Closed Reduction for DDH
Closed Reduction for Developmental Dysplasia of the Hip: Early-term Results From a Prospective, Multicenter Cohort
Wudbhav N. Sankar, MD, Alex L. Gornitzky, BS, Nicholas M.P. Clarke, MD, José A. Herrera-Soto, MD, Simon P. Kelley, MD, Travis Matheney, MD, Kishore Mulpuri, MBBS, MS (Ortho), MHS, Emily K. Schaeffer, PhD, Vidyadhar V. Upasani, MD, Nicole Williams, MD, Charles T. Price, MD, and International Hip Dysplasia Institute.
Background: Closed reduction (CR) is a common treatment for infantile developmental dysplasia of the hip. The purpose of this observational, prospective, multicenter study was to determine the early outcomes following CR.
Methods: Prospectively collected data from an international multicenter study group was analyzed for patients treated from 2010 to 2014. Baseline demographics, clinical exam, radiographic/ultrasonographic data, and history of previous orthotic treatment were assessed. At minimum 1-year follow-up, failure was defined as an IHDI grade 3 or 4 hip and/or need for open reduction. The incidence of avascular necrosis (AVN), residual dysplasia, and need for further surgery was assessed.
Results: A total of 78 patients undergoing CR for 87 hips were evaluated with a median age at initial reduction of 8 months (range, 1 to 20 mo). Of these, 8 hips (9%) were unable to be closed reduced initially. At most recent follow-up (median 22 mo; range, 12 to 36 mo), 72/79 initially successful CRs (91%) remained stable. The likelihood of failure was unaffected by initial clinical reducibility of the hip (P=0.434), age at initial CR (P=0.897), or previous treatment in brace (P=0.222). Excluding those hips that failed initial CR, 18/72 hips (25%) developed AVN, and the risk of osteonecrosis was unaffected by prereduction reducibility of the hip (P=0.586), age at CR (P=0.745), presence of an ossific nucleus (P=0.496), or previous treatment in brace (P=0.662). Mean acetabular index on most recent radiographs was 25 degrees (Å}6 degrees), and was also unaffected by any of the above variables. During the followup period, 8/72 successfully closed reduced hips (11%) underwent acetabular and/or femoral osteotomy for residual dysplasia.
Conclusions: Following an initially successful CR, 9% of hips failed reduction and 25% developed radiographic AVN at early term follow-up. History of femoral head reducibility, previous orthotic bracing, and age at CR did not correlate with success or chances of developing AVN. Further follow-up of this prospective, multicenter cohort will be necessary to establish definitive success and complication rates following CR for infantile developmental dysplasia of the hip.
Level of Evidence: Level II—prospective observational cohort.
Key Words: developmental dysplasia of the hip, DDH, closed reduction, avascular necrosis, AVN, hip dysplasia.
Encompassing a spectrum from mild dysplasia to frank dislocation, developmental dysplasia of the hip (DDH) is one of the most common developmental deformities of the lower extremities and one of the leading causes of future osteoarthritis and hip arthroplasty.1,2 The goal of treatment is to achieve a stable, concentric reduction to facilitate normal femoral head development and continued acetabular growth and remodeling.2-4 For those infants who fail initial nonsurgical management, or those who present late, treatment often involves closed reduction (CR) and spica casting.2,5 Although this procedure is generally successful,2 concerns persist about the risk for iatrogenic avascular necrosis (AVN) of the femoral head and its associated impact on long-term out-comes and function.3,6-8 Other adverse outcomes may include redislocation and the need for further corrective surgery to address residual dysplasia.2,5
Although a variety of demographic and baseline characteristics have been proposed to inﬂuence outcomes, conﬂicting reports have often been described for each. Because of the mixed results following a failure of re-duction, as well as the long-term negative impact of AVN, the role of CR in the management of DDH is increasingly being questioned. There is, however, a relative dearth of prospectively collected data guiding such decision-making. Therefore, the purpose of this multicenter study group was to establish a representative cohort of dislocated hips that may be followed prospectively to ascertain accurate long-term success and complications rates following CR for infantile DDH. For this investigation, the authors sought to determine early-term outcomes following CR, as well as potential risk factors for failure of therapy and the future development of AVN.
The data for this study were queried from a pro-spectively collected database of a multicenter, interna-tional study group [International Hip Dysplasia Institute (IHDI)]. Institutional review board approval was ob-tained by all sites before patient enrollment. At each en-counter, contributing surgeons collected and uploaded all relevant patient information and pertinent clinical details to a central database, including most recent hip radio-graphs. This deidentiﬁed database was then queried be-tween 2009 and December 2014 to identify all patients below 18 months of age at initial diagnosis who were treated with CR for DDH. All hips were frankly dislocated by ultrasound (≤35% femoral head coverage)9,10 and/or radiograph (IHDI grade Z3)11 at the time of CR. Bilateral hips were counted separately. Fourteen hips were excluded for <12 months radiographic follow-up postreduction. As demonstrated in Figure 1, patients treated with open reduction at any time following the index CR were noted and separated from the ﬁnal cohort.
The database was designed to collect patient in-formation and clinical details for each included hip from the time of initial consultation through most recent follow-up. All radiographic measurements, except AVN, were utilized as originally entered by the treating surgeon. The presence or absence of AVN by most recent follow-up was determined by independent, blinded assessment of follow-up radiographs by 3 authors from 3 diﬀerent institutions (W.N.S., K.M., C.T.P.); group consensus was reached for any discrepancies. Each reviewer based their designations on the Salter criteria,12-14 but for the purposes of this manuscript each hip was only classiﬁed with a simple “yes” versus “no” for radiographic evidence of AVN to reduce subtype variability (Figs. 2A–C). Sec-ondary outcomes included residual dysplasia (acetabular index and IHDI grade11), redislocation rate, and the need for further corrective surgery. Failure of CR was defined as either an open reduction at any time following the index reduction and/or an IHDI grade 3 or 4 hip at most recent radiographic follow-up.
Comparative analyses were performed using the Fisher exact/w2 tests, Student t tests/Mann-Whitney U test, or ANOVA. Multivariate regression was used to assess outcomes when controlling for given patient characteristics and/or clinical details. Categorical variables are reported as frequency and percentage; continuous variables are presented with a measure of central tendency (mean or median) and spread (SD or range). All comparative analyses were 2-tailed with a set at 0.05. Sta-tistical analysis was performed using Stata (StataCorp, College Station, TX).
A total of 78 patients (87 hips) across 7 institutions were evaluated. Median age at initial CR was 8 months (range, 1 to 20 mo; Fig. 1). Of these, 8 hips (9%) could not be successfully closed reduced on initial attempt. Hips that were irreducible on clinical and/or ultrasonographic exam were more likely to have an unsuccessful initial CR. We found no other signiﬁcant diﬀerences between those that were and were not able to be successfully closed reduced (Table 1).
Among infants with an initially successful CR (n = 79; 91%), 34% were previously treated in brace (Pavlik harness and/or rigid abduction orthosis) for a median of 3 weeks (range, 1 to 25 wk). An adductor te-notomy was performed in 85% of infants, a psoas release in 14%, and prereduction traction in 13%. When used, prereduction traction was for an average duration of 1 week. For those with preoperative ultrasound (n = 32, 41%), median femoral head coverage was 10% (range, 0% to 40%), and median a angle was 48 degrees (range, 22 to 60 degrees). For those with preoperative radio-graphs (n = 57, 72%), the proportion of infants with a prereduction IHDI grade of 1, 2, 3, or 4 was 0%, 12%, 56%, and 32%, respectively. An ossiﬁc nucleus was present in 56% of hips. Following the initially successful CR, 90% of hips were placed into a hip spica cast (median 12 wk; range, 4 to 30 wk), and 10% were placed directly into a Denis Browne splint (median 26 wk; range, 9 to 53 wk). Of those treated with spica casting, 68% of hips were treated to some extent with a postcasting abduction orthosis (eg, Rhino cruiser, broomsticks, Denis Browne splint, etc.).
At most recent follow-up (median 22 mo; range, 12 to 36 mo), 7/79 initially successful CRs (9%) went on to fail (6 requiring open reduction; 1 with an IHDI grade of 3) (Table 2). The failures occurred at a median of 4 months (range, 1 to 6 mo) following the index reduction. Within our limited numbers, the likelihood of failure was unaﬀected by initial clinical reducibility of the hip (P = 0.434), age at initial CR (P = 0.897), or previous treatment in brace (P = 0.222). There was a non-signiﬁcant trend toward fewer adductor releases (57% vs. 88%, P = 0.065) and higher initial IHDI grades in hips that failed initial reduction.
At most recent follow-up, 67/72 (93%) hips were IHDI grade 1 and 5/72 (7%) were grade 2. According to consensus review of the radiographs, 18/72 (25%) of hips developed radiographic evidence of AVN (Table 3). The risk of osteonecrosis was unaﬀected by prereduction reducibility of the hip (P = 0.586), age at CR (P = 0.745), presence of an ossiﬁc nucleus at the time of reduction (P = 0.496), previous treatment in brace (P= 0.662), or prereduction traction (P = 1.000). Hips that developed AVN had a signiﬁcantly longer duration of postreduction spica casting (median 15 wk vs. 12 wk, P= 0.011). There was a nonsigniﬁcant trend toward increased AVN with higher initial IHDI grade.
At latest follow-up, mean acetabular index was 25 degrees (± 6 degrees), and was unaﬀected by any baseline characteristics or treatment details, including the presence of AVN on ﬁnal radiographs (P= 0.220). The Spearman correlations revealed no statistically signiﬁcant linear association between age at CR and ﬁnal acetabular index (P = 0.553). During the follow-up period, 8/72 success-fully reduced hips (11%) underwent acetabular and/or femoral osteotomy for residual dysplasia (Supplementary Table 1, Supplemental Digital Content 1, BPO Link). Hips treated with further corrective surgery were older at the time of CR (median 14 mo vs. 8 mo, P = 0.003), more likely to have an ossiﬁc nucleus present (100% vs. 52%, P = 0.017), and less likely to have been treated with a prereduction brace (0% vs. 36%, P = 0.049). Hips with AVN were not more likely to un-dergo further corrective surgery (P = 1.000). Increasing severity of initial disease as measured by IHDI grade did not appear to aﬀect the rate of further corrective surgery for residual dysplasia (P= 0.362).
As one of the primary treatment options for infants with DDH, CR typically includes examination under anesthesia, arthrogram with/without an adductor tenot-omy, and postreduction immobilization of the aﬀected hip (eg, spica cast). Although largely successful, CR may also lead to a number of adverse complications, including AVN and loss of reduction. Most outcome data for CRs, however, are based on retrospective series. Speciﬁcally, a systematic review by Novais et al15 evaluating both open and closed management of DDH noted that “the majority of studies included were rated as being of poor methodologic quality” and that “higher quality evidence is needed to better understand” the potential risk factors for future AVN.
In addition to establishing a representative multi-center cohort that can be followed prospectively, the purpose of this investigation was to report early outcomes following CR for DDH. At early-term follow-up (median 22 mo; range, 12 to 36 mo), our cohort had a 91% initial success rate with a 25% incidence of AVN, a 9% in-cidence of redislocation following an initially successful CR, and an 11% incidence of further corrective surgery. Interestingly, the likelihood of either loss of reduction or the development of AVN was unaﬀected by a number of characteristics frequently cited to negatively inﬂuence early outcomes, including femoral head reducibility before CR, previous orthotic bracing, age, or presence of the ossiﬁc nucleus at the time of reduction. Conversely, infants that were older at the time of CR were more likely to need further corrective surgery. Importantly, this study’s multicenter design and primary analysis by in-vestigators not aﬃliated with the treating institutions limited the publication bias inherent to any single-center, self-authored cohort.
Osteonecrosis is one of the most concerning com-plications following CR. In recent studies (published since 2000), the incidence of AVN following CR varies widely between 2% and 36%,4,5,13,14,16-27 with Cooper et al2 noting that this disparity is “largely caused by diﬀerences in the deﬁnitions of AVN and the timing of follow-up.” Comparing closed versus open treatment, Novais et al15 reported in their recent meta-analysis an incidence of signiﬁcant AVN (grade Z2) of 8% following CR versus 19% following OR, with Clarke et al19 reporting a similar pattern of 7% and 14%, respectively. Here, our reported incidence of AVN was 25% at a median follow-up of approximately 2 years. Although some have suggested that prolonged follow-up for 10 to 12 years is necessary to ascertain the true incidence of AVN,3,6,7 others argue that follow-up beyond 2 years will not identify additional cases.14,28-30 A variety of classiﬁcation criteria have also been used, and many authors diﬀerentially include low-grade changes.2 This variation in AVN deﬁnitions and lengths of follow-up highlight the importance of establishing a prospective cohort that can be followed longitudinally to determine accurate data.
Multiple potential risk factors for AVN have been reported, with the presence of the ossiﬁc nucleus at the time of reduction remaining one of the most con-troversial. Although some authors argue that the ossiﬁc nucleus is protective against AVN,17,19,27,31 others note no such eﬀect.18,23,24,26,29 Relatedly, whereas some have linked older age at reduction to a higher incidence of AVN but a lower rate of severe disease,3,7,18,32 others have noted that age at reduction has no eﬀect on AVN risk.4,6,15,24,29,31 In this cohort, neither age at reduction nor the presence of an ossiﬁc nucleus was associated with the presence of AVN. Consistent with prior studies, sex, bilateral disease, family history, previous treatment with Pavlik harness/abduction orthosis, and the performance of an adductor tenotomy also did not impact disease frequency.3,6,13,17,21,29,31 Although a trend was appre-ciated, the limited number of patients with prereduction radiographs may have constrained our ﬁndings regarding the severity of initial dislocation (as measured by IHDI grade), which prior reports suggest may inﬂuence AVN risk.3,18,32 We did note an increased length of spica cast immobilization following CR in those who developed AVN versus those who did not. Whereas this diﬀerence was statistically signiﬁcant, its clinical signiﬁcance is un-clear. It is possible that longer periods of casting were an indirect reﬂection of greater disease severity, or that there is an intrinsic ill eﬀect on perfusion from prolonged immobilization of an infantile hip.
Achieving, and maintaining, a concentric reduction of the hip is one of the primary drivers of future acetabular growth and development.4,33,34 Although potential overlap is likely, rates of failed initial reduction across the literature vary between 6% and 25%,2,13,16,20,25,35,36 and rates of loss of reduction from 0% to 18%.5,13,16,20,23–25,35–37 In this cohort, 91% of hips were successfully reduced on the initial attempt, with 9% going on to fail CR at latest follow-up. Importantly, however, rates of failed initial reduction are likely to vary consid-erably based upon the treating surgeon’s threshold for proceeding directly to open reduction. Nevertheless, our results fall within the lower half of the previously reported ranges, suggesting reasonable success may be achieved following CR.
Several factors have been proposed to increase risk of residual dysplasia following CR. In this study, we chose to measure residual dysplasia in 2 ways: objective acetabular measurements (acetabular index and IHDI grade) and the incidence of further corrective surgery. Regarding the former, no patient-speciﬁc factors in this cohort were predictive of worsened acetabular index, including pre-reduction IHDI grade, age at reduction, or ﬁnal AVN status. Regarding the latter, 11% of hips in this study were treated with a subsequent acetabuloplasty and/or femoral osteotomy, although that rate will likely rise with con-tinued follow-up. In comparison, rates of further corrective surgery across the literature range as high as 57% to 100%,19,22,38 with the majority of published studies identifying a frequency ≤35%.5,16,21,23,25,36,37,39 In contrast to the acetabular index, we did observe that increased age at the time of CR predicted a higher rate of further corrective surgery, which is consistent with Luhmann et al’s37 observation that delaying CR may increase the need for later intervention.
This study has a number of limitations. First, longer follow-up may lead to diﬀerent values for AVN, acetabular dysplasia, and further corrective surgery. The purpose of this study was to report early results from a prospective cohort of infants undergoing CR. Certainly, we aim to follow this cohort longitudinally to provide long-term success and complication rates. Second, AVN is diﬃcult to unilaterally deﬁne, as there are a variety of diﬀerent classiﬁcation systems in use across the literature. Given the potential for these variations to impact the reported frequency of AVN within our database, we chose to retrospectively assess AVN using a blinded panel of surgeons each applying the Salter criteria in a yes/no manner, a step the authors felt was imperative to improve the accuracy and consistency of our ﬁndings. Third, due to the large number of surgeons contributing to the da-tabase, there is considerable variation in treatment prac-tices. Regardless, the authors view this heterogeneity as a strength in that it more accurately reﬂects real-world practice. Fourth, given the smaller number of hips with certain conditions and/or outcomes, the P-values in our tables should be interpreted with caution. Fifth, we did not measure abduction angle in postreduction immobilization, which has been suggested to inﬂuence AVN rates,26-40 as practice variation in postoperative imaging and treatment limited the availability of such data. Finally, the quality of our results is dependent on the completeness and accuracy of data reporting by the individual centers, which is a limitation of any multicenter database study.
In conclusion, this prospective, multicenter cohort of infants undergoing CR for DDH demonstrated a 9% failure rate after initially successful reduction and a 25% rate of AVN at early-term follow-up. In this study, older age at the time of CR and longer immobilization in a spica cast were predictive of needing further corrective surgery for residual dysplasia and developing osteonecrosis, re-spectively. No other patient-speciﬁc factors related to history, treatment, or outcome were associated with the future development of either AVN, residual dysplasia, or failed reduction. Further follow-up of this cohort will be necessary to establish accurate long-term success and complication rates following CR for infantile DDH.