Multidimensionality of Oral Health in Dentate Adults

Author(s): Gregg H. Gilbert, R. Paul Duncan, Marc W. Heft, Teresa A. Dolan and W. Bruce Vogel
Source: Medical Care, Vol. 36, No. 7 (Jul., 1998), pp. 988-1001
Published by: Lippincott Williams & Wilkins
Stable URL: http://www.jstor.org/stable/3767359
Accessed: 22-02-2016 00:46 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Lippincott Williams & Wilkins is collaborating with JSTOR to digitize, preserve and extend access to Medical Care.

http://www.jstor.org

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MEDICAL CARE
Volume36, Number7, pp 988-1001
?Lippincott-RavenPublishers
of OralHealthin DentateAdults
Multidimensionality
DDS,*t R. PAULDUNCAN,PHD,*$MARCW. HEFT,DMD, PHD,*t
GREGGH. GILBERT,
TERESAA. DOLAN,DDS, MPH,*?ANDW. BRUCEVOGEL,PHD*:
OBJECTIVES.An understanding of the valid-
ity and usefulness of self-reported measures
(as distinct from clinically determined meas-
ures) of oral health is emerging. These self-
reported measures include self-rated oral
health (SROH). Three objectives were to: (1)
describe self-rated oral health in dentate
adults, (2) quantify associations between
self-rated oral health and other measures of
oral health (oral disease and tissue damage,
pain and discomfort, functional limitation,
and disadvantage), and (3) assess the con-
struct validity of a model of oral health pro-
posed herein.
METHODS. The Florida Dental Care Study is
a longitudinal study of oral health, which in-
cluded at baseline 873 subjects who had at
least one tooth, were 45 years or older, and
who participated for an interview and clini-
cal examination.
RESULTS.The prevalence of self-rated oral
health decrements was substantial; approxi-
mately one fourth of subjects reported their
Recognizing that there is more to health than
the lack of disease, health status measures have
progressed from those based solely on mortality
or morbidity to those that include self-reported
social, psychological, functional, and behavioral
*From the Claude D. Pepper Center for Research on
Oral Health in Aging, University of Florida,Gainesville.
tFrom the Division of Oral Medicine, College of Den-
tistry,University of Florida,Gainesville.
tFrom the Department of Health Services Administra-
tion, College of Health Professions, University of Florida,
Gainesville.
?From the Division of Public Health Services and Re-
search, College of Dentistry, University of Florida,
Gainesville.
This investigation was supported by DE-12587, DE-
11020, and DE-00392. Additional support was provided
by funds from the University of Florida.
The opinions and assertions contained herein are
988
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
oral health as only fair or poor. Bivariate and
multivariate results provided consistent evi-
dence of the construct validity of the pro-
posed model of oral health. Additionally, the
salience of one measure of dental appearance
suggests that persons may use esthetic cues
when rating their oral health.
CONCLUSIONS. The proposed multidimen-
sional model of oral health has construct
validity. Self-rated oral health is affected
by oral disease and tissue damage, oral pain
and discomfort, oral functional limitation,
and oral disadvantage. These self-reported
measures and the proposed model should
provide useful information for dental care
effectiveness research. General health
status has been disaggregated into the
"physical" and the "mental;" an additional
separation into the "oral" aspects of health
seems warranted.
Key words: oral health; self-rated health;
quality of life; attitudes; dental care; epide-
miology. (Med Care 1998;36:988-1001)
measures.1-3 Efforts to assess general health
status, functional status, and other outcomes of
health care in adults have led to several instru-
ments now in widespread use.4 No single instru-
ment, however, is in widespread use in oral health
those of the authors and are not to be construed as nec-
essarily representing the views of the University of Florida
or the National Institutes of Health. The informed con-
sent of all human subjects who participatedin this inves-
tigation was obtained after the nature of the procedures
had been explained fully.
An internet home page devoted to details on the FDCS
can be found at http://www.nerdc.ufl.edu/-gilbert.
Address correspondence to: Gregg H. Gilbert, DDS,
Claude Pepper Center for Research on Oral Health in Ag-
ing, PO Box 100416, Gainesville, FL 32610-0416; e-mail:
gilbert@nervm.nerdc.ufl.edu.
Received May 23, 1997; initial review completed July
23, 1997; final acceptance December 8, 1997.
Vol. 36, No. 7
status measurement. Despite this lag, some infor-
mation on the multiple dimensions of oral health
is emerging, and it suggests that self-reported
measures are more informative of how oral dis-
ease affects the daily lives of individuals and
populations than are clinically derived meas-
ures.5-10
We propose a model of oral health (shown in
Fig. 1) that was conceptualized by adapting, with
revision, the work of two groups of investigators:
(1) Locker,11who adapted to the oral health con-
text a model by the World Health Organization,12
and (2) Johnson and Wolinsky,13who modified a
model by Nagi,14 predominately by adding self-
rated health as a construct. The model in Figure 1
posits a sequential causal process that involves
specific antecedents and consequents and
strongly parallels the biomedical concept of the
natural history of disease.13
We tested the validity of our model of oral
health within the context of the "Florida Dental
Care Study" (FDCS), which identified a repre-
sentative community-based, diverse sample of
dentate (at least one remaining natural tooth)
adults in north Florida. Three objectives of the
FDCS specific to the current report were to: (1)
describe self-rated oral health (SROH) in dentate
middle-aged and older adults; (2) quantify the as-
sociations between SROH and other measures of
oral health, namely, self-rated and clinically de-
rived measures of oral disease and tissue damage,
pain, functional limitation, and disadvantage; and
(3) assess the construct validity of our proposed
model of oral health.
Methods
Sampling Methods, Recruitment,Interview,
and Clinical Methods
The goal of the sampling design was to ensure
that a large number of persons at a hypothesized
increased risk for oral health decrements would
FIG.1. Multidimensional
conceptual model of oral
health specifying relations
between dimensionsof oral
health. Adapted with revi-
sion from Locker11 and
JohnsonandWolinsky13'
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MULTIDIMENSIONALITYOF ORAL HEALTH
be included (namely, black adults, residents of ru-
ral areas, persons 45 years old or older, and the
poor, who were defined as having income below
the US poverty level).15Details of sampling meth-
odology and selection were provided in an earlier
publication.16The 873 subjects who participated
at baseline resulted in a sample of only modest
bias within the population of interest.16Also, this
sample had a history of dental care similar to re-
cent National Health Interview Survey (NHIS)
data, and conclusions drawn from the FDCS and
the NHIS about determinants of dental care re-
cency were the same.16 Subjects participated for a
baseline in-person interview that immediately
was followed by a clinical dental examination. The
actual wording of questionnaire items is available
from the authors on request. We have provided
methodological details about the interview and
examination in earlier publications.16-22Briefly,
interexaminer reliability for surface-specific
crown caries was kappa = 0.89, for root caries it
was kappa = 0.62, and for periodontal attachment
loss + 2 mm, 97% of pairs were concordant. Test-
retest reliability for questionnaire items ranged
from 77% to 100%, expressed as percentage of
pairs concordant.
Operationalizingthe Proposed
Model of Oral Health
The in-person interview and clinical examina-
tion were used to obtain information about the
broad range of oral health measures (Table1) that
is suggested by the model in Figure 1. Data on
current oral disease and tissue damage were gath-
ered by direct clinical examination and by selected
self-reported items that queried whether the sub-
ject had the disease and tissue damage at the time
of the interview and, for those who answered af-
firmatively,the duration of that disease and tissue
damage. Oral pain and discomfort were measured
by asking subjects to report the presence and du-
989
 Dental Care Stu
TABLE1. Oral Health Status Measures in the Florida

Current Oral Pain and Oral Functional

Limitation Oral Disadvantage
Discomfort
CurrentDisease and Tissue Damage
Difficulty speakin or Due to disease and ti
Measure by clinical
d examination Toothache pain icutv spea bcag Avoided laughing o
Measuredby clinicalexaminationteeth Toothache pain Avoidedlaughingo
Dental sensitivity pronouncing because
Presence,location of remaining Denture soreness of mouth (ever and Avoidedtalkingbe
Dental decay (caries)and restorations Been embarrassed
(fillingsrecent) mouth
Current chewing difficulty
Fracturesof restorations
Dental cusp/incisaledge fractures Due to pain
Severeroot surfacedefects Troublesleeping b
mouth pain
Severetooth mobility
Mouth pain/disco
Periodontalattachmentloss
doing normald
Self-reportedmeasures Due to function
Teeththat are stained or look bad Due to function
Badbreath Avoidedchewing
Bad breath
with food catchingbetween because of mo
Trouble
Been preventedfr
teeth or dentures(everand recent)
because of mo
Abscessedtooth Avoidedeating w
Brokendenture because of ch
Infectedor sore gums
Bleedinggums
Has and wears maxillaryfull denture
Has and wears maxillarypartial
denture
Has and wears mandibularpartial
denture
Drymouth
months.
recent, presence within the previous 6
ever, presence within one's adult lifetime;

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
Vol. 36, No. 7
ration of toothache pain, root sensitivity to
hot/cold changes, and denture soreness. Oral
functional limitation was measured by asking
subjects about current chewing ability and about
difficulty speaking or pronouncing words because
of problems with their teeth, mouths, or den-
tures.22 Oral disadvantage was measured with
eight questions that asked subjects to report on
whether mouth problems caused them to avoid
certain activities and the frequency of that disad-
vantage within the previous 6 months.21'23Self-
rated oral health was measured using three ques-
tions that queried self-rated oral, dental, and
periodontal health. Oral health was measured by
asking, "Compared to others your age, how would
you rate the health of your mouth overall? Would
you say the health of your mouth overall is...?,"
with these response categories: excellent, very
good, good, fair, or poor. Dental health was meas-
ured by asking "...the present condition of your
teeth...,"and periodontal health was measured by
asking "...the present condition of your gums..."
The conceptual basis and response categories
were taken from the relatively large literature on
self-rated general health and adapted to the oral
health context.
Our analysis of cross-sectional data in this re-
port relied on the proposed temporal ordering of
antecedents and consequents to make inferences
about the validity of the proposed model and re-
lied on construct validity because of the lack of a
single criterion or "gold standard."Construct va-
lidity is defined as the extent to which a particular
measure correlates with other measures in a man-
ner consistent with theoretically derived hy-
potheses. We used oral diseases and conditions
(measured by direct clinical examination) and
questions (self-reported oral conditions and be-
haviors) that should theoretically be correlated
with SROH.
McDowell and Newell4 have critiqued publica-
tions of health status measurements that have re-
lied on the construct validation method and em-
phasized the importance of specifying not only
the direction of correlations expected, but also the
expected strength of the correlations, as well as
the rationale for these expectations. Concurring
with their recommendations, we expected that
measures in each of the four constructs related to
SROH (disease and tissue damage, pain, func-
tional limitation, and disadvantage) would be as-
sociated with SROH at the bivariate level. We
theorized that persons use oral health-related
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MULTIDIMENSIONALITYOF ORAL HEALTH
cues at the disease and tissue damage, pain, func-
tion, and disadvantage levels to make an overall
assessment of their SROH; thus, our expectation
was that the magnitude of any single correlation
between SROH and any single explanatory corre-
late would be moderate (0.4 < r < 0.7) or small (r
< 0.4), depending on the measure. These statisti-
cally significant correlations would provide sup-
porting evidence for the construct validity of the
measurement battery as a whole and would pro-
vide evidence on an empirical basis that the
SROH construct and the model proposed in Fig-
ure 1 are valid.
Disease and tissue damage were measured by
direct clinical examination. It was expected that
subjects would rate their oral health lower if they
had fewer natural teeth, more decayed tooth sur-
faces, more root fragments, more bulk fractures of
restorations (fillings), more fractures of the tooth's
cusp and/or incisal edge, more teeth that had se-
vere root defects, more teeth that were severely
mobile, and more teeth that had severe periodon-
tal attachment loss.
Effect magnitudes were anticipated at the bi-
variate level: [number of remaining teeth = root
fragments = decayed tooth surfaces = severely
mobile teeth] > [teeth with severe attachment
loss -tooth fractures] > [bulk fractures of resto-
rations = severe root defects], ranging from ap-
proximately 0.5 to 0.4 for the first group of vari-
ables to 0.1 for the last of the three groups. The
rationale for hypothesizing these effect magni-
tudes requires some explanation. The presence of
teeth, root fragments, large decayed areas of the
teeth, and severely loose teeth were expected to
be obvious to most subjects and therefore more
highly correlated with SROH than less obvious
types of clinical disease and tissue damage. Teeth
with severe attachment loss may not have been ob-
vious to subjects unless that attachment loss was
due to obvious gum recession; the instances where
the attachment loss was due to deep pockets around
the teeth would not have been obvious. Although
tooth fracturesmay have been obvious in some in-
stances, our diagnostic criterionof 2 mm was not as
severe as that used for some of the other clinical
measures. Bulkrestorationfracturesare typicallyless
noticeable than tooth fracturesexceeding 2 mm, and
we therefore expected a lower correlationbetween
bulk fractures and SROH. Additionally,tooth frac-
tures and bulk restoration fractures are often the
long-term sequelae of successively largerdental res-
torations, restorations that more often are de-
991
GILBERT ET AL
manded by those with a stronger interest in main-
taining their oral health (thereby being associated
with a higher rating for SROH, not a lower one).
Lastly, although a 2-mm criterion for root defects
is a severe criterion, these defects often are caused
by excessively vigorous toothbrushing, a behavior
that actually should be associated with higher
SROH, not lower SROH.
The expectation at the multivariate level was
that self-reported measures of disease and tissue
damage would predominate over disease and tis-
sue damage determined clinically for two reasons:
(1) subjects would be less likely than dentists to
notice (and therefore report) clinical signs that are
less severe, and (2) self-reported oral health dec-
rements should be more strongly associated with
self-rated oral health than would decrements re-
ported by dentists. Our expectation was that,
among those who had a full and/or partial den-
ture, the ones who actually wore that denture(s)
would report better SROH because they have bet-
ter chewing ability. We expected that the magni-
tude of the difference in SROH would be greatest
for those who have and wear a full denture as
compared with those who wear a partial denture
because the full denture wearers presumably
would experience the greatest improvement in
chewing ability.
We expected that oral pain and discomfortwould
be associated with poorer SROH. Toothache pain,
tooth sensitivity to hot/cold, and tooth abscess were
expected to be more strongly correlatedwith SROH
than having a sore denture (oralpain) and/or a bro-
ken denture because the health of living tissue
should be rated as more important than the condi-
tion of a prosthetic device. Oral functional limita-
tion, reported as chewing difficultyor speaking dif-
ficulty, was expected to be associated with poorer
SROH, and the strength of these two associations
was expected to be approximatelyequal. The num-
ber of decrements due to oral disadvantage was ex-
pected to be associated with poorer SROH. Because
oral disadvantage measures impact on daily activi-
ties, we expected a strong correlationbetween oral
disadvantage and SROH.
Although not specified in Figure 1, we also hy-
pothesized that exogenous factors affect SROH,
such as attitudes toward dental health/dental care
and sociodemographic factors. These exogenous
factors are not specified in Figure 1 for simplicity
of graphical presentation. In this report, we spe-
cifically tested these exogenous factors: dental at-
titudes, age group, sex, race, poverty status, area of
992
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MEDICAL CARE
residence, and level of formal education. The item
content and psychometric properties of the atti-
tudes scale have been described elsewhere.19 The
dental attitudes scale originally was included in
the FDCS to determine the effects of attitudes on
the use of dental care. Therefore, this amounts to
a post hoc analysis with respect to the original in-
tent of the FDCS. Nonetheless, there is a strong
theoretical basis for expecting that health atti-
tudes have an important influence on self-ratings
of health independent of clinically determined
ratings.24-27Consequently, we hypothesized that
positive dental attitudes would be associated with
higher ratings of SROH.
Statistical Methods
Results were weighted using methods de-
scribed previously.16 Statistical analyses were
done using SAS.28,29 Multicollinearity was
measured using a procedure described by Bel-
sley et al.30 Several explanatory covariates in-
troduced problems with multicollinearity. Only
when "sore denture" and "broken denture" were
combined into one variable, when "toothache
pain,""abscessed tooth,"and "dental sensitivity to
hot/cold"were combined into one variable, when
"infected or sore gums," "bleeding gums," and
"loose tooth" were combined into one variable,
and only when the eight measures of oral disad-
vantage were combined into one variable, was
the problem with multicollinearity eliminated.
These combined variables were used in the re-
gression in Table 5 (for exact coding of these
variables, please see the legend in Table 5).
Some recoded variables combined measures of
disease/tissue damage with measures of pain;
this was necessary for the stated empiric rea-
sons, not for theoretical reasons.
Results
For self-rating of the health of the mouth over-
all, 20% of subjects said excellent, 31% said very
good, 29% said good, 13% said fair, and 8% said
poor. When responses to the three self-ratings
(overall, dental, and periodontal self-rated health)
were added to form the composite variable called
SROH, the weighted mean + SD was 10.0 + 3.3,
with a possible and observed range of 3 to 15. Five
percent of persons had a score of 3 ("poor" for
each of the three measures), 10% had a score of
15 ("excellent" for each of the three measures),
Vol. 36, No. 7 MULTIDIMENSIONALITY OF ORAL HEALTH

and 17% had the modal score of 9. The distribu-
tion of SROH approximated a normal distribution
(skewness = -0.14; kurtosis = -0.74; Shapiro-Wilk
W = 0.94). A common factor analysis (FACTOR
procedure, SAS28) of the three SROH measures
was done and suggested a single factor that ac-
counted for 84% of the total variance; factor load-
ings for these three measures ranged from 0.88 to
0.94, with Cronbach's a internal consistency reli-
ability of 0.90, providing good empiric justifica-
tion for our decision to model SROH as a single
outcome.31
Assessment of Construct Validity
Table 2 presents correlations between SROH
and aspects of oral health that were quantified
on an ordinal, interval, or ratio scale. With one
exception (root defects), the directions of effect
were as expected. None of the magnitudes of
these correlations were large, and the correlation
between SROH and oral disadvantage was the
largest. The numbers of remaining teeth, decayed
tooth surfaces, and severely mobile teeth were
the most highly correlated with SROH of the
clinical measures (r > 0.3), although our expecta-
tion about root fragments was not met. Although
the expectation that bulk fractures and root de-
fects would have the lowest correlations was met,
the correlation between SROH and cusp frac-
tures was lower than expected, perhaps because
the dentists'2-mm criterion was not sufficiently
severe to elicit a lower SROH.
Table 3 shows the mean SROH values for sub-
jects with oral pain and self-reported oral dis-
ease/tissue damage when queried at the nominal
level. All the differences in mean SROH values
were in the expected directions. All the mean
SROH values between those with and those
without the condition were significantly differ-
ent, with the exception of three of the prosthetic
variables (broken denture, wearing an upper or
lower partial denture). As expected, the effect of
wearing the maxillary full denture was greater
than the effect due to wearing either a maxillary
or mandibular partial denture.
We also evaluated the relative validity of the
three individual measures of self-rated oral
health by observing how the probability of hav-
ing a decrement in each of the three measures
varied with decrements in the other aspects of
oral health (results not shown, but tables avail-
able from the authors on request). As an exam-
TABLE 2. Weighted Pearson Correlations
Between SROH and Other Measures of Oral
Health and Dental Attitudes That Were
Quantified on an Ordinal, Interval,or Ratio Scale*
Pearson r
Oral disadvantage
Oral disadvantage (8-40 scale) -0.53t
Oral functional limitation
Number of chewing difficulties -0.43t
(0-5 scale)
-0.31t
Speaking difficulty (1-5 scale)
Clinical disease and tissue damage
Number of teeth remaining 0.33t
Number of decayed tooth surfaces -0.31t
Number of severely mobile teeth -0.30t
Number of teeth with severe LOA -0.29t
Number of root fragments -0.23t
Number of teeth with severe root 0.04 (NS)
defects
Number of teeth with cusp fractures -0.03 (NS)
Number of teeth with bulk fractures -0.02 (NS)
Dental attitudes
Cost has influenced treatment received 0.46t
Quality of dental care 0.32t
Eventuality of dental decline 0.28t
Importance of dental care to prevent 0.18t
dental problems
Personal influence on oral health 0.18'
Effectiveness of dental care 0.17t
Cynicism toward dentists 0.12t
NS, not statistically significant; LOA, loss of
periodontal attachment.
*The outcome of interest was self-rated oral health
(SROH), which was the sum of scores of three
variables: (1) self-rated health of the mouth overall, (2)
self-rated dental health and (3) self-rated periodontal
health. Each variable was coded as 1 = poor; 2 = fair; 3
= good; 4 = very good; 5 = excellent. SROH ranged
from 3 to 15.
tp < 0.05.
ple, although the correlation between SROH and
root fragments in Table2 was only -0.23, the prob-
ability of reporting fair or poor self-rated dental
health increased from 23% with no root frag-
ments to 37% with one root fragment and to 74%
with two or more root fragments. The probability
of reporting fair or poor dental health was higher
for root fragments, a strictly "dental"health prob-
lem, than it was for periodontal health. A similar

993

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
GILBERT ET AL MEDICAL CARE

TABLE3. Weighted Mean (SD) SROH, by the Measures of

Oral Health That Were Quantified on a Nominal Scale*

SROHScores

Yes,Has the No, Does Not Difference

Condition Have the Condition in Means

Oral pain and discomfort

Has toothache pain 7.9 (2.8) 10.2 (3.2) -2.3t
Has dental sensitivity to hot/cold 8.9 (3.2) 10.5 (3.2) -1.6t
Has a sore denturet 7.9 (2.5) 9.3 (2.9) -1.4t
Self-reported oral disease and tissue damage
Has teeth that are stained or look bad 8.2 (2.9) 11.2 (2.9) -3.0t
Has cavities now 7.4 (2.9) 10.8 (3.0) -3.4t
Has a loose tooth 7.1 (2.8) 10.5 (3.1) -3.4t
Has bleeding gums 7.1 (2.8) 10.4 (3.1) -3.3t
Has a broken tooth or cap 7.8 (2.9) 10.5 (3.2) -2.7t
Has infected/sore gums 7.3 (2.6) 10.4 (3.2) -3.1t
Has bad breath problem 8.2 (3.2) 10.5 (3.1) -2.3t
Has a broken filling 8.2 (3.2) 10.3 (3.2) -2.1t
Has an abscessed tooth 6.8 (2.8) 10.1 (3.3) -3.3t
Has dry mouth sometimes 9.2 (3.2) 10.2 (3.3) -1.0t
Has a broken denturet 8.3 (2.8) 9.2 (2.9) -0.9 (NS)
Has and wears maxillary full denture? 8.8 (2.7) 6.1 (2.3) 2.7t
Has and wears maxillary partial denturell 9.5 (3.7) 8.9 (2.9) 0.6 (NS)
Has and wears mandibular partial denture[ 9.6 (3.4) 9.4 (2.9) 0.2 (NS)

SD, standard deviation, NS, not statistically significant.

*The outcome of interest was self-rated oral health (SROH), which was the sum of scores of three variables: (1)
self-rated health of the mouth overall, (2) self-rated dental health and (3) self-rated periodontal health. Each
variable was coded as 1 = poor; 2 = fair; 3 = good; 4 = very good; 5 = excellent. SROH ranged from 3 to 15.
tp < 0.05; differences between means tested using Wilcoxon rank sum test.
tOnly includes persons who reported having and currently wearing a removable denture; n = 211.
?Only includes persons who upon clinical examination were edentulous in the maxilla; n = 89.
IlOnlyincludes persons who reported ever having a maxillary partial denture and who upon clinical examination
had 1-15 teeth in the maxillary arch; n = 143.
9Only includes persons who reported ever having a mandibular partial denture and who upon clinical
examination had 1-15 teeth in the mandibular arch; n = 186.

circumstance existed for other strictly dental MultivariateFindings

measures of clinical disease and tissue damage,
such as decayed tooth surfaces, cusp fractures,
and bulk fractures of fillings.
Table 4 shows that approach to dental care,
race and poverty status, and level of formal edu-
cation were significantly associated with SROH,
consistent with oral disease being more preva-
lent in certain of these groups.17-20The amount
of oral disease was similar between rural and ur-
ban residents, so their similarity in SROH was
not surprising. Middle-aged adults and women
reported worse SROH.
A linear multiple regression was done to iden-
tify independent determinants of SROH using
the variables in Table 1. Table 5 shows results
from a single (nonstepwise) linear regression of
SROH. With the exception of oral pain, each con-
struct in Figure 1 had a statistically significant di-
rect effect on SROH. With the other conditions
taken into account, having oral disadvantage,
speaking difficulty, six measures of self-reported
oral disease and tissue damage, and having fewer
teeth remaining were significantly associated

994

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
Vol. 36, No. 7 MULTIDIMENSIONALITY OF ORAL HEALTH

TABLE 4. Weighted Mean (SD) SROH for the Sample Overall, by

Approach to Dental Care, and Selected Sociodemographic Characteristics*
Mean (SD) Difference
Characteristic(n) SROH Score in Means

All subjects (873) 10.0 (3.3)

Approach to dental care
Never go or go only when have a problem (400) 8.5 (2.9)
Go to a dentist whether or not have a problem 11.3 (3.1) 2.8t
(474)
Age group
45-54 years ago (269) 9.3 (3.8)
55-64 years old (243) 9.8 (3.6)
65-74 years old (257) 10.6 (2.9)
75 years old or older (104) 10.8 (2.3) 1.5t
Sex
Female (491) 9.7 (2.9)
Male (383) 10.4 (3.9) 0.7t
Race and poverty status
Poor black (75) 8.5 (2.1)
Poor white (57) 8.6 (2.8)
Nonpoor black (135) 9.3 (3.4)
Nonpoor white (553) 10.5 (4.0) 2.0t
Missing (53)
Ability to pay an unexpected $500 dental bill
Not able to pay the bill (122) 7.5 (2.7)
Able to pay, but with difficulty (342) 9.5 (3.0)
Able to pay comfortably (406) 11.1 (3.3) 3.6t
Missing (3)
Present financial situation
I really can't make ends meet (36) 7.4 (3.5)
I manage to get by (386) 8.9 (2.8)
I have enough to manage, plus some extra (306) 11.0 (3.4)
Money is not much of a problem; I can buy about 11.4 (3.3) 4.0t
whatever I want (138)
Missing (8)
Area of residence
Rural (436) 9.9 (3.3)
Urban (437) 10.1 (3.3) 0.2 (NS)
Highest level of formal education
Did not graduate from high school (184) 8.5 (2.7)
Graduated from high school (689) 10.4 (3.4) 1.9t
Missing (1)

SD, standard deviation; NS, not statistically significant.

*The outcome of interest was self-rated oral health (SROH), which was the sum
of scores of three variables: (1) self-rated health of the mouth overall, (2) self-rated
dental health, and (3) self-rated periodontal health. Each variable was coded as 1
= poor; 2 = fair; 3 = good; 4 = very good; 5 = excellent. SROH range from 3 to 15.
Some sample sizes do not add to 873 persons due to weighted rounding.
tP < 0.05; differences between means tested using Wilcoxon rank sum test or
Kruskal-Wallis test.

995

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
GILBERTET AL
with poorer SROH. With the oral health measures
taken into account, older persons and persons
with more positive dental attitudes were more
likely to report better SROH. The majority of the
variance in SROH was accounted for by the
model.
Table 6 shows how the percent variance of
SROH varied with successive, stepwise inclusion
of each oral health construct, as well as dental at-
titudes and sociodemographic variables. Oral dis-
advantage alone accounted for more than one
fourth of the variance in SROH. Oral pain and
oral functional limitation successively added
small percentages to the R2 value. Self-reported
disease and tissue damage added a substantive
portion to the R2value. Although most of the den-
tal attitudes variables and one of the sociode-
mographic variables were statistically significant,
each construct improved the R2 value by only a
small amount when the oral health constructs
were taken into account.
Discussion
Our factor analyses suggest that SROH is best
described as one factor. This suggests either that
subjects do not differentiate oral health and den-
tal health from periodontal health to any great ex-
tent, or that the same persons have the conditions
affecting self-rated oral, dental, and periodontal
health. The latter explanation can be diminished
largely by an analysis of the persons who have
dental decay, severe attachment loss, and multiple
missing teeth. Although the correlations between
dental decay, severe attachment loss, and number
of remaining teeth were statistically significant,
they were modest in strength (Spearman's r <
0.25), and only 8% of the subjects met these three
criteria:(1) had at least one tooth surface with de-
cay; (2) had at least one tooth with severe attach-
ment loss; and (3) had 16 or fewer teeth.
The bivariate and multivariate findings pro-
vided consistent evidence of the construct va-
lidity of the proposed model of oral health. It is
clear from this report and from our previous re-
ports that subjects with oral disease and tissue
damage did have poorer SROH and exhibited
higher prevalence of pain, functional limitation,
and disadvantage.21'22 Subjects with multiple
diseases had the greatest likelihood of decre-
ments in SROH. Additionally, these measures
discriminated between groups with differing
extents of disease within single diagnostic
996
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MEDICAL CARE
groups. It should be noted that the measures of
pain, functional limitation, disadvantage, and al-
most all of the disease/tissue damage measures
used in the FDCS are measures of extent and du-
ration, not severity. Because the correlations be-
tween constructs may have been larger if meas-
ures of severity were used, we recommend that
further refinement include gradations of severity.
Although the correlations among constructs in
the model provided consistent evidence of con-
struct validity, their lack of strong correlation sug-
gested that these constructs represent distinct
and complementary dimensions of oral health.32
This has been a common finding in the nonoral
health literature, in which the correlation be-
tween symptoms (reported by subjects) and dis-
ease severity (measured by clinicians and diag-
nostic tests) has been weak or modest.26 Our
model of oral health implies that oral health is
multidimensional. Much understanding of oral
health is lost if measurement is limited only to
conditions that can be examined clinically by a
dentist. Instead, a more comprehensive under-
standing is garnered by measuring subjects' re-
ports of how these conditions affect their daily
lives, the improvement of which is, after all, the
predominant reason for the existence of dental
care.
Unlike other recent reports, our measurement
battery of oral health is not disease-specific, condi-
tion-specific, or treatment-specific.33'34Rather, like
some others, it is organ-specificand specific to a sin-
gle organ system.35This offers the advantage of be-
ing more comprehensive when assessing dental
care at large,which is an organ-specificsubset of the
health care system. Our organ-specific battery is
composed of disease-specific and condition-specific
components, which offers the potential for disag-
gregation when assessing disease-specific and con-
dition-specific dental services (eg, periodontal sur-
gery or root canal therapy).
Understanding which factors are inde-
pendently associated with SROH helps deter-
mine what aspects of oral health are important to
patients as compared with those important to the
dentists who provide care. As we hypothesized,
oral disadvantage was strongly associated with
SROH. Combined with the other results, this sug-
gests that persons use multidimensional cues,
with direct effects from each of the dimensions of
oral health, with the exception of oral pain. Oral
pain had a direct effect on oral disadvantage, but
an indirect effect on SROH.21 Also, because of
Vol. 36, No. 7 MULTIDIMENSIONALITYOF ORAL HEALTH

5. Weighted Linear Regression of Self-Rated Oral Health*

TABLE
Parameter Standard
Covariate Estimate Error P

Intercept -31.4 1.79 0.01

Disadvantage
Disadvantage -0.12t 0.03t 0.01t
Functionallimitation
Difficultyspeaking -0.32t 0.12t 0.01t
Chewingdifficulty -0.18 0.29 0.55
Painand discomfort
Sore/brokendenture 0.42 0.34 0.22
Painfultooth 0.03 0.10 0.74
Self-reporteddiseaseandtissue
damage
Stainedteeth -1.10t 0.20t 0.01t
Gumproblems -0.94t 0.17t 0.01t
Broken filling -0.83t 0.29t 0.01t
Cavities -0.76t 0.27t 0.01t
Broken tooth -0.63t 0.27t 0.02t
Foodcatching -0.17t 0.08t 0.04t
Badbreath -0.15 0.25 0.56
Drymouth -0.24 0.23 0.29
Maxillary fulldenture 0.67 0.45 0.14
Maxillary partialdenture -0.33 0.34 0.33
Mandibular partialdenture -0.52 0.31 0.10
Clinicaldiseaseandtissue
damage
Numberof teeth 0.43t 0.15t 0.01t
Carioustoothsurfaces -0.01 0.04 0.70
Rootfragments -0.10 0.19 0.62
Bulkfractures 0.17 0.20 0.40
Cuspfractures 0.02 0.21 0.91
Severerootdefects 0.28 0.18 0.12
Severelymobileteeth -0.19 0.14 0.18
Severeattachmentloss -0.03 0.06 0.64
Dentalattitudes
Eventuality 0.41t 0.18t 0.02t
Quality 0.36t 0.11t 0.01t
Cost 0.22t 0.09t 0.02t
Importance 0.21t 0.10t 0.04t
Cynicism -0.12 0.13 0.38
Sociodemographic
Agegroup 0.74t 0.19t .01t
Sex -0.33 0.19 0.07
Race 0.24 0.24 0.33
Povertystatus 0.53 0.27 0.05
Areaof residence 0.15 0.19 0.41
Education 0.21 0.27 0.43

(Continues)

problems with multicollinearity, measures of Other studies have found related results, al-
"gum problems" had to be combined, one of though, in our judgment, the FDCS had the most
which ("infected or sore gums") may have meas- comprehensive array of clinical and self-reported
ured pain. measures. Matthias et a136observed a correlation

997

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
GILBERT ET AL
TABLE5.
Coding of explanatory covariates:
Disadvantage: sum of scores of 8 variables that
measured avoidance of certain activities because of
mouth/dental problems: (1) avoided laughing/smiling;
(2) avoided talking; (3) been embarrassed; (4) trouble
sleeping; (5) mouth pain/discomfort kept from doing
normal daily activities; (5) avoided chewing hard things;
(6) been prevented from eating foods would like to eat;
(8) avoided eating with others. Each variable coded as: 5
= very often within previous 6 mo; 4 = fairly often within
previous 6 mo; 3 = sometimes within previous 6 mo; 2 =
at least once in adult lifetime, but never within previous
6 mo; 1 = never in adult lifetime. The oral disadvantage
variable ranged from 8 to 40.
Difficulty speaking: frequency that had difficulty
speaking or pronouncing any words because of
problems with teeth, mouth, or dentures; 5 = very
often in previous 6 mo; 4 = fairly often in previous 6
mo; 3 = sometimes in previous 6 mo; 2 = never in
previous 6 mo; 1 = never have had difficulty.
Chewing difficulty: 0 = does not have one or more
current chewing difficulties; 1 = does.
Sore/broken denture: number of these problems
currently has: (i) broken denture; (ii) sore denture. 3 =
both; 2 = 1 of the 2; 1 = neither of the 2. Note that this
recoded variable combines a measure of disease and
tissue damage with a measure of pain.
Painful tooth: number of these problems currentlyhas:
(i) toothache or painful tooth; (ii) abscessed tooth; (iii)
tooth with sensitivity to hot/cold drinksor sweets. 3 = all 3;
2 = 2 of the 3; 1 = 1 of the 3; 0 = none of the 3. Note that
this recoded variable combines a measure of disease and
tissue damage with two measures of dental pain.
Stained teeth: 0 = does not report having teeth that
are stained or look bad; 1 = does.
Gum problems: sum of scores of three variables: (1)
Do you have infected or sore gums?; (2) Do you have
bleeding gums?; (3) Do you have a loose tooth? Each
variable was coded as 0 = No; 1 = Yes.
Broken filling: response to "Do you have a broken
filling?"0 = No; 1 = Yes.
Cavities: response to "Do you have any cavities?"0 =
No; 1 = Yes.
Broken tooth: response to "Do you have a broken
tooth or a broken cap?"0 = No; 1 = Yes.
Food catching: frequency in previous 6 mo that had
trouble with food catching in teeth or dentures; 4 =
very often; 3 = fairly often; 2 = sometimes; 1 = never.
Bad breath: 0 = does not report having a problem
with bad breath; 1 = does.
Dry mouth: 0 = does not report having a dry mouth;
1 = does.
*The outcome of interest was self-rated oral health (SROH), which was the sum of scores of three variables. (1)
self-rated health of the mouth overall, (2) self-rated dental health, and (3) self-rated periodontal health. Each variable
was coded as 1 = poor; 2 = fair; 3 = good; 4 = very good; 5 = excellent. SROH ranged from 3 to 15, with higher scores
connoting better health. n = 680; model F value 22.3; 35 df; P < 0.0001; R2 = 52% (adjusted for the number of variables
included in the regression.
tStatistically significant.
998
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MEDICALCARE
(Continued)
Maxillary full denture: 0 = does not have and wear a
maxillary full denture; 1 = does.
Maxillary partial denture: 0 = does not have and
wear a maxillary partial denture; 1 = does.
Mandibular partial denture: 0 = does not have and
wear a mandibular partial denture; 1 = does.
Number of teeth: number of natural teeth remaining
upon clinical examination; 0 = 1-8 teeth; 1 = 9-16
teeth; 3 = 17-24 teeth; 4 = 25-32 teeth.
Carious tooth surfaces: number of remaining coronal
and root tooth surfaces that are carious upon clinical
examination.
Root fragments; number of remaining teeth that are
root fragments upon clinical examination.
Bulk fractures: number of remaining teeth that have
a bulk restoration fracture upon clinical examination.
Cusp fractures: number of remaining teeth that have
a cusp or incisal edge fracture upon clinical
examination.
Severe root defects: number of remaining teeth that
have a severe root defect upon clinical examination.
Severely mobile teeth: number of remaining teeth
that are severely mobile upon clinical examination.
Severe attachment loss: number of remaining teeth
that have attachment loss of 7 mm or more (relative to
the cementoenamel junction) upon clinical
examination.
Eventuality: mean response to "eventuality of dental
decline" scale.
Quality: mean response to "quality of recent dental
care"scale.
Cost: mean response to "influence of dental care
costs on past dental treatment"scale.
Importance: mean response to "importance of dental
visits to prevent dental problems" scale.
Cynicism: mean response to "cynicism toward
dentists and dental care"scale.
Age group: 0 = 45-64 years old; 1 = 65 years old or
older.
Sex: 0 = male; 1 = female.
Race: 0 = white; 1 = black.
Poverty status: 0 = at or above U.S. poverty threshold;
1 = below poverty threshold. Subjects with missing data
(n = 137) were assigned values randomly within race and
sex groupings. Any misclassification would lessen the
apparent effect of the variable on SROH.
Area of residence: 0 = rural; 1 = urban.
Education: highest level of formal education
attained; 0 = did not graduate high school; 1 = did
graduate high school.
Vol. 36, No. 7 MULTIDIMENSIONALITYOF ORAL HEALTH

TABLE6. Summary of Variancein SROH Accounted for Using Linear Regression, by Stepwise Addition of
VariousOral Health Constructs and Modifying Variables

(P)That
Probability
R2for Overall Model* FValueof Variable(s) Additional Parameter(s)
Construct(s) Included in Model (%) Added Jointly = 0

Oral disadvantage 27.8

Oraldisadvantage
Oralfunctional limitationt 30.3 12.4 <0.001

Oraldisadvantage
Oral functional limitation
Oral pain and discomfortt 32.0 9.6 <0.001

Oral disadvantage
Oral functional limitation
Oral pain and discomfort
Self-reportedoral disease/tissuedamaget 48.3 28.9 <0.001

Oral disadvantage
Oral functional limitation
Oral pain and discomfort
Self-reported oral disease/tissue damage
Clinical oral disease/tissuedamaget 49.5 2.9 <0.005

Oral disadvantage
Oral functional limitation
Oral pain and discomfort
Self-reported oral disease/tissue damage
Clinical oral disease/tissue damage
Dental attitudest 51.5 7.0 <0.001

Oral disadvantage
Oralfunctionallimitation
Oral pain and discomfort
Self-reported oral disease/tissue damage
Clinical oral disease/tissue damage
Dental attitudes
Sociodemographict 52.3 3.8 <0.001

*The R2 value is adjusted for the number of variables included in the regression.
'Variables were the ones added at that step.

between self-rated dental health and missing
teeth that was nearly identical to that found in the
FDCS. A study of subjects in the Rand Health In-
surance Experiment observed that each of four
clinical measures was significantly associated
with the "subjective dental health index," which
comprised three questions on pain, worrying
about the teeth, and conversation avoidance.37
Locker and Slade32reported results using the Oral
Health Impact Profile (OHIP), a weighted index
that includes, among other constructs, oral pain,
functional limitation, and oral disadvantage. They
found that missing teeth was the only important
clinical variable correlated with the Oral Health
Impact Profile in a multivariate model.
Our results from the FDCS are the first that we
know of that specifically link dental health atti-
tudes with SROH. Although the magnitude of the

999

This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
GILBERT ET AL
association was small, it suggests that even with a
comprehensive accounting of oral diseases, con-
ditions, and symptoms, having more negative
dental attitudes predisposes one to lower SROH.
An additional explanation may be that decre-
ments in SROH preceded a decline in dental atti-
tudes.38 Unfortunately, conclusions from our
cross-sectional data would be speculative.
Our finding that age group was the only so-
ciodemographic factor independently associated
with SROH, with the other covariates taken into
account, has been addressed to a limited extent in
other studies.9,32'36'39
In the FDCS, black subjects,
poor persons, and women reported worse SROH,
but with oral health conditions taken into ac-
count, no significant differences were evident. The
finding about age group suggests that evaluations
of dental care effectiveness should be age group-
specific.
The salience of the "teeth that are stained or
look bad"variable in the multivariate model in Ta-
ble 5 suggests that subjects use not only health
cues to rate their SROH, but also esthetic cues. Al-
though dental and oral health clearly can affect
one's medical/general health (for a review, see
Hollister and Weintraub40),the importance that
the face and mouth have on interpersonal rela-
tions may motivate subjects to include dental ap-
pearance as a subset of dental "health."Therefore,
this may be an additional reason to structure den-
tal/oral health as a separate component of health.
For example, just as "general health" has been
separated into the "physical"and the "mental,"an
additional separation to include the "oral"seems
warranted.l
Acknowledgments
The authorsthankD. E.Antonson,U. Foerster,D. W.
Legler,and M. L. Ringelberg,who, in additionto T. A.
Dolan and G. H. Gilbert,servedas clinicalexaminersfor
the study.The authorsalso thank J. L. Earlsand B. A.
Ringelberg,who servedas regionalcoordinatorsfor the
baselinephase of the study.
References
1. Ware JE Jr. The status of health assessment
1994.Ann RevPublicHealth1995;16:327.
2. Patrick DL, Deyo RA. Genericand disease-
specificmeasuresin assessinghealthstatusand quality
of life.Med Care1989;27:S217.
1000
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MEDICAL
CARE
3. Ware JE Jr, Kosinski M, Bayliss MS, McHor-
ney CA, Rogers WH, Raczek A. Comparison of meth-
ods for the scoring and statistical analysis of SF-36 health
profile and summary measures: Summary of results from
the Medical Outcomes Study. Med Care 1995;33:AS264.
4. McDowell I, Newell C. Measuring health: A
guide to rating scales and questionnaires. 2nd edition.
New York:Oxford University Press, Inc., 1996.
5. Atchison KA, Dolan TA. Development of the
Geriatric Oral Health Assessment Index. J Dent Educ
1990;680.
6. Strauss RP, Hunt RJ. Understanding the value
of teeth to older adults: Influences on the quality of life. J
Am Dent Assoc 1993;124:105.
7. Locker D, Miller Y. Evaluation of subjective
oral health status indicators. J Public Health Dent
1994;54:167.
8. Slade GD, Spencer AJ. Development and
evaluation of the Oral Health Impact Profile. Commu-
nity Dent Health 1994;11:3.
9. Hunt RJ, Slade GD, Strauss RP. Differences
between racial groups in the impact of oral disorders
among older adults in North Carolina. J Public Health
Dent 1995;55:205.
10. Atchison KA, Gift HC. Perceived oral health in
a diverse sample. Adv Dent Res 1997;11:272.
11. Locker D. Measuring oral health: A conceptual
framework. Community Dent Health 1988;5:3.
12. World Health Organization. International
classification of impairments, disabilities, and handicaps.
Geneva, Switzerland: WHO, 1980.
13. Johnson RJ, Wolinsky FD. The structure of
health status among older adults: Disease, disability,
functional limitation, and perceived health. J Health So-
cial Behav 1993;34:105.
14. Nagi SZ. An epidemiology of disability among
adults in the United States. Milbank Q 1976;54:439.
15. US Bureau of the Census, Census of Population
and Housing, 1990. Public Use Microdata Samples US
Technical Documentation. Washington, DC: US Bureau
of the Census, 1992.
16. Gilbert GH, Duncan RP, Kulley AM, Coward
RT, Heft MW. Evaluation of bias and logistics in a sur-
vey of adults at increased risk for oral health decrements.
J Public Health Dent 1997;54:48.
17. Gilbert GH, Antonson DE, Mjor IA, et al.
Coronal caries, root fragments, and restoration and cusp
fracturesin US adults. Caries Res 1996;30:101.
18. Ringelberg ML, Gilbert GH, Antonson DE, et
al. Root caries and root defects in urban and rural
adults. J Am Dent Assoc 1996;127:885.
19. Gilbert GH, Duncan RP, Heft MW, Coward
RT. Dental health attitudes among dentate black and
white adults. Med Care 1997;35:255.
Vol. 36, No. 7
20. Dolan TA, Gilbert GH, Ringelberg ML, et al.
Behavioral risk indicators of attachment loss in adult Flo-
ridians. J Clin Periodontol 1997;24:223.
21. Gilbert GH, Duncan RP, Heft MW, Dolan TA,
Vogel WB. Oral disadvantage among dentate adults.
Community Dent Oral Epidemiol 1997. In press.
22. Foerster U, Gilbert GH, Duncan RP. Oral func-
tional limitation in dentate adults. Submitted manuscript.
23. Cushing AM, Sheiham A, Maizels J. Devel-
oping sociodental indicators: The social impact of dental
disease. Community Dent Health 1986;3:3.
24. Mechanic D. Social psychological factors af-
fecting the presentation of bodily complaints. N Engl J
Med 1972;286:1132.
25. Watson D, Pennebaker JW. Health com-
plaints, stress, and distress: Exploring the central role of
negative affectivity. Psychol Rev 1989;96:234.
26. Barsky AJ, Cleary PD, Klerman GL. Determi-
nants of perceived health status of medical outpatients.
Soc Sci Med 1992;34:1147.
27. Wilson IB, Cleary PD. Linking clinical vari-
ables with health-related quality of life: A conceptual
model of patient outcomes. JAMA 1995;273:59.
28. SAS Institute, Inc. SAS/STAT User's Guide:
Version 6. 4th ed. Cary,NC: SAS Institute, Inc., 1989.
29. SAS Institute, Inc. SAS/STAT Software:
Changes and enhancements for release 6.12. Cary,NC:
SAS Institute, Inc., 1996.
30. Belsley DA, Kuh E, Welsch RE. Regression di-
agnostics: Identifying influential data and sources of col-
linearity.New York:Wiley, 1980.
This content downloaded from 132.239.1.230 on Mon, 22 Feb 2016 00:46:03 UTC
All use subject to JSTOR Terms and Conditions
MULTIDIMENSIONALITY OF ORAL HEALTH
31. Cronbach LJ. Coefficient alpha and the internal
structure of tests. Psychometrika 1951;16:297.
32. Locker D, Slade G. Association between clini-
cal and subjective indicators of oral health status in an
older adult population. Gerodontol 1994;11:108.
33. Rothman ML, Revicki DA. Issues in the
measurement of health status in asthma research. Med
Care 1993;31:MS82.
34. Bombardier C, Melfi CA, Paul J, et al. Com-
parison of a generic and a disease-specific measure of
pain and physical function after knee replacement sur-
gery. Med Care 1995;AS131.
35. Mangione CM, Phillips RS, Seddon JM, et al.
Development of the Activities of Daily Vision scale: A
measure of visual functional status. Med Care
1992;30:1111.
36. Matthias RE, Atchison KA, Lubben JE, De-
Jong F, Schweitzer SO. Factors affecting self-ratings of
oral health. J Public Health Dent 1995;55:197.
37. Gooch BF, Dolan TA, Bourque LB. Correlates
of self-reported dental health status upon enrollment in
the Rand Health Insurance Experiment. J Dent Educ
1989;53:629.
38. Douglass CW, Cole KO. Utilization of dental
services in the United States. J Dent Educ 1979:43:223.
39. Reisine ST, Bailit HL. Clinical oral health
status and adult perceptions of oral health. Soc Sci Med
1980;14A:597.
40. Hollister MC, Weintraub JA. The associations
of oral status with systemic health, quality of life, and
economic productivity.J Dent Educ 1993;57:901.
1001