Changing society, changing research: integrating gender to better understand physical and psychological treatments use in chronic pain management

Supplemental Digital Content is Available in the Text.

Identifying as a man and having masculine and feminine personality traits decrease the odds of reporting the use of physical/psychological treatments for chronic pain management.

Abstract

Treatment of chronic pain should be multimodal and include pharmacological, physical, and psychological treatments. However, because various barriers to physical and psychological treatments (PPTs) exist, a better understanding of biopsychosocial factors leading to their use is relevant. This study aimed to explore the association between gender identity, gender-stereotyped personality traits, and the use of PPTs in chronic pain management. The ChrOnic Pain trEatment cohort, a self-reported data infrastructure resulting from a web-based recruitment of 1935 people living with chronic pain (Quebec, Canada) was analyzed. Gender identity was operationalized as women, men, and nonbinary. Gender-stereotyped personality traits were measured using the Bem Sex-Role Inventory (feminine, masculine, androgynous, undifferentiated). A checklist of 31 types of PPTs that can be used for chronic pain management was presented to participants (yes/no). From the 1433 participants, 85.5% reported using at least one PPT. Hot–cold therapies (43.4%), exercise (41.9%), and meditation (35.2%) were the most frequently used PPTs, but most popular PPTs were not the same among women and men. Women reported a significantly higher use of PPTs in general (87.2% vs 77.2%; P < 0.001). Multivariable and interaction analyses showed that identifying as a man decreased the odds of reporting the use of PPTs (odds ratio: 0.32, 95% confidence interval: 0.11-0.92) but only among participants who scored high on both masculine and feminine personality traits (those classified as androgynous). The high prevalence of PPTs use found in our study is positive. Our results are relevant for a more personalized promotion of PPTs for chronic pain management.

1. Introduction

People who live with chronic pain (ie, pain that persists or recurs for more than 3 months)⁶³ must cope with the biopsychosocial impacts.¹⁰ This condition alone affects 20% of the Canadian and Quebec populations.^2,52,56 With population aging, the prevalence of chronic pain will increase further in the coming years.¹⁰ Societal costs are reaching more than 40 billion dollars per year in Canada.¹¹ Treatments offered must be more effective, accessible, and personalized to the patients for any hope for change.⁵⁰

Currently, a multimodal treatment approach combining pharmacological, physical, and psychological¹⁰ therapies is recommended to manage chronic pain. Achieving this balance is, however, fraught with challenges, and physical/psychological treatments (PPTs) are sometimes underused.^4,44 Some barriers exist such as the difficulty of access because of the region of residence, costs, or treatment requiring active patient involvement.⁴ To improve the management of chronic pain, factors associated with the use of PPTs must be considered to ensure that these types of treatments are promoted and used effectively. In fact, because pain is a personal and subjective experience,⁴⁸ treatment strategies must take into account patient-specific characteristics such as biological, psychological, and social factors that can affect the pain experience.^24,37 Among these factors, sex and gender may exercise a significant influence on the pain experience and use of PPTs.

Despite the increasing attention given to the importance of considering both sex and gender in health research,^17,30,47 these terms are often used inconsistently and interchangeably in the literature,^8,66 including in the pain field.⁴⁶ Biological sex is defined as a “set of biological attributes in humans and animals. It is primarily associated with physical and physiological features including chromosomes, gene expression, hormone levels, and reproductive/sexual anatomy.”²⁸ It is usually categorized as female, male, or intersex.²⁸ Gender touches on the “socially constructed roles, behaviours, expressions and identities of girls, women, boys, men, and gender-diverse people.”²⁸ It refers to the socially constructed norms that dictate and determine the roles, relationships, and power positions of all people throughout their lives.^28,54 Many studies have identified biological sex as a predictor of pharmacological pain treatment use,⁸ but the association between PPT use and gender is less clear. In fact, although some studies reported differences in PPTs use between men and women (women being more inclined)^21,33, the literature is scarce and to our knowledge, no previous study has examined the influence of gender identity and gender-stereotyped personality traits on the prevalence of PPT use in chronic pain management. Men are known to use health services less (psychosocial barriers).²³ Because nonpharmacological treatment of pain involves self-management, but also consultations with physical and psychological health care professionals, we hypothesized that gender was likely to influence the use of PPT and that it was relevant to understand the situation to better harness the modifiable determining factors. This study thus examined gender identity and gender-stereotyped personality traits as potential determinants of PPTs use in chronic pain management.

2. Methods

2.1. Data source and study population

This study used data from the ChrOnic Pain trEatment (COPE) cohort,³⁵ a data set that aims to provide insight into the use of pharmacological, physical, and psychological treatments among people living with chronic pain. In total, 1935 French-speaking adults from Quebec (Canada) with chronic pain (persistent or recurrent pain lasting more than 3 months) participated in the COPE cohort study. The study was conducted between June and October 2019. A web-based recruitment strategy (social media, pain organization websites and newsletters, email snowball sampling) made possible the recruitment of a community sample of participants. They were invited to complete a web-based questionnaire that included previously used items and validated composite scales, including the minimum data set recommended by the Canadian Registry Working Group of the Strategy for Patient-Oriented Research (SPOR) Chronic Pain Network (CPN).¹⁵ The complete methodology of the COPE cohort implementation is described elsewhere.³⁵ This study was conducted among a sample of participants who completed the physical and psychological pain treatment section of the questionnaire (n = 1433). The Université du Québec en Abitibi-Témiscamingue Research Ethics Board approved the study (#2018-05—Lacasse, A). The first page of the web-based landing website provided details allowing participants to make an informed decision, and electronic consent was obtained from each participant before starting the questionnaire. ChrOnic Pain trEatment cohort participants were previously found to be comparable with random samples of Canadians living with chronic pain regarding age, employment, education, and pain characteristics.³⁵ Women are, however, overrepresented (84% vs 55%-65% in Canadian chronic pain random samples),³⁵ justifying gender-stratified or gender-standardized statistics and multivariable analyses.

2.2. Study variables

2.2.1. Physical and psychological treatments

In the COPE study, the use of PPTs for pain management was our primary dependent variable of interest and was measured dichotomously (current use for pain yes or no). A standardized list of 31 treatment options, inspired by Canadian Agency for Drugs and Technologies in Health (CADTH)⁴⁴ and Canadian Pain Task Force¹⁰ reports, was presented to participants. Participants were asked “Apart from medication, do you currently use any other types of treatment for your pain?” The PPTs were listed as follows: acupuncture/acupressure, aquatic therapy, aromatherapy, biofeedback, hot–cold, chiropractic care, ergotherapy, exercises/physical activity, support groups, hypnosis, neurostimulator implantation, group intervention, massage therapy, meditation, music therapy, osteopathic treatments, physiotherapy, homeopathic products, psychotherapy (miscellaneous), virtual/augmented reality, reflexology, reiki, transcutaneous electrical nerve stimulation (TENS), tai chi, tapping-elastic bands, relaxation/respiration, cognitive behavioural therapy, behavioural therapy, yoga, and animal-assisted therapy. Participants were asked to check off treatments currently used (participants could check more than one treatment). A category named “other” (open-ended question) allowed participants to indicate whether a treatment was not listed, and a thorough cleaning process was applied and validated by more than one member of the research team. For the individual treatments, some PPTs were recategorized according to their nature. We regrouped the variables “meditation” and “relaxation/respiration” under the label “meditation.” We combined the variables “psychotherapy,” “cognitive behavioural therapy,” and “behavioural therapy” under “psychotherapy.” The label “yoga” combined the “yoga” and “tai chi” variables. In addition to the overall use of PPTs, 4 other nonmutually exclusive variables were created for the purpose of our analysis—(1) use of active treatments, as defined in the literature as treatments that require active patient involvement¹⁰: aquatic therapy, biofeedback, exercises or physical activity, support group, meditation, physiotherapy, psychotherapy, virtual or augmented reality, yoga, ergotherapy, and group intervention; (2) use of passive treatments, which include all treatments not requiring active patient involvement (travelling to and from destinations does not constitute active involvement): acupuncture, aromatherapy, hot–cold therapies, chiropractic care, homeopathic products, hypnosis, neurostimulator implantation, massage therapy, music therapy, osteopathic treatments, reflexology, tapping-elastic bands, transcutaneous electrical nerve stimulation, and animal-assisted therapy; (3) use of a multimodal approach, which includes participants who combined pharmacological treatment and at least one PPT; and (4) use of do-it-yourself (DIY) treatments which we defined, for the purpose of our study, as treatments that can be performed at home, without the help of a health care professional, and at nearly no cost (aromatherapy, hot–cold therapies, exercises, music therapy, meditation, and yoga).

2.2.2. Gender identity

Gender identity can be defined as how a person defines themselves.²⁹ In the questionnaire, participants could indicate whether they identify as a woman, man, undetermined, or unknown. The question was drawn from the minimum data set of the National Institutes of Health (NIH) Task Force on Research Standards for Chronic Low Back Pain.¹⁹ In the final analysis, we combined “indeterminate” and “unknown” into “nonbinary” because it is now a more appropriate term. Because specific gender identities of individuals who do not identify as a man or a woman were not available in the COPE cohort data, the term “nonbinary” used in our study represents an umbrella term encompassing all gender-diverse identities.²⁰

2.2.3. Gender-stereotyped personality traits

To measure this construct, we used the Bem Sex-Role Inventory (BSRI),⁵ an instrument that assesses various personality traits stereotyped as feminine or masculine.²⁶ The original BSRI comprises 60 items, but several short versions exist, including an 18-item French version published by Fontayne et al.²² in 2000 that was used in the COPE cohort. This French adaptation was chosen because it is brief and considered appropriate for people with different literacy levels (items are understood by adolescents).²² Each item was scored on a 7-point Likert scale (1 = never true; 7 = always true).⁵ Items were averaged to obtain a feminine subscale score (10 items) and a masculine subscale score (8 items).²² These 2 scores were then used to create 4 gender-stereotyped personality trait subgroups using the split median approach^7,18 (median were calculated for the entire sample without regard to gender identity): (1) Participants scoring above or equal to the median of the feminine subscale and below the median of the masculine subscale were classified as having dominant feminine traits; (2) those scoring below the median of the feminine subscale and above or equal to the median of the masculine subscale were classified as having dominant masculine traits; (3) those scoring above or equal to the median on both subscales were classified as having androgynous traits; and (4) those scoring below the median on both subscales were classified as having undifferentiated traits. Specifically, participants categorized as feminine are more likely to describe themselves as tender and sensitive to others; participants categorized as masculine are more likely to describe themselves as athletic, leaders, and self-confident. Participants are categorized as androgynous when they score high on all these traits and undifferentiated when they score low on all of them. The classification should therefore be interpreted in this sense. In our COPE cohort, the internal consistency and factor structure of the short version of the BSRI Fontayne et al.⁴⁵ were found to be adequate, ie, Cronbach's alphas of 0.90 [95% confidence intervals (95% CI) = 0.89-0.91] and 0.82 (95% CI = 0.81-0.84) were obtained for the feminine and masculine scales, respectively; confirmatory factor analysis reproduced the 5 first-order factors (tenderness, sensitivity to others, athletic, leadership, self-confidence) and 2 second-order factors (feminine, masculine) of the theoretical model published by Fontayne et al.²² with acceptable goodness-of-fit indices; c² (125) = 1202.62, P < 0.0001, Goodness of Fit Index (GFI) = 0.9008, Comparative Fit Index (CFI) = 0.9147, Root Mean Square Error of Approximation (RMSEA) = 0.0823. Although criticized by various authors, the BSRI remains the most widely used instrument in gender research literature.²⁶ The BSRI has been validated in multiple populations and settings, as evidenced by a systematic review of 23 validation studies.¹³

2.2.4. Covariables

The following pain-related variables were analyzed to build the profile of participants: various pain locations as dichotomous variables, generalized pain and multisite pain (ie, 2 or more locations), frequency of pain (continuous or occasional), pain duration in years, pain intensity (0-10 numerical rating scale measuring the average pain intensity in the last 7 days), pain catastrophizing (single item of the NIH Minimum Data Set¹⁹: agree/disagree with the following statement “I feel that my pain is terrible and it's never going to get any better”), neuropathic pain (a DN4 Interview part score of ≥3/7 indicates a likely presence of neuropathic pain⁹), and pain interference (Brief Pain Inventory [BPI] Interference Scale¹⁴ which ranges from 0 to 10). We also included variables related to pharmacological pain treatment (over-the-counter medication, prescribed medication, cannabis), private drug insurance, and access to a trusted health care professional. Other covariates measured in the COPE Cohort included sociodemographic profile (age, country of birth, employment, education level, region of residence), physical functioning (SF-12 subscale items⁴⁰), general health (SF-12 subscale item⁴⁰), feeling the need to reduce alcohol or drug consumption, smoking habits, and psychological distress (Patient Health Questionnaire PHQ-4).³⁴

2.3. Statistical analysis

A gender-based analysis was conducted according to best practice guidelines (ie, stratification of descriptive results by gender, evaluation of the statistical significance of these variables and associated interaction terms in multivariate models, and report any nonsignificant results).^25,41

The characteristics of the study population were summarized using descriptive statistics (means, standard deviations, counts, percentages). Because only 4 participants identified as nonbinary, they were included in the overall study population description, but it was impossible to form a statistically sound subgroup for all our analyses. The distribution of gender-stereotyped personality trait subgroups in participants who self-identified as women and men were depicted in a graph. Proportion of PPT users (overall and for each individual treatment) was first calculated in the whole study population.

The 3 most used treatments, in addition to the overall use of PPTs, use of active treatments, passive treatments, multimodal approach, and DIY treatments were then compared across gender identity subgroups (women, men, nonbinary) and gender-stereotyped personality trait subgroups (feminine, masculine, androgynous, undifferentiated). Bivariable analyses (χ² test) were performed, in addition, to post hoc multiple comparisons analysis (Turkey-style multiple comparisons of proportions test) for gender-stereotyped personality trait subgroup pairwise comparisons.

A multivariable logistic regression model was used to investigate the association between the use of PPTs (dependent variable), gender identity, and gender-stereotyped personality traits (independent variables) while accounting for potential confounding factors. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were computed. The reference category for the interpretation of gender-stereotyped personality trait dummy variables was “feminine.” COPE cohort variables that could potentially be associated with gender identity, gender-stereotyped personality traits, or PPTs (potential confounders) were identified a priori and included in the regression analysis: pain location, multisite pain, pain frequency, pain duration, pain intensity, feeling that pain is terrible and it is never going to get any better, evidence of neuropathic pain, pain interference, private drug insurance, use or over-the-counter or prescription pain medications, access to a trusted health care professional for pain management, age, country of birth, employment, education level, living in a remote region, polypharmacy, psychological distress, and cannabis use for pain management. The choice of variables was based on existing literature, clinical considerations, and a review of 2 existing models: (1) the Anderson model,¹ which is widely used in health care utilization studies, allowing to identify predisposing, enabling need factors and (2) the intersectionality model,⁴¹ which is useful for examining variables potentially relevant to sex or gender research.³ The a priori selection of variables to include in our final regression model follows the latest recommendation in that regard.⁵⁷ Owing to our substantial sample size, this method was favoured over criticized selection techniques such as relying on bivariate regression analysis P-values⁵⁷ or stepwise selection.³¹ Multicollinearity problems were screened (variance inflation factors below 5),⁶⁵ and Hosmer-Lemeshow tests (P > 0.05) supported the goodness of fit of the models. Sensitivity analyses were performed to assess whether different analytic approaches affected our results: (1) use of continuous feminine and masculine BSRI subscale scores instead of the categorical gender-stereotyped personality traits variable obtained using the split median approach, (2) multiple imputation of missing values,³¹ and (3) use of Least Absolute Shrinkage and Selection Operator (LASSO) for the selection of independent variables to be included in the multivariable model (prevents overfitting). In a second model, interaction terms (gender identity × gender-stereotyped personality traits dummy variables) were also tested. In case of statistical significance, it was planned to better map and evaluate the direction of effect modification by stratifying the gender identity multivariable regression coefficient across gender-stereotyped personality trait subgroups. All analyses were performed using SPSS Statistics for Windows version 27 (IBM Corp, Armonk, NY) and SAS version 9.4 (SAS Institute, Cary, NC).

3. Results

Of the 1935 participants in the COPE cohort, 1433 answered the questions about treatment use. Those participants were clinically comparable with nonincluded ones (n = 502) regarding proportion of individuals reporting moderate to severe pain in the past 7 days (68.4% vs 69.3%) and living with chronic pain for ≥10 years (51.6% vs 56.1%).

In total, 85.5% of participants reported using at least one PPT to manage their pain. The study population characteristics are reported in Table 1. Mean age was 49.8 years; 83.8% self-identified as women, 15.9% as men, and 0.3% as nonbinary (n = 4; it was thus impossible to form a statistically sound subgroup for all subsequent analyses). Regarding gender-stereotyped personality traits subgroups, 22.4% of participants were classified as having feminine traits (tenderness and sensitivity to others), 19.8% as having masculine traits (leadership, athleticism, self-confidence), 30.6% as androgynous (high score on masculine and feminine traits), and 27.2% as undifferentiated (low scores for both masculine and feminine traits). More than half (51.5%) had been living with pain for more than 10 years, 88.8% were experiencing multisite pain (≥2 sites), 87.0% were living with pain continuously, and 14.8% reported severe pain intensity. In addition, most participants (79.6%) used prescribed pain medications.

Table 1.

Sample characteristics.

Characteristics (n = 1433)	No. (%) of participants*
Age (y)—mean ± SD	49.8 ± 13.3
Gender identity
Women	1180 (83.8)
Men	224 (15.9)
Nonbinary	4 (0.3)
Gender-stereotyped personality traits
Feminine	284 (22.4)
Masculine	251 (19.8)
Androgynous	389 (30.6)
Undifferentiated	345 (27.2)
Country of birth
Canada	1333 (95.9)
Other	57 (4.1)
Employment
Worker (full-time or part-time)	504 (36.3)
Unemployed	886 (63.7)
Education level
Postsecondary education	1313 (94.8)
No postsecondary education	72 (5.2)
Region of residence
Remote†	332 (23.9)
Urban	1060 (76.1)
Pain duration (y)
<1	44 (3.1)
1-4	324 (22.7)
5-9	324 (22.7)
≥10	738 (51.5)
Pain intensity in the past 7 d (score 0-10)
Mild (1-4)	447 (31.6)
Moderate (5-7)	759 (53.6)
Severe (8-10)	210 (14.8)
Three most common pain locations‡
Back	891 (62.2)
Neck	640 (44.7)
Shoulders	627 (43.8)
Pain frequency
Continually	1242 (87.0)
Occasionally	186 (13.0)
Prescribed pain medication use
Yes	1136 (79.6)
No	292 (20.4)
Over-the-counter pain medication use
Yes	959 (67.1)
No	470 (32.9)
Naturals product use
Yes	242 (16.9)
No	1188 (83.1)

The proportion of missing data across the presented variable ranges between 0% and 11.44%.

^*Unless stated otherwise.

^†Remote resource regions as defined by Revenu Quebec (ie, the provincial revenue agency): Bas-Saint-Laurent, Saguenay—Lac-Saint-Jean, Abitibi-Témiscamingue, Côte-Nord, Nord-du-Québec, Gaspésie—Îles-de-la-Madeleine. Nonremote regions are near a major urban center.

^‡Nonmutually exclusive categories.

BSRI, Bem Sex-Role Inventory; CÉGEP, Collège d'enseignement général et professionnel (in Quebec [Canada], a CEGEP is a public school that provides the first level of postsecondary education).

Figure 1 presents the distribution of the 4 gender-stereotyped personality traits categories among women and men (gender identity). Among women, generally well-distributed frequency of each gender-stereotyped personality traits category was observed, with 24.2% as having feminine traits, 18.6% as masculine, 30.8% as androgynous, and 26.4% as undifferentiated. Among men, these proportions were 12.2%, 26.5%, 29.1%, and 32.1% respectively.

Figure 1.

Distribution (%) of gender-stereotyped personality trait subgroups in women (left) and men (right). Feminine: described themselves as tender and sensitive to others. Masculine: described themselves as athletic, having leadership, and being self-confident. Androgynous: scored high on all these traits. Undifferentiated: scored low on all these traits.

3.1. Overall prevalence of physical and psychological treatment use

The overall proportion of participants using PPTs for chronic pain management was 85.5%. The proportion of users of individual treatments is presented in Figure 2. Hot–cold therapies (43.4%), exercise (41.9%), and meditation (35.2%) were the most frequently used PPTs, while virtual/augmented reality, group intervention, and biofeedback were the less frequently used (<1%).

Figure 2.

Prevalence of physical and psychological pain treatments. PPTs, physical and psychological treatments; TENS, transcutaneous electrical nerve stimulation. Other mostly included kinesiology, nutrition diet, rest, art therapy, and compressive clothing.

3.2. Prevalence of physical and psychological treatment use across gender subgroups

The 3 most popular treatments in each subgroup defined by gender identity (women, men, nonbinary) and gender-stereotyped personality traits (feminine, masculine, androgynous and undifferentiated traits) are presented in Table 2. Women and people with feminine traits, androgynous traits, and undifferentiated traits reported using hot–cold therapies more than other treatments, while men, nonbinary individuals, and persons with masculine traits seem to favour exercise as a principal treatment to help manage their pain. The complete picture of each treatment's prevalence of use by gender identity and gender-stereotyped personality traits subgroups is available in Supplementary Content 1 (available at http://links.lww.com/PAIN/B924).

Table 2.

Use of physical and psychological pain treatments across gender identity and gender-stereotyped personality trait subgroups.

	Gender identity subgroups n (%)
	Women (n = 1185)	Men (n = 227)	Nonbinary (n = 4)
3 most used treatments	Hot–cold (46.3)	Exercise (34.2)	Exercise (75.0)
	Exercise (43.3)	Hot–cold (29.7)	Meditation (75.0)
	Meditation (37.4)	Massage therapy (27.9)	Hot–cold/physiotherapy/psychotherapy (50.0)

	Gender-stereotyped personality trait subgroups n (%)
	Feminine (n = 284)	Masculine (n = 253)	Androgynous (n = 390)	Undifferentiated (n = 346)
3 most used treatments	Hot–cold (45.2)	Exercise (53.0)	Hot–cold (47.6)	Hot–cold (41.0)
	Meditation (36.4)	Hot–cold (42.2)	Exercise (46.0)	Exercise (39.0)
	Exercise (33.9)	Massage therapy (39.0)	Massage therapy (40.4)	Massage therapy (31.7)

The proportion of missing data across the presented variable ranges between 1.95% and 11.65%.

The prevalence of use of different treatments by gender identity is presented in Figure 3. Women, compared with men, reported a significantly higher use (P < 0.001) of PPTs in general (87.2% vs 77.2%), of active treatments (71.5% vs 58.1%), of passive treatments (79.0% vs 66.2%), of multimodal treatment (82.3% vs 67.9%), and of DIY treatments (75.2% vs 60.4%). Regarding gender-stereotyped personality traits subgroups (Fig. 4), no statistically significant difference in the overall prevalence of PPTs use was observed between subgroups in the bivariable analysis. Differences were however detected regarding the proportion of participants reporting the use of active treatments, passive treatments, multimodal treatment, or DIY treatments (participants classified as masculine or androgynous showing higher prevalence of use).

Figure 3.

Use of physical and psychological treatments (PPTs) across gender identity subgroups. All P-values < 0.001 (χ² tests comparing women and men), CI 95%. CI 95%, 95% confidence interval.

Figure 4.

Use of physical and psychological treatments (PPTs) across gender-stereotyped personality trait subgroups. The lines indicate a significant difference (P-value < 0.05; χ² tests) between the 2 corresponding variables (post hoc multiple comparisons analysis [Turkey-style multiple comparisons of proportions test]).

3.3. Gender identity and gender-stereotyped personality traits as predictors of the use of physical and psychological treatments

Table 3 shows the main results of the multivariable model used to assess the association between gender identity, gender-stereotyped personality traits, and the use of PPTs for chronic pain (with and without interaction terms). In the first model, gender identity and gender-stereotyped personality traits were both associated with the use of PPTs after adjusting for potential confounders: (1) Identifying as a man (gender identity) decreased the odds of reporting the use of PPTs for chronic pain management (OR: 0.57; 95% CI: 0.35-0.94), and (2) being classified as androgynous (vs feminine) increased the odds of reporting the use of PPTs for chronic pain management (OR: 1.75; 95% CI: 1.01-3.00). Similar associations between gender identity, gender-stereotyped personality traits, and the use of PPTs were found when using continuous BSRI scores, multiple imputation of missing values, or LASSO variable selection. Because some interaction terms (gender identity × gender-stereotyped personality traits dummy variables) reached statistical significance (Table 3), gender identity multivariable regression coefficients were disaggregated across gender-stereotyped personality traits subgroups to better map and evaluate the direction of effect modification (Table 4). Gender identity was only associated with the use of PPTs among participants classified as androgynous (OR: 0.32, 95% CI: 0.11-0.92). In other words, identifying as a man decreased the odds of reporting the use of PPTs for chronic pain management only among participants who scored high on both masculine and feminine personality traits (those classified as androgynous).

Table 3.

Multivariable models exploring associations between gender identity, gender-stereotyped personality traits, and the use of physical and psychological pain treatments.

Variable	Odds ratio (OR)—adjusted β	P	95% CI
Model without interaction terms
Gender identity (men vs women)	0.570	0.027	0.346-0.937
Gender-stereotyped personality traits (vs feminine)
Undifferentiated	1.310	0.305	0.782-2.196
Masculine	1.817	0.065	0.964-3.426
Androgynous	1.745	0.045	1.014-3.003
Model with interaction terms
Gender identity (men vs women)	1.984	0.336	0.491-8.021
Gender-stereotyped personality traits (vs feminine)
Undifferentiated	1.482	0.169	0.846-2.596
Masculine	2.576	0.014	1.207-5.498
Androgynous	2.029	0.019	1.122-3.671
Interaction terms
Gender identity × undifferentiated	0.282	0.126	0.056-1.425
Gender identity × masculine	0.146	0.032	0.025-0.843
Gender identity × androgynous	0.244	0.092	0.047-1.260

P-values < 0.05 are reported in bold.

The multivariable analysis was adjusted for the following covariables: pain location (generalized pain and back pain), multisite pain, pain frequency, pain duration, pain intensity in the past 7 days, feeling that pain is terrible and it is never going to get any better, evidence of neuropathic pain, pain interference, private drug insurance, pharmacological pain treatment use (over-the-counter and prescription drugs), access to a trusted health care professional for pain management, age, country of birth, employment, education level, living in a remote region, polypharmacy (use of ≥5 medications), psychological distress, and cannabis use for pain management. In total, 925 participants (64.5%) were included in the final model (508 missing data; 35.5% were excluded).

Table 4.

Gender-stereotyped personality trait-stratified multivariable results.

Association between gender identity and the use PPTs
		Adjusted OR*	P	95% CI
Undifferentiated	Men vs Women	0.814	0.651	0.333-1.987
Feminine	Men vs Women	1.982	0.409	0.390-10.062
Masculine	Men vs Women	0.422	0.243	0.099-1.794
Androgynous	Men vs Women	0.317	0.034	0.110-0.915

P-values < 0.05 are reported in bold.

^*Adjusted for the same variables listed in Table 3 footnotes.

95% CI, 95% confidence interval; PPT, physical and psychological treatment; OR, odds ratio.

Apart from gender, other variables associated with a greater likelihood of using PPTs included longer pain duration, use of over-the-counter pain medications, and having a postsecondary diploma. Variables associated with fewer chances of reporting the use of PPTs included feeling that pain is terrible and that it is never going to get any better, use of prescribed pain medication, and being older. The complete multivariable analysis is presented in Supplementary Content 2 (available at http://links.lww.com/PAIN/B924).

4. Discussion

This study analyzed the association between gender and the use of PPTs in chronic pain. A great majority of participants (86%) reported using at least one PPT to manage their pain. When controlling for potential confounders, being a woman was associated with a greater likelihood of using PPTs compared with men but only among participants scoring high on both masculine and feminine stereotyped personality traits.

In a study among veterans living with musculoskeletal pain in the United States, only 36% of the women and 26% of the men had used at least one PPT to manage their pain.²¹ Regional differences in the use of nonpharmacological treatments can occur (eg, differences in insurance coverage, distribution of allied health care professionals^44,53). In fact, an Australian study conducted among persons living with chronic pain using opioids revealed a prevalence of PPT use of 92%.²⁷ Canadian studies estimated that 70% to 97% of persons living with chronic pain were using PPTs.^6,36 It is encouraging to find levels as large in our study as well, showing that despite the existing barriers,⁴ people with chronic pain use such approaches.

Hot–cold, exercise, meditation, and massage therapy were the most popular PPTs (>34% of participants). The 3 first ones are low-cost treatments that can be performed easily at home and an interesting avenue of treatment for patients. Cold treatment was indeed found to be the most commonly used treatment in other studies.²¹ Regarding exercise, it is still the most recommended PPT by health care professionals⁶⁴: and evidence suggests that it could help improved function and pain.¹ As for meditation, our numbers are clearly above the prevalence of use in the general population (4%).¹⁶ Although we do not have information on when treatments were started or what treatments were attempted but stopped, treatments currently being used could be reflective of treatment maintained by participants (eg, because they are easy to implement, effective, etc).

When comparing treatment use across gender identity subgroups, our study revealed that participants self-identifying as women tended to use hot–cold therapies and meditation more. Men tended to use exercise and massage therapy more, and nonbinary people tended to use exercise and meditation more. This is consistent with the literature suggesting that men tend to do more exercise than women⁶⁴ and that women are more likely to use spiritual and distraction treatments.³³ Further studies could be conducted to better understand barriers to the use approaches such as meditation among men.

Statistically significant differences were observed in the overall prevalence of PPT use between men and women, even after adjusting for potential confounders. However, this was not true for everyone (only among participants reporting androgynous personality traits). Keeping in mind that other socioeconomic factors can be different between men and women and not captured in this study could have contributed to this association (eg, salary,⁶⁰ help-seeking tendencies,^42,49 health literacy⁴⁹), gender identity can play a central role in the way individuals perceive and respond to pain³² and can even have a greater impact on pain than biological sex itself.⁶¹ Social factors can influence pain expression, recognition, treatment, and the need to communicate pain.⁴³ Beliefs about pain also seem to influence patients' vision of pain and would therefore be linked more to gender identity, rather than biological sex and hormonal/anatomical components.⁶¹

The fact that the effect of gender identity varied across strata of gender-stereotyped personality traits challenge a one-size-fits-all approach and certain beliefs and prejudices held in the community. Will an athletic person or a man necessarily do more active treatment such as exercise? Is it appropriate to suggest meditation rather than exercise to women or more calm, spiritual, and sensitive people? In light of our results, health care professionals need to take time with their patients to offer and explain a range of treatments that may interest them using a neutral approach. In other words, we cannot guess what is appropriate for a patient based on our perception of that patient's gender. There is an association between unconscious bias, the way health care professionals treat patients, and health equity,⁶⁶ and a neutral approach is essential to help create a more inclusive and just context of care. As the consultation progresses, the clinician can tailor the education and promote PPTs to the patient needs (eg, if the patient is more or less open to psychotherapy, open to a learning curve, or more interested in a more passive approach).

Based on our results, concrete recommendations can be made to further promote the use of physical and psychological approaches to chronic pain management. First, better communication between health care professionals and people living with chronic pain is essential. Indeed, knowing that there are differences between men and women in the use of PPTs, it is important to adapt the suggested treatments to patients to aim for a more personalized treatment approach. The one size fits all is not appropriate and does not seem to help patients maintain treatment over time.³⁸ Physical and psychological treatments are essential treatment options, and follow-ups should be as regular as we would do for any pharmacological treatment. Second, better strategies to promote the use of physical and psychological approaches among some men living with chronic pain should be developed. Knowing that some men seem to use fewer PPTs in general, it is essential for health care professionals to adapt their discourse when promoting these treatments. The possibility of explaining treatments in another form may help patients to engage. For example, how can we associate meditation with a moment of relaxation of importance to the person in front of us (eg, reading, time in the garage, outdoors, hunting) rather than the classic image of meditation that people get when they hear this word (meditation hands/hand mudras). A more adapted treatment, better maintained by the patients according to their interest and their definition, could directly help their engagement.

Our study has several strengths including the measurement of our outcome using an exhaustive literature-based list of 31 PPTs and its large sample of participants recruited throughout the entire province of Quebec, Canada. Because some treatments have both physical and psychological components (eg, interdisciplinary group therapy) and a physical treatment can have impacts on psychological well-being (eg, exercise⁵⁵), we chose to not separate these 2 broad types of approaches and they were analyzed as a whole (ie, PPTs). Further studies could assess the association between gender and the use of specific treatments. It was not possible to know exactly whether PPTs were used as continuous or occasional treatments, and it was also not possible to know the exact chronic pain diagnoses because it is difficult to ascertain in the context of a self-reported study.⁵⁹ The COPE cohort was shown to be representative of random samples of persons living with chronic pain in terms of age, employment status, education level, and pain characteristics.³⁵ However, it oversampled women, probably because the web-based recruitment methods and questionnaire administration reach more women known to use Facebook¹² and to work in online environments more than men.³⁹ However, we have taken several precautions to address this limitation (a priori sample size calculation, stratify the results by gender identity, verification of goodness of fit indices, and the width of confidence intervals in the multivariable analysis). Our sample enabled us to study individuals across a diverse spectrum of gender-stereotyped personality traits (Fig. 1). A limitation of our study is related to the underrepresentation of nonbinary persons (n = 4) and racialized persons. However, aware of the importance of visibility for this already marginalized group, descriptive results of the nonbinary population are available in Supplementary Content 3 (available at http://links.lww.com/PAIN/B924). This study needs to be extended through qualitative studies examining the experiences of these underrepresented groups or by large quantitative studies that use more targeted recruiting methods. Other limits are inherent to the use of the BSRI.^8,26 There has been an evolution in the notions of societal roles, traits, and expectations concerning gender in recent years. It is important to interpret the results for what is really being measured by BSRI. Other ways of operationalizing gender during data collection such as questionnaire items about income, gender relations, children caregiving, division of labour in the household, and type of employment could have been very relevant.⁶²

5. Conclusion

This study represents an advancement in our understanding of the biopsychosocial factors linked to the use of PPTs. We strongly advise clinicians to refrain from stereotyping and instead present a variety of treatment options that align with patients' preferences. One suggestion is to use tools such as a board displaying a range of options in the form of images, ensuring a neutral approach that encourages patient engagement and prompts inquiries. It is crucial to dismantle biases and make treatments appealing to both women and men. For instance, this can be achieved by elucidating how a particular treatment can be integrated into an individual's daily life. Our results also emphasize the importance of including both gender identity and gender-stereotyped personality traits when studying the use of PPTs.

Conflict of interest statement

The authors have no conflicts of interest to declare.