Psychological and Lifestyle Factors That Influence the Serial Reporting of Postconcussion-like Symptoms in a Non-concussed Population

Accepted Manuscript Psychological and lifestyle factors that influence the serial reporting of post concussive-like symptoms in a non-concussed population Dr Arun Prasad Balasundaram, PhD, Dr Josie Athens, PhD, Professor Anthony Gerard Schneiders, PhD, Assoc/Prof Paul McCrory, MD, PhD, Professor Stephen John Sullivan, PhD PII:

S1934-1482(17)30131-4

DOI:

10.1016/j.pmrj.2017.01.004

Reference:

PMRJ 1845

To appear in:

PM&R

Received Date: 22 March 2016 Revised Date:

22 January 2017

Accepted Date: 29 January 2017

Please cite this article as: Balasundaram AP, Athens J, Schneiders AG, McCrory P, Sullivan SJ, Psychological and lifestyle factors that influence the serial reporting of post concussive-like symptoms in a non-concussed population, PM&R (2017), doi: 10.1016/j.pmrj.2017.01.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

TITLE Psychological and lifestyle factors that influence the serial reporting of post concussive-like

RI PT

symptoms in a non-concussed population Author Information 1) Dr Arun Prasad Balasundaram, PhD

2) Dr Josie Athens, PhD

M AN U

University of Otago, Dunedin, New Zealand

SC

Centre for Health, Activity and Rehabilitation Research, School of Physiotherapy,

Department of Preventive and Social Medicine, Dunedin School of Medicine, University of Otago, Dunedin, New Zealand

3) Professor Anthony Gerard Schneiders, PhD

Australia

TE D

School of Human, Health & Social Sciences, Central Queensland University, Branyan,

4) Assoc/Prof Paul McCrory, MD, PhD

EP

The Florey Institute of Neuroscience and Mental Health, Melbourne Brain Centre, University of Melbourne, Victoria, Australia

AC C

5) Professor Stephen John Sullivan, PhD Centre for Health, Activity and Rehabilitation Research, School of Physiotherapy, University of Otago, Dunedin, New Zealand 6) Corresponding Author: Dr Arun Prasad Balasundaram Mailing address: Forskningsveien 3A, Harald Schjelderups hus, 0373, Oslo, University of Oslo, Norway, Telephone: 47 228-45397, Email ID: [email protected]

ACCEPTED MANUSCRIPT 1

TITLE

2

Psychological and lifestyle factors that influence the serial reporting of post concussive-

3

like symptoms in a non-concussed population

RI PT

4

5

SC

6

M AN U

7

8

9

13

14

15

16

EP

12

AC C

11

TE D

10

17

18

1

ACCEPTED MANUSCRIPT ABSTRACT

20

Background: Symptoms related to concussion are generally found to be non-specific in

21

nature as they are also reported by non-concussed individuals. What is currently not

22

known is whether the symptoms vary over time, and whether they are also influenced

23

by a multitude of factors.

24

Objective: To investigate the potential influence of psychological, lifestyle, and

25

situational factors on the change in postconcussion-like symptoms reported over seven

26

consecutive days in a cohort of normal individuals.

27

Design: Longitudinal observational study.

28

Setting: Real-world context.

29

Participants: Convenience sample of 180 non-concussed university students were

30

enrolled. Of these, 110 participants only provided data for the entire period of the study.

31

Methods: An experience sampling methodology was employed to document the

32

symptoms reported over time. Stepwise multivariate linear mixed-effects modeling

33

performed to identify the predictors that contribute to the serially reported symptoms.

34

Independent Variables: Gender, time of the day, location, primary activity and type of

35

interactant (person) of the participant, physical activity status, trouble sleeping, alcohol

36

consumption, caffeine consumption, stress, anxiety, depression, mental and physical

37

fatigue and life stressors.

38

Main Outcome Measures: The key outcome measures were the change in total

39

symptom score (TSSchange) and symptom severity score (SSSchange) reported over 7

40

consecutive days.

41

Results: The predictors of location at the time of reporting, physical fatigue (Estimate =

42

- 0.98, p<. 001) and mental fatigue (Estimate = - 0.53, p<. 001) contributed to the

AC C

EP

TE D

M AN U

SC

RI PT

19

2

ACCEPTED MANUSCRIPT TSSchange. Post-hoc analysis of the variable of location at the time of reporting revealed

44

that participants reported increased TSSchange when they were at a café/restaurant

45

compared to flat/university.

46

Conclusions: A number of factors within the context of daily life influenced the

47

postconcussion-like symptoms reported over time. These findings indicate that

48

clinicians need to be cautious when interpreting the serially assessed symptom scores to

49

track the recovery profile of a concussed athlete in order to make decisions on return-to-

50

play. Additional investigation is warranted to examine the change in symptom scores

51

reported over time by concussed individuals considering that this study was conducted

52

in a non-concussed cohort.

53

Key words: Experience sampling methodology; Return-to-play; Real-world context;

54

Serial assessment.

M AN U

SC

RI PT

43

INTRODUCTION

TE D

55

Clinicians frequently use the serial assessment of symptoms to track the recovery of a

57

concussed athlete [1]. The symptoms are expected to resolve spontaneously in 80% of

58

athletes following a concussion within 10 days [2]. Athletes are monitored over this

59

timeframe and sometimes beyond with the premise that the resolution of symptoms

60

reflects changes in the underlying neurological status of a concussive brain injury. This

61

information is used to assist in making decisions on the appropriate time to return-to-

62

play (RTP) or training [3]. However, it is well established that symptoms are non-

63

specific in nature [4], thus changes in the symptom scores of recovering athletes may

64

not necessarily reflect their true neurological status. Moreover, postconcussion-like

65

symptoms are often reported by non-concussed college and high school athletes [5, 6]

66

and normal healthy individuals [7, 8]. Additionally, the reporting of postconcussion-like

AC C

EP

56

3

ACCEPTED MANUSCRIPT symptoms is shown to be influenced by a myriad of factors such as stress and

68

depression [9], sleep problems [1, 10], physical fatigue [11] and anxiety [12].

69

Nevertheless, these previous findings reported, including in recent research [13] have

70

been in cross-sectional studies; therefore, it is still not known whether these same

71

factors would have an impact on the postconcussion-like symptoms when reported

72

consecutively over a period of time. There is also a need to explore the contribution of

73

additional factors such as the location, the primary activity and the person who interacts

74

with the participant at the time of reporting symptoms. This is because associations

75

have been reported between these factors and physical symptoms at the time of

76

reporting [14].

77

A recent study documented symptom scores across different (random) times within the

78

same day (intra-day variability) and concluded that time of the day did not contribute to

79

symptom reporting in concussed individuals [15]. However, it is not known whether the

80

time of day would have an impact on the postconcussion-like symptoms reported across

81

multiple days (inter-day variability) in a non-concussed cohort. Gender differences in

82

the reporting of symptoms in non-concussed individuals have been found in several

83

cross-sectional studies [5, 16, 17, 18], where females reported an increased severity of

84

symptoms (SSS) compared to males. Furthermore, there was an increase in the change

85

of symptom scores reported by females compared to males when tested on only two

86

different occasions (test-retest) seven days apart in a sample of non-concussed

87

undergraduate students [19]. Nonetheless, the influence of gender on the

88

postconcussion-like symptoms reported over an extended period of time in non-

89

concussed individuals is yet to be investigated.

AC C

EP

TE D

M AN U

SC

RI PT

67

4

ACCEPTED MANUSCRIPT The contribution of these factors to the reported symptoms, which are likely to vary on

91

a day-to-day basis, has to be understood within a real-world context [20] as it may

92

depend on the situation and time they are reported. By not accounting for these potential

93

factors when interpreting serially assessed symptoms, there is a possibility that

94

clinicians could assume that a concussed athlete has attained asymptomatic status,

95

thereby leading to a premature return-to-play (RTP) or conversely a decision that would

96

prevent them from returning-to-play. Thus, the purpose of this study was to investigate

97

whether the postconcussion-like symptoms serially reported by non-concussed

98

individuals are influenced by gender, time of the day, and various psychological,

99

lifestyle and situational factors experienced and occurring within the context of daily

SC

M AN U

100

RI PT

90

life.

METHODS

101

Study Design

103

A longitudinal observational design was used, where an experience sampling

104

methodology was adopted [21] and employed to capture postconcussion-like symptoms

105

reported over a 7-day period.

106

Setting

107

The postconcussion-like symptoms were measured over time in a real-world context.

108

The recruitment and study period was between June and October 2012.

109

Subjects

110

One hundred and eighty university students aged 18-30 years, who reported not having

111

an episode of concussion in the previous 3 months, were included in the study.

112

Participants who did not have a mobile phone and access to the internet for the entire

AC C

EP

TE D

102

5

ACCEPTED MANUSCRIPT study period were excluded. The recruitment of participants followed the main strategy

114

of soliciting expression of interest by using the sign-up sheets, where they provided

115

their email address to contact them with study details. Secondary strategies included the

116

advertisement of posters on the notice boards throughout the university, and in the

117

university Facebook page. The Human Ethics Committee of the University of Otago

118

approved this study, and informed consent was obtained from all participants by

119

sending a link containing an electronic consent form to their email address.

120

Measures

121

The panel survey [22] was specifically developed for this study that measured the same

122

constructs from the same cohort over 7 consecutive days. The dependent variable and

123

the factors within the context of daily life (predictor variables) used in this study are

124

detailed below.

125

Dependent variable

126

The SCAT2 post-concussion symptom scale [23] was used, which was the current

127

version at the time of the study. This scale consists of 22-symptom items, each rated on

128

a 7-point Likert scale from 0 to 6, which allows the calculation of two scores, the total

129

symptom score (TSS) and symptom severity score (SSS). The TSS is calculated by

130

counting the number of symptoms endorsed >0 (maximum score = 22) and the SSS is

131

summing up the rating of each of the 22 symptoms (maximum score=132; 22 x 6). This

132

scale appears to be developing as a reference scale due to its widespread use. Although

133

not formally established, researchers have reported this scale to have good reliability as

134

well as content and face validity [24].

AC C

EP

TE D

M AN U

SC

RI PT

113

135

6

ACCEPTED MANUSCRIPT Predictor Variables

137

In order to explore the key factors of stress [25], anxiety and depression [26], physical

138

fatigue [27] and mental fatigue [28], specific and relevant items were extracted from

139

each of these recognised scales to include in the panel survey. These variables were

140

framed in relation to the participant’s behaviour in the previous 2 hours. Additionally,

141

brief custom-designed questions were developed for variables of physical activity status

142

(yes/no), alcohol (yes/no) and caffeine consumption (yes/no) in the previous 2 hours,

143

trouble sleeping the previous night (yes/no), and life stressors (e.g. relationship

144

problems, exam stress) experienced at the time of reporting. Furthermore, the variables

145

of location, primary activity and the type of interactant (e.g. family or friend) of the

146

participant were also included based on the work of previous researchers [21, 29]. These

147

authors have suggested their inclusion, when employing a study design similar to the

148

one used in this study.

149

There may be a certain degree of overlap regarding the relationships between constructs

150

such as stress, anxiety and depression with the items in the postconcussion-like

151

symptoms scale. However, these associations have only been shown in studies using

152

cross-sectional data, and thus, there is a lack of evidence with longitudinal data. Apart

153

from the variables of stress and anxiety, this study also included a range of lifestyle and

154

situational factors. There is no study to date, which has systematically investigated the

155

association between these predictor variables and the serially reported symptoms

156

(repeated measurements).

AC C

EP

TE D

M AN U

SC

RI PT

136

157 158 159

7

ACCEPTED MANUSCRIPT Procedures

161

A pilot testing was undertaken with a small group (n=12) of participants in order to

162

establish the face validity of the panel survey and its ease of use. The procedures of the

163

experience sampling methodology were implemented based on those outlined by

164

Conner and colleagues [30], and this is detailed from hereon. Following the enrolment

165

of the participants, they were arbitrarily but randomly divided into one of six groups

166

with each group comprising 30 participants. The procedure of this type was carried out

167

for practical reasons, as it is realistically only possible to obtain the symptomology data

168

for 30 participants on each day. This is because the primary investigator had to

169

manually send out the text messages and also monitor the symptom data obtained from

170

all 30 participants. For each group, the first day of the study started with the signalling

171

of each participant by sending a text message to their mobile phone at a designated time

172

asking them to check their email to find a link to complete the panel survey.

173

Each participant was selectively sent a text message between 9am and 8pm (NZST), and

174

in order to control for the expected variability associated with the time of day, they

175

received it at semi-random times of the day on all 7 days [31]. Each day was equally

176

divided into three time zones: morning (9.00am to 12.59pm), afternoon (1.00pm to

177

4.59pm) and evening (5.00pm to 8.59pm). For pragmatic reasons, ten participants were

178

assigned to each of the three (i.e. Morning, afternoon and evening) time zones on each

179

day. For example, on day 1, the first 10 participants in a particular group received the

180

text message in the morning zone; the next 10 received it in the afternoon and the last

181

10 in the evening and this order was randomly changed on the subsequent days.

AC C

EP

TE D

M AN U

SC

RI PT

160

182

8

ACCEPTED MANUSCRIPT The participants were asked to complete the panel survey ‘as close in time’ as possible

184

upon receiving the text message on all 7 consecutive days. However, it was decided a

185

priori to accept only those responses that were submitted within 4 hours on the same

186

day they received their text message [30]. Each panel survey consisted of a total of 21

187

questions and the time taken to complete the entire survey was approximately 5 minutes

188

on each occasion. All those participants who met the study criteria on all 7 days were

189

entered into a prize draw to receive one of five $40 NZD iTunes store vouchers.

190

Statistical Analyses

191

A total of 180 participants (6 groups’ x 30 each) were enrolled in this study; however,

192

only the data of those participants who complied with the a priori criterion were

193

included in the analyses. The symptom score (TSS & SSS) reported on day 1 was

194

subtracted from each of the symptom score (TSS & SSS) reported for days 2 to 7 (e.g.,

195

t1 minus t0, t2 minus t0...... t6 minus t0), where t0 represents day 1, t1 is day 2 and so forth

196

on each day for each participant. The change in postconcussion-like symptoms was

197

computed in this way for each participant for each day. Following this, the descriptive

198

statistics (mean & SD) were calculated from the computed values of change score for

199

each day. In the final step, this calculated average difference in the symptom score,

200

which is expressed as TSSchange and SSSchange respectively was used as the outcome

201

measure for all subsequent analyses.

202

The baseline linear mixed-effects model was established by conducting a series of linear

203

mixed-effects models. Initially, a null model or unconditional linear mixed-effect model

204

(no predictor) was conducted to estimate the intra-class correlation (ICC) which

205

explains the between subject variability and the average correlation between repeated

AC C

EP

TE D

M AN U

SC

RI PT

183

9

ACCEPTED MANUSCRIPT measurements [32]. Furthermore, the estimation of the ICC is also needed to

207

substantiate the necessity to use the linear mixed-effects modelling in any study [32].

208

Following the generation of the null model, a random intercept model with the inclusion

209

of the variable of time (days) was conducted, followed by a random slope model with

210

the inclusion of the same variable of time (days). Once the random slope model was

211

obtained, an auto-regressive covariance structure was added due to the correlation of

212

repeated measurements. Additionally, in the subsequent random slope model with the

213

already existing covariance structure, heteroscedasticity was accounted for decreasing

214

variances over time. Thus, a baseline linear mixed-effects model was derived through

215

these series of steps.

216

During the process of establishing a baseline linear mixed-effects model, the model

217

comparisons for a series of linear mixed-effects models were based on the information

218

criterion of -2 log likelihood which follows a Chi-Squared distribution [33]. Maximum

219

likelihood was used as an estimation method in all models leading to the establishment

220

of a baseline linear mixed-effects model [33]. Using the baseline model, a univariate

221

linear mixed-effects model was conducted with a threshold value of p < 0.20 set to

222

identify the significance of the fixed-effects component of each predictor variable.

223

This initial univariate linear mixed-effect analysis was conducted in order to filter the

224

predictors to be included in the subsequent multiple linear mixed-effects analyses. All

225

the predictor variables were included in the model as time-varying covariates with the

226

exception of gender, which was added as a time-invariant variable.

AC C

EP

TE D

M AN U

SC

RI PT

206

227

10

ACCEPTED MANUSCRIPT Following the univariate model, a stepwise multiple linear mixed-effects model with a

229

backward elimination method [34] was performed. A level of significance was set at p <

230

0.05 to identify the individual predictor contribution to the fixed-effects component of

231

the multivariate model. A post-hoc analysis was performed for any significant

232

categorical predictor variable by adjusting the confidence intervals using the Westfall

233

method [35]. All the data analyses were performed using the R statistical software,

234

version 3.0.2 [36]. RESULTS

235

SC

RI PT

228

Of the 180 participants recruited, only 110 participants (26 males and 84 females) met

237

the a priori criterion for inclusion in the analyses. A total of 70 participants did not

238

comply with the study criterion of responding within 4 hours from the time of text

239

message. Participants were excluded from analyses for not responding within 4 hours

240

(n=12), albeit on the same day, completed the following day (n=29) or did not complete

241

the questionnaire at all (n=29). The details of frequencies for both the compliant and

242

non-compliant participants are provided in a supplementary table (Table S1).

243

The mean age of all participants was 20.4 ± 2.3. The average change in the TSS and

244

SSS for all participants was –1.7 (SD=4.0) & –3.6 (SD=12.6), for males it was –1.62

245

(SD=3.62) & –4.08 (SD=11.3), and for females –1.75 (SD=4.15) & –3.48 (SD=13.0).

246

The descriptive results for all predictors that include both the categorical and continuous

247

variables obtained on the first day (baseline=t0) are presented as supplementary tables

248

(Tables S2-S4).

249

Univariate linear mixed-effects results for the TSSchange and SSSchange

250

The results for each predictor variable modelled in the univariate linear mixed-effects

251

analyses for the TSSchange and SSSchange are presented in Table 1. On examination of the

AC C

EP

TE D

M AN U

236

11

ACCEPTED MANUSCRIPT results, the predictor variables of gender, time of the day, type of interactant (person),

253

physical activity status and caffeine consumption were found to be non-significant, thus

254

omitted from further analyses.

255

Multiple linear mixed-effects and post-hoc results for the TSSchange

256

Three predictor variables that include, location of the participant at the time of reporting,

257

and physical and mental fatigue were shown to contribute to the TSSchange (Table 2).

258

Initially, an average of 2.19 unit increase in the TSSchange was found when participants

259

reported while they were at a café/restaurant. A subsequent post-hoc analysis revealed

260

that participants reported increased TSSchange when they were at a café/restaurant

261

compared to flat/college or university (Table 3).

M AN U

SC

RI PT

252

262

Multiple linear mixed-effects and post-hoc results for the SSSchange

264

The predictor variables of anxiety, physical and mental fatigue all contributed to the

265

SSSchange. There was a 6.82 unit average increase in the SSSchange when the participants

266

felt ‘anxious frequently’ (Table 4). Overall, the post-hoc results suggest that the

267

participants reported increased SSSchange when they experienced anxiety ‘a great deal of

268

time’ compared to feeling anxious either only ‘from time to time, but not too often’ or

269

‘only occasionally’ (Table 5).

EP

AC C

270

TE D

263

DISCUSSION

271

This study investigated the influence of factors within the context of daily life, gender

272

and time of the day on the changes in TSS and SSS reported over 7 days. Specifically,

273

the predictor variables of physical and mental fatigue contributed to the changes in

274

symptoms reported over time. This finding may be attributed to academic workloads

275

during university study which involves completing assignments, attending lectures and

12

ACCEPTED MANUSCRIPT laboratories. It has been shown in undergraduate nursing students that increased

277

academic workloads during the course of study led them to report excessive physical

278

and mental fatigue [37]. In certain situations, the cause of fatigue in university students

279

can also be due to sleep deprivation [38, 39]. It must be noted that approximately one

280

third of the participants in this study were recruited during the period leading up to their

281

university examinations.

282

Hence, this could be explained in part that a situation may have been created where the

283

participants had a change in their sleep patterns while approaching end of the semester,

284

which in turn may have led them to be fatigued. However, the results of our study

285

suggest that when participants reported more physical and mental fatigue they simply

286

have endorsed fewer TSS and SSS across time. In this instance, the role of practice

287

effects cannot be ruled out considering that the participants completed the panel survey

288

repeatedly for 7 consecutive days. This could be indicative of either habitual, repeated

289

responding or greater self-awareness and accuracy [40]. Practice/learning effects have

290

also been shown to occur when balance and/or neuropsychological tests were repeated

291

[41, 42].

292

Although this study found mental and physical fatigue to be predictive of symptom

293

change, these symptoms overlap with the 22 items on the SCAT2 symptom scale. This

294

implies that it is possible to rule out concussion as the cause of fatigue. However, it is

295

difficult to establish a relationship between fatigue and concussion symptomatology in

296

case of concussed individuals. More precisely, it is not possible to determine whether

297

the reported fatigue is due to the consequence of the concussion directly or

298

psychological factors, or combination of both.

AC C

EP

TE D

M AN U

SC

RI PT

276

13

ACCEPTED MANUSCRIPT The location of the participant at the time of reporting contributed only to the change in

300

TSS reported over time, but not to the SSS. Specifically, the participants reported an

301

increased TSSchange while they were at university as opposed to other locations. These

302

findings were anticipated considering the amount of stress in an academic setting. Our

303

study found that participants demonstrated decreased change in TSS when they were at

304

a flat/residential college, which may be considered a more relaxed environment as they

305

are in the company of their friends and fellow resident/flatmates.

306

It was also identified that the predictor variable of anxiety contributed only to the

307

change in SSS. In university students, anxiety is one of the most common mental health

308

problems [43, 44], and the reasons cited include family issues, financial and relationship

309

problems [45], and the prevalence is highest in females aged between 18-34 years [46].

310

Given the similar demographics of our participants, one or more of these issues may be

311

the reasons that apply to them.

312

The predictor variable that is, time of the day did not contribute to the change scores of

313

TSS and SSS. These findings are similar to the study which found that time of day did

314

not have an influence on the symptoms (SSS) reported [15]. However, direct

315

comparisons cannot be made because the present study recruited non-concussed

316

participants, whereas the previous study [15] involved concussed athletes. Furthermore,

317

they examined the influence of time of day by documenting symptoms (SSS) each day

318

for 30 days, whereas this study obtained symptom data for only 7 days. An explanation

319

for this non-significant finding may be because participants were asked to respond

320

within 4 hours from the time of receiving their text message. Hence, there could have

321

been an overlap in the responses provided by the participants despite sending the text

322

messages out at a pre-defined schedule for each of the three time zones of the day.

AC C

EP

TE D

M AN U

SC

RI PT

299

14

ACCEPTED MANUSCRIPT The gender of the participants did not influence the changes of either the TSS or SSS.

324

These results are contrary to those reported in cross-sectional studies [5, 16, 17, 18],

325

which have found that gender differences exist for the reporting of symptoms.

326

Additionally, Maria et al [19] explored the reporting of postconcussion-like symptoms

327

in non-concussed participants and noted that females had an increase in the change

328

symptom scores compared to males. These contrasting results may be due to different

329

time frames, where this study examined the change in symptoms reported for 7

330

consecutive days, while others [19] investigated the change in symptom score (SSS) at

331

two time points (days 1 & 7). The disparity in the sample (26 males & 84 females) in

332

this study may have also resulted in these findings.

333

While a number of predictor variables were significant in the univariate analyses;

334

however, they were non-significant when included in the multivariate model. This is

335

likely due to the adoption of a conservative approach using the Bayesian information

336

criterion [47], which usually removes a maximum number of predictors to derive a final

337

best model. In general, these findings may be attributed to the different methodology

338

adopted in this study, where data was obtained in a real-world context. Specifically, all

339

participants reported the symptoms for the entire study period at the same time while

340

they were engaged in their routine daily life activities.

341

To the best of our knowledge, this is the first study which examined the influence of

342

factors within the context of daily life on the symptoms reported over a definitive period

343

of time in non-concussed individuals. This is a major contribution to the literature,

344

because this could act as a precursor for future studies to further investigate the same

345

aims followed in this study in persons with a concussive brain injury. In doing so,

346

researchers would be able to identify the factors which they need to account for while

AC C

EP

TE D

M AN U

SC

RI PT

323

15

ACCEPTED MANUSCRIPT interpreting the serially reported symptoms in concussed athletes. Another important

348

strength of this study is the use of a novel approach (experience sampling methodology)

349

to document the symptoms. This allowed all the participants to report their symptoms at

350

the same moment they experienced it, thus reducing any recall bias often associated

351

with reporting at a later time. The experience sampling methodology has been widely

352

used in the field of psychology, but this is the first of its kind in the area of exercise and

353

sports sciences.

354

The generalisability of the study findings may be a concern considering it was

355

conducted using the real-world paradigm. Questions may be raised that the variation

356

and determinants of symptomology are not representative of the typical settings in

357

which clinicians conduct assessment. This is true to a certain degree; however, it must

358

be noted that with the advent of technologies such as smart phones and other similar

359

devices (e.g. iPAD), these days, athletes can be asked to report their symptoms using

360

concussion apps [48], and the symptom reports are subsequently sent to the clinician’s

361

office. These concussion apps are specially designed to be used by athletes, parents and

362

coaches for the ease of obtaining symptoms at a location of their convenience, including

363

when serial (7-10 days and beyond) reporting of symptoms are undertaken to monitor

364

recovery. In such instances, athletes are not always assessed for symptomology in a

365

clinical setting rather in a real-world setting where contextual factors could have an

366

influence on the reporting of symptoms. Thus, these findings could act as a precursor

367

until a study is conducted in athletes with a concussion, for the results to be used in a

368

clinical scenario.

AC C

EP

TE D

M AN U

SC

RI PT

347

369

16

ACCEPTED MANUSCRIPT Although, this study had taken a rigorous methodological approach, it is not without

371

some potential limitations. The data with missing occasions were excluded from the

372

analyses since an a priori decision was to include only those participants who had met

373

the study criterion. In particular, we did not conduct analyses for making comparisons

374

between the included vs excluded participants considering it was not included as one of

375

the aims of the study. It is possible to have yielded different results with the inclusion of

376

missing data considering that mixed-effects models can handle them. However, such an

377

approach would have only resulted in the deviation from complying with the criterion

378

set for the purposes of the study. The present data were highly skewed towards female

379

participants, where the recruitment did not control for this to occur considering the

380

study was designed using a real-world approach. Nonetheless, it is likely that the

381

findings could have been different with the inclusion of equal numbers of male and

382

female participants.

383

Custom-designed questions were developed and utilized for measuring the predictor

384

variables of interest. The rationale for not using standardized questionnaires for these

385

measures was to minimize the burden for the participants. This is likely to occur with

386

the participants having to complete lengthy questionnaires in a real-world context for a

387

period of 7 days. The experience sampling methodological approach adopted in this

388

study followed the guidelines of obtaining data in the real-world [30]. The guidelines

389

[30] state that it is preferable to keep the questions to a minimum number in order to

390

prevent dropout rates, where there is a possibility of participants failing to comply with

391

the study requirements. Nevertheless, we acknowledge that our findings may be limited

392

due to not using standardized/recognized questionnaires. Finally, the lack of a power

393

analysis is also a limitation of the study.

AC C

EP

TE D

M AN U

SC

RI PT

370

17

ACCEPTED MANUSCRIPT CONCLUSION

394

The serially reported symptoms were influenced by a number of factors within the

396

context of daily life in a non-concussed cohort. Therefore, the current practice of serial

397

assessment of symptoms needs to be approached with caution by the clinicians when

398

making decisions on RTP. This is required considering that it could be speculated that

399

the same number of factors identified in this study could also impact the symptom

400

reporting in concussed athletes. In order to investigate this, a study is warranted in

401

future to explore the influence of the same predictor variables used in this study on the

402

serially reported symptoms in a concussed cohort. This study had an adult population;

403

hence there is a need to determine whether the change in symptoms (TSS & SSS)

404

reported over time is different in other population groups (e.g. children, adolescents).

M AN U

SC

RI PT

395

405

ACKNOWLEDGEMENTS

TE D

406

Special thanks to Mr. Bruce Knox, research technical advisor from the School of

408

Physiotherapy, University of Otago, Dunedin, New Zealand for the technical

409

development of the panel survey platform and its associated databases.

411

REFERENCES

AC C

410

EP

407

412

1. McClure DJ, Zuckerman SL, Kutscher SJ, Gregory AJ, Solomon GS. Baseline

413

neurocognitive testing in sports-related concussions: The importance of a prior

414 415

night's sleep. Am J Sports Med 2013; 42:472-478.

2. McCrory P, Meeuwisse WH, Aubry M, et al. Consensus statement on concussion in

416

sport: the 4th International Conference on Concussion in Sport held in Zurich,

417

November 2012. Br J Sports Med 2013; 47:250-25.

18

ACCEPTED MANUSCRIPT 418

3. McCrory P, Johnston K, Meeuwisse WH, et al. Summary and agreement statement

419

of the 2nd international conference on concussion in sport, Prague 2004. Br J Sports

420

Med 2005; 39:196-204. 4. Iverson GL, Brooks BL, Ashton VL, Lange RT. Interview versus questionnaire

RI PT

421 422

symptom reporting in people with the post-concussion syndrome. J Head Trauma

423

Rehabil 2010; 25:23-30.

5. Jinguji TM, Bompadre V, Harmon KG, et al. Sport Concussion Assessment Tool 2:

SC

424

Baseline values for high school athletes. Br J Sports Med 2012; 46:365-370.

426

6. Valovich McLeod TC, Barr WB, McCrea M, Guskiewicz KM. Psychometric and

M AN U

425

427

measurement properties of concussion assessment tools in youth sports. J Athl Train

428

2006; 41:399-408.

430

7. Iverson GL, Lange RT. Examination of "post concussion-like" symptoms in a healthy sample. Appl Neuropsychol 2003; 10:137-144.

TE D

429

8. Sullivan KA, Edmed SL. An examination of the expected symptoms of post-

432

concussion syndrome in a nonclinical sample. J Head Trauma Rehabil 2011;

433

27:293-301.

435 436 437 438

9. Edmed SL, Sullivan KA. Depression, anxiety, and stress as predictors of post concussion-like symptoms in a non-clinical sample. Psychiatry Res 2012; 200:41-45.

AC C

434

EP

431

10. Mihalik JP, Lengas E, Register-Mihalik JK Oyama S, Begalle RL, Guskiewicz KM. The effects of sleep quality and sleep quantity on concussion baseline assessment. Clin J Sport Med 2013; 23:343-348.

439

11. Piland SG, Ferrara MS, Macciocchi SN, Broglio SP, Gould TE. Investigation of

440

baseline self-report concussion symptom scores. J Athl Train 2010; 45:273-278.

19

ACCEPTED MANUSCRIPT 441

12. Iverson GL. A biopsychosocial conceptualization of poor outcome from mild

442

traumatic brain injury. In Vasterling JJ, Bryant RA, Keane TM, eds. PTSD and Mild

443

Traumatic Brain Injury. New York, NY: Guilford Press; 2012, 37-60. 13. Balasundaram, AP, Josie A, Schneiders AG, McCrory P, Sullivan SJ. The influence

RI PT

444 445

of psychological and lifestyle factors on the reporting of postconcussion-like

446

symptoms. Arch Clin Neuropsychol 2016; 31:197-205

449

SC

448

14. Pennebaker JW, ed. The psychology of physical symptoms. New York, NY: Springer-Verlag; 1982.

15. Anthony CA, Peterson AR. Utilization of a text-messaging robot to assess intraday

M AN U

447

450

variation in concussion symptom severity scores. Clin J Sport Med 2015; 25:149-

451

152.

452

16. Covassin T, Swanik CB, Sachs M, et al. Sex differences in baseline neuropsychological function and concussion symptoms of collegiate athletes. Br J

454

Sports Med 2006; 40:923-927.

456 457

fitness level affect baseline concussion symptoms? J Athl Train 2013; 48:654-658. 18. Schneider KJ, Emery CA, Kang J, Schneider GM, Meeuwisse WH. Examining sport concussion assessment tool ratings for male and female youth hockey players with

461

AC C

458

17. Mrazik M, Naidu D, Lebrun C, Game A, Matthews-White J. Does an individual's

EP

455

TE D

453

462

mild head injury status. Arch Clin Neuropsychol 2001; 16:133-140.

459 460

463 464

and without a history of concussion. Br J Sports Med 2010; 44:1112-1117.

19. Maria MPS, Pinkston JB, Miller SR, Gouvier WD. Stability of postconcussion symptomatology differs between high and low responders and by gender but not by

20. Shiffman S, Stone AA, Hufford MR. Ecological momentary assessment. Annu Rev Clin Psychol 2008; 4:1-32.

20

ACCEPTED MANUSCRIPT 465 466 467

21. Mehl MR, Conner TS, eds. Handbook of research methods for studying daily life. 1st ed. New York, NY: Guilford Press; 2011. 22. Dillman DA, Smyth JD, Christian LM, eds. Internet, Mail, and Mixed-Mode Surveys: The Tailored Design Method. 3rd ed. Hoboken, NJ: Wiley Publishing;

469

2008.

470

RI PT

468

23. McCrory P, Meeuwisse WH, Johnston K, et al. Consensus statement on concussion in sport 3rd International conference on concussion in sport held in Zurich,

472

November 2008. Clin J Sport Med 2009; 19:185-200.

24. Valovich McLeod TC, Bay RC, Lam KC, Chhabra A. Representative baseline

M AN U

473

SC

471

474

values on the Sport Concussion Assessment Tool 2 (SCAT2) in adolescent athletes

475

vary by gender, grade, and concussion history. Am J Sports Med 2012; 40:927-933.

476

25. Cohen S, Williamson G. Perceived stress in a probability sample of the United States. In Spacapan S, Oskamp S, eds. The social psychology of health. Newbury

478

Park, CA: Sage; 1988:31-67.

481 482 483 484 485 486 487

Scand 1983; 67:361-370.

EP

480

26. Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr

27. Stewart AL, Hays RD, Ware JEJ. Heath perceptions, energy/fatigue and health distress measures. In Stewart AL, Ware JEJ, eds. Measuring Functioning and Well-

AC C

479

TE D

477

Being: The Medical Outcomes Study Approach. Durham, NC: Duke University Press; 1992:55-377.

28. Chalder T, Berelowitz G, Pawlikowska T, et al.. Development of a fatigue scale. J Psychosom Res 1993; 37:147-153. 29. Hormuth SE. The sampling of experiences in situ. J Pers 1986; 54:262-293.

21

ACCEPTED MANUSCRIPT 488 489 490

30. Conner Christensen T, Barrett LF, Bliss-Moreau E, Lebo K, Kaschub C. A practical guide to experience-sampling procedures. J Happiness Stud 2003; 4:53-78. 31. Kernis MH, Cornell DP, Sun CR, Berry A, Harlow T. There's more to self-esteem than whether it is high or low: the importance of stability of self-esteem. J Pers Soc

492

Psychol 1993; 65:1190-1204.

497 498 499 500 501 502 503 504 505 506 507 508 509 510 511

SC

NY: Routledge; 2010.

M AN U

496

33. Hox JJ, ed. Multilevel analysis: Techniques and Applications. 2nd ed. New York,

34. Cohen J, Cohen P, West SG, eds. Applied multiple regression/correlation analysis for the behavioral sciences. 3rd ed. Mahwah, NJ: Lawrence Erlbaum; 2003. 35. Westfall PH, Young SS, eds. Resampling-Based Multiple Testing: Examples and Methods for P-Value Adjustment. 1st ed. New York, NY: Wiley; 2003.

TE D

495

SPSS. 1st ed. New York, NY: Routledge; 2013.

36. R Core Team (R: A Language and Environment for Statistical Computing). Version 3.0.2. Vienna, Austria: R Foundation for Statistical Computing; 2013. 37. Amaducci Cde M, Mota DD, Pimenta CA. Fatigue among nursing undergraduate

EP

494

32. Heck RH, Thomas SL, Tabata LN. Multilevel and Longitudinal Modeling with IBM

students. Rev Esc Enferm USP 2010; 44:1052-1058. 38. Oginska H, Pokorski J. Fatigue and mood correlates of sleep length in three age-

AC C

493

RI PT

491

social groups: School children, students, and employees. Chronobiol Int 2006; 23:1317-1328.

39. Valpiani EM, Brown RF, Thorsteinsson EB. Poor sleep quality mediates between depression to fatigue in a university student sample. Psychol Educ 2011; 48:59-71. 40. Scollon C, Kim-Prieto C, Diener E. Experience Sampling: Promises and Pitfalls, Strengths and Weaknesses. J Happiness Stud 2003; 4:5-34.

22

ACCEPTED MANUSCRIPT 512

41. Collie A, Maruff P, Darby DG, McStephen M. The effects of practice on the

513

cognitive test performance of neurologically normal individuals assessed at brief

514

test-retest intervals. J Int Neuropsychol Soc 2003; 9:419-428. 42. Valovich McLeod TC, Perrin DH, Guskiewicz KM, Shultz SJ, Diamond R,

RI PT

515

Gansneder BM. Serial administration of clinical concussion assessments and

517

learning effects in healthy young athletes. Clin J Sport Med 2004; 14:287-295.

518

43. Andrews B, Wilding JM. The relation of depression and anxiety to life-stress and

519

SC

516

achievement in students. Br J Psychol 2004; 95:509-521.

44. Said D, Kypri K, Bowman J. Risk factors for mental disorder among university

521

students in Australia: findings from a web-based cross-sectional survey. Soc

522

Psychiatry Psychiatr Epidemiol 2013; 48:935-944.

523

M AN U

520

45. Gallagher RP. National survey of college counseling, [The International Association of Counseling Services, Inc], 2012. Available at

525

http://www.collegecounseling.org/wp-content/uploads/NSCCD_Survey_2012.pdf.

526

Accessed June 6, 2015.

528

general population data. Aust Psychol 2010; 45:249-257. 47. Moreno E, Girón FJ. On the frequentist and Bayesian approaches to hypothesis

532

AC C

529

46. Stallman HM. Psychological distress in university students: A comparison with

EP

527

TE D

524

533

Br J Sports Med 2015; 49:499-505

530 531

testing. SORT 2006; 30:3-28.

48. Lee H, Sullivan SJ, Schneiders AG, et al. Smartphone and tablet apps for concussion road warriors (team clinicians): a systematic review for practical users.

534 535

23

ACCEPTED MANUSCRIPT TABLES

536 537

Table 1

538

Univariate Linear Mixed-Effects Analyses Predicting the TSSchange and SSSchange

RI PT

Univariate linear mixed-effects analysesa TSSchange Predictor variable

p value

SC

p value

SSSchange

.90

.85

Time zones of day

.29

.25

.01

.06

.02

.007

.47

.66

Physical activity status in the last 2 hours

.48

.28

Trouble sleeping the previous night

.001

.02

Alcohol consumption in the last 2 hours

.05

.70

Caffeine consumption in the last 2 hours

.76

.76

Stress

.001

.001

.001

.001

.001

.001

Mental fatigue

.001

.001

Physical fatigue

.001

.001

Life stressors (e.g. exam stress)

.004

.001

Location of the participant Primary activity

EP

Anxiety

TE D

Type of interactant (person)

M AN U

Gender

AC C

Depression

539

Note. TSS=total symptom score; SSS=symptom severity score

540

a

p < .20 was set as a criterion to filter the predictors from the univariate model

541 542

24

ACCEPTED MANUSCRIPT

Multiple Linear Mixed-Effects Analyses Predicting the TSSchange Predictor variable

Estimate

Time (days)

-0.29***

[-0.40, -0.18]

University

1.06**

[0.36, 1.75]

Flat/college

0.38

[-0.29, 1.05]

Café/Restaurant

2.19***

[0.84, 3.55]

M AN U

Location of participanta

Sports/recreation centre

0.30

[-0.83, 1.45]

Other place

4.04***

[3.08, 5.01]

-0.98***

[-1.21, -0.74]

-0.53***

[-0.77, -0.29]

Physical fatigue

TE D

Mental fatigue 545

Note. CI=confidence interval

546

a

547

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

549 550 551

EP

Other place (reference category)

AC C

548

95% CI

RI PT

544

Table 2

SC

543

552 553 554

25

ACCEPTED MANUSCRIPT 555

Table 3

556

Post-hoc Analyses for the Predictor Variable of Location of the Participant Estimate

University vs Other place

1.06 *

[0.11, 2.00]

Flat/college vs Other place

0.38

[-0.53, 1.30]

Café/Restaurant vs Other place

2.19 *

-0.67 *

[-1.29, -0.05]

Café/Restaurant vs University

1.13

[-0.62, 2.90]

Sports/recreation centre vs University

-0.75

[-2.15, 0.64]

Café/Restaurant vs Flat/college

1.81 *

[0.07, 3.55]

Café/Restaurant vs Sports/recreation centre

-1.88

[-1.43, 1.29]

-0.07

[-4.02, 0.24]

TE D

Sports/recreation centre vs Flat/college Note. CI=confidence interval

558

*p < .05

AC C

562

EP

557

561

SC

[-1.25, 1.87]

Flat/college vs University

560

[0.35, 4.04]

0.30

M AN U

Sports/recreation centre vs Other place

559

95% CI

RI PT

Predictor variable

563 564 565

26

ACCEPTED MANUSCRIPT 566

Table 4

567

Multiple Linear Mixed-Effects Analyses Predicting the SSSchange

-0.53**

[-0.86, -0.19]

-0.98

[-2.76, 0.78]

Anxietya From time to time,

A great deal of time A lot of the time Only occasionally Physical fatigue

568

Note. CI=confidence interval

569

a

570

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

572 573 574

1.17

[-1.28, 3.62]

13.21***

[10.02, 16.40]

-2.73 ***

[-3.44, -2.02]

-1.73***

[-2.47, -0.99]

EP

Only occasionally (reference category)

AC C

571

[3.63, 10.01]

TE D

Mental fatigue

6.82***

M AN U

but not too often

95% CI

RI PT

Time (days)

Estimate

SC

Predictor variable

575 576

27

Table 5

578

Post-hoc Analysis for the Predictor Variable of Anxiety Estimate

From time to time, but not too often vs Only occasionally A great deal of time vs Only occasionally

1.17

[-2.00, 4.35]

7.81*** [4.10, 11.52]

TE D

A great deal of time vs From time to time, but not too often

2.16

A lot of the time vs A great deal of time

-5.65*** [-9.15, -2.15]

580

***p < .001

AC C

Note. CI=confidence interval

[-0.46, 4.79]

EP

A lot of the time vs From time to time, but not too often

579

582

[-3.28, 1.30]

6.82*** [2.69, 10.90]

A lot of the time vs Only occasionally

581

-0.98

95% CI

M AN U

Predictor variable

SC

577

RI PT

ACCEPTED MANUSCRIPT

583

28

ACCEPTED MANUSCRIPT

585

Table S1

586

Demographic Comparison of Compliant and Non-Compliant Participants

RI PT

SUPPLEMENTARY TABLES

584

Outside the time

framea

a

588

b

Females

n

n

Group 1

19.7 (2.1)

5

Group 2

19.7 (2.6)

Group 3

Males

Females

Combined

n

n

n

n

20

25

1

4

5

6

13

19

4

7

11

20.4 (2.7)

5

7

12

3

15

18

Group 4

21.5 (2.4)

4

11

15

6

9

15

Group 5

20.5 (2.6)

2

20

22

1

7

8

Group 6

20.7 (1.5)

4

13

17

2

11

13

TE D

M (SD)

Combined

Number of participants who completed the study within 4 hours of sending the text message

AC C

589

Males

EP

587

Age (Years)

frameb

M AN U

Groups

SC

Within the time

Number of participants who did not comply with the study criterion

590

29

ACCEPTED MANUSCRIPT

Table S2

592

Percentages and Frequencies for the Categorical Predictor Variables (2 levels) at Baseline (t0)

593

Participants 594

Females (n=84)

595

Yes

No

Yes

% (n)

% (n)

% (n)

597

Variable 598 599

Physical activity status

26.9 (7)

73.1 (19)

Trouble sleeping

38.5 (10)

Alcohol consumption Caffeine consumption

% (n)

% (n)

76.4 (84)

61.5 (16)

40.5 (34)

59.5 (50)

40.0 (44)

60.0 (66)

0.0 (0)

100.0 (26)

2.4 (2)

97.6 (82)

1.8 (2)

98.2 (108)

26.9 (7)

73.1 (19)

16.7 (14)

83.3 (70)

19.1 (21)

80.9 (89)

TE D

EP AC C

606

No

23.6 (26)

602

605

Yes

77.4 (65)

601

604

% (n)

Combined (N=110)

22.6 (19)

600

603

No

M AN U

596

SC

Males (n=26)

RI PT

591

607

30

ACCEPTED MANUSCRIPT

Table S3

609

Percentages and Frequencies for the Categorical Predictor Variables (>2 levels) at Baseline (t0)

RI PT

608

Participants

65.4 (17) 26.9 (7) 3.8 (1) 3.8 (1)

35.7 (30) 48.8 (41) 4.8 (4) 10.7 (9)

42.7 (47) 43.6 (48) 4.5 (5) 9.1 (10)

53.8 (14) 3.8 (1) 34.6 (9) 3.8 (1) 0.0 (0) 3.8 (1) 0.0 (0)

29.8 (25) 9.5 (8) 11.9 (10) 10.7 (9) 4.8 (4) 2.4 (2) 31.0 (26)

35.5 (39) 8.2 (9) 17.3 (19) 9.1 (10) 3.6 (4) 2.1 (3) 23.6 (26)

AC C

EP

TE D

Location of the participant University/classroom/labs/library Flat/college (residential) Sports/recreation facility Other Primary activity of the participant Studying Watching TV Activity in classroom/laboratory Sports/leisure activities Sleeping Working Other

Females (n=84) Combined (N=110) % (n) % (n)

SC

Males (n=26) % (n)

M AN U

Variable

31

ACCEPTED MANUSCRIPT

615 616 617 618 619 620 621 622 623 624

Other issues None

7.7 (2) 11.5 (3) 0.0 (0) 3.8 (1) 7.7 (2)

3.6 (3) 10.7 (9) 2.4 (2) 10.7 (9) 15.5 (13)

4.5 (5) 10.9 (12) 1.8 (2) 9.1 (10) 13.6 (15)

57.1 (48)

60.0 (66)

26.9 (7)

47.6 (40)

27.3 (30)

19.2 (5) 7.7 (2)

19.0 (16) 11.9 (10)

42.7 (47) 19.1 (21)

46.2 (12)

21.4 (18)

10.9 (12)

42.3 (11) 15.4 (4) 7.7 (2) 34.6 (9)

52.4 (44) 15.5 (13) 7.1 (6) 25.0 (21)

50.0 (55) 15.5 (17) 7.3 (8) 27.3 (30)

69.2 (18)

Anxiety From time to time, but not too often A great deal of time A lot of the time Only occasionally Depression Sometimes Very often Nearly all the time Not at all

RI PT

34.5 (38) 2.7 (3) 19.1 (21) 43.6 (48)

SC

614

36.9 (31) 2.4 (2) 15.5 (13) 45.2 (38)

M AN U

613

26.9 (7) 3.8 (1) 30.8 (8) 38.8 (10)

TE D

612

EP

611

Type of interactant of the participant Friends/acquaintances Co-workers Classmates/flat mates Alone Life stressors Exam related issues Study-related (e.g., assignments) Family issues Health-related

AC C

610

625

32

626

Table S4

627

Descriptive Results for the Continuous Predictor Variables

RI PT

ACCEPTED MANUSCRIPT

Participants

Females (n=84)

Day 1 score Day 7 score Difference

M AN U

Day 1 score Day 7 score Difference

SC

Males (n=26)

scorea M (SD)

-.12 (.15)

3.34 (1.18) 3.55 (1.35)

.21 (.17)

2.99 (1.20) 3.16 (1.35)

.17 (.15)

(Max score=5)

Physical fatigue

M (SD)

2.35 (2.01) 2.23 (2.16)

(Max score=8)

Mental fatigue

M (SD)

628

Note. M=mean; SD=standard deviation.

629

a

Day 7 score minus Day 1 score

2.65 (1.73) 2.65 (1.92) 3.00 (1.41) 3.06 (1.50) 2.48 (1.28) 2.78 (1.43)

Day 1 score Day 7 score Difference

scorea M (SD)

scorea M (SD)

M (SD)

M (SD)

.00 (.19)

2.58 (1.79) 2.55 (1.98)

-.03 (.19)

.06 (.09)

3.08 (1.36) 3.19 (1.44)

.11 (.08)

.30 (.15)

2.60 (1.27) 2.87 (1.42)

.27 (.15)

AC C

(Max score=5)

M (SD)

TE D

Stress

M (SD)

EP

Variable

Combined (N=110)

630

33