Content uploaded by Regan L Bailey
Author content
All content in this area was uploaded by Regan L Bailey on May 02, 2014
Content may be subject to copyright.
This article was downloaded by: [76.26.48.132]
On: 23 April 2014, At: 08:13
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Nutrition in Gerontology and
Geriatrics
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/wjne21
Body Mass Index, Poor Diet Quality,
and Health-Related Quality of Life Are
Associated With Mortality in Rural Older
Adults
Dara W. Ford PhD, RD a , Terryl J. Hartman PhD, MPH, RD b ,
Christopher Still DO c , Craig Wood MS c , Diane C. Mitchell MS, RD a
, Pennifer Erickson PhD d , Regan Bailey PhD, RD e , Helen Smiciklas-
Wright PhD a , Donna L. Coffman PhD f & Gordon L. Jensen MD, PhD a
a Department of Nutritional Sciences , The Pennsylvania State
University , University Park , Pennsylvania , USA
b Department of Epidemiology , Rollins School of Public Health,
Emory University , Atlanta , Georgia , USA
c Geisinger Obesity Institute, Geisinger Health System , Danville ,
Pennsylvania , USA
d Department of Public Health Sciences , Penn State Hershey ,
Hershey , Pennsylvania , USA
e Office of Dietary Supplements, National Institutes of Health ,
Rockville , Maryland , USA
f The Methodology Center , The Pennsylvania State University , State
College , Pennsylvania , USA
Published online: 05 Mar 2014.
To cite this article: Dara W. Ford PhD, RD , Terryl J. Hartman PhD, MPH, RD , Christopher Still
DO , Craig Wood MS , Diane C. Mitchell MS, RD , Pennifer Erickson PhD , Regan Bailey PhD, RD ,
Helen Smiciklas-Wright PhD , Donna L. Coffman PhD & Gordon L. Jensen MD, PhD (2014) Body
Mass Index, Poor Diet Quality, and Health-Related Quality of Life Are Associated With Mortality
in Rural Older Adults, Journal of Nutrition in Gerontology and Geriatrics, 33:1, 23-34, DOI:
10.1080/21551197.2014.875819
To link to this article: http://dx.doi.org/10.1080/21551197.2014.875819
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [76.26.48.132] at 08:13 23 April 2014
23
Journal of Nutrition in Gerontology and Geriatrics, 33:23–34, 2014
Copyright © Taylor & Francis Group, LLC
ISSN: 2155-1197 print/2155-1200 online
DOI: 10.1080/21551197.2014.875819
Address correspondence to Dara W. Ford, PhD, RD, 3001 SW 27th Ave., Apt. 308, Miami,
FL 33133, USA. E-mail: Djw5083@gmail.com
Original Research
Body Mass Index, Poor Diet Quality, and
Health-Related Quality of Life Are Associated
With Mortality in Rural Older Adults
DARA W. FORD, PhD, RD
Department of Nutritional Sciences, The Pennsylvania State University,
University Park, Pennsylvania, USA
TERRYL J. HARTMAN, PhD, MPH, RD
Department of Epidemiology, Rollins School of Public Health,
Emory University, Atlanta, Georgia, USA
CHRISTOPHER STILL, DO, and CRAIG WOOD, MS
Geisinger Obesity Institute, Geisinger Health System,
Danville, Pennsylvania, USA
DIANE C. MITCHELL, MS, RD
Department of Nutritional Sciences, The Pennsylvania State University,
University Park, Pennsylvania, USA
PENNIFER ERICKSON, PhD
Department of Public Health Sciences, Penn State Hershey,
Hershey, Pennsylvania, USA
REGAN BAILEY, PhD, RD
Office of Dietary Supplements, National Institutes of Health,
Rockville, Maryland, USA
HELEN SMICIKLAS-WRIGHT, PhD
Department of Nutritional Sciences, The Pennsylvania State University,
University Park, Pennsylvania, USA
DONNA L. COFFMAN, PhD
The Methodology Center, The Pennsylvania State University,
State College, Pennsylvania, USA
Downloaded by [76.26.48.132] at 08:13 23 April 2014
24 D. W. Ford et al.
GORDON L. JENSEN, MD, PhD
Department of Nutritional Sciences, The Pennsylvania State University,
University Park, Pennsylvania, USA
In an aging population, potentially modifiable factors impacting
mortality such as diet quality, body mass index (BMI), and health-
related quality of life (HRQOL) are of interest. Surviving members
of the Geisinger Rural Aging Study (GRAS) (n = 5,993; aged ≥ 74
years) were contacted in the fall of 2009. Participants in the pres-
ent study were the 2,995 (1,267 male, 1,728 female; mean age
81.4 ± 4.4 years) who completed dietary and demographic ques-
tionnaires and were enrolled in the Geisinger Health Plan over
follow-up (mean = 3.1 years). Cox proportional hazards multi-
variate regression models were used to examine the associations
between all-cause mortality and BMI, diet quality, and HRQOL.
Compared to GRAS participants with BMIs in the normal range,
a BMI < 18.5 was associated with increased mortality (HR 1.85
95% CI 1.09, 3.14, P = 0.02), while a BMI of 25–29.9 was associ-
ated with decreased risk of mortality (HR 0.71 95% CI 0.55, 0.91,
P = 0.007). Poor diet quality increased risk for mortality (HR 1.53
95% CI 1.06, 2.22, P = 0.02). Finally, favorable health-related qual-
ity of life was inversely associated with mortality (HR 0.09 95% CI
0.06, 0.13, P < 0.0001). Higher diet quality and HALex scores, and
overweight status, were associated with reduced all-cause mortal-
ity in a cohort of advanced age. While underweight (BMI < 18.5)
increased risk of all-cause mortality, no association was found
between obesity and all-cause mortality in this aged cohort.
KEYWORDS aging, body mass index (BMI), diet quality, health-
related quality of life, mortality
INTRODUCTION
The associations between body mass index (BMI) and all-cause mortality
remain highly controversial in the aging population (1). Recent findings sug-
gest that overweight (BMI 25–29.9) in older persons (≥ 65 years) may be asso-
ciated with decreased risk for all-cause mortality (1, 2). While in younger and
middle-aged adults overweight and obesity are associated with increased risk
for all-cause mortality, this association appears to be attenuated with advanced
age (2). Underweight (BMI < 18.5) is associated with increased all-cause mor-
tality risk in younger, middle-aged, and older adults, but this relationship is
especially strong among those of advanced age (3, 4). The 2010 U.S. Census
Downloaded by [76.26.48.132] at 08:13 23 April 2014
BMI, Poor Diet Quality, and Health-Related Quality of Life 25
figures reveal that 5.4% of the population was aged ≥ 80 years and when con-
trasted with the 2000 figures, these old older persons were the fastest growing
segment of the U.S. population (5). Therefore, determining the relationship
between BMI and all-cause mortality in this age group is of particular interest.
Evidence regarding the association between diet quality and mortality is
mixed, and results vary based on methods that are used to assess diet quality
(6). Research in adults aged 65 years and older has shown that individuals
have decreased rates of mortality who adhere to a Mediterranean pattern, or
eat what is deemed a “healthy” pattern derived through statistical methods
including cluster and factor analysis (6). However, this is not a consistent
finding, with some studies finding no association between diet quality and
mortality outcomes in adults aged ≥ 75 years (7).
In addition to health-related factors such as BMI and diet quality,
health-related quality of life (HRQOL) is also of interest in the aging popula-
tion. HRQOL is a broad concept that encompasses subjective assessments of
overall quality of life (QOL) as well as perceptions of physical and mental
health that may impact quality of life (8). Evaluating measures for moni-
toring HRQOL in the United States is a goal of Healthy People 2020 (9),
and improvements in the subjective HRQOL may be meaningful in altering
the perceived health status of older adults. Poorer HRQOL assessed using
the SF-36 Physical and Mental Component Scores has been associated with
increased risk for mortality among adults aged 60 and older (10). The rela-
tionship between the Health and Activity Limitation Index (HALex), a specific
measure of HRQOL, and mortality, does not appear to have been previously
reported. The objectives of this investigation were to examine the associa-
tions of BMI, diet quality, and HRQOL with mortality in an aged cohort of
rural persons.
SUBJECTS AND METHODS
More than 20,000 adults aged older than 65 years who were enrolled in a
Medicare managed health maintenance organization were recruited for the
Geisinger Rural Aging Study (GRAS) in 1994 (11). Detailed information on sub-
ject recruitment has been previously published (11). Over time, the participants
have been followed as a longitudinal cohort with data repeatedly collected
on height, weight, living situation, functional status, self-rated health, and diet
among other characteristics. All participants are community-dwelling individu-
als in rural northeastern and central Pennsylvania, residing in an area where
the population density is 14–475 individuals/mile2 (5–183 individuals/km2) (12).
The cohort is primarily non-Hispanic White, and high school educated.
Surviving members of the GRAS (n = 5,993 aged ≥74 years) were mailed
surveys in the fall of 2009 regarding demographic and health information as
well as the Dietary Screening Tool (DST). Information regarding medication
Downloaded by [76.26.48.132] at 08:13 23 April 2014
26 D. W. Ford et al.
use was queried in the survey questionnaire. Participants reported use of less
than three, or three or more prescription medications daily. Of the survivors,
2,995 participants returned completed surveys and were followed forward in
time using electronic medical record and National Death Index data through
February 25, 2013. The study protocol was approved by both the Office of
Research Protections at the Pennsylvania State University and the Human
Research Protection Program of the Geisinger Health Systems Institutional
Review Board. Consent was implied by survey completion.
DST, BMI, and HALex
The DST has been described in detail previously (11, 13). Briefly, this tool
consists of 25 food-based questions which have been validated to assess
diet quality in older adults (13). The questions on the DST were derived
from information from multiple 24-hour recalls conducted with a subset of
the GRAS population. Questions were formatted to capture usual intake and
point breakdown was based on the major dietary component breakdown of
the Healthy Eating Index (HEI)-2005 (14). Questions were tested to ensure
understandability for the target population through cognitive interviewing,
and diet quality was established through comparison with both nutrient
and food group intakes (11, 13). Scores range from 0–100, with 5 “bonus”
points available for multivitamin/mineral supplement use (score could not
exceed 100). Based on the previously validated scoring algorithm, a score of
< 60 was considered “unhealthy,” 60–75 “borderline,” and > 75 “healthy” (13).
Good test-retest reliability of the DST has been previously demonstrated
with a coefficient of 0.83 (P < 0.0001). Individuals consuming a healthy diet
(DST > 75) served as the referent group for all analyses.
BMI was calculated from self-reported height and weight as reported
in the fall of 2009 and was categorized based on the National Institutes of
Health (NIH) guidelines (15). A BMI < 18.5 kg/m2 was considered under-
weight, 18.5–24.9 healthy, 25–29.9 overweight, 30–34.9 obese class I, and ≥ 35
combined obesity classes II and III. The healthy BMI range (18.5–24.9) was
the referent for all statistical analyses of BMI as a categorical variable.
HRQOL was assessed using the Health and Activity Limitation Index
(HALex) (16). This measure combines self-reported functional limitation and
self-rated health into a single HRQOL measure scored on a continuum from
0.0 (deceased) to 1.0 (optimal health) (17). An individual with no functional
limitations (i.e., no limitations in activities of daily living (ADLs) or instru-
mental activities of daily living (IADLs)) who self-reports excellent health
would receive a score of 1.00. Placement on the score matrix is derived
from five categories of self-rated health (excellent, very good, good, fair, and
poor) and six levels of activity limitation (not limited, limited- other, lim-
ited- major, unable- major, limited in IADL, limited in ADL) (17). The HALex
was developed using the National Health Interview Survey data and was
Downloaded by [76.26.48.132] at 08:13 23 April 2014
BMI, Poor Diet Quality, and Health-Related Quality of Life 27
validated in 41,104 persons older than 18 years standardized to 10,000 per-
sons by age and gender group (16). Because age was a consideration on the
original tool for score placement based on functional limitation, only those
categories befitting the ages of the present cohort (≥74 years) were used in
order to model findings as closely as possible to original development (16).
The range of scores in the present analysis is 0.1 to 1.0.
Morbidity and Mortality
A disease burden variable was created based on the sum of obesity-related
comorbidities extracted from electronic medical records including diabe-
tes mellitus, hypertension, coronary artery disease, osteoarthritis, obstruc-
tive sleep apnea, depression, and liver disease. The presence of disease at
baseline was extracted from electronic medical records. Because of the age
of our cohort, the presence of these diseases was considered prevalent, and
number of conditions present were summed and treated as a continuous
variable. This disease burden variable, based on obesity-related comorbidi-
ties, potentially falls on the pathway between diet, BMI, and mortality, and
so was not considered as a covariate in the analysis presented (18).
Deaths were identified using electronic medical records and the Social
Security Death Index data (19). The last date of data extraction for the pres-
ent analysis was February 25, 2013. All individuals surviving beyond that
point were censored at that date.
Statistical Analysis
All data were analyzed using the Statistical Analysis Software Package 9.3
(SAS Institute Inc., Cary, NC, USA). Demographic and descriptive data are
presented by gender as means with standard deviations for continuous vari-
ables and percentages for categorical variables. Cox proportional hazards
regression models were used to estimate multivariate adjusted hazard ratios
of mortality for different categories of BMI, DST, and continuous HALex
scores. Although tests of the proportional hazards assumption suggested
the relationship varied quantitatively over time (P < 0.05), the associations
were qualitatively similar, thus we present results including the full follow-
up period. Adjusted hazard ratios (HR) and their 95% confidence intervals
(95% CI) are reported. For each participant, follow-up time accrued from
administration of the baseline survey (October 28, 2009) until date of death
or end of the study period (February 25, 2013) if death did not occur. Three
separate models were used to assess the hazard ratio for mortality with each
predictor of interest (BMI, DST, and HALex), while controlling for the demo-
graphic covariates. A fourth model was used including BMI, DST, and HALex
within the same model to assess associations with mortality, while adjusting
for the two predictors not being analyzed in addition to the demographic
Downloaded by [76.26.48.132] at 08:13 23 April 2014
28 D. W. Ford et al.
covariates. The covariates included age (continuous), sex, smoking status
(ever/never), and recent weight loss or gain (greater than 5 pounds (2.3
kg) self-reported in previous six months yes/no). Self versus proxy report-
ing, living situation, physical activity, and education were all considered as
covariates but did not contribute significantly to the models and so were
excluded from analysis. Results are presented as hazard ratios (HR) with
95% confidence intervals (95% CI). Interactions between predictors of inter-
est (BMI, DST, and HALex score) and each covariate (age, sex, smoking
status, weight loss, and gain) were assessed by including each individual
factor (e.g., age) and its cross-product term in separate models. Interactions
between predictors of interest were also assessed. Significance was consid-
ered at P < 0.05.
RESULTS
Descriptive data are presented in Table 1. Of the 4009 participants who
provided completed demographic and dietary questionnaires in the fall of
2009, approximately three-quarters (n = 2,995; 1,267 male, 1,728 female;
mean age 81.3 years) were enrolled in the Geisinger Health Plan (GHP)
for at least part of the follow-up. Follow-up time ranged from 71–1,201
days, with a mean follow-up of 1,144 days (>3 years). Those who did
not respond, returned incomplete surveys, or were not enrolled in the
Health Plan were excluded from primary analysis, but were available for
comparison by demographic characteristics and mortality data. Compared
to participants with complete information, nonresponders and those with
incomplete information were significantly older (82.8 ± 4.8 y, vs. 81.4 ± 4.4
y, P < 0.05) and more likely to be female (OR 1.1 95% CI 1.0, 1.3, P <0.05).
In the entire remaining GRAS sample of 5,993, there were 1,015 (17%)
deaths during follow-up. In contrast, 360 (12%) occurred for the subset of
2,995 respondents with complete data included in the present analyses.
Incomplete or nonresponders were also more likely to be deceased at the
end of follow-up (OR 2.1 95% CI 1.8, 2.4, P < 0.05). Of those with com-
plete information, women reported significantly higher diet quality scores
(61.6 ± 12.6 vs. 57.8 ± 12.4, P <0.05) and had a higher BMI (27.3 ± 5.5 vs.
27.1 ± 4.1, P < 0.05) than their male counterparts.
In models adjusted for age, sex, smoking status, weight gain, and weight
loss, low BMI (<18.5) was associated with an increased hazard ratio for mor-
tality (HR 1.85 95% CI 1.09, 3.14, P = 0.02). In this model, being overweight
(BMI 25–29.9) was associated with a decreased hazard ratio for mortality
(HR 0.71 95% CI 0.55, 0.91, P = 0.007), while obesity was not associated with
mortality (BMI 30–34.9; HR 0.82 95% CI 0.60, 1.11, P = 0.19; BMI ≥ 35; HR 0.89
95% CI 0.62, 1.51, P = 0.89). We did not observe any meaningful effect modi-
fication for any of these associations by age, sex, education, smoking status,
Downloaded by [76.26.48.132] at 08:13 23 April 2014
BMI, Poor Diet Quality, and Health-Related Quality of Life 29
weight change, diet quality, living situation, or self versus proxy reporting.
In addition, controlling for self-reported medication use as indicated by use
of three or more prescription medications did not alter the findings so this
variable was excluded from further analyses.
In the adjusted models, individuals with an unhealthy (low) DST score
had an increased hazard ratio for mortality (HR 1.53 95% CI 1.06, 2.22,
P = 0.02) compared to those with higher quality diets. Intakes classified as
borderline were not associated with mortality, and again no significant inter-
actions were observed between DST score and age, sex, education, smoking
status, weight change, BMI, living situation, or self versus proxy report.
Finally, we assessed the relationship between HALex score and mor-
tality. In the adjusted models, a higher (more favorable) HALex score was
significantly associated with a decreased hazard ratio (HR 0.09 95% CI 0.06,
TABLE 1 Demographic and Personal Characteristics by Gender
for the Geisinger Rural Aging Study (GRAS)
Male
(n = 1,267)
Female
(n = 1,728)
Age in years, mean ± SD 81.2 ± 4.2 81.5 ± 4.8
Smoking status n (%)
Ever 51 (4.1) 63 (3.7)
Never 1,195 (95.9) 1,636 (96.)
Education n (%)
< High school 349 (28.1) 437 (25.9)
≥ High school 894 (71.9) 1,250 (74.1)
Source n (%)
Self 1,117 (89.1) 1,586 (93.0)
Proxy 137 (10.9) 120 (7.0)
Weight gain n (%)
Yes 143 (11.3) 248 (14.4)
No 1,124 (88.7) 1,480 (85.6)
Weight loss n (%)
Yes 339 (26.8) 450 (26.0)
No 923 (73.2) 1,278 (74.0)
BMI n (%)
< 18.5 12 (1.0) 50 (2.9)
18.5–24.9 315 (24.8) 508 (29.4)
25–29.9 610 (48.1) 638 (36.9)
30–34.9 271 (21.4) 368 (21.3)
35–39.9 53 (4.2) 124 (7.2)
≥ 40 6 (0.5) 40 (2.3)
DST score n (%)
Unhealthy 693 (54.7) 725 (42.0%)
Borderline 459 (36.2) 723 (41.8)
Healthy 115 (9.1) 280 (16.2)
HALex score, mean ± SD 0.76 ± 0.20 0.73 ± 0.21
Deceased n (%) 185 (14.6) 175 (10.1)
Note. DST = Dietary Screening Tool; HALex = Health and Activity Limitation
Index.
Downloaded by [76.26.48.132] at 08:13 23 April 2014
30 D. W. Ford et al.
0.13, P < 0.0001). The HALex model also contained no significant interactions
with age, sex, education, smoking status, weight change, BMI, diet quality,
living situation, or self versus proxy reporting.
Significant findings were somewhat attenuated when all predictors
were considered within the same model (Table 3). Findings indicate that
the unique attributable portion of risk remains significant for individuals
who are classified as overweight, with overweight associated with marginally
TABLE 3 Associations Between BMI, DST, and HALex Score With All-Cause Mortality in
Geisinger Rural Aging Study (GRAS); All Predictors in the Same Model
Variable
Hazard ratio
(95% CI) P value
BMI*
<18.5 1.47 (0.84, 2.47) 0.18
18.5–24.9 Ref –
25–29.9 0.78 (0.60, 1.00) 0.05
30–34.9 0.80 (0.59, 1.10) 0.17
≥35 0.75 (0.48, 1.17) 0.20
DST score+
Unhealthy 1.34 (0.91, 1.97) 0.14
Borderline 1.13 (0.76, 1.68) 0.54
Healthy Ref –
HALex score−0.09 (0.06, 0.14) <0.0001
Note. BMI = body mass index; DST = Dietary Screening Tool; HALex = Health and Activity LimitationIndex.
*Controlling for age, sex, smoking status, weight gain, weight loss, DST score (categorical), andHALex.
+Controlling for age, sex, smoking status, weight gain, weight loss, BMI (categorical), andHALex.
−Controlling for age, sex, smoking status, weight gain, weight loss, BMI (categorical), and DST (categorical).
TABLE 2 Associations Between BMI, DST, and HALex score With
All-Cause Mortality in Geisinger Rural Aging Study (GRAS; N = 2995)
Variable
Hazard ratio
(95% CI) P value*
BMI
<18.5 1.85 (1.09, 3.14) 0.02
18.5–24.9 Ref –
25–29.9 0.71 (0.55, 0.91) 0.007
30–34.9 0.82 (0.60, 1.11) 0.19
≥35 0.89 (0.62, 1.51) 0.89
DST score
Unhealthy 1.53 (1.06, 2.22) 0.02
Borderline 1.19 (0.81, 1.74) 0.39
Healthy Ref –
HALex score 0.09 (0.06, 0.13) <0.0001
Note. BMI = body mass index; DST = Dietary Screening Tool; HALex = Health
and Activity LimitationIndex.
*Each predictor was analyzed in a separate model, controlling for age, sex,
smoking status, weight gain, and weight loss.
Downloaded by [76.26.48.132] at 08:13 23 April 2014
BMI, Poor Diet Quality, and Health-Related Quality of Life 31
significant decreased risk of all-cause mortality (HR 0.78 95% CI 0.60, 1.00,
P = 0.05). In addition, increased HALex score remains highly associated with
decreased risk of all-cause mortality, even when controlling for BMI and DST
score (HR 0.09 95% CI 0.06, 0.14, P < 0.0001). Associations between under-
weight and poor quality diet with mortality were attenuated in this model,
but the anticipated trends remained.
DISCUSSION
We found that hazard ratios for all-cause mortality increased with BMI
<18.5 and poor diet quality, while they decreased with a BMI of 25–29.9
and with more favorable HALex scores. Our findings are in line with other
research demonstrating that in older adults, a low BMI is associated with
greater rates of all-cause mortality, even when controlling for weight loss
and conditions that may cause weight loss (3, 4, 20). In addition, our find-
ing that hazard ratios for all-cause mortality decreased with overweight
in adults aged 74years and older supports growing evidence that over-
weight status may confer reduced risk for all-cause mortality in persons of
advanced age (2, 3, 21). These findings persisted even when controlling for
dietary intake and health related quality of life. Mild obesity did not appear
to increase risk of all-cause mortality in this age group, further suggesting
that the excess mortality associated with obesity in older age is primar-
ily attributable to BMI class II and III (1). The current NIH recommenda-
tions for healthy BMI may be less relevant for this age group compared to
younger persons in relation to all-cause mortality. This observation does
not rule out that there may remain associations between overweight and
other adverse health outcomes. In addition, it is not clear that these find-
ings may be extended to populations younger than 74 years. Those persons
who make it to the advanced age of our cohort already reflect selection as
survivors.
In the current analyses, we did not statistically control for comorbidi-
ties because adjusting for intermediate variables (i.e., disease) may result in
biased estimates (18). However, in preliminary models where disease burden
(prevalent cases of the obesity-related chronic diseases diabetes mellitus,
hypertension, coronary artery disease, osteoarthritis, obstructive sleep apnea,
depression, and liver disease) was adjusted for as a continuous covariate, the
increased hazard ratio for mortality with low BMI remained (HR 1.89 95%
CI 1.11, 3.21, P = 0.02), while both overweight (HR 0.66 95% CI 0.51, 0.85,
P = 0.002) and class I obesity (HR 0.72 95% CI 0.53, 0.98, P = 0.04) were asso-
ciated with reduced hazard ratios. The interaction between BMI and disease
burden was not significant.
The present analysis indicates that poor diet quality is associated with
an increased hazard ratio for mortality. While previous work regarding this
Downloaded by [76.26.48.132] at 08:13 23 April 2014
32 D. W. Ford et al.
topic had been inconclusive (6, 7), our investigation is one of the first to use
a diet quality tool that was developed and validated for use in the targeted
age population (7). Again, in the models that we present, adjustments for
disease burden were not made due to concerns for biased results. In prelimi-
nary models in which disease burden was utilized as a covariate, although
moderately attenuated, the significant relationship between poor diet and
mortality remained (HR 1.47 95% CI 1.01, 2.13, P = 0.04). The majority of
previous studies controlled for disease burden (6) when examining associa-
tions between diet quality and mortality. Poor diet quality may serve as a
proxy for disease, thus significant associations may diminish when disease
burden is entered into a model. When BMI is considered as a covariate, the
significance of diet is diminished, but the trend toward increased mortality
with poor diet remains.
Finally, we found a dramatic association between the HRQOL measure
of HALex and hazard ratios for mortality. This relationship also remained
the same in models controlling for disease burden (HR 0.11 95% CI 0.07,
0.17, P < 0.0001), and when diet and BMI were covariates (HR 0.09 95% CI
0.06, 0.14, P < 0.0001). Previous studies have shown that low baseline physi-
cal and mental component scores were associated with increased mortality
over three years of follow-up in community-dwelling adults aged ≥ 65 years
in urban and rural areas of Taiwan (10). Otero-Rodriguez and colleagues
(22) also showed that a decline in HRQOL over time was associated with
increased rates of mortality in a population-based Spanish cohort of adults
at least 60 years of age. These studies, combined with the present results,
suggest that poor HALex score carries strong prognostic import for mortality
outcomes. To our knowledge our investigation is one of the first studies to
examine the association of the HALex as a specific measure of HRQOL to
mortality.
Major strengths of this study include a robust sample representing
an understudied population of aging adults (≥ 74 years). Health care
data were obtained from electronic medical records. Descriptive data for
nonresponders was also available for comparison analysis. Some limita-
tions must also be noted. Height, weight, dietary information, functional
limitations, weight loss or gain, and health rating were all self-reported,
potentially allowing for recall bias. However, for 2,221 individuals mea-
sured height and weight were available from electronic medical records
and the correlation between self-reported and measured BMI for these
individuals was strong (Pearson correlation = 0.91, P < 0.0001). Dietary
data were self-reported, but the DST measures dietary patterns and has
previously been shown to correlate well with markers of nutrient ade-
quacy in older adults (11). The DST was developed and thus far has
been used in a rather homogenous sample. Further investigation with
populations of greater diversity will be necessary to broaden applicabil-
ity of this tool.
Downloaded by [76.26.48.132] at 08:13 23 April 2014
BMI, Poor Diet Quality, and Health-Related Quality of Life 33
CONCLUSIONS
Higher diet quality, overweight status, and especially higher HALex score
were associated with reduced all-cause mortality in a cohort of advanced
age. The strong association between HALex and mortality subsisted even in
the model fully adjusted for BMI and diet quality. These findings highlight
the priority to better understand the health-related characteristics of “old
older” persons in relation to mortality outcomes as they may not be appro-
priately extrapolated from those of younger samples.
TAKE AWAY POINTS
• Both a BMI < 18.5 and poor diet quality increased risk for all-cause mortality.
• A BMI in the overweight category (25.0–29.9) was associated with a
reduced risk for all-cause mortality in older adults.
• Health-related quality of life is strongly inversely associated with risk of
all-cause mortality.
FUNDING
This work was supported by the United States Department of Agriculture
(USDA #1950-51530-010-00).
REFERENCES
1. Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality
with overweight and obesity using standard body mass index categories: a sys-
tematic review and meta-analysis. JAMA. 2013; 309(1):71–82.
2. Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL. The effect of age
on the association between body-mass index and mortality. N Engl J Med. 1998;
338(1):1–7.
3. Locher JL, Roth DL, Ritchie CS, Cox K, Sawyer P, Bodner EV, etal. Body mass
index, weight loss, and mortality in community-dwelling older adults. J Gerontol
A Biol Sci Med Sci. 2007; 62(12):1389–92.
4. Takata Y, Ansai T, Soh I, Aki fusa S, Sonoki K, Fujisawa K, etal. Association
between body mass index and mortality in an 80-year-old population.
J Am Geriatr Soc. 2007; 55(6):913–7.
5. Werner CA. The older populatio n: 2010; 2010 ce nsus briefs. 2011. Accessed at
http://www.census.gov/newsroom/releases/archives/2010_census/cb11-cn192.
html on 14 July 2013.
6. Ford DW, Jensen GL, Hartman TJ , Wray L, Smiciklas-Wright H. Association
between dietary quality and mortality in older adults: a review of the epidemio-
logical evidence. J Nutr Gerontol Geriatr. 2013; 32(2):85–105.
Downloaded by [76.26.48.132] at 08:13 23 April 2014
34 D. W. Ford et al.
7. Hsiao PY, Mitchell DC, Coffman DL, Craig Wood G, Hartman TJ, Still C, etal.
Dietary patterns and relationship to obesity-related health outcomes and mortal-
ity in adults 75 years of age or greater. J Nutr Health Aging. 2013; 17(6):566–72.
8. Health-Related Quality of Life ( HRQOL). Centers for Disease Control and
Prevention. March 17 AA, 2013. Accessed at http://www.cdc.gov/hrqol/concept.
htm on 1 July 2013.
9. Amarantos E, Martinez A, Dwyer J. Nutrition and quality of life in older adults. J
Gerontol A Biol Sci Med Sci. 2001; 56 (Spec No 2):54–64.
10. Tsai SY, Chi LY, Lee CH, Chou P. Health-related quality of life as a predictor
of mortality among community-dwelling older persons. Eur J Epidemiol. 2007;
22(1):19–26.
11. Bailey RL, Mitchell DC, Miller CK , Still CD, Jensen GL, Tucker KL, etal. A dietary
screening questionnaire identifies dietary patterns in older adults. J Nutr. 2007;
137(2):421–6.
12. Jensen GL, Kita K, Fish J, Heydt D , Frey C. Nutrition risk screening characteristics
of rural older persons: relation to functional limitations and health care charges.
Am J Clin Nutr. 1997; 66(4):819–28.
13. Bailey R, Miller PE, Mitchel DC, Ha rtman TJ, Lawrence FR, Sempos CT, etal.
Dietary screening tool identifies nutritional risk in older adults. Am J Clin Nutr.
2009; 90:1–7.
14. Guenther PM. Healthy Eating Index—20 05. Alexandria, VA: USDA; 2008.
15. U.S. Department of Health and Human S ervices. Classification of overweight
and obesity by BMI WC, and associated disease risks. Accessed at http://www.
nhlbi.nih.gov/health/public/heart/obesity/lose_wt/bmi_dis.htm on 25 August
2013.
16. Erickson P. Evaluation of a populatio n-based measure of quality of life: the
Health and Activity Limitation Index (HALex). Qual Life Res. 1998; 7(2):101–14.
17. Erickson P, Wilson R, Shannon I. Years of healthy life. Healthy People 2000 Stat
Notes. 1995; 7:1–15.
18. Robins JM, Greenland S. Identifiability and exchangeability for direct and indi-
rect effects. Epidemiology. 1992; 3(2):143–55.
19. Social Security Administration. Death M aster File. Washington, DC: U.S.
Department of Commerce National Technical Information; 2013.
20. Luchsinger JA, Patel B, Tang MX, Schupf N, Mayeux R. Body mass index, demen-
tia, and mortality in the elderly. J Nutr Health Aging. 2008; 12(2):127–31.
21. Flegal KM, Graubard BI, Williamson DF, Ga il MH. Cause-specific excess
deaths associated with underweight, overweight, and obesity. JAMA. 2007;
298(17):2028–37.
22. Otero-Rodriguez A, Leon-Munoz LM, Balboa-C astillo T, Banegas JR, Rodriguez-
Artalejo F, Guallar-Castillon P. Change in health-related quality of life as a predic-
tor of mortality in the older adults. Qual Life Res. 2010; 19(1):15–23.
Downloaded by [76.26.48.132] at 08:13 23 April 2014