Introduction

Supracondylar fractures of the humerus are the most common type of elbow fracture seen in children, comprising 55–80% of pediatric elbow fractures and 3% of all childhood fractures [1,2,3,4,5,6]. Over 50% of all pediatric supracondylar humerus (SCH) fractures are non-displaced [7]. The Gartland classification, which is the most widely used system for this type of fracture, categorizes non-displaced SCH fractures as type 1 of 4 based on fracture severity [7]. Type 1 SCH fractures do not involve any translation or rotation of the humerus, and under radiographic examination, the anterior humeral line transects the capitellum [7].

As type 1 SCH fractures are stable, immobilization for one to four weeks will typically produce good functional outcomes. Once the fracture is healed, motion of the limb should be allowed [8]. Type 1 SCH fractures can be treated by a variety of immobilization methods, including long-arm splinting and above-elbow circumferential casting. In both of these treatment methods, the elbow is immobilized in flexion at 90° with the forearm in a supine neutral position [6, 7]. The most commonly preferred treatment for type 1 SCH fractures is circumferential casting with immobilization for three to four weeks [8,9,10]. However, a review by Cuomo et al. suggests that long-arm splints can also be an effective treatment for type 1 SCH fractures [11]. To our knowledge, there are no studies comparing the efficacy of long-arm splinting to above-elbow casting for type 1 SCH fractures.

The objective of this randomized controlled trial (RCT) was to assess and compare the long-term radiographic and functional outcomes, complication rates, and patient-reported outcomes between long-arm splinting and above-elbow casting treatments for type 1 SCH fractures. Our hypothesis was that the radiographic and functional outcomes of long-arm splinting would not be inferior to the above-elbow casting treatment. Due to recruitment and follow-up challenges, we have resolved to treat this as a pilot study that can inform future robust, multi-center RCTs.

Materials and Methods

Study Design

We conducted a RCT comparing long-arm splinting and above-elbow casting for Gartland’s type 1 SCH fractures in children. This study received approval from the research ethics board at the treating center and was registered with the U.S. National Institutes of Health (ClinicalTrials.gov, #NCT01912365). A non-inferiority study design was used to evaluate whether long-arm splint treatment resulted in comparable radiographic and functional outcomes to the standard above-elbow casting for this fracture type.

Study Population and Setting

Our study included children ages three to 12 years old, who sought care for an injury consistent with the diagnosis of a type 1 SCH fracture in the emergency department (ED) at BC Children’s Hospital. Research staff approached families to describe potential risks, benefits, treatment options, and their ability to withdraw consent at any time. Families were then given adequate time to decide whether to enroll or not.

This ED is at a quaternary care referral center with over 45,000 annual visits. Type 1 SCH fractures could be identified as a definitive fracture seen on radiography, or a combination of an extension injury, elbow tenderness, regional swelling, and posterior fat pad on radiographs without a definitive fracture. Patients with neurovascular compromise, underlying metabolic or structural bone disease, other fracture of the ipsilateral upper extremity, or a previous fracture at the same site were excluded.

Blinding

Blinding was not feasible as there were clear and visible differences in appearance between treatment types that made a blinded study design impossible. Both the patients and physicians were aware of the chosen treatment group following the randomization process. However, all patients were assigned a de-identified subject ID upon enrollment. The investigator assessing patient outcome data was blinded, as only the de-identified subject ID was provided, and thus unaware of the treatment group each patient was placed in.

Randomization

After obtaining patient consent, an orthopedic research assistant or ED team member used a random number software package on a designated computer to randomize (with equal probability) the patient into one of the two treatment groups.

Patients who did not consent to randomization were given the option to consent to be enrolled into an observational sub-group. Patients in this group received treatment corresponding to the decision made with their attending physician, and follow-up data was collected at their 6-month follow-up visit.

Interventions

Patients in the first treatment group received an above-elbow cast. The injured elbow was positioned at 90° in flexion with the forearm in a semi-prone position, and a fiberglass cast was applied extending from the proximal metacarpal to midpoint of the humeral shaft. The cast was then placed in a cuff-and-collar to relieve tension. Patients in the cast group were scheduled for a follow-up appointment at three weeks post-injury for cast removal, X-rays, and clinical assessment.

Subjects randomized into the second treatment group received a long-arm splint. The injured elbow was again positioned at 90° in flexion with the forearm in a semi-prone position, and a long-arm splint was applied extending from the proximal metacarpal to midpoint of the humeral shaft. The splint was then placed in a cuff-and-collar for support. Patients and parents in the splint group were given instruction for proper care and splint removal at three weeks post-injury. A follow-up phone call was conducted by an orthopedic nurse or surgical fellow at one and three weeks post-injury to assess for immediate complications and address any questions about management.

Diagnostic and Primary Outcome Measures

The primary outcome measure was loss in position as measured by a change in Baumann’s angle between initial and 6-month follow-up radiographs. Demographic data, including age and sex, and information regarding the injury, such as side of involvement, mechanism of injury, and time of injury, were collected at enrollment. Anteroposterior (AP) and lateral radiographs of the affected elbow were collected at initial visit and at each follow-up visit to the orthopedic clinic. All patients were scheduled for a 6-month follow-up visit in the orthopedic trauma clinic, which involved a clinical assessment of range of motion according to Flynn’s criteria and a radiographic assessment of Baumann’s angle [12]. Patients aged five years and older were also asked to complete the Activities Scale for Kids performance (ASKp) questionnaire to report activity and functioning level.

Sample Size and Statistical Analysis

A target sample size of 40 patients was determined based on an acceptable difference in mean loss of position (Baumann’s angle) of less than 6° with an α = 0.05 and β = 0.10 [13]. This non-inferiority margin of 6° was chosen as the normal Baumann’s angle has been shown to vary by 6° for every 10° of humeral rotation on the AP radiograph [14, 15]. A minimum sample size of eight patients per group was estimated assuming a standard deviation of 4° from literature. This sample size was increased to 20 patients per group, as a larger sample size would ensure sufficient data for analysis could be obtained and the risk of a type 2 error failing to detect an existing difference between treatment groups would be minimized. Additionally, a sample size of 20 patients per group was seemingly attainable due to the common nature of type 1 SCH fractures.

The statistical analysis for the primary outcome of mean difference in Baumann’s angle between the two treatment arms is based on the lower bound of the 95% Wald confidence interval. In this case, if the lower bound of the mean difference between groups is less than 6°, then we can infer non-inferiority (see Fig. 1) [16].

Fig. 1
figure 1

Margins for inferring inferiority, equivalence, and superiority between treatment types

Full size image

Results

A total of 34 patients were enrolled in our study. Of these 34 patients, 13 participants consented to enrol in the randomized portion of the study, while 21 opted for the observational arm. Five patients in the randomized arm and 10 in the observational arm completed the 6-month follow-up and provided analyzable data for our primary objective. Within this final cohort, five patients received a long-arm splint and 10 received an above-elbow cast.

Demographic data for patients who were lost to follow-up and completed the 6-month follow-up are in Tables 1 and 2, respectively. In the lost to follow-up group, demographic and treatment data were not retained from six patients in the randomized group and one patient in the non-randomized group at the patients’ discretion. Demographic and outcome data for the cohort of patients with 6-month follow-up are displayed in Table 3, with separation into four sub-groups by randomization arm and treatment type (see Fig. 2). We did not observe any notable differences in sex, age at enrollment, or affected elbow between the sub-groups.

Table 1 Demographic data for patients lost to follow-up
Full size table
Table 2 Demographic data for patients who completed their 6-month follow-up visit
Full size table
Table 3 Demographic and 6-month follow-up data from the final cohort
Full size table
Fig. 2
figure 2

Separation of enrolled patients into the final cohort. Patients in the final cohort obtained 6-month follow-up data and are separated by randomization arm and treatment type

Full size image

In each of the four sub-groups, we observed a change in Baumann’s angle of 3° or less at the 6-month follow-up visit. The lowest change was seen in the randomized splinting group at 0.5° (SD = 0.71), and the highest change was seen in the observational splinting group at 3.0° (SD = 1.0). The observational and randomized casting groups had average changes in Baumann’s angle of 2.86° (SD = 3.02) and 2.0° (SD = 1.73), respectively.

At final follow-up, the average functional limb score based on the ASKp questionnaire was higher in the randomized and observational splinting groups at 3.74 (SD = 0.26) and 3.69 (SD = 0.27), respectively, compared to 3.20 (SD = 0.44) in the randomized casting group and 3.22 (SD = 0.51) in the observational casting group. Flynn’s criteria score at long-term follow-up was excellent in all five splinting patients in terms of both change in carrying angle, and flexion and extension. In the randomized casting group, there were three excellent Flynn’s criteria scores, one good score, and two scores that were not recorded, while the observational casting group had 11 excellent scores, one good score, and two scores that were not recorded. In an analysis of the rate of complications, three patients in the splinting group experienced a device breakdown, one was switched to an above-arm cast, and one reported a case of significant itchiness. There were no complications reported in the casting group.

Discussion

RCT-based evidence supporting casting over splinting treatments for type 1 SCH fractures in pediatric patients is lacking. Our preliminary findings suggest that functional and radiological outcomes following the long-arm splinting treatment may not be inferior to those following above-elbow casting. However, we are unable to make any definitive conclusions due to our small sample size, the difficulties in recruiting patients consenting to be randomized, and lack of long-term follow-up. Our difficulties recruiting patients for the randomized arm and obtaining long-term follow-up data can be used similarly to a pilot study, to inform the development and study designs of future RCTs.

The lack of long-term follow-up in our study was partially due to the disconnect between the ED and orthopedic research department. Although a collaborative research study between departments is an appealing design, we observed that this disconnect influenced the consistency and completeness of recruitment and follow-up visits. This observation demonstrates a potential pitfall of collaboration between busy hospital departments and suggests that these types of clinical studies are best conducted in single departments to avoid miscommunication and discontinuity. In regard to pediatric type 1 SCH fractures, clinical studies may be preferably conducted solely in the ED, as orthopedic consult is often not required for this type of injury.

Another limitation of our study is the small patient cohort. Despite recruiting 34 patients and nearly reaching our target recruitment goal, only a small portion of patients returned for long-term follow-up and could be used in meaningful data analysis. As type 1 SCH fractures typically undergo satisfactory recovery and are often healed before the 6-month post-injury time point, many patients and families opted to not return for a long-term follow-up visit if it was not required by the physician. Of those who were included in the final analysis, the majority opted to be enrolled into the observational arm rather than the randomized arm, which also limited the capacity of this study to produce randomized results.

A significant advantage of the long-arm splinting treatment is that splints can be removed in the patients’ home. This eliminates the need for the additional hospital visit that is required in the above-elbow casting treatment for cast removal. Families are subsequently spared from stressors associated with hospital visits, including long wait times, travel expenses, and missing school, work, and other activities for appointments. Decreasing clinic visits is especially relevant in the time of COVID-19, as non-essential travel and in-person appointments should be reduced where possible to mitigate the risk of infection for patients and physicians alike. If shown to be non-inferior to above-elbow casting, the implementation of long-arm splinting as a standard form of management for type 1 SCH fractures would be beneficial to support physical distancing regulations and promoting safe practices in pediatric orthopedic care.

Another option to minimize return visits is the use of a soft cast. This material holds the fracture as it unites in a secure manner, with similar stability to an above-elbow full cast. The soft cast has a small tab shown to the parents upon application, which can be used at three weeks to unravel the cast at home. This results in the convenience of a long-arm splint in terms of at-home removal with the comfort of an above-elbow cast. It also avoids the risk of device breakdown, which occurred in three patients in our splinting group.

This study faced limitations that prevent any conclusive findings from being made. Further investigations with larger sample sizes and complete long-term follow-up are needed to compare the efficacy of long-arm splinting to the standard above-elbow casting treatment. Our experience and challenges recruiting patients in the randomized arm and obtaining long-term follow-up can be used to inform future trials and serve the purpose of a pilot study. Long-arm splinting presents a possible alternative form of management for pediatric type 1 SCH fractures and encompasses advantages that are especially relevant in the time of COVID-19.