Trevor Dummer

Associate Professor

Relevant Thesis-Based Degree Programs


Graduate Student Supervision

Doctoral Student Supervision

Dissertations completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest dissertations.

Diabetes medication use and risk of breast, colorectal, and pancreatic cancer : a population-based cohort study (2023)

Background: Certain diabetes medications have been linked to reduced risk of breast, colorectal, and pancreatic cancers, although findings are inconclusive, reflecting limitations in the methodology of previous studies. Objective: To evaluate associations between classes, subclasses and individual diabetes medications, and risks of breast, colorectal, and pancreatic cancers. Methods: Using linked administrative health data from the Province of British Columbia, a cohort of over 3 million people aged 35-100 (1996-2019) was constructed. Diabetes medication use was parameterized as binary, continuous, and categorical variables. Duration-response and dose-response associations were determined using cumulative duration and cumulative dose (years receiving the defined daily doses). Medication use was treated as time-varying, with a 1-year lag, over the follow-up period. Cox proportional hazards models were used to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for associations between new use of diabetes medications and occurrence of breast, colorectal, and pancreatic cancers. In a subset of participants from the British Columbia Generations Project (BCGP), risk estimates were further adjusted for sociodemographic and lifestyle variables. An active comparator analysis was employed to mitigate confounding by indication. Results: Metformin (HR=1.04, 95%CI=1.04-1.05) and insulin secretagogues (HR=1.02, 95%CI=1.01-1.02), especially gliclazide and glyburide, were found to be linked to an increased risk of colorectal cancer. A higher dose of basal insulin was associated with a higher pancreatic cancer risk (HR=1.05, 95%CI=1.04-1.06). Conversely, ever use of SGLT-2 inhibitors was related to a lower breast cancer risk than never use (HR=0.64, 95%CI=0.48-0.85). Many of these associations persisted after adjusting for sociodemographic and lifestyle factors in the BCGP subset. In the active comparator analysis, similar associations were identified when compared to alternative diabetes medications. Conclusion: Multiple diabetes medications were associated with breast, colorectal, and pancreatic cancer risk, highlighting the possible need to consider long-term health implications of these commonly prescribed medications.

View record

Colorectal cancer and health behaviours: primary prevention, early detection, and improved survivorship (2021)

Background: Health behaviours play a pivotal role in the primary and secondary prevention of colorectal cancer (CRC), and are a modifiable determinant of survivorship in CRC patients. However, adherence to recommended health behaviours, including physical activity and healthy body mass index (BMI) maintenance, is sub-optimal. This dissertation contributes to the understanding of individual-level factors that influence CRC prevention-related behaviours, the impact of a CRC diagnosis on these behaviours, and how these behaviours might be improved in CRC survivors. Methods: Baseline data from the British Columbia Generations Project (BCGP) were used to examine adherence to World Cancer Research Fund (WCRF) recommendations for cancer prevention. These data were also combined with follow-up data, and cancer registry data, to examine longitudinally the impact of a diagnosis of cancer, CRC or otherwise, on a subset of these behaviours. Additionally, a systematic mixed studies review was conducted to synthesize available literature concerning behaviour change interventions in CRC survivors. Results: Overall adherence to WCRF recommendations was relatively high in the BCGP cohort, though it was markedly lower for diet-related behaviours. Some clustering of low adherence with other risk factors for CRC, such as low socioeconomic status, was identified. The longitudinal analysis found CRC was not predictive of significant change in two cancer prevention recommendations regarding alcohol consumption and BMI between baseline and follow-up. The systematic mixed studies review identified tentative evidence that behaviour change interventions can improve CRC survivor health behaviours, and that members of this group have clear preferences regarding intervention content. However, a paucity of available evidence and a tendency towards poor specification of intervention content was common. Conclusion: These results offer novel contributions regarding targets for future interventions and research to improve health behaviours in the general population and in CRC survivors. Further research is needed to ensure the design of any such interventions are grounded in theory and evidence, and informed by patient preferences and contextual considerations.

View record

The relationships between livestock and human wealth, health, and wellbeing in a rural Maasai community of southwestern Kenya (2020)

Livestock are critical to the livelihood of up to two billion global poor and thus represent an ideal focus for poverty amelioration. For traditional keepers, livestock are: culturally significant, nutritionally important, and serve as “daily currency” and household “savings”. However, they may also increase infectious disease risk, especially via zoonoses which can reduce both human and livestock health and quality of life. Although many studies exist on livestock-dependent communities, including the Maasai and other pastoralists, significant knowledge gaps persist regarding the relationships between traditional livestock-keeping and human wellbeing.This dissertation investigated associations between pastoral livestock and owner health through a series of cross-sectional and longitudinal studies conducted in Olkoroi, a rural Maasai community. The objectives were to: 1) review the literature on connections between livestock health and productivity, and human wealth, health and wellbeing; 2) describe Olkoroi sociodemography and capital; 3) assess local human and livestock disease priorities and livelihood challenges; 4) conduct longitudinal studies of livestock growth, livestock and human infectious disease; 5) measure adult psychological wellbeing; and 6) use the collected data to build predictive models of human wellbeing, herd size, livestock growth, livestock and human infectious disease frequency.I found livestock were the primary livelihood and predicted psychological wellbeing, but 40% of households, primarily female-headed, had insufficient animals to support themselves. Men and women identified similar factors affecting wellbeing but differed in proportional attribution: women uniquely spoke of restrictions on autonomy. Community disease prioritizations were similar to national priorities, however, disease management was inconsistent and causal understanding was low. Households self-rated husbandry practices highly, but felt financial constraints prevented adoption of best practice. Household variables were associated with herd size, but climate was the best predictor of livestock growth, and livestock and human infectious disease: livestock disease prevalence did not predict human disease. My results suggest livestock research must prioritize gender and local context to better understand livestock-human health relationships. Claims about the contribution of livestock to human disease burdens must also be clarified through more consistent research frameworks which allow inter-study comparisons, and more longitudinal studies to better identify causal relationships between exposures and disease incidence.

View record

Master's Student Supervision

Theses completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest theses.

Built environment and post-menopausal breast cancer risk: analysis of a linked British Columbian cohort (2021)

Background: Breast cancer is the most common type of cancer worldwide and the second leading cause of cancer deaths in women. Up to 50% of breast cancer cases are preventable, underscoring the importance of research into underlying risk factors, especially for post- menopausal women. With urbanization rates increasing in recent decades, the built environment may contain an important yet understudied set of modifiable breast cancer risk factors.Objectives: To evaluate the impact of three factors of the built environment– traffic-related air pollution (TRAP), measured using NO2, walkability, and residential greenness – on risk of breast cancer in post-menopausal women in the Lower Mainland of British Columbia (BC). Methods: This research was conducted using BC Generations Project cohort and linked CANUE environmental datasets. Descriptive statistics summarized socio-demographic, behavioural and health indicators in relation to the built environment. Cox proportional hazard regression was used to model cancer risk for three built environmental factors, while adjusting for relevant confounders. A change-in-effect model building strategy was used. Results: The study included 7,330 participants, including 122 incident breast cancer cases. The HR for a 10-ppb increase in baseline NO₂ was 1.45 (95% CI=0.90, 2.33; p=0.12), whereas the HR for NO₂ averaged over the years 1980-2012 was 1.41 (95% CI=0.95, 2.08; p = 0.09), both adjusting for body mass index and social deprivation. The walkability model had HRs adjusted for social deprivation, ranging from 1.67 to 2.53 for quintile (Q) 2 though Q5 (Q5 being most walkable), with the highest HR being for Q3 (test for trend p=0.05). The HR (unadjusted) for baseline greenness was 0.96 for a 1-interquartile range increase (p=0.76), The HR for greenness averaged over 1982-2016 adjusted for social deprivation was 0.80 (p=0.07). Conclusions: Although statistically non-significant, the magnitude and direction of TRAP HR was similar to previous studies. This study was the first study to our knowledge to assess whether walkability and greenness are associated with breast cancer risk. We found that those residing in less walkable communities; or in neighbourhoods with more greenness had lower risk of breast cancer (statistically non-significant). More research into these associations is warranted.

View record

Recent trends in cancer screening uptake amongst immigrants in Canada between 2005-2015 (2019)

INTRODUCTION: In 2016 alone, an estimated 202,400 Canadians developed cancer, and 78,800 died of the disease. In Canada, the use of reliable screening techniques for three types of cancer – breast, cervical, and colorectal – could help decrease cancer burden on a national scale. However, the literature also suggests that cancer screening uptake may be hampered in specific subpopulations, including among Canadian immigrants. The current study seeks to examine the association between immigration and breast, cervical, and colorectal cancer uptake.METHODS: This analysis uses ten years of data, 2005 – 2015, from the Canadian Community Health Survey (CCHS). Survey cycles were pooled to create an average pseudo-population and a bootstrap resampling technique was used to estimate variance. Age and sex-standardized rates were used to examine breast, cervical, and colorectal cancer screening rates between recent immigrants, long-term immigrants, and Canadian-born individuals. Multivariate logistic regression was used to evaluate the impact of immigration status on non-adherence and never screening, as well as to look at differences among immigrant subgroups and between screening years.RESULTS: Results indicate that recent immigrants (residing in Canada for 0 – 9 years) may have higher odds of never screening and non-adherence for breast cancer screening (AOR 2.15 (CI 0.89 – 5.20) and AOR 1.73 (CI 0.90 – 3.33), respectively), for cervical cancer screening (AOR 1.27 (CI 0.70 – 2.29) and AOR 1.47 (CI 0.98 – 2.21), respectively), and for colorectal cancer screening (AOR 1.75 (CI 1.11 – 2.77) and AOR 1.54 (CI 0.98 – 2.44), respectively) compared to Canadian-born individuals, although most results were not statistically significant. Importantly, several sociodemographic factors were significantly associated with never screening and non-adherence, including higher income and higher educational attainment. There was some evidence of differential uptake in immigrants by world region of birth and racial origin. The risk of non-adherence and never-screening among immigrants did not change significantly between 2005 – 2015.CONCLUSIONS: This study indicates that immigrants residing in Canada for 9 years or less are at higher risk of never screening and non-adherence to breast, cervical, and colorectal cancer screening guidelines. This study supports targeted interventions to increase preventative cancer screening use among newcomers to Canada.

View record


If this is your researcher profile you can log in to the Faculty & Staff portal to update your details and provide recruitment preferences.


Follow these steps to apply to UBC Graduate School!