Introduction for Clinical

Database
陳勁辰
2003/06/02
 Introduction
 Database management system (DBM
S)
 Actual study information

 Administrative information

 Relational database model

 Relational database model
 Multiple tables
 Entities: rows

 Attributes: columns

 Primary key

 Foreign key

 Principle of table normalization

 Relational database model
 Example: Infant Jaundice Study: whe
ther neonatal jaundice affects neurop
sychiatric scores at five years of age
 Table relations: one-to-many, many-
to-many, and one-to-one
 Need for a multi-table: repeated mea
surements on individual subjects.
Jaundice vs non-jaundice: t-test p=0.46
One-to-many
 Normal Form Violation
 Repeating columns: violate the First
Normal Form (1NF)
 Redundancy in column values: violat
es the Second Normal Form (2NF)
Many-to-many
One-to-one
 Referential integrity
 Referential Integrity in a normalized, relati
onal database
 Not allow creation of an exam record for a
subject who does not already exist in the “
Baby” table, and it will not allow assigning
an exam to a doctor who does not already
exist in the “Doctor” table.
 A subject may not be deleted unless and u
ntil all that subject’s examinations have al
so been deleted.
 One-to-many: parent-children
 Forbids the creation of “orphans”
Undesirability of Storing Calculated
Values
 Inconsistencies result if one of the “r
aw-data” fields is updated without up
dating the calculated field
 Solution: recalculating the value in a
query
 Storing calculated values such as “Ag
eInMonths” : violate the Third Norma
l Form (3NF)
Data Dictionaries, Data Types, and
Domains
 Data dictionary: a table of informatio
n about the database itself
 “Metadata”

 Data types: text, number, dates, etc.

 Domain: range of allowed values

 Object data type: BLOB (Binary Larg

e Object), eg. photo (*.jpeg); cannot
be sorted or sought
Extracting Data from the Database
(Queries)
 Structured Query Language or SQL
 A query can join data from two or m
ore tables, display only selected field
s, and filter for records that meet cer
tain criteria
 SQL has 3 sublanguages: DDL – Dat
a Definition Language, DML – Data M
anipulation Language, and DCL – Dat
a Control Language
 SQL
SELECT Baby.SubjectID, Baby.DOB, Ex
am.ExDate
FROM Baby INNER JOIN Exam ON Bab
y.SubjectID = Exam.SubjectID
WHERE Exam.ExDate Between #1/1/2
010# And #2/28/2010#
ORDER BY Exam.ExDate;
Guidelines for Database Managem
ent for Clinical Research
1. Establish the database tables, their rows and col
umns, and their relationships correctly at the out
set.
2. Establish and follow naming conventions for colu
mns and tables.
3. Obtain baseline demographic and clinical informa
tion about members of the study population from
existing computer databases.
4. Minimize the extent to which study measuremen
ts are recorded on paper forms.
5. Follow standard data entry conventions.
6. Back up the database regularly.
Guidelines for Database Managem
ent for Clinical Research
 Establish the database tables, their
rows and columns, and their relatio
nships correctly at the outset.
 A poorly organized database makes
data maintenance and retrieval nea
rly impossible. Make sure the data
are normalized. Avoid data structur
es that require duplicate data entry
or redundant storage.
Guidelines for Database Managem
ent for Clinical Research
 Establish and follow naming conventions
for columns and tables.
 Short field names without spaces or unde
rscores are convenient for programming,
querying, and other manipulations. Inst
ead of spaces or underscores, use “Intra
Caps” (upper case letters within the varia
ble name) to distinguish words, e.g. “Stu
dyID”, “FName”, or “ExamDate”. Table n
ames should be singular, e.g. “Baby” inst
ead of “Babies”, “Exam” instead of “Exa
ms”.
Guidelines for Database Managem
ent for Clinical Research
 Obtain baseline demographic and clinical inform
ation about members of the study population fr
om existing computer databases.
 Avoid re-entering data which are already availa
ble (in digital format) from other sources. In th
e Infant Jaundice Study, the patient demograph
ic data and contact information are obtained fro
m the hospital database. Computer systems ca
n almost always produce character-delimited or
fixed-column-width text files that the database
management system can import.
Guidelines for Database Managem
ent for Clinical Research
 Minimize the extent to which study measureme
nts are recorded on paper forms.
 Enter data directly into the computer database
or move data from paper forms into the comput
er database as close to the data collection time
as possible. When you define a variable in a co
mputer database, you specify both its format an
d its domain or range of allowed values. Using
these format and domain specifications, comput
er data entry forms give immediate feedback ab
out improper formats and values that are out of
range. The best time to receive this feedback is
when the study subject is still on site.
Guidelines for Database Managem
ent for Clinical Research
 Follow standard data entry conventions.
 Several conventions for data entry and display h
ave developed over time. Although most users o
f screen forms are not aware of these convention
s, they have come to expect them subconsciousl
y. For example, a series of mutually exclusive, c
ollectively exhaustive choices is usually displayed
as an “option group” consisting of several differe
nt “radio buttons”, whereas choices which are no
t mutually exclusive are displayed as check boxe
s
Guidelines for Database Managem
ent for Clinical Research
 Back up the database regularly and
check the adequacy of the back up
procedure by periodically restoring
a file from the back up medium.