Correlation between drug treatment adherence and lithium treatment attitudes
and knowledge by bipolar patients
Adriane Ribeiro Rosa
a, Marion Marco
Airton Tetelbom Stein
a, Jandyra M.G. Fachel
a, Helena M.T. Barros
b, Flávio Kapczinski
aDivision of Phamacology, Fundação Faculdade Federal de Ciências Médicas de Porto Alegre, Brazil
bDepartment of Statistics, Institute of Mathematics, Universidade Federal do Rio Grande do Sul, Brazil
cDepartment of Psychiatry, Universidade Federal do Rio Grande do Sul, Brazil
Available online 18 September 2006
Background: Non-adherence should always be investigated when there is a failure in bipolar treatments, since it is known that reported non-
adherence rates in bipolar disorder treatment for long-term prophylactic pharmacotherapy range from 18% to 52%, with a median prevalence of
44.7%. Several factors are related to the poor adherence and reduction of medication efficiency, such as the different types of bipolar disorder, the
presence of side effects, medication interactions, level of patient's knowledge about the disorder and their attitude towards treatment, complexity
of medical regimens and the doctor–patient relationship.
Methods: Bipolar disorder outpatients under lithium treatment from the Hospital de Clínicas and Materno Infantil Presidente Vargas of Porto
Alegre were recruited. All the patients had bipolar disorder and gave informed consent to participate in a clinical interview (106), answered the
Lithium Attitudes Questionnaire (LAQ), Lithium Knowledge Test (LKT), Medication Adherence Rating Scale (MARS) and had plasma and red
blood cells lithium measurements to assess their medication adherence and the factors that influenced it.
Results: 85.6% of bipolar disorder were adherent to lithium treatment showing plasma lithium between 0.6 and 1.2 mmol/L. There was an inverted
correlation between the total LAQ score with plasma and red blood cells, a positive correlation between LKT and MARS with plasma and red
blood cell lithium.
Conclusion: These results confirmed that knowledge level is directly correlated to treatment adherence and patients' attitudes, lower adherence,
general opposition to prophylaxis, fear of side effects, denial of therapeutic effectiveness and illness severity.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Adherence scales; Attitudes; Bipolar disorder; Knowledge; Lithium treatment
prevalence rate ranging from 1% from studies based primarily on
ascertaining history of mania, up to 8% when the concept of
bipolar disorder is expanded to include mania, hypomania, brief
hypomania, cyclothymia (Akiskal et al., 2000).
Lithium is the main treatment for bipolar disorder prophy-
laxis, as demonstrated in several clinical trials. It reduces manic
symptoms in 73% of the treated patients, twice as much as
placebo treatment and prevents the recurrences of bipolar
disorder (Burgess et al., 2002; Maj, 2000; Schou, 1999).
Furthermore, anti-manic efficacy correlates directly to the
lithium serum concentration: 32% of the patients with serum
lithium concentrations between 0.8 and 1.0 mmol/L present
recurrences, compared to 53% of the patients with lithium
concentrations between 0.4 and 0.6 mmol/L (Sproule, 2002;
Hopkins and Gelenberg, 2000; Gelenberg et al., 1989).
However, a marked gap has been noted between the efficacy
of lithium in clinical trials and its effectiveness in ordinary
clinical practice, this difference being almost certainly due to
poor treatment adherence (Dharmendra and Eagles, 2003; Scott
and Pope, 2003, 2002). Rates of lithium non-adherence are
high, ranging from 18% to 52% (Scott and Pope, 2002), with a
Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
Abbreviations: LAQ, Lithium Attitudes Questionnaire; LHS, Lithium
Hazard Score; LKT, Lithium Knowledge Test; MARS, Medication Adherence
Rating Scale; SEQ, Side Effects Questionnaire.
⁎Corresponding author. Rua Sarmento Leite, 245. Porto Alegre, RS, Brazil.
Tel.: +55 51 32248822x151; fax: +55 51 32248822x129.
E-mail address: firstname.lastname@example.org (H.M.T. Barros).
0278-5846/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
median prevalence of 44.7% (Schumann et al., 1999). First
discontinuation of prophylaxis has been described in 43.2%
patients during the first 6 months (Schumann et al., 1999). Poor
adherence to lithium is very common and is the most frequent
cause of recurrence during prophylactic lithium treatment
(Dharmendra and Eagles, 2003; Colom et al., 2003). While
side effects may be one cause for non-adherence, the
phenomenon is undoubtedly a complex one, involving clinical,
personal and interpersonal factors (Dharmendra and Eagles,
2003; Schumann et al., 1999; Wong et al., 1999).
Several explanations for the non-adherence to lithium
treatment are proposed, such as being in the first year of
treatment with few previous episodes of the disease, being a
male and young patient, presenting a history of grandiosity,
elevated mood and complaints of ‘missing high’, medication
side effects, lithium effectiveness, knowledge level, negative
attitudes, dislike of having his/her mood regulated by a drug, or
as a result of anti-drug pressure (Lingam and Scott, 2002;
Schou, 1997). Since compliance with treatment of any sort
tends to be enhanced by education (Dharmendra and Eagles,
2003; Colom et al., 2003; Schaub et al., 2001), it is entirely
logical to investigate the effect of educational interventions on
adherence to lithium treatment. However, patients could acquire
knowledge about lithium successfully, but the effect of
knowledge on enhancing adherence was equivocal in a small
trial of 60 patients (Harvey, 1991).
Is has also been suggested that knowledge about lithium had
no association with adherence (Sing Lee et al., 1992).
Schumann et al. (1999) and Scott and Pope (2002) found that
attitudes towards lithium treatment affected adherence, while
Wong et al. (1999) could not detect relationship between insight
and adherence. Therefore, since the existing evidence is
conflicting, a better understanding of the impact that proper
knowledge about the disease and its treatment may bring into
the adherence of bipolar patients is still to be determined.
This study was conducted to determine the adherence to
lithium treatment among patients that receive intensive
monitoring and education. We investigated the degree of non-
adherence, the reasons for non-adherence and its consequences,
and evaluated specific attitudes towards lithium long-term
treatment in a sample of bipolar patients. To elucidate the
factors associated with non-adherence, the aims of the present
study are to determine plasma and red blood cell lithium
concentrations in bipolar patients at the same time as estimating
attitudes and knowledge about lithium treatment in adherence
One hundred and six bipolar disorder patients, 18–75 years
old, under lithium treatment for at least 1 month, regularly
complying with the weekly visits scheduled in two psychiatric
outpatient services specialized in mood disorders in the city of
Porto Alegre, Brazil, agreed to take part in the survey. The study
was approved by the Ethics Committees on Research of the
Hospital Materno Infantil Presidente Vargas and Hospital de
Clínicas of Porto Alegre, where the research took place. In the
weekly consultations, the patients were evaluated for both their
symptoms and general state. All patients participated in monthly
psycho-educational groups coordinated by a qualified psychi-
atric nurse, where the standard package of psycho-education
included written information about the disorder, monthly
meetings with the psychiatric nurse, occasional discussions
with invited speakers and encouragement to join the local
bipolar disorder patients association. For this study, the subjects
were contacted 7 days before the scheduled appointment and
were instructed to be at the hospital early in the morning,
without having taken their lithium dose in the morning because
the lithium blood levels were to be measured, respecting a 12-
h interval between the last dose of lithium and the blood
sampling. The patients who agreed to participate in the study
gave informed consent, were interviewed and had blood drawn
immediately for the assessment of their blood level of lithium,
rates of self-reported adherence, attitudes and knowledge of
lithium treatment. Patients were submitted to a structured
questionnaire, which included demographic data and details
about their disorder and treatment.
2.2.1. Demographical Data
Every subject gave information about marital status, work,
age, gender, education level, habits such as smoking, drinking
coffee or tea and medication used. Patients were diagnosed as
bipolar type I or II through the Structured Clinical Interview for
DSM-IV-APA/1994 (SCID-I) (Del-Ben et al., 1996). Their
mood state was evaluated for follow-up through the Hamilton
Rating Scale for Depression and Young Mania Rating Scale.
The Lithium Attitudes Questionnaire consists of 19 items for
rating of attitudes towards prophylactic lithium treatment.
Seven subscales assess the resistance to prophylaxis in general
(LAQ 1), the denial of therapeutic effectiveness of lithium as a
prophylactic agent (LAQ 2), the fear of side effects (LAQ 3), the
difficulties with the daily routine medication intake (LAQ 4),
the denial of the severity of the illness (LAQ 5), the negative
attitudes toward drugs in general (LAQ 6) and the lack of
information about lithium prophylaxis (LAQ 7). The items are
in Yes/No format with low scores indicating positive attitudes
and high scores indicating negative attitudes (Dharmendra and
Eagles, 2003; Harvey, 1991).
It comprises 10 questions related to the non-adherence to the
daily intake or specified days for taking the medication, which
include attitudes about the failure to use lithium when
improvement or worsening is perceived by the patient. High
scores would be associated to a better adherence by the patient.
We considered that the cutoff point in this scale is seven. Scores
higher than seven were considered high adherence (Thompson
et al., 2000).
218 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
It comprises 20 items about side effects associated with
lithium such as nausea, vomiting, diarrhea, tremor, polyuria and
polydipsia, goiter and hypothyroidism, etc. The frequency of
the side effects was verified (Ghose, 1977).
The Lithium Knowledge Test evaluates aspects of patients'
practical and pharmacological knowledge which are important
if therapy is to be safe and effective, one point being scored for
each correct answer and one deducted for a wrong answer,
giving a total Lithium Knowledge Score. Some incorrect
answers, since they constitute potential hazards to patients on
lithium, are added up to give the LHS (Dharmendra and Eagles,
2003; Harvey, 1991).
Before their use, these scales were translated to Portuguese,
back-translated to English, and this English version was
evaluated by five psychiatrists who are proficient in English.
2.2.6. Assessment of levels of lithium in plasma and red blood
Venous blood was collected from each patient into two
Vacutainer tubes containing edetic acid. For whole blood
lithium measurement, 99 μl of blood was diluted 1/20 in
distilled, deionised water. The remaining blood was then
centrifuged, 1600 rpm for 10 min, and the plasma removed by
aspiration. A 1/20 dilution in water was made of 99 μl of
plasma. Lithium concentrations were measured on the whole
blood and plasma dilutions by the indirect method, using an
Instrumentation Laboratory CELM Flame Photometer. Assays
were performed in duplicate. The formula used to calculate the
red cell lithium by the indirect method was: Red Cell
Lithium=Whole Blood Lithium−(1−HTC)×Plasma Lithi-
um/HTC. The difference between the direct and indirect
method was that for the direct method, 200 μl of the re-
maining packed red cells were dispersed into 1 ml of 150 mM
choline chloride, which was layered on 0–2 ml of dibutyl
phthlate in a 1.5-ml microfuge tube. Dibutyl phthalate has a
density between that of water and erythrocytes. The
erythrocytes therefore dropped to the bottom of the tube and
passed through the dibutyl phthalate removing the adherent
plasma. A 1/20 dilution of the packed cells was made and
mixed thoroughly to ensure complete haemolysis and lithium
concentrations were measured on the whole blood and plasma
dilutions by the indirect method, using an Instrumentation
Laboratory CELM Flame Photometer. The correlation coeffi-
cient between the direct and indirect method was 0.97 (Harvey
et al., 1989).
2.2.7. Statistical analysis
The procedures for univariated statistics included the t-test
for independent samples and the t-test for paired samples for
comparisons of non-adherent and adherent patients. Clinical
states were compared in respect to the adherence parameters
with a one-way ANOVA, followed by the Dunnet's test when
appropriate. Bivariate correlations were calculated with the
(parametric) Pearson product–moment correlation. All vari-
ables exhibiting statistical significance in the univariate
analyses were subjected to multiple regression with LKT,
LHS, LAQ, MARS, Plasma Lithium and Red Blood Cell
Lithium as dependent variables. At each step, the variable with
the lowest t-value was removed until only variables that were
statistically and significantly related to the dependent variables
remained. Confounders such as sex, age, years of education and
the use of multiple drugs were controlled for using a linear
The answers to the Lithium Knowledge Test (LKT), Lithium
Attitudes Questionnaire (LAQ), Medication Adherence Rating
Scale (MARS), and Side Effects were obtained from 106
bipolar patients. Plasma lithium and red blood cell lithium
concentration was obtained from 99 patients (Table 1). The
sample included 73 (68.9%) women, mean age 43.56±9.83 and
33 men with a mean age of 41.61±9.72. Their education level
was 9.91±3.23 years of study for men and 8.81±3.44 for
women. Habits showed 40 (37.7%) were smokers, 70 (66%)
used coffee on a regular basis, and 68 (64.2%) had tea on a
regular basis. There were 71 euthymic patients, 11 hypomanic
and 17 depressed patients in the sample, according to the
evaluation on the day of data collection. The clinical status of
the patients (euthymia vs depression or hypomania) did not
induce significant changes in blood levels of lithium or in scores
of the LKT, LKT Hazard and MARS scales (Table 2). Although
the mean scores in the LAQ scale were within the cut-off for
adherence there was a significant overall difference between the
clinical state of patients: hypomanic and depressive patients
presented higher LAQ scores (F(2,103)=3.25; P<0.05); howev-
er, post hoc tests did not detect if there were differences with
respect to euthymic patients, probably due to the low number of
patients within these two classifications.
For subjects being prescribed lithium (n=106) the mean
score on the LKT was 9.0±0.75 for men and 8.74±0.44 for
women and LKT Hazard was 4.06±0.19 for men and 3.96
Demographics and scores for lithium treatment
Red cell lithiuma
Parameters in 106 bipolar patients.
aInformation available for 99 of 106 patients.
219 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
±0.15 for women. The mean score on the LAQ TOTAL was
4.12±0.55 for men and 4.10±0.41 for women. The mean score
on the MARS was 7.24±0.32 for men and 6.92±0.18 for
women. Mean lithium levels in the plasma were 0.85±0.048 for
men and 0.85±0.029 for women and lithium levels in the red
blood cells were 0.87±0.048 for men and 0.87±0.029 for
women. Therefore, gender does not influence any of the
adherence parameters evaluated by the scales or lithium plasma
or red cell levels. In plasma lithium concentration 85 (85.6%) of
the patients were adherents. According to the MARS, 70
(66.94%) adherence (Table 3), the LAQ scale pointed to
positive attitudes among 66 (61.32%) patients and 89 (84%)
patients presented good knowledge levels according to the LKT
scale. In contrast, 33 (33.96%) patients presented high chances
of intoxication because of lower knowledge levels observed in
the answers to questions such as: what should you avoid while
taking lithium and what you should do if you develop acute
diarrhea and vomiting?
All the patients (n=106) presented at least 1 of the 20
adverse effects listed in SEQ. The most common side effects
were weight gain (79.2%), polyuria (77.4%), tremor (67.9%),
fatigue or discouragement (66%), skin problems (62.3%), slow
movements (57.5%), polydipsia (53.8%), sleepiness (52.8%),
diarrhea (45.3%), muscular weakness (42.5%), nausea (41.5%),
dizziness (38.7%), sexual problems (37.7%), vomiting (20.8%),
and thyroid problems (19.8%).
3.1. Simple Correlations
There was a significant negative correlation concerning age
with LKT scores (r=−0.2; p=0.04). Age was positively
correlated with LKT Hazard scores (r=0.366; p=0.001) and
with MARS scale score (r=0.216; p=0.026). Such results
suggest that elderly patients present lower scores in the LKT
scale, that is, less knowledge about lithium, and at the same time
present a higher toxicity risk during treatment and higher
MARS scores. Age did not correlate with lithium plasma or red
blood cell levels.
A negative correlation of LKT and LAQ was found (r=
−0.269; p=0.005). When LAQ subscores and LKTscores were
compared, only LAQ4 and LAQ6 (r=−0.205; p=0.035 and
r=−0.286; p=0.003), respectively, maintained significant
correlations with LKT, suggesting that more treatment knowl-
edge may overcome difficulties in maintaining pill taking
routines and cultural opposition to drug treatment.
There was a negative correlation between MARS and LAQ
(r=−0.6; p<0.001), as expected, which suggests that the
higher MARS scores and lower LAQ scores confirm the
patients' better adherence. The study of associations between
MARS and the LAQ subscores shows that the only statistically
significant correlation was with LAQ3, which contains ques-
tions regarding fear of side effects (r=−0.300; p=0.002). There
was a positive correlation between MARS and the plasma
lithium (r=0.259; p=0.009).
Correlations are presented in respect to plasma lithium and
cell lithium concentration. There was a significant positive
correlation between LKT with plasma lithium (r=0.232;
p=0.020). As seen in Fig. 1A, the patients with higher LKT
scores are more likely to have plasma levels within the
therapeutic levels of lithium. At the same time, there was a
negative correlation between LKTand LKT Hazard (r=−0.629;
p<0.0001), suggesting that the patients with lower LKT scores
are more uninformed about intoxication risks.
There was a significant negative correlation between the total
LAQ score and plasma lithium (r=−0.198; p=0.048), as seen
in Fig. 1B. LAQ5 subscores are negatively related to the
concentration of plasma lithium (r=0.237; p=0.018), suggest-
ing that disease denial is the principal factor, which results in
patients presenting lower lithium therapeutic levels. Patients
were subdivided according to their plasma lithium levels being
within the therapeutic range (plasma Li: 0.6 mmol/L–1.2 mmol/
L; n=85) or not (Li<0.6 mmol/L or Li>1.2 mmol/L; n=14).
The only statistical difference was found with LAQ1 (t=2.365;
p=0.02), suggesting that patients opposed to continuing lithium
treatment influences lithium plasma levels monitoring. Disease
denial, evaluated in LAQ5, shows a trend of difference when
these two groups of patients were compared (t=1.953; 0.054).
Sixty seven of the patients reported habitually using caffeine
beverages. When lithium plasma levels of these patients were
compared to those of patients not using caffeine no significant
difference was detected (0.84±0.26×0.87±0.22). Fourteen
patients of the 106 that were examined were only using lithium
at the time of the study. Very few patients were using
medications for any clinical condition: 5 patients were using
Adherence or non-adherence to lithium, according to the cutoffs on the scales on
therapeutic lithium levels
Parameters (cut-off)Numbers of non-adherent/
Plasma lithium (0.6–1.2 mmol/L)
Red cell lithium (0.6–1.2 mmol/L)
LAQ Total (score<6)
Adherence parameters to lithium, according to the clinical states
Clinical state EuthymicHypomanicDepressive
Red cell lithium.65±.26
220 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
captopril, 6 patients used hydrochlorothiazide, while verapamil
or diclofenac were used by 1 patient each. None of the patients
using these medications had significant changes in lithium
plasma or blood cell levels. The concomitant use of mood
stabilizing drugs depicted that antipsychotic agents (used by
44.30% of the patients), valproic acid (29.20%), antidepres-
sants (26.40%), anxyolitic agents (18.90%), carbamazepine
(14.20%) were the most frequently drugs prescribed with
lithium for this group of patients. No difference in plasma
lithium levels was seen in respect to patients using psychiatric
drugs or not.
There was a positive correlation between red cell lithium
and LKT (r=0.236; p=0.019), Fig. 1C, a negative correlation
between the total LAQ score (r=−0.220; p=0.029), Fig. 1D,
and a positive correlation between MARS (r=0.276; p=0.006).
The LAQ2 subscores, for accepting lithium effectiveness (r=
−0.207; p=0.040) and LAQ5 subscores show that the disease
denial has negative correlation with the red blood cell levels of
lithium (r=−0.243; p=0.016). By comparing results of red
blood cells and plasma lithium with adherence scales, this study
registered that red cells and plasma lithium show equal results
(r=1; p=0), and therefore, red blood cells may be used as an
adherence marker for patients with high adherence rate. It is
necessary to further investigate if the lithium levels in the red
blood cells can be considered adherence markers for patients
with low adherence rates.
Fig. 1. The correlations between A-LKTand lithium plasma levels; B-LAQ and lithium plasma levels; C-LKTand red blood cell lithium levels; D-LAQ and red blood
cell lithium levels. The results suggest that LKTscore >4 indicates adherence, with a positive correlation with plasma and red blood cell lithium (p<0.005) and that
LAQ score <6 indicates adherence, with a positive correlation with plasma and red blood cell lithium (p<0.005).
221 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
3.2. Multiple Regression
Using plasma lithium as a dependent variable and LAQ,
MARS, LKT, and LKT Hazard as predictors, only LKT
(t=2.844; p=0.005) showed significant correlation. The
knowledge level was directly related to the better adherence
levels of the patients. Using red cell lithium as a dependent
variable and LAQ, LKT, LKT Hazard, AGE as predictors, only
the LKT (t=2.540; p=0.013) showed positive correlation.
Also, a multiple linear regression was carried out using the
number of additional medications currently used as the
independent variable and plasma or cell lithium levels or the
adherence scales as the outcome measure; within this model,
sex, age and year of education were considered as confounders.
The only significant result found showed that the number of
medications currently used were associated with higher scores
in the LKT scale (p=0.008). The knowledge level was directly
related to the better adherence levels of the patients.
In the present study, 86% of bipolar patients attending a
specialized mood disorder clinic and educational support
activities were adherent to lithium, showing plasma levels of
lithium between 0.6 and 1.2 mmol/L. These adherence levels
were higher than other studies, whose values were of 53% and
59.2%, respectively (Scott and Pope, 2002; Schumann et al.,
1999). Our high adherence level can be due to the fact that our
patients receive periodic consultations for regular monitoring of
plasma levels of lithium and for clinical evaluation, and also
because non-adherent patients are more likely to be
Different from other reports, the blood levels of lithium did
not differ according to the clinical status of the patients
(Sproule, 2002). Not finding a difference between the clinical
state may be reflecting the fact that, within our sample, most
patients were euthymic, and very few were composing the
group of hypomanic or depressed clinical state. It is not unlikely
that larger samples with a broader spectrum of clinical severity
of illness would unveil the correlation of full-blown episodes or
even subsyndromal symptoms with the blood levels of lithium.
In this same vein, the present study showed that hypomanic or
depressed patients tend to present more negative attitudes
towards treatment than patients in euthymia. This is consistent
with the clinical observation that patients with hypomanic or
depressed symptoms are more likely to deny the severity of their
illness and are, therefore, more likely to loosen their adherence
to treatment (Lingam and Scott, 2002).
Because it was reported that withdrawal from caffeine
increases lithium blood levels (Mester et al., 1995) in bipolar
patients, one could expect a difference in lithium levels between
patients who take caffeine containing drinks or not. In the
present study, the use of caffeine did not alter the blood levels of
lithium. Also, no changes were seen in lithium levels between
patients using other substances known to interact with lithium,
such as diuretics. This may be related to the fact that patients
were carefully monitored for their blood levels of lithium and a
good amount of care was delivered in order to prevent
deleterious drug interactions.
In this study, different indexes of attitudes and knowledge
about lithium were measured as additional adherence markers,
because such indexes were described to have correlated results
(Dharmendra and Eagles, 2003; Scott and Pope, 2002;
Schumann et al., 1999). Effectively, our results confirm that
lithium levels monitoring and adherence questionnaire answers
are correlated. However, not all three indexes of adherence
turned out to be sensitive to determine the same levels of
adherence prevalence. LKT measures knowledge about lithium,
and indicates adherence prevalence to the treatment withlithium
of 84%, the same of the lithium plasma level values. Although it
was not constructed to measure treatment adherence, it has been
applied in some studies with this purpose and it is considered to
be an efficient predictor, as confirmed in this study (Dharmen-
dra and Eagles, 2003; Scott and Pope, 2002). Also, it is
noteworthy that patients who were using a higher number of
psychiatric drugs were found to have more awareness of the
details related to their lithium treatment. This may reflect the
fact that patients with more severe forms of the disorder – which
may require the use of multiple drugs – tend to receive more
attention from the health care providers and, therefore, better
The answers to MARS index, elaborated to measure
adherence to treatments in patients with psychotic disorders
(Thompson et al., 2000), indicate an adherence prevalence 15%,
lower than the plasma levels, therefore being less efficient to
measure adherence levels to lithium. We propose that it is not
possible to verify adherence to lithium, because the adherence
questions do not characterize drug intake timing in a specific
way. As the mood of the bipolar patient oscillates, his/her
adherence to the treatment also oscillates (Waters et al., 1982). It
is our opinion that for an index to be effective, it should define a
time for adherence question to medication such as the previous
week or month. In further studies, we should test this
hypothesis, modifying the questions to the patients.
LAQ is useful to understand patients' attitudes toward
lithium treatment. It presented adherence frequency of around
60%, also lower than lithium levels. In spite of LAQ not being
able to precisely measure adherence prevalence, it is funda-
mental in the investigation of factors responsible for poor
adherence (Dharmendra and Eagles, 2003; Scott and Pope,
2002; Schumann et al., 1999). Among the most frequent
attitudes of non-adherent patients were non-acceptance of the
therapeutic lithium effectiveness, opposition to the treatment,
denial of the disease and fear of side effects. Our findings
confirm Schumann's hypothesis, which related the opposition
to chronic treatments as the main difference between adherent
and non-adherent groups, and that this attitude may directly
influence lithium effectiveness (Scott and Pope, 2002; Schu-
mann et al., 1999).
As opposed to MARS and LAQ, red blood cell levels of
patients closely followed the results of the plasma levels. This
proves that there was in these patients a dynamic balance
between intra- and extracellular lithium, which is characteristic
in patients under chronic treatment (Taylor et al., 2002). A
222A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
previous study showed that after an acute lithium dose, patients
presented plasma levels of lithium higher than the red blood cell
levels, and that the changes in the plasma levels indicate
omission of dosages in the days prior to the examination or
administration of extra dosages before the assessment and such
changes slightly modify the red blood cell levels (Frazer et al.,
1978). On the other hand, after chronic use of lithium, we
obtained red blood cell levels parallel to the plasma lithium
levels, which means that changes in the red blood cell levels
would indicate a persistence in the regular use and could
characterize the lack of adherence, for example, skipping pills
for many days or important changes of dosage or medication
schedule. Therefore, results suggest that lithium red blood cell
levels can be used as an adherent marker for the maintenance
treatment with lithium (Taylor et al., 2002; Frazer et al., 1978).
The different indexes used to evaluate adherence have shown
some factors more related to non-adherence to treatment among
these patients. We would like to call one's attention to the fact
that we are working with a group of patients that are possibly
differentiated in that they are very motivated for the treatment
because they are intensively followed through continuous
monitoring. The non-acceptance of the therapeutic lithium
effectiveness showed by some patients results in modifications
of the ingested lithium dosages, usually with a reduction of the
dosages, increase of symptomatology or, very rarely, increase of
the dosages with consequent side effects (Scott and Pope, 2002;
Claxton et al., 2001). Denial of the disease is linked to
opposition to the treatment. These factors seem to be a more
important issue in: the beginning of the bipolar disorder, young
patients, long periods of treatment, poor doctor–patient
relationships and when the patient has little information and
knowledge about the treatment (Scott and Pope, 2002; Claxton
et al., 2001). According to the same authors, longer therapies
with lithium strongly influence adherence, and the doctor–
patient relationship is positively associated to patients' attitudes
and adherence to treatment (Sing Lee et al., 1992), as might be
the case in our group of patients. Fear of side effects, seen with
LAQ, was associated to low adherence in our patients. On the
other hand, every patient, adherent or not to the treatment,
presented side effects to lithium, differing from other studies,
which points to an association of side effects as the main cause
of non-adherence (Jonhson and McFarland, 1996; Nilsson and
Axelsson, 1990; McCreadie et al., 1985).
We did not take into account the patients co-morbidity
diagnosis in respect to Axis II and III, which could be of
influence on adherence to treatment.
Because the Portuguese version of the LAQ, LKT and
MARS scales are not yet validated, some cultural differences
between our patients and those of English speaking countries
could have been mixed.
Finally, patients' knowledge level about lithium and the
disease proved to be a direct influence to adherence. The fact
that patients are informed about the disease, the treatment and
the risks of not treating it positively influences adherence,
because it facilitates their acceptance of the disease and
maintenance therapy. It is important to give special attention
to elderly people as they presented less knowledge and less
ability to retain information, presenting higher risks of toxicity.
Our findings confirm that the higher the knowledge level, the
higher the adherence and the lower the toxicity risks
(Dharmendra and Eagles, 2003; Scott and Pope, 2002; Clarck
and Pickles, 1994). It also indicates that education and
motivation of patients towards their lithium treatment increases
adherence. It is to be determined how to extend these
motivational attitudes to the other patients who, in spite of
attending the clinical environment still do not fully comply with
The authors offer thanks for the help in intensive patient care
from Hospital de Clínicas e Hospital Materno Infantil
Presidente Vargas. F.K and H.M.T. B. received Researcher
Scholarships from CNPq.
Akiskal HS, Bourgeois ML, Angst J, Post R, Möller H, Hirschfeld R. Re-
evaluating the prevalence of and diagnostic composition within the broad
clinical spectrum of bipolar disorders. J Affect Disord 2000;59(1):S5–S30.
Burgess, S, Geddes, J, Hawton, K, Townsend, E, Jamison, K, Goodwin, G.
Lithiumfor maintenance treatmentof mooddisorders, In Cochrane Database
Syst. Ver. Oxford. 2002. www.bireme.br.
Clarck DJ, Pickles J. Lithium treatment for people with learning disability:
patients' and carers' knowledge of hazards and attitudes to treatment. J
Intellect Disabil Res 1994;38:187–94.
Claxton AJ, Cramer J, Courtney PA. Systematic review of the association
between dose regimens and medication compliance. Clin Ther
Colom F, Vieta E, Reinares M, Martinez-Aran A, Torrent C, Goikolea JM, et al.
A randomized trial on the efficacy of group psychoeducation in the
prophylaxis of recurrences in bipolar patients whose disease is in remission.
Arch Gen Psychiatry 2003;60:402–7.
Del-Ben CM, Rodrigues CR, Zuardi AW. Reliability of the Portuguese version
of the structured clinical interview for DSM-III-R (SCID) in a Brazilian
sample of psychiatric outpatients. Braz J Med Biol Res 1996;29
Dharmendra MS,EaglesJM. Factors associatedwithpatients' knowledge of and
attitudes towards treatment with lithium. J Affect Disord 2003;75:29–33.
Frazer A, Mendels J, Brunswuck D, London J, Pring M, Ramsey A, et al.
Erythrocyte concentrations of the lithium ion: clinical correlates and
mechanisms of action. Am J Psychiatry 1978;9:1065–9.
Gelenberg AJ, Kane JM, Keller MB, Lavori P, Rosenbaum JF, Cole K, et al.
Comparison of standard and low serum levels of lithium for maintenance
treatment of bipolar disorder. N Engl J Med 1989;321:1489–93.
Ghose K. Lithium salts: therapeutic and unwanted effects. Br J Hosp Med
Harvey SN. The development and descriptive use of the lithium attitudes
questionnaire. J Affect Disord 1991;22:211–9.
Harvey NS, Summerton AM, Forrest ARW. New direct method for measuring
red cell lithium. J Clin Pathol 1989;42:435–7.
Hopkins HS, Gelenberg AJ. Serum lithium levels and the outcome of
maintenance therapy of bipolar disorder. Bipolar Disord 2000;2:174–9.
Jonhson RE, McFarland BH. Lithium use and discontinuation in a health
maintenance organization. Am J Psychiatry 1996;153:993–1000.
Lingam R, Scott J. Treatment non-adherence in affective disorders. Acta
Psychiatr Scand 2002;105:164–72.
223 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224
Maj M. The impact of lithium prophylaxis on the course of bipolar disorder: a Download full-text
review of the research evidence. Bipolar Disord 2000;2:93–101.
McCreadie RG, McCormick M, Morrison DP. The impact of lithium in south-
west Scotland. Br J Psychiatry 1985;146:70–80.
Mester R, Toren P, Mizrachi I, Wolmer L, Karni N, Weizman A. Caffeine
withdrawal increases lithium blood levels. Biol Psychiatry 1995;37:348–50.
Nilsson A, Axelsson R. Lithium discontinuers—I. Clinical characteristics and
outcome. Acta Psychiatr Scand 1990;82:433–8.
Schaub RT, Berghoefer A, Müller-Oerlinghausen B. What do patients in a
lithium outpatient clinic know about lithium therapy? J Psychiatry Neurosci
Schou M. The combat of non-compliance during prophylactic lithium treatment.
Acta Psychiatr Scand 1997;95:361–3.
Schou M. Perspectives on lithium treatment of bipolar disorder: action, efficacy,
effect on suicidal behavior. Bipolar Disord 1999;1:5–10.
Schumann C, Lenz G, Berghofer A, Muller-Oerlinghausen B. Non-adherence
with long-term prophylaxis: a 6-year naturalistic follow-up study of
affectively ill patients. Psychiatr Res 1999;89:247–57.
Scott J, Pope M. Nonadherence with mood stabilizers: prevalence and
predictors. J Clin Psychiatry 2002;63:384–90.
Scott J, Pope M. Do clinicians understand why individuals stop taking lithium? J
Affect Disord 2003;74:287–91.
Sing Lee Wing YK, Wing, Wong KC. Knowledge and compliance towards
lithium therapy among Chinese psychiatry patients in Hong Kong. Aust N Z
J Psychiatry 1992;26:444–8.
Sproule B. Lithium in bipolar disorder. Clin Pharmacokinet 2002;41(9):639–60.
Taylor R, Mallinger AG, Frank E, Rucci P, Stat D, Thase M, et al. Variability of
erythrocyte and serum lithium levels correlates with therapist treatment
adherence efforts and maintenance treatment outcome. Neuropsychophar-
Thompson K, Kulkarni J, Sergejew AA. Reability and validity of a new
medication adherence rating scale (MARS) for the psychoses. Schizophr
Waters B, Lapierre Y, Gagnon A. Determination of the optimal concentration of
lithium for the prophylaxis of manic-depressive disorder. Biol Psychiatry
Wong SSL, Lee S, Wat KHY. A preliminary communication of an insight scale
in the assessment of lithium non-adherence among Chinese patients in Hong
Kong. J Affect Disord 1999;55:241–4.
224 A.R. Rosa et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 31 (2007) 217–224