Migraine, quality of life, and depression

Population sample and survey.

The population surveys were conducted in demographically diverse areas of the United States (Philadelphia, PA) and mainland England (excluding the Isle of Wight, Wales, and Scotland). In both the United States and the United Kingdom, telephone-based sampling procedures were used to select representative samples.

In the United States, Genesys Random Sampling System provided the call list. Eligible participants were between 18 and 65 years of age. In the United Kingdom, the call list was provided by Sample Answers, Ltd., a UK-based geodemographic firm. In the United Kingdom, eligible participants were between 16 and 65 years of age. Results of telephone interview surveys are generalizable only to those who have telephones.

In both samples, eligible participants had to be a permanent resident of the household, able to converse easily in English, and mentally competent to be interviewed. Data for this study were obtained from the eligible individuals selected at random from within the household. Before proceeding to the interview, oral informed consent was obtained and the purpose of the survey was explained to the respondent.

The CATI interview has been described in detail elsewhere.18 The US and UK versions differed slightly to reflect differences in English usage in the two countries. Participants were first asked if they had at least one headache not due to a head injury, hangover, pregnancy, or an illness such as a cold or flu. Detailed questions were then asked regarding up to two different self-defined types of headache that occurred in the last 12 months, including all of the diagnostic features of migraine specified by the International Headache Society (IHS).19 Each headache feature reported in the affirmative (ever versus never) was followed by a question about how often the feature occurred with the specific headache type. The response options included never, rarely, less than half the time, and half the time or more. When nausea, photophobia, or phonophobia was reported in the affirmative, a follow-up question was also asked about the severity (mild, moderate, or severe) of the symptom.

The validity of a CATI-based diagnosis of migraine was previously assessed in the United States18 and United Kingdom.20 A random sample of individuals who met CATI criteria for migraine and a sample who did not were independently evaluated by two clinicians who used a semi-structured interview for headache diagnosis and followed IHS criteria in assigning a diagnosis.20 The diagnostic sensitivity and specificity of the CATI for migraine were 85% and 96% in the United States and 91% and 97% in the United Kingdom.18,20⇓

Definition and selection of migraine cases and controls.

Migraine diagnosis was based on the IHS criteria for migraine with or without aura.19 Eligible subjects with migraine had six or more migraine headaches reported in the last year but 15 or fewer headaches (of any type) per month in the last year. Imposing the lower bound on attack frequency helped ensure that individuals met criteria for migraine (five lifetime attacks are required for migraine without aura19) and that their attacks are sufficiently frequent so as to constitute a chronic health problem. The upper bound was intended to exclude those with transformed migraine, a disorder not included in the IHS criteria, whose relationship to migraine remains controversial.19 Controls were selected from participants who completed the CATI, did not meet the IHS criteria for migraine, and were willing to complete the SF-12 and PRIME-MD in a follow-up interview. The control population was group matched to the case population on gender, age, income, and household size.

In the US study, follow-up interviews were conducted by telephone. In the UK study, follow-up interviews were conducted in person. During the follow-up interview, in both studies, cases and controls completed the SF-12 and PRIME-MD. The SF-12 measures HRQoL in two domains: a physical component (PCS-12) and a mental component (MCS-12). This 12-item questionnaire, which contains a subset of questions from the SF-36, was designed to reproduce the Physical Component Summary Score and Mental Component Summary Score of the SF-36.21 The MCS-12 evaluates the frequency of feelings of nervousness, depression, happiness, and calmness. The PCS-12 measures the impact of health on limitations to any physical activity including climbing stairs, moving heavy objects, household work, and low impact sports such as bowling or playing golf.

The PRIME-MD is a case-finding and diagnostic instrument.22,23⇓ The first part is a patient questionnaire consisting of 25 yes–no questions designed to trigger clinical assessments of mood, anxiety, alcohol abuse or dependence, and eating and somatoform disorders. PRIME-MD has been validated using a telephone interview by a mental health specialist as the gold standard.23 It has also been implemented and validated as a computerized phone interview.24 In this study, we administered the mood module from the patient questionnaire by CATI to identify depression.

Analysis.

Demographic and selected headache characteristics were summarized for the total migraine population, the total nonmigraine population, and migraine cases and controls for each country; t-tests were used to compare the MCS-12 and PCS-12 scores between the cases and controls for both the United States and United Kingdom.

Because the data collection methods were similar in the United States and United Kingdom, and the demographic features, headache characteristics, and MCS-12 and PCS-12 scores were also comparable, we sought to combine the samples. To determine whether the data could be pooled, multivariate models for both the MCS-12 and PCS-12 were run with a main effect term for country, along with separate models with case by country interaction terms. Neither the main effect term nor the interaction term was significant in the MCS-12 or the PCS-12 models. Therefore, the samples were judged to be comparable and pooled.

Linear regression was used to model PCS-12 and MCS-12 scores, simultaneously adjusting for gender, age (30 to 39, 40 to 49, 50 to 65), education, migraine status, and depression status. Three education groups (GCSE/“O” [UK] or high school [US], technical training/HND/“A” levels [UK] or some college [US], university+ [UK] or college+ [US]) were defined. Interaction terms for migraine and depression were also evaluated.

Linear regression was also used to identify which headache characteristics were associated with HRQoL. The following headache characteristics were included in the model: headache frequency (<13 headaches per year, 13 to 48 headaches per year, 49 to 365 headaches per year), headache duration (<6 hours, 6 to 23 hours, 24 to 47 hours, 48+ hours), and headache pain intensity on a 0 to 10 scale, where 0 is no pain and 10 is pain that is as bad as it can be (<7, 7 to 8, 9 to 10). Headache interference with daily activities was rated on a three-point scale (never/rarely, <half time, ≥half time). SAS 7.0 was used for all analyses (SAS Institute Inc., Cary, NC).

Study participants.

During the initial phone surveys, 4376 interviews were completed in the United States (participation rate 62%) and 4007 were completed in the United Kingdom (participation rate 76.5%). In follow-up interviews, SF-12 and PRIME-MD data were obtained from 246 cases and 242 controls in the United States and 143 cases and 137 controls in the United Kingdom. The final sample with complete data includes 238 cases and 230 controls in the United States and 133 cases and 128 controls in the United Kingdom.

In samples from both the United States and United Kingdom, demographic features and headache characteristics of the cases and controls were generally similar to those of the larger populations (i.e., total migraine and total nonmigraine) from which they were selected (for additional information, please visit www.neurology.org and scroll down the Table of Contents for the September 12 issue to find the title link for this article). In the United States and United Kingdom, migraine participants who completed a follow-up interview resembled the entire migraine sample with regard to gender, age, race, household size, and pain intensity scores. However, headaches were slightly more frequent among cases when compared with the total migraine population.

Similarly, the US and UK controls who completed the SF-12 and PRIME-MD were comparable to the total nonmigraine population with regard to age, race, and household size. Gender differences between the controls and total nonmigraine populations occurred because migraine is more common in females and the controls were gender-matched to the migraine group. The controls were also similar to the nonmigraine population with regard to headache frequency and pain intensity score (for additional information, please visit www.neurology.org and scroll down the Table of Contents for the September 12 issue to find the title link for this article). Overall, these data indicate that the migraine cases and controls who participated in the follow-up interviews were comparable to the population samples from which they were derived.

Migraine and HRQoL.

In the United States, migraine cases had significantly lower scores on both the MCS-12 (cases 42.6, controls 51.1, p < 0.001) and PCS-12 (cases 45.5, controls 49.6, p < 0.001) scales (table 1). A similar pattern was observed in the United Kingdom (see table 1), where migraine cases again had significantly lower scores on both the MCS-12 (cases 44.8, controls 50.4, p < 0.001) and PCS-12 (cases 44.6, controls 50.3, p < 0.001). Statistically significant differences persisted after pooling US and UK data (MCS-12: cases 43.4, controls 50.9, p < 0.001; PCS-12: cases 45.2, controls 49.8, p < 0.001). For men, case and control differences did not reach statistical significance for the PCS-12 in the United States or United Kingdom alone, but did so when data were pooled (see table 1).

To control for the potential effects of gender, age, and education, a series of linear regression models were completed using the combined US and UK data. After controlling for these covariates, HRQoL scores remained significantly lower among migraine cases compared to controls for both the MCS-12 and PCS-12 (model not shown). For the MCS-12, the adjusted mean difference (case score minus control score) after controlling for gender, age, and education was −7.37 (SE = 9.36, p < 0.001). The corresponding adjusted mean difference for the PCS-12 was −5.31 (SE = 0.69, p < 0.001).

Migraine, depression, and HRQoL.

In the combined US and UK data, depression and migraine were highly comorbid; 47% (183/389) of migraineurs had PRIME-MD defined depression compared to 17% (64/379) of the controls (adjusted prevalence ratio 2.7, 95% CI 2.1 to 3.5).

The figure summarizes HRQoL scores for the MCS-12 and PCS-12 for migraine cases and controls with and without depression. Overall, depressed subjects had lower HRQoL scores than their nondepressed counterparts. There was no significant difference in MCS-12 scores between migraine cases and controls among depressed subjects (migraine 36.1, control 38.5, unadjusted p = 0.14). There was, however, a significant difference in MCS-12 scores between nondepressed migraine cases and controls (migraine 49.7, control 53.3, unadjusted p = 0.0001). A similar pattern was observed for PCS-12 scores. Among depressed subjects, the PCS-12 score was similar for migraine cases and controls (migraine 42.5, control 44.5, unadjusted p = 0.25). However, among nondepressed subjects, there was a significant difference between migraine cases and controls (migraine 47.5, control 50.9, unadjusted p = 0.0001).

• Download figure
• Open in new tab
• Download powerpoint

Figure. Mental and Physical Health Summary Scores by migraine (migraine [gray bars] versus control [black bars]) and depression (depressed versus not depressed) status. HRQoL = Health-related quality of life.

To further assess the independent and joint influence of migraine and depression on HRQoL, we separately modeled the MCS-12 and PCS-12 scores, adjusting for sociodemographic variables (table 2). Both depression and migraine were significantly and independently associated with lower MCS-12 and PCS-12 scores. Depression had a larger effect than migraine both on MCS-12 and PCS-12 scores. Migraine–depression interaction terms did not substantially alter model results (models not shown).

We also compared HRQoL for subjects with migraine with and without depression after adjusting for age, gender, and education. We found lower MCS-12 (p < 0.0001) and PCS-12 (p < 0.0001) scores for subjects with migraine and depression than for subjects with migraine without depression.

To assess the influence of headache features on HRQoL, MCS-12 and PCS-12 were modeled in relation to headache frequency and duration, pain intensity, and activity limitations (temporary disability) (table 3). Attack frequency >48 days per year was associated with a 5.6-point decrement in the MCS-12 score (p = 0.001) relative to subjects with 6 to 12 attacks per year. Disability (headache interfered with daily activities half the time or more) was associated with an independent 5-point decrement in MCS-12 (p = 0.0035) relative to subjects whose headaches never or rarely interfered with daily activity. In models that excluded the covariate for disability, pain intensity was a significant predictor of the MCS-12 score (model not shown). For PCS-12, the highest category of attack frequency was associated with a 4.7-point decrement (p = 0.0003) relative to subjects with 6 to 12 attacks per year; other attack characteristics did not reach statistical significance (see table 3). When depression was added to these models, the overall relationships between headache features and MCS-12 and PCS-12 did not change substantially (models not shown).