Background and Purpose: Although a criterion reference of 10,000 steps per day is recommended for health benefits, a literature search yielded no normative standard for daily steps. Therefore, a meta-analysis was used to estimate the number of pedometer-documented steps that adults take daily.
Methods: Studies were retrieved using electronic databases and cross-referencing from retrieved studies. A random-effects model and 95% confidence intervals (CIs) were used for all analyses of daily step data.
Results: Seventy-one outcomes, representing 6,199 subjects from 42 studies, met the inclusion criteria. Excluding Amish subjects, the daily step count averaged 9,448 (95% CI=8,899–9,996) and was greater in subjects younger than 65 years of age (mean=9,797, 95% CI=9,216–10,377) compared with subjects 65 years of age or older (mean=6,565, 95% CI=4,897–8,233).
Discussion and Conclusion: This study provides an estimate of the number of daily steps taken by adults. The number is less than the recommended 10,000 steps per day and is especially low in adults 65 years of age or older.
The benefits of physical activity are many and well-established. Although there are numerous modes of aerobic physical activity in which individuals can engage, walking is the most widely practiced aerobic physical activity in the United States.1 Of 1,816 adult respondents to the US Physical Activity Study, Eyler et al2 noted that 34% walked regularly (ie, 5 days per week for at least 30 minutes) and 46% walked occasionally (ie, not regularly but 10 minutes per session). Among elderly people, walking may be particularly preferred as a mode of exercise.3,4
With the advent of pedometers, it has become economically feasible to accurately quantify the number of walking steps that individuals actually take on a daily basis in a real-world setting. Currently, 10,000 steps per day is widely promoted as a target for achieving health-related benefits.5 Individuals who meet the goal are more likely to meet the current activity guideline of an accumulated 30 minutes of moderate physical activity on most, if not all, days per week.6 The goal of 10,000 steps is supported by Iwane et al,7 who found that walking 10,000 steps was effective for lowering blood pressure and increasing exercise capacity in patients with hypertension, and by Schneider et al, who demonstrated that a “10,000 step per day exercise prescription resulted in weight loss over 36 weeks in previously sedentary, overweight/obese adults.”8(p85) The usefulness of a criterion reference of 10,000 steps per day notwithstanding, its comparability to actual walking volume is not known. Moreover, it does not provide a standard by which an individual's behavior can be compared with that of his or her peers. For such comparisons, normative reference values are required.9
Since the commercial introduction of pedometers in the 1980s, many studies have been published that describe walking activity on the basis of the number of steps taken per day by individuals who apparently are healthy. However, the mean or median number of steps taken per day varies widely (Tab. 1).10–51 Given the variability found in these studies, a need exists to consolidate available information to provide an estimate of the number of steps taken per day by adults and to examine whether estimates vary among subgroups. Although I am aware of one review of the literature on this topic, the review involved a traditional, narrative approach.52
Meta-analysis is an alternative approach in which individual studies addressing a common issue are statistically combined to arrive at a conclusion about the body of research.53,54 Meta-analysis provides the opportunity to: (1) increase power for primary outcomes and subgroup analysis, (2) help resolve uncertainty when studies disagree, (3) improve estimates of treatment effectiveness, and (4) answer questions not posed at the start of individual trials.54 Thus, considering the potential health relevance of quantifying the steps taken per day by adults, the variability in the literature regarding the number of daily steps taken by adults, and the absence of meta-analytic work summarizing the number of daily steps by adults, the purpose of this study was to use a meta-analytic approach to summarize the number of steps taken per day by adults.
Potentially relevant studies were identified via electronic searches of MEDLINE, CINAHL, EMBASE, Science Citation Index, and PsycINFO. The key words “pedometer” and “pedometry” were used in the searches. To capture studies published prior to January 2005, the last search was conducted in February 2005. The reference lists of all retrieved studies were examined to locate other potentially relevant articles not identified by the database searches. Thereafter, I asked experts on the topic (Catrine Tudor-Locke and David R Bassett) to review the reference list for omissions.
Study Selection and Data Retrieval
Studies were included in the meta-analysis if they met the following criteria: (1) were published prior to January 2005; (2) included adult humans, aged 18 years and older, who apparently were healthy; (3) were disseminated in English-language journals; and (4) reported data (mean and standard deviation) for the number of steps taken as assessed by a pedometer. I selected all studies included in the meta-analysis. Studies were excluded if they: (1) involved children, (2) were limited to patients or individuals with pathologies or abnormalities (eg, hypertension, obesity), or (3) contained no relevant data. Studies published in a foreign language were not included because of concern about the translation and interpretation of findings.
A coding sheet was developed that could hold information on: (1) study characteristics (eg, year of publication), (2) subject characteristics (eg, age, sex), (3) pedometer assessment characteristics (eg, make and model of pedometer), and (4) the primary outcome (number of steps taken per day). For intervention studies that described the number of steps taken at multiple time points (eg, baseline, final), only baseline data were used. For studies involving both patients and matched nonpatients, only data from the nonpatients were used. In cases where authors published multiple articles and the data were determined to not be from unique samples, redundant data were excluded. If the uniqueness of subjects could not be determined across studies, the authors were contacted to clarify such. Such a contact resulted in the exclusion of one study. Sample sizes, means, and dispersion statistics (eg, standard deviations) were abstracted for each outcome from each study.
In cases where step data (mean and standard deviation) were not provided or could not be estimated from the information provided in the study, contact with the authors was attempted to obtain the data. Authors of 2 articles could not be reached, authors of 1 article were contacted but did not provide information, and authors of 1 article provided information.
I abstracted and coded all retrieved studies. In addition, an assistant independently abstracted and coded the data from the studies. Our results were compared, and discrepancies were reconciled.
Pooled outcomes (mean and 95% confidence intervals [CIs] for steps per day) were calculated by assigning weights equal to the inverse of the variance for the number of steps taken per day. A random-effects model that controls statistically for heterogeneity was used to pool results.53 However, I also examined for heterogeneity based on a fixed-effects model using the Q statistic.53 Because the Q statistic suffers from low power, the alpha level for statistically significant heterogeneity was set at P≤.10 versus P≤.05.54 In order to examine the sensitivity of my overall results, I partitioned the data according to several selected strata when such partitioning was possible. The strata were sex (male, female), mean years of age (≥65, <65), days of sampling (<3, ≥3), and country (United States, Japan). The strata differentiation for days of sampling was based on the conclusion of Tudor-Locke et al that “any 3 days can provide a sufficient estimate.”55(p293) The number of studies outside the United States and Japan was deemed insufficient to warrant stratification. If the 95% CIs for comparisons were not overlapping, the difference between strata was considered to be statistically significant. All analyses were conducted using SPSS for Windows (version 11.0)* and the meta-analytic syntax for SPSS developed by Wilson.56
The electronic searches identified between 31 (PsycINFO) and 193 (Science Citation Index) potentially relevant articles. My examination of the articles revealed that 42 articles published between 1983 and 2004 were appropriate for inclusion.10–51 A description of the studies is shown in Table 1. The number of subjects participating in individual studies ranged from 8 to 1,151. Most subjects were North American or Japanese, but some were from Europe or Australia. Studies did not always categorize (subdivide) subjects by sex, age, race or ethnicity, level of function or exercise, and nature of work. The type of pedometer used to monitor steps was not always identified or described thoroughly. Of the pedometers identified, the Yamax Digi-Walker SW-200† (or some name variation) was used most often. The time over which steps were sampled ranged from 1 day to 1 year, with 7 days being the mode (22 studies).
Outcomes for the entire sample and those of selected strata are presented in Table 2. For the entire sample of 42 studies (6,199 subjects), the mean number of daily steps ranged from a low of 3,766 for Americans 65 or more years of age12 to a high of 18,425 for Amish men.51 The results from these different studies were homogeneous. That is, their nonsignificant Q score suggests that the studies represent samples from a common population and that coalescing of their data was justified (Q=67.9, P=.550). The overall number of steps per day for all of the studies averaged 9,501. The weighted standard deviation was 2,295 daily steps, and the 95% CI was 8,955 to 10,047 daily steps. The homogeneity of the studies notwithstanding, the mean number of steps taken by Amish men (18,425) and women (14,196) was more than 2 standard deviations beyond the mean for the remainder of the sample and thus were excluded from comparisons among strata. Data for each stratified variable (eg, age) also were homogeneous, indicating that subjects within the strata were from a common population. Nonetheless, there was a statistically significant difference between age strata. Specifically, the group with a mean age of less than 65 years took significantly more steps than the group with a mean age of ≥65 years.
Although a criterion reference of 10,000 daily steps has been recommended5 and numerous studies have been published that described the number of steps taken daily by adults, I believe that this is the first article to quantitatively summarize literature in an effort to estimate the number of steps that adults actually take. The mean steps for some groups described in the literature (most notably the Amish)21,24,43–51 exceeded the 10,000-step criterion. For the Amish, who eschew some labor-saving technology such as automobiles, a larger step number should be expected. Using the classification scheme of Tudor-Locke and Bassett,5 the average number of steps for all subjects from all studies together in this meta-analysis would place them within the “somewhat active” category (7,500–9,999 steps per day).
One of the most notable findings of this study was the lower mean number of steps taken per day (6,565) by adults with a mean age of 65 years or older. The mean number of steps of these older adults would place them in the “low active” category (5,000–7,499 steps per day) of Tudor-Locke and Bassett.5 Although the lower number recorded for older adults could be influenced by their slower gait, which is less likely to trigger the pedometer to record a step,57 I doubt that this explains all or most of the lower step number in the “ostensibly healthy older adult population” incorporated into this meta-analysis. Patient- or client-related instruction, one of the roles of physical therapists,58 might include information on the health benefits of walking for older adults. Among such benefits are lowered blood pressure, weight loss, reduced waist circumference, and improved glucose tolerance,7,8,59–61 Once a pedometer has been used to quantify daily their steps, they will have access to accurate information that will allow them to compare the number of steps they take relative to both the normative values presented in this article and the 10,000-step criterion.
Meta-analysis, like any type of research or review, has limitations. One limitation was the samples of data contributing to this meta-analysis. To the best of my knowledge, no study contributing to the analysis utilized a random sample from the population. Such a sample would provide a more representative indication of “normal” than the convenience samples used in the included studies. In any case, this meta-analysis should provide a better estimate of daily steps than is provided by any single study.
A second limitation of this study was the different pedometers used to measure steps as well as the lack of specificity provided by some studies about the pedometer used. Previous research41 has shown that the validity and reliability of data obtained with pedometers vary. The largest number of studies used either the Yamax Digi-Walker SW-200 or the Yamax Digi-Walker SW-701,† both of which have been shown to accurately reflect the number of steps taken.41,62 However, some studies did not specifically indicate the pedometer used, or they used other pedometers to measure steps. The former factor made sensitivity analysis in regard to pedometer use impracticable.
A third limitation may have been the number of days over which steps were monitored. Tudor-Locke et al55 suggested that a single day of step monitoring is not acceptable but that 3 days provide “a sufficient estimate.” Several of the studies included in this meta-analysis monitored steps for only 1 or 2 days. The sensitivity analysis, however, did not find any difference when results were partitioned according to whether the data were collected for less than 3 days or for 3 or more days.
A fourth limitation of the study was its inability to provide complete summary data for specific strata. Often the necessary breakdown of information (eg, men versus women) from the contributing studies was not provided. When information was provided, it was not always specific. Consequently, I had to use discrete categories (ie, mean age of <65 years versus ≥65 years) rather than more specific categories (eg, decades) that often are used to establish normative reference values.
More representative and refined normative reference values than could be derived from my meta-analysis are warranted. Such values would need to be obtained from a large population-based sample that could be divided into strata based on variables such as age, sex, ethnicity, and living environment.
In conclusion, the number of steps taken per day by adults is typically less than the 10,000 that have been recommended. This is particularly the case for elderly people. This information can be used by physical therapists and others engaged in health promotion.
- Received February 1, 2006.
- Accepted July 20, 2007.
- Physical Therapy