scispace - formally typeset

Journal ArticleDOI

Public transit, obesity, and medical costs: assessing the magnitudes.

Ryan D. Edwards1
01 Jan 2008-Preventive Medicine (Elsevier)-Vol. 46, Iss: 1, pp 14-21

TL;DR: While no silver bullet, walking associated with public transit can have a substantial impact on obesity, costs, and well-being.

AbstractObjective. This paper assesses the potential benefits of increased walking and reduced obesity associated with taking public transit in terms of dollars of medical costs saved and disability avoided. Methods. I conduct a new analysis of a nationally representative U.S. transportation survey to gauge the net increase in walking associated with public transit usage. I translate minutes spent walking into energy expenditures and reductions in obesity prevalence, estimating the present value of costs and disability that may be avoided. Results. Taking public transit is associated with walking 8.3 more minutes per day on average, or an additional 25.7–39.0 kcal. Hill et al. [Hill, J.O., Wyatt, H.R., Reed, G.W., Peters, J.C., 2003. Obesity and the environment: Where do we go from here? Science 299 (5608), 853–855] estimate that an increase in net expenditure of 100 kcal/day can stop the increase in obesity in 90% of the population. Additional walking associated with public transit could save $5500 per person in present value by reducing obesity related medical costs. Savings in quality-adjusted life years could be even higher. Conclusions. While no silver bullet, walking associated with public transit can have a substantial impact on obesity, costs, and well-being. Further research is warranted on the net impact of transit usage on all behaviors, including caloric intake and other types of exercise, and on whether policies can promote transit usage at acceptable cost.

Topics: Population (51%), Present value of costs (51%), Poison control (50%)

Summary (2 min read)

Introduction

  • A topic of much recent interest is the degree to which public transportation may increase exercise through walking.
  • Because residents typically select their communities, much remains unclear about the causal influence of environment on activity (Handy and Mokhtarian, 2005; Ogilvie et al., 2006).
  • The amount of additional physical activity associated with public transportation appears potentially significant.

Methods

  • Estimating additional walking associated with public transit Part of the 2001 NHTS included a daily travel diary in which household respondents were asked to self-report all trips, their purposes, starting and ending times, and the means of transportation during an assigned travel day.
  • I can only compare my estimates to those of Wener and Evans (2007), who collect objective measures of extra walking using pedometers.
  • I interpret β as the additional walking associated with transit use.
  • Of the 105,942 individuals in the adult subsample, 39,782 filled out the entire survey and have a sample weight, and 28,771 records contain all covariates.

Forecasting obesity prevalence

  • These statistics are reported by Flegal et al. (2002) and Ogden et al. (2006), who examine data from the 1960–1962 National Health Examination Survey (NHES) and subsequent NHANES waves.
  • “College degree” includes the bachelor but not the associate.
  • It is convenient to combine the estimates of Sturm, who examines additional costs under age 65, with those of Lakdawalla et al., who explore costs over 70.
  • In each year of the projection, I apply the forecast obesity prevalence rate to the additional per capita spending associated with obesity.
  • In addition to increased medical costs, obesity also threatens the quality of health and well-being, most notably later in life, and I measure these costs as well.

Results

  • Additional walking through transit Table 1 describes the characteristics of the weighted NHTS sample of adults, where the observations are person-days.
  • An almost equally large share, 1.9%, reported walking as their sole means of transit on their travel day, but these were primarily recreational walkers rather than commuters.
  • Column 1 reports ordinary least squares estimates, while columns 2–6 re marginal effects of each Tobit are the partial derivatives of the expected value o significance, with ⁎⁎⁎ at the 1% level, ⁎⁎ at 5%, and ⁎ at the 10% level.
  • No el of daily walking time as shown in Eq. (1) in the text, with standard errors in port Tobit estimates under various alternative specifications, where the reported f the observed walking time variable.

Reductions in obesity

  • The distribution of excess energy stored reported by Hill et al. (2003) reveals that these levels of additional expenditure could eliminate weight gain in approximately 43%, 50%, or 60% of the population.
  • The assumed QALY weight of a life-year spent disabled is 0.8.

Obesity prevalence scenarios

  • An OLS regression line through the historical obesity prevalence data in Fig. 1 has a slope equal to roughly 0.5% per year, significant at the 1% level.
  • These three scenarios are depicted in the right-hand side of Fig. 1 beneath the baseline projection.
  • Present value per person depending on the intensity, and about 80% of the savings is public money.

Discussion

  • The objective of this paper was to explore the potential benefits of shifting an average U.S. citizen from driving to using public transit.
  • Results hinge on an unknown that is difficult to estimate with great confidence: the additional physical activity associated with public transit.
  • It seems unlikely that anyone rationally chooses to be obese (Cutler et al., 2003), so anti-obesity policies in general are unlikely to trigger rational behavior that directly counteracts the intent, as might be an issue with anti-smoking policies if addiction is rational.
  • It could be that public transit indeed causes more physical activity, but individuals offset the potential health effects by eating more.
  • I have chosen a simple linear extrapolation of the historical trend, which has been a 0.5% linear rate of increase since the inception of the NHES/NHANES survey in 1960.

Conclusion

  • Use of public transit is associated with more walking, by about 8.3 extra minutes per day.
  • This is not enough walking to halt the spread of obesity, but it could substantially reduce it.
  • The present value of medical expenditure savings per person could be $5500, while the value of reduced disability could be even greater.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

Public transit, obesity, and medical costs: Assessing the magnitudes
Ryan D. Edwards
Queens CollegeCity University of New York, 300-S Powdermaker Hall, 65-30 Kissena Blvd., Flushing, NY 11367, USA
Available online 18 October 2007
Abstract
Objective. This paper assesses the potential benefits of increased walking and reduced obesity associated with taking public transit in terms of
dollars of medical costs saved and disability avoided.
Methods. I conduct a new analysis of a nationally representative U.S. transportation survey to gauge the net increase in walking associated with
public transit usage. I translate minutes spent walking into energy expenditures and reductions in obesity prevalence, estimating the present value
of costs and disability that may be avoided.
Results. Taking public transit is associated with walking 8.3 more minutes per day on average, or an additional 25.739.0 kcal. Hill et al. [Hill,
J.O., Wyatt, H.R., Reed, G.W., Peters, J.C., 2003. Obesity and the environment: Where do we go from here? Science 299 (5608), 853855]
estimate that an increase in net expenditure of 100 kcal/day can stop the increase in obesity in 90% of the population. Additional walking
associated with public transit could save $5500 per person in present value by reducing obesity-related medical costs. Savings in quality-adjusted
life years could be even higher.
Conclusions. While no silver bullet, walking associated with public transit can have a substantial impact on obesity, costs, and well-being.
Further research is warranted on the net impact of transit usage on all behaviors, including caloric intake and other types of exercise, and on
whether policies can promote transit usage at acceptable cost.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Obesity; Walking; Exercise; Transportation; Health expenditures
Introduction
A topic of much recent interest is the degree to which public
transportation may increase exercise through walking. Other
things equal, an increase in exercise could then improve health
outcomes by lowering obesity, which many view as a loom ing
but potentially manageable threat to public health (Hill et al.,
2003; Olshansky et al., 2005; Preston, 2005). The New
Urbanism movement of the 1990s, which continues today,
calls for development of denser, grid-based neighborhoods in
order to increase walking, bicycling, and use of transit (Cervero
and Radisch, 1996). More recently, the entire January 2007 issue
of Environment and Behavior was devoted to examining how
the built environment relates to diet and exercise, and thus, to
obesity. A joint study by the Transportation Research Board and
the Institute of Medicine (2005) surveyed the state of knowledge
regarding the built environment and physical activity. Although
the study showed that the built environment, including access to
public transit, can help or hinder the choice to engage in physical
activity, it emphasized how the lack of good data and inadequate
study design have significantly hampered inference.
Because residents typically select their communities, much
remains unclear about the causal influence of environment on
activity (Handy and Mokhtarian, 2005; Ogilvie et al., 2006).
Longitudinal panel studies of relocated families and their
behavior and outcomes, such as the nascent RESIDE project in
Perth, Australia, are designed to untangle this issue. Without a
clear sense of how urban form and the availability of mass
transit can actually produce more exercise, researchers are
limited in advocating specific policy interventions. Still, it is
worth assessing the potential magnitudes of the influences of
public transit on health, in order to gauge the plausible scope for
policy and motivate further research.
The amount of additional physical activity associated with
public transportation appears potentially significant. Besser and
A
vailable online at www.sciencedirect.com
Preventive Medicine 46 (2008) 14 21
www.elsevier.com/locate/ypmed
All errors and opinions are those of the author alone and do not reflect the
views of Queens College or the City University of New York.
Fax: +1 718 997 5466.
E-mail address: redwards@qc.cuny.edu.
0091-7435/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.ypmed.2007.10.004

Dannenberg (2005) report that half of the roughly 3% of adults
in the 20 01 U.S. National Household Travel Survey (NHTS)
who walked to and from public transit spent 19 min or more in
total walking time, and almost a third exceeded 30 min. Wener
and Evans (2007) find that the average New York City train
commuter walked about 9500 steps per day, roughly 2000 or
30% more steps than the average car commuter. Several papers
associate form of transit with obesity directly. Frank et al.
(2004) report that obesi ty around Atlanta, as measured by body
mass index (BMI), is associated positively with time spent in
cars and negatively with mixed land-use and with walking.
Gordon-Larsen et al. (2005) reveal that non-overweight young
adults in the Add Health survey were more likely to engage in
active transportation like walking or bicycling, possibly in
addition to taking public transit. Rundle et al. (2007) find BMI
to be inversely associated with the density of bus stops, subway
stops, and population around New York City.
However, is the additional walking associated with mass
transit large enough to reduce obesity and associ ated healt h care
costs? If yes, by how much? In this paper, I address this question
by modeling daily time spent walking based on characteristics
including transit use, and I then translate those differences into
extra net energy expenditure and reductions in obesity.
Methods
Estimating additional walking associated with public transit
The quantity of interest is the additional amount of physical activity
associated with taking public transit as opposed to driving. Wener and Evans
(2007) measure this directly by asking a sample of car and train commuters
around New York City to wear pedometers, and then comparing total steps for
each group. No comparable study exists at the national level, but the 2001
National Household Travel Survey (U.S. Dept. of Transportation and Federal
Highway Administration, 2001) contains similar data that are nationally
representative.
Part of the 2001 NHTS included a daily travel diary in which household
respondents were asked to self-report all trips, their purposes, starting and
ending times, and the means of transportation during an assigned travel day.
Individuals or their proxies were asked to fill out the travel diaries on their travel
day, and then to relay that information during follow-up telephone interviews.
The goal was to obtain travel information for each and every member of the
household, but when members were unavailable, their trips went unmeasured.
Walking trips undertaken for any reason, whether part of daily commutes, chores
or errands, recreation or exercise, were part of the universe of daily trips
recorded, as were bicycle trips. No other forms of physical activity were directly
measured, but trips were also classified by purpose and could include trips to the
gym, to exercise, or to play sports. All legs of all trips on the travel day were
categorized by means of transportation. I define a public transit user as anyone
who reports using it for any reason during the travel day.
Walking time in the NHTS is a limited measure of total physical activity in at
least three respects. First, and most obviously, physical activity other than
bicycling and walking goes unmeasured. Second, because the survey covers
only one travel day per individual, it cannot measure behavior on other days that
may be related. If a transit user walks more on Monday through Friday, she may
choose to walk less on Saturday and Sunday because she is worn out. Third, self-
reported walking time may not be a good objective measure of walking. As
reviewed by Tudor-Locke and Myers (2001), the literature examining objective
and subjective measures of walking typically reveals that individuals under-
report total walking. To address the first two limitations, I explore the available
information on other exercise in the NHTS, and I also examine how excluding
weekend trips affects the results. The third limitation will bias my estimate of
additional walking if under-reporting is correlated with public transit use. This
seems unlikely but is untestable. I can only compare my estimates to those of
Wener and Evans (2007), who collect objective measures of extra walking using
pedometers.
I constructed total reported daily walking in minutes for each individual in
the NHTS, and then I estimated the following model for individual i:
walktime
i
¼ a
i
þ b pubtrans
i
þ B
Y
d X
Y
i
þ e
i
; ð1Þ
where α
i
is a fixed effect based on geography, pubtrans is a dummy variable
indicating public transit use,
Y
X
i
is a vector of socioeconomic and demographic
controls, and ε
i
is a white-noise error. I interpret β as the additional walking
associated with transit use.
Estimation is complicated by several concerns. Because public transit use is
a choice, endogeneity may render estimates of β biased and inconsistent. With
neither a controlled experiment nor good instrumental variables, little can be
done other than to acknowledge this problem and work toward improving future
study design. A more tractable problem is the fact that walktime exhibits severe
response pooling, with 85% of NHTS respondents reporting no walking at all.
(Table 1 reports characteristics of the weighted sample.) I therefore estimate
Eq. (1) using the Tobit, a standard model for dealing with truncated data.
The version of the dataset I downloaded from the Inter-university Consortium
for Political and Social Research contains a total of 140,915 individuals with at
least partial travel diaries. Roughly 50,000 completed the entire survey and thus
also have a sample weight. Sample weights were calculated based on the char-
acteristics of all sampled households and Census data, where controls included
geographic, socioeconomic, and demographic variables. For comparability to
Besser and Dannenberg (2005), I define covariates similarly and restrict my
analysis to respondents 18 years old and over. Of the 105,942 individuals in the
adult subsample, 39,782 filled out the entire survey and have a sample weight,
and 28,771 records contain all covariates.
Estimating changes in obesity based on walking
I convert my estimate of β, minutes of additional walking, into reductions in
obesity prevalence in three steps. First, I translate minutes of walking into
kilocalories (kcal) of energy expended using the basal metabolic rates (BMR)
reported by Morabia and Costanza (2004): slow walking expends 3.1 kcal/min,
moderate walking 3.9 kcal/min, and fast walking 4.7 kcal/min. Then I convert
additional kilocalories expended into reductions in stored energy using the
efficiency factor of 50% cited by Hill et al. (2003). Finally, I match reductions in
stored energy with percentiles of the empirical distribution of excess energy stored
reported by Hill et al., who examine recent waves of the National Health and
Nutrition Examination Survey (NHANES). The result is a percentage that
represents the share of Americans for whom weight gain would be eliminated by
the given amount of extra walking. This can also be interpreted as the percentage
reduction in the percentage increase in obesity prevalence for the average American.
Forecasting obesity prevalence
My baseline forecast of obesity prevalence is a simple extrapolation of past
trends in U.S. obesity rates since 1960. These statistics are reported by Flegal
et al. (2002) and Ogden et al. (2006), who examine data from the 19601962
National Health Examination Survey (NHES) and subsequent NHANES waves.
They are depicted in the left side of Fig. 1, which plots historical and projected
adult obesity prevalence. I produce alternative forecasts of obesity prevalence by
multiplying the baseline annual increase in obesity by the percentage for whom
walking eliminates weight gain, which I estimate as described above.
Estimating medical cost savings and other benefits
Obesity is costly along several dimensions. Chronic illnesses such as
diabetes and musculoskeletal disorder associated with obesity (Must et al., 1999)
cost additional dollars of medical expenditure. These and other diseases also
reduce the quality of life (Cutler and Richardson, 1997), and they can also
shorten life (Olshansky et al., 2005).
I forecast total medical cost savings per person by first projecting obesity
prevalence for each remaining year of the average U.S. citizen's life using the
15R.D. Edwards / Preventive Medicine 46 (2008) 1421

extrapolation technique and the intervention estimates described above. The
Census Bureau estimates the median age in the United States to be about 36, and
a person of that age can expect to live 46 more years (Bell and Miller, 2005).
Next, I obtain estimates of additional medical costs associated with obesity,
which are assessed by a number of researchers, including Allison et al. (1999),
Sturm (2002), Finkelstein et al. (2003), and Lakdawalla et al. (2005).Itis
convenient to combine the estimates of Sturm, who examines additional costs
under age 65, with those of Lakdawalla et al., who explore costs over 70. I
translate each of their estimates into 2007 dollars by assuming an annual rate of
real growth in per capita medical spending of 3%, per Lakdawalla et al., plus an
additional 2.7% per year in general price inflation, for total annual growth of
5.7% in the nominal amounts. Sturm's estimate becomes $650 per year for those
under 65, whereas that of Lakdawalla et al. becomes $46,000 for ages 70 and
over. I assume that additional spending between age 65 and 70 is $650.
In my next steps, I follow the lead of Lakdawalla et al., who explain their
assumptions about future cost growth and discounting in greater detail. First, I
project additional real spending in each future year by assuming that real per
capita costs grow 3% per year. In each year of the projection, I apply the forecast
obesity prevalence rate to the additional per capita spending associated with
obesity. Then I compute the present discounted value of future amounts using a
rate of time discounting also equal to 3% per year. Finally, I allocate spending
between public and private funding sources by assuming that all spending at ages
under 65 is privately funded, whereas 92.3% of spending at 65 and over is public.
In addition to increased medical costs, obesity also threatens the quality of
health and well-being, most notably later in life, and I measure these costs as
well. Lakdawalla et al. (2005) and Reynolds et al. (2005) both argue that obese
elderly can expect to live roughly the same number of remaining years as the
non-obese, but that their quality of life will be eroded through obesity-related
disability. Lakdawalla et al. expect obese 70-year-olds to enjoy 4 years of
disability-free life, or 2.8 fewer than the non-obese. Reynolds et al. estimate the
obese will live more like 8 years free of disability, but still about 2 years less than
the non-obese. I assume 2.5 fewer years of disability-free life for obese elderly,
and I ignore impacts earlier in life, about which less is known. The value of a
life-year spent in disability, or a quality-adjusted life-year (QALY) weight, is
about 80% of a life-year in perfect health (Cutler and Richardson, 1997), and I
use that factor along with estimates of the value of a life-year (Viscusi and Aldy,
2003) to gauge the welfare costs of obesity-related disability.
Results
Additional walking through transit
Table 1 describes the characteristics of the wei ghted NHTS
sample of adults, where the observations are person-days.
Typical respondents are nearly 50 years old, roughly split
between men and women, predominantly white, and typically
hold only high school degrees. Average household income in the
dataset is roughly twice per capita income because the data file
frequently includes both adults in a typical household. Eighty
percent of respondents own their own home. The average
population densi ty among these respondents, at just under 4000
people per square mile, is relatively high and comparable to
levels around San Diego, Sacramento, and Portland, OR. But
Fig. 1. Adult obesity rates in the United States, historical and projected. Data are
the rate of obesity, defined as body mass index (BMI) 30 or greater, for the adult
population measured in the National Health Examination Survey (NHES) and
National Health and Nutrition Examination Survey (NHANES). Obesity rates
are calculated by Flegal et al. (2002) and Ogden et al. (2006), appear as circles in
the figure, and are plotted at the midpoint of the NHANES examination period.
The dashed line represents a simple forward extrapolation from the most recent
data point of the average annual increase in the rate, 0.5%, which is estimated by
ordinary least squares regression of all historical data on time.
Table 1
Sample characteristics of adults in the 2001 National Household Travel Survey
Variable Average
Age 48.2
Male 47.7%
White 83.4%
African American 5.7%
Other race 10.8%
Hispanic 6.3%
Less than high school degree 8.0%
High school graduate 55.9%
College graduate 27.2%
Graduate or professional degree 8.8%
Household income $65,473
Homeownership 80.2%
Census division of residence:
New England 5.5%
Middle Atlantic 13.0%
East North Central 18.4%
West North Central 9.1%
South Atlantic 18.1%
East South Central 5.8%
West South Central 8.7%
Mountain 7.5%
Pacific 14.0%
Population per square mile, block group 3933
Lives in an MSA with rail 12.3%
Number of household vehicles 2.2
Total daily walking time (min) 5.1
Total daily biking time (min) 0.4
Any walking 14.6%
Any biking 1.0%
Used public transit 2.7%
Used bus 2.0%
Used rail 0.9%
Only reported walking 1.9%
Only reported walking, to work 0.4%
Reported trips to gym/exercise/play sports 12.3%
Sample size 28,771
The sample is adults age 18 years and over, and sample weights are used to
adjust for representativeness and incomplete survey response. A user of public
transit is defined as anyone who reported using public transit at any time during
his or her travel day. The Other racial group includes those not listed and all
those self-identifying as mixed-race. Hispanic derives from a separate
ethnicity question. College degree includes the bachelor but not the associate.
Advanced degree includes any graduate or professional degree. Household
income is measured over the previous 12 months.
16 R.D. Edwards / Preventive Medicine 46 (2008) 1421

average rail access is low at only 12.3%, which is also roughly
the percentage in the Pacific division, and the average number of
household vehicles is 2.2.
Total walking time during the travel day averages just
5.1 min, with only 14.6% reporting any walking at all. Time
spent bicycling is even less common. Only about 2.7% reported
any use of public transit, with most using buses as opposed to
trains, although some used both. An almost equally large share,
1.9%, reported walking as their sole means of transit on their
travel day, but these were primarily recreational walkers rather
than commuters. Although they reported little walking, which
as discussed above may be an underestimate of actual walking,
12.3% of respondents reported a trip whose purpose was to
facilitate engaging in physical exercise, like driving to the gym
or to a softball game.
Table 2 presents the results of estimating Eq. (1), the model of
daily walking time, using several alternative specifications. The
first column presents simple ordinary least squares (OLS)
estimates, which do not correct for the response heaping at zero
minutes of walking time. The other five columns employ the
Tobit estimation technique, which corrects for the 85% change
of reporting zero walking. In each set of Tobit results, I report the
marginal effects on the observed (truncated) variable rather than
the underlying behavioral parameters, which are roughly equal
to the reported coefficients divided by the probability of positive
walking time, 0.15. The latter would be only appropriate if zero
walking times actually repres ented truncated negatives, which
seems implausible.
Results are fairly robust across specifications, with public
transit use significant at the 1% level in each and associated with
between 8 and 10 additional minutes of walking per day. The
Tobit specification produces a slightly smaller point estimate
than OLS, and adding Census region fixed effects in column 3
lowers it a little more. Respondents in New England, the mid-
Atlantic states, Mountain, and Pacific regions reported the most
walking, whereas residents of states in East and West South
Central regions, which are located along the Gulf, walked the
least. Specifying state fixed effects yielded similar resul ts.
When I included an indicator variable for train use, shown in
the fourth column, a significant difference between train and
bus commuters emerges. Users of public transit who do not use
trains walk only an additional 6 min compared with non-users,
whereas those who use trains walk another 4.5 min more, for a
total of 10.5 extra minutes per day.
In the fifth column, I remove the train dummy and re-estimate
the fixed effects model after excluding observations in which no
other mode of transportation other than walking was reported.
Public transit obviously cannot increase walking among those
who report no other form than walking, so it is useful to estimate
β without them. As shown, dropping exclusive walkers increases
the point estimate somewhat to 10.4.
In the sixth and final column, I dropped weekend obser-
vations and found little change relative to column 3. My
preferred estimate is the 8.3 additional min utes that appears in
the third column.
Other physical activity and transit
The NHTS provides only two other measures of physical
activity, bicycling time and trips taken to the gym, to exercise,
Table 2
Marginal effects of characteristics on total daily walking time in the 2001 National Household Transport Survey
123456
OLS Tobit Tobit Tobit Tobit Tobit
Use public transit 9.50 (1.21)
⁎⁎⁎
8.74 (0.89)
⁎⁎⁎
8.26 (0.87)
⁎⁎⁎
5.86 (0.85)
⁎⁎⁎
10.36 (0.94)
⁎⁎⁎
8.23 (0.92)
⁎⁎⁎
Use train –––4.50 (1.31)
⁎⁎⁎
––
Age 0.02 (0.01)
⁎⁎⁎
0.02 (0.01)
⁎⁎⁎
0.02 (0.01)
⁎⁎⁎
0.02 (0.01)
⁎⁎⁎
0.02 (0.00)
⁎⁎⁎
0.03 (0.01)
⁎⁎⁎
Male 0.47 (0.21)
⁎⁎
0.52 (0.17)
⁎⁎⁎
0.51 (0.17)
⁎⁎⁎
0.53 (0.17)
⁎⁎⁎
0.44 (0.15)
⁎⁎⁎
0.54 (0.19)
⁎⁎⁎
African American 0.85 (0.49) 0.83 (0.31) 0.57 (0.32)
0.58 (0.32)
0.92 (0.29)
⁎⁎⁎
0.28 (0.37)
Other race 0.36 (0.53) 0.53 (0.38) 0.52 (0.38) 0.52 (0.38) 0.45 (0.36) 0.42 (0.43)
Hispanic 0.43 (0.69) 0.43 (0.54) 0.45 (0.55) 0.46 (0.55) 0.20 (0.50) 0.71 (0.63)
Less than high school degree 1.12 (0.36)
⁎⁎⁎
0.81 (0.29)
⁎⁎⁎
0.75 (0.29)
⁎⁎⁎
0.73 (0.29)
⁎⁎
0.97 (0.27)
⁎⁎⁎
0.72 (0.33)
⁎⁎
College graduate 1.00 (0.29)
⁎⁎⁎
1.09 (0.24)
⁎⁎⁎
1.11 (0.24)
⁎⁎⁎
1.11 (0.24)
⁎⁎⁎
1.04 (0.22)
⁎⁎⁎
0.80 (0.26)
⁎⁎⁎
Graduate or professional degree 1.59 (0.48)
⁎⁎⁎
1.40 (0.37)
⁎⁎⁎
1.38 (0.37)
⁎⁎⁎
1.36 (0.37)
⁎⁎⁎
1.20 (0.34)
⁎⁎⁎
1.47 (0.42)
⁎⁎⁎
Log of Household Income 0.17 (0.15) 0.09 (0.11) 0.14 (0.11) 0.16 (0.11) 0.09 (0.10) 0.07 (0.13)
Log population per square mile,
block group
0.42 (0.06)
⁎⁎⁎
0.50 (0.05)
⁎⁎⁎
0.42 (0.05)
⁎⁎⁎
0.42 (0.05)
⁎⁎⁎
0.34 (0.05)
⁎⁎⁎
0.43 (0.06)
⁎⁎⁎
Own home 0.62 (0.30)
⁎⁎
0.81 (0.24)
⁎⁎⁎
0.72 (0.24)
⁎⁎⁎
0.72 (0.24)
⁎⁎⁎
0.15 (0.22) 0.72 (0.27)
⁎⁎⁎
Number of household vehicles 0.78 (0.10)
⁎⁎⁎
0.86 (0.10)
⁎⁎⁎
0.86 (0.10)
⁎⁎⁎
0.86 (0.10)
⁎⁎⁎
0.57 (0.09)
⁎⁎⁎
0.83 (0.11)
⁎⁎⁎
N 28,771 28,771 28,771 28,771 28,217 21,326
Census region fixed effects? No No Yes Yes Yes Yes
Include those who only
reported walking?
Yes Yes Yes Yes No Yes
Include weekends? Yes Yes Yes Yes Yes No
See notes to Table 1. Each column reports marginal effects from estimation of a model of daily walking time as shown in Eq. (1) in the text, with standard errors in
parentheses. Column 1 reports ordinary least squares estimates, while columns 26 report Tobit estimates under various alternative specifications, where the reported
marginal effects of each Tobit are the partial derivatives of the expected value of the observed (truncated) walking time variable. Asterisks denote statistical
significance, with
⁎⁎⁎
at the 1% level,
⁎⁎
at 5%, and
at the 10% level.
17R.D. Edwards / Preventive Medicine 46 (2008) 1421

or to play sports. Bicycling is extremely rare in the NHTS, as
shown in Table 1, and its association with public transit is small
and of borderline significance. In a Tobit regression of bicycling
time on the same covariates and region fixed effects listed in
Table 2, I find transit use to be associated with just 0.2 fewer
minutes of bicycling per day, significant at the 10% level.
Transit use is negatively associated with taking trips to go to
the gym, to exercise, or to play sports. I ran a probit model on
the probability of reporting any such trip on the covariates in
Table 2, and I found transit use was associated with a reduction
in the probability of trips for exercise of 2.7%, significant at the
5% level (estimates not shown). Since only 12.3% of the
sample reported any such trips (Table 1), this seems relatively
large.
Reductions in obesity
Walking expends 3.1, 3.9, or 4.7 kcal/min depending on
whether the walking is slow, moderate, or brisk (Morabia and
Costanza, 2004), so the 8.3 min of additional walking
associated with transit use could represent 25.7, 32.4, or 39.0
additional kcal expended each day. At a 50% efficiency rate,
those numbers translate into 12.9, 16.2, and 19.5 fewer kcal
stored per day. The distribution of excess energy stored reported
by Hill et al. (2003) reveals that these levels of additional
expenditure could eliminate weight gain in approximately 43%,
50%, or 60% of the population.
Obesity prevalence scenarios
An OLS regression line through the historical obesity
prevalence data in Fig. 1 has a slope equal to roughly 0.5%
per ye ar, significant at the 1% level. The baseline forecast, with
future obesity prevalence increasing 0.5% each year, is shown at
top-right in the figure. With 8.3 additional minutes of slow
walking per day, 43% of the 0.5% annual increase is offset,
leaving 0.29% per year. Similarly, moderate walking reduces
growth to 0.25%, whereas brisk walking leaves 0.2%. These
three scenarios are depicted in the right-hand side of Fig. 1
beneath the baseline projection.
Medical cost and QALY savings
Table 3 compares the present value of future medical costs in
the baseline scenario to five alternatives: the three I have
presented thus far; one based on the Wener and Evans (2007)
estimate of 2000 extra steps; and for comparison, a scenario
with 100 additional kcal/da y expended, per the suggestion of
Hill et al. (2003). At a rate of 0.04 kcal/step (Hill et al., 2003;
Tudor-Locke and Bassett, 2004), 2000 steps translates into
80 kcal, or 20.5 min of moderate walking.
At baseline, I project that obesity will generate an extra
$34,200 in health costs per person, $24,400 of which will be
borne by Medicare and other public sources. An additional
8.3 min of daily walking could save $4800, $5500, or $6600 in
Table 3
The effect of additional walking on the present value of additional health care spending associated with obesity, 2007 dollars
Additional energy
expenditure per
day (kcal)
Annual growth
in obesity
prevalence (%)
Present value of
additional health
spending per person
Present value of
savings relative
to baseline
Present value of
additional public health
spending per person
Present value of
public savings
relative to baseline
Baseline 0.0 0.5 $34,200 $24,400
8.3 additional minutes of walking:
slowly (3.1 kcal/min) 25.7 0.29 $29,400 $4800 $20,600 $3800
moderately (3.9 kcal/min) 32.4 0.25 $28,700 $5500 $19,900 $4500
briskly (4.7 kcal/min) 39.0 0.20 $27,600 $6600 $19,000 $5400
2000 additional steps of walking,
or 20.5 min at 3.9 kcal/min
80.0 0.07 $24,700 $9500 $16,700 $7700
100 kcal/day 100.0 0.05 $24,300 $9900 $16,400 $8000
Savings are presented per person, not per obese person. Forecasts assume a 36-year-old in 2007 who lives another 46 years. The obesity prevalence rate is assumed to
be 34% in 2007. From age 36 to 69, obese individuals incur $650 extra each year in medical expenditures, inferred from Sturm (2002) according to the text. Individuals
70 and over incur $46,000 extra in present value, from Lakdawalla et al. (2005) as described in the text, 92.3% of which is paid by Medicare. That share of spending
between age 65 and 70 is also categorized as public. All real medical spending grows at 3% per year. The real discount rate is also 3% per year.
Table 4
The effect of additional walking due to public transit on health status and its value by rate of future increase in obesity
Average obesity prevalence
over age 70 (%)
Expected QALYs lost
due to obesity
Expected present value of
QALYs lost due to obesity
Savings relative to
baseline
Baseline 54 0.27 $54,000
8.3 additional minutes of walking:
slowly (3.1 kcal/min) 45 0.23 $45,400 $8600
moderately (3.9 kcal/min) 44 0.22 $44,000 $10,000
briskly (4.7 kcal/min) 42 0.21 $42,000 $12,000
2000 additional steps of walking,
or 20.5 min at 3.9 kcal/min
37 0.18 $36,800 $17,200
100 kcal/day 36 0.18 $36,000 $18,000
See notes to Table 3. The assumed QALY weight of a life-year spent disabled is 0.8. The present value of a life-year in perfect health is assumed to be $200,000.
18 R.D. Edwards / Preventive Medicine 46 (2008) 1421

Citations
More filters

Journal ArticleDOI
TL;DR: A meta-analysis of the built environment-travel literature existing at the end of 2009 is conducted in order to draw generalizable conclusions for practice, and finds that vehicle miles traveled is most strongly related to measures of accessibility to destinations and secondarily to street network design variables.
Abstract: Problem: Localities and states are turning to land planning and urban design for help in reducing automobile use and related social and environmental costs. The effects of such strategies on travel demand have not been generalized in recent years from the multitude of available studies. Purpose: We conducted a meta-analysis of the built environment-travel literature existing at the end of 2009 in order to draw generalizable conclusions for practice. We aimed to quantify effect sizes, update earlier work, include additional outcome measures, and address the methodological issue of self-selection. Methods: We computed elasticities for individual studies and pooled them to produce weighted averages. Results and conclusions: Travel variables are generally inelastic with respect to change in measures of the built environment. Of the environmental variables considered here, none has a weighted average travel elasticity of absolute magnitude greater than 0.39, and most are much less. Still, the combined effect o...

2,893 citations


Journal ArticleDOI
Abstract: Background—Poor lifestyle behaviors, including suboptimal diet, physical inactivity, and tobacco use, are leading causes of preventable diseases globally. Although even modest population shifts in risk substantially alter health outcomes, the optimal population-level approaches to improve lifestyle are not well established. Methods and Results—For this American Heart Association scientific statement, the writing group systematically reviewed and graded the current scientific evidence for effective population approaches to improve dietary habits, increase physical activity, and reduce tobacco use. Strategies were considered in 6 broad domains: (1) Media and educational campaigns; (2) labeling and consumer information; (3) taxation, subsidies, and other economic incentives; (4) school and workplace approaches; (5) local environmental changes; and (6) direct restrictions and mandates. The writing group also reviewed the potential contributions of healthcare systems and surveillance systems to behavior change...

421 citations


Journal ArticleDOI
TL;DR: A systematic review of how much time is spent in physical activity among adults using public transport and the potential effect on the population level of physical activity if inactive adults in NSW, Australia, increased their walking through increased use of public transport is explored.
Abstract: Active travel, particularly walking and cycling, has been recommended because of the health benefits associated with increased physical activity. Use of public transport generally involves some walking to bus stops or train stations. This paper is a systematic review of how much time is spent in physical activity among adults using public transport. It also explores the potential effect on the population level of physical activity if inactive adults in NSW, Australia, increased their walking through increased use of public transport. Of 1,733 articles, 27 met the search criteria, and nine reported on absolute measures of physical activity associated with public transport. A further 18 papers reported on factors associated with physical activity as part of public transport use. A range of 8-33 additional minutes of walking was identified from this systematic search as being attributable to public transport use. Using "bootstrapping" statistical modelling, if 20% of all inactive adults increased their walking by only 16 minutes a day for five days a week, we predict there would be a substantial 6.97% increase in the proportion of the adult population considered "sufficiently active". More minutes walked per day, or a greater uptake of public transport by inactive adults would likely lead to significantly greater increases in the adult population considered sufficiently active.

232 citations


Cites background from "Public transit, obesity, and medica..."

  • ...(2008) [16] USA Cross-sectional n = 28,771...

    [...]

  • ...3 more minutes walking per day than did people who relied on cars [16]....

    [...]


Journal ArticleDOI
TL;DR: Generalized estimating equations, conducted on 5000 randomly chosen licensed drivers aged 25-64 in Salt Lake County, Utah, relate lower BMIs to older neighborhoods, components of a 6-category land use entropy score, and nearby light rail stops to healthy weight.
Abstract: Few studies compare alternative measures of land use diversity or mix in relationship to body mass index. We compare four types of diversity measures: entropy scores (measures of equal distributions of walkable land use categories), distances to walkable destinations (parks and transit stops), proxy measures of mixed use (walk to work measures and neighborhood housing ages), and land use categories used in entropy scores. Generalized estimating equations, conducted on 5000 randomly chosen licensed drivers aged 25-64 in Salt Lake County, Utah, relate lower BMIs to older neighborhoods, components of a 6-category land use entropy score, and nearby light rail stops. Thus the presence of walkable land uses, rather than their equal mixture, relates to healthy weight.

227 citations


Cites background from "Public transit, obesity, and medica..."

  • ...A recent cost benefit analysis estimated that rail stop users can accrue 8.3 min of walking per day walking to transit, which over time may prevent weight gain and prevent estimated expenditures of $5500 per person in additional health costs (Edwards, 2008)....

    [...]


Journal ArticleDOI
TL;DR: The positive psychological wellbeing effects identified in this study should be considered in cost–benefit assessments of interventions seeking to promote active travel.
Abstract: Data from the BHPS were supplied by the Economic and Social Research Council (ESRC) Data Archive. Neither the original collectors of the data nor the Archive bear any responsibility for the analysis or interpretations presented here. The work was undertaken by the Centre for Diet and Activity Research (CEDAR), a UKCRC Public Health Research Centre of Excellence. Funding from the British Heart Foundation, Cancer Research UK, the ESRC, the Medical Research Council, the National Institute for Health Research, and the Wellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.

214 citations


Cites background from "Public transit, obesity, and medica..."

  • ...A partial explanation for our finding could be that public transport journeys typically feature physical activity when accessing bus stops or railway stations (Besser and Dannenberg, 2005; Edwards, 2008; Laverty et al., 2013; MacDonald et al., 2010;Morabia et al., 2010; Rissel et al., 2012)....

    [...]


References
More filters

Journal ArticleDOI
05 Apr 2006-JAMA
TL;DR: These estimates suggest that the increases in body weight are continuing in men and in children and adolescents while they may be leveling off in women; among women, no overall increases in the prevalence of obesity were observed.
Abstract: ContextThe prevalence of overweight in children and adolescents and obesity in adults in the United States has increased over several decades.ObjectiveTo provide current estimates of the prevalence and trends of overweight in children and adolescents and obesity in adults.Design, Setting, and ParticipantsAnalysis of height and weight measurements from 3958 children and adolescents aged 2 to 19 years and 4431 adults aged 20 years or older obtained in 2003-2004 as part of the National Health and Nutrition Examination Survey (NHANES), a nationally representative sample of the US population. Data from the NHANES obtained in 1999-2000 and in 2001-2002 were compared with data from 2003-2004.Main Outcome MeasuresEstimates of the prevalence of overweight in children and adolescents and obesity in adults. Overweight among children and adolescents was defined as at or above the 95th percentile of the sex-specific body mass index (BMI) for age growth charts. Obesity among adults was defined as a BMI of 30 or higher; extreme obesity was defined as a BMI of 40 or higher.ResultsIn 2003-2004, 17.1% of US children and adolescents were overweight and 32.2% of adults were obese. Tests for trend were significant for male and female children and adolescents, indicating an increase in the prevalence of overweight in female children and adolescents from 13.8% in 1999-2000 to 16.0% in 2003-2004 and an increase in the prevalence of overweight in male children and adolescents from 14.0% to 18.2%. Among men, the prevalence of obesity increased significantly between 1999-2000 (27.5%) and 2003-2004 (31.1%). Among women, no significant increase in obesity was observed between 1999-2000 (33.4%) and 2003-2004 (33.2%). The prevalence of extreme obesity (body mass index ≥40) in 2003-2004 was 2.8% in men and 6.9% in women. In 2003-2004, significant differences in obesity prevalence remained by race/ethnicity and by age. Approximately 30% of non-Hispanic white adults were obese as were 45.0% of non-Hispanic black adults and 36.8% of Mexican Americans. Among adults aged 20 to 39 years, 28.5% were obese while 36.8% of adults aged 40 to 59 years and 31.0% of those aged 60 years or older were obese in 2003-2004.ConclusionsThe prevalence of overweight among children and adolescents and obesity among men increased significantly during the 6-year period from 1999 to 2004; among women, no overall increases in the prevalence of obesity were observed. These estimates were based on a 6-year period and suggest that the increases in body weight are continuing in men and in children and adolescents while they may be leveling off in women.

9,175 citations


"Public transit, obesity, and medica..." refers background in this paper

  • ...These statistics are reported by Flegal et al. (2002) and Ogden et al. (2006), who examine data from the 1960–1962 National Health Examination Survey (NHES) and subsequent NHANES waves....

    [...]


Journal ArticleDOI
09 Oct 2002-JAMA
TL;DR: The increases in the prevalences of obesity and overweight previously observed continued in 1999-2000, and increases occurred for both men and women in all age groups and for non-Hispanic whites, non- Hispanic blacks, and Mexican Americans.
Abstract: ContextThe prevalence of obesity and overweight increased in the United States between 1978 and 1991. More recent reports have suggested continued increases but are based on self-reported data.ObjectiveTo examine trends and prevalences of overweight (body mass index [BMI] ≥25) and obesity (BMI ≥30), using measured height and weight data.Design, Setting, and ParticipantsSurvey of 4115 adult men and women conducted in 1999 and 2000 as part of the National Health and Nutrition Examination Survey (NHANES), a nationally representative sample of the US population.Main Outcome MeasureAge-adjusted prevalence of overweight, obesity, and extreme obesity compared with prior surveys, and sex-, age-, and race/ethnicity–specific estimates.ResultsThe age-adjusted prevalence of obesity was 30.5% in 1999-2000 compared with 22.9% in NHANES III (1988-1994; P<.001). The prevalence of overweight also increased during this period from 55.9% to 64.5% (P<.001). Extreme obesity (BMI ≥40) also increased significantly in the population, from 2.9% to 4.7% (P = .002). Although not all changes were statistically significant, increases occurred for both men and women in all age groups and for non-Hispanic whites, non-Hispanic blacks, and Mexican Americans. Racial/ethnic groups did not differ significantly in the prevalence of obesity or overweight for men. Among women, obesity and overweight prevalences were highest among non-Hispanic black women. More than half of non-Hispanic black women aged 40 years or older were obese and more than 80% were overweight.ConclusionsThe increases in the prevalences of obesity and overweight previously observed continued in 1999-2000. The potential health benefits from reduction in overweight and obesity are of considerable public health importance.

6,439 citations


"Public transit, obesity, and medica..." refers background in this paper

  • ...Obesity rates are calculated by Flegal et al. (2002) and Ogden et al. (2006), appear as circles in the figure, and are plotted at the midpoint of the NHANES examination period....

    [...]

  • ...These statistics are reported by Flegal et al. (2002) and Ogden et al. (2006), who examine data from the 1960–1962 National Health Examination Survey (NHES) and subsequent NHANES waves....

    [...]

  • ...These statistics are reported by Flegal et al. (2002) and Ogden et al. (2006), who examine data from the 1960–1962 National Health Examination Survey (NHES) and subsequent NHANES waves....

    [...]


Journal ArticleDOI
27 Oct 1999-JAMA
TL;DR: A graded increase in the prevalence ratio (PR) was observed with increasing severity of overweight and obesity for all of the health outcomes except for coronary heart disease in men and high blood cholesterol level in both men and women.
Abstract: ContextOverweight and obesity are increasing dramatically in the United States and most likely contribute substantially to the burden of chronic health conditions.ObjectiveTo describe the relationship between weight status and prevalence of health conditions by severity of overweight and obesity in the US population.Design and SettingNationally representative cross-sectional survey using data from the Third National Health and Nutrition Examination Survey (NHANES III), which was conducted in 2 phases from 1988 to 1994.ParticipantsA total of 16,884 adults, 25 years and older, classified as overweight and obese (body mass index [BMI] ≥25 kg/m2) based on National Institutes of Health recommended guidelines.Main Outcome MeasuresPrevalence of type 2 diabetes mellitus, gallbladder disease, coronary heart disease, high blood cholesterol level, high blood pressure, or osteoarthritis.ResultsSixty-three percent of men and 55% of women had a body mass index of 25 kg/m2 or greater. A graded increase in the prevalence ratio (PR) was observed with increasing severity of overweight and obesity for all of the health outcomes except for coronary heart disease in men and high blood cholesterol level in both men and women. With normal-weight individuals as the reference, for individuals with BMIs of at least 40 kg/m2 and who were younger than 55 years, PRs were highest for type 2 diabetes for men (PR, 18.1; 95% confidence interval [CI], 6.7-46.8) and women (PR, 12.9; 95% CI, 5.7-28.1) and gallbladder disease for men (PR, 21.1; 95% CI, 4.1-84.2) and women (PR, 5.2; 95% CI, 2.9-8.9). Prevalence ratios generally were greater in younger than in older adults. The prevalence of having 2 or more health conditions increased with weight status category across all racial and ethnic subgroups.ConclusionsBased on these results, more than half of all US adults are considered overweight or obese. The prevalence of obesity-related comorbidities emphasizes the need for concerted efforts to prevent and treat obesity rather than just its associated comorbidities.

4,809 citations


"Public transit, obesity, and medica..." refers background in this paper

  • ...Part of the 2001 NHTS included a daily travel diary in which household respondents were asked to self-report all trips, their purposes, starting and ending times, and the means of transportation during an assigned travel day....

    [...]

  • ...Chronic illnesses such as diabetes and musculoskeletal disorder associated with obesity (Must et al., 1999) cost additional dollars of medical expenditure....

    [...]


OtherDOI
01 Jan 2003

4,617 citations


Journal ArticleDOI
09 Oct 2002-JAMA
TL;DR: The prevalence of overweight among children in the United States is continuing to increase, especially among Mexican-American and non-Hispanic black adolescents.
Abstract: ContextThe prevalence of overweight among children in the United States increased between 1976-1980 and 1988-1994, but estimates for the current decade are unknown.ObjectiveTo determine the prevalence of overweight in US children using the most recent national data with measured weights and heights and to examine trends in overweight prevalence.Design, Setting, and ParticipantsSurvey of 4722 children from birth through 19 years of age with weight and height measurements obtained in 1999-2000 as part of the National Health and Nutrition Examination Survey (NHANES), a cross-sectional, stratified, multistage probability sample of the US population.Main Outcome MeasurePrevalence of overweight among US children by sex, age group, and race/ethnicity. Overweight among those aged 2 through 19 years was defined as at or above the 95th percentile of the sex-specific body mass index (BMI) for age growth charts.ResultsThe prevalence of overweight was 15.5% among 12- through 19-year-olds, 15.3% among 6- through 11-year-olds, and 10.4% among 2- through 5-year-olds, compared with 10.5%, 11.3%, and 7.2%, respectively, in 1988-1994 (NHANES III). The prevalence of overweight among non-Hispanic black and Mexican-American adolescents increased more than 10 percentage points between 1988-1994 and 1999-2000.ConclusionThe prevalence of overweight among children in the United States is continuing to increase, especially among Mexican-American and non-Hispanic black adolescents.

4,204 citations


Frequently Asked Questions (1)
Q1. What contributions have the authors mentioned in the paper "Public transit, obesity, and medical costs: assessing the magnitudes" ?

This paper assesses the potential benefits of increased walking and reduced obesity associated with taking public transit in terms of dollars of medical costs saved and disability avoided. Further research is warranted on the net impact of transit usage on all behaviors, including caloric intake and other types of exercise, and on whether policies can promote transit usage at acceptable cost.