This is the presentation I gave describing the new database I created to help Baltimore City Health Department workers track city mortality statistics.
1. The New Baltimore
City Health Statistics
Database
By Nonye Madu
Summer Epidemiology Internship
September 2, 2014
2. Purpose of the new database
• The new database was created to not only
streamline data entry, but also to increase
the utility of the data that exists in the
database for reporting purposes.
• The old database, while functional for data
entry, was not suitable for generating
reports due to shortcomings in the design.
7. Changes in the new database
• Field types were changed to more accurately reflect the type
of data being entered. This process involved extensive data
cleaning.
Old Database New Database
The ages, sex codes, and race codes were converted to numeric values, and the date
of death text field was converted to a date field. New SexValue and RaceValue fields
were added to display the decoded sex and race values, respectively. These values
are generated upon making a sex or race selection and are not visible on the data
entry form.
8. Changes in the new database
• Links to data entry forms for data before 2010 were consolidated on
the Main Menu screen by decade. The Main Menu in the previous
database was constructed using the Switchboard function, which
only allows for 8 options per screen, thus leading to the use of
arbitrary time intervals for each screen.
Old Selection Screen for 1998-2004 New Selection Screen for 1991-1999
9. Changes in the new database
• All of the tables with data from 2010 (the oldest year for which
paper records are available) onwards were combined into one
table to allow for more efficient aggregate queries to be
generated. All future data will also be entered in this new table.
New Combined Table
Old Tables Separated by Year
10. Changes in the new database
• The data entry forms were updated to more accurately reflect
data contained on Maryland death certificates.
Old Form
11. Changes in the new database
New Form
New fields specifying the manner of death, the lethal agent in the case of an overdose, and the zip code where the
death occurred were added. The census tract code year was changed from 1990 to 2010 to reflect the most recent
census year. The race categories were also changed to reflect the options typically recognized on Maryland state
death certificates. In addition, the age unit field was edited so that individuals over 100 are not counted separately.
13. New Reports
• Several reports were added should the
data entry specialist be interested in
discerning particular trends within the
mortality data. Due to potentially differing
quality control measures, these numbers
are unofficial and may not exactly match
figures generated by the Department
epidemiologists, though they should be
relatively similar. Reports can be generated
for any date range from January 1, 2010.
14. New Reports
• Listing Reports: basic listing of all deaths in a specified date range.
Listing reports can be restricted by race, gender, or manner of death.
15. New Reports
Sample of a listing report showing all deaths ruled as accidents during
the specified date range (certificate numbers omitted for privacy).
16. New Reports
• Count Reports:
aggregate mortality
statistics reports that
can be grouped by
race, age, or gender.
Also included is a
report that lists the
top ten causes of
death and a report
showing the number
of overdose deaths
by zip code.
17. New Reports
Sample of a count report showing the total number of deaths and
leading causes of death by race during the specified date range.
18. New Reports
Sample of an overdose
report showing the
number of overdose
deaths by the zip code
where the death
occurred.
19. New Reports
Sample of a report
showing the top ten
causes of death during
the specified date
range.
20. New Reports
• The specific cause of death for each individual
is generated by a query linking the ICD-10
code specified in the mortality data table
with another table containing decoded ICD-
10 values. All of the codes are not
represented, but codes were selected in a
way as to capture the vast majority of
mortality causes. If needed, the table with
the decoded ICD-10 values can be used to
generate reports about any cause of interest.
21. New Reports
Example cells from the “Death Codes” table with decoded ICD-
10 values. This table was created in Microsoft Excel and
imported into Microsoft Access.
22. New Reports
The following query was used to link the ICD-10 code in the mortality
data table with the ICD-10 code in the imported table with decoded
values:
SELECT tbl_ALLDeathsSince2010.*, [tbl_ICD].SpecCause
FROM tbl_ALLDeathsSince2010
LEFT JOIN tbl_ICD
ON tbl_ALLDeathsSince2010.CAUSE Like [tbl_ICD].ICDCode & '*';
tbl_ALLDeathsSince2010 is the table with the mortality data and tbl_ICD is the
table imported from Excel containing the decoded ICD-10 values. The code is
written so that all derivatives of the specified ICD-10 codes are captured (for
example, both C153 and C505 are classified as malignant neoplasms in the query
results despite malignant neoplasms only being denoted as the letter “C” in
tbl_ICD). This query is saved under the name qryICDecode, and is referenced by
most of the count reports.
23. New Reports
• In addition to being printable, all of the
reports can be exported from Microsoft
Access to Microsoft Excel, allowing for use
in other applications such as SAS and
ArcGIS.
24. With this new database,
hopefully all interested parties
can gain new insight about
mortality in Baltimore City.
25. Thank you for your attentionand best
wishes going forward!
Nonye Madu
2015 MPH Candidate
Drexel University School of Public Health
crm99@drexel.edu