Study | Study objective | Study design | Study population | Symptom measure | Psychometric evidence |
McCaffrey (2007)30 | Determine the relation between force caused by head impact and subsequent neurocognitive or balance symptoms | Double-blind, repeated measures | 43 non-concussed college athletes (age mean = 20.7, SD = 1.62)Sport: football | GSC; 18 items | Reliability: not reportedValidity: increase in number of symptoms (but not severity) following 90 g force taken but not “low impact” (<60 g force); 2 symptoms on average. Minimal change in symptoms following >90 g of force was not corroborated by change in balance or cognitionSerial use: two administrations, 2 week interval. No RCI |
Register-Mihalik (2007)31 | Examine the effects of pre-season baseline headaches and post-traumatic headaches on neurocognitive performance | Retrospective repeated measures study | 247 concussed high school and college athletes (age mean = 16.65, sd = 1.87)Sport: mixed | GSC; 20 items | Reliability: not reportedValidity: presence and severity of symptoms correlated with baseline headaches and post-traumatic headachesSerial use: symptom total score and symptom severity was used across post-injury testing times considered relative to headache at baseline |
Piland (2006)32 | Evaluate the factorial validity of a self-reported measure of concussion symptom severity | Cross sectional study of normative cases | 1089 high school athletes (mean age = 16.3, SD = 0.9)Sport: football | GSC; 20 items | Reliability: not reportedValidity: analysis showed 3-factor (eg, somatic, neurobehavioural and cognitive) fit for the 16 items; better 3 factor fit for a model comprised of 9 items. Single second-order factor deemed “concussion symptomotology” foundSerial use: not reported |
McCrea (2003)33 | Prospectively measure immediate effects and natural recovery course relating to symptoms, cognitive functioning and postural stability following sport-related concussion | Prospective cohort with matched case controls. | 1631 college football players (94 injured and 56 controls). Concussion mean age = 20, SD = 1.36, control mean age = 19.20, SD = 1.45Sport: football | GSC; 17 items | Reliability: not reportedValidity: group differences between concussed and non-concussed controls (through 5 days post injury); average symptom resolution 7 days. Symptom recovery curves similar to recovery curves on balance and neurocognitive functionSerial use: GSC administered serially post-injury, with greater group differences persisting through day 5 |
McCrea (2005)34 | Use standard regression-based methods, with baseline and serial testing paradigm to measure individual rates of cognitive/functional impairment following concussion | Prospective study with matched case controls. | 150 college athletes (94 injured and 56 controls) Concussion mean age = 20.04, SD = 1.36; control mean age = 19.20, SD = 1.45Sport: football | GSC; 17 items | Reliability: not reportedValidity: sensitivity and specificity were reported on the GSC immediately following injury (0.89 and 1.00, respectively). Specificity remained at 1.00 through day 7 post-injury; sensitivity declined successively. Symptom recovery generally preceded recovery on balance and cognitive testingSerial use: immediate post-injury, day 7 |
Guskiewicz (2003)35 | Estimate the incidence of concussion and time to recovery after concussion | Prospective longitudinal cohort study; | 2905 college athletes (184 injured)Sport: football | GSC; 17 items | Reliability: not reportedValidity: GSC used serially post-injury, demonstrated linear recoverySerial use: baseline, 7 post-injury (3 hours to 90 days); no RCI |
Iverson & Goetz, (2004)36 | Describe practical considerations for interpreting change following brain injury | Book chapter analying prospective cohort | 200 collegiate football players | PCS; 22 items | Reliability: good internal consistency (Cronbach α = .87) and moderate reliability (test-retest = 0.55) at beginning and end of seasonValidity: interpretation of post- concussion symptom scores must be based on baseline ratesSerial use: pre-season to post-season for test-retest; no RCI |
Lovell (2006)37 | Present psychometric and clinical properties of the PCS | Cross-sectional study with normative and clinical samples | 1746 high school (n = 707) and college (n = 1,039) athletes (260 injured)Sport: unspecified | PCS; 22 items | Reliability: high internal consistency in normal samples (Cronbach α range from 0.88 to 0.94) and for concussed sample (r = 0.93)Validity: demonstrated differences in symptoms between concussed and normalsSerial use: 52 concussed athletes endorsed decreasing levels of symptoms across 3 visits |
McClincy (2006)36 | Serially assess cognitive and symptom-report of concussed athletes | Prospective cohort | 104 concussed high school and college athletes (age mean = 16.11, SD = 2.22)Sport: mixed (football n = 83) | PCS; 22 items | Reliability: not reportedValidity: discriminated between baseline and post-injury on visits 1 and 2 but not 3 (day 14 on average). Symptom report and recovery pattern similar to neurocognitive performanceSerial use: day 2, 7, 14: demonstrated expected pattern of post-injury symptom resolution across visits |
Collins (1999)39 | Prospectively assess premorbid variables with neuropsychological performance and to evaluate post-concussion recovery | Prospective cohort with normative and clinical samples | 393 non-concussed collegiate athletes (age age mean = 20.4, SD = 1.7), with concussed sample of 16Sport: football | PCS; 20 items | Reliability: not reportedValidity: increase in symptom report related to number of previous concussionsSerial use: none |
Collins (2003)40 | Examine relationship between on-field markers of concussion severity and post-injury neuropsychological and symptom presentation | Case control | 78 concussed high school and college athletes (ages = 14–22)Sport: mixed (football n = 64) | PCS; 22 items | Reliability: not reportedValidity: symptom scores similar to good versus poor outcome classification based on neurocognitive performanceSerial use: none |
Field (2003)41 | Evaluate symptoms and neurocognitive recovery patterns | Case control | 92 high school (ages 14–18) and college (ages 17–25) athletes (54 total injured)Sport: football and soccer (football) | PCS (20 item version) | Reliability: not reportedValidity: increased symptoms in concussed sample relative to baseline and controls at 24 hours and 3 days post-injury and at 5 days for the high school athletesSerial use: 1, 3, 5 days; increased symptoms immediately post-injury with gradual decline |
Echemendia (2001) 42 | Evaluate neuropsychological test performance prior to and following mTBI | Prospective cohort with matched controls | 49 college athletes (29 injured)Sport: mixed (football n = 12) | PCS; 22 items | Reliability: not reportedValidity: demonstrated group differences: symptoms reported higher in concussed athletes versus controls at 2 hours post-injury but not at 48 hours post-injury. Agreement with neuropsychological data at 2 hours but not at 48 hoursSerial use: significant differences were found between controls and injured at 2 hours post-injury but not at 48 hours |
Schatz (2006)43 | Identify sensitivity and specificity of ImPACT (including symptom report) for identifying athletes with concussion | Case control cohort | 138 high school athletes (72 injured). Concussed age mean = 16.5 (SD = 2.3), control age mean = 17.3 (SD = 1.7)Sport: mixed (football 73%) | PCS; 22 items | Reliability: not reportedValidity: demonstrated group differences: significant differences between concussed and non-concussed samples. Good sensitivity (82%) and specificity (89%) for identifying individuals with concussion when combined with neurocognitive test composites on ImPACTSerial use: not reported |
VanKampen (2006)44 | Evaluate the individual and combined sensitivity and specificity of player symptom reporting | Case control | 192 high school and college athletes (122 injured age 12 to 27 and 70 controls age 14–22)Sport: mixed (68% football) | PCS; 22 items | Reliability: not reportedValidity: limited change in symptoms post-injury (64% of sample). Neurocognitive testing resulted in a net increase in sensitivity of 19%; conclude that reliance on patients’ self-reported symptoms post-concussion is likely to result in under-diagnosis of concussionSerial use: 64% of concussed athletes reported post-injury an increased symptom score that exceeded reliable change compared with 9% in the control group |
Iverson (2003)45 | Examine stability of test scores, calculate reliable change confidence intervals for test-retest difference scores on ImPACT | Prospective longitudinal cohort | (1) 56 high school and college student-athletes; (2) 41 high school/college student-athletes post-injury | PCS; 22 items | Reliability: test-retest moderate (r = 0.65)Validity: group differences reportedSerial use: preseason, post-concussion. RCI calculated |
Iverson & Goetz (2004)36 | Describe practical considerations for interpreting change following brain injury | Book chapter analysing prospective cohort | 113 athletes (age range 15–20)Sport: junior hockey league | RPCSQ; 16 items | Reliability: good internal consistency (Cronbach α = 0.89) and split-half reliability (r = 0.88). Low test-retest reliability (Pearson r = 0.24, Spearman = 0.23)Validity: interpretation of post- concussion scores must be considered relative to level of baseline reportingSerial use: one-year time period for test-retest |
Wilde (2008)46 | Examine the association between self-reported concussion symptoms and neuroimaging techniques (diffusion tensor imaging tractography) in adolescents with GCS = 15 and normal CT scan | Case control | 10 adolescents ages 14–17 years (mean = 15.7)Sport: 70% of sample was from “low impact” (eg, sports-related, bicycle, fall or assault) | RPCSQ; 16 items | Reliability: not reportedValidity: discriminated between concussed and non-concussed groups; higher post-concussion symptoms in concussed group versus control group (cohen’s d = 1.57). Group differences found on all three subscales (cognitive, affective and somatic) with large effect sizes. Imaging measures not significantly correlated with symptom measureSerial use: none |
GSS, graded symptom scale; ImPACT, immediate post-concussion assessment and cognitive testing; mTBI, mild traumatic brain injury; PCS, post-concussion symptom scale; RPCSQ, Rivermead post-concussion symptom questionnaire; RCI, reliable change indices.