Walking speed, physical activity, and breast cancer in postmenopausal women

Abstract

Higher self-reported physical activity is associated with lower breast cancer incidence and mortality. Objectively measured timed walking speed, predictive of longevity in older adults, has been associated with ambulatory physical activity in small studies but definitive assessment of the association is lacking. Participants were a subset of 14 719 postmenopausal women in the Women’s Health Initiative study who, at entry, had 10 m, timed walking speed determined. After 12.4 years [mean (SD) (3.5)] follow-up, 762 invasive breast cancers were diagnosed in this group. In addition, 8162 of these women self-reported physical activity. Simple linear regression was used to examine the relationship between timed walking speed and self-reported physical activity. A Cox proportional hazard model was used to estimate age-adjusted hazard ratios and 95% confidence intervals for the association between timed walking speed and invasive breast cancer incidence. Although a linear regression model for self-reported physical activity [log metabolic equivalent task (MET) h/week] versus 10 m, timed walking speed had a statistically significant slope (coefficient=0.03, P<0.0001, correlation=0.20), the magnitude of the relationship was not clinically useful. Timed walking speed quintile was not associated with breast cancer incidence in age-adjusted or multivariant analyses (P for trend=0.60). Timed walking speed was not associated with self-reported physical activity in a clinically useful manner or with breast cancer incidence. Our findings do not support use of timed walking speed as an objective surrogate for self-reported physical activity.

Introduction

Observational studies have shown statistically significant associations between higher levels of self-reported physical activity and lower breast cancer incidence and breast cancer mortality (^{McTiernan et al., 2010}). Timed walking speed represents an easily determined, objective physical performance measure that has been associated with lower chronic disease risk (^{McGinn et al., 2008}) and longer survival in older individuals (^{Cooper et al., 2010; Studenski et al., 2011}). Although timed walking speed has been associated with ambulatory physical activity in older women with hip disorders (^{Soichiro et al., 2004}), this association has not been confirmed in larger populations of otherwise healthy postmenopausal women. Taken together, these results suggested the hypothesis that, if timed walking speed could be significantly associated with self-reported physical activity, it could represent an objective physical activity surrogate, which would then be significantly associated with breast cancer incidence. We evaluated this concept in a subgroup of postmenopausal women participating in the Women’s Health Initiative (WHI) clinical trials and observational study.

Methods

The study design and conduct of the WHI study have been previously described (^{Anderson et al., 2003}). Briefly, between 1993 and 1998, postmenopausal women, aged 50–79 years with anticipated 3 years survival, were eligible for and entered either one or more randomized, controlled clinical trials evaluating strategies for chronic disease risk reduction (menopausal hormone therapy trials evaluating estrogen alone or estrogen plus progestin or a dietary modification trial) (n=68 132) or an observational study cohort (n=96 676). Eligible for the current study were a subset of 14 719 WHI participants who had timed walking speed determined at study entry with a 10-m walk test. Additional analyses were performed for 8162 women who had information not only on walking speed but also provided self-report of ambulatory physical activity at entry and were able to walk outside the home for more than 10 min without stopping. Women with prior breast cancer or implausible walking speeds were excluded.

The WHI studies had institutional review board approval from all participating institutions and written informed consent was obtained from all participants. The WHI study is registered with clincialtrials.gov, NCT000000611.

The timed 10-m walking speed was measured using a 10 m straight line walk, with a static start, timed by a stop watch and reported as velocity (m/s). Participants were told to use their ‘usual walking speed’ and no verbal encouragement during the test was provided (^{McGinn et al., 2008}). For physical activity assessment, self-reported information on current walking frequency, duration, and speed and frequency and duration of strenuous, moderate, and mild recreational physical activity was collected. Minutes of activity were multiplied by frequency separately for strenuous, moderate, and mild recreational physical activity, and three intensities of walking. Metabolic equivalent task (MET) values were assigned as previously reported (^{McTiernan et al., 2003}) and used to generate MET h/week categories. Clinical outcomes were determined at annual clinic visits and semiannual contracts. Breast cancers were verified by central medical record and pathology report review (^{McTiernan et al., 2003}).

Statistical analysis

Simple linear regression was used to evaluate the relationship between timed walking speed and log transformed self-reported moderate to vigorous physical activity from walking and recreation. Cox proportional hazards age-adjusted and multivariant adjusted models were used to estimate hazard ratios and 95% confidence intervals for invasive breast cancer incidence by quintiles of timed walking speed (Table 1). All P-values are two sided. All analyses were performed using SAS for Windows, v 9.2 (Cary, North Carolina, USA).

Table 1: Association between timed walking speed quintiles and invasive breast cancer in 14 719 postmenopausal women

Results

Women with faster walking speed were younger, had better self-reported health status and were more commonly White (data not shown). The relationship between total energy expenditure from self-reported physical activity from walking and recreation (total METs) was compared with timed, 10-m walking speed in a regression model (Fig. 1). The regression line for walking speed (m/s)=1.08+0.03×log (energy expenditure from physical activity, in MET h/week plus 0.05); P<0.0001, adjusted R²=0.04, correlation=0.20. Although the slope of the relationship between timed walking speed and reported physical activity was statistically significant, no clinically useful association between the two variables emerged as substantial overlap was observed between walking speed values and reported physical activity.

Fig. 1: Timed walking speed vs. energy expenditure from self-reported physical activity in 8162 postmenopausal women. ^aEnergy expenditure is reported on the natural log scale. A value of 9 MET h/week would correspond to a value of 2.2 on the natural log scale. The regression line is: walking speed (m/s)=1.08+0.03×log (energy expenditure from physical activity, MET h/week+0.05); P<0.0001, adjusted R ²=0.04, correlation=0.20.

After 12.4 years (mean; SD 3.5) of median follow-up, 762 invasive breast cancers were diagnosed among 14 719 women who had timed walking speed at study entry. The association between subsequent invasive breast cancer incidence and initial timed walking speed quintiles is outlined in both age-adjusted and multivariant adjusted models as shown in Table 1. As observed, breast cancer incidence was not associated with walking speed quintile.

Discussion

In a subgroup of 8162 postmenopausal participants in the WHI, timed walking speed did not have a clinically meaningful relationship with self-reported physical activity energy expenditure. In addition, timed walking speed in 14 719 WHI participants was not significantly associated with breast cancer incidence.

Although increasing evidence supports an association between higher physical activity and lower breast cancer incidence and mortality, these findings are based on self-reported physical activity (^{McTiernan et al., 2003, 2010; Irwin et al., 2011}). Although the self-reported physical activity measures used in such studies have been validated, a recent review of direct, as compared with self-report, measures for assessing physical activity in adults found only moderate degrees of correlation (^{Prince et al., 2008}). For example, the WHI self-reported measure of physical activity had a 0.73 correlation with results from women based on accelerometer findings (^{Johnson-Kozlow et al., 2007}). Corroboration of self-reports with an objective, surrogate measure of physical activity would provide additional support for the favorable association seen between higher levels of physical activity and lower breast cancer incidence in observational study reports.

The reported association between timed walking speed and clinical outcomes, including a statistically significantly lower stroke risk observed in these WHI participants (^{McGinn et al., 2008}), suggested that timed walking speed could potentially represent a surrogate for self-reported physical activity related to walking and recreation and provide an easy to implement, objective test for use in clinical studies and practice. Our findings, however, do not support such an association. It seems women who can walk fast do not more commonly choose to exercise more frequently. Walking speed’s association with stroke perhaps reflects instead central or peripheral neurologic dysfunction.

Walking speed has been assessed by several methods including the 10-m walk test used in the current report, walking speed calculated from 6-min walk test distance, and longer duration walk assessments. At this time evidence is insufficient to determine whether any of the methods represent a reliable test for physical activity assessment (^{Johnson-Kozlow et al., 2007; Prince et al., 2008}).

Study strengths include the size of the well-characterized study population, central training of clinical staff and quality assurance visits to ensure uniform data collection across centers, and central confirmation of breast cancers. A limitation was the measurement of walking speed only at one period.

We conclude that timed walking speed is not a useful surrogate for self-reported physical activity and is not significantly associated with breast cancer incidence in postmenopausal women.

Acknowledgements

Conflicts of interest

There are no conflicts of interest.