Abstract
Objective:
Inactive lifestyles contribute to health problems and premature death and are influenced by the physical environment. The primary objective of this study was to quantify patterns of physical inactivity in New York City and the United States by combining data from surveys and accelerometers.
Methods:
We used Poisson regression models and self-reported survey data on physical activity and other demographic characteristics to predict accelerometer-measured inactivity in New York City and the United States among adults aged ≥18. National data came from the 2003-2004 and 2005-2006 National Health and Nutrition Examination Surveys. New York City data came from the 2010-2011 New York City Physical Activity and Transit survey.
Results:
Self-reported survey data indicated no significant differences in inactivity between New York City and the United States, but accelerometer data showed that 53.1% of persons nationally, compared with 23.4% in New York City, were inactive (P < .001). New Yorkers reported a median of 139 weekly minutes of transportation activity, compared with 0 minutes nationally. Nationally, 50.0% of self-reported activity minutes came from recreation activity, compared with 17.5% in New York City. Regression models indicated differences in the association between self-reported minutes of transportation and recreation and accelerometer-measured inactivity in the 2 settings.
Conclusions:
The prevalence of physical inactivity was higher nationally than in New York City. The largest difference was in walking behavior indicated by self-reported transportation activity. The study demonstrated the feasibility of combining accelerometer and survey measurement and that walkable environments promote an active lifestyle.
Keywords: accelerometer measurements, sedentary lifestyle, place-based activity
Physical inactivity is the fourth leading risk factor for premature death in the United States, and it contributes to the second, third, and fifth leading risk factors—high blood pressure, obesity, and high blood glucose, respectively—for premature death.1 The largest effect of physical activity on health occurs when individuals shift from no activity to some activity.2 Yet physical inactivity is difficult to measure, and the 2 most common instruments for measurement—surveys (self-reported data) and accelerometers (objective devices)—produce different results.3
Self-reported survey data may be subject to overreporting, particularly among less active persons, because of social desirability bias or recall difficulties. Survey data, compared with accelerometer data, show higher levels of physical activity and lower levels of inactivity.4-8 These differences may relate to overreporting in survey data. However, accelerometer studies are expensive and difficult to administer on a population level and, thus, are rare. Accelerometer data also may undercount activity and may therefore overestimate inactivity because accelerometers, which focus only on forward motion, do not accurately record certain types of physical activity.4
Several previous studies examined the effect of the physical (or built) environment on physical activity. Walking behavior is linked to urban features, such as street connectivity, residential density, and land-use mix.9 New York City differs from the rest of the country in its density, well-developed public transportation systems, and difficulty of car transportation, all of which promote active transportation.
To better understand the relationship between physical inactivity and the built environment, we compared physical activity measured in New York City and nationally. We examined responses to survey questions on domain-specific activity from self-reported surveys and analyzed objectively measured accelerometer data. The primary objective of this study was to quantify patterns of physical inactivity in New York City and the United States by combining survey and accelerometer data.
Methods
Data Collection and Processing
This study used data from 2 sources: the New York City Physical Activity and Transit (PAT) survey, described in detail elsewhere,4,10 and the National Health and Nutrition Examination Survey (NHANES).11,12 The PAT survey was conducted in 2 waves, in September–November 2010 and in March–November 2011; it used a random-digit-dial (both cell and landline) telephone sample design to recruit a representative sample of mobile (able to walk >10 feet) New York City adults aged ≥18. PAT survey data were weighted to be representative of all New York City adults. All survey respondents were asked to complete a survey that asked about physical activity in 3 domains: home, recreation, and transportation. Transportation-related physical activity was defined as walking and bicycling. The survey recorded activity in bouts, or episodes, of ≥10 minutes. The survey asked whether recreation-related activity was moderate-intensity or vigorous-intensity activity. All home-related and transportation-related activity was considered to be moderate-intensity activity.
All participants in the second wave of the PAT survey (n = 2488) were invited to participate in a follow-up study in which they would wear a waist-worn Actigraph GT3x accelerometer and carry a DG Globatstat 100 Global Positioning System (GPS) device for 7 days. Participants who did not wear the accelerometer for ≥4 days, 10 hours per day, were excluded from analysis.4,10 A total of 1134 participants (46%) agreed to take part in the follow-up device study. Of these, 803 (71%) participants returned the device with any GPS data and 679 (60%) participants provided ≥4 ten-hour days of valid accelerometer data. A total of 674 participants had complete information for our analyses; 5 participants did not have information on body mass index and were excluded. Data were weighted by ABT-SRBI, a market research firm, to represent the adult population of New York City.
We analyzed national data from the 2003-2004 and 2005-2006 cycles of the NHANES, which are, to our knowledge, the most comprehensive national publicly available accelerometer data. NHANES uses a multistage probability sampling design to generate a sample representative of the US civilian noninstitutionalized population.11,12 We combined these 2 cycles of data for adults aged ≥20; 20 is the minimum age in the adult NHANES sample; 5776 participants provided valid accelerometer and survey data. NHANES used a waist-worn Actigraph AM-7164 accelerometer.
Both surveys asked similar questions about home, recreation, and transportation activity, and NHANES asked detailed questions about recreation activity. NHANES first asked about participation in various types of recreation (eg, running, basketball) and then asked questions to collect data on minutes of vigorous-intensity activity and minutes of moderate-intensity activity for each type of recreation. The PAT survey asked a question about minutes of total vigorous-intensity recreation activity and a question about minutes of moderate-intensity recreation activity. The PAT survey asked separate questions about minutes of walking and biking activity, whereas NHANES asked a single question. The 2003-2004 and 2005-2006 NHANES asked participants to recall physical activity in the past 30 days, whereas the PAT survey asked questions based on a 7-day period. We scaled NHANES 30-day data in our study to 1 week.
The protocols for accelerometer data for the PAT survey and NHANES were identical: nonwear time was defined as an interval of ≥60 consecutive minutes of 0 counts, allowing for intervals of 1 or 2 minutes with a count value of <100, where an active minute had ≥2020 counts.11,12 A 10-minute bout of activity was defined as a consistent period of ≥10 minutes of activity consisting of ≥2020 counts per minute. To match with the 2008 Physical Activity Guidelines for Americans,2 which recommend ≥150 minutes per week of moderate-intensity physical activity or ≥75 minutes of vigorous-intensity physical activity, we estimated weekly physical activity in each data set. For these estimates, we averaged data for each valid day and then multiplied the average physical activity on each valid day by 7. All participants in the final sample had ≥4 days of valid data per NHANES protocol.12
Statistical Analysis
To assess differences in physical inactivity between the United States and New York City, we first tabulated data on socioeconomic characteristics for the New York City sample and the US sample. We next summarized participants in each setting by domain. We calculated the percentage (with 95% confidence intervals [CIs]) for no overall activity and no activity in the 3 domains, and we used 2-sample t tests to test for differences in percentages between New York City and the United States. New York City is part (approximately 3%) of the US population, and the t test assumes these are distinct populations. Yet if the subgroup is sufficiently small (<10% of the overall group), no adjustment is needed.13 We tabulated mean and median values and 95% CIs. We used 2-sample t tests to test for differences between New York City and the United States in mean minutes of domain-specific activity. We calculated minutes of moderate-to-vigorous physical activity as minutes of moderate-intensity activity plus 2 times the minutes of vigorous-intensity activity. For the accelerometer data, we calculated mean and median values (and 95% CIs) for moderate-to-vigorous physical activity in 10-minute bouts for New York City and the United States.
We examined the associations between self-reported activity and accelerometer-measured activity in New York City and the United States. We first calculated prevalence ratios (PRs) and 95% CIs. To model the association, we used a Poisson regression with a log link function and a generalized estimating equation to estimate robust standard errors. We chose a Poisson regression because our outcomes were common, often representing >20% of outcomes. To avoid overestimating the standard errors, we adjusted the weights to sum to the size of the sample rather than to the adult population of New York City. The dependent variable was an accelerometer measurement of no activity, defined as no 10-minute bouts of moderate-intensity or vigorous-intensity physical activity in 1 week. The independent variables of interest were self-reported recreation, transportation, and home activity, each categorized as 0 minutes per week, 1-149 minutes per week, and ≥150 minutes per week, to parallel the cutoffs from the 2008 Physical Activity Guidelines for Americans.2 We adjusted the models for sex, race/ethnicity, age group, education, poverty status (household income <200% federal poverty level [FPL], 200%-400% FPL, or >400% FPL),14 and categories of body mass index (underweight/normal weight, overweight, obese).15
The institutional review board of the New York City Department of Health and Mental Hygiene approved the PAT study as human subjects research. We analyzed data by using SAS version 9.2 and SUDAAN version 11.0 to adjust for the complex survey design and clustering.16,17 We conducted the analysis in March 2018.
Results
In New York City, the percentage of Hispanic participants (25.6%), non-Hispanic black participants (22.1%), and participants living at <200% FPL (37.0%) was significantly higher than in the US sample (Table 1). Although 73.8% (95% CI, 72.7%-74.9%) of the US sample was non-Hispanic white, only 36.3% (95% CI, 29.8%-43.4%) of the New York City sample was non-Hispanic white (P < .001). Participants in the US sample were significantly more likely than participants in the New York City sample to be obese.
Table 1.
Characteristics of study populations in a comparison of accelerometer and survey data on patterns of physical inactivity in the United States (2003-2006) and New York City (2010 and 2011)a
| Characteristic | % (95% CI) | P Value for Differenceb | |
|---|---|---|---|
| United States (n = 5776) | New York City (n = 674) | ||
| No overall physical activity | 11.1 (10.3-12.1) | 8.2 (4.7-14.0) | .21 |
| Domain-specific inactivity | |||
| No home activity | 28.8 (27.5-30.2) | 66.6 (59.8-72.8) | <.001 |
| No recreation activity | 31.0 (29.6-32.4) | 62.5 (55.2-69.3) | <.001 |
| No transportation activity | 75.7 (74.4-77.1) | 16.9 (12.0-23.3) | <.001 |
| Race/ethnicityc | |||
| Non-Hispanic white | 73.8 (72.7-74.9) | 36.3 (29.8-43.4) | <.001 |
| Non-Hispanic black | 9.6 (9.1-10.2) | 22.1 (17.2-28.0) | <.001 |
| Hispanic | 11.2 (10.5-12.0) | 25.6 (19.6-32.7) | <.001 |
| Asian | 0 | 12.7 (7.9-19.8) | <.001 |
| Otherd | 5.3 (4.6-6.1) | 3.2e (1.3-8.0) | .18 |
| Educationc | |||
| <High school diploma | 15.3 (14.3-16.2) | 19.5 (13.8-26.7) | .20 |
| High school diploma/some college | 57.7 (56.1-59.2) | 47.6 (40.4-55.0) | .01 |
| College degree | 27.1 (25.7-28.6) | 32.9 (26.6-40.0) | .10 |
| Percentage of federal poverty levelf | |||
| <200 | 26.6 (25.3-27.8) | 37.0 (30.4-44.2) | .004 |
| 200-400 | 32.0 (30.5-33.5) | 19.2 (13.8-26.1) | <.001 |
| >400 | 37.4 (35.9-39.0) | 31.1 (24.9-38.1) | .07 |
| Age group,c y | |||
| 18-29 | 14.7 (13.7-15.8) | 19.3 (13.9-26.2) | .15 |
| 30-44 | 29.9 (28.4-31.4) | 35.4 (28.4-43.0) | .15 |
| 45-64 | 37.9 (36.3-39.4) | 31.2 (25.3-37.7) | .04 |
| ≥65 | 17.6 (16.6-18.6) | 14.2 (10.4-19.1) | .13 |
| Sex | |||
| Male | 48.5 (46.9-50.1) | 46.4 (39.1-53.8) | .58 |
| Female | 51.5 (49.9-53.1) | 53.6 (46.2-60.9) | .58 |
| Body mass index, kg/m2 | |||
| Normal weight (18.5-24.9)g | 32.7 (31.2-34.2) | 43.0 (35.8-50.5) | .01 |
| Overweight (25.0-29.9) | 35.1 (33.6-36.6) | 31.7 (25.3-38.8) | .33 |
| Obese (≥30.0) | 32.2 (30.7-33.7) | 25.3 (19.7-31.8) | .03 |
aData sources: For New York City, the New York City Physical Activity and Transit survey, conducted in 2 cycles (September–November 2010 and March–November 2011) among adults aged ≥184, 10; for the US sample, the 2003-2004 and 2005-2006 cycles of the National Health and Nutrition Examination Survey (NHANES) among adults aged ≥20.11,12 Activity does not include activity performed as part of paid labor. All data were self-reported, except data on body mass index from NHANES, which were collected by NHANES examiners.
bP values are based on a 2-sample t test, with P < .05 considered significant.
cCategories may not total to 100% because of rounding.
dOther includes Native American, multiple race, “don’t know,” and respondents who did not answer the question.
eEstimate should be interpreted with caution: its relative standard error (a measure of estimate precision) is greater than 30%, the half-width of the 95% CI is greater than 10, or the sample size is too small, making the estimate potentially unreliable.
fPercentages do not add to 100% because of missing data.
gIncludes small percentage of respondents who were underweight (body mass index <18.5 kg/m2).
Although we found no significant difference in self-reported overall physical inactivity between the US sample and New York City (US sample, 11.1% [95% CI, 10.3%-12.1%] vs New York City, 8.2% [95% CI, 4.7%-14.0%]), we found significant differences in physical activity by domain (Table 1). In the US sample, 28.8% (95% CI, 27.5%-30.2%) reported no home activity, compared with 66.6% (95% CI, 59.8%-72.8%) in New York City (P < .001). In the US sample, 31.0% (95% CI, 29.6%-32.4%) reported no recreation activity, compared with 62.5% (95% CI, 55.2%-69.3%) in New York City (P < .001). In the US sample, 75.7% (95% CI, 74.4%-77.1%) reported no transportation activity, compared with 16.9% (95% CI, 12.0%-23.3%) in New York City (P < .001).
In New York City, 97% of transportation-related physical activity came from walking and 3% came from bicycling. Accelerometer data showed that 53.1% (95% CI, 51.5%-54.7%) of the US sample had no physical activity (in 10-minute bouts) in a 1-week period, compared with 23.4% (95% CI, 17.7%-29.0%) in New York City (P < .001).
The US sample reported an average of 392 (95% CI, 375-410) minutes per week of overall activity, compared with 594 (95% CI, 452-736) minutes per week in New York City (P = .01) (Table 2). Most of this difference was attributable to transportation activity. New York City participants reported 304 (95% CI, 223-386) minutes per week of transportation activity, 6.6 times the activity of the US sample (46 minutes per week; 95% CI, 39-52) (P < .001). In contrast, the US sample reported almost twice the level of recreation activity (196 minutes per week; 95% CI, 186-206) than that of New York City (104 minutes per week; 95% CI, 69-139) (P < .001).
Table 2.
Mean and median minutes of weekly physical activity, collected by survey and accelerometer, in a study on patterns of physical inactivity in the United States (2003-2006) and New York City (2010 and 2011)a
| Type of Physical Activity | United States (n = 5776) | New York City (n = 674) | P Value for Difference in Meansb | ||
|---|---|---|---|---|---|
| Mean (95% CI) | Median (95% CI) | Mean (95% CI) | Median (95% CI) | ||
| Self-reported via survey | |||||
| Overall activity | 392 (375-410) | 221 (209-232) | 594 (452-736) | 311 (260-362) | .01 |
| Homec | 151 (141-162) | 53 (47-58) | 185 (104-267) | 0 | .41 |
| Transportationc | 46 (39-52) | 0 | 304 (223-386) | 139 (118-160) | <.001 |
| Recreation | 196 (186-206) | 80 (70-90) | 104 (69-139) | 0 | <.001 |
| Moderate | 132 (124-140) | 42 (32-45) | 60 (33-86) | 0 | <.001 |
| Vigorous | 64 (59-69) | 0 | 44 (26-63) | 0 | .04 |
| Accelerometer measurement of moderate-to-vigorous physical activity in 10-minute bouts | |||||
| Overall activity | 47 (44-50) | 0 | 118 (98-139) | 62 (39-84) | <.001 |
aData sources: For New York City, the New York City Physical Activity and Transit survey, conducted in 2 cycles (September–November 2010 and March–November 2011) among adults aged ≥184,10; for the US sample, the 2003-2004 and 2005-2006 cycles of the National Health and Nutrition Examination Survey among adults aged ≥20.11,12 Activity does not include activity performed as part of paid labor.
bP value based on a 2-sample t test, with P < .05 considered significant. Significance testing was not performed for medians.
cConsidered a moderate activity.
In New York City, the median number of recreation minutes per week was 0 and the median number of minutes of transportation activity per week was 139 (95% CI, 118-160). Nationally, the median values were 80 (95% CI, 70-90) minutes per week for recreation activity and 0 minutes per week for transportation activity (Table 2).
Overall, 50.0% (196 of 392) of all minutes in the national sample came from recreation activity, compared with only 17.5% (104 of 594) of all minutes among participants in New York City. In contrast, in New York City, 51.2% (304 of 594) of overall activity minutes were transportation minutes, compared with 11.7% (46 of 392) nationally.
The accelerometer data showed large overall differences between the US sample and New York City in the amount of moderate-to-vigorous physical activity (Table 2). Accelerometer-measured inactivity was 23.4% in New York City and 53.1% nationally. The mean value for weekly accelerometer-measured activity, in 10-minute bouts, was 118 (95% CI, 98-139) minutes in New York City and 47 (95% CI, 44-50) minutes nationally (P < .001). The median values for weekly accelerometer-measured activity were 62 (95% CI, 39-84) minutes in New York City and 0 minutes nationally (Table 2).
In both the United States and New York City, participants with any transportation activity were less likely than participants with no transportation activity to have an inactive accelerometer measurement. A high level (≥150 minutes) of self-reported transportation activity was associated with a lower prevalence of accelerometer-measured inactivity in New York City (PR = 0.26; 95% CI, 0.15-0.44) than nationally (PR = 0.68; 95% CI, 0.58-0.79) (Table 3). Thus, the association between self-reported transportation and accelerometer minutes was stronger in New York City than in the US sample. A high level (≥150 minutes) of self-reported recreation activity was associated with a lower prevalence of accelerometer-measured inactivity nationally (PR = 0.77; 95% CI, 0.74-0.81) but not in New York City (PR = 1.04; 95% CI, 0.73-1.48). A medium level (1-149 minutes per week) of recreation activity was not significantly associated with accelerometer-measured inactivity in either population. In New York City and the US sample, women were significantly more likely than men to be inactive. In the US sample, but not New York City, obese participants were significantly more likely than normal-weight participants to be inactive.
Table 3.
Likelihood of accelerometer-measured inactivitya according to self-reported demographic characteristics and physical activity in a study on patterns of physical inactivity in the United States (2003-2006) and New York City (2010 and 2011)b
| Characteristic | Prevalence Ratioc (95% CI) | |
|---|---|---|
| United States (n = 5776) | New York City (n = 674) | |
| Race/ethnicity | ||
| Non-Hispanic white | 1 [Reference] | 1 [Reference] |
| Non-Hispanic black | 1.03 (0.97-1.08) | 0.94 (0.65-1.35) |
| Hispanic | 0.83 (0.78-0.89) | 0.86 (0.61-1.22) |
| Asian | NA | 1.03 (0.50-2.11) |
| Otherd | 1.17 (1.07-1.29) | 0.97 (0.52-1.80) |
| Education | ||
| <High school diploma | 1.04 (0.99-1.10) | 0.87 (0.55-1.39) |
| High school diploma/some college | 1.12 (1.08-1.16) | 1.24 (0.92-1.67) |
| ≥College | 1 [Reference] | 1 [Reference] |
| Percentage of federal poverty level | ||
| <200 | 1 [Reference] | 1 [Reference] |
| 200-400 | 1.01 (0.97-1.07) | 0.70 (0.40-1.23) |
| >400 | 0.93 (0.88-0.98) | 1.03 (0.69-1.53) |
| Age group | ||
| 18-29 | 0.84 (0.78-0.91) | 0.90 (0.59-1.37) |
| 30-44 | 0.90 (0.85-0.95) | 0.86 (0.58-1.26) |
| 45-64 | 1.01 (0.96-1.06) | 1.16 (0.83-1.63) |
| ≥65 | 1 [Reference] | 1 [Reference] |
| Sex | ||
| Male | 0.92 (0.90-0.95) | 0.61 (0.47-0.80) |
| Female | 1 [Reference] | 1 [Reference] |
| Body mass index, kg/m2 | ||
| Normal weight (18.5-24.9)e | 0.88 (0.84-0.92) | 0.77 (0.59-1.01) |
| Overweight (25.0-29.9) | 0.98 (0.94-1.02) | 1.30 (0.97-1.75) |
| Obese (≥30.0) | 1 [Reference] | 1 [Reference] |
| Self-reported transportation activity, min | ||
| 0 | 1 [Reference] | 1 [Reference] |
| 1-149 | 0.80 (0.73-0.88) | 0.61 (0.37-1.02) |
| ≥150 | 0.68 (0.58-0.79) | 0.26 (0.15-0.44) |
| Self-reported recreation activity, min | ||
| 0 | 1 [Reference] | 1 [Reference] |
| 1-149 | 1.06 (1.02-1.11) | 1.20 (0.88-1.63) |
| ≥150 | 0.77 (0.74-0.81) | 1.04 (0.73-1.48) |
| Self-reported home activity, min | ||
| 0 | 1 [Reference] | 1 [Reference] |
| 1-149 | 0.99 (0.95-1.03) | 1.10 (0.75-1.61) |
| ≥150 | 1.00 (0.95-1.04) | 1.13 (0.87-1.46) |
Abbreviation: NA, not applicable.
aDefined as no 10-minute bouts of moderate-intensity or vigorous-intensity physical activity in 1 week, per the 2008 Physical Activity Guidelines for Americans.2
bData sources: For New York City, the New York City Physical Activity and Transit survey, conducted in 2 cycles (September–November 2010 and March–November 2011) among adults aged ≥184,10; for the US sample, the 2003-2004 and 2005-2006 cycles of the National Health and Nutrition Examination Survey (NHANES) among adults aged ≥20.11,12 Activity does not include activity performed as part of paid labor. All data were self-reported, except data on body mass index from NHANES, which were collected by NHANES examiners.
cGeneralized estimating equations were used to estimate robust standard errors, and the association was modeled using a Poisson regression with a log link function.
dOther includes Native American, multiple race, “don’t know,” and respondents who did not answer the question.
eIncludes small percentage of respondents who were underweight (body mass index <18.5 kg/m2).
Discussion
To our knowledge, the New York City PAT survey is the first study to use accelerometers in a sample population designed to represent a single entire city, and these survey data present a unique opportunity to compare objective measurements of physical activity in 2 settings.
We found that, on average, adults in New York City engaged in 2.5 times more minutes of weekly physical activity than adults nationally (118 vs 47 minutes per week). In New York City, 51.1% of overall activity minutes were transportation minutes, compared with 11.7% nationally, and 75.7% of participants in New York City, compared with 16.9% nationally, reported some transportation activity. Thus, we found striking differences in the activity patterns in these 2 settings.
In addition, 23.4% of New Yorkers were inactive via accelerometer (no 10-minute bouts of moderate or vigorous activity in 1 week), compared with 53.1% nationally. The self-reported data suggest that these findings were driven by differences in transportation activity. This finding supports existing research that urban design promotes walking behavior, and highly walkable neighborhoods are associated with lower body mass index.18
Accelerometer measurements and survey measurements provide different information, and research shows that using both sources provides a more complete picture of physical activity patterns than using 1 source alone.8 We found no difference in inactivity between New York City and the US sample via self-report, but we found large differences via accelerometer. In addition, we found that the accelerometer measurement was strongly correlated with transportation activity and, thus, inactivity. We saw that high levels of recreation activity were moderately associated with accelerometer measurements in the US sample, but not in New York City. The comparison between self-report and accelerometer data suggests that accelerometers are more likely to accurately measure transportation or walking activity than they are to accurately measure recreation activity.
Our results complement the research that demonstrates associations between the physical characteristics of a place with patterns and levels of physical activity.18 New York City has been rated as the most walkable city in the United States—transportation in many parts of the city can be accomplished by walking (www.walkscore.come/cities-and-neighborhoods). Research on the built environment shows that such factors can promote active transportation, whereby persons walk or bicycle as part of their daily routine.18,19 Because walking for transportation is more common in New York City than it is nationally, respondents who did not walk in New York City were more likely than respondents nationally to be inactive. Nationally, 50.0% of all activity minutes, compared with 17.5% in New York City, came from recreation minutes. Self-reported recreation activity was significantly associated with lower accelerometer-measured inactivity nationally but not in New York City. The activity profiles of the 2 settings differed, and these differences appear to be driven by large amounts of walking in New York City, which relates to the factors in the physical environment, such as ease and necessity of walking for transportation.
Limitations
Our study had several limitations. First, NHANES and the PAT survey asked similar questions about physical activity, but the questions were asked in a slightly different way. NHANES asked about past-month activity, and the PAT survey asked about past 7-day activity. Our adjustment to the NHANES data may have resulted in small differences between the 2 sets of data. Second, 803 participants nationally were excluded because of incomplete survey data, and these exclusions could have led to biased results if the excluded participants had less activity than the included participants. However, the demographic characteristics of excluded participants did not differ from those of included participants. Third, we did not have information on the level of walkability of the geographic regions included in NHANES.12 Fourth, physical activity data are difficult to measure, and fifth, survey questions may have been subject to overreporting and social desirability bias.
Our study also had several strengths. Both data sources had similar domain-based questions, and both surveys collected accelerometer data by using the same methods, which were developed by NHANES in 2003-2004. The NHANES and the PAT survey comprise 2 of the largest population-based samples for examining physical activity in the United States. In addition, the accelerometer data for New York City are unique in representing a local geographic area.
Conclusion
Although self-reported survey data indicated no significant differences between New York City and the United States in the proportion of the population who were inactive, the objectively measured accelerometer data indicated large differences. New Yorkers reported 6.6 times more transportation activity than NHANES participants nationally. In addition, 75.7% nationally reported no transportation activity, compared with 16.9% of New Yorkers. The association of transportation activity with physical activity was strongly confirmed by the accelerometer measurement, whereas the association between recreation activity and physical activity was only moderately confirmed by accelerometer measurement.
The survey data showed differences in domain-specific inactivity, whereas the accelerometer measured differences in physical inactivity. Our study illustrates the potential to combine accelerometer data and survey data to better understand patterns of physical activity.
A small percentage of New Yorkers were physically inactive, whereas more than half of persons nationally were inactive, according to objective accelerometer measurements. Overall, our results show that physical inactivity may be strongly related to place, including aspects of its built environment, with New Yorkers getting substantially more activity than persons nationwide. These findings have health consequences because an inactive lifestyle contributes to chronic disease and premature death.
Acknowledgments
The authors thank Brett Wyker and Enver Holder-Hayes for processing the accelerometer data, James McClain and Richard Troiano for consultation on the accelerometer data, and Hannah Gould and Amber Levanon Seligson for comments and editorial suggestions. Donna L. Eisenhower is now an independent contractor, Tiffany Harris is now affilated with Columbia University, and Karen K. Lee is now affiliated with the University of Alberta.
Footnotes
Declaration of Conflicting Interests: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research for this study was provided by the Centers for Disease Control and Prevention’s Communities Putting Prevention to Work (CPPW) grant no. 3U58DP002418-01S1 (PAT survey) and no. 1U58DP002418-01 (PAT biometric component).
References
- 1. Danaei G, Ding EL, Mozaffarian D, et al. The preventable causes of death in the United States: comparative risk assessment of dietary, lifestyle, and metabolic risk factors. PLoS Med. 2009;6(4):e1000058 doi:10.1371/journal.pmed.1000058 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. US Department of Health and Human Services. Physical Activity Guidelines for Americans, First Edition. Washington, DC: US Department of Health and Human Services; 2008; https://health.gov/paguidelines/2008/pdf/paguide.pdf. Accessed March 13, 2018. [Google Scholar]
- 3. Lim S, Wyker B, Bartley K, Eisenhower D. Measurement error of self-reported physical activity levels in New York City: assessment and correction. Am J Epidemiol. 2015;181(9):648–655. doi:10.1093/aje/kwu470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Wyker B, Bartley K, Holder-Hayes E, Immerwahr S, Eisenhower D, Harris TG. Self-Reported and Accelerometer-Measured Physical Activity: A Comparison in New York City. Queens, NY: New York City Department of Health and Mental Hygiene; 2013; https://www1.nyc.gov/assets/doh/downloads/pdf/epi/epiresearch-pa_measures.pdf. Accessed March 13, 2018. [Google Scholar]
- 5. Dyrstad SM, Hansen BH, Holme IM, Anderssen SA. Comparison of self-reported versus accelerometer-measured physical activity. Med Sci Sports Exerc. 2014;46(1):99–106. doi:10.1249/MSS.0b013e3182a0595f [DOI] [PubMed] [Google Scholar]
- 6. Slootmaker SM, Schuit AJ, Chinapaw MJ, Seidell JC, van Mechelen W. Disagreement in physical activity assessed by accelerometer and self-report in subgroups of age, gender, education and weight status. Int J Behav Nutr Phys Activity. 2009;6:17 doi:10.1186/1479-5868-6-17 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Tucker JM, Welk GJ, Beyler NK. Physical activity in U.S. adults: compliance with the physical activity guidelines for Americans. Am J Prev Med. 2011;40(4):454–461. doi:10.1016/j.amepre.2010.12.016 [DOI] [PubMed] [Google Scholar]
- 8. Beyler N, Nusser S, Fuller W, Welk G. Relating self-report and accelerometer physical activity with application to NHANES 2003-2004. Proceedings of the Survey Research Methods Section. Boston: American Statistical Association; 2008. [Google Scholar]
- 9. Frank LD, Andresen MA, Schmid TL. Obesity relationships with community design, physical activity, and time spent in cars. Am J Prev Med. 2004;27(2):87–96. doi:10.1016/j.amepre.2004.04.011 [DOI] [PubMed] [Google Scholar]
- 10. Immerwahr S, Wyker B, Bartley K, Eisenhower D. The Physical Activity and Transit Survey Device Follow-up Study: Methodology Report. Queens, NY: New York City Department of Health and Mental Hygiene; 2012; https://www1.nyc.gov/assets/doh/downloads/pdf/episrv/PAT-device-methodology.pdf. Accessed March 13, 2018. [Google Scholar]
- 11. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008;40(1):181–188. doi:10.1249/mss.0b013e31815a51b3 [DOI] [PubMed] [Google Scholar]
- 12. Centers for Disease Control and Prevention, National Center for Health Statistics. National Health and Nutrition Examination Survey: questionnaires and related documentation [survey operations manuals, consent documents, brochures, and interview and exam manuals online]. https://wwwn.cdc.gov/nchs/nhanes/default.aspx. Accessed February 12, 2019.
- 13. Hayes LJ, Berry G. Comparing the part with the whole: should overlap be ignored in public health measures? J Public Health (Oxf). 2006;28(3):278–282. doi:10.1093/pubmed/fdl038 [DOI] [PubMed] [Google Scholar]
- 14. Toprani A, Hadler JL. Selecting and Applying a Standard Area-Based Socioeconomic Status Measure for Public Health Data: Analysis for New York City. Queens, NY: New York City Department of Health and Mental Hygiene; 2013. https://www1.nyc.gov/assets/doh/downloads/pdf/epi/epiresearch-SES-measure.pdf. Accessed February 5, 2019. [Google Scholar]
- 15. Flegal KM, Carroll MD, Ogden CL, Johnson CL. Prevalence and trends in obesity among US adults, 1999-2000. JAMA. 2002;288(14):1723–1727. doi:10.1001/jama.288.14.1723 [DOI] [PubMed] [Google Scholar]
- 16. SAS [computer program]. Version 9.2. Cary, NC: SAS Institute Inc; 2008. [Google Scholar]
- 17. SUDAAN [computer program]. Version 11.0. Research Triangle Park, NC: Research Triangle Institute; 2005. [Google Scholar]
- 18. Rundle A, Neckerman KM, Freeman L, et al. Neighborhood food environment and walkability predict obesity in New York City. Environ Health Perspect. 2009;117(3):442–447. doi:10.1289/ehp.11590 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Lund H, Wilson RW, Cervero R. A re-evaluation of travel behavior in California TODs. J Archit Plann Res. 2006;23(3):247–263. [Google Scholar]