Increased and decreased bicycling and weight change. A, Increased bicycling: includes only women who did not initially bicycle (0 min/d) at baseline (1989) (n = 9556). The figure reflects the slope of weight change if women remained in the nonbicycling category in 2005 (reference) or if they increased their bicycling in 2005. B, Decreased bicycling: includes only women who initially bicycled for more than 15 min/d at baseline (1989) (n = 1506). The figure reflects the slope of weight change if women remained in the high bicycling activity category in 2005 (reference) or if they decreased their bicycling in 2005. Error bars represent the standard error for weight change. *Adjusted for 1989 baseline age, weight, plus the covariates listed in Table 2 footnote c.
Lusk AC, Mekary RA, Feskanich D, Willett WC. Bicycle Riding, Walking, and Weight Gain in Premenopausal Women. Arch Intern Med. 2010;170(12):1050-1056. doi:10.1001/archinternmed.2010.171
Copyright 2010 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2010
To our knowledge, research has not been conducted on bicycle riding and weight control in comparison with walking. Our objective was to assess the association between bicycle riding and weight control in premenopausal women.
This was a 16-year follow-up study of 18 414 women in the Nurses' Health Study II. Weight change between 1989 and 2005 was the primary outcome, and the odds of gaining more than 5% of baseline body weight by 2005 was the secondary outcome.
At baseline, only 39% of participants walked briskly, while only 1.2% bicycled for more than 30 min/d. For a 30-min/d increase in activity between 1989 and 2005, weight gain was significantly less for brisk walking (−1.81 kg; 95% confidence interval [CI], −2.05 to −1.56 kg), bicycling (−1.59 kg; 95% CI, −2.09 to −1.08 kg), and other activities (−1.45 kg; 95% CI, −1.66 to −1.24 kg) but not for slow walking (+0.06 kg; 95% CI, −0.22 to 0.35 kg). Women who reported no bicycling in 1989 and increased to as little as 5 min/d in 2005 gained less weight (−0.74 kg; 95% CI, −1.41 to −0.07 kg; P value for trend, <.01) than those who remained nonbikers. Normal-weight women who bicycled more than 4 h/wk in 2005 had a lower odds of gaining more than 5% of their baseline body weight (odds ratio, 0.74; 95% CI, 0.56 to 0.98) compared with those who reported no bicycling; overweight and obese women had a lower odds at 2 to 3 h/wk (odds ratio, 0.54; 95% CI, 0.34 to 0.86).
Bicycling, similar to brisk walking, is associated with less weight gain and an inverse dose-response relationship exists, especially among overweight and obese women. Future research should focus on brisk walking and greater time spent bicycling.
In 1995, the Centers for Disease Control and Prevention and the American College of Sports Medicine recommended that every US adult accumulate 30 minutes or more of moderate-intensity activity on most, preferably all, days of the week to address rising obesity and improve health1; similar recommendations were issued in 2008.2 Nevertheless, 66% of adults are overweight or obese, 16% of children and adolescents are overweight, and 34% of children and adolescents are at risk of being overweight in the United States.3
Extensive research has been conducted on walking in relation to weight control,4- 6 but less has examined bicycling.7 Of the studies conducted on bicycling, many have included primarily men,8- 11 combined walking and bicycling,12- 15 and been conducted in countries with different bicycle environments. In the United States, roads, lanes, and shared-use paths are the recommended facilities,16- 18 while in the Netherlands, facilities include barrier-protected and bicycle-exclusive cycle tracks.19 In the Netherlands, where bicycle riding is actively supported by an extensive network of bicycle exclusive cycle tracks, 27% of the population bicycles. Of this population of bicyclists, 55% are female.20 In contrast, in the United States, only 0.5% of the commuting population 16 years and older bicycles, of which only 23% are female.21 In 2007, only 48% of the US population engaged in the recommended levels of physical activity22 compared with 64% in the Netherlands.23 While based on different measures and not demonstrating causality, the obesity prevalence of men and women is 23.9% in the United States and 8.1% in the Netherlands.24
In previous articles, we reported that brisk walking was beneficial for the prevention of weight gain among women with normal weight25 and for maintenance after weight loss,26 whereas nonbrisk walking (ie, slow walking) was of little benefit except for women with excessive weight. We are extending this research to examine bicycle riding in association with weight gain in premenopausal women using data from the Nurses' Health Study II.
The Nurses' Health Study II (NHSII) is an ongoing prospective study of 116 608 US female nurses aged 25 to 42 years in 1989, who responded to a mailed questionnaire about their medical history, lifestyle, and health-related behaviors. Follow-up questionnaires have been mailed biennially. Body weight is assessed on every questionnaire, and physical activity (PA) has been assessed periodically, including in the 1989 and 2005 questionnaires. A food frequency questionnaire has been included every 4 years starting 1991. The overall response rate has been approximately 90% over the years of follow-up.
This investigation included the NHSII women who were premenopausal through 2005 (n = 56 716). From these, we excluded women who were pregnant or lactating within 12 months of reporting weight (n = 15 728); did not report their PA, walking pace, and weight in 1989 and 2005 (n = 2820); reported extreme weight values (<37 kg or >182 kg) (n = 1291); reported extreme weight changes (>40 kg lost or >60 kg gained) (n = 15); had extreme baseline body mass index (BMI <15 or >45 [calculated as weight in kilograms divided by height in meters squared]) (n = 670); had physical chronic conditions impairing exercise (n = 3380); were unable to walk in 1989 or 2005 (n = 106); or reported more than 240 min/d of total PA (n = 2285). Further exclusions included reporting myocardial infarction, stroke, angina, or diabetes (n = 8759) or cancer (n = 3248) through 2005. After these exclusions, 18 414 healthy premenopausal women remained in the analysis.
Participants were asked in 1989 and 2005 to report the average time spent per week in the previous year in each of the following recreational activities: walking or hiking outdoors, jogging (<10 min/mile), running (≥10 min/mile), bicycling (including stationary machine), calisthenics/aerobics/aerobic dance/rowing machine, tennis/squash/racquetball, lap swimming, or other aerobic activity (eg, lawn mowing). For each activity, women chose 1 of 10 duration categories, which ranged from 0 to 11 h/wk or more. For walking, women reported their usual pace in 1989 and 2005: easy (<2.0 mph), average (2.0-2.9 mph), brisk (3.0-3.9 mph), very brisk (≥4.0 mph), or unable to walk. For simplicity slow and average walking paces were collapsed under “slow walking” and brisk and very brisk walking paces, under “brisk walking.” Women were also asked in 1989 and 2005 to report the average number of flights of stairs climbed daily. Stair climbing (minutes per day) was then estimated. Total discretionary activity (minutes per day) was the sum of the duration reported in each of the 9 activities reported at baseline and in 2005. The questionnaire has been validated in a random representative sample of NHSII participants (n = 147).27 When we used past-week activity recalls and 7-day activity diaries as the referent methods, the correlation (r) between activity reported on questionnaires and that of recalls was 0.79 and that reported on diaries was 0.62.
Duration of sitting at home was used as a measure of sedentary behavior (ie, inactivity). In 1989, total sitting at home was collected with one general question (“How many hours per week do you spend sitting at home?”), which was later expanded to 2 specific questions in 2005 (“How many hours per week do you spend: (1) Sitting at home while watching TV/VCR? (2) Other sitting at home (eg, reading, meal times, and at desk)?”). Because inactivity and, in particular, TV watching has been associated with obesity,28 which might confound the associations with PA, total sitting at home was included in our models.
Our primary outcome was defined as 16-year weight change between 1989 and 2005. The secondary outcome was defined as gaining more than 5% of weight from baseline in 1989 to the 16-year follow-up. Weight and height were assessed at baseline (1989), and weight was assessed on each follow-up questionnaire. Self-reported weight was strongly correlated in adults with measured weight (r = 0.97).29 Baseline BMI was calculated from self-reported baseline height and weight.
Diet components that have previously been observed to be predictive of weight gain30- 33 were included in the analysis. Using a validated food frequency questionnaire, we included 1991 and 2003 intakes of sugar-sweetened beverages, energy-adjusted trans fats, and energy-adjusted dietary fiber in the model to account for changes in these covariates. Alcohol consumption at baseline (1989) and 2003 was also included. The other risk factors for weight change in the multivariate models were smoking status (never, past, or current), oral contraceptive use (never, past, or current), parity (nulliparous, 1, 2, or ≥3 births), and antidepressant intake (never, past, or current) at baseline and in 2005.
In the first set of analyses, multiple linear regression was used to assess the relation between change in total activity and change in weight over 16 years. Weight change was modeled as weight in 2005 as the outcome and 1989 baseline weight as a predictor. Both 1989 and 2005 activity were predictors in the model, therefore the β for 2005 activity represented 16-year change. In a second model, total activity was partitioned among slow walking, brisk walking, bicycling, and other activities (the other 7 assessed activities). Total activity and specific activities were modeled per an increase of 30 min/d, while holding the levels of other activities and covariates constant. We also conducted a stratified analysis by baseline weight status (underweight and normal weight [BMI <25] vs overweight or obese [BMI ≥25]) to assess effect modification. There were no important differences across the age strata; thus, the results were not stratified by age.
In addition, bicycling in 1989 and 2005 was categorized into 4 groups (0, ≤5, >5-15, and >15 min/d). The top and bottom categories in 1989 were cross-classified with the 4 categories in 2005 and were assessed in 2 other linear regression models.
To assess a dose-response relationship between bicycling and weight gain, logistic regression was used to estimate the odds ratio (OR) of gaining more than 5% of baseline body weight at the 16-year follow-up for categories of time spent bicycling at the end of this follow-up period (2005). Bicycling in 2005 was categorized into 5 groups using the collapsed categories from the questionnaire (0, 1-59 min/wk, 1-1.5 h/wk, 2-3 h/wk, and ≥4 h/wk). Mean (SE) weight changes were calculated for each of the bicycling categories.
The 1989 baseline characteristics of the study population by levels of slow walking, brisk walking, and bicycling are displayed (Table 1). Of the women, 50% reported they spent time walking slowly, 39% reported they spent time walking briskly, and 48% reported they spent time bicycling. For our sample, the mean time spent on slow walking was 7.8 min/d, brisk walking 8.5 min/d, and bicycling 4.6 min/d.
In 2005, on average, participants reported spending more time walking briskly (9.0 min/d), some time walking slowly (5.9 min/d), and the least amount of time bicycling (2.5 min/d). The mean time spent sitting at home was 5 times (153 min/d) as much as total time spent in total activity (30.5 min/d), especially if they were overweight (175 min/d). Compared with lean women, overweight women spent more time walking slowly (7.6 vs 5.5 min/d) and less time walking briskly (5.4 vs 10 min/d). Even though most women did not spend considerable time bicycling, on average, overweight women (2 min/d) and lean women (2.7 min/d) bicycled comparable amounts of time.
Between 1989 and 2005, all women gained a mean (SD) 9.3 (9.7) kg. For a BMI less than 25, women gained a mean (SD) 8.4 (7.9) kg; for a BMI of 25 or greater, women gained a mean (SD) 12.6 (13.4) kg. In parallel, women decreased the mean (SD) time spent on their total discretionary activity by 8.6 (43.3) min/d, and more specifically, decreased the mean (SD) times spent on slow walking by 1.9 (21.8) and bicycling by 2.1 (12.2) min/d. Only the mean (SD) time spent on brisk walking slightly increased by 0.5 (23.1) min/d.
In linear regression models using change in PA as a continuous variable and controlling for baseline activity and other risk factors, a 30-min/d increase in total discretionary activity between 1989 and 2005 was associated with less weight gain (−1.31 kg, 95% CI, −1.44 to −1.18), whereas a 30-min/d increase in total time sitting at home was associated with greater weight gain (+0.21 kg, 95% CI, +0.18 to +0.25) (Table 2). When we partitioned the change in total discretionary activity, the estimated weight gain for a 30-min/d increase in all women was significantly less for brisk walking (−1.81 kg; 95% CI, −2.05 to −1.56), bicycling (−1.59 kg; 95% CI, −2.09 to −1.08), and for other activities (−1.45 kg; 95% CI, −1.66 to −1.24), whereas an increase in slow walking was not associated with less weight gain (+0.06 kg; 95% CI, −0.22 to +0.35). In our cohort, only a few women (1.2%) actually attained this increase of 30 min/d in bicycling. The benefits of brisk walking, bicycling, and other activities were significantly stronger among overweight and obese women (BMI ≥25) compared with lean women (BMI <25) (P value for interaction, <.01 for all), whereas slow walking continued to show no benefit even among overweight and obese women (−0.60 kg, 95% CI, −1.35 to +0.16) (P value for interaction, .74).
In women who did not bicycle in 1989 and increased their bicycling in 2005, less weight gain was seen even for 5-min/d or less increase in bicycling (−0.74 kg; 95% CI, −1.41 to −0.07), and even lower weight gain was seen with greater duration of bicycling (P value for trend, <.001) (Figure, A). In comparison, women who initially bicycled for more than 15 min/d at baseline and who decreased their bicycling time in 2005 to more than 5 to 15 min/d gained more weight (+2.13 kg; 95% CI, +0.35 to +3.92), which accrued with more reduction in bicycling (P value for trend, .005) (Figure, B).
To assess if a dose-response relationship was evident, we re-examined change in activity as a categorical variable, with 2005 activity as the predictor controlled for 1989 baseline activity. Compared with women who did not bicycle in 2005, those who engaged in bicycling 4 h/wk or more were less likely to gain weight (OR, 0.71; 95% CI, 0.55 to 0.93) after controlling for other covariates (Table 3). There was a significant inverse dose-response relationship between increased time spent bicycling in 2005 and odds of weight gain (P value for trend, <.001). The results appeared to be stronger in women with excess baseline weight compared with lean women. The mean [SE] weight gain was the smallest (5.5 [0.4] kg) in women who engaged in 4 h/wk or more of bicycling compared with women who bicycled for less time.
In this large 16-year prospective cohort study of premenopausal women, an increase in time spent bicycling was associated with a significantly lower change in weight, and this relationship was stronger among women with excess weight. For women who did not bicycle in 1989, less weight gain was evident for even a small increase to 5 min/d or less in 2005. Conversely, women who bicycled for 15 min/d or more in 1989 were at a higher risk of weight gain if they decreased or stopped bicycling in 2005.
Although brisk walking, unlike slow walking, has been suggested as a beneficial PA, only 39% of the women reported that they walked briskly at baseline, while 50% reported they walked slowly. Walking briskly can be difficult, especially for women who are overweight or obese or those with arthritis or other disabilities.34 Overweight women spent half the time walking briskly (5.4 min/d) compared with lean women (10 min/d), while overweight and lean women spent comparable times bicycling (2.0 and 2.7 min/d, respectively).
Unlike discretionary gym time, bicycling could replace time spent in a car for necessary travel of some distance to work, shops, or school as activities of daily living. Bicycling could then be an unconscious form of exercise because the trip's destination, and not the exercise, could be the goal.35
Research on bicycling in addition to walking is relatively new.36 Although bicycling was found to be inversely associated with weight gain, fewer studies have included women, many studies have combined walking and bicycling, and several studies have been conducted in countries with bicycle environments different from the United States. Our findings agree with Littman et al,37 who conducted a study in western Washington State and found that fast bicycling, and not slow walking, in nonobese men between the ages of 53 and 57 years was associated with weight attenuation. Though the age range was similar to the women in our study, this finding was only for lean men. In France and Northern Ireland, Wagner et al38 studied 8865 men aged 50 to 59 years and found that men who walked or bicycled to work for more than 30 min/d had a lower BMI of 0.3, a smaller waist circumference of 1 cm, and BMI change of 0.06 compared with men who did not walk or bicycle to work. While this study agreed with our positive association between brisk walking, bicycling, and obesity reduction, the study only included men and walking was not differentiated between slow and brisk, even though the walking time was more than 30 min/d.
In an Australian study that included men (n = 3810) and women (n = 3022), Wen and Rissel39 found that men who bicycled to work (n = 93) were significantly less likely to be overweight and obese (39.8%) compared with those who drove to work (60.8%), but these inverse relationships were not evident in the women studied (n = 10), which could be owing to the lack of power. In men (n = 195) and women (n = 216), walking to work was not inversely associated with overweight or obesity, and the authors suggested that this may be because walking was not sufficiently vigorous or the distance was not great enough to affect weight. Their findings concur with ours. Bicycling by men in that study was associated with less weight gain.
Hemmingsson et al36 conducted a randomized trial in abdominally obese women (age 30-60 years) with success defined as bicycling 2 km/d or more (primary) or walking 10 000 steps per day (secondary) for an 18-month duration. The intervention group members were given bicycles and followed a PA prescription of walking or bicycling, while the control group members were given program support and pedometers. The intervention group was more likely to bicycle than the control group (38.7% vs 8.9%), while both groups had the same compliance rates for walking. Both groups achieved similar waist reductions (−2.1 cm and −2.6 cm, respectively; P = .72). Though none of the participants reported bicycling at baseline, 29% of the women bicycled as part of the intervention group. In our study, overweight and obese women bicycled for approximately the same amount of time as lean women but did not walk briskly for the same amount of time as lean women. The study conducted in Sweden supports our finding that overweight and obese women will bicycle and can then achieve weight control.
While our research found significant associations between bicycling and less weight gain, dose-response associations, and greater benefits for overweight and obese women, as previously shown in other studies,25,26,40 our research also revealed how few women bicycled for a substantial period. Though 48% indicated that they bicycled and may have been on a stationary machine, they bicycled on average for only 2.5 min/d. Of these bicyclists, only 13% bicycled for 10 min/d or more at baseline and only 1.2% bicycled for 30 min/d or longer. Perhaps more women did not bicycle for longer periods because of a lack of bicycle environments comfortable to them and an emphasis in the United States on walking. Compared with bicycling,7 multiple studies have been conducted on walking, described as the “near perfect form of exercise.”5 Perhaps walking has been identified as beneficial because it has been compared, in the US car-centric nation, with not walking.41,42 The research that has been conducted on bicycling in the United States has included bicycle environments based on the American Association of State Highway and Transportation Officials (AASHTO) bicycle guidelines18 and the Federal Highway Administration (FHWA) teachings,43 which favor roads, lanes, and shared-use paths. The guidelines have been based primarily on the perceptions of male bicyclists perhaps because more men bicycle,21 and the studies have thus included a higher percentage of male respondents.8- 10 Research conducted in Minnesota, Canada, and Australia has suggested that women have a greater preference for separation from vehicle traffic.44- 47
Compared with the United States, the Dutch use 50- to 60-year-old male and female bicyclists as the design models in their bicycle facility guidelines,19(pp39-56) and these guidelines detail bicycle-exclusive cycle tracks and cycle track intersection and curb cut treatments.19,48 On Dutch roads with car speeds of 80 km/h (49.71 mph), a separate cycle track is recommended parallel to the road, and on Dutch roads, which bicyclists share with cars, the recommended car speed is 30 km/h (18.6 mph).19 Though the Netherlands might be acculturated as a bicycle country, Canada's Technical Handbook of Bikeway Design handbook features cycle tracks and other European bicycle facilities.49
As a result of being car-centric or being overly cautious about trying different bicycle facilities, in the United States, 9% of the population walks for commuting, whereas only 0.5% commutes by bicycle.24 In the Netherlands, 22% of the population walks and 27% commutes by bicycle.20 Individuals who have available comfortable bicycle infrastructure may still require individual determinants, such as self-efficacy and interest in bicycling, but they have, as a start, the infrastructure. Notably, if individuals have comfortable bicycle environments and they then bicycle, they are less likely to have medical risk factors and likely to have lower overall mortality.50- 53
There are several limitations to this study. First, the sample was not a random sample from the United States and the women in the study were better educated (all nurses) and primarily white. Second, our PA measurements are inevitably imperfect, which would tend to attenuate the association between PA and weight control. While objective measures of PA may have been desirable, the validity of our self-reported PA questions has been documented.27,54 Third, the intensity of bicycling was not recorded, and the assessment did not discriminate between regular biking and riding a stationary bicycle. Fourth, only recreational PA was assessed and not total PA, (ie, not time spent in activity doing housework or nursing work). Finally, although the different activities were analyzed in terms of minutes per day, information about the frequency of bicycling is lacking from our questionnaires such that we are unable to determine if there is any difference between someone engaging in 70 min/wk all at once vs someone bicycling 10 minutes every day.
These limitations notwithstanding, there are several strengths. First, this research included a large sample size followed with repeated measurements over 16 years. Second, women with conditions affecting weight such as pregnancy-related or postpartum weight gain were excluded. Finally, information on a wide variety of potentially confounding behavioral and demographic variables was collected at every assessment, which allowed us to assess different activity types and weight change associations independent of these potential confounders.
In conclusion, bicycling, like brisk walking, is associated with reduced weight gain in premenopausal women, especially among overweight and obese women. In the United States, brisk walking should not only be encouraged, but additional research should also be conducted to determine which bicycle environments might be preferred by the largest percentage of the population, as in the Netherlands. If facilities were designed based on women's requests, the outcome might lead to bicycle facilities in the United States that are comfortable to more people and facilitate greater weight control.
Correspondence: Anne C. Lusk, PhD, Department of Nutrition, Harvard School of Public Health, 655 Huntington Ave, Bldg II, Room 314, Boston, MA 02115 (AnneLusk@hsph.harvard.edu).
Accepted for Publication: December 16, 2009.
Author Contributions: Dr Mekary had full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of data analysis. Drs Lusk and Mekary contributed equally to the study. Study concept and design: Lusk and Mekary. Acquisition of data: Mekary. Analysis and interpretation of data: Lusk, Mekary, Feskanich, and Willett. Drafting of the manuscript: Lusk. Critical revision of the manuscript for important intellectual content: Lusk, Mekary, Feskanich, and Willett. Statistical analysis: Mekary, Feskanich, and Willett. Administrative, technical, and material support: Lusk. Study supervision: Lusk.
Financial Disclosure: None reported.
Funding/Support: Dr Lusk is supported by Ruth L. Kirschstein National Research Service Award F32 HL083639 from the National Institutes for Health, National Heart, Lung, and Blood Institute. Dr Mekary is supported by grant R25 CA098566 from the National Institutes for Health. The Nurses' Health Study II: Risk Factors for Breast Cancer in Younger Nurses was supported by grant R01CA050385 from the National Cancer Institute.
Role of the Sponsors: The funding sources had no role in the study conduct and analysis and are not responsible for the views expressed herein.
Additional Contributions: Stephanie Chiuve, ScD, provided insightful comments on this topic. Lisa Li, MD, MPH, conducted the SAS program review.