J Clin Epidemiol Vol. 45, No. 8, pp. 877-883, Printedin Great Britain.All rights reserved
0895-4356/92 55.00+ 0.00 Copyright0 1992PergamonPressLtd
1992
COMPARISON OF TWO PHYSICAL ACTIVITY QUESTIONNAIRES, WITH A DIARY, FOR ASSESSING PHYSICAL ACTIVITY IN AN ELDERLY POPULATION B. CARTMEL* and T. E. MOON Arizona Cancer Center and the Department of Family and Community Medicine, University of Arizona College of Medicine, Tucson, AZ 85724, U.S.A. (Received in revised form 29 January 1992)
Abstract-Measurement of physical activity in epidemiological studies is usually achieved by means of a questionnaire. Little work has heen done to determine which questionnaire format has greater validity in an elderly population. In this study of elderly subjects, physical activity as reported in two self-administered questionnaires (A and B), which differed in format and length, were compared to activity reported in a 4 day diary. As compared with the diary, moderate/heavy activity was more accurately reported in Questionnaire A (mean difference 5 min), the longer more detailed questionnaire, than B (mean difference 170 min). Light activity was under reported in Questionnaire A (mean difference 68 min) and over reported in B (mean difference 88 min) as compared with the diary. In contrast, time spent sitting was more accurately reported in Questionnaire B (mean difference 40 min) than in A (mean difference 230 min) as compared with the diary. The longer more detailed questionnaire was the more accurate instrument for assessing moderate/heavy activity in this elderly population. The shorter questionnaire was more accurate for assessing time spent sitting. Elderly
Physical activity
Questionnaire
INTRODUCTION Physical activity has been reported as, directly or indirectly, a protective factor for many diseases common in the elderly. A direct association has been found between physical inactivity and cardiovascular disease, osteoporosis and colon cancer [l-4]. Physical activity is also an important factor in weight control, obesity is a risk factor for diabetes mellitus, cancer and cardiovascular disease [ 5-71. Measurement of physical activity in free-living individuals can be achieved in three ways: measurement by a mechanical monitor [8], by the individual maintaining a record of activity, or by recall of activity. In large epidemiological studies *All correspondence should be addressed to: Dr B. Cartmel, The Arizona Cancer Center, 15I5N Campbell, Tucson, AZ 85724, U.S.A.
the latter method is the most widely used, with both interview and self-administered question-
naires developed for assessing past or present physical activity. Questionnaires for assessing activity differ in format and length [9]. Since detailed questionnaires can be more frustrating to the elderly than to younger persons [ 10, 111, a short physical activity questionnaire may be more readily accepted and more accurately completed by elders than a longer more detailed one. In this study activity from two self-administered physical activity recall questionnaires (A and B), of differing length and detail, were compared with activity recorded in a 4 day diary, in a population of men and women aged 59 and over. The questionnaires were adapted from those of the Minnesota Leisure-Time Physical Activity Questionnaire (MLTPAQ) [ 121 (Questionnaire A) and the 7 Day Physical 877
878
B. CARTMELand T. E. MWN
Activity Questionnaire (SDPAQ) as used in the Five Cities Project [13] (Questionnaire B). The MLTPAQ, with a reference period of the last year, has been validated for interview administration using several different criteria. Leon et al. [14] reported a significant correlation of 0.41 between total leisure time physical activity (LTPA) and a graded treadmill test, and 0.44 between heavy LTPA and the graded treadmill test. In contrast, DeBacker et al. [15] reported values of 0.08 and 0.11, for total and heavy LTPA, respectively, as measured by the MLTPAQ, when correlated with aerobic capacity in a population of 3179 Belgian males. Similarly, correlations of 0.10 and 0.09, were reported for total and heavy LTPA, respectively, with cardiorespiratory fitness in 238 1 male Slovaks [16]. Validation of the questionnaire with a Large Scale Activity Monitor was conducted by LaPorte et al. [ 171 in 20 male undergraduates and resulted in a correlation of 0.69 for total LTPA. A reliability study of the MLTPAQ was conducted by Folsom et al. [ 181 in men and women aged 25-74; rank-order correlation coefficients of 0.88, 0.79, 0.86 and 0.82 were reported for total, light, moderate, and heavy LTPA scores, respectively. LTPA as assessed by the MLTPAQ has been shown to be positively associated with high density lipoprotein (HDL) cholesterol (0.09 for males and 0.10 for females) in a survey of 878 men and 738 women aged 25-74 [19]. Among males aged 25-75, increased levels of heavy LTPA were inversely associated with a reduced risk of primary cardiac arrest [20]. The SDPAQ as its name suggests has a reference period of 7 days. Validation of the interview administered SDPAQ using a daily self-report log was conducted by Taylor et al. [21] in 30 male subjects (aged 34-69) and, similarly by Dishman and Steinhardt [22] in 158 university students; both groups reported similar correlations of 0.81 and 0.82-0.87, respectively, for total LTPA. Validation of the SDPAQ using energy expenditure, measured by 24 h dietary recall, in 495 males and 545 females, resulted in correlations of 0.16 in males and 0.09 in females [23]; validation of the SDPAQ against maximal oxygen uptake ( VOtmaxin 8 1 sedentary males resulted in a correlation of 0.61. The test-retest reliability of the SDPAQ, administered in weeks 1 and 5, and 5 and 9 to participants in an exercise program, resulted in correlations of 0.58 and 0.63, respectively [22].
The energy expenditure assessed by the SDPAQ has been shown to be significantly associated with body mass index (r = - 0.16) and the ratio of HDL to low density lipoprotein (LDL) (r = 0.15) in a study of female adults. However similar significant relationships were not seen in males [24]. The MLTPAQ and SDPAQ questionnaires differed in format and the time required for completion but both assess physical activity quantitatively. In this study it was found that Questionnaire A, the longer more detailed questionnaire, was the more accurate instrument for use in measuring moderate/heavy activity. However, Questionnaire B was the more accurate instrument for use in assessing time spent sitting. METHODS
Nine white males, median age 68.5 (66 to 80) and 15 white females, median age 74 (59 to 83), were recruited as volunteers for the study. All subjects were retired and lived in the Sun Cities Community, an Arizonan retirement community. They were recruited via an advertisement in a community newsletter. All subjects had a minimum of 12 yr education. On entering the study each participant completed the first physical activity questionnaire (A) and were asked to keep a record of all activity performed over 4 consecutive days. Six weeks after starting the study they completed physical activity Questionnaire B. Both questionnaires were selfadministered and were completed with reference to the previous year. Activity diary
Each subject recorded their activities in 6-h periods, along with a brief description of the level of exertion if this was not self-evident. An example of a 6-h record was supplied to each subject to ensure that they knew the detail required when recording activity. Questionnaire A
Questionnaire A was based on the MLTPAQ of Taylor et al. [12], but was modified to include common household activities and sedentary activities such as reading at home and watching television. Written instructions were provided on how to complete the questionnaire. First the subject read through the list of 61 activities and checked “Yes” to any activity that they had performed in the last 12 months. The
879
Comparison of Two Physical Activity Questionnaires Table 1. Time spent sitting, in light activity and in heavy/moderate activity, as reported in each of three instruments(n = 24) Sitting
Diary Ques. A @es. B
Light activity
Heavy/moderate activity
Mean
SE
Mean
SE
Mean
SE
513 284 474
36.1 34.2 48.4
205 137 293
15.7 21.6 31.9
66 66 236
7.6 10.9 48.8
subject then returned to those activities performed, checking the months the activities were performed, “the average number of times per month” and the “time per occasion”. Questionnaire B
The format of Questionnaire B was similar to that used in the Five City Project [13]. Participants recorded how much time during an average weekday and weekend day he/she spent sleeping, sitting, doing light, medium and heavy activity. Subjects were asked to record their typical activity for a week in the previous year. Examples of activities were given on the questionnaire for each level of activity. The questionnaire format encouraged subjects to report duration of activities, including sleep, which summed to 24 hr. Analysis
Activities reported in the diary and Questionnaire A were divided into three categories: sitting, light activity (< 3.0 METS) and moderate/ heavy activity (> 3.0 METS). 1 MET approximates to 1 kcal/min for a 60 kg person [13]. Previously reported sources of intensity codes were used [12,25,26]. If an intensity code, MET or equivalent, could not be found in the literature for a specific activity the activity was assigned to a category based on similarity to other activities which have documented intensity codes. All coding was performed by one coder with review by a second coder. Activities specified in Questionnaire A are listed in the Appendix by intensity category:
sitting, light activity and moderate/heavy activity. The statistical analysis was based on the number of reported minutes per day for each of the three activity categories, and used the paired t-test. RESULTS The mean times reported as spent sitting, in light activity and moderate/heavy activity are shown in Table 1. The results from all three instruments showed that subjects spent most time sitting, and least time in moderate/heavy activity. The mean individual differences in time reported for each category of activity are shown in Table 2. Comparison of the diary and Questionnaire A
The number of minutes per day spent sitting and in light activity were significantly under reported in Questionnaire A as compared with the diary by an average of 230 (p < 0.001) and 68 (p = 0.008) min, respectively. However, time spent in moderate/heavy activity was similar in Questionnaire A and the diary, with only an average difference of 5 min/day (p = 0.59). Comparison of the diary and Questionnaire B
The time spent sitting was similar in Questionnaire B as compared with the diary (p = 0.47). However both light and moderate/ heavy activity were over reported as compared with the diary, an average of 88 (p = 0.019) and 170 (p = 0.018) min, respectively. The average time reported for moderate/heavy activity in Questionnaire B was three and a half times
Table 2. Mean individual differences in time (min) reported for sitting, light activity and heavy/moderate activity between the diary and two questionnaires (n = 24) Activity Sitting Light Moderate/heavy
Instrument comparison Diary Diary Diary Diary Diary Diary
vs vs vs vs vs vs
Ques. Ques. Ques. Ques. Ques. C&es.
A B A B A B
Mean difference
f
P
230 40 68 -88 -5 -170
5.16 0.72 2.91 -2.52 -0.55 -3.53
0895-4356/92 55.00+ 0.00 Copyright0 1992PergamonPressLtd
1992
COMPARISON OF TWO PHYSICAL ACTIVITY QUESTIONNAIRES, WITH A DIARY, FOR ASSESSING PHYSICAL ACTIVITY IN AN ELDERLY POPULATION B. CARTMEL* and T. E. MOON Arizona Cancer Center and the Department of Family and Community Medicine, University of Arizona College of Medicine, Tucson, AZ 85724, U.S.A. (Received in revised form 29 January 1992)
Abstract-Measurement of physical activity in epidemiological studies is usually achieved by means of a questionnaire. Little work has heen done to determine which questionnaire format has greater validity in an elderly population. In this study of elderly subjects, physical activity as reported in two self-administered questionnaires (A and B), which differed in format and length, were compared to activity reported in a 4 day diary. As compared with the diary, moderate/heavy activity was more accurately reported in Questionnaire A (mean difference 5 min), the longer more detailed questionnaire, than B (mean difference 170 min). Light activity was under reported in Questionnaire A (mean difference 68 min) and over reported in B (mean difference 88 min) as compared with the diary. In contrast, time spent sitting was more accurately reported in Questionnaire B (mean difference 40 min) than in A (mean difference 230 min) as compared with the diary. The longer more detailed questionnaire was the more accurate instrument for assessing moderate/heavy activity in this elderly population. The shorter questionnaire was more accurate for assessing time spent sitting. Elderly
Physical activity
Questionnaire
INTRODUCTION Physical activity has been reported as, directly or indirectly, a protective factor for many diseases common in the elderly. A direct association has been found between physical inactivity and cardiovascular disease, osteoporosis and colon cancer [l-4]. Physical activity is also an important factor in weight control, obesity is a risk factor for diabetes mellitus, cancer and cardiovascular disease [ 5-71. Measurement of physical activity in free-living individuals can be achieved in three ways: measurement by a mechanical monitor [8], by the individual maintaining a record of activity, or by recall of activity. In large epidemiological studies *All correspondence should be addressed to: Dr B. Cartmel, The Arizona Cancer Center, 15I5N Campbell, Tucson, AZ 85724, U.S.A.
the latter method is the most widely used, with both interview and self-administered question-
naires developed for assessing past or present physical activity. Questionnaires for assessing activity differ in format and length [9]. Since detailed questionnaires can be more frustrating to the elderly than to younger persons [ 10, 111, a short physical activity questionnaire may be more readily accepted and more accurately completed by elders than a longer more detailed one. In this study activity from two self-administered physical activity recall questionnaires (A and B), of differing length and detail, were compared with activity recorded in a 4 day diary, in a population of men and women aged 59 and over. The questionnaires were adapted from those of the Minnesota Leisure-Time Physical Activity Questionnaire (MLTPAQ) [ 121 (Questionnaire A) and the 7 Day Physical 877
878
B. CARTMELand T. E. MWN
Activity Questionnaire (SDPAQ) as used in the Five Cities Project [13] (Questionnaire B). The MLTPAQ, with a reference period of the last year, has been validated for interview administration using several different criteria. Leon et al. [14] reported a significant correlation of 0.41 between total leisure time physical activity (LTPA) and a graded treadmill test, and 0.44 between heavy LTPA and the graded treadmill test. In contrast, DeBacker et al. [15] reported values of 0.08 and 0.11, for total and heavy LTPA, respectively, as measured by the MLTPAQ, when correlated with aerobic capacity in a population of 3179 Belgian males. Similarly, correlations of 0.10 and 0.09, were reported for total and heavy LTPA, respectively, with cardiorespiratory fitness in 238 1 male Slovaks [16]. Validation of the questionnaire with a Large Scale Activity Monitor was conducted by LaPorte et al. [ 171 in 20 male undergraduates and resulted in a correlation of 0.69 for total LTPA. A reliability study of the MLTPAQ was conducted by Folsom et al. [ 181 in men and women aged 25-74; rank-order correlation coefficients of 0.88, 0.79, 0.86 and 0.82 were reported for total, light, moderate, and heavy LTPA scores, respectively. LTPA as assessed by the MLTPAQ has been shown to be positively associated with high density lipoprotein (HDL) cholesterol (0.09 for males and 0.10 for females) in a survey of 878 men and 738 women aged 25-74 [19]. Among males aged 25-75, increased levels of heavy LTPA were inversely associated with a reduced risk of primary cardiac arrest [20]. The SDPAQ as its name suggests has a reference period of 7 days. Validation of the interview administered SDPAQ using a daily self-report log was conducted by Taylor et al. [21] in 30 male subjects (aged 34-69) and, similarly by Dishman and Steinhardt [22] in 158 university students; both groups reported similar correlations of 0.81 and 0.82-0.87, respectively, for total LTPA. Validation of the SDPAQ using energy expenditure, measured by 24 h dietary recall, in 495 males and 545 females, resulted in correlations of 0.16 in males and 0.09 in females [23]; validation of the SDPAQ against maximal oxygen uptake ( VOtmaxin 8 1 sedentary males resulted in a correlation of 0.61. The test-retest reliability of the SDPAQ, administered in weeks 1 and 5, and 5 and 9 to participants in an exercise program, resulted in correlations of 0.58 and 0.63, respectively [22].
The energy expenditure assessed by the SDPAQ has been shown to be significantly associated with body mass index (r = - 0.16) and the ratio of HDL to low density lipoprotein (LDL) (r = 0.15) in a study of female adults. However similar significant relationships were not seen in males [24]. The MLTPAQ and SDPAQ questionnaires differed in format and the time required for completion but both assess physical activity quantitatively. In this study it was found that Questionnaire A, the longer more detailed questionnaire, was the more accurate instrument for use in measuring moderate/heavy activity. However, Questionnaire B was the more accurate instrument for use in assessing time spent sitting. METHODS
Nine white males, median age 68.5 (66 to 80) and 15 white females, median age 74 (59 to 83), were recruited as volunteers for the study. All subjects were retired and lived in the Sun Cities Community, an Arizonan retirement community. They were recruited via an advertisement in a community newsletter. All subjects had a minimum of 12 yr education. On entering the study each participant completed the first physical activity questionnaire (A) and were asked to keep a record of all activity performed over 4 consecutive days. Six weeks after starting the study they completed physical activity Questionnaire B. Both questionnaires were selfadministered and were completed with reference to the previous year. Activity diary
Each subject recorded their activities in 6-h periods, along with a brief description of the level of exertion if this was not self-evident. An example of a 6-h record was supplied to each subject to ensure that they knew the detail required when recording activity. Questionnaire A
Questionnaire A was based on the MLTPAQ of Taylor et al. [12], but was modified to include common household activities and sedentary activities such as reading at home and watching television. Written instructions were provided on how to complete the questionnaire. First the subject read through the list of 61 activities and checked “Yes” to any activity that they had performed in the last 12 months. The
879
Comparison of Two Physical Activity Questionnaires Table 1. Time spent sitting, in light activity and in heavy/moderate activity, as reported in each of three instruments(n = 24) Sitting
Diary Ques. A @es. B
Light activity
Heavy/moderate activity
Mean
SE
Mean
SE
Mean
SE
513 284 474
36.1 34.2 48.4
205 137 293
15.7 21.6 31.9
66 66 236
7.6 10.9 48.8
subject then returned to those activities performed, checking the months the activities were performed, “the average number of times per month” and the “time per occasion”. Questionnaire B
The format of Questionnaire B was similar to that used in the Five City Project [13]. Participants recorded how much time during an average weekday and weekend day he/she spent sleeping, sitting, doing light, medium and heavy activity. Subjects were asked to record their typical activity for a week in the previous year. Examples of activities were given on the questionnaire for each level of activity. The questionnaire format encouraged subjects to report duration of activities, including sleep, which summed to 24 hr. Analysis
Activities reported in the diary and Questionnaire A were divided into three categories: sitting, light activity (< 3.0 METS) and moderate/ heavy activity (> 3.0 METS). 1 MET approximates to 1 kcal/min for a 60 kg person [13]. Previously reported sources of intensity codes were used [12,25,26]. If an intensity code, MET or equivalent, could not be found in the literature for a specific activity the activity was assigned to a category based on similarity to other activities which have documented intensity codes. All coding was performed by one coder with review by a second coder. Activities specified in Questionnaire A are listed in the Appendix by intensity category:
sitting, light activity and moderate/heavy activity. The statistical analysis was based on the number of reported minutes per day for each of the three activity categories, and used the paired t-test. RESULTS The mean times reported as spent sitting, in light activity and moderate/heavy activity are shown in Table 1. The results from all three instruments showed that subjects spent most time sitting, and least time in moderate/heavy activity. The mean individual differences in time reported for each category of activity are shown in Table 2. Comparison of the diary and Questionnaire A
The number of minutes per day spent sitting and in light activity were significantly under reported in Questionnaire A as compared with the diary by an average of 230 (p < 0.001) and 68 (p = 0.008) min, respectively. However, time spent in moderate/heavy activity was similar in Questionnaire A and the diary, with only an average difference of 5 min/day (p = 0.59). Comparison of the diary and Questionnaire B
The time spent sitting was similar in Questionnaire B as compared with the diary (p = 0.47). However both light and moderate/ heavy activity were over reported as compared with the diary, an average of 88 (p = 0.019) and 170 (p = 0.018) min, respectively. The average time reported for moderate/heavy activity in Questionnaire B was three and a half times
Table 2. Mean individual differences in time (min) reported for sitting, light activity and heavy/moderate activity between the diary and two questionnaires (n = 24) Activity Sitting Light Moderate/heavy
Instrument comparison Diary Diary Diary Diary Diary Diary
vs vs vs vs vs vs
Ques. Ques. Ques. Ques. Ques. C&es.
A B A B A B
Mean difference
f
P
230 40 68 -88 -5 -170
5.16 0.72 2.91 -2.52 -0.55 -3.53
