Hip morphology on initial ultrasound predicts hip morphology at one year in developmental dysplasia of the hip

Abstract

Purpose:

This study aimed to determine whether point-of-care ultrasound performed during the Pavlik method treatment of developmental dysplasia of the hip predicts acetabular morphology at 12 months of age.

Methods:

We reviewed the medical records, ultrasounds, and radiographs of patients treated successfully with the Pavlik method between 2017 and 2019. We performed sonographic measurements on point-of-care ultrasound at the initial presentation, the Pavlik discontinuation, and an additional sonographic follow-up. We measured the acetabular index on a plain anteroposterior radiograph of the pelvis obtained at a minimum of 12 months of age. Spearman’s rank correlation coefficient was used to analyze for correlation between sonographic measurements and the acetabular index.

Results:

A total of 72 patients were included in the final analysis. There were no residual or late dysplasia cases at the last radiographic follow-up (mean age = 14.8 ± 2.7 months). Sonographic parameters at the initial presentation significantly correlated with the acetabular index at the final radiographic follow-up (p < 0.001). Patients with worsening degrees of developmental dysplasia of the hip based on stability on sonographic testing (stable, unstable, or dislocated) had higher acetabular indices at the final radiographic follow-up (p < 0.05).

Conclusion:

Point-of-care ultrasound at initial presentation and the Pavlik discontinuation significantly correlate with acetabular morphology at 1–1.5 years of age. At initial presentation, hips that were unstable or dislocated on point-of-care ultrasound had significantly greater acetabular indices than stable hips at the final follow-up.

Level of evidence:

level IV case series.

Introduction

Developmental dysplasia of the hip (DDH) is the most common orthopedic disorder in newborns, and, when diagnosed early, it can typically be successfully treated non-operatively with a Pavlik harness. The consequences of late acetabular dysplasia are significant for a patient who may require operative intervention. Patients with late-detected dysplasia suffer from accelerated joint degeneration and can sometimes develop chronic pain. ¹ This underscores the importance of early detection and treatment.

While it is clear that follow-up is needed to detect patients who may have recurrent hip dysplasia, there is controversy regarding the timing of follow-up, especially in patients who are treated successfully with normalization of their acetabular morphology after the Pavlik harness treatment. Some studies have shown that patients who start with stable hips on ultrasound do not require long-term follow-up to assess deterioration.^2–4 Other studies have suggested that patients with significant risk factors, such as breech presentation, should be evaluated with radiographs despite normal ultrasound imaging. ⁵ One systematic review of 17 studies and 6029 hips found a 4.6% rate of persistent dysplasia at a mean follow-up of 5.29 years (range = 1–20 years), but it was unclear at what time point dysplasia tended to recur. Unnecessary radiographs increase infants’ exposure to radiation and incur extra time and costs. A better understanding of predictors of acetabular development would help identify which patients are at increased risk for late or residual dysplasia and which patients are at lower risk and might not need routine imaging at 12 months, as some authors have suggested recently. ⁶

In addition to the timing of follow-up imaging, there is disagreement over which imaging modality is best. The femoral head ossification center first appears around 4–6 months of age, and radiographs are often first obtained if there is a concern for hip dysplasia. ⁷ Ultrasound, however, is a reliable method of diagnosing and classifying hip dysplasia that does not incur any radiation exposure. ⁸ Ultrasound is a reliable way to measure and document the response to the Pavlik method treatment and can be performed in newborns before ossification of the femoral head. Moreover, the treating orthopedic surgeon can perform an ultrasound at the point of care, further reducing time and cost. Growing advances in point-of-care ultrasound (POCUS) could reduce the need for radiographic follow-up in the first year of life for low-risk patients successfully treated with a Pavlik harness. Further studies are needed to determine whether sonographic parameters on POCUS performed during and after the Pavlik harness treatment predict acetabular development before recommending its widespread use.

This study aimed to determine whether POCUS performed during and after the Pavlik method treatment for DDH can predict the incidence of recurrent dysplasia after successful Pavlik harness treatment of DDH. We hypothesized that sonographic parameters on POCUS, specifically α and β angles and percentage of bony acetabular coverage, would correlate with the acetabular index at subsequent radiographic follow-up.

Materials and methods

Before the commencement of this study, Institutional Review Board approval was obtained. We reviewed a single surgeon series of all patients who underwent successful treatment of DDH with the Pavlik method between September 2017 and April 2019 at an academic medical center. Patients were referred to this surgeon directly from primary care providers due to an abnormal Barlow or Ortolani exam. Patients with risk factors, including breech delivery or abnormal hip exam findings during their newborn evaluation, were referred directly to this surgeon.

Inclusion criteria included a diagnosis of DDH, treatment with the Pavlik method with normalization of sonographic parameters, and minimum radiographic follow-up of 12 months. Exclusion criteria included prior treatment for DDH, lack of follow-up ultrasound examination or radiograph, and lack of normalization of sonographic parameters.

The diagnosis of DDH was made on ultrasound imaging using static and dynamic ultrasound imaging. POCUS was performed by the single treating surgeon during patient visits using a handheld portable ultrasound probe (Butterfly, New York City, NY, USA) in all cases, and images were uploaded to the electronic medical record (Epic, Madison, WI, USA) for later review and measurements. This method has been previously described and found to have high interobserver reliability.^9,10 Sonographic parameters included measures of the α and β angles obtained using a modified version of the minimum dynamic method described by Graf, Harcke, and Clarke. ¹¹ Normal parameters were defined as an α angle >60 or β angle <55. We measured the degree of coverage with the transducer in the coronal view with a perfectly flat ilium, drawing a line across the lateral border of the ilium and measuring the percentage of the femoral head inferior to this line. This was calculated as a percentage, and normal was defined as 50% or greater coverage. The initial stability of the hip was evaluated as stable, unstable, or dislocated. Hips were defined as either being ultrasonographically stable if there was less than 4 mm of displacement of the femoral head from the acetabulum when placed under stress or unstable if the head was located in the acetabulum initially but displaced more than 4 mm under stress testing (akin to Barlow testing). ¹² Based on prior literature, ¹³ hips were defined as dislocated if the femoral head coverage was less than 33% at rest. Ultrasound exams were performed at the initial presentation and at 2- to 4-week intervals during initial treatment until Pavlik discontinuation. Patients were treated in a Pavlik harness until ultrasonographic imaging demonstrated normal hips, meaning 50% bony coverage and an α angle of at least 60°. A post-treatment ultrasound was subsequently performed at an interval between 3 and 6 months after discontinuation to evaluate for residual or late-presenting dysplasia, and we used the same criteria as mentioned above to define normal, specifically an α angle >60°, β angle <55°, and femoral head coverage ≥50%. We measured the acetabular index on a plain anteroposterior radiograph of the pelvis, obtained at a minimum of 12 months of age. The normal values for the acetabular index were defined based on recently published population-based values for average acetabular index based on age, sex, and laterality and defined as being no greater than one standard deviation (SD) above the mean acetabular index for a given demographic (Table 1). ¹⁴ All sonographic and radiographic measurements were performed by the treating pediatric orthopedic fellowship-trained surgeon. Measurements for both normal contralateral hips and the involved abnormal hips were included in the analysis. We based this decision on multiple prior studies that noted late-presenting or subtle dysplasia in the contralateral hip in patients with unilateral hip dysplasia.^15,16 The sonographic measurements at the initial presentation, the Pavlik discontinuation, and the post-treatment follow-up visit were analyzed. This table depicts the mean acetabular index for females and males based on age and laterality. These data were derived from Novais et al. 14 A hip was defined as normal if its acetabular indices were no more than one standard deviation above the reported mean for its age range, patient sex, and laterality.

Table 1.

Normal values of acetabular index.

Age (years)	Females		Males
	Right ± SD	Left ± SD	Right ± SD	Left ± SD
0–0.5	24.04 ± 3.7	25.64 ± 4.0	24.14 ± 1.8	23.43 ± 3.0
0.5–1	24.60 ± 4.2	25.67 ± 3.9	23.41 ± 3.7	23.91 ± 4.0
1–2	23.84 ± 3.4	25.46 ± 4.0	22.95 ± 3.9	23.00 ± 4.0
2–3	21.48 ± 3.8	21.81 ± 3.6	19.82 ± 4.0	19.87 ± 4.1

SD: standard deviation.

This table depicts the mean acetabular index for females and males based on age and laterality. These data were derived from Novais et al. ¹⁴ A hip was defined as normal if its acetabular indices were no more than one standard deviation above the reported mean for its age range, patient sex, and laterality.

Statistical analysis

Descriptive statistics, including mean values and SDs, were calculated for all continuous variables, including age, α, and β angles, and the acetabular index. We determined frequency counts and percentages for categorical variables, including sex and stability, on the initial exam. We tested for the normality of the data for continuous variables using the Shapiro–Wilk test. After data collection, we found that distributions for all continuous variables were non-parametric and, therefore, used Spearman’s correlation coefficient to analyze correlations between continuous variables, such as initial sonographic measurements and the final acetabular index. To qualify the strength of each correlation, the following scale was used for the absolute value of the correlation coefficient (r): weak (r < 0.3), moderate (0.3 ≤ r < 0.5), and strong (0.50 ≤ r ≤ 1.0). We used a Kruskal–Wallis test to compare final acetabular indices between patients based on their initial stability on the sonographic exam (stable, unstable, or dislocated). We performed statistical analyses with SPSS (v24; IBM Corp., Armonk, NY, USA). Significance was set at p < 0.05.

Results

We collected data sets on 73 patients with DDH but subsequently excluded one patient because she had undergone previous Pavlik harness treatment complicated by a femoral nerve palsy before being treated at our institution. Thus, a total of 72 patients were included in the final analysis. Unless otherwise noted, all data are reported as a mean ± SD (range).

Demographics for the cohort are shown in Table 2. All patients were infants with DDH who underwent successful Pavlik harness treatment with normalized clinical and sonographic parameters. Data are reported as a mean ± SD unless otherwise stated. Treatment started at a mean age of 22 ± 16 days (range = 4–81 days), and patients spent a mean of 8.8 ± 1.9 weeks (range = 6.0–15.0 weeks) in the Pavlik harness. Of the 72 patients, there were 16 cases of bilateral DDH and 56 cases of unilateral DDH. As mentioned earlier, fifty-six normal contralateral hips in the patients with unilateral DDH were included in the analysis. Of the 144 included hips, 100 were stable or Barlow negative, 22 were unstable or Barlow positive, and 22 were dislocated. At the first sonographic follow-up, which took place at the time of Pavlik harness discontinuation, and the second sonographic follow-up, no patients had evidence of residual or late-presenting DDH based on the previously defined sonographic parameters. The mean age at the second sonographic follow-up was 4.5 ± 1.7 months (range = 2.5–16.1 months), and the mean time between the Pavlik discontinuation and the second sonographic follow-up was 7.5 ± 7.0 weeks (range = 0.7–60.7 weeks). The sonographic measurements at initial presentation, the Pavlik harness discontinuation, and the sonographic follow-up after the Pavlik discontinuation are shown in Table 3.

Table 2.

Demographic data for the 72 included patients.

Sex	Female 66 (92%)	Male 6 (8%)
Breech	Yes 8 (11%)	No 64 (89%)
Family history	Yes 4 (6%)	No 68 (94%)
Laterality	Unilateral 56 (76%)	Bilateral 16 (24%)

All data are represented as n (%).

Table 3.

Sonographic measurements.

Measurement	Subgroup	Initial presentation	Pavlik discontinuation	Second sonographic follow-up
α angle	Total (n = 144)	52.7 ± 8.8 (35–67)	65.7 ± 3.3 (60–74)	67.2 ± 3.1 (60–74)
	Stable (n = 100)	56.8 ± 6.8 (40–67)	66.5 ± 3.1 (60–74)	67.7 ± 2.9 (61–74)
	Unstable (n = 22)	46.7 ± 4.7 (35–55)	64.2 ± 2.4 (60–68)	65.8 ± 2.7 (60–70)
	Dislocated (n = 22)	40.2 ± 3.1 (35–46)	63.4 ± 3.2 (60–70)	65.9 ± 3.3 (60–72)
β angle	Total	55.2 ± .8 (45–70)	48.1 ± 3.3 (40–56)	47.5 ± 3.7 (40–56)
	Stable	52.4 ± 3.7 (45–62)	66.5 ± 3.1 (40–56)	67.7 ± 2.9 (40–56)
	Unstable	58.5 ± 3.4 (55–65)	50.4 ± 2.2 (45–55)	49.6 ± 3.7 (40–55)
	Dislocated	64.6 ± 3.7 (58–70)	49.8 ± 2.8 (45–55)	49.6 ± 2.8 (45–55)
Acetabular coverage	Total	38.2 ± 16.6 (0–55)	53.6 ± 3.1 (50–60)	54.5 ± 2.9 (45–60)
	Stable	46.2 ± 7.8 (20–55)	54.3 ± 3.0 (50–55)	55.1 ± 2.6 (50–60)
	Unstable	35.5 ± 8.2 (10–45)	51.9 ± 2.3 (50–60)	53.0 ± 3.0 (45–57)
	Dislocated	4.8 ± 5.5 (0–15)	52.0 ± 2.9 (50–60)	53.0 ± 2.7 (50–57)

All data are depicted as a mean value ± standard deviation (range).

At the final radiographic follow-up, which took place at a mean age of 14.8 ± 2.7 months (range = 12–24 months), all hips had an acetabular index within one SD from the mean for their respective ages, and there were no cases of late-presenting or recurring DDH. The mean acetabular indices at final follow-up for stable, unstable, and dislocated hips were 20.8 ± 2.1 (range = 17–26), 21.6 ± 2.1 (range = 18–26), and 24 ± 2.2 (range = 20–27). The Kruskal–Wallis testing suggested that the patients with worsening degrees of DDH based on stability on sonographic testing (stable, unstable, or dislocated) had higher acetabular indices at the final radiographic follow-up (p < 0.05). Post hoc testing showed that hips that were dislocated at the initial presentation had a significantly higher mean acetabular index at final follow-up than unstable and stable hips (24.0 vs 21.6 vs 20.8°, respectively, p < 0.05). There was no significant difference in the acetabular index at the final radiographic follow-up between unstable and stable hips (21.6 vs 20.8°, respectively, p = 0.103).

An analysis of the correlation between sonographic parameters of hip dysplasia and the acetabular index at the final follow-up is shown in Table 4. The acetabular index at last follow-up was moderately inversely correlated with the α angle (ρ = −0.37, p < 0.001) and acetabular coverage (ρ = −0.45, p < 0.001) at the initial presentation. The acetabular index at the final follow-up was moderately correlated with the β angle at the initial presentation (ρ = 0.33, p < 0.001) and the Pavlik discontinuation (ρ = 0.17, p < 0.05). A trend toward correlation did not ultimately reach statistical significance between α angle and acetabular coverage at the Pavlik discontinuation and second sonographic follow-up and acetabular index at the final follow-up.

Table 4.

Analysis of correlation between sonographic parameters and acetabular index at final follow-up.

Sonographic parameter	Initial presentation	Pavlik discontinuation	Second sonographic follow-up
α angle	−0.37 (p < 0.001)*	−0.14 (p = 0.076)	−0.14 (0.098)
β angle	0.35 (p < 0.001)*	0.21 (p = 0.014)*	0.18 (0.034)*
Acetabular coverage	−0.45 (p < 0.001)*	−0.15 (0.081)	−0.12 (0.15)

All data are depicted as a correlation coefficient (p-value). All correlations were analyzed using Spearman’s rank correlation coefficient. Statistical significance was set at p < 0.05.

^*The values denote statistically significant correlations.

Discussion

The most important finding of this study was that hip morphology on POCUS at initial presentation was significantly correlated with the radiographic acetabular index at approximately one year. More specifically, patients whose hips are dislocated at the initial presentation had higher acetabular index at the final follow-up. These findings highlight the importance of considering a patient’s initial sonographic exam on POCUS when evaluating their risk for late or residual dysplasia rather than their most recent POCUS. This finding is essential for clinical practice as it provides evidence to help identify infants at higher risk for residual dysplasia.

While prior studies have noted that ultrasound screening at birth can be used to predict the incidence of later dysplasia,^17,18 the hip ultrasounds in many of these studies were performed by a variety of professionals, including specialized ultrasound facilities, technicians, as well as physicians. This study utilized POCUS performed by a single surgeon, which has several benefits over traditional sonography in a radiology department, including increased convenience for patients and their families by avoiding two separate appointments and improved availability. Regarding the reliability and benefits of POCUS, one recently published study found high interobserver reliability with POCUS across various education levels after a 2-h training course. ¹⁰ The authors’ institution also recently performed a separate study which found that POCUS, compared to traditional sonography in the radiology department, led to significantly reduced visit time (42 vs 92 min, p = 0.002) and costs (US$121.13 vs US$339.38, p = 0.002). ¹⁹ While this study’s findings could likely be extrapolated to traditional sonography, this study’s findings add to the growing body of literature that POCUS, when performed by the treating orthopedic surgeon, is a safe, reliable, and cost-saving tool in evaluating pediatric orthopedic patients, ¹² especially those with DDH. ²⁰

Treatment was started at a mean age of 22 days. Hip ultrasound performed within four weeks of birth has been shown to yield a higher rate of false positives compared to ultrasound between 4 and 8 weeks after birth,^21,22 and the early age at treatment in this cohort could have led to a higher rate false positives, that is, the inclusion of patients whose hips would have normalized without treatment. The evaluation ages reflect the practice patterns at our hospital, where parents are instructed to return between 2 and 6 weeks due to the difficulty of having the first visit four weeks after discharge. This article suggests, however, that despite the incidence of false positives, sonographic parameters on POCUS even before four weeks are predictive of acetabular morphology at one year. Patients were only included if their hip morphology normalized after the Pavlik harness treatment. This did exclude the more severe cases of hip dysplasia that might require further treatment, including bracing or reduction and spica casting. However, this study’s goal was to evaluate the utility of POCUS in predicting hip morphology in the more significant subset of patients successfully treated.

While multiple prior studies had found a correlation between the radiographic acetabular index and sonographic parameters, including acetabular coverage and alpha angle, when both forms of imaging were obtained concurrently,^23,24 they did not examine whether sonographic parameters predict the acetabular index later in development. Similar to this study, another recently published study found that a lower initial alpha angle and positive family history for DDH predicted a higher acetabular index at one year; however, this study only included patients with stable DDH. ²⁵ This study included patients with stable, unstable, and dislocated hips and healthy hips; thus, this study’s findings may apply to a more general population. Given persistent disagreement on the need for universal ultrasound screening for DDH, this study’s findings may lay the groundwork for future studies looking at whether sonographic measurements at birth may be predictive of the future development of hip dysplasia, even when initial measurements are within normal limits. This is important because some patients have normal hips on clinical and ultrasound examinations but develop hip dysplasia later. ⁵ While the etiology for this is poorly understood, identifying risk factors for the late development of DDH could reduce the incidence of complications of untreated DDH, including early-onset hip osteoarthritis.

The α and β angles and acetabular coverage on the sonographic exam at both initial presentation and Pavlik discontinuation were both significantly correlated with the acetabular index at the final radiographic follow-up, which took place at a mean age of 14.8 months—suggesting that the degree of dysplasia at the initial exam and the Pavlik discontinuation is predictive of the acetabular index at 1–1.5 years of age. The sonographic exam at the Pavlik discontinuation can identify patients at risk for residual acetabular dysplasia, requiring close follow-up. The degree of acetabular coverage and α angle at the initial presentation had the strongest correlations with the acetabular index at the final follow-up. It should be considered the most predictive factor for evaluating future dysplasia risk. Patients whose hips were dislocated at the initial presentation had a significantly higher mean acetabular index at the final follow-up. These patients may also benefit from closer follow-up.

The degree of dysplasia on the second sonographic exam, which took place at a mean of 7.6 weeks after Pavlik discontinuation and a mean of 4.5 months of age, did not correlate with the final acetabular index. This finding supports the theory that acetabular morphology continues to normalize even after the Pavlik discontinuation. ²⁶

Over the 10–24 months of follow-up in the present cohort, there were no cases of late-presenting or residual dysplasia in the current cohort. In recent literature, recurrence rates at one year for dysplasia have ranged from 2.6%–34%.^6,27 There are several possible explanations for this wide range of reported rates, including varying definitions of dysplasia and varying techniques for the Pavlik harness treatment and variation in specific patient populations and risk factors. Sarkissian et al. ⁶ published a retrospective case series of 115 infants with DDH. They found a 17% incidence of residual acetabular dysplasia at 6.6 months in infants who had previously normalized on ultrasonographic and clinical parameters at 3.1 months of age and a 33% incidence at 12.5 months in untreated infants. Given this high incidence of residual dysplasia, they concluded that the risks of radiation exposure were outweighed by the importance of identifying these cases of early residual dysplasia. Out of 75 patients who successfully underwent the Pavlik harness treatment, David et al. ²⁷ found two instances of residual dysplasia (2.7%); however, one of their cases involved a patient who started treatment late at four months of age. The lack of residual dysplasia in the present cohort may be because most patients began treatment at an early age, with a mean age at the start of treatment of 3 weeks of age and the latest being 11.6 weeks of age.

This study’s strengths include being single surgeon series drawn from a general and diverse patient population. The normal acetabular index values at various ages were based on recent normative data from a study that included over 1700 patients. This is important as many studies continue to base their definition of dysplasia on the data on the normal acetabular index published by Tönnis in 1975, ²⁸ which is less precise considering the accumulation of data on morphometric measurements of the developing hip.

The most significant limitations of this study include that it was retrospective. Moreover, there were no cases of residual dysplasia in the present cohort. Thus, no definitive conclusions can be drawn about whether sonographic examination can predict residual dysplasia. The mean follow-up in this study was only 14.8 months; longer-term follow-up will be valuable in determining the typical time point at which DDH might recur. The accuracy and reliability of ultrasound in the evaluation of DDH are user-dependent, ²⁹ and, while ultrasound can be taught, ³⁰ the applicability of these results may vary based on who performs sonographic examinations. Two advantages of POCUS of the hip are that it reduces the variability that arises from having multiple examiners and non-physicians performing the evaluation. It also provides the physician and family with the immediate information necessary to make clinical decisions.

Conclusion

Sonographic exams at the initial presentation and the Pavlik discontinuation significantly correlate with acetabular morphology at 1–1.5 years of age. Despite successful Pavlik harness treatment with normalization of sonographic parameters, hips dislocated on POCUS at initial presentation had significantly greater acetabular indices than stable hips at the final follow-up.