The Postconcussion Symptom Scale
The Postconcussion Symptom Scale
There were 438 patients who met the inclusion criteria. Thirty-six were excluded because they did not have a completed PCSS form and/or they had conditions that suggested an alternate cause of symptoms, leaving a total of 402 patients for analysis. The mean age of the included patients was 14.27 ± 2.67 yr (Table 1). The majority were male (66.8%), and the most frequently reported mechanism of injury was football (20.7%). Forty-three percent reported a prior history of head injury. The average time between injury and evaluation in the sports medicine clinic was approximately 3 wk.
Three factors showed an eigenvalue >1, and there was one factor before the "elbow" of the scree plot, indicating one dominant factor and two additional potential factors among the 19 items. Variance explained by these three factors was 73%, 16.1%, and 10.9% for factors 1, 2, and 3, respectively. Table 2 shows the factor loading of each item for these three potential factors. Factor 1 consists of eight items that may be described as neurocognitive symptoms, including "balance problems," "dizziness," "drowsiness," "numbness or tingling," "difficulty concentrating," "feeling as if 'in a fog'," "difficulty remembering," and "feeling slowed down." Factor 2 consists of six items that may be described as somatic symptoms, including "headache," "nausea," "sleeping more than usual," "sensitivity to light," "sensitivity to noise," and "trouble falling asleep." Factor 3 consists of four items that may be described as emotional symptoms, including "irritability," "nervousness," "sadness," and "more emotional than usual." "Vomiting" did not have loading >0.30 on any of these three factors and thus was not included in the following CFA.
Results confirmed the unidimensionality of factors 2 and 3 with CFI = 0.985 and 0.999, RMSEA = 0.09 and 0.051, for factors 2 and 3, respectively. Modification indices were all <10, residual correlations were all <0.15, and all but one item had R > 0.30. "Sleeping more than usual" had a borderline R = 0.256. For factor 1, most fit indices met the criteria (CFI = 0.975, modification indices <10, residual correlations <0.15, and R > 0.3) except RMSEA (0.121). Internal consistency of the factors was acceptable, with Cronbach α = 0.88, α = 0.81 and α = 0.82, and item-total correlations ranging from 0.44 to 0.72, from 0.42 to 0.69, and from 0.56 to 0.73 for factors 1, 2, and 3, respectively. On the basis of these fit indices, we concluded that these three factors could be considered as unidimensional and that summation scores could be produced for each factor. Given the relatively high RMSEA of factor 1, however, interpretation of factor 1 summation scores should be used with caution.
Factor 1 (neurocognitive symptoms) and factor 2 (somatic symptoms) were highly correlated, r = 0.78, P < 0.01. Factor 3 (emotional symptoms) was moderately correlated with factor 1 (neurocognitive symptoms) and factor 2 (somatic symptoms), r = 0.54 and 0.53, respectively. Patients were grouped into either ≤14 d since injury (n = 236, 58.71%) or >14 d since injury (n = 166, 41.29%). Scores for factors 1 and 2 did not differ significantly between the two groups. However, patients who were injured >14 d prior had significantly higher (worse) factor 3 scores than those who were injured ≤14 d prior, t = -2.15, P = 0.03. This difference remained significant even after adjusting for preexisting mood disorder (F = 4.25, P = 0.0399) and ADD (F = 9.4, P = 0.0023) but not after adjusting for preexisting anxiety disorder (F = 2.65, P = 0.1044). Patients with anxiety disorders had higher scores on all three factors compared to those without anxiety disorders (F = 5.97, P = 0.015; F = 9.95, P = 0.0017; and F = 17.46, P < 0.0001 for factors 1, 2, and 3, respectively). Scores for all three factors did not differ significantly between patients with and without a history of ADD. However, patients with a history of a mood disorder had significantly higher (worse) scores for factor 3 than those without a mood disorder (F = 8.23, P = 0.0044). Compared to males, females had significantly higher (worse) scores on all three factors (t= -3.34, t = -4.80, and t = -4.49, P < 0.01, for factors 1, 2, and 3, respectively). Patients with one or more previous concussions did not differ significantly on any of the three factors from patients without previous concussions (t = 0.11, t = 0.46, and t = -0.05 for factors 1, 2, and 3, respectively).
Results
Patient Characteristics
There were 438 patients who met the inclusion criteria. Thirty-six were excluded because they did not have a completed PCSS form and/or they had conditions that suggested an alternate cause of symptoms, leaving a total of 402 patients for analysis. The mean age of the included patients was 14.27 ± 2.67 yr (Table 1). The majority were male (66.8%), and the most frequently reported mechanism of injury was football (20.7%). Forty-three percent reported a prior history of head injury. The average time between injury and evaluation in the sports medicine clinic was approximately 3 wk.
EFA
Three factors showed an eigenvalue >1, and there was one factor before the "elbow" of the scree plot, indicating one dominant factor and two additional potential factors among the 19 items. Variance explained by these three factors was 73%, 16.1%, and 10.9% for factors 1, 2, and 3, respectively. Table 2 shows the factor loading of each item for these three potential factors. Factor 1 consists of eight items that may be described as neurocognitive symptoms, including "balance problems," "dizziness," "drowsiness," "numbness or tingling," "difficulty concentrating," "feeling as if 'in a fog'," "difficulty remembering," and "feeling slowed down." Factor 2 consists of six items that may be described as somatic symptoms, including "headache," "nausea," "sleeping more than usual," "sensitivity to light," "sensitivity to noise," and "trouble falling asleep." Factor 3 consists of four items that may be described as emotional symptoms, including "irritability," "nervousness," "sadness," and "more emotional than usual." "Vomiting" did not have loading >0.30 on any of these three factors and thus was not included in the following CFA.
CFA
Results confirmed the unidimensionality of factors 2 and 3 with CFI = 0.985 and 0.999, RMSEA = 0.09 and 0.051, for factors 2 and 3, respectively. Modification indices were all <10, residual correlations were all <0.15, and all but one item had R > 0.30. "Sleeping more than usual" had a borderline R = 0.256. For factor 1, most fit indices met the criteria (CFI = 0.975, modification indices <10, residual correlations <0.15, and R > 0.3) except RMSEA (0.121). Internal consistency of the factors was acceptable, with Cronbach α = 0.88, α = 0.81 and α = 0.82, and item-total correlations ranging from 0.44 to 0.72, from 0.42 to 0.69, and from 0.56 to 0.73 for factors 1, 2, and 3, respectively. On the basis of these fit indices, we concluded that these three factors could be considered as unidimensional and that summation scores could be produced for each factor. Given the relatively high RMSEA of factor 1, however, interpretation of factor 1 summation scores should be used with caution.
ANOVA and Regression Analysis
Factor 1 (neurocognitive symptoms) and factor 2 (somatic symptoms) were highly correlated, r = 0.78, P < 0.01. Factor 3 (emotional symptoms) was moderately correlated with factor 1 (neurocognitive symptoms) and factor 2 (somatic symptoms), r = 0.54 and 0.53, respectively. Patients were grouped into either ≤14 d since injury (n = 236, 58.71%) or >14 d since injury (n = 166, 41.29%). Scores for factors 1 and 2 did not differ significantly between the two groups. However, patients who were injured >14 d prior had significantly higher (worse) factor 3 scores than those who were injured ≤14 d prior, t = -2.15, P = 0.03. This difference remained significant even after adjusting for preexisting mood disorder (F = 4.25, P = 0.0399) and ADD (F = 9.4, P = 0.0023) but not after adjusting for preexisting anxiety disorder (F = 2.65, P = 0.1044). Patients with anxiety disorders had higher scores on all three factors compared to those without anxiety disorders (F = 5.97, P = 0.015; F = 9.95, P = 0.0017; and F = 17.46, P < 0.0001 for factors 1, 2, and 3, respectively). Scores for all three factors did not differ significantly between patients with and without a history of ADD. However, patients with a history of a mood disorder had significantly higher (worse) scores for factor 3 than those without a mood disorder (F = 8.23, P = 0.0044). Compared to males, females had significantly higher (worse) scores on all three factors (t= -3.34, t = -4.80, and t = -4.49, P < 0.01, for factors 1, 2, and 3, respectively). Patients with one or more previous concussions did not differ significantly on any of the three factors from patients without previous concussions (t = 0.11, t = 0.46, and t = -0.05 for factors 1, 2, and 3, respectively).