randomization to at least two or three treatment arms in order to assess
the therapeutic response to a single radical cure regimen.
In the instance of a trial including a reinfection control group, the groups
would resemble these in design and intent:
A) Blood schizontocide only for therapy of acute vivax malaria (relapse
and reinfection control).
B) Blood schizontocide plus high-dose primaquine (reinfection control).
C) Blood schizontocide plus standard primaquine (evaluation arm).
This approach assumes that the blood schizontocide(s) has complete efficacy
in all arms, and that the high-dose primaquine has complete efficacy against
relapse in the reinfection control. Few therapeutic options today come with
such certainty. Nonetheless, if such treatments were applied, new parasitae-
mias among these groups may be classified as follows:
A = parasitaemias arising from relapse and reinfection;
B = parasitaemias arising from reinfection;
C = parasitaemias arising from reinfection and failure of primaquine;
where A − B = risk of relapse without primaquine, and C − B = risk of
primaquine treatment failure, then [(A − B) − (C − B)]/(A − B)) = propor-
tion of parasitaemias from relapse prevented by the primaquine in the evalu-
ation arm, i.e. the therapeutic response to standard primaquine therapy. This
approach may only apply to first relapse and cannot manage multiple relapses
in any given subject. The model also does not account for possible confound-
ing by long- vs. short-elimination half-life blood schizontocidal treatments.
Alternatively, a design without reinfection control would look like this:
X = parasitaemias arising from reinfection and relapse (no primaquine);
Y = parasitaemias arising from reinfection and failure of primaquine;
Where X − Y = relapses from failure of primaquine; X − Y/X = propor-
tion of parasitaemias prevented by primaquine therapy. A low rate of reinfec-
tion, e.g. <5% of recurrences, would have minimal impact on the estimate
of efficacy. However, as risk of reinfection rises, the estimate of efficacy
becomes increasingly confounded in the negative direction.
Whether using reinfection control or not, in any given subject of study in
endemic settings, a single infection cannot be unambiguously classified as
a relapse. Phenotyping resistance to primaquine for any isolate is not cur-
rently possible in this setting.
188.8.131.52. Primaquine resistance in returned travellers
Removing reinfection as a confounder by evaluating primaquine thera-
peutic responses in travellers repatriated to non-endemic zones eases the