Delayed Presentation of Seymour Fractures: A Single Institution Experience and Management Recommendations

Abstract

Background: Seymour fractures in children are prone to complications without prompt and appropriate treatment. This study investigated outcomes of Seymour fractures with delayed presentations; specifically, if deep infection predisposed to operative treatment, if antibiotic administration improved fracture healing, and if oral clindamycin had fewer treatment failures than oral cephalexin. Methods: A single-institution retrospective cohort study was performed of patients with delayed Seymour fracture presentations (defined as greater than 24 hours post-injury) between 2009 and 2017. Data collected included demographics, time to presentation, infection on presentation, operative treatment, antibiotic use and duration, fracture union, and complications. Statistical testing used logistic regression and Fisher’s exact test, with results reported as P-values (P), odds ratios (ORs), and 95% confidence intervals (CIs). Results: There were 73 patients with delayed Seymour fracture presentations, with mean age of 11.1 years (standard deviation: 2.9), with 56 (77%) males, and median time to presentation of 7 days (interquartile range: 3-17). Deep infection on presentation was a risk factor for operative intervention (OR = 34.4, P = .0001, CI, 5.5-217.2). Antibiotic administration protected against the development of a nonunion or delayed union (OR = 0.11, P = .008, CI, 0.021-0.57). Time to antibiotics did not protect against nonunion or delayed union (OR = 0.77, P = .306, CI, 0.37-1.3). Clindamycin had fewer treatment failures than cephalexin (P = .039). Conclusions: Deep infection is a risk factor for operative treatment of Seymour fractures with delayed presentations. Clindamycin is a better antibiotic choice for Seymour fractures that present in delayed fashion.

Introduction

Phalangeal fractures are common injuries in pediatric patients, with an incidence of 185 per 100 000 individuals between 5 and 14 years of age.¹ Seymour² described a specific epiphyseal fracture occurring in the distal phalanx. These fractures, now termed “Seymour fractures,” often present clinically as a mallet-type deformity with elongation of the lunula or blood at the nailfold.^3-5 The dorsal displacement of the fracture fragments can cause a nailbed laceration and disrupt the soft tissue envelope around the fracture site, leading to an open fracture.^5,6 The recommended management of Seymour fractures has been prompt recognition with antibiotic administration, followed by nailplate removal, extraction of the germinal matrix entrapped at the fracture site, irrigation and debridement (I&D), fracture reduction, nailbed repair, and immobilization.^5-10 If the reduction is unstable with a splint or cast, operative stabilization with Kirschner wire fixation is advised.⁶

However, Seymour fractures have been frequently unrecognized when patients present with these injuries, possibly due to their similarity to mallet finger injuries and the subtlety of soft tissue disruptions.^6,11 A recent investigation by Reyes and Ho⁶ found higher infectious complications when the treatment was delayed (defined as treatment rendered greater than 24 hours after injury) or with only partially treated injuries (defined as incomplete administration of the triad of I&D, fracture reduction, and antibiotic administration). Superficial infections that develop in Seymour fractures could be addressed with commonly used antibiotics directed against skin flora (such as cephalexin),¹² but deep infection treatment may benefit from antibiotics such as clindamycin that have reliable bone penetration.¹³ A synthesis of literature to date for treatment recommendations was performed by Krusche-Mandl et al¹⁴ in their review of 9 nonoperatively treated and 15 surgically treated Seymour fractures. Although this treatment algorithm advocates for operative treatment for frankly open injuries and those with unstable closed reductions,¹⁴ no study has rigorously compared treatments of Seymour fractures with delayed presentations.

We performed a retrospective review of Seymour fractures with a delayed presentation. Our primary null hypothesis is that there is no association between evidence of deep infection (either gross purulence communicating to the fracture site or radiographic evidence of osteomyelitis) and formal operative intervention in a Seymour fracture with delayed presentation. Secondary aims were to determine: (1) if antibiotic administration was associated with a lack of fracture healing; and (2) whether there was a difference in treatment failure rates with oral clindamycin compared to oral cephalexin.

Methods

Study Population

After obtaining institutional review board approval for this investigation, patients were retrospectively identified in a single institution (a tertiary care pediatric hospital) by the use of relevant International Classification of Diseases (ICD) codes in their electronic medical records (EMRs). Diagnoses identified included ICD-9 codes 816.00 (closed fracture of unspecified phalanx or phalanges of hand), as well as 816.02 and 816.12 (closed or open fracture of distal phalanx or phalanges of hand, respectively). In addition, ICD-10 diagnoses of S62.521A-S62.637A and S62.521B-S62.637B (closed or open displaced fractures of the distal phalanx, respectively) and S62.525A-S62.666A and S62.525B-S62.666B (closed or open nondisplaced fractures of the distal phalanx, respectively) were utilized. Additional inclusion criteria used were: (1) clinical evidence of a Seymour fracture with nailbed injury; (2) radiographic evidence of a Seymour fracture (manifesting as Salter-Harris type I or II fractures or transverse juxtaepiphyseal fractures of the distal phalanx);^2,9 and (3) documented treatment that was initiated more than 24 hours after injury (using the definition of Reyes and Ho⁶).

Demographic, Treatment, and Outcome Data Collection

Demographic data recorded included age, sex, affected digit, time from injury to presentation, and length of follow-up. In addition, the presence of superficial or deep infection at initial evaluation, use of immobilization, and antibiotics utilized (along with route of administration and duration) was evaluated. Superficial infection was defined by the presence of clinical signs of infection (eg, warmth or erythema) without gross purulence emanating from the fracture site or radiographic evidence of osteomyelitis (eg, periosteal reactions, sclerotic bone, and bone erosion). Infections with gross purulence or osteomyelitis were classified as deep infections, per the criteria of Harley et al¹⁵. Operative treatment was defined as treatment of the Seymour fracture in an operating room environment with sedation or general anesthesia provided by an anesthesiologist, digital block of the affected finger, I&D of the fracture site, open fracture reduction, nailbed repair if indicated, and stabilization with a Kirschner wire if splinting/casting immobilization inadequately maintained reduction. The choice of 24 hours as a timepoint for evaluating outcomes for treatment of open fractures (such as Seymour fractures) is supported by Swanson et al,.¹⁶ demonstrating an increased rate of infection following open hand fractures from 6% overall (in 154 patients) to 28.5% (in 7 patients) following treatment greater than 24 hours post-injury.

Outcomes included the absence of fracture healing on radiographs (indicated by trabecular bone or callus bridging across the fracture site) at the final follow-up visit, whether formal operative debridement was required, and antibiotic complications. Antibiotic treatment failure was defined as the either the absence of fracture healing on radiographs at the final follow-up visit or failure to eliminate infection with the prescribed antibiotic. Unplanned operative interventions due to complications or inadequate outcomes from the initial rendered treatment (either surgical or nonsurgical) were also recorded.

Statistical Analysis of Data

All statistical tests were performed with a standard software package (STATA 15.0, StataCorp, College Station, Texas). Descriptive statistics were provided for demographic data and treatments rendered in the form of mean and standard deviation for normal interval data, median and interquartile range (IQR) for non-normal interval data, and proportions for categorical data. The method of univariate logistic regression was used to test associations between: (1) presence of deep infection at initial evaluation (dichotomous predictor variable); and need for formal operative treatment (dichotomous outcome variable); (2) antibiotic administration (dichotomous predictor variable) and lack of fracture healing at final radiographic follow-up (dichotomous outcome variable); and (3) time from injury to antibiotic administration (interval predictor variable) and lack of fracture healing at final radiographic follow-up. In addition, multivariate logistic regression controlling for operative treatment was used to re-evaluate the association between antibiotic administration and lack of fracture healing at final follow-up. Fisher’s exact test was used to evaluate the difference in treatment failure rates with oral clindamycin compared to oral cephalexin (due to low frequencies of expected outcomes). Results of statistical testing were reported as P-values, ORs, and confidence intervals (CIs) at the 95% level. The significance level was α = 0.05.

Results

Demographic Data

There were 73 patients with Seymour fractures presenting in delayed fashion. The mean patient age was 11.1 ± 2.9 years, with 56 (77%) males, and 17 females (23%). The median time from injury to presentation was 7 days (IQR: 3-17), summarized in Table 1. The most commonly injured digits were the thumb (30%) and small finger (26%), followed by the long finger (18%), ring finger (15%), and index finger (11%). The most common mechanisms of injury were impacting fingers against another child or ball during play (45.2%), crushing injuries by falling objects (16.4%), and slamming doors or windows on fingers (13.7%). Patients in clinic were followed for a median of 28 days (IQR: 14-41) after the day they were initially diagnosed with a Seymour fracture in our institution.

Table 1.

Demographics of Patients With Delayed Presentations of Seymour Fractures (N = 73).

Variable	Value (mean ± standard deviation, median [interquartile range], or proportion)a
Age (years)	11.1 ± 2.9
Male Sex	56 (77%)
Affected digit
Thumb	22 (30%)
Index	8 (11%)
Long	13 (18%)
Ring	11 (15%)
Small	19 (26%)
Time from injury to presentation (days)	7 (3-17)
Duration of follow-up (days)	28 (14-41)

Note. A summary of demographic data obtained for all patients included in the study (a total of 73 patients).

^aMean and standard deviation are given for normal interval variables, number and proportion are given for categorical variables, and median and interquartile range (IQR) are given for non-normal interval data.

Treatments Provided After Initial Presentation

Four patients (5%) had a superficial infection, and 8 patients (11%) had a deep infection of the affected digit at presentation to our institution. Example clinical photos and radiographs of a deep infection with hyperemic granulation tissue, purulence at the fracture site, and radiographic signs of osteomyelitis treated with surgical management are shown in Figure 1. Eight patients (11%) were treated with surgical management in the operating room (with 2 operative interventions being unplanned), and 65 (89%) were treated nonoperatively (Table 2). Seven of the 8 (87.5%) operatively treated patients and 8 of the 65 (12.3%) nonoperatively treated patients underwent nail plate removal and repair of the nailbed. In 5 of the 8 patients with deep infection at presentation (62.5%), surgical management in the operating room was performed, whereas only 3 of the 65 patients without deep infection at presentation (4.6%) had intervention in the operating room. The presence of deep infection at presentation was a significant risk factor for requiring formal operative intervention in our cohort (OR = 34.4, P = 0.0001, CI: 5.5-217.2).

Figure 1.

Example case of a 5-year-old female patient presenting 21 days after sustaining a Seymour fracture to the left long finger, complicated by mallet-type deformity with hyperemic tissue and purulence communicating to the fracture site.

Note. Shown are (a) dorsal (top) and lateral (bottom) clinical views of the affected finger. (b) anteroposterior and lateral radiographs of the affected finger demonstrating osteomyelitis and no evidence of fracture healing, and (c) repeat anteroposterior and lateral radiographs of the affected finger after formal operative treatment with nailplate removal, irrigation, debridement, open reduction, nailbed repair, and Kirschner wire fixation.

Table 2.

Treatments and Outcomes of Patients With Delayed Presentations of Seymour Fractures (N = 73).

Variable	Value (median [interquartile range], or proportion)a
Patients undergoing operative treatment	8 (11%)
Patients treated nonoperatively	65 (89%)
Proportion immobilized in splint or cast	66 (90%)
Duration of immobilization, days	25 (15-35)
Duration of antibiotic treatment, days	14 (7-28)
Evidence of fracture healing at final follow-up visit	53 (84%)
Superficial infection at presentation	4 (5%)
Deep infection at presentation	8 (11%)
Unplanned operative procedure	2 (3%)
Patients with antibiotic side effects	3 (7%)

Note. Summaries of treatments rendered and outcomes of interest, including complications such as unplanned operative procedures and antibiotic side-effects. Fracture healing is determined by radiographic signs of trabecular bone bridging or callus formation. Deep infections are characterized by either clinical examination demonstrating gross purulence communicating with the fracture site or radiographic signs of osteomyelitis.

^aNumber and proportion are given for categorical variables, and median and interquartile range (IQR) are given for non-normal interval data.

Other adjunctive treatments included immobilization in 66 patients (90%) for a median of 25 days (IQR: 15-35). A total of 30 patients (41%) received no antibiotics at initial presentation, due to no clinical or radiographic evidence of infection. For the remaining 43 patients (59%) who received antibiotics, the median duration of treatment was 14 days (IQR: 7-28). Oral clindamycin (n = 18) was the most common treatment, with weight-based dosing (mg/kg) per standard pharmaceutical guidelines divided into 3 times daily administration, and a median duration of 27 days (IQR: 14-28). Oral cephalexin (n = 16) was the next most common treatment, also weight-based 3 times daily dosing, and a median duration of 9 days (IQR: 7-11).

Outcomes of Interventions

Evidence of fracture healing at final follow-up was seen in 53 of 63 patients (84%). In 24 of these 63 patients, antibiotics were not administered, and 8 patients (33.3%) showed no evidence of fracture healing. Only 2 of the remaining 39 patients who received antibiotics (5.1%) did not show evidence of fracture healing. The administration of antibiotics was a significant protective factor against lack of fracture healing (OR = 0.11, P = 0.008, CI: 0.021-0.57). Controlling for operative intervention in multivariate logistic regression, antibiotic administration remained a significant protective factor (OR = 0.065, P = 0.013, CI: 0.0074-0.56). Time to antibiotic treatment was not significant risk factor for lack of fracture healing (OR = 0.77, P = 0.306, CI: 0.37-1.3).

Three nonoperatively treated patients with radiographic evidence of osteomyelitis on initial presentation were successfully treated with oral clindamycin (75% of nonoperatively treated patients with deep infection). In addition, 2 nonoperatively treated patients with superficial infections were successfully treated with oral clindamycin (100% of nonoperatively treated patients with superficial infection). Treatment failures were observed in 4 of 16 cases with oral cephalexin treatment, and 0 of 18 cases with oral clindamycin treatment; this difference was statistically significant (P = 0.039).

Complications of Treatment

Three patients using oral clindamycin reported mild adverse reactions: 1 had a rash that resolved with Benadryl, 1 had a self-limited nausea episode, and 1 had mild diarrhea, with other family members also having gastrointestinal issues. No cases of Clostridium difficile–associated diarrhea were reported. No adverse reactions were reported with cephalexin.

Of the 2 patients who required an unplanned operative intervention, 1 had osteomyelitis at presentation and failed outpatient oral trimethoprim-sulfamethoxazole and oral amoxicillin-clavulanate therapies (the patient could not take oral clindamycin due to intolerance after 1 dose). A magnetic resonance imaging (MRI) study in this patient demonstrated a dorsal abscess and persistent osteomyelitis. Following operative I&D, the patient was treated with oral linezolid for coagulase-negative Staphylococcus and had resolution of the infection. The second patient was treated operatively, including Kirschner wire fixation, due to an unstable closed reduction.

Discussion

After an extensive literature search, we did not find reports detailing specific treatment outcomes and recommendations for management for Seymour fractures with a delayed presentation.^5-9,14 Thus, this served as the focus of the current study. Our findings demonstrated evidence of deep infection at presentation was a risk factor for operative management. Secondarily, we established that: (1) antibiotic administration was protective against nonunion; however, (2) delayed antibiotic therapy was not a risk factor for nonunion; and (3) oral clindamycin had fewer treatment failures than oral cephalexin.

Patients with Seymour fractures may have a variety of presentations, from a subtle nailbed injury with a minimally displaced fracture on radiographic imaging to a grossly displaced and exposed fracture on clinical exam and imaging. Therefore, the severity of these injuries may go unnoticed by patients and even some providers, leading to delayed presentations of Seymour fractures. This finding was supported by our data showing 23 of the 73 patients (32%) had a misdiagnosis of their fracture by a clinician prior to presentation at our institution. Reyes and Ho⁶ found 31% of patients with Seymour fractures had delayed presentations (greater than 24 hours after injury). They reported an overall infection rate of 0% for the acutely, appropriately treated cohort versus 45% for the delayed treatment cohort.⁶ Krusche-Mandl et al¹⁴ recommended operative management for Seymour fractures with unstable closed reductions or obvious open injury (eg, injury proximal to the nailfold).

Evaluating the specific outcomes of our investigation, we noted that the presence of deep infection at initial evaluation was a risk factor for undergoing operative intervention. It is noteworthy that all 3 patients with deep infection treated nonoperatively had no purulence at the level of the fracture site. While the presence of purulent fluid is not necessarily a marker of the severity of osteomyelitis, it does typically represent active bacterial infection¹⁷ and hence elucidates the need for formal surgical management. Kargel et al¹⁸ demonstrated that 13 of 21 pediatric patients with acute hand osteomyelitis were treated successfully with oral antibiotics alone, and the 4 patients treated with surgical debridement at initial presentation had gross purulent drainage, supporting oral antibiotic therapy for primary treatment. In addition, Hamdy et al¹⁹ advocated for conservative antibiotic treatment as first-line therapy for subacute hematogenous osteomyelitis in 44 pediatric patients. Time to antibiotic treatment was not found to be a significant risk factor for lack of fracture healing, analogous to the findings of Al-Arabi et al.²⁰ However, the provision of antibiotics did decrease the risk of progression to nonunion. The absence of fracture healing on radiographs may be a proxy for ongoing low-grade infection and hence may explain the role of antibiotics in facilitating healing. This is supported by the lack of complications of nonoperative treatment without antibiotics in our study population if no signs of infection and evidence of fracture healing were present. Finally, we found fewer treatment failures with oral clindamycin compared to oral cephalexin, which is congruent with previous studies demonstrating high osseous penetration and oral bioavailability of clindamycin^13,21 and data indicating success in treating osteomyelitis in pediatric patients with clindamycin.^22,23

On the basis of other investigations in the literature and findings in our study, we propose a treatment algorithm for Seymour fractures with a delayed presentation accounting for the evidence of deep infection, closed reduction success, and evidence of fracture healing (Figure 2). If there is either clinical evidence of deep infection (ie, the presence of gross purulence communicating directly with the fracture site) or instability following closed fracture reduction and immobilization,^6,14 formal surgical management would be recommended. This surgical management would entail ensuring tetanus immunization, perioperative antibiotic administration, I&D of the fracture, open reduction, nailplate removal and nailbed injury repair (if present), and Kirschner wire fixation if open reduction had any signs of instability.^5-9,14 Interposed germinal matrix at the fracture site would be removed, using adequate radial and ulnar skin tissue flaps at the eponychium. The eponychial skin flaps and soft tissue lacerations would be well-approximated after satisfactory fracture reduction is confirmed. Postoperatively, the patient would be treated with a 4-week course of oral clindamycin (weight-based and may range from 15 to 20 mg/kg/day divided into every 8 hour dosing) due to osseous penetration^13,21 and success in treating osteomyelitis.^22,23 Alternatively, another antibiotic may be chosen on the basis of speciation and sensitivity data from intra-operative cultures. If the patient had neither clinical evidence of deep infection nor unsuccessful closed reduction, the patient could undergo nonoperative treatment. The presence of radiographic findings of osteomyelitis (eg, periosteal reactions, sclerotic bone, and bone erosion) would prompt a 4-week course of oral clindamycin for treatment.^13,21-23 Lack of improvement or worsening infection while on this treatment wound indicate advanced imaging, such as MRI to evaluate the osseous extent of infection and presence of abscesses,²⁴ in addition to consideration of formal surgical management. Patients with either no evidence of fracture healing or superficial infection could be treated empirically with a 2-week course of oral clindamycin.²⁵ In all cases, patients should be followed closely (eg, minimum of 2-week intervals) until the infection is resolved and signs of fracture union appear.

Figure 2.

Proposed algorithm for determining appropriate treatment of Seymour fractures with a delayed presentation, on the basis of findings published in the literature (denoted by numerical superscripts corresponding to articles cited in this publication) and results of this study (denoted by an asterisk).

Note. Diamond-shaped boxes represent major treatment decision points based on diagnostic findings listed within these boxes. Rectangular boxes represent treatment pathways. Clinical evidence of deep infection is the presence of gross purulence communicating directly with the fracture site. Radiographic findings of osteomyelitis include periosteal reactions, sclerotic bone, and bone erosion. Signs of fracture healing including trabecular bone or callus bridging across the fracture site. Clindamycin dosing is weight-based and typically ranges from 15 to 20 mg/kg/day divided into every 8-hour dosing. PO = per os, oral route of administration; MRI = magnetic resonance imaging.

This study reports the largest cohort of patients with delayed presentations of Seymour fractures to date, thus permitting comparisons between certain interventions, such as antibiotic regimens and operative versus nonoperative treatments. Moreover, patient assessments and treatments were rendered by fellowship-trained hand surgeons or resident physicians (in orthopedic surgery or plastic surgery) under direction of the attending surgeons. Thus, a reliable appraisal of patient outcomes and complications following Seymour fracture treatment could be made. Finally, the results reported by our investigation are congruent with previous recommendations for Seymour fracture and general open fracture treatment.

However, complete interpretation of our findings and recommendations necessitates consideration of this study’s limitations, mainly due to the small sample of patients. Due to the limited number of Seymour fractures with delayed presentation, especially those undergoing operative intervention, we were not able to perform comparisons between cohorts, such as time to operative treatment, surgical technique, and postoperative protocols. Larger patient cohorts should be used in the future to definitively establish significant associations, such as deep infection requiring operative treatment. Moreover, the limited number of patients treated with oral regimens of clindamycin and cephalexin did not permit a detailed stratification of antibiotics by dosing amount and frequency in order to determine optimal dosing recommendations. An insufficient number of patients were treated with alternative antibiotics to cephalexin and clindamycin to permit analysis, and this may be explored in future investigations. One must also consider that this study was a retrospective review limited by data collected for clinical and billing purposes, but not for dedicated research and may introduce errors such as information or observation biases. These biases may arise from lack of a priori standardization of initial patient evaluation, treatment protocols, and outcome assessments (such as visual analog scale pain scores, range of motion, and modified Kapandji index scores)²⁶ in our patient sample. Furthermore, because this was not a single surgeon study, varying approaches to treatment could have impacted the overall results. Finally, the lack of longer term patient follow-up (eg, 1 year or greater) did not permit definitive tabulation of later complications such as nonunions, physeal disturbance, and nail dystrophy.

In conclusion, our investigations found that patients with delayed presentations of Seymour fractures and either evidence of deep infection or inability to attain a stable closed reduction required formal surgical management. However, those with only radiographic evidence of infection could benefit from a course of antibiotics alone and have a reduced risk of fracture nonunion. The preferred antibiotic for delayed Seymour fracture treatment is clindamycin, likely due to the improved osseous penetration improving its effectiveness against osteomyelitis and facilitating fracture healing. With the aforementioned strengths and limitations in mind, future research may be directed to a prospective multi-center randomized and controlled trial with detailed protocols for initial patient examination, imaging, treatments (including operative technique and antibiotic dosing regimens), and outcome assessments.