Abstract
Background:
Adherence to the World Cancer Research Fund/American Institute for Cancer Research cancer prevention recommendations has been associated with reduced cancer risk; however, Canadian evidence remains limited. We examined associations between adherence to these recommendations and cancer incidence and mortality in a Canadian cohort.
Methods:
We conducted a prospective cohort analysis of 16,053 participants in the British Columbia Generations Project with no cancer diagnosis at recruitment. Participants were followed from 2009 to 2022 through linkage with the British Columbia Cancer Registry. Adherence to five behavioral factors (body weight, physical activity, fruit and vegetable intake, alcohol consumption, and smoking) was scored (1 = full, 0.5 = partial, and 0 = nonadherence) to generate a composite adherence score (range, 0–5). Multivariable Cox proportional hazards models estimated hazard ratios (HR) and 95% confidence intervals (CI) for cancer incidence and mortality.
Results:
During follow-up, 2,171 incident cancers and 516 cancer deaths were identified. Compared with nonadherence, full adherence was associated with lower overall cancer incidence (HR = 0.68; 95% CI, 0.54–0.78) and cancer mortality (HR = 0.66; 95% CI, 0.49–0.88). In sex-specific analyses, full adherence was associated with lower cancer incidence and mortality among both females and males. Full adherence was also associated with reduced breast, colorectal, and lung cancer risk. Findings were robust in sensitivity analyses excluding cancers diagnosed within the first 2 years of follow-up.
Conclusions:
Greater adherence to cancer prevention recommendations was associated with lower cancer incidence and mortality.
Impact:
These findings support integrated, multi-behavior cancer prevention strategies in Canadian populations.
Introduction
Cancer remains a major public health concern worldwide, with nearly 20 million new cases and approximately 10 million deaths reported annually (1–3). In Canada, cancer is the leading cause of death, and the number of new diagnoses continues to rise due to population aging and growth (4, 5). Although advances in early detection and treatment have improved survival, a substantial proportion of cancers are attributable to modifiable behavioral factors (6, 7). Current estimates suggest that around 30% to 50% of cancer diagnoses in Canada are linked to tobacco use, excess body weight, alcohol consumption, poor diet, and insufficient physical activity (8, 9). These findings underscore the critical importance of primary prevention strategies in reducing the cancer burden.
In response to growing evidence linking behavioral factors to cancer risk, the World Cancer Research Fund (WCRF) and the American Institute for Cancer Research (AICR) published the third evidence-based expert report for cancer prevention in 2018 (10). These recommendations emphasize maintaining a healthy body weight; engaging in regular physical activity; consuming a diet rich in whole grains, fruit, and vegetables; and limiting fast foods, other energy-dense processed foods, and alcohol. The WCRF/AICR framework promotes an integrated approach to cancer prevention, recognizing that health behaviors often cluster within individuals and may jointly influence cancer risk (11). Although smoking is not included as a scored component in the 2018 WCRF/AICR recommendations, it remains one of the most important modifiable cancer risk factors and was incorporated as an additional behavioral factor in the present analysis.
Greater adherence to cancer prevention recommendations has been associated with lower cancer incidence and mortality in several prospective cohorts (12, 13). However, evidence from Canadian populations remains limited. This evidence gap is significant because behavioral patterns, sociodemographic composition, healthcare access, and cancer screening practices differ across populations and regions, potentially influencing both exposure distributions and cancer outcomes. Therefore, generating evidence from diverse regions of the world is crucial. Generating Canadian evidence is therefore important for assessing the relevance of existing cancer prevention frameworks in this population context.
In addition, most studies have focused on either individual behaviors or composite adherence scores, with less attention to how combinations of behaviors jointly influence cancer risk. Examining joint behavioral patterns may therefore offer additional insight into how clustered risk factors collectively influence cancer incidence and mortality in real-world populations. To address these gaps, we examined the association between adherence to cancer prevention recommendations and overall and site-specific cancer incidence and mortality in a prospective cohort of Canadian adults. We also evaluated joint behavioral patterns to assess how combinations of smoking status, alcohol consumption, body weight, and physical activity were associated with cancer risk.
Materials and Methods
Study population
Data were obtained from the British Columbia Generations Project (BCGP), a population-based prospective cohort established to investigate environmental and behavioral determinants of chronic disease (14). Details of the cohort design and recruitment procedures have been described previously (15). Briefly, between 2009 and 2016, 29,179 adults 35 to 69 years of age were recruited across British Columbia, Canada. Participants completed baseline health and behavioral questionnaires and consented to linkage with provincial administrative health databases.
After excluding individuals with a prior cancer diagnosis (excluding nonmelanoma skin cancer), pregnancy at baseline, or missing baseline covariate data, 23,482 participants were eligible. Participants with incomplete adherence component or multivariable covariate data were further excluded, resulting in a final analytic cohort of 16,053 individuals (Fig. 1). Participants were followed from baseline and continued until incident cancer diagnosis, death, or December 31, 2022, whichever occurred first. All participants provided written informed consent. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the University of British Columbia Research Ethics Board (H20-02493). Baseline demographic and behavioral characteristics, including age, sex, body mass index (BMI), education, household income, physical activity, fruit and vegetable intake, alcohol consumption, and smoking status, were assessed at baseline. Participants were not randomized because this was an observational, prospective cohort study.
Figure 1.
Flow diagram of participant selection and cancer outcome ascertainment in the British Columbia Generations Project, 2009–2022.
Ascertainment of cancer outcomes
Incident cancer diagnoses and cancer-related deaths were identified through linkage with the British Columbia Cancer Registry (BCCR; http://www.bccancer.bc.ca/health-professionals/professional-resources/bc-cancer-registry), a population-based registry that captures all cancer diagnoses and cancer deaths in the province. Cancer diagnoses were classified according to the International Classification of Diseases, 10th Revision (codes C00–C99). The primary outcomes were overall cancer incidence and cancer-specific mortality. Secondary outcomes included site-specific incident cancers with sufficient case numbers for analysis.
Assessment of adherence to cancer prevention recommendations
Adherence was assessed using five self-reported behavioral factors measured at BCGP baseline: body weight, physical activity, fruit and vegetable intake, alcohol consumption, and smoking status. Body weight, physical activity, fruit and vegetable intake, and alcohol consumption were selected based on the 2018 WCRF/AICR cancer prevention recommendations. Although smoking is not included as a scored component in the 2018 WCRF/AICR recommendations, it was incorporated as an additional behavioral factor in this analysis because of its established role as a major modifiable cancer risk factor.
Each behavior was assigned a score of 1 for full adherence, 0.5 for partial adherence, or 0 for nonadherence. BMI was calculated from self-reported height and weight and categorized as healthy weight (18.5–24.9 kg/m2; score = 1), overweight (25–29.9 kg/m2; score = 0.5), or obesity (≥30 kg/m2; score = 0). Physical activity was assessed using self-reported weekly minutes of moderate-to-vigorous physical activity and categorized as ≥150 minutes/week (score = 1), 75 to 149 minutes/week (score = 0.5), or <75 minutes/week (score = 0). Fruit and vegetable intake was assessed using self-reported daily servings and categorized as ≥5 servings/day (score = 1), 3 to 4 servings/day (score = 0.5), or ≤2 servings/day (score = 0). Alcohol consumption was assessed using self-reported average daily intake over the past 12 months and categorized as no alcohol use (score = 1), moderate intake (>0 to ≤2 drinks/day for males or >0 to ≤1 drink/day for females; score = 0.5), and heavy intake (≥3 drinks/day for males or ≥2 drinks/day for females; score = 0). Smoking status was classified as never-smoker (score = 1), former smoker (quit ≥6 months prior to baseline; score = 0.5), or current smoker (score = 0). The operationalization of each behavioral component and adherence scoring criteria is shown in Table 1.
Table 1.
Operationalization of WCRF/AICR cancer prevention recommendations and adherence scoring system.
| Modifiable behavioral factor | WCRF/AICR recommendation | Operationalization | Full adherence (score = 1) | Partial adherence (score = 0.5) | Nonadherence (score = 0) |
|---|---|---|---|---|---|
| Body weight | Maintain a healthy body weight throughout life | BMI (kg/m2) | Healthy weight (18.5–24.9) | Overweight (25–29.9) | Obesity (≥30) |
| Physical activity | Be physically active as part of everyday life | Weekly minutes of moderate-to-vigorous physical activity | ≥150 minutes/week | 75–149 minutes/week | <75 minutes/week |
| Fruit and vegetable intake | Eat more fruit and vegetables | Servings per day | ≥5 servings/day | 3–4 servings/day | ≤2 servings/day |
| Alcohol consumption | Avoid alcohol consumption | Average daily alcohol intake over the past 12 months | No alcohol use | 0 to ≤2 drinks/day for males or 0 to ≤1 drink/day for females | ≥3 drinks/day for males or ≥2 drinks/day for females |
| Smoking status | Live smoke-freea | Smoking status | Never-smoker | Former smokerb | Current smoker |
NOTE: The total adherence score ranged from 0 to 5, with higher scores indicating greater adherence to the behavioral factors included in this analysis.
Smoking status was included as an additional behavioral factor in this analysis because of its established role as a major modifiable cancer risk factor; it is not a scored component of the 2018 WCRF/AICR cancer prevention recommendations.
Former smoker defined as cessation ≥6 months before baseline.
Component scores were summed to generate a composite adherence score ranging from 0 to 5, with higher scores indicating greater adherence. For primary analyses, the composite score was categorized into three groups: nonadherence, partial adherence, and full adherence. In secondary analyses, the score was modeled as a continuous variable to estimate hazard ratios (HR) per one-point increase. Scoring methodology and cutoffs were adapted from previously published WCRF/AICR adherence scoring approaches with established content validity through alignment between score components and the underlying cancer prevention recommendations (10).
Covariates
Covariates were selected a priori based on known or plausible associations with behavioral factors and cancer outcomes. These included age (continuous), sex, marital status, education level, household income, and first-degree family history of cancer (16). Sex was excluded from sex-specific cancer models where appropriate. Smoking was not included as a covariate in the primary composite score models because it was incorporated into the exposure definition. In sensitivity analyses excluding smoking from the composite score, smoking status was included as a covariate.
Statistical analysis
Baseline demographics and behavioral characteristics were summarized across adherence categories using means (standard deviations) for continuous variables and counts (percentages) for categorical variables. Differences across adherence categories were assessed using analysis of variance for continuous variables and χ2 tests for categorical variables.
Cox proportional hazards regression models were used to estimate HRs and 95% confidence intervals (CI) for the association between adherence and (i) overall cancer incidence, (ii) cancer mortality, and (iii) site-specific cancer incidence. Attained age was used as the time scale, with entry defined as age at baseline and exit defined as age at event or censoring. Multivariable models were adjusted for age, sex where applicable, marital status, education level, household income, and first-degree family history of cancer. Tests for trend were conducted by modeling adherence category as an ordinal variable.
To assess the independent contribution of individual behavioral factors, models including all five components simultaneously were fitted. Several prespecified sensitivity analyses were conducted. First, cancers diagnosed within the first 2 years of follow-up were excluded to assess potential reverse causation. Second, a modified composite score excluding smoking was constructed, with smoking status included separately as a covariate. Third, analyses were restricted to never-smokers. Complete-case analysis was used for multivariable models. No formal a priori power calculation was conducted because the study included all eligible participants available within the cohort. The available sample size was considered sufficient to detect clinically meaningful associations between adherence and cancer outcomes.
To evaluate joint behavioral patterns, combinations of smoking status, alcohol consumption, BMI, and physical activity were examined using their original categorical levels to assess dose–response relationships. These behaviors were selected based on a previously developed Cancer Risk Index designed to represent major cancer risk factors and because they had sufficient case numbers for stable estimation (17). The reference group consisted of participants who were fully adherent to all four behavioral factors (never-smoker, no alcohol use, healthy BMI, and ≥150 minutes/week of physical activity). Bonferroni-adjusted significance thresholds were applied for multiple comparisons in the joint behavioral analyses (18).
The proportional hazards assumption was evaluated using Schoenfeld residuals and graphical inspection of log-minus-log survival plots, and no meaningful violations were observed. Outcome ascertainment through linkage with the BCCR was conducted independently of baseline exposure assessment. Investigators performing statistical analyses were blinded to participant identifiers and used coded data. All analyses were conducted using R version 4.4.0 (R Foundation for Statistical Computing; https://www.r-project.org/).
Results
Participant characteristics
Baseline demographic and behavioral characteristics across adherence categories are presented in Table 2. Participants with higher adherence to WCRF/AICR recommendations were more likely to be female, have higher educational attainment, and report higher household income. As expected, distributions of BMI, physical activity, fruit and vegetable intake, alcohol consumption, and smoking status varied across adherence categories. Full adherence to all five behavioral factors was observed in 39% of females and 23% of males.
Table 2.
Baseline characteristics of BCGP participants by adherence category.
| Characteristic | Nonadherence (n = 1,897) |
Partial adherence (n = 8,657) |
Full adherence (n = 5,499) |
P value |
|---|---|---|---|---|
| Age, mean (SD), years | 57.6 (8.3) | 56.5 (8.5) | 54.6 (8.9) | <0.001 |
| Sex, n (%) | | | | <0.001 |
| Female | 960 (50.6) | 5,792 (66.9) | 4,374 (79.5) | |
| Male | 937 (49.4) | 2,865 (33.1) | 1,125 (20.5) | |
| Education, n (%) | | | | <0.001 |
| High school or less | 551 (29) | 1,731 (20) | 803 (14.6) | |
| College or trade | 693 (36.5) | 2,815 (32.5) | 1,529 (27.8) | |
| University degree or higher | 653 (34.5) | 4,111 (47.5) | 3,167 (57.6) | |
| Household income, n (%) | | | | <0.001 |
| <$50,000 | 542 (28.6) | 1,844 (21.3) | 1,121 (20.4) | |
| $50,000–149,999 | 1,106 (58.3) | 5,385 (62.2) | 3,437 (62.5) | |
| ≥$150,000 | 249 (13.1) | 1,428 (16.5) | 941 (17.1) | |
| Marital status, n (%) | | | | 0.042 |
| Single/divorced | 474 (25) | 1,818 (21) | 1,045 (19) | |
| Married/living with partner | 1,423 (75) | 6,839 (79) | 4,454 (81) | |
| First-degree family history of cancer, n (%) | | | | <0.001 |
| Yes | 456 (24) | 2,424 (28) | 1,705 (31) | |
| No | 1,441 (76) | 6,233 (72) | 3,794 (69) | |
| BMI (kg/m2), n (%) | | | | |
| Healthy weight (18.5–24.9) | 228 (12) | 3,027 (35) | 3,979 (72.4) | |
| Overweight (25–29.9) | 684 (36.1) | 3,680 (42.5) | 1,342 (24.4) | |
| Obesity (≥30) | 985 (51.9) | 1,950 (22.5) | 178 (3.2) | |
| Physical activity (minutes/week), n (%) | | | | |
| <75 | 908 (47.9) | 1,325 (15.3) | 99 (1.8) | |
| 75–149 | 725 (38.2) | 3,401 (39.3) | 1,166 (21.2) | |
| ≥150 | 264 (13.9) | 3,931 (45.4) | 4,234 (77) | |
| Alcohol consumption, n (%) | | | | |
| No alcohol use | 328 (17.3) | 2,702 (31.2) | 3,328 (60.5) | |
| 0 to ≤2 drinks/day (males) or 0 to ≤1 drink/day (females) | 516 (27.2) | 2,926 (33.8) | 1,705 (31) | |
| ≥3 drinks/day (males) or ≥2 drinks/day (females) | 1,053 (55.5) | 3,029 (35) | 466 (8.5) | |
| Fruit and vegetables (servings/day), n (%) | | | | |
| ≤2 servings/day | 917 (48.3) | 1,057 (12.2) | 55 (1) | |
| 3–4 servings/day | 698 (36.8) | 3,040 (35.1) | 820 (14.9) | |
| ≥5 servings/day | 282 (14.9) | 4,560 (52.7) | 4,624 (84.1) | |
| Smoking status, n (%) | | | | |
| Never-smoker | 315 (16.6) | 3,860 (44.6) | 4,321 (78.6) | |
| Former smokera | 1,157 (61) | 4,389 (50.7) | 1,144 (20.8) | |
| Current smoker | 425 (22.4) | 408 (4.7) | 34 (0.6) | |
NOTE: Values are presented as n (%) unless otherwise indicated. P values were calculated using analysis of variance for age and χ2 tests for categorical variables. P values are presented for baseline covariates only and not for components of the adherence score.
Former smoker defined as cessation ≥6 months prior to baseline.
Adherence and overall cancer incidence and mortality
During follow-up, 2,171 incident cancer cases and 516 cancer deaths were identified. In multivariable-adjusted Cox proportional hazards models, greater adherence to cancer prevention recommendations was associated with lower overall cancer incidence and cancer mortality (Table 3). Compared with nonadherence, full adherence was associated with a 32% lower hazard of cancer incidence (HR, 0.68; 95% CI, 0.54–0.78), whereas partial adherence was associated with a 22% lower hazard of cancer incidence (HR, 0.78; 95% CI, 0.64–0.95). A significant dose–response relationship was observed across adherence categories (P trend = 0.001). When modeled continuously, each one-point increase in adherence score was associated with an 11% lower hazard of cancer incidence (HR = 0.89; 95% CI, 0.81–0.98).
Table 3.
Adjusted HRs and 95% CIs for associations between adherence to cancer prevention recommendations and overall cancer incidence and mortality.
| Outcome | Cases (n) | Full adherence | Partial adherence | Nonadherence (Ref) | P trend |
|---|---|---|---|---|---|
| Overall cancer incidence | | | | | |
| Females | 1,389 | 0.67 (0.50–0.91) | 0.74 (0.56–0.98) | 1 | 0.010 |
| Males | 782 | 0.66 (0.45–0.97) | 0.76 (0.64–0.94) | 1 | 0.002 |
| Total | 2,171 | 0.68 (0.54–0.78) | 0.78 (0.64–0.95) | 1 | 0.001 |
| Cancer mortality | | | | | |
| Females | 330 | 0.57 (0.37–0.88) | 0.69 (0.46–0.81) | 1 | 0.005 |
| Males | 186 | 0.77 (0.51–0.85) | 0.81 (0.63–0.93) | 1 | 0.003 |
| Total | 516 | 0.66 (0.49–0.88) | 0.72 (0.56–0.89) | 1 | 0.001 |
NOTE: HRs were estimated using Cox proportional hazards models and adjusted for age (continuous, years), sex (where applicable), marital status (living with partner vs. without partner), education level (high school or less, college/trade, and university), annual household income (<$50,000; $50,000–149,999; and ≥$150,000), and first-degree family history of cancer (no/yes). Smoking status was not included as a covariate because it was incorporated into the composite adherence score. P trend was calculated by modeling adherence category as an ordinal variable.
Abbreviation: Ref, reference category.
Sex-stratified analyses showed similar inverse associations among females and males. Among females, full adherence was associated with lower overall cancer incidence (HR = 0.67; 95% CI, 0.50–0.91), whereas partial adherence was also associated with lower incidence (HR = 0.74; 95% CI, 0.56–0.98). Among males, full adherence was associated with lower overall cancer incidence (HR = 0.66; 95% CI, 0.45–0.97), as was partial adherence (HR = 0.76; 95% CI, 0.64–0.94).
A similar inverse association was observed for cancer mortality. Compared with nonadherence, full adherence was associated with a 34% lower hazard of cancer death (HR = 0.66; 95% CI, 0.49–0.88; P trend = 0.001), and partial adherence was associated with a 28% lower hazard of cancer death (HR = 0.72; 95% CI, 0.56–0.89). The inverse association was observed among both females and males. The results were consistent when adherence was modeled as a continuous score.
Site-specific cancer outcomes
In site-specific analyses, greater adherence to cancer prevention recommendations was associated with reduced incidence of several major cancers (Table 4). Compared with nonadherence, full adherence was associated with lower risk of breast cancer (HR, 0.79; 95% CI, 0.61–0.96), colorectal cancer (HR, 0.59; 95% CI, 0.48–0.87), lung and bronchus cancer (HR, 0.72; 95% CI, 0.60–0.85), and non–Hodgkin lymphoma (HR, 0.74; 95% CI, 0.56–0.89). No clear dose–response associations were observed for prostate, uterine, bladder, pancreatic, or kidney cancer. A modest inverse association was also observed for all other cancers combined (HR, 0.73; 95% CI, 0.47–0.92).
Table 4.
Adjusted HRs (95% CIs) for the association between adherence to cancer prevention recommendations and site-specific incident cancers.
| Cancer site/type | Cases (n) | Full adherence | Partial adherence | Nonadherence (Ref) | P trend |
|---|---|---|---|---|---|
| Breast | 467 | 0.79 (0.61–0.96) | 0.91 (0.47–1.33) | 1 | 0.005 |
| Colorectal | 159 | 0.59 (0.48–0.87) | 0.80 (0.68–0.94) | 1 | 0.005 |
| Lung and bronchus | 161 | 0.72 (0.60–0.85) | 0.84 (0.60–0.97) | 1 | 0.001 |
| Non–Hodgkin lymphoma | 97 | 0.74 (0.56–0.89) | 0.86 (0.61–1.03) | 1 | 0.020 |
| Prostate | 231 | 0.93 (0.65–1.10) | 1.02 (0.53–1.65) | 1 | 0.739 |
| Uterus (body) | 91 | 0.56 (0.21–1.51) | 0.71 (0.29–1.75) | 1 | 0.245 |
| Bladder | 83 | 0.85 (0.54–0.99) | 0.95 (0.61–1.08) | 1 | 0.210 |
| Pancreas | 58 | 0.62 (0.49–1.02) | 0.91 (0.33–2.46) | 1 | 0.366 |
| Kidney | 49 | 0.79 (0.52–1.30) | 0.97 (0.81–1.37) | 1 | 0.734 |
| All other cancers combineda | 775 | 0.73 (0.47–0.92) | 0.80 (0.63–1.30) | 1 | 0.001 |
NOTE: HRs were estimated using Cox proportional hazards models and adjusted for age (continuous, years), sex (where applicable), marital status (living with partner vs without partner), education level (high school or less, college/trade, and university), annual household income (<$50,000; $50,000–149,999; and ≥$150,000), and first-degree family history of cancer (no/yes). Smoking status was not included as a covariate because it was incorporated into the composite adherence score. P trend was calculated by modeling adherence category as an ordinal variable (nonadherence = 0, partial adherence = 1, and full adherence = 2).
Abbreviation: Ref, reference category.
All other cancers include cancers of the brain, thyroid, cervix, esophagus, head and neck (larynx and oral cavity), Hodgkin lymphoma, leukemia, liver, melanoma, multiple myeloma, stomach, and testis.
Component-level and sensitivity analysis
The results were robust across multiple sensitivity analyses (Table 5). Excluding cancers diagnosed within the first 2 years of follow-up did not materially alter the findings; full adherence remained significantly associated with lower cancer incidence (HR = 0.68; 95% CI, 0.52–0.88). When smoking was excluded from the composite score and included separately as a covariate, full adherence remained significantly associated with reduced cancer incidence (HR = 0.78; 95% CI, 0.63–0.95). Current smoking was independently associated with increased cancer risk. In analyses restricted to never-smokers, associations were directionally consistent but attenuated and did not reach statistical significance, likely due to reduced sample size and fewer events.
Table 5.
Sensitivity analyses for the association between adherence to cancer prevention recommendations and overall cancer incidence.
| Sensitivity analysis | n | Events | Partial adherence HR (95% CI) | Full adherence HR (95% CI) | HR per 1-point increase (95% CI) |
|---|---|---|---|---|---|
| Excluding the first 2 years of follow-up | 12,752 | 1,977 | 0.77 (0.62–0.97) | 0.68 (0.52–0.88) | 0.89 (0.81–0.98) |
| Composite score excluding smokinga | 12,928 | 1,753 | 0.90 (0.74–1.09) | 0.78 (0.63–0.95) | 0.90 (0.82–0.99) |
| Restricted to never-smokers | 6,830 | 734 | 0.74 (0.46–1.18) | 0.65 (0.41–1.04) | 0.88 (0.77–1.01) |
NOTE: HRs were estimated using Cox proportional hazards models and adjusted for age (continuous, years), sex (where applicable), marital status (living with partner vs. without partner), education level (high school or less, college/trade, and university), annual household income (<$50,000; $50,000–149,999; and ≥$150,000), and first-degree family history of cancer (no/yes). Participants diagnosed with cancer within the first 2 years of follow-up were excluded from the reverse causation analysis..
Smoking status was not included as a covariate in the primary composite model because it was incorporated into the composite adherence score. In the composite score excluding smoking analysis, smoking status was included as a covariate.
In mutually adjusted models that included all five behavioral components simultaneously, smoking status and BMI demonstrated the strongest independent associations with overall cancer incidence. Lower physical activity and lower fruit and vegetable intake showed modest positive associations with cancer incidence, whereas alcohol consumption showed attenuated associations after adjustment for the other behavioral factors (Supplementary Table S1).
Behavioral pattern analysis
Joint behavioral analyses showed substantial heterogeneity in cancer risk across combinations of smoking status, alcohol consumption, BMI, and physical activity. For cancer incidence, higher risks were generally observed among participants with multiple behaviors that did not meet recommendations. Individuals with three or more behaviors not meeting recommendations had markedly higher cancer incidence and mortality risks (Supplementary Figs. S1 and S2). The highest risks were observed among participants who were current smokers and frequently consumed alcohol, particularly when combined with overweight or obesity (HRs up to 2.43). In contrast, participants who were never-smokers, physically active, had a healthy BMI, and consumed little or no alcohol generally had the lowest observed cancer risks (HRs ranging from 0.69–0.88).
Cancer mortality followed similar patterns. The highest mortality risks were observed among participants with obesity who were current smokers and consumed alcohol frequently (HRs up to 5.31), whereas the lowest risks were generally observed among participants with a healthy BMI who were never-smokers and were physically active and consumed little or no alcohol (HRs ranging from 0.31–0.44; Supplementary Fig. S3). After Bonferroni correction for multiple comparisons, associations involving current smoking remained statistically significant, whereas most other behavioral combinations did not.
Discussion
In this prospective cohort of Canadian adults, greater adherence to cancer prevention recommendations was associated with lower overall cancer incidence and cancer mortality. Participants who were fully adherent to five behavioral recommendation factors experienced approximately one-third lower cancer incidence and mortality compared with those with nonadherence. A graded inverse association was observed across adherence categories, and continuous modeling demonstrated that each one-point increase in adherence score was associated with reduced cancer risk. These findings suggest that incremental improvements in adherence may contribute to progressively lower cancer risk. They also support integrated behavioral prevention strategies that address multiple modifiable factors together rather than in isolation.
The magnitude of association observed in this study is consistent with previous prospective cohort studies from other countries that have reported lower cancer incidence and mortality among individuals with greater adherence to cancer prevention recommendations (19, 20). The present study extends this evidence to a Canadian population–based cohort, where data remain limited. This is important because behavioral patterns, sociodemographic composition, healthcare access, and cancer screening practices may differ across populations and could influence both exposure distributions and cancer outcomes. Although the effect sizes observed in this study were modest, particularly when comparing full adherence with nonadherence, they remain meaningful from a population health perspective. Cancer has a multifactorial etiology, and behavioral factors represent only one component of overall cancer risk. Therefore, even modest reductions in risk associated with behavioral adherence may translate into important public health benefits when applied across populations.
Significant inverse associations were observed for several site-specific cancers, including breast cancer, colorectal cancer, lung and bronchus cancer, and non–Hodgkin lymphoma. These findings are broadly consistent with evidence linking body weight, physical activity, diet, alcohol consumption, and smoking to cancer risk. The association with colorectal cancer is biologically plausible given the established roles of adiposity, physical inactivity, alcohol intake, and diet in colorectal carcinogenesis (12). The inverse association observed for lung cancer was expected, given the strong contribution of smoking to lung cancer risk, although sensitivity analyses excluding smoking from the composite score suggested that the overall association was not entirely driven by smoking.
The association with breast cancer aligns with prior prospective studies reporting lower breast cancer risk among women with greater adherence to cancer prevention recommendations (21, 22). However, the findings should be interpreted cautiously because menopausal status was not available in the analytic dataset. This is important because the association between body weight and breast cancer risk may differ before and after menopause. As a result, the breast cancer estimates, including the joint behavioral pattern estimates, should be interpreted as overall associations among women in the cohort rather than menopause-specific associations. A study by Kabat and colleagues (12) and meta-analyses by Kohler and colleagues (23) reported similar site-specific reductions among individuals with higher adherence to cancer prevention recommendations.
In contrast, no statistically significant associations were observed for prostate, pancreatic, kidney, bladder, or uterine cancer, consistent with previous cohort studies reporting limited or null associations for these cancer sites (22–25). These null findings may reflect differences in cancer etiology, weaker associations between the behavioral factors assessed and some cancer sites, or limited statistical power for less common cancers. Although the overall analytic cohort included 16,053 participants, the number of cases for some site-specific cancers was relatively small, limiting precision in subgroup and site-specific analyses. Longer follow-up may also be needed to observe the full impact of behavioral adherence on cancers with long latency periods or lower incidence in the cohort. Overall, the heterogeneity observed across cancer sites likely reflects differences in etiology, hormonal influences, statistical power, and the relative contribution of modifiable behavioral factors.
The inverse association observed for cancer mortality suggests that greater adherence to behavioral recommendations was associated not only with lower cancer incidence but also with lower cancer-related death. Prior prospective studies have similarly reported reductions in cancer-specific mortality among individuals with higher adherence scores. In the present study, excluding cancers diagnosed within the first 2 years of follow-up and using alternative modeling strategies did not materially alter the mortality associations, supporting the robustness of the findings. However, because behaviors were assessed only at baseline, potential behavioral changes during follow-up or following cancer diagnosis were not captured.
Behavioral risk factors frequently cluster within individuals, and evaluating these jointly may better reflect real-world exposure patterns. The joint behavioral analyses provided additional insight into how combinations of behavioral factors were associated with cancer risk. Higher cancer incidence and mortality were generally observed among participants with multiple behaviors that did not meet recommendations, particularly combinations involving current smoking and alcohol consumption. These findings are consistent with the possibility that smoking and alcohol may act together or cluster with other behavioral risk factors to increase cancer risk. Although the analysis was not designed to formally test biological interaction, the observed patterns highlight the relevance of addressing co-occurring behavioral risks. From a prevention perspective, this supports comprehensive approaches that consider smoking, alcohol consumption, body weight, and physical activity together rather than as isolated targets.
The findings also reinforce the need to consider broader social and environmental determinants of cancer prevention behaviors. Factors such as social connectedness, marital status, socioeconomic position, and neighborhood environments may shape opportunities for physical activity, dietary patterns, alcohol use, smoking behaviors, and access to preventive health services. Although the present analysis adjusted for selected sociodemographic factors, including marital status, education, and household income, it did not directly examine social connectedness or neighborhood-level characteristics. Future studies incorporating built environment, social support, and area-level deprivation measures may provide a more complete understanding of how behavioral cancer prevention recommendations operate within real-world social contexts.
Strengths and limitations
This study has several strengths, including its prospective design, use of a population-based Canadian cohort, comprehensive assessment of multiple behavioral factors, and linkage to the BCCR, supporting robust outcome ascertainment. The use of attained age as the time scale and adjustment for relevant sociodemographic covariates strengthens internal validity. Examination of both composite adherence scores and joint behavioral pattern analyses provides a more detailed assessment of behavioral clustering than single-factor analyses and allows us to assess overall adherence as well as combinations of behaviors. Sensitivity analyses excluding cancers diagnosed within the first 2 years of follow-up, excluding smoking from the composite score, and restricting analyses to never-smokers supported the robustness of the main findings.
Despite the strengths, the study has several limitations that should be considered. The behavioral factors were self-reported at baseline and may be subject to measurement error and misclassification. In addition, behaviors were assessed only at recruitment, and changes in body weight, physical activity, fruit and vegetable intake, alcohol consumption, or smoking status during follow-up were not captured. This may have attenuated observed associations and limited our ability to assess the impact of behavioral changes over time.
The dietary component was operationalized using available fruit and vegetable intake data and did not capture the full scope of the WCRF/AICR dietary recommendations, which include broader guidance on whole grains, red and processed meat, sugar-sweetened beverages, and ultra-processed foods. Smoking status was also included as an additional behavioral factor in this analysis because of its established role as a major modifiable cancer risk factor, although it is not a scored component of the 2018 WCRF/AICR recommendations.
Breast cancer findings should be interpreted cautiously because menopausal status was not available, and analyses could not be stratified by premenopausal and postmenopausal status. Given that associations between body weight and breast cancer risk may differ by menopausal status, the breast cancer estimates, including the joint behavioral pattern estimates, should be interpreted as overall associations among women in the cohort rather than menopause-specific associations.
Statistical power was also limited for some site-specific cancers and for joint behavioral combinations, particularly in cancer mortality analyses, in which some estimates were based on small numbers of deaths. These estimates should therefore be interpreted cautiously. Finally, residual confounding from unmeasured or imperfectly measured factors, including social connectedness and neighborhood-level characteristics, cannot be excluded.
Implications and conclusion
Overall, these findings reinforce the evidence that greater adherence to cancer prevention behavioral recommendations is associated with lower cancer incidence and mortality. Importantly, this study extends the literature by demonstrating that combinations of behavioral factors are associated with meaningful differences in cancer risk, supporting prevention strategies that address multiple behaviors together rather than in isolation. Given that a relatively small proportion of participants were fully adherent, substantial opportunities remain for population-level cancer prevention through supportive public health policies and interventions that address multiple behavioral factors together.
In conclusion, greater adherence to cancer prevention behavioral recommendations was associated with lower cancer incidence and mortality in this Canadian cohort. Future research with repeated behavioral measures, longer follow-up, menopause-specific breast cancer analyses, and broader social and neighborhood-level data would help clarify how behavioral and contextual factors jointly shape cancer risk.
Supplementary Material
Supplementary Table S1 shows mutually adjusted associations between individual behavioural components and overall cancer incidence
Supplementary Figure S1 shows hazard ratios (HRs) of site-specific cancer incidence according to combinations of smoking status, alcohol consumption, body mass index, and physical activity. Panels show: A, lung cancer incidence; B, breast cancer incidence; C, colorectal cancer incidence; D, prostate cancer incidence. Exposure categories were defined as described in Table 1. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, healthy BMI, and ≥150 minutes/week of physical activity. All HRs were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at a Bonferroni-adjusted level (α = 0.000617). NE, not estimable due to insufficient case numbers. Estimates based on small case numbers should be interpreted cautiously.
Supplementary Figure S2 shows hazard ratios of overall cancer incidence according to behavioural factor combinations. Behavioural combinations were defined by smoking status, alcohol consumption, body mass index, and physical activity. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, had healthy BMI, and reported ≥150 minutes/week of physical activity. Hazard ratios were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at the Bonferroni-adjusted level. HR, hazard ratio; BMI, body mass index; NE, not estimable due to insufficient case numbers.
Supplementary Figure S3. Hazard ratios of overall cancer mortality according to behavioural factor combinations. Behavioural combinations were defined by smoking status, alcohol consumption, body mass index, and physical activity. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, had healthy BMI, and reported ≥150 minutes/week of physical activity. Hazard ratios were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at the Bonferroni-adjusted level. HR, hazard ratio; BMI, body mass index; NE, not estimable due to insufficient case numbers. Estimates based on small numbers of cancer deaths should be interpreted cautiously.
Acknowledgments
This study was funded by the Canadian Institutes of Health Research (funding reference #170248), supporting T.J.B. Dummer. The data used in this research were made available by the BCGP. The BCGP was created with financial support from the Canadian Partnership Against Cancer and by Health Canada and the BC Cancer Foundation. The views expressed herein represent the views of the authors and do not necessarily represent the views of Health Canada. T.J.B. Dummer is the Canadian Cancer Society Chair in Cancer Primary Prevention. The authors acknowledge the use of ChatGPT, a generative Artificial Intelligence tool, to assist with refining sentence structure, improving clarity, and ensuring conciseness in the manuscript. All intellectual content, study design, analyses, interpretation, and conclusions remain the responsibility of the authors.
Footnotes
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).
Data Availability
The data used in this study are available through application to the BCGP. Further information on data access is available at https://bcgpresearch.ca or by contacting bcgpaccess@bccrc.ca. Analytic code is available from the corresponding author upon reasonable request. Access is subject to approval by the BCGP and applicable data governance requirements.
Authors’ Disclosures
R.A. Murphy reports personal fees from Pharmavite outside the submitted work. T.J.B. Dummer reports grants from the Canadian Institutes of Health Research during the conduct of the study. No disclosures were reported by the other authors.
Authors’ Contributions
S.S. Fiati Kenston: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft. R.A. Murphy: Conceptualization, methodology, writing–review and editing. J.M. Spiegel: Data curation, validation, investigation, writing–review and editing. T.J.B. Dummer: Conceptualization, data curation, supervision, writing–review and editing.
References
- 1. Miranda-Filho A, Bray F, Charvat H, Rajaraman S, Soerjomataram I. The world cancer patient population (WCPP): an updated standard for international comparisons of population-based survival. Cancer Epidemiol 2020;69:101802. 10.1016/j.canep.2020.101802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. World Health Organization . Global cancer burden growing, amidst mounting need for services. 2024[cited 2024 May 27]. Available from:https://www.who.int/news/item/01-02-2024-global-cancer-burden-growing--amidst-mounting-need-for-services. [PMC free article] [PubMed]
- 3. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209–49. 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 4. Brenner DR, Poirier AE, Walter SD, King WD, Franco EL, Demers PA, et al. Estimating the current and future cancer burden in Canada: methodological framework of the Canadian population attributable risk of cancer (ComPARe) study. BMJ Open 2018;8:e022378. 10.1136/bmjopen-2018-022378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Poirier AE, Ruan Y, Walter SD, Franco EL, Villeneuve PJ, King WD, et al. The future burden of cancer in Canada: long-term cancer incidence projections 2013–2042. Cancer Epidemiol 2019;59:199–207. 10.1016/j.canep.2019.02.011. [DOI] [PubMed] [Google Scholar]
- 6. Brenner DR, Weir HK, Demers AA, Ellison LF, Louzado C, Shaw A, et al. Projected estimates of cancer in Canada in 2020. CMAJ 2020;192:E199–205. 10.1503/cmaj.191292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Poirier AE, Ruan Y, Volesky KD, King WD, O’Sullivan DE, Gogna P, et al. The current and future burden of cancer attributable to modifiable risk factors in Canada: summary of results. Prev Med 2019;122:140–7. 10.1016/j.ypmed.2019.04.007. [DOI] [PubMed] [Google Scholar]
- 8. Pader J, Ruan Y, Poirier AE, Asakawa K, Lu C, Memon S, et al. Estimates of future cancer mortality attributable to modifiable risk factors in Canada. Can J Public Health 2021;112:1069–82. 10.17269/s41997-020-00455-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Whelan HK, Xu JY, Vaseghi S, Lo Siou G, McGregor SE, Robson PJ. Alberta’s Tomorrow Project: adherence to cancer prevention recommendations pertaining to diet, physical activity and body size. Public Health Nutr 2017;20:1143–53. 10.1017/S1368980016003451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Shams-White MM, Brockton NT, Mitrou P, Kahle LL, Reedy J. The 2018 World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) score and all-cause, cancer, and cardiovascular disease mortality risk: a longitudinal analysis in the NIH-AARP diet and health study. Curr Dev Nutr 2022;6:nzac096. 10.1093/cdn/nzac096. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Vineis P, Wild CP. Global cancer patterns: causes and prevention. Lancet 2014;383:549–57. 10.1016/S0140-6736(13)62224-2. [DOI] [PubMed] [Google Scholar]
- 12. Kabat GC, Matthews CE, Kamensky V, Hollenbeck AR, Rohan TE. Adherence to cancer prevention guidelines and cancer incidence, cancer mortality, and total mortality: a prospective cohort study. Am J Clin Nutr 2015;101:558–69. 10.3945/ajcn.114.094854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Jung AY, Miljkovic I, Rubin S, Kritchevsky SB, Klepin HD, Newman AB, et al. Adherence to cancer prevention guidelines among older white and black adults in the health ABC study. Nutrients 2019;11:1008. 10.3390/nu11051008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Zanif U, Chu J, Simkin J, Dummer T, Woods R, Belanger E, et al. The BC Generations Project as a tumor tissue resource for cancer research. Curr Oncol 2022;29:1262–8. 10.3390/curroncol29020107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Dhalla A, McDonald TE, Gallagher RP, Spinelli JJ, Brooks-Wilson AR, Lee TK, et al. Cohort profile: the British Columbia Generations Project (BCGP). Int J Epidemiol 2019;48:377–8k. 10.1093/ije/dyy160. [DOI] [PubMed] [Google Scholar]
- 16. Walter S, Tiemeier H. Variable selection: current practice in epidemiological studies. Eur J Epidemiol 2009;24:733–6. 10.1007/s10654-009-9411-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Colditz GA, Atwood KA, Emmons K, Monson RR, Willett WC, Trichopoulos D, et al. Harvard report on cancer prevention volume 4: Harvard Cancer Risk Index. Risk Index Working Group, Harvard Center for Cancer Prevention. Cancer Causes Control 2000;11:477–88. 10.1023/a:1008984432272. [DOI] [PubMed] [Google Scholar]
- 18. Armstrong RA. When to use the Bonferroni correction. Ophthalmic Physiol Opt 2014;34:502–8. 10.1111/opo.12131. [DOI] [PubMed] [Google Scholar]
- 19. Pichardo MS, Esserman D, Ferrucci LM, Molina Y, Chlebowski RT, Pan K, et al. Adherence to the American Cancer Society Guidelines on nutrition and physical activity for cancer prevention and obesity-related cancer risk and mortality in Black and Latina Women’s Health Initiative participants. Cancer 2022;128:3630–40. 10.1002/cncr.34428. [DOI] [PubMed] [Google Scholar]
- 20. Jankovic N, Geelen A, Winkels RM, Mwungura B, Fedirko V, Jenab M, et al. Adherence to the WCRF/AICR dietary recommendations for cancer prevention and risk of cancer in elderly from Europe and the United States: a meta-analysis within the CHANCES project. Cancer Epidemiol Biomarkers Prev 2017;26:136–44. 10.1158/1055-9965.EPI-16-0428. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Catsburg C, Miller AB, Rohan TE. Adherence to cancer prevention guidelines and risk of breast cancer. Int J Cancer 2014;135:2444–52. 10.1002/ijc.28887. [DOI] [PubMed] [Google Scholar]
- 22. Oh J-K, Han M, Kim B, Park EY. Adherence to cancer prevention guidelines and cancer incidence and mortality: a population-based cohort study. Cancer Res Treat 2023;55:15–27. 10.4143/crt.2021.1031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Kohler LN, Garcia DO, Harris RB, Oren E, Roe DJ, Jacobs ET. Adherence to diet and physical activity cancer prevention guidelines and cancer outcomes: a systematic review. Cancer Epidemiol Biomarkers Prev 2016;25:1018–28. 10.1158/1055-9965.EPI-16-0121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Xu J-Y, Vena JE, Whelan HK, Robson PJ. Impact of adherence to cancer-specific prevention recommendations on subsequent risk of cancer in participants in Alberta’s Tomorrow Project. Public Health Nutr 2019;22:235–45. 10.1017/S1368980018002689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Kaluza J, Harris HR, Håkansson N, Wolk A. Adherence to the WCRF/AICR 2018 recommendations for cancer prevention and risk of cancer: prospective cohort studies of men and women. Br J Cancer 2020;122:1562–70. 10.1038/s41416-020-0806-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Table S1 shows mutually adjusted associations between individual behavioural components and overall cancer incidence
Supplementary Figure S1 shows hazard ratios (HRs) of site-specific cancer incidence according to combinations of smoking status, alcohol consumption, body mass index, and physical activity. Panels show: A, lung cancer incidence; B, breast cancer incidence; C, colorectal cancer incidence; D, prostate cancer incidence. Exposure categories were defined as described in Table 1. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, healthy BMI, and ≥150 minutes/week of physical activity. All HRs were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at a Bonferroni-adjusted level (α = 0.000617). NE, not estimable due to insufficient case numbers. Estimates based on small case numbers should be interpreted cautiously.
Supplementary Figure S2 shows hazard ratios of overall cancer incidence according to behavioural factor combinations. Behavioural combinations were defined by smoking status, alcohol consumption, body mass index, and physical activity. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, had healthy BMI, and reported ≥150 minutes/week of physical activity. Hazard ratios were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at the Bonferroni-adjusted level. HR, hazard ratio; BMI, body mass index; NE, not estimable due to insufficient case numbers.
Supplementary Figure S3. Hazard ratios of overall cancer mortality according to behavioural factor combinations. Behavioural combinations were defined by smoking status, alcohol consumption, body mass index, and physical activity. The HR = 1.00 cell represents the reference group: participants who were never smokers, reported no alcohol consumption, had healthy BMI, and reported ≥150 minutes/week of physical activity. Hazard ratios were adjusted for age, education level, marital status, annual household income, and first-degree family history of cancer. Bold font indicates statistical significance at the Bonferroni-adjusted level. HR, hazard ratio; BMI, body mass index; NE, not estimable due to insufficient case numbers. Estimates based on small numbers of cancer deaths should be interpreted cautiously.
Data Availability Statement
The data used in this study are available through application to the BCGP. Further information on data access is available at https://bcgpresearch.ca or by contacting bcgpaccess@bccrc.ca. Analytic code is available from the corresponding author upon reasonable request. Access is subject to approval by the BCGP and applicable data governance requirements.