A Meta-analysis of Surgical Versus Nonsurgical Treatment of Primary Patella Dislocation

Meta-analysis

A Meta-analysis of Surgical Versus Nonsurgical Treatment of Primary Patella Dislocation Gherardo Pagliazzi, M.D., Francesca Napoli, M.D., Davide Previtali, M.D., Giuseppe Filardo, M.D., Ph.D., Prof. Stefano Zaffagnini, M.D., and Prof. Christian Candrian, M.D.

Purpose: To compare outcomes after surgery versus nonsurgical treatment in the management of primary lateral patellar dislocation (LPD) through a meta-analysis of randomized controlled trials (RCTs) in terms of redislocation rate and clinical outcome, investigating both short-term (<6 years) functional recovery and overall benefit over time (>6 years). Methods: A systematic search of the literature was performed in November 2018. Risk of bias and quality of evidence were evaluated according to the Cochrane guidelines. RCTs investigating differences between surgery and nonsurgical treatment in primary LPD were included. The outcomes evaluated were redislocation rate, reinterventions, and Kujala score at short-, mid-, and long-term follow-up, with subanalyses for the pediatric population. Results: We included 510 patients from 10 RCTs in the meta-analysis. Redislocation rate was 0.40 (0.25 to 0.66; P < .001) and 0.58 (0.29 to 1.15; P ¼ .12) at the short- and mid-term follow-ups, respectively, and the risk ratio for the need for further operations at 6 to 9 months’ follow-up was 0.14 (0.02 to 1.03; P ¼ .05), all favoring surgery. Concerning the Kujala score, an advantage of the surgical approach of 10.2 points (1.6 to 18.7; P ¼ .02) at short-term follow-up was seen, whereas long-term follow-up results were similar between the groups. The subanalysis of the pediatric population at heterogeneous follow-up confirmed a lower risk of recurrence in surgery, with a risk ratio of 0.60 (0.26 to 1.37; P ¼ .22), although not significant. Conclusion: The literature documents a low number of high-level trials. The meta-analysis of RCTs underlined that the redislocation rate is higher with the nonsurgical approach compared with the surgical one. Moreover, when looking at the clinical outcome, more favorable findings were found with the surgical approach up to 6 years, whereas results seems to be similar at a longer follow-up after either surgical or nonsurgical treatment of primary LPD. Level of evidence: II, meta-analysis of level I and level II randomized clinical trials.

See commentary on page 2482

L

ateral patellar dislocation (LPD), defined as an acute patellar luxation from its normal position in the trochlear groove, is a common event, especially in females during adolescence and young adulthood, accounting for 2% to 3% of all knee injuries. LPD holds an annual risk during the second and third decades of life of 29 and 9 per 100,000, respectively.1,2 LPD involves the rupture of the medial patellofemoral ligament (MPFL), whose deficiency, among other

underlying abnormalities such as trochlear dysplasia, leg malalignment, and patella alta, is linked with LPD recurrence.3-7 MPFL is indeed the most important structure involved in medial patellar restraint, contributing 60% of the total restraining force at 20 flexion by exerting a tensile strength of w200 N.8-11 Several surgical techniques have been proposed for LPD, mainly focused on the treatment of MPFL rupture by either repairing it or reconstructing it with autografts

From the Orthopaedic and Traumatology Unit (G.P., F.N., D.P., G.F., C.C.), Ospedale Regionale di Lugano, EOC, Lugano, Switzerland; 2nd Orthopaedic and Traumatologic Clinic, IRCCS Istituto Ortopedico Rizzoli, (S.Z.), Bologna, Italy; and Applied and Translational Research Center (G.F.), IRCCS Istituto Ortopedico Rizzoli, Bologna, Italy. The authors report the following potential conflicts of interest or sources of funding: G.F. reports grants from Finceramica Faenza SPA, Fidia Farmaceutici SPA, CartiHeal Ltd, EON Medical SRL, IGEA Clinical Biophysics, Biomet, Kensey Nash. S.Z. reports personal fees from Iþ SRL, Fidia Farmaceutici SPA, CartiHeal Ltd, IGEA Clinical Biophysics, Biomet, Kensey Nash; patent from Springer with royalties paid. C.C. reports grants from Medacta

International SA, Johnson & Johnson, Lima Corporate, Zimmer Biomet, Oped AG. Full ICMJE author disclosure forms are available for this article online, as supplementary material. Received December 3, 2018; accepted March 4, 2019. Address correspondence to Davide Previtali, M.D., Ospedale Regionale di Lugano, Via Tesserete 46, 6900 Lugano, Switzerland. E-mail: davide. [email protected] Ó 2019 by the Arthroscopy Association of North America 0749-8063/181444/$36.00 https://doi.org/10.1016/j.arthro.2019.03.047

Arthroscopy: The Journal of Arthroscopic and Related Surgery, Vol 35, No 8 (August), 2019: pp 2469-2481

2469

2470

G. PAGLIAZZI ET AL.

(hamstring,12 quadriceps,13 or patellar tendon14) or allografts (biological15 or synthetic16).17 Good results have been reported for all these techniques, with low recurrence rate and high knee clinical outcome scores up to 5 years, although reconstruction seems to provide better results.18,19 However, high-quality studies are still lacking, and the indication for a surgical approach, especially when dealing with primary dislocations, remains controversial.20 In this light, surgical indication has been mainly reserved for recurrent cases.17 After a primary LPD, the treatment approach may involve immobilization in a plaster cast, a posterior splint, or a brace; partial weightbearing with crutches; and physiotherapy, aimed at vastus medialis strengthening and range of motion (ROM) maintenance.21 Nonetheless, despite the good results reported after conservative treatment, a certain degree of pain, ROM limitation, strength deficiency, and patellar instability (27%) still persists.22,23 Therefore, in the last decade, many authors have advocated the surgical approach as primary LPD treatment after the first dislocating event.24-26 Accordingly, a growing number of randomized controlled trials (RCTs) have compared the clinical outcomes of surgical and nonsurgical approaches in treating primary LPD.25,27,28 Thus, the aim of this study was to compare outcomes after surgery versus nonsurgical treatment in the management of primary LPD through a meta-analysis of RCTs in terms of redislocation rate and clinical outcome, investigating both short-term (<6 years) functional recovery and overall benefit over time (>6 years). We hypothesized that surgical treatment can provide better results, preventing redislocation and improving functional outcome.

Methods Search Strategy and Article Selection A systematic search of the literature was performed on November 27, 2018, on PubMed, Web of Science, Cochrane library, and gray literature (isrctn.org, clinicaltrials.gov, greylit.org, and opengrey.eu) databases with the following string: ((medial patellofemoral ligament OR MPFL OR primary patellar dislocation) AND (conservative treatment OR surgery OR repair OR reconstruction)). The electronic database search was supplemented by manual search of the reference list of the selected articles. Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines were used.29 The following inclusion criteria were used for article selection: RCTs evaluating the differences between surgery and nonsurgical treatment in primary LPD and articles written in English language. Exclusion criteria were: biomechanical, in vitro, or preclinical studies;

review articles and meta-analyses; surgical technique articles; case reports, cohort studies, letters to the editor, and editorials; nonrandomized clinical trials; studies not available in English, and records about recurrent patellar dislocation. At first, articles were screened by title and abstract. The full text of the articles was obtained and evaluated if eligibility could not be assessed from the first screening. Article selection was performed independently by an orthopaedic fellow and an orthopaedic researcher (F.N. and D.P.), with a final summary obtained by consensus. When necessary, discrepancies were solved by an orthopaedic surgeon experienced in the treatment of LPDs (G.P.). Data Extraction and Synthesis and Outcomes Measurement An electronic form for data extraction was created before the study. Information about demography of included patients such as age, sex, and body mass index (BMI) or details about study design such as inclusion and exclusion criteria, number of patients included, surgical and nonsurgical treatments performed, rehabilitation protocol, and follow-up duration were extracted from the selected papers. Furthermore, data about redislocation and subluxation rate, reoperation rate, patient-recorded outcome measures (Kujala clinical score, Tegner activity index, visual analogue scale [VAS], Lysholm score, and knee injury and osteoarthritis outcome score [KOOS]), complications, patient satisfaction, apprehension test, patellar tilt, lateral shift ratio, and thigh hypotrophy were documented. When possible, data were analyzed through a meta-analysis, and different time points were considered to better characterize the results over time and limit the heterogeneity: short-term (<6 years), medium-term (6 to 9 years), and long-term (>9 years) follow-up. Assessment of Risk of Bias and Quality of Evidence The risk of bias was evaluated according to the Cochrane Risk of Bias tool.30 The overall quality of evidence for each outcome was graded as high, moderate, low, and very low, according to Grading of Recommendations Assessment, Development, and Evaluation (GRADE) guidelines.31 The risk of bias and quality of evidence assessment was done independently for all outcomes with pooled data by 2 authors (F.N., D.P.), and discrepancies, if present, were solved by a third author (G.P.). Statistical Analysis The meta-analysis was conducted using RevMen software.32 Treatment effects were assessed by a z test on the pooled risk ratio for binomial outcomes such as redislocation rate (primary outcome) and resurgery rate and on the pooled mean difference for continuous outcomes such as Kujala score, with their

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION

corresponding 95% confidence intervals (CIs). To limit heterogeneity, separate analyses were performed for short-term (<6 years), medium-term (6-9 years), and long-term (>9 years) follow-ups according to a previously published Cochrane review.33 The short-term follow-up analysis represent stabilized results after treatment, whereas the mid- and long-term analyses are representative of the persistence over time of these results. A subanalysis considering only adolescents (<16 years old) was performed. Heterogeneity was assessed using the Cochran’s Q statistic and I2 metric statistical tests, and data were pooled using the fixed effect method when I2 < 25%.34 Otherwise, the random effect method was used, and prediction intervals (PIs) were estimated for mean differences using the formulas of Higgins et al.35 Subanalyses considering only the studies with fewer methodological weaknesses (valid randomization and allocation methods, balanced untreated predisposing factors between groups) and with similar surgical treatments were performed to limit

2471

heterogeneity. The risk of publication bias was assessed using Begg’s funnel plot. A P value of 0.05 was considered significant. Intrarater reliability for the risk of bias assessment was analyzed with Cohen’s kappa. The following cutoffs were considered: <0, no agreement; 0 to 0.20, poor; 0.20 to 0.40, fair; 0.40 to 0.60, moderate; 0.60 to 0.80, good; and >0.80, very good.

Results Article Selection and Characteristics The initial search resulted in 2,093 articles being selected. Of these, 728 articles were found to be duplicates, leaving 1,365 original records. After application of the selection criteria, a pool of 15 papers met the inclusion criteria. Three articles were excluded because they presented the same series of patients, and 2 records were excluded from a noncompleted clinical trial.36-41 Therefore, 10 studies were included in the meta-analysis (Fig 1). All the studies were RCTs with 2

Fig 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart of the study selection process.

2472

G. PAGLIAZZI ET AL.

arms: patients treated with surgery or with nonsurgical treatment. All the studies included were focused on primary LPD in patients with no previous injuries or surgery in the indexed knee or large osteochondral fragments. All but 3 studies included both adults and adolescents. The study of Palmu et al.42 presented a 6and 14-year follow-up subanalysis of patients <16 years old who were part of the series of Nikku et al.40 Thus, this study was included only in the age subgroup analysis. Because of practical and ethical considerations, none of the studies was blinded. Patient Characteristics and Treatments The number of patients included in the selected studies was 510; 492 were examined at final follow-up. The minimum number of patients per study at final follow-up was 30 in Regalado et al.,28 and the highest number was 125 in Nikku et al.40 The mean follow-up was 60 months (median 43, range 24 to 168). The proportion of males and females varied among studies (overall, 51% men, 49% women). The mean age ranged from 13.0 years in the nonsurgical group of Askenberger et al.25 to 27.2 years in the nonsurgical group of Petri et al.43 Only 3 studies reported data about BMI.28,40,44 In all the included studies, there were no differences between surgical and nonsurgical groups at baseline for sex, age, or BMI (Table 1). Regarding predisposing factors, they were exclusion criteria in the study of Ji et al. (trochlear dysplasia type B-D, InsallSalvati index >1.2, >20-mm TT-TG distance)27 and Petri et al. (significant anatomic abnormalities).43 Moreover, in 4 studies there was no difference at baseline between the 2 arms in the prevalence of predisposing factors (low trochlear depth, abnormal patella height, and abnormal TT-TG distance),25,28,44,45 whereas in Nikku et al.40 and Palmu et al.42 only baseline Q-angle and Insall-Salvati index were analyzed, with no significant difference between the 2 arms.43 Bitar et al.37 found a higher incidence of crossing sign in the nonsurgical group, and Camanho et al.46 found at least 1 predisposing factor (trochlear dysplasia, patellar dysplasia, patella alta, or valgus knee) in 16 patients in the nonsurgical group and 10 patients in the surgical group. Predisposing factors were not surgically addressed in the included studies. With regard to surgery, LPDs were addressed with different surgical techniques including MPFL repair, MPFL reconstruction, and Roux-Goldthwait procedure, sometimes combined with lateral release, medial plication, or augmentation with adductor magnus. Nonsurgical treatments and postsurgical rehabilitation protocols were similar for all studies, with periods of immobilization, partial weightbearing, and physiotherapy aimed at vastus medialis strengthening, ROM maintenance, and muscular trophism recovery (Table 1).

Outcomes of Surgery Versus Nonsurgical Treatment The redislocation rate was documented in 7 of the included RCTs at short-term follow-up, 3 at mediumterm follow-up, and only by Palmu et al.42 at longterm follow-up. A lower number of recurrences was documented in the surgical group at short-term followup, with a risk rate of 0.40 (CI 0.25 to 0.66; P < .001; I2 ¼ 0%) (Fig 2A), whereas no significant difference was detected at medium-term follow-up (Fig 2B). Although it was relevant (0.58, CI 0.29 to 1.15; I2 ¼ 45%), the risk ratio reduction at medium-term followup was not statistically significant, likely because it was documented in a low number of studies. The need for further operations, evaluated in only 2 studies at medium-term follow-up, highlighted the advantages of restoring MPFL, with a risk ratio of 0.14 (CI 0.02 to 1.03; P ¼ .05; I2 ¼ 0%) (Fig 3). An analysis of the number of reoperations performed at the shorter or longer follow-ups was not possible owing to lack of data. Complications after surgery were documented in Askenberger et al.,25 Nikku et al.39 and Regalado et al.28 and consisted of 26 ROM limitation (23 in Nikku et al.39), 3 superficial wound infections, 2 neural problems, 2 skin lesions, and 1 arthritis development. Concerning the functional scores, a pooled analysis was possible only for the Kujala score at the short-term follow-up, with an advantage of the surgical approach consisting of a mean difference of 10.2 points (CI 1.6 to 18.7; P ¼ .02; I2 ¼ 92%; PI e12.8 to 33.1) that is both statistically and clinically significant47,48 (Fig 4). Although a meta-analysis was not possible because of the lack of information, data about the Kujala score at the medium-term follow-up showed no significant difference between surgical and nonsurgical treatments in both Nikku et al. (88 vs 90, P ¼ .6)39 and Sillanpää et al. (90 vs 91, P ¼ .8),45 and also at the long-term follow-up in the pediatric cohort of Palmu et al.42 (83 vs 84, P ¼ .7). Regarding the other outcomes, a pooled analysis was not possible because of lack of data. No significant difference between groups was found in the degree of satisfaction of the included patients, when documented.27,28,39 To better investigate the topic and to limit data heterogeneity, a subanalysis of the pediatric population (<16 years old) was performed, even though it was possible only for the redislocation rate outcome and without a stratification for different follow-up time points. Despite the lower risk ratio of 0.60 (CI 0.26 to 1.37) in surgery, no significant difference in terms of recurrence was shown (P ¼ .22; I2 ¼ 62%) (Fig 2C). In the subgroup analyses, considering only studies with limited methodological weaknesses and performing only MPFL repair, the only outcome whose significance was confirmed was the redislocation rate at short-term follow-up, with a risk ratio favoring

Table 1. Characteristics of the Included Studies Demographic* Study Askenberger et al., 201825

Level of Evidence Level I

Patients (Knees) Initial 74

Sex (M;F)

Age (yr)

Final 74

NR

57

Regalado et al., 201628

Level II

36

30

Petri et al., 201343

Level I

24

20 NR

Bitar et al., 201236

Level I

39 (41)

39 (41)

Camanho et al., 200946

Level II

33

33

14;22 NR

13.5  1

13;7 NR

42

NR

72

NS, 27  9; S, 22  6

24

NR

44 (24 to 61)

13;20 13;20 NR

NS, 26.8 (12 NS, 36.3; to 74); S, 40.4 S, 24.6 (15 to 33)

Nonsurgical Treatment Surgical Treatment Arthroscopy þ 4 wk Arthroscopyknee brace assisted MPFL allowing full repair weightbearing physiotherapy

Type of Patients Adolescents (<14 yr)

Arthroscopy þ brace Arthroscopy if loose Adolescents immobilization bodies or and adults for 3 wk, hemarthrosis and physiotherapy, open MPFL repair partial weightbearing

Immobilization and physiotherapy

Immobilization and physiotherapy

LRR for Fulkerson type I (n ¼ 3), Roux-Goldthwait for Fulkerson types II, III, IV (n ¼ 13) MPFL repair and optional lateral release

Adolescents

Adolescents and adults

Brace for 3 wk þ physiotherapy þ cryotherapy þ electrical stimulation, full weightbearing 3 wk later

MPFL reconstruction with patellar tendon

Adolescents and adults

3 wk of splint immobilization physiotherapy

MPFL repair

Adolescents and adults

(continued)

2473

20;21 20;21 NS, 24 (18 to 38); S, 23 (12 to 37)

Inclusion Criteria Age 9, 14 yr, acute primary LPD with hemarthrosis <12 h after injury, no limb disability, no previous injuries Acute primary LPD, no other injuries, no predisposing factors, <3 wk after trauma, þ apprehension test, MRI-confirmed lesion, no contralateral problems Primary acute LPD, no other or previous injuries or surgeries, no osteochondral fragment requiring surgery Primary, unilateral, isolated LPD; no deformities; age >15 and <40 yr; no osteochondral fragment requiring surgery; no pregnancy Acute LPD, <3 wk after injury, age >12 yr, no surgeries or lesions, no other ligament lesions, no large osteochondral fragments, no neuromuscular disease Primary traumatic dislocation, need for reduction maneuver, no fracture or

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION

Ji et al., 201727

Initial Final Initial Final 36;38 36;38 NS, 13.03  NS, 1.14; 13.03  S, 13.19  1.14; S, 1.08 13.19  1.08 56 NR 20;36 NR NR

Trial Duration Months 24

2474

Table 1. Continued Demographic* Study

Level of Evidence

Patients (Knees) Initial

Final

Sex (M;F) Initial

Final

Level I

40

38

Christiansen et al., 200844

Level I

80

77 NR

Palmu et al., 200842

Level II

62

58 16; 46 NR

NIkku et al., 200540

NR

37;3 NR

125 NR

Initial

NS, 20 (19 to 21); S, 20 (19 to 22)

42;35 NR

13  2

43;82 NR

Final

NR

NS, 20 (13 to 39); S, 20 (14 to 30)

13  2

NS, 20  8; S, 20  9

Trial Duration Months

Nonsurgical Inclusion Criteria Treatment Surgical Treatment previous surgery, >25-mo follow-up, no other ligament lesions Arthroscopy þ MPFL repair 84 (60 Primary acute physiotherapy (n ¼ 14), Rouxto 108) traumatic LPD, no Goldthwait previous injuries or (n ¼ 4) underlying deformities, no osteochondral fragments 24 Primary dislocation, age Arthroscopy þ knee Arthroscopy for >13 and <30 yr, no brace for 2 wk fixation or anterior knee pain removal of osteochondral fragment þ MPFL repair 168 Primary acute LPD, age Arthroscopy þ Suture repairs <16 yr, no previous immobilization, (n ¼ 29) þ lateral injuries or surgeries, physiotherapy retinaculum no other ligament release (n ¼ 25), lateral release lesions, no only (n ¼ 7) osteochondral fragments 86 (68 Primary, acute (within Arthroscopy þ Osteochondral to 109) 14 days) LPD, no immobilization, fragment removal surgery or other physiotherapy (n ¼ 19), repairs ligament injuries no (n ¼ 63; 54 LRR), osteochondral MPFL sutures fractures (n ¼ 39), MR plication (n ¼ 18), adductor magnus augmentation (n ¼ 6), LRR only (n ¼ 7)

Type of Patients

Adolescents and adults

Adolescents and adults

Adolescents (<16 yr)

Adolescents and adults

NOTE. Data are presented as n or mean  SD. F, female; M, male; LPD, lateral patellar dislocation; LRR, lateral retinaculum release; MPFL, medial patellofemoral ligament; MR, medial retinaculum; MRI, magnetic resonance imaging; NR, not reported; NS, nonsurgical; S, surgical.

G. PAGLIAZZI ET AL.

Sillanpää et al., 200945

127

Age (yr)

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION

2475

Fig 2. Forest plot of the redislocation rate at short-term follow-up (range 24 to 44 months) (A), at medium-term follow-up (72 to 86 months) (B), and for the adolescence subgroup (24 to 168 months) (C).

surgery of 0.59 (CI 0.34 to 1.00; P ¼ .05; I2 ¼ 0%) and 0.51 (CI 0.31 to 0.85; P ¼ .009; I2 ¼ 0%), respectively. In all the other outcomes, the low number of studies included limited the significance of the results without a decrease in heterogeneity. Risk of Bias and Quality of Evidence A high risk of bias was detected in almost all studies; only the study of Askenberger et al.25 can be considered as having a moderate risk of bias. In particular, blinding was lacking in all the studies, which could affect the subjective outcome measurements. Selective reporting bias could influence the results as well, since only 1 trial was formerly registered (Fig 5). Interrater reliability was good (kappa ¼ 0.71).

According to the Cochrane GRADE guidelines, the aforementioned risk of bias influenced the level of evidence. In fact, for the Kujala score, the lack of blinding was highly suspected to influence the outcome, so the level of evidence was downgraded 2 points. For redislocation rate and reintervention rate, the evidence was downgraded only 1 point, since they are objective outcomes and the influence of lack of blinding is very limited. Regarding inconsistency, the high heterogeneity of the results entailed a further downgrading of the level of evidence for the Kujala score and for the redislocation rate in the pediatric subgroup. Another point was lost by all the outcomes owing to indirectness related to high heterogeneity in terms of surgery offered and population characteristics. The level of

Fig 3. Forest plot of the resurgery rate at medium-term follow-up (72 to 84 months).

2476

G. PAGLIAZZI ET AL.

Fig 4. Forest plot of the Kujala score at short-term follow-up (24 to 44 months). IV, inverse-variance method.

evidence was not downgraded for imprecision, since the difference in the Kujala score at the short-term follow-up was higher than the minimal clinically

important difference of 10 points, and the risk ratios for redislocation and reintervention were <0.75.47,48 No points were lost owing to the risk of publication bias

Fig 5. Risk of bias summary of the included studies. F-U, follow-up.

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION

Fig 6. Funnel plot to assess publication bias for the redislocation rate. RR, risk ratio.

(Fig 6). Thus, in the overall population, the level of evidence was low for redislocation and reintervention rate and very low for the Kujala score, whereas in the subgroup of adolescents, the level of evidence was very low for the redislocation rate.

Discussion The main finding of this meta-analysis was that surgery provided a lower redislocation rate compared with nonsurgical treatment for primary LPD up to 6 years, with better short-term clinical improvement for the surgical approach but, at a longer follow-up, similar clinical results for the two groups. The redislocation rate was considered the main outcome in studies on primary LPD, with all articles reporting this data for both surgical and nonsurgical treatments. The focus on this outcome measure underlines the advantages of a more aggressive approach to restore the anatomy of the MPFL, which has been proven to exert a key role in patellar stabilization.8 In this light, it is not surprising that, in addition, the analysis of the need for further operations supports the benefit of primary surgery, as this allows a better restoration of the conditions favoring patellar stabilization. On the other hand, with the nonsurgical approach, the rates of redislocations and need for further operations in the short-term follow-up have been shown to range from 9.9% to 37.5% and from 14.2% to 25.0%, respectively, which implies that the majority of patients may have a benefit, with no need for further surgeries. However, despite these benefits, the nonsurgical approach is questionable when taking clinical scores into account. In fact, this meta-analysis underlined a significant advantage in terms of symptoms/functional results for up to 6 years of follow-up with the surgical approach. Clinical scores are crucial to quantify differences in parameters that influence quality of life such as pain, perceived instability, or activity limitations. The analysis of the Kujala anterior knee pain score, even with high heterogeneity among the included studies and with

2477

large prediction intervals, supported the surgical treatment at 1 to 6 years of follow-up. This could result from the number of redislocations in the nonsurgical group over time and the time needed to completely recover from injury. There was high heterogeneity in the surgical treatments performed in the included studies, and MPFL repair was the treatment of choice in the majority of these. Even though MPFL repair is considered an effective treatment after primary LPD, MPFL reconstruction is advocated as superior by many surgeons.18 New high-level RCTs comparing the best surgical treatment available (MPFL reconstruction) and conservative treatment could help to confirm the clinically and statistically significant difference in the shortterm Kujala score favoring surgical treatment reported in the present meta-analysis. It was impossible to analyze other clinical scores such as Tegner activity index, KOOS, VAS, and Lysholm score because of the insufficient data reported. A better characterization of the activity level of the included patients could be of interest to clarify whether the differences observed in the Kujala score were correlated with differences in patients’ physical demand. Nonetheless, this metaanalysis highlighted key aspects in this field, which could not have been underlined in previous studies based on a smaller number of patients. Previous efforts aimed at analyzing the literature on surgical treatment after primary LPD presented limitations in terms of either low number or heterogeneity of the included studies, leading to conflicting findings. In 2014, Zheng et al.,49 also analyzing 2 nonrandomized trials50,51 and with no follow-up time subanalysis, concluded that surgical treatment can lead to a lower redislocation rate but also to a worse clinical outcome. In 2015, Yao et al.52 completed a meta-analysis on 6 independent RCTs, showing better results for surgery in the short-term follow-up (<5 years), but opposite results were found in the long-term follow-up studies, favoring nonsurgical treatment. In 2016, Saccomanno et al.53 and Wang et al.54 performed 2 meta-analyses including only 7 independent RCTs and found better results for surgery in terms of redislocation rate and clinical scores, such as Hughson VAS and KOOS, although there was no significant difference in Kujala score. More recently, Longo et al.26 extended the analysis by including studies of all levels of evidence and concluded that surgical treatment is superior to nonsurgical treatment at short-term follow-up, with a slight but nonsignificant difference in the long term. Besides the low number of papers and the low level of studies included, that analysis also presented other methodological limitations, such as the inclusion of papers reporting the same population, which may represent a bias. This explains some differences observed with respect to the findings previously described by Smith et al.,33 who in 2015 completed an

2478

G. PAGLIAZZI ET AL.

intervention review and meta-analysis following the Cochrane guidelines. In their analysis of 6 RCTs, they observed a lower redislocation rate with a higher Kujala score for surgery in the short-term follow-up (5 years) but a better Kujala score for the nonsurgically treated patients in the medium-term follow-up. Although methodologically correct, the low number of included studies hindered the possibility to further evaluate differences in terms of other clinical outcomes or followup times, as well as to perform subgroup analysis based on patient characteristics.33 The present meta-analysis included 10 RCTs, which allowed evaluation of 510 patients treated with either a surgical or nonsurgical approach for primary LPD. Findings about redislocation rate are in line with the ones previously published and actually provide stronger evidence favoring surgery, also in terms of the need for later surgery. Moreover, a subgroup analysis for age was possible due to recent studies including adolescents. Although the effectiveness and the safeness of the surgical approach after recurrent LPD in skeletally immature patients has been shown,55 only 1 systematic review previously focused specifically on primary LPD in young patients: Nwachukwu et al.24 observed lower redislocation rate, higher clinical scores, and better return to sport levels for patients who underwent MPFL surgery. However, they included papers of all levels of evidence and did not perform a meta-analysis. In the present meta-analysis, the redislocation rate could be evaluated in this subgroup of patients: a tendency favoring surgery was confirmed, despite not reaching statistical significance, probably because of the lower number of studies in the subanalysis. Further studies are needed to clearly understand the advantage of primary surgical treatment, especially in young, highly demanding patients with open physis. This meta-analysis also investigated the outcomes with respect to the follow-up time, revealing interesting underlying aspects. The adult population presented a redislocation rate after the surgical approach ranging from 0% to 17% and 0% to 31% at short- and medium-term follow-up, respectively; redislocations rose to 10% to 37% and 29% to 39% at short- and medium-term follow-up with the nonsurgical approach. The adolescent population presented a redislocation rate after the surgical approach ranging from 0% to 22% and 0% to 50% at short- and medium-term follow-up, respectively; redislocations rose to 35% to 43% and 54% to 73% at short- and medium-term follow-up with the nonsurgical approach. Although trends were similar, the higher redislocation rate in adolescents, together with the variability of this outcome, suggests the large heterogeneity of the analyzed LPD patients. Nonetheless, despite some limitations, some key aspects can be underlined. Primary LPD can be effectively

treated with both a surgical and a nonsurgical approach, and surgeons should consider advantages and disadvantages of both when discussing the indication for surgery with patients. Surgery is a more invasive approach, which entails a risk of complications of 26%, with the need of further interventions in 4% of cases.56 However, it significantly reduces the redislocation rate. On the other hand, nonsurgical treatment presents a higher risk of redislocation, an increasing rate of interventions, and lower short-term results, but with similar outcomes at the long-term follow-up. Limitations Patients with primary LPD present a high heterogeneity, which complicates the analysis on the choice between surgical and nonsurgical treatments and its impact on the clinical outcome. Another limitation in the analysis on this field is the heterogeneity of surgical approaches used in the included studies, with only 1 study performing MPFL reconstruction, which is considered by some surgeons the most effective technique in restoring MPFL.18,19,36,57,58 In this light, a subanalysis including only MPFL repair was performed, but the low number of included studies limited the significance of all results except the redislocation rate, whose significance was confirmed. Another important confounding factor was the presence in the majority of the included studies of untreated predisposing factors such as trochlear dysplasia, patellar dysplasia, patella alta, genu valgum, ligament hyperlaxity, lateralized tibial tuberosity, and limb malrotation.59 These play a key role in LPD incidence and recurrence, should be addressed with specific treatments, and, since they were not treated in the included studies, could have influenced the results of the present analysis.7,60-62 In addition, the low number of studies with mid- to longterm follow-ups limits the strength of the subanalyses about these time points. Finally, another limitation is the overall medium to low quality of papers, with moderate to high risk of bias. One of the main problems was the lack of blinding; because of the characteristics of the intervention and the disease, blinding of patients and assessor is very difficult if not impossible. As discussed earlier, the level of evidence of the pooled outcomes was low to very low.

Conclusions The literature documents a low number of high-level trials. The meta-analysis of RCTs underlined that the redislocation rate is higher with the nonsurgical approach compared with the surgical one. Moreover, when looking at the clinical outcome, more favorable findings were found with the surgical approach up to 6 years, whereas results seems to be similar at a longer follow-up after either surgical or nonsurgical treatment of primary LPD.

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION

Acknowledgments The authors thank Giorgio Treglia and Elettra Pignotti for their help in the statistical analysis.

16.

17.

References 1. Stefancin JJ, Parker RD. First-time traumatic patellar dislocation: A systematic review. Clin Orthop Relat Res 2007;455:93-101. 2. Fithian DC, Paxton EW, Stone ML, et al. Epidemiology and natural history of acute patellar dislocation. Am J Sports Med 2004;32:1114-1121. 3. Elias DA, White LM, Fithian DC. Acute lateral patellar dislocation at MR imaging: Injury patterns of medial patellar soft-tissue restraints and osteochondral injuries of the inferomedial patella. Radiology 2002;225:736-743. 4. Jaquith BP, Parikh SN. Predictors of recurrent patellar instability in children and adolescents after first-time dislocation. J Pediatr Orthop 2017;37:484-490. 5. Allen MM, Krych AJ, Johnson NR, Mohan R, Stuart MJ, Dahm DL. Combined tibial tubercle osteotomy and medial patellofemoral ligament reconstruction for recurrent lateral patellar instability in patients with multiple anatomic risk factors. Arthroscopy 2018;34:2420-2426.e2423. 6. Parikh SN, Lykissas MG, Gkiatas I. Predicting risk of recurrent patellar dislocation. Curr Rev Musculoskelet Med 2018;11:253-260. 7. Hevesi M, Heidenreich MJ, Camp CL, et al. The recurrent instability of the patella score: A statistically based model for prediction of long-term recurrence risk after first-time dislocation. Arthroscopy 2019;35:537-543. 8. Desio SM, Burks RT, Bachus KN. Soft tissue restraints to lateral patellar translation in the human knee. Am J Sports Med 1998;26:59-65. 9. Mountney J, Senavongse W, Amis A, Thomas N. Tensile strength of the medial patellofemoral ligament before and after repair or reconstruction. Bone Joint J 2005;87:36-40. 10. Zaffagnini S, Colle F, Sharma B, et al. Medial patellofemoral ligament influence on patellofemoral joint kinematics: An in-vitro analysis. Arthroscopy 2013;10: e116-e117. 11. Woodmass JM, Johnson NR, Cates RA, Krych AJ, Stuart MJ, Dahm DL. Medial Patellofemoral ligament reconstruction reduces radiographic measures of patella alta in adults. Orthop J Sports Med 2018;6: 2325967117751659. 12. Carmont MR, Maffulli N. Medial patellofemoral ligament reconstruction: A new technique. BMC Musculoskelet Disord 2007;8:22. 13. Steensen RN, Dopirak RM, Maurus PB. A simple technique for reconstruction of the medial patellofemoral ligament using a quadriceps tendon graft. Arthroscopy 2005;21:365-370. 14. Camanho GL, Bitar AC, Hernandez AJ, Olivi R. Medial patellofemoral ligament reconstruction: a novel technique using the patellar ligament. Arthroscopy 2007;23:108. e101-108.e104. 15. Hohn E, Pandya NK. Does the utilization of allograft tissue in medial patellofemoral ligament reconstruction in

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

2479

pediatric and adolescent patients restore patellar stability? Clin Orthop Relat Res 2017;475:1563-1569. Berruto M, Ferrua P, Uboldi F, et al. Medial patellofemoral ligament reconstruction with bioactive synthetic ligament is an option. A 3-year follow-up study. Knee Surg Sports Traumatol Arthrosc 2014;22:2419-2425. Mistry JB, Bonner KF, Gwam CU, Thomas M, Etcheson JI, Delanois RE. Management of injuries to the medial patellofemoral ligament: A review. J Knee Surg 2018;31:439-447. Lee D-Y, Park Y-J, Song S-Y, Hwang S-C, Park J-S, Kang D-G. Which technique is better for treating patellar dislocation? A systematic review and meta-analysis. Arthroscopy 2018;34:3082-3093.e3081. McNeilan RJ, Everhart JS, Mescher PK, Abouljoud M, Magnussen RA, Flanigan DC. Graft choice in isolated medial patellofemoral ligament reconstruction: A systematic review with meta-analysis of rates of recurrent instability and patient-reported outcomes for autograft, allograft, and synthetic options. Arthroscopy 2018;34: 1340-1354. Schneider DK, Grawe B, Magnussen RA, et al. Outcomes after isolated medial patellofemoral ligament reconstruction for the treatment of recurrent lateral patellar dislocations: A systematic review and meta-analysis. Am J Sports Med 2016;44:2993-3005. Moiz M, Smith N, Smith TO, Chawla A, Thompson P, Metcalfe A. Clinical outcomes after the nonoperative management of lateral patellar dislocations: A systematic review. Orthop J Sports Med 2018;6: 2325967118766275. Mäenpää H, Lehto M. Patellar dislocation. The long-term results of nonoperative management in 100 patients. Am J Sports Med 1997;25:213. Magnussen RA, Verlage M, Stock E, et al. Primary patellar dislocations without surgical stabilization or recurrence: How well are these patients really doing? Knee Surg Sports Traumatol Arthrosc 2017;25:2352-2356. Nwachukwu BU, So C, Schairer WW, Green DW, Dodwell ER. Surgical versus conservative management of acute patellar dislocation in children and adolescents: A systematic review. Knee Surg Sports Traumatol Arthrosc 2016;24:760-767. Askenberger M, Bengtsson Mostrom E, Ekstrom W, et al. Operative repair of medial patellofemoral ligament injury versus knee brace in children with an acute first-time traumatic patellar dislocation: A randomized controlled trial. Am J Sports Med 2018:363546518770616. Longo UG, Ciuffreda M, Locher J, Berton A, Salvatore G, Denaro V. Treatment of primary acute patellar dislocation: Systematic review and quantitative synthesis of the literature. Clin J Sports Med 2017;27:511-523. Ji G, Wang S, Wang X, Liu J, Niu J, Wang F. Surgical versus nonsurgical treatments of acute primary patellar dislocation with special emphasis on the MPFL injury patterns. J Knee Surg 2017;30:378-384. Regalado G, Lintula H, Kokki H, Kroger H, Vaatainen U, Eskelinen M. Six-year outcome after non-surgical versus surgical treatment of acute primary patellar dislocation in adolescents: a prospective randomized trial. Knee Surg Sports Traumatol Arthrosc 2016;24:6-11.

2480

G. PAGLIAZZI ET AL.

29. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med 2009;6: e1000097. 30. Higgins JP, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. Br Med J 2011;343:d5928. 31. Ryan R, Hill S. How to GRADE the quality of the evidence. Cochrane Consumers and Communication Group. Available at: http://cccrg.cochrane.org/author-resources. Version 3.0 December 2016. 32. Review Manager (RevMan). [computer program]. Version 5. 3. Copenhagen: The Nordic Cochrane Centre: The Cochrane Collaboration, 2014. 33. Smith TO, Donell S, Song F, Hing CB. Surgical versus non-surgical interventions for treating patellar dislocation. Cochrane Database Syst Rev 2015:Cd008106. 34. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. Br Med J 2003;327:557. 35. Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-analysis. J R Stat Soc Ser A Stat Soc 2009;172:137-159. 36. Bitar AC, Demange MK, D’Elia CO, Camanho GL. Traumatic patellar dislocation: Nonoperative treatment compared with MPFL reconstruction using patellar tendon. Am J Sports Med 2012;40:114-122. 37. Bitar AC, D’Elia CO, Demange MK, Viegas AC, Camanho GL. Randomized prospective study on traumatic patellar dislocation: Conservative treatment versus reconstruction of the medial patellofemoral ligament using the patellar tendon, with a minimum of two years of follow-up. Revista Brasileira Ortop 2011;46:675-683. 38. Bitar A, Demange M, D’ELIA C, Camanho G. Reconstruction of the medial patellofemoral ligament was effective for traumatic patellar dislocation. J Bone Joint Surg Am 2012;94:2093. 39. Nikku R, Nietosvaara Y, Kallio PE, Aalto K, Michelsson JE. Operative versus closed treatment of primary dislocation of the patella. Similar 2-year results in 125 randomized patients. Acta Orthop Scand 1997;68:419-423. 40. Nikku R, Nietosvaara Y, Aalto K, Kallio PE. Operative treatment of primary patellar dislocation does not improve medium-term outcome: A 7-year follow-up report and risk analysis of 127 randomized patients. Acta Orthop 2005;76:699-704. 41. Pediatric and Adolescent Patellar Instability. Available at: https://clinicaltrials.gov/show/nct02333825. Published 2015. Accessed November 27, 2018. 42. Palmu S, Kallio PE, Donell ST, Helenius I, Nietosvaara Y. Acute patellar dislocation in children and adolescents: A randomized clinical trial. J Bone Joint Surg Am 2008;90: 463-470. 43. Petri M, Liodakis E, Hofmeister M, et al. Operative vs conservative treatment of traumatic patellar dislocation: results of a prospective randomized controlled clinical trial. Arch Orthop Trauma Surg 2013;133:209-213. 44. Christiansen SE, Jakobsen BW, Lund B, Lind M. Isolated repair of the medial patellofemoral ligament in primary dislocation of the patella: A prospective randomized study. Arthroscopy 2008;24:881-887.

45. Sillanpää PJ, Mattila VM, Maenpaa H, Kiuru M, Visuri T, Pihlajamaki H. Treatment with and without initial stabilizing surgery for primary traumatic patellar dislocation. A prospective randomized study. J Bone Joint Surg Am 2009;91:263-273. 46. Camanho GL, Viegas Ade C, Bitar AC, Demange MK, Hernandez AJ. Conservative versus surgical treatment for repair of the medial patellofemoral ligament in acute dislocations of the patella. Arthroscopy 2009;25:620-625. 47. Bennell K, Bartam S, Crossley K, Green S. Outcome measures in patellofemoral pain syndrome: test retest reliability and inter-relationships. Phys Ther Sport 2000;1:32-41. 48. Crossley KM, Bennell KL, Cowan SM, Green S. Analysis of outcome measures for persons with patellofemoral pain: Which are reliable and valid? Arch Phys Med Rehabil 2004;85:815-822. 49. Zheng X, Kang K, Li T, Lu B, Dong J, Gao S. Surgical versus non-surgical management for primary patellar dislocations: An up-to-date meta-analysis. Eur J Orthop Surg Traumatol 2014;24:1513-1523. 50. Apostolovic M, Vukomanovic B, Slavkovic N, et al. Acute patellar dislocation in adolescents: Operative versus nonoperative treatment. Int Orthop 2011;35:1483-1487. 51. Sillanpää PJ, Mäenpää HM, Mattila VM, Visuri T, Pihlajamäki H. Arthroscopic surgery for primary traumatic patellar dislocation: A prospective, nonrandomized study comparing patients treated with and without acute arthroscopic stabilization with a median 7-year follow-up. Am J Sports Med 2008;36:2301-2309. 52. Yao LW, Zhang C, Liu Y, et al. Comparison operative and conservative management for primary patellar dislocation: An up-to-date meta-analysis. Eur J Orthop Surg Traumatol 2015;25:783-788. 53. Saccomanno MF, Sircana G, Fodale M, Donati F, Milano G. Surgical versus conservative treatment of primary patellar dislocation. A systematic review and metaanalysis. Int Orthop 2016;40:2277-2287. 54. Wang SN, Qin CH, Jiang N, Wang BW, Wang L, Yu B. Is surgical treatment better than conservative treatment for primary patellar dislocations? A meta-analysis of randomized controlled trials. Arch Orthop Trauma Surg 2016;136:371-379. 55. Vavken P, Wimmer MD, Camathias C, Quidde J, Valderrabano V, Pagenstert G. Treating patella instability in skeletally immature patients. Arthroscopy 2013;29: 1410-1422. 56. Shah JN, Howard JS, Flanigan DC, Brophy RH, Carey JL, Lattermann C. A systematic review of complications and failures associated with medial patellofemoral ligament reconstruction for recurrent patellar dislocation. Am J Sports Med 2012;40:1916-1923. 57. Matic GT, Magnussen RA, Kolovich GP, Flanigan DC. Return to activity after medial patellofemoral ligament repair or reconstruction. Arthroscopy 2014;30: 1018-1025. 58. Song J-G, Kang S-B, Oh S-H, et al. Medial soft-tissue realignment versus medial patellofemoral ligament reconstruction for recurrent patellar dislocation: Systematic review. Arthroscopy 2016;32:507-516. 59. Tompkins MA, Arendt EA. Patellar instability factors in isolated medial patellofemoral ligament

META-ANALYSIS ON PRIMARY PATELLA DISLOCATION reconstructionsdwhat does the literature tell us? A systematic review. Am J Sports Med 2015;43:2318-2327. 60. Burnham JM, Howard JS, Hayes CB, Lattermann C. Medial patellofemoral ligament reconstruction with concomitant tibial tubercle transfer: A systematic review of outcomes and complications. Arthroscopy 2016;32: 1185-1195. 61. Balcarek P, Rehn S, Howells NR, et al. Results of medial patellofemoral ligament reconstruction compared with

2481

trochleoplasty plus individual extensor apparatus balancing in patellar instability caused by severe trochlear dysplasia: A systematic review and meta-analysis. Knee Surg Sports Traumatol Arthrosc 2017;25:3869-3877. 62. Longo UG, Berton A, Salvatore G, et al. Medial patellofemoral ligament reconstruction combined with bony procedures for patellar instability: Current indications, outcomes, and complications. Arthroscopy 2016;32: 1421-1427.