The Role of Fear of Pain in Headache
The Role of Fear of Pain in Headache
Of the 908 retained participants, 382 (42.1%) denied headache (ie, non-headache controls), 237 (26.1%) met diagnostic criteria for episodic TTH (ETTH), 232 (25.6%) for episodic migraine (EM; 167 [18.4%] without aura and 65 [7.2%] with aura), 38 (4.2%) for chronic migraine (CM), and 19 (2.1%) for chronic TTH. Participants with a primary headache disorder reported experiencing headache, on average, 6.7 days per month (SD = 5.3), and mean severity of these headaches was 4.7 out of 10 (SD = 1.7). Headache participants reported a mean score of 52.5 (SD = 9.1) on the HIT-6, with nearly 36% endorsing substantial to very severe headache-related disability. As expected, FOP had significant moderate associations with both anxiety, r = 0.40, P < .001, and depression, r = 0.31, P < .001, and females reported significantly higher FOP on the PASS-20 than males (M total score = 21.8 [15.2] vs 16.3 [11.8], P < .001). As such, anxiety, depression, and gender were used as covariates in subsequent analyses. Table 1 displays means on the DASS-21 scales and the HIT-6 as a function of headache diagnosis.
The 1-way MANOVA on PASS subscales across diagnostic groups yielded a significant omnibus effect, Pillai's V = 0.16, F (20, 3608) = 7.53, P < .001. Given the significance of the overall test, the univariate main effects for headache diagnosis were assessed and are presented in Table 2. Scores on all 4 PASS-20 subscales differed significantly between groups: cognitive anxiety, F (5, 902) = 20.33, P < .001; escape and avoidance, F (5, 902) = 16.17, P < .001; fearful appraisals, F (5, 902) = 12.94, P < .001; and physiological anxiety, F (5, 902) = 24.12, P < .001. Pertinent to hypothesis 1, Tukey post-hoc tests confirmed that migraineurs reported greater FOP than non-headache controls. Those with CM and EM with aura evidenced higher FOP on all PASS-20 subscales than those without headache; episodic migraineurs without aura reported higher anxiety than controls on 2 subscales (cognitive anxiety and escape and avoidance). Individuals with TTH, whether episodic or chronic, however, did not report higher FOP than those without headache. Pertinent to hypothesis 2, those with CM reported significantly higher FOP than those with TTH on all 4 subscales. In addition, chronic migraineurs endorsed higher FOP than episodic migraineurs without aura on all subscales (Ps < .05) except escape/avoidance (P = .052). Compared with those with CM, episodic migraineurs with aura reported similarly high levels of FOP, except on the physiological anxiety subscale (P < .05). Chronic migraineurs differed significantly from TTH sufferers on all 4 subscales (Ps < .05).
The subsequent MANCOVA controlling for gender, depression, and anxiety yielded a significant omnibus effect, Pillai's V = 0.09, F (20, 3488) = 4.06, P < .001, indicating that overall group differences remained after incorporating these covariates. Univariate tests confirmed that scores on all 4 PASS-20 subscales differed significantly between groups: cognitive anxiety, F (5, 872) = 7.80, P < .001; escape and avoidance, F (5, 872) = 8.40, P < .001; fearful appraisals, F (5, 872) = 5.19, P < .001; and physiological anxiety, F (5, 872) = 12.10, P < .001. Specifically, migraineurs reported greater FOP than non-headache controls on all 4 PASS-20 subscales (Ps < .05), and individuals with ETTH differed from controls only on the escape/avoidance subscale (P < .05). In general, the major group differences from the MANOVA remained, except those with CM and EM without aura no longer differed on cognitive anxiety (P = .066), and EM participants with aura scored significantly higher than those without aura on physiological anxiety (P < .05).
Separate linear regression analyses among headache sufferers confirmed that FOP, as measured by the PASS-20 total score, significantly "predicted" headache severity and frequency from the SDIH-R, accounting for 6.1% and 4.5% of variance in these variables, respectively (Ps < .001). In addition, FOP was an even stronger predictor of headache-related disability, accounting for 17.5% (P < .001) of variance in HIT-6 scores. After controlling for gender, anxiety, and depression, FOP remained a significant but smaller unique predictor of headache severity (ΔR = 2.7%; P < .001), frequency (ΔR = 1.0%; P < .001), and disability (ΔR = 8.7%; P < .001).
For the mediation analysis, the effect of headache severity on disability (HIT-6) was assessed directly and indirectly through FOP (ie, PASS-20 total score) using the INDIRECT procedure for SPSS (see Figure). The total effect (path c) indicated that for every 1-point increase in headache severity, HIT-6 scores increased on average by more than 3 points (3.29, P < .001, 95% CI = 2.92–3.65), indicating that individuals who experienced more severe pain also reported greater headache disability. The direct effect (path c') of headache severity on disability indicated that when controlling for FOP, a 1-point increase in headache severity was associated with a 2.90 unit increase in HIT-6 scores (P < .001, 95% CI = 2.54–3.26).
(Enlarge Image)
Figure.
Path coefficients for simple mediation analysis on headache-related disability. Note: c denotes the total effect, while c' denotes the effect of headache severity on disability when fear of pain is not included as a mediator. *P < .001.
The estimate of the indirect effect (ab) of headache severity on disability through FOP was quantified as the product of 2 ordinary least squares (OLS) regression coefficients, one estimating FOP from headache severity (path a) and the other estimating headache disability from FOP while controlling for severity (path b). Path a indicates that a unit increase in headache severity was associated with a 2.21 unit increase in FOP (a = 2.21, 95% CI = 1.46–2.95), and path b shows that a unit increase in FOP was associated with a 0.17 unit increase in disability (b = 0.17, 95% CI = 0.13–0.21). The difference between the total and direct effects (c – c') is the indirect effect (ab) through FOP, which yielded a point estimate of 0.38 (95% CI = 0.23–0.57). This bias-corrected and accelerated bootstrap CI for the indirect effect did not include zero; thus, the null hypothesis that the total indirect effect equals zero was rejected, and the mediation effect was statistically significant. Although a small proportion of the variance in FOP was explained by changes in headache severity (R = 6.1%), a substantial amount of variance in HIT-6 scores was accounted for by both FOP and headache severity (R = 44.5%). These results provide support for the hypothesis that headache severity is associated with headache-related disability partly through its effect on FOP.
Sensitivity Analyses. To address whether inclusion of adults >25 years of age influenced the obtained results, prior analyses were re-run after excluding the 17 individuals ≥ age 26 (1.9% of entire sample). The MANOVA post-hoc tests revealed that those with EM without aura now scored higher than controls on all 4 PASS-20 subscales and were now significantly lower than those with CM on the escape/avoidance subscale. The pattern of major findings thus remained similar and excluding these relatively older participants made the diagnostic group differences in FOP even more striking.
Results
Of the 908 retained participants, 382 (42.1%) denied headache (ie, non-headache controls), 237 (26.1%) met diagnostic criteria for episodic TTH (ETTH), 232 (25.6%) for episodic migraine (EM; 167 [18.4%] without aura and 65 [7.2%] with aura), 38 (4.2%) for chronic migraine (CM), and 19 (2.1%) for chronic TTH. Participants with a primary headache disorder reported experiencing headache, on average, 6.7 days per month (SD = 5.3), and mean severity of these headaches was 4.7 out of 10 (SD = 1.7). Headache participants reported a mean score of 52.5 (SD = 9.1) on the HIT-6, with nearly 36% endorsing substantial to very severe headache-related disability. As expected, FOP had significant moderate associations with both anxiety, r = 0.40, P < .001, and depression, r = 0.31, P < .001, and females reported significantly higher FOP on the PASS-20 than males (M total score = 21.8 [15.2] vs 16.3 [11.8], P < .001). As such, anxiety, depression, and gender were used as covariates in subsequent analyses. Table 1 displays means on the DASS-21 scales and the HIT-6 as a function of headache diagnosis.
The 1-way MANOVA on PASS subscales across diagnostic groups yielded a significant omnibus effect, Pillai's V = 0.16, F (20, 3608) = 7.53, P < .001. Given the significance of the overall test, the univariate main effects for headache diagnosis were assessed and are presented in Table 2. Scores on all 4 PASS-20 subscales differed significantly between groups: cognitive anxiety, F (5, 902) = 20.33, P < .001; escape and avoidance, F (5, 902) = 16.17, P < .001; fearful appraisals, F (5, 902) = 12.94, P < .001; and physiological anxiety, F (5, 902) = 24.12, P < .001. Pertinent to hypothesis 1, Tukey post-hoc tests confirmed that migraineurs reported greater FOP than non-headache controls. Those with CM and EM with aura evidenced higher FOP on all PASS-20 subscales than those without headache; episodic migraineurs without aura reported higher anxiety than controls on 2 subscales (cognitive anxiety and escape and avoidance). Individuals with TTH, whether episodic or chronic, however, did not report higher FOP than those without headache. Pertinent to hypothesis 2, those with CM reported significantly higher FOP than those with TTH on all 4 subscales. In addition, chronic migraineurs endorsed higher FOP than episodic migraineurs without aura on all subscales (Ps < .05) except escape/avoidance (P = .052). Compared with those with CM, episodic migraineurs with aura reported similarly high levels of FOP, except on the physiological anxiety subscale (P < .05). Chronic migraineurs differed significantly from TTH sufferers on all 4 subscales (Ps < .05).
The subsequent MANCOVA controlling for gender, depression, and anxiety yielded a significant omnibus effect, Pillai's V = 0.09, F (20, 3488) = 4.06, P < .001, indicating that overall group differences remained after incorporating these covariates. Univariate tests confirmed that scores on all 4 PASS-20 subscales differed significantly between groups: cognitive anxiety, F (5, 872) = 7.80, P < .001; escape and avoidance, F (5, 872) = 8.40, P < .001; fearful appraisals, F (5, 872) = 5.19, P < .001; and physiological anxiety, F (5, 872) = 12.10, P < .001. Specifically, migraineurs reported greater FOP than non-headache controls on all 4 PASS-20 subscales (Ps < .05), and individuals with ETTH differed from controls only on the escape/avoidance subscale (P < .05). In general, the major group differences from the MANOVA remained, except those with CM and EM without aura no longer differed on cognitive anxiety (P = .066), and EM participants with aura scored significantly higher than those without aura on physiological anxiety (P < .05).
Separate linear regression analyses among headache sufferers confirmed that FOP, as measured by the PASS-20 total score, significantly "predicted" headache severity and frequency from the SDIH-R, accounting for 6.1% and 4.5% of variance in these variables, respectively (Ps < .001). In addition, FOP was an even stronger predictor of headache-related disability, accounting for 17.5% (P < .001) of variance in HIT-6 scores. After controlling for gender, anxiety, and depression, FOP remained a significant but smaller unique predictor of headache severity (ΔR = 2.7%; P < .001), frequency (ΔR = 1.0%; P < .001), and disability (ΔR = 8.7%; P < .001).
For the mediation analysis, the effect of headache severity on disability (HIT-6) was assessed directly and indirectly through FOP (ie, PASS-20 total score) using the INDIRECT procedure for SPSS (see Figure). The total effect (path c) indicated that for every 1-point increase in headache severity, HIT-6 scores increased on average by more than 3 points (3.29, P < .001, 95% CI = 2.92–3.65), indicating that individuals who experienced more severe pain also reported greater headache disability. The direct effect (path c') of headache severity on disability indicated that when controlling for FOP, a 1-point increase in headache severity was associated with a 2.90 unit increase in HIT-6 scores (P < .001, 95% CI = 2.54–3.26).
(Enlarge Image)
Figure.
Path coefficients for simple mediation analysis on headache-related disability. Note: c denotes the total effect, while c' denotes the effect of headache severity on disability when fear of pain is not included as a mediator. *P < .001.
The estimate of the indirect effect (ab) of headache severity on disability through FOP was quantified as the product of 2 ordinary least squares (OLS) regression coefficients, one estimating FOP from headache severity (path a) and the other estimating headache disability from FOP while controlling for severity (path b). Path a indicates that a unit increase in headache severity was associated with a 2.21 unit increase in FOP (a = 2.21, 95% CI = 1.46–2.95), and path b shows that a unit increase in FOP was associated with a 0.17 unit increase in disability (b = 0.17, 95% CI = 0.13–0.21). The difference between the total and direct effects (c – c') is the indirect effect (ab) through FOP, which yielded a point estimate of 0.38 (95% CI = 0.23–0.57). This bias-corrected and accelerated bootstrap CI for the indirect effect did not include zero; thus, the null hypothesis that the total indirect effect equals zero was rejected, and the mediation effect was statistically significant. Although a small proportion of the variance in FOP was explained by changes in headache severity (R = 6.1%), a substantial amount of variance in HIT-6 scores was accounted for by both FOP and headache severity (R = 44.5%). These results provide support for the hypothesis that headache severity is associated with headache-related disability partly through its effect on FOP.
Sensitivity Analyses. To address whether inclusion of adults >25 years of age influenced the obtained results, prior analyses were re-run after excluding the 17 individuals ≥ age 26 (1.9% of entire sample). The MANOVA post-hoc tests revealed that those with EM without aura now scored higher than controls on all 4 PASS-20 subscales and were now significantly lower than those with CM on the escape/avoidance subscale. The pattern of major findings thus remained similar and excluding these relatively older participants made the diagnostic group differences in FOP even more striking.
Source...