1. Introduction
Major depressive disorder (MDD) and bipolar disorder (BD) are recurrent and chronic mood disorders associated with multi-domain functional impairment.Reference Greer, Kurian and Trivedi 1 , Reference McIntyre, Berk, Brietzke, Goldstein, López-Jaramillo and Kessing 2 At the population level, MDD and BD are the leading causes of disability globally.Reference James, Abate, Abate, Abay, Abbafati and Abbasi 3 Moreover, cost-of-illness studies report that indirect costs (eg, presenteeism and absenteeism) make up a large proportion of the yearly US$200 billion costs of MDD and BD, reflecting the substantial impairments in workplace function associated with these disorders.Reference Ho, Mak, Chua, Ho and Mak 4 –Reference Zuelke, Luck, Schroeter, Witte, Hinz and Engel 7 The foregoing observatin is even more salient in light of rising presenteeism in the digital workplace during the COVID-19 pandemic.Reference Ferreira, Mach, Martinez and Miraglia 8 Functional recovery has thus become well-recognized by both clinicians and workplaces alike as a key therapeutic target in mood disorder populations.Reference Cao, Xu, Chen, Wang, Lee and Rosenblat 9 –Reference Sheehan, Harnett-Sheehan, Spann, Thompson and Prakash 12
Replicated evidence has established a strong association between functional impairment in the workplace and several features of MDD and BD, including depressive symptoms and cognitive impairment.Reference Maj, Stein, Parker, Zimmerman, Fava and De Hert 13 , Reference McIntyre, Alda, Baldessarini, Bauer, Berk and Correll 14 Depressive symptom severity has been associated with loss of productive employment, increased disability payments, and poor time management.Reference Bender and Farvolden 15 –Reference Woo, Kim, Hwang, Frick, Choi and Seo 18 Additional evidence suggests that cognitive impairment is a robust predictor of workplace outcomes and role functioning in MDD and BD populations. Cognitive impairment remains a strong predictor even after controlling for the relationship between cognitive impairment and mood symptoms.Reference Cha, Carmona, Subramaniapillai, Mansur, Lee and Lee 19 –Reference McIntyre, Soczynska, Woldeyohannes, Alsuwaidan, Cha and Carvalho 22
Both depressive symptoms and cognitive function have been identified as modifiable targets to improve workplace performance in BD and MDD populations.Reference Toyoshima, Inoue, Shimura, Masuya, Ichiki and Fujimura 23 There are few treatments with demonstrated efficacy for cognitive impairment in populations with mood disorders.Reference Bortolato, Miskowiak, Köhler, Maes, Fernandes and Berk 24 –Reference Zuckerman, Pan, Park, Brietzke, Musial and Shariq 26 However, some antidepressants have been shown to improve workplace outcomes through reductions in depressive symptoms and, in some cases, through pro-cognitive effects.Reference Finkelstein, Berndt, Greenberg, Parsley, Russell and Keller 27 –Reference McIntyre, Lophaven and Olsen 29 For example, vortioxetine, a multimodal antidepressant with effects on norepinephrine, serotonin, histamine, and cholinergic neurotransmitter systems, has well-established pro-cognitive effects in mood disorder populations.Reference Cao, Park, Subramaniapillai, Lee, Iacobucci and Mansur 30 –Reference Rosenblat, Kakar and McIntyre 32 Emerging evidence also suggests that ketamine, an N-methyl-D-aspartate receptor antagonist with rapid efficacy in treatment-resistant depression (TRD), and infliximab, a tumor necrosis factor antagonist with efficacy in BD and TRD, may also demonstrate pro-cognitive effects.Reference Dadsetan, Balzano, Forteza, Cabrera-Pastor, Taoro-Gonzalez and Hernandez-Rabaza 33 –Reference Pan 38
Extant literature suggests that cognitive impairment mediates the relationships between depressive symptoms and both psychosocial and functional outcomes.Reference Kiosses and Alexopoulos 39 –Reference Paelecke-Habermann, Pohl and Leplow 41 More specifically, recent studies suggest that cognitive impairment mediates the relationship between depressive symptoms and workplace outcomes.Reference Toyoshima, Inoue, Shimura, Masuya, Ichiki and Fujimura 23 , Reference Liu, Zhou, Zheng, Wang, Zhan and Li 42 , Reference Woo, Rosenblat, Kakar, Bahk and McIntyre 43 Notwithstanding, there have been relatively fewer studies that have evaluated the specific cognitive domains involved in this relationship and still fewer that have focused on BD populations.Reference Bonnín, González-Pinto, Solé, Reinares, González-Ortega and Alberich 44 –Reference Knight and Baune 46 Herein, we further substantiate the relationship between depressive symptoms, cognitive function, and workplace outcomes in populations with MDD or BD. Specifically, we assess how overall cognitive function, executive function, and subjective cognitive function mediate the relationship between depressive symptoms and workplace function in BD and MDD populations receiving vortioxetine, ketamine, or infliximab therapy.
2. Methods
2.1. Dataset study design
The data used in the analysis herein were combined from three separate datasets: an open-label vortioxetine trial with individuals with MDDReference McIntyre, Subramaniapillai, Park, Zuckerman, Cao and Lee 47 (ClinicalTrials.gov Identifier: NCT03053362), an observational study of outpatients with TRD receiving intravenous (IV) ketamine at the Canadian Rapid Treatment Center of Excellence (CRTCE)Reference McIntyre, Carvalho, Lui, Majee, Masand and Gill 48 (ClinicalTrials.gov Identifier: NCT04209296), and a double-blind placebo-controlled infliximab trial with individuals with bipolar I/II depressionReference McIntyre, Subramaniapillai, Lee, Pan, Carmona and Shekotikhina 37 (ClinicalTrials.gov Identifier: NCT02363738). The three studies will hereafter be referred to by the intervention evaluated in the study (ie, the vortioxetine trial, the ketamine trial, and the infliximab trial). The methodologies used in the foregoing studies are described extensively in their original reports. An outline of key study features is provided in Table 1.
Dataset Characteristics of the Vortioxetine Trial, Ketamine Trial, and Infliximab Trial
2.1.1. Vortioxetine trial
The vortioxetine trial is an open-label trial of vortioxetine in adults with MDD as defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM) and healthy controls (HC). Participants were included in the study if they were experiencing a current major depressive episode (MDE), and at least one other MDE in the past.Reference McIntyre, Subramaniapillai, Lee, Pan, Carmona and Shekotikhina 37 Exclusion criteria included current alcohol and/or substance use disorder, concurrent diagnosis with a psychiatric disorder other than MDD, and medication intended for, or presumed to, affect cognitive function.
A total of 158 participants were included in the study (n MDD = 100; n HC = 58). Patients received 10–20 mg/day of open-label vortioxetine for 8 weeks. The primary outcome was a change in cognition from baseline to study endpoint, as measured by the THINC-integrated tool (THINC-it), a screening tool for cognition and cognitive impairment. Assessments were completed at weeks 0, 2, 4, and 8 of the study.
2.1.2. Ketamine trial
Data were also obtained from the CRTCE, a clinical and research facility located in Mississauga, Canada.Reference McIntyre, Rodrigues, Lee, Lipsitz, Subramaniapillai and Gill 49 The CRTCE provides intravenous (IV) ketamine infusion treatment to eligible adults with TRD of least Stage 2 in severity and a primary diagnosis of DSM-5-defined MDD or BD. Ketamine infusions are administered adjunctively to current medications for most patients; however, those taking an irreversible monoamine oxidase inhibitor (MAOI), naltrexone, or benzodiazepines discontinued use prior to ketamine therapy. Psychiatric comorbidities and suicidal ideation were not considered exclusion criteria for treatment.
The data from the CRTCE analyzed herein were collected from July 2018 to July 2020 and comprised a total sample of 238 patients. Patients received a total of four IV ketamine infusions over a period of 1.5–2 weeks for acute treatment. Patients received the first two ketamine infusions at 0.5 mg/kg diluted in 0.9% saline solution over a period of 40–45 minutes. If a suboptimal decrease in depressive symptom severity was observed, the dosing was optimized to 0.75 mg/kg for the following two infusions. The primary outcome was a change in depressive symptoms as measured by the 16-item Quick Inventory of Depressive Symptomatology-Self Report (QIDS-SR16). Measures were completed at the first infusion, fourth infusion, and post-initiation treatment visit.
2.1.3. Infliximab trial
The infliximab trial is a double-blind, placebo-controlled randomized trial of the antidepressant efficacy of adjunctive infliximab in outpatient adults with BD I/II experiencing a current MDE. Participants were eligible if they received a minimum of two treatments for BD at least 4 weeks before participation in the trial. Participants were not excluded based on current medications.
Sixty participants were randomized to receive a placebo (n = 31) or 5 mg/kg IV infliximab infusions (n = 29). Infusions were administered over 120 minutes at weeks 0, 2, and 6. The primary outcome was a change in depressive symptoms between baseline and endpoint, as measured by the Montgomery–Åsberg Depression Rating Scale (MADRS). Clinical assessments were completed at weeks 0, 1, 2, 3, 4, 6, 8, 10, and 12. Cognitive assessments were completed as secondary outcomes at weeks 0, 2, and 12.
2.2. Variable operationalization
2.2.1. Depressive symptoms
In the vortioxetine and infliximab trials, depressive symptoms were measured using MADRS.Reference Montgomery and Asberg 50 In the ketamine trial, depressive symptoms were measured using the 16-item Quick Inventory of Depressive Symptomatology-Self Report (QIDS-SR16) assessments.Reference Trivedi, Rush, Ibrahim, Carmody, Biggs and Suppes 51 For the purposes of the analysis herein, QIDS-SR16 scores were converted into MADRS scores using a standardized table.Reference Carmody, Rush, Bernstein, Brannan, Husain and Trivedi 52
2.2.2. Workplace function
The Sheehan Disability Scale (SDS) is a self-rated scale that assesses functional impairment in the domains of work–school, social life, and family life/home responsibilities.Reference Sheehan, Harnett-Sheehan and Raj 53 It is a validated measure of functional outcomes in depression and is sensitive to treatment effects.Reference Sheehan and Sheehan 54 The patient rates 3 items on an analog scale of 0 to 10 to indicate the degree to which their symptoms have interfered with each of the 3 functional domains. A rating of 0 corresponds to “not at all” disruptive and a rating of 10 corresponds to “extremely” disruptive. A score over 5 is associated with significant functional impairment. The three items may be summed to obtain a global measure of functional impairment, ranging from 0 to 30. The analysis herein used the work–school item of the SDS as a measure of function in the workplace and/or educational setting.
2.2.3. Cognitive function
Cognitive function was measured using the 5-item version of the Perceived Deficits Questionnaire (PDQ-5), the Digit Symbol Substitution Test (DSST), and/or the Trail Making Test Part B (TMT-B). The PDQ-5 is a patient-reported measure of cognitive function originally developed for patients with multiple sclerosis.Reference Sullivan, Edgley and Dehoux 55 More recently, the PDQ-5 has been used to assess subjective cognitive function in other populations, including MDD and BD.Reference Lawrence, Roy, Harikrishnan, Yu and Dabbous 56 The DSST and TMT-B are 2 objective assessments of cognitive function. The DSST is a pencil-and-paper cognitive test that involves the use of a key to match a series of symbols to numbers.Reference Jaeger 57 , Reference McLeod, Griffiths, Bigelow and Yingling 58 The DSST is a polyfactorial cognitive test; the time taken to complete the task and the number of matching errors made are indicators of cognitive function in domains including attention, visuospatial processing, working memory, and psychomotor speed.Reference Amaresha, Danivas, Shivakumar, Agarwal, Kalmady and Narayanaswamy 59 The TMT-B is also a polyfactorial cognitive test. This test requires the individual to draw lines to connect alternating numbers and letters.Reference Tombaugh 60 The time taken to correctly complete the task reflects psychomotor speed, visuospatial processing, and attention.Reference Kortte, Horner and Windham 61
2.3. Inclusion and exclusion criteria
A summary of the inclusion criteria for the original studies is outlined in Section 2.1. In the analysis herein, further inclusion–exclusion criteria were applied. Participants were excluded if they did not receive an antidepressant intervention (ie, participants receiving a placebo in the infliximab trial and healthy controls in the vortioxetine trial). Participants were also excluded if data were not available for MADRS, SDS work, and at least one test of cognitive function at baseline and endpoint measures.
2.4. Data extraction
Two reviewers (JKT and DH) compiled data from the three datasets. Data were collected for primary measures of interest (ie, MADRS total score, SDS work score, TMT-B time, PDQ-5 total score, DSST errors, DSST time) and demographic variables (ie, age, sex, race, marital status, primary diagnosis, BMI), where available. The ketamine trial did not report race or marital status. The infliximab trial did not assess cognition using the TMT-B. The infliximab trial also employed the full PDQ assessment instead of PDQ-5, and thus these scores were not included in the analysis herein.
2.5. Statistical analysis
The statistical analysis was conducted using SPSS Statistics Software, Version 25. Descriptive statistics were generated to characterize sociodemographic characteristics (ie, age, sex, race, marital status, BMI, primary diagnosis, education, and employment status). Chi-squared and Kruskall–Wallis tests were conducted to compare demographic measures between the three datasets.
The mediation model was generated using PROCESS Macro for SPSS, Version 3.5 (Hayes, 2013). Model 4 was used to determine if the relationship between MADRS and SDS work scores was mediated by changes in PDQ-5 total scores, TMT-B time, DSST error scores, and DSST symbols scores from baseline to endpoint. Each model was controlled by covariates (age, BMI, sex). A bias-corrected 95% confidence interval (CI) was calculated using bootstrapping of 5000 samples. The significance of associations was defined as a two-tailed P-value of < .05. Sub-analyses were originally planned to be conducted for the primary diagnosis (ie, MDD and BD) and treatment type (ie, vortioxetine, ketamine, or infliximab) subgroups. However, these sub-analyses were not sufficiently powered to complete and thus were not included in the current study.
3. Results
3.1. Baseline demographic and clinical characteristics
The combination of the 3 datasets yielded a total sample of 456 individuals. There were 31 participants excluded as they received a placebo in the infliximab trial. There were an additional 58 participants excluded as healthy controls of the vortioxetine trial. Of the remaining participants, there was insufficient data for 21 of the vortioxetine trial participants, 229 of the ketamine trial participants, and 9 of the infliximab trial participants. This resulted in a final sample of 108 individuals.
A summary of the sample sociodemographic characteristics is located in Table 2. Overall, the sample was mostly female (67%), white (69%), and single (49%), with an average BMI of 29.4 (SD = 7.7) and an age of 40 (SD = 12.5). The mean baseline MADRS score was 32.8 (SD = 7.5), corresponding to moderate to severe depressive symptoms. The mean baseline SDS work score was 7.5 out of 10 (SD = 2.80).
Clinical and Demographic Characteristics at Baseline
3.2. Mediation analysis
The mediation analysis was conducted to compare how changes in PDQ-5 total score, DSST symbols score, DSST error score, and TMT-B time mediated the association between changes in MADRS scores and changes in SDS work scores.
Changes in PDQ-5 total scores from baseline to endpoint partially mediated the association between changes in MADRS scores and changes in SDS work scores (Figure 1). Increases in MADRS scores were significantly associated with increases in PDQ-5 total scores (a = 0.32, P < .05, SE = 0.03) and PDQ-5 total scores correlated with SDS work scores (b = 0.15, P < .05, SE = 0.05). Both indirect (a × b = 0.05, 95% CI [0.01, 0.08]) and direct (c′ = 0.12, t(156) = 4.82, P < .05, 95% CI [0.07, 0.17]) effects of changes in MADRS on changes in SDS work scores were significant. Overall, the change in PDQ-5 total score was a significant partial mediator of the relationship between changes in MADRS and changes in SDS Work scores.
Direct and indirect effects (via subjective/objective cognitive measures) of depression on workplace function.
Changes in TMT-B time from baseline to endpoint did not significantly mediate the association between changes in MADRS scores and changes in SDS work scores (Figure 1B). Increases in MADRS scores were significantly associated with increases in TMT-B times (a = 0.98, P < .05, SE = 0.22), but the effect of changes in TMT-B time on SDS work scores was insignificant (b = 0.00, P > .05, SE = 0.06). Indirect (a × b = 0.00, 95% CI [−0.01, 0.01]) effects of changes in MADRS on changes in SDS work scores were insignificant. However, the direct effects of changes in MADRS on SDS work scores were significant (c′ = 0.17, t(170) = 9.25, P < .05, 95% CI [0.13, 0.20]). Overall, change in TMT-B time was not a significant mediator of the relationship between changes in MADRS and changes in SDS work scores.
Changes in DSST symbols scores from baseline to endpoint partially mediated the association between changes in MADRS scores and changes in SDS work scores (Figure 1C). Increases in MADRS scores were associated with decreases in DSST symbols scores (a = −0.40, P < .05, SE = 0.08) and increases in DSST symbols scores were associated with decreases in SDS work scores (b = −0.04, P < .05, SE = 0.15). Both indirect (a × b = 0.02, 95% CI [0.00, 0.03]) and direct effects of changes in MADRS on changes in SDS work scores were significant (c′ = 0.17, t(209) = 10.90, P < .05, 95% CI [0.14, 020]). Overall, change in PDQ-5 total scores was a significant partial mediator of the relationship between changes in MADRS and changes in SDS work scores.
Finally, changes in DSST error score from baseline to endpoint did not significantly mediate the association between changes in MADRS scores and changes in SDS work scores (Figure 1D). Increases in MADRS scores were significantly associated with an increased DSST error (a = 0.01, P < .05, SE = 0.00), but the effect of changes in DSST error scores on changes in SDS work scores was insignificant (b = −0.09 P > .05, SE = 0.26). Indirect (a × b = −0.00, 95% CI [−0.00, 0.00]) effects of changes in MADRS on changes in SDS work scores were insignificant. However, the direct effects of MADRS on SDS work scores were significant (c′ = 0.19, t(208) = 12.02, P < .00001, 95% CI [0.15, 0.22]). Overall, the change in DSST error score was not a significant mediator of the relationship between changes in MADRS and changes in SDS work scores.
4. Discussion
Our analysis further substantiates the mediational role of cognitive impairment on workplace function in persons with MDD and BD. Data were obtained from 3 studies: an open-label vortioxetine trial with patients with MDD, a double-blind placebo-controlled infliximab trial with patients with bipolar I/II depression, and a trial with outpatients with TRD receiving IV ketamine. The results of our analysis suggest that both objective cognitive function and subjective cognitive function partially mediate the relationship between depressive symptoms and workplace outcomes.
One measure of objective cognitive function, the DSST symbols score, was a partial mediator of workplace function in our analyses herein. The DSST is a polyfactorial measure of objective cognitive function that encapsulates psychomotor speed, attention, learning/memory, and executive function.Reference Tombaugh 60 , Reference Jaeger 62 The foregoing findings cohere with the literature on the role of objective cognitive function in functional outcomes.Reference Bonnín, Martínez-Arán, Torrent, Pacchiarotti, Rosa and Franco 63 –Reference McIntyre, Cha, Soczynska, Woldeyohannes, Gallaugher and Kudlow 66 Notably, a study of 182 participants with remitted MDD and healthy controls found that, of the 6 cognitive domains assessed, executive functioning was the only cognitive domain that significantly predicted workplace function.Reference Knight, Air and Baune 65 Taken together, this finding reiterates the role of objectively evaluated cognitive function and, specifically, executive function, as mediators of workplace function in mood disorder populations.
The PDQ-5 was employed as a measure of subjective cognitive function in the analysis herein and was found to be a significant mediator of workplace function in our model. The mediational role of subjective cognitive function must be differentiated from objective cognitive function as the two measures are not always in alignment. For example, an individual may perceive their cognition as impaired while objectively being cognitively intact, and vice versa.Reference Fava, Mahableshwarkar, Jacobson, Zhong, Keefe and Olsen 67 , Reference Maor, Olmer and Mozes 68 Subjective cognitive impairment was also reported to mediate workplace function in a non-clinical sample of 477 Japanese adult workers using the Cognitive Complaints in Bipolar Disorder Rating Assessment (COBRA) measure of subjective cognitive function .Reference Toyoshima, Inoue, Shimura, Masuya, Ichiki and Fujimura 23 Furthermore, a post hoc analysis of 260 patients with MDD noted that subjective cognitive function is a significant predictor of occupational outcomes.Reference McIntyre, Soczynska, Woldeyohannes, Alsuwaidan, Cha and Carvalho 22 Overall, this study substantiates the importance of recognizing the contributions of subjective cognitive function to workplace outcomes.
Two measures of objective cognitive function, the TMT-B time and DSST error scores, were not significant mediators in the current model. The TMT-B and DSST overlap in the cognitive domains they reflect (ie, executive function, psychomotor agility, and visuospatial processing). As such, it is unclear why the DSST symbols score was a significant mediator in the current model, and not the TMT-B or DSST errors score. The DSST involves a greater memory/learning component compared to the TMT-B, which may suggest a role for memory in workplace functionReference Jaeger 57. However, it is challenging to draw conclusions regarding differences between cognitive tests due to their lack of specificity to cognitive domainsReference Jaeger 57. Other possible explanations for the lack of significant mediational relationships relate to limitations to the data. For example, the DSST scores data had a small range; most participants had a score of 0 (ie, made no errors in the test), and those who did make errors did so infrequently. Regardless, further research to evaluate the specific cognitive domains implicated in the relationship between depressive symptoms and workplace outcomes is warranted.
The findings of the analysis herein must be understood within the context of several limitations. One such limitation was that SDS work scores and cognitive assessments were measured as secondary or tertiary outcomes in the ketamine and infliximab trials. Furthermore, as vortioxetine and ketamine were delivered to patients open-label, awareness of the medication may have influenced outcomes. It should also be noted that there was significant heterogeneity in the patient sample and between subgroups, which included a variety of diagnoses and degrees of disease severity. Finally, subgroup analyses of BD and MDD populations, as well as separate analyses of the contributions of individual antidepressant agents could not be completed given the lack of power. Future studies would be warranted to elucidate the differential impacts of these specific mediations and differences between MDD and BD populations. Attention may also be paid to patient populations with TRD, which demonstrate a higher incidence and severity of cognitive impairments compared to non-treatment-resistant populations.Reference Gregory, Torres, Ge, Blumberger, Downar and Daskalakis 69 Finally, the analysis herein included mechanistically different agents (ie, vortioxetine, ketamine, infliximab), which serve as both a strength and limitation. It is a strength as the results suggest a common effect across diagnoses and antidepressant agents, but also a weakness as the results may not be ascribed to any specific agent.
5. Conclusion
This study presents further evidence in support of the partial mediational role of both objective and subjective cognitive function in the relationship between depressive symptoms and workplace outcomes. Our results further instantiate the promise of improved psychosocial functioning and workplace performance in persons receiving treatments that improve cognitive functions. Further research is required to demonstrate improvement in cognitive function by functional neuroimaging based on other cognitive paradigms (eg, verbal fluency test).Reference Husain, Tang, Tam, Tran, Ho and Ho 70 , Reference Husain, Yu, Tang, Tam, Tran and Quek 71
Author contribution
The authors confirm their contribution to the paper as follows: Conceptualization and methodology: JT, DH, NR, and RSM; Data curation: JT, DH, NR, RBM, MS, and YL; Formal analysis: JT, DH, NR, and RS; Writing - original draft: JT and DH; Writing – review & editing: All authors. All authors reviewed the results and approved the final version of the manuscript.
Disclosures
DEJ receives salary support from the Canadian Institutes of Health Research (CIHR), the Canada Graduate Scholarship - Master’s (CGS-M), and the University of Toronto. JDR has received research grant support from the Canadian Cancer Society, Canadian Psychiatric Association, American Psychiatric Association, American Society of Psychopharmacology, University of Toronto, University Health Network Centre for Mental Health, Joseph M. West Family Memorial Fund, and Timeposters Fellowship and industry funding for speaker/consultation/research fees from Allergan, Lundbeck, and COMPASS. He is the medical director of a private clinic providing off-label ketamine infusions for depression. LMWL has received personal fees from Braxia Scientific Corp. RSM has received research grant support from CIHR/GACD/National Natural Science Foundation of China (NSFC) and the Milken Institute; speaker/consultation fees from Lundbeck, Janssen, Alkermes, Neumora Therapeutics, Boehringer Ingelheim, Sage, Biogen, Mitsubishi Tanabe, Purdue, Pfizer, Otsuka, Takeda, Neurocrine, Sunovion, Bausch Health, Axsome, Novo Nordisk, Kris, Sanofi, Eisai, Intra-Cellular, NewBridge Pharmaceuticals, Viatris, Abbvie, Atai Life Sciences. YL received salary support from the Global Alliance for Chronic Diseases/Canadian Institutes of Health Research (CIHR)/National Natural Science Foundation of China’s Mental Health Team Grant and the CIHR Frederick Banting and Charles Best Canada Graduate Scholarship; personal fees from Braxia Scientific Corp. BC, DEJ, DH, FC, JDV, JKT, JV, MS, NR, RBM, RH, and RS declare no competing interests.
Open access
