‘Compliance’ is futile but is ‘concordance’ between rheumatology patients and health professionals attainable?

The extent to which patients take prescribed medications ‘as directed’ and why they do not have been important issues in health research for many decades [1] and subject to recent review [2–6]. These issues are important in rheumatic diseases, given the primary reliance upon medication to control symptoms and improve patients’ functional ability and longevity [7]. Not taking required medications can therefore have both a personal health impact and health-economics consequences [3]. In recent years patients have shown increased interest in their health-care; this is partially due to a boom in access to information technology coupled with increased health coverage in traditional media sources [8]. A fresh approach to medication prescribing is required, one that evolves away from the paternalistic approach of patient compliance or adherence towards the more recently introduced philosophy of patient–professional concordance. Previous research has lacked consistent definition and measurement of medication-taking and, particularly in rheumatology, has rarely investigated biopsychosocial predictors in a longitudinal or interventional fashion. In this editorial we reiterate the definitions of patient ‘compliance’ and ‘adherence’ and comment on the implications of using these terms. We review some of the relevant research findings on patient compliance/adherence in rheumatology and finish by outlining the more recent philosophical and clinical move towards patient–professional concordance along with some of the practical implications for its impending application in routine clinical practice. We highlight potential predictors of medication taking that should be integrated in future research. We also examine how medication taking and its potential predictors should be measured in research by detailing key methods and identifying areas in which they can be applied. Definitions of key concepts and our recommended measures of these are summarized in Table 1. Definitions and recommended measures of key concepts around medication-taking in rheumatology aThe ASES measures perceived control over pain, fatigue, function and mood; extension of this scale to cover patients and health-care professionals’ perceived ability to achieve concordance/adherence would be theoretically and operationally useful. Definitions and recommended measures of key concepts around medication-taking in rheumatology aThe ASES measures perceived control over pain, fatigue, function and mood; extension of this scale to cover patients and health-care professionals’ perceived ability to achieve concordance/adherence would be theoretically and operationally useful. The conceptualization and operationalization of the ‘correct’ taking of prescription medications often reflects the dominant paternalistic approach to health-care. As Charavel et al. [9] put it, ‘the physician–patient relationship has always implied that the physician alone makes the treatment decision, while the passive and dependent patient is obliged to comply with the physician's prescription.’ (p. 93, emphasis added.) The verb ‘comply’ implies enforcement through its association with ‘bending the will of another’ ([10], etymology section). This has led to increased use of the verb ‘adhere’, which literally means ‘to stick to’ ([11], etymology section), to describe medication-taking. ‘Adherence’ is described as ‘persistence in a practice or tenet; steady observance or maintenance.’ ([12], section 3). This definition captures the routine that people with chronic rheumatic disease ideally engage in when taking prescription medication. The classic estimate of adherence is that only around 50% of patients adhere to prescription medications [1]. Ten days after a new prescription has been filled, 30% of patients with chronic diseases have missed at least one dose of the medication not on medical advice, half of them unintentionally, the others intentionally [13]. Reasons for missing a dose include side-effects, concerns about the medication and an expressed need for further information about the disease or medication itself, particularly among intentionally non-adherent patients. Medication compliance/adherence may be operationalized (i.e. conceptualized and measured) in many different ways [5], including multi-item questionnaire scales, individual ordinal/categorical questionnaire items, patients’ and physicians’ categorizations and independent observations such as prescription refill monitoring, electronic dispenser monitoring or metabolite/tracer measuring. We will now describe how estimates of adherence may differ depending on the ways in which adherence is assessed, as well as the exact disease and medication investigated. We also provide details of example studies that raise some of the central issues requiring attention in this field. In general rheumatology clinics, Berry et al. [14] found global non-adherence (answering ‘no’ to the question ‘Have you taken medicine regularly as prescribed or directed?’) more common among new (28%) than follow-up patients (1%). No new patients were on disease-modifying anti-rheumatic drugs (DMARDs), but they were more likely than follow-up patients to be on NSAIDs. As NSAIDs are suitable for use according to symptoms, up to a maximum directed dose, this may explain the global difference in adherence. The ‘as needed’ use of NSAIDs and analgesia needs to be clearly defined in this type of research so that it is considered in terms of the directions given to patients, some of whom may define themselves as non-adherent if they believe they have to take an NSAID regularly despite being given ‘as needed’ directions. Among lupus patients, Petri et al. [15] used a physician's opinion to categorize patients as adherent or not, with some evidence in hospital records only required to confirm non-adherence. Patients of Afro-Caribbean origin were more commonly categorized as non-adherent (56%) than white patients (34%), and this was thought to explain poorer renal outcomes in the Afro-Caribbean group. In contrast, in Mosley-Williams et al.'s study [16], in which patients themselves stated whether they failed to take prescribed medication all the time or did not fill any prescriptions, no difference between Afro-Caribbean and white lupus patients was observed. Only 10% of both racial groups were globally non-adherent. This gross difference in racial variability in adherence between these studies is likely to be an artefact of the completely different operationalizations and requires comparison in future samples. Cultural differences in adherence are possible if socio-economic differences in material wealth and location impact on access to health-care [15], or if cultural beliefs and perceptions about medication taking vary across groups. The extent of adherence among lupus patients also appears to depend on the medications prescribed. McElhone et al. [17] reported that the prevalence of perfect adherence (always taking the medication as prescribed) among white women with lupus ranged from 100% for azathioprine and 94% for oral steroids to 68% for NSAIDs and 61% for bone-protective medications. Lower adherence may relate to stringent recommendations for the way the latter should be taken [18] and the way the relative importance of medications is communicated by health professionals and perceived by patients. Adherence estimates in rheumatoid arthritis (RA) patients are similarly variable. Neame and Hammond [19] found 92% of RA patients disagreed with the single statement ‘I often do not take my medicines as directed’. Adherence rates among RA patients also depend on the type of medication. Across a 3-yr period, Viller et al. [20] found the proportion of RA patients who answered they stick ‘very strictly’ to both dose and timing of their medications was 55% for NSAIDs, 59% for DMARDs and 65% for oral steroids. Using electronic dispenser monitoring over one month, Park et al. [21] found only 38% of RA patients showed perfect adherence across all medications, with no dose omissions or commissions in any one day. This may also vary between medication classes: using the same definition, called ‘correct dosing’, de Klerk et al. [22] found adherence of around 70% for NSAIDs, 55% for sulphasalazine and 81% for methotrexate over 6 months. A more stringent assessment of adherence using ‘timing compliance’ (the number of inter-dose intervals of the prescribed duration ±25%) found that adherence rates were around 40–50% for NSAIDs, 25% for sulphasalazine and 83% for methotrexate. Electronic dispenser monitoring is costly and methodologically complex [22], and when participants know they are being observed they may increase their actual adherence or fake it. However, the less demanding, weekly regimen for methotrexate may facilitate enhanced adherence. In addition, these figures suggest that patients may not be as dependent on NSAIDs as is thought, and their sparing use may actually be advantageous in the light of recent concerns about the cardiovascular safety of continuous high-dose use of these drugs [23]. The presented prevalence figures for adherence to anti-rheumatic medications are not adjusted for disease activity, which is likely to be high among patients requiring DMARDs. In our cross-sectional investigation of this issue [24], we found weak, non-significant correlations between self-reported adherence and inflammation (C-reactive protein levels and erythrocyte sedimentation rate) in RA patients. This does not discount the possibility of more subtle processes that may lead to fluctuating associations of disease activity and adherence over time within the same individual. Consistent adherence would be expected for the highly desired anti-TNF therapy, but may depend on the route of administration [25]. Around 25% of RA patients on home-administered subcutaneous anti-TNF reported not taking it regularly in an Italian study [26]. The reasons for this non-adherence are unclear, but more than 60% of those who missed a dose did not contact a health professional beforehand. Lack of an arranged communication procedure may amalgamate with the presence of a side-effect or other concern to lead to non-adherence to anti-TNF treatment. There are currently a number of self-report, multi-item questionnaire-based measures of medication adherence employed in research on this topic. Morisky et al. [27] devised a scale consisting of four questions on reasons for non-adherence, which are answered with a yes/no response. Internal consistency of the four varied questions tends to be lower than acceptable, hence researchers resort to using individual items or specific combinations of items when using this measure [28]. A more recent measure is Horne et al.'s Medication Adherence Report Scale (MARS [29]; available at: http://www.medicines-partnership.org/projects/evaluation-toolkit), which consists of five non-adherence behaviours that are mainly intentional and are rated for frequency on a five-point scale. The MARS has good internal consistency [29]. However, neither of these self-report measures performed well compared with electronic dispenser monitoring (over 6 weeks) in transplant recipients taking oral steroids and immunosuppressants (medications which many rheumatology patients also take) [29]. The Compliance Questionnaire – Rheumatology (CQR [30, 31]) is currently the only rheumatology-specific compliance/adherence measure. The 19-item CQR was derived from patients’ statements by de Klerk et al. and is answered on a 4-point Likert scale. Although less parsimonious than the MARS and Morisky scale, the CQR compares well with electronic dispenser monitoring over 6 months [31]. Responses to the CQR items are summed to provide a total score that can be treated statistically as interval level data, allowing more powerful statistical comparisons of predictors or effects of medication adherence than simple dichotomy responses or unbalanced categorizations. Beliefs about medications (Table 1) have social and psychological roots, and are consistently important for adherence. Among RA patients, those who believe their medications are necessary display greater adherence cross-sectionally compared with patients who do not hold such beliefs [19, 24, 32]. Research in this area employs the Beliefs about Medicines Questionnaire (BMQ), which measures five independent beliefs, two specific (personal) and three general, across 22 items answered on a five-point Likert scale [33, 34]. The two specific beliefs are how necessary people believe their medications are and the concerns they have about their use (e.g. disruption, dependency, long-term effects) [33]. There are three general beliefs: whether medications are inherently dangerous; whether they are prescribed too commonly [33]; and the more recently added perceptions of the general benefits of medications [34]. Interviews with RA patients demonstrate that their beliefs about DMARDs do reflect these five factors [35]. Future research in this area would benefit from interviewing RA patients in much greater depth about their beliefs and perceptions of their medications, to further theoretical and conceptual issues and assist in developing rheumatology-specific scales of medication beliefs. Other valid questionnaire measures of medication beliefs exist, including the Beliefs About Medication Scale [36] and the Drug Therapy Concerns Questionnaire [37], but these are yet to be employed and evaluated among rheumatology patients. Improving RA patients’ self-efficacy (Table 1) has been linked with better medication adherence. Strong interventional evidence was produced from Hill et al.'s [38] unique randomized controlled trial. Self-efficacy education had a positive effect on adherence, which was measured using a tracer in standardized medication. This is an impartial way to assess adherence but is also costly and cannot fully determine timing [22]. Elliott et al.'s recent systematic review [3] concluded that further high-quality studies are required to substantiate the claim that medication adherence interventions are cost-effective. Such interventions may be even more effective if reframed around the concordance philosophy [39] that we describe below. The term ‘adherence’, although slightly less paternalistic than ‘compliance’, also implies an imbalance of power, wherein the health professional recommends a particular course of action to be followed, regardless of the patient's opinion. Generally, the health professional has the greatest disease knowledge and therefore advises the patient on the best known treatment appropriate for the level of illness. This decision is balanced against risks and grounded in the latest evidence-based, locally tailored guidelines and experience (e.g. [40]). However, ultimately it is the patient who has their health compromised and who has to live with the disease and medication-taking. The use of the term ‘concordance’ is growing in the context of medication-taking and literally means ‘together–heart’ and refers to ‘agreement’ or ‘harmony’ ([41], section 1) or establishing a treaty ([41], section 2). This eloquently captures the interactional decisional processes needed to promote partnership between patients and health professionals that the NHS is currently working towards [42, 43] by creating a therapeutic alliance to ‘optimise health gain from the best use of medicines, compatible with what the patient desires and is capable of achieving.’ (p. 12, original emphasis [42]). Although discordance can exist in a consultation, this must be resolved through compromise or agreeing to disagree [44], the patient having the final say on the ‘concordat’ (which philosophically replaces the prescription). The term ‘intentional non-compliance’ is thus futile and non-adherence to the concordat is only possible through failures in communication, accurate recording or memory, from either party. A patient who chooses to stop a medication if there is a potentially serious side-effect (e.g. infection) without seeking medical advice can be seen as both non-adherent and non-concordant. Side-effect concealment has been found to occur among RA patients on DMARDs and anti-TNF therapy [26, 35, 45]. Concealment is likely to reflect a perception that the necessity of the medication outweighs the side-effect [35, 45], and may indicate a lack of effective communication with patients and other health professionals, especially general practitioners (GPs). Careful guidelines and information for patients could facilitate more optimal health outcomes in scenarios such as these. At present, concordance may not be a familiar concept (or descriptor) for either patient or health-care audiences, so it needs to be explained in order to facilitate the process itself and distinguish it from other meanings (e.g. concordance of diseases in twins or between raters of a phenomenon). Concordance is particularly relevant when two or more possible interventions exist and have approximately equal chances of benefit [46, 47], and when the potential benefits and risks of these interventions cannot be accurately predicted for the individual (as is the case in most of clinical practice). Concordance is therefore very applicable in rheumatology, particularly in the anti-TNF era [48]. Despite this, research touching on concordance is rare in rheumatology (e.g. [49–51]) even though the topic has been subject to media attention (e.g. [52]). A patient's decision about their level of involvement is important to the concordance process [53]. Some patients want to defer decisions to their health professionals or is therefore to about patients’ desired level of involvement and whether patients who defer their treatment decision actually take the medication they are recommended as accurately as patients who have to the Research by and that medical and a in their treatment A of medical patients a while a of patients a passive Such research extension in a patients would be an appropriate to this among as they often from by medication-taking to requiring Research with RA patients by Neame et al. that patients should to decisions about medical but should medical 50% patients should along with this decision even if they disagreed with it (i.e. a in research has the importance of involvement in for RA patients the study by et al. found that 94% of RA patients they were in the only of these patients the or frequency of their medications. this would be conceptualized as non-adherence but may reflect it would not be seen as as patient–professional communication is have been to measure the of without any patient et al.'s Concordance scale available at: items answered on a Likert scale. The is an consistent scale Research this scale among patients their concordance is to a greater extent among women than and than patients, but its does not relate to the number of over the et al.'s Scale items answered on a five-point Likert scale. The has four consistent to and should These scales also rheumatology-specific but will be for comparison with Such scales need to be both in studies and of concordance interventions in important issue is to highlight the of participants patients and health-care in studies of concordance. may be to researchers are a to clinical to facilitate of that may or may not be the traditional is with towards and prescribing of these will be an given the different location and duration of they may The paternalistic approach to health-care as by the term ‘compliance’, is not high rates of and this may have implications for the individual patient and for health-care In the era of and access to information and a and socio-economic level across the the concept of concordance to be a However, in order to facilitate the very process of we need to how to and to both patients and health This is in the and has been shown to be effective in an for medicine may also be for professional in rheumatology, but needs to be in this specific context and more the concept of concordance in with to in the NHS and other health we will not know the of its and benefits for rheumatology research is and evidence on the process and its concordance will to be only for patients with diseases or particular of treatment or for patients and not the concordance approach may be more when with or medications or among patients with good knowledge about the and treatment or concordance is likely to be a and Concerns about professional may also be an issue when the concordance health professionals may about their for outcomes a concordat when their are by the the other do health professionals a paternalistic which to a patient's the is the same in both with the latter it much to the of the individual In the and implications of to the concordance approach need to be considered and accurately in future The research is on concordance in rheumatology, the better for all We would to three for their on this The have no of of of of of NHS and of