Income, Poverty, Health Insurance Coverage Technical Briefing Part 2 of 2


Michael Cook: Okay we’re going to go ahead
and get started. We’re going to go ahead and get started. I’d like to welcome everyone
back to the second half of our technical meeting on plans for the upcoming releases of income,
poverty and health insurance coverage estimates for federal sources. Our second half of the agenda consists of
health insurance coverage estimates and other source information from Mr. Stephen Blumberg
the Associate Director for Science in the Division of Health, Interview Statistics at
NCHS. And also, we’ll hear from our own Alfred Gottschalck
the Assistant Division Chief for Small Area and Longitudinal Estimates here at the U.S.
Census Bureau. So without further delay here is Stephen. Stephan Blumberg:All right welcome back. Now
we turn to the National Health Interview Survey. The National Health Interview Survey is conducted
by the National Center for Health Statistics and is one of our principal sources of information
on the health of the US population. It provides estimates for monitoring progress
towards public health goals and for addressing specific issues of current public health concern
including the health insurance coverage of the US population and its relationship with
health status and health care access and use. Like the CPS the National Health Interview
Survey the health household survey of the civilian noninstitutionalized population conducted
by interviewers from the US Census Bureau. The HIS however is a cross sectional survey
which means that we generally interview each family only once. Interviewing is continuous throughout the
year with a goal of completing interviews in at least 35,000 households annually and
often more if funding permits. The National Health Interview Survey has been
collecting data continuously since 1957. Questions about the health insurance coverage of family
members have been part of NHIS since 1959. The monitoring was periodic until 1968 then
every two years until 1986 and annually since 1989. Since 1997 which was the last time that the
NHIS questionnaire was redesigned the health insurance section has begun with a general
question about whether anyone in the family is covered by any kind of health insurance
or some other kind of health care plan. If so then we ask what kind of health insurance
or health care coverage each family member has. For each type of coverage we then ask
a series of detailed questions about that coverage. So examples include questions about how the
plan was obtained, who pays for it, whether it’s a high deductible health plan and whether
it has managed care features. And then we collect the full names of all
private and public plans preferably from a health plan card or other communication from
the health plan. Health insurance as most of you know is a
complex topic. Some inconsistencies in survey response are expected therefore before producing
statistics on coverage the NHIS looks at the responses to the entire battery of insurance
questions. If the follow-up questions clearly suggest
that the original coverage type recorded was incorrect the follow-up questions are then
used to assign the coverage type. Now before I go further I want to highlight
several strengths of the National Health Interview Survey health insurance data. The NHIS data are collected in the context
of extensive data on the health and healthcare of the individual. We collect extensive follow-up
data including plan names to help us verify public or private coverage. And because we have been – and because the
data has been collected using the same general approach since 1997 observed changes in coverage
over time can be considered reliable. We also have sufficient sample sizes to permit
annual coverage estimates for a majority of states and in fact with 2014 data we were
quite excited that it was the first time we had sufficient sample sizes to produce estimates
for all 50 states and the District of Columbia. This map showing state estimates of uninsurance
from 20 using 2014 data was published this past June. Dark green identifies the states with the
lowest uninsurance rates Hawaii had the lowest percentage of uninsured individuals under
age 65 in 2014 followed by Massachusetts, Delaware and Iowa. The District of Columbia
also had a similarly low uninsurance rate. The dark purple show the states with the highest
uninsurance rates for persons under age 65 in 2014. The highest rates were observed in
Texas and Oklahoma followed by Alaska and Florida. Now as I mentioned the National Health Interview
Survey is conducted continuously throughout the year and that’s based on monthly random
samples. We take the 12 monthly samples over across
the calendar year, aggregate them together into annual data files and after processing
and weighting release those files in June of each year so 2014 data were released just
this past June. And we recognize however that for key topics
health insurance included many people want to see preliminary early data from the National
Health Interview Survey. And with that goal in mind the National Center
for Health Statistics developed the HIS early release program. Through this program early release products
are developed to provide early access to the most recent information. Now these products are produced prior to final
processing and weighting and therefore the estimates that are in these reports can best
be considered preliminary. I’ll talk a little bit more about that in a moment. Turning back to what exactly the program produces
every three months the early release program produces a report on 15 key health indicators,
a report on health insurance coverage, Web tables that present quarterly health insurance
estimates and also we release through our research data centers preliminary micro data
files that are used for these reports so that other researchers may be able to analyze the
data themselves. The early release health insurance estimates
represents an average over the months included in the report. They are in that sense cumulative. So historically
in September we have released estimates that are based on data collected from January through
March. Then in December we’ve released data collected
or data from the first six months of the year, in March 9 months and in June the full 12
months of the year. However with funding and encouragement from
the Assistant Secretary for Planning and Evaluation in the Department of Health and Human Services
we have been working to improve the timeliness of our releases. And in fact the release from
the first three months of 2015 came out two weeks ago on August 12, 2015. We anticipate that across, you know, for the
next three reports that we will be releasing them roughly a month ahead of our historic
schedule that is that the January to June data will be released in November, the January
to September data will be released in February and the full year data will be released in
May that’s for health insurance estimates only. Again the full year 2015 data sets and reports
based on that for other than health insurance we anticipate coming in June 2016. Now this slide shows the first page of the
report released two weeks ago based on data from January to March 2015. Those of you who are familiar with our reports
may notice that we changed the format. The new report has more figures and less text
but all of the estimates that were included in early reports are still there they’re just
now in there as appendix tables. The big highlight from this report was that
seven million fewer persons lacked health insurance coverage in the first three months
of 2015 compared to 2014. This estimate is a point in time estimate
reflecting reports of insurance status at the time of interview. Most of the estimates
in the report are point in time estimates but we do present three different uninsurance
measures a point in time, and estimate of uninsurance at some time in the previous 12
months and an estimate of the percentage of persons who have been uninsured for more than
one year. As you can see on this slide since 2013 the
percentage of uninsured has decreased for all three measures of the uninsured. Now getting back to the point in time uninsurance
estimates I want to show you a selection of figures from this report that was released
two weeks ago. So this graph shows changes in the uninsurance
rate point in time uninsurance rate by age. You can see that the prevalence of uninsured
adults has recently decreased for all age groups. The decrease occurred for all income groups
though more sharply for near poor adults. That is adults living in households or living
in families with incomes between 100% and 200% of the federal poverty level. And recent
declines were steepest for Hispanic and non-Hispanic Asian adults. The report also presents estimates of private
and public insurance for adults shown here and for children. Note that among children under age 18 the
percentage with private insurance increased from about 52% in 2013 to 56% in the first
three months of 2015 reversing a 14 year trend of declining rates of private coverage for
children. Newly released reports also present the percentage
with private coverage obtained through the health insurance marketplace or state based
exchanges. This percentage has increased from the first
quarter of 2014 to the fourth quarter of 2014 and again in the first quarter of 2015. As I said that’s just a selection of some
of the figures that are presented in our early release health insurance report. Some of the other information that’s in there
includes the percentage of persons in high deductible health plans. They’re also are several figures representing
subnational coverage estimates for instance by region, by state Medicaid expansion status
and by state health insurance marketplace type. We also present estimates for selected states
though in the recent report there are no state estimates because sample size is just not
sufficient after three months of data collection to present state estimates. But the November and the February reports
we anticipate we’ll have estimates for probably ten to 12 states and then for the May report
we hope that with the full year 2015 data we will again be able to present estimates
for all 50 states and the District of Columbia. Now as I mentioned before the health insurance
estimates that we present in the early release reports are preliminary and that’s because
they’re based on a streamlined version of the final processing procedures. They’re more automated than the manual checks
that we do through all the follow-up data when we’re doing final processing. They’re also based on the prior year’s list
of health insurance plan names though we do update with the latest exchange plans for
the August release. And we also don’t attempt in the early release
reports to distinguish between individual types of public programs so for instance trying
to distinguish between Medicaid and the Children’s Health Insurance Program. Despite the fact that these estimates are
preliminary there are generally close to the final estimates. Our comparisons indicate
that they are within .1 percentage point of for the proportion of uninsured and within,
you know, two or 3/10 of a percentage point for estimates of private and public coverage. So what’s coming? Well next week on September
1 we will be releasing our report on the 15 key health indicators based on data from the
first three months of 2015. Now one of those key health indicators is
health insurance but it is just a subset of the information that’s in the report that
we released two weeks ago. There’s nothing new on health insurance in next week’s report. However next week’s report does contain a
number of variables that are related to health insurance. So for instance the proportion
of persons who have a usual place for healthcare and the proportion of persons who have had
financial problems obtaining care. Then as you’ve heard on September or on September
16 jointly with the Census Bureau we will be releasing an updated comparison of the
NHIS and CPS point in time estimates. We did this last year comparing the estimates
from the January through March NHIS to the CPS point in time estimates which are from
February through April. Last year we didn’t find much in the way of
significant differences. We’ll be updating those estimates and roughly two weeks I guess
it is. And then as I’ve already mentioned the next
early release of health insurance coverage estimates will be in November presenting data
from the first six months of 2015. If you’re looking for these reports the best
place to go is the National Health Interview Survey Web site. That address is at the top
of this slide. Typically you can find our most recent reports
over in the What’s New column on the right. But if you’re – if it’s not there or you’re
looking for an older report you can go where that red arrow is pointing on the left-hand
side to the NHIS early release program, click there and you’ll get to the early release
program Web site. And the second bullet on that page refers to the health insurance reports. Also if you want to receive announcements
via email about our early releases and other HIS data releases you can join our Listserv
using the address on this slide. Thank you. Alfred Gottschalck:Good morning and thank
you for taking the time today to meet with us. I am the Assistant Division Chief for
Small Area and Longitudinal Estimates the area that oversees the Small Area Health Insurance
estimates SAHIE program. Today I would like to provide an update on
the current SAHIE production plans concerning our next SAHIE release. SAHIE is one of the estimates of health insurance
that the Census Bureau releases. And for the 2014 SAHIE estimates we have some changes
to consider and we would welcome your feedback. First I would like to provide a brief overview
of the SAHIE program. The Census Bureau releases model based health insurance coverage estimates
for all US counties on an annual basis through the SAHIE program. We also release health insurance coverage
estimates for states as well with race and Hispanic origin detail. We produce SAHIE estimates because they provide
the only single year estimates of health insurance coverage for every county in the United States. We use a model to produce estimates that typically
have lower variances than the survey estimates. This past March we released the 2013 SAHIE
estimates. The SAHIE program is partially funded by the
Centers for Disease Control and Prevention’s, National Breast and Cervical Cancer Early
Detection Program. The CDC have a congressional mandate to provide
such screening services through this program. The CDC has been a strong partner over the
years and we are very grateful for their support. In addition SAHIE data are also used by other
government agencies and researchers interested in examining health insurance coverage. SAHIE data represents ACS health insurance
coverage estimates that are enhanced with administrative data to create the model based
estimates of health insurance coverage. SAHIE has used ACS data as a base since 2008
hence SAHIE reflects annual changes over time from 2008 to 2013. SAHIE data can be used to analyze geographic
variations in health insurance coverage across states and counties. Furthermore SAHIE data can be used to examine
differences in coverage by race or ethnicity, sex, age and income levels that reflect thresholds
for state and federal assistance programs. To demonstrate one of the key strengths of
SAHIE data its geographic coverage please see this map. This map shows that shows the
counties that are and are not published with 2014 ACS one year estimates. The green areas are counties that are available
with ACS one year estimates and the white areas are the counties not available via ACS
one year estimates. Given SAHIE ‘s statistical power SAHIE is
able to provide estimates for every county in the country on an annual basis. ACS publishes one year estimates for geographies
with population 65,000 or greater. This covers approximately 20%, 26% of all counties or
85% of the total population. With five your estimates ACS provides data
for all geographies. The next ACS five your release will occur this December and cover
the period 2010 to 2014. Several types of data are used to create SAHIE
. The input for SAHIE are data from the American Community Survey both one year and five year
estimates, data from Census 2010 as well as data from the Census Bureau’s population estimates
and County business patterns. We also use other data in the model such as
information from tax returns, supplemental nutrition assistance programs participation
records and Medicaid and children’s Health Insurance Program participation records. So how does the SAHIE model work? We combined
ACS one year published and unpublished estimates with model estimates using statistical techniques
which weight the relative contribution of the two components based on their relative
precision. If the ACS estimate has a smaller variance
for example due to a large sample size it contributes more to the final estimate. In
this case the survey data have high relative precision compared with the model estimate. Otherwise the model estimate has more weight
since modeling produces more reliable and stable estimates we can publish one year SAHIE
estimates for all counties every year. Now I would like to discuss 2014 SAHIE in
relation to the changing healthcare landscape. States can choose whether or not to expand
their eligibility criteria for Medicaid participation. For 2014 SAHIE we need to rethink the inputs.
Typically the Medicaid data used in the model are lagged by a few years. For instance the SAHIE estimate that we released
this past March used Medicaid data from 2011. We used lagged Medicaid data because it is
a later latest data set available that has all the information we need for the model. For the 2014 SAHIE estimates we used lagged
Medicaid data using for 2014 SAHIE estimates using lagged Medicaid data may present an
issue. Obviously using 2012 Medicaid data to model
health insurance for 2014 may be a problem. We expect that health insurance coverage will
change due to the change in Medicaid eligibility. So we really need to ensure that the data
in our model can capture this change. Currently 28 states and the District of Columbia
have chosen to expand — indicated in the light color in the map — and 22 states have
not expanded indicated in the dark color of the map. Given this change in Medicaid eligibility
it is important for the SAHIE program to effectively incorporate the most recent available data
on Medicaid coverage into the SAHIE model. Consequently we need to rethink where we get
the Medicaid data or rethink the data that go into our model. Specifically we need to evaluate alternative
Medicaid data sources. Currently SAHIE uses data from the Medicaid Statistical Information
System or MSIS. These data contain information of all those
eligible and receiving services under the Medicaid and CHIP programs for every state
and territory. And we are very familiar with the characteristics and features of these
data given we have used these data for many years. T-MSIS will be replaced by the Transformed
Medicaid Statistical Information System or T-MSIS. This transition is occurring on a
phased basis as states begin submitting data under this new system. As more states begin submitting their data
under this new system we will have a better understanding of the characteristics relative
to the prior MSIS system. As a result we are currently evaluating three
other data sources that provide state Medicaid enrollment data. The first is the Medicaid and CHIP application
eligibility determination and enrollment data also from CMS, the Medicaid enrollment data
collected through the Medicaid budget and expansion system again from CM and Medicaid
and CHIP enrollment data snapshot reports from the Kaiser family foundation. To summarize we need to effectively capture
the changes in Medicaid eligibility. We need to address the lag in our Medicaid input data. And lastly as the transition from MSIS to
T-MSIS continues we need to gain an understanding of the difference between the two systems. So our challenge is to evaluate the underlying
data to better understand the strengths and weaknesses of each data source. Once we more fully understand the characteristics
and features of each data source we then must determine the impact on the model estimates
of making the change to Medicaid data component on the SAHIE model. Given the outcome of these
evaluations we will then determine whether a modeling change is warranted. Consequently we have some options to consider
this current production cycle. Our first option would be to make no change in the Medicaid
data source where we would use the same data source and lagged. This data would be from
the 2012 T-MSIS. Another option would be to change the source
Medicaid data this production cycle as justified by our evaluations previously mentioned. Lastly we could use our evaluations to inform
the decision as the Medicaid data source for 2015 SAHIE and beyond. In regards to upcoming SAHIE release products
2014 SAHIE data will be released in 2016. If a change in modeling is made for the 2014
SAHIE data an updated version of 2013 SAHIE data will be released for comparison purposes. Lastly detailed documentation of any new data
sources and modeling will be released with the 2014 SAHIE data along with our data interactive
tool. For further information concerning the SAHIE
program please see our SAHIE Web site at census.gov. You may also contact us via email. We welcome any feedback you may have concerning
the issues I have outlined today. Please direct this feedback to our SAHIE email address.
Thank you for your time and interest in our health insurance data. Michael Cook:Thanks Al. That concludes the
presentation from the second half of our technical meeting this morning and so we’ll go ahead
and open up the lines for those that are watching online and also those in the room if you have
any questions we’ll take them now at this point in time. Operator do we have any questions on the phone? Operator:On the audio portion to ask a question
please press Star 1. And just one moment please. Sir I am showing no questions at this time. Michael Cook:Do we have any questions in the
room? As we wait to see if anybody is trying to login and prepare themselves for questions
in the room we’ve seen it before but I’ll show it again the timeline of upcoming releases. We have the income and poverty in the United
States, health insurance coverage in the United States and the sub-community poverty measure
for 2014 all being released September 16 at 10 o’clock via Web cast. They will include detailed tables of income
poverty and health insurance coverage from 2014, the public use file and also the measure
of current health insurance coverage release with the National Center for Health Statistics. And then that same day at 12 o’clock the ACS
American Community Survey one year estimates including information on the nation, states
and all geographic areas over 65,000 will go into embargo at noon. Let embargo will last until 12:01 AM on September
17 where that information will be released and made available publicly. Also forthcoming we have income consistent
research file based on the full sample and model income, Americas families for 2015 on
the horizon, geographic mobility from 2014 to 2015 and then as Al mentioned the Small
Area Health Estimate Insurance Estimates for 2014. Operator do we have any questions on the phone? Operator:I am showing no questions at this
time sir. Michael Cook:And are there any questions in
the room? Yes go ahead sir. Donald Larch:This is a question related to
the SAHIE estimates and data sources that you use. I know that historically you’ve used
data from IRS aggregate data. Are there plans to secure the data on recipients
of the premium tax credit in that aggregate data which would help identify low income
households with insurance through the marketplace? Alfred Gottschalck:No. I’m not aware of any
of those plans no. Michael Cook:Thank you for that question.
Any other questions in the room? And operator do we have any questions on the phone? Operator:I am showing no questions sir. Michael Cook:In conclusion I’d like to remind
folks that are watching online as well as those in the room that if you are from the
media you can contact the public information office for more information. You can contact
us by dialing 301-763-3030 or by email at [email protected] If you are nonmedia I ask that you contact
our customer service center by dialing 1-800-932-8282 or 301763 info. I’d like to thank everyone all of our panelists
and discussants and you for tuning in online and being here in the room for participating
in today’s Web cast on the upcoming releases of income poverty and health insurance coverage
estimates from federal sources. This concludes our technical meeting. Thank
you everyone. Operator:Thank you. And this does conclude
today’s conference. You may disconnect at this time.

Tags:, ,

Add a Comment

Your email address will not be published. Required fields are marked *