In Australia, survey instruments for the

Comparison of surveys used to measure physical activity Abstract Objective: To compare the level of agreement in results obtained from four physical activity (PA measurement instruments that are in use in Australia and around the world. Methods: 1,280 randomly selected participants answered two sets of PA questions by telephone. 428 answered the Active Australia (AA and National Health Surveys, 427 answered the AA and CDC Behavioural Risk Factor Surveillance System surveys (BRFSS, and 425 answered the AA survey and the short International Physical Activity Questionnaire (IPAQ. Results: Among the three pairs of survey items, the difference in mean total PA time was lowest when the AA and NHS items were asked (difference=24 (SE:17 minutes, compared with 144 (SE:21 mins for AA/BRFSS and 406 (SE:27 mins for AA/IPAQ. Correspondingly, prevalence estimates for sufficiently active were similar for AA and NHS (56% and 55% respectively, but about 10% higher when BRFSS data were used, and about 26% higher when the IPAQ items were used, compared with estimates from the AA survey. Conclusions: The findings clearly demonstrate that there are large differences in reported PA times and hence in prevalence estimates of sufficient activity from these four measures. Implications: It is important to consistently use the same survey for population monitoring purposes. As the AA survey has now been used three times in national surveys, its continued use for population surveys is recommended so that trend data over a longer period of time can be established. (Aust N Z J Public Health 2004; 28: 128-34 Wendy Brown School of Human Movement Studies, University of Queensland Adrian Bauman, Tien Chey School of Public Health and Community Medicine, University of New South Wales Stewart Trost School of Human Movement Studies, University of Queensland Kerry Mummery School of Health and Human Movement Studies, Central Queensland University, Queensland In Australia, survey instruments for the assessment of physical activity (PA have evolved over time, from those used in the early cardiovascular disease risk factor surveys 1 to those now used for population monitoring. As is the case in other countries, several different instruments are used to assess population levels of activity in Australia. For example, at the national level, the Australian Bureau of Statistics (ABS included nine items to assess PA in its fiveyearly National Health Surveys (NHS in 1989, 1995 and 2001, 1 while a different PAspecific survey was developed to evaluate the Active Australia campaign in 1997, 1999 and 2000. 2 Both these surveys offer the opportunity to assess trends in PA over time. However, in some States, there has been interest in using the PA items that form part of the US CDC Behavioural Risk Factor Surveillance System (BRFSS, 3 and in the International Physical Activity Questionnaire (IPAQ, a newer survey developed by a group of international researchers (including several Australians that has undergone reliability and validity testing in 12 countries. 4 Each instrument includes consideration of frequency and duration of a range of activities of defined intensity, such as walking, moderate or vigorous activity. Frequency is usually assessed as the number of days or number of sessions of activity in a defined period, such as last week, last seven days, a usual week or the last two weeks, etc. Duration is usually reported for either the same time period (e.g. last week, last two weeks or for a typical day when the activity was undertaken in that time period. It is generally reported in hours and minutes and is usually summed across the domains of walking, moderate and vigorous-intensity activity, to provide an indication of total physical activity time over the reporting period. Some measures also ask about selfrated intensity of participation in the nominated activities. The primary measurement goal in all these surveys is to provide an estimate of whether respondents reach an energy expenditure threshold which is sufficient to be of health benefit. In Australia, this is usually defined as participation for a specified minimum time (150 minutes in activities of at least moderate intensity (i.e. with energy expenditure at least three times the resting rate, or 3 metabolic equivalents or METs 5 accumulated over at least five days each week. Submitted: June 2003 Revision requested: September 2003 Accepted: October 2003 Correspondence to: Professor Wendy Brown, School of Human Movement Studies, University of Queensland, St Lucia, Queensland 4072. Fax: (07 3365 6877; e-mail: wbrown@hms.uq.edu.au 128 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 2

Comparison of PA surveys Very little is known about the comparability of different surveys in terms of estimating the proportion of the population that is sufficiently active for health benefit. The main aim of this study was therefore to compare the level of agreement in prevalence estimates derived from four instruments that are in use in Australia and elsewhere around the world. Methods Recruitment and data collection The sample was drawn at random from the regularly updated electronic database of telephone numbers in Rockhampton, Queensland, with replacement of duplicate, mobile and business numbers, as well as of numbers for nursing homes and other collective dwellings. Calls were made using a CATI system in August and September 2001, as part of the baseline data collection for a physical activity intervention study. Each interviewer asked consenting participants to answer the Active Australia survey and one of the other three surveys: either the BRFSS, the short form of the IPAQ or the NHS physical activity questions. The order of administration of the two surveys was randomly generated by the Ci3 CATI system (Sawtooth Software, Evanston, Illinois, with a random half of each group answering the AA questions first, and the remainder answering the other questions first. The study protocol was approved by the Human Ethics Research Review Panel at Central Queensland University. Data management and scoring protocols Four different measures were used to compare the results of each survey, based on different interpretations of the evidence on the health benefits of physical activity. Total time in physical activity: Total minutes of walking, moderate and/or vigorous-intensity physical activity were summed, with time spent in vigorous activity weighted by a factor of two, to account for its greater intensity. 2 150 minutes: Based on the National PA Guidelines, 6 sufficient activity for health was first defined as accumulation of 150 minutes or more of at least moderate intensity physical activity in a week (with time in vigorous activity weighted by two. 150 minutes and five sessions: A second definition of sufficient activity was developed to take account of the current guidelines, which advise that 30 minutes of moderate intensity physical activity on most days of the week is sufficient to confer health benefits. 2,7 For this definition, the criterion for sufficient activity was defined as 150 minutes of walking, moderate and vigorous activity (with time in vigorous activity weighted by a factor of two accrued in at least five sessions or five days of activity in a week. Sedentariness: The final measure was based on the proportion of people who were sedentary or physically inactive, defined as no participation (i.e. no sessions 10 minutes in any walking, moderate or vigorous physical activity. Coding and truncation rules were developed to derive each of these four measures from each survey, taking into account Table 1: Coding and truncation rules for derivation of the four measures from each survey. Active Australia National Health Behavioural Risk International Physical Survey Factor Surveillance Activity System Questionnaire short version Measure 1 (# of times x time (# of times x total (# days x usual time (# days x time per day Total time calculation for spent in each activity in time in each activity in each activity in the in each activity in a each activity the last week in last two weeks/2 last 7 days usual week Measure 2 W (t + W (r + M + (Vx2 W + M + (Vx2 M + (Vx2 W (all + M + (Vx2 150 minutes 150 minutes/week 150 minutes/week 150 minutes/week 150 minutes/week or or or W (t + W (r + M + (Vx2 + VG W + M + (Vx2 W (mv + M + (Vx2 150 minutes/week 150 minutes/week 150 minutes/week Measure 3 W (t + W (r + M + (Vx2 W + M + (Vx2 M+(Vx2 W (all +M+(Vx2 150 mins and 5 sessions 150 minutes/week AND 150 minutes/week 150 minutes/week 150 minutes/week 5 sessions AND 5 sessions AND 5 sessions AND 5 sessions or or or W (t + W (r + M + (Vx2+VG W + M + (Vx2 W (mv + M + (Vx2 150 minutes/week AND 150 minutes/week 150 minutes/week 5 sessions AND 5 sessions AND 5 sessions Measure 4 Proportion reporting NO activity ( 10 minutes at a time Inactivity Truncation rules For each activity (walking, moderate, vigorous activity weekly minutes truncated to a maximum of 840 minutes (2 hours per day or 14 hours per week. The sum of total weekly time in all activities truncated 1,680 minutes (4 hours per day or 28 hours per week. Notes: W = time spent walking; [W (t = walking to/from places (AA; W (r = walking for recreation or leisure (AA; W (all = all walking (BRFSS and IPAQ; W (mv = moderate and vigorous paced walking (IPAQ]. M = time spent in moderate intensity activities. V = time spent in vigorous activities. 2004 VOL. 28 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 129

Brown et al. differences in the individual items asked in each survey (see Table 1. For example, in the NHS, the number of minutes and sessions reported in the last two weeks was divided by two to give weekly times/sessions. In some cases, more than one interpretation of each measure was derived. For example, in the AA survey, one interpretation of the first three measures did not include time spent in gardening or heavy yardwork [four item AA ( ], while a second interpretation did [five item AA ( ]. Similarly, for the BRFSS, one interpretation included time spent in an additional walking question [three item BRFSS ( ], while the second only included walking as part of the moderate intensity question [two item BRFSS ( ]. For IPAQ, the first three measures were calculated (a including all paces of walking (IPAQ and (b including only walking reported to be at moderate or vigorous pace (. These variations are summarised in Table 1. Note that for the AA survey, the walking question was split into two (walking to and from places and for recreation or pleasure and the responses were summed. Table 2: Numbers of participants in each age and sex group who completed each pair of surveys. Statistical analyses Means and quartiles were calculated for total time in activity derived from each survey and for the total number of days/ sessions of reported activity. Differences between the three pairs of measures were computed and displayed graphically using mean and difference plots. 8 For the three categorical measures, per cent agreement was calculated for both measures of sufficiently active (150 minutes, and 150 minutes and five sessions and for sedentariness. The consistency with which participants were classified as sufficiently active or insufficiently active and as sedentary or not-sedentary on each survey, correcting for chance agreement, was also assessed using Cohen s Kappa. 9 All analyses were conducted using SAS version 8.02 (SAS Institute Inc., Cary NC, USA 1999-2001. Results Active Active Active Australia Australia Australia and NHS and BRFSS and IPAC 18-44 years Men 112 112 111 Women 119 118 119 45-59 years Men 42 43 42 Women 45 45 45 60-75 years Men 47 47 48 Women 63 62 61 Total 428 427 425 Participants The overall response rate was 51%. Of the 1,280 participants, 428 answered the Active Australia and NHS surveys, 427 answered the AA and BRFSS and 425 answered the AA and IPAQ surveys. There was a similar distribution of men and women in each of three age groups in each survey group (see Table 2. Their education levels were similar to those reported for a random Queensland sample in 2001, with 39% having fewer than 12 years of education (37% in the Queensland sample, 18% having completed year 12 (19%, 24% with a trade or similar qualification (26% and 19% with a diploma or degree (18%. 10 The Table 3: Means and quartiles for total time and frequency of activity derived from each survey. Comparisons n Total time (weekly reported minutes Frequency (days/sessions a of measures and distributional quartiles (minutes and quartiles (days Mean (SE 25% 50% 75% Mean (SE 25% 50% 75% 428 327 (18.4 60 180 480 5.9 (6.5 1 4 8 406 (19.8 100 270 600 6.6 (6.6 2 5 9 NHS 303 (17.4 36 180 428 4.6 (4.9 1 3 6.5 427 317 (16.8 40 210 480 5.5 (5.2 2 5 7 399 (19.9 90 253 590 6.4 (5.6 2 5 9 461 (21.1 90 310 840 5.0 (4.5 2 5 7 761 (27.4 240 720 1180 8.9 (6.1 5 9 12 425 353 (18.9 53 210 585 5.3 (4.9 2 4 8 450 (20.8 95 300 720 6.4 (5.2 2 5 9 IPAQ 918 (27.9 430 870 1500 8.8 (5.8 4 8 13 856 (28.1 360 840 1380 8.2 (5.9 3 7 12 Notes: (a Note that more than 7 sessions/days can be accrued per week, by summing across walking, moderate and vigorous questions. = four item Active Australia (both walking questions, moderate and vigorous activities, not including gardening. = five item Active Australia (both walking questions, moderate and vigorous activities and gardening. NHS= National Health Survey, walking, moderate and vigorous. = two item BRFSS, moderate and vigorous activities only. = three item BRFSS, including walking, moderate and vigorous activities. IPAQ = all paces of walking, plus moderate and vigorous activities. = only moderate and vigorous paced walking, and other moderate and vigorous activities. SE = standard error. 130 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 2

Comparison of PA surveys respondents were similar to the Australian population in terms of the proportion categorised as adequately active for health benefit on the AA survey (43.2% of this sample and 45.2% nationally. Total time and number of days/sessions Means and quartiles for total activity time and for frequency of days/sessions reported for each pair of surveys are shown in Table 3. The difference in mean total activity time between the NHS and the four item AA survey (, without vigorous yard work or gardening was 24 minutes (SE:17. When the AA survey was compared with the BRFSS, the differences were much larger (see middle panel of Table 3. The BRFSS total from the moderate and vigorous questions, without the additional double counted walking question, was 144 (SE:21 minutes more than the total. However, when the time spent in gardening item was added to the AA total ( the mean difference was only 62 (SE:21 minutes. The addition of the BRFSS walking question increased the total time estimate by 362 or 444 minutes when compared with or respectively. The differences in mean total time from the AA and short IPAQ surveys are shown in the bottom panel of Table 3. When only vigorous and moderately paced walking time was included in the IPAQ total, the difference (compared with was 468 (SE:27 minutes, and when all walking was included in the IPAQ total the difference was 406 (SE:27 minutes. Differences between these two surveys amounted to more than 500 minutes per week when results were used. These differences between the times reported in the and the other three surveys are illustrated graphically using Bland and Altman plots in Figure 1. From these, it is clear that there was reasonable agreement between the and NHS surveys (mean difference close to zero, but the funnel shape that is evident in all three graphs indicates that absolute agreement worsened as the average time reported increased. Given that most of the health benefits of PA are seen with PA durations of up to one hour per day (about 400 minutes per week, it is of less concern in terms of PA measurement that the poorest agreement was seen for levels of about 800 minutes per week. The same pattern was evident for the other two comparisons, but with a much larger mean difference for the and comparison. The data in Table 3 also show the mean number of times (AA, NHS or days (IPAQ, BRFSS reported in each pair of surveys. The patterns of difference between the surveys were similar to those described for total minutes (above, with the lowest frequency reported for the NHS, and highest for the BRFSS and IPAQ. Analysis of the order of survey administration found very little ordering effect of the questions in the comparisons between AA and both NHS and BRFSS. However, in the AA and IPAQ comparison, there was a trend for AA to score more total minutes and number of days/sessions when IPAQ was asked first, compared with the results when AA was asked first, but the trend was not statistically significant (data not shown. Prevalence data sufficient physical activity Measures of agreement for the prevalence of sufficiently active and sedentariness from the three pairs of surveys are shown in Table 4. This table shows the proportions meeting the 150- minute criterion (top panel, the proportions meeting both the 150 minutes and five days/sessions criteria (middle panel, and the proportions reporting no activity (bottom panel. Per cent agreement coefficients and kappa coefficients with 95% confidence intervals are shown for the paired data in each row. The first row of data shows that prevalence estimates from AA and NHS were similar, especially when data were used. Compared with, prevalence was about 10% higher when BRFSS data were used. However, there was closer agreement between prevalence estimates from BRFSS and, but the Kappa scores were in the modest range. When walking was double counted in the five item BRFSS, prevalence estimates were much higher but the per cent agreement and Kappa scores were not markedly changed. The relationship between the AA and the short IPAQ questionnaire indicates that use of the IPAQ survey resulted in much higher prevalence estimates than the AA instrument, across all modes of measurement. For example, there was a prevalence difference of around 26% between the and the short IPAQ on the criterion of 150 minutes per week. When the 150 minutes and five sessions criteria were used, the relationships between the three pairs of surveys were similar to those described above, but the prevalence estimates were between 13% and 19% lower than when the single 150 minutes criterion was used (see Table 4, middle panel. Prevalence data sedentariness The data in the lower panel of Table 4 show that, in general, the more items included in each version of each survey, the lower the prevalence of sedentariness, with overall estimates of sedentariness higher when the AA, NHS and two item BRFSS were used. Per cent agreement scores were higher than those shown for activity but Kappa scores were lower for sedentariness than for activity in the AA/NHF and AA/BRFSS comparisons. Discussion The aim of this study was to compare agreement between prevalence estimates for sufficient activity derived from four survey instruments, using a randomly selected population sample. The AA survey was used as a reference as it is in current population use for PA surveillance in Australia. The comparison instruments were the traditional Australian NHS measure, the US BRFSS, and the recently developed IPAQ. The findings clearly demonstrated that there were large differences in the reported PA times and hence in the prevalence estimates of sufficient activity from these four measures. The highest level of agreement was between the AA and NHS and the poorest was between the AA and IPAQ. This reflects the different domains of activity included in each survey. Traditionally, PA surveys have focused on leisure-time PA (LTPA, because, until the mid 1990s, the health benefits of PA were largely attributed to 2004 VOL. 28 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 131

Brown et al. Figure 1a-c: Plots of differences in total weekly minutes for pairs of comparison surveys (y axis, against the average time reported in the two surveys (x axis. 1a: comparison of NHS v AA4; 1b: comparison of BRFSS2 v AA4; 1c: comparison of IPAQmv v AA4. Figure 1a Weekly minutes (NHS, Average in minutes (NHS and Figure 1b Weekly minutes (, Average in minutes ( and Figure 1c Weekly minutes (, Difference in minutes ( - Difference in minutes ( - Difference in minutes (NHS- Average in minutes ( and 132 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 2

Comparison of PA surveys participation in more vigorous forms of activity such as organised sport and fitness. More recently, as interest in the obesity epidemic has focused our attention on total energy expenditure, there have been attempts to include different domains of activity, such as transport and occupational activity, in PA surveys. 1 In some cultures, energy expended in work and transport contributes much more to total energy expenditure than LTPA, and the IPAQ survey probably does best at capturing activity in all these domains. As a result, the minutes of PA reported in response to the IPAQ survey are higher than for the other three surveys, and prevalence estimates are concomitantly higher. This has implications for policy, as prevalence estimates are used to derive the population attributable risks of behaviours, which are in turn used to determine priority areas for public health action. For example, using the NHS,, and we would estimate that about half the population is not sufficiently active for health benefit, compared with 20-30% using IPAQ. More research into the relative health benefits of activity in each PA domain is required to assess whether the threshold for categorising activity should be changed if more domains are included, and criterion-based validity data are urgently required for all four of these surveys. One contributing factor to the higher times reported in the IPAQ may be that this survey was the only one of the four to ask respondents to think about a usual week. While this will overcome any problems relating to the fact that the last week may be Table 4: Per cent agreement and Kappa scores for the proportion of participants categorised as sufficiently active (a using the criterion of 150 minutes/week (top panel, (b using the criterion of 150 minutes and 5 sessions/week (middle panel, and (c for the proportion of participants categorised as sedentary (no physical activity of any kind reported, bottom panel for each pair of surveys. Active Alternative n AA Alternative % Kappa Australia prevalence prevalence agreement (95% CI (a NHS 428 56.0 55.1 75.3 0.50 (0.42-0.58 150 min/week 427 58.5 68.3 69.2 0.34 (0.25-0.43 427 58.5 82.9 67.1 0.26 (0.18-0.34 IPAQ 425 59.6 85.9 70.2 0.31 (0.23-0.39 425 59.6 83.3 70.5 0.32 (0.24-0.40 NHS 428 66.6 55.1 72.9 0.44 (0.35-0.52 427 66.7 68.3 71.6 0.35 (0.26-0.45 427 66.7 82.9 72.4 0.29 (0.20-0.38 IPAQ 425 69.5 85.9 78.4 0.40 (0.30-0.49 425 69.5 83.3 78.1 0.41 (0.31-0.50 (b NHS 428 43.1 35.8 74.3 0.47 (0.38-0.55 150 min/week 427 42.7 48.7 68.2 0.36 (0.27-0.45 PLUS 427 42.7 74.0 59.2 0.24 (0.17-0.31 5 sessions IPAQ 425 42.9 72.9 65.4 0.35 (0.28-0.42 425 42.9 69.9 66.6 0.37 (0.29-0.44 NHS 428 51.2 35.8 69.6 0.40 (0.31-0.48 427 50.2 48.7 69.0 0.38 (0.29-0.47 427 50.2 74.0 63.3 0.26 (0.18-0.35 IPAQ 425 53.6 72.9 73.3 0.45 (0.37-0.53 425 53.6 69.9 73.6 0.46 (0.37-0.54 (c NHS 428 15.1 19.6 83.2 0.41 (0.29-0.52 Sedentary 427 16.5 16.8 79.5 0.26 (0.14-0.37 427 16.5 8.4 83.5 0.25 (0.13-0.37 IPAQ 425 16.2 4.5 86.7 0.31 (0.18-0.43 425 16.2 7.3 86.7 0.37 (0.24-0.50 NHS 428 8.3 19.6 83.1 0.30 (0.19-0.42 427 11.0 16.8 81.3 0.21 (0.09-0.33 427 11.0 8.4 87.1 0.26 (0.11-0.40 IPAQ 425 10.1 4.5 91.3 0.36 (0.20-0.52 425 10.1 7.3 91.3 0.45 (0.30-0.60 Notes: = four item Active Australia (both walking questions, moderate and vigorous activities, not including gardening. = five item Active Australia (both walking questions, moderate and vigorous activities and gardening. NHS= National Health Survey, walking, moderate and vigorous activities. = two item BRFSS, moderate and vigorous activities only. = three item BRFSS, including walking, moderate and vigorous activities. IPAQ = all paces of walking, plus moderate and vigorous activities. = only moderate and vigorous paced walking, and other moderate and vigorous activities. CI = confidence interval 2004 VOL. 28 NO. 2 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 133

Brown et al. atypical due to illness or holidays, it is likely that this form of questioning will result in more socially acceptable, and therefore more inflated responses. It may also be true that between survey differences are attributable in part to the assessment of activity duration. Notably, the BRFSS and the IPAQ require respondents to estimate the amount of time spent in each activity domain on a single day. In contrast, the AA and NHS ask respondents to estimate the time spent in each activity domain for the entire recall period. Given that physical activity behaviour exhibits considerable day-to-day variability, 4,11 it is possible that estimating the usual or average duration of activity on a single day may be inherently more difficult for respondents than recalling the duration of each activity session performed and summing over the entire recall period. One of the strengths of this study was that we used a randomly selected population sample with PA patterns similar to those previously reported in Australia. 2 A limitation of the study was that participants were asked to respond to two surveys on one occasion, and their responses to one survey may have influenced their responses to the other. However, the order of administration of the surveys did not greatly affect the results. Overall, this study has demonstrated the difficulty of intersurvey comparison when surveys are designed to include different domains of physical activity. Each survey captures different dimensions of total physical activity, ranging from the relatively narrow leisure time PA focus of the NHS to the more inclusive IPAQ survey, which attempts to have respondents think about all types of daily activity at work, at home and during recreation. Given that the variability between methods does not show consistent patterns, it is not possible to develop algorithms to translate prevalence estimates obtained from different surveys. In light of this, and of the public health need to track trends in PA over time, it is important to consistently use the same survey for population monitoring purposes. In Australia, the NHS items have been used by the ABS since 1985, but prevalence estimates are lower than for the other three surveys because of the focus on more structured leisure-time activities. The AA survey was developed with a view to capturing more of the moderate intensity activity associated with today s lifestyle. As it has now been used three times in national surveys and as it appears to perform as well as any of the available alternatives, its continued use for population surveys is recommended so that trend data over a longer period of time can be established. The IPAQ survey may be useful for evaluation of intervention trials, when more details about changes in different domains of physical activity are required. Acknowledgements This work was supported by a grant from the Australian Commonwealth Department of Health and Ageing. References 1. Bauman A, Merom D. Measurement and surveillance of physical activity in Australia an introductory guide. Australas Epidemiol 2002;9(2:2-5. 2. Armstrong T, Bauman A, Davies J. Physical Activity Patterns of Australian Adults. Canberra (ACT: Australian Institute of Health and Welfare; 2000. AIHW Catalogue No.: CVD 10 3. Macera CA, Pratt M. Public health surveillance of physical activity. Res Q Exerc Sport 2000;71:S97-103. 4. Craig CL, Marshall AL, Sjostrom M, et al. International Physical Activity Questionnaire: 12 country reliability and validity. Med Sci Sports Exerc 2003;35(3:1381-96. 5. Ainsworth BE, Haskell WL, Whitt MC, Irwin ML, Swartz AM, Strath SJ, et al. Compendium of physical activities: classificiation of energy costs of human physical activities. Med Sci Sports Exerc 2000;32(9:S498-S515. 6. National Physical Activity Guidelines. Canberra (ACT: Commonwealth Department of Health and Aged Care; 1999. 7. United States Department of Health and Human Services. Physical Activity and Health [report]. Atlanta (GA: Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; 1996. 8. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;327:307-10. 9. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960;20:37-46. 10. Australian Institute of Health and Welfare. Physical Activity Patterns of Queensland Adults. Trends from 1999 to 2001. Brisbane (Qld: Queensland Health; 2003. 11. Washburn R, Heath G, Jackson A. Reliability and validity issues concerning large scale surveillance of physical activity. Res Q Exerc Sci 2000;71:104-13. 134 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2004 VOL. 28 NO. 2