Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

The inverse relationship between number of steps per day and obesity in a population-based sample – the AusDiab study



Physical activity (PA) is inversely associated with obesity but the effect has been difficult to quantify using questionnaires. In particular, the shape of the association has not yet been well described. Pedometers provide an opportunity to better characterize the association.


Residents of households over the age of 25 years in randomly selected census districts in Tasmania were eligible to participate in the AusDiab cross-sectional survey conducted in 1999–2000. 1848 completed the AusDiab survey and 1126 of these (609 women and 517 men) wore a pedometer for 2-weekdays. Questionnaire data on recent PA, TV time and other factors were obtained. The outcomes were waist circumference (in cm) and body mass index (BMI) (kg/m2).


Increasing daily steps were associated with a decline in the obesity measures. The logarithmic nature of the associations was indicated by a sharper decline for those with lower daily steps. For example, an additional 2000 steps for those taking only 2000 steps per day was associated with a reduction of 2.8 (95% confidence interval (CI): 2.1,4.4) cm in waist circumference among men (for women; 2.2 (95% CI: 0.6, 3.9 cm)) with a baseline of only 2000, steps compared to a 0.7 (95% CI 0.3, 1.1) cm reduction (for women; 0.6 (95% CI: 0.2, 1.0)) for those already walking 10 000 steps daily. In the multivariable analysis, clearer associations were detected for PA and these obesity measures using daily step number rather than PA time by questionnaire.


Pedometer measures of activity indicate that the inverse association between recent PA and obesity is logarithmic in form with the greatest impact for a given arithmetic step number increase seen at lower levels of baseline activity. The findings from this study need to be examined in prospective settings.


Recent increases in obesity and type 2 diabetes in many countries, including the US and Australia, have been attributed to declining physical activity (PA).1, 2, 3, 4 Evidence for this comes from comparisons of time trends in PA and in obesity and diabetes, and also from both cross-sectional and prospective cohort data on the associations in individuals.2, 5, 6 The extent of the contribution of PA to obesity, however, has been difficult to quantify because questionnaire measures of PA, which have been the means of measuring PA,7 are not able to adequately capture the more dispersed incidental activity, such as short periods of walking around the home or workplace.8 That there might be an important contribution from this source is indicated by the finding in a number of studies that television watching predicts obesity or diabetes independently of PA estimated from questionnaires.9, 10, 11 This possibly reflects the fact that time spent TV watching is a proxy for inactivity during periods when, otherwise, incidental PA may have been performed.

The development of relatively inexpensive movement monitors – pedometers – which objectively measure the number of steps taken per day by an individual – has made it possible to objectively estimate total daily PA from subjects in epidemiological studies.12, 13 To this point, pedometers have only been used in relatively small or non-representative samples or in intervention settings, limiting inference about the quantitative relationship between steps taken per day and health outcomes in a free-living population.2, 14, 15, 16 Data providing information on the shape of the relationship and whether a threshold might exist, for example, is likely to be important in the development of public health messages.

In this report, we aim to (i) describe the association between recent PA (measured by both pedometer and questionnaire) and waist circumference and body mass index (BMI) and (ii) to compare the association of number of steps per day measured by pedometer with waist circumference and body mass index (BMI) to that of PA time over the past week by questionnaire and these measures of obesity. We do this by using data available on these measures and related factors, such as television watching time and alcohol intake, from the Tasmania sample of those who were enrolled in the AusDiab study.4


Study sample

The AusDiab Study estimated the national prevalence of diabetes and its risk factors in a representative sample of adults aged 25 years and older from the six States and the Northern Territory of Australia during 2000. Methods and central findings for the study have been reported in detail previously.2, 14, 17 The study had ethics committee approval. Essential details of the study design in Tasmania were as follows: six census collection districts (CDs) were sampled at random from the Tasmanian list, based on the 1996 Australian Census. Within those CDs all houses were approached and all adults 25 years and over were asked to participate. Approximately, 40% of contacted eligible subjects took part. The sample included 1848 Tasmanian adults who were invited to also wear pedometers. In all, 1126 provided additional pedometer data for 2 week days. Twenty-nine other subjects who wore a pedometer on at least one weekend day were excluded from analyses. Among the 1126 subjects, 59% were from families with a single study subject and 41% came from families with two or more subjects participating in the study. Among families with more than one participant 93% had one male and one female subject. The study was approved by the International Diabetes Institute ethics committee.17


The pedometer protocol for this study was developed at the Menzies Research Institute based on the literature18 and on experience with administration of the pedometers over time. The pedometers used were the Omron HJ-003 and Omron HJ-102. These were regularly checked for accuracy of count. Research staff demonstrated to each participant the correct placement and operation of the pedometer, as well as providing them with printed instructions. A diary was supplied for recording of daily steps. We collected 2 days of pedometer readings for each subject. A previous study found an intraclass correlation of 0.71–0.84 for any 2 days of recording, and two consecutive days captured 89% of the variance of a 7-day recording period.19

Assessment of recent PA time, TV time, diet and demographic attributes

Self-reported PA, TV time, diet, alcohol, cigarette consumption and demographic attributes, were assessed using an interviewer-administered questionnaire. Participants reported the frequency and duration of their PA for the previous week using the Active Australia Survey questionnaire.21 For this study, PA was estimated by the time spent walking for recreation or transport, undertaking moderate intensity non-work, non-domestic activity and vigorous activity, usually of a sporting nature, during the last week. These questions have been found to provide reliable and valid estimates of adult PA.21, 22, 23 Vigorous activity was given double the weighting of moderate activity when calculating the total hours of PA time.21 In addition, participants also self-reported the total time they spent watching TV over the past week. This measure provides a reliable and valid estimate of TV time among adults.24 Participants completed a dietary questionnaire with a semiquantitative food frequency section as well as questions relating to alcohol intake for the calculation of total daily energy intake (kJ/day).25 Further questions relating to alcohol consumption and tobacco use and number of children for female subjects were asked in a structured interview.

Assessment of anthropometric measures of obesity

Waist circumference was measured halfway between the lower border of the ribs and the iliac crest on a horizontal plane. Two measurements using a steel measuring tape to the nearest 0.5 cm were recorded. If the measurements varied by more than 2 cm, a third measurement was taken. The mean of the two closest measurements was calculated. Height without shoes was measured to the nearest 0.005 m (0.5 cm) using a stadiometer. Weight was measured in light clothes on a mechanical beam balance, and was recorded to the nearest 0.1 kg. BMI was calculated as weight (kg)/height2 (m2).

Statistical analysis

Statistical analyses were performed for the two obesity-related outcomes, waist circumference and BMI. Analyses were performed using either SPSS Version 11.026 or Stata Version 8.0.27 SPSS was used to compute descriptive statistics (means and standard errors), tests and significance levels comparing gender differences shown in Table 1 and BMI groups within gender in Table 2. Stata was used to perform all linear regression modeling. The linear regression modeling took account of correlation of observations of subjects within families by using a robust estimator of standard errors obtained by defining families as clusters. The regression analyses were weighted to account for the AusDiab sampling. Rather than use a complicated single model containing multiple gender by covariate interactions we chose to use gender-specific models developed using the following five step procedure for each outcome variable: (1) parallel but separate gender-specific multivariable models containing all the covariates in Table 1 were fitted, with the exception that number of children was considered for women but not men as the number of children was not collected for men. Robust estimates of the standard errors of the estimated coefficients were used in the computation of significance levels for the estimated regression coefficients; (2) the models were simplified by successively deleting the covariate with the least significant estimated coefficient until all the estimated coefficients were significant in one or both gender-specific models. No variable not included in the model changed the estimate of the coefficient of any variable in the model by more than 15 percent; (3) the assumption of linearity in the model of continuous covariates was checked using the method of fractional polynomials.28 These analyses were performed without weighting or clustering observations as Stata's implementation of fractional polynomials does not support these options; (4) interactions were created between main effects in the model. These were added one at a time to the main effects model resulting at step 3 and tested for significance. There were no significant interactions; (5) sensitivity to individual observations was checked using measures of leverage, residual and influence. Model assumptions were evaluated using a test for normality of the residuals and a test for homogeneity of error variance.29 Suspected influential or outlying observations were identified from box plots of the diagnostic statistics. If the data values were judged to be implausible or highly unusual using standard criteria29 then the subject was excluded from the fit of the final model. We note that two men and six women were identified as being outliers for waist circumference and four men and five women for BMI. In all cases the reason for exclusion was an unrealistic number of hours of TV, average number of steps or alcohol per day relative to the rest of the subjects in their gender group.

Table 1 Estimated means and s.e. or percents for study variables for men and women
Table 2 Distribution of average number of steps per day


Gender-specific means and standard errors for the obesity measures, pedometer data and demographic and lifestyle variables are shown in Table 1 along with the P-value from the two sample t-test for continuous variables or χ2-test for current smoking status. Significant gender differences were observed for all variables except average daily steps, current smoking status and weekly hours of TV. For all variables where there was a significant gender difference, men had higher values than women. While men reported more time in PA than women (men: 6.4 (0.33), women: 4.5 (0.23), P<0.01); the average number of steps (in 000's) taken per day was not significantly different (men: 10.9 (0.23), women: 11.2 (0.21), P=0.34). The relatively greater time in PA reported by men was not explained by their participation in vigorous activity.

The overall distribution of the average number of steps per day among male and female subjects is shown in Table 2. The distributions are not significantly different (P=0.56).

Men who did not wear a pedometer (n=306) were not significantly different to those who did with respect to age, waist circumference, weekly hours of exercise and of TV watching. Women who did not wear a pedometer (n=387) were significantly older than those who did (mean 53.0 years, P<0.001) and had a larger mean waist circumference (87.3 cm, P<0.01), but were not significantly different in their weekly hours of exercise and TV watching.

Gender-specific means and standard errors or percentages according to BMI groups for the anthropometric measures, pedometer data, demographic and lifestyle variables are shown in Table 3a and b. For men, significant differences were observed for average daily steps and weekly hours of TV. Difference in average daily steps from the lowest to the highest category was observed for women, it was not significant.

Table 3 (a) Estimated means and s.e. or percents for study variables for three BMI groups for men and (b) estimated means and s.e. or percents for study variables for three BMI groups for women

As many of the associations in Table 3a and b were different for men and women, a single regression model would require multiple gender by covariate interactions. Hence, we decided to build gender-specific but parallel regression models for each outcome (waist circumference and BMI) – that is, separately fit models but ones that contained the same covariates. The models were obtained by applying the five steps described in statistical methods and are shown in Table 4 for waist circumference and in Table 5 for BMI. The common set of variables included average number of steps, hours of TV time, alcohol consumption per day, hours of PA, age and height (for waist circumference). The number of children was significant in the univariable models for waist circumference and BMI but not in the multivariable models for women. Application of the method of fractional polynomials showed that the models for both outcomes were linear in the log transformation of average steps and hours of PA. This is reflected in the fact that these variables are denoted by Ln (*) in Tables 4 and 5.

Table 4 Multivariable regression models showing associations with waist circumference
Table 5 Multivariable regression models showing associations with BMI

The results for waist circumference in Table 4 show that the association of daily steps was the most highly significant of the PA measures in the models for both men and women: the estimated coefficient (effect) for men being about 28% larger than that for women (men: −4.1, women −3.2). The estimated coefficients for Ln (PA Time), TV time and alcohol consumption per day were significant for men but not for women. The effects of age and height were significant in each model.

Table 5 shows that the association between daily step number and BMI was significant for men and marginally significant for women (P=0.062). In contrast, recent PA time measured by questionnaire was not significantly associated with BMI. Note that in Table 5 an additional term for Age2 was also included, because age has a significant quadratic effect for both men and women (with those of middle age having higher BMI than younger and older study subjects). Food frequency questionnaire measurement of daily energy intake from all sources was not associated with either waist circumference or BMI after the variables shown in Tables 3 and 4 were taken into account.

Another way to present the effect of number of steps is the estimate for, say, a 50% increase in the number of steps. As the model for waist circumference and BMI are linear in log steps this estimate does not depend on the number of steps. For waist circumference the estimate is a decrease of 1.7 cm (95% confidence interval (CI) 0.77, 2.56) (men) and 1.3 cm (95% CI 0.82, 1.79) (women) The estimate for BMI is a decrease of 0.66 (95%CI. 32, 1.0) (men) and 0.42 (95% CI 0.02 increase to 0.87 decrease) (women). Specific examples of the estimated difference in means for an increase of 2000 steps per day are shown in Table 6. These results demonstrate that most of the differences in mean waist circumference and BMI associated with a difference of 2000 daily steps occurs in subjects walking less than 10 000 steps per day.

Table 6 Estimated decrease in mean waist circumference and mean BMI associated with an additional 2000 steps per day varies by the baseline of number of steps per day


The inverse association between PA and obesity has been difficult to characterize using self-report activity data. Here, using pedometer measures of recent PA in a large population-based sample, an increasing number of average daily steps was inversely associated with waist circumference and BMI. The inverse association was logarithmic in form. The estimated coefficient for the effect of average number of steps in logarithmic form for men was 22% larger than that for female subjects for waist circumference and was 35% larger for BMI. After taking into account other factors associated with obesity, such as the time spent watching TV, independent associations for daily steps remained but PA was significant only for waist circumference in men. Although the average daily step number was similar for female and male subjects, males reported significantly more PA time.

In the 1995 recommendation on PA and public health by the Centres for Disease Control and Prevention and the American College of Sports Medicine,30 it was stated that the lower the baseline PA status, the greater will be the health benefit associated with a given increase in PA.30 Here, this concept has now been demonstrated quantitatively for average daily steps and waist circumference as well as BMI. The decline in these obesity measures associated with an additional 2000 daily steps varied according to baseline activity with a larger decline for those who were sedentary or with a low daily step count. These results suggest baseline activity should be taken into account when studying the health effect of an arithmetic exercise ‘dose’ – defined by time, intensity and or daily steps or when providing an exercise prescription.

The associations between PA and obesity here are cross-sectional and thus the underlying connection cannot be assumed to be causal. For example, high BMI has been associated with subsequent physical inactivity.31 However, although people who are obese are usually inactive,32 body composition, in terms of percentage body fat, is not a powerful predictor of PA habits.30 In an intervention, baseline BMI was not significantly related to changes in step number.16 That is, BMI did not appear to influence the intervention-associated increase in step number. Previously cross-sectional associations between self-reported PA and BMI2, 33 have also been found in prospective studies.6, 10, 34, 35 It seems reasonable to infer that the cross-sectional associations observed for PA measured by pedometer rather than questionnaire report are likely to similarly reflect cause. Nonetheless, prospective data on large samples with concurrent measurement of other determinants of anthropometry, such as diet, are required. Such studies are currently underway.

Strength of these data is that they were from a large representative sample of a defined population within Australia – a country that has a standard of living comparable to North America and Western Europe, ready access to modern manufactured goods, and shares a growing problem of obesity. The size of the data set make it possible to estimate with reasonable validity the shape of the line or curve of best fit, which has not been performed previously. While it has been suggested that pedometer readings be captured for more than 2 days to validly estimate habitual activity,20 a 2-day capture was not only more feasible in a large free-living sample such as this, it also accounts for 89% of the variance of 7 days of recording (compared with 94% for 3 days of readings).20 The association was nonlinear and best fitted by a log function. The estimates from this model indicate that one would be likely to see prospectively a greater reduction in obesity by increasing number of steps for those who are at the lower end of daily energy expenditure, particularly among those taking fewer than 10 000 steps per day.

The stronger independent associations observed for PA and obesity when assessing activity by pedometer rather than questionnaire may have reflected an improvement in measurement validity due to increased precision and/or to the inclusion of incidental activity in the daily step tally. The gender difference in PA found in questionnaire data, but not in pedometer data, may reflect a higher level of incidental activity in women picked up by the pedometer. Past work showing weak or absent prospective associations detected between questionnaire measures of PA and waist circumference36 or walking less than 4 h/week and BMI37 may partly reflect these measurement issues. Here, possible inaccuracies in measuring energy intake by questionnaire and the fact that energy intake is positively correlated with PA might have obscured the association between daily energy intake and obesity in this study.

Recommendations on exercise are a major public health priority but definitive data are lacking. In 2002, a consensus meeting concluded that 45–60 min a day of moderate PA per day was required to prevent the transition to overweight or obesity,38 substantially more than the 1995 recommendation that every adult should accumulate 30 min or more of moderate-intensity PA on most, preferably all, days of the week.30 Future prospective studies incorporating pedometer or accelerometer measures of activity should assist the evidence base for future recommendations.

In the USA in 1999–2000, more than 60% of adults aged 20–74 years were classified as overweight or obese.39 As of 2001, more than 25% were not active at all in their leisure time.39 The findings here suggest that clinicians working with overweight patients would be justified in recommending even modest increases in daily activities, such as walking, among inactive patients, to achieve benefits. Pedometers have been shown in other studies to effectively motivate individuals to increase and maintain additional levels of PA.40 They also assist the individual and the physician to monitor their efforts to increase PA. The increasing availability of information to assist the clinician to use pedometer devices in structured programs41 is making their introduction to clinical practice more feasible.

In conclusion, this study has characterized the inverse association between PA and obesity using a pedometer measure of average daily steps. It provides quantitative evidence to support the proposal that the health effect of a given increase in activity will vary depending on baseline activity level. A greater reduction in obesity by increasing daily step number is likely for those who are at the lower end of daily energy expenditure, particularly in the range under 10 000 steps per day.


  1. 1

    Mokdad AH, Serdula MK, Dietz WH, Bowman BA, Marks JS, Koplan JP . The spread of the obesity epidemic in the United States, 1991–1998. JAMA 1999; 282: 1519–1522.

    CAS  Article  Google Scholar 

  2. 2

    Cameron AJ, Welborn TA, Zimmet PZ, Dunstan DW, Owen N, Salmon J et al. Overweight and obesity in Australia: the 1999–2000 Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Med J Aust 2003; 178: 427–432.

    PubMed  Google Scholar 

  3. 3

    Centers for Disease Control and Prevention (CDC). Prevalence of diabetes and impaired fasting glucose in adults–United States, 1999–2000. Morb Mortal Week Rep 2003; 52: 833–837.

  4. 4

    Dunstan DW, Zimmet PZ, Welborn TA, De Courten MP, Cameron AJ, Sicree RA et al. The rising prevalence of diabetes and impaired glucose tolerance: the Australian Diabetes, Obesity and Lifestyle Study. Diabetes Care 2002; 25: 829–834.

    Article  Google Scholar 

  5. 5

    Fitzgerald SJ, Kriska AM, Pereira MA, de Courten MP . Associations among physical activity, television watching, and obesity in adult Pima Indians. Med Sci Sports Exerc 1997; 29: 910–915.

    CAS  Article  Google Scholar 

  6. 6

    Ching PL, Willett WC, Rimm EB, Colditz GA, Gortmaker SL, Stampfer MJ . Activity level and risk of overweight in male health professionals. Am J Public Health 1996; 86: 25–30.

    CAS  Article  Google Scholar 

  7. 7

    Sallis JF, Saelens BE . Assessment of physical activity by self-report: status, limitations, and future directions. Res Q Exerc Sport 2000; 71: S1–S14.

    CAS  Article  Google Scholar 

  8. 8

    McCormack G, Giles-Corti B, Milligan R . The test–retest reliability of habitual incidental physical activity. Aust NZ J Public Health 2003; 27: 428–433.

    Article  Google Scholar 

  9. 9

    Jakes RW, Day NE, Khaw KT, Luben R, Oakes S, Welch A et al. Television viewing and low participation in vigorous recreation are independently associated with obesity and markers of cardiovascular disease risk: EPIC-Norfolk population-based study. Eur J Clin Nutr 2003; 57: 1089–1096.

    CAS  Article  Google Scholar 

  10. 10

    Hu FB, Leitzmann MF, Stampfer MJ, Colditz GA, Willett WC, Rimm EB . Physical activity and television watching in relation to risk for type 2 diabetes mellitus in men. Arch Intern Med 2001; 161: 1542–1548.

    CAS  Article  Google Scholar 

  11. 11

    Dunstan DW, Salmon J, Owen N, Armstrong T, Zimmet PZ, Welborn TA et al. Physical activity and television viewing in relation to risk of undiagnosed abnormal glucose metabolism in adults. Diabetes Care 2004; 27: 2603–2609.

    Article  Google Scholar 

  12. 12

    Chan CB, Spangler E, Valcour J, Tudor-Locke C . Cross-sectional relationship of pedometer-determind ambulatory activity to indicators of health. Obes Res 2003; 11: 1563–1570.

    Article  Google Scholar 

  13. 13

    Tudor-Locke C, Bassett DR . How many steps/day are enough? Preliminary pedometer indices for public health. Sports Med 2004; 34: 1–8.

    Article  Google Scholar 

  14. 14

    Miyatake N, Nishikawa H, Morishita A, Kunitomi M, Wada J, Suzuki H et al. Daily walking reduces visceral adipose tissue areas and improves insulin resistance in Japanese obese subjects. Diabetes Res Clin Pract 2002; 58: 101–107.

    CAS  Article  Google Scholar 

  15. 15

    Bassett DR, Schneider PL, Huntington GE . Physical activity in an Old Order Amish community. Med Sci Sports Exerc 2004; 36: 79–85.

    Article  Google Scholar 

  16. 16

    Chan CB, Ryan DA, Tudor-Locke C . Health benefits of a pedometer-based physical activity intervention in sedentary workers. Prev Med 2004; 39 (6): 1215–1222.

    Article  Google Scholar 

  17. 17

    Dunstan DW, Zimmet PZ, Welborn TA, Cameron AJ, Shaw J, de Courten M et al. The Australian diabetes, obesity and lifestyle study (AusDiab) – methods and response rates. Diabetes Res Clin Pract 2002; 57: 119–129.

    Article  Google Scholar 

  18. 18

    Tudor-Locke CE, Myers AM . Methodological consideration for researchers and practioners using pedometers to measure physical (ambulatory) activity. Rese Quart Exerc Sport 2001; 72: 1–12.

    CAS  Article  Google Scholar 

  19. 19

    Tudor-Locke C, Burkett L, Reis JP, Ainsworth BE, Macera CA, Wilson DK . How many days of pedometer monitoring predict weekly physical activity in adults? Prev Med 2005; 40 (3): 293–298.

    CAS  Article  Google Scholar 

  20. 20

    Australian Institute for Health and Welfare (AIHW). The Active Australia Survey: A Guide and Manual for Implementation, Analysis and Reporting. AIHW: Canberra, 2003.

  21. 21

    Timperio A, Salmon J, Bull F, Rosenberg M . Validation of Adult Physical Activity Questions for use in Australian Population Surveys. University of Western Australia: Perth, 2002.

    Google Scholar 

  22. 22

    Brown W, Trost S, Bauman A, Mummery K, Owen N . Test- Re-Test Reliability Study of Physical Activity Questions for Use in Australian Population Surveys. University of Queensland: St Lucia, 2002.

    Google Scholar 

  23. 23

    Armstrong T, Bauman A, Davies J . Physical Activity Patterns of Australian adults. Results of the 1999 National Physical Activity Survey. Australian Institute of Health and Welfare: Canberra, Australia, 2000.

    Google Scholar 

  24. 24

    Salmon J, Owen N, Crawford D, Bauman A, Sallis JF . Physical activity and sedentary behaviour: a population-based study of barriers, enjoyment, and preference. Health Psychol 2003; 22: 178–188.

    Article  Google Scholar 

  25. 25

    Ireland P, Jolley D, Giles G, O'Dea K, Powles J, Rutishauser I et al. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pacific J Clin Nutr 1994; 3: 19–31.

    CAS  Google Scholar 

  26. 26

    SPSS Inc. SPSS Statistical Software: Release 11.0. SPSS Corp: Chicago, IL, 2003.

  27. 27

    Stata Corp. Statistical Software: Release 8.0. Stata Corporation: College Station, TX, 2003.

  28. 28

    Royston P, Ambler G, Sauerbrei W . The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol 1999; 28: 964–974.

    CAS  Article  Google Scholar 

  29. 29

    Kleinbaum DG, Kupper LL, Mueller KE, Nazim A . Applied Regression Analysis and Other Multivariable Methods 3rd edn. Duxbury: Belmont, CA, 1998.

    Google Scholar 

  30. 30

    Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA, Bouchard C et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995; 273: 402–407.

    CAS  Article  Google Scholar 

  31. 31

    Petersen L, Schnohr P, Sorensen TI . Longitudinal study of the long-term relation between physical activity and obesity in adults. Int J Obes Relat Metab Disord 2004; 28 (1): 105–112.

    CAS  Article  Google Scholar 

  32. 32

    Bouchard C, Depres JP, Tremblay A . Exercise and obesity. Obesity Res 1993; 1: 133–147.

    CAS  Article  Google Scholar 

  33. 33

    Pratt M, Macera CA, Blanton C . Levels of physical activity and inactivity in children and adults in the United States: current evidence and research issues. Med Sci Sports Exerc 1999; 31: S526–S533.

    CAS  Article  Google Scholar 

  34. 34

    Garrow JS, Summerbell CD . Meta-analysis: effect of exercise, with or without dieting, on the body composition of overweight subjects. Eur J Clin Nutr 1995; 49: 1–10.

    CAS  PubMed  PubMed Central  Google Scholar 

  35. 35

    Ballor DL, Keesey RE . A meta-analysis of the factors affecting exercise-induced changes in body mass, fat mass and fat-free mass in males and females. Int J Obes Relat Metab Disord 1991; 15: 717–726.

    CAS  Google Scholar 

  36. 36

    Eck LH, Pascale RW, Klesges RC, Ray JA, Klesges LM . Predictors of waist circumference change in healthy young adults. Int J Obes Relat Metab Disord 1995; 19: 765–769.

    CAS  PubMed  Google Scholar 

  37. 37

    Kahn HS, Tatham LM, Rodriguez C, Calle EE, Thun MJ, Heath Jr CW . Stable behaviors associated with adults' 10-year change in body mass index and likelihood of gain at the waist. Am J Public Health 1997; 87: 747–754.

    CAS  Article  Google Scholar 

  38. 38

    Saris WH, Blair SN, van Baak MA, Eaton SB, Davies PS, Di Pietro L et al. How much physical activity is enough to prevent unhealthy weight gain? Outcome of the IASO 1st Stock Conference and consensus statement. Obes Rev 2003; 4: 101–114.

    CAS  Article  Google Scholar 

  39. 39

    Morrill AC, Chinn CD . The obesity epidemic in the United States. J Public Health Policy 2004; 25: 353–366.

    Article  Google Scholar 

  40. 40

    Marshall A, Ferney S . Are pedometers a measure or motivator for physical activity? Active Living – All Together Better National Physical Activity Conference, Fremantle, Western Australia 2003.

  41. 41

    Steps to a Better Health Manual

Download references


We are most grateful to the following for their financial support of the study: The Commonwealth Department of Health and Aged Care, Abbott Australasia Pty Ltd, Alphapharm Pty Ltd, Aventis Pharmaceutical, AstraZeneca, Bristol-Myers Squibb Pharmaccuticals, Eli Lilly (Australia) Pty Ltd, GlaxoSmithKline, Janssen-Cilag (Australia) Pty Ltd, Merck Lipha s.a., Merck Sharp & Dohme (Australia), Novartis Pharmaceutical (Australia) Pty Ltd, Novo Nordisk Pharmaceutical Pty Ltd, Pharmacia and Upjohn Pty Ltd, Pfizer Pty Ltd, Roche Diagnostics, Sanofi Synthelabo (Australia) Pty Ltd, Servier Laboratories (Australia) Pty Ltd, BioRad Laboratories Pty Ltd, HITECH Pathology Pty Ltd, the Australian Kidney Foundation, Diabetes Australia, Diabetes Australia (Northern Territory), Queensland Health, South Australian Department of Human Services, Tasmanian Department of Health and Human Services, Territory Health Services, Victorian Department of Human Services, and Health Department of Western Australia. DD is supported by a National Health and Medical Research Council (NHMRC) Post-Doctoral Research Fellowship. JS is supported by a Victorian Health Promotion Foundation Public Health Research Fellowship. A-L Ponsonby provided comments on the manuscript draft. Also, for their invaluable contribution to the field activities of AusDiab, we are enormously grateful to Annie Allman, Marita Dalton, Adam Meehan, Clare Reid, Alison Stewart, Robyn Tapp and James Dilger. The AusDiab Steering Committee consisted of Dr B Atkins, Dr S Bennett, Dr S Chadban, Professor S Colagiuri, Dr M de Courten, Dr M D'Embden, Dr D Dunstan, Professor T Dwyer, Dr D Jolley, Dr P Magnus, Professor J Mathews, Dr D McCarty, Professor K O'Dea, Dr P Phillips, Dr P Popplewell, Mr I Kemp, Professor H Taylor, Professor T Welborn and Professor P Zimmet. Role of the Funding Source: The funding source had no such involvement in study design, collection, analysis and interpretation of data, in the writing of the report and in the decision to submit the paper for publication. Contributors: Terence Dwyer, Alison Venn, Leigh Blizzard and Paul Zimmet designed and implemented the study. Jenny Cochrane contributed to study implementation. David Hosmer, Trina Hosmer and Leigh Blizzard conducted the statistical analyses. All authors contributed to the interpretation of the study findings and the writing and revision of manuscript. Terence Dwyer is the guarantor of the paper.

Author information



Corresponding author

Correspondence to T Dwyer.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Dwyer, T., Hosmer, D., Hosmer, T. et al. The inverse relationship between number of steps per day and obesity in a population-based sample – the AusDiab study. Int J Obes 31, 797–804 (2007).

Download citation


  • physical activity
  • pedometer
  • body mass index
  • waist circumference

Further reading


Quick links