Highlights
-
• Few studies in Canada have examined the relationship between migraine-related disability and productivity loss.
-
In this cross-sectional study, employed adults living with migraine across Canada completed the Valuation of Lost Productivity questionnaire.
-
• After adjusting for relevant covariates, greater migraine-related disability was associated with more total, paid and unpaid productivity loss.
Introduction
Migraine poses a significant socioeconomic burden on society. It is estimated that over 1 billion people are affected by migraine worldwide, Reference Safiri, Pourfathi and Eagan1 and of all medical conditions, migraine is responsible for the second-greatest number of years lived with disability. 2 Migraine impairs quality of life and has been linked to several chronic conditions, including insomnia, depression, anxiety and gastric ulcers. Reference Buse, Reed and Fanning3 The direct healthcare costs related to health resource utilization and treatment of migraine are significant. Reference Gilligan, Foster and Sainski-Nguyen4–Reference García-Azorín, Moya-Alarcón and Armada6
In economic evaluations, the indirect costs related to work productivity loss are also an important consideration, given that migraine disproportionately impacts working-age adults. Reference Schramm, Obermann and Katsarava7–Reference Seddik, Branner and Ostwald9 Multiple studies have shown that the productivity loss associated with migraine is substantial. Reference Seddik, Branner and Ostwald9–Reference Kim, Han and Suh17 However, additional research is needed for several reasons. First, there is a paucity of observational studies in North America that have captured data examining the impact of migraine on presenteeism (reduced productivity while at work), which may be a greater contributor to migraine-related productivity loss than absenteeism. Reference Wong, Alias and Bhoo-Pathy15,Reference Amoozegar, Khan and Oviedo-Ovando16,Reference Haw, Cabaluna and Kaw18,Reference Ishii, Schwedt and Dumkrieger19 Second, studies have not estimated productivity loss related to unpaid work (such as childcare and housework), which is an important consideration given that migraine is more common in women. Reference Schramm, Obermann and Katsarava7,Reference Amiri, Kazeminasab and Nejadghaderi8,Reference Graves, Gerber and Berrigan20,Reference Delaruelle, Ivanova and Khan21 Third, the productivity loss associated with different categories of migraine-related disability or severity is rarely evaluated. Reference Wong, Alias and Bhoo-Pathy15,Reference Kim, Han and Suh17–Reference Ishii, Schwedt and Dumkrieger19 Fourth, most studies assessing productivity loss from migraine have used the Work Productivity and Activity Impairment (WPAI) questionnaire, which quantifies productivity losses as a percent impairment. Reference García-Azorín, Moya-Alarcón and Armada6,Reference Wong, Alias and Bhoo-Pathy15,Reference Ishii, Schwedt and Dumkrieger19,Reference Ford, Ye and Ayer22–Reference Sumelahti, Sumanen and Sumanen27 Estimating productivity loss in hours would provide a more direct quantification of the cost burden. Reference Zhang, Bansback and Boonen28
Further comprehensive and patient-centered valuations of productivity loss from migraine would be valuable for assessing the economic impact of this condition, particularly when considering the perspective of the employer and society. These data could also be used in cost-effectiveness analyses as new migraine therapies reach the market. Accordingly, we conducted a cross-sectional study that examined productivity loss among individuals with migraine across Canada. The study’s primary objective was to examine the association between different levels of migraine-related disability and productivity loss.
Methods
Study design and participants
This was a cross-sectional study in which participants completed an online questionnaire. Participants were recruited from throughout Canada from an Ipsos market research panel. Potentially eligible members from the Ipsos iSay rewards community were invited to participate via the Ipsos iSay website, the mobile app and/or text message (depending on the member’s preferences). To be eligible, participants were required to be 19 or older, employed, a resident of Canada, have a history of migraine and be able to comprehend English or French. The questionnaire was administered electronically by Qualtrics (Provo, Utah). Participants completed eligibility screening questions through the Ipsos iSay platform before electronically accessing the main study questionnaire. Through the eligibility screening, participants were considered to have a history of migraine if they reported being previously diagnosed by a clinician. We targeted 450 total participants for this study and set Qualtrics quotas to ensure an approximately equal distribution of respondents for different levels of migraine-related disability. Some participants were prevented from completing the questionnaire if their responses deemed them ineligible (e.g., unemployed) or the predetermined quota had already been met.
This study was designed and executed in collaboration with a patient partner living with migraine and two additional patient partners with chronic disease (one living with atopic dermatitis and one with alopecia areata). A draft of the questionnaire was piloted in three people with a history of migraine, three people with atopic dermatitis and one person with alopecia areata. Questions related to productivity loss and demographics were the same for the three diseases. The questions related to disease history, severity and treatment were disease-specific. After they completed the draft questionnaire, participants were interviewed for feedback, and appropriate revisions were made. The final questionnaire was available to study participants in English and French. Based on feedback from our patient partner, it was presented in dark mode to reduce possible migraine exacerbation from photophobia. Reference Tian, Xu and Han29–Reference Choi, Oh and Kim31
This study was approved by the University of British Columbia Research Ethics Board (REB # H22-03211). Recruitment for this study occurred between December 4, 2023, and February 12, 2024. Participants provided electronic consent before starting the questionnaire. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for reporting observational studies. Reference von Elm, Altman and Egger32
Migraine-related disability
Migraine-related disability was assessed using the Migraine Disability Assessment (MIDAS) questionnaire. Reference Stewart, Lipton and Dowson33 The first five items of the MIDAS ask about the number of days in the past 3 months that were affected by migraine: the number of missed work or school days; missed household chores days; missed non-work activity days; days at work or school where productivity was reduced by half or more; and days in which household work was reduced by half or more. Reference Stewart, Lipton and Dowson33 The total MIDAS score was derived by summing the total number of days affected by migraine. Reference Stewart, Lipton and Dowson33 Using previously established cutoffs, we categorized participants as having little to no disability (MIDAS score 0–5), mild disability (MIDAS score 6–10), moderate disability (MIDAS score 11–20) or severe disability (MIDAS score ≥21). Reference Stewart, Lipton and Dowson33
The MIDAS questionnaire has two additional items. The sixth item asks about the number of days the participant has experienced headaches over the past 3 months, and the seventh item asks about the average severity of the headaches (on a scale of 0–10). These two items were used in sensitivity analyses.
Outcomes
Productivity loss was measured using the Valuation of Lost Productivity (VOLP) questionnaire. Reference Zhang, Bansback and Kopec34 The VOLP consists of questions about health-related paid and unpaid productivity loss. The questions about productivity loss refer to health in general and are not migraine-specific. Paid productivity loss comprises absenteeism (number of absent workdays due to health) and presenteeism (hours actually taken to complete all work relative to the hours taken to complete the same work if not experiencing any health problems). Unpaid work loss represents the hours of paid and unpaid help received for unpaid work activities (such as childcare and housework) due to health. Reference Zhang, Bansback and Kopec34 It has been validated and used in other chronic conditions to estimate health-related productivity loss in hours over the preceding 3 months. Reference Zhang, Bansback and Kopec34–Reference Gelfand, Sou and Sawatzky38 In this study, the primary outcome was total hours of productivity loss, calculated as the sum of paid productivity loss (from absenteeism and presenteeism) and unpaid productivity loss. The hours of total paid productivity loss, hours lost due to absenteeism, hours lost due to presenteeism and hours of unpaid productivity loss were evaluated as secondary outcomes. Details regarding calculating the productivity loss outcomes using the VOLP are described in the supplemental methods.
An additional secondary outcome was the percent overall work impairment and percent activity impairment due to health as measured by the WPAI – General Health questionnaire. The WPAI is a validated measure that assesses the impact of health on work productivity and impairment of regular activities in the prior week. Reference Ford, Ye and Ayer22,Reference Zhang, Bansback, Boonen, Young, Singh and Anis39,Reference Reilly, Zbrozek and Dukes40 Calculations for work and activity impairment using the WPAI are outlined in the supplemental methods.
Statistical analysis
Mean values for each outcome were calculated for different levels of migraine-related disability based on MIDAS responses. Reference Stewart, Lipton and Dowson33 We then used ordinary least squares (OLS) regressions to measure the association between migraine-related disability levels and the outcomes while adjusting for potential confounding variables. These additional covariates were prespecified based on a review of the literature and were captured from questionnaire responses. Reference Amiri, Kazeminasab and Nejadghaderi8,Reference Landy, Runken and Bell11,Reference Zhang, Li and Gobis35,Reference Rodriguez Llorian, Zhang and Khakban37,Reference Buse, Manack, Serrano, Turkel and Lipton41 These included age, gender, ethnicity, marital status, education level, work income, household income, employment status (part of VOLP), work habits and number of comorbidities. We chose to use OLS models for productivity loss outcomes based on previously published practical recommendations for regression model selection in productivity loss analyses. Reference Zhang and Sun42
We conducted sensitivity analyses based on responses to the two additional MIDAS items to determine if outcomes were associated with (1) the number of days with migraine over the past 3 months and (2) the average severity of migraine.
Statistical tests were two-sided, and the threshold for significance was p < 0.05. Analyses were performed using R statistical software version 4.3.3 and Stata version 15.1 (StataCorp LLC, College Station, TX).
Results
In total, 441 participants were included in the analyses. Due to incomplete or invalid responses, 10 participants were excluded from the VOLP analyses, and 17 were excluded from the WPAI analyses.
Characteristics of the study population are described in Table 1. The mean (SD) age was 37.7 (10.9); 60.1% were women, 75.5% were White, 81.6% worked full-time and 50.1% were sedentary at work. Of note, the no to little migraine-related disability level group had the greatest proportion of participants at the highest work income level (43.1% with over $100,000), and the severe migraine-related disability level group had the greatest proportion (45.5%) of participants with two or more comorbidities.
Table 1. Characteristics of the study population
All percentages represent column proportions unless otherwise indicated.
MIDAS = Migraine Disability Assessment; SD = standard deviation.
a ”Non-binary person” was an option provided for gender, but no participants selected this.
b Atlantic Canada includes the provinces Nova Scotia, New Brunswick, Prince Edward Island and Newfoundland and Labrador.
c Territiories include Yukon, Northwest Territories and Nunavut.
d Other race/ethnicity includes South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.), Chinese, First Nations, Southeast Asian (e.g., Vietnamese, Cambodian, Malaysian, Laotian, etc.), West Asian, Filipino, Latin American, Métis, Korean, Japanese, Arab, Inuit, Black, Indigenous/Aboriginal (not included elsewhere), Other and mixed (i.e., more than one) ethnicities.
e Comorbidities include asthma, arthritis or osteoporosis, back problems, cancer, cardiovascular disease, chronic obstructive pulmonary disease, diabetes, mental health conditions, neurologic conditions, digestive diseases, fibromyalgia or chronic fatigue syndrome, kidney disease, liver disease or gallbladder problems.
The migraine preventative strategies, treatments and workplace accommodations reported by participants are reported in Table S1. The most commonly used strategies by participants to prevent migraine were lifestyle changes (73.0%) and oral medications (68.7%); the most common migraine treatment was oral medications (87.3%). Concerning workplace accommodations for health conditions, 38.1% reported being granted paid leave, and 28.8% had been granted flexible work arrangements.
The mean (SD) hours of total productivity loss in the past 3 months were higher at greater levels of migraine-related disability (61.0 [120.4] hours per 3 months for little to no disability, 105.9 [128.7] for mild disability, 132.3 [148.8] for moderate and 196.5 [214.5] for severe) (Table 2). Specifically, paid productivity loss (including absenteeism and presenteeism) increased with migraine-related disability level (47.6 [106.4], 64.8 [99.1], 85.0 [96.3] and 119.8 [109.4], hours per 3 months, respectively) and so did the mean (SD) hours of unpaid productivity loss (16.5 [61.0], 40.4 [89.5], 46.8 [97.6] and 76.0 [166.9] hours per 3 months, respectively). The mean (SD) hours of absenteeism increased greatly with migraine-related disability levels (7.0 [13.4], 13.9 [16.2], 29.4 [48.2] and 50.9 [56.2] hours per 3 months, respectively). However, the increase in mean (SD) hours of presenteeism across disability levels was not as pronounced (40.4 [102.1], 50.8 [96.1], 55.5 [84.0] and 68.7 [88.5] hours per 3 months, respectively), and for all levels, presenteeism contributed more to paid productivity loss than absenteeism.
Table 2. Productivity loss and percentage impairment by migraine-related disability level
MIDAS = Migraine Disability Assessment; SD = standard deviation; VOLP = Valuation of Lost Productivity; WPAI = Work Productivity and Activity Impairment.
†Sample size N = 431 and 10 participants did not provide valid answers for questions related to presenteeism.
* Sample size for the WPAI percent work impairment outcome N = 424 and 17 participants had valid question skip patterns (not currently employed (working for pay) or 0 hours missed because of health problems and 0 hours worked in the past 7 days).
The mean (SD) WPAI percent overall work impairment in the prior 7 days reported by participants also increased with migraine-related disability level (23.1 [22.4]% for little to no disability, 37.9 [26.2]% for mild disability, 49.5 [26.5]% for moderate and 65.4 [22.4]% for severe disability), as did percent activity impairment (23.2 [22.2]%, 35.9 [23.9]%, 46.7 [23.0]% and 58.5 [22.1]%, respectively).
In our multiple regression models, moderate migraine-related disability was also associated with greater total and paid productivity loss compared with little to no disability (54.1 [10.2, 98.1] adjusted hours, p < 0.05; 32.4 [3.1, 61.8] adjusted hours, p < 0.05, respectively) (Table 3; Table S2). Furthermore, having severe migraine-related disability was associated with greater total productivity loss (110.5 [65.5, 155.6] adjusted hours, p < 0.001), paid productivity loss (61.6 [31.5, 91.7] adjusted hours, p < 0.001) and unpaid productivity loss (43.5 [12.7, 74.3] adjusted hours, p < 0.01) compared with little to no disability. Similarly, in models evaluating overall work impairment derived from WPAI responses, greater levels of migraine-related disability were associated with greater percent impairment (13.1% [6.2, 20.0], p < 0.001 for mild; 23.0% [16.3, 29.8], p < 0.001 for moderate; and 37.3% [30.3, 44.2], p < 0.001 for severe disability compared with little to no disability) (Table S3).
Table 3. Multiple regression models for productivity loss and percentage impairment by migraine-related disability level
Models are adjusted for gender, age, ethnicity, marital status, education, household income, employment status, work habits and the number of comorbidities reported. Complete models are reported in Table S2 and Table S3. MIDAS = Migraine Disability Assessment; WPAI = Work Productivity and Activity Impairment; VOLP = Valuation of Lost Productivity.
* p < 0.05;
** p < 0.01;
*** p < 0.001.
In sensitivity analyses, more headache days over the past 3 months and greater average migraine severity were associated with greater total productivity loss, paid productivity loss, unpaid productivity loss, overall work impairment and activity impairment (Table S4).
Discussion
In this cross-sectional study involving participants from across Canada, we compared productivity loss between individuals with different levels of migraine-related disability. We found that components of paid and unpaid productivity loss (as measured by VOLP), as well as work and activity impairment (as measured by WPAI), were higher in individuals with more disability from migraine.
This is one of the first observational studies to examine productivity loss among people with migraine in a Canadian context. As part of their study on the overall economic burden of migraine, Amoozegar et al. estimated the percentage of patients who had productivity loss after administering the WPAI questionnaire to 287 patients with migraine. Reference Amoozegar, Khan and Oviedo-Ovando16 Our study builds on this work by including a larger cohort, estimating productivity loss in hours, measuring unpaid losses and stratifying by migraine-related disability.
Our findings also contribute to accumulating evidence that migraine-related disability has a significant impact on work productivity loss. Reference Wong, Alias and Bhoo-Pathy15,Reference Haw, Cabaluna and Kaw18,Reference Domingues, Picon and VESCOVi43 For example, a recent study by Wong et al. evaluated WPAI outcomes by MIDAS level in employees within the banking sector in Malaysia. Reference Wong, Alias and Bhoo-Pathy15 Compared to this study, we observed that the percent overall work impairment and activity impairment for little to no, mild and moderate disability levels were lower, but we observed greater impairment for severe disability. Similar to our study, this Malyasian study also reported significant levels of productivity loss associated with just minimal levels of migraine-related disability. Reference Wong, Alias and Bhoo-Pathy15 Based on VOLP responses, individuals in our study with little to no migraine-related disability had an average of 61 hours of productivity loss over the prior 3 months. This finding is an indication that even mild or treated migraine disorder may result in significant occupational impairment.
Like the Malaysian study, our results showed that presenteeism (productivity loss at work) is significant among persons with migraine. Regardless of migraine-related disability level, presenteeism contributed more to paid productivity loss than absenteeism. This is a relevant finding from the employer’s perspective, as individuals with migraine have experienced stigma, and there is potential for migraine exacerbation in the workplace. Reference Haw, Cabaluna and Kaw18,Reference Parikh, Kempner and Young44–Reference Vicente-Herrero, Burke and Laínez48 Indeed, employers have become increasingly aware of the importance of developing work environments and programs that support people with migraine. Reference Vicente-Herrero, Burke and Laínez48–Reference Begasse de Dhaem, Gharedaghi, Bain, Hettie, Loder and Burch50 Studies have suggested that reducing screen time, implementing migraine-specific disease management programs, safe/ dark rooms and referrals to occupational health could be beneficial for people with migraine. Reference Haw, Cabaluna and Kaw18,Reference Begasse de Dhaem, Gharedaghi, Bain, Hettie, Loder and Burch50 However, further research is needed to evaluate whether these interventions can reduce productivity loss. Reference Vicente-Herrero, Burke and Laínez48,Reference Begasse de Dhaem, Gharedaghi, Bain, Hettie, Loder and Burch50
Our study also observed that unpaid work significantly contributes to productivity loss in people with migraine. Unpaid work, such as caregiving, can affect mental health, impair health-related quality of life and have significant societal value. Reference Pinquart and Sörensen51,Reference Seedat and Rondon52 . However, unpaid losses are not routinely considered in economic analyses and have not been accounted for in migraine productivity loss assessments until our study. Women are estimated to spend 2–10 times more time on unpaid work activities than men. Reference Ferrant, Pesando and Nowacka53 In the context of migraine – which is at least twice as prevalent in women – it is imperative to consider unpaid losses when examining the economic benefits of an intervention. Reference Schramm, Obermann and Katsarava7,Reference Amiri, Kazeminasab and Nejadghaderi8,Reference Nicolas, Aikaterini and Tomas54
The results of this study highlight the economic value of developing effective migraine treatments. For example, recent randomized trials showed that 3 months of treatment with the calcitonin-gene-related peptide (CGRP) antagonist galcanezumab resulted in MIDAS score improvements of over 20 points. Reference Tepper, Ailani and Ford55,Reference Ford, Tassorelli and Leroux56 This level of improvement is enough to reduce migraine-related disability from severe to little or no symptoms; based on our data, this would represent an adjusted total productivity loss improvement of 110.5 hours in 3 months and an adjusted paid productivity loss improvement of 61.6 hours (i.e., nearly two full workweeks).
The productivity loss valuations reported in this study could be applied to future cost-effectiveness analyses. Over two decades ago, productivity loss valuations played a role in demonstrating the efficacy of triptans; Reference Schulman, Cady and Henry57–Reference Cady, Ryan and Jhingran59 similar assessments will be required for CGRP receptor antagonists and other new migraine therapies. Reference Zobdeh, ben Kraiem and Attwood60 While recent cost-effectiveness analyses of CGRP receptor antagonists have used WPAI outcome data, Reference Begasse de Dhaem, Gharedaghi, Bain, Hettie, Loder and Burch50,Reference Seedat and Rondon52 the VOLP should be considered as it was designed for use in economic evaluations or cost of illness studies and provides a more comprehensive assessment from a societal perspective. Reference Zhang, Bansback and Boonen28 Unlike WPAI, the VOLP estimates paid and unpaid work productivity loss in terms of time, which can then be valued in monetary terms. Reference Zhang, Bansback and Boonen28
However, it is prudent to consider the limitations of our study. As it was a cross-sectional analysis, causal relationships between migraine-related disability levels and the outcomes cannot be established. Since we relied on online convenience sampling of participants and set quotas to ensure a similar number of participants for each disability level group, the study population should not be taken to represent all employed Canadian residents with migraine. In addition, VOLP and WPAI captured productivity loss due to health (any physical, mental or emotional problems or symptoms) as opposed to migraine-specific productivity loss. The VOLP was developed as a generic health instead of a disease-specific questionnaire because patients may have difficulty attributing their sick leaves or reduced work productivity to a specific disease, especially when they have multiple chronic health conditions, and because they are less likely to attribute the related treatment side effects or comorbidities to a specific disease. Reference Zhang, Bansback and Boonen28,Reference Zhang, Bansback and Anis61 The severe disability group was more likely to have at least two comorbidities and thus tended to have higher health-related productivity loss. Thus, the findings on the adjusted differences between different disability levels have more practical implications than the outcomes for a given disability level. Furthermore, we relied solely on self-report (as opposed to clinical records) to ascertain migraine diagnosis, which may have led to the inclusion of individuals who did not truly have a migraine disorder. Similarly, comorbidity information was captured from questionnaire responses and was not comprehensive; this may have resulted in unmeasured confounding.
Our study has several strengths. We captured data in two languages from regions across Canada and included participants from various socioeconomic backgrounds and workplaces. In contrast to previous productivity loss assessments of migraine in Canada, our study included a larger sample size, and recruitment was not limited to specific clinics or patients with particular treatment profiles. Reference Amoozegar, Khan and Oviedo-Ovando16 The diversity of our study population increases the generalizability of our findings – an important consideration given that productivity loss from migraine has been shown to differ by occupation and region. Reference Rondinella and Silipo12 Furthermore, all our study outcomes were patient-reported, and we applied a patient-oriented approach by engaging patient partners, which helped ensure that the procedures and results were centered on the values of individuals with migraine and other chronic diseases. Lastly, a major strength of our study was the selection of the outcome measures. Although the VOLP has not been previously applied to individuals with migraine, it has been used for several other diseases and permitted a comprehensive valuation of productivity loss, including paid and unpaid losses. Reference Zhang, Bansback and Kopec34,Reference Rodriguez Llorian, Zhang and Khakban37,Reference Gelfand, Sou and Sawatzky38 This was complemented by including the WPAI outcomes, allowing comparisons with other studies. Reference García-Azorín, Moya-Alarcón and Armada6,Reference Wong, Alias and Bhoo-Pathy15,Reference Ishii, Schwedt and Dumkrieger19,Reference Ford, Ye and Ayer22,Reference Spierings, Ning and Ramirez Campos24–Reference Sumelahti, Sumanen and Sumanen27,Reference Doane, Gupta and Vo62
Conclusion
In conclusion, greater migraine-related disability was associated with greater total, paid and unpaid productivity loss among employed adults. These findings demonstrate the economic impact of migraine and highlight the potential societal value of effective interventions.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/cjn.2024.337.
Acknowledgments
The authors would like to thank Ms. Amanda Brar for her insights as a patient partner during the development of this study. Dr Naik would like to acknowledge the support of the University of British Columbia Clinician Investigator Program and a CAN-TAP-TALENT and Michael Smith Health Research BC Postdoctoral Fellowship. Dr Trenaman would like to acknowledge the support of the Leo Greenawalt Endowed Professorship in Health Policy. Dr Zhang would like to acknowledge the support of the Michael Smith Health Research BC Scholar Award.
Author contributions
LT, LL and WZ conceived the study. HN, AT and WZ were involved in data acquisition. AT conducted the statistical analysis. All authors were involved in data interpretation. HN wrote the first draft of the manuscript. All authors were involved in critically reviewing the manuscript for important intellectual content. All authors approved of the final manuscript.
Funding statement
This study was supported by a grant-in-aid from Pfizer Canada.
The funders had no role in the design and conduct of the study, data collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication.
Competing interests
None.
Ethical approval
The University of British Columbia Research Ethics Board (REB # H22-03211) approved this study. All participants provided electronic consent.
Availability of data and materials
The datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.
Open access
