Society for
Epidemiologic Research (SER) Presidential Addresses
SER President
Discusses Issues Related to Epidemiology and Policy
Szklo Underscores
Importance of Biology in Epidemiology
“There is a need for “more
incisive” discussion of the concepts and methods of epidemiology
relevant to policy formulation. This was the theme of the presidential
address at the recent SER meeting in Birmingham. Outgoing president
Moyses Szklo presented his views on some of the
issues related to the interface between epidemiology and policy and on
the importance of biological thinking in all phases of epidemiologic
research.
Why Discuss?
In explaining the need for such
discussion, Dr. Szklo noted that the contributions of statistical
methods to epidemiology, which have added enormously to the growth of
epidemiology, have received considerable attention in graduate
programs. However, the specific relevance of these methods to policy
recommendations has not received comparable attention. He attributed
this to the fact that epidemiologic expertise in recent decades has
tended to move away from service-oriented agencies such as health
departments and has gravitated to academic institutions where the
pressure to develop policy is less.
Issues Worth Discussing
In his view, the well known
concepts and methods of epidemiology... “acquire a special meaning
when seen in the specific context of the policy implications of
research.” Among the examples he cited was in the area of determining
causality. Identifying a truly causal association is important for
establishing etiology and for designing primary prevention strategies.
However, perhaps less well appreciated is the fact that it is also
important for epidemiologists to help distinguish between valid,
non-causal associations and truly spurious ones. The former may be
useful for identifying risk groups for secondary prevention purposes
such as screening.
Replicability
Dr. Szklo also focused on the
importance of replicability in establishing whether or not reported
associations are truly causal. Noting the potential contributions of
meta-analysis in this regard, he warned the audience that published
studies may not be representative of all findings because of the
documented bias towards publishing positive findings. One remedy
proposed for this problem is the establishment of registries which
would track studies regardless of outcome.
Another warning he gave was that
discrepancies between study findings were not necessarily evidence of
lack of association. Study populations are not homogeneous. In
considering policy recommendations, prevention strategies might be
able to be tailored to subgroups at genuine increased risk.
Models
Dr. Szklo criticized the
assessment of interaction only for statistical modeling purposes. He
underlined the importance of the additive model of interaction for
public health purposes and proposed that its effects should be
systematically reported regardless of its usefulness for other
methodologic purposes.
Weak Associations
Weak associations are difficult
for epidemiology to deal with, yet their importance may be
considerable in public health terms when disease is relatively common
and exposure widespread. Meta-analysis can help by pooling results
from several studies; however, small sample size is also a risk factor
for non-submission or non-publication of results. “...Clearly a
rethinking of the implications of insufficient statistical power due
to weak associations is in order in this era of meta-analysis,”
according to Szklo.
Epi and Biology
Dr. Szklo closed his talk by
emphasizing the need to strengthen biological thinking in
epidemiologic research. He identified several components of
epidemiologic work which can be enhanced by sound biological
knowledge. For example, problems in exposure misclassification which
tend to bias studies towards a lower risk estimate may result in the
failure to detect weak associations which could have public health
importance. Likewise, the failure of some reported risk factors to
explain a larger share of disease occurrence might be remedied in part
by more accurate classification of disease outcomes.
Kelsey Committee
Dr. Szklo noted that a committee
headed by Jennifer Kelsey of Columbia has been formed
to examine the need for strengthening biological thinking in
epidemiology. If this committee finds there is a need, he believes the
profession should move swiftly to discuss ways in which this can be
accomplished. This will be important to increase the credibility of
epidemiology, to improve its ability to establish causal
relationships, and ultimately to make more incisive contributions to
establishing health policies, he said.
Published July 1989
Postscript 2000
The decade that
followed this presidential address has seen the confirmation of the
importance of both biologic thinking and translational activities in
epidemiology. The extraordinary growth of molecular biology
and genetics has led to a
proliferation of genetic epidemiology studies aiming at elucidating
the genetic etiology of a number of disease areas, particularly those
representing the major killers, coronary heart disease and cancer. An
important challenge is that the focus has been on isolated genes or
polymorphisms, rather than on the more interesting gene-environment
interactions, which become of paramount interest given the fact that
pure genetic etiology cannot explain more than a small fraction of
disease in the population. Another challenge posed by the development
of genetic epidemiology studies relates to the strategy for prevention
at the population-level. Hopefully, identification of highly
susceptible individuals on a genetic basis will not lead to a shift
from the clearly superior population-based to the high risk strategy
in prevention.
The 1990s also saw
an increase in the use of epidemiologic data for policy purposes, with
the best example of this given by the anti-tobacco fight.
Meta-analysis has been increasingly used as an instrument for the
shaping of health policies and the development of recommendations
pertaining to public health strategies. An example is the recent
meta-analysis of results of studies dealing with treatment of nicotine
dependence, which has led to a series of recommendations from the
Surgeon General office published a few months ago. The strengthening
of the Cochrane database is yet another important development in the
implementation of evidence-based public health and medical actions.
In summary, the last
decade has seen strong movements towards integrating biologic thinking
into epidemiologic research and translating its findings into public
health action. From the latter viewpoint, epidemiology seems to be
moving back to its origins as a basic tool of public health, meant to
act as the scientific basis for the development of health policy.
|
