Pain Medication Beliefs in Individuals with Headache

Abstract

Objective:

To evaluate pain medication beliefs in a community sample of individuals with headache.

Background:

Previous studies of medication adherence for individuals with headache have identified a high rate of prescription nonfulfillment, frequent medication discontinuation, and widely varying levels of medication-related satisfaction. Still, there is a limited understanding of how these individuals view their medications and their relationships with healthcare providers. Insight into these perceptions could prove useful in explaining medication adherence behaviors.

Methods:

In this secondary analysis of a cross-sectional study, data from N = 215 adults with headache were analyzed. Participants completed the Pain Medication Attitudes Questionnaire (PMAQ), Center for Epidemiologic Studies Depression Scale (CES-D), State-Trait Anxiety Inventory Form Y-2 (STAI-T), Weekly Stress Inventory Short Form (WSI-SF), and Migraine Disability Scale (MIDAS). These participants also provided a list of their current pain medications.

Results:

Using the PMAQ, participants could be characterized as having medication beliefs that were ‘Trusting and unconcerned’ (83/215 [38.6%]), ‘Skeptical and somewhat worried’ (99/215 [46.0%]), or ‘Skeptical and concerned’ (n = 33/215 [15.3%]). Individuals with skeptical and concerned beliefs expressed elevated concerns (z > 1.15) about side effects, scrutiny, perceived need, tolerance, withdrawal, and addiction. Individuals who were trusting and unconcerned expressed low levels (z < −0.40) of these beliefs. Increasing levels of mistrust and medication concerns were correlated with higher depression scores on the CES-D, with values ranging from r = 0.23 to r = 0.38.

Conclusions:

Subgroups of pain medication beliefs were identified, including two groups of patients with at least some concerns about their medical providers. Beliefs ranged from a lack of concern about using pain medications to worries about scrutiny and harm. It is unclear if poor experiences with pain medications cause these beliefs or if they prevent individuals from effectively utilizing medications. Additionally, more negative beliefs about pain medications were associated with more depressive symptoms.

Plain Language Summary

Studies have shown that while most people with headaches use medicines to treat their attacks, they are not always happy with these medicines, and they often do not fill their prescriptions. To better understand why this may be, we looked at beliefs about medication in 215 individuals who have headaches and experience using pain medicines. We found that some individuals reported little concern about their medications, while others reported high levels of worry about side effects or becoming dependent on them.

Introduction

Medication use is integral to managing symptoms for most individuals with primary headache disorders.^1–3 With 98% of individuals reporting the use of abortive medications and 12.4–16.8% reporting the use of preventative medications,^2,4 most individuals with headache have some experience with the use of pain-relieving medications. However, little is known about how individuals view these medications’ benefits, risks, and social implications.

Several sources of evidence support the notion that individuals experiencing headache disorders have a complicated view of their medications. Previous investigations into medication adherence for those experiencing primary headache disorders have found a high rate of nonfulfillment of prescriptions,^5,6 frequent medication discontinuation or switching,⁶ and a wide variety of medication-related satisfaction levels.⁷ However, the reasons behind these phenomena are less well known. A few studies in individuals with headache have considered various attitudes or beliefs about the role of medications as a ‘problem solving’ method for dealing with pain⁸ or as part of an illness schema⁹ and the elements of decision-making related to medication usage.¹⁰ These studies examined how individuals can assess the experience of pain and related disability as a problem to be solved while viewing medication-taking behavior as a decision that balances immediate reward (e.g., alleviation of the current attack) with future consequences (e.g., running out of prescribed triptan doses before the next attack).

How patients with headache view their medications in the context of several common pain medication-related attitudes or beliefs (e.g., fear of addiction, fear of tolerance, social stigma) is not well understood. We postulate that pain-medication attitudes, beliefs, and fears are multifaceted and could be related to direct experience with the medications. They also could be indirectly acquired through information from media, family, or friends, and they could be amplified by internal states such as anxiety or depression. Insight into these perceptions could prove useful in explaining medication adherence behaviors and treatment-seeking preferences. This study aims to evaluate various pain medication beliefs using the Pain Medication Attitudes Questionnaire (PMAQ)¹¹ in a community sample of participants with primary headache disorders. Based on previous work,^8–10 we hypothesized that subgroups of individuals with similar medication beliefs could be identified, that these individuals would differ meaningfully on these beliefs, and that these beliefs would be associated with measures of anxiety and depression.

Methods

This is a post hoc secondary analysis of data collected in the Headache and Cognition Study (HACOGS). The four pre-planned HACOGS sub-studies and one post hoc secondary analysis have been previously published.^12–16 Data collection for this cross-sectional study was conducted from May 2012 to May 2014. In a single research laboratory visit, participants completed questionnaires about their headache activity and related cognitions. The study was approved by the Institutional Review Board of the institution where data collection occurred, and all participants provided written informed consent prior to beginning study procedures.

Participants were recruited through local advertisements or headache specialist referrals. Eligible individuals were 18 years old or older and had experienced more than five migraine, tension-type, or cluster headache attacks in their lives. They also had experience using over-the-counter or prescription headache medications. Individuals were excluded from participation if they did not meet the age requirement, had not experienced the required number of headaches, or did not have experience using headache medications.

Questionnaires

As part of the study protocol, participants completed several questionnaires and listed their current pain medications. The Pain Medication Attitudes Questionnaire (PMAQ)¹¹ was the focus of this study. This 47-item scale was developed to assess the beliefs and concerns about pain-related medications in individuals with chronic pain, including individuals with headache. The questionnaire was designed to be broadly applicable to beliefs about a wide range of medications, including those used for prophylactic therapy, acute pain treatment, and abortive or palliative treatments. Each item of the PMAQ is scored from 0 (‘Never True’) to 5 (‘Always True’), with higher scores indicating greater concern about a specific medication belief. After consulting with focus groups and pain-treatment experts and after completing a validation study, the developers identified seven thematic concerns.

Addiction (ɑ = 0.85; score range: 0 to 25):

With items such as “I worry about becoming addicted to my pain medication/s,” this scale measures an individual’s concern about the addictive potential of their medications or about the potential to lose control over their use.

Need (ɑ = 0.82; score range: 0 to 40):

With items such as “I rely on my pain medication/s” and “I fear that I may run out of my pain medication/s,” this scale measures an individual’s perceived need of medication for successful management of their pain disorder.

Scrutiny (ɑ = 0.82; score range: 0 to 40):

With items such as “I worry about how other people view my use of pain medication/s” and “I try to hide the fact that I use pain medication/s,” this scale measures an individual’s concern about the social ramifications and presumed scrutiny for how medication use is viewed by others.

Side Effects (ɑ = 0.84; score range: 0 to 35):

With items such as “I have concerns about the side effects from my pain medication/s” and “I worry that my pain medication/s have an effect on my ability to think,” this scale measures an individual’s concern about unwanted side effects and the perceived harm of their pain medications.

Tolerance (ɑ = 0.77; score range: 0 to 30):

With items such as “I worry that over time I will need more pain medication/s” and “I worry that my pain medication/s will stop working,” this scale measures an individual’s concern about the loss of efficacy of their medication or about the need to escalate the dose to retain therapeutic effects.

Mistrust of Doctors (ɑ = 0.82; score range: 0 to 35):

With items such as “I worry that I have been told different information about my pain medication/s by different doctors” and “I am afraid that I am being prescribed the wrong pain medication/s,” this scale measures an individual’s lack of trust in their doctor and prescribed treatment regimen.

Withdrawal (ɑ = 0.84; score range: 0 to 30):

With items such as “I worry that I will have some withdrawal symptoms if I stop my medication” and “I am afraid that stopping my pain medication/s will cause me to feel ill,” this scale measures an individual’s concern that they would experience withdrawal effects or have an adverse response to stopping their pain medications.

Four additional questionnaires were administered to evaluate symptoms of affective distress, stress, and headache-related disability. The 20-item Center for Epidemiologic Studies Depression Scale (CES-D) was used to assess depressive symptoms, with a score > 16 used as a clinical cutoff for the risk of major depression.¹⁷ The State-Trait Anxiety Inventory Form Y-2 (STAI-T), another 20-item questionnaire, was employed to measure trait anxiety, with a score > 40 used as a cutoff for the risk of clinical anxiety.¹⁸ To assess minor stressful events participants had experienced in the past week, the 25-item Weekly Stress Inventory Short Form (WSI-SF) was used.¹⁹ The sample distribution was used to determine “low” (z < 0), “moderate” (z near 0), or “high” (z > 0) scores on the WSI-SF. Finally, migraine-related disability was assessed with the Migraine Disability Scale (MIDAS), with scores <=5 interpreted as “little or no disability,” 6 to 10 as “mild,” 11 to 20 as “moderate,” and >= 21 as “severe”.²⁰ The reliability and validity of these questionnaires have been well studied in research and clinical settings and have been demonstrated to be suitable for such uses.^17–19,21

Sample Size

The sample size for the HACOG study was based on estimating the precision of a 95%CI for proportions.^12,13 The original study included N = 300 participants. However, the PMAQ questionnaire was added to the assessment battery after enrollment had begun and was not completed by all participants. This analysis was based on the available data from the N = 215 individuals who completed all of the questionnaires, including the PMAQ.

Statistical Analyses

Descriptive statistics that included median [25th, 75th] for continuously scaled variables and frequency counts (%) for categorical variables were used to summarize sample characteristics. To examine the individual beliefs and concerns about pain medications in the sample, a hierarchical cluster analysis was conducted on the standardized (z-scores) subscales of the PMAQ. Euclidian distances were used to estimate distances, and a cluster solution was selected using Ward’s agglomeration method. The subgroup profiles were then plotted by displaying the cluster scores from an accompanying principal component analysis (PCA) and characterized based on their mean and SD of z-scores for all of the questionnaires. Omnibus differences across cluster groups were evaluated using Kruskal-Wallis or chi-square tests. Pearson correlation coefficients and scatter plots were used to estimate and display the correlations between the PMAQ subscales. All analyses were two-tailed with a p < 0.05 interpreted for statistical significance. The statistical analyses were conducted using R 4.2.1 and R-studio 2023.06.1+524.

Results

During the study, N = 215 participants completed the questionnaires that included the PMAQ. As previously described, to increase anonymity and decrease participant burden, we collected minimal demographic, headache diagnostic, and protected health information.^12–15 There were no missing data for any item on any of the administered instruments for these individuals. Table 1 displays the descriptive statistics for each of the questionnaires, as well as the frequency of various medications reported to be used by the sample. There was substantial variability in each of the instruments and in medication usage, so analyses aimed at describing the associations across the PMAQ subscale measurements were employed.

Table 1.

Questionnaire Data and Self-Reported Medication Use Stratified by Cluster Group

	Skeptical and somewhat worried	Trusting and unconcerned	Skeptical and concerned	p-value
	n = 99	n = 83	n = 33
MIDAS	24.0 [11.0, 46.8]	13.0 [6.0, 31.0]	27.0 [13.0, 59.0]	0.001
CES-D	16.0 [7.5, 27.0]	8.4 [5.5, 16.0]	22.0 [16.0, 32.0]	<0.001
STAI-T	41.0 [31.0, 50.0]	33.0 [27.0, 39.5]	45.0 [37.0, 55.0]	<0.001
WSI-SF	37.0 [19.0, 61.0]	30.0 [17.5, 44.0]	49.0 [35.0, 85.0]	0.001
PMAQ
Addiction	1.0 [0.0, 3.0]	0.0 [0.0, 1.5]	9.0 [7.0, 13.0]	<0.001
Need	13.0 [7.0, 16.0]	6.0 [3.0, 10.0]	22.0 [17.0, 27.0]	<0.001
Scrutiny	8.0 [4.5, 12.0]	3.0 [1.0, 5.5]	15.0 [10.0, 19.0]	<0.001
Side effects	7.0 [3.0, 12.0]	4.0 [2.0, 8.0]	14.0 [11.0, 16.0]	<0.001
Tolerance	5.0 [2.0, 11.0]	3.0 [0.0, 6.0]	17.0 [11.0, 20.0]	<0.001
Mistrust	11.0 [8.0, 14.0]	1.0 [0.0, 5.0]	11.0 [8.0, 17.0]	<0.001
Withdrawal	2.0 [0.0, 6.0]	0.0 [0.0, 3.0]	14.0 [11.0, 18.0]	<0.001
Self-reported prescription medications*
Opioids	43 (43.4)	23 (27.7)	12 (36.4)	0.089
NSAIDs	52 (52.5)	32 (38.6)	18 (54.5)	0.115
Triptan	40 (40.4)	28 (33.7)	15 (45.5)	0.445
Muscle relaxant	21 (21.2)	13 (15.7)	10 (30.3)	0.205
Sedative	11 (11.1)	10 (12.0)	10 (30.3)	0.018
Barbituate	12 (12.1)	8 (9.6)	3 (9.1)	0.820
Antidepressant	5 (5.1)	0 (0.0)	4 (12.1)	0.011
Anti-nausea	1 (1.0)	2 (2.4)	1 (3.0)	0.678
Prophylactic	11 (11.1)	3 (3.6)	8 (24.2)	0.004
Ergot	3 (3.0)	2 (2.4)	2 (6.1)	0.598

Data are reported as median [25^th, 75^th] or as frequency counts (%). The p-value is calculated from an omnibus test of differences across cluster groups using Kruskal-Wallis or chi-square tests.

^*Self-reported medications grouped into classes. Prophylactic medications included beta-blockers (e.g., propranolol) and antiepileptics (e.g., topiramate, sodium valproate, gabapentin) but were separated from anti-depressants.

The distributions of each of the PMAQ subscales, their bivariate scatter plots, and the corresponding Pearson correlations are displayed in Figure 1. All of the PMAQ domains were positively skewed, with most individuals reporting low scores and a smaller number of individuals reporting higher scores. The correlations between the PMAQ domains were all positive and ranged from r = 0.21 (between Withdrawal and Mistrust) to r = 0.72 (between Withdrawal and Addiction). No participant reported ‘0’ (i.e., no concerns) for all domains, and no participant reported maximum scores on all domains, though one participant reported maximum scores on three domains.

Figure 1.

Histograms, scatter plots, and correlations of Pain Medication Attitudes Questionnaire (PMAQ) domains.

Descriptive Cluster Analysis

A three-cluster solution was selected as the simplest but most interpretable representation of the participant pain medication belief subgroups (See Figure 2A). Individuals within these clusters reported medication attitudes and beliefs that were generally more similar to each other than across cluster groups (See Figure 2B). However, the reported medication usage was similar across clusters. Figure 3 displays the profile of each of these clusters across all pain medication attitude domains. Each cluster is described below.

Figure 2.

A dendrogram of the descriptive cluster solution where 3 clusters were identified by red boxes (A). The height at which participants are joined by horizontal lines reflects their similarity to each other on the PMAQ scales. The distance between clusters on two components (B) that were created by interpreting the first two components of a principal component analysis (PCA).

Figure 3.

A profile plot of the mean and SD (error bars) for the z-scores for each of the three clusters across the individual PMAQ subscales.

Cluster 1: “Trusting and unconcerned” (n = 83 [38.6%])

Individuals in this subgroup reported a high level of trust in their prescribing physicians (Mistrust: 1.0 [0.0, 5.0]), only very infrequently reporting attitudes or concerns related to mistrust of their providers or medication choices. They also reported relatively low levels (i.e., z < −0.40 SD below the mean of the sample) of fear of addiction, perceived need of medication, unfavorable scrutiny by others, adverse side effects, fear of tolerance, and withdrawal. Individuals in this cluster also reported experiencing moderate levels of disability (MIDAS: 13.0 [6.0, 31.0]), low levels of stress (WSI-SF: 30.0 [17.5, 44.0]), depressive symptoms below the threshold for risk of clinical depression (CES-D: 8.4 [5.5, 16.0]), and anxiety symptoms below the cutoff for probable clinical anxiety (STAI-T: 33.0 [27.0, 39.5]).

Cluster 2 “Skeptical and somewhat worried” (n = 99 [46.0%])

Individuals in this subgroup reported concerns about their physicians’ role in prescribing appropriate pain medications and in properly communicating relevant information about these medications (Mistrust: 11.0 [8.0, 14.0]. However, despite these concerns, individuals in this cluster had concerns near the sample mean level (in relation to the rest of the sample, −0.25 < z < 0.10) about side effects, scrutiny, perceived need, tolerance, withdrawal, and addiction. Individuals in this cluster also reported experiencing severe levels of disability (MIDAS: 24.0 [11.0, 46.8]), moderate levels of stress (WSI-SF: 30.0 [17.5, 44.0]), depressive symptoms near the threshold for risk for clinical depression (CES-D: 16.0 [7.5, 27.0]), and anxiety symptoms above the cutoff for probable clinical anxiety (STAI-T: 41.0 [31.0, 50.0]).

Cluster 3 “Skeptical and concerned” (n = 33 [15.3%])

Individuals in this smallest subgroup reported concerns about their physicians’ role in prescribing appropriate pain medications and in properly communicating relevant information about these medications (Mistrust: 11.0 [8.0, 17.0]. Additionally, individuals in this cluster had elevated concerns (in relation to the rest of the sample, z > 1.15) about side effects, scrutiny, perceived need, tolerance, withdrawal, and addiction. Individuals in this cluster also reported experiencing severe levels of disability (MIDAS: 27.0 [13.0, 59.0]), high levels of stress (WSI-SF: 30.0 [17.5, 44.0]), depressive symptoms above the threshold for risk for clinical depression (CES-D: 22.0 [16.0, 32.0]), and anxiety symptoms above the cutoff for probable clinical anxiety (STAI-T: 45.0 [37.0, 55.0]).

Relationship between Cluster Groups and Other Measures

Although not used in clustering the participants into medication belief groups, the CES-D, STAI-T, WSI-SF, and MIDAS scores were associated with cluster membership. Figure 4 displays the z-scores for each measure by cluster group, and Table 1 displays descriptive statistics for the measure scores. Individuals in the ‘Skeptical and concerned’ group had numerically higher mean values than individuals in the other two groups on all measures. In contrast, individuals in the ‘Trusting and unconcerned’ group had numerically lower scores on all measures. Increasing levels of mistrust and medication concerns were correlated with CES-D scores, with r = 0.31 for mistrust and other concerns ranging from r = 0.22 (side effects) to r = 0.37 (withdrawal). A similar pattern of correlations was observed with STAI-T scores, with r = 0.27 for mistrust and other concerns ranging from r = 0.19 (side effects) to r = 0.40 (withdrawal). Correlations with the MIDAS were numerically lower, with r = 0.11 for mistrust and other concerns ranging from r = 0.12 (scrutiny) to r = 0.28 (addiction). Finally, correlations with the WSI-SF were r = 0.25 for mistrust and other concerns ranging from 0.20 (side effects) to r = 0.39 (tolerance).

Figure 4.

A profile plot of the mean and SD (error bars) for the z-scores for each of the three clusters across CES-D, STAI-T, WSI-SF, and MIDAS.

Table 1 also displays the patterns of self-reported prescription medication use by cluster membership. Although the three groups of participants reported similar medication use for common medications such as triptans, they did report differences in several types of medications. For example, individuals in the skeptical and concerned cluster reported greater use of antidepressants, sedatives, and headache prophylaxis medications than the other groups (Table 1).

Discussion

This study aimed to explore how individuals experiencing headache view their medications and relationships with their healthcare providers. Using the Pain Medication Attitudes Questionnaire,¹¹ we characterized our sample into three clusters of patients’ medication beliefs and attitudes. Of the sample, 38.6% fell under the ‘Trusting and unconcerned’ cluster, in which individuals were overall trusting of their healthcare provider and less concerned or anxious about their medications or health. The second cluster contained 46.0% of individuals who lacked confidence in their providers and were mildly concerned about medication side effects, scrutiny, perceived need, tolerance, withdrawal, and addiction. The last and smallest cluster contained 15.3% of individuals and represented those who were skeptical of their providers, very concerned about their medications, and at higher risk for disability, depression, and anxiety. Interestingly, no identified cluster of patients was both trusting of their providers and concerned about their medications.

Although medication concerns and beliefs were moderately correlated within individuals, there was substantial variability in pain medication beliefs across individuals. While some individuals appeared to be very satisfied with their relationships with providers and medications (38.6%), most individuals (61.4%) were at least somewhat concerned about some aspect of their pain medications. Medications typically used in chronic pain (e.g., opioids, NSAIDs) are commonly associated with concerns about addiction or withdrawal.²² However, the medications used by individuals with headache in this study appeared to result in low levels of concern about addiction or withdrawal. Conversely, many individuals reported concerns about the perceived ‘need’ of their medications for the successful management of their headache pain. Concerns over side effects were extremely variable, with many individuals expressing at least moderate concern.

Because of the influence of medication beliefs on medication adherence,^23–25 the evaluation of these beliefs may be useful to incorporate into routine headache management when the setting and available time permit. When feasible for use, the PMAQ can be publicly accessed at no charge, and to address the potential lack of time for completion, work has been done to develop a short form of the questionnaire for faster assessment of the constructs.²⁶ When possible, providers could use the PMAQ to open dialog with their patients about their medication concerns or as part of an assessment battery to measure ongoing satisfaction with treatment (see: ²⁷). In this way, patient medication concerns could be explicitly incorporated into a treatment plan, thoughtfully addressed, and possibly reduced, improving trust in a medical provider. Additionally, because these beliefs are also associated with affective distress, addressing them could facilitate the treatment of depression and anxiety, as medication concerns in one set of medications may generalize to others.

In this sample, individuals from different clusters reported taking different patterns of medication classes. Individuals in these clusters also reported a different pattern of medication concerns, though it is not clear if these concerns arise from exposure to the medications themselves. It is also unclear if these concerns would change over time in response to improved treatment or modified management. Studying medication adherence and cluster membership could be informative, but we did not have such data available for this analysis. Still, by targeting these beliefs and concerns with a behavioral intervention, it may be possible to improve headache management by enhancing the effectiveness of prescribed medications.²⁸ Trust of providers is associated with better outcomes; encouraging discussion of patient beliefs and uneasiness with treatment may help to improve trust of providers and alleviate apprehension about medications.²⁹ Reducing concerns about medication through education or treatment change could then result in improvements in adherence, thereby improving medication effectiveness.

Limitations of this study include potential sampling bias of those patients included within the HACOGS cohort. The HACOG study utilized convenience sampling of people who volunteered to participate after viewing the posted advertisements or who were referred by their headache specialist. This non-random sample could potentially limit the generalizability of our cluster findings, as these community-based participants may not be representative of individuals who are seeking treatment for their headache disorder and who may be more comfortable with using medications. Additionally, we did not assess headache diagnostic information, non-prescription medication use, adherence to medication prescriptions, or changes in medication use over time. In this sample, the rate of opioid use appears high for the population of individuals with headache. This may correspond with the practice patterns at the time of data collection and with the presumed chronic pain comorbidity in the sample. Despite these limitations, we believe our findings can be generalized to the larger population of people with headache, as we included both individuals referred by specialists and those who volunteered on their own account.

Additionally, with sensitive information included in the study questionnaires, such as addiction, physician mistrust, and depression, participants may not have shared their true feelings about their relationships with physicians or medications, perhaps to avoid socially non-desirable responses or in fear of non-anonymity of the information. We contend that this social desirability bias is minimized due to the clear separation between the research environment and the participant’s clinical care. As part of the study’s informed consent process, participants were reminded of their rights as a study participant and that their answers would not be shared with their healthcare providers.

Conclusions

After assessing various patient medication beliefs, descriptive subgroups of individuals with similar beliefs were identified. Among these subgroups, two groups constituting 61.4% of the sample reported at least some concerns about their medications or their medical providers. Across all groups, beliefs ranged from a lack of concern about using pain medications to high concern about side effects and social scrutiny. An association between medication beliefs and measures of affective distress was identified, with more negative beliefs correlated with more depressive symptoms. It is unclear if these medication beliefs stem from poor experiences with the medications themselves or if these beliefs influence a patient’s future medication use or adherence.

Abbreviations:

HACOGS

Headache and Cognition Study

PMAQ

Pain Medication Attitudes Questionnaire

CES-D

Center for Epidemiologic Studies Depression Scale

STAI-T

State-Trait Anxiety Inventory Form Y-2

WSI-SF

Weekly Stress Inventory Short Form

MIDAS

Migraine Disability Scale