Background
Walking is associated with better mental and physical health and reduced mortality [
1,
2] and, when used for transport, with reduced air and noise pollution [
3]. In contrast to other forms of physical activity, walking has the advantage of being accessible to most people. For these reasons, promotion of walking has become more prominent in public health campaigns [
4‐
6]. With rare exceptions [
7,
8], the possibility of gender differences in walking practices has largely been ignored, but an understanding of whether, how and why walking practices differ between men and women would help inform such campaigns.
Within the widely used ecological model of health behaviour, gender is usually identified as a personal characteristic that may modify, at the individual level, the impact of wider social and environmental influences on behaviours such as walking [
9,
10]. However, gender is social construct with greater power than this [
8,
11]. Gender codes the body, structures social relations (e.g. the division of labour and responsibilities) and affects access to resources [
12]. Transport and exercise related activities, technologies such as the car, and environments such as ‘the street’ also have gendered meanings [
12]. As a consequence, daily mobility is powerfully shaped by gender [
12], as exemplified in the consistent finding in countries such as the USA and Germany that women make, on average, shorter trips than men (partly because they tend to work closer to home than men [
13,
14]). Gender differences in the types of physical activity undertaken for leisure are also widely recognised, for example more men than women participate in team sports in the USA and across the European Union [
15,
16]. We predict that walking, which is undertaken both for transport and for leisure, will also be gendered.
The aim of this systematic review is to assess the current evidence on gender differences in walking in high income countries. Because gender is routinely used to organise data presentation, and is often included in multivariate models in studies examining other (usually environmental) determinants of walking, data on gender differences in walking are available. We hypothesised that there are gender differences in participation in walking for leisure, for transport, and in total walking. We also set out to examine whether gender differences change across the life-course. Both walking practices and the impact of gender on everyday life vary across societies and we restricted the review to high income countries to reduce such variation, while remaining alert to the likely relevance of geographical variation in gender differences.
Methods
Eligibility criteria
Studies were eligible if they met the following inclusion criteria: 1) provides quantitative data on both men’s and women’s walking in everyday life in a form allowing assessment of effect size; 2) participants at least 18 years old; 3) published 1995-2015 reporting data collected in 1990 or later (to limit variation due to changes in gender roles and walking practices over time); 4) written in English 5) study setting must be a high-income country as defined by the World Bank (
http://www.data.worldbank.org/about/country-and-lending-groups#High_income). Exclusion criteria were: 1) studies including only people with certain characteristics, e.g. particular ethnic groups, employed people, patients (other than groups defined by age or ability to walk); 2) inclusion only of people who engage in walking or in particular types of walking; 3) intervention-based studies unless pre-intervention data were available. As we were interested only in ‘walks’ rather than total number of steps taken, we did not include studies that assessed only number of steps.
A search of the literature was conducted in February 2016 using Web of Science Core Collections, PubMed, and the Transportation Research International Documentation (TRID) databases. The search strategy included a combination of terms for walking (walk, walking, pedestrian*, active travel) and adults (men AND women, gender*, adult, adults), with exclusion terms for certain types of studies (e.g., clinical trials, reviews). The grey literature (
http://www.opengrey.eu,
http://www.greylit.org), the reference lists of all included papers, and the archives of the
Journal of Physical Activity and Health and the
International Journal of Behavioral Nutrition and Physical Activity were also searched.
Study selection
Study selection occurred in two phases. First, titles and abstracts of studies were screened for relevance. This task was shared between the authors, and agreement between authors was checked in an initial sample of 50 articles to ensure consistency. Articles with relevant abstracts were divided between the authors and the full text was consulted to determine eligibility. As with abstracts, agreement between authors was checked at the beginning. All full texts deemed eligible by one author were confirmed by the other author. In cases of disagreement between authors, the differences were discussed until a joint decision was made. Where more than one paper used the same dataset or survey to report on the same type of walking (leisure, travel or total), a joint decision about which paper to include in the review was made, with preference given to data from more recent survey years.
Quality assessment
The quality of each study was assessed taking the Critical Appraisal Skills Programme (CASP) as a starting point. We assessed the quality of the study design including recruitment strategy, the use of appropriate methodology to assess walking, correct use of statistical models where relevant, and the clarity and completeness of reporting of findings. This involved assessing risk of bias relevant to gender having been introduced in the recruitment strategy or in assessment of walking. We applied these criteria specifically to data on gender differences in walking. Some studies used methods that were appropriate for their research aims but did not provide good quality data on gender differences and were therefore excluded. The quality of each study was assessed independently and confirmed by both researchers. Discrepant assessments were discussed until agreement was reached.
For all articles deemed eligible for inclusion in the review, we extracted data on location, date of data collection, methodology, characteristics of sample, and results for gender differences in leisure, transportation, or total walking. Where participation in walking was reported in more than one way in a paper, we extracted results for the variable closest to the prevalence of any walking in a week. Data were extracted independently by one author and checked for each paper by the other author.
Synthesis of results
The findings of this review are presented as a numerical and textual summary, rather than through meta-analysis, due to the wide variation in methodology and measures in the included studies, and also because gender roles may vary across societies.
For prevalence data, odds ratios and 95% confidence intervals comparing the proportion of women who walked with the proportion of men were taken from the paper or, if necessary, calculated. An odds ratio of above 1 indicates that more women walked than men and below 1 that fewer women walked than men. Unadjusted odds ratios were preferred; if only adjusted odds ratios were given in the original papers and it was possible to calculate unadjusted odds ratios, we did so. Forest plots were constructed including studies that reported the prevalence of up to 30 min of walking over a specified time period. Logarithmic odds ratios were plotted on a linear scale so that effect sizes were visually equivalent above and below 1. Where papers reported data separately for different age groups, we plotted these odds ratios in a separate plot. Odds ratios for gender differences not included in the forest plots are given in the text in the form OR (95% confidence intervals). The results of tests for gender differences for other outcomes are also given in the text.
In the synthesis, studies are cited using the first author’s name for identification and giving the location of the study and age range of participants, to facilitate interpretation.
Discussion
This review reveals that women report a higher prevalence than men of walking for leisure, for exercise, and for fun when all ages are considered together. This finding was largely consistent across countries. However, an international study found no gender difference in duration of walking for leisure in adults aged between 20 and 65. Thus it is possible that while more women than men walk for leisure, when men walk for leisure they walk further than women. There appears to be no gender difference in the prevalence of dog-walking. Our most striking finding is a consistent pattern whereby more women than men walk for leisure at young adult ages, a difference that progressively declines with age and then reverses, until in the oldest age groups (aged 60-70 and above) more men than women walked for leisure.
Walking for leisure is an activity that women can undertake with children [
53] and it is possible that child-care plays a role in the relatively high levels of walking for leisure in younger women. In many high-income countries more of young and middle-aged women’s leisure time than of men’s leisure time is combined with unpaid work, usually in the form of child care, as shown by multinational time-use data from a subset of ‘Modern Western’ countries [
54]. The fact that walking requires less confidence in the body’s capacity than most other forms of leisure physical activities may also encourage walking for leisure amongst women [
55].
The decrease in gender difference in participation in walking for leisure across younger and middle-aged groups derives mainly from a gradual increase in participation in walking for leisure with age by men [
22,
25,
31,
32,
47]. Young men’s relatively high levels of participation in sports and exercise decline with age, as reported for the UK [
56] and the USA [
57], and it is possible that men adopt walking for leisure as a replacement for more vigorous activities as they get older.
In the oldest age groups, the proportion of men walking for leisure declines, but the proportion of women walking for leisure declines more [
22,
25,
31,
32,
47]. This pattern may reflect differences in ability to walk in older age. A British study found that “mobility limitation” rises faster with age in women than in men [
58], probably because of higher levels of morbidity in older women than in older men, including musculoskeletal problems [
59‐
61]. It is also plausible that increased physical limitations exacerbate gendered concerns about personal safety, which appear to constrain women’s walking [
62]; older women with limited mobility may feel particularly vulnerable in public spaces. On the other hand, it is also possible that a reluctance to walk contributes to a decline in functional mobility in older women and further research is needed to disentangle the causes of the gender difference in the decline in walking for leisure in older age.
There was no evidence for a consistent gender difference in participation in walking for transport and no gender difference in the duration or the frequency of walking for transport was observed. However, there was some evidence that the purpose of men’s and women’s walking for transport may differ. In the one study (from Japan) that reported gender differences in walking to work without controlling for employment status, more men than women walked to work [
29]. This raises the possibility that more men walk to work than women because more men are in employment [
63], although such an effect would vary across countries since gender differences in labour force participation are variable (and relatively large in Japan).
More women than men walked for errands, in line with a general trend for women to devote more time and make more trips than men to serve their household. For example, time-use data collected in 1995 in the USA found that in European American dual-earner married couples both part-time and full-time employed women spent significantly more time per day than employed men serving household needs outside the home [
64]. Data from the USA in the 2000s show that women in couples with children made 2.0 trips to serve children’s needs for every 1 trip made by a man, and 1.7 trips to buy groceries for every 1 trip by a man [
65]. In Germany, while the ratio of trips made for shopping by women compared to men fell sharply between 1976 and 2008, it was still 1.28 in 2008 [
14]. In this context, it is not surprising that more women than men report walking for the purpose of running errands.
There is no gender difference in participation in total walking in the USA when all ages are considered together. There is some evidence of an age-related pattern whereby at younger ages more women walk than men, but at older ages this gender difference is much smaller, or may reverse (consistent with patterns for walking for leisure and with the results for walking for transport in the oldest age groups). As for walking for leisure, this change in the gender difference with age appears to arise because the proportion of women who walk declines faster with age than the proportion of men [
41,
47]. There was also evidence that in Australia more women walk than men. A tentative suggestion, therefore, is that, at least at younger ages, more women than men walk (in total), but that there appears to be geographical variation in gender differences in total walking in middle and older age groups.
This review has some limitations. All of the data were self-reported and while some studies used validated instruments such as the IPAQ, others used similar methods of assessment without specific validation. It is possible that the self-reports of men and women in relation to walking are subject to different biases, which would affect the validity of our findings. Most studies reported on the proportion of men and women who walked during a given period of time. There was limited information on the frequency or duration of walking, which would allow more insight into walking patterns. All the data were collected cross-sectionally and patterns of changing differences over the life-course may therefore reflect cohort effects as well as aging effects. The fact that relevant data were usually reported incidentally in papers focused on other topics means that some eligible studies may not have been retrieved by the search strategy. We suggest there is relatively little risk of publication bias having affected the availability of data for this review, largely because gender was not the main focus of the vast majority of included studies.
Conclusions
The main contribution of this review is to reveal and draw attention to previously largely unrecognised but apparently consistent gender differences in walking for different purposes amongst adults, principally higher levels of walking for leisure and walking for errands by women. We also identified a clear age-related pattern in gender differences in walking for leisure, whereby at younger ages more women than men walk for leisure, a difference that gradually declines and then reverses, so that in the oldest age groups, more men walk for leisure than women.
We have offered some possible explanations for these findings, noting the strongly gendered nature of daily life as reflected in time-use studies, labour force participation statistics and in surveys of general mobility. It seems plausible that women’s childcare responsibilities, perceptions of risk and functional limitations in older life affect their opportunities for, and experiences of, walking. However, the gendered patterns of walking revealed here suggest that more research is needed to improve our understanding of how walking fits into the lives of women and men across the life-course. Particular attention should be paid to investigating the causes of gender differences in the impact of aging on walking, with a view to designing successful and possibly gender-differentiated interventions to maintain walking levels into older age. Maintaining physical activity in older age is important for health and wellbeing and walking has great potential to help older adults achieve recommended levels of physical activity [
66].We suggest that approaches to understanding walking as a social practice, embedded in, and shaped by, complex social worlds [
67,
68], in which gender plays a powerful role, are likely to be particularly helpful.
Acknowledgements
We would like to thank Frances Hillier-Brown, Adetayo Kasim, Kate Hampshire, and Helen Ball for their feedback, which greatly improved the manuscript.