Article Text
PDF
Abstract
Objective Poor concussion knowledge in the community has been linked to reduced injury identification. This study investigated if concussion knowledge could be improved by providing standard postinjury advice (written brochure).
Methods This study was a prospective, controlled study, with random allocation of 199 Australian adults to receive either a concussion information (CI, n=101), or non-CI (n=98). All participants completed the Rosenbaum Concussion Knowledge and Attitudes Survey on three occasions: pre-education and posteducation, and 1 week later.
Results A 2 (condition) × 3 (occasion) mixed analysis of variance with concussion knowledge as the dependent variable did not find a statistically significant interaction (p>0.05). This result was unchanged: (1) with the covariate addition of background education and; (2) in a subgroup analysis (individuals with initially ‘low’ self-rated knowledge). Some key misconceptions about concussion were identified.
Conclusion The community knowledge of concussion was not significantly improved by the concussion advice. Since injury recognition relies ton an extent on community knowledge, the identified misconceptions should be addressed. This could occur via public health messaging. In clinical settings and for future research, the next steps should also include regular updating of concussion information to keep pace with advances in the field.
- Concussion
- Recreation / Sports
- Education
- Traumatic Brain Injury
Data availability statement
Data are available on reasonable request.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Concussion knowledge in the community can support the initial injury identification, especially among community sport athletes, but the extent of this knowledge is not widely known.
WHAT THIS STUDY ADDS
We found that some concussion signs and symptoms were correctly identified by the community, which could aid injury identification. However, some important misconceptions were also present, and these were not changed by providing concussion information.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE AND/OR POLICY
If community members are to play a role in athlete monitoring, and by extension improved concussion identification, they require access to information that corrects misconceptions about the causal events and potential grave consequences of injury. This information will need regular review and must be rigorously tested. Once identified and deployed, future studies can investigate if a better-informed public supports the identification of injuries that would otherwise be missed.
Introduction
Concussion is a common injury. This injury can affect people of all demographics as a result of falls, workplace accidents and physical assault.1 Although most people make a full recovery from a first-event concussion,2 there is a 10%–15% chance of symptom persistence.3 For repetitive injury, the outcomes are debated, but they include serious incapacity and even death.2 4 5 These adverse potential outcomes for concussion underscore the importance of accurate injury identification and management, and have led to the recognition that concussion research is a public health priority.4
Concussion affects athletes, especially contact-sport athletes, at community to elite levels. There are differences in the supports available to those injured during sport, including for injury identification, rehabilitation and prevention of reinjury.6 In non-professional and non-elite contexts, community members can play an important role in initial injury identification, especially the parents, partners, friends of junior and adult community-sport athletes, because they can monitor the athlete for symptoms.
Several national and international groups have called for improvements in public knowledge of concussion.7–9 In Australia, for example, the peak government sporting body recently recommended increased concussion education for athletes and health professionals, as well as parents and teachers.7 This recommendation was based on two independent Australian surveys that found ‘modest’ concussion knowledge among approximately 900 community sport coaches and trainers10 and 146 members of the public (50% self-identified sports fans), respectively.11 A further study of junior Australian Rules Football players and their parents found that concussion knowledge could be improved, especially knowledge of specific return-to-play criteria and selected signs and symptoms (eg, disturbed sleep).12 Thus, the findings from empirical studies strongly suggest that public knowledge of concussion needs to be improved, but a focused study in this group has yet to be undertaken.
Overseas studies of concussion knowledge typically reveal specific gaps in public knowledge such as, underestimating cognitive or affective concussion symptoms.7 13 14 14 15 15 It is unclear whether such gaps can be addressed through targeted education. For example, in a quasi-experimental (single-group) study of 61 Dutch professional footballers that compared pre-education and posteducation scores on the 25-item Rosenbaum Concussion Knowledge and Attitudes Scale (RoCKAS) Concussion Knowledge Index (CKI), no change was found to the average CKI scores, after viewing a concussion education module.16 By comparison, Caron et al 17 studied 35 Canadian high-school student athletes who completed the RoCKAS before (T1), and twice after four interactive oral educative presentations, with the posteducation evaluations occurring immediately (T2) and 2 months (T3) after the presentations. In Caron et al’s single group study, a significant CKI score increase of approximately 2.5 points from T1 to T2 was found, with learning maintained at 2 months. Taken together, the divergent findings in these studies suggest the need for additional research. Further, since neither of these studies excluded participants if they showed poor test effort despite this being recommended,18 19 nor did they control for potential covariates, such as the participant’s level of general education,20 improving the research design could help resolve uncertainties.
This study had two aims. First, using valid protocols only, this study investigated adult concussion knowledge in the Australian community using a established tool, the RoCKAS. We expected that knowledge might be limited in some areas, particularly the ‘difficult’ CKI items.21 The second aim was to perform the first parallel groups, randomised controlled test of concussion information (CI) to improve RoCKAS CKI scores with repeat testing after a 1-week period.
Methods
The study targeted Australian adults from the general community. The study was promoted via broadcast emails within a metropolitan university and on social media. Eligible participants were required to have English language proficiency. Ineligible participants: (1) were aged <18 years; (2) did not reside in Australia at the time of participation; or (3) submitted an invalid or incomplete protocol. Invalid protocols were formally identified using the RoCKAS Validity Scale (vs; see below).
An online survey collected background information, such as participant age and concussion exposure (eg, prior concussions (yes or no), self-rated concussion knowledge (low, medium, high)). Conditional logic was employed for follow-up questions. For example, if the participant had a prior concussion, we then asked about the number of prior concussions; the injury context (eg, sport, fall); recency (eg, 6–12 months ago, >12 months ago); and, treatment (yes or no). As a proxy measure for knowledge translation, we recorded if a concussion was reported (yes or no; and if yes, to whom (open-ended)). A yes or no question was asked to assess if participants had any prior concussion education or training, and if so, they were asked to nominate the prior information sources (eg, social media, classroom, brochures).
Concussion knowledge was assessed with the RoCKAS CKI.18 The CKI has 25 true-false items. One-point is awarded for each correct response, for a total score of 25. The CKI comprises 14 statements (eg, A concussion can only occur…[from a]…direct hit to the head), three scenarios (application of knowledge), and eight ‘commonly reported postconcussion symptoms (Rosenbaum, p 47)’18 Based on past literature, a CKI score of 18 or 19 is typical.21 The RoCKAS 3-item true-false VS was also used. A VS score of <2 is indicative of ‘poor/inconsistent effort’ (Rosenbaum, p 47).18 As per precedent,19 minor item modifications were made to improve comprehension or update terms (eg, ‘freshmen’ is an uncommon term in Australia; see table 1). The RoCKAS was selected for this study because it has been psychometrically evaluated and it is the most used survey for measuring changes in concussion knowledge, facilitating comparison with prior research. Further, the CKI items from the RoCKAS are not sport specific.
Original and modified wording of three RoCKAS items
An existing two-page brochure was used to deliver CI.22 This brochure was titled: Mild brain injury discharge advice (adults). The brochure comprised infographics and text about symptoms and postacute (postdischarge) care (eg, when to seek further medical advice). An analysis of the content determined that the brochure answered approximately 50% of the CKI items (answerability). In a prior comparative evaluation of such advice, this resource was highly rated, including on dimensions such as readability.23 A brochure with similar attributes (eg, length) was selected for the non-CI (NCI) condition.24 This material was titled: What is a fracture. It was also labelled as a hospital Emergency Department Fact Sheet. Both materials were branded government health resources.
All participants received study information prior to enrolling. This information included the use of a T3 (delayed) test phase, sample questions from the survey, and an explanation of the study discontinuation and withdrawal options. The data were collected November 2019 to June 2020.
Consenting participants created a unique identifier. This identifier was used for data linkage (T1 to T3). The participants provided background information and completed the RoCKAS (T1). A Qualtrics function randomly allocated the participants to a condition (CI or NCI). The accompanying instruction was to: ‘… closely read the … fact sheet…[imagining] that you are a patient … and you really want to know about your injury. Please pay very close attention to the information because you may be tested on it later’. Immediately following education, the RoCKAS was repeated (T2). After 7 days, an automated email was sent to each participant inviting T3 completion as optional participation. This interval was used in response to recent calls for studies of the retention of information following concussion education.25 All participants were offered a token of appreciation for their study involvement (eg, opportunity to win a gift voucher).
Data analysis
The data were exported to IBM SPSS V.25. Unless otherwise stated, an alpha level of 0.05 was used to determine statistical significance. Descriptive analyses were performed on the CKI summary score, and the percent correct for RoCKAS items was determined. A 2×3 mixed analysis of variance (ANOVA) was used to analyse the effect of condition (information type) on concussion knowledge over time (T1 to T3). The dependent variable was the CKI summary score. ANOVA assumptions of normality, homogeneity of variance and sphericity were examined. Inspection of histograms revealed outliers and negative skew on all CKI scores. Levene’s test for equality of error variances and Mauchley’s test of sphericity were nonsignificant (all conditions, p>0.05). Analyses were conducted with and without outliers and as results did not change, the full sample was used. A second ANOVA included the level of general education as a covariate. A Bonferroni correction was applied to the post hoc tests.
Results
Five hundred and twenty-five people commenced the study. After eligibility and data screening (figure 1), 199 people remained. The major reason for exclusion was failure to return for the optional T3 evaluation (n=221). The mean T2 to T3 retest interval was 8.9 days (SD=8.5).
Flow chart for the selection of the study participants.
Table 2 shows the sample characteristics. The final sample had a mean age of 27.1 years (SD=10.5, range=18–61; 22% men). Most respondents had not had a prior concussion (77%), and of those who had, half (54%) did not report it at the time of injury. The self-rated concussion knowledge of the sample was low (53%) to medium (43%). Most respondents (73%) reported that they did not receive prior concussion education. Of those who had, most had received formal classroom training (16%), or advice from medical (9%) or sports professionals (eg, coaches, trainers; 7%).
Sample characteristics
The mean CKI score for the NCI group was: T1=20.0 (SD=1.9); T2=19.7 (SD=2.1) and T3=20.2 (SD=2.2). The comparable data for the CI group was: 20.5 (SD=1.7); 20.3 (SD=2.0); and 20.8 (SD=2.0). Figure 2 charts the mean CKI score for each group and occasion. This figure highlights the minimal within-group and between-group variability in the CKI score. The ANOVA revealed a non-significant interaction between occasion and condition on the mean CKI score, F(2,394)=0.15 p=0.865, 
=0.001. There was a significant but small main effect for occasion, F(2,394)= 8.20, p<0.001, 
=0.040, and condition, F(1,197)=5.35, p=0.022, 
=0.026. The post hoc Bonferroni-adjusted tests for these main effects revealed that, when averaged across occasions, the NCI group (estimated marginal mean (EMM)=20.0) had a lower CKI score than the CI group (EMM=20.5, p=0.022); and, when averaged across condition, only the T2 and T3 scores differed (T3 EMM 20.5>T2 EMM 20.0, p<0.001). This pattern of results was unchanged when (1) the level of general education was included as a covariate, and (2) for a subgroup analysis using data from individuals with low self-rated concussion knowledge (CI: n=49; NCI: n=56; p’s>0.05).
Results of mixed design ANOVA showing CKI scores for two groups (CI and NCI) on the three test occasions (T1), posteducation (T2) and 1 week later (T3)). ANOVA, analysis of variance; CI, concussion information; CKI, Concussion Knowledge Index; NCI, non-CI.
An exploratory item-level analysis was performed to identify the T1 CKI items answered incorrectly by ≤50% of the sample, and whether these items changed with CI. Table 3 shows these items and the % of the sample who correctly answered them. These items were cross-referenced according to item difficulty (low through high)18 and answerability (table 3). Separate paired t-tests (T1, T2) for each of these items found that only one of the responses was significantly improved by CI.
The five RoCKAS items that were the least likely to be answered correctly (≤50% of the sample) at T1, by correct endorsement frequency (T1, T2), item difficulty (as identified by Rosenbaum and Arnett)18 and answerability from our content analysis of the brochure
Discussion
This study determined the level of concussion knowledge in a community sample, and whether this could be improved with CI. Knowledge was comprehensively assessed with a standardised, psychometrically sound measure, and the effect was examined both with and without controls for general education, and in a self-rated low-knowledge subgroup. This study employed a rigorous method to remove invalid responses. We predicted that concussion knowledge would be improved after CI (written discharge advice), and that the gain would be maintained after the delay.
Contrary to expectations, this study did not find that CI improved CKI scores. This was unexpected because at least one previous report found that CKI scores are improved by concussion education.17 However, such improvements are not always shown.16 This could mean that CKI scores can be improved by education, but the effect might depend on baseline knowledge of the sample or method of intervention.17 If a high level community knowledge of concussion is vital to improve injury outcomes as proposed,7 26 we must understand the contexts in which a public education strategy can succeed.
There are several reasons why CI might not have improved overall CKI scores. First, the T1 CKI score was higher than anticipated, despite over half of the sample (53%) self-rating their knowledge as low. If a ceiling effect was present, it would mean that the study could not show a benefit of education. The current study found that approximately 80% of participants answered all CKI items correctly. This is (1) comparable to the findings from three prior US studies of physical education teachers, collegiate athletes and football coaches (eg, 82%27–29) and (2) higher than reported in studies of contact-sport athletes in the UK and South Africa (eg, 66%,19 67%,30 70%31); all of whom could be expected to be more knowledgeable about concussion than members of the general public. This could indicate that this study has a sample bias (eg, the volunteers were particularly interested in concussion). Nevertheless, if replicated in a larger, stratified sample, the knowledge findings are quite promising, and this suggests that educational efforts can be focussed on addressing the residual gaps (ie, the remaining misconceptions).
In this study, CKI scores might not have improved with CI because the chosen resource was not ideal for the study context or because of the choice of outcome measure (RoCKAS), or because of these factors combined (eg, answerability). The strategy of employing a brief publicly available, expert-developed educational brochure could have expedited the translation of results if a benefit was shown. This does not mean that this measure is not fit for purpose as discharge advice, simply that it did not improve community knowledge. In addition, on further inspection and analysis, this resource did not provide full coverage of the RoCKAS items, including the ‘high difficulty’ items, which limited the ability to test this particular aim. Given that these high difficulty items are also clinically relevant, this is important. A modified resource with extended coverage of issues could be employed in future studies to determine if this adjustment leads to improved CKI scores, without the need for a major reinvestment to develop a new resource. Alternatively, if the improvement was not observed because of the choice of outcome measure (eg, limitations of the measure), this could have affected the results. This is an empirical question that awaits testing in future research. To the best of our knowledge the RoCKAS questions we employed were both current and applicable to concussion from a range of causes, but an independent review of these aspects would benefit future research.
The RoCKAS item analysis found that participants had good symptom recognition. Item analyses are helpful because they indicate specific misunderstandings. Consistent with previous studies16, most participants (>95%) correctly recognised three concussion symptoms: headache, dizziness and difficulty concentrating. Most participants (98%) also correctly answered the item ‘concussions can…[contribute]…emotional disruptions’, an item that has received low scores in prior RoCKAS studies (see19 30). Despite this, at T1, five items were rarely (25% of the sample) or somewhat rarely answered correctly (~50% of the sample). For example, only 21% of participants understood that ‘when…knocked out…[the person is]…experiencing a coma’, and only 24% of individuals correctly disagreed that ‘after a concussion, people can forget who they are….but be perfect in every other way’. The two most divisive items concerned the increased concussion risk if there is a prior concussion; and that, for adults, after 10 days the symptoms are usually completely gone. CI improved the latter item. There was some uncertainty about how concussion is caused (only 67% of participants correctly disagreed that ‘a concussion can only occur…[from]…a direct hit to the head’). Collectively, these responses suggest that a sizeable percentage of this sample did not fully understand the grave consequences that can follow from concussion, or the events that can cause it.
Limitations
This study has several limitations. First, it did not use purposive sampling; hence, it did not obtain a representative sample of the Australian adult population. This sample included a larger number of women volunteers, yet men and women each make up about 50% of the Australian population. One third of our sample had an undergraduate degree, and while almost 50% of Australian women aged 25–34 years have an undergraduate degree, our sample was relatively highly educated and this could have inflated their concussion knowledge scores. Most participants resided in Queensland, even though other geographical regions were sampled. There was high attrition from T2 to T3, and while this could confer a selection bias (eg, that only confident responders returned), the results were unchanged in the low self-reported knowledge analysis. Indeed, an attrition effect was confirmed through a follow-up t-test which found that T1 CKI scores were slightly higher in those participants who did vs did not return for the delay test (p<0.05, Cohen’s d =~0.2). There was also some loss of participants due to invalidity; but this is expected when screening is used. For example, in the RoCKAS development study, 23 participants failed the VS. Nevertheless, this sample did not retain all enrolees and the reasons for the discontinuations are unknown. Considering these factors, this study does not provide a national snapshot of concussion knowledge in the community; this important work remains to be done. Second, this study used a RoCKAS variant. While the changes were minimal (minor item wording alterations) this could affect cross-study comparisons. A third limitation is that the CI was not customised for this study, and it contained answers to just over half of the RoCKAS questions. Further, its use in this context could be questioned since it was not designed for community education. Future studies could evaluate other concussion education resources, including video based materials.32 It is possible that in a different sample, such as one with lower T1 scores, or with a clear injury-related motivation to learn (eg, parent of a contact sport athlete), this same material would have improved knowledge. Finally, some participants might have ignored the resource; despite a mean study completion time of 40 min.
Conclusion
This study’s objective was to examine concussion knowledge in a community sample and determine if this could be improved via a cost-effective, low-intensity method (existing brochure). This study found important gaps in the community’s knowledge of concussion. These gaps were unable to be addressed by providing people access to CI. A targeted public awareness campaign is needed to improve community knowledge of concussion, and in turn this could improve early and accurate injury identification and management, especially in community sport contexts. In clinical contexts, the discharge advice to patients (written and verbal) could be checked to ensure coverage of those issues that are commonly misunderstood. Alternative (or modified) CI is needed for the purpose of public education. Unlike the present study, very few concussion education evaluation studies have included a retention phase, used a control group, screened for validity, or used controls for background eductation: we recommend including these design aspects in future research. The search for solutions to improve concussion outcomes is continuing.26 Larger scale, multi-dimensional studies are now needed to determine how best to improve concussion knowledge in the community and its role in improving injury prevention and management .
Data availability statement
Data are available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by QUT HREC ID number 1900000328. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors thank Catherine Kennon for technical assistance and software support.
References
Footnotes
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
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.
Provenance and peer review Not commissioned; externally peer reviewed.