Biomedical Engineering Reference
In-Depth Information
Their estimation result showed that for many participants, AZT yielded the
highest level of utility though it had the smallest impact of the publicly observed
CD4 counts. This fi nding was in contrast with previous evaluations, which con-
cluded that combination therapies were superior to using AZT alone because of the
greater improvement in CD4 counts. Their results suggested that using the standard
evaluation criteria alone may generate misleading conclusions regarding effective-
ness of alternative treatments. They also found that AZT + ddC had the most severe
side effects on average, although it had the highest impact on CD4 counts by the end
of the trial; and AZT + ddi and ddI had rather mild side effects and positive impact
on CD4 counts, therefore were the preferred treatment for the majority of patients.
In addition, the authors also found substantial learning throughout the trial, suggest-
ing that early attrition was primarily driven by side effects, while later dropout was
by treatment effectiveness.
In a related study, Lamiraud and Geoffard (
2007
) also used participants' adher-
ence to medication regimen to estimate their subjective preferences. They devel-
oped a binary choice model of participants' regimen adherence behavior as the
result of trade-off between treatment effectiveness and side effects. They then esti-
mated the model using data from CNAF3007, a multicenter, randomized Phase IIIB
trial, to compare the safety and effi cacy of two HIV therapies, a new simplifi ed
4-pill daily or two-intake regimen (referred as CBV/ABC) vs. the standard 11-pill
daily or 3-intake regimen (referred to as CBV/NFV). Health status measures such
as viral load and CD4 counts, occurrence of treatment-related side effects, and treat-
ment adherence were recorded for each participant at the beginning of the trial and
then at weeks 4, 8, 16, 24, 36, 40, and 48 of the trial. Although the treatment effec-
tiveness (measured by both antiretroviral activity and the rise in median CD4
counts) and the occurrence of side effects were both comparable between these two
regimens, the percentage of compliant participants was higher in CBV/ABC group
than in CBV/NFV group. The authors found that the probability of compliance
would decrease when side effects occurred, and increase when a positive change in
the CD4 count was observed, in the previous period. Using the estimated coeffi -
cients associated with the side effects occurrence and the change in CD4 count, they
quantifi ed patients' trade-off between treatment effectiveness and side effects, i.e.,
how severe should the side effects be for patients not to comply with a treatment and
how effective should the treatment be for patients to adhere to a treatment despite of
the side effects?
6.3
Measuring Treatment Effectiveness and Side
Effects from Prescription Data
In this section, we will fi rst discuss the limitations of using clinical trials data, which
calls for further using post-marketing prescription data to address some important
issues related to treatment evaluation. We then provide a review of related literature
Search WWH ::
Custom Search