Abstract
Background:
Although risk factors for redisplacement after nonoperative treatment of pediatric supracondylar humeral fractures (SCHFs) are well described, the temporal distribution of this risk remains poorly understood. This study investigated temporal patterns of redisplacement and identified independent predictors using time-to-event analysis.
Methods:
This retrospective cohort study included 218 children (aged 3–14 years) with Gartland type II (n = 142) and selectively treated type III (n = 76) SCHFs managed with closed reduction and custom-molded triplanar splinting between September 2020 and August 2023. Type III fractures were included only when fluoroscopic stress testing confirmed post-reduction stability. Redisplacement was assessed radiographically and analyzed using Kaplan–Meier and Cox regression methods.
Results:
Redisplacement occurred in 39 patients (17.9%). Cumulative incidence demonstrated a nonlinear temporal pattern, with most events occurring early after reduction: 87% within the first 14 days, with apparent peaks at days 3–4 and 7–14. In multivariable analysis, initial lateral displacement percentage >85% (hazard ratio [HR] 3.52, 95% confidence interval (CI): 1.82–6.83) and severe soft-tissue swelling (HR 3.08, 95% CI: 1.61–5.89) were independently associated with redisplacement, whereas Gartland classification was not.
Conclusions:
Redisplacement risk after nonoperative management of pediatric SCHFs appears concentrated in early post-reduction periods. Displacement magnitude and severity of soft-tissue swelling may provide more clinically relevant prognostic information than fracture classification alone. The observed temporal pattern may be influenced by discrete follow-up intervals and requires prospective validation.
Keywords
Introduction
Supracondylar humeral fractures (SCHFs) are the most common elbow injuries in children, accounting for approximately 3% of all pediatric fractures. 1 For displaced fractures, closed reduction and percutaneous pinning represent the gold standard, particularly for Gartland type III injuries.2 –5 However, nonoperative treatment remains a therapeutic option for most type II fractures, and type III fractures meeting specific stability criteria are selected.6 –9 It should be noted that treatment practices vary considerably worldwide—surgical fixation is strongly preferred for type III fractures in most North American and European centers, and conservative management of these injuries is not considered standard care in such settings.10,11
For patients undergoing nonoperative treatment, maintaining anatomic alignment throughout healing is critical, as redisplacement may necessitate delayed intervention. Surveillance monitoring, therefore, determines treatment success. Substantial research has identified risk factors for redisplacement, including Gartland classification, rotational instability, displacement degree, and reduction quality.12,13 However, these studies predominantly treat redisplacement as a binary outcome, overlooking a clinically important question: when does risk peak? Since fracture healing is a dynamic biomechanical process, redisplacement risk likely fluctuates nonlinearly over time.
This temporal knowledge gap has practical consequences. For low-risk patients, uniform frequent follow-up may impose unnecessary burden and radiation exposure.14,15 For high-risk patients, fixed-interval surveillance may miss redisplacement between visits, leading to delayed intervention. Understanding the temporal dynamics of risk could inform more rational, individualized surveillance strategies.
This study employed time-to-event analysis to explore temporal patterns of redisplacement risk following standardized nonoperative treatment. We aimed to characterize when redisplacement occurs and identify independent predictors, providing hypothesis-generating data for the future development of risk-stratified surveillance protocols.
Methods
Study design and ethical approval
This single-center retrospective cohort study received institutional ethics committee approval, with a waiver of informed consent granted owing to the retrospective design and complete anonymization of all the data. Reporting adhered strictly to the STROBE statement.
Study population
We consecutively enrolled all children aged 3–14 years with closed SCHFs who received nonsurgical treatment at our institution’s orthopedic department from September 1, 2020, to August 31, 2023.
The inclusion criteria were as follows: (1) aged 3–14 years; (2) had Gartland type II or III fractures; (3) satisfactory alignment confirmed radiographically following standardized closed reduction; for Gartland type III fractures, inclusion required demonstration of sufficient inherent stability via stress testing under C-arm fluoroscopy, defined as maintained alignment during gentle passive flexion-extension (30°–120°) and pronation-supination maneuvers; and (4) complete follow-up radiographic documentation.
The exclusion criteria were as follows: (1) open, pathological, or chronic fractures; (2) concomitant progressive neurovascular compromise requiring urgent surgical intervention; (3) unsatisfactory initial reduction or instability on stress testing necessitating surgical conversion (n = 42 type III fractures during the study period); and (4) incomplete clinical or radiographic records.
Standardized treatment protocol
All enrolled children received a unified, standardized nonoperative treatment protocol consisting of closed reduction and external immobilization. All reduction procedures were either performed or directly supervised by attending or associate chief physicians with >10 years of pediatric orthopedic trauma experience, ensuring high technical consistency.
Closed reduction technique
Under conscious sedation, standardized closed reduction was executed using the following biomechanical principles: (1) sustained longitudinal traction with the elbow in slight flexion to correct overlap and release soft tissues; (2) correction of coronal (varus/valgus) and sagittal (flexion/extension) angulation and rotation; and (3) precise reduction via three-point mechanics (lift–squeeze–push), with reduction quality confirmed via C-arm fluoroscopy to meet acceptable standards.
Immobilization technique
Following successful reduction, custom-molded triplanar splinting (posterior long-arm/sugar-tong splint) was immediately applied. 16 The splint, which was fabricated from cedar bark material and offers excellent malleability, moldability, and breathability, was hand-contoured by the operating physician according to each child’s specific elbow anatomy into a U-shaped structure covering the posterior, medial, and lateral aspects. The elbow was immobilized at approximately 120° flexion with the forearm in neutral rotation and secured with multiple cotton layers and elastic bandage compression.
Neurovascular monitoring and post-reduction protocol
Following splint application, all patients underwent serial neurovascular assessments, including radial pulse palpation, capillary refill, and sensory-motor examination. Patients with severe soft-tissue swelling were observed for a minimum of 4–6 h with hourly neurovascular monitoring prior to discharge. In cases of pulse diminution, elbow flexion was reduced to 90°–100° until vascular status normalized. Caregivers received standardized written and verbal instructions on warning signs that necessitate immediate return. Anteroposterior and lateral radiographs were routinely obtained immediately after splint application to confirm maintained reduction. All 218 enrolled patients achieved satisfactory alignment and were discharged without same-day surgical conversion. No cases of compartment syndrome occurred in this cohort.
Data collection and variable definitions
Data were extracted from the hospital electronic medical records system and Picture Archiving and Communication System. The baseline variables included age, sex, injured side, and Gartland classification.
The key predictor variables were defined as follows:

Schematic illustration of the measurement of the initial LDP on an anteroposterior radiograph.
Primary endpoint definition
The primary endpoint was defined as the first radiographic redisplacement documented during follow-up from post-reduction to fracture union. Redisplacement required meeting at least one of the following objective criteria: (a) absolute change in the Baumann angle from immediate post-reduction imaging >10°17,18; (b) the anterior humeral line failing to intersect the capitellum or only touching its anterior third 19 ; (c) new or increased coronal plane angulation >10°; and (d) absolute LDP >50% or increase ≥15% from the immediate value. These thresholds were established by synthesizing the literature regarding radiographic measurement error ranges and clinically significant displacement magnitudes. 20
Statistical analysis
All analyses were conducted via R software (version 4.2.2). Given that E = 39 events and k = 4 covariates, the events-per-variable ratio ≈9.8 approached the accepted threshold; therefore, ridge-penalized Cox regression was performed as a sensitivity analysis alongside conventional Cox modeling. Continuous variables are presented as medians and interquartile ranges (IQRs); categorical variables are presented as frequencies and percentages. The Kaplan–Meier method was used to plot survival curves describing cumulative temporal risk evolution, with log-rank tests comparing curves across subgroups. To identify independent redisplacement risk factors, univariable and multivariable Cox proportional hazards models were constructed, and hazard ratios (HRs) with 95% CIs were calculated. The proportional hazards assumption was validated using Schoenfeld residuals testing, confirming assumption validity; ridge-penalized Cox models yielded HRs consistent in direction and significance with the primary models, supporting the robustness of the results. All statistical tests were two-sided, with p < 0.05 considered statistically significant.
Results
During the study period, 374 children presenting with SCHFs underwent initial evaluation. Following screening, 218 children meeting all inclusion criteria were enrolled (Figure 2). The cohort had a median age of 5.8 years (IQR 4.5–7.5), consistent with the known epidemiological peak of supracondylar fractures in early childhood; 78.4% (171/218) of patients were aged ≤8 years. Males comprised 60.1% (131/218) of the sample. Gartland type II fractures accounted for 65.1% (142/218) and type III for 34.9% (76/218). Among predictor variables, 9.2% (20/218) demonstrated an initial LDP >85%, and 18.3% (40/218) exhibited severe soft-tissue swelling. Baseline characteristics are summarized in Table 1.

CONSORT flow diagram of patient selection.
Baseline demographics and fracture characteristics by redisplacement status.
LDP: lateral displacement percentage. Continuous variables are presented as the median (interquartile range); p values were calculated using the Mann–Whitney U test for continuous variables and the χ² test or Fisher’s exact test for categorical variables. Redisplacement rates for Gartland type II and type III fractures were 14.1% (20/142) and 25.0% (19/76), respectively.
During a median follow-up of 8.5 weeks (IQR 6.2–12.1), 39 patients (17.9%) experienced radiographic redisplacement. Redisplacement rates differed by fracture type: 14.1% (20/142) for type II versus 25.0% (19/76) for type III fractures (p = 0.04). The redisplacement group had significantly higher proportions of Gartland type III fractures, initial LDP >85%, and severe soft-tissue swelling than the nonredisplacement group (all p < 0.05). Notably, these risk factors were intercorrelated: type III fractures were more likely to present with LDP >85% (15.8% vs. 5.6% in type II, p = 0.008) and severe swelling (26.3% vs. 14.1%, p = 0.024).
Temporal distribution of redisplacement risk
Kaplan–Meier analysis revealed a nonlinear temporal pattern of redisplacement risk (Figure 3). Rather than declining steadily over time, the cumulative risk curve exhibited two phases of rapid ascent followed by plateauing. By post-reduction day 4, the cumulative redisplacement incidence was 6.9%; by day 14, this increased to 15.1%, accounting for approximately 87% (34/39) of all events. After day 14, the curve plateaued, indicating a transition to a lower-risk phase.

Kaplan–Meier curve of cumulative redisplacement risk.
The temporal distribution was further characterized using kernel density estimation (Figure 4). Two apparent high-risk windows emerged: an early peak at days 3–4, when daily event incidence was highest, and a broader second peak spanning days 7–14, during which total event volume was maximal. Beyond day 14, event density declined substantially. However, it should be noted that events could only be detected at scheduled follow-up visits, and this pattern may be partially influenced by the timing of radiographic surveillance (typically at days 3–5, 7–10, and 14–21 at our institution). Of the 39 redisplacement events, 15 (38.5%) were detected at Visit 1 (days 3–5), 19 (48.7%) at Visit 2 (days 7–14), and 5 (12.8%) at Visit 3 or later (after day 14).

Temporal distribution of redisplacement events.
Stratified survival analysis
To investigate which populations drove the observed temporal pattern, stratified Kaplan‒Meier analyses were performed (Figure 5). Children with an initial LDP >85% demonstrated a significantly higher cumulative redisplacement risk than those with an LDP ≤85% (35.0% vs. 14.0%, log-rank p < 0.001). Similarly, severe soft-tissue swelling was associated with a significantly higher risk than nonsevere swelling (33.3% vs. 13.1%, p < 0.001). Gartland type III fractures also showed a higher risk than type II fractures (25.0% vs. 14.1%, p = 0.021). Notably, high-risk subgroups demonstrated not only higher cumulative risk but also earlier risk accrual, with steeper curve descent during the first week.

Stratified Kaplan–Meier curves by key predictor variables.
Independent predictors of redisplacement
Univariable Cox regression identified Gartland type III (HR 1.92, 95% CI: 1.04–3.55, p = 0.04), initial LDP >85% (HR 4.28, 95% CI: 2.26–8.11, p < 0.001), and severe soft-tissue swelling (HR 3.64, 95% CI: 1.94–6.83, p < 0.001) as significant predictors (Table 2). In multivariable analysis adjusting for age, sex, and other covariates, initial LDP >85% (HR 3.52, 95% CI: 1.82–6.83, p < 0.001) and severe soft-tissue swelling (HR 3.08, 95% CI: 1.61–5.89, p < 0.001) remained independently predictive. Gartland classification lost statistical significance after adjustment (HR 1.45, 95% CI: 0.76–2.78, p = 0.26), suggesting that its prognostic effect may be largely explained by the more direct biomechanical indicators of displacement magnitude and soft-tissue injury severity.
Univariable and multivariable Cox proportional hazards regression analysis for redisplacement risk.
HR: hazard ratio; CI: confidence interval; LDP: lateral displacement percentage. The reference indicates the baseline comparison category for each variable (HR = 1.00). Variables with p < 0.10 in the univariable analysis were included in the multivariable model. The proportional hazards assumption was verified using Schoenfeld residuals (p > 0.05 for all covariates). Ridge-penalized Cox regression yielded consistent results. Events-per-variable ratio = 9.8; total events = 39; total sample = 218.
Sensitivity analysis
To validate robustness, the analysis was repeated in the subgroup of 142 Gartland type II fractures only, excluding potentially more unstable type III fractures. The temporal pattern with two apparent high-risk windows remained evident, with timing consistent with the primary analysis. This supports that the observed pattern is not driven solely by the inclusion of type III fractures. Ridge-penalized Cox regression yielded HRs consistent in direction and significance with those of the primary models, further supporting the robustness of the results.
Discussion
This study analyzed the temporal dynamics of redisplacement risk in a standardized nonoperative treatment cohort of pediatric SCHFs. Three principal findings emerged: (1) redisplacement risk exhibited a nonlinear temporal pattern, with two apparent high-risk windows at days 3–4 and 7–14 post-reduction rather than a steady linear decline; (2) initial lateral displacement >85% and severe soft-tissue swelling were independent predictors of redisplacement; and (3) Gartland classification, while significant in univariable analysis, lost predictive value after adjusting for these direct biomechanical indicators.
Biomechanical interpretation
The observed temporal pattern provides a plausible biomechanical framework for understanding early fracture-healing dynamics. The first apparent risk window (days 3–4) coincides with the peak inflammatory response and tissue edema.21,22 During this early phase, fracture stability depends almost entirely on external immobilization, and physiological swelling fluctuations may disrupt splint–limb interface mechanics, potentially triggering micromotion sufficient to cause redisplacement. 23 The second apparent risk window (days 7–14) represents a transitional period when swelling resolution reduces limb circumference and splint conformity, while biological internal stabilization (fibrocartilaginous callus) remains insufficient to resist physiological stresses.24,25 This period may represent a critical “race” between external constraint decay and internal stabilization formation.26,27 However, we emphasize that this mechanistic interpretation is a speculative inference based on observational data, and its validity requires verification through prospective studies. Notably, because our institution utilized custom cedar bark splinting with distinct biomechanical properties, whether this pattern would be observed under standard fiberglass or plaster casting remains uncertain.
The attenuation of the Gartland classification effect in multivariable analysis deserves attention. This finding suggests that the traditionally emphasized type II/III distinction may serve as a proxy for more fundamental biomechanical parameters—displacement magnitude and soft-tissue injury severity—rather than an independent risk determinant. Clinically, quantitative assessment of the initial displacement percentage and swelling grade may provide more actionable prognostic information than categorical classification alone, potentially enabling more individualized risk stratification.
Clinical implications
We deliberately refrain from proposing a specific follow-up algorithm, as doing so based on single-center retrospective data would be premature. However, should the temporal pattern be validated in prospective multicenter studies using standard casting, a conceptual framework for risk-stratified surveillance might include high-risk patients (LDP >85% or severe swelling) assessed at days 3–4, 7–10, and 14, and standard-risk patients assessed at days 5–7 and 14, with earlier evaluation if clinical concerns arise. Such protocols require rigorous validation—ideally through randomized trials—before clinical implementation.28,29
Limitations
Several limitations must be acknowledged. First, this study utilized custom-molded triplanar cedar bark splinting, and the results cannot be directly extrapolated to settings using standard circumferential casting. Whether the observed temporal pattern represents an intrinsic phenomenon of fracture healing or a specific product of our immobilization method remains a key unanswered question. Second, redisplacement could be detected only at discrete follow-up intervals (typically days 3–5, 7–10, and 14–21) rather than through continuous monitoring; the apparent clustering of events may partially reflect detection timing rather than true biological risk peaks. Third, the single-center retrospective design may introduce unknown confounders. Fourth, our selective inclusion of stable type III fractures—a practice not standard in most North American and European centers—limits generalizability.10,11 The 25% redisplacement rate in our type III subgroup may be considered unacceptably high where surgical resources are readily available.
Future directions
Future research should address (1) prospective multicenter studies using standard casting to determine whether the temporal pattern is generalizable; (2) studies with standardized frequent radiographic protocols to disentangle biological risk patterns from detection artifacts; and (3) randomized trials comparing risk-stratified versus uniform surveillance with respect to redisplacement rates, functional outcomes, and resource utilization.
Conclusions
In this single-center cohort, redisplacement risk following nonoperative treatment of pediatric SCHFs appeared concentrated in 2 early post-reduction windows (days 3–4 and 7–14). Initial displacement magnitude and soft-tissue swelling severity were stronger predictors than the Gartland classification. However, this pattern may reflect discrete follow-up intervals, and generalizability requires validation through prospective multicenter studies using standard cast immobilization.
Supplemental Material
sj-pdf-1-cho-10.1177_18632521261433873 – Supplemental material for Temporal patterns of redisplacement risk following nonoperative treatment of pediatric supracondylar humeral fractures: A retrospective cohort study
Supplemental material, sj-pdf-1-cho-10.1177_18632521261433873 for Temporal patterns of redisplacement risk following nonoperative treatment of pediatric supracondylar humeral fractures: A retrospective cohort study by Jie Zhang, Chaohua Li and Ying Zhao in Journal of Children's Orthopaedics
Footnotes
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.*
Author contributions
Jie Zhang, as the corresponding author, was responsible for the overall conceptualization and design of the study, the statistical analysis of the data, the drafting of the initial manuscript, and the final review of the manuscript. Chaohua Li and Ying Zhao were jointly responsible for the collection, organization, and verification of the clinical case data and participated in revising the manuscript. All authors have read and approved the final manuscript.
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Thank you for your consideration.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Youth Talent Fund Project of the Zhejiang Provincial Science and Technology Program of Traditional Chinese Medicine (Grant No. 2021ZQ078).
Ethical approval and consent to participate
This retrospective cohort study was conducted as part of a larger project funded by the Youth Talent Fund Project of the Zhejiang Provincial Science and Technology Program of Traditional Chinese Medicine (Grant No. 2021ZQ078). The Ethics Committee of Hangzhou Fuyang Hospital of Orthopedics of Traditional Chinese Medicine determined that this retrospective analysis fell within the scope of the original project’s approval and oversight, and therefore did not require a separate ethical application or a new approval number. The committee also waived the requirement for written informed consent due to the anonymized nature of the data. The study was conducted in accordance with the Declaration of Helsinki.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
