ISSN 0017-8748 doi: 10.1111/head.12503 Published by Wiley Periodicals, Inc.
Headache © 2014 American Headache Society
Research Submission Pain Beliefs and Perceptions Inventory: A Cross-Sectional Study in Chronic and Episodic Migraine Cecilia Condello, MD; Virginie Piano, PhD; Daria Dadam, PhD; Lorenzo Pinessi, MD; Michel Lantéri-Minet, MD, PhD
Objective/Background.—This study aims to compare differences in pain beliefs according to headache type, chronic vs episodic migraine, in a large cohort of patients, using the Pain Beliefs and Perceptions Inventory (PBPI), and to identify possible predictive factors of the same pain beliefs. Methods.—All patients referring for the first time at our center in 2011 were screened using PBPI and the Hamilton Anxiety and Depression Scale (a total of 1032 patients). PBPI is a 4-subscale questionnaire that explores a patient’s personal beliefs on their subjective experience of pain. Headache patients also completed the Headache Impact Test (HIT-6) and a 30-day headache diary. For all participants, age, gender, duration of pain were collected. The sample was narrowed down to 899 as we experienced a nonresponse rate of 12.8%. For the purpose of this study, 2 groups were identified: chronic and episodic migraine, consisting of 116 and 126 patients, respectively, which were compared using Student’s t-test; correlation analyses were conducted to investigate the relationship between variables before running a model selection based on Akaike’s Information Criterion to identify possible predictive factors of different pain beliefs. Patients below 18 years of age and those diagnosed with other painful conditions were excluded from the analysis. Results.—Beliefs from chronic and episodic migraine patients were very similar, with only a difference in beliefs related to constancy of pain (Mean value ± SD 0.5 ± 1.1 vs −0.6 ± 1.1, P < .001). Predictive factors were depression and HIT-6 scores for all PBPI subscales apart from Self-Blame, which showed a stronger relation to anxiety scores. Number of days with headache per month was correlated to higher constancy values. Diagnosis was a predictive factor for any particular belief. Discussion.—This is the first study, to our knowledge, that addresses differences and predictive factors in pain beliefs according to headache diagnosis. A deeper knowledge of beliefs pattern in patients could lead to better-tailored psychological management. Key words: Pain Beliefs and Perceptions Inventory, chronic migraine, episodic migraine, depression, disability (Headache 2015;55:136-148)
Address all correspondence to C. Condello, Department of Neuroscience, Head Center, University of Torino, Via Cherasco 15 10100, Turin, Italy, email: cecilia.condello@ gmail.com
Certain pain-related beliefs, like faith in the persistence of pain, seem to impact on coping and compliance, while regarding pain as mysterious and inexplicable is related to worse outcome, psychological distress, and somatization.1-3 It has been shown that headache patients with a low perception of selfefficacy (ability to take actions to influence the course of headache), or convinced that the origin of their headache was due to chance or fate, were more prone
Accepted for publication October 9, 2014.
Disclosures: No funding was received for this study, and the authors declare no conflict of interest.
From the Department of Neuroscience, Headache Center, University of Torino, Turin, Italy (C. Condello and L. Pinessi); Pain Department, CHU Pasteur, Nice, France (V. Piano and M. Lantéri-Minet); Statistics, British Trust for Ornithology, Thetford, UK (D. Dadam).
136
Headache to inadequate coping, with worse outcome on overall disability.4 In light of this, early systematic assessment of pain beliefs was promoted.5 Pain beliefs are defined as “patients’ own conceptualization of what pain is and what pain means for them.”1 The main dimensions of these beliefs are eventuality of pain lifting, the mysterious nature of pain, wonder at the cause of pain, lifestyle altering because of pain, possibility of personal control over pain, blame, and constant vs intermittent nature of pain experience. A useful tool to assess pain beliefs is the Pain Beliefs and Perceptions Inventory (PBPI), a 16-question Likert-type scale addressing 4 dimensions of pain beliefs: seeing pain as mysterious (Mystery), holding oneself responsible for pain (SelfBlame), regarding one’s condition as lingering in the future (Permanence), and/or continuous over time (Constancy). PBPI was initially designed around 3 main cores, the dimensions Mystery, Self-Blame, and Time, but the Australian and German replication of the scale showed an improvement in the method by adding a fourth dimension (Constancy and Permanence instead of Time).6,7 This change provided a further tool to explore disorders not associates with intermittent pain, like headache, in which is likely that Constancy and Permanence are regarded differently by patients.8 In this new version, PBPI was validated in the United States by its developers,9 in England,8 in Norway,10 in France,11 in China,12 and in Italy.13 It is worth noting that it is not possible to define a score of normality for PBPI subscales, because mental representations can never be regarded as pathological in themselves. PBPI has been used to assess pain beliefs in heterogeneous samples of patients suffering from chronic back, limbs, head, and abdominal pain.8,9,11 In these studies, beliefs that pain would be enduring and constant are associated with increased reported pain intensity, regardless of actual pain duration.1,9,12 Chronic patients create a mental representation of pain as persistent and mysterious, with high scores in Permanence and Mystery scales of PBPI, which in turn is related to more catastrophizing and worse coping and diability.3,7,14 A correlation has also been found between pain beliefs of permanence, mystery
137 and self-blame, and psychological distress, often displayed with anxiety and depression symptoms.9 Research on clinical psychology of headache has investigated comorbidity with psychiatric syndromes ranging from mood, anxiety,15 and somatoform disorders16 to problems with personality,17,18 impulse control, eating,19 and abuse of substances.20 A variety of cognitive and affective factors have been linked to headache, like self-efficacy, locus of control, catastrophizing, or anger, with particular regard to impact on headache disability.21-24 Pain beliefs in headache disease were addressed with the Illness Behavior Questionnaire (IBQ) in a study that did not find significant differences between tension-type headache and migraine sufferers. In the entire headache group, the 3 scales of IBQ that showed high scores were affecting inhibition, denial, and irritability.25 Self-efficacy beliefs and internal locus of control beliefs seemed to be related to less important headache-related disability,4 but they were never studied in relation to items investigated with PBPI. The content of mental representation may vary depending on the illness, therefore PBPI assess causal beliefs with Mystery and Self-Blame subscales, and the 2 constituents of the time line with Permanence and Constancy.8,26 Debate on episodic migraine (EM) and chronic migraine (CM) being part of a same spectrum of disease or 2 different entities is still open.27,28 The notion of CM is quite recent. CM patients appear more debilitated, focused on their illness and with poorer functional status, but we do not know if their beliefs are different from EM patients.29 The main objective of this study was to compare pain beliefs between CM and EM patients, and to identify possible predictive factors of pain beliefs in migraine patients according to PBPI.
METHODS Context.—This study was performed in the Pain Department of the Nice University Hospital, which is one of the certified French multidisciplinary pain centers and which participates in the French “Observatoire of Migraine and Headaches,” a nationwide survey network on headaches and facial pains set up in 2002 by the French Society for the Study of Migraine and Headaches (SFEMC).30 The setting up
138 of the observatory database was declared to, and approved by, the French Commission on Data Processing and Liberties. However, since the present study had no impact on disease management, no subjects’ written informed consent was required, and the study did not need to be approved by an Ethics Committee. This study was performed using daily practice data collected in the Pain Department. According to French guidelines on the chronic pain management,31 patients referring to the Pain Department received a self-administered 2-part questionnaire before the first consultation. In the first part, the assessment questionnaire contained generic questions aimed at gathering socio-demographic data, medical history, and self-evaluation of pain (daily activities interference, emotional impact, beliefs, and quality of life). In the second part, questions are asked about the type of pain. All answers were checked by the practitioner during the first consultation, and a form summarizing these data was included in the patient’s medical file. All data used in the study were extracted from these summary forms. Subjects.—In this cross-sectional study, all patients with chronic pain referring for the first time to our Pain Department from 1 January 2011 to 31 December 2011 were screened, and we selected all patients of age affected by CM and EM. Data concerning patients affected by nonheadache painful conditions will be the object of a future analysis, meant to compare migraine patients pain beliefs with beliefs of persons with different pain settings, as inflammatory nociceptive pain or neuropathic pain (data not shown). Patients with non-migrainous episodic headache were excluded: Tension type headache (TTH) and most secondary headaches often pose a risk because of the differential diagnosis with cervical pain and osteoarthritis. Individuals with cluster headache were also excluded because of the nonhomogeneous frequency of attacks and the particular psychological setting related to the associated violent pain. Headache specialist physicians made all diagnoses. In particular, all diagnoses of different types of headache were made according to the Second Edition of the International Classification of Headache Disorders (ICHD-II).32
January 2015 For this study, cases were divided into 2 groups. The first one (Group A) was composed of all patients affected by CM (defined as the ICHD-II criteria, 1.5.1). The second group (Group B) included all patients suffering of EM according to ICHD-II criteria (1.1).32 Data.—For all patients it was decided a priori to collect data on sex, age, and duration of pain; the latter was calculated, for CM cases, from the beginning of chronic migraine and not from the start of possible migraine disease that had preceded it. The French version of the Hospital Anxiety Depression Scale (HADS) was used to evaluate the symptoms of anxiety and depression.33 This scale was included in our pain department selfquestionnaire because it is currently the only one that can be used in a postal survey to evaluate this impairment and the impact of pain on everyday life. For these reasons, it is recommended by French Guidelines on chronic pain evaluation.34 The HADS involves 7 anxiety items alternating with 7 depression items, each one with scores ranging from 0 to 3. Final scores for both anxiety and depression would therefore range between 0 and 21. Anxiety and depression impairments are commonly defined by anxiety (HAD-Anx) and depression (HAD-Dep) scores > 7,35 although even if in clinical everyday approach we used this cutoff, scores of HADS were used as a continuous scale, and not in a dichotomic way, in order to explore better a possible influence on beliefs. Pain beliefs were also collected using the French version11 of the PBPI.9 This inventory was included in our pain department self-questionnaire, because it is adapted overall to chronic pains, and it is the only tool validated to assess pain beliefs in French population. A scoring method for the PBPI consists of four subscales: Mystery, Self-Blame, Pain Permanence, and Pain Constancy. Each question had a score ranged that goes from −2 to +2, without a 0 value. Mystery explores the belief that pain is a poorly understood experience (for example: “I don’t know enough about my pain”). An example of items for Self-Blame, Permanence, and Constancy are, respectively: “I am the cause of my pain,” “My pain is here to stay,” “It seems like I wake up with pain and I go to sleep with pain.”8
Headache In the French version, Mystery is addressed with 5 questions (mean 0.47, standard deviation [SD] 0.96, Cronbach’s α 0.74); Self-Blame with 3 questions (mean −1.13, SD 0.88, Cronbach’s α 0.63), Permanence and Constancy with 4 questions each (mean 0.06, SD 1.11, Cronbach’s α 0.70; mean 0.85, SD 1, Cronbach’s α 0.79, respectively). All means, SD, and Cronbach’s α reported refer to the French version used in the present study.11 In the Permanence subscale, there are 3 questions for which scoring must be reversed, and another reversed scoring question is to be found in Constancy. For both groups, data on the number of days with headache were collected using the headache diary, recommended by French guidelines on migraine management,36 and included in the pain department self-questionnaire. The headache diary was systematically completed by each patient from 30 days prior to the first scheduled consultation in our department. Patients were asked explicitly to indicate each day during which headache pain was present. Usually, a headache diary should be completed with an intensity scoring of the headache attack, but in our sample, patients never filled in this part systematically. They often reported intensity of just a few random attacks or of the more intense attacks that would have constituted an important bias. Thus, this variable was not included in our study. Data on headache impact were collected using the French version37 of the short version of Headache Impact Test (HIT-6), which was developed by applying Item Response Theory and other psychometric techniques to widely used questionnaires of headache impact.38 HIT-6, which is an assessment tool recommended by the SFEMC,39 measures the impact and severity of headache using 6 questions. It also considers a broader spectrum of the measurement of headache impact to include social-role functioning, pain, emotional distress and well-being, cognitive functioning, and vitality. Clinicians and patients easily interpret it because it quantifies headache impact by a score varying from 36 to 78. This score relates to the degree of headache impact: from 36 to 49 is equivalent to little or no impact, 50 to 55 equates to some impact, 56 to 59 substantial
139 impact, and above 60 indicates a very severe impact.40 Statistical Analysis.—A descriptive analysis of patients considered in the study was performed. Student’s t-test was conducted to compare directly pain beliefs between the 2 groups. A Levene’s test was performed to check for violation of the assumption of homogeneity of variances between groups. In the case of heteroscedasticity, Mann– Whitney U-test was used. Pearson correlation was conducted between potential covariates, which included variables theoretically or empirically associated with pain beliefs in prior research or ordinary clinical activity, to assess their level of correlation. In order to look for possible predictive factors of each of the 4 PBPI subscale in each group, parameters considered meaningful a priori were considered as possible independent factors. The inclusion of these variables was based either on previous studies that had already found a correlation between those parameters (especially anxiety and depression9) and PBPI or they were thought to have the potential to influence personal beliefs on pain, as it was the case in this study. In contrast with previous studies, we also introduced the variable “diagnosis” to underline possible differences related specifically to EM or CM condition. Model selection based on Akaike’s Information Criterion (AIC)41 was used on a set of candidate models, starting from a global generalized linear model, with normal error distribution, which included all variables deemed to have a potential effect on each of the separate 4 PBPI subscales: diagnosis, age, duration of pain, sex, HAD-Anx and HAD-Depr, HIT-6, and days of headache/month scores. Due to the relative small sample size compared to the number of factors, AICc was used to compare candidate models, after removing all missing data to render all models comparable.41,42 The best models, identified by having a ΔAICc less than 2, were averaged to provide an average model41-43 in which significant parameters were identified by confidence intervals around the estimate not including zero. The data analysis plan was determined at the beginning of the study when the study protocol was defined.
140
January 2015
t-Test and Pearson’s correlation were performed using SPSS 20 (IBM, Armonk, NY, USA), while AIC model selection was conducted in R 3.0.2 (R core Team, Vienna, Austria).
RESULTS A total of 1032 patients were screened among those referred for the first time to our Pain Department within the period considered. Out of those that received the questionnaire, 543 patients were excluded as their painful condition was unrelated to headache, 133 people did not fill the questionnaires as asked (nonresponse rate 12.8%), 16 patients were younger than 18 years of age, 16 were undiagnosed, and 82 were diagnosed with non-migrainous headache (Fig. 1. Precise diagnosis of non-migrainous headache patients is given). Following these exclusion criteria, 242 cases were therefore included in the analyses, with 116 patients for CM and 126 for EM. 1032 133 did not fill questionnaires 899 16 < 18 years old 883 543 patients with non-headache pain 340 16 undiagnosed
A significant difference in age was found between the 2 groups: EM patients were younger than CM ones (P < .01). Duration of pain was longer in EM Group than CM one (P < .001). There were lower HAD-Depr scores in EM Group (P < .001). There was no significant difference in scores of HIT-6 between the 2 groups. PBPI-Mystery, Permanence and Self-Blame scores did not show significant differences between the groups, whereas lower Constancy scores were found in the EM group (P < .001). Sample features and results of comparison are summarized in Table 1. Correlation studies were performed on the 2 groups. Most explanatory variables included in model selection showed weak correlation44 apart from one moderate correlation medium HAD-Anx and HADDep. Results are presented in Table 2. The best ranking models for each PBPI subscales are explicated in Table 3. The results after model averaging in the 4 subscales showed Mystery was positively correlated with HIT-6, while Self-Blame increased with Anxiety. Permanence was positively correlated with HIT-6 and depression, while Constancy was also associated with increased depression and increasing number of days with headache per month (Table 4). The relationship between Constancy scores and the number of days with headache per month is outlined by the regression line in Figure 2.
324 82 with non-migrainous headache* 242
Fig 1.—Flow chart of the study sample. Out of the 1032 screened cases, 242 were eligible for data analysis, according to inclusion and exclusion criteria. *Out of 82 patients with nonmigrainous headache, 29 were affected by Cluster Headache, 27 by Tension-Type Headache, 7 by Classical Trigeminal Neuralgia, 3 by Classical Glossopharyngeal Neuralgia, 3 by Cervicogenic Headache, 3 by psychiatric disorders (in particular, 2 headache attributed to somatization disorders and 1 to psychotic disorder), 3 by headache attributed to increased intracranial pressure caused by neoplasm (1 disembryoplastic frontal tumor, 1 pineal tumor, 1 pituitary macroadenoma), 2 by chronic post-traumatic headache, 1 by Hypnic Headache, 1 by Orgasmic Headache, and 3 by other rare secondary headaches attributed to disorders of sinuses, teeth, and facial bones.
DISCUSSION In this cross-sectional descriptive study, we compared pain beliefs according to PBPI in patients suffering from CM and EM; we completed our observation by exploring the predictive factors for the 4 subscales of PBPI in the whole population of patients that had met eligibility criteria. In the direct comparison of the 2 groups, the only difference in the PBPI scores was for Constancy, which was strongly correlated to the number of days with headache per month. Furthermore, the beliefs remained similar even in the presence of a significant difference in the HAD-Depr scores, more elevated in the CM group. HIT-6 scores did not differ between the 2 groups, suggesting that the migraine attacks
38.83 12.93 7.22 65.68 8.9 5.5 0.7 −1.3 −0.2 −0.6
242 242 230
226 235 235 240 238 219 218
N
EM Mean
66.33 9.7 7.3 0.8 −1.3 −0.2 0.5
44.33 5.50 24.84
CM Mean
66.26 9.3 6.3 0.7 −1.3 −0.2 −0.1
41.44 9.42 15.88
Overall mean
242 (100) 201 (83.05)
5.89 4.1 4.3 0.8 0.8 0.9 1.1
13.55 12.07 3.42
EM St. Dev
6.61 4.3 4.7 0.9 0.7 0.9 1.1
15.04 7.65 5.65
CM St Dev
−0.59 −1.28 −2.91 0.12 0.09 −0.72 −7.15
−2.98 5.53 −27.12
T
126 (52.06) 107 (84.9)
EM
.539 .200 .004 .503 .926 .469 .000
.003 .000* .000*
Student’s t test P
2.979 0.012 0.799 2.343 2.453 1.147 0.018
0.037 24.371 32.387
Levene’s Test F
0.086 0.912 0.372 0.127 0.119 0.285 0.894
0.848 0.000* 0.000*
Levene’s Test P
116 (47.94) 94 (81.0)
CM
— — — — — — —
– .000 .000
Mann– Whitney P
1.01 0.54 0.58 0.58 0.30 0.49 0.60
1.85 1.34 0.64
Standard error
*In case of Levene’s test positivity, Mann–Whitney test was added. t-Test was conducted to compare the 2 groups of our sample; no difference was remarked on HIT-6 scores; of the PBPI subscales, only Constancy showed significant difference between the headache subpopulation. Levene’s test was used to test the equality of variance: when positive, Mann–Whitney test was added. CM = chronic migraine; EM = episodic migraine; HAD-Anx = anxiety score at hospital anxiety depression; HAD-Dep = depression score at hospital anxiety depression; HIT-6 = Headache Impact Test score.
Age Duration of pain Days with headache/ month HIT-6 HAD-Anx HAD-Dep Mystery Self-blame Pain permanence Pain constancy
Number (%) Females (%)
Total sample
Table 1.—Descriptive Statistics of Our Sample and Comparison Between CM and EM Groups
Headache 141
0.00** 0.00** 0.00** 0.88 0.01* — Correlation Matrix among demographic and clinical features of our sample. Significance is underlined in the Table with *P < .05 or **P < .01).
0.01* 0.03* 0.10 0.05 — — 0.00** 0.02* 0.38 — — — 0.01* 0.01* — — — — 0.00** — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —
— — — — — —
0.98 0.50 0.19 0.04* 0.02* 0.27 0.53 0.83 0.58 0.91 0.31 0.52 0.12 0.49 0.91 0.66 0.01* 0.18 0.00** 0.01* 0.25 0.68 0.00** 0.02* 0.03* 0.19 0.44 0.18 0.00** 0.05 0.01* 0.00** 0.01* 0.52 0.01* — 0.50 0.16 0.39 0.55 — — 0.21 0.42 0.25 — — — 0.00** 0.00** — — — — 0.00** — — — — — — — — — — —
1. Age 2. Etiology 3. Duration of Pain 4. Sex 5. HIT-6 6. Days with headache/ month 7. HAD-Anx 8. HAD-Dep 9. Mystery 10. Selfblame 11. Pain permanence 12. Pain constancy
Variables
0.07 0.00** 0.15 0.20 0.00** 0.00**
January 2015 6. Days 3. Duration with headache/ 7. HAD- 8. HAD11. Pain 12. Pain 1. Age 2. Etiology of pain 4. Sex 5. HIT-6 month Anx Dep 9. Mystery 10. Selfblame Permanence Constancy
Table 2.—Correlation Analysis
142
were perceived as something extremely disturbing regardless of their frequency. The difference between the groups on age and duration of pain was perhaps unsurprising given the reported epidemiology of the 2 diseases29 and the decision to count years of illness starting from the beginning of chronic migraine for the CM group. The high scores in HIT-6 were related to the higher scores in Mystery, and HAD-Depr scores to Constancy, and the number of days with migraine was strongly positively correlated with higher Constancy. Finally, anxiety showed a direct relationship with SelfBlame. To our knowledge, this is the first study that explores the issue of possible differences in pain beliefs profile in different headache settings; in particular, CM was never previously directly compared to EM. A qualitative analysis on cognitive representations of migraine through migraineurs’ discourses showed a “successive attacks” representation of migrainous conditions, more than a long-term progressive disease,45 but in this study, EM and CM were not distinguished. There was only one study that had compared pain beliefs between episodic and chronic tension type headache, but it was based on small samples sizes (20 and 40 patients, respectively). The authors had reported higher scores on Mystery and Constancy in the chronic group.46 In our study, we did not find the difference in Mystery score, but the 2 studies are difficult to compare because of the difference in sample size and in diagnosis. Very few studies correlated characteristics of patients with pain to PBPI subscales scores. Williams et al observed a relationship between high scores on anxiety–depression questionnaires and Mystery and Permanence subscales, and they noticed a higher depression linked to higher Self-Blame score,9 results that are only partly confirmed in our study. However, Williams et al used the Beck questionnaires, which render their results difficult to compare to those found in the current study. When HADS were used, a correlation was found between HAD-Anx with Mystery, Constancy and Self-Blame, and between HAD-Depr and all PBPI subscales, in patients with chronic painful disorders.11
Headache
143 Table 3.—Candidate Models Evaluation
a. Candidate models Mystery 1 HAD-Dep + HIT-6 + 1 2 HAD-Anx + HIT-6 + 1 3 Age + HAD-Dep + HIT-6 + 1 4 HAD-Anx + HAD-Dep + HIT-6 + 1 5 Age + HAD-Anx + HIT-6 + 1 6 HAD-Dep + HIT-6 + Sex + 1 7 Age + HIT-6 + 1 8 DaysMigr + HAD-Dep + HIT-6 + 1 b. Candidate models Self-blame 1 HAD-Anx + 1 2 Diagnosis + HAD-Anx + 1 3 DaysMigr + Diagnosis + HAD-Anx + 1 4 DaysMigr + HAD-Anx + 1 5 Age + HAD-Anx + 1 6 DP + HAD-Anx + 1 7 HAD-Anx + Sex + 1 8 HAD-Anx + HAD-Dep + 1 c. Candidate models Permanence 1 DP + HAD-Dep + HIT-6 + Sex 2 DP + HAD-Dep + HIT-6 + 1 3 HAD-Dep + HIT-6 + Sex + 1 4 HAD-Dep + HIT-6 + 1 5 DP + HAD-Anx + HAD-Dep + HIT-6 + Sex + 1 d. Candidate models Constancy 1 DaysMigr + HAD-Dep + HIT-6 2 Age + DaysMigr + HAD-Dep + HIT-6 3 DaysMigr + HAD-Dep + HIT-6 + Sex 4 DaysMigr + Diagnosis + HAD-Dep + HIT-6 5 DaysMigr + HAD-Anx + HAD-Dep + HIT-6 6 Age + DaysMigr + HAD-Dep + HIT-6 + Sex 7 Age + DaysMigr + Diagnosis + HAD-Dep + HIT-6 8 DaysMigr + Diagnosis + HAD-Dep + HIT-6 + Sex
k
L
AICc
Δ AICc
w
4 4 5 5 5 5 4 5
−230.57 −231.05 −230.01 −230.19 230.23 −230.38 −231.45 −230.44
469.35 470.31 470.35 470.70 470.78 471.09 471.11 471.21
0.00 0.96 1.00 1.35 1.43 1.74 1.76 1.86
0.22 0.14 0.14 0.11 0.11 0.09 0.09 0.09
3 4 5 4 4 4 4 4
−210.78 −209.86 −208.82 −210.64 −210.69 −210.69 −210.69 −210.73
427.69 427.93 427.96 429.49 429.59 429.59 429.60 429.68
0.00 0.24 0.26 1.80 1.90 1.90 1.91 1.99
0.21 0.19 0.19 0.09 0.08 0.08 0.08 0.08
6 5 5 4 7
−234.91 −236.24 −236.51 −237.99 −234.81
482.29 482.80 483.34 484.19 484.23
0.00 0.52 1.05 1.91 1.95
0.32 0.25 0.19 0.12 0.12
5 6 6 6 6 7 7 7
−256.50 −255.82 −255.86 −256.03 −256.26 −255.21 −255.28 −255.34
523.32 524.10 524.19 524.52 524.99 525.03 525.17 525.30
0.00 0.78 0.87 1.20 1.67 1.71 1.85 1.98
0.22 0.15 0.14 0.12 0.10 0.09 0.09 0.08
Best ranked models. In (A) the 8 models whose Δi
Headache © 2014 American Headache Society
Research Submission Pain Beliefs and Perceptions Inventory: A Cross-Sectional Study in Chronic and Episodic Migraine Cecilia Condello, MD; Virginie Piano, PhD; Daria Dadam, PhD; Lorenzo Pinessi, MD; Michel Lantéri-Minet, MD, PhD
Objective/Background.—This study aims to compare differences in pain beliefs according to headache type, chronic vs episodic migraine, in a large cohort of patients, using the Pain Beliefs and Perceptions Inventory (PBPI), and to identify possible predictive factors of the same pain beliefs. Methods.—All patients referring for the first time at our center in 2011 were screened using PBPI and the Hamilton Anxiety and Depression Scale (a total of 1032 patients). PBPI is a 4-subscale questionnaire that explores a patient’s personal beliefs on their subjective experience of pain. Headache patients also completed the Headache Impact Test (HIT-6) and a 30-day headache diary. For all participants, age, gender, duration of pain were collected. The sample was narrowed down to 899 as we experienced a nonresponse rate of 12.8%. For the purpose of this study, 2 groups were identified: chronic and episodic migraine, consisting of 116 and 126 patients, respectively, which were compared using Student’s t-test; correlation analyses were conducted to investigate the relationship between variables before running a model selection based on Akaike’s Information Criterion to identify possible predictive factors of different pain beliefs. Patients below 18 years of age and those diagnosed with other painful conditions were excluded from the analysis. Results.—Beliefs from chronic and episodic migraine patients were very similar, with only a difference in beliefs related to constancy of pain (Mean value ± SD 0.5 ± 1.1 vs −0.6 ± 1.1, P < .001). Predictive factors were depression and HIT-6 scores for all PBPI subscales apart from Self-Blame, which showed a stronger relation to anxiety scores. Number of days with headache per month was correlated to higher constancy values. Diagnosis was a predictive factor for any particular belief. Discussion.—This is the first study, to our knowledge, that addresses differences and predictive factors in pain beliefs according to headache diagnosis. A deeper knowledge of beliefs pattern in patients could lead to better-tailored psychological management. Key words: Pain Beliefs and Perceptions Inventory, chronic migraine, episodic migraine, depression, disability (Headache 2015;55:136-148)
Address all correspondence to C. Condello, Department of Neuroscience, Head Center, University of Torino, Via Cherasco 15 10100, Turin, Italy, email: cecilia.condello@ gmail.com
Certain pain-related beliefs, like faith in the persistence of pain, seem to impact on coping and compliance, while regarding pain as mysterious and inexplicable is related to worse outcome, psychological distress, and somatization.1-3 It has been shown that headache patients with a low perception of selfefficacy (ability to take actions to influence the course of headache), or convinced that the origin of their headache was due to chance or fate, were more prone
Accepted for publication October 9, 2014.
Disclosures: No funding was received for this study, and the authors declare no conflict of interest.
From the Department of Neuroscience, Headache Center, University of Torino, Turin, Italy (C. Condello and L. Pinessi); Pain Department, CHU Pasteur, Nice, France (V. Piano and M. Lantéri-Minet); Statistics, British Trust for Ornithology, Thetford, UK (D. Dadam).
136
Headache to inadequate coping, with worse outcome on overall disability.4 In light of this, early systematic assessment of pain beliefs was promoted.5 Pain beliefs are defined as “patients’ own conceptualization of what pain is and what pain means for them.”1 The main dimensions of these beliefs are eventuality of pain lifting, the mysterious nature of pain, wonder at the cause of pain, lifestyle altering because of pain, possibility of personal control over pain, blame, and constant vs intermittent nature of pain experience. A useful tool to assess pain beliefs is the Pain Beliefs and Perceptions Inventory (PBPI), a 16-question Likert-type scale addressing 4 dimensions of pain beliefs: seeing pain as mysterious (Mystery), holding oneself responsible for pain (SelfBlame), regarding one’s condition as lingering in the future (Permanence), and/or continuous over time (Constancy). PBPI was initially designed around 3 main cores, the dimensions Mystery, Self-Blame, and Time, but the Australian and German replication of the scale showed an improvement in the method by adding a fourth dimension (Constancy and Permanence instead of Time).6,7 This change provided a further tool to explore disorders not associates with intermittent pain, like headache, in which is likely that Constancy and Permanence are regarded differently by patients.8 In this new version, PBPI was validated in the United States by its developers,9 in England,8 in Norway,10 in France,11 in China,12 and in Italy.13 It is worth noting that it is not possible to define a score of normality for PBPI subscales, because mental representations can never be regarded as pathological in themselves. PBPI has been used to assess pain beliefs in heterogeneous samples of patients suffering from chronic back, limbs, head, and abdominal pain.8,9,11 In these studies, beliefs that pain would be enduring and constant are associated with increased reported pain intensity, regardless of actual pain duration.1,9,12 Chronic patients create a mental representation of pain as persistent and mysterious, with high scores in Permanence and Mystery scales of PBPI, which in turn is related to more catastrophizing and worse coping and diability.3,7,14 A correlation has also been found between pain beliefs of permanence, mystery
137 and self-blame, and psychological distress, often displayed with anxiety and depression symptoms.9 Research on clinical psychology of headache has investigated comorbidity with psychiatric syndromes ranging from mood, anxiety,15 and somatoform disorders16 to problems with personality,17,18 impulse control, eating,19 and abuse of substances.20 A variety of cognitive and affective factors have been linked to headache, like self-efficacy, locus of control, catastrophizing, or anger, with particular regard to impact on headache disability.21-24 Pain beliefs in headache disease were addressed with the Illness Behavior Questionnaire (IBQ) in a study that did not find significant differences between tension-type headache and migraine sufferers. In the entire headache group, the 3 scales of IBQ that showed high scores were affecting inhibition, denial, and irritability.25 Self-efficacy beliefs and internal locus of control beliefs seemed to be related to less important headache-related disability,4 but they were never studied in relation to items investigated with PBPI. The content of mental representation may vary depending on the illness, therefore PBPI assess causal beliefs with Mystery and Self-Blame subscales, and the 2 constituents of the time line with Permanence and Constancy.8,26 Debate on episodic migraine (EM) and chronic migraine (CM) being part of a same spectrum of disease or 2 different entities is still open.27,28 The notion of CM is quite recent. CM patients appear more debilitated, focused on their illness and with poorer functional status, but we do not know if their beliefs are different from EM patients.29 The main objective of this study was to compare pain beliefs between CM and EM patients, and to identify possible predictive factors of pain beliefs in migraine patients according to PBPI.
METHODS Context.—This study was performed in the Pain Department of the Nice University Hospital, which is one of the certified French multidisciplinary pain centers and which participates in the French “Observatoire of Migraine and Headaches,” a nationwide survey network on headaches and facial pains set up in 2002 by the French Society for the Study of Migraine and Headaches (SFEMC).30 The setting up
138 of the observatory database was declared to, and approved by, the French Commission on Data Processing and Liberties. However, since the present study had no impact on disease management, no subjects’ written informed consent was required, and the study did not need to be approved by an Ethics Committee. This study was performed using daily practice data collected in the Pain Department. According to French guidelines on the chronic pain management,31 patients referring to the Pain Department received a self-administered 2-part questionnaire before the first consultation. In the first part, the assessment questionnaire contained generic questions aimed at gathering socio-demographic data, medical history, and self-evaluation of pain (daily activities interference, emotional impact, beliefs, and quality of life). In the second part, questions are asked about the type of pain. All answers were checked by the practitioner during the first consultation, and a form summarizing these data was included in the patient’s medical file. All data used in the study were extracted from these summary forms. Subjects.—In this cross-sectional study, all patients with chronic pain referring for the first time to our Pain Department from 1 January 2011 to 31 December 2011 were screened, and we selected all patients of age affected by CM and EM. Data concerning patients affected by nonheadache painful conditions will be the object of a future analysis, meant to compare migraine patients pain beliefs with beliefs of persons with different pain settings, as inflammatory nociceptive pain or neuropathic pain (data not shown). Patients with non-migrainous episodic headache were excluded: Tension type headache (TTH) and most secondary headaches often pose a risk because of the differential diagnosis with cervical pain and osteoarthritis. Individuals with cluster headache were also excluded because of the nonhomogeneous frequency of attacks and the particular psychological setting related to the associated violent pain. Headache specialist physicians made all diagnoses. In particular, all diagnoses of different types of headache were made according to the Second Edition of the International Classification of Headache Disorders (ICHD-II).32
January 2015 For this study, cases were divided into 2 groups. The first one (Group A) was composed of all patients affected by CM (defined as the ICHD-II criteria, 1.5.1). The second group (Group B) included all patients suffering of EM according to ICHD-II criteria (1.1).32 Data.—For all patients it was decided a priori to collect data on sex, age, and duration of pain; the latter was calculated, for CM cases, from the beginning of chronic migraine and not from the start of possible migraine disease that had preceded it. The French version of the Hospital Anxiety Depression Scale (HADS) was used to evaluate the symptoms of anxiety and depression.33 This scale was included in our pain department selfquestionnaire because it is currently the only one that can be used in a postal survey to evaluate this impairment and the impact of pain on everyday life. For these reasons, it is recommended by French Guidelines on chronic pain evaluation.34 The HADS involves 7 anxiety items alternating with 7 depression items, each one with scores ranging from 0 to 3. Final scores for both anxiety and depression would therefore range between 0 and 21. Anxiety and depression impairments are commonly defined by anxiety (HAD-Anx) and depression (HAD-Dep) scores > 7,35 although even if in clinical everyday approach we used this cutoff, scores of HADS were used as a continuous scale, and not in a dichotomic way, in order to explore better a possible influence on beliefs. Pain beliefs were also collected using the French version11 of the PBPI.9 This inventory was included in our pain department self-questionnaire, because it is adapted overall to chronic pains, and it is the only tool validated to assess pain beliefs in French population. A scoring method for the PBPI consists of four subscales: Mystery, Self-Blame, Pain Permanence, and Pain Constancy. Each question had a score ranged that goes from −2 to +2, without a 0 value. Mystery explores the belief that pain is a poorly understood experience (for example: “I don’t know enough about my pain”). An example of items for Self-Blame, Permanence, and Constancy are, respectively: “I am the cause of my pain,” “My pain is here to stay,” “It seems like I wake up with pain and I go to sleep with pain.”8
Headache In the French version, Mystery is addressed with 5 questions (mean 0.47, standard deviation [SD] 0.96, Cronbach’s α 0.74); Self-Blame with 3 questions (mean −1.13, SD 0.88, Cronbach’s α 0.63), Permanence and Constancy with 4 questions each (mean 0.06, SD 1.11, Cronbach’s α 0.70; mean 0.85, SD 1, Cronbach’s α 0.79, respectively). All means, SD, and Cronbach’s α reported refer to the French version used in the present study.11 In the Permanence subscale, there are 3 questions for which scoring must be reversed, and another reversed scoring question is to be found in Constancy. For both groups, data on the number of days with headache were collected using the headache diary, recommended by French guidelines on migraine management,36 and included in the pain department self-questionnaire. The headache diary was systematically completed by each patient from 30 days prior to the first scheduled consultation in our department. Patients were asked explicitly to indicate each day during which headache pain was present. Usually, a headache diary should be completed with an intensity scoring of the headache attack, but in our sample, patients never filled in this part systematically. They often reported intensity of just a few random attacks or of the more intense attacks that would have constituted an important bias. Thus, this variable was not included in our study. Data on headache impact were collected using the French version37 of the short version of Headache Impact Test (HIT-6), which was developed by applying Item Response Theory and other psychometric techniques to widely used questionnaires of headache impact.38 HIT-6, which is an assessment tool recommended by the SFEMC,39 measures the impact and severity of headache using 6 questions. It also considers a broader spectrum of the measurement of headache impact to include social-role functioning, pain, emotional distress and well-being, cognitive functioning, and vitality. Clinicians and patients easily interpret it because it quantifies headache impact by a score varying from 36 to 78. This score relates to the degree of headache impact: from 36 to 49 is equivalent to little or no impact, 50 to 55 equates to some impact, 56 to 59 substantial
139 impact, and above 60 indicates a very severe impact.40 Statistical Analysis.—A descriptive analysis of patients considered in the study was performed. Student’s t-test was conducted to compare directly pain beliefs between the 2 groups. A Levene’s test was performed to check for violation of the assumption of homogeneity of variances between groups. In the case of heteroscedasticity, Mann– Whitney U-test was used. Pearson correlation was conducted between potential covariates, which included variables theoretically or empirically associated with pain beliefs in prior research or ordinary clinical activity, to assess their level of correlation. In order to look for possible predictive factors of each of the 4 PBPI subscale in each group, parameters considered meaningful a priori were considered as possible independent factors. The inclusion of these variables was based either on previous studies that had already found a correlation between those parameters (especially anxiety and depression9) and PBPI or they were thought to have the potential to influence personal beliefs on pain, as it was the case in this study. In contrast with previous studies, we also introduced the variable “diagnosis” to underline possible differences related specifically to EM or CM condition. Model selection based on Akaike’s Information Criterion (AIC)41 was used on a set of candidate models, starting from a global generalized linear model, with normal error distribution, which included all variables deemed to have a potential effect on each of the separate 4 PBPI subscales: diagnosis, age, duration of pain, sex, HAD-Anx and HAD-Depr, HIT-6, and days of headache/month scores. Due to the relative small sample size compared to the number of factors, AICc was used to compare candidate models, after removing all missing data to render all models comparable.41,42 The best models, identified by having a ΔAICc less than 2, were averaged to provide an average model41-43 in which significant parameters were identified by confidence intervals around the estimate not including zero. The data analysis plan was determined at the beginning of the study when the study protocol was defined.
140
January 2015
t-Test and Pearson’s correlation were performed using SPSS 20 (IBM, Armonk, NY, USA), while AIC model selection was conducted in R 3.0.2 (R core Team, Vienna, Austria).
RESULTS A total of 1032 patients were screened among those referred for the first time to our Pain Department within the period considered. Out of those that received the questionnaire, 543 patients were excluded as their painful condition was unrelated to headache, 133 people did not fill the questionnaires as asked (nonresponse rate 12.8%), 16 patients were younger than 18 years of age, 16 were undiagnosed, and 82 were diagnosed with non-migrainous headache (Fig. 1. Precise diagnosis of non-migrainous headache patients is given). Following these exclusion criteria, 242 cases were therefore included in the analyses, with 116 patients for CM and 126 for EM. 1032 133 did not fill questionnaires 899 16 < 18 years old 883 543 patients with non-headache pain 340 16 undiagnosed
A significant difference in age was found between the 2 groups: EM patients were younger than CM ones (P < .01). Duration of pain was longer in EM Group than CM one (P < .001). There were lower HAD-Depr scores in EM Group (P < .001). There was no significant difference in scores of HIT-6 between the 2 groups. PBPI-Mystery, Permanence and Self-Blame scores did not show significant differences between the groups, whereas lower Constancy scores were found in the EM group (P < .001). Sample features and results of comparison are summarized in Table 1. Correlation studies were performed on the 2 groups. Most explanatory variables included in model selection showed weak correlation44 apart from one moderate correlation medium HAD-Anx and HADDep. Results are presented in Table 2. The best ranking models for each PBPI subscales are explicated in Table 3. The results after model averaging in the 4 subscales showed Mystery was positively correlated with HIT-6, while Self-Blame increased with Anxiety. Permanence was positively correlated with HIT-6 and depression, while Constancy was also associated with increased depression and increasing number of days with headache per month (Table 4). The relationship between Constancy scores and the number of days with headache per month is outlined by the regression line in Figure 2.
324 82 with non-migrainous headache* 242
Fig 1.—Flow chart of the study sample. Out of the 1032 screened cases, 242 were eligible for data analysis, according to inclusion and exclusion criteria. *Out of 82 patients with nonmigrainous headache, 29 were affected by Cluster Headache, 27 by Tension-Type Headache, 7 by Classical Trigeminal Neuralgia, 3 by Classical Glossopharyngeal Neuralgia, 3 by Cervicogenic Headache, 3 by psychiatric disorders (in particular, 2 headache attributed to somatization disorders and 1 to psychotic disorder), 3 by headache attributed to increased intracranial pressure caused by neoplasm (1 disembryoplastic frontal tumor, 1 pineal tumor, 1 pituitary macroadenoma), 2 by chronic post-traumatic headache, 1 by Hypnic Headache, 1 by Orgasmic Headache, and 3 by other rare secondary headaches attributed to disorders of sinuses, teeth, and facial bones.
DISCUSSION In this cross-sectional descriptive study, we compared pain beliefs according to PBPI in patients suffering from CM and EM; we completed our observation by exploring the predictive factors for the 4 subscales of PBPI in the whole population of patients that had met eligibility criteria. In the direct comparison of the 2 groups, the only difference in the PBPI scores was for Constancy, which was strongly correlated to the number of days with headache per month. Furthermore, the beliefs remained similar even in the presence of a significant difference in the HAD-Depr scores, more elevated in the CM group. HIT-6 scores did not differ between the 2 groups, suggesting that the migraine attacks
38.83 12.93 7.22 65.68 8.9 5.5 0.7 −1.3 −0.2 −0.6
242 242 230
226 235 235 240 238 219 218
N
EM Mean
66.33 9.7 7.3 0.8 −1.3 −0.2 0.5
44.33 5.50 24.84
CM Mean
66.26 9.3 6.3 0.7 −1.3 −0.2 −0.1
41.44 9.42 15.88
Overall mean
242 (100) 201 (83.05)
5.89 4.1 4.3 0.8 0.8 0.9 1.1
13.55 12.07 3.42
EM St. Dev
6.61 4.3 4.7 0.9 0.7 0.9 1.1
15.04 7.65 5.65
CM St Dev
−0.59 −1.28 −2.91 0.12 0.09 −0.72 −7.15
−2.98 5.53 −27.12
T
126 (52.06) 107 (84.9)
EM
.539 .200 .004 .503 .926 .469 .000
.003 .000* .000*
Student’s t test P
2.979 0.012 0.799 2.343 2.453 1.147 0.018
0.037 24.371 32.387
Levene’s Test F
0.086 0.912 0.372 0.127 0.119 0.285 0.894
0.848 0.000* 0.000*
Levene’s Test P
116 (47.94) 94 (81.0)
CM
— — — — — — —
– .000 .000
Mann– Whitney P
1.01 0.54 0.58 0.58 0.30 0.49 0.60
1.85 1.34 0.64
Standard error
*In case of Levene’s test positivity, Mann–Whitney test was added. t-Test was conducted to compare the 2 groups of our sample; no difference was remarked on HIT-6 scores; of the PBPI subscales, only Constancy showed significant difference between the headache subpopulation. Levene’s test was used to test the equality of variance: when positive, Mann–Whitney test was added. CM = chronic migraine; EM = episodic migraine; HAD-Anx = anxiety score at hospital anxiety depression; HAD-Dep = depression score at hospital anxiety depression; HIT-6 = Headache Impact Test score.
Age Duration of pain Days with headache/ month HIT-6 HAD-Anx HAD-Dep Mystery Self-blame Pain permanence Pain constancy
Number (%) Females (%)
Total sample
Table 1.—Descriptive Statistics of Our Sample and Comparison Between CM and EM Groups
Headache 141
0.00** 0.00** 0.00** 0.88 0.01* — Correlation Matrix among demographic and clinical features of our sample. Significance is underlined in the Table with *P < .05 or **P < .01).
0.01* 0.03* 0.10 0.05 — — 0.00** 0.02* 0.38 — — — 0.01* 0.01* — — — — 0.00** — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —
— — — — — —
0.98 0.50 0.19 0.04* 0.02* 0.27 0.53 0.83 0.58 0.91 0.31 0.52 0.12 0.49 0.91 0.66 0.01* 0.18 0.00** 0.01* 0.25 0.68 0.00** 0.02* 0.03* 0.19 0.44 0.18 0.00** 0.05 0.01* 0.00** 0.01* 0.52 0.01* — 0.50 0.16 0.39 0.55 — — 0.21 0.42 0.25 — — — 0.00** 0.00** — — — — 0.00** — — — — — — — — — — —
1. Age 2. Etiology 3. Duration of Pain 4. Sex 5. HIT-6 6. Days with headache/ month 7. HAD-Anx 8. HAD-Dep 9. Mystery 10. Selfblame 11. Pain permanence 12. Pain constancy
Variables
0.07 0.00** 0.15 0.20 0.00** 0.00**
January 2015 6. Days 3. Duration with headache/ 7. HAD- 8. HAD11. Pain 12. Pain 1. Age 2. Etiology of pain 4. Sex 5. HIT-6 month Anx Dep 9. Mystery 10. Selfblame Permanence Constancy
Table 2.—Correlation Analysis
142
were perceived as something extremely disturbing regardless of their frequency. The difference between the groups on age and duration of pain was perhaps unsurprising given the reported epidemiology of the 2 diseases29 and the decision to count years of illness starting from the beginning of chronic migraine for the CM group. The high scores in HIT-6 were related to the higher scores in Mystery, and HAD-Depr scores to Constancy, and the number of days with migraine was strongly positively correlated with higher Constancy. Finally, anxiety showed a direct relationship with SelfBlame. To our knowledge, this is the first study that explores the issue of possible differences in pain beliefs profile in different headache settings; in particular, CM was never previously directly compared to EM. A qualitative analysis on cognitive representations of migraine through migraineurs’ discourses showed a “successive attacks” representation of migrainous conditions, more than a long-term progressive disease,45 but in this study, EM and CM were not distinguished. There was only one study that had compared pain beliefs between episodic and chronic tension type headache, but it was based on small samples sizes (20 and 40 patients, respectively). The authors had reported higher scores on Mystery and Constancy in the chronic group.46 In our study, we did not find the difference in Mystery score, but the 2 studies are difficult to compare because of the difference in sample size and in diagnosis. Very few studies correlated characteristics of patients with pain to PBPI subscales scores. Williams et al observed a relationship between high scores on anxiety–depression questionnaires and Mystery and Permanence subscales, and they noticed a higher depression linked to higher Self-Blame score,9 results that are only partly confirmed in our study. However, Williams et al used the Beck questionnaires, which render their results difficult to compare to those found in the current study. When HADS were used, a correlation was found between HAD-Anx with Mystery, Constancy and Self-Blame, and between HAD-Depr and all PBPI subscales, in patients with chronic painful disorders.11
Headache
143 Table 3.—Candidate Models Evaluation
a. Candidate models Mystery 1 HAD-Dep + HIT-6 + 1 2 HAD-Anx + HIT-6 + 1 3 Age + HAD-Dep + HIT-6 + 1 4 HAD-Anx + HAD-Dep + HIT-6 + 1 5 Age + HAD-Anx + HIT-6 + 1 6 HAD-Dep + HIT-6 + Sex + 1 7 Age + HIT-6 + 1 8 DaysMigr + HAD-Dep + HIT-6 + 1 b. Candidate models Self-blame 1 HAD-Anx + 1 2 Diagnosis + HAD-Anx + 1 3 DaysMigr + Diagnosis + HAD-Anx + 1 4 DaysMigr + HAD-Anx + 1 5 Age + HAD-Anx + 1 6 DP + HAD-Anx + 1 7 HAD-Anx + Sex + 1 8 HAD-Anx + HAD-Dep + 1 c. Candidate models Permanence 1 DP + HAD-Dep + HIT-6 + Sex 2 DP + HAD-Dep + HIT-6 + 1 3 HAD-Dep + HIT-6 + Sex + 1 4 HAD-Dep + HIT-6 + 1 5 DP + HAD-Anx + HAD-Dep + HIT-6 + Sex + 1 d. Candidate models Constancy 1 DaysMigr + HAD-Dep + HIT-6 2 Age + DaysMigr + HAD-Dep + HIT-6 3 DaysMigr + HAD-Dep + HIT-6 + Sex 4 DaysMigr + Diagnosis + HAD-Dep + HIT-6 5 DaysMigr + HAD-Anx + HAD-Dep + HIT-6 6 Age + DaysMigr + HAD-Dep + HIT-6 + Sex 7 Age + DaysMigr + Diagnosis + HAD-Dep + HIT-6 8 DaysMigr + Diagnosis + HAD-Dep + HIT-6 + Sex
k
L
AICc
Δ AICc
w
4 4 5 5 5 5 4 5
−230.57 −231.05 −230.01 −230.19 230.23 −230.38 −231.45 −230.44
469.35 470.31 470.35 470.70 470.78 471.09 471.11 471.21
0.00 0.96 1.00 1.35 1.43 1.74 1.76 1.86
0.22 0.14 0.14 0.11 0.11 0.09 0.09 0.09
3 4 5 4 4 4 4 4
−210.78 −209.86 −208.82 −210.64 −210.69 −210.69 −210.69 −210.73
427.69 427.93 427.96 429.49 429.59 429.59 429.60 429.68
0.00 0.24 0.26 1.80 1.90 1.90 1.91 1.99
0.21 0.19 0.19 0.09 0.08 0.08 0.08 0.08
6 5 5 4 7
−234.91 −236.24 −236.51 −237.99 −234.81
482.29 482.80 483.34 484.19 484.23
0.00 0.52 1.05 1.91 1.95
0.32 0.25 0.19 0.12 0.12
5 6 6 6 6 7 7 7
−256.50 −255.82 −255.86 −256.03 −256.26 −255.21 −255.28 −255.34
523.32 524.10 524.19 524.52 524.99 525.03 525.17 525.30
0.00 0.78 0.87 1.20 1.67 1.71 1.85 1.98
0.22 0.15 0.14 0.12 0.10 0.09 0.09 0.08
Best ranked models. In (A) the 8 models whose Δi
![[Progression from episodic migraine to chronic migraine].](/assets/img/hold.png)
