It is has been established that several clinical features
are related to increased risk of relapse. Factors
associated with shorter survival times include
increased number of previous episodes, decreased
interval between episodes, and persistence of affective
symptoms.21 The results from the present study
showing higher discharge CGI-BP-S scores associated
with a higher risk of rehospitalization also confirm
previous reports that residual affective symptoms are
associated with outcomes.21
Several possible weaknesses in our study design
should be considered. The main limitations were
the retrospective design and small sample. Moreover,
we included only patients with at least 12
months of follow-up care who continued to receive
the same combination regimen as was given at discharge.
Hence, the results are most applicable to
patients who can tolerate treatment with lithium or
valproate in combination with an atypical antipsychotic
and who are largely adherent to this regimen.
Medication adherence is a well-known predictor of
relapse in bipolar patients.21 The rehospitalization
rates reported in this study may be underestimates,
as we excluded patients who did not receive
follow-up care for 12 months. Moreover, we did not
use objective adherence monitoring and could not
exclude the possibility that adherence issues substantially
affected relapse or rehospitalization rates.
Another concern was that clinical variables that
can affect rehospitalization rates, such as number
of previous depressive episodes, age at onset, and
number of previous psychiatric hospitalizations,
were not considered. Additionally, we did not
examine plasma concentrations of lithium and
valproate. Hence, the lithium and valproate plasma
concentrations of some patients may have been suboptimal
during the follow-up period. All researchers
participating in this study were members of the
Committee for a Korean Medication Algorithm for
Bipolar Disorder.22 The Korean medication algorithm
for bipolar disorder recommends lithium or
valproate in combination with an atypical antipsychotic
as the first-line strategy for acute mania. This
may have affected clinical practice and contributed
to the relatively even distribution of clinical variables
and treatment preferences observed in the
study data. Finally, we did not distinguish
rehospitalizations due to the failure of acute treatment
from those due to the failure of maintenance
treatment. Some proportion of rehospitalized
patients were experiencing a failure of acute treatment
rather that a failure of maintenance treatment.
However, the efficacy of lithium is comparable to
that of valproate for the treatment of acute mania,23
but about 34% more patients relapse due to any
kind of mood episode while receiving lithium than
while receiving valproate.24 The results of the
present study showing that valproate reduced the
risk of rehospitalization are consistent with the
results of a previous maintenance study.