The New England Journal of Medicine

EFFECTS OF WALKING ON MORTALITY AMONG NONSMOKING RETIRED MEN

AMY A. HAKIM, M.S., HELEN PETROVITCH, M.D., CECIL M. BURCHFIEL, PH.D., G. WEBSTER ROSS, M.D.,
BEATRIZ L. RODRIGUEZ, M.D., PH.D., LON R. WHITE, M.D., KATSUHIKO YANO, M.D., J. DAVID CURB, M.D.,
AND ROBERT D. ABBOTT, PH.D.

ABSTRACT mortality.7,8 At the time of study enrollment, when the subjects

were 45 to 68 years old, they received a complete physical exam-
Background The potential benefit of low-intensity ination. The procedures followed were in accordance with insti-
activity in terms of longevity among older men has tutional guidelines and approved by the Kuakini Medical Center
not been clearly documented. We examined the as- Research and Institutional Review Committee. Informed consent
sociation between walking and mortality in a cohort was obtained from the study participants.
of retired men who were nonsmokers and physically The follow-up period for this study began at the time of a base-
capable of participating in low-intensity activities on line examination, which occurred between 1980 and 1982, when
a daily basis. walking was first assessed. The sample comprised 1379 members
of a cohort of men who were among the original participants in
Methods We studied 707 nonsmoking retired men, the Cooperative Lipoprotein Phenotyping Study.9 That study and
61 to 81 years of age, who were enrolled in the Hon- the sample included in this report have been described else-
olulu Heart Program. The distance walked (miles per where.10
day) was recorded at a base-line examination, which After the base-line examination, 12 years of follow-up were
took place between 1980 and 1982. Data on overall available in which to assess the relation between the distance
mortality (from any cause) were collected over a 12- walked and the risk of death on the basis of comprehensive sur-
year period of follow-up. veillance of death certificates, hospital admissions, and obituary
Results During the follow-up period, there were notices. As of 1990, 62 of the 8006 men in the original cohort
208 deaths. After adjustment for age, the mortality had moved off the island of Oahu, resulting in an out-migration
rate of about 1 per 1000 men per year. The current survival status
rate among the men who walked less than 1 mile of only five men is unknown.
(1.6 km) per day was nearly twice that among those At the time the follow-up began (1980 to 1982), the men were
who walked more than 2 miles (3.2 km) per day (40.5 asked about the average distance they walked per day. Overall
percent vs. 23.8 percent, P  0.001). The cumulative measures of activity at a variety of intensities were also assessed
incidence of death after 12 years for the most active by recording the number of hours per day spent at each of five
walkers was reached in less than 7 years among the levels of activity with the use of questionnaires similar to those of
men who were least active. The distance walked re- the Framingham 4 and Puerto Rico11 heart studies. The five levels
mained inversely related to mortality after adjust- of activity were basal (sleeping or lying down), sedentary (sitting
ment for overall measures of activity and other risk or standing), slight (e.g., casual walking), moderate (e.g., light
carpentry or gardening), and heavy (e.g., lifting or shoveling).
factors (P  0.01). Only physically capable men were included in this study. Such
Conclusions Our findings in older physically ca- men were included for follow-up if they reported undertaking at
pable men indicate that regular walking is associat- least one hour of slight, moderate, or heavy activity on a daily ba-
ed with a lower overall mortality rate. Encouraging sis. All the men also indicated that they were retired. Men who
elderly people to walk may benefit their health. were working full time or part time and those seeking employ-
(N Engl J Med 1998;338:94-9.) ment were excluded from follow-up. In addition, because ciga-
©1998, Massachusetts Medical Society. rette smoking is known to confound the relation between physical
activity and mortality,12-15 only nonsmoking men were included in
the study. After the exclusions, 707 men, 61 to 81 years old,
remained in the study sample; their average age (SD) was
68.95.1 years.

T
HE benefits of low-intensity activity in re-
Statistical Analysis
ducing the risk of cardiovascular disease
and increasing longevity have not been To help isolate the independent effect of the distance walked
clearly identified.1-6 Even less is known on overall mortality, the statistical analysis included adjustments
about the effects of such activities in older people.
We undertook this study to examine the association
between walking and mortality after adjustment for
several concomitant risk factors. Special attention was
given to a cohort of older nonsmoking men who From the Division of Biostatistics, University of Virginia School of Med-
icine, Charlottesville (A.A.H., R.D.A.); the University of Minnesota Med-
were retired and physically capable of participating ical School, Minneapolis (A.A.H.); and the Department of Medicine, John
in low-intensity activities on a daily basis. A. Burns School of Medicine, University of Hawaii (H.P., C.M.B., G.W.R.,
B.L.R., J.D.C., R.D.A.); the Honolulu Heart Program, Kuakini Medical
METHODS Center (H.P., B.L.R., K.Y., J.D.C., R.D.A.); the Honolulu Epidemiology
Research Section, Epidemiology and Biometry Program, National Heart,
Study Population and Procedures Lung, and Blood Institute (C.M.B.); the Department of Veterans Affairs
(G.W.R.); and the National Institute on Aging (L.R.W.) — all in Honolu-
Since 1965, the Honolulu Heart Program has followed 8006 lu. Address reprint requests to Dr. Abbott at the Division of Biostatistics,
men of Japanese ancestry living on the island of Oahu, Hawaii, Box 600, University of Virginia School of Medicine, Charlottesville, VA
for the development of cardiovascular disease and cause-specific 22908.

94  Ja nuar y 8 , 1 9 9 8

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.
 E F F E C TS O F WA L K I NG O N M O RTA L I T Y A M O NG NO NS M O K I NG R ET I R E D M E N

for several possible risk factors. These included age, concentra-

tions of total and high-density lipoprotein (HDL) cholesterol, TABLE 1. UNADJUSTED AND AGE-ADJUSTED 12-YEAR CUMULATIVE
blood pressure, whether there was diabetes (on the basis of the MORTALITY ACCORDING TO DISTANCE WALKED PER DAY.
medical history or the use of insulin or oral hypoglycemic thera-
py), and alcohol intake. With the use of 24-hour dietary-recall
methods,16 nutritional risk factors were recorded, including total CAUSE OF DEATH AND
calories consumed per day, the percentage of calories from fat, DISTANCE WALKED
(MILES/DAY)* AGE† MORTALITY RATE
protein, and carbohydrates, and the strength of the subjects’ pref-
UNADJUSTED AGE-ADJUSTED
erence for a Japanese diet (calculated as a percentage of foods
consumed). Further description of the risk factors is provided % (no. of deaths/
elsewhere.7,8,16 yr total no.) %
Efforts to examine the effects of the distance walked per day
on mortality also included adjustments for an overall physical- All causes
activity index calculated as a weighted sum of the number of 0.0–0.9 69.85.2 43.1 (65/151) 40.5
hours spent at each of the five activity levels. In general, the high- 1.0–2.0 69.05.1 27.7 (105/379)‡ 27.4§
2.1–8.0 67.94.6 21.5 (38/177)‡ 23.8¶
er values of the index corresponded to more active lifestyles, P value for trend 0.001 0.002
whereas lower indexes corresponded to lifestyles that were more Coronary heart disease
sedentary.4,11,15,17 The physical-activity index provided a means of or stroke
isolating an independent relation between walking and mortality 0.0–0.9 69.35.0 6.6 (8/122) 6.2
by adjusting for the overall intensity of activity that might coexist 1.0–2.0 68.95.1 5.6 (18/321) 5.5
with the various distances walked. 2.1–8.0 67.94.6 2.1 (3/145) 2.3
To describe the way the distance walked might vary with the P value for trend 0.102 0.143
overall physical-activity index and other confounding risk factors, Cancer**
0.0–0.9 69.65.2 13.4 (19/142) 12.8
we calculated age-adjusted means for each of the factors across 1.0–2.0 68.85.0 9.4 (34/361) 9.4
the ranges of distances walked (in miles per day). The procedures 2.1–8.0 67.74.4 5.3 (9/169)†† 5.6‡‡
for adjustment were based on analysis-of-covariance methods that P value for trend 0.008 0.013
used general linear and logistic-regression models.18
Among the ranges of distance walked, the cumulative incidence *To convert distances to kilometers, multiply by 1.609.
of mortality was also derived from estimated Kaplan–Meier sur- †Mean (SD) ages are those at the base-line examination (1980–1982).
vival curves19 and calculated as a cumulative percentage of deaths
‡P0.001 for the comparison with the men who walked less than 1 mile
across the 12 years of follow-up. Proportional-hazards regression per day.
models20 were also used to examine the independent effect of
walking on the risk of death and to provide estimates of the rel- §P  0.003 for the comparison with the men who walked less than
1 mile per day.
ative risk among the ranges of distance walked. All reported P val-
ues were based on two-sided tests of significance. ¶P  0.001 for the comparison with the men who walked less than
1 mile per day.
RESULTS Preexisting cases of coronary heart disease and stroke were excluded
from cause-specific follow-up.
Among the 707 men included in this study, the **Preexisting cases of cancer were excluded from cause-specific follow-up.
average (SD) distance walked was 1.81.3 miles ††P  0.01 for the comparison with the men who walked less than
(2.92.1 km) per day. All the men participated in 1 mile per day.
‡‡P  0.02 for the comparison with the men who walked less than
at least 1 hour of slight activity on a daily basis 1 mile per day.
(4.51.8 hours on average). Most of the men (589)
participated in at least 1 hour of moderate daily ac-
tivity (2.91.7 hours on average), whereas only 45
reported undertaking any heavy activity on a daily
basis (1.50.9 hours on average). continuous variable, it was positively related to a de-
During the 12 years of follow-up, there were 208 creased risk of death (P0.001 without adjustment
deaths. Thirty-three were due to coronary heart dis- for age and P  0.002 after adjustment for age).
ease, 19 to stroke, 68 to cancer, and 88 to other The effects of walking on death due to coronary
causes. The median time to death was seven years. heart disease could not be adequately distinguished
Table 1 shows the unadjusted and age-adjusted from its effects on death due to stroke, since mortal-
cumulative mortality over the 12-year follow-up pe- ity from each cause occurred too infrequently in the
riod among the men according to the range of dis- study sample. Nevertheless, walking appeared to pro-
tance walked per day. For the men who walked less tect against these events, although the findings were
than 1 mile (1.6 km) per day, the unadjusted mor- not statistically significant. After the exclusion of the
tality rate was 43.1 deaths per 100 men. For the men men with known coronary heart disease and stroke
who walked more than 2 miles (3.2 km) per day, the at the base-line examination, 6.6 percent of those
unadjusted mortality was more than halved (21.5 who walked the least died of one of these causes in
per 100). The 12-year cumulative mortality rate was the course of follow-up, as compared with 2.1 per-
significantly lower among the men who walked a cent of the men who walked more than two miles
mile or more per day than among those who walked per day.
shorter distances (P0.001). Adjustment for age Cancer was the most common cause of death;
had a negligible effect on the observed incidence of 13.4 percent of the men who walked less than one
death. When the distance walked was modeled as a mile per day died of cancer, as compared with 5.3

Vol ume 338 Numbe r 2  95

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.
 The New England Journal of Medicine

percent of those who walked more than two miles per To help determine whether the risk of death could
day. The difference in the risk of cancer between these be attributed to an association between walking and
groups was statistically significant both with and with- the other risk factors listed in Table 2, proportional-
out adjustment for age (P  0.01 and P  0.02, respec- hazards regression models were estimated to control
tively). for possible confounding influences of these factors.
Figure 1 shows the overall cumulative incidence of The results of these analyses are shown in Table 3,
mortality from all causes over time according to with comparisons of the expected risks of death for
ranges of distance walked per day. Throughout most the ranges of distance walked per day.
of the follow-up period there is a clear ordering of After adjustment for risk factors, the risk of death
the incidence curves across the ranges of distance among the men who walked less than one mile per
walked. Men who walked the least (1 mile per day) day was 1.8 times that among the men who walked
had the highest incidence of death, followed by the more than two miles per day (P  0.009). The risk
men who walked 1 to 2 miles per day and those who of death among the men who walked the least
walked more than 2 miles per day. The cumulative (1 mile per day) was 50 percent greater than it was
incidence of death in 12 years among the most ac- among the men who walked 1 to 2 miles per day
tive walkers was reached in less than 7 years among (P  0.008). The risk of death among the men who
the men who were least active. walked 1 to 2 miles per day was also greater than
Table 2 shows the associations between the other that among the men who walked longer distances
risk factors and the distance walked. As expected, (2 miles per day), although the difference was not
the trend for the physical-activity index was signifi- statistically significant.
cant (P0.001). In contrast, the distance walked per In addition to selecting men who were physically
day was unrelated to total cholesterol concentration, capable, we also attempted to control for possible
blood pressure, or alcohol intake. The body-mass in- influences of preexisting illness and subclinical dis-
dex (the weight in kilograms divided by the square ease by excluding the men who died within a year
of the height in meters) and the concentration of after follow-up began. The exclusions did little to al-
HDL cholesterol increased, however, and the per- ter the findings.
centage of men with diabetes tended to decrease The distance walked had a significant inverse rela-
with increases in the distance walked, although the tion with the risk of death from cancer but not with
trends were not significant. The distance walked was the risk of death from coronary heart disease or
unrelated to the nutritional variables. stroke. After adjustment for risk factors, the risk of
death from cancer among the men who walked the
least was 2.4 times that among those who walked
the most (P  0.03).
45
DISCUSSION
40 Ranges of distance
In this study we investigated the effects of low-
walked (miles/day)
intensity activity (walking) on overall mortality in a
35 0.0 – 0.9
cohort of nonsmoking, physically capable older men
1.0 – 2.0
who participated in the Honolulu Heart Program.
2.1 – 8.0
Total Mortality (%)

30 Our results suggest that walking is associated with a

lower risk of death among such men.
25 Although the relative risks appear to be small, the
effects of walking on the absolute differences in
20 mortality are actually large. After 12 years of follow-
up, 43.1 percent of the men who walked less than
15 one mile per day had died, as compared with 21.5
percent of the men who walked more than two miles
10
per day. To quantify further the effect of walking on
mortality, we modeled the distance walked as a con-
tinuous variable; the results suggest that the risk of
5
death can be reduced by 19 percent when the dis-
tance walked is increased by one mile per day.
0 In this study we specifically examined the relation
0 2 4 6 8 10 12
between the low-intensity activity of walking and
Year of Follow-up the risk of death among nonsmoking, retired, phys-
Figure 1. Cumulative Mortality According to Year of Follow-up ically capable men. The purpose of limiting the fol-
and Distance Walked per Day. low-up to retired men was to reduce potential bias
To convert distances to kilometers, multiply by 1.609. due to the inclusion of men who continued to work

96  Ja nuar y 8 , 1 9 9 8

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.
 E F F E C TS O F WA L K I NG O N M O RTA L I T Y A M O NG NO NS M O K I NG R ET I R E D M E N

TABLE 2. AGE-ADJUSTED RISK FACTORS AT BASE LINE, TABLE 3. RELATIVE RISK OF DEATH ACCORDING TO
ACCORDING TO DISTANCE WALKED PER DAY.* DISTANCE WALKED PER DAY, WITH ADJUSTMENT FOR AGE
AND FOR OTHER RISK FACTORS.

RISK FACTOR DISTANCE WALKED†

CAUSE OF DEATH AND
0.0 –0.9 1.0–2.0 2.1–8.0 DISTANCE WALKED
MILE/DAY MILES/DAY MILES/DAY (MILES/DAY)* RELATIVE RISK (95% CI)
RISK FACTOR–
Physical-activity index 29.93.5 30.93.2§ 31.02.6
AGE-ADJUSTED ADJUSTED†
(metabolic equivalents)‡
Total cholesterol (mg/dl) 21435.1 20934.4 21435.7 All deaths
HDL cholesterol (mg/dl) 45.912.5 46.012.2 47.812.4 0.0–0.9 vs. 2.1–8.0 1.9 (1.3–2.9)‡ 1.8 (1.2–2.7)§
Body-mass index** 23.43.1 23.73.0 23.82.7 0.0–0.9 vs. 1.0–2.0 1.6 (1.2–2.2)¶ 1.5 (1.1–2.1) .

1.0–2.0 vs. 2.1–8.0 1.2 (0.8–1.7) 1.1 (0.8–1.7)

Hypertension (%) 41.9 40.9 48.9 P value for trend 0.002 0.01
Diabetes (%) 28.9 26.0 23.5 Coronary heart disease
Total caloric intake 1776566 1799534 1790556 or stroke**
(kcal/day) 0.0–0.9 vs. 2.1–8.0 3.1 (0.8–11.9) 2.6 (0.7–10.3)
Protein (% of calories) 16.23.9 16.64.2 16.84.4 0.0–0.9 vs. 1.0–2.0 1.2 (0.5–2.8) 1.1 (0.4–2.5)
1.0–2.0 vs. 2.1–8.0 2.6 (0.8–8.9) 2.5 (0.7–8.6)
Fat (% of calories) 30.310.3 30.19.5 30.210.0 P value for trend 0.14 0.32
Carbohydrates (% of calories) 51.411.4 51.710.5 51.711.5 Cancer††
Preference for Japanese diet 48.519.0 47.117.6 47.120.7 0.0–0.9 vs. 2.1–8.0 2.5 (1.1–5.6)‡‡ 2.4 (1.1–5.4)§§
(% of foods) 0.0–0.9 vs. 1.0–2.0 1.5 (0.9–2.6) 1.5 (0.9–2.7)
Alcohol consumption 12.124.5 10.518.9 10.921.9 1.0–2.0 vs. 2.1–8.0 1.7 (0.8–3.5) 1.6 (0.8–3.4)
(g/day) P value for trend 0.01 0.02

*To convert distances to kilometers, multiply by 1.609.

*Plus–minus values are means SD.
†Values are the relative risks adjusted for age, total and HDL cholesterol,
†To convert distances to kilometers, multiply by 1.609. hypertension, diabetes, alcohol use, the overall physical-activity index, and
‡P0.001 by test for trend. preference for a Japanese diet (percentage of foods). CI denotes confidence
interval.
§P  0.002 for the comparison with the men who walked less than 1 mile
per day. ‡P0.001 for the comparison between groups.
¶P0.003 for the comparison with the men who walked less than §P0.009 for the comparison between groups.
1 mile per day. ¶P0.003 for the comparison between groups.
To convert values to millimoles per liter, multiply by 0.02586. P0.008 for the comparison between groups.
**The body-mass index was calculated as the weight in kilograms divid- **Preexisting cases of coronary heart disease and stroke were excluded
ed by the square of the height in meters. from cause-specific follow-up.
††Preexisting cases of cancer were excluded from cause-specific follow-up.
‡‡P0.02 for the comparison between groups.
§§P0.03 for the comparison between groups.

away from home and to focus specifically on activity fits of walking in this sample are harder to evaluate.
during retirement. Excluding men who were physi- Although walking was inversely related to total mor-
cally incapable of participating in low-intensity activ- tality and possibly to death from cancer, associations
ities on a daily basis also allowed us to focus on men were weaker than in the sample of men who were re-
for whom changes in physical activity would be pos- tired nonsmokers.
sible. Our inclusion of only men who were physically Unfortunately, observational studies often have a
capable makes it less likely that the levels of activity limited ability to describe relations between physical
observed in the Honolulu Heart Program were re- activity and the risk of disease because of difficulties
lated to mortality through associations with disabil- in quantifying highly variable behavioral patterns on
ity and physical impairment. Even among the men the basis of self-reported information and individual
who reported walking at least a half-mile a day, the recall. Selection bias may also exist among older
distance walked continued to be inversely associated members of the Honolulu cohort, since morbidity
with mortality after adjustment for age and the oth- and mortality may have removed men who were per-
er risk factors (P  0.04). haps less fit, leaving a group of healthy survivors
The distance walked also appeared to have a ben- who were more robust.
eficial effect on mortality among the men who were Documenting the consistency of behavioral pat-
excluded from follow-up. In this group, most were terns over time is also difficult, particularly in those
still working (455), and a large proportion smoked who die before their behavior can be reassessed. Of
cigarettes (259). Because of the potential influence the original sample of 707 nonsmoking and retired
of diverse work environments and the confounding men, however, 422 were reexamined 10 years later
effects of cigarette smoking,12-15 however, the bene- (1991 to 1993). Among those who walked the most

Vol ume 338 Numbe r 2  97

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.
 The New England Journal of Medicine

at the time of the base-line examination (2 miles population into the early part of the next century.1-6
per day), 29 percent (34 of 119) continued to do so Of the studies published, however, few controlled
after 10 years, and 60 percent (71 of 119) continued for confounding risk factors that are relevant to old-
to walk a mile or more per day. Among those who er retired men, including physical capabilities and
walked the least at base line (1 mile per day), 74 the use of alcohol and tobacco. Other studies tend-
percent (53 of 72) continued to walk less than a ed to define activity more broadly or to focus less
mile per day 10 years later. Although levels of activ- narrowly on older retired men who were physically
ity may be expected to decline as age increases in the capable than our study did. Only the Harvard Alum-
more active men, such data suggest that daily walk- ni Study has addressed the effects of walking on
ing over a period of 10 years was not uncommon mortality rates.3 The results of that study indicated
and may be a factor in reducing the risk of early that men 35 to 74 years of age who walked 1.3 miles
mortality among older men. (2.1 km) or more per day had a 22 percent lower
Information on specific forms of activity other than risk of death than men who walked less than 0.3
walking in the Honolulu Heart Program is also limit- mile (0.5 km) per day. The Harvard investigators,
ed. In our sample, only 29 men reported that they however, did not limit their study to retired non-
jogged. Although data were available on the numbers smoking men who were physically capable of low-
of flights of stairs climbed, there was no association intensity activity.
between climbing stairs and the distance walked or Of course, the effects on longevity of intentional
mortality. The numbers of flights of stairs climbed efforts to increase the distance walked per day by
could have little meaning here, since the value of such physically capable older men cannot be addressed in
information might be influenced by the number of our study. Our findings do, however, provide some
stories in a home as well as by physical ability. evidence that mortality is reduced when the distance
Information about the intensity of walking by the walked is increased. In the light of previous evidence
men in this study also was not available. Presumably, that active lifestyles reduce the risk of cardiovascular
however, the intensity was less variable and possibly disease and other adverse outcomes in younger and
lower in this group of men than it might be in more diverse groups of people, increasing the amount
groups that are more heterogeneous. Walking in of low-intensity activity is likely to benefit the health
Hawaii may also be more easily sustained and uni- of the elderly as well. In addition, compliance with
formly practiced throughout the year because of the recommendations to increase the time spent in sim-
mild climate. Concomitant high-intensity activity is ple activities such as walking, which require only
probably not a confounding factor, since only a few modest amounts of effort, may be easier to achieve
men reported undertaking any heavy activity (45 of than compliance with recommendations of more
707). After these men were excluded, the associa- vigorous exercise.
tions between walking and mortality remained sig-
nificant. Supported by a contract (NO1-HC-05102) with the National Heart,
Since walking appears to have a positive effect in Lung, and Blood Institute and by a Research Centers in Minority Institu-
reducing the risk of death from cancer and cardio- tions Award (P20 RR/AI 11091) from the National Institutes of Health.
vascular disease in addition to its effect on overall
mortality, the explanatory mechanisms are probably REFERENCES
multifactorial. Presumably, walking reduces the risk
1. Sherman SE, D’Agostino RB, Cobb JL, Kannel WB. Does exercise re-
of death from coronary heart disease and stroke duce mortality rates in the elderly? Experience from the Framingham Heart
through mechanisms related to cardiovascular fit- Study. Am Heart J 1994;128:965-72.
ness, including effects on hypertension, lipid pro- 2. Simonsick EM, Lafferty ME, Phillips CL, et al. Risk due to inactivity
in physically capable older adults. Am J Public Health 1993;83:1443-50.
files, and clotting mechanisms. Although such an as- 3. Paffenbarger RS Jr, Hyde RT, Wing AL, Lee I-M, Jung DL, Kampert
sociation would be difficult to observe, the levels of JB. The association of changes in physical-activity level and other lifestyle
characteristics with mortality among men. N Engl J Med 1993;328:538-
walking in this study may have been associated with 45.
healthier lifestyles that were adhered to throughout 4. Kannel WB, Sorlie P. Some health benefits of physical activity: the Fra-
life. Although the effects of walking on mortality mingham Study. Arch Intern Med 1979;139:857-61.
5. Kaplan GA, Seeman TE, Cohen RD, Knudsen LP, Guralnik J. Mortality
were independent of other risk factors, the men who among the elderly in the Alameda County Study: behavioral and demo-
walked the least may have been more prone to graphic factors. Am J Public Health 1987;77:307-12. [Erratum, Am J Pub-
changes over time in risk factors that affect progres- lic Health 1987;77:818.]
6. Lindsted KD, Tonstad S, Kuzma JW. Self-report of physical activity and
sion to cardiovascular disease (e.g., the development patterns of mortality in Seventh-Day Adventist men. J Clin Epidemiol
of hypertension) than those who walked greater dis- 1991;44:355-64.
7. Kagan A, Harris BR, Winkelstein W Jr, et al. Epidemiologic studies of
tances. coronary heart disease and stroke in Japanese men living in Japan, Hawaii
In addition to the Honolulu Heart Program, oth- and California: demographic, physical, dietary and biochemical character-
er studies have also focused on the benefits of phys- istics. J Chronic Dis 1974;27:345-64.
8. Yano K, Reed DM, McGee DL. Ten-year incidence of coronary heart
ical activity in older men and women, particularly in disease in the Honolulu Heart Program: relationship to biologic and life-
the light of the projected increases in the elderly style characteristics. Am J Epidemiol 1984;119:653-66.

98  Ja nuar y 8 , 1 9 9 8

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.
 E F F E C TS O F WA L K I NG O N M O RTA L I T Y A M O NG NO NS M O K I NG R ET I R E D M E N

9. Castelli WP, Cooper GR, Doyle JT, et al. Distribution of triglyceride and mortality: a 27-year follow-up of middle-aged men. JAMA 1993;270:
and total LDL and HDL cholesterol in several populations: a cooperative 2823-8.
lipoprotein phenotyping study. J Chronic Dis 1977;30:147-69. 15. Abbott RD, Rodriguez BL, Burchfiel CM, Curb JD. Physical activity
10. Heilbrun LK, Kagan A, Nomura A, Wasnich RD. The origins of epi- in older middle-aged men and reduced risk of stroke: the Honolulu Heart
demiologic studies of heart disease, cancer and osteoporosis among Hawaii Program. Am J Epidemiol 1994;139:881-93.
Japanese. Hawaii Med J 1985;44:294-6. 16. Yano K, Reed DM, Curb JD, Hankin JH, Albers JJ. Biological and
11. Garcia-Palmieri MR, Costas R Jr, Cruz-Vidal M, Sorlie PD, Havlik RJ. dietary correlates of plasma lipids and lipoproteins among elderly Japanese
Increased physical activity: a protective factor against heart attacks in Puer- men in Hawaii. Arteriosclerosis 1986;6:422-33.
to Rico. Am J Cardiol 1982;50:749-55. 17. Donahue RP, Abbott RD, Reed DM, Yano K. Physical activity and cor-
12. Garrison RJ, Feinleib M, Castelli WP, McNamara PM. Cigarette onary heart disease in middle-aged and elderly men: the Honolulu Heart
smoking as a confounder of the relationship between relative weight and Program. Am J Public Health 1988;78:683-5.
long-term mortality: the Framingham Heart Study. JAMA 1983;249: 18. Lane PW, Nelder JA. Analysis of covariance and standardization as in-
2199-203. stances of prediction. Biometrics 1982;38:613-21.
13. Harris T, Cook EF, Garrison RJ, Higgins M, Kannel WB, Goldman L. 19. Kaplan EL, Meier P. Nonparametric estimation from incomplete ob-
Body mass index and mortality among nonsmoking older persons: the Fra- servations. J Am Stat Assoc 1958;53:457-81.
mingham Heart Study. JAMA 1988;259:1520-4. 20. Cox DR. Regression models and life-tables. J R Stat Soc [B] 1972;34:
14. Lee IM, Manson JE, Hennekens CH, Paffenbarger RS Jr. Body weight 187-202.

Vo l u m e 3 3 8 Nu m b e r 2  99

The New England Journal of Medicine

Downloaded from nejm.org at STOCKHOLMS UNIVERSTITETSBIBL on October 31, 2019. For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.