Life Course Predictors of Young Men Surviving to Age 90 in a Cohort Study: The Manitoba Follow-up Study.

The purpose of this study was to identify factors at various time points in life that are associated with surviving to age 90. Data from men enrolled in a cohort study since 1948 were considered in 12-year intervals. Logistic regression models were constructed with the outcome of surviving to age 90. Factors were: childhood illness, blood pressure (BP), body mass index (BMI), chronic diseases, and electrocardiogram (ECG) findings. After 1996, the Short Form-36 was added. A total of 3,976 men were born in 1928 or earlier, and hence by the end of our study window in 2018, each had the opportunity of surviving to age 90. Of these, 721 did live to beyond his 90th birthday.The factors in 1948 which predicted surviving were: lower diastolic BP, lower BMI, and not smoking. In 1960, these factors were: lower BP, lower BMI, not smoking, and no major ECG changes. In 1972, these factors were lower BP, not smoking, and fewer disease states. In 1984, these factors were lower systolic BP, not smoking, ECG changes, and fewer disease states. In 1996, the factors were fewer disease states and higher physical and mental health functioning. In 2008, only higher physical functioning predicted survival to the age of 90. In young adulthood, risk factors are important predictors of surviving to age 90; in mid-life, chronic illnesses emerge, and in later life, functional status becomes predominant.

Patterns of Functional Impairment over Time Amongst Older Men-the Manitoba Follow-up Study.

BACKGROUND: To determine the incidence and prevalence patterns of activity of daily living (ADL) impairments in ageing men. METHODS: 3,983 men were enrolled in the Manitoba Follow-up Study (MFUS) cohort study in 1948. From 1996 onwards, functional status was measured. We classified basic (BADL) and instrumental (IADL) into mutually exclusive categories as a time dependant factor after the second survey wave as: First survey response; no limitation; incident (first episode of disability); persistent (limitation which was seen on all questionnaires after the incident episode); resilient (noted in previous surveys but not present); and recurrent (noted in present survey, and limitations noted as present and absent in previous surveys). RESULTS: There were 1,745 participants in 1996 at a mean age of 76 years. Incident BADL limitations increased substantially with age: from 1% at age 75 to 15% at age 95. Similarly, persistent limitations increased with age: from 0.4% at age 75 to 18% at age 95. However, BADL function was fluid, with many individuals grouped within the resilient and recurrent patterns. Similar age effects and variability were noted in IADLs. CONCLUSION: New and persistent disabilities are highly associated with age. However, there is considerable change in functional status over time.

Associations between concussion and risk of diagnosis of psychological and neurological disorders: a retrospective population-based cohort study.

OBJECTIVE: To investigate associations between concussion and the risk of follow-up diagnoses of attention-deficit hyperactivity disorder (ADHD), mood and anxiety disorders (MADs), dementia and Parkinson's disease. DESIGN: A retrospective population-based cohort study. SETTING: Administrative health data for the Province of Manitoba between 1990-1991 and 2014-2015. PARTICIPANTS: A total of 47 483 individuals were diagnosed with a concussion using International Classification of Diseases (ICD) codes (ICD-9-CM: 850; ICD-10-CA: S06.0). All concussed subjects were matched with healthy controls at a 3:1 ratio based on age, sex and geographical location. Associations between concussion and conditions of interest diagnosed later in life were assessed using a stratified Cox proportional hazards regression model, with adjustments for socioeconomic status and pre-existing medical conditions. RESULTS: 28 021 men (mean age ±SD, 25±18 years) and 19 462 women (30±21 years) were included in the concussion group, while 81 871 men (25±18 years) and 57 159 women (30±21 years) were included in the matched control group. Concussion was associated with adjusted hazard ratios of 1.39 (95% CI 1.32 to 1.46, p<0.001) for ADHD, 1.72 (95% CI 1.69 to 1.76; p<0.001) for MADs, 1.72 (95% CI 1.61 to 1.84; p<0.001) for dementia and 1.57 (95% CI 1.41 to 1.75; p<0.001) for Parkinson's disease. CONCLUSION: Concussion was associated with an increased risk of diagnosis for all four conditions of interest later in life.

Self-Rated Health Predicts Mortality in Very Old Men-the Manitoba Follow-Up Study.

BACKGROUND: Self-rated health (SRH) predicts death, but there are few studies over long-time horizons that are able to explore the effect age may have on the relationship between SRH and mortality. OBJECTIVES: 1. To determine how SRH evolves over 20 years; and 2. To determine if SRH predicts death in very old men. METHODS: We analyzed a prospective cohort study of men who were fit for air crew training in the Second World War. In 1996, a regular questionnaire was administered to the 1,779 surviving participants. SRH was elicited with a 5-point Likert Scale with the categories: excellent, very good, good, fair and poor/bad. We examined the age-specific distribution of SRH in these categories from the age of 75 to 95 years, to the end of the follow-up period in 2018. We constructed age-specific Cox proportional hazard models with an outcome of time to death. RESULTS: SRH declined with age. The gradient in risk of death persisted across all ages; those with poor/fair/bad SRH had consistently higher mortality rates. However, the discrimination between good and excellent was less in those aged 85+. CONCLUSIONS: SRH declines with advancing age, but continues to predict death in older men.

Differential Expression of Genes for Ubiquitin Ligases in Medulloblastoma Subtypes.

Using publically available datasets on gene expression in medulloblastoma (MB) subtypes, we selected genes for ubiquitin ligases and identified statistically those that best predicted each of the four major MB subgroups as separate disease entities. We identify a gene coding for an ubiquitin ligase, ZNRF3, whose overexpression alone can predict the WNT subgroup for 100% in the Pfister dataset. For the SHH subgroup, we identify a gene for a regulatory subunit of the protein phosphatase 2A (PP2A), PPP2R2C, as the major predictor among the E3 ligases genes. The ubiquitin and ubiquitin-like conjugation database (UUCD) lists PPP2R2C as coding for a Cullin Ring ubiquitin ligase adaptor. For group 3 MBs, the best ubiquitin ligase predictor was PPP2R2B, a gene which codes for another regulatory subunit of the PP2A holoenzyme. For group 4, the best E3 gene predictors were MID2, ZBTB18, and PPP2R2A, which codes for a third PP2A regulatory subunit. Heatmap analysis of the E3 gene data shows that expression of ten genes for ubiquitin ligases can be used to classify MBs into the four major consensus subgroups. This was illustrated by analysis of gene expression of ubiquitin ligases of the Pfister dataset and confirmed in the dataset of Cavalli. We conclude that genes for ubiquitin ligases can be used as genetic markers for MB subtypes and that the proteins coded for by these genes should be investigated as subtype specific therapeutic targets for MB.

Older Men's Definitions of Frailty - The Manitoba Follow-up Study.

ABSTRACTBackground: There is little empirical research into lay definitions of frailty. OBJECTIVES: (1) To explore the definitions of frailty among older men, and (2) to explore if these definitions match commonly used clinical definitions of frailty. METHODS: Analysis of open-ended questions to survey data from a prospective cohort study of older airmen. The definitions of frailty were elicited, and grouped according to themes. RESULTS: 147 men responded (mean age: 93). There was considerable heterogeneity in older men's' definitions of frailty, and no theme of frailty was predominant. The most common theme was impairment in activities of daily living. Older men's' definition of frailty was not consistent with any commonly used medical theory of frailty. CONCLUSIONS: Most older men think frailty is important, but their definitions are not consistent. Frailty may be a heterogeneous experience, which different people experience differently.

Lifetime Pattern of Atrial Fibrillation and the Risks of Stroke and Death in a Population-based Cohort of Men (from The Manitoba Follow-Up Study).

Atrial fibrillation (AF) is associated with stroke and mortality. The arrhythmia can be sustained or intermittent. Previous studies that have used fixed covariates and short-time horizons to examine the relation between the pattern of AF and the occurrence of events have produced conflicting results. The Manitoba Follow-Up Study includes 3,983 originally healthy men who have been followed with routine examinations since 1948. AF status during each visit was classified into the following patterns: free of AF, newly diagnosed; intermittent AF-in sinus;intermittent AF-in AF; sustained AF. We created adjusted Cox proportional hazards models with time-dependent covariates to estimate risks for stroke and death according to AF pattern. After 167,982 person-years of follow-up and 66,297 electrocardiograms (ECGs), 548 men had at least 1 ECG documenting AF, 799 had a stroke, and 3173 died. Relative to men free of AF, sustained and newly diagnosed AF were associated with stroke (hazard ratio [HR] 1.85, 95% confidence interval [CI] 1.33 to 2.59 and HR 1.71, 95% CI 1.10 to 2.66, respectively) and death (HR 2.48, 95% CI 2.11 to 2.92 and HR 2.03, 95% CI 1.64 to 2.52, respectively). Intermittent AF was associated with death (HR 2.41 95% CI 1.58 to 3.68 in AF and HR 1.71 95% CI 1.44 to 2.03 in sinus), but not with stroke (HR 0.68, 95% CI 0.22 to 2.13 in AF and HR 1.02 95% CI 0.72 to 1.45 in sinus). Antithrombotic therapy was associated with a reduced risk of the outcomes. In conclusion, longitudinal analysis of patterns of AF evolving over time provided evidence that the associated risks of stroke and death vary considerably with rhythm classification on serial ECGs.

Quality of Life Trajectories Predict Mortality in Older Men: The Manitoba Follow-Up Study.

OBJECTIVE: To describe quality of life trajectories of older men over a 10-year time frame in mental and physical health domains, and to determine if these trajectories predict death over a subsequent 9-year period. METHOD: A cohort study of Royal Canadian Air Force aircrew veterans. We used Short Form-36 (SF-36) measures of mental and physical functioning collected prospectively at six time points between 1996 to 2006 (734 men with a mean age of 85.5 [ SD 3.0] years in 2006) to determine trajectories. Continued contact with the cohort from 2006 to 2015 determined subsequent mortality. RESULTS: Men were more likely to maintain high levels of mental functioning than physical functioning. Thirty-seven percent of participants maintained a high level of both mental and physical functioning. Declining function in either mental or physical function was associated with lower survival. CONCLUSION: Men who maintain physical and mental functioning have a lower mortality rate.

Construct Validation of Three Nutrition Questions Using Health and Diet Ratings in Older Canadian Males Living in the Community.

Brief nutrition screening tools are desired for research and practice. Seniors in the Community: Risk Evaluation for Eating and Nutrition (SCREEN-II, 14 items) and the abbreviated version SCREEN-II-AB (8 items) are valid and reliable nutrition screening tools for older adults. This exploratory study used a retrospective cross-sectional design to determine the construct validity of a subset of 3 items (weight loss, appetite, and swallowing difficulty) currently on the SCREEN-II and SCREEN-II-AB tools. Secondary data on community-dwelling senior males (n = 522, mean ± SD age = 86.7 ± 3.0 years) in the Manitoba Follow-up Study (MFUS) study were available for analysis. Participants completed the mailed MFUS Nutrition Survey that included SCREEN-II items and questions pertaining to self-rated health, diet healthiness, and rating of the importance of nutrition towards successful aging as the constructs for comparison. Self-perceived health status (F = 14.7, P < 0.001), diet healthiness (ρ = 0.17, P = 0.002) and the rating of nutrition's importance to aging (ρ = 0.10, P = 0.03) were correlated with the 3-item score. Inferences were consistent with associations between these construct variables and the full SCREEN-II. Three items from SCREEN-II and SCREEN-II-AB demonstrate initial construct validity with self-perceived health status and diet healthiness ratings by older males; further exploration for criterion and predictive validity in more diverse samples is needed.

Variations over four decades in body mass index trajectories prior to diagnosis of diabetes mellitus: The Manitoba Follow-up Study, 1948-2007.

OBJECTIVE: The aim of the study was to explore the variations in body mass index (BMI) trajectories during the 20 years before diagnosis of type 2 diabetes mellitus (T2DM) over four decades between 1968 and 2007. METHODS: Longitudinal measurements of BMI from 437 men, all with a diagnosis of T2DM, were used in the analysis. A mixed method approach was used to fit individual patterns of BMI measurements during the 20 years before diagnosis of T2DM. RESULTS: The mean BMI at diagnosis was 26.7 kg/m2 (95% confidence interval, 26.4-27.1). Compared with men whose condition was diagnosed between 1968 and 1977, for men with a diagnosis between 1978 and 2007 the mean BMI about 10 years before diagnosis significantly increased by 0.92 to 1.54 BMI units. Decades also varied in how long there was a persistent increase in BMI during the 20 years before diagnosis. The rate of change in mean BMI among men whose T2DM was diagnosed in the most recent two decades increased by 8.8% to 22.6% during the 10-year interval before diagnosis, but there was no significant difference among men given a diagnosis between 1978 and 1987. The quadratic trend of BMI prior to diagnosis was also significantly affected by age at diagnosis. CONCLUSION: The BMI trajectories during the 20 years leading up to T2DM varied by decade of diagnosis. The increase in BMI persisted for much longer among relatively younger men with a diagnosis in more recent decades. Strategies to prevent T2DM, informed by the pattern of BMI trajectories, should be customized to consider a potential age-period effect.

Cohort Profile: The Manitoba Follow-up Study (MFUS).

The Manitoba Follow-up Study (MFUS) is Canada's longest running study of cardiovascular disease and ageing. The MFUS cohort consists of 3983 men recruited from the Royal Canadian Air Force at the end of World War II. At entry to the study, 1 July 1948, their mean age was 31 years, with 90% between ages 20 and 39 years. All study members were free of clinical evidence of ischaemic heart disease. The protocol of MFUS was to obtain routine medical examinations from these men at regular intervals over time. The research goal of the study was to examine the role that any abnormalities detected on routine electrocardiograms from apparently healthy men might play in the prediction of subsequent diagnoses of cardiovascular disease. Over the course of 65 years, about 35% of the cohort has documented evidence of ischaemic heart disease. The research focus was expanded in 1996 to explore the roles of physical, mental and social functioning in support of healthy and successful ageing. On 1 July 2013, 429 original cohort members were alive with a mean age of 92 years. Collaborative research with others outside the in-house team is welcomed.

Nutritional risk in community-dwelling older men: the Manitoba follow-up study.

PURPOSE: The role of nutrition in older men's health and successful aging has been inadequately studied. We examined the relationships among nutritional risk, self-rated health, and successful aging in community-dwelling Canadian older men. METHODS: The surviving cohort of the Manitoba Follow-up Study (n=690, mean age = 86.8 years) were sent a self-administered nutrition survey in December 2007. The survey consisted of the Seniors in the Community: Risk Evaluation for Eating and Nutrition, version II (SCREEN II), a validated tool for assessing nutritional risk of cognitively intact community-living older adults, and questions about successful aging and health. RESULTS: Of the 553 surveys returned (80% response), 522 with complete SCREEN II data were included in the analysis. Forty-four percent of respondents were at high nutritional risk, 24% were at moderate risk, and 32% were at low risk. Significant relationships were found between nutritional risk and self-rated health (P<0.0001) and successful aging (P=0.008), with greater nutritional risk associated with lower self-ratings of health and successful aging. Higher use of prescription medication was related to greater nutritional risk (P=0.004). CONCLUSIONS: Nutritional screening programs for community-dwelling older men are warranted as two-thirds of the study participants were at nutritional risk. Identifying older men at nutritional risk is a critical step in the process of nutritional assessment, and subsequent nutrition interventions and follow-up are required to prevent further health decline.

Response shift recalibration and reprioritization in health-related quality of life was identified prospectively in older men with and without stroke.

OBJECTIVES: To prospectively consider evidence for response shift (RS) in health-related quality of life (HRQL) in older men who experienced stroke or remained stroke free. RS is a change in the meaning of self-evaluation; this includes recalibration, reprioritization, and reconceptualization. STUDY DESIGN: A cohort of 3,983 male World War II Royal Canadian Air Force recruits has been followed since 1948. There were three prospectively determined groups: stroke survivors (n = 168; mean age, 80.1 years); older stroke-free group (n = 254; mean age, 82.8 years); and younger stroke-free group (n = 323; mean age, 74.7 years). The Short Form-36 (SF-36) was used to evaluate HRQL. Longitudinal structural equation models were developed using SF-36 subscales and three latent variables. Measurement invariance over two time points for each of the three groups was evaluated to identify RS. RESULTS: All RS models had reasonable fit: stroke [root mean square error of approximation (RMSEA), 0.069; 90% confidence interval (CI): 0.052, 0.086], older (RMSEA, 0.055; 90% CI: 0.041, 0.068), and younger (RMSEA, 0.062; 90% CI: 0.051, 0.074). Recalibration of physical function occurred in all three groups. Reprioritization of role limitations due to physical health happened in both stroke-free groups. CONCLUSION: This study is unique in our ability to prospectively identify RS recalibration and reprioritization in HRQL in aging men with stroke and remaining free of stroke. Changes in the meaning of self-evaluation of HRQL occur not only with stroke but also in men who remain free of stroke.

Themes from older men's lay definitions of successful aging as indicators of primary and secondary control beliefs over time: The Manitoba Follow-up Study.

Constructs of control have theoretically been equated to successful aging in the psychology literature. Hence, we used themes from lay definitions of successful aging to quantify the prevalence of primary and secondary control beliefs over time. In doing so we hoped to shed new light upon the virtually uncharted area of older men's primary and secondary control beliefs over time. Using successful aging narratives spanning a 10-year timeframe from the Manitoba Follow-up Study cohort, we mapped themes from older men's lay definitions of successful aging onto Rothbaum, Weisz, and Snyder's (1982) constructs of primary and secondary control. We then examined the prevalence of the constructs of control over 10 years and found that some men emphasized primary control, some emphasized secondary control, and others emphasized both, prospectively. Counter to what had previously been theorized, many older men continued to emphasize primary control as important well into late life. As expected, secondary control became more important with age. Furthermore, among those men who endorsed both primary and secondary control, significantly more men switched emphasis from primary to secondary control beliefs as they aged. This finding supported Rothbaum et al.'s (1982) surmise that individuals could switch from one type of control to another, presumably as life circumstances dictated. Knowing which types of control beliefs older men emphasize as they age has theoretical and practical implications. Theoretically, it sheds new light on the under-researched area of control beliefs in older men. Practically, it is informative for anyone interested in enhancing older men's perceptions of control in very late life, particularly in the face of otherwise uncontrollable age-related decline and imminent demise.

Nutritional risk and 5-year mortality of older community-dwelling Canadian men: the Manitoba Follow-Up Study.

This study examines nutritional risk and 5-year mortality rates of community-dwelling older Canadian men participating in the Manitoba Follow-Up Study. The surviving cohort (n = 690; mean age = 86.8 years) was mailed a self-administered Nutrition Survey consisting of SCREEN II, a validated nutrition risk screening tool and health-related questions. Five hundred fifty-three completed surveys (80% completion rate) were returned, with 522 (94%) containing complete responses necessary to score nutritional risk, comprising the participants for this analysis. Forty-four percent of the 522 respondents were scored as high risk, 24% at moderate risk, and 32% at low risk. Over five years from 2007 to 2012, 212 (41%) of the men died, with 1-year, 2-year, 3-year, and 4-year survival rates of 92%, 86%, 77%, and 66%, respectively. Men in the lowest 40th percentile of the nutritional risk distribution accounted for half of all deaths. Adjusted for other characteristics, Cox proportional hazard models demonstrated that with each unit decline on the nutritional risk scale there was a 4% greater risk of mortality (hazard ratio = 0.96 [95% CI 0.94,0.98]). Early identification of older men at nutritional risk and timely nutrition interventions are essential in delaying the progression of morbidity and mortality.

Older men's lay definitions of successful aging over time: the Manitoba follow-up study.

The concept of "successful aging" has become widely accepted in gerontology, yet continues to have no common underlying definition. Researchers have increasingly looked to older individuals for their lay definitions of successful aging. The present analysis is based on responses to five questionnaires administered to surviving participants of the male Manitoba Follow-up Study cohort (www.mfus.ca) in 1996, 2000, 2002, 2004, and 2006 (n = 2,043 men were alive at a mean age of 78 years in 1996). One question on each survey asked: "What is YOUR definition of successful aging?" Applying content analysis to the 5,898 narratives received over the 11 years, we developed a coding system encompassing 21 main themes and 86 sub-themes defining successful aging. We quantitatively analyzed trends in prevalence of themes of successful aging prospectively over time. Our findings empirically support colleagues' past suggestions to shift from defining successful aging in primarily biomedical terms, by taking lay views into account.

The epidemiology of hepatitis C in a Canadian Indigenous population.

BACKGROUND: An estimated 1% to 1.9% of North Americans are infected with the hepatitis C virus (HCV). Although Indigenous peoples are considered to bear the highest burden, there are only limited data regarding the demographic features and epidemiology of hepatitis C in this population. OBJECTIVES: To document the demographic characteristics, rates of newly diagnosed hepatitis C cases and prevalence of HCV infection in a Canadian First Nations population, and to compare the findings with an infected non-First Nations population. METHODS: A research database spanning 1991 to 2002 was developed, linking records from multiple clinical and administrative sources. Over a 12-year period, 671 First Nations and 4347 non-First Nations HCV-positive Canadians were identified in the province of Manitoba. Demographics, residence and time trends were compared between infected First Nations and non-First Nations persons. RESULTS: HCV-infected First Nations individuals were younger (mean [± SD] age 33.0±0.4 years versus 39.7±0.2 years; P<0.0001), more often female (60% versus 40%; P<0.0001) and more often resided in urban centres (73% versus 27%; P<0.001). The rate of newly diagnosed HCV cases was 2.5-fold (91.1 per 100,000 versus 36.6 per 100,000; P<0.000) and prevalence 2.4-fold (801.7 per 100,000 versus 334.8 per 100,000; P<0.000) higher among the First Nations relative to non-First Nations populations. CONCLUSIONS: The results of the present large population-based study indicate that the First Nations population with hepatitis C is characteristically different from infected non-First Nations persons. The results also describe higher rates of newly diagnosed cases and prevalence of HCV infection in the First Nations population. These findings should serve as an important baseline for future primary prevention and therapeutic intervention strategies in this high-risk population.

Clinical diagnoses before age 75 and men's survival to their 85th birthday: the Manitoba follow-up study.

PURPOSE: Of all Canadian and American men who live to age 75 years, about half can expect to live to age 85. Our objective is to examine how clinical diagnoses made before age 75 relate to a man's survival to age 85 years. DESIGN AND METHODS: Since 1948, a cohort of 3,983 young men (mean age of 31 years at entry) has been followed with routine contact and medical examinations to prospectively document incident disease. Over 62 years of follow-up, 2,414 of the cohort lived to celebrate their 75th birthday. Of these survivors, 1,060 (44%) died before their 85th birthday. Cox proportional hazard models were used to examine the effects of ischemic heart disease, cancer, cerebrovascular disease, diabetes mellitus, peripheral arterial disease, and chronic obstructive pulmonary disease on all-cause mortality between age 75 and 85 years. RESULTS: Modeled as six binary risk factors at age 75 years, all were significantly (p < .01) and independently related to 10-year mortality. Multivariate risk ratios ranged from 1.36 to 1.46 except for chronic obstructive pulmonary disease with a risk ratio of 1.85 (95% CI: 1.38, 2.49). The cumulative 10-year probability of survival from age 75 to 85 among men with none of these diagnoses was 63%, 52% for any one diagnosis, 39% for two diagnoses, and 22% for three or more diagnoses. IMPLICATIONS: Joint independence of these six common clinical diagnoses implies that each is important and their effects on mortality are cumulative.

Rural practice and the personal and educational characteristics of medical students: survey of 1269 graduates of the University of Manitoba.

OBJECTIVE: To describe the relationships between rural practice and the personal and medical education characteristics of medical students and residents. DESIGN: Cross-sectional, mailed survey. SETTING: Manitoba. PARTICIPANTS: Of 2578 physician graduates of the University of Manitoba from 1965 to 2000 who were surveyed, 1269 (49%) responded. MAIN OUTCOME MEASURES: Whether physicians had ever practised in rural settings, and their demographic characteristics and adolescent, medical school, and residency training experiences. Multivariate logistic regression models were used to determine variables jointly and independently associated with rural practice. RESULTS: Of 1269 respondents, 39% had practised in rural settings, including 58% of the 362 respondents who identified family practice as their primary career activity, and 32% of the 907 respondents whose primary activities were other than family practice. For all graduates, being male (P = .0289), having lived in a rural community (P < .0001), having attended a rural high school (P < .0001), and having rural educational experiences during medical school (P = .0068) or during postgraduate training (P < .0001) were significantly related to a greater likelihood of rural practice. In the final multivariate model, graduates of rural high schools, compared with those from urban public schools, were 1.57 times (95% CI 1.09 to 2.26) more likely to have practised in rural settings. Graduates who undertook part of their undergraduate training in rural settings were 1.34 times (95% CI 1.09 to 1.75) more likely to practise in rural locations. For both undergraduates and residents, the distance of their rural education experiences from Winnipeg and the likelihood of rural practice were directly related. For both FPs and non-FPs, being male and undertaking rural education during residency training were associated with a greater likelihood of rural practice, as was the distance of the training experience from the urban setting. For non-FPs a similar association was observed with undergraduate rural experiences. CONCLUSION: This large survey of graduates from a Canadian medical school demonstrated that attending a rural high school (P < .0001) and having rural educational exposure during medical school and residency training (P = .0068) were significantly associated with a physician practising in a rural location. That is, rural educational experiences on the continuum from high school through residency training appeared to be associated with rural practice.