Section 6: Other Considerations for Recommendations
6.1 Consideration of Starting and Stopping Times for Screening
In considering the issues of starting/stopping points, the Task Force believes there
are weaknesses in the system of using a chronologic age cutpoint. Clearly, factors beyond
chronologic age (including risk of the target condition, risk of harms from the preventive
intervention, and efficacy of the intervention) are involved in determining when screening
an individual should begin or stop.
The Task Force developed a list of factors to be considered when recommending starting
or stopping times. These are given in
Appendix X.
6.2 Consideration of Screening Interval
Because clinicians are very interested in the recommended interval for screening tests
and yet the evidence for one screening test compared with another is often lacking,
the issue of what interval to recommend can be problematic. For some topics a decision
analysis may be used by the Task Force to help inform its recommendation (go to
section 5.4.3). The Task
Force may not recommend an interval when the evidence is insufficient. It has, however,
adopted a set of factors to consider in determining whether to recommend an interval.
These factors are given in
Appendix XI.
6.3 Outcomes Tables
The Task Force adopted the concept of having "outcomes tables" prepared by
the topic team to assist the Task Force in deliberating about the potential net benefits
and evidence gaps (Appendix IX).
The topic teams are encouraged to provide quantitative information in such metrics as
"number needed to treat" (NNT) or "number needed to screen" (NNS)
where this is possible; to provide ranges of values where this is most precise; and
to write "not available" where this is necessary. The tables are clear about
what is known, at what level of precision, and what is not known.
The Task Force recognizes that the NNT or NNS metric does not itself allow comparison
of benefits across topics, as the outcomes and times involved are quite different. It
does provide, however, a useful first step in making sure the Task Force is as clear
as possible about the actual effects of a proposed intervention.
Although the Task Force understands that complete standardization of Outcomes Tables
is impossible, it encourages the use of the standards in
Appendix IX where possible.
6.4 Extrapolation Policy
"Extrapolation" is a term that is sometimes used synonymously with
"generalizability." To the Task Force, however, these terms have different
meanings. While generalizability is a synonym for external validity, extrapolation
refers instead to a separate question in the Task Force's interpretation of evidence.
The issue raised by extrapolation is the extent to which the Task Force can make
inferences across evidence gaps within the Analytic Framework to reach a complete
chain of evidence connecting the population with health benefits. Examples include
whether the Task Force judges that it can extrapolate evidence about intermediate
endpoints to health endpoints. Another example is the question of whether the Task
Force can judge long-term health outcomes based on shorter-term outcomes. Still another
example is whether the Task Force can judge the effects of screening based on RCTs
of treatment. In these cases, the Task Force is not considering the question of
whether a study applies to a different population, situation, or provider. Instead,
it is judging whether a gap in the evidence within the Analytic Framework can be
overcome with logic and biologic plausibility. These 2 factors—logic and
biologic plausibility—play the greatest role in the decision about extrapolation.
In general, the Task Force extrapolates only when the case for extrapolation is strong.
Return to Contents
Proceed to Next Section