Medication nonadherence in bipolar disorder: a narrative review

Abstract

A number of effective maintenance medication options exist for bipolar disorder (BD) and these are regarded as the foundation of long-term treatment in BD. However, nonadherence to medication is common in BD. For example, a large data base study in the United States of America (USA) showed that approximately half of patients with BD were nonadherent with lithium and maintenance medications over a 12 month period. Such nonadherence carries a high risk of relapse due to the recurrent nature of the illness and the fact that abrupt cessation of treatment, particularly lithium, may cause rebound depression and mania. Indeed, medication nonadherence in BD is associated with significantly increased risks of relapse, recurrence, hospitalization and suicide attempts and a decreased likelihood of achieving remission and recovery, as well as with higher overall treatment costs. Factors associated with nonadherence include adverse effects of medication, complex medication regimens, negative patient attitudes to medication, poor insight, rapid-cycling BD, comorbid substance misuse and a poor therapeutic alliance. Clinicians should routinely enquire about nonadherence in a nonjudgmental fashion. Potential steps to improve adherence include simple pragmatic strategies related to prescribing including shared decision-making, psychoeducation with a clear focus on adherence, reminders (traditional and digital), potentially using a depot rather than an oral antipsychotic, managing comorbid substance misuse and improving therapeutic alliance. Financial incentives have been shown to improve adherence to depot antipsychotics, but this approach raises ethical issues and its long-term effectiveness is unknown. Often a combination of approaches will be required. The strategies that are adopted need to be patient specific, reflecting that nonadherence has no single cause, and chosen by the patient and clinician working together.

Introduction

The United Kingdom (UK)’s National Institute for Health and Clinical Excellence (NICE) guidance on medication adherence defines adherence as ‘the extent to which the patient’s behaviour matches agreed recommendations from the prescriber’.¹ Adherence, from the Latin adhaerere (to cleave or stick to), is considered to suggest sustained active decision-making and ownership and is the term used in this manuscript rather than the more paternalistic ‘compliance’, from complere (to fill up, fill out or complete). Nonadherence does not only involve medication but can occur with other health care recommendations including nonpharmacological interventions, for example lifestyle changes and psychological treatment, and with scheduled appointments.² Nonadherence to medication is a major problem in all chronic medical and psychiatric disorders. This narrative review discusses the impact of medication nonadherence in bipolar disorder (BD) including the extent of the problem, contributory factors and potential management solutions. The published evidence used for this narrative review was sourced primarily using PubMed with additional use of EMBASE, Medline and PsycINFO. Initial searches consisted of keywords ‘Bipolar’ AND ‘Adherence’ OR ‘Nonadherence’. Systematic reviews, randomized control trials and observational studies that were published in the English language were reviewed. Qualitative studies were considered for patient-centred factors influencing adherence. Case reports and case studies were not included in our literature search. The focus was on adults over the age of 18 years.

General features of nonadherence

Adherence occurs on a spectrum. At one end of the spectrum is total adherence that is, all doses of medication are taken at the frequency specified on the prescription. At the other end of the spectrum is total nonadherence that is, none of the prescribed medication is taken. In between are varying degrees of partial adherence, or partial nonadherence, that is, some, but not all, of the prescribed medication is taken. Adherence is dynamic and a patient who adheres well at one point in time may adhere poorly at another time. Adherence can be selective in that it only affects some prescribed medications, or it may relate to all medications prescribed at that time point. Nonadherence can also refer to patients taking more than the dose prescribed but in this review, we restrict ourselves to the issue of less than the prescribed amount of medication being taken as this is a more common problem. In most research studies adherence is dichotomized, with patients being considered as adherent or nonadherent, though the definitions used to make this categorization vary. As a result, it can be difficult to compare studies. For example, Hong defined nonadherent patients as those who took medication for at least half the time.³ In contrast, other researchers have defined nonadherence as the patient missing at least 20% of the total prescribed medication over a specified time period.⁴ Definitions apart, another methodological issue relates to how adherence is measured. A distinction can be made between subjective measures, such as a patient’s diary, report or questionnaires, and more objective measures, such as pill counting, monitoring drug serum levels or medication containers with inbuilt electronics to record when the container is opened.⁵ In the latter case, it is assumed that opening the container equates to taking medication but in reality this is not always the case, so even this apparent ‘gold standard’ method may overestimate adherence.

In clinical practice, nonadherence may be covert or overt. Sometimes nonadherence is rather simplistically divided into deliberate and nonintentional.⁶ Cited examples of nonintentional nonadherence include forgetting to take medication, the inability to meet the financial cost of the prescription or misunderstanding the instructions regarding medication-taking. Misunderstandings are considered to occur in various ways, for example due to language barriers, a patient with poor eyesight who cannot read the instructions on the medication container label, or a doctor who did not explain the instructions clearly, especially if the patient was distracted in the consultation. Deliberate nonadherence is said to occur when the individual makes a conscious decision to take less than the prescribed amount of medication. In practice such categorizations are unhelpful and adherence reflects the person’s attitudes and beliefs about the medication, the underlying illness and the relative importance they give to the pros and cons of taking medication.

Costs of nonadherence in BD

Although many psychiatric disorders follow a relapsing–remitting condition, BD has a higher frequency of relapses than, for example, unipolar depression and the relapse risk remains constant throughout the life span.⁷ As such, on average, it is reasonable to assume that nonadherence in BD is more likely to be associated with relapse over a given time period than nonadherence in a less recurrent illness such as unipolar depression. While some patients with BD undoubtedly have long spells of recovery, long-term follow up shows that, on average, patients are symptomatic around 50% of the time.^8,9 During much of this time patients are suffering from subsyndromal symptoms,^8,9 but even low levels of symptoms are associated with impaired psychosocial functioning.^10,11 Prophylactic medication, ideally supported by psychosocial interventions such as psychoeducation,^12,13 is regarded as the cornerstone of managing BD.^14,15

A recurrence in BD can have a high cost for the individual. BD has the highest suicide rate of any psychiatric condition.¹⁶ Most suicides occur in depressive episodes, though mixed episodes also carry significant risk. Manic episodes can result in inappropriate behaviour that can have a marked impact on a person’s future in terms of employment and personal life, particularly if the inappropriate behaviour involves finance or sexual behaviour. Long-term pharmacological treatment of BD has been shown to be associated with a significantly lower suicide rate¹⁷ which may in part relate to the specific antisuicidal effect of lithium.¹⁸ The effects of medication nonadherence in BD were investigated in an analysis using data from the EMBLEM (European Mania in Bipolar Longitudinal Evaluation of Medication) study, a 21-month prospective observational study of patients who had experienced a manic or mixed episode. Adherence was based on clinician ratings. Nonadherence was associated with a significantly increased risk of relapse, recurrence, hospitalization and suicide attempts, and a decreased likelihood of achieving remission and recovery.³

Nonadherence in BD has a major economic cost for health services. The annual cost to the National Health Service (NHS) in the UK was estimated to be £342 million at 2009/2010 prices, with 60% of this accounted for by inpatient admissions.¹⁹ In the EMBLEM study, the total costs incurred by nonadherent (ranging from never taking medications to taking them approximately half the time) compared with adherent (almost always taking their medications) BD patients over a 21-month period of follow up was £10,231 compared with £7379 (p < 0.05).³ The difference was in large part due to the costs of inpatient care, which averaged £4796 for the nonadherent group but £2150 for the adherent group.³ Consistent with this data, Gianfrancesco and colleagues conducted a claims database study in the USA that showed that improved adherence to antipsychotic medication in BD was associated with lower total and outpatient mental health care expenditures.²⁰

Lithium remains the gold standard medication for long-term prophylaxis of BD.^14,15 However, abrupt discontinuation of lithium is associated with an increased risk of relapse.²¹ Baldessarini and colleagues assessed the effect of discontinuing treatment abruptly (1–14 days) or gradually (15–30 days) in patients with BD.²² The median time to recurrence ± SE was five-fold greater in the rapid discontinuation group (20.0 ± 5.8 versus 4.0 ± 0.7 months; p < 0.0001). The increased relapse risk was seen for both depressive and manic relapses and in people with bipolar I and II disorders, though it was greater in bipolar II patients. These observations led Goodwin to argue that lithium’s benefits are only likely to be realized if taken for at least 2 years with complete adherence.²³ Observational data from a 6 year study of 1594 lithium users aged 15 year and above from a Health Maintenance Organization based in the USA suggest that the median duration of continuous lithium adherence is just 72 days²⁴ and a more recent nationwide study in Denmark found a median of just 181 days before lithium was discontinued.²⁵

Prevalence of medication nonadherence in BD

Prevalence of nonadherence will be influenced by the methodology adopted to assess it as well as the characteristics of the patients being studied. This probably explains the wide variation in rates of nonadherence reported in the literature across medical disorders, with BD^26,27 being no exception. However, nonadherence rates of approximately 50% are commonly reported for patients with BD^28–31 and in a survey of over 2000 psychiatrists, based in eight European countries, the respondents estimated that the majority of patients (57%) with BD were nonadherent or partially adherent to treatment.³²

The medication possession ratio (MPR) is a method of monitoring adherence based on pharmacy refills of medications. Since it does not directly record tablet ingestion it is likely to overestimate adherence. Sajatovic and colleagues used the MPR to examine a patient register in the USA to assess adherence to lithium and anticonvulsant mood stabilizers in BD and found that over half the sample had an MPR of less than 0.50.³⁰ Patients were divided into three groups according to whether they were nonadherent (MPR < 0.50; 54% of the sample), partially adherent (MPR 0.50–0.80; 24%) or fully adherent (MPR > 0.80; 21%). In an earlier study, Scott and Pope interviewed 98 patients with affective disorders who were prescribed a mood stabilizer (i.e. lithium, carbamazepine, or valproate). The majority of the patients had a diagnosis of BD (n = 78) and the remainder (n = 20) had a diagnosis of major depressive disorder. Approximately half reported some degree of medication nonadherence in the last 2 years and nearly one-third reported missing 30% or more of their prescribed medication in the last month.²⁹ Being only partially adherent, and having subtherapeutic plasma levels, were the most significant predictors of admission to hospital over the following 18 months (Cox regression analysis, p = 0.001). Interestingly, there was little correlation between subjective reporting of adherence and objective measurement with drug plasma levels.²⁹

The high prevalence of nonadherence in BD and the high associated costs make it imperative that clinicians routinely and nonjudgementally, enquire about, quantify, and explore nonadherence during patient review.

Factors influencing adherence

Sociodemographic factors

Sociodemographic factors, including socioeconomic status, do not appear to be major determinants of adherence. However an online survey suggests that younger patients are more likely to be nonadherent to lithium and anticonvulsants with adherence increasing after the age of 41 years.³³ Kessing and colleagues have also identified that adherence was lowest among patients below the age of 40 years, however rates were also lower above the age of 60 years.²⁵ Other social factors, include being non-White, unmarried, homeless and having a diagnosis of substance misuse disorder have been reported as being associated with nonadherence.³⁰

The role of sex and gender identity in adherence has been explored by Sajatovic and colleagues in a cross-sectional study of BD patients. Overall there appears to be no significant differences in adherence between men and women. However, men with high masculinity attitudes and perception scores had lower adherence to medications compared with other men. For women, there was no relationship between masculinity scores and adherence.³⁴ This is slightly at odds with a study of 225 patients with bipolar I or schizoaffective disorder which found that women were more likely to be represented in the nonadherent subgroup.³⁵

Attitudes to medication

Given that general demographic features have a limited ability to predict levels of adherence,³⁶ coupled with a move away from the concept of ‘compliance’ to the more collaborative ‘adherence’,³⁷ has led to a shift from an ‘illness-centred’ to a ‘patient-centred’ approach to medication usage. Such a shift has been argued to improve understanding of nonadherence by emphasizing the importance of negative attitudes, beliefs and stigma.³⁸

The perception of the strength of the ‘therapeutic alliance’ between patient and care giver, and the optimization of this collaborative relationship, can itself be seen as a management approach to BD. Stronger alliance has been associated with a reduction in manic symptoms with less negative attitudes about medication and reduced perception of stigma regarding BD.³⁹

Patients knowledge about their illness and medications can also impact on adherence. In a study of 223 patients using the Satisfaction with Information about Medicines Scale those whose perception that they have received less than satisfactory knowledge about their medications presented with significantly lower adherence rates.⁴⁰ A belief that antipsychotic medications are exclusively for schizophrenia, and a reluctance to discuss side effects such as weight gain, can also potentially lead to nonadherence.⁴¹

In a study using the Lithium Attitudes Questionnaire (LAQ) and the Lithium Knowledge Test (LKT), Rosa and colleagues found that in a sample of 106 outpatients with BD, better pharmacological knowledge on the LKT was associated with lithium levels being in the therapeutic range. In contrast, the main factor of ‘denial of disease’ in the LAQ significantly contributed to lower plasma levels. Interestingly, younger patients had better understanding of lithium.⁴² The relationship between negative attitudes to medication and nonadherence has also been shown in studies examining other medication in BD.^43,44 Importantly, patients with positive attitudes towards medication have significantly greater clinical stability.⁴⁴

In addition to patient attitudes, the patient’s perceived criticism from family attitudes and knowledge towards medication may also influence adherence and thus clinical outcome.⁴⁵ This is supported by data showing that positive medication attitudes were seen in patients with BD who had good social networks and who had an external locus of control, holding the belief that others such as clinicians and family determined their clinical outcomes.⁴⁶ Similarly, it has been reported that if a caregiver is emotionally over-involved with a patient with BD, this is associated with reduced medication adherence.⁴⁷

Insight

Studies in schizophrenia suggest that lack of insight is the most common driver for poor adherence.⁴⁸ Consistent with this, nonadherent BD patients have been shown to have lower levels of insight compared with adherent patients⁴³ and that better insight is associated with higher adherence.⁴⁹ However, some studies in BD have failed to find this relationship.^44,50 A limitation of these studies is that they primarily used subjective reporting of adherence.

Course of illness

It is important to consider that the level of adherence to medication may fluctuate with changes in the mental state of patients with BD.⁵¹ Poor adherence has been associated with rapid cycling, previous suicide attempts, earlier onset of illness, active anxiety or alcohol use disorder, a greater number and severity of affective symptoms, not being in full remission, having comorbid obsessive compulsive disorder, and having suffered from a recent manic or hypomanic episode.^52–54

Substance misuse

Substance misuse is a common comorbidity in BD^55,56 that can have a negative impact on outcome. For example, alcohol misuse is related to increased mood fluctuation⁵⁷ and increased suicide risk.⁵⁸ Multiple studies have shown an association between nonadherence to prescribed medication and substance misuse. Montes found a greater prevalence of comorbid substance abuse and dependence (of alcohol, cocaine and cannabis) in patients with BD who had poor medication adherence.⁵⁹ Similarly, Manwani and colleagues compared adherence with mood stabilizers between BD patients with and without comorbid substance misuse and found significantly lower adherence rates in those who misused illicit substances.⁶⁰ The association between nonadherence and substance misuse is complex, with several different paths of causality. For example, some patients who do not adhere to prescribed mood stabilizers will relapse as a consequence and the deterioration in their mental state can lead to substance misuse through attempts to alleviate symptoms (self-medication) or through increased impulsivity and recklessness (for example, when manic). Substance misuse may also destabilize the mental state which then leads to nonadherence. Another pathway between substance misuse and nonadherence is that the substance misuse leads to a chaotic lifestyle that makes regular medication-taking difficult.

Cognition

BD is associated with a range of well described cognitive dysfunctions as detailed in an individual patient data meta-analysis of 2876 patients.⁶¹ It is seen across all clinical subgroups,⁶² including in patients when euthymic,^63,64 and there is little evidence of progression of dysfunction over time.⁶⁵ The dysfunction may relate to underlying attentional and processing speed abnormalities⁶⁶ but deficits are seen across all cognitive domains including memory.⁶¹ Around 40% of patients with BD score below the 5^th percentile for their age and sex in two or more cognitive domains.⁶⁷ These may particularly be patients who are experiencing a sleep disorder, insomnia, hypersomnia, circadian rhythm disorder or obstructive sleep apnoea (McAllister-Williams and colleagues, 2018 under review).

There have been two previous studies involving euthymic BD outpatients examining the relationship between cognition and adherence. The first studied 103 patients and found impaired ability on the trail-making task and spatial memory deficits were associated with reduce adherence rates.⁶⁸ The second study of 353 euthymic outpatients found a relationship between poor inhibitory control and reduced adherence.⁶⁹ However it is not possible to conclude the direction of causality between cognitive dysfunction and adherence from either of these studies.

Complexity of the medication regimen

In other areas of medicine, increased complexity of medication regimes is associated with decreased adherence rates. For example, a systematic review of 76 studies (mainly in cardiovascular disease) that used electronic monitoring devices found significantly higher adherence with once daily dosing compared with three times a day, though not twice daily dosing schedules.³⁶ If nonadherence is not recognized by the prescriber, and it is wrongly assumed that a patient is treatment resistant, additional drugs may be added to the regimen which may reduce adherence further.

Medication side effects

Side effects can be an important factor contributing to nonadherence. However, it is the patient’s perception of side effects and how this compares with their perception of the benefits of medication that is key in determining whether side effects impact on adherence. An online survey of 469 patients with self-reported BD identified that weight gain within 3 months was the side effect most likely to impact on adherence to bipolar medications, followed by cognitive side effects.³³ A 2005 study comparing divalproex with lithium in rapid-cycling BD found that in the double-blind monotherapy phase, tremors, polydipsia and polyuria were significantly more common with lithium compared with divalproex. However, the study did not establish significant differences in the discontinuation rates due to side effects between these treatment arms.³⁴ The side effect profiles of medications used in BD differ markedly and this provides scope for patients and clinicians to work together to prevent or alleviate certain side effects.

Assessing medication side effects can be problematic, both in clinical practice and in clinical trials. Symptoms of BD can sometimes be misattributed to side effects, and vice versa. For example, sedation from a medication may be difficult to differentiate from lethargy associated with a depressive episode. Furthermore, mood states such as depression can impact on the subjective reporting of side effects. Moreover, being on multiple medications will have an impact on the patient’s overall side effect profile. Side effects may also change during the course of treatment. For example, with lithium polydipsia and polyuria become increasingly problematic in the longer term, with rates of up to 70%.³⁵

Improving adherence in BD

As described above, poor adherence can have many causes. It is therefore important to understand at an individual patient level why nonadherence has occurred as this can help the clinician to formulate a specific treatment plan. Potential strategies to improve adherence in BD are reviewed below though it should be noted that these strategies often overlap.

Simple prescribing strategies

There are a number of simple prescribing strategies that can be employed to try to maximize adherence.⁵ These include adopting shared decision-making, using as simple a medication regimen as possible in terms of both the number of medications and the frequency of medicating taking, and effective management of side effects. At the outset, it is vital to as far as possible have full and frank discussions about treatment and the different medication options that are available, as well as the benefits and risks of each. This also applies to individuals considering not altering their current medication or not starting medication at all. A mutual agreement should be reached on the treatment plan. The NICE guidelines for BD Assessment and Management advise that, for pharmacological interventions, clinicians must take into account any advance statements, the patient’s preference, and clinical factors such as comorbid physical ill health and previous adverse reactions to treatment.¹⁴

Managing medication side effects is a crucial part of good clinical care. It starts before medication is prescribed with the patient and clinician considering the likely impact of side effects for that person, the side effect profiles of potential medication and factoring this in to a shared decision about the most appropriate medication(s) to use. Throughout treatment side effects should be screened for regularly and managed when they occur. It is outside the remit of this paper to consider the management of side effects, but we briefly mention weight gain as an example, as this is particularly distressing to patients and occurs with many medications used in BD including antipsychotics, lithium and valproate. Antipsychotics differ significantly in their propensity to cause weight gain and will usually be a factor that patients and prescribers consider when making joint decisions about selecting medication. Switching to an antipsychotic with lower risk of weight gain can help reduce antipsychotic associated weight gain.⁷⁰ Weight management, including dietary change and encouraging physical activity, are effective in reducing antipsychotic associated weight gain. Metformin is effective in reducing antipsychotic induced weight gain⁷¹ and can be considered on a case by case basis though it is important to highlight that this is an off-label indication.

Improving therapeutic alliance

The therapeutic alliance between patients and health care professionals has a key role in supporting adherence to medication. A study of 3037 patients with BD found that patients’ perceptions of collaboration, empathy, and accessibility of their psychiatrist were significantly associated with adherence to treatment.⁷² Similarly, it has also been shown that a positive therapeutic alliance is associated with better adherence.⁷³ In terms of improving adherence, assessment of a patient’s understanding of their diagnosis, symptoms and the potential risks associated with stopping medications, is likely to be critical. The communication style of the doctor is important to their ability to forge a therapeutic alliance. Alliance may be weakened when a patients’ care lacks clinical continuity, for example, they see different doctors at different appointments or receive their care from different teams at different time points, for example, the community mental health team, inpatient team and crisis home treatment team. Sometimes the involvement of multiple teams is unavoidable, and it may carry some advantages for example in terms of specialization and greater input at a time of high clinical need. However, services need to ensure a seamless transfer between teams with timely handovers.

An additional concern that may affect adherence is that regarding the impact of medication in pregnancy for women with BD, particularly given the concerns of teratogenic and other adverse outcomes for infants exposed in utero to a number of commonly used BD medications.⁷⁴ It is important for prescribers to anticipate such concerns in all women of child-bearing potential with BD.

Psychoeducation

There is a lack of strong evidence supporting specific psychotherapies for BD.¹⁵ However, there is some evidence that psychoeducation improves medication adherence and short term knowledge about medication⁷⁵ and that this can have long-term beneficial effects on patient outcomes.¹² Psychoeducation is very broad concept and cover many areas; it is more likely to improve adherence if there is a clear focus on adherence and incorporating behavioural interventions.⁷⁶

Family and carer attitudes towards BD and its treatment can have a major impact on patient adherence.⁷⁷ It is therefore potentially important to consider some form of psychoeducation for significant individuals in a patient’s life if this is at all possible or practical.⁷⁸ A recent study of 270 patients, of which 136 in the experimental group received five sessions of motivational interviewing and psychoeducation with family members, identified increased adherence rates when using the Medication Adherence Rating Scale.⁷⁹ Addressing self-stigmatization with patients requires not only educating individual patients but also families, care providers and wider society.⁴¹

Managing substance misuse

Comorbid alcohol and illicit substance misuse can complicate all psychiatric conditions and BD is no different. The evidence that substance misuse can be associated with nonadherence⁶⁰ is another reason why it is important to actively address substance use and BD at the same time. This may require collaborative work between general adult psychiatry and addiction services. Increasing patient awareness of substance misuse as a potential relapse indicator, and ensuring this leads to an urgent assessment, is also important irrespective of the direction of causality between relapse and substance misuse.

Depot antipsychotics

A significant advantage of a depot antipsychotic (also known as antipsychotic long-acting injections; LAIs) over oral antipsychotic medication is that covert nonadherence is impossible.⁵ All clinical guidelines for the management of schizophrenia regard depots as being an option to manage prior nonadherence with oral medication. However, the prescription of depot does not necessarily guarantee adherence as the patient must be agreeable to taking it. A number of guidelines advocate the use of depot treatment in prophylaxis against manic relapses and in particular those patients who are nonadherent to oral formulations (see Table 1). It is important to highlight that in general placebo-controlled randomized controlled trial (RCT) evidence for the long-term use of antipsychotics in BD suggest that they primarily risk of recurrence of mania and not depression, with the exception that quetiapine⁸⁰ (and possibly lurasidone⁸¹) protects against depressive relapse as well. Neither quetiapine nor lurasidone are available in a LAI formulation. It is also assumed that the prophylactic effect of an antipsychotic against relapse of mania is more likely if there is evidence that an individual’s previous acute manic episodes have responded to that antipsychotic (usually given in oral form).

Table 1.

Summary of advice on use of antipsychotic long-acting injectables in the maintenance treatment of bipolar disorder in key guidelines.

Guidelines	Recommendations
CANMAT and ISBD, 2013 update82	Maintenance treatment for bipolar mania. Risperidone LAI considered as the first line.
Maudsley Guidelines, 12th edition, 201583	The use of FGA depots is commonly used in clinical practice but is associated with higher risk of depression. Risperidone LAI shows efficacy in treatment of mania, but due to pharmacokinetics it is considered unsuitable choice for acute treatment of mania.
RANZCP, 201584	Maintenance treatment for prevention. LAI risperidone has some support for manic and depressive relapses.
NICE, 2016 update14	The Guideline Development Group stress that lithium treatment has the most robust evidence as a first-line treatment, but that the NICE Guidance on Psychosis and Schizophrenia 2014 can be adapted for recommendation for bipolar patients and include risperidone LAI in the review of the evidence but include no other LAI.
BAP, 2016 update15	Good practice to initiate effective and tolerated dose with oral formulation before switching to LAI. Only risperidone LAI has licensed RCT support and is consistently positive for treating bipolar mania and not depression. Fluphenazine and haloperidol decanoate, olanzapine pamoate, paliperidone and aripiprazole depot can be used. Case by case clinical practice review is advised when considering these alternatives.

CANMAT, Canadian Network for Mood and Anxiety Treatments; FGA, first-generation antipsychotic; ISBD, International Society for Bipolar Disorders; LAI, long-acting injection; NICE, National Institute for Health and Clinical Excellence; RCT, randomized controlled trial.

In comparison with schizophrenia the evidence base for the use of depots in the maintenance treatment of BD is smaller. The efficacy of risperidone LAI in the prevention of relapse in bipolar I disorder was assessed in a randomized, placebo-controlled study.⁸⁵ Following an open-label treatment period with risperidone LAI, stable patients entered an 18-month double-blind period in which they were randomized to continue treatment with risperidone LAI or switch to placebo or oral olanzapine. Time to recurrence of any mood episode was significantly longer with risperidone LAI versus placebo. However, the difference was significant for time to recurrence of elevated mood episodes but not depressive episodes. This is consistent with the fact that antipsychotics, other than quetiapine and possibly lurasidone, have only been shown to have a consistent maintenance effect in BD in terms of prevention of manic episodes and not depressive episodes. Risperidone LAI was approved as a maintenance treatment for BD by the United States Food and Drug Administration (US FDA) in 2009 as a monotherapy as well as an adjunct to sodium valproate or lithium.⁸⁶ To date it is the only depot approved for use in BD. It is not licensed for this indication in Europe.

The efficacy, safety, and tolerability of aripiprazole once-monthly (AOM) 400 mg LAI (400 mg) has recently been assessed as maintenance treatment for bipolar I disorder in a double-blind, placebo-controlled, 52-week randomized withdrawal study following a manic episode and stabilization on oral aripiprazole.⁸⁷ The risk of recurrence of any mood episode over 1 year was reduced by approximately half with AOM 400 compared with placebo, with the effects observed predominantly on manic and mixed episodes but not depressive episodes. AOM was generally safe and well tolerated.

A small case series has described the use of paliperidone palmitate LAI in BD. Overall, three patients with severe psychotic BD, poorly compliant with oral treatment, were started on (open-label) paliperidone palmitate LAI (100–150 mg monthly) and followed up for 12 months. None of the patients relapsed during follow up or experienced adverse effects or significant metabolic changes.⁸⁸

The choice between using a first-generation antipsychotic (FGA) depot or a second-generation antipsychotic (SGA) LAI in BD raises several issues. Firstly, no FGA antipsychotic, oral or depot, is approved for the maintenance treatment of BD. In contrast, several SGA oral antipsychotics (including aripiprazole, olanzapine, quetiapine and risperidone) are approved for the maintenance treatment of BD if an acute bipolar episode has responded to that drug (other than with quetiapine, this acute episode is defined as a manic episode). The efficacy of risperidone LAI has been compared with FGA depots in a BD retrospective cohort design of 3916 patients in Taiwan.⁸⁹ The findings demonstrated that compared with risperidone LAI, the use of FGA depots was associated with a higher rate of hospitalization for any mood episode and for major depressive episode. Observations were limited to severe episodes requiring hospitalization, so the findings may not be generalizable to milder mood episodes. A second issue relates to risk of extrapyramidal side effects (EPS). Meta-analyses indicate that patients with BD are more prone to develop extrapyramidal side effects when treated with FGAs than are patients with schizophrenia and that the vulnerability of BD patients to develop EPS is particularly marked during depressive episodes schizophrenia.^90,91 A third issue is that FGAs may be depressogenic in BD⁹² whereas this has not been shown with SGAs. A meta-analysis of RCTs concluded that there was a 42% lower risk of switching to depression after treatment of acute mania with a second-generation antipsychotic (SGA) compared with the use of haloperidol.⁹³ On the basis of their potential depressogenic effect, Bond and colleagues argued that ‘depot FGAs should be avoided in patients with a high burden of illness from depressive symptoms and particularly in those judged to be at high risk of suicide’.⁹⁴ In summary, current data suggest that SGA LAIs confer several advantages over FGA depots in the maintenance phase of BD. The main disadvantage of SGA LAIs is their higher acquisition cost relative to FGA depots.

It should be noted that, even in schizophrenia, the evidence base for the relative effectiveness of oral and depot antipsychotics is far from clear, with observational studies tending to be more supportive of depot whereas most RCTs find oral and depot to be equally effective. The fact that different trial designs give different results is generally regarded as reflecting methodological issues.⁵ In particular, RCTs are likely to recruit relatively adherent patients. Furthermore, in a double-blind design, those in the oral treatment arm will be reviewed at a frequency equal to that of those in the depot arm in order to ensure blinding. In reality, a stable patient on oral medication would be reviewed far less frequently than a patient on depot who will be typically be seen every 2–4 weeks for the administration of their depot. As such the double-blind design distorts the ecological validity of the findings and the outcome cannot be assumed to represent that seen with oral antipsychotic treatment in clinical practice.⁵

Reminders

Reminders are more likely to be effective if nonadherence is due to forgetfulness as opposed to intentional. There is a lack of research evidence in mental illness comparing the effect of simpler reminder interventions from blister packs or multi-compartmental devices, telephone reminders, or face-to-face reminders on medication adherence rates. In recent years attention has focussed on the use of electronic technology systems to assess and improve adherence.

Medication event monitoring system (MEMS) use micro-technology to detect when a medication container is opened. Typically, this would equate to a medication bottle cap being removed, though the technology can be adapted to other containers including multi-compartmental containers labelled with the day and time that each medication is due. There is then an assumption that an opened container equates to medication being taken. Clearly there is the possibility that patients open their medication bottle, extract the medication and throw this away.⁵ The use of MEMS may still have to be used in conjunction with self-reporting measures such as the Tablets Routine Questionnaire where issues of cost and polypharmacy and forgetfulness become an issue with the MEMS device.⁹⁵ Such data support MEMS as a measure of adherence rather than an intervention to help improve this. Further research is required to explore the potential for systems that feedback to patients if a bottle opening is not detected. This would require remote monitoring and the use of feedback systems to the patient such as text messaging. Such systems are now available.

An alternative approach to the remote monitoring of patients is the use of digital sensors that can be imbedded within a tablet.⁹⁶ Following ingestion, the sensor is activated by gastric acid and sends a signal to a receiver worn as a patch by the patient. This receiver links by Bluetooth to the patient’s smart phone which in turn sends a signal to a remote monitoring centre and feedback on adherence (i.e. ingestion of the sensor) can then be sent to the patient or the medical team looking after them. Feasibility and acceptability has been confirmed in a small study of 12 nonparanoid patients with BD and 16 with schizophrenia.⁹⁷ Ongoing clinical feasibility trials which do not use paranoia as an exclusion factor will help elucidate the acceptability of this approach in a broader cohort of patients.

Financial incentives

Financial incentives have been shown to reduce tobacco and illicit drug use in people with serious mental illness.⁹⁸ A recent RCT showed that financial incentives improved adherence to antipsychotic depots.⁹⁹ This study primarily recruited patients with schizophrenia, though those with schizoaffective disorder and BD were also eligible to enter. The study only lasted 12 months and so the effectiveness of continuing financial benefits beyond this is unknown. The approach also raises potential ethical issues.

Conclusion and further research

Nonadherence is a common problem in BD and is strongly associated with poorer clinical outcomes. It also places an increased demand on mental health services and increases service costs. It is reasonable to assume that the total costs of BD incurred by society is increased by poor adherence.

Addressing nonadherence needs to be on an individual patient basis, as factors that contribute to its occurrence may differ significantly between patients. Further research is required to establish a better understanding of the complexities of medication adherence in BD.¹⁰⁰ However, there are general principles that clinicians can adopt to reduce the likelihood of nonadherence. These include ensuring patients have adequate knowledge about their illness and its management, that any misperceptions are addressed, and that there is a positive therapeutic alliance.

Prescribing ideally should occur within a shared decision-making process, with the medication regimen kept as simple as possible. It is important to aim for total symptom control, as a patient is unlikely to take medication if it is perceived to be of little benefit. Comorbidities, especially substance misuse, need to be managed. Screening for side effects, and appropriate intervention when they are detected, needs to be a routine part of management. Throughout treatment there needs to be nonjudgemental exploration of adherence to prevent it being covert. When nonadherence is identified the factors leading to it need to be explored as these will guide the intervention that is adopted. A range of additional interventions are available including psychoeducation, reminders and adherence aids.

A depot antipsychotic may have a role where there has been nonadherence with an oral antimanic maintenance agent. The current data available support the use of risperidone and aripiprazole LAI for the prophylactic treatment of mania in patients nonadherent to oral medication. For patients with BD at risk of self-harm or in whom depressive symptoms are or have been clinically problematic, we would suggest a FGA depot should probably not be considered due to their potential depressogenic risk. If a LAI antipsychotic is considered appropriate for a patient with BD (because of poor oral adherence) a SGA LAI should be chosen as a first-line treatment over a FGA antipsychotic. Although the majority of the SGA LAI literature in BD is on risperidone LAI, there is recent support for aripiprazole monthly LAI, and paliperidone LAI (monthly or 3-monthly) may be considered based on its pharmacology related to risperidone, tolerability, and patient acceptability. For those rare patients whose BD includes only manic episodes, it may be reasonable to consider a FGA depot. However, the evidence base for their effectiveness is inferior to that for SGAs and on these grounds a SGA LAI is likely to be the more evidence-based choice. Conclusions drawn from the clinical comparability literature between FGA depots and SGA LAIs in schizophrenia cannot be extrapolated uncritically to BD.

In terms of future research, further exploration of the attitudes and knowledge of carers regarding BD and medication used to treat it, and its relationship to patient adherence, would be valuable.¹⁰¹ Further research is needed on the role of depot antipsychotics in BD, including the relative risks and benefits of different FGA and SGA medications. Digital technology, including smart phone apps and smart medication containers, can be used as platforms to monitor and facilitate adherence¹⁰² but this work is in its infancy.¹⁰³ The effectiveness and acceptability of these approaches requires further work but may appeal to a younger generation given the younger onset of age of BD patients.