The total study population included in the statistical analysis consisted of 332 participants (231 female, mean age 47.2 ± 12.0 years). There was no significant difference in age (p = 0.61) or sex (p = 0.38) between the CRPS (n = 184) and control group (n = 148). CRPS characteristics and headache prevalence did not differ between CRPS patients recruited from support groups and CRPS patients recruited from specialized pain centers. The mixed (n = 91) and central (n = 70) phenotypes were more common than the peripheral (n = 25) phenotype. Further sociodemographic, headache and disease characteristics of the study population are summarized in Table 1 and elaborated in more detail in the following paragraphs.

Table 1 Patient’s demographics and clinical characteristicsHeadache characteristics and comparison between patients with CRPS and a control group (H1)

Seventy percent of CRPS patients (n = 129) experienced headaches at least once in the last year. Based on the questionnaire, 32% (n = 60) of the patients were classified to suffer from migraine (58% episodic), 7% (n = 12) to tension type headache (80% episodic) and 3% (n = 6) to TAC (50% < 15 d/m). A single diagnosis could not be made in twenty patients (MIG/TTH: 6% (n = 11); MIG/TAC 4% (n = 7), TTH/TAC: 1% (n = 2)) and in 31 patients, headache characteristic did not correspond to any of these primary headache diagnoses (17%, non-classified”).

Compared to the control group, CRPS patients were more likely to suffer from migraine (OR: 3.23, 95% CI 1.82–5.85), TAC (OR: 8.07, 95% CI 1.33–154.79), or non-classified headaches (OR: 3.68, 95% CI 1.88–7.49) and reported significant higher burden of headache as indicated by the HIT-6 (mean difference: 8.61, 95% CI 7.49–9.73). Patients in the migraine subgroup reported significantly more monthly headache days (12.8 ± 7.7 SD vs. 6.0 ± 4.4 SD, p = 0.002), and consequently, they also exhibited a higher proportion of chronic migraine compared to the control subjects (42.1% vs. 7.1%, p = 0.014).

Significant differences in the same direction were present when comparing sex and age-adjusted data to normative data from the German Headache Consortium Study, which used the same questionnaire (migraine SMR: 1.71, 95% CI 1.32–2.19; TAC SMR: 21.74, 95% CI 8.81–45.21; non-classified SMR: 1.62, 95% CI 1.18–2.18). In addition, CRPS patients were more likely to have experienced any kind of headache (SMR: 1.21, 95% CI 1.01–1.43) and less likely to have TTH (SMR: 0.52, 95% CI 0.28–0.89) or mixed migraine and TTH (SMR: 0.45, 95% CI 0.23–0.80). The control group did not significantly differ from population data, after adjusting for sex and age, see Table 2. Further details on headache characteristics can be found in the supplementary material.

Table 2 Prevalence and 95% confidence intervals of headache phenotypes in the general population and in the current CRPS patientRelationship of headache and disease onset of CRPS (H2)

Based on pathophysiological considerations and insufficient statistical power of the TAC group, headache phenotypes were grouped as MIG/TAC, mixed/unclassifiable, TTH and no headache for further analyses.

ANOVA revealed a significant association of headache phenotype and age at CRPS onset (F(3,173) = 8.24, p < 0.001). Post-hoc test yielded significantly younger age at CRPS onset in patients suffering MIG/TAC than patients without headache (37.2 ± 11.1 SD vs. 46.9 ± 13.4 SD years; − 9.6 years, 95% CI − 5.1 to − 14.2)) and patients with TTH (50.9 ± 7.5 SD years, p = 0.002; − 13.7, 95% CI 9.9–17.5). (Fig. 2).

Fig. 2

Violin plot with boxplot for age at CRPS onset dependent on headache diagnosis. Red dot indicates mean. MIG/TAC combined group of patients with migraine and trigeminal autonomic cephalalgias, TTH tension type headache, **Indicate p < 0.01, *** indicate p < 0.001

Patients with migraine and TAC show more often a central phenotype of CRPS (H3).

Chi-squared test resulted in significant differences between headache and CRPS phenotypes (chi-squared(6) = 36.4, p < 0.001).

Post-hoc analyses showed significantly more central CRPS phenotypes for patients with MIG/TAC (60.6% vs. 37.0% overall, p < 0.001) and less mixed CRPS phenotypes (25.8% vs. 49.5% overall, p < 0.001). Patients with mixed or non-classified headache diagnosis were more likely to have a mixed CRPS phenotypes (76.5% vs. 49.5% Overall, p < 0.001) and less often central CRPS phenotypes (11.8% vs. 37.0% Overall, p < 0.001).

Allodynia differed significantly between headache diagnosis (p = 0.045). A binomial regression analysis also revealed that allodynia was more frequent in CRPS-patients with MIG/TAC (OR: 3.45, 95% CI 1.4–9.4), but not in patients with TTH (OR: 4.85, 95% CI 0.8–92.9) or mixed or non-classified headache (OR: 2.15, 95% CI 0.9–5.7).

CRPS severity, occurrence of anxiety and depression and quality of life differs depending on the headache status in CRPS (H4).

There was a significant association between the severity of CRPS (indicated by the adapted CSS) and headache type (chi-squared (3) = 9.57, p = 0.023). Dunn’s test for multiple comparisons with Holm’s correction resulted in significant differences between mixed/non-classified headache group and patients without headache (6.8 ± 1.2 vs 5.7 ± 2.1, p = 0.012).

Rating of movement and resting pain differed between the groups (movement pain: chi-squared (3) = 9.57, p = 0.023; rest pain: F(3,179) = 4.78, p = 0.003). Patients with concomitant MIG/TAC reported significantly higher pain levels at rest (5.3 ± 2.4 vs 3.8 ± 2.6, p = 0.006) and on movement (7.2 ± 2.1 vs 5.8 ± 2.8, p = 0.01) compared to patients without headache. Patients with mixed or non-classified headaches also reported higher pain levels at rest (5.3 ± 2.0 vs. 3.8 ± 2.6, p = 0.01) compared to patients without headache. Results are summarized in Fig. 3a–c.

Fig. 3

a–c Violin plot with boxplot showing CRPS severity (CSS), movement pain and resting pain dependent on headache status. Red dot indicates mean. MIG/TAC combined group of patients with migraine and trigeminal autonomic cephalalgias, TTH tension type headache. *Indicate p < 0.05, ** indicate p < 0.01

ANOVA revealed a significant association between headache phenotypes and depression (F(3,180) = 7.79, p < 0.001), anxiety (F(3,180) = 7.74, p < 0.001), the VAS (F(3,178) = 4.18, p = 0.007), and quality of life (F(3,169) = 6.26, p < 0.001). The Post-hoc test demonstrated that CRPS patients with MIG/TAC and mixed headache syndromes had a higher HADS-depression score compared to patients without headache (MIG/TAC: 10.2 ± 5.3 vs 6.5 ± 4.1, p < 0.001; mixed/non-classified: 10.08 ± 4.47 vs 6.5 ± 4.1, p < 0.001). In these patients HADS-anxiety score was also higher compared to patients without headache (MIG/TAC: 9.9 ± 4.7 vs 6.4 ± 3.4, p < 0.001; mixed/non-classified: 9.5 ± 4.3 vs 6.4 ± 3.4, p = 0.001).

Patients with MIG/TAC also rated their individual health status significantly lower than patients without headache (43.6 ± 21.5 vs 55.2 ± 20.5, p = 0.009) and reported poorer quality of life (0.35 ± 0.30 vs 0.59 ± 0.30, p < 0.001). Quality of life was also lower in patients with mixed/non-classified headaches (0.42 ± 0.28 vs 0.59 ± 0.30, p = 0.022) than patients without concomitant headache (Fig. 4a–c).

Fig. 4

a–c Violin plot with boxplot showing occurrence of anxiety, depression and quality of life depending on headache status. Higher values in The Hospital Anxiety and Depression Scale (HADS) and in EQ-5D-5L indicating more severe anxiety, depression, and better quality of life respectively. Red dot indicates mean. MIG/TAC combined group of patients with migraine and trigeminal autonomic cephalalgias, TTH tension type headache. *Indicate p < 0.05, ** indicate p < 0.01, *** indicate p < 0.001

When analyzing the patient subgroup with migraine we found a strong correlation of MHD with resting pain (r = 0.52, 95% CI 0.23–0.68), a moderate correlation with movement pain (r = 0.44, 95% CI 0.20–0.63), and HADS depression score (r = 0.44, 95% CI 0.20–0.63), as well as a small correlation with CSS (r = 0.27, 95% CI 0.01–0.49). There was no significant correlation between MHD and HADS anxiety score (r = 0.22, 95% CI − 0.05 to 0.45) or HIT-6 score (r = 0.06, 95% CI − 0.20 to 0.31). (Fig. 5a–d). In addition, it is noteworthy that almost all CRPS patients with chronic migraine (n = 22, one “not sure”) reported allodynia.

Fig. 5

a–d Dot plots with regression line showing correlation of monthly headache days with resting pain, movement pain, CRPS severity, and depression scores in the subgroup of CRPS patients with migraine. Higher values in The Hospital Anxiety and Depression Scale (HADS) indicate more severe depression. Orange line indicates direction of correlation with grey areas around it representing its 95% confidence interval. The vertical dashed line represents the division between episodic headache (less than 15 monthly headache days) and chronic headache (15 or more monthly headache days). Darker dots represent multiple entries with the same values

Effect of therapeutic headache treatment on CRPS (H5)

About twenty percent (n = 16) of participants reported either previous or current use of preventive medication for their headaches. Seven patients used CGRP mAbs for migraine prevention, and six of which indicated a good therapeutic effect on headache symptoms whereas one patient reported no effect. All patients reported a positive effect of CGRP mAbs on CRPS symptoms.

Five patients reported a current or previous intake of topiramate as a preventive therapy (good effect: n = 3, moderate: n = 1; no effect: n = 1). One of these reported a beneficial effect on CRPS due to topiramate. Three patients reported intake of flunarizine or/and metoprolol/bisoprolol and reported moderate effects on headache and lack of effect on CRPS.