## Abstract

Self-reported symptoms are an integral part of the assessment and management of a sports-related concussion. However, postconcussion-like symptoms are reported by non-concussed individuals. Moreover, the current best practice in the reporting of symptoms does not take into account the potential influence of psychological and lifestyle factors. This study aimed to explore the influence of these factors on the reporting of postconcussion-like symptoms. University students (N= 603) completed the Sport Concussion Assessment Tool 2 postconcussion symptom scale along with other predictor variables via a cross-sectional web-based survey. Linear regression analyses revealed six modifiers contributing to the total symptom score with the strongest being alcohol consumption (Estimate = 2.75, p < .001). Following these findings, clinicians need to exercise caution when interpreting the symptom scores for making decisions on the return-to-play (RTP). A failure to do so may lead the health professional to either prematurely RTP or not clear the concussed athlete to resume their sport.

## Introduction

The self-reporting of symptoms (SRS) is widely used in the multifaceted approach to diagnose a concussed athlete and as one of the components for making informed decisions on return-to-play (RTP) Guskiewicz et al., 2013). Recent consensus statements developed and disseminated by the international Concussion in Sport (CIS) group (McCrory et al., 2013) and the National Athletic Trainer's Association Position Statement (Broglio et al., 2014) have stated that pre-season baseline measurement of symptoms can be helpful in interpreting the post-injury symptoms to diagnose a concussive brain injury. Therefore, the current best practice in many settings is to obtain symptom scores at baseline for comparison with the postconcussion symptom scores (Broglio et al., 2014). In order to document the symptoms, a number of postconcussion symptom scales (Iverson, Zasler, & Lange, 2007; Lovell et al., 2006; McCrory et al., 2009; Piland, Ferrara, Macciocchi, Broglio, & Gould, 2010) have been used with the most widely being the Sport Concussion Assessment Tool 2 (SCAT2) postconcussion symptom scale (McCrory et al., 2009).

This practice of reporting symptoms at baseline and following concussion for comparative purposes may appear to be an ideal strategy considering that an individual acts as his own control. However, it must be noted that several studies have reported the presence of postconcussion-like symptoms in normal healthy (non-concussed) individuals at rest (Iverson & Lange, 2003; Sullivan & Edmed, 2011; Zakzanis & Yeung, 2011) and following exercise (Balasundaram, Sullivan, Schneiders, & Athens, 2013). Moreover, the presence of postconcussion-like symptoms have also been identified in individuals with clinical conditions such as chronic pain (Gasquoine, 2000), psychiatric problems (Fox, Lees-Haley, Earnest, & Dolezal-wood, 1995), orthopedic injuries (Ettenhofer & Barry, 2012), and following trauma (Meares et al., 2008).

While these studies have reported the presence of postconcussion-like symptoms, it is not yet known whether their reporting can be modified by selected psychological (e.g., physical fatigue, anxiety) and lifestyle factors (e.g., alcohol and caffeine consumption). For example, both stress and depression have been shown to modify the reporting of postconcussion-like symptoms in university students (Edmed & Sullivan, 2012). Given the lack of investigation of other modifying factors, there is a need to explore the influence of selected modifiers on the reporting of postconcussion-like symptoms. This is pertinent because a failure to account for the influence of psychological and lifestyle factors on the reporting of symptoms may result in a potential misinterpretation by the clinicians. This may impact the decisions made on the diagnosis and/or RTP when SRS is used along with other measures of concussion assessment.

In addition to exploring these potential modifiers, there is also a possibility of differences in symptom reporting between males and females. Studies that investigated gender differences on symptom reporting have had varied findings (Covassin et al., 2006; Lovell et al., 2006; Mrazik, Naidu, Lebrun, Game, & Matthews-White, 2013; Valovich McLeod, Bay, Lam, & Chhabra, 2012; Zuckerman et al., 2012). In general, a few studies have identified that female collegiate (Covassin et al., 2006), high school and student athletes (Lovell et al., 2006), and undergraduate students (Mrazik et al., 2013) report more symptoms than males during baseline testing. Conversely, other studies did not find any gender differences in high school athletes (Valovich McLeod et al., 2012), and high school soccer players (Zuckerman et al., 2012) at baseline. Therefore, there is a need to investigate whether gender, when combined with potential modifiers, has an impact on symptom reporting. Collectively, the purpose of this study was to investigate the influence of selected psychological and lifestyle factors on the reporting of postconcussion-like symptoms in a non-concussed cohort.

## Methods

### Participants

Six hundred and three university students (154 males and 449 females) aged 18–30 years volunteered to complete the survey. Participants were recruited via visits to classes, email distribution lists and posters within the university. Exactly 65% (392/603) of the participants reported playing one or more recreational sports on three or more days per week, and the participants came from the four academic divisions of the university. This study was approved by the institutional review board of the University of Otago. The informed consent was provided by all participants by voluntarily participating in the cross-sectional web-based survey (Dillman, Smyth, & Christian, 2008) and their identity remained anonymous.

### Survey Instrument

The survey incorporated the SCAT2 postconcussion symptom scale (McCrory et al., 2009) together with recognized scales for psychological and customized questions for lifestyle factors as predictor variables. The factors of stress, anxiety, depression, physical fatigue, and trouble sleeping were selected as potential predictor variables based on the bio-psychosocial conceptual model of poor outcome following mild traumatic brain injury (Iverson, 2012). Additionally, other factors were considered potentially important based on the previous work from other researchers and these include gender (Covassin et al., 2006), physical activity/exercise (Alla, Sullivan, McCrory, Schneiders, & Handcock, 2010; Balasundaram et al., 2013), and mental fatigue (Johansson, Starmark, Berglund, Rodholm, & Ronnback, 2010). The phraseology of instructions for the recognized scales were slightly modified where appropriate to suit the needs of this study and population. Additional questions regarding ethnicity and the primary sport played were also included for descriptive purposes.

### Psychological Factors

Each of these is detailed below.

#### Stress

The Perceived Stress Scale 4 measures stress and is appropriate for use in situations that require a brief measure of the individual's perception of stress (Cohen & Williamson, 1988). It consists of four-items, each rated on a five-point Likert scale and the score of each item is summed to provide the composite scores with higher scores representing an increased level of stress (Cohen & Williamson, 1988). This scale has shown to have internal reliability (r = .60) and predictive validity (Cohen & Williamson, 1988).

#### Physical fatigue

The energy/fatigue scale is an adapted subscale of the Medical Outcomes Study measures (Stewart, Hays, & Ware, 1992). This scale contains five-items, each rated on a six-point Likert scale and the score of each item is summed and averaged to provide a summary score with increased scores representing more energy and less fatigue (Stewart et al., 1992). The scale has demonstrated excellent internal consistency (0.89) and test–retest reliability (0.85).

#### Mental fatigue

A two-item mental fatigue scale was customized, modeled on the mental component of the Chalder Fatigue Scale (Chalder et al., 1993), which is shown to have good clinical validity in the general population (Chalder et al., 1993). A separate scale was used to differentiate it from the physical fatigue component because an individual can have a lack of concentration without being physically tired and vice versa. Each item on this scale is rated on a six-point Likert scale and the score of each item is summed and averaged to provide the summary score with higher scores indicating greater mental fatigue.

#### Anxiety and depression

The Hospital Anxiety and Depression Scale (HADS) measures both anxiety and depression (Zigmond & Snaith, 1983) and consists of a total of 14-items with a seven-item subscale for each of the anxiety and depression constructs. The participant is instructed to rate each item on a four-point Likert scale and the score of each item of the respective subscales for anxiety and depression is summed to provide the composite scores with scores higher than 8 indicative of greater levels of anxiety and depression (Zigmond & Snaith, 1983). In general, the HADS subscales have been reported as having good to excellent psychometric properties for different population groups (Bjelland, Dahl, Haug, & Neckelmann, 2002).

### Lifestyle Factors

Customized (dichotomous) questions were developed to measure the lifestyle factors of physical activity status (e.g., Did you engage in any exercise in the last 24 h?), trouble sleeping the previous night (e.g., Did you have trouble sleeping the previous night?), and alcohol (e.g., Did you consume alcohol in the last 24 h?) and caffeine consumption (e.g., Did you consume caffeine in the last 24 h?). The variables of alcohol and caffeine consumption and physical activity status represent the previous 24 h. Our aim was only to use a single customized question for each of the lifestyle factors and not to measure these variables as an entire construct.

### SCAT2 Postconcussion Symptom Scale

The postconcussion symptom scale which is one of the components of the SCAT2 scale (McCrory et al., 2009) was used as a dependent variable. The SCAT2 scale also consists of objective measures such as balance examination, tests for cognitive assessment, and co-ordination tests (McCrory et al., 2009). The symptom component of the SCAT2 scale consists of 22-symptom items, each rated on a seven-point Likert scale from 0 to 6, where the total symptom score (TSS) is calculated by adding the number of symptoms rated >0 for a maximum possible score of 22. Additionally, the symptom severity score (SSS) is calculated by summing the rated score of each of the 22 symptoms (i.e., 22 × 6, maximum possible score is 132). This symptom scale appears to be the emerging reference standard due to its widespread use and incorporation into the whole SCAT2 scale (McCrory et al., 2009) and subsequently the SCAT3 scale (McCrory et al., 2013) for which the symptom component remained unmodified. This scale demonstrates acceptable reliability and face validity (Valovich McLeod et al., 2012), and was also endorsed by the CIS group (McCrory et al., 2009).

### Pretesting of the Survey Instrument

In order to establish the content validity, the questions were scrutinized by the research team and later pilot tested with a small (n= 12) group of participants from the same sampling frame who provided feedback on their comprehension of questions, the web-layout, and the time taken to complete the survey. Following feedback the participants, the web-survey was amended where required prior to its formal launch.

An email containing a link to the survey was sent to the email address of all participants who expressed interest in participating in the survey. A reminder email was sent to non-respondents a week following the initial email in order to enhance the participation rate. The question items in the survey instrument were randomized in order to reduce any bias associated with participants identifying the constructs. An adaptive questioning format was used; wherein, only certain items were displayed based on the participants’ response to previous questions in order to minimize the number of questions on each screen page. To prevent multiple entries from the same participant, a registration method was adopted where the participants had to first login to start the survey. Participants completed the web-based survey voluntarily and were entered into a prize draw to receive one of ten NZ\$20 iTunes store vouchers. The survey instrument had 44 items in total, and took ∼8–10 min to complete as reported by participants during pilot testing. Data were collected between July and October 2012.

### Data Analyses

This study subscribed to the design and reporting requirements identified in the CHERRIES guidelines (Eysenbach, 2004) for web-based surveys. The data from only fully completed questionnaires were included in the analyses. Considering the importance and value of calculating both the TSS and SSS from a clinical perspective (Valovich McLeod et al., 2012), this study analyzed and reported on both the TSS and SSS. Frequencies, percentages for dichotomous variables, and descriptive statistics for continuous variables were generated for combined participants and by gender.

The effect of the predictor variables (k= 10) on the outcome measures of TSS and SSS was estimated using separate multiple linear regression models. Prior to this modeling, a univariate linear regression analysis was conducted separately for the TSS and SSS with the level of significance set at p < .20 in order to filter the predictors to be included in the subsequent multiple linear regression models. Following this, a stepwise multiple linear regression analysis (backward elimination method) was performed separately for both the TSS and SSS. All possible two-way interactions between the predictor variables (k = 45) were also included in the models. Post hoc probing of the significant interactions was conducted by plotting the interactions to examine the simple slope.

Model fit selection was based on the Bayesian information criterion (BIC), a conservative approach that penalizes models for the number of parameters estimated (Moreno & Girón, 2006). A level of significance was set at p < .05 for individual predictor contribution to the TSS and SSS models. Stein's formula (Field, 2009) was computed (manually) separately for the TSS and SSS models in order to determine the predictive ability (cross-validate) of the overall model fit. Cross-validation is usually performed to assess how accurate a model is across different samples from the same population (Field, 2009). All data analyses were performed using the R statistical software, version 3.0.2 (R Core Team, 2013).

## Results

The survey completion rate (Eysenbach, 2004) was 95.1% (603/634). The mean age for the males and females was 20.85 years (SD= 2.45 years) and 21.04 years (SD= 2.83 years), respectively. The descriptive statistics for symptom scores (TSS and SSS) and psychological factors and frequencies for lifestyle factors are presented in Tables 1–3.

Table 1.

Descriptive statistics for total symptom scores and symptom severity scores

Symptom scores Males (N= 154)

Females (N= 449)

Combined (N= 603)

M SD 95% CI M SD 95% CI M SD 95% CI
TSS 8.94 5.88 [8.0, 9.8] 10.02 6.03 [9.4, 10.5] 9.74 6.01 [9.2, 10.2]
SSS 19.50 17.49 [16.7, 22.2] 23.73 21.30 [21.7, 25.7] 22.65 20.47 [21.0, 24.2]
Symptom scores Males (N= 154)

Females (N= 449)

Combined (N= 603)

M SD 95% CI M SD 95% CI M SD 95% CI
TSS 8.94 5.88 [8.0, 9.8] 10.02 6.03 [9.4, 10.5] 9.74 6.01 [9.2, 10.2]
SSS 19.50 17.49 [16.7, 22.2] 23.73 21.30 [21.7, 25.7] 22.65 20.47 [21.0, 24.2]

Notes: CI = confidence interval; TSS = total symptom score; SSS = symptom severity score.

Table 2.

Descriptive statistics for psychological factors

Predictor variable Males (N= 154)

Females (N = 449)

Combined (N = 603)

M SD Min Max M SD Min Max M SD Min Max
Stress (Max score = 16) 5.76 2.99 13 6.83 3.18 16 6.55 3.17 16
Physical fatigue (Max score = 5) 2.57 0.55 3.8 2.51 0.55 1.2 3.8 2.52 0.55 3.8
Mental fatigue (Max score = 5) 1.95 1.15 2.16 1.17 2.11 1.17
Anxiety (Max score = 21) 6.86 3.53 16 8.20 3.89 19 7.86 3.84 19
Depression (Max score = 21) 3.90 3.00 12 4.38 3.25 17 4.25 3.19 17
Predictor variable Males (N= 154)

Females (N = 449)

Combined (N = 603)

M SD Min Max M SD Min Max M SD Min Max
Stress (Max score = 16) 5.76 2.99 13 6.83 3.18 16 6.55 3.17 16
Physical fatigue (Max score = 5) 2.57 0.55 3.8 2.51 0.55 1.2 3.8 2.52 0.55 3.8
Mental fatigue (Max score = 5) 1.95 1.15 2.16 1.17 2.11 1.17
Anxiety (Max score = 21) 6.86 3.53 16 8.20 3.89 19 7.86 3.84 19
Depression (Max score = 21) 3.90 3.00 12 4.38 3.25 17 4.25 3.19 17

Notes: Min = minimum; Max = maximum.

Table 3.

Frequencies and percentages for lifestyle factors

Predictor variable Males (N= 154)

Females (N= 449)

Combined (N = 603)

Yes
n (%)
No
n (%)
Yes
n (%)
No
n (%)
Yes
n (%)
No
n (%)
Physical activity status 83 (53.9) 71 (46.1) 197 (43.9) 252 (56.1) 280 (46.4) 323 (53.6)
Trouble sleeping 57 (37.0) 97 (63.0) 176 (39.2) 273 (60.8) 233 (38.6) 370 (61.4)
Alcohol consumption 31 (20.1) 123 (79.9) 48 (10.7) 401 (89.3) 79 (13.1) 524 (86.9)
Caffeine consumption 90 (58.4) 64 (41.6) 232 (51.7) 217 (48.3) 322 (53.4) 281 (46.6)
Predictor variable Males (N= 154)

Females (N= 449)

Combined (N = 603)

Yes
n (%)
No
n (%)
Yes
n (%)
No
n (%)
Yes
n (%)
No
n (%)
Physical activity status 83 (53.9) 71 (46.1) 197 (43.9) 252 (56.1) 280 (46.4) 323 (53.6)
Trouble sleeping 57 (37.0) 97 (63.0) 176 (39.2) 273 (60.8) 233 (38.6) 370 (61.4)
Alcohol consumption 31 (20.1) 123 (79.9) 48 (10.7) 401 (89.3) 79 (13.1) 524 (86.9)
Caffeine consumption 90 (58.4) 64 (41.6) 232 (51.7) 217 (48.3) 322 (53.4) 281 (46.6)

### Modeling Results for the TSS

Alcohol consumption (p < .001) and trouble sleeping (p < .001) were the strongest predictors in the model for TSS followed by mental fatigue, stress, anxiety, and depression (Table 4). On average, there was an increase of 2.75 units of the TSS for participants who reported consumption of alcohol in the last 24 h compared with non-consumers. Gender, caffeine consumption, physical activity, and physical fatigue were significant in the univariate model; however, they did not contribute to the stepwise multiple linear regression analyses.

Table 4.

Multiple regression analysis predicting the total symptom scores

Predictor variable Estimate 95% CI
Intercept −1.75 [−3.36, 0.15]
Trouble sleepinga 2.27*** [1.48, 3.05]
Alcohol consumptiona 2.75*** [1.44, 4.07]
Mental fatigue 1.54*** [1.13, 1.95]
Stress 0.59*** [0.33, 0.84]
Anxiety 0.57*** [0.34, 0.80]
Depression 0.44*** [0.28, 0.59]
Stress × anxiety −0.05*** [−0.07, −0.02]
Trouble sleeping × alcohol consumption −2.73* [−4.86, −0.60]
Predictor variable Estimate 95% CI
Intercept −1.75 [−3.36, 0.15]
Trouble sleepinga 2.27*** [1.48, 3.05]
Alcohol consumptiona 2.75*** [1.44, 4.07]
Mental fatigue 1.54*** [1.13, 1.95]
Stress 0.59*** [0.33, 0.84]
Anxiety 0.57*** [0.34, 0.80]
Depression 0.44*** [0.28, 0.59]
Stress × anxiety −0.05*** [−0.07, −0.02]
Trouble sleeping × alcohol consumption −2.73* [−4.86, −0.60]

Notes: N= 603. CI = confidence interval; $Radj2=0.47$ (p < .001).

aNo = (reference category), yes.

*p < .05, ***p < .001.

Two interaction terms were found to be significant in the model for TSS. Plots for interaction revealed a positive relationship of stress on TSS at lower values of anxiety and a negative correlation for higher values of anxiety (Fig. 1a). A negative correlation between trouble sleeping and the TSS was identified for respondents who reported having consumed alcohol (Fig. 1b).

Fig. 1.

Plots for the interacting predictors. (a) Two-way interaction plot between stress and anxiety for the TSS. (b) Two-way interaction plot between trouble sleeping and alcohol consumption for the TSS. (c) Two-way interaction plot between depression and mental fatigue for the SSS.

Fig. 1.

Plots for the interacting predictors. (a) Two-way interaction plot between stress and anxiety for the TSS. (b) Two-way interaction plot between trouble sleeping and alcohol consumption for the TSS. (c) Two-way interaction plot between depression and mental fatigue for the SSS.

### Modeling Results for the SSS

On average, the SSS was higher for respondents who reported to have had trouble sleeping (Estimate = 5.82, p < .001) the previous night compared with those who did not (Table 5). Gender, caffeine consumption, stress, physical activity, and physical fatigue predictors were found to be significant in the univariate model; however, they were non-significant when included in the stepwise multiple linear regression model. Further exploration of the significant interaction in this model revealed depression to be positively associated with SSS for both lower and higher values of mental fatigue (Fig. 1c).

Table 5.

Multiple regression analysis predicting the symptom severity scores

Predictor variable Estimate 95% CI
Intercept −1.71 [−5.54, 2.10]
Trouble sleepinga 5.82*** [3.37, 8.27]
Alcohol consumptiona 5.55** [2.15, 8.94]
Mental fatigue 3.61*** [1.72, 5.49]
Anxiety 0.77*** [0.36, 1.18]
Depression 0.28 [−0.62, 1.20]
Depression × mental fatigue 0.57*** [0.28, 0.86]
Predictor variable Estimate 95% CI
Intercept −1.71 [−5.54, 2.10]
Trouble sleepinga 5.82*** [3.37, 8.27]
Alcohol consumptiona 5.55** [2.15, 8.94]
Mental fatigue 3.61*** [1.72, 5.49]
Anxiety 0.77*** [0.36, 1.18]
Depression 0.28 [−0.62, 1.20]
Depression × mental fatigue 0.57*** [0.28, 0.86]

Notes: N= 603. CI = confidence interval; $Radj2=0.51$ (p < .001).

aNo = (reference category), yes.

**p < .01, ***p < .001.

### Cross-Validation of the Models for the TSS and SSS

Cross-validation of the models revealed that both the TSS and SSS models achieved high prediction accuracy with the shrunken values of R2 of 0.46 and 0.50, respectively. As these R2 values obtained through cross-validation of models were very close to that of the values of adjusted R2 (0.47 and 0.51) estimated from our study sample, it can be concluded that the predictive power of the TSS and SSS models is high. Therefore, with this high predictive ability of both the TSS and SSS models, it is most likely that the findings from our study could be generalizable to other future samples.

## Discussion

This study explored the contribution of psychological and lifestyle factors on the reporting of postconcussion-like symptoms in a cohort of university students. We found that the reporting of postconcussion-like symptoms was influenced by a number of psychological and lifestyle factors. Specifically, six and four predictor variables contributed to the TSS and SSS, respectively.

The findings from our study that trouble sleeping the previous night resulted in increased symptoms is in accordance with recent studies, which investigated the effect of sleep quality (Mihalik et al., 2013) and quantity (McClure, Zuckerman, Kutscher, Gregory, & Solomon, 2013; Mihalik et al., 2013) on symptom reporting. These studies postulated that university students tend to sleep less due to their academic demands, expectations and social lives. These same reasons most likely apply to the findings in our study.

The significant interaction between trouble sleeping and alcohol consumption for TSS found in this study reflects their close relationship. This association has also been seen where the sleep schedules of college students were affected due to frequent alcohol consumption in a typical week (Singleton & Wolfson, 2009). Although not measured directly in our study, another study conducted at the same university with a similar cohort identified that many of the students were moderate to heavy drinkers of alcohol (Polak & Conner, 2012). Thus, it is possible that individuals who consumed alcohol in the previous 24 h period were likely to endorse a greater number and/or increased severity of symptoms. However, this associative inference is very difficult to ascertain, because to our knowledge, this is the first study which has explored the relationship between alcohol consumption and the reporting of postconcussion-like symptoms.

The predictors of mental fatigue, stress, depression, and anxiety also contributed to symptom reporting. It is possible that the participants in our study may have experienced mental health problems (e.g., depression) due to pressures associated with their university study and student lifestyle. In a recent report, Gallagher (2012) noted that the number of students seeking medical help for mental health problems has increased in recent years and attributed this to family issues, relationship, and financial problems. In addition, other factors that have been identified as contributing to mental health problems in the student population are being female and aged between 18 and 34 years (Stallman, 2010), the predominant demographics of our sample (females = 449, age = 18–30 years). Another reason for increased levels of stress, anxiety, and depression in university students may be due to their living away from traditional social support networks such as family and friends (Kontos, Covassin, Elbin, & Parker, 2012).

The findings for SSS in this study do not concur with those of Edmed and Sullivan (2012), who used a similar multivariate approach where they concluded that both stress and depression each predicted the SSS. However, these authors used the BC-PSI (16-items) which is different in content to the SCAT2 (22-items) postconcussion symptom scale (McCrory et al., 2009). Moreover, they did not test all possible two-way interactions between the predictors in the regression model, thus, making direct comparisons between studies difficult. In contrast, our study included all possible two-way interactions between the predictors and also explored the post hoc probing of the resultant interactions for both the TSS and SSS.

The results from this study do not support previous studies that found gender (Covassin et al., 2006), physical activity/exercise (Alla et al., 2010), and physical fatigue (Piland et al., 2010) to act as modifiers of symptom reporting. While these predictor variables were significant in the univariate linear regression analyses, they were not significant when included in the multiple linear regression models. The fact that gender did not contribute to either of the models (TSS & SSS) may be due to the disproportionate sample size, where there were a greater number of females (n= 449) compared with males (n= 154). The explanation for these variables not contributing to the TSS and SSS could be due to the fact that this study adopted a conservative approach of using BIC to derive parsimonious models for the TSS and SSS. Specifically, using BIC to deduce a final model would result in the removal of a maximum number of predictor variables (Moreno & Girón, 2006). Moreover, other studies (Alla et al., 2010; Covassin et al., 2006; Piland et al., 2010) applied less robust statistical methods; however, conducted in keeping with their aims. In addition, the exact reasons for these differences are difficult to ascertain due to heterogeneity in methodologies (e.g., aims, study designs, and statistical methods) between this and the earlier studies.

Although this study took a pragmatic approach to obtain data from participants in a real-world environment to investigate the influence of factors on the reported symptoms, it is not without limitations. We included all possible two-way interactions between predictors in the linear regression models to further explore and understand the study findings when the predictors are closely related. However, we acknowledge the fact that in statistical parlance the main effects of the predictors have to be interpreted with caution due to the presence of its interactive terms. We did not conduct a formal and rigorous validation process for the questions that were custom developed to address the specific aims of this study, thus they may lack a certain degree of rigor in their use. The predictor variable of alcohol consumption was captured only as a “key variable” rather than a “construct” and so other measures of alcohol use were not used. A question remains unanswered whether a participant who consumed one drink in the past 24 h had the same moderation effect on symptom reporting as oppose to a person who consumed several drinks. Therefore, the findings related to this variable are limited due to a lack of information on the level of alcohol use (i.e., quantity and frequency), which is sufficient to moderate symptom reporting.

The reporting of postconcussion-like symptoms was influenced by a range of psychological and lifestyle factors. Therefore, the current practice of using symptom scores as a part of concussion assessment and management should be exercised with caution considering the influence of these factors on the reporting of symptoms. An individual should be screened for both psychological and lifestyle factors which are likely to impact symptom reporting as a failure to do so might result in a spurious interpretation of symptom scores. Specifically, there is a possibility that athletes may be allowed to prematurely RTP with symptoms still existing. On the contrary, they may be withheld from participating in the sport despite the resolution of symptoms. The next step in the direction of this study would be to determine whether the associations between psychological/lifestyle factors and the symptoms reported in the early postconcussion phase differ (positively or negatively) while making decisions on RTP.

## Conflict of Interest

P.M. is a co-investigator, collaborator, or consultant on grants relating to mild TBI funded by several governmental organisations. He is Co-Chair of the Australian Centre for Research into Sports Injury and its Prevention (ACRISP), which is one of the International Research Centres for Prevention of Injury and Protection of Athlete Health supported by the International Olympic Committee (IOC). He has a clinical and consulting practice in general and sports neurology. He receives book royalties from McGraw-Hill and was employed in an editorial capacity by the British Medical Journal Publishing Group from 2001 to 2008. He has been reimbursed by the government, professional scientific bodies, and sporting bodies for presenting research relating to mild TBI and sports-related concussion at meetings, scientific conferences, and symposiums. He received consultancy fees in 2010 from Axon Sports (USA) for the development of educational material (which was not renewed) and has received support since 2001 from CogState Inc for research costs and the development of educational material. He is a cofounder and shareholder in two biomedical companies (involved in eHealth and Compression garment technologies) but does not hold any individual shares in any company related to concussion or brain injury assessment or technology.

## Acknowledgements

Our sincere thanks to Mr Bruce Knox, Research Technical Adviser at the Centre for Health, Activity and Rehabilitation Research, School of Physiotherapy, University of Otago, Dunedin, New Zealand for technical development of this web-based survey. We would also like to extend our thanks to the teaching faculty and departmental administrators across the University of Otago for their assistance in the recruitment of participants. Thanks to the Graduate Research Committee of the University of Otago, Dunedin, New Zealand for their contribution in the form of Post Graduate Publishing Bursary (Doctoral) for the preparation of this manuscript.

## References

Alla
S.
,
Sullivan
S. J.
,
McCrory
P.
,
Schneiders
A. G.
,
Handcock
P.
(
2010
).
Does exercise evoke neurological symptoms in healthy subjects?
Journal of Science and Medicine in Sport
,
13
(1)
,
24
26
.
Balasundaram
A. P.
,
Sullivan
S. J.
,
Schneiders
A. G.
,
Athens
J.
(
2013
).
Symptom response following acute bouts of exercise in concussed and non-concussed individuals —A systematic narrative review
.
Physical Therapy in Sport
,
14
(4)
,
253
258
.
Bjelland
I.
,
Dahl
A. A.
,
Haug
T. T.
,
Neckelmann
D.
(
2002
).
The validity of the Hospital Anxiety and Depression Scale. An updated literature review
.
Journal of Psychosomatic Research
,
52
(2)
,
69
77
.
Broglio
S. P.
,
Cantu
R. C.
,
Gioia
G. A.
,
Guskiewicz
K. M.
,
Kutcher
J.
,
Palm
M.
et al
. (
2014
).
National Athletic Trainers’ Association Position Statement: Management of sport concussion
.
Journal of Athletic Training
,
49
(2)
,
245
265
.
Chalder
T.
,
Berelowitz
G.
,
Pawlikowska
T.
,
Watts
L.
,
Wessely
S.
,
Wright
D.
et al
. (
1993
).
Development of a fatigue scale
.
Journal of Psychosomatic Research
,
37
(2)
,
147
153
.
Cohen
S.
,
Williamson
G.
(
1988
).
Perceived stress in a probability sample of the United States
. In
Spacapan
S.
,
Oskamp
S.
(Eds.),
The social psychology of health
(pp.
31
67
).
Newbury Park, CA
:
Sage
.
Covassin
T.
,
Swanik
C. B.
,
Sachs
M.
,
Kendrick
Z.
,
Schatz
P.
,
Zillmer
E.
et al
. (
2006
).
Sex differences in baseline neuropsychological function and concussion symptoms of collegiate athletes
.
British Journal of Sports Medicine
,
40
(11)
,
923
927
.
Dillman
D. A.
,
Smyth
J. D.
,
Christian
L. M.
(
2008
).
Internet, mail, and mixed-Mode surveys: The tailored design method
,
3rd ed
.
Hoboken, NJ
:
Wiley Publishing
.
Edmed
S. L.
,
Sullivan
K. A.
(
2012
).
Depression, anxiety, and stress as predictors of postconcussion-like symptoms in a non-clinical sample
.
Psychiatry Research
,
200
(1)
,
41
45
.
Ettenhofer
M. L.
,
Barry
D. M.
(
2012
).
A comparison of long-term postconcussive symptoms between university students with and without a history of mild traumatic brain injury or orthopedic injury
.
Journal of the International Neuropsychological Society
,
18
(03)
,
451
460
.
Eysenbach
G.
(
2004
).
Improving the quality of web surveys: The checklist for reporting results of internet E-surveys (CHERRIES)
.
Journal of Medical Internet Research
,
6
(3)
,
e34
.
Field
A.
(
2009
).
Discovering statistics using SPSS
,
3rd ed
.
London
:
Sage
.
Fox
D. D.
,
Lees-Haley
P. R.
,
Earnest
K.
,
Dolezal-wood
S.
(
1995
).
Post-concussive symptoms: Base rates and etiology in psychiatric patients
.
The Clinical Neuropsychologist
,
9
(1)
,
89
92
.
Gallagher
R. P.
(
2012
).
National survey of college counseling, 2012. (9T)
.
Alexandria, VA
:
The International Association of Counselling Services Inc
. .
Gasquoine
P. G.
(
2000
).
Postconcussional symptoms in chronic back pain
.
Applied Neuropsychology
,
7
(2)
,
83
89
.
Guskiewicz
K. M.
,
Register-Mihalik
J.
,
McCrory
P.
,
McCrea
M.
,
Johnston
K.
,
Makdissi
M.
et al
. (
2013
).
Evidence-based approach to revising the SCAT2: Introducing the SCAT3
.
British Journal of Sports Medicine
,
47
(5)
,
289
293
.
Iverson
G. L.
(
2012
).
A biopsychosocial conceptualization of poor outcome from mild traumatic brain injury
. In
Vasterling
J. J.
,
Bryant
R. A.
,
Keane
T. M.
(Eds.),
PTSD and mild traumatic brain injury
(pp.
37
60
).
New York
:
Guilford Press
.
Iverson
G. L.
,
Lange
R. T.
(
2003
).
Examination of “postconcussion-like” symptoms in a healthy sample
.
Applied Neuropsychology
,
10
(3)
,
137
144
.
Iverson
G. L.
,
Zasler
N.
,
Lange
R.
(
2007
).
Post-concussive disorder
. In
Zasler
N.
,
Katz
D.
,
Zafonte
R.
(Eds.),
Brain injury medicine: Principles and practice
(pp.
373
406
).
New York, NY
:
Demos Medical Publishing
.
Johansson
B.
,
Starmark
A.
,
Berglund
P.
,
Rodholm
M.
,
Ronnback
L.
(
2010
).
A self-assessment questionnaire for mental fatigue and related symptoms after neurological disorders and injuries
.
Brain Injury
,
24
(1)
,
2
12
.
Kontos
A. P.
,
Covassin
T.
,
Elbin
R. J.
,
Parker
T.
(
2012
).
Depression and neurocognitive performance after concussion among male and female high school and collegiate athletes
.
Archives of Physical Medicine and Rehabilitation
,
93
(10)
,
1751
1756
.
Lovell
M. R.
,
Iverson
G. L.
,
Collins
M. W.
,
Podell
K.
,
Johnston
K. M.
,
Pardini
D.
et al
. (
2006
).
Measurement of symptoms following sports-related concussion: Reliability and normative data for the post-concussion scale
.
Applied Neuropsychology
,
13
(3)
,
166
174
.
McClure
D. J.
,
Zuckerman
S. L.
,
Kutscher
S. J.
,
Gregory
A. J.
,
Solomon
G. S.
(
2013
).
Baseline neurocognitive testing in sports-related concussions: The importance of a prior night's sleep
.
The American Journal of Sports Medicine
,
42
(2)
,
472
478
.
McCrory
P.
,
Meeuwisse
W. H.
,
Aubry
M.
,
Cantu
B.
,
Dvorak
J.
,
Echemendia
R. J.
et al
. (
2013
).
Consensus statement on Concussion in Sport: The 4th International Conference on Concussion in Sport held in Zurich, November 2012
.
British Journal of Sports Medicine
,
47
(5)
,
250
258
.
McCrory
P.
,
Meeuwisse
W. H.
,
Johnston
K.
,
Dvorak
J.
,
Aubry
M.
,
Molloy
M.
et al
. (
2009
).
Consensus statement on Concussion in Sport: The 3rd International conference on Concussion in Sport held in Zurich, November 2008
.
British Journal of Sports Medicine
,
43
(Suppl
.
1)
,
i76
i84
.
Meares
S.
,
Shores
E. A.
,
Taylor
A. J.
,
Batchelor
J.
,
Bryant
R. A.
,
Baguley
I. J.
et al
. (
2008
).
Mild traumatic brain injury does not predict acute postconcussion syndrome
.
Journal of Neurology, Neurosurgery, and Psychiatry
,
79
(3)
,
300
306
.
Mihalik
J. P.
,
Lengas
E.
,
Register-Mihalik
J. K.
,
Oyama
S.
,
Begalle
R. L.
,
Guskiewicz
K. M.
(
2013
).
The effects of sleep quality and sleep quantity on concussion baseline assessment
.
Clinical Journal of Sport Medicine
,
23
(5)
,
343
348
.
Moreno
E.
,
Girón
F. J.
(
2006
).
On the frequentist and Bayesian approaches to hypothesis testing
.
Statistics and Operations Research Transactions
,
30
(1)
,
3
28
.
Mrazik
M.
,
Naidu
D.
,
Lebrun
C.
,
Game
A.
,
Matthews-White
J.
(
2013
).
Does an individual's fitness level affect baseline concussion symptoms?
Journal of Athletic Training
,
48
(5)
,
654
658
.
Piland
S. G.
,
Ferrara
M. S.
,
Macciocchi
S. N.
,
Broglio
S. P.
,
Gould
T. E.
(
2010
).
Investigation of baseline self-report concussion symptom scores
.
Journal of Athletic Training
,
45
(3)
,
273
278
.
Polak
M. A.
,
Conner
T. S.
(
2012
).
Impairments in daily functioning after heavy and extreme episodic drinking in university students
.
Drug and Alcohol Review
,
31
(6)
,
763
769
.
R Core Team
. (
2013
).
R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. From http://www.R-project.org
.
Singleton
R. A. J.
,
Wolfson
A. R.
(
2009
).
Alcohol consumption, sleep, and academic performance among college students
.
Journal of Studies on Alcohol and Drugs
,
70
(3)
,
355
363
.
Stallman
H. M.
(
2010
).
Psychological distress in university students: A comparison with general population data
.
Australian Psychologist
,
45
(4)
,
249
257
.
Stewart
A. L.
,
Hays
R. D.
,
Ware
J. E. J.
(
1992
).
Heath perceptions, energy/fatigue and health distress measures
. In
Stewart
A. L.
,
Ware
J. E.
Jr
(Eds.),
Measuring functioning and well-Being: The medical outcomes study approach
(pp.
155
377
).
Durham, NC
:
Duke University Press
.
Sullivan
K. A.
,
Edmed
S. L.
(
2011
).
An examination of the expected symptoms of postconcussion syndrome in a nonclinical sample
.
,
27
(4)
,
293
301
.
Valovich McLeod
T. C.
,
Bay
R. C.
,
Lam
K. C.
,
Chhabra
A.
(
2012
).
Representative baseline values on the Sport Concussion Assessment Tool 2 (SCAT2) in adolescent athletes vary by gender, grade, and concussion history
.
The American Journal of Sports Medicine
,
40
(4)
,
927
933
.
Zakzanis
K. K.
,
Yeung
E.
(
2011
).
Base rates of post-concussive symptoms in a nonconcussed multicultural sample
.
Archives of Clinical Neuropsychology
,
26
(5)
,
461
465
.
Zigmond
A. S.
,
Snaith
R. P.
(
1983
).
The hospital anxiety and depression scale
.
Acta Psychiatrica Scandinavica
,
67
(6)
,
361
370
.
Zuckerman
S. L.
,
Solomon
G. S.
,
Forbes
J. A.
,
Haase
R. F.
,
Sills
A. K.
,
Lovell
M. R.
(
2012
).
Response to acute concussive injury in soccer players: Is gender a modifying factor?
Journal of Neurosurgery Pediatrics
,
10
(6)
,
504
510
.

## Author notes

The author to whom proofs/reprints should be addressed.