Article Text
PDF
Abstract
Objectives We evaluated the association between timing of presentation and postconcussive symptoms (PCS) at 1, 4 and 12 weeks after injury.
Methods This was a secondary analysis of a prospective cohort study conducted in nine Canadian paediatric EDs in 2013–2015 (5P study). Participants were children who suffered a head injury within the preceding 48 hours and met Zurich consensus concussion diagnostic criteria. The exposure was the time between head injury and ED presentation. The primary outcome was the presence of PCS at 1 week defined by the presence of at least three symptoms on the Post-Concussion Symptom Inventory (PCSI). Secondary outcomes evaluated PCS at 4 and 12 weeks. Multivariable logistic regression analyses were adjusted for ED PCSI and other potential confounders.
Results There were 3041 patients with a concussion in which timing of the injury was known. 2287 (75%) participants sought care in the first 12 hours, 388 (13%) 12–24 hours after trauma and 366 (12%) between 24 and 48 hours. Compared with children who sought care >24 hours after trauma, children who sought care in the first 12 hours had a significantly lower incidence of PCS at 1 week (OR: 0.55 (95% CI 0.41 to 0.75)) and 4 weeks (OR: 0.74 (95% CI 0.56 to 0.99)) but not at 12 weeks (OR: 0.88 (95% CI 0.63 to 1.23)).
Conclusions Patients who present early after a concussion appear to have a shorter duration of PCS than those presenting more than 12 hours later. Patients/families should be informed of the higher probability of PCS in children with delayed presentation.
- trauma, head
- paediatric injury
- paediatric emergency med
Statistics from Altmetric.com
Key messages
What is already known on this subject
Most children will have symptoms during 1–4 weeks following a head concussion. In a prior study we validated a score that would predict duration of postconcussion symptoms (PCS). However, we did not consider time to evaluation as potential risk factor for longer duration of PCS.
What this study adds
In this secondary analysis of prospectively collected data at nine paediatric EDs in 2013–2015, we found that children who consult the ED more than 12 hours following a head injury and subsequently diagnosed with concussion are more likely to have prolonged symptoms compared with children with earlier evaluation. Patients/families should be informed of the high probability of PCS in children with delayed presentation.
Introduction
Concussion is an important health problem commonly evaluated in the ED. Paediatric concussions account for up to 90% of all concussions1 with an annual incidence for adolescents varying from 10.5/1000 to 16.5/1000.2 Between 55% and 90% of patients who have sustained a concussion exhibit postconcussion symptoms (PCS) at 1 week after injury3–5 and 30% of these children continue to have persistent PCS at 4 weeks.3 ,5
Children are commonly brought to the ED for an evaluation of a head injury and many of these children will subsequently be diagnosed with a concussion.6 In the course of these presentations, it has been noted that some children present to the ED on the same day as the head injury, while others have a delayed presentation.5
Not all children diagnosed with a concussion are immediately seen in the ED.7 A number will first be evaluated on the sidelines of athletic events, or in a primary care provider’s office. Landmark studies reporting the persistence of postconcussive symptoms from ED presentation include patients who sought care at different times following injury.5 7 Time to ED presentation may be associated with the risk of postconcussive symptoms. For example, children who seek care immediately after the accident may do so because of the ‘accident’ or following a protocol while those who present after a 24 hours’ delay may consult for persistent symptoms. What remains unknown is if there is an association with the timing of the presentation to the ED with a concussion and the duration of postconcussive symptoms.
Knowing if there is an association between time to presentation in the ED and the duration of postconcussive symptoms after injury will help ED physicians to prognosticate the outcome of their patients. To evaluate a potential association, we conducted a secondary analysis of the 5P study, a prospective cohort study of 3063 children with head injury and subsequently diagnosed with concussion conducted in 2013.5
Methods
Design
This was an unplanned post hoc secondary analysis of data from the 5P study, which was conducted in nine Canadian EDs.5 The primary aim of the 5P study was to derive and validate a clinical risk score predicting PCS in children at 1 month following concussion. The idea for the current subanalysis emerged following publication of the first study. In the original 5P study over 46 variables were selected a priori as predictors of PCS at 1 month. However, we did not consider time from injury to ED triage as a potential predictor.
Setting
The EDs of nine Canadian paediatric hospitals recruited patients from 2013 until 2015. All participating sites are members of the Paediatric Emergency Research Canada group.8 All participants/families provided written informed consent, and assent as appropriate.
Participants
Study participants were children and adolescents aged 5.00–17.99 years who presented to a participating ED within 48 hours of a head trauma, and who met diagnostic criteria for concussion consistent with the Zurich consensus statement.9 Concussion was defined as a pathophysiological process caused by a mechanical energy transfer to the head, resulting in a brain injury with one or more symptoms in one or more of the following clinical domains: somatic, cognitive, emotional/behavioural or sleep. Patients with language barriers, abnormal neuroradiological findings, those who required neurosurgical operative intervention, intubation or intensive care were excluded. In addition, those with multisystem injuries requiring admission to hospital, operating room or procedural sedation, neurological developmental delay, or intoxication were excluded.
Independent variable
The exposure of interest was the time between injury and presentation to the participating ED’s triage. This was defined by the interval between the hour of the accident as reported by the patients/parents during the initial interview and the time of arrival to the ED. The variable was categorised a priori as <12.0 hours, 12.0 to <24.0 hours and 24.0 to <48.0 hours.
Outcomes
The primary outcome was the presence of PCS 1 week following ED presentation. As defined in the 5P study, PCS involved an increase of at least three symptoms on the ED child Post-Concussion Symptom Inventory (PCSI) as compared with the child’s baseline behaviour prior to the injury.5 The PCSI is a validated symptom scale that queries symptoms reflecting physical, cognitive, emotional and sleep domains.10 11 Evaluation of PCS was conducted at 1, 4 and 12 weeks using an electronic survey. Research assistants telephoned non-responders and those opting for telephone follow-up to complete the survey orally. Given that persistent PCS is defined by symptom persistence at 4 weeks or 3 months, we also included the 4-week and 12-week time points as an outcome of interest. All study outcomes were expressed as binary variables (ie, presence of PCS vs absence of PCS).
Demographic and clinical characteristics
Other baseline characteristics were collected during ED visit and were considered potential confounders. These included demographic characteristics (age, sex), premorbid history (eg, history of concussion, migraine, learning disability, attention deficiency and hyperactivity disorder, depression, anxiety, sleeping disorders), event characteristics (loss of consciousness, mechanism of injury, seizure), extent of acute concussion-related symptoms measured using the ED child PCSI items at presentation in the ED and performance on standardised ED tests (Balance Error Scoring System tandem stance test). These variables were used to derive the 5P score in the prior study, and were selected based on conceptual relevance, existing evidence5 6 or expectation of potential associations with our proposed outcomes.
Analysis
Descriptive statistics were used to characterise the distribution of baseline demographic, clinical and study outcome variables after stratification into the three time intervals (<12 hours, 12–24 hours, >24 hours). Univariate and multivariable logistic regression analyses were used to examine the relationship between time to presentation (categorised) and PCS at 1, 4 and 12 weeks. Two multivariable analyses were conducted. The first included all 16 potential confounders (age, sex, previous number of concussions and maximum symptom duration, duration of loss of consciousness, personal history of migraine, family history of migraine, premorbid history variables (learning disability, attention deficit hyperactivity disorder, depression, anxiety, sleep disorders, other psychiatric disorders), seizure following injury, mechanism of injury, tandem stance, ED PCSI) as model covariates to predict PCS. A second model included only the 5P score as a covariate. The 5P score is a clinical score composed of nine items (age, sex, migraine history, previous concussion duration, answering slowly to questions, balance problem, headache, sensitivity to noise and fatigue). The score goes from 0 points (4% risk of PCS at 1 month) to a maximum of 12 points (81% risk of PCS at 1 month).
From the models, the unadjusted or adjusted OR and associated p value for time to presentation were estimated along with the c-statistic to provide an indication of model discrimination performance. To gain further insights into the relationship of time to presentation with the study outcomes, we refitted the multivariable regressions with time operationalised as a continuous variable and applied restricted cubic splines with 4 knots to allow for non-linearity. To characterise insights from these model fits, the predicted probability of postconcussive symptoms was plotted over the range of times to presentation. In all multivariable models, listwise deletion was performed for participants with missing data on any covariate or outcome variable for the associated model. All analyses were performed using R V.3.5.2.12
Patient and public involvement
No patient and public were involved in this study.
Results
Characteristics of the participants
Among the 3063 children included in the 5P study, 3041 participants had data on the timing of the injury (n=3041, 99.3% of initial sample) and were thus included in this analysis. Of these, 2601 (86%) were followed at 1 week, 2554 (84%) at 4 weeks and 2240 (74%) at 12 weeks. Baseline characteristics of children with missing data on outcome were not significantly different from those with complete data.5 Among those reached in follow-up, PCS was present in 1503/2601 (57.8%) at 1 week, it persisted in 798/2554 (31.2%) at 4 weeks and in 443/2240 (19.8%) at 4 weeks.
Of the 3041 participants, 2287 (75%) sought care in the first 12 hours, 388 (13%) between 12 and 24 hours and 366 (12%) sought care after 24 hours (figure 1). Children who sought care earlier were younger and were more likely to be male (table 1). There was no difference between the three groups for mean PCSI at initial evaluation. However, timing of initial presentation to the ED was associated with the 5P score: children who were evaluated earlier had a lower 5P score and were thus estimated to be at lower risk of PCS at 1 month.
Flow diagram of participant recruitment and selection.
Participants’ baseline demographic characteristics (n=3041 participants) by time to ED presentation
Comparison of <12 hours vs 24–48 hours
In comparison to children who sought care 24–48 hours after injury, those who sought care within 12 hours were less likely to report postconcussive symptoms at 1 week (OR: 0.45 (95% CI 0.34 to 0.58)), 1 month (OR: 0.54 (95% CI 0. 42 to 0.69)) and 12 weeks (OR: 0.65 (95% CI 0. 48 to 0.87)) in univariate analysis (tables 2 and 3 and online supplementary material). After adjustment for potential confounders, these associations remained statistically significant at 1 week (OR: 0.55 (95% CI 0.41 to 0.75)) and 1 month (OR: 0.74 (95% CI 0.56 to 0.99)) but not at 12 weeks (OR: 0.88 (95% CI 0. 63 to 1.23)). Similar findings were seen with the second model, when time to evaluation was adjusted for the 5P score only; the associations were significant at 1 and 4 weeks, but not at 12 weeks (tables 2 and 3).
Supplemental material
Association between time to ED presentation and risk of postconcussion symptoms at 1 week (n=2601)
Association between time to ED presentation and risk of postconcussion symptoms at 4 weeks (n=2554) and 12 weeks (n=2240)
Comparison of 12–24 hours vs 24–48 hours
In unadjusted analysis, PCS was present in significantly fewer children who sought care 12–23.99 hours after injury compared with children who sought care 24–48 hours after injury (table 2) (OR: 0.70 (95% CI 0.50 to 0.97)) but there was no statistical significance at 4 and 12 weeks (table 3). Moreover, there was no statistical difference between these two groups at 1, 4 or 12 weeks when adjusted for multiple confounders or for the 5P score (tables 2 and 3).
Comparison of <12 hours vs 12–24 hours
Children who consulted in the first 12 hours were less likely to report PCS at 1 week (OR: 0.64 (95% CI 0.51 to 0.82)) and 4 weeks (OR: 0.62 (95% CI 0.49 to 0.79)) in comparison to children who consulted 12–24 hours following trauma. These associations remained significant after adjustment using either the model with multiple confounders, or the model using 5P as a covariate. However, there was no statistical difference in presence of PCS at 12 weeks (OR: 0.78 (95% CI 0.56 to 1.09)).
PCS on presentation
Figure 2 demonstrates a positive linear association between time to ED presentation and the duration of PCS. Duration of PCS increased linearly for each hour delay before ED presentation.
Probability of postconcussive symptoms at 1 and 4 weeks according to the timing of visit (with 95% CI).
Discussion
This study demonstrates that children who seek care rapidly after a head injury and subsequently diagnosed with concussion are at a lower risk of prolonged PCS as compared with children who seek care at 12–48 hours after the injury. Furthermore, this relationship appears to be linear: the greater the delay to presentation, the greater the probability of suffering from persistent PCS. While this may be secondary to a self-selection bias because children who seek care earlier had a lower 5P score, the association between timing of visit and PCS remained significant after adjustment for the 5P score and other potential confounders. Also, timing was not associated to the PCSI score at initial visit.
We are not aware of any previous study evaluating the association between PCS and timing of ED presentation. Previous studies reporting the proportion of patients having PCS at 1 month following concussion differ depending on the study site. While it was 30% for children evaluated in the ED3 5 and 58%–62% in specialised clinics,7 13 persistent PCS occurred only in 1.5% of children diagnosed with concussion on the field.14 On a similar topic, a recent study demonstrated that children who seek care after 24 hours for a head injury are at a higher risk of finding a positive traumatic brain injury on CT scan as compared with children who sought care in the first 24 hours (3.8% vs 1.2%).15 A recent systematic review of the predictors of clinical recovery from concussion did not evaluate time to medical care as a predictor variable. However, the study reported that development of subacute problems (new concussion symptoms in the days following presentation) was associated with persistent PCS.16 While children who seek care in the first few hours may do so because of specific concern for intracranial bleed and as a precaution following the worrying circumstances of an accident, those who seek care after a delay of 24 hours may do so because of the number, severity and total burden of symptoms. However, the latter was not identified in our study as PCSI scores reported at the moment of ED consult were similar in all three groups and this was included as a potential confounder in the multivariate analysis. Nonetheless, time to presentation is multifactorial and merits further investigation.
Given this study’s observational design, we cannot evaluate the causality of the association between time to presentation and PCS. However, considering the limited healthcare provided for concussion, the identified association is likely related to a selection bias: children who continue to have symptoms beyond 24 hours have more PCS and more prone to consult a healthcare facility than those who have no symptoms. Based on this, emergency physicians evaluating children seeking care for a head injury and subsequently diagnosed with concussion after a delay should be aware of the higher association with PCS and appropriate management strategies should be implemented.
The study has several limitations. Among them, the use of an observational design is prone to bias by indication. As mentioned, the higher frequency of PCS may be related to the presence of comorbid conditions that were not included in the model. A second limitation is the absence of information related to the reasons behind the delayed consult. This could have helped to understand the relationship between timing of consult and outcome. There were missing data and patients who were lost to follow-up who might have improved, which would lead to overestimation of PCS. Finally, while the study identifies a potential predictor of poor outcome, it does not lead to a modifiable action unless treatments can alter the natural history. This would need to be studied in a subsequent study.
In conclusion, in this secondary analysis of the 5P study, we identified a strong positive relationship between the time to ED presentation following head injury subsequently diagnosed with concussion in children and presence of PCS at 1 week and 1 month. This risk factor should be considered when prognosticating the risk of PCS in these children.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors JG designed the study, supervised the data analysis, drafted the manuscript and finalised the submission. AAL, KOY, MO, WC, KB, EB, GS, ASD, DB and RZ participated in the study design, revised the manuscript and provided approval for the final version. KT participated in the study design, conducted the initial statistical analysis and provided feedback on the manuscript. RZ was the instigator of the primary study (5P study). All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Funding This work was supported by the Canadian Institute of Research Health (grant number 293380).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Patient consent for publication Not required.
Ethics approval The initial study and subsequent studies were approved by the Research Ethics Board of the nine participating institutions.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available. Data from this study are not publicly available. It may be available for researchers upon request.