Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 104
Statement Concerning Application of
the Recommended Method
D. M. Hegsted
The subcommittee is in agreement that under idea,
conditions the probability approach which is recommended
appears to offer the best available approach to esti-
mating the proportion of the population at risk of nut ri
tonal deficiency from dietary survey data. As the
report notes, dietary surveys are used as a basis for
determining the magnitude of inadequate nutrition, serve
as a basis for food assistance programs, etc., and a
general instrument for formulation of nutrition policy.
I believe that the recommendation that the probability
approach be adopted is premature. The method has not
been adequately tested. It appears likely that this
approach will grossly overestimate the extent of inade-
quate nutrition in the country and, thus, lead to inap-
propriate policy decisions.
-
The two major requirements for the application of
the probability approach are a) that there be available
reasonably accurate estimates of the mean requirement
for each nutrient and an estimate of the range of require-
ments within the population group, and b) that the survey
methodology yield satisfactory estimates of the usual nutri-
ent intakes of the group.
~ -Dissenting statements prepared by individual members of
a committee are not subject to the normal National Research
Council review processes, nor are they subject to committee
or staff editing or review. They appear exactly as the dis-
senting committee members prepare them. The Research Coun-
cil neither endorses nor takes responsibility for the con-
tent of the statements.
104
OCR for page 105
105
NUTRIENT REQUIREMENT INFORMATION
Currently available estimates of the mean require-
ment for any nutrient are confined to few age-sex groups,
usually young adult men or women. Presumably the dietary
survey data will be analyzed according to the format of
the Recommended Dietary Allowances which provide values for
17 or more age-sex groups.
If so, "requirement curves for
each nutrient for each of these age-sex groups must be
inferred from the limited data available. Just how this
can be done to yield reliable values is unclear and one
must assume that such derived values will be less accurate
than values based upon actual data.
There are even less data on the range of require-
ments within any age-sex group. The report places con-
siderable reliance on the assumption that protein require-
ments determined by nitrogen balance have a standard devia-
tion of about 15% of the mean value. It must be empha-
sized that this value is the total variance of require-
ments determined by this technique which includes the
variation due to error as well as biologic variation. It
is the latter which is of importance. Since the estimated
mean requirement for nitrogen balance may vary by as much
as 50% depending upon the experimental protocol, particu-
larly the prior diet of the individual subjects, it is
apparent that the error term must be large and one must
assume that the true biologic variation is relatively
small. In all estimates of the variability in nutrient
requirements derived from compilation of values in the
literature the error term must be similarly large since
the experimental protocol, background of the subjects,
etc., differ.
The other example of differences in requirements is
based upon blood losses in women of the childbearing age
which can be translated into differences in iron require-
ments. The utility of this example, however, is compro-
mised by many factors which influence iron absorption,
such as the adaptive response of the individual, the form
of iron in the diet, and the amounts of various factors
which promote or inhibit iron absorption. Indeed, it
appears that such factors are more important in deter-
mining the availability of iron than the amount of iron
actually consumed. One may make general assumptions about
average levels of iron and of these f actors in the diet in
establishing recommended levels of iron intake, but these
OCR for page 106
106
have little applicability for the individual. We have no
methodology for including these factors in the probability
approach, and estimates of the proportion of the population
at risk of iron deficiency inevitably suffer. The recent
analysis of the extent of iron deficiency in the American
population for the NHANES data indicates that the proba-
bility approach overestimates the extent of the problem.
Currently available information on the nutritional
status of Americans and recent developments in nutrition
inevitably shift the interest in dietary surveys and how
they should be interpreted. Nutritional deficiency
disease is clearly not a major public health problem.
Ate major chronic diseases are the major public health
problem, and these are nutrition related. Interest in
dietary surveys in the future should focus on these
problems rather than nutritional deficiency. Thus, to
varying degrees, the report focuses attention on the
wrong problem and fails to indicate how the survey data
should be examined relative to these more important
issues.
Nutritional standards related to excessive consump-
tion of fat, cholesterol, sugar, or salt can probably
not be derived by attempting to develop curves related
to requirements or toxicity. Rather, these are devel-
oped by attempting to accommodate desirability, feasibil-
ity, acceptability, etc. Whether approaches similar to
those outlined for the probability approach can be adapted
to these issues is unclear at this time.
Recent evidence implicates vitamin A, carotene, and
vitamin C in the etiology of cancer. If intake of these
materials is relevant to cancer, it will have a profound
effect upon the interpretation of dietary surveys since
even modest effects upon cancer prevalence are immeasur-
ably more important than the defined nutritional defi-
ciencies. Yet the likelihood that one can develop reli-
able estimates of requirements to prevent cancer seems
relatively remote at this time, and the probability
approach can probably not be applied.
The subcommittee notes that for a considerable nether
of nutrients the information on requirements and/or range
of requirements is virtually nonexistent, the probability
approach cannot be applied, and estimates of the proportion
of the population at risk cannot be made. No doubt this
OCR for page 107
107
is true, but it seems rather unlikely that these nutrients
can be simply ignored in evaluating dietary survey data.
Finally, the subcommittee recommends that research to
improve knowledge of nutrient requirements be expanded in
order to make the application of the probability approach
more feasible. No doubt this is desirable, but given the
fact that nutritional deficiency does not appear to be a
major public health issue, it is quite unlikely that such
research can be a national research priority except for a
few nutrients. Thus it seems unlikely that the necessary
data required for more accurate application of the proba-
bility approach will be available any time in the near
future.
ESTIMATES OF NUTRIENT INTAKE
The report recognizes that the reporting of food intake
is subject to gross errors but, in my opinion, falls to
adequately consider the impact of such errors on the relia-
b~lity of the estimated extent of malnutrition. The
application of the probability approach rests upon the
assumption that the errors in reported food intake and
calculated nutrient intake are randomly distributed, i.e.,
that it is equally likely that under- and overreporting
occur with equal frequency and extent. No doubt the
reliability of the data collected in any survey will
depend upon the methodology used, but most of the data
available do not support the supposition and, even if it
is true that mean intakes are correctly reported (over-
and underreporting are equal), the method will still
apparently overestimate the extent of undernutrition.
Chapter 6 refers to many of the reports which have
attempted to validate the reliability of dietary recall or
food records. While the conclusions vary, a large number
of investigators find that such methods often yield grossly
low estimates of energy consumption, in some cases to the
surprising extent of 30 to 40 percent underestimate.
It should be noted that most such estimates refer to energy
intake only. It can be assigned that total energy intake
will be similar whether the subjects are free-living or
under controlled conditions if weight is maintained and
physical activity is similar. It is much more difficult,
or perhaps impossible, to obtain reliable estimates of
intake of other nutrients in free-living subjects. It is
OCR for page 108
108
reasonable to assume that most subjects will be more likely
to forget to recall or record food eaten than they will
list foods not eaten. Furthermore, most of the literature
reports have obviously been obtained with relatively well-
motivated subjects. This cannot be assumed in large scale
surveys where motivation may be minimal, and instructions
to the subjects will generally be less extensive than in
well-designed studies.
However, even if the mean intake of a group is cor-
rectly recorded, it still seems likely that the proba-
bility approach will yield excessive estimates of the
extent of malnutrition. It is certain that a substantial
part of the survey population will underestimate their
individual average intake either because of error or
because the two or three days derring which intake is
measured were atypically low. When nutrient intakes vary
greatly from one day to the next as they do in the U.S.,
often 100%, it is inevitable that a substantial number of
atypical intakes will be recorded. The probability
approach (or any other approach that has been suggested)
is concerned only with the nether of low intakes, the
"tail end" of the distribution of intakes. These remain
to be counted as in risk of deficiency even though over-
estimates might yield a correct mean value for the group.
The calculation of "usual intakes" by subtracting the
intraindividual variance from the total variance no doubt
diminishes the extent of erroneously recorded low intakes,
but when data are available for only two or three days, it
is not eliminated. The consumption, for example, of a
potent source of carotenoids once a week might provide an
adequate intake of vitamin A.
Another criticism of the probability approach is that
it provides no estimate of the severity of deficiency.
Subjects whose consumption is only modestly below standard
are grouped with those well below standard. The report
recommends that multiple standards be applied in order to
provide an estimate of severity but, at this time, only
the RDAs are available as standards. We can anticipate
obvious problems if agencies establish their own "require-
ment curves to meet their own needs or expectations.
OCR for page 109
109
CONCLUSION
It appears to me that the probability approach rests
upon a weak foundation both with respect to the data on
nutrient requirements and, especially, the survey meth-
odology. Most of the errors, biases, and variability in
the data collected are likely to result in substantial
overestimates of the extent of undernutrition. The cal-
culated examples in the body of the report do not appear
to be reassuring since, except for protein, they indicate
large numbers at risk of apparently all other nutrients
examined. his seems inconsistent with the reasonably
well-based conclusion that nutritional deficiencies are
not a public health problem of any magnitude in this
country.
The probability approach gives the impression that the
estimates of undernutrition are rather precise since the
utility of the extensive statistical calculations is not
apparent otherwise. The identification of such "problems"
will presumably call for public health solutions. Unless
the "problems" are real, which I doubt, this will result
in large expenditures of money and effort with little or
no benefit. Indeed, no solution appears to be in sight
nor are there other data indicating that the problems
exist.
Dietary surveys are useful in indicating food pat-
terns and provide data for rough comparisons of differ-
ences between groups, time trends in food consumption,
etc. Until some data are available which indicate that
the estimates of undernutrition obtained by application of
the probability approach are available, however, I do not
believe we should recommend its application. The most
appropriate reck foundations of the subcommittee should be
that the probability approach deserves further study but,
at this tome, the extent of under- or malnutrition can not
be determined from dietary survey data.
Representative terms from entire chapter:
dietary surveys
