Active commuting and cardiovascular disease risk: The CARDIA study

Abstract

Background

There is little research on the association of lifestyle exercise, such as active commuting (walking or biking to work) with obesity, fitness, and cardiovascular disease (CVD) risk factors.

Methods

This cross-sectional study included 2,364 participants who worked outside the home from 2005–06 of the CARDIA study. Associations between walking/biking to work (self-reported time, distance, and mode of commuting) with: body weight (measured height and weight); obesity (BMI≥30 kg/m²); fitness (symptom-limited exercise stress testing); objective moderate-vigorous physical activity (accelerometry); and CVD risk factors [blood pressure (oscillometric systolic and diastolic); and serum measures (fasting measures of lipids, glucose, and insulin)] were separately assessed by sex-stratified multivariable linear (or logistic) regression modeling.

Results

16.7% of participants used any means of active commuting to work. Controlling for age, race, income, education, smoking, exam center, and physical activity index excluding walking, men with any active commuting (versus none) had reduced likelihood of obesity (OR=0.50; 95% CI 0.33, 0.76), reduced CVD risk: ratio of geometric mean triglycerides: 0.88 (0.80, 0.98); ratio of geometric mean fasting insulin: 0.86 (0.78, 0.93); difference in mean diastolic blood pressure (mmHg):−1.67; −3.20, −0.15)], and higher fitness; mean difference in treadmill test duration {seconds}: 50.0 (31.5, 68.6) in men and 28.8 (11.6, 45.9) in women.

Conclusions

Active commuting was positively associated with fitness in men and women and inversely associated with body mass, obesity, triglycerides, blood pressure and insulin in men. Active commuting should be investigated as a modality for maintaining or improving health.

Introduction

Because of its flexibility and accessibility, ^{1, 2} walking is generally reported as the most popular leisure-time physical activity for adults^3–5 and has been specifically promoted as a targeted activity to achieve national physical activity recommendations.^{1, 6} For most adults walking 60 minutes per day at a brisk pace is sufficient to meet the Institute of Medicine’s physical activity guidelines for avoiding weight gain.^{7, 8} One potentially effective means of increasing physical activity is through alternative, non-leisure forms of physical activity such as active commuting (walking or biking to work).

Selected research has suggested an inverse relationship between leisure-time walking and adiposity^{9, 10} and cardiovascular disease (CVD) risk factors.^11–14 A recent meta-analysis showed modest reductions in cardiovascular outcomes related to active commuting,¹⁵ but the majority of these studies were conducted in Scandinavian samples. The frequency of active commuting is likely to vary by region. Further, the prevalence of obesity and CVD risk factors are higher in the U.S than in Scandinavia. In addition, many previous analyses have not included detailed behavioral and clinical control variables that would permit investigations of whether active commuting has an independent effect on cardiovascular health. Thus, research is needed in U.S. population-based cohorts with rich behavioral data and clinically measured CVD risk factors.

We use population-based data from the CARDIA study to examine the association between active commuting (defined as walking or biking to work) with obesity, fitness, and CVD risk factors (blood pressure, lipids, and blood glucose, insulin) to understand whether active commuting is a feasible target for maintaining or improving health. We hypothesized that active commuting is positively associated with lower obesity, higher fitness, and favorable CVD risk factor profile.

Methods

Setting and Participants

The Coronary Artery Risk Development in Young Adults (CARDIA) Study is a population-based prospective epidemiologic study of the determinants and evolution of cardiovascular disease risk factors among young adults. At baseline (1985–6), 5,115 eligible participants, aged 18–30 years, were enrolled with balance by race, gender, education (high school or less and more than high school) and age (18–24 and 25–30) from the populations of Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA. Specific recruitment procedures are described elsewhere.¹⁶ Six follow-up examinations were conducted over 20 years. We use data from the Year 20 exam (2005–06), with the year 20 retention rate for surviving cohort members of 72%.

From the initial 3,549 study subjects at year 20, we excluded one transgendered respondent (n=1) and women who were pregnant at time of exam (n=6). We further excluded participants who reported that they did not work outside of the home (n=507) or for whom data on work outside of the home was missing (n=567) and those missing outcome or covariate data (n=104). The final analysis sample included 2,364 individuals with complete exposure, outcome and covariate data. Among those meeting inclusion criteria, whites, nonsmokers, and those with high income and education and high physical activity levels were more likely to have complete data and thus included in analysis. Missing data also varied by study site, with those in Minneapolis less likely to have complete data. This secondary data analysis was approved by the CARDIA Steering committee and the Institutional Review Board of University of North Carolina at Chapel Hill.

Exposure Measure: Active Commuting

At the year 20 exam, subjects reported (in minutes and miles) how long it takes to get from home to their place of work and the percentage of trip by car, public transportation (bus, train, subway), walking, or bicycling. Active commuting was defined as any walking or biking during the trip from home to work.

Outcome Measures

BMI and obesity

Measurements of weight and height, with subjects in light clothing and without shoes, were obtained according to standardized protocol described previously.¹⁷ Body mass index (BMI) was calculated as a measure of weight (kg)/height (m)² and obesity was classified as a BMI ≥ 30.0.

Leisure time and occupational physical activity

At each examination, self-reported physical activity was ascertained by an interviewer-administered questionnaire designed for CARDIA. Participants were asked about the frequency of participation in 13 different activity categories (eight vigorous and five moderate) of recreational sports, exercise, leisure, and occupational activities over the previous 12 months. Vigorous activities included running, racquet sports, bicycling faster than 10 miles per hour, swimming, vigorous exercise classes, sports (e.g., basketball, football), heavy lifting, carrying or digging on the job, and home activities such as snow shoveling or lifting heavy objects. Moderate activities included non-strenuous sports (e.g., softball), walking, bowling/golf, home maintenance (e.g., gardening, raking), and calisthenics. Because participants were not asked explicitly about duration of activity, physical activity scores are expressed in exercise units (EU), from which duration can be estimated.¹⁸ Scores were computed by multiplying the intensity of the activity by the number of months of participation, weighted by a factor proportional to lesser or greater frequency and duration. Separate scores were obtained for heavy (i.e., vigorous) and moderate activities. The two subscores were summed for a total physical activity score. As an example, a score of 100 EU is roughly equivalent to participation in a vigorous activity, 2 or 3 hours per week for six months of the year, calculated as [6 MET*(3*6 months of high volume activity)]. The reliability and validity of the instrument is comparable to other activity questionnaires.^{18, 19}

Using the physical activity scoring algorithm, we created two physical activity measures. First, we created a specific leisure-walking score derived from walking items in the physical activity questionnaire described above. We used the continuous walking score, ranging from 0–144 units, to categorize 12-month walking patterns at 3 metabolic equivalents (METs) defined as multiples of the resting metabolic rate: none (0 units), intermittent (1–143 units) and regular (144 units, approximating walking ≥ 4 hours/wk over 12 months) to capture subjects with no, moderate, and high levels of walking. Second, we created a physical activity score that excluded walking, which was dichotomized into low (below the median) and high. This “non-walking” activity variable was used as a control variable in our multivariable regression models in order to statistically control for physical activity other than walking for transit or leisure in models using active transit physical activity exposures.

Accelerometer-measured physical activity

Total daily minutes of moderate and vigorous physical activity were obtained from at least four days of accelerometer recordings. Moderate-vigorous physical activity (MPA) cutpoints were established during a treadmill walking session using Freedson cut points (1952–5725 counts per minute). Participants were instructed to wear the accelerometer (ActiGraph model 7164) around the waist for seven days, except when sleeping, bathing, or engaging in water activities. The epoch was set at one minute, and periods of non-wear were identified by 60 or more consecutive zero counts. At least four days of valid data (≤720 minutes of inactive time) were required for inclusion in analyses. MPA minutes per day was dichotomized (<24.0, ≥24.1 minutes/day), equivalent to the recommended 5 MPA bouts/week and examined as a CVD-related health behavior (outcome). Average accelerometer-measured vigorous physical activity (VPA) minutes per day was dichotomized into meeting (versus not meeting) VPA recommendations and used to exclude those meeting VPA recommendations (n=102 men, 98 women of those with valid accelerometer data) in models examining accelerometer-measured MPA. The rationale for the exclusion was to tie findings directly with the recommendation for moderate physical activity.^20,21

Treadmill Fitness Test Duration

A symptom-limited maximal Graded Exercise Test was administered using a modified Balke protocol,²² including nine 2-minute stages of increasing difficulty with participants encouraged to exercise to exhaustion, followed by a recovery period at a speed of 3.2 km/hour at 0% grade. Fitness was indicated by the treadmill test duration in seconds. Primary exclusion criteria for exercise testing included a resting systolic or diastolic pressure >160 or >100 mmHg, or being febrile at time of examination.

Lipid, Glucose, and Insulin Measurements

Samples of blood lipids, glucose, and insulin were collected according to standardized CARDIA protocols and were processed at central laboratories as described previously. ^23–26 Individuals fasting for less than eight hours were excluded from these analyses. Insulin was measured by radioimmunoassay.²⁷ We created the following measures: high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), triglycerides (TRIG) [all three measures reported in mg/dL and exclude subjects reporting cholesterol-lowering medications (n=194), and fasting glucose and fasting insulin [both measures reported in mg/dL and exclude subjects reporting diabetes medications (n=85)].

Blood Pressure Measurements

Three systolic and diastolic blood pressure measurements were obtained by a trained technician using a standard automated blood pressure measurement monitor (Omron HEM907XL) after a 5-minute seated rest. The average of the second and third measurements was used for analysis. Participants were asked to fast for at least 12 hours and not to smoke or engage in heavy physical activity for at least 2 hours prior to the measurement. We used systolic and diastolic blood pressure calibrated to be comparable with random-zero sphygmomanometers used in prior CARDIA exam periods (systolic blood pressure calibrated to the random zero level was estimated as 3.74+0.96* the Omron value; diastolic blood pressure as 1.30+0.97*the Omron value) and reported in mg/dL and excluding subjects reporting blood pressure-lowering medications (n=374).

Control variables

Sociodemographic and behavioral characteristics were measured by self- and interviewer-administered questionnaires. Age (years), race (black, white), income tertiles (<$50,000, $50,000–99,999, ≥$100,000), years of education (≤ high school, any college, graduate/professional training), and clinic site (Birmingham, Chicago, Minneapolis, Oakland) were used as control variables in all statistical models. Smoking status was classified as: never smoker, former smoker, or current smoker, and alcohol intake was classified as: no consumption, ≤12 ml/day (sample median), >12 ml/day).

Statistical Analysis

Statistical analyses were conducted using Stata (version 9.2, College Station, TX).²⁸ Descriptive statistics were computed for commuting patterns, non-walking physical activity, smoking, alcohol consumption, and sociodemographic factors and presented by active versus non-active commuting (any walking or biking during the trip from home to work) and sex. Percentages were calculated for categorical variables. Continuous variables are presented either as means and standard errors or median and interquartile range (for skewed measures).

Associations between walking or biking to work and body mass, fitness, and CVD risk factors were separately assessed by sex-stratified multivariate regression (linear, logistic, or multinomial logistic) modeling. If necessary, outcome variables were transformed or categorized based on their sample distribution. Skewed variables were natural log transformed to achieve approximate normality or categorized into ordinal variables if transformation was not adequate. Leisure walking and accelerometer-measured leisure MPA were examined as categorical variables in order to explicitly examine policy-relevant categories of physical activity. All models adjusted for sociodemographics (age, race, income, years of education, and exam center). Leisure-time walking models also adjusted for non-walking PA score (to hold all non-walking physical activity constant). Accelerometer-measured MPA models excluded those meeting accelerometer-measured VPA recommendations (>8min/day). Two sets of fitness, obesity, BMI models were conducted: Model 1 adjusted for sociodemographics and Model 2 adjusted for sociodemographics and health-related behaviors (alcohol consumption, smoking, and non-walking PA score). Models for lipids, blood pressure, and fasting glucose and insulin measurements additionally controlled for BMI in order to examine BMI as a potential mediator.

Measures of effect varied across models, depending upon the outcome measure. For categorical outcomes, adjusted odds ratios were used. For continuous natural log transformed outcomes, we calculated the ratio of the outcome in its reported scale for those who actively commute relative to those who do not. For continuous untransformed outcome, we calculated the difference in outcome between those who actively commute versus those who do not.

Interactions between: active commuting and moderate to vigorous physical activity other than walking were tested by including the appropriate cross-product terms in the model and assessing likelihood ratio tests (p≤0.10). Final models were stratified by sex. Variables were retained in models if backward elimination resulted in a >10% change in the estimated effect measures or if variables were conceptually relevant (e.g., control for clinic site).

Results

Descriptive Characteristics

Of the 2,364 respondents who worked outside of the home, 16.7% of the sample (men: 18.0%; women: 15.6%) used any means of active commuting to work. In both sexes, active commuters were generally of higher education levels, with variation across exam center (particularly low active commuting in Birmingham, AL). Among women, active commuting was higher among whites and those with higher non-walking physical activity levels. Among men, active commuting was higher in those with greater alcohol intake (Table 1).

Table 1.

Descriptive statistics^* at Exam Year 20 (2005–06) of the Coronary Artery Risk Development in Young Adults (CARDIA) Study, by sex and commuting^† status

	Men		Women
	Non-Active Commuters (n=873)	Active Commuters (n=192)	Non-Active Commuters (n=1096)	Active Commuters (n=203)
age [mean (SE)]	45.2 (0.1)	45.3 (0.2)	45.1 (0.1)	45.6 (0.3)
Race [% (SE)]
Black	39.3 (1.7)	35.4 (3.5)	50.5 (1.5)‡	37.9 (3.4)‡
White	60.7 (1.7)	64.6 (3.5)	49.5 (1.5)‡	62.1 (3.4)‡
Education [% (SE)]
<=High school	38.8 (1.7)‡	32.8 (3.4)‡	36.2 (1.5)‡	26.1 (3.1)‡
Any college	42.6 (1.7)‡	34.4 (3.4)‡	44.0 (1.5)‡	50.7 (3.5)‡
Graduate/Prof	18.6 (1.3)‡	32.8 (3.4)‡	19.8 (1.2)‡	23.2 (3.0)‡
Income Tertile [% (SE)]
1st	23.4 (1.4)	27.6 (3.2)	36.0 (1.5)	37.4 (3.4)
2nd	37.9 (1.6)	35.4 (3.5)	38.2 (1.5)	34.0 (3.3)
3rd	38.7 (1.6)	37.0 (3.5)	25.8 (1.3)	28.6 (3.2)
Exam Center [% (SE)]
1	29.8 (1.5)‡	7.3 (1.9)‡	25.2 (1.3)‡	4.9 (1.5)‡
2	20.6 (1.4)‡	31.8 (3.4)‡	21.4 (1.2)‡	27.1 (3.1)‡
3	24.6 (1.5)‡	30.7 (3.3)‡	22.7 (1.3)‡	30.5 (3.2)‡
4	25.0 (1.5)‡	30.2 (3.3)‡	30.7 (1.4)‡	37.4 (3.4)‡
Smoking Status [% (SE)]
never	65.4 (1.6)	67.2 (3.4)	60.7 (1.5)	61.6 (3.4)
former	17.5 (1.3)	17.7 (2.8)	22.2 (1.3)	21.7 (2.9)
current	17.1 (1.3)	15.1 (2.6)	17.2 (1.1)	16.7 (2.6)
Alcohol [% (SE)]
None	39.5 (1.7)‡	26.0 (3.2)‡	52.5 (1.5)	45.3 (3.5)
≤ 12 ml/day§	22.7 (1.4)‡	31.8 (3.4)‡	26.2 (1.3)	27.1 (3.1)
>12 ml/day§	37.8 (1.6)‡	42.2 (3.6)‡	21.4 (1.2)	27.6 (3.1)
PA Other than Walking** [% (SE)] Gender-specific Tertiles
1st tertile	29.7 (1.5)	28.1 (3.3)	31.1 (1.4)‡	24.1 (3.0)‡
2nd tertile	35.9 (1.6)	33.3 (3.4)	35.3 (1.4)‡	30.0 (3.2)‡
3rd tertile	34.5 (1.6)	38.5 (3.5)	33.6 (1.4)‡	45.8 (3.5)‡

^*Limited to respondents with no missing covariates

^†Active commuting was defined as any walking or biking during the trip from home to work

^‡Statistically different (p<0.05) between Active Commuters vs. Non-Active Commuters, within sex (chi-square)

^§Cutpoint is equal to the median value among those with values >0

^**PA=Physical Activity; index derived from a standard physical activity questionnaire

Patterns of commuting behavior, shown in Table 2, were reported for the total trip to work (e.g., subjects reported percentage of trip made by walking, bike, car, public transportation). Average miles and minutes of the commute to work varied between active and non-active commuters, with medians of 5 miles (men and women) and 20 (men) and 17 (women) minutes for those who actively commuted to work (distance and minutes of commuting may not correspond due to combined modes of transportation). Considerably higher proportions of subjects used walking versus biking for their active commuting. There was variation across modes of transit, even for those who used active means of commuting, with highest proportions of commuters using cars for some portion of their commute. Of note are low overall rates of active commuting.

Table 2.

Commuting patterns^* at Exam Year 20 (2005–06) of the Coronary Artery Risk Development in Young Adults (CARDIA) Study, by sex and commuting status^†

	Men		Women
	Non-Active Commuters (n=873)	Active Commuters (n=192)	Non-Active Commuters (n=1096)	Active Commuters (n=203)
Miles to work [median (IQR)]‡	14 (7, 22)§	5 (2, 13.5)§	10 (5, 20)§	5 (1, 10)§
Minutes to work [median (IQR)]‡	20.5 (15, 35)§	20 (10, 40)§	20 (15, 30)§	17 (10, 30)§
Walked to work [% (SE)]	NA†	64.1 (3.5)†	NA†	82.8 (2.7)†
Biked to work [% (SE)]	NA†	49.5 (3.6)†	NA†	28.1 (3.2)†
% trip by bike [median (IQR)]	NA†	0 (0, 10)†	NA†	0 (0, 1)†
% trip by car [median (IQR)]	100 (100, 100)§	50 (0, 90)§	100 (100, 100)§	50 (0, 90)§
%trip by public transit [median (IQR)]	0 (0, 0)§	0 (0, 32.5)‡	0 (0, 0)§	0 (0, 50)§
%trip by walking [median (IQR)]	NA†	5 (0, 22.5)†	NA†	10 (1,30)†

^*Active commuting was defined as any walking or biking during the trip from home to work. Given the very low rates of active commuting, the distribution of these variables includes a substantial proportion of the sample reporting no active commuting (or 0% of the trip made by biking, walking, or public transport).

^†Limited to respondents with no missing covariates. Medians are reported as [median (IQR=interquartile range)] Walking and biking to work statistics are not applicable to non-active commuters; differences in walking and biking to work statistics were therefore not tested between active and non-active commuters.

^‡Distance and minutes of commute reported by percentage of trip made by walking, bike car, public transportation thus values are not expected to correspond

^§Statistically different (p<0.001) between Active Commuters vs. Non-Active Commuters, within sex (t-test on ln transformed variable)

Likelihood of leisure walking was positively related to active commuting, with strongest association seen for the regular walker versus non-walker comparison (Table 3). Similarly, among those not meeting VPA recommendations, accelerometer-measured MPA was positively related to active commuting, although statistically significant for women only.

Table 3.

Association between active commuting physical activity^* and two forms of leisure physical activity: walking and objectively measured moderate and vigorous physical activity at Exam Year 20 (2005–06) of the Coronary Artery Risk Development in Young Adults (CARDIA) Study^†

		Men			Women
Outcome	Model (effect measure)	n	% (SE)	effect (95% CI)	n	% (SE)	effect (95% CI)
Leisure Walking‡	mlogit (OR)	1065			1299
non-walker (referent outcome)				--			--
intermittent vs. non-walker				1.96 (1.25, 3.08)			2.82 (1.64, 4.86)
regular vs. non-walker				3.26 (1.95, 5.43)			5.62 (3.10, 10.18)
Accelerometer-measured leisure MPA§		721			930
Meets recommendations	logit (OR)			1.16 (0.71, 1.90)			1.83 (1.25, 2.69)

^*Active commuting was defined as any walking or biking during the trip from home to work.

^†Bold indicates significant association (p<0.05).

^‡Odds ratios obtained multinomial logistic regression models, controlling for age, race, income, education, exam center, and physical activity index excluding walking (self-report)

^§MPA=Moderate intensity physical activity. MPA recommendations (≥ 30 minutes/day on 5 or more days each week) corresponds to average of 30*5/7=24.1 minutes/day, within or without MPA bouts. Odds ratios obtained logistic regression models, controlling for age, race, income, education, and exam center, among individuals without accelerometer-measured VPA. Excluded individuals with less than four valid accelerometry days.

Treadmill fitness test duration (seconds) was higher among men and women who actively versus non-actively commute to work in models with adjustment for sociodemographics only (Model 1) and then adding smoking, alcohol, and leisure physical activity excluding walking (Model 2) (Table 4). Similarly, BMI and likelihood of obesity were lower among men who were active (versus non-active) commuters. When analyses were restricted to those living within two miles of their place of work, results were similar (with the exception of women for BMI and obesity).

Table 4.

Association between active commuting^* and fitness, BMI, and obesity at Exam Year 20 (2005–06) of the Coronary Artery Risk Development in Young Adults (CARDIA) Study^†

		Men				Women
	Model (effect measure)	n	mean/% (SE)	Model 1 effect (95% CI)	Model 2 effect (95% CI)	n	mean/% (SE)	Model 1 effect (95% CI)	Model 2 effect (95% CI)
Treadmill time (TT, sec) ‡	(TTactive)- (TTnon-active)	1038	520.3 (136.8)	58.58 (38.67, 78.49)	50.02 (31.45, 68.59)	1250	364.2 (141.4)	34.72 (16.40, 53.04)	28.77 (11.61, 45.92)
BMI§	(BMIactive)/(BMInon-active)	1063	28.7 (5.2)	0.94 (0.92, 0.97)	0.95 (0.92, 0.97)	1295	29.4 (7.3)	1.00 (0.97, 1.03) ¶	1.01 (0.97, 1.04) ¶
Obesity||	logit (OR)	1063	31.2% (1.4)	0.51 (0.34, 0.76)	0.50 (0.33, 0.76)	1295	39.1% (1.4)	0.87 (0.62, 1.22) ¶	0.91 (0.64, 1.29) ¶

^*Active commuting was defined as any walking or biking during the trip from home to work.

^†Model 1 adjusted for age, race, income, education, and exam center. Model 2 adjusted for smoking, alcohol consumption, physical activity index excluding walking (self-report), and Model 1 variables. Bold indicates significant association (p<0.05).

^‡Differences obtained from linear regression of treadmill fitness test duration on active commuting and control variables

^§Ratios obtained by exponentiation of coefficients from linear regression of natural log-transformed BMI on active commuting and control variables.

^||Odds ratios obtained logistic regression of obesity on active commuting and control variables.

^¶When limiting to subjects who reside within two miles of their work location estimated associations were similar across sexes and outcomes, with the exception of women: BMI Model 1 0.93 (0.86, 0.99); BMI Model 2 0.92 (0.86, 0.99); Obesity Model 1: 0.47 (0.22, 1.01), Obesity Model 2: 0.45 (0.20, 1.02)

Table 5 contrasts associations between active commuting and CVD risk factors in two models: Model 1 (adjusting for sociodemographics) and Model 2 (adjusting for Model 1 covariates plus smoking, alcohol, and leisure physical activity). In men, active commuting was inversely associated with triglycerides, diastolic blood pressure and fasting insulin, and positively associated with HDL. Results varied depending upon statistical adjustment, with HDL become non-significant in Model 2. Across all outcomes, statistical significance disappeared with BMI adjustment (results not shown). Active commuting was not statistically associated with any CVD risk biomarkers in women.

Table 5.

Association between active commuting^* and CVD risk biomarkers at Exam Year 20 (2005–06) of the Coronary Artery Risk Development in Young Adults (CARDIA) Study^†

		Male				Female
	Model (effect measure)	n	mean/% (SE)	Model 1 effect (95% CI)	Model 2 effect (95% CI)	n	mean/% (SE)	Model 1 effect (95% CI)	Model 2 effect (95% CI)
HDL‡	(HDL active)/(HDL non -active)	883	47.8 (14.2)	1.05 (1.00, 1.10)	1.03 (0.99, 1.08)	1120	60.1 (16.0)	1.02 (0.98, 1.06)	1.01 (0.97, 1.06)
LDL‡	(LDL active)/(LDL non -active)	862	116.9 (31.7)	0.99 (0.94, 1.04)	0.99 (0.93, 1.04)	1117	108.5 (29.2)	1.01 (0.96, 1.06)	1.01 (0.97, 1.06)
TRIG‡	(TRIG active)/(TRIG non -active)	883	127.0 (97.5)	0.88 (0.80, 0.98)	0.88 (0.80, 0.98)	1120	90.9 (60.1)	1.02 (0.95, 1.11)	1.04 (0.96, 1.12)
Diastolic Blood Pressure (DBP)§	(DBP active)-(DBP non -active)	913	71.6 (9.2)	−1.54 (−3.07, −0.01)	−1.67 (−3.20, −0.15)	1076	69.8 (10.7)	−0.18 (−1.84, 1.47)	−0.15 (−1.81, 1.51)
Systolic Blood Pressure (SBP) §	(SBP active)-(SBP non -active)	913	116.9 (11.3)	−1.39 (−3.29, 0.52)¶	−1.60 (−3.51, 0.32)¶	1076	111.4 (13.7)	0.93 (−1.20, 3.07)	0.74 (−1.39, 2.87)**
Fasting glucose (FG)||	(FG active)/(FG non –active)	965	99.9 (17.4)	0.98 (0.96, 1.00)	0.98 (0.96, 1.01)	1143	94.2 (16.1)	1.00 (0.98, 1.02)	1.00 (0.98, 1.02)
Fasting insulin (FI)||	(FI active)/(FI non -active)	964	14.5 (9.4)	0.84 (0.77, 0.92)	0.86 (0.78, 0.93)	1143	13.4 (8.3)	1.00 (0.92, 1.08)	1.00 (0.93, 1.09)

^*Active commuting was defined as any walking or biking during the trip from home to work.

^†Model 1 adjusted for age, race, income, education, and exam center. Model 2 adjusted for smoking, alcohol consumption, physical activity index excluding walking (self-report), and Model 1 variables. Bold indicates significant association (p<0.05).

^‡Ratios obtained by exponentiation of coefficients from linear regression of natural log-transformed lipid measure on active commuting and control variables. Excludes individuals reporting cholesterol-lowering medications or who fasted less than eight hours prior to blood draw

^§Differences obtained from linear regression of blood pressure measure on active commuting and control variables. Excludes individuals reporting blood pressure-lowering medications

^||Ratios obtained by exponentiation of coefficients from linear regression of natural log-transformed fasting insulin measure on active commuting and control variables.

^¶Excludes individuals reporting diabetes medications or who fasted less than eight hours prior to blood draw.

^**When limiting to subjects who reside within two miles of their work location estimated associations were similar across sexes and outcomes, with the exception of systolic blood pressure: Men, Model 1: 3.92 (−0.74, 8.59), Model 2: 4.40 (−0.24, 9.04); women, significant interaction with physical activity, High physical activity: 2.39 (−0.50, 5.29), Low physical activity:−1.17 (−4.26, 1.92)

Discussion

Few participants in this population-based cohort reported any walking or biking to work. In men, active commuting was inversely associated with body mass, obesity, triglycerides, diastolic blood pressure, and fasting insulin and positively associated with walking, HDL, and fitness. In women, walking and treadmill time were positively associated with active commuting. However, statistical associations between active commuting and all CVD risk biomarkers in men disappeared with adjustment for BMI, suggesting that BMI is a potential mediator between active commuting and CVD risk. Results were similar when restricted to those living within a 2-mile distance from their place of work.

Associations were clearer for men, who had relatively higher rates and distance of active commuting, thus suggesting that efforts to increase active commuting in women may be particularly relevant for increasing overall physical activity. While the association of active commuting with walking behavior and fitness are clear for women, associations between active commuting and measures of CVD risk were less clear for women. The lack of associations for women could be because women have lower levels of active commuting, or they may have lower intensity of activity during active commuting.

While the positive association between walking and CVD risk has been well investigated for leisure walking (e.g., reviews by Hamer and Chida²⁹ and Murphy, et al.³⁰), there is less research on the associations with non-leisure forms of physical activity, such as walking for utilitarian purposes.³¹ A study of Finnish adults observed an inverse association between daily active commuting and ischemic stroke, with highest risk reduction at greater time in active commuting.³² Interestingly, in that study, associations were evident only in the pooled sample of men and women (with BMI and other risk factor adjustment), but not in each separately, indicating relatively modest magnitude of effect. This same research group similarly found reductions in type 2 diabetes,³³ other cardiovascular risk factors,^{34, 35} including reductions in mortality among diabetic men and women³⁶ and hypertensive women³⁷ patients who used active forms of commuting, such as walking or biking. A study in Danish adults observed positive associations between active commuting and high-density lipoprotein cholesterol and negative associations with low-density lipoprotein cholesterol, triglycerides, waist circumference and body mass index..³⁸ There is limited such work in U.S. samples.

The strengths of this study include: extensive CVD risk biomarker data, objective physical activity measures, detailed active commuting data, and additional measures of leisure physical activity. Further, most studies relating walking to CVD risk factors do not adequately control for adiposity and leisure physical activity²⁹ as we have in the present study. Even with these strengths, there are limitations. The CARDIA data are observational in nature and our results do not imply causality. Further, the present study is cross-sectional. Yet, our results suggest that any portion of the commute made by walking or biking is important for maintaining or improving health, regardless of the direction of causation. Unfortunately, the low rates of active transit preclude analyses of dose-response, and thus reduce power to detect effects. Even using the lowest possible threshold (i.e., “any active commuting”) to define active commuting, there were favorable associations with several CVD risk factors in men. Thus, associations could be underestimated due to low variability, and higher levels of active commuting could produce stronger associations with CVD risk factors.

A major limitation is the potential self-selection of active transportation: individuals who are more inclined to be active may be more likely to use active forms of transportation. Indeed, Williams³⁹ has shown that self-selection bias plays a role in the inverse associations between adiposity and walking (leaner individuals selecting to walk greater distances and at higher intensity). Similarly, there is evidence of higher walking among individuals who prefer and live in walkable neighborhoods⁴⁰. However, many of the associations in this study remained after controlling for other forms of physical activity.

We are further limited by self-report commuting data and other lifestyle factors, and cannot completely control for misreporting though non-differential measurement error would tend to bias our results toward the null. While our active commuting measure has face validity and was related to fitness levels, no psychometric evaluation was conducted.

While these data do not fully resolve the role of active commuting in health, they contribute information that adds to current thought that additional active commuting would have several benefits. Walking is a particularly good form of activity to target. Among leisure walkers, walking is the sole source of their leisure physical activity.⁴¹ Indeed, adherence to physical activity recommendations is higher when considering both leisure and non-leisure forms of physical activity.³¹ Further, walking can be integrated into other activities beyond leisure into active transportation or commuting^42–45 and overall lifestyle activity.^{1, 6–8} Public support for policies that encourage active commuting has been shown, particularly for individuals with experience using active commuting and with positive attitudes towards walking and biking.⁴⁶ Intervention research to promote active commuting, reviewed by Ogilvie,⁴⁷ indicates that the majority of such interventions consistently report a net increase in proportion of trips by foot and in walking in general. Further, increasing active commuting will have dual benefits of increasing population health and in reduction of greenhouse gas emissions.⁴⁸ Environmental supports for commuting, such as physical environment^49–51 and sociocultural^{49, 50} factors, have been shown to promote active forms of commuting.

These findings support previous studies of health benefits of leisure-time walking and extend these findings to active commuting. Future investigation into the link between active commuting and health outcomes should address the amount of commuting needed for positive health benefit. There is a major need for development of more precise measures of active commuting. Most importantly, the use of longitudinal designs to address selectivity and reverse causality is strongly encouraged. Similarly, research aimed at unraveling the selectivity in active commuting behaviors and understanding whether those who choose to actively commute are healthier and more active is of major importance.