Quantitative signal detection
The basic principle of quantitative signal detection is to fi nd combinations of birth defects and specifi c
drugs or drug classes that are more frequently present than expected in the database compared to
other birth defects, by calculating cross tabulations between selected birth defects and selected drugs.
Inclusion criteria
Birth defects are included if they present either as isolated defects or in combination with other, but
unrelated, defects (without an overall or syndrome diagnosis) . In the latter situation, the defects are
analysed as separate defects. Defects associated with a chromosomal or single gene disorder are not
included. A birth defect associated with another defect (for instance, a clubfoot associated with a neural
tube defect) is only included in the primary birth defect category. Defects as part of a non-genetic
syndrome are only included in the syndrome category. Statistical considerations mean that only birth
defects occurring in 10 or more aff ected subjects from the study population are selected.
Only specifi c drugs and drug classes that are chemically related within the same anatomical and
therapeutic setting (fi rst 5 positions of the ATC code) are included if 10 or more subjects from the study
population were exposed to them in the fi rst trimester of pregnancy.
If a drug class and an underlying specifi c drug comprise the same number of exposed subjects
for a specifi c malformation or birth defects category, only the specifi c drug is included in the analyses.
The same applies to the malformations: if the more specifi c malformation (for instance, spina bifi da)
comprises the same number of subjects exposed to a specifi c drug as in the malformation group defi ned
more broadly (for instance, neural tube defects), only the most specifi c malformation is included in the
analyses.
Analysis
If a combination of a specifi c birth defect and drug or drug class contains at least three cases, the
Fisher Exact test is performed to test whether the number of observed cases diff ers from the number
of expected cases. The odds ratio (OR) and the 95% confi dence interval (CI) are calculated as a measure
for the strength of the signal. The OR is calculated as the ratio of the exposure odds among cases with
a specifi c malformation to the exposure odds among all other registrations with other birth defects. .
The basic principle of quantitative signal detection is to fi nd combinations of birth defects and specifi c
drugs or drug classes that are more frequently present than expected in the database compared to
other birth defects, by calculating cross tabulations between selected birth defects and selected drugs.
Inclusion criteria
Birth defects are included if they present either as isolated defects or in combination with other, but
unrelated, defects (without an overall or syndrome diagnosis) . In the latter situation, the defects are
analysed as separate defects. Defects associated with a chromosomal or single gene disorder are not
included. A birth defect associated with another defect (for instance, a clubfoot associated with a neural
tube defect) is only included in the primary birth defect category. Defects as part of a non-genetic
syndrome are only included in the syndrome category. Statistical considerations mean that only birth
defects occurring in 10 or more aff ected subjects from the study population are selected.
Only specifi c drugs and drug classes that are chemically related within the same anatomical and
therapeutic setting (fi rst 5 positions of the ATC code) are included if 10 or more subjects from the study
population were exposed to them in the fi rst trimester of pregnancy.
If a drug class and an underlying specifi c drug comprise the same number of exposed subjects
for a specifi c malformation or birth defects category, only the specifi c drug is included in the analyses.
The same applies to the malformations: if the more specifi c malformation (for instance, spina bifi da)
comprises the same number of subjects exposed to a specifi c drug as in the malformation group defi ned
more broadly (for instance, neural tube defects), only the most specifi c malformation is included in the
analyses.
Analysis
If a combination of a specifi c birth defect and drug or drug class contains at least three cases, the
Fisher Exact test is performed to test whether the number of observed cases diff ers from the number
of expected cases. The odds ratio (OR) and the 95% confi dence interval (CI) are calculated as a measure
for the strength of the signal. The OR is calculated as the ratio of the exposure odds among cases with
a specifi c malformation to the exposure odds among all other registrations with other birth defects. .
No comments:
Post a Comment