|
Interactive Atlas of Reproductive
Health: Frequently Asked Questions |
|
After reading the FAQ's below, if you have additional
questions, please Contact Us for more information and select
"Reproductive Health Atlas" in the subject drop-down box.
Overview
The
Reproductive Health Atlas gives users the capability to make
direct comparisons of data using indicators that span time,
geography and population groups. Users who are not familiar with
either GIS technology or statistical methods should understand
the benefits and limitations of using the Atlas
Reading this section of the Web site will help new users become
comfortable with the Atlas and its features. A few examples of
actual Atlas use are also included. Also included in this
section are “tips” on how to use the Atlas for specific tasks.
To contribute to the "tips" section, please
contact us.
New Users:
Q:
Why have a Reproductive Health Atlas?
A:
Program managers in public and private organizations
noted difficulty in acquiring reliable indicators of key
reproductive health events so the Atlas was developed using
the innovative GIS technology. The Atlas can automatically
generate maps, tables, and charts with data specific to
users’ needs.
The Atlas can perform complex statistical and data
management methods and procedures. The Atlas interface (this
Web site) lets users retrieve, review, and analyze data with
little training or experience.
Q: Why is the
Atlas important for me?
A: This Atlas allows users to compare data across time, jurisdictions,
geographic locations, and populations.
Q: How has this
Atlas been used?
A: Two
examples come from our pre-release test users. These are people either working in
reproductive health settings or have responsibility to plan health programs.
Here are some examples:
A user wanted to generate tables and charts for their state, with a
focus on teen pregnancy in a specific population. As they reviewed the
data, it was apparent the target population resided far away from reproductive
health service centers. Using the maps and tables, this can now determine
whether to establish new facilities or collaborate with existing programs
and coalitions to deliver health services.
A women’s organization wanted to support local women’s health promotions.
A member used the Atlas to generate information about key indicators
in the state and in several counties in a neighboring state. The Atlas
helped identify priorities for community outreach and education for
affiliates in each county.
Examples of use:
Q: I get different
information when I use my county health department's health statistics Web
site. Why?
A: We used data from a national source, the CDC's
National Center for Health Statistics
(NCHS). Local health departments report data to the state/territorial
health department, which in turn reports data to NCHS. This creates
a delay in reporting national health statistics.
Also, some data at a local or state level may be considered “provisional”
meaning data might be updated or corrected after initial reporting.
The Atlas uses the best available NCHS datasets as these are considered
“final” and likely to have minimal errors.
Consider the Atlas as one of several sources that help inform users
of health trends and problems. It is not intended to replace publications
and information from local and state health officials, but rather expand
access to and understanding of data sources.
Q: I work in
a state legislator's office. Do I need any special software to use this
service?
A: No.
The data and applications reside on CDC's servers. Users do not need
additional software unless exporting images or text to other computer
programs.
Q: Because of
my work, I know there were two infant deaths in my area during one of the
years listed on this Atlas. Why am I not able to see these cases reflected
in my Atlas table?
A: A
key feature of the Atlas allows users maximum flexibility in comparing
and contrasting key reproductive health indicators. To support this,
complex statistical and data management methods "smooth"
these data and allow comparisons, even when there are relatively few
events in a place, time, or group.
Users will find the Atlas’ statistical and data management systems provide
“rates” rather than the number of actual events. This not only protects
the privacy of an individual but also provides more accuracy for comparisons
of time, geography, and population groups.
Experienced Users:
Defining the
analysis
Q: What is
the map layer (step 4)?
A:
The map layer is the collection of map
features—such as counties—that
comprise the geographic area defined by the map view and target.
These features become the units for the spatial analysis.
Q: What is
the map view and target (steps 5 and 6)?
A:
The view and target geographically
define the population to be analyzed, as well as define the visual
map area to be displayed. View defines the type of area such as
states or regions, and target defines the specific area such as
Texas or New England.
Q: Why do
the selections in subsequent steps revert back to the default if I change
my selection in a preceding step?
A:
The dataset selection procedure is based on a hierarchical model,
therefore selection options in subsequent steps are dependent on
choices made in the previous steps. For example, if you choose states
as your analysis unit then state is not a valid choice as the data
set. This is why following the steps in numeric order is important.
Interpreting the results
Q: Why does
the spatial pattern of the map change so much when I change the map
classification from equal interval to quantiles?
A:
The pattern revealed is a function of the distribution of the rates.
This is particularly obvious when viewing large geographic areas
that are divided into many small units such as a view of a census
region divided into counties. If the rates are distributed normally,
then the patterns revealed using either method are very similar.
However, if the rates are skewed, the cutpoints and patterns will
reflect the nature of the skew.
For the
quantile method, the units are divided into equal numbers of units.
The cutpoints will be based on the maximum rates for each group,
and the pattern will have an equal number of members for each classification
group based on the hierarchical rank of each unit. The result is
that all classification groups are represented on the map, but the
magnitude of the ranges of each classification group may be very
different depending on the distribution of the rates. For example,
if a state has 120 counties and the map legend has 4 quantile classifications
then there will be 30 counties in each group, but the range of the
rates in the groups may be 5 for the first group and 10 for the
last group.
Alternatively, for the equal interval method, the cutpoints are
based on equal intervels between the maximum and minimum rates of
the entire group. Therefore, the magnitude of the range is the same
for each group, but all groups (except for the highest and lowest)
may not be represented. For example, if the minimum rate is 4.0
and the maximum rate is 12.0 then the cutpoints would be 6.0, 8.0
and 10.0, but if most of the rates fall between 4.0 and 7.0, the
last classification group will have only a few units in it.
Care must be taken when viewing map results to appreciate these
characteristics and interpret the patterns accordingly.
Q: What is
the difference between no events and no population?
A:
Events and population represent the numerator and denominator of
the rate calculation respectively. For example, to get an infant
mortality rate, a summary count of events (infant deaths) within
a geographic area (Dade county) and time span (2002) are summed
and divided by a count of the relevant population (liveborn infants)
in the same geographic area (Dade county) and time span (2002),
and multiplied by 1,000 (or other power of 10) to get a rate. In
this example, no events would tell us that among all the liveborn
infants born in Dade county in 1995 there were no infant deaths.
Whereas, no population would indicate that there were no infants
born in Dade county during 2002.
Q: Why do
most of my counties display insufficient data?
A:
Insufficient data means that although there were relevant events
that occurred, the relative standard error (RSE) generated for the
rate equation indicated the rate was statistically unstable. This
check is done to avoid displaying data that may vary wildly from
time period to time period—usually due to small numbers in the numerator
(events)—and therefore would not represent a true and fair evaluation
of that indicator for the chosen time and area.
Q: How do
I get around insufficient data?
A:
There are several options for getting around insufficient data.
First, insufficient data usually indicates small numbers in the
numerator of the rate calculation, but is also dependent on the
denominator; therefore, the goal is to increase the denominator.
This can be done by increasing the time period from one year to
cumulative years, or increasing the geographic area (dataset choice)
of the analysis. Another option available when analyzing counties
is to use smoothed counties.
Q:
What are smoothed counties?
A:
Smoothed counties are county-level rates that have been spatially
adjusted by their neighboring counties. The rate actually reflects
the rate of the county plus its neighboring counties. The technique
is similar to that of using moving averages over time except that
the smoothing occurs spatially rather than chronologically.
Q: How can
I compare demographic groups when the legend ranges and map classifications
change for each demographic group that I choose?
A:
You can compare maps representing two different demographic
groups using the legends and visual perspective. The legend gives
you the range of rates as well as the highest and lowest rate. Comparison
of the overlap (or non-overlap) of the two ranges may yield an indication
of demographic disparity. The visual patterns revealed by the map
classifications indicate where high and low rates occur geographically.
Overlap may indicate service problems. Non-overlap may signal demographic
disparity.
Q: Can I
interpret that the area with the highest map classification is undesirable
for some reason?
A:
You must take care to interpret the legend ranges and geographic
patterns in context with the area and national rates displayed below
the map. The highest map classification may actually be below the
national rate or the Healthy People 2000 target rate. Also,
a relatively narrow range between the highest and lowest rate on
the legend indicates that the rates are not significantly different
and that the geographic pattern will probably not be consistent
over time.
Q: Can I
rank the map features by their rate?
A:
Yes, you can. However ranking is to be discouraged because rates
are often not significantly different from each other, and in some
cases, even the highest and lowest rates are not significantly different.
A better choice for comparison are the Healthy People 2000
and 2010 target rates, the area rate, or the national
statistics located below each map.
Q: Why is
the smoothed county rate for Washington, D.C., so different from the actual
county rate and the state rate?
A:
Washington, D.C., occupies a unique position in the data because
it is the only county-equivalent that is also a state-equivalent
and is surrounded by other counties. Because D.C. is considered both
a county and a state in this data set, the county rate always equals
the state rate. There are other instances of county-equivalents
that are also state equivalents in the data such as Guam or American
Samoa, but these island territories do not have surrounding counties.
Because D.C. is surrounded by other counties, the "smoothed" county
rate is affected by the other counties and is often significantly
different from D.C.'s actual state-county rate.
Page last reviewed: 7/28/08
Page last modified: 1/29/07
Content source: Division
of Reproductive Health, National
Center for Chronic Disease Prevention and Health Promotion
|
|
|