Ecological momentary assessment of the relationship between headache pain intensity and pain interference in women with migraine and obesity

Abstract

Background

While pain intensity during migraine headache attacks is known to be a determinant of interference with daily activities, no study has evaluated: (a) the pain intensity-interference association in real-time on a per-headache basis, (b) multiple interference domains, and (c) factors that modify the association.

Methods

Participants were 116 women with overweight/obesity and migraine seeking behavioral treatment to lose weight and decrease headaches in the Women’s Health and Migraine trial. Ecological momentary assessment, via smartphone-based 28-day headache diary, and linear mixed-effects models were used to study associations between pain intensity and total- and domain-specific interference scores using the Brief Pain Inventory. Multiple factors (e.g. pain catastrophizing (PC) and headache management self-efficacy (HMSE)) were evaluated either as independent predictors or moderators of the pain intensity-interference relationship.

Results

Pain intensity predicted degree of pain interference across all domains either as a main effect (coeff = 0.61–0.78, p < 0.001) or interaction with PC, allodynia, and HMSE (p < 0.05). Older age and greater allodynia consistently predicted higher interference, regardless of pain intensity (coeff = 0.04–0.19, p < 0.05).

Conclusions

Pain intensity is a consistent predictor of pain interference on migraine headache days. Allodynia, PC, and HMSE moderated the pain intensity-interference relationship, and may be promising targets for interventions to reduce pain interference.

Keywords

Migraine obesity ecological momentary assessment pain intensity pain interference

Introduction

Migraine is characterized by recurrent attacks of moderate-to-severe headache pain and associated symptoms that interfere with daily functioning (1 –4). However, the degree of interference (pain-related disability) and subsequent impairment of health-related quality of life (HRQoL) vary considerably among individuals with migraine (5). In one population-based study, participants with migraine and moderate-to-high levels of interference with work activities had lower HRQoL compared to those with low interference (2).

Pain intensity, perhaps more so than other migraine headache features, may help explain differences in pain interference. Pain intensity has been shown to account for twice the amount of variance as headache frequency in pain interference, measured via the Migraine Disability Assessment (MIDAS) (6). Data from the American Migraine and Prevalence Study revealed that pain severity predicted pain interference, as assessed by the Headache Impact Test-6 (HIT-6), both for episodic and chronic migraine (7). In a clinic-based sample, higher pain-intensity levels, but not higher headache frequencies, were associated with greater pain interference across several areas of functioning according to the Multidimensional Pain Inventory (8).

The abovementioned studies demonstrate a robust headache pain intensity-interference association, but are limited by certain measurement aspects. First, retrospective questionnaires were used to measure pain interference, which are likely biased because of imperfect recall (9,10). Second, the environments in which measures were administered (e.g. physician’s office) may have influenced response patterns, thereby introducing additional bias (9 –11). Similar limitations accompany measurement of migraine headache features in the above studies. Two studies assessed headache features retrospectively and one study used a paper-and-pencil diary without verification of compliance, which can affect the validity of the data (11). Finally, none of the studies conducted a thorough evaluation of factors that independently predict pain interference and/or affect the strength of the relationship between headache pain intensity and interference.

The present study involved smartphone-based ecological momentary assessment (EMA) of pain intensity and pain interference related to multiple domains of functioning (i.e. work, relationships, walking capacity, sleep, mood, and enjoyment of life). EMA counters limitations of retrospective measures (i.e. bias/inaccuracy and poor ecological validity) by collecting date- and time-verified data, in real-time, in the natural environment (9 –11). Also, because participants provide information about their headache pain every day during the monitoring period, it is possible to study dynamic patterns of the relationship between pain intensity and pain interference, thus yielding richer data (9 –11). Using this assessment approach in 116 women with both migraine and overweight/obesity, we conducted exploratory analysis of the: 1) association between pain intensity and pain interference on days that migraine headaches occurred; and 2) extent to which pain interference within each domain was predicted by demographic/anthropometric (age, race/ethnicity, body mass index (BMI)), headache (i.e. attack frequency, duration, and cutaneous allodynia), and psychosocial (i.e. depression, pain catastrophizing, and headache management self-efficacy) characteristics, either independently or in an interaction with pain intensity.

Methods

Participants and procedures

This study, involving smartphone-based EMA of migraine headache activity, included 116 women participants who were seeking behavioral treatment to lose weight and reduce headache attacks as part of the Women’s Health and Migraine (WHAM) trial (12). Participants were aged 18–50 years with overweight/obesity (BMI ≥ 25 kg/m²) who had migraine with or without aura according to International Classification for Headache Disorders third edition beta criteria (13). Participants were recruited between November 2012 and May 2015 from communities (via direct mailings, Internet sites, and social media outlets) and neurological medicine clinics. Of 673 women who contacted the research center to express interest, 515 were reached by telephone to be screened for initial eligibility. Of the 515, more than half (53.6%) were ineligible, primarily for not meeting study criteria for migraine (n = 83) or weight status (n = 62), and unwillingness to commit to the research protocol (n = 66). Of the remaining 239 individuals deemed eligible, 133 attended a study orientation/baseline visit that involved obtaining of informed consent, confirmation of migraine diagnosis by the study neurologist (JR), and objective verification of height and weight status. At this same visit, participants completed questionnaires assessing headache, psychosocial, and demographic characteristics, and began recording their headache activity for 28 consecutive days using a smartphone headache diary. Of the 133 participants who initially consented, 116 were declared fully eligible after completing the 28-day diary protocol. All measures were completed before randomization and initiation of treatment. The study protocol was approved by the Rhode Island Hospital Institutional Review Board (Providence, RI, USA). The first and last authors (DSB and JGT) assume full responsibility for the integrity of the data.

Measures

Daily monitoring of headache pain intensity, pain interference, and other clinical features

Participants recorded their migraine activity at the end of each day for 28 consecutive days using a smartphone equipped with a Web-based diary application (12). This analysis evaluates 840 diary entries on days that migraine headaches occurred. Participants responded to questions regarding migraine headache occurrence (yes/no), headache duration (hours), and the variables of primary interest—i.e. maximum pain intensity (0 (no pain) to 10 (pain as bad as you can imagine)) and pain interference. Questions used to determine level of pain interference were derived from the pain interference subscale of The Brief Pain Inventory (14). For this subscale, participants rated the degree to which pain interfered with seven domains (i.e. general activity, mood, walking ability, normal work both outside and inside the home, relations with other people, sleep, and enjoyment of life) during the past 24 hours on a 10-point scale (0 (no interference) to 10 (complete interference)). Higher scores indicate greater interference due to pain. In the current study, the association between pain intensity and total pain interference scores (ranging from 0 to 70) was of primary interest. Given that the focus of previous research on pain interference in migraine patients has been primarily limited to work and social domains, we also explored associations between pain intensity and each of the seven domains contributing to the composite pain interference score. As a result, this study provides a more global assessment of migraine burden by examining the extent to which pain interferes with daily functioning in other domains relevant to migraine including emotional status, physical activity, sleep quality, and general well-being. Moreover, obtaining more detailed information on these domains could help to inform interventions to reduce domain-specific levels of interference.

The Brief Pain Inventory has demonstrated validity and reliability in samples with different chronic pain syndromes (15). Participants’ data were time stamped and automatically transmitted each day to research staff members. In cases of missing data, research staff immediately followed up via telephone to obtain data, thereby ensuring 100% compliance.

Questionnaire-based assessment of headache, psychosocial, and demographic/ anthropometric characteristics

Headache characteristics

Cutaneous allodynia

The validated 12-item Allodynia Symptom Checklist (ACS-12) measured frequency of cutaneous allodynia, or pain in response to application of an innocuous stimuli to the skin or scalp (e.g. “taking a shower”), during headache (16). Scores of 0–2, 3–5, 6–8, and≥9 indicate none, mild, moderate, and severe allodynia, respectively. ACS-12 scores are associated with headache features and related risk factors (e.g. obesity and pain catastrophizing) in patients with migraine (17,18).

Psychosocial characteristics

Depression

The 20-item Center for Epidemiologic Studies Depression Scale (CES-D) assessed how often over the past week participants experienced depressive symptoms (19). Scores range from 0 to 60, with higher scores indicating greater depressive symptoms, and a cutoff score ≥16 indicating clinically significant depression symptoms. This measure has high internal consistency, reliability, and both predictive and concurrent validity within the context of chronic pain (20).

Pain catastrophizing

The Pain Catastrophizing Scale (PCS) evaluated catastrophic thinking related to pain (21). The PCS rates how often 13 thoughts or feelings reflecting key dimensions of catastrophizing (i.e. rumination, magnification, and helplessness) occurred while experiencing pain over the past month. PCS total scores range from 0 to 52, with higher scores indicating higher levels of pain catastrophizing, and a cutoff score of ≥30 representing clinical levels of pain catastrophizing. The PCS has established reliability and construct validity and has previously been used to evaluate levels of pain catastrophizing in patients with migraine (18,22).

Headache Management Self-Efficacy

The 25-item Headache Management Self-Efficacy Scale (HMSE) was used to assess level of confidence regarding ability to manage and prevent headache pain (23). Scores on the HMSE range from 25 to 175, with higher scores reflecting more frequent use of positive psychological coping strategies regarding headache pain. The HMSE has high internal consistency and demonstrates negative associations with headache severity, headache disability, and pain catastrophizing in patients with migraine (18,22).

Demographic/Anthropometric characteristics

Participants’ age, race/ethnicity, and level of education were assessed via questionnaire. Participants’ height and weight were measured objectively using a wall-mounted Harpenden stadiometer (Holtain Ltd., Crosswell, Crymyh, Pembs, UK) and calibrated digital scale (Tanita BWB 800: Tanita Corporation of America Inc, Arlington Heights, IL, USA). Participants were weighed in light street clothing, without shoes, and to the nearest 0.1 kg. BMI was calculated using the formula: BMI (kg/m²) = weight (kg) / (height [m])².

Analytic approach

Analyses were conducted in 2015 using IBM SPSS Statistics for Windows, Version 20.0 (IBM Corp., 2011, Armonk, NY; http://www.spss.com). Demographic/anthropometric, headache, and psychosocial characteristics were summarized using means and standard deviations (SD) or counts with percentages, as appropriate. Unconditional linear mixed-effects models with intercepts treated as a random effect were used to calculate means and standard errors (SE) for ratings of maximum pain intensity, the seven domains of pain interference, and a pain interference total score (mean of the seven domains) (14). Given that ratings of pain interference were captured daily, on days that headaches occurred, each participant could have contributed up to 28 ratings of maximum pain intensity and pain interference to the analyses. Linear mixed-effects models allow for such repeated measures designs with an unequal number of observations per participant (24).

A series of linear mixed-effects models were used to examine predictors of pain interference. First, each domain of pain interference and the pain interference total score were modeled using maximum pain intensity from the concurrent diary entry as the only predictor. Variance estimates for intercepts, and the maximum pain intensity slopes, were statistically significant (p < 0.05) in all cases and were therefore treated as random effects in all analyses. Second, additional predictors were added to the model including three demographic/anthropometric characteristics (age, Non-Hispanic white race/ethnicity versus other, BMI), three headache characteristics (total number of headache days and average duration of headaches during the 28-day monitoring period; ACS-12), three psychosocial measures (CES-D, PCS, HMSE), and their interaction with maximum pain intensity. A backward elimination procedure was used to produce final models with only those predictors that were statistically significant at p < 0.05 with two-tailed tests for each domain of pain interference and the pain interference total score (25). The model building process was conducted manually to ensure that an automated algorithm did not drop key variables of interest and that the model did not become over parameterized.

Results

Demographic/anthropometric, headache, and psychosocial characteristics of the 116 participants are reported in Table 1. Participants were mixed in terms of age, race, ethnicity, and level of education. The mean CES-D score of 16.2 (11.3) met or exceeded the cutoff of ≥16, which is indicative of clinically significant depressive symptoms; 51 (44.0%) participants exceeded this cutoff. Likewise, 29 (25.0%) participants met or exceeded the cutoff of ≥ 30 for clinically significant pain catastrophizing, and 46 (40.0%) met or exceeded the cutoff of ≥ 6 for moderate or greater allodynia.

Table 1.

Participant characteristics.

Demographic and anthropometric characteristics (n = 116)
Age, mean (±SD), years	38.2	(8.1)
Race, n (%)
African American	11	(9.5%)
White	90	(77.6%)
Other	12	(10.3%)
Mixed	3	(2.6%)
Ethnicity, n (%)
Non-Hispanic	97	(83.6)
Hispanic	19	(16.4)
Education, n (%)
< 4 year college/university degree	53	(45.7%)
≥ 4 year college degree	63	(54.3%)
Body mass index, mean (±SD), kg/m²	35.0	(6.4)
Headache characteristics
Migraine headache days/month, mean (±SD), number	8.8	(5.5)
Total migraine headache duration/month, mean (±SD), hour	113.4	(100.1)
Maximum pain intensity, mean (±SE), 0–10 scale	6.0	(0.1)
Cutaneous allodynia, mean (±SD)	5.1	(3.9)
Pain Interference, mean (±SE)
Total interference	3.5	(0.2)
General activity	4.0	(0.2)
Mood	4.6	(0.2)
Walking ability	1.9	(0.2)
Normal work	3.6	(0.2)
Relations with other people	3.4	(0.2)
Sleep	2.9	(0.2)
Enjoyment of life	4.3	(0.2)
Psychosocial characteristics
Pain catastrophizing, mean (±SD)	22.4	(10.8)
Depressive symptoms, mean (±SD)	16.2	(11.3)
Headache management self-efficacy, mean (±SD)	98.3	(22.9)

Adherence to the protocol for daily monitoring of headache pain intensity, pain interference, and other clinical headache features was 100% because of the daily adherence checks described above. Each participant contributed a mean (SD) of 7.3 (5.1) ratings of maximum pain intensity and pain interference to the analysis. This is less than the average of 8.8 (5.5) migraine headache days reported by each participant during the 28-day monitoring period because some headaches spanned more than one day but were rated only at the end of the headache.

Predictors of pain interference

Results of the linear mixed-effects models used to examine predictors of pain interference are reported in Table 2. We first evaluate predictors of total pain interference and then examine predictors of each separate domain of pain interference in a series of linear mixed-effects models.

Table 2.

Results of linear mixed effects models used to predict pain interference (total and seven individual domains) from maximum pain intensity, demographic/anthropometric, headache, and psychosocial characteristics.

	Total interference				General activity				Mood				Walking ability
	Coeff	SE	p		Coeff	SE	p		Coeff	SE	p		Coeff	SE	p
Intercept	–2.897	0.958	0.003		–2.564	1.125	0.024		–0.310	0.516	0.549		–2.606	0.889	0.005
MPI	0.501	0.096	< 0.001		0.659	.0118	< .001		0.777	0.057	< 0.001		0.045	0.129	0.73
Age (years)	0.049	0.018	0.007		0.055	0.021	0.009		–	–	–		0.037	0.013	0.006
BMI	–	–	–	–	–	–	–	–	–	–	–		0.033	0.016	0.047
PCS	–0.025	0.025	0.320		–0.045	0.030	0.140		–	–	–		-0.043	0.022	0.048
ASC-12	0.125	0.038	0.001		–	–	–	–	–	–	–		-0.032	0.056	0.570
CES-D	–	–	–	–	–	–	–		0.060	0.019	0.002		–	–	–
Migraine days	–	–	–	–	–	–	–		–0.083	0.038	0.029		–	–	–
MPI*PCS	0.010	0.004	0.007		0.011	0.005	0.017		–	–	–		0.013	0.005	0.009
MPI*ASC-12	–	–	–	–	–	–	–	–	–	–	–		0.030	0.013	0.027

	Normal work				Relations with other people				Sleep				Enjoyment of life
	Coeff	SE	p		Coeff	SE	p		Coeff	SE	p		Coeff	SE	p
Intercept	–3.775	1.239	0.003		–2.473	1.518	0.110		–4.477	0.979	< 0.001		–3.961	1.020	< .001
MPI	0.599	0.135	< 0.001		1.147	0.234	< 0.001		0.608	0.061	< 0.001		0.778	0.058	< .001
Age (years)	0.058	0.021	0.008		–	–	–		0.044	0.022	0.049		0.072	0.024	0.003
PCS	–0.019	0.035	0.590		–	–	–		–	–	–	–	–	–	–
ASC-12	0.105	0.045	0.021		0.119	0.047	0.013		0.192	0.017	< 0.001		0.155	0.049	0.002
CES-D	–	–	–		–	–	–		0.063	0.017	< 0.001		–	–	–
HMSE	–	–	–		0.012	0.014	0.410		–	–	–	–	–	–	–
MPI*PCS	0.011	0.005	0.042		–	–	–		–	–	–	–	–	–	–
MPI*HMSE	–	–	–		–0.005	0.002	0.040		–	–	–	–	–	–	–

BMI: body mass index; MPI: maximum pain intensity; PCS: Pain Catastrophizing Scale; ASC: Allodynia Symptom Checklist-12; CES-D: Center for Epidemiologic Studies Depression Scale Revised; HMSE: Headache Management Self-Efficacy Scale.

Total interference

Mean levels of total interference were higher for individuals who were older (coeff = 0.049, SE = 0.018, p = 0.007) and those who reported higher levels of allodynia (coeff = 0.125, SE = 0.038, p = 0.001). An interaction between maximum pain intensity and pain catastrophizing showed that higher pain intensity contributed to higher total interference to a greater degree as catastrophizing increased (coeff = 0.010, SE = 0.004, p = 0.007; Figure 1).

Figure 1.

Model estimates of total pain interference as predicted by maximum pain severity, pain catastrophizing, and their interaction. Pain catastrophizing is represented at three levels: high (one standard deviation above the mean), moderate (the mean), and low (one standard deviation below the mean). The association between maximum pain severity and pain interference was stronger at higher levels of pain catastrophizing.

Interference with general activity

Mean levels of interference with general activity were higher for older individuals (coeff = 0.055, SE = 0.021, p = 0.009). An interaction between maximum pain intensity and pain catastrophizing showed that higher pain intensity contributed to higher interference with general activity to a greater degree as catastrophizing increased (coeff = 0.011, SE = 0.005, p = 0.017).

Interference with mood

Mean levels of interference with mood were higher for individuals who reported more severe depressive symptoms (coeff = 0.060, SE = 0.019, p = 0.002). Greater maximum pain intensity was associated with higher levels of mood impairment (coeff = 0.777, SE = 0.057, p < 0.001). There was a negative effect of the total number of migraine days, such that individuals with more migraine days during the 28-day monitoring period reported lower mean levels of mood interference (coeff = –0.083, SE = 0.038, p = 0.029).

Interference with walking ability

Mean levels of interference with walking ability were higher for older individuals (coeff = 0.037, SE = 0.013, p = 0.006) and those with higher BMIs (coeff = 0.033, SE = 0.016, p = 0.047). An interaction between maximum pain intensity and pain catastrophizing showed that higher pain intensity contributed to higher interference with walking ability to a greater degree as catastrophizing increased (coeff = 0.013, SE = 0.005, p = 0.009). A similar pattern was observed for the interaction between maximum pain intensity and allodynia (coeff = 0.030, SE = 0.013, p = 0.027).

Interference with normal work

Mean levels of interference with normal work were higher for older individuals (coeff = 0.058, SE = 0.021, p = 0.008) and those who reported higher levels of allodynia (coeff = 0.105, SE = 0.045, p = 0.021). An interaction between maximum pain intensity and pain catastrophizing showed that higher pain intensity contributed to higher interference with normal work to a greater degree as catastrophizing increased (coeff = 0.011, SE = 0.005, p = 0.042).

Interference with relations with other people

Mean levels of interference with relations with other people were higher for individuals who reported higher levels of allodynia (coeff = 0.119, SE = 0.047, p = 0.013). An interaction between maximum pain intensity and headache management self-efficacy showed that the contribution of pain intensity to interference with relations with other people became weaker as a function of increasing self-efficacy (coeff = –0.005, SE = 0.002, p = 0.040).

Interference with sleep

Mean levels of interference with sleep were higher for older individuals (coeff = 0.044, SE = 0.022, p = 0.049) and those who reported higher levels of allodynia (coeff = 0.192, SE = 0.017, p < 0.001) and depressive symptoms (coeff = 0.063, SE = 0.017, p < 0.001). Greater maximum pain intensity was also associated with higher levels of sleep interference (coeff = 0.608, SE = 0.061, p < 0.001).

Interference with enjoyment of life

Mean levels of interference with enjoyment of life were higher for older individuals (coeff = 0.072, SE = 0.024, p = 0.003) and those who reported higher levels of allodynia (coeff = 0.155, SE = 0.049, p = 0.002). Greater maximum pain intensity was also associated with higher levels of interference with enjoyment of life (coeff = 0.778, SE = 0.058, p < 0.001).

Discussion

This study is the first to naturalistically examine the daily relationship between pain intensity and pain interference in women with migraine via a smartphone-based EMA over a 28-day period. Additionally, we assessed the influence of multiple demographic, anthropometric, headache, and psychosocial characteristics on pain interference and the relationship between pain intensity and pain interference. Findings showed that pain intensity was associated with pain interference across all domains either as an independent positive predictor or in combination with moderators including pain catastrophizing, allodynia, or headache management self-efficacy. Moreover, with the exception of mood-related pain interference, for which attack frequency emerged as an independent predictor (possibly signaling that individuals with more frequent attacks, who also have higher rates of depression (26), have less mood-related interference given that their headache and mood patterns are more predictable), pain intensity was the only headache feature consistently associated with pain interference. These findings concur with previous research that shows pain intensity is an important determinant of pain interference across several domains of functioning (6 -8). Taken together, the above findings suggest that in women with migraine and overweight/obesity, higher levels of pain intensity contribute to higher levels of daily pain interference, irrespective of attack frequency (6-8). In other words, pain interference on a particular headache day is not associated with the total number of migraine days for most measures.

Another pattern of findings was that only the variables most proximally related to pain (i.e. pain catastrophizing, allodynia) and headache management self-efficacy moderated the pain intensity-interference relationship. Specifically, higher levels of catastrophizing interacted with higher levels of pain intensity to predict higher levels of interference related to general activity and overall. Additionally, catastrophizing and allodynia each interacted with pain intensity to adversely affect the extent to which pain interfered with walking ability. This suggests that individuals with migraine who engage in more frequent catastrophizing and/or have more severe allodynia experience more pain interference. Reasons for these interactions are not entirely clear, but catastrophizing and allodynia, which promote pain hypervigilance, may reduce the ability to detach from pain (even after a headache has ended), thereby perpetuating the pain experience and ultimately resulting in more pain interference, particularly with respect to ambulatory function and activities of daily living. While pain catastrophizing and allodynia are demonstrated predictors of both headache features and pain interference in individuals with migraine (16,17), including those with obesity (18), this study is the first to show that these variables reliably contribute to the influence of pain intensity on pain-related interference in multiple domains of functioning on a per headache basis. Results showing that higher levels of pain intensity and lower levels of headache management self-efficacy interacted to predict more pain interference related to social relationships suggest that the need to avoid social activities may be greater when headache pain is more severe and poorly managed. Also, given that individuals with migraine tend to use maladaptive stress-coping strategies, such as social isolation, to a greater degree than controls (27,28), this tendency may be amplified when pain intensity is high and perceived ability to cope with headache pain is low.

Results revealed several other independent predictors of pain interference, the most consistent of which were allodynia and age. Older age and greater allodynia contributed to more pain interference across several individual domains and overall, regardless of pain intensity. Allodynia has been associated with longer headache duration and a reduced response to treatment, possibly contributing to these findings (16,17,29). More-severe depressive symptoms also independently predicted more pain-related sleep and mood interference. Previous research suggests that sleep problems are common in people with migraine and that depression symptoms are partially responsible (30). Future EMA studies involving more detailed daily assessments of mood and sleep habits, along with migraine activity, are warranted to elucidate the complex interrelationships among these variables. Finally, higher BMI contributed to greater interference with walking ability, independent of pain intensity. This finding supports our previous work showing that higher BMI, but not migraine features, are related to lower objectively measured daily physical activity levels in women with migraine (31).

Taking a broader perspective, the current study raises the question of why and how the level of pain intensity experienced by any given patient with migraine could affect different types of pain interference to a different degree. Our findings suggest that the mechanisms by which pain intensity contributes to pain interference may vary for different types of pain interference and/or that the pathways are modulated by different biological (e.g. allodynia) and psychosocial (e.g. pain catastrophizing) factors. It may therefore be important to consider that two patients with the same level of migraine pain may experience substantially different interference profiles, and that it may be possible to capture and explain this phenomenon via a sophisticated measurement and modeling approach such as the one employed in this study.

This study has important strengths. It investigates dynamic patterns of pain intensity and pain interference among women with migraine in the natural environment via EMA, thereby countering issues related to bias and ecological validity inherent to use of retrospective measures. Additionally, novel information provided by this study regarding psychosocial and headache-related factors that modify the migraine pain intensity-interference relationship carries potentially important clinical and research implications. For example, our findings warrant future experimental studies to test whether interventions targeting high pain catastrophizing can effect reductions in pain intensity and related interference. Moreover, given the influence of allodynia on pain intensity and related interference in the present study and potential links between increased catastrophizing and increased pain sensitivity in individuals with migraine (18), it is especially important to understand whether such reductions occur through favorable changes in central pain sensitization. Finally, while the exploratory nature of the analysis does not warrant swift changes to clinical practice, our findings do suggest that routine screening of patients for catastrophizing and allodynia may be helpful in developing a more effectively tailored treatment plan that addresses both psychological and biological contributors to pain and related interference.

This study also has certain limitations. Past research shows that obesity in people with migraine is associated with greater pain intensity and pain interference (18,32,33), as well as several of the factors that moderated the pain intensity-interference relationship in the current study (17,18,34,35). The failure of BMI to emerge as a more consistent predictor of pain intensity and related interference may be partially accounted for by our sample, which was restricted to women who had overweight/obesity. Consequently, future similar studies should include normal-weight women with migraine. Similarly, given that this study was limited to women of reproductive ages, it is unclear whether the findings generalize to women who are older and men who have migraine. The failure of migraine days to predict pain interference is likely the result of the measurement model, in which pain interference was measured only on days that migraine occurred. Thus, our model tested whether the total number of migraine days was associated with pain interference only on days that migraine occurred. Because ratings of pain interference were not available on days that migraine did not occur, we cannot test for an association between total migraine days and cumulative pain interference across the full 28-day monitoring period. Additionally, because participants rated maximum pain intensity and pain interference concurrently, the potential for causal inference is limited. Finally, it is not known if pain may have affected cognitive functioning and possibly subsequently interfered with ability to report on pain experiences in real-time and in the natural environment. While we received no anecdotal reports of this occurring in the study, this could be an important topic for future research employing EMA in patients with migraine and other types of pain.

It is also important to acknowledge the exploratory nature of the analysis, which involved a large number of statistical comparisons without correction for type I error. This approach had the advantage of preserving statistical power to detect the interactions that were of particular interest, but it raised the risk of spurious associations. In addition, the backward selection procedure used to build the statistical models can sometimes produce results that are sample specific and that do not fully generalize to other samples or populations (25). Therefore, additional research using other samples is needed to validate the findings of this study, and confirm their utility for clinical applications.

Conclusion

In a sample of women with migraine and obesity, this study employed a smartphone-based EMA to examine the relationship between pain intensity and pain interference on migraine days over a 28-day monitoring period. The main findings of this study were: 1) pain intensity was a consistent predictor of pain interference across multiple domains of functioning on migraine days; 2) only variables most proximally associated with pain (i.e. pain catastrophizing, allodynia) and headache management self-efficacy interacted with pain intensity to predict greater pain interference; and 3) older age and more severe allodynia consistently predicted higher levels of pain interference, regardless of levels of pain intensity. This work sets the stage for experimental research aimed at determining whether interventions targeting pain catastrophizing, allodynia, and/or headache management self-efficacy could reduce the negative impact of headache pain on pain interference in individuals with migraine.

Article highlights

Use of an ecological momentary assessment approach implemented via smartphone technology allows for the study of dynamic patterns in associations between pain intensity and pain interference.

Pain intensity is a consistent predictor of the degree to which pain interferes with multiple domains of daily functioning on migraine headache days.

Allodynia, pain catastrophizing, and headache management self-efficacy moderate the pain intensity-interference relationship, and may be promising targets for interventions to reduce pain interference.

Footnotes

Acknowledgments

The authors wish to thank Tiffany Leblond and Krystal Defaria for assisting with data collection.

ClinicalTrials.gov Identifier: NCT01197196.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: J. Graham Thomas, PhD, received research support from the National Institutes of Health (NIH) R42 DK103537 (principal investigator), R01 DK095779 (principal investigator), R41 HL114046 (principal investigator), R01 NS077925 (investigator), Weight Watchers International Inc; and serves as a consultant to Applied VR, KetoThrive, and Weight Watchers International Inc. Jelena Pavlovic, MD, PhD, received consulting honoraria from Allergan Inc. Richard B. Lipton, MD, received research support from the NIH (PO1 AG03949 (program director), PO1AG027734 (project leader), RO1AG025119 (investigator), RO1AG022374-06A2 (investigator), RO1AG034119 (investigator), RO1AG12101 (investigator), the National Headache Foundation, and the Migraine Research Fund; serves on the editorial board of Neurology and as senior advisor to Headache; has reviewed for the National Institute on Aging (NIA) and National Institute of Neurological Disorders and Stroke (NINDS); holds stock options in eNeura Therapeutics; and serves as consultant, advisory board member, or has received honoraria from: Allergan, American Headache Society, Autonomic Technologies, Boston Scientific, Colucid, Eli Lilly, Endo, eNeura Therapeutics, Novartis, and Teva. Dawn C. Buse, PhD, received grant support and honoraria from Allergan Pharmaceuticals, The American Headache Society and the National Headache Foundation. Dale S. Bond, PhD, received research support from the NIH/NINDS, R01 NS077925 (principal investigator) and travel support from The American Headache Society. Julie Roth, MD, Lucille Rathier, PhD, Kevin O’Leary, MS, and E. Whitney Evans, PhD, declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: grants from the National Institutes of Health (R01 NS077925, PI: Bond and P01 AG003949, PI: Lipton).

References

Lipton

Stewart

Diamond

. Prevalence and burden of migraine in the United States: Data from the American Migraine Study II. Headache 2001; 41: 646–657.

Lipton

Liberman

Kolodner

. Migraine headache disability and health-related quality-of-life: A population-based case-control study from England. Cephalalgia 2003; 23: 441–450.

Lipton

Bigal

Kolodner

. The family impact of migraine: Population-based studies in the USA and UK. Cephalalgia 2003; 23: 429–440.

Raggi

Leonardi

Bussone

. Value and utility of disease-specific and generic instruments for assessing disability in patients with migraine, and their relationships with health-related quality of life. Neurol Sci 2011; 32: 387–392.

Stewart

Shechter

Lipton

. Migraine heterogeneity. Disability, pain intensity, and attack frequency and duration. Neurology 1994; 44(6 Suppl 4): S24–S39.

Stewart

Lipton

Kolodner

. Migraine disability assessment (MIDAS) score: Relation to headache frequency, pain intensity, and headache symptoms. Headache 2003; 43: 258–265.

Buse

Manack

Serrano

. Headache impact of chronic and episodic migraine: Results from the American Migraine Prevalence and Prevention study. Headache 2012; 52: 3–17.

Magnusson

Becker

. Migraine frequency and intensity: Relationship with disability and psychological factors. Headache 2003; 43: 1049–1059.

Stone

Broderick

. Real-time data collection for pain: Appraisal and current status. Pain Med 2007; 8(Suppl 3): S85–S93.

10.

Kikuchi

Yoshiuchi

Miyasaka

. Reliability of recalled self-report on headache intensity: Investigation using ecological momentary assessment technique. Cephalalgia 2006; 26: 1335–1343.

11.

Gendreau

Hufford

Stone

. Measuring clinical pain in chronic widespread pain: Selected methodological issues. Best Pract Res Clin Rheumatol 2003; 17: 575–592.

12.

Bond

O’Leary

Thomas

. Can weight loss improve migraine headaches in obese women? Rationale and design of the Women’s Health and Migraine (WHAM) randomized controlled trial. Contemp Clin Trials 2013; 35: 133–144.

13.

Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia 2013; 33: 629–808.

14.

Cleeland

Ryan

. Pain assessment: Global use of the Brief Pain Inventory. Ann Acad Med Singapore 1994; 23: 129–138.

15.

Tan

Jensen

Thornby

. Validation of the Brief Pain Inventory for chronic nonmalignant pain. J Pain 2004; 5: 133–137.

16.

Lipton

Bigal

Ashina

. Cutaneous allodynia in the migraine population. Ann Neurol 2008; 63: 148–158.

17.

Bigal

Ashina

Burstein

. Prevalence and characteristics of allodynia in headache sufferers: A population study. Neurology 2008; 70: 1525–1533.

18.

Bond

Buse

Lipton

. Clinical pain catastrophizing in women with migraine and obesity. Headache 2015; 55: 923–933.

19.

Radloff

. The CES-D Scale: A self-report depressive scale for research in the general population. J Appl Psychol Meas 1977; 1: 385–401.

20.

Wong

Chen

Yap

. Assessing depression in patients with chronic pain: A comparison of three rating scales. J Affect Disord 2011; 133: 179–187.

21.

Sullivan

Bishop

Pivik

. The Pain Catastrophizing Scale: Development and validation. Psychol Assess 1995; 7: 524–532.

22.

Holroyd

Drew

Cottrell

. Impaired functioning and quality of life in severe migraine: The role of catastrophizing and associated symptoms. Cephalalgia 2007; 27: 1156–1165.

23.

French

Holroyd

Pinell

. Perceived self-efficacy and headache-related disability. Headache 2000; 40: 647–656.

24.

Woltman

Feldstain

MacKay

. An introduction to hierarchical linear modeling. TQMP 2012; 8: 52–69.

25.

Núñez

Steyerberg

Núñez

. Regression modeling strategies [article in Spanish]. Rev Esp Cardiol 2011; 64: 501–507.

26.

Zwart

Dyb

Hagen

. Depression and anxiety disorders associated with headache frequency. The Nord-Trøndelag Health Study. Eur J Neurol 2003; 10: 147–152.

27.

Huber

Henrich

. Personality traits and stress sensitivity in migraine patients. Behav Med 2003; 29: 4–13.

28.

Stronks

Tulen

Pepplinkhuizen

. Personality traits and psychological reactions to mental stress of female migraine patients. Cephalalgia 1999; 19: 566–574.

29.

Burstein

Yarnitsky

Goor-Aryeh

. An association between migraine and cutaneous allodynia. Ann Neurol 2000; 47: 614–624.

30.

Walters

Hamer

Smitherman

. Sleep disturbance and affective comorbidity among episodic migraineurs. Headache 2014; 54: 116–124.

31.

Bond

Thomas

O’Leary

. Objectively measured physical activity in obese women with and without migraine. Cephalalgia 2015; 35: 886–893.

32.

Bigal

Liberman

Lipton

. Obesity and migraine: A population study. Neurology 2006; 66: 545–550.

33.

Bigal

Tsang

Loder

. Body mass index and episodic headaches: A population-based study. Arch Intern Med 2007; 167: 1964–1970.

34.

Tietjen

Peterlin

Brandes

. Depression and anxiety: Effect on the migraine-obesity relationship. Headache 2007; 47: 866–875.

35.

Bond

Roth

Nash

. Migraine and obesity: Epidemiology, possible mechanisms and the potential role of weight loss treatment. Obes Rev 2011; 12: e362–e371.