this study enables the visualization of the arterial supply to the femoral head after the onset of epiphyseal ossification.

In asymptomatic hips, the duplex Doppler waveform resembles a low-resistance vascular bed with a mean RI of 0.58, which was fairly constant. The 97.5 percentile was 0.7 and considered the upper limit of normal. In this context, it is important to identify the posterior layer of the anterior joint capsule because vessels in the anterior layer (Fig. 1B) characteristically show a high resistive type waveform (Fig. 7) and do not contribute to the blood supply of the femoral head. Especially in hips without effusion, special attention must be paid to correctly identify the posterior layer.

View larger version (141K)

Fig. 7. —3-year-old boy with asymptomatic hip. Duplex Doppler sonogram shows artery in anterior layer of joint capsule. Note high-resistance vascular bed. Artery does not supply blood to femoral head. F = femoral neck.

In this study, the RI in hips with transient synovitis was significantly higher than that of other disease groups. The RI in other disease groups did not show mutually significant differences. The difference between transient synovitis and other diseases can be explained by a relatively large effusion in patients with transient synovitis (Table 1). In fact, it is probable that effusion is the sole cause of the increase of the RI for three reasons. First, we noted a linear correlation between RI and the amount of effusion (r = 0.73, p < 0.001) (Fig. 6), and increasing amounts of effusion cause a progressive decrease of diastolic velocity, resulting in the arrest in diastolic flow (I = 1) and, eventually, a reversal of diastolic flow (I > 1) (Figs. 3, 4, and 5A). Second, RI values returned to normal when the effusion had disappeared in patients with healed transient synovitis; and third, virtually all symptomatic hips without effusion had a normal RI, regardless of underlying disease.

We considered distention of the capsule to represent the volume of effusion because there is a linear correlation between capsular thickening and the volume of effusion. A capsular distention of more than 2 mm is a threshold value beyond which the RI changes. The RI is constant (mean, 0.58) when the amount of effusion is less than 2 mm; however, if the capsule distention exceeds 2 mm, a positive correlation between the width of the effusion and the RI develops. This finding supports the assumption of many studies suggesting that a distention of the anterior joint capsule of more than 2 mm is the threshold between physiologic and abnormal amounts of effusion [8, 19, 22, 27,28,29,30,31].

Although we noted a linear correlation between the amount of effusion and the RI, no increase in RI was identified in several patients, despite a large effusion (Table 2 and Fig. 6). To evaluate this finding, we divided hips with an abnormal effusion (>2 mm) into two groups: group A included patients (n = 26; mean age, 5 years 2 months) with an increased RI (≥0.70), and group B included patients (n = 11; mean age, 6 years 1 month) with a normal RI (<0.70) (Table 2). The composition of diseases within both groups differed significantly: group A consisted of only one hip with Perthes' disease (4%), 24 hips with transient synovitis, and one hip with nonspecific synovitis, whereas group B consisted of four hips with Perthes' disease (36%), six hips with transient synovitis, and one hip with nonspecific synovitis.

Apparently, unlike transient synovitis, a certain amount of effusion does not affect the vascularization of hips with Perthes' disease. An explanation of this finding is that the compliance of the capsule increases because of long-standing effusion (median duration of symptoms was 28 days in Perthes' disease, versus 2 days in transient synovitis), resulting in less pressure per milliliter of effusion [32]. This hypothesis is supported by animal studies that found a decay of intraarticular pressure caused by nonelastic capsular stretching [33]. According to Wingstrand et al. [26], a positive correlation between age and capsular compliance exists, but this does not contribute to the differences between groups A and B because we noted no significant difference in age (p = 0.31).

None of the hip joints with effusion was punctured because none of the patients had clinical signs or symptoms of septic arthritis; therefore, RI values could not be correlated with intraarticular pressure. Although this may be a limitation of our study, we have several reasons supporting our hypothesis that RI values positively correlate with intraarticular pressure. First, RI values show a positive correlation with the amount of effusion. This is analogous to the positive correlation between the amount of effusion and intraarticular pressure, as found by Kallio et al. [25] and Kallio and Ryoppy [32] (r = 0.52). Second, our patients' RI values and intraarticular pressures [26, 34] returned to normal when the effusion disappeared (Fig. 5A,5B). Third, our findings show no correlation between RI and amount of effusion in hips with Perthes' disease. This finding is analogous to the lack of correlation between intracapsular pressure and aspirated volume of fluid in patients with Perthes' disease [35], and it also indicates a correlation between RI and intraarticular pressure.

The mean number of ascending cervical arteries that could be evaluated was one. There were neither statistically significant differences among diseases nor between symptomatic and asymptomatic hips. Therefore, we did not find evidence of hyperemia of the joint capsule in patients with synovitis, a finding that is in accordance with the study of Strouse et al. [36]. These authors made a qualitative evaluation of perfusion by visual comparison of both hips using power Doppler sonography, and found no difference in flow between both hips in children with unilateral transient synovitis. We even found evidence of decreased flow with a mean RI of 0.92. Apparently, power Doppler sonography cannot depict these differences in flow characteristics.

In conclusion, we determined the feasibility of the in vivo evaluation of the vascularization of the femoral head in children. This technique provides a tool for further studies on hemodynamic changes in diseased and healthy hips.