CHAPTER 3 IMPROVING MEDICATION ADHERENCE IN BIPOLAR DISORDER: A SYSTEMATIC
3.1 B ACKGROUND
In assessing the need for a novel intervention, it is essential to review the existing literature on interventions in this field. From this, it is possible to identify whether high-quality, effective interventions have been conducted within the population in question. It can also identify which intervention content and delivery methods are worthwhile of development in order that previous work is built upon as opposed to developing new interventions from scratch each time (Campbell et al., 2000; Coyne, Thombs, & Hagedoorn, 2010; Liberati et al., 2009). In addition, important information can be obtained on how best to target interventions and whether there is evidence to suggest who might benefit most and who might be unlikely to respond.
Reviewing the existing evidence prior to developing an intervention is recommended by MRC guidelines for developing complex interventions (Craig et al., 2008) and the more detailed, Intervention Mapping (IM) process which is described in Chapter 5 (Bartholomew et al., 2011).
This systematic review forms part of the first stage of the process of conducting a needs assessment to identify what methods have been used to attempt to improve outcomes, what has worked in terms of population samples and contexts, and what gaps need to be addressed by conducting new empirical research.
Systematic reviews should also provide an assessment of the quality of studies included, therefore whether the findings concerning the intervention effectiveness is likely to be subject to bias (Petticrew & Gilbody, 2004). In addition to assessing intervention effectiveness, it is crucial also to examine the reporting of interventions in order to identify whether potentially promising interventions could be replicated (Hoffmann et al., 2014).
In the same way that a novel intervention would not be conducted if an effective, appropriate high quality one existed, prior to conducting a review, it is necessary to conduct a scoping review in order to determine if a high-quality, up to date review had already been published.
This identifies firstly if a review is necessary and secondly to identify areas which may not be covered either by the scope of previous reviews, specificity of inclusion criteria and whether an adequate assessment of intervention and study quality has been conducted.
3.1.1 Limitations of previous reviews
Systematic reviews of trials of programmes to improve outcomes in the area of severe mental health have been conducted however, they have limitations which reduce their use in
identifying effective techniques to improve medication adherence. Some focussed only on psychoeducational or psychosocial interventions (Batista, Baes, & Juruena, 2011; L. Berk et al., 2010; Crowe, Porter, et al., 2012; C. Depp, Moore, Patterson, Lebowitz, & Jeste, 2008;
Gaudiano, Weinstock, & Miller, 2008). Others did not examine adherence outcomes (Lolich, Vazquez, Alvarez, & Tamayo, 2012), provide only a narrative review (Busby & Sajatovic, 2010;
Colom & Lam, 2005; Miklowitz, 2006), or were not specific to BD (Desplenter, Simoens, &
Laekeman, 2006; Fernandez, Evans, Griffiths, & Mostacchi, 2006). Together these provide a limited picture of the adherence intervention literature in BD, as they do not systematically quantify the magnitude of intervention effects across studies, or test what components of study and intervention design may be most effective. The effects of inadequately powered studies may also be over-estimated in narrative reviews. Meta-analysis provides a
quantification of effect and we can test whether particular components may significantly contribute to the effect. Reviews are also limited by the fact that they investigate what is ostensibly a single illness, but what is in reality a highly variable condition, with different ways of characterising the diagnosis and different experiences and beliefs held by participants. In the same way, they group together ‘adherence’ as a behaviour, which is in reality a highly variable behaviour, measured using different techniques and different cut-offs are applied.
3.1.2 Need for a comprehensive review in BD
It is important to systematically synthesise the evidence and conduct meta-analysis, where appropriate, to quantify the effect of interventions. In addition, this method can test whether particular components of the study design or intervention may significantly contribute to effects and whether effects depend on study quality, which is not possible in narrative reviews.
Interventions which are taken forward and used in clinical practice need to be clinically effective, but also cost and time efficient for providers and patients. As such, they should include only the necessary components and be delivered only for as long as needed. They need to be appropriately targeted to ensure they are delivered to participants who might benefit and not suffer unwanted effects. The intervention and study components tested in this review reflect current thinking of what may be useful in improving adherence and by conducting a meta-analysis, this allows for confident recommendations to be made on what makes for an effective intervention.
Certain interventions components may be important, for example; tailoring (L. Berk et al., 2010; Crowe, Porter, et al., 2012; Desplenter et al., 2006; Horne et al., 2005), content and delivery (ABC Project Team, 2013) specifically the focus given to adherence, the length of the programme (Batista et al., 2011; Gaudiano et al., 2008; Rouget & Aubry, 2007), whether it is educational, includes a focus on beliefs and cognitions (Desplenter et al., 2006; Fernandez et al., 2006; Gaudiano et al., 2008; Horne et al., 2005; Leclerc et al., 2013; Sajatovic, Davies, &
Hrouda, 2004). Targeting and delivery variables which are explored include involving partners and family members (Gaudiano et al., 2008), stage of illness, whether patients were euthymic and also whether they were inpatients or outpatients (L. Berk et al., 2010; Lolich et al., 2012;
Rouget & Aubry, 2007). In addition to content and delivery, certain process variables are key to assessing how interventions were implemented such as fidelity (Craig et al., 2008) and what comprises standard clinical care (de Bruin, Viechtbauer, Hospers, Schaalma, & Kok, 2009;
Gaudiano et al., 2008).
3.1.3 Use of Systematic review, Intervention reporting and Quality assessment guidelines This review follows the guidelines set out for reporting systematic reviews and meta-analysis of studies that evaluate healthcare interventions, the PRISMA statement (Preferred Reporting Items for Systematic reviews and Meta-analyses) (Moher, Liberati, Tetzlaff, & Altman, 2009).
This ensures that it is transparent and complete in that it contains all of the details necessary for replication (see Appendix A for completed PRISMA checklist).
Published guidelines on reporting behavioural interventions were used to systematically describe content and delivery of the studies included in the review (Davidson et al., 2003) as well as guidance from the Consolidated Standards of Reporting Trials (CONSORT) (Boutron, Moher, Altman, Schulz, & Ravaud, 2008; Schulz, Altman, & Moher, 2010).
This review provides an assessment of the methodological quality and reporting of studies (Higgins et al., 2011). The problems of inconsistent and inadequate reporting have recently begun to be addressed. CONSORT guidelines state that interventions must be reported with
‘sufficient details to allow replication, including how, and when they were actually administered’ (Schulz et al., 2010). A recent study of intervention reporting in non-pharmacological trials concluded that 61% of interventions were reported inadequately (Hoffmann, Erueti, & Glasziou, 2013). To address this issue the Template for Intervention Description and Replication (TIDieR) checklist and guide has been published (Hoffmann et al., 2014). TIDieR is intended as a tool for study authors and systematic reviewers to describe interventions accurately and consistently, with enough detail to allow replication. In the
present review all interventions were assessed for the quality of intervention descriptions provided in published reports using TIDieR.