Study participants and data collection
The headache specialists at the headache clinic of the TVGH surveyed patients presented with headache during the period from May 2010 to July 2020. All participants completed a structured questionnaire to assess their headache profiles, comorbidities, mood, sleep, photophobic scale and suicidal ideation or attempts (see below) at their first visit. A semi-structured visual phenomenon questionnaire was used to assess the characteristics of TVDs. Later, they underwent a thorough clinical interview for the ascertainment of headache diagnosis as well as the questionnaires. Information collected from the questionnaires was de-identified and entered into the TVGH headache registry. No a priori statistical power calculation was conducted and the sample size was based on the available data.
Diagnoses of migraine
The diagnoses of MA, MO, and chronic migraine (CM) were based on the International Classification of Headache Disorders, 2nd and 3rd edition (ICHD-2 and ICHD-3) criteria (MO: code 1.1; MA: code 1.2; CM: code 1.3) 1,2, made through face-to-face interviews by experienced headache specialists. Patients with MA were excluded from this retrospective cohort study.
Definition of TVD
TVDs were defined as transient visual phenomena related in time to the occurrence of a migraine headache attack, but not visual aura. All of the participants received the five items of the Visual Aura Rating Scale (VARS) embedded in the questionnaires. The score is the weighted sum of the five-item scale: duration 5–60 minutes (three points), developing gradually ≥ 5 minutes (two points), scotoma (two points), zig-zag lines (two points) and unilateral visual field (one point). A cutoff VARS score of five or more points is reportedly highly sensitive and specific to diagnose MA 14. In order to exclude potential aura, we also explored the VARS cutoff value in our cohort. MA diagnoses made by neurologists based on ICHD 1,15 were used as the gold standard. Compared to physician diagnosis, a VARS score ≥ five had a sensitivity of 68.4% and a specificity of 88.7%, while a VARS score ≥ 4 had a sensitivity of 77.1% and a specificity of 79.9% to identify typical migraine aura. By using the criteria, i.e. a VARS score of ≥ 4, the positive predictive value (PPV) was 0.29 and the negative predictive value (NPV) was 0.97 for MA. Hence, in this study, we included only MO patients (diagnosed by neurologists) with VARS < 4 to ensure that their TVDs are unlikely to be visual aura.
Demographic data, including age, sex, occupation, education level, marital status, and medical history were collected. A validated headache questionnaire was used to specifically inquire headache frequency (days/ month), intensity (Numerical Rating Scale 0–10), duration, location, characteristics, accompanying symptoms, frequency of acute abortive medications usage (days/month), and disease duration of migraine (years), as well as TVD symptoms 13,16. A visual phenomenon questionnaire was used to assess the TVDs, including patterns (zigzag flashes, flickering dots/lines, or blurred/foggy vision), laterality of the visual fields, colors, presence of movement, development time, duration and temporal relationship with headaches.
The questions were:
1.Have you ever seen zigzag flashes before or during the headache?
2. Have you ever seen flickering dots or lines before or during the headache?
3. Before or when the headache started, did you have blurred or foggy vision?
4. What colors are these zigzag flashes, flickering dots or lines?
5. Do these TVDs (zigzag flashes, flickering dots or lines, or blurred vision) occur every time you have a headache?
6. Are these TVDs that you have seen unilateral, bilateral or different every time?
7. Did your TVDs move?
8. How long did your TVDs develop?
9. How long did your TVDs last?
10. Did these TVDs develop before, after or during the onset of the headache?
Afterwards, the MO patients were divided in to 2 subgroups based on the presence (MwTVD) or absence (MwoTVD) of TVDs. Additionally, the questionnaires also include the Migraine Disability Assessment (MIDAS), six-item Headache Impact Test (HIT-6), Migraine Photophobia Score (MPS), Hospital Anxiety and Depression Scale (HAS, HDS), Beck Depression Inventory (BDI), Perceived Stress Scale (PSS), and Pittsburgh Sleep Quality Index (PSQI), presence of MOH, suicidal ideation and attempts to collect relevant clinical information. Suicidal ideation and attempts were evaluated by 2 separate directly asked questions: 1. Have you ever had ideational thoughts of engaging in suicidal behavior? 2. Have you ever had engaged in any self-injurious behavior with the intent to die? Unanswered questionnaires were interpreted as missing data. The responses to these questions were validated by experienced headache specialists with face-to-face interview.
The Migraine Disability Assessment (MIDAS) questionnaire assesses headache-related disability in a 3-month period and the six-item Headache Impact Test (HIT-6) measures the severity of headache pain and adverse impact of headache 17,18. Both MIDAS and HIT-6 have been well accepted and widely utilized to evaluate the disability and impact caused by migraine. Headache intensity influences HIT-6 score more than the MIDAS, whereas the MIDAS is influenced more by headache frequency 19. Migraine Photophobia Score (MPS) is a self-administered, eight-question questionnaires to evaluate the scale of photophobia in migraine patients. Adding the total number of “yes” responses generates the MPS 20. Hospital Anxiety and Depression Scale is a self-administered instrument to detect psychiatric comorbidity in the setting of a hospital outpatient clinic. Anxiety was defined as a Hospital Anxiety Scale (HAS) score ≥ 11, and depression was defined as Hospital Depression Scale (HDS) score ≥ 11 21. The Beck Depression Inventory (BDI) is a 21-item self-report measure that evaluates major depression symptoms according to diagnostic criteria listed in the Diagnostic and Statistical Manual for Mental Disorders 22. The Perceived Stress Scale (PSS) is a 14-item self-reported questionnaire that was designed to measure “the degree to which individuals appraise situations in their lives as stressful” 23. Pittsburgh Sleep Quality Index (PSQI) evaluates the quality and patterns of sleep in the past one month. Poor sleep quality was defined as a PSQI score of > 5 24. Suicidal ideation and attempts were evaluated by 2 separate directly asked questions: (1) whether they, once in their lifetime, had ideational thoughts of engaging in suicidal behavior, and (2) whether they had engaged in any self-injury behavior with the intent to die. Unanswered questionnaires were interpreted as missing data.
Migrainous features, the prevalence of visual disturbance, and the severity of photophobia
Migrainous features (including moderate to severe intensity; pulsating quality; unilaterality; aggravation by physical activity; nausea or vomiting; photophobia and photophobia) of each subject were evaluated. The “yes” responses to each features were summed up to a total score of migrainous features, ranging from 0 to 6. To evaluate the migrainous feature in patients with photophobia, we re-calculated numbers of migrainous features ranged from 0 to 5, including the same features mentioned above except photophobia and phonophobia.
The descriptive data were presented as means ± SDs or percentages. The Chi square test was used to test the difference in categorical data. Normality was checked with histograms before conducting parametric tests. Continuous data between groups were analyzed using two-tailed independent sample t-test. Mann–Whitney U test was used to compare variables that were not distributed normally, including headache frequency, disease duration, BDI, HDS, MIDAS, and MPS. Bonferroni correction was done for the 16 variables (i.e., age, gender, disease duration of migraine (years), headache frequency (days/month), MPS, MIDAS, HIT-6, HAS, HDS, BDI, PSS, PSQI, chronic migraine, medication overuse headache, suicidal ideation, and suicide attempt). For post-hoc subgroup analysis, logistic regression was performed to test the interaction effect. The risks associated with comorbid suicidal ideation and attempts were analyzed separately by 3 layers of models: (1) no controlling for any covariates; (2) controlling for demographics; (3) controlling for demographics and clinical characteristics. These 3 layers of models were performed with “Enter” method. That is, the independent variables in each layer were fitted in the regression model simultaneously. Demographics included patients’ age, sex, and marital status (married vs. single/separated/divorced/widowed). Clinical characteristics included headache frequency (days/month), disease duration, HIT-6 score, depression (HDS score ≥ 11), anxiety (HAS score ≥ 11), poor sleep quality (PSQI >5), PSS, MOH, and CM. The risk factors were presented as odds ratio (OR) with 95% confidence interval. Results were considered significant for p value < 0.05. Statistical analyses were performed with R for Mac OS (Version 3.6.3; R Core Team, Vienna, Austria.)
Ethical approval and consent to participate
This retrospective study analyzed these de-identified data included in the TVGH headache registry. Because this study involved secondary analysis of existing de-identified data, informed consent has been waived by the institutional review board of TVGH, which approved the whole study protocol (TVGH IRB-2021-04-121-CC). The study was performed using ethical principles for medical research involving human subject in accord with the Declaration of Helsinki. The corresponding authors had full access to all of the data in the study and had final responsibility for the decision to submit for publication.