Evidence Required for Adoption of New Vector Control Methods in Public Health - Biological and Environmental Control of Disease Vectors - page 195

the intermediate entomological ef ects of
interventions are translated into the fi nal
epidemiological outcomes, as the ento-
epidemiological mode of action of a vector control
intervention. For example, in the case of IRS for
malaria control, Macdonald's formula for
malaria transmission (Macdonald, 1957)
provides not only a checklist of entomological
parameters (mosquito density, longevity,
feeding behaviour, etc.), but also the analytical
framework needed to identify and defi ne the
epidemiological modes of action of alternative
forms of vector control.
The next logical step is to list the various
biological and contextual factors that this mode
of action might be dependent upon, or
vulnerable to. For example, does it depend on the
host-preferences of the local vector species of
Anopheles ? Does it depend on where they tend to
rest after feeding? Does it assume that people
sleep inside houses, and that the houses have
walls? Then, for each parameter in turn, we
consider whether this factor might be dif erent if
the interventions were deployed in other ways
and in other contexts.
If we decide that, in practice, an inter-
vention is likely to encounter variation in a given
factor when used in a range of settings, then we
can consult the fi eld trial data to see whether it
includes observations of such variation within
or between trials. If so, we may be able to con-
clude that the intervention has been consistently
ef ective despite this variation. Conversely, we
may conclude that ef ectiveness is indeed
infl uenced or constrained by this factor, in which
case we will try to defi ne these constraints as
precisely as possible. In other cases, however, we
may fi nd that in the fi eld trials so far, this factor
has been deliberately held constant, or has been
naturally invariant. Here, we may have to
conclude that no fi rm recommendation can yet
be made about the expected ef ectiveness of the
intervention in settings where this factor is very
dif erent.
To illustrate practical examples of the
theoretical considerations discussed above, the
following paragraphs list some of the factors
relevant to ITNs/LLINs and to anti-larval
interventions. These are presented together with
notes on the available evidence about these
factors, and how this evidence has been
interpreted in the past.

11.5.1 ITNs/LLINs

Transmission intensity

At the time of the early ITN trials, there was a lot
of public debate about the epidemiological
consequences of reducing immunity to malaria,
and concerns that vector control might be less
benefi cial (and perhaps even counter-
productive) in conditions of the highest intensity
'saturation' transmission (Snow and Marsh,
1995; Trape and Rogier, 1996). Thus, when a
coordinated second wave of trials was being
planned, the trial sites were explicitly chosen to
represent the range of transmission intensities
found in Africa (Lengeler, 2004). This strategy
was eventually successful in confi rming that
ITNs would be ef ective in Africa across the range
of transmission conditions found in that region.

Coverage
The relationship between intervention coverage
and ei cacy is often non-linear, and a basic
question for all forms of vector control is whether
a minimum degree of coverage is needed for
ef ectiveness. In the case of ITNs, early
entomological studies (Lines et al ., 1987; Lines,
1996) had predicted that ITNs could have two
distinct ento-epidemiological modes of action:
(i) by isolating the sleeper from the local
mosquito population (personal protection); and
(ii) by reducing local vector population's mean
longevity (mass ef ect). This was later confi rmed
by both entomological and epidemiological
observations (Magesa et al ., 1991; Binka et al .,
1998). In the case of personal protection, we
would expect the epidemiological benefi ts to be
independent of overall coverage but confi ned to
the user; in contrast, we would expect the
benefi ts of the mass ef ect to be dependent on a
minimum level of coverage, but be shared by
everyone in the community. All but one of the
trials studied full-coverage community-level
interventions, and all showed an impressive
degree of protection from ITNs (Lengeler, 2004).
Unfortunately, however, the question of the
mass ef ect, and its relative importance
compared to personal protection, was not
addressed in a consistent way, and the outcome
was not clear. Some of the sites showed that a
strong mass ef ect was present, but did not