Integration of sex and gender in a continuing professional development course on diabetes and depression: a mixed methods feasibility study

Abstract

Objectives

Assess the feasibility and impact of a continuous professional development (CPD) course on type 2 diabetes and depression on health professionals’ intention to include sex and gender considerations in patient care.

Design and setting

In collaboration with CPD organisations and patient-partners, we conducted a mixed-methods feasibility controlled trial with postintervention measures in three Canadian provinces.

Participants

Of 178 eligible health professionals, 127 completed questionnaires and 67 participated in semistructured group discussions.

Intervention and comparator

An interactive 1 hour CPD course, codesigned with patient-partners, on diabetes and depression that included sex and gender considerations (innovation) was compared with a similar course that did not include them (comparator).

Outcomes

Feasibility of recruitment and retention of CPD organisations and patient-partners throughout the study; adherence to planned activities; health professionals’ intention to include sex and gender considerations in patient care as measured by the CPD-Reaction questionnaire; and barriers and facilitators using the Theoretical Domains Framework.

Results

All recruited CPD organisations and patient-partners remained engaged throughout the study. All planned CPD courses occurred. Overall, 71% of eligible health professionals participated (63% under 44 years old; 79.5% women; 67.7% practising in French; 66.9% practising in Quebec; 78.8% in urban practice). After training, mean intention scores for the innovation (n=49) and control groups (n=78) were 5.65±0.19 and 5.19±0.15, respectively. Mean difference was −0.47 (CI −0.95 to 0.01; p=0.06). Adjusted for age, gender and practice settings, mean difference was −0.57 (CI −1.09 to −0.05; p=0.03). We identified eight theoretical domains related to barriers and six related to facilitators for providing sex-adapted and gender-adapted diabetes and depression care.

Conclusions

CPD training on diabetes and depression that includes sex and gender considerations is feasible and, compared with CPD training that does not, may prompt health professionals to modify their care. Addressing identified barriers and facilitators could increase intention.

Trial registration number

NCT03928132 with ClinicalTrials.gov; Post-results.

Strengths and limitations of this study.

• Continuous professional development courses that included sex and gender considerations were codesigned with patients experiencing diabetes and/or depression.

• Outcome measures were informed by theory.

• This mixed-method controlled trial used postintervention measures only, as preintervention measures were not feasible. Although randomised allocation of participants was not possible, it was feasible to conduct a mixed-method controlled trial.

Introduction

A variety of research initiatives are attempting to reduce health inequities between men and women.^{1 2} Research that includes sex-based and gender-based analysis results in more accurate evidence, more relevant recommendations, more specifically-targeted interventions and better outcomes.^3–6 Sex differences are biology-linked differences between females and males caused by different sex chromosomes, sex-specific gene expression of autosomes, sex hormones and their effects on organ systems.⁷ Gender differences arise from sociocultural processes such as the different behaviours of women and men, their exposure to environmental influences, impacts of nutrition, lifestyles or stress and attitudes towards illness, treatment and prevention.⁷ Gender roles and gender identity are influenced by a complex interplay between genetic, endocrinal and social factors.⁸ Finally, sex and gender are not straightforward binary categories. Many femininities and masculinities exist and can influence other important sociodemographic variables.⁹

During their lifetime, women are twice as likely as men to be diagnosed with depression. In contrast, three times as many men commit suicide.^{5 10 11} Recent evidence supports a link between type 2 diabetes (T2D) and depression, and shows that sex and gender are influential factors in this comorbidity.^{7 9} The prevalence of depression in patients with diabetes is higher in women than men (23.8% and 12.8%, respectively).⁷ On the other hand, a pooled result from 32 studies stated that the risk of developing T2D in patients diagnosed with depression was higher in men than in women (Odds ratio (OR)=1.63 vs OR=1.29, respectively).^{7 12 13} These differences are explained by biological differences and psychosocial factors such as body mass index, differences in the distribution of types of adipose tissue, an imbalance of sex hormones, socioeconomic status, psychosocial stress and sleep deprivation.^{7 9} Comorbidity and mortality associated with the complications of T2D and depression are also different for men and women. For instance, men develop diabetic foot syndrome at earlier ages and are more likely to have complications leading to amputations.^{7 14} Women, on the other hand, have a higher risk of metabolic syndrome and fatal coronary heart disease than men.^{7 15 16} T2D and depression are also affected by gender differences, explained in part by the different behaviours associated with representations of men and women, as well as their different perceptions of stress.^17–19

Despite the impacts of sex and gender differences on prevalence, diagnosis, treatment, outcomes and equity, evidence on the importance of these differences has yet to be translated adequately into clinical training or practice.^{2 5 20} For example, a 2017 review suggested that only 35% of studies on Canadian practice guidelines, a cornerstone of knowledge translation, reported screening, diagnosis or management considerations specific to sex or gender, and only 25% used the terms ‘sex’ and ‘gender’ correctly.²¹

Continuing professional development (CPD) is another cornerstone of knowledge translation as it mobilises professional and regulatory bodies as well as educational institutions to foster changes in clinical practice.^{22 23} We argue that integrating sex and gender considerations into CPD is a promising avenue for addressing the inequities between men and women.⁵ We define CPD as all educational activities serving to maintain or increase the knowledge, skills, work performance and relationships that a clinician needs to serve patients, the public or the profession.^{5 24 25} Courses should be informed by theory-based factors known to influence the adoption of a given behaviour. Although other factors influence behaviour change, such as organisational constraints, intention is considered an acceptable proxy. Indeed, according to Godin’s integrated model for health professional behaviour change, behavioural intention is the central influencing factor on behaviour adoption. In turn, this intention is under the influence of a number of other sociocognitive factors.²⁶ We aimed to assess the feasibility and impact of including sex and gender considerations in a CPD course on T2D and depression on health professionals’ intention to include sex and gender considerations in patient care.

Methods

Study design and setting

We conducted a non-randomised mixed-methods study with a concurrent embedded design: (1) a two-arm non-randomised controlled trial with postintervention measures only; and (2) semistructured group discussions (GDs) following the CPD course. We used the Theory of Planned Behaviour for quantitative analysis,^{27 28} the Theoretical Domains Framework (TDF) for qualitative analysis,^{29 30} and the Capability, Opportunity, Motivation and Behaviour (COM-B) model to triangulate findings.³¹ We followed the Consolidated Standards of Reporting Trials extension for Pilot and Feasibility Trials Checklist to report results.³²

This project is one of six that were funded by the Canadian Institutes of Health Research to explore sex and gender issues in knowledge translation,³³ gender transformative approaches to knowledge translation and sex-based and gender-based analysis.^{5 33}

A multidisciplinary team was created of 25 researchers, including two sex and gender specialists, three patient-partners with experience with T2D and/or mental health issues (two men and one woman), two physicians, one nurse, two CPD managers, one research assistant and two trainees. An executive committee of 12 team members (including all patient-partners) held monthly meetings addressing the main concerns in each research phase. They chose the clinical topic of the course based on needs expressed by CPD providers (see Innovation below). They then adapted an existing diabetes and depression CPD course to include sex and gender considerations and contacted CPD providers in three Canadian provinces to collaborate on implementing the courses.

Patient involvement

Three patient-partners, core members of the executive committee, contributed to governance (eg, attending meetings and courses, making executive decisions) and innovation design. They contributed their experience to the CPD course, helped collect data and interpret results, coauthored this paper and advised us on plain language use for our presentations.

Participants and recruitment

All health professionals working in the clinical settings where our CPD course was advertised, including hospitals and family medicine groups, or participating in the continuing medical education (CME) conference where the course was to be offered, were invited to participate. Invitations were by email and through the Internet registration platforms of CME conferences in three Canadian provinces (Quebec, Ontario, New-Brunswick). Participants stayed in their respective groups for the semistructured GDs that immediately followed the CPD course. Inclusion criteria were: practising health professionals available to participate in person for the whole course; and fluent in French (all our CPD courses were in French).

Innovation

Informed by a CME needs assessment by our key CPD stakeholder and partner, Médecins francophones du Canada (data not published), we chose patients with T2D and depression combined as the clinical topic, as physicians felt there was a gap in their education about this comorbidity. There is growing evidence of a link between T2D and depression and the importance of sex as a risk factor for this comorbidity.^34–36 The team adapted an existing T2D and depression CPD course to include evidence-based sex and gender considerations. The original course, a 1-hour classroom-based activity, describes links between T2D and depression, reviews CANMAT (Canadian Network for Mood and Anxiety Treatments) 2016 Depression Guidelines and reviews pharmacological and non-pharmacological treatment of T2D and depression. This original course was used in the control group. Participants in the innovation group attended the same course but adapted to integrate sex-specific and gender-specific content including: (1) definitions and differences between the concepts of sex and gender, (2) epidemiological data on the differences in incidence, prevalence, morbidity and mortality between men and women with T2D and depression and (3) a video explaining sex biases associated with these two conditions. The adapted CPD course (innovation) kept the original duration (1 hour) and medical content of the original course (comparator). Links between T2D and depression were explained together with sex and gender differences, and reviews of pharmacological and non-pharmacological treatments were condensed. As per patient-partners’ recommendations, we also held 30-min semistructured GDs with both the innovation and control group immediately following the course. In the GD, we presented a clinical case vignette on managing a patient with T2D and depression in which the health professional’s behaviour exhibited various divergences with best clinical practices. We asked participants to write down the main divergence and to categorise it within five categories determined by our team: (1) failure to mention positive factors for recovery, (2) failure to engage the patient in their health-related decision, (3) sex and gender biases, (4) failure to take into account notions of sex and gender and (5) cannot be categorised. We prompted participants to discuss their perception of sex and gender considerations by linking them to the clinical vignette and to their clinical experience of integrating sex and gender considerations in general.

Depending on the setting (hospitals, family medicine groups, CME conferences), we either (1) assigned the participants to the control or innovation group on their arrival to achieve a balanced number of participants in both groups or (2) the participants registered in one group or the other, both groups being blinded to the innovation and control group. Thus participants entered the classroom for whichever course they signed up for. There was no communication between these groups, as the two courses were given simultaneously. Participants had all received the same invitation to attend a course on T2D and depression. There was no mention of sex and gender content before participants entered the room. Efforts were made to equally divide groups regarding number and gender of participants. At registration, participants were told that it was a research project that required their consent. Participants could attend the course and receive CME credits whether they chose to participate in the study or not. All CPD courses were delivered by the same two physicians (one man, assigned to the control group, and one woman, assigned to the innovation group) in all the research settings. We planned to offer six courses (three innovation and three control), two in each province (control and innovation simultaneously). Each course (both control and innovation) was a 45-min lecture on T2D and depression followed by 15 min to fill in the CPD-Reaction questionnaire. An additional 30 min was planned for the semistructured GD.

Outcome measures

We assessed three feasibility outcome measures: recruitment, retention and adherence: (1) recruitment of >90 course participants for six courses and study participation rate of >70%,^{28 37} (2) retention of CPD organisations, collaborators and patient-partners throughout the project and (3) the holding of all planned CPD courses in all three provinces. Sample size was based on consultations with clinic managers and CPD providers and on practical considerations (eg, average size of CPD courses, venues, the course being provided in French only).

We used CPD-Reaction (French version) to measure participants’ behavioural intention to include sex and gender considerations in patient care. CPD-Reaction is a self-administered questionnaire (Cronbach α 0.79–0.89).^{38 39} Twelve items measure five constructs determined through a systematic review of theory-driven studies of behaviour change in health professionals: (1) behavioural intention, (2) beliefs about capabilities, (3) social influences, (4) beliefs about consequences and (5) moral norm.³⁷ The score for each construct is computed as the average of each item (Likert scale of 1–7), except for social influence, which is rated on a Likert scale of 1–5.²⁸ There is no global score. Finally, in GDs, we identified barriers and facilitators to including sex and gender considerations in caring for patients with T2D and depression and mapped them onto the TDF. The TDF was developed through a consensus of experts who consolidated 33 psychosocial theories of behaviour change to generate 14 domains.⁴⁰

Data collection

Quantitative data were collected postintervention with the CPD-Reaction questionnaire and sociodemographic questions.³⁸ Semistructured qualitative discussion took place in both innovation and control groups after the questionnaires were completed so as not to influence quantitative results. In both innovation and control groups, discussions were recorded and transcribed.

Analysis

Quantitative analysis

Categorical variables were described by reporting absolute (n) and relative (%) frequencies. Continuous variables were described by their measure of central tendency (mean and/or median) and dispersion (SD and percentiles). Covariance analysis was used to compare the scores of the innovation and control groups. As the intention did not have a perfectly Gaussian distribution, we also compared intention scores using Wilcoxon’s non-parametric analysis and used the Kruskal-Wallis test to compare medians. We used Spearman’s rank test to assess the correlation between the intention scores and psychosocial factors (social influence, beliefs about capabilities, moral norms, beliefs about consequences). We used general linear models to assess whether the intention score varied significantly from the control group to innovation group after adjusting for confounding factors. These factors were identified using the 10% change in the regression coefficient associated with the exposure variable.^{41 42} However, to increase the appearance validity of the model, we constructed a separate model in which we forced age, gender and practice environment. SAS software (V.9.4) was used for all statistical analyses. The empirical significance threshold (p value) was set at 0.05 in bilateral analysis.

Qualitative analysis

The discussion transcripts were imported into N’Vivo V.12 for analysis. Using the TDF as a guide, two researchers reviewed and agreed on codes and data were simultaneously coded using a thematic deductive approach (ADT, AGo).²⁹ Data were then refined into TDF domains. As the discussion occurred in French, all illustrative quotes were translated into English by a master’s student (ADT) and reviewed by a scientific translator. We calculated the frequency of each barrier and facilitator by recording the number of times it was mentioned in the four GDs (GDs 1–4).

Triangulating quantitative and qualitative data

We triangulated quantitative and qualitative data to propose practical theory-driven recommendations for improving our CPD innovation.⁴³ We compared the five psychosocial determinants measured in the CPD-Reaction questionnaire to the domains of the TDF. We observed where quantitative and qualitative data converged, where they offered additional information on the same constructs, and where they diverged. We derived recommendations using the COM-B model of behaviour.⁴⁴ COM-B proposes three criteria essential for a behaviour to occur: capacity, opportunity and motivation.⁴⁵ The subcategories of these criteria can be linked to the TDF domains and their associated barriers or facilitators. The COM-B also proposes nine intervention functions assigned to TDF domains that can prompt behaviour change: education, persuasion, incentivisation, coercion, training, restriction, environmental restructuring, modelling and enablement.^{31 44 46} Recommendations were made by identifying which of these intervention functions matched our results and then selecting relevant function-associated behaviour change techniques.⁴⁴

Results

Recruitment and participant characteristics

We offered the 12 CPD courses (ie, six innovation/control pairs) in each of three Canadian provinces: Quebec, Ontario and New Brunswick. Four pairs of courses were held in Quebec (two in Montreal, 10 October 2018 and 30 October 2019, and two in Quebec City, 17 October 2019 and 29 January 2019), one in Ontario (Ottawa, 8 November 2019) and one in New-Brunswick (Moncton, 4 October 2019).

Figure 1 illustrates the flow of participants. The participation rate (ratio of users who participated in the study to those who took the training) was 71% (127/178). Forty-nine of 92 questionnaires were analysed from the innovation groups and 78 of 86 from the control groups. Most participants were under 44 years old (n=80, 63%), women (n=101, 79.5%), practised in French (n=86, 67.7%), in Quebec (n=85, 66.9%) and in an urban setting (n=100, 78.8%) (table 1).

Figure 1.

Flow chart of participants. *This is an approximate figure given the changing dynamics of the hospital’s professional environment; an email was sent to 2000 employees including healthcare professionals, others were invited using posters in the training sites, oral communication at a meeting with the organising team of the clinical setting, and announcements in Médecins francophones du Canada’s conference calendar. CPD, continuing professional development.

Table 1.

Sociodemographic characteristics of the participants in innovation and control groups

	Total	Innovation group	Control group
No. of participants	127	49	78
Age (years)*
<44	80 (63.0)	28 (57.1)	52 (66.7)
≥45	42 (33.1)	19 (38.8)	23 (29.5)
Missing data	5 (3.9)	2 (4.1)	3 (3.8)
Gender*
Women	101 (79.5)	40 (81.6)	61 (78.2)
Men	19 (15.0)	7 (14.3)	12 (15.4)
Missing data	7 (5.5)	2 (4.1)	5 (6.4)
Language of practice*
French	86 (67.7)	32 (65.2)	54 (69.2)
Other	36 (28.3)	15 (30.6)	21 (26.9)
Missing data	5 (4.0)	2 (4.1)	3 (3.9)
Province of practice
Quebec	85 (66.9)	31 (63.2)	54 (69.3)
Ontario	18 (14.2)	9 (18.4)	9 (11.5)
New Brunswick	16 (12.6)	7 (14.3)	9 (11.5)
Missing data	8 (6.3)	2 (4.1)	6 (7.7)
Practice environment*
Urban	100 (78.8)	39 (79.6)	61 (78.2)
Rural	14 (11.0)	4 (8.2)	10 (12.8)
Missing data	13 (10.2)	6 (12.2)	7 (9.0)

*n (%).

Quantitative results

Feasibility

We recruited a total of 127 participants, a 41% increase from our target of 90 participants. Collaborators and executive committee members remained involved throughout the project. We held monthly executive committee meetings as planned. Our CPD trainings were held in the three provinces as planned. We gave 12 courses instead of the 6 initially planned, as additional organisations in Quebec City (n=1) and Montreal (n=2) showed interest. Due to time constraints imposed by CME settings, completing 1.5 hours (45-min course, 15-min evaluation and 30-min discussion) in all settings was not possible; therefore, we held the GDs in only two out of the six settings (Montreal and Ottawa).

Behavioural intention

The innovation aims to influence behaviour by modifying intention and its psychosocial determinants. For example, the innovation could change beliefs about capabilities (or confidence) by increasing health professionals’ knowledge about the desired behaviour. Table 2 shows scores for intention and its psychosocial determinants for innovation and control groups as evaluated using the CPD-Reaction questionnaire. Mean difference (MD) between innovation and control scores for the four psychosocial determinants of behaviour change influencing intention was: MD=0.16 for social influence (95% CI −0.26 to 0.58), MD=0.63 for belief about capabilities (95% CI 0.21 to 1.06), MD=0.25 for moral norm (95% CI −0.21 to 0.72) and MD=0.22 for belief about consequences (95% CI −0.23 to 0.67). The mean intention score for including sex and gender considerations in patient care was higher in the innovation than in the control group, that is, 5.65 (±0.19) versus 5.19 (±0.15), on a scale from 1 (low) to 7 (high). The mean difference between the two groups was −0.47 (95% CI −0.95 to 0.01), with a p value of 0.06 (online supplemental table 1). No statistically significant differences were observed for the remaining four psychosocial determinants. Bivariate analysis showed that the higher median for intention was significantly associated with age over 45 (p=0.03) and a rural practice environment (p=0.02) (online supplemental table 1). After adjusting for age, gender and practice environment, the mean difference in intention between the two groups was statistically significant: −0.57 (95% CI −1.09 to –0.05), with a p value of 0.03 (table 3).

Table 2.

Continuing professional development-Reaction questionnaire mean scores

	Total	Innovation	Control	Absolute mean difference (95% CI)
No. of participants	127	49	78	–
Psychosocial determinants—score range (1–7)*
Social influence	4.62 (4.42 to 4.83)	4.72 (4.44 to 5.00)	4.56 (4.27 to 4.85)	0.16 (–0.26 to 0.58)
Beliefs about capabilities	5.1 (4.90 to 5.33)	5.50 (5.27 to 5.74)	4.87 (4.56 to 5.17)	0.63 (0.21 to 1.06)
Moral norm	5.90 (5.69 to 6.13)	6.06 (5.80 to 6.32)	5.81 (5.48 to 6.14)	0.25 (–0.21 to 0.72)
Beliefs about consequences	5.68 (5.46 to 5.90)	5.82 (5.52 to 6.11)	5.60 (5.28 to 5.91)	0.22 (–0.23 to 0.67)
Intention*	5.37 (5.13 to 5.60)	5.65 (5.36 to 5.95)	5.19 (4.85 to 5.52)	0.47 (–0.01 to 0.95)

*Mean (95% CI).

Table 3.

Mean difference of the intention score between innovation and control groups

	Model 1*		Model 2†		Model 3‡
	β (95% CI)	P value	β (95% CI)	P value	β (95% CI)	P value
Innovation	Reference		Reference		Reference
Control	−0.47 (−0.95 to 0.01)	0.057	−0.61 (−1.10 to −0.12)	0.015	−0.57 (−1.09 to −0.05)	0.031

*Non-adjusted.

†Adjusted for age and gender.

‡Adjusted for age, gender and environment of practice.

Qualitative findings

Due to time constraints imposed by CME settings, we held the GDs in two out of the six settings, Montreal, 30 October 2019 and Ottawa, 8 November 2019. Thus, 4 semistructured GDs (GD1, GD2, GD3, GD4) were conducted and 67 health professionals participated, reporting a variety of barriers and facilitators (table 4).

Table 4.

Mapping facilitators and barriers to the Theoretical Domains Framework (TDF) with illustrative quotes and frequencies

TDF domain	Facilitator/barrier	Illustrative quotes*	Frequencies† (N=4 groups)
Skills	The health professional acknowledges different treatment methods by gender (facilitator)	‘Dominique, is that a man or a woman? … Because they are probably not treated the same’ (GD4)	4
	The health professional acknowledges different clinical representation by gender (facilitator)	‘…I work as a nurse in cardiac and pulmonary rehabilitation, and … it is a fact, that women come less [to rehabilitation programs] in general than men. Women often will quit [rehabilitation] or they won’t come because they’re taking care of everyone. But something happens [illness] and then they don’t have time to take care of themselves, because it’s too much’ (GD3)	1
	The health professional made assumptions about the gender of the patient when analysing a clinical vignette (barrier)	‘I assumed that it was a guy’ (GD3) / ‘I presumed that it was a girl’ (GD4)	3
Beliefs about capabilities	The health professional feels he/she can accurately observe the phenotype of the patient (facilitator)	‘At the first contact we have with a patient … we see the phenotype there without talking about gender, it’s one of the things that jumps out at you when you’re taking notes.’ (GD3)	3
Social influences	The health professionals assume the patient’s gender based on his/her societal role (barrier)	‘I heard ‘civil servant’, I don’t know, in my head I was like ‘civil servant’, so it’s a man.’ (GD4)	3
Knowledge	The health professional knows the differences between sex and gender in scientific literature (facilitator)	‘Yes, that’s it actually, the biological aspect you certainly take into account in the study, but we are talking about the [social] categories of sex and gender… And menopause, and on the other hand [there’s] also andropause’ (GD2)	2
	The health professional did not ask the gender of the patient when analysing a clinical vignette (barrier)	‘Well, I don’t know why we didn’t note it [the gender of the patient], I don’t have the answer to that. But … when we talk about the clinical context it is systemically noted in the first … sentence, in the first two words (of notes documenting a consultation]. It’s hard to say that we ignore it [gender of the patient]. We didn’t notice it here, but in clinical practice, have you ever met a patient without identifying their gender?’ (GD3)	2
	The health professional is not aware of the concepts of sex and gender when analysing a clinical vignette (barrier)	‘…but in the seminar, there was no emphasis on that, so it didn’t jump out at us,’ (GD3)	1
Beliefs about consequences	The health professional mentions that they would not change their therapeutic approach according to the patient’s gender (barrier)	‘I would say that I didn’t see the need to know if it was a man or a woman…I never asked myself the question…’ (GD1)	2
Environmental context and resources	The patient’s sex is routinely recorded in medical notes (facilitator)	‘…in the clinical context it’s [the sex of the patient] systematically noted in the first lines in every consultation. In the first sentence, in the first two words. It’s hard to say that we ignore it.’ (GD3)	2
	The androcentric nature of the French language (the use of masculine generic language to refer to men and women, as well as other gender representation) (barrier)	‘In French everything is masculine until you know, like in the room here [mostly women participants] we’ll say like “ils ont fait ça” [ils is a masculine pronoun] because you are the only men, but…’ [generalizing to the masculine pronoun] (GD3) / ‘The language doesn’t help … [to differentiate between men and women].’ (GD3)	1
	The healthcare professional perceives that the language used by physicians towards a patient may be different according to sex and gender (barrier)	‘Well it’s about when you say ‘our diabetes’ and ‘your depression’, if it had been a woman would we have said the same thing?…‘your depression’ ‘our diabetes’…’ (GD2) [referring to the bias in the language to describe ‘your’ depression versus ‘our’ diabetes]	1
Social/professional role and identity	The health professional reflects positively on his/her relationship with the patient (facilitator)	‘I work in an exclusively white environment, and I am the only black person, and I have no problem whether [the patient] is male, female or a child’ (GD3)	1
Intentions	The health professional has the intention to change his/her therapeutic approach by considering the differences of gender (facilitator)	‘With the information that I have here [clinical description of vignette], if I had ‘menopaused woman’, then I think I would have researched more, but with what I had here, I didn’t [see the need].’ (GD4)	1
	The health professional does not have the intention to change his/her therapeutic approach by considering the differences of gender (barrier)	“With what I have here [descriptive information of the clinical vignette], I am not sure to what extent I would have changed my approach’ (GD4)	1
Goals	The health professional does not perceive the integration of the concepts of sex and gender in clinical practice as a priority (barrier)	‘It wasn’t important … the most important, [but] that doesn’t mean that [the lack of sex and gender consideration in the clinical vignette] wasn’t perceived” (GD4)	1
Memory, attention and decision processes	The health professional does not consider that sex and gender are necessary parts of the decision-making process (barrier)	‘If it is not obvious, we are not inclined to do it… [take into consideration the sex and gender of the patient]’ (GD2)	1

*Free translation from French.

†The number of times that the barrier/facilitator appeared in the transcript.

GD, group discussion.

Barriers and facilitators mapped to the TDF domains

Ten barriers mapped to eight of the 14 TDF domains and seven facilitators mapped onto six of the domains. The most frequent barriers were related to Skills (eg, failing to consider a patient’s gender) (n=3) and to Social influence (eg, making gender assumptions about employment) (n=3). The most frequent facilitators were also related to Skills (n=4) (table 4).

We mapped to the Skills domain when the participants asked whether their patient was a woman or man before analysing the clinical vignette, or else failed to ask the question (the fictive name of the patient—Dominique—was strategically ambiguous). Thus, failure to ask was coded as a barrier, and asking was coded as a facilitator. Discussion about information on sex and/or gender was coded as a facilitator in the Knowledge domain, but reporting differentiating between women and men patients in clinical practice was coded as a facilitator in the Skills domain. When participants reported not needing to know the patient’s gender because this information would not have changed their intervention, we mapped the barrier to Beliefs about consequences domain. Participants documented some differences between men and women patients in their clinical practice, demonstrating ability acquired through practice to include sex and gender considerations. Participants also reported they did not ask the sex of the patient in the clinical vignette as they automatically observe a patient’s sex in practice, so did not feel the need to mention it in this context. This facilitator was mapped to the domain beliefs about capabilities (n=3). Some participants reported that they routinely observe and record a patient’s sex when taking notes. This facilitator was mapped to the domain environmental context and resources, since it this is an institutional practice reflecting an organisational clinical culture, and could foster further awareness and consideration of sex and gender (table 4).

Triangulation

CPD-Reaction psychosocial variables matched barriers that mapped onto to the TDF domains beliefs about consequences, social influence and intentions. CPD-Reaction psychosocial variables also matched facilitators that mapped onto to the TDF domains beliefs about capabilities and intentions. We identified six additional psychosocial variables from the TDF: knowledge, skills, goal, memory, attention and decision processes, environmental context and resources, social/professional role and identity. Results of triangulation were summarised with consequent recommendations (online supplemental table 2). Recommendations for improving the CPD training were based on behaviour change techniques associated with the following functions: modelling, training, environmental restructuring, enablement, education and goal settings (online supplemental table 2).⁴⁴ Training (n=5) and education (n=4) were the most frequent functions used in the recommendations.

Discussion

We assessed the feasibility and impact of including sex and gender considerations in a CPD course on T2D and depression care on health professionals’ intention to include sex and gender considerations in patient care. Recruited CPD organisations, collaborators and patient-partners stayed engaged throughout the study. All planned activities occurred and 71% of targeted health professionals participated. The intention to include sex and gender considerations in patient care was higher in the innovation group, and statistically significant when controlling for age, gender and practice sites. Barriers were mostly related to skills and social influence and facilitators to skills and beliefs about capabilities. We triangulated results and produced recommendations for improving the CPD course. The following observations could enable CPD organisations to systematically improve CPD by integrating sex and gender considerations into their existing material.

First, all our predetermined feasibility criteria were met. In fact, due to increased interest in the topic, we recruited more participants and gave more CPD activities than planned. Recruitment may also have improved because we involved stakeholders early on in the research process, including in applying for the grant. Early engagement of stakeholders has been associated elsewhere with more successful recruitment.⁴⁷ Therefore, elements that should be considered when designing similar CPD activities include, but are not limited to: (1) successful collaboration and cocreation with CPD organisations early on including during grant writing, (2) offering CME accreditation for the CPD activities, (3) the duration of the training and (4) the evidence base relevant to the clinical topic.⁴⁸

Second, the CPD course that included sex and gender considerations increased health professionals’ intention to include sex and gender considerations in patients’ care. This may suggest a significant knowledge gap among participants. Studies show that health professionals lack knowledge of sex and gender differences in disease manifestation and outcomes and fail to recognise the gender constraints that their patients face.^49–52 For example, in a cross-sectional survey of physicians (71% male), 55% said that the medical curriculum did not adequately prepare them for dealing with sexual health problems, particularly those of female patients.⁴⁹ In another study, only 49% of primary care physicians (n=200, 65% male) and 59% of cardiologists (n=100, 85% male) reported that their training prepared them to assess female patients’ cardiovascular risk.⁵¹ Our study represents a promising avenue for rectifying these gaps. Furthermore, bivariate analyses of the between-group difference in the intention scores yielded significant results in older, but not younger, participants and in those practising in rural area. Their age and geographical isolation perhaps reduced their exposure to sex and gender issues, which have only been included in medical curricula since they qualified.⁵² They may also have less access to CPD training due to isolation, poor technological resources, low financial support^{53 54} and geographical variations in medical practice styles.^{55 56} Future studies could further investigate the perceptions of health professionals in rural settings on age and gender. They could also document if patients experience geographical differences in care regarding sex and gender. Training could target older and rural health professionals, who seemed more open to modifying their clinical practice.

Third, beliefs about capabilities as a facilitator showed the strongest mean difference between the innovation and control groups. These results are consistent with a literature review of 277 studies showing that the mechanisms of action most frequently associated with behaviour change techniques are beliefs about capabilities and intention.⁵⁷ Adding a practical component to the CPD course could strengthen beliefs about capabilities. Also, several barriers and facilitators to considering sex and gender in patient care were identified. Our qualitative analysis showed that participants did not consider integrating sex and gender into clinical practice as a priority, with social influences emerging as an important barrier. The social influence score as measured by CPD-Reaction also showed the lowest impact (MD=0.16), suggesting that the training did not address this factor (table 2). A CPD course could offer a reflective segment on how social influence could be affecting their clinical practice.^{56 58} Furthermore, belief about consequences had one of the lowest MD (0.22) of the five psychosocial determinants, and one associated barrier (n=2). This could be remedied by focusing more on the consequences of not integrating sex and gender into clinical practice.⁵⁰

Finally, in spite of the low priority given to sex and gender by our participants, qualitative analysis demonstrated that opportunities already exist for integrating these considerations into practice, such as the routine documenting of the patient’s sex. CPD strategies could make more of these opportunities.⁵⁹ For example, CPD activities could advocate for sex-adapted and gender-adapted care when treating men and women for diabetes and depression. Indeed, specific attention could be given to diabetic foot care when treating men, while specific attention could be given to blood-glucose regulation and to family and lifestyle issues when treating women.^{7 60}

This innovation could be adapted to medical fields other than T2D and depression, and to other countries and areas outside French-speaking provinces of Canada. While many of the barriers participants mentioned were culture-specific and language-specific to the Quebec or francophone context, many other languages (eg, Spanish, German, Italian and Portuguese) also generalise everything to the masculine gender, suggesting shared linguistic barriers. However, each culture has highly specific sex and gender norms affecting physicians’ clinical assumptions.⁶¹ Our qualitative results highlight the fact that CPD on sex and gender considerations must be tailored to specific cultural contexts¹⁷ and incorporate sex-based and gender-based analysis tools.⁶²

Our study has a few limitations. As we used a single postintervention measure, we cannot attribute the difference between the two groups solely to the innovation. However, our analysis suggests that those who completed the innovation increased their intention, as well as increasing all four psychosocial predictors, suggesting an association with the innovation. Second, the fact that participants could choose which course to attend (according to conference guidelines), and hence the non-randomised nature of the study, may have biased our feasibility findings. Third, the training was given by teachers of different genders for the innovation and control groups (a woman in the innovation group and a man in the control group). As a bias could have been introduced owing to differences in communication styles between men and women, the teaching teams practised the courses several times to ensure that teaching methods were equivalent. In addition, we ensured the teachers stayed with their respective groups for the six data collections. Fourth, due to ethics guidelines, we only analysed questionnaires completed by participants who had also signed consent forms. Although the human resources for both groups were the same (trainer, research-assistant and patient-partners), the control group had an extra team member, resulting in unequal numbers of participants who signed consent in each group. The presence of this extra member could also explain the difference in the number of questionnaires collected in the two groups. Fifth, our study had low participation rates, although it did meet our feasibility target sample size given the logistical and contextual constraints. Recruitment followed the way CPD activities are usually publicised in large organisations (a scattershot approach that includes posters, calendars, mass emailing); thus, the participation rate did not necessarily reflect a lack of interest. Our study approach was pragmatic, that is, it took place in a real CPD training setting. This pragmatic study will inspire other health services researchers and implementation scientists to collaborate with CPD stakeholders and knowledge users to embed their studies in real CPD training settings. Sixth, although there is evidence that intention is an effective determinant for measuring behaviour change,³⁹ it is limited as a proxy. Finding other reliable measures of behaviour change is challenging.⁶³ However, identifying barriers and facilitators to change is a first step.⁶³ Semistructured GDs using a clinical vignette have also been shown to contribute to clinical behaviour change.⁶³ Methods such as audit and feedback, as well as ‘commitment to change statements’ could reduce the intention-behaviour gap and strengthen the understanding of clinical changes following CPD activities.^{64 65} Last, our GDs attracted many participants, limiting both participants’ opportunity to speak and the depth of the discussion. Our mixed-methods approach is a strength of this study and our findings support the feasibility of a randomised trial informed by identified barriers and facilitators.

Conclusion

A CPD course with sex and gender considerations was feasible and well received by health professionals. The significant between-group difference in the intention scores suggests the innovation had a favourable impact on health professionals’ intention to include sex and gender considerations when caring for their patients with T2D and depression. However, caution is required in interpreting our results as this effect may be attributed to other sources given the non-randomised nature of our study. Future randomised controlled trials are needed to control for potential selection biases to confirm our results and identify barriers and facilitators in sex-adapted and gender-adapted diabetes and depression care. Our findings will inform future CPD initiatives that address this topic and other inequities in healthcare pertaining to sex and gender.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by Centre intégré universitaire de santé et de services sociaux de la Capitale-Nationale (CIUSSS-CN) Ethics Board (2017-2018-16 MP), the Hôpital Montfort Research ethics board (19-20-05-009), Vitalité health network research ethics board (CER-2019-18). Participants gave informed consent to participate in the study before taking part.