Synthesis Analysis is a statistical procedure used to
combine regression coefficients for a common outcome or
disease, which have been extracted from independent
research studies. BioSignia has applied Synthesis Analysis
statistics to the development of evidence-based disease
risk models and has improved existing disease risk models
with recent epidemiological findings.
Synthesis Analysis is different in purpose from
meta-analysis, in that meta-analysis can only combine
similar findings from independent studies if the variables
are the same. Meta-analysis summarizes from many research
studies the effect of a single risk factor on a specific
disease.
Synthesis Analysis combines the research on multiple risk
factors for a single disease. For example, many studies
have shown that high blood pressure (hypertension) can
lead to heart attack. A meta-analysis might combine all
the relevant research on the effect of hypertension for
heart disease and may conclude that hypertension increased
the risk of heart attack four-fold. Another meta-analysis
might summarize all of the studies on the effect that
smoking has on heart disease. They in turn might conclude
that smokers have a three-fold increased risk of heart
disease.
Synthesis Analysis, on the other hand, combines risk
factors such as hypertension and smoking into one multiple
risk factor equation for heart disease. Synthesis Analysis
builds on meta-analyses. The problem of combining separate
risk factors is that people who have one risk factor may
also have another. In the example above, smokers tend to
have higher blood pressure.
If these factors were combined without accounting for this
overlap the assessment of risk from the combined factors
would be overstated. It might be concluded that smokers
with hypertension have a twelve-fold increased risk
(3x4=12) over non-smokers with normal blood pressure. In
actuality, the risk for this group is most likely closer
to six or seven. Synthesis Analysis adjusts for the
collinearity (overlap) of risk factors so that they can be
combined without overestimation of risk.
|