LISA WAGNER: Welcome, and
thank you for standing by.
All participants are
in a listen-only mode.
Today's call is being recorded.
If you have any objections, you
may disconnect at this time.
My name is Lisa Wagner, and I
am on the policy team at the
National Center for Health
Statistics, or NCHS.
I am pleased to introduce
today's speakers, Dr. Lara
Akinbami, and Dr.
Bryan Stierman.
Lara is a pediatrician and
medical officer with the
Division of Health and Nutrition
Examination Surveys at NCHS.
Her research has focused on
childhood asthma and allergies
and health disparities.
Bryan is an epidemic
intelligence service officer
assigned to the Division of
Health and Nutrition Examination
Surveys.
He is a pediatrician and
medical officer specializing in
environmental health and has
authored several reports using
National Health and Nutrition
Examination Survey, or NHANES,
data.
Jay Clark will join Lara
and Bryan during the Q-and-A
session.
Jay is a senior statistician at
Westat with more than 20 years
of experience in the statistical
aspects of survey research and
data-quality control.
He has worked on sampling and
weighting tasks for NHANES for
more than 15 years and currently
oversees the daily statistical
activities to assist
NHANES with this study.
Lara and Bryan will present on
the newly released data set from
NHANES, from 2017 to March 2020.
This data set provides
nationally representative
estimates on selected health
outcomes prior to the COVID-19
pandemic.
The presentation
will be followed by a
question-and-answer session.
Just a reminder: The audience is
currently in a listen-only mode.
Questions or comments may be
entered through the Q-and-A
feature and we will address
them as time permits during the
Q-and-A session.
And now I turn it over to Lara.
DR. LARA AKINBAMI: Thank you,
Lisa, for that introduction.
And good afternoon, everyone,
and thank you for joining.
In this webinar, we will discuss
how the COVID-19 pandemic
affected data collection for the
National Health and Nutrition
Examination Survey,
also known as NHANES.
We will provide an overview of
how a public-release file was
prepared with data from
2017 through March 2020.
We will also present estimates
for prevalent health conditions
that were calculated
using these data.
First, we'll start with a
brief overview of NHANES.
NHANES is designed to assess the
health and nutritional status of
adults and children
in the United States.
It is a complex
survey in two ways.
First, it uses complex sampling
design to ensure that the data
collected are nationally
representative.
And, second, it combines
information collected across
several different
survey components.
During a home interview,
participants provide information
on demographic characteristics,
health conditions, and risk
factors and behaviors, such as
which dietary supplements and
prescription medications they
are using, as shown here.
Then, participants are invited
to travel to a nearby mobile
examination center to
participate in a health exam,
undergo tests, provide lab
specimens, and take part in
additional interviews.
After the exam, participants
may be contacted again to
participate in
post-exam content.
This can involve activities such
as dietary recall, interviews,
or wearing a physical
activity monitor.
The survey content varies over
time and covers a wide variety
of health conditions and
public-health topics.
Conducting examinations, along
with health interviews, provides
data that is invaluable
for public health.
But it also poses operational
and statistical challenges.
NHANES is conducted in 15 sites
per year due to the intricate
field operations of the survey.
The mobile exam centers, or
MECs, must be driven to each new
site and set up and
maintained according to exact
specifications.
(Audio break) -- include
interviewers, clinicians,
technicians, and engineers live
in the field full time as they
travel among the different
survey locations over the year.
Each MEC contains a mobile
laboratory with all the
equipment needed for specimen
processing and storage until
specimens can be shipped to
laboratories for testing.
On-site testing is also
performed for some health
measures to provide immediate
results to participants.
There's also other equipment
for medical testing in the MECs.
For example, a spirometer
has been used to measure lung
function and a sound-isolating
room is used to test hearing.
Depending on which health
exams and measures are being
performed, equipment can be
swapped in and out of the MECs.
The range of pre-pandemic
activities that occurred in the
MECs was broad.
These included body measures,
such as weight and height,
blood-pressure measurements,
DEXA scans to assess bone
density, oral health exams, and
phlebotomy and urine collection
to collect specimens for
a wide array of lab tests.
In addition, participants
responded to additional
interviews, such as
audio computer-assisted
self-interviews for more
sensitive topics that included
reproductive health and
alcohol and substance use.
The MECs provide a way
to standardize protocols,
equipment, and exam environments
across different locations and
across time, so that
results are comparable.
This allows for more accurate
interpretation of health
differences between groups
and of trends over time.
NHANES began continuous
field operations in 1999.
And, although data collection
continued from year to year,
data were released
in two-year cycles.
An example of the 2017-to-2018
data-cycle release on the NHANES
website is shown here.
Each two-year cycle is drawn
from a multiyear sample design.
These sample designs have
changed over time to keep up
with changes in the U.S.
population.
For instance, the 2015-to-2018
sample design selected 60
locations to be visited
over four years.
The 2015-to-2016 data-collection
cycle visited the other 30.
Although data are available for
two-year cycles, NHANES advises
combining cycles together into
four-year data sets to calculate
reliable estimates
for subgroups.
For example, estimating the
prevalence of a health condition
by age group separately for
men and women, or for race and
Hispanic-origin groups among
children, is best done with a
four-year data set.
So, like almost everything
else, NHANES was affected by the
COVID-19 pandemic.
The 2019-to-2022 sample design
also chose 60 locations to be
sampled over four years.
NHANES entered the field in
2019 with a plan of visiting the
first 30 locations in the
2019-to-2020 data collection
cycle.
In March -- sorry, in March
2020, that is, a growing number
of cases of COVID-19 disease
were being reported to CDC.
This suggested that community
spread was occurring.
Widespread shutdowns had not yet
occurred, but the environment
was starting to change.
For example, this mobility data
show that, during March, normal
patterns and movements
started to decline.
The NHANES program needed to
decide whether continuing field
operations posed a risk of
coronavirus transmission to
participants and staff
and their close contacts.
On March 16, field operations
for NHANES were suspended.
And, although it wasn't clear at
the time, this meant that the 30
locations planned for the
2019-to-2020 cycle would not all
be visited.
When field operations were
stopped in March of 2020, the
survey had been to 18
of 30 planned locations.
And, as 2020 progressed, it was
clear that there was no feasible
way to resume in-person exams.
The potentially long pause
before field operations could be
resumed raised questions about
how a break in data collection
would affect estimates
of health conditions.
Resuming data collection when
it was safe to do so would mix
pre-pandemic data and pandemic
data together and potentially
introduce bias into the
estimates, especially for a
two-year cycle that would
have to be extended.
Therefore, it was decided not
to collect more data for this
cycle.
Because no additional data would
be collected, the 2019-to-March
2020 sample was not
nationally representative.
There was no method to create
sample weights using the
2019-to-2022 sample design.
Additionally, publicly releasing
the data for fewer than 30
locations could pose disclosure
risks for participants.
However, the data that were
collected represented a
significant investment by
survey participants, the federal
government, and collaborators;
and simply not using the data
wasn't an option.
So, a solution was found in the
creation of a pre-pandemic data
file.
The 2017-to-2018 two-year cycle
contained a complete sample and
was nationally representative.
It could be used to
build a larger data set.
And methodology to combine
a probability sample with a
nonprobability sample was used
but adapted to this situation.
The probability sample in
this case was the 2017-to-2018
sample.
And, rather than a
nonprobability sample, the
2019-to-March 2020 sample was
a partial probability sample,
because it was selected based on
the 2019-to-2022 sample design.
So here's an overview of how a
2017-to-March 2020 pre-pandemic
data set was created and some
analytic considerations when
working with the data.
In this slide, I've highlighted
the sample design years in
colored font to distinguish
these four-year sample designs
from the two-year data
collection cycles.
The 2015-to-2018 sample design
specified the locations chosen
for the 2017-to-2018
data collection cycle.
And, as we mentioned previously,
all 30 locations were visited in
2017 to 2018.
The 2019-to-2022 sample design
specified 30 locations that were
supposed to have been visited in
the 2019-to-2020 data collection
cycle and only 18 were visited.
Combining the 2017-to-2018
sample with the 2019-to-March
2020 sample posed a problem.
The 2015-to-2018 and the
2019-to-2022 sample designs were
different because the
2019-to-2022 sample design was
updated to reflect the
changing United States.
So the chosen solution was
to pick one of these sample
designs.
Because the 2017-to-2018 data
collection cycle fully adhered
to the 2015-to-2018 sample
design, this design was chosen.
The 18 sites that were visited
in 2019 to March 2020 were
reassigned to the
2015-to-2018 sample design.
Now that a design was chosen,
the sample weights could be
calculated.
However, there were still
some issues that needed to be
resolved.
The 2019-to-March 2020 locations
didn't line up exactly with the
2015-to-2018 sample design.
The result was that some
portions of the country were
underrepresented in the data.
An adjustment factor was used to
equalize representation over the
sites visited from
2017 to March 2020.
And, once that was done,
interview weights and exam
weights were then calculated
using previous methodology.
Extensive assessments confirmed
that the final sample was
nationally representative by
making demographic comparisons
to the American Community
Survey, which is a population
survey administered by the U.S.
census.
So here are some analytic
considerations for using these
data.
The resulting 2017 to March 2020
pre-pandemic data can be used to
calculate nationally
representative estimates of
health conditions and behaviors.
It can be used like the
previously released data sets
for two-year cycles.
However, the data from the
partial 2019-to-March 2020 cycle
by themselves are not
nationally representative.
Therefore 2019-to-2018 data --
sorry, 2017-to-2018 data cannot
be compared to the
2019-to-March 2020 data.
And remember that, because the
2019-to-March 2020 data did not
conform to the 2019-to-2020
survey design, no separate
survey weights could be
constructed for this cycle.
It is not appropriate to use the
2017-to-March 2020 pre-pandemic
waits for the partial sample
collected in 2019 to March 2020.
The weight adjustment that was
applied to the 2017-to-March
2020 data was designed for
overall estimates but not
necessarily for subgroups.
So, therefore, when
2017-to-March 2020 estimates for
subgroups are compared to
earlier estimates, trends should
be interpreted with caution.
For example, when the adjustment
factor and other measures were
applied to the survey weights,
national representation by sex
was achieved and so is
representation by age.
But some sex-specific age
groups, for example, may have
larger variation in estimates
depending on how the
participants are distributed
across survey locations.
The 2017-to-March 2020 data set
represents a 3.2-year sample
rather than a two-year sample.
This means that sample sizes are
larger than for two-year samples
and that standard errors for
estimates are generally smaller
than those calculated for
-- (inaudible) -- data sets.
In other words, estimates made
with a 3.2 year sample will
generally be more precise than
those from a two-year sample.
Conversely, standard errors for
the 3.2-year 2017-to-March 2020
data set will likely be larger
than for four-year data sets
made by combining
previous two-year cycles.
So, speaking of combining data
cycles together, analysts may
wish to do this with the
2017-to-March 2020 data in
certain circumstances, such as
when assessing a less common
health condition or when
analyzing subgroups with smaller
sample sizes.
Previous analytic guidelines had
formulas to use to apply to the
survey weights when
combining cycles.
Adjusting the weights to account
for the different time period
for the 2017-to-March 2020 cycle
is necessary when combining
these data with prior
two-year cycles.
So, for example, when the
2015-to-2016 two-year cycle is
combined with 2017-to-March 2020
3.2-year data, this data set
represents a 5.2-year
period when combined.
Therefore, analysts would want
to use a multiplier for sample
weights for each cycle.
This should be the fraction
represented by each cycle of the
average annual population of
the entire 5.2-year period.
Or another way to think about it
is that one would be calculating
the population in the
middle year of the period.
For 2015-to-2016 sample weights,
the multiplier would be the
fraction 2 over 5.2.
And, for 2017-to-March 2020
sample weights, it would be the
fraction 3.2 over 5.2.
Likewise, combining three
successive cycles would
represent a 7.2-year period.
The multipliers for the survey
weights for each cycle are shown
here.
And note that, in each case, the
numerators for the multipliers
across the cycles add up to the
duration of the entire period.
Analysts also need to keep in
mind some issues involved in
combining cycles together.
One major consideration is if
the condition being analyzed is
changing prevalence over the
time period being assessed.
Combining cycles together would
mask any trend over the period.
Analysts also need to check if
the variables being assessed
remained the same over
successive cycles by consulting
the survey documentation
for each cycle.
So here's a brief overview
of how to access the data.
The 2017-to-March 2020
pre-pandemic data will be
released in stages just
as done in prior cycles.
Components with currently
available data have active
links, components that have not
yet been released or indicated
with hatching.
And clicking on any of the live
components on the NHANES website
will show the currently
available data files.
So here's the current list of
data files that are available
for downloading.
These include the demographic
file, questionnaire data on
blood pressure, diabetes, oral
health, and examination and lab
data related to measuring these
conditions, as well as weight
status.
Again, these files can be used
like those from previously
released two-year data cycles.
Information about using
the data is also provided.
Currently, there's a brief
overview with analytic guidance
that was posted with the data
when it was released at the end
of May.
A longer, more detailed report
will be released later in 2021.
And a report presenting
estimates of selected health
conditions was released
earlier this week.
These resources are both
available using the first two
links under the highlighted
section using the data that is
on the data-release page.
So here's a glimpse of the brief
report and the direct link at
which it is available.
And now I will hand over the
presentation to Dr. Bryan
Stierman, who will present
an overview of the recently
published report on health
estimates from this data set.
DR. BRYAN STIERMAN:
Thank you, Lara.
We present here prevalence
estimates for several selected
health outcomes calculated
from the available NHANES
2017-through-March 2020
pre-pandemic data set.
Next slide, please.
These estimates are available
in more detail in the National
Health Statistics Report that
was released this week and that
is available on
the NCHS website.
Next slide.
The health outcomes selected for
estimates include for children,
obesity and dental caries; for
adults, hypertension, obesity,
severe obesity, and diabetes;
and for older adults, complete
tooth loss.
These health outcomes were
selected for estimates because
they were able to be calculated
from the files currently
released publicly available
on the NHANES website.
Next slide.
Today we present estimates by
several covariates including
sex, age groups, race and
Hispanic origin, and family
income.
Other covariates and
stratification by sex are
included in the accompanying
National Health Statistics
Report publication.
Next slide.
As is usual with NHANES
analyses, to calculate these
estimates we accounted for the
complex, multistage probability
design of NHANES, including
the unequal probability of
selection.
Provided sample weights
were used for calculations.
For estimates for diabetes,
fasting sample weights were
used.
For all other estimates,
examination simple weights for
used.
Standard errors were
estimated using Taylor series
linearization.
And adult estimates were
directly age-adjusted to the
2000 projected U.S.
census population.
Next slide.
As would be expected, the
overall estimates for each
health outcome calculated for
2017 through March 2020 are
similar to those from
2017 through 2018 alone.
This reflects both the
methodological adjustments, as
well as the patterns in the
prevalence estimates, which
typically are not expected to
vary by large amounts from when
one year to the next in NHANES
due to the relatively small
sample size and a
one-year data collection.
Next slide.
The data from 2017 through March
2020 provide an increase in
sample size, generally about 1.5
to 2 times the sample size of
that from 2017
through 2018 alone.
Next slide.
As expected, this increase in
sample size generally leads to
smaller standard errors, as can
be seen with all health outcomes
here except for
complete tooth loss.
However, for some estimates
in some demographic subgroups,
increased variation in the
sampling weights, increased
variation in the true underlying
population values of the health
outcomes from the data added
from 2019 through March 2020, or
both may result in equivalent or
increased variance of estimates,
as seen here with complete
tooth loss, which has equivalent
standard errors from
both time periods.
Next slide.
We found that 19.7% of children
aged 2 through 19 years had
obesity, defined as a body mass
index greater than or equal to
the 95th percentile
for age and sex.
There was no difference
in obesity by sex.
Obesity increased with
increasing age groups.
The highest prevalence of
obesity was among non-Hispanic
Black and Hispanic children.
While non-Hispanic Asian
children had a lower prevalence
of obesity than other race and
Hispanic origin groups, obesity
decreased with
increasing family income.
Next slide.
Dental caries in childhood was
defined here as untreated or
restored dental caries in one or
more primary or permanent teeth.
46% of children aged 2
through 19 had dental caries.
There is no difference
in dental caries by sex.
Dental caries increased
with increasing age groups.
Hispanic children had the
highest prevalence of dental
caries among children.
And dental caries decreased
with increasing family income.
Next slide.
For hypertension, the estimates
are based on a different
methodology than those
previously published for NHANES.
Prior NHANES hypertension
estimates have used an
auscultatory protocol for
blood-pressure measurements.
During 2017 through 2018, both
an auscultatory protocol, which
utilizes a manually obtained
blood pressure with a mercury
sphygmomanometer, and an
oscillometric protocol, which
utilizes an automated machine
to obtain blood pressure, were
used.
However, during 2019
through March 2020, only an
oscillometric protocol was used.
Therefore, blood-pressure
measurements and hypertension
estimates for the combined
2017-through-March 2020
pre-pandemic data required
the use of the oscillometric
protocol.
The differences in these
protocols and a comparison of
the blood-pressure values from
each protocol are available in a
separate Series 2
report from NCHS.
Next slide.
We define hypertension here as
meeting any of the following
three conditions: a mean
systolic blood pressure of
greater than or equal to 130
millimeters of mercury, a mean
diastolic blood pressure of
greater than or equal to 80
millimeters of mercury, or
taking a medication to lower
blood pressure.
Again, the blood pressure
measurements were taken using an
oscillometric protocol.
During 2017 through March
2020, 45.1% of adults had
hypertension.
More men had
hypertension then woman.
Hypertension increased
with increasing age.
And Non-Hispanic Black adults
had a higher prevalence of
hypertension than other race
and Hispanic origin groups.
Next slide.
We found that 41.9% of adults
had obesity, defined as a body
mass index greater than or
equal to 30 kilograms per meter
squared.
There was no difference in
obesity by sex or by age.
Non-Hispanic Black adults
had the highest prevalence of
obesity.
Non-Hispanic Asian adults had a
lower prevalence of obesity than
other race and
Hispanic origin groups.
Next slide.
Severe obesity was defined here
as a body mass index of greater
than or equal to 40
kilograms per meter squared.
During 2017-from-March 20, 9.2%
of adults had severe obesity.
More women had severe
obesity than men.
Severe obesity was less common
in those aged 60 and above,
compared to those aged 20 to 39,
and those aged 40 through 59.
The prevalence of severe obesity
was highest among non-Hispanic
Black adults, and the least
among non-Hispanic Asian adults.
Severe obesity was lowest among
those with a family income of
greater than 350% of the
federal poverty level.
Next slide.
Diabetes was defined here as
having previously been given a
diagnosis of diabetes, having
a fasting plasma glucose of
greater than or equal to 126
milligrams per deciliter, or
having a hemoglobin A1C
greater than or equal to 6.5%.
Fasting sample weights were used
to calculate these estimates.
14.8% of adults had diabetes.
The prevalence of diabetes was
higher among men than women.
The prevalence of diabetes
increased with increasing age
but decreased with increasing
family income and the prevalence
of diabetes was lower in
non-Hispanic White adults
compared to other race and
Hispanic origin groups.
Next slide.
Complete tooth loss among adults
aged 65 years and older was
defined here as having no
natural tooth, dental root
fragment nor implanted tooth and
was based on 28 teeth, excluding
third molars.
The prevalence of complete
tooth loss was 13.8%.
The prevalence did not differ by
sex but did increase with age.
Tooth loss was higher among
non-Hispanic Black adults than
non-Hispanic White adults but
otherwise did not differ by race
and Hispanic origin and tooth
loss decreased with increasing
family income.
Next slide.
So with regards to the
future, more data releases are
anticipated.
These data releases will occur
in several different forms.
Other combined
2017-through-March 2020
pre-pandemic data are expected
to be released on the NHANES
website and would be treated
like a probability sample.
And provided sample weights
should be used for analysis with
these data.
Next slide.
In the future, this data would
be released on the NHANES
website along with the currently
available data, which can be
found under the NHANES 2017
through March 2020 Pre-pandemic
data page.
Next slide.
In some cases,
2017-through-March 2020
pre-pandemic data determined
to have disclosure risk will be
released through the NCHS
Research Data Center to ensure
additional measures to
protect confidentiality.
For these data, which are
treated like a probability
sample, the provided sample
weights should also be used for
analyses.
Next slide.
For those data released as
limited access data files, once
released, information about the
variables will be available on
the NHANES website under
limited-access files, under the
2017-to-March 2020
Pre-pandemic data page.
However, the actual data will
only be available through NCHS's
Research Data Center.
Next slide.
There are some measures that are
unique to the 2019-through-March
2020 NHANES data collection.
These cannot be combined
with 2017-through-2018.
And, for these measures,
nationally representative
estimates are not possible.
These data will instead be
released as a -- (audio break)
-- through the NCHS
Research Data Center.
Next slide.
For these data, released as
limited access data files, once
released, information about the
variables will be available on
the NHANES website under
limited-access files under the
2019-through-2020 data page.
However, again, the actual data
will only be available through
NCHS's Research Data Center.
Next slide.
And this can be found
on the NCHS website.
Information about accessing
restricted data, including
submission of research
proposals, can be found here.
Next slide.
Thank you.
MS. WAGNER: We are now entering
the question-and-answer session.
As time allows, the presenters
will address questions from the
Q-and-A feature.
Please submit your questions
through the Q-and-A feature now.
If your question or comment is
not addressed, please direct it
to paoquery@cdc.gov.
That's P-A-O-Q-U-E-R-Y@cdc.gov.
Okay.
So we definitely
have some questions.
One question is: Will all
measures collected be released?
For example, for example, will
the dietary data eventually be
released?
DR. AKINBAMI: I can
take that question.
Yes, we plan to definitely
release all data.
It will be in stages as the data
are processed and evaluated,
kind of similar to the schedules
used for previous cycles.
When they are released, there'll
be under the "What's New"
section of our webpage, and we
anticipate the next data release
to be towards the end of summer.
MS. WAGNER: Great.
We have another question
on the poverty level.
How many participants -- how
many participants had missing
values of FPL?
Is an imputation approach
required for analyzing the FPL?
DR. STIERMAN: Yes, so
that's a good question.
I could take that one.
So not all participants had
their federal poverty level
necessarily available, though
it was a very high percentage.
I don't have the exact
number available right now.
The sample sizes that we had for
these estimates are available on
the NCHS National Health
Statistics Report that I was
referring to earlier.
And I believe I'd have
to double-check the exact
percentages.
But I think that we had a high
enough percentage -- and it's
pretty close to complete -- that
we didn't think that imputation
was necessary here.
Yeah.
MS. WAGNER: Okay.
Great.
We have another question
about restarting NHANES.
So do you all have any
additional information around
when the NHANES will restart?
DR. AKINBAMI: Yes, we're very
excited that we'll be restarting
operations this summer.
So we plan to -- we're actively
planning and preparing to return
to the field just in a couple of
weeks and hopefully we will have
smooth sailing to the end
of the cycle this time.
MS. WANER: Great.
We have one question about
obesity, the obesity measures I
think, Bryan, you referred to
particularly -- the question is:
Could you show the slide with
the findings again, on the
2017-to-2018 or -- the
2017-to-March 2020 findings?
So I don't know, Lara, if you
can do that, or if you guys can
recount those particular
measures or findings.
And then the -- I think they
had the same question on when we
would be restarting.
Would it be a
2021-to-2022 data set?
DR. STIERMAN: Sorry, I was
trying to unmute myself there.
Can you all see this slide now?
MS. WAGNER: Yes.
Yes.
DR. STIERMAN: With
the obesity estimates?
Okay.
Yes.
So this was for adult obesity,
again, greater than or equal to
20 years.
That's a body mass index of
greater than or equal to 30
kilograms per meter squared.
So I'm showing it here.
But, again, for people who want
to be able to look at any of
these slides, or want to see the
estimates in even more detail,
there is a National Health
Statistics Report, which I can
show again the link to a little
bit later, but easily found on
the NCHS website, or
on the NHANES website.
There's a link to it as well.
It has all of these
estimates plus more.
MS. WAGNER: And then the second
part of that question was around
the data set.
So, once we restart NHANES,
will that be a -- will we be
considering that the
2021-to-2022 data set?
Or is there still
discussions about that?
DR. AKINBAMI: That would
definitely be considered
'21-to-'22 data set.
So we anticipate release of that
with the normal schedule that we
have for production
after data processing.
MS. WAGNER: Okay.
And the -- this question --
also, could you go back to the
previous one, previous slide?
DR. STIERMAN: So here's one
for hypertension, which was the
previous slide.
MS. WAGNER: It says, with all
the measures, obesity, adult
diabetes.
DR. STIERMAN: Okay.
MS. WANER: Okay.
Yeah.
DR. STIERMAN: I can go back.
That is several slides previous.
Here are the estimates
for each of the measures.
I'll just kind of highlight
a little bit more.
So, in orange, what you can see
here is the 2017-through-March
2020 pre-pandemic estimates.
And for comparison's sake,
in blue, we provide the
2017-through-2018
alone estimates.
So just to -- kind of as a
reminder, the 2017-through-March
2020 contains the
same participants as
2017-through-2018, along with
those that were enrolled in 2019
through March of 2020.
So they are not
independent samples.
MS. WAGNER: Okay.
All right.
We have another question here
on trend testing for several of
your sample designs.
So the question is: Could
you please clarify your
recommendations regarding trend
testing for 1999 to 2000, 2003
to 2006, etc.? 2013 to 2016,
and then 2017 to March 2020?
DR. AKINBAMI: Yeah,
that's a great question.
We have analytic guidelines on
our webpage that covers some of
those questions specific to
NHANES, as well as a general
NCHS trend report that has
recommendations on analyzing
trends of our data.
The thing that you need to take
into account when looking at the
'17-to-'20 data is that it's a
3.2 year-period, as opposed to a
two-year period.
And therefore, the time points
would be unequally spaced across
those data points.
So you would need to use some
indicator in your model of the
midpoint of each data period.
So one way we do it is to take
the time period, and divide it
by 2, and then add that to the
beginning of the time period.
So, for example, if you were
doing a trend from, say, 2013 to
2020, the first time
period would be 2014.
Then, the next would be 2016.
And then the next for the
3.2-year cycle would be 2018.6.
I think that's right.
And that would allow your model
to know that those data are
coming from unequally
spaced periods in time.
For the weighting, again,
the previous weighting
recommendations are in
our analytic guidelines.
And I refer you to those because
it's a little difficult to relay
all of those multipliers for the
sample weights, but, the ones
that we presented today here,
we'll make sure we release those
on our website soon so you'll
have those for reference.
And I believe this recording
will also be available for that.
MS. WAGNER: Great.
Great.
We have a question
regarding weights again.
So the question is: When
calculating weights for combined
data sets -- so, for example,
between the 2015-to-2016, with
the 2017-to-March 2020, you
indicated that the weight for
the 2015-to-2016 cycle should
be the two divided by 5.2.
And, for 2017-to-March 2020
should be 3.2 divided by 5.2.
So the question is: Does that
mean, when coding for the new
weight, it would be done
differently within the same
combined data set for data
from the individual cycles?
Does that makes sense?
DR. AKINBAMI: I'm not sure but
the -- I guess the way I would
approach that is to combine the
data and then to multiply the
weights and then to do any
estimates from your combined
data set.
So you would -- so, for example,
if you're interested in just the
'15-'16, you would subset that
sample when you're looking at
that estimate but use the
entire combined data set.
I know -- Jay, do you have
another take on that question?
I'm not sure I interpreted
it completely correctly.
JAY CLARK: I think
you answered it fine.
I mean, the main thing is just
instead of two-year cycles now
-- before each data
set was for two years.
This new one for 2017-to-March
'20 is for 3.2 years, so you
just have to make the
appropriate adjustment.
And I think it is spelled out
more clearly in the analytic
guidelines.
MS. WAGNER: Right.
So we have another question.
The question starts
with a statement.
You said with some of the
measures the 2019-to-2020
findings could not be merged
with the 2017-to-2018 findings.
Could you give an example
or examples of them -- which
measures those were?
DR. AKINBAMI: So those measures
would include things that were
newly introduced in 2019-to-2020
that wouldn't have any
counterpart in '17-'18.
And I'm -- I think some of
them might be nutrition for
0-to-2-year-old children.
That may have started in -- I
have to go look to make sure I
get -- I'm getting this right.
But, for example, there would
not be a '17-'18 variable to
combine those data with.
So we can only release the data
that were consistently measured
in the '17-'18 cycle and the
'19-2020 -- March 2020 cycle.
MS. WAGNER: Okay.
We have another question.
What impacts could this combined
file have on data linkage or
analyses using linked files?
DR. AKINBAMI: So that's
a great question.
We are definitely in
conversation with the linkage
group at NCHS.
And we anticipate that the
linkage would continue pretty
much the same way.
As for other files, again, we
would probably do linkage for
the '17-March 2020 file to
protect confidentiality and to
make the linkages more robust.
MS. WAGNER: We have
a question as well.
Are you all thinking about how
the next iteration of NHANES
might incorporate any
COVID-19-related data?
DR. AKINBAMI: Yes.
It's a different survey than,
say, real-time surveillance but
it's -- when you combine those
COVID data potentially with all
the rich data we have, it's
a very powerful data set.
So we're asking about
immunization as well as zero
status.
So we're collecting serology
data, both to indicate whether
you are naturally or
participants were naturally
infected versus obtaining
antibodies through vaccination.
MS. WAGNER: Okay.
So I think this might be a
follow-up to the trend-analysis
question.
Can we use a regression to model
an outcome by two-year periods
plus the 2019-to-March 2020
period as a continuous variable?
DR. AKINBAMI: Yes, and in that
-- in that case, you would use
kind of the midpoint of the time
period that I was talking about
earlier.
So, for example, if you started
with 2013-to-2014, the midyear
of that would be 2014
when you round up.
The mid-year of
'15-to-'16 would be 2016.
And then the midyear of
'17-to-March 2020 would be
2018.6.
So those are the values you
would put in your regression.
But you could also just start
with one if you wanted to, just
as long as those have the
same spacing as the years.
You could use essentially any
number you wanted to represent
the data points in time.
MS. WAGNER: Okay.
We have a follow-up question
from the COVID-19 related
question.
So are the samples that were
collected in Q4 2019 and Q1 2020
being checked for any antibodies
or other markers of COVID-19
exposure?
DR. AKINBAMI: That's
a great question.
We have not checked those data
because, as I mentioned before,
we're a very clustered survey.
So the places that we were --
just because it's a partial
sample in an entire sample
design would not be nationally
representative.
So it would be difficult to
interpret how to use those
estimates if we were
to check those data.
So, while we have stored samples
and would be able to do that, in
the future, if it looks like it
would be -- it would reveal some
useful public health data, we
haven't undertaken that yet.
MS. WAGNER: Okay.
Another question: Can you
compute subnational estimates
with these data?
DR. AKINBAMI: That's a
question we get frequently.
And, unfortunately, it's very
difficult because of our design.
In order to examine enough
people, we're limited into how
many places we can visit per
year because you have to have --
to stay in one area long enough
for people to be interviewed and
then to travel to the
MEC to get examined.
That means you're in very -- you
have very sparse data per year.
And we don't visit the
same places every year.
So it would be difficult to be
able to combine enough cycles
for any one place to get
a subnational estimate.
The exception is LA County,
which is so large that it is
usually included in our sample.
And we do have data from
that county available as
limited-release data.
You can find that
on our website.
But that is the only example
really we were able to go
subnational.
MS. WAGNER: Okay.
We have a question about -- how
can the public best know what
happened throughout most of
2020 and early 2021 in terms of
trends regarding the U.S.
obesity, diabetes, hypertension
and other measures because there
will be no NHANES
data for that period?
Should or can people look to
other surveys, like the National
Health Interview Survey
for some of those data?
DR. AKINBAMI: I would say yes.
Bryan, you know what?
I can -- I can chime in and,
Bryan, you can chime in and talk
with me.
So for -- yes, there are other
surveys that release those data
based on self-report.
And there are some good
studies out there that show how
self-report measures of that --
basically, BMI calculated from
self-report measures differed
from measured values.
So, in general, people tend to
overestimate their height and
underestimate their weight.
It would be like squeezing
your zoom picture a little bit.
So there's a lot of literature
about adjustment equations and
those type of methods that you
could use to adjust for the mode
effect of people measured versus
self-reporting their weight and
height.
But the health interview
survey collects those data.
The behavioral risk factor --
BRFSS -- behavioral risk factor
surveillance survey
also collect those data.
And those are both
available at CDC.
Great.
DR. STIERMAN: I didn't
really have much else to add.
I think you answered
that question well.
Yeah, the -- some of those are
NCHS surveys and some of them
can be found through -- for
instance, for obesity, going to
the obesity website on CDC.
They have a lot of linkage to
different surveys that they use
to come up with
estimates as well.
MS. WAGNER: Great.
So we have a question now -- a
couple of questions regarding
the restart of the survey or,
I guess, the planned restart of
the survey.
Are there general concerns with
the program around restarting,
particularly with
response rates?
And are there any sort
of plans to address that?
And, then, I guess the second
part of that question that might
help with this is I guess: What
sort of precautions are being
taken into consideration
as you're restarting?
DR. AKINBAMI: Yeah, great
questions and ones that
concerned us greatly
through the past year.
So it's true that the response
rates for NHANES as well as all
federal surveys have been
declining since about 2011.
And it is a concerted effort
across federal statistical
agencies to understand and
address those falling response
rates because, as we know, when
you have a lower response rate,
that's believed to possibly
impact the data quality.
We did an extensive evaluation
in -- of the 2017-to-2018 data
of how nonresponse
affected our estimates.
And that report is also
available on our website.
We used all of those methods
developed to really assess and
address bias with the
2017-to-March 2020 weighting as
well.
When we go back into the field,
we are really looking to all the
various methods to really
interest people in responding to
the survey to increase response
rates and we are trying many
different things together.
Part of that is -- as the second
part of the question alluded to,
is ensuring that participants
are -- that they can safely
participate.
So, when they visit the mobile
examination center, there is a
screening protocol in place.
There's information about that
on our website, as well, for
participants.
If anyone listening happens
to be chosen, you can get that
information there.
It's very similar to what's
being done in medical centers,
including the same type of
precautions with the full
masking while you're in the MEC.
You know, Jay, was there any
other aspects that you're aware
of that I didn't mention?
MR. CLARK: I think you
covered it really well.
MS. WAGNER: Great.
For Bryan's question, how do
hypertension estimates differ
with the two new protocols
-- or the two protocols, the
oscillometric one and the
auscultatory, the new one being
used?
DR. STIERMAN:
That's very helpful.
Excellent question.
So the -- I'd say for the most
comprehensive answer, I would
refer the person who asked the
question to the -- there's an
NCHS Series 2 report by
Ostchega, et al, which is
entitled "Comparing Blood
Pressure Values Obtained by Two
Different Protocols," and it was
released in April of this year.
The report compares the two
blood-pressure protocols during
2017 through 2018, when they
were both taken among the same
participants.
And, briefly, the mean systolic
and diastolic blood pressures
did differ a little bit with
higher-mean systolic blood
pressure, and lower-mean
diastolic blood pressure using
the auscultatory protocol
compared to the oscillometric
protocol.
However, the prevalence of
stage-one hypertension, which is
what we showed here, did not
really differ significantly
between the two protocols and
there's generally good agreement
between those two protocols.
Just to give a little bit more
of a background on these two
different measurement protocols,
I think something that can be a
little bit helpful is to -- and
I touched on it a little bit.
But the auscultatory protocol
uses mercury sphygmomanometers
and was taken with clinicians
or physicians, whereas the
oscillometric blood-pressure
protocol use automated devices
and are done with
health technicians.
So there were some minor
differences in the actual
protocol as well as the
differences in the equipment
that were being used.
And some more detail on this,
again, it's in that Series 2
report that I've
mentioned a few times now.
MS. WAGNER: Great.
So we have another question
on the COVID-19 data potential
additions to the next
iteration of NHANES.
Would we get a self-reported
history of COVID-19 infection as
a part of that next iteration?
Is that one of the
considerations?
DR. AKINBAMI: There is a
COVID-19 questionnaire that goes
along with the measurements.
And I believe there are
questions about self-reported
infection to correlate with
the serology data that will be
collected.
MS. WAGNER: Okay.
Great.
We have no additional questions.
We have no additional
questions currently.
Any other questions from our
attendees, please type them in
the Q-and-A feature now.
All right.
I'm not seeing any.
So thank you all for attending
today's NCHS webinar on the
National Health and Nutrition
Examination Survey 2017-to-March
2020 pre-pandemic data release.
If you have any questions or
comments that were not addressed
during this webinar, please
email them to PAOquery@cdc.gov.
That's P-A-O-Q-U-E-R-Y@cdc.gov.
And I just want to say a special
thanks to Lara, Bryan, and Jay
for joining us today.
So thank you all.