CHAPTER 3 IMPROVING MEDICATION ADHERENCE IN BIPOLAR DISORDER: A SYSTEMATIC
3.5 R ESULTS
3.5.1 Description of studies
Studies are described in Tables 3.9 and 3.9 and intervention characteristics across all studies are summarised in Table 3.1 and Table 3.2.
3.5.1.1 Intervention content & tailoring
Five interventions had adherence as the primary focus of the intervention content (Cochran, 1984; Dogan & Sabanciogullari, 2003; Elixhauser, Eisen, Romeis, & Homan, 1990; N. S. Harvey
& Peet, 1991; Peet & Harvey, 1991; Sajatovic, Davies, et al., 2009), 18 were multi-focus programmes, (Ball et al., 2006; D. Castle et al., 2007; Clarkin, Carpenter, Hull, Wilner, & Glick, 1998; Colom et al., 2003; D'Souza, Piskulic, & Sundram, 2010; Eker & Harkin, 2012; Frank et al., 2005; Frank et al., 1999; Gilbert, 2000; Javadpour et al., 2013; Lam et al., 2000; Lam, Hayward, Watkins, Wright, & Sham, 2005; Lam et al., 2003; Lenz, 2010; Miklowitz, George, Richards, Simoneau, & Suddath, 2003; Miklowitz et al., 2000; Rea et al., 2003; Reinares et al., 2008; van Gent & Zwart, 1991; Zaretsky, Lancee, Miller, Harris, & Parikh, 2008). We were unable to classify one study due to the limited information provided in the paper (Bordbar et al., 2009).
Fifteen interventions were described as being or including psychoeducation (Ball et al., 2006;
Bordbar et al., 2009; Clarkin et al., 1998; Colom et al., 2003; D'Souza et al., 2010; Eker &
Harkin, 2012; Gilbert, 2000; Javadpour et al., 2013; Lenz, 2010; Miklowitz et al., 2003;
Miklowitz et al., 2000; Rea et al., 2003; Reinares et al., 2008; Sajatovic, Davies, et al., 2009; van Gent & Zwart, 1991; Zaretsky et al., 2008). Five classified themselves as cognitive-behavioural therapy (CBT) or cognitive therapy (Ball et al., 2006; Cochran, 1984; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003; Zaretsky et al., 2008). When coding interventions based on the content described in the paper, 13 clearly included all elements of psychoeducation (Ball et al., 2006; D. Castle et al., 2007; Clarkin et al., 1998; Colom et al., 2003; D'Souza et al., 2010; Eker &
Harkin, 2012; Gilbert, 2000; Javadpour et al., 2013; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003; Lenz, 2010; Miklowitz et al., 2003; Miklowitz et al., 2000; Rea et al., 2003; Sajatovic, Davies, et al., 2009).
As described, four interventions appeared to be information provision only without a specific behavioural or psychosocial element (Bordbar et al., 2009; Dogan & Sabanciogullari, 2003; N.
S. Harvey & Peet, 1991; Javadpour et al., 2013; Peet & Harvey, 1991). Three had a specific focus on family communication (Gilbert, 2000; Miklowitz et al., 2003; Miklowitz et al., 2000;
Rea et al., 2003). Elixhauser and colleagues (1990) used an electronic adherence monitoring device, adherence feedback and education.
Fifteen interventions included more than one type of tailoring. Sixteen interventions used micro-tailoring of components of the intervention such as action planning in relation to participants lifestyle and goals or information on risk tailored to individual characteristics or history (Ball et al., 2006; D. Castle et al., 2007; Cochran, 1984; Colom et al., 2003; D'Souza et al., 2010; Eker & Harkin, 2012; Frank et al., 1999; Gilbert, 2000; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003; Miklowitz et al., 2003; Miklowitz et al., 2000; Rea et al., 2003; Reinares et al., 2008; Sajatovic, Davies, et al., 2009; van Gent & Zwart, 1991; Zaretsky et al., 2008).
Sixteen studies involved participation where the participants had active input, for example own difficulties with adherence, deciding on strategies and evaluating the outcomes of strategies used (Ball et al., 2006; D. Castle et al., 2007; Cochran, 1984; Colom et al., 2003;
D'Souza et al., 2010; Eker & Harkin, 2012; Frank et al., 1999; Gilbert, 2000; Javadpour et al., 2013; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003; Miklowitz et al., 2003; Miklowitz et al., 2000; Rea et al., 2003; Reinares et al., 2008; Sajatovic, Davies, et al., 2009; van Gent &
Zwart, 1991; Zaretsky et al., 2008).
Six interventions used individualization, where the content of sessions was individualised in response to participants’ needs, their questions or queries (Ball et al., 2006; Dogan &
Sabanciogullari, 2003; Eker & Harkin, 2012; Gilbert, 2000; N. S. Harvey & Peet, 1991; Peet &
Harvey, 1991; Rea et al., 2003).
Five interventions were tailored by varying the level of intervention contacts depending on participants’ needs or preferences (Frank et al., 2005; Frank et al., 1999; Gilbert, 2000; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003) with the decisions either based on participant choice (k=1) (Gilbert, 2000), HCP/ clinical need (k=2) (Lam et al., 2000; Lam et al., 2003) or both patient and practitioner choice (k=2) (Frank et al., 2005; Frank et al., 1999).
One intervention used tailoring at a macro-level, i.e. determining the intervention to be received by pre-tested characteristics, in this case the levels of symptoms, where participants must be classed as asymptomatic before proceeding (Frank et al., 1999) and one tailored feedback according to a mid-point assessment of adherence (Elixhauser et al., 1990).
3.5.1.2 Delivery (provider, setting, target, duration, frequency of sessions)
A range of providers were involved in intervention delivery; most commonly psychologists (k=5) and psychiatrists (k=6), but also mental health nurses (k=3), social workers (k=3) and
‘therapists’ (k=3) (Table 3.1).
Five interventions were delivered through specialist affective, mood or bipolar disorder clinics (Cochran, 1984; Colom et al., 2003; Eker & Harkin, 2012; N. Harvey & Peet, 1991; Reinares et al., 2008). Other settings included hospital outpatient mental health departments and University psychology or psychiatry departments (not specifying specialism in affective disorders) (Table 3.1). In six cases, the exact location of delivery of the intervention was not specified (David Castle et al., 2010; Clarkin et al., 1998; D'Souza et al., 2010; Lenz, 2010; van Gent & Zwart, 1991; Zaretsky et al., 2008), however it might be assumed that they were conducted at the outpatient clinics or University departments where authors were based.
Most interventions targeted either the patient alone (k=11) or the patient and their family (k=10), three interventions targeted the family or patients significant others only (Bordbar et al., 2009; Reinares et al., 2008; van Gent & Zwart, 1991) (Table 3.1). Interventions were
delivered to only groups (k=8), solely to individuals (k=9), or delivered to the patient and family members or as a couple (k=4).
The median number of intervention contacts as per protocol was 17 (range 1-56), with a mean contact time (estimated from data available for 16 studies) of 19.7 (sd 11.67) hours. The median duration of interventions was 6 months (IQR=0.5-12), ranging from a single intervention to 2 ½ years.
The spacing of intervention contacts was highly variable including, single-sessions (Bordbar et al., 2009) or two contacts (Elixhauser et al., 1990; Peet & Harvey, 1991), but most were delivered as a weekly programme for at least 6 sessions (k=17). Seven interventions started with regular weekly sessions, and before reducing the frequency. The family-focussed therapy programmes were conducted for 12 weekly sessions, reducing to bi-weekly, then 3 monthly (Miklowitz et al., 2000; Rea et al., 2003). Clarkin and colleagues (1998) psychoeducation for couples ran weekly for 10 sessions, then bi-monthly for remaining 15 sessions. Javadpour and colleagues (2013) conducted weekly face to face sessions for eight weeks then monthly telephone follow-up until 18 months. Booster sessions after the main intervention were a feature of three studies (Lam et al., 2000; Lam et al., 2003; Lenz, 2010).
3.5.1.3 Control group
Thirteen studies compared the intervention to TAU or standard care only (low or high
intensity), eight compared standard care plus an additional component and four described the comparison group as another form of therapy. TAU was classified into low intensity general care or intensive support (structured specialist support).
The comparison group comprised only low intensity general care in six studies (Clarkin et al., 1998; Harvey & Peet, 1991; Javadpour et al., 2013; Lam et al., 2000; Lam et al., 2005; Lam et al., 2003; Peet & Harvey, 1991; Reinares et al., 2008) and only high intensity in three studies (Bordbar et al., 2009; D'Souza et al., 2010; Sajatovic, Davies, et al., 2009). Two studies did not provide detail on what TAU constituted (Dogan & Sabanciogullari, 2003; van Gent & Zwart, 1991).
Where the comparison group received additional components, these included additional clinician training (Ball et al., 2006), very brief medication training (Eker & Harkin, 2012) and attention matched group therapy sessions or phone calls to control for contact time (D. Castle et al., 2007; Cochran, 1984; Colom et al., 2003). Additional education or psychoeducation sessions were provided to control groups, for example seven sessions of psychoeducation (the IG received the same and the addition of 13 sessions of CBT) (Zaretsky et al., 2008), individual patient treatment sessions with a therapist and medication management sessions with a psychiatrist for a year (Rea et al., 2003), the provision of a self-help book and additional group sessions (Lenz, 2010).
The electronic monitoring device intervention provided by Elixhauser and colleagues (1990) was compared against TAU and individual compliance feedback based on lithium levels and suggestions for improving compliance.
For the four studies where the comparison group received an alternative intervention, this took the form of a programme of therapy differing in content or focus: Intensive Clinical Management (ICM), Interpersonal and Social Rhythm Therapy (IPSRT), ‘Crisis management’, Family Focussed Therapy and ‘Individual patient management’ (FFT) (Frank et al., 2005; Frank et al., 1999; Gilbert, 2000; Miklowitz et al., 2003; Miklowitz et al., 2000).
3.5.1.4 Study & sample characteristics
The median number of participants in the studies was n=66 (IQR=45-102) and participants randomised to intervention groups was n=31 (IQR=19-54). The length of time of follow-up varied across studies, participants were followed up for a median of 6 months (IQR=0.5-12) after last intervention contact (Table3.2).
Over one-third of studies (k=9) were based in the United States, three in the UK, three in Australia, two in Spain, two in Turkey and Iran and one each in Canada, Austria and the Netherlands (Tables 3.8 and 3.9). Data was available in 21 studies for the proportion of males and females recruited to the study. Overall there was a mean of 55.72% (sd=12.21) females in
the sample, ranging from 25% to 82%. Seventeen studies did not specify the ethnicity of participants.
To determine diagnosis eligibility different criteria were use; DSM (k=17); Research Diagnostic criteria (k=2); MINI (k=2); Schedule for Affective Disorders and Schizophrenia (SADS (k=2);
unspecified (k=1). In 11 studies researchers recruited patients who were euthymic, 5 during episodes, 3 during manic episodes only, 1 recruited participants in all states, 4 studies did not report this (Table 3.2).
3.5.1.5 Adherence assessment and primary outcome
Thirteen studies reported the use of more than one assessment method, of which seven combined these into one adherence score or categorisation. Seven studies used a composite measure of adherence, combining a number of assessment methods which were then either rated for the level of adherence, or defined by pre-set criteria (Table 3.2). Subjective
techniques included self-report (k=11), informant report (k=3) and physician report (k=4).
Objective techniques (other than biological measures) were used in two studies (D'Souza et al., 2010; Elixhauser et al., 1990), nine studies used biological measures for example serum
medication levels.
Seventeen studies either assessed and reported adherence overall for all medications participants were prescribed (Ball et al., 2006; Bordbar et al., 2009; D. Castle et al., 2007;
Clarkin et al., 1998; D'Souza et al., 2010; Eker & Harkin, 2012; Gilbert, 2000; Javadpour et al., 2013; Lam et al., 2000; Lam et al., 2003; Lenz, 2010; Sajatovic, Davies, et al., 2009), or assessed them separately and then reported an overall adherence result (Frank et al., 2005; Miklowitz et al., 2000; Rea et al., 2003; Reinares et al., 2008; Zaretsky et al., 2008). Within these 17 studies, data was collected and reported for mood stabilisers only in four (Ball et al., 2006;
Frank et al., 2005; Gilbert, 2000; Lenz, 2010), for mixed medications such as mood stabilisers, anti-depressants and benzodiazepines in nine studies (Bordbar et al., 2009; Clarkin et al., 1998;
D'Souza et al., 2010; Lam et al., 2000; Lam et al., 2003; Miklowitz et al., 2000; Rea et al., 2003;
Reinares et al., 2008; Zaretsky et al., 2008). Four studies did not report what medications were assessed (D. Castle et al., 2007; Eker & Harkin, 2012; Javadpour et al., 2013; Sajatovic, Davies, et al., 2009). Only one study measured and reported results individually for different mood stabilisers, and this was only at the two year follow-up paper (Colom et al., 2005). Six studies only reported on lithium prescriptions (Cochran, 1984; Dogan & Sabanciogullari, 2003;
Elixhauser et al., 1990; Frank et al., 1999; N. S. Harvey & Peet, 1991; van Gent & Zwart, 1991).
The primary outcome included adherence (k=8) (three studies had adherence as the only primary outcome and five had multiple primary outcomes including adherence), relapse/
recurrence, symptoms (k=14), wellbeing/ functioning (k=7), coping with prodromes (k=1) and attitudes towards treatment (k=1), and was not specified for two studies (Clarkin et al., 1998;
Lenz, 2010).
Table 3.1: Intervention components/ delivery Intervention Component Number of
studies (k)
Electronic device, phone & mail Group & telephone
Study Component Number of
studies (k) n. participants IG, median (IQR) Follow-up from last intervention contact, median (IQR) mths
66 (45-102) 31 (19-54) 6 (0.5--12)
a Diagnostic and Statistical Manual of Mental Disorders
b MINI International Neuropsychiatric Inventory
c Schedule for Affective Disorders and Schizophrenia
d Treatment as usual
3.5.1.6 Quality Indicators (Retention, Fidelity, Intervention description, Study design – risk of bias)
Many studies reported a good participant retention rate in multi-session programmes, for example, 76% completing the full programme of nine months of treatment (Rea et al., 2003), 73% of the intervention group adhering to the programme (missed no more than five out of 21 sessions) (Colom et al., 2003), 97% taking part in the two session intervention sessions (Harvey
& Peet, 1991; Peet & Harvey, 1991), 89% participating in at least four psychoeducation sessions (average 7 sessions) (Javadpour et al., 2013). Sajatovic and colleagues (2009) found that 37% never participated in the sessions, 49% did complete at least four of the six sessions.
This data was not reported for nine studies (Table 1). Where the programme was delivered to caregivers, a good retention rate was reported, for example, 100% of families agreeing to participate completing the intervention (Bordbar et al., 2009), 95% of caregivers attending at least eight out of 12 sessions (Reinares et al., 2008).
Eighteen studies specified that the intervention was manualised although only nine studies mentioned some form of fidelity assessment. For five studies this was formal assessment by independent raters using published or adapted coding systems or scales (Ball et al., 2006;
Frank et al., 2005; Miklowitz et al., 2003; Miklowitz et al., 2000; Rea et al., 2003). The fidelity assessments reported good adherence to the manuals, compliance with the principles of the therapy and competent delivery. Fidelity assessments which were not formally scored or rated were used in four studies (Clarkin et al., 1998; Lam et al., 2000; Lam et al., 2003; Sajatovic, Davies, et al., 2009) for example, Sajatovic and colleagues (2009) reported that fidelity was assessed by the co-principle investigator delivering feedback at the end of each session.
Using the TIDieR checklist and guidelines, only two studies provided all detail required to allow replication (items 1-9) (Colom et al., 2003; Miklowitz et al., 2000). Adequate reporting was most common in; Brief name (100%), Why (rationale, theory or goal) (92%) and What (procedures used) (88%). Less well described items were Where (location) (54%), When and how much (58%) and What (materials used) (58%) (Figure 3.2). Reasons for coding items as
‘inadequately described are reported in Table 3.3.
Figure 3.2: TiDier reporting
Table 3.3: Summary of reasons for inadequate reporting of interventions TIDieR Item Studies adequately
reporting (%)
Example reasons for ‘inadequate reporting’ coding
1. Brief name 100.0 n/a
2. Why (rationale,
theory or goal) 91.7 Paper doesn’t link theory or rationale with actual intervention content or techniques.
3. What (materials
used) 58.3 No information was provided on where or how to access intervention content or materials, or manuals not referenced.
4. What
(procedures used) 87.5
Paper doesn’t provide sufficient detail to replicate the procedure, key details on timing, or how to actually carry out the
intervention is unclear.
5. Who provided
54.2
Information is required on who delivered the intervention, i.e.
job role, training or expertise and background. Papers often reported only job title.
6. How (mode of
delivery) 79.2 The descriptions provided did not report the size of the groups in group delivered interventions.
7. Where (location)
54.2
Either the context of the research (to provide information on the country and healthcare context) or the exact location of delivery of the intervention was not reported (for example in an outpatient clinic, hospital or research setting).
8. When and how
much 58.3 The duration of intervention sessions or the spacing of contacts was not reported.
9. Tailoring
66.7
Tailoring techniques are mentioned as being part of the therapy, but unclear how they were used or personalised or whether there was a two-way dialogue.
10. Modifications 8.3 n/a
11. How well (planned)
50.0 The study was manualised but no actual assessment of fidelity, or sessions were recorded, but no mention of whether or not an assessment of recordings was conducted.
12. How well (actual)
20.8 The information provided was not specific, ie ‘adherence to manual was high’ or the outcome of any fidelity checks was not reported.
The method of randomisation was judged to have low risk of bias in nine studies, high risk in one study (Dogan & Sabanciogullari, 2003) but was inadequately described to make an assessment in 14 studies. Three studies demonstrated low risk of bias in allocation
concealment, (Frank et al., 2005; Lam et al., 2003; Miklowitz et al., 2000), one demonstrated high risk of bias (Dogan & Sabanciogullari, 2003), in 20 studies and was judged to be unclear.
Descriptions of blinding of study personnel was inadequately described to judge risk of bias in nine studies and in 14 studies there was high risk of bias. Studies were judged to have a high risk of bias in participant blinding in 10 cases. Only one study each were judged to have low risk of bias for personnel blinding (Castle et al., 2007) and participant blinding (Reinares et al., 2008) (Figure 3.3). In terms of adherence outcome assessment, 10 studies included an outcome measure with low risk of bias, however, 12 studies only included measures where there was a high risk of bias (for example self-report).
Figure 3.3: Risk of bias in included studies (k=23).