Background
For several decades, attention of specialists has been attracted by a group of heterogeneous chronic pain syndromes (fibromyalgia, migraine, irritable bowel syndrome, tension-type headache), which cannot be explained by somatic diseases or damage to nervous system [1]. The American rheumatologist Muhammad Yunus studied these syndromes and defined them as central sensitivity syndromes in English-language literature. Indeed, all these syndromes are characterized by high sensitivity of central nervous system to various stimuli including non-painful one [1—9].
Obviously, impairment of multiple CNS functions (concentration, attention, memory, emotions, sleep and perception) typical for central sensitivity syndromes and combined with impaired pain sensation gives grounds to define them as dysfunctional pain syndromes [10—12]. In dysfunctional pain syndromes, excitation of afferent nociceptive system can occur indirectly. At the same time, light, sound, smells, emotions (i.e. ordinary stimuli) can activate nociceptors and trigger the mechanisms of peripheral and central sensitization (CS) [11]. In clinical practice, these syndromes result appropriate symptoms. Patients experience permanent pain in various parts of the body; they difficultly concentrate. Patients are also absent-minded, forgetful. Sleep is disturbed. These complaints are often combined with autonomic and affective disorders (anxiety and depression). In these circumstances, patients have trouble in normal activities; they need outside assistance and catastrophize pain. Behavior of avoiding of various actions including fear of physical activity develops. All this depresses patients and contributes to impairment of adaptation to usual physical, psychological and social stress.
Various questionnaires have been developed for rapid identifying pain syndromes. Screening disease-specific questionnaires include questionnaires for fibromyalgia (FiRST) and migraine (ID-MigraineTM) [13–15]. Each of these tools contains a minimum number of key statements (6 and 3, respectively) characterizing fibromyalgia and migraine.
General type questionnaires include the Central Sensitization Inventory. This title describes the relationship between central sensitivity syndromes and underlying pathophysiological mechanism, i.e. central sensitization [2, 4, 16–21]. According to the authors, the Central Sensitization Inventory is conceived as “a screening tool to identify patients with key symptoms associated with central sensitization and quantify severity of these symptoms” [4, 22].
Since the Central Sensitization Inventory is designed as a general questionnaire, it is designed to detect symptoms of central sensitivity in various dysfunctional pain syndromes (interstitial cystitis, fibromyalgia, chronic fatigue syndrome, migraine, tension-type headache, multiple hypersensitivity syndrome, temporomandibular joint dysfunction, irritable bowel syndrome, restless legs syndrome, chronic pelvic pain syndrome, etc.) [4]. Filling out the questionnaire, patients emphasize the symptoms persisting for a long time in addition to those arose at the office. Considering these data, physician can timely identify the maximum number of comorbid dysfunctional syndromes and correctly determine treatment strategy.
Psychometric properties of general type questionnaires, including the Central Sensitization Inventory, can depend on the properties of the study group. Therefore, it is important to study the psychometric properties of CSI in various homogeneous groups of patients with dysfunctional pain syndromes, in particular, chronic headache (migraine, TTH and cervicogenic headache) [4, 23, 24]. These forms of headache, especially TTH and migraine, are fairly common in population. They are also often associated with non-specific neck pain. For example, the prevalence of neck pain in patients with migraine is 76.2%, with TTH — 88.4%, with their combination — 89.3% [25]. On the one hand, non-specific neck pain as one of the triggers of migraine can contribute to chronic course of this pain [24–29]. On the other hand, migraine can begin with a neck pain not related to lesion of cervical spine and soft tissue in this area [24, 25].
Central sensitization is essential in pathogenesis of chronic headaches, in particular migraine and tension-type headache [12, 16–19, 22, 25]. Moreover, a recent study revealed that an experimental protocol based on the use of slowly repeated evoked pain (SREP) can also be a marker of CS in episodic migraine [30]. The total CSI score in episodic migraine is higher than in healthy volunteers without headache [31]. In this regard, validation of the Russian-language CSI in patients with headache seems to be relevant. This will allow the validated questionnaire to be used in practical and scientific work.
The purpose of the study was to evaluate the psychometric properties of the Russian-language version of the Central Sensitization Inventory in patients with non-specific neck pain associated with migraine and tension-type headache.
Material and methods
Study patients
Patient groups were formed in accordance with the International Classification of Diseases (10th edition) and diagnostic criteria of the International Classification of Headache Disorders (3rd edition) [32]. The 1st group included patients with non-specific neck pain (ICD-10 M54.2) not accompanied by headache. The 2nd and 3rd groups included patients with non-specific neck pain combined with headaches (migraine and TTH). Inclusion criterion for the 2nd and 3rd groups was episodic and chronic headache, respectively.
Chronic headache was determined by attacks lasted ≥ 15 days per month for > 3 months (ICHD, section 1.3) [32]. Episodic headache was recognized in patients with pain attacks < 15 days per month (ICHD, sections 2.1, 2.2) [32].
We analyzed patients aged 18 — 25 years who signed an informed consent.
Exclusion criteria: 1) neck pain accompanied by symptoms of radiculopathy or neuropathy; 2) neck pain due to specific causes such as trauma, rheumatoid arthritis, metastases, etc. We excluded patients with cancer, structural lesion of nervous system, dementia, severe mental disorders, as well as patients under 18 and over 65 years of age.
Questionnaires
Central Sensitization Inventory
The questionnaire consists of two parts (A and B). Part A includes 25 statements devoted to violation of various general (sleep) and specific (concentration, memory, emotions, etc.) functions of various body systems (nervous, musculoskeletal, urogenital, etc.). These statements make up 25 items of the questionnaire. There are 5 response options for each question: never (0 score), rarely (1 score), sometimes (2 scores), often (3 scores), always (4 scores). The patient is offered to answer each of statement choosing one the most appropriate answer. Scores are summed up, and the total score is calculated. CS severity is assessed as follows: subclinical CS — 0–29 scores, mild — 30–39 scores, moderate — 40–49 scores, severe — 50–59 scores, extreme — 60–100 scores [4].
Hospital Anxiety and Depression Scale
This scale is intended for screening anxiety and depression in inpatients. It includes 14 statements that make up 2 subscales — A (anxiety) and D (depression). Each subscale contains 7 statements. Each statement has 4 possible answers. Each answer is assigned an appropriate score from 0 to 4. After calculating the total score of all subscales, the results are interpreted as follows: normal — 0—7 scores, subclinical anxiety / depression — 8—10 scores, clinically significant anxiety / depression — ≥11 scores.
Content validity
Content validity refers to representativeness of items in relation to the questionnaire concept. To avoid erroneous interpretation of the questionnaire title, one must remember that it does not directly assess CS, but allows one to quantify severity of symptoms typical for chronic dysfunctional pain syndromes [4].
Internal consistency
To assess internal consistency, we analyzed Cronbach's α and inter-item correlation coefficients for the entire questionnaire and each factor identified in factor analysis. Cronbach's α values were interpreted as follows: 0.6<α<0.7 — questionable value, 0.7<α<0.8 — sufficient value, 0.8<α<0.9 — good value, α>0.9 — very good value. Optimal inter-item correlation coefficient does not exceed 0.5 (optimal — 0.35, satisfactory range — 0.2-0.5).
Retest reliability
To assess retest reliability, we analyzed intraclass correlation coefficient (ICC) with a 95% confidence interval (95% CI). Two-sided variance analysis was chosen as a model, ICC2,1 type as intraclass correlation coefficient, ICC2,1 for absolute agreement as a definition for coefficient. ICC2,1 value was interpreted as follows: moderate correlation — 0.50–0.77, good correlation — 0.75–0.90, excellent correlation — >0.90. The interval between the first and second filling of the questionnaire ranged from 2 to 5 days. Evaluating test-retest reliability, we considered parameters of only those patients whose condition remained the same throughout this interval.
Factor analysis
We used 3 methods in factor analysis: 1) selection of the main components; 2) scree plot; 3) maximum likelihood [33–35]. Assuming the presence of a correlation between the selected CSI factors, we applied an oblique rotation (promax). The identified factor had to satisfy the following 3 conditions: 1) inflection point can be easily determined on the graph using scree method; 2) eigenvalue of the factor >1; 3) the factor should explain >10% of variance [34, 35].
Intervals for interpretation of factor loadings were as follows: weak load — 0.32-0.44, moderate load — 0.45-0.54, good load — 0.55-0.62, very good load — 0.63-0 .70, excellent load — > 0.71. Factor loading value of 0.40 was defined as the cut-off point [36, 37].
Minimal detectable changes
Minimal detectable changes (MDC) were calculated using the equation MDC=SEM×√2×1.96, where SEM — standard error of measurement (SEM), 1.96 — coefficient for 95% confidence interval. Standard error of measurement is a reliability index indicating variability of values from measurement to measurement [37]. SEM was calculated using the equation SEM=Sd/√2, where Sd — standard deviation of CSI total score difference obtained by test-retest analysis.
Floor and ceiling effect
To assess the floor and ceiling effect, we analyzed the percentage of responses with a minimum and maximum value for each CSI statement. The floor and ceiling effect is present when more than 15% of respondents choose the answer with minimum (“floor effect”) or maximum (“ceiling effect”) value [37].
Convergent validity
Convergent validity refers to convergence of CSI data and measurements of other questionnaires — the Neck Disability Index (NDI-RU), the Short-Form McGill Pain Questionnaire (SF-MPQ-2), Hospital Anxiety and Depression Scale (HADS). We estimated convergent validity using the Pearson correlation coefficient r (p<0.05). Pearson's r values were interpreted as follows: high correlation — r ≥ 0.5, moderate correlation — r = 0.3–0.5, low correlation — r < 0.3 [38].
Results
A total of 204 patients (173 women and 31 men aged 38.4±10.5 years) were enrolled in the study. Duration of disease averaged 6.2±5.9 years (range 3 months — 30 years). Most of the respondents were working (62.3%) women (84.4%) with higher (68.3%) or secondary specialized (23.8%) education.
The entire sample was divided into 3 groups: patients with non-specific neck pain not accompanied by headache (the 1st group), patients with non-specific neck pain combined with episodic headache (the 2nd group), patients with chronic headache (the 3rd group). Descriptive characteristics of both groups are summarized in Table 1.
Table 1. Descriptive characteristics of the groups (M±SD or Me (Q1; Q3))
|
Variable |
Group |
Significance of between-group difference |
||||
|
1 |
2 |
3 |
1—2 |
1—3 |
2—3 |
|
|
chronic cervicalgia |
chronic cervicalgia + episodic headache |
chronic cervicalgia + chronic headache |
p-value |
|||
|
Number of patients |
37 |
30 |
137 |
|||
|
Female/male ratio, % |
73/27 |
73/27 |
79/21 |
— |
— |
— |
|
Mean age, years |
39.6±11.6 |
37.9±10.6 |
37.8±10.1 |
— |
— |
— |
|
Neck pain, score |
3 (1; 5) |
4 (3; 5) |
4 (3; 6) |
— |
=0.02 |
— |
|
Headache, score |
0 (0; 1) |
3 (2.5; 4) |
5 (3; 7) |
<0.01 |
<0.01 |
— |
|
NDI-RU score |
11±6 |
17±9 |
16±7 |
<0.005 |
<0.001 |
— |
|
CSI score |
32±13 |
35±11 |
45±13 |
— |
<0.0001 |
<0.0003 |
All groups were comparable in age and sex of patients. Neck pain intensity in the 1st group (mild) was lower compared to the 2nd and 3rd groups.
According to inclusion criteria, patients of the 1st group had no headache. Intensity of headache in the 2nd group was lower than in the 3rd group. No significant differences in disability were found between the 2nd and 3rd groups.
Central sensitization was mild and similar in the 1st and 2nd groups. Total CSI score was significantly higher in the 3rd group compared to the 1st and 2nd groups.
NDI-RU score was mild in the 1st group and moderate in the 2nd and 3rd groups. Degree of disability was significantly lower in the 1st group compared to the 2nd and 3rd groups.
Reliability — Internal Consistency
Internal consistency of CSI was good (Cronbach's α = 0.88), inter-item correlation coefficient — 0.24. Distribution of values for each CSI statement and correlation of each statement with other statements of the questionnaire are presented in Table 2.
Table 2. Descriptive data and distribution of responses for CSI-RU and Pearson correlation between item score and total score, n=204
|
CSI item |
Correlation of each statement with other statements of the questionnaire |
Cronbach's α if statement deleted |
|
1. I feel tired and unrefreshed when I wake from sleeping |
0.52 |
0.88 |
|
2. My muscles feel stiff and anchy |
0.47 |
0.88 |
|
3. Anxiety attacks |
0.49 |
0.88 |
|
4. I grind or clench my teeth |
0.40 |
0.88 |
|
5. I have problems with diarrhoea and/or constipation |
0.29 |
0.88 |
|
6. I need help in performing my daily activities |
0.48 |
0.88 |
|
7. I am sensitive to bright lights |
0.42 |
0.88 |
|
8. I get tired very easily when I am physically active |
0.56 |
0.88 |
|
9. I feel pain all over my body |
0.53 |
0.88 |
|
10. I have headaches |
0.43 |
0.88 |
|
11. I have discomfort in my bladder and/or burning when I urinate |
0.29 |
0.88 |
|
12. I do not sleep well |
0.51 |
0.88 |
|
13. I have difficulty concentrating |
0.54 |
0.88 |
|
14. I have skin problems |
0.32 |
0.88 |
|
15. Stress makes my physical symptoms get worse |
0.47 |
0.88 |
|
16. Depression |
0.54 |
0.88 |
|
17. I have low energy |
0.64 |
0.87 |
|
18. I have muscle tension in my neck and shoulders |
0.50 |
0.88 |
|
19. I have pain in my jaw |
0.41 |
0.88 |
|
20. Certain smells make me feel dizzy or nauseated |
0.57 |
0.88 |
|
21. I have to urinate frequently |
0.34 |
0.88 |
|
22. My legs feel uncomfortable |
0.51 |
0.88 |
|
23. I have difficulty remembering things |
0.49 |
0.88 |
|
24. I suffered trauma as a child |
0.34 |
0.88 |
|
25. I have pain in my pelvic area |
0.35 |
0.88 |
We found a good correlation of each CSI statement with the entire questionnaire in most cases.
Competitive validity
Assessing the competitive validity of the CSI, we found a high correlation with NDI-RU (rS=0.57, p<0.05) and the HADS (rS=0.57, p <0.05 for anxiety subscale; rS=0.56, p<0.05 for depression subscale). Moreover, there was a moderate correlation between the CSI and the McGill Pain Questionnaire (SF-MPQ-2) (rS=0.46, p<0.05).
Minimal detectable changes
Minimal detectable changes for the CSI amounted 10 scores.
Floor and ceiling effects for each statement of the questionnaire
We obtained all categories of responses for all statements of the questionnaire. Minimal values expressed by the adverb "never" were obtained for complaints that are not typical for non-specific neck pain accompanied by headache. Nevertheless, there were no “floor and ceiling effects” for the entire CSI.
Retest reliability
Analysis of CSI retest reliability is shown in Table 3. Retest reliability was excellent for the entire questionnaire (ICC=0.91) and moderate-to-good for its items.
Table 3. ICC values for each item and the whole questionnaire, n=204
|
CSI item |
ICC |
95% CI |
|
1. I feel tired and unrefreshed when I wake from sleeping |
0.69 |
0.53—0.80 |
|
2. My muscles feel stiff and anchy |
0.69 |
0.49—0.76 |
|
3. Anxiety attacks |
0.78 |
0.69—0.84 |
|
4. I grind or clench my teeth |
0.84 |
0.78—0.89 |
|
5. I have problems with diarrhoea and/or constipation |
0.85 |
0.79—0.89 |
|
6. I need help in performing my daily activities |
0.81 |
0.75—0.87 |
|
7. I am sensitive to bright lights |
0.82 |
0.74—0.87 |
|
8. I get tired very easily when I am physically active |
0.76 |
0.68—0.83 |
|
9. I feel pain all over my body |
0.71 |
0.61—0.79 |
|
10. I have headaches |
0.74 |
0.65—0.81 |
|
11. I have discomfort in my bladder and/or burning when I urinate |
0.61 |
0.48—0.71 |
|
12. I do not sleep well |
0.70 |
0.60—0.79 |
|
13. I have difficulty concentrating |
0.77 |
0.68—0.83 |
|
14. I have skin problems |
0.86 |
0.78—0.89 |
|
15. Stress makes my physical symptoms get worse |
0.88 |
0.82—0.91 |
|
16. Depression |
0.76 |
0.67—0.83 |
|
17. I have low energy |
0.69 |
0.53—0.79 |
|
18. I have muscle tension in my neck and shoulders |
0.73 |
0.62—0.81 |
|
19. I have pain in my jaw |
0.79 |
0.71—0.85 |
|
20. Certain smells make me feel dizzy or nauseated |
0.82 |
0.75—0.87 |
|
21. I have to urinate frequently |
0.84 |
0.77—0.88 |
|
22. My legs feel uncomfortable |
0.79 |
0.71—0.85 |
|
23. I have difficulty remembering things |
0.74 |
0.65—0.81 |
|
24. I suffered trauma as a child |
0.80 |
0.73—0.86 |
|
25. I have pain in my pelvic area |
0.79 |
0.71—0.85 |
|
Total score |
0.91 |
0.87—0.98 |
Factor analysis
The sample was adequate for exploratory factor analysis (Kaiser-Meyer-Olkin coefficient 0.864, χ2 test 566.15, number of degrees of freedom = 275).
Exploratory factor analysis carried out by the method of principal component selection revealed 8 factors with eigenvalues >1. The scree method indicated a one-factor CSI model: significant decrease of eigenvalues of the factors occurred during transition from the 1st factor (6.20) to the 2nd factor (1.71) (Figure).
Scree plot Exploratory Factor Analyses.
A significant decrease in the eigenvalues takes place during the transition from the 1st factor to the 2nd factor.
In this model, the 1st factor explained 27.5% of the variance. However, 8 out of 25 statements had factor loading < 0.40 (Table 4). These statements reflect complaints that are not typical for patients with neck pain and headache.
Table 4. Factor loadings for 1-factors model of the CSI-RU, n=204
|
CSI item |
Factor loads |
|
1. I feel tired and unrefreshed when I wake from sleeping |
0.603 |
|
2. My muscles feel stiff and anchy |
0.511 |
|
3. Anxiety attacks |
0.526 |
|
4. I grind or clench my teeth |
0.386 |
|
5. I have problems with diarrhoea and/or constipation |
0.280 |
|
6. I need help in performing my daily activities |
0.514 |
|
7. I am sensitive to bright lights |
0.440 |
|
8. I get tired very easily when I am physically active |
0.613 |
|
9. I feel pain all over my body |
0.572 |
|
10. I have headaches |
0.476 |
|
11. I have discomfort in my bladder and/or burning when I urinate |
0.263 |
|
12. I do not sleep well |
0.566 |
|
13. I have difficulty concentrating |
0.588 |
|
14. I have skin problems |
0.322 |
|
15. Stress makes my physical symptoms get worse |
0.521 |
|
16. Depression |
0.611 |
|
17. I have low energy |
0.715 |
|
18. I have muscle tension in my neck and shoulders |
0.528 |
|
19. I have pain in my jaw |
0.394 |
|
20. Certain smells make me feel dizzy or nauseated |
0.584 |
|
21. I have to urinate frequently |
0.330 |
|
22. My legs feel uncomfortable |
0.524 |
|
23. I have difficulty remembering things |
0.519 |
|
24. I suffered trauma as a child |
0.316 |
|
25. I have pain in my pelvic area |
0.337 |
In addition to one-factor model, we also studied the four-factor model proposed by the developers for original CSI. We obtained such a model. However, this model showed a fairly high cross-correlation between the 1st and the 2nd factors (0.565), the 1st and the 4th factors (0.547), the 2nd and the 4th factors (0.524). In our case, the one-factor model turned out to be optimal for describing the structure of CSI.
Discussion
The purpose of our study is to validate the Russian version of the CSI in patients with non-specific neck pain associated with headaches (migraine and TTH). In these groups, the CSI has good psychometric properties (content, construct and convergent validity, retest reliability).
Central sensitization was mild in patients with non-specific neck pain that is consistent with literature data (non-specific neck pain is not characterized by central sensitization) [39].
Patients with chronic headache had moderate central sensitization. This is also consistent with literature data. Indeed, central sensitization is essential in pathogenesis of chronic migraine and TTH [12, 16–19].
Patients with episodic headache had mild central sensitization (higher than in cervicalgia but lower than in chronic headache). This is also consistent with recent (2020) studies: central sensitization occurs in episodic migraine, although not to the same extent as in chronic migraine [30, 31].
We obtained a good internal consistency of the CSI (Cronbach's α = 0.88), that correlates with internal consistency of the original questionnaire and its adapted versions. We also observed an excellent retest reliability of the CSI (ICC2.1=0.91), that is also consistent with the results of other researchers. Evaluation of convergent validity revealed a strong correlation of the CSI with NDI-RU and moderate correlation with the Short-Form McGill Pain Questionnaire.
In our sample, we found a strong correlation between anxiety/depression (HADS) and central sensitization (CSI). This can be explained by the fact that anxiety and depression are comorbid with headache (migraine and TTH) [40–43]. Moreover, central sensitization is essential in pathogenesis of migraine and TTH [12, 16-19, 44-45]. Our findings are justified since our sample included patients with both migraine and TTH.
Various methods of factor analysis applied to assess the CSI structure revealed 2 possible models — multifactorial and unifactorial. Multifactorial model is consistent with the structure of original English-language questionnaire [4], Dutch [45], French [47] and Brazilian [48] versions, as well as Russian version adapted in patients with non-specific neck and back pain [49]. It is known that exploratory factor analysis, which revealed multifactorial model of CSI, is recommended if it is necessary to reduce the number of statements in the questionnaire or to single out separate subscales [36, 50]. However, reducing the number of statements would violate CSI integrity, and allocation of several subscales would hardly be appropriate.
The unifactorial model is more in line with the concept of CSI designed to identify symptoms typical for various dysfunctional pain syndromes [4, 36, 50]. In our case, unifactorial model optimally describes the structure of the questionnaire, which corresponds to its Spanish [36] and German [50] versions.
Conclusion
The validated Russian-language version of the Central Sensitization Inventory showed good psychometric properties in patients with non-specific neck pain accompanied by headache (migraine and tension type headache). This reliable tool can be used in practical work and clinical trials.
The authors declare no conflicts of interest.