Original paper
Impact of scoring algorithm on physical activity prevalence estimates in Australian adults

https://doi.org/10.1016/j.jsams.2010.05.003Get rights and content

Abstract

Public health recommendations for physical activity are operationalised by defining thresholds for frequency (sessions/week), duration (min/week), or volume (MET-min/week). This study compared estimates of meeting physical activity recommendations when scoring algorithms varied in specifications for frequency and duration but were comparable in volume. Data were obtained from 13,105 Australian adult respondents to the 2006 Exercise, Recreation and Sport Survey (ERASS). Prevalence estimates were calculated using algorithms defined by (i) frequency only (≥5 sessions/week); (ii) duration only (≥150 min/week); (iii) duration only when minutes of vigorous activity were weighted by 2 (≥150 weighted-min/week); (iv) frequency and duration (≥5 sessions/week, ≥150 min/week); (v) volume only (≥600 MET-min/week); and (vi) volume and frequency (≥600 MET-min/week, ≥5 sessions/week). The proportion of adults who met recommendations operationalised without a frequency requirement was twice the proportions obtained for algorithms with frequency requirements. Volume or duration-based algorithms yielded higher estimates for men than women, and for the younger age groups (<35 years) than the older groups, with the opposite observation for frequency-based algorithms. Consistent for all algorithms, people classified at the highest educational attainment had the highest prevalence of meeting recommendations. Agreement in achieving 600 MET-min/week when activities were categorised using activity-specific MET values versus median MET values was 98.3%. Prevalence rates based on 600 MET-min/week were similar to 150 weighted-min/week. In conclusion, varying frequency and duration requirements of scoring algorithms can yield different population estimates and patterns by population subgroup of physical activity for a health benefit.

Introduction

There are many different ways to collect and interpret physical activity data, with little consensus about how to derive estimates consistent with meeting public health guidelines.1 A challenge is translating public health recommendations that prescribe an amount of moderate-to-vigorous physical activity into a measure quantifiable by surveillance methods.

The public health recommendations in Australia are straightforward: all adults should engage in at least 30 min of moderate-intensity physical activity on most, preferably all, days of the week, with each session lasting at least 10 min, and where possible they should also engage in some regular, vigorous activity.2 However, researchers have developed different protocols for categorising survey respondents that yield different estimates of the proportion of the population who achieve the recommended level of physical activity.3, 4 In public health surveillance the ability to correctly categorise individuals is important because measures that lack sensitivity may result in incorrect measures of association, and those that lack specificity may result in estimates with high standard errors and broad confidence intervals.5

The best method for combining moderate-intensity and vigorous-intensity activity into a single indicator of physical activity level has not been determined, but one method is to weight moderate and vigorous activity by relative energy expenditure (i.e., metabolic equivalent of task (MET) values) and calculate volume by multiplying energy expenditure, activity frequency, and duration.6 The general practice in scoring surveillance data has been to assume vigorous activity results in twice the energy expenditure of moderate activity, either by assigning METs of 3.0 or 4.0 for moderate and 6.0 or 8.0 METs for vigorous activities,7, 8 assuming these values are representative of the average MET values for moderate and vigorous activities in populations. An equivalent scoring practice is applied when populations are assessed using the Active Australia questionnaire, in which case minutes of vigorous activity are weighted by a factor of two to approximate a doubling in energy expenditure.9

This work used data from the 2006 Exercise, Recreation and Sport Survey (ERASS), to compare estimates of meeting recommendations for physical activity when six scoring algorithms varied in specifications for frequency and duration but were comparable in volume. The median MET values for moderate- and vigorous-intensity activities were calculated and agreement in categorising respondents based on activity-specific versus median MET values was assessed to determine the impact of weighting by an average MET value on population estimates.

Section snippets

Methods

Participants were respondents to the 2006 ERASS, a survey conducted by the Australian Sports Commission to monitor leisure time physical activities among Australian adults. Respondents were selected using a list-assisted, random-digit-dial of households, and informed consent was obtained from respondents by their willingness to complete the telephone survey. Surveys were completed by 13,708 respondents aged 15–99 (42% response rate). The final analytic sample consisted of 5544 men and 7561

Results

Among respondents who participated in at least one activity of moderate- or vigorous-intensity in the 2 weeks prior to answering the survey, men and women engaged in the same median sessions/week of exercise, recreation, and sport activities, but median min/week and median MET-min/week were higher in men compared to women (Table 1). Those aged 65 years and over had the highest median sessions/week but the lowest median MET-min/week. Persons aged 15–19 years had the lowest median sessions/week

Discussion

The population prevalence of meeting physical activity recommendations varied by the algorithm used to define this indicator. Scoring algorithms with a frequency requirement were the most stringent in classifying individuals. Algorithms based only on MET-minutes or that weighted vigorous-intensity activity to account for additional physiological benefits were the least stringent in classifying individuals. Consistent for all algorithms, people classified at the highest educational attainment

Practical implications

  • The population prevalence of meeting physical activity recommendations varies by the scoring algorithm researchers apply to define these indicators.

  • The lowest prevalence rates appear when indicators are restricted to achieving threshold of frequency; these also tend to minimize gender and age differences.

  • Assigning activity-specific intensity codes (METs) are not likely to improve the accuracy of indicators that are based on one intensity code assigned to groups of activities at the moderate or

Acknowledgements

We wish to thank the Australian Sports Commission (Canberra) and state based departments of Sport and Recreation for permission to use the 2006 ERASS data. NM was employed by the NSW Department of Health on the NSW Biostatistical Officer Training Program at the time this work was conducted. There has been no financial assistance with the project.

References (18)

  • A.L. Dunn et al.

    Lifestyle physical activity interventions. History, short- and long-term effects, and recommendations

    Am J Prev Med

    (1998)
  • A.L. Dunn et al.

    Exercise treatment for depression: efficacy and dose response

    Am J Prev Med

    (2005)
  • R.E. LaPorte et al.

    Assessment of physical activity in epidemiologic research: problems and prospect

    Public Health Rep

    (1985)
  • Australian Institute of Health and Welfare

    Australia's health 2008. Cat. no. AUS 99

    (2008)
  • J.A. Sarkin et al.

    Self-report measures and scoring protocols affect prevalence estimates of meeting physical activity guideline

    Med Sci Sports Exerc

    (2000)
  • R.C. Brownson et al.

    Measuring physical activity with the behavioral risk factor surveillance system

    Med Sci Sports Exerc

    (2000)
  • S.D. Walter et al.

    Effect of exposure misclassification on the mean squared error of population attributable risk and prevented fraction estimates

    Stat Med

    (2007)
  • A.M. Kriska et al.

    Introduction to a collection of physical activity questionnaires

    Med Sci Sports Exerc

    (1997)
  • Center for Disease Control

    Compliance with physical activity recommendations by walking for exercise-Michigan, 1996 and 1998

    MMWR

    (2000)
There are more references available in the full text version of this article.

Cited by (11)

  • Self-Reported Physical Activity Using the International Physical Activity Questionnaire (IPAQ) in Australian Adults with Type 2 Diabetes, with and Without Peripheral Neuropathy

    2016, Canadian Journal of Diabetes
    Citation Excerpt :

    No other studies appear to have reported the proportion of people with type 2 diabetes and peripheral neuropathy who meet physical activity recommendations, but several have reported adherence to physical activity level recommendations in people with type 2 diabetes to be between 25% and 41.7% (17–20). However, these studies had substantially larger sample sizes (between n=648 and n=20,000) compared to the current study and, furthermore, comparison of such studies should be done with caution because researchers have found that varying the frequency and duration requirements of the scoring algorithms can yield different results (21,22). It is recommended that further research be conducted using a larger sample by targeting people with type 2 diabetes and peripheral neuropathy.

  • How diverse was the leisure time physical activity of older Australians over the past decade?

    2012, Journal of Science and Medicine in Sport
    Citation Excerpt :

    Given that exposure to a variety activities and preferences for formal/informal modes of participation were found to influence seniors’ initiation and maintenance of exercise,8 these are important indicators for this population. In order to guide promotional strategies for older Australians, this research uses the Exercise Recreation and Sport Surveys (ERASS) to complement existing public health surveillance.9–11 It is the only surveillance system in Australia that reports specific activities.

  • "Where have all the bicycles gone?" Are bicycle sales in Australia translated into health-enhancing levels of bicycle usage?

    2012, Preventive Medicine
    Citation Excerpt :

    The data used to answer this question are Australian representative data on cycling based on the Exercise, Recreation and Sport Survey (ERASS) data, which is a joint initiative of the Australian Sports Commission and the state and territory government agencies responsible for sport and recreation. ERASS has been previously used for estimating physical activity population prevalence and trends in Australia (Mealing et al., 2011; Merom et al., 2010). ERASS are continuous telephone surveys, collected from around 3500 adult Australians (15 years and older) each quarter (Australian Sports Commission, 2011).

View all citing articles on Scopus
View full text