RESEARCH DOSSIER
Medication Adherence, Treatment Burden, and Care Continuity
Abstract. Medication adherence, treatment burden, and care continuity form an interdependent field of health psychology and health-services research. Adherence describes the initiation, implementation, and persistence of an agreed medication regimen; treatment burden describes the work of navigating treatment and the consequences of that work for everyday life; and care continuity describes the experienced coherence of care across clinicians, organizations, and time. This fundamental research monograph reconstructs the development of these concepts and critically synthesizes taxonomic, psychometric, observational, qualitative, experimental, and comparative-effectiveness evidence. It examines behavioral and cognitive mechanisms, access and affordability, therapeutic relationships, multimorbidity, prescription dispensing, measurement error, deprescribing decisions, and transitions between settings. Foundational evidence includes the ABC adherence taxonomy, pharmacy-record methodology, treatment-burden measures, the cumulative complexity model, Cochrane adherence-intervention evidence, the MI FREEE experiment, CATIE, X:BOT, care-transition trials, and studies of relational continuity. The synthesis establishes that appropriate medication-taking is contingent on an acceptable regimen, sufficient resources, accurate information, and reliable service organization. It distinguishes intentional deviation, practical inability, treatment discontinuation after informed deliberation, and clinician-directed change. The evidence supports multicomponent, context-specific approaches while exposing limited causal knowledge concerning long-term burden reduction, real-world persistence, and continuity mechanisms. Clinical recommendations are interpreted as research evidence, not individualized medical instructions.
Keywords: medication adherence; medication initiation; medication implementation; persistence; discontinuation; treatment burden; treatment workload; patient capacity; cumulative complexity; minimally disruptive medicine; continuity of care; relational continuity; informational continuity; management continuity; psychopharmacology; chronic disease; multimorbidity; patient-centered outcomes; shared decision-making; health literacy; health services research; pharmacy dispensing; proportion of days covered; patient-reported measures; care transitions; medication reconciliation; health equity; behavioral medicine; longitudinal research
Introduction and Scientific Problem
A medication order is a clinical event; sustained treatment is a sequence of actions distributed across days, people, places, organizations, and changing circumstances. Between a clinician's recommendation and a pharmacological effect lie prescription transmission, insurance or payment arrangements, dispensing, comprehension, remembering, administering, recognizing adverse effects, attending reviews, obtaining refills, and responding to changes in illness. Each step is an observable possible point of interruption. Understanding these interruptions requires a theory of behavior and an analysis of the institutional conditions in which behavior occurs.
Medication adherence has often been investigated as the degree to which an individual follows a prescribed regimen. This formulation is useful for measuring exposure, yet the clinical prescription itself may change, the patient may have sound reasons to question a treatment, and the prescribed work may exceed the resources available to complete it. The contribution of treatment-burden research is to specify the required work and its effects on life. The contribution of continuity research is to explain how clinical information, responsibility, and the therapeutic relationship persist across episodes of care. Together they provide a better-defined object than an undifferentiated label of nonadherence.
The primary research questions are how medication use unfolds over time; which psychological, socioeconomic, treatment-specific, and service-system processes shape it; how treatment workload interacts with practical and psychological capacity; how different kinds of care continuity affect execution and safety; and which interventions have produced credible improvements in medication use or patient-important outcomes. The published classification of adherence by Vrijens et al. (2012), the patient-complexity model of Shippee et al. (2012), and the continuity synthesis of Haggerty et al. (2003) establish complementary starting points.
This Research Dossier belongs to Health Psychology and Behavioral Medicine and Psychopharmacology. It addresses the process of treatment participation across medical specialties, including mental-health care. The pharmacodynamic action of individual drugs, comparative efficacy within every disease category, and diagnostic definitions of each condition belong to their respective scientific investigations. The subject here is the human and organizational trajectory through which an indicated and acceptable intervention is initiated, sustained, modified, interrupted, or ended.
Research Method and Evidence Appraisal
This work is a critical, structured narrative synthesis of identifiable studies and published scientific frameworks. It integrates original randomized trials, psychometric validation research, observational cohorts, qualitative studies, methodological articles, official clinical guidance, and systematic reviews. Literature was identified by targeted searches for concept definitions, instrument validation, trials, treatment discontinuation, continuity, burden, and cross-cutting mechanisms, supplemented by direct reading of available original and review records. A source's epistemic contribution is limited to the design and information actually accessible. The work has no registered protocol, exhaustive database search, dual independent selection process, risk-of-bias table for all eligible studies, or pooled effect estimate and therefore makes no claim to be a systematic review or meta-analysis.
The evidence is organized by three levels. Descriptive and construct research identifies the concepts, behavioral stages, reported experiences, and measurement procedures. Analytic observational research estimates associations between psychological or service variables and medication use or outcomes. Randomized interventions establish contrasts between specified assignments under defined settings, with further questions concerning fidelity, mechanisms, sustainability, and transportability. Published reviews are used to locate converging or conflicting patterns, rather than to convert heterogeneity into an unsupported consensus. The Cochrane synthesis by Nieuwlaat et al. (2014) and the clinical review by Kini & Ho (2018) are treated as syntheses of their search periods.
Four principles guide interpretation. First, possession of dispensed medication is an imperfect proxy for ingestion, and ingestion is an imperfect proxy for pharmacological exposure. Second, an association between adherence and favorable health may reflect disease severity, health behavior, access, or the quality of accompanying care. Third, burdens are multidimensional and vary by treatment, person, and time, so a total burden score should not automatically be read as a universal quantity. Fourth, the clinical appropriateness of a course of treatment is distinct from its execution: a carefully negotiated decision to stop a medication may be clinically appropriate despite reducing a mechanical persistence indicator.
The scientific record is evaluated for construct alignment, selection and information biases, confounding, follow-up duration, patient-centered outcomes, adverse-event ascertainment, and applicability to people with multimorbidity or limited access. Source links in the text identify the scholarly work supporting each substantial factual claim. This monograph does not reconstruct participant-level analyses or reproduce unverified numerical results.
Conceptual Definitions and Boundaries
Medication adherence is the process by which a person takes medication as agreed with a healthcare provider, commonly analyzed as initiation, implementation, and discontinuation, with persistence spanning the interval from initiation to discontinuation. Agreement is integral to the ethical and clinical interpretation of the process. Implementation concerns correspondence between actual dosing behavior and the specified regimen while treatment continues. Persistence concerns duration until discontinuation. A person may obtain medication yet never initiate it; may initiate yet miss or alter doses; or may implement an agreed regimen accurately until a later planned or unplanned stop. These are distinct research outcomes (Vrijens et al. (2012)).
Treatment burden is the workload imposed by healthcare and the impact of that workload on the patient’s functioning, resources, and life. Tasks include obtaining information, arranging appointments, performing monitoring, managing multiple medications, navigating institutions, paying direct and indirect costs, and organizing relationships around care. Burden may be experienced by family caregivers as well as by the person receiving treatment. It is theoretically possible for a regimen to be clinically efficacious and yet unsustainable because its total workload exceeds available capacity (May et al. (2009); Tran et al. (2015)).
Care continuity is the experienced connectedness and coherence of healthcare events with the patient's needs and context. Informational continuity concerns the use of past information to inform present care; management continuity concerns the consistency and responsiveness of a care plan across time and providers; relational continuity concerns an ongoing therapeutic relationship. These dimensions can vary independently. A patient may see the same clinician repeatedly while medication lists remain inconsistent, or receive highly coordinated information transfer without longitudinal contact with one clinician (Haggerty et al. (2003)).
Noninitiation, nonimplementation, treatment interruption, informed refusal, administrative supply failure, and authorized deprescribing require separate labels. The ethical distinction between intentional and unintentional divergence does not automatically describe its cause: a person may deliberately reduce doses because costs or side effects make the regimen unmanageable, and an apparently inadvertent omission may be produced by fragmented services rather than defective memory. The terms "compliance," "adherence," and "concordance" possess different historical associations, but terminology alone cannot substitute for a precise definition of the observed behavior and the agreed treatment target.
Historical Formation of Adherence Research
Early research commonly adopted the vocabulary of patient compliance, treating correspondence to medical recommendations as its central dependent variable. As pharmacy data systems expanded, adherence became measurable through prescriptions filled and gaps in possession. The methodological review by Steiner & Prochazka (1997) documented the potential and limitations of refill records, presenting evidence that acquisition patterns could be studied at population scale without direct observation of consumption. Their contribution was methodological as much as clinical: it distinguished a usable administrative signal from the underlying behavioral process.
By the beginning of the twenty-first century, meta-analytic studies addressed relationships between adherence and outcomes. Dimatteo et al. (2002) examined associations across a range of medical regimens, emphasizing methodological differences in assessing adherence and outcomes. Research on cost-related underuse and communication brought structural constraints and unspoken patient concerns into focus. In a survey-based study, Piette et al. (2004) investigated which treatments chronically ill people forgave and who faced greater financial risk.
The institutional turn in adherence research followed changes in the understanding of chronic disease. Multiple treatment recommendations accumulated for the same patient, health systems divided tasks among organizations, and management increasingly depended on self-care between consultations. May et al. (2009) argued for minimally disruptive medicine: reducing the unnecessary work imposed by healthcare while protecting valuable outcomes. Shippee et al. (2012) formalized the reciprocal relation between workload and capacity as a functional model of patient complexity.
A parallel conceptual shift differentiated phases of medication use. The ABC project's Vrijens et al. (2012) provided a standardized taxonomy for initiation, implementation, and discontinuation. Raebel et al. (2013) clarified measurement terminology for electronic prescribing and dispensing databases. In care continuity, Haggerty et al. (2003) consolidated terminology across disciplines; subsequent trials of discharge and transition interventions investigated whether failures of communication and coordination could be modified experimentally.
This history represents a progressive expansion of the explanatory unit from a single medication-taking act to a patient–regimen–service system evolving through time. It also demands methodological caution. New labels and models explain phenomena only when their constructs can be measured distinctly and the proposed pathways can be tested.
The ABC Taxonomy and Treatment Trajectories
The ABC taxonomy defines initiation as taking the first prescribed dose, implementation as the subsequent alignment of actual use with the prescribed regimen, and discontinuation as taking the last dose before stopping treatment. Persistence is the duration between initiation and discontinuation. A trajectory can therefore contain delays, variable daily implementation, temporary interruptions, schedule changes, and eventual stopping. Clinical decisions may change the target regimen during observation, making the target itself time-varying (Vrijens et al. (2012)).
The distinction matters because interventions operate through different mechanisms. A barrier at initiation may concern fear, uncertainty about necessity, inability to afford a first prescription, or failure of prescription transmission. A barrier during implementation may concern complex timing, interruptions of routine, adverse effects, travel, or the cognitive burden of multiple tasks. Shortened persistence may arise from changing benefit–harm judgments, clinical discontinuation, recurring supply problems, or deterioration of therapeutic trust. An intervention successful at one phase should not be assumed to resolve the others.
In observational research, phase mixing produces interpretive errors. A cohort constructed only from people who collected a prescription excludes people who never filled it. A persistence analysis that ignores clinical switching may classify appropriate medication changes as failure. A refill metric cannot necessarily distinguish prescribing discontinuation from patient disengagement. Raebel et al. (2013) argue for definitions matched to the data-generating process and the question being studied.
The comparative review by Kardas et al. (2013) documents determinants across systematic reviews and reinforces the need to distinguish potentially modifiable treatment, patient, and system factors. The longitudinal unit of analysis also matters. Individual average adherence can conceal extended consecutive gaps. Two people with similar annual proportions of days covered may have different temporal patterns: one misses sporadic doses; another experiences a prolonged interruption followed by sustained use. Their expected clinical consequences can differ according to the drug's properties, indication, and timing of outcome measurement. A serious analysis treats adherence as a trajectory and distinguishes descriptive exposure from any assumption about therapeutic benefit.
Treatment Burden as a Research Object
Treatment burden is composed of the work that treatment requires and the interference that work produces in life. Its domains extend from the pharmacological regimen to appointments, monitoring, service navigation, paperwork, transportation, financial expenditure, learning about disease, and the coordination of professional advice. Tran et al. (2015) used a multi-country qualitative study to construct a taxonomy of patients' experiences, documenting the range of tasks and consequences that conventional adherence indicators leave unmeasured.
Workload has temporal characteristics. A new diagnosis may create intense learning and scheduling demands. Stable treatment may become routinized, reducing cognitive effort even when doses remain numerous. New symptoms, caregiver changes, insurance transitions, or hospitalization can reorganize the entire workload. A person may report relatively low medication-taking difficulty while experiencing severe exhaustion from travel and administrative demands. Treating all these conditions as a single adherence attitude discards actionable distinctions.
Burden is relational: the same treatment task has different meaning and feasibility depending on personal capacity and surrounding infrastructure. The empirical work of Vijan et al. (2005) on diabetes therapy illustrates why regimen design must include patient experience. Medication burden can be reduced by a feasible change to treatment organization, yet such modifications must also preserve appropriate clinical oversight and the intended therapeutic benefit.
Burden research further changes the normative question. An intervention that increases the proportion of measured doses but expands administrative work, anxiety, or dependence on constant surveillance can create a worse overall experience. The appropriate evaluation includes treatment effectiveness, burden, safety, control over decisions, and participation in ordinary life. The central objective of minimizing unnecessary burden is analytical and clinical rather than a demand to minimize treatment intensity regardless of expected health benefits.
Cumulative Complexity and Patient Capacity
The cumulative complexity model describes patient complexity through the relation between workload and capacity. Workload arises from healthcare and life demands; capacity includes physical, cognitive, emotional, social, financial, and practical resources for meeting them. When demands accumulate or capacity diminishes, the person may be unable to perform all requested tasks even when motivated to do so (Shippee et al. (2012)).
Capacity is dynamic. Fatigue, worsening symptoms, bereavement, changes in work, migration, financial shocks, and transitions between services may reduce the resources available to manage an unchanged prescription. Conversely, accessible transportation, consistent pharmacy supply, clearer instructions, social assistance, or simplified administration can increase effective capacity without changing personal motivation. The model is valuable because it identifies multiple potential intervention points rather than attributing every lapse to an individual disposition.
A recursive process is plausible. Illness increases workload; excessive workload disrupts treatment or self-care; disruption may contribute to poor outcomes and further workload. Yet the existence and magnitude of each directional pathway must be tested in longitudinal data. A conceptual feedback diagram is not itself causal evidence. Repeated measurements of task demands, capacity, adherence trajectories, treatment changes, and health outcomes would permit stronger assessment than one-time cross-sectional correlations.
The model also informs study design. Capacity should be measured with instruments that distinguish physical function, literacy, material resources, social help, and competing role demands. Workload should include patient tasks outside the clinic. Analyses should consider treatment indication, illness severity, clinical benefit, and whether patients had realistic alternatives. Without these distinctions, the observation that overwhelmed patients take fewer medicines tells little about the best response.
Care Continuity and Its Three Dimensions
Continuity concerns coherence experienced over time, not simply repeated visits. Informational continuity uses knowledge of prior diagnoses, treatment effects, preferences, and decisions. Management continuity means that a plan responds coherently as needs change. Relational continuity rests on a sustained connection in which both patient and clinician can develop knowledge and trust. Haggerty et al. (2003) articulated these dimensions in a multidisciplinary review.
Medication management depends on all three. Information may be accurate but no professional may assume responsibility for reconciling conflicting recommendations. A consistent management plan may be implemented by different clinicians, although loss of relational trust can still impair communication about side effects or financial distress. A long-standing clinician–patient relationship may be reassuring but insufficient when transfer documents omit recent medication changes. The dimensions therefore require separate indicators.
Continuity can be interrupted by discharge, changes in insurance, professional turnover, referral delays, fragmented health records, migration, and the division of care between mental-health and general medical services. These disruptions are experienced at the level of the individual care trajectory, whereas service evaluations often count encounters by organization. A scientifically adequate unit of analysis links the patient's sequence of events with the sequence of prescribing and dispensing decisions.
This framework also clarifies responsibility. Continuity should not mean that the patient alone must discover every inconsistency across multiple prescribers. A service can organize reliable informational handoffs, clear ownership of follow-up, and timely clarification of medication changes. The clinical quality of these mechanisms must be judged with reference to preventable harm, comprehension, access, and patient preferences.
An Integrated Temporal Account
The three constructs form a dynamic analytical system. At time of treatment initiation, beliefs about necessity and risk interact with affordability, the clinician's explanation, and the first available dispensing opportunity. During implementation, a patient repeatedly coordinates medication with daily work, symptoms, meals, sleep, changing environments, and the availability of supplies. Over longer periods, monitoring visits, side-effect evaluation, hospitalization, and changing goals alter the regimen and the resources available to complete it.
Treatment burden acts on the feasibility and personal cost of implementation; continuity influences the coherence, accessibility, and credibility of treatment instructions and follow-up. Medication adherence can in turn affect health and workload, although the direction and strength of effects vary with treatment and condition. A useful analytical representation therefore includes bidirectional pathways rather than a single arrow from patient motivation to clinical outcome.
For causal research, the system must be decomposed. The key longitudinal quantities are changes in regimen requirements, measured burden, capacity, continuity, adherence phase, and patient-important outcomes. Confounding may arise because severe illness increases both the need for encounters and the risk of medication interruption. Selection may arise when only people remaining in treatment are measured. A transition intervention may influence safety chiefly through information reconciliation rather than the volume of medications consumed.
The integrated account provides testable predictions: when burden is reduced without compromising effective care, persistence may improve among people whose previous interruptions were workload-related; when informational continuity is strengthened, discrepancies between discharge instructions and outpatient medication lists should decline; and when a patient receives an informed choice between treatments with different administration demands, decisions may better reflect preferences. These are hypotheses to test with suitable designs, not established universal effects.
Motivation, Beliefs, and Perceived Necessity
Medication-taking requires repeated action sustained by expectations about illness, treatment benefit, and tolerability. The person must decide whether the recommendation makes sense, whether anticipated benefit merits effort and risk, and how the regimen fits personal goals. These judgments may be explicitly verbalized or embedded in routines. Qualitative work synthesized by Kvarnström et al. (2021) indicates the importance of patient perspectives, communication, health-system context, and everyday feasibility.
The distinction between motivation and capacity is crucial. A highly motivated person may lack funds or access. Another may have adequate medication supply but experience uncertainty about its long-term necessity. A third may view adverse effects as more immediately consequential than a future reduction in risk. Because these cases have different causes, a generic motivational message is unlikely to be adequate for all.
Beliefs also change over time. Early concerns may disappear as symptoms improve, or treatment may become less attractive when adverse effects accumulate or perceived benefit remains uncertain. Trust can be strengthened by transparent discussion of uncertainty and revised through experience. Research that measures baseline beliefs and an endpoint adherence score without intermediate observations risks overlooking these changes.
Intention measures should not be interpreted as direct evidence of ingestion. Interviews can explain a person's narrative; validated self-report instruments can measure stated behavior or beliefs; dispensing data can measure acquisition. A coherent explanatory model uses several sources and treats disagreement between them as information about the complexity of medication use.
Memory, Prospective Action, and Routines
Health literacy is another measurable component of treatment capacity: the systematic review by Berkman et al. (2011) relates limited health literacy to important health outcomes, although it does not identify one universal pathway to missed medication doses. Medication implementation often depends on prospective memory: the ability to perform an intended action at the appropriate future time. Routine environments can provide cues, reducing reliance on conscious rehearsal. However, cue stability is affected by variable work schedules, travel, care transitions, hospitalization, and changing daily roles. An implementation failure may reflect the absence of a reliable cue rather than low endorsement of treatment.
Regimen characteristics modify prospective memory demands. Multiple dose times, distinct administration instructions, refills on different dates, monitoring requirements, or interactions with food create coordination tasks. Routines can convert repeated decisions into stable action sequences, but a previously adaptive routine may fail after a dose change or a medication switch. It follows that prospective memory should be considered alongside comprehension, habit structure, and available assistance.
Electronic reminders and packaging aids target aspects of the implementation problem; they cannot independently resolve an unaffordable prescription, an intolerable adverse effect, or an unresolved clinician disagreement. This explains why effects observed on opening a device, responding to an alert, or recorded refill behavior should be interpreted through the specific stage measured. The review by Kini & Ho (2018) emphasizes the diversity and limited generalizability of adherence interventions across settings.
Experimental studies should distinguish a reminder's effect on a recorded response from its effect on therapeutic outcomes. Measurement devices may alter behavior merely by being present, and a notification can impose new attentional work. Research should examine adherence sustainability after withdrawal of external prompts and the acceptability of monitoring to the people affected.
Emotional Distress, Cognition, and Illness Experience
Distress, depressive symptoms, anxiety, cognitive impairment, fatigue, and fluctuating illness symptoms can affect the acquisition and implementation of medication. The mechanism may involve concentration, planning, confidence, hopelessness, adverse-effect interpretation, or reduced practical capacity. Yet these constructs also relate to illness severity, social adversity, and service access. A cross-sectional association between distress and low adherence cannot by itself establish a psychological causal pathway.
The meta-analysis by Dimatteo et al. (2002) demonstrates the importance of comparing outcomes across different medical regimens and measurement strategies. A broad evidence base should not be treated as evidence that one affective mechanism has the same direction and magnitude in every disease category. Treatment burden itself can increase distress, producing possible reciprocal relations.
Medication effects may also influence cognition, sleep, or daily functioning. Clinical assessments must distinguish symptoms of the condition, adverse drug reactions, interactions, and the consequences of missed treatment. In psychopharmacology, the experience of sedation, activation, movement symptoms, sexual side effects, or weight changes may directly affect whether continuing treatment is acceptable. The content of these effects depends on the drug and patient context.
Clinical research benefits from repeated measures collected before and after changes in treatment or care conditions. Designs should account for the possibility that poor health and low medication use are joint consequences of unobserved social disadvantage. They should also preserve patient reports of reasons for changes rather than reducing all deviations to a numerical deficit.
Costs, Supply, and Structural Access
Financial and structural constraints are mechanisms of medication interruption. Out-of-pocket expenses, transportation, time away from work, pharmacy availability, language barriers, reimbursement rules, and the administrative complexity of obtaining refills can affect all adherence phases. Piette et al. (2004) examined cost-related medication underuse among chronically ill adults, establishing cost as a substantive empirical object rather than a peripheral patient preference.
A person who cannot obtain a prescription is not adequately characterized by an inventory of beliefs or motivation alone. The correct observational unit must include access to dispensing and the economic conditions of that access. Insurance coverage can change during follow-up; generic substitution, stock shortages, or delays in authorization may create interruptions that appear in pharmacy claims without revealing the responsible mechanism.
Structural determinants also interact with clinical communication. Patients may be reluctant to disclose financial trouble, may receive prescriptions without understanding payment conditions, or may have several clinicians who assume another service is responsible for assistance. These mechanisms turn otherwise routine tasks into cumulative treatment workload. An intervention offering information without improving actual access may produce little change.
The randomized insurance-coverage experiment known as MI FREEE offers direct evidence that lowering patient costs can change medication-use behavior in a defined population (Choudhry et al. (2011)). Its results should be interpreted for the drugs, benefits design, health system, and follow-up assessed. Broader claims about universal effects of medication coverage require additional comparative evidence.
Complex Regimens and Multimorbidity
Multimorbidity often generates multiple recommendations that are clinically reasonable when considered separately and difficult to sustain in combination. A person may be asked to coordinate drug regimens, lifestyle changes, specialist appointments, laboratory tests, symptom monitoring, and rehabilitation activities. Time and financial resources are finite, and recommendations can compete with family responsibilities or the need to remain employed.
Guthrie et al. (2012) examined how clinical guidelines should be adapted when several conditions coexist. The generalizable insight is that disease-specific optimization can produce aggregate workload and potential interactions that must be evaluated at patient level. Shippee et al. (2012) provide the complementary analytic model of workload and capacity.
Medication count by itself is a crude proxy for burden. A stable, well-tolerated regimen with reliable access may demand less attention than a single medication requiring complex monitoring or difficult administration. Likewise, reducing the number of drugs is not necessarily beneficial if it removes effective therapy. Regimen evaluation needs appropriate indication, expected benefit, patient priorities, administration demands, safety, and ongoing review.
Research on multimorbidity should include combinations of conditions rather than treating each diagnosis as an independent additive factor. Clinical teams may face inconsistent guidance, while patients manage the practical synthesis. Studies capable of evaluating redesign should measure the total workload and its distribution among patients, family, pharmacies, and clinicians.
Family, Caregivers, and Social Organization
Medication-taking is sometimes distributed across a network of people. Family members may arrange appointments, obtain refills, prepare doses, interpret instructions, or monitor changing symptoms. Social assistance can extend a patient's capacity, but it can also create new dependencies, privacy concerns, conflicts about responsibility, and caregiver burden.
The analytic question is who performs each task and under what conditions. A person may remain autonomous in deciding whether to continue treatment while relying on someone else for transport or packaging. Another may physically take medicine independently but have limited control over purchasing or interpreting care plans. A single self-report adherence score generally cannot capture this division of labor.
For studies of caregiver involvement, measurement should distinguish practical assistance, coercion, shared deliberation, and the patient's ability to contest decisions. Family participation may improve the availability of medication while reducing perceived control if roles are poorly negotiated. Appropriate study designs must consider the patient's preferences, consent, and capacity as well as measurable implementation.
Continuity extends to the caregiver network when responsibilities transfer across hospital, home, and community care. Unclear instructions or changing prescriptions can impose substantial reconciliation work on families. The mechanism remains a service-system issue even when its consequences are observed in the home.
Aging, Disability, and Life-Course Change
Population averages can conceal life-course differences in regimen demands. Older adults may face multimorbidity, sensory changes, cognitive impairment, or dependence on several services, while some retain highly stable routines and extensive expertise in managing treatment. The systematic review by Gellad et al. (2011) examined barriers to medication adherence among older adults beyond cost and regimen complexity.
Disability modifies access to pharmacy services, packaging, communication, transportation, and self-administration. These barriers depend on environmental design as well as individual functioning. A research framework that attributes all disability-related interruptions to personal incapacity misses the possibility of accessible systems and task redistribution.
Transitions such as moving into assisted living, leaving education, entering employment, developing a new chronic condition, or recovering from hospitalization can destabilize established medication routines. In a longitudinal analysis, age is a contextual variable rather than a sufficient mechanism. Detailed information about life events, impairment, support, and service context is necessary.
Measures should accommodate different modes of medication administration and different kinds of assistance without automatically treating assisted administration as poor adherence. Cultural and linguistic differences matter because the meaning of an agreed regimen and the division of responsibility can vary across institutions and households. Generalization requires samples that reflect these varied circumstances.
Medication Adherence Measurement
Adherence is a latent behavioral process measured through imperfect indicators. Common sources include self-report, pill count, pharmacy dispensing, electronic medication-event monitoring, clinician assessment, biomarkers, and direct observation in selected settings. Each observes something different. Self-report may capture reasons and patient interpretation but is vulnerable to recall and social desirability. Electronic opening records show device interaction rather than ingestion. Pharmacy data show medicines supplied rather than consumed.
Steiner & Prochazka (1997) established early methodological principles for refill measurement. Raebel et al. (2013) later clarified the relationship among prescribing, initial filling, subsequent dispensing, and persistence. Measures such as proportion of days covered depend on the defined observation window, days' supply, overlap rules, hospitalizations, dose changes, and allowable gaps. A familiar numerical threshold is an analytic convention, not a universal physiological boundary.
A refill-based proportion does not reveal whether the patient took doses at appropriate times. Electronic event monitoring can provide finer temporal resolution but creates device-related missingness and may alter behavior. Biomarker concentrations depend on pharmacokinetics and sampling time; therapeutic monitoring is not automatically an adherence gold standard. The validity of an indicator is conditional on the research question and the medication.
Measurement studies should prespecify initiation, implementation, and persistence separately and document how switching, deprescribing, discontinuation after adverse events, hospitalization, and death are handled. Analyses require care with denominator choice, censoring, immortal time, and selection of people who survive or remain eligible for observation. Triangulating reports, dispensing, and clinical records can strengthen interpretation, provided disagreements are investigated rather than averaged away.
Treatment-Burden Measurement
Treatment burden requires direct assessment of tasks and their life consequences. Tran et al. (2014) adapted and validated the Treatment Burden Questionnaire in English, and their subsequent Tran et al. (2015) broadened the conceptual taxonomy. These sources demonstrate the need for wording that patients recognize and for domain coverage extending beyond medication counts.
The Patient Experience with Treatment and Self-management measure, or PETS, was developed and validated by Eton et al. (2017). The reported validation sample included 332 completed survey responses from patients with multimorbidity, and factor analyses supported domains including medical information, medications, appointments, monitoring, service difficulties, financial expense, and interference with roles and well-being. These findings provide psychometric support for multidimensional assessment in the studied context, not proof that the same factor structure applies unchanged to every population.
Construct validity involves several questions. Does the instrument measure workload itself, perceived difficulty, or life impact? Do scales distinguish burden caused by illness from burden caused by treatment? Are patients asked about the relevant recall interval? Do scores respond to changes in actual service organization? Does the measure work comparably across disability status, language, age, and health system? A burden measure optimized for multimorbid adults may omit context-specific tasks relevant to children, caregivers, or complex psychiatric care.
Burden measurement also creates a methodological challenge: asking patients to complete repeated lengthy surveys adds research-related work to already demanding treatment. Sampling schedules and instrument length should be justified by their value for valid measurement. Studies of burden-reducing interventions must measure both reduction in work and preservation of clinically valuable outcomes.
Continuity Measurement and Causal Identification
Continuity has been operationalized through repeated contact with the same clinician, distribution of visits across professionals, information transfer, coordination indicators, and patient-reported coherence. These operationalizations are related but nonidentical. A visit-concentration score can be high despite poor informational transfer; reliable record exchange can occur without a sustained personal therapeutic relationship (Haggerty et al. (2003)).
Observational associations between continuity and outcomes require particular caution. Patients with greater severity may receive more specialist care and simultaneously face higher event risk. People who remain in one care system may differ in financial security, geographic stability, health behavior, or access from those who change clinicians. Reverse causality can occur if deteriorating health increases service fragmentation. Mortality studies also must consider observation windows and how continuity scores are conditioned on survival or healthcare use.
A causal study should define the continuity intervention precisely: assigning a named coordinator, improving discharge information transfer, strengthening relational follow-up, or ensuring timely medication reconciliation. These interventions are not interchangeable. The comparison condition, adherence to the intervention, and target outcome must be specified. A randomized care-transition intervention can test a defined service package but cannot alone identify which component caused the effect.
Patient-reported measures are important because continuity is experienced rather than merely administratively registered. However, satisfaction and continuity ratings can be affected by current health and prior treatment outcomes. Triangulation among patient experience, clinical records, and objective transition events helps separate service process from retrospective impression.
The Evidence on Adherence Interventions
Adherence interventions include education, counseling, reminders, regimen simplification, pharmacist support, financial assistance, feedback, motivational approaches, and coordinated multicomponent services. Their effects depend on what problem the intervention addresses, the medication and population studied, and the outcome used. An intervention developed to improve short-term tablet taking among stable patients cannot be assumed to improve long-term clinical outcomes for people experiencing unstable housing or severe multimorbidity.
The Cochrane review by Nieuwlaat et al. (2014) found that interventions for long-term medications were often complex and that evidence for simultaneously improving adherence and clinical outcomes remained limited. This conclusion reflects the trials and methods available at the review's cutoff, not a claim that every later intervention is ineffective. The Kini & Ho (2018) review likewise treats adherence improvement as a heterogeneous intervention field requiring attention to mechanism and context.
Several design problems recur. Studies may enroll only people already engaged with care, measure adherence using self-report alone, follow participants briefly, or compare a resource-intensive intervention with minimally supported usual care. Some report improved dispensing without adequate information on health, adverse effects, quality of life, or economic sustainability. Intervention fidelity and ongoing staffing costs matter for implementation outside trials.
A stronger research program would diagnose barriers before assigning interventions, specify which phase of adherence is targeted, use longitudinal behavioral measures, and include patient-important outcomes. Trials should evaluate the actual burden created by the intervention itself. When trials are too small for meaningful clinical outcome contrasts, findings should be identified as behavioral or feasibility evidence rather than proof of improved health.
Randomized Evidence on Financial Access
The MI FREEE trial provides a distinctive example of modifying a structural condition rather than primarily asking patients to change their beliefs. Choudhry et al. (2011) randomized insurance-plan sponsors to full coverage or usual coverage for selected cardiovascular medicines after myocardial infarction. The design tested whether removal of out-of-pocket charges could improve use and clinical outcomes in the participating insurance setting.
The published findings support the conclusion that financial design can influence medication use. The experiment is especially instructive because it addresses an observable constraint outside the patient's immediate control. Its endpoint distinctions also matter: changes in adherence, primary clinical outcomes, and secondary events should not be collapsed into one broad claim of success. Follow-up duration, the covered drug classes, the enrolled population, and the structure of United States insurance influence transportability.
Financial assistance may also interact with service continuity. If coverage changes frequently, repeated authorization and refill requirements impose treatment workload. An intervention that reduces direct prices without simplifying acquisition may leave access problems unresolved. Conversely, a stable and transparent supply pathway can reduce cognitive and administrative work in addition to monetary cost.
The study demonstrates the value of randomized policy-level evidence. Its broader lesson is methodological: adherence research can intervene on pharmacy, insurance, and organizational conditions and should therefore treat such conditions as explanatory variables rather than background descriptions.
Psychopharmacology and Treatment Discontinuation
Psychotropic medication introduces distinctive questions of perceived benefit, tolerability, stigma, symptom fluctuation, clinician relationship, and long-term monitoring. A treatment's pharmacology and the person's experience of that treatment influence persistence, but persistence is not a standalone measure of efficacy or ethical appropriateness. Clinical deterioration, adverse effects, voluntary preference changes, and planned switching can all affect discontinuation.
The CATIE trial compared antipsychotic treatments in chronic schizophrenia and used time to discontinuation as a pragmatic outcome reflecting effectiveness and tolerability under trial conditions (Lieberman et al. (2005)). Such an endpoint deliberately combines multiple reasons for discontinuation. It is valuable for comparative effectiveness but should not be interpreted as a pure measure of an individual's medication adherence or of molecular drug action.
The X:BOT randomized comparative-effectiveness trial contrasted extended-release naltrexone with buprenorphine–naloxone for opioid relapse prevention (Lee et al. (2018)). Its induction and retention challenges show why treatment initiation and the feasibility of entering a regimen are part of real-world comparative effectiveness. The trial does not authorize broad conclusions about one form of long-acting treatment across other conditions or about individualized medication choice.
Research on long-acting formulations should specify whether the measured outcome is appointment attendance, timely administration, persistence, symptom outcome, adverse event, or patient preference. A depot or long-acting schedule may reduce daily dosing tasks while adding clinic-visit demands and altering the experience of control. Observational advantages may reflect selection of patients, service organization, or clinical indication. Current regulatory labels and individualized benefit–risk assessment remain necessary for treatment guidance, which is outside the purpose of this research monograph.
Care Transitions and Medication Safety
Transitions between inpatient care, outpatient care, community treatment, and home are points where medication plans may change rapidly. Information can be delayed, duplicated, or contradictory; a person may leave with altered instructions that an existing pharmacy or clinician has not yet received. Kripalani et al. (2007) reviewed deficits in communication and information transfer at hospital discharge, identifying a service-level problem relevant to medication safety and follow-up.
The Care Transitions Intervention randomized trial by Coleman et al. (2006) investigated a support package intended to improve patient and caregiver ability to manage transitions. The reengineered discharge intervention studied by Jack et al. (2009) tested a structured discharge program. Both demonstrate that care transitions are suitable for experimental intervention rather than merely descriptive quality measurement. Their mechanisms, sites, and endpoints differ.
Medication reconciliation aims to establish an accurate and clinically reviewed list across settings. The research distinction is between finding a discrepancy and preventing a harmful outcome. Discrepancy counts may reflect detection intensity as well as true error incidence. A service that identifies more discrepancies may be performing better surveillance, so outcome interpretation must consider both the opportunities for detection and the seriousness of detected problems.
The patient and caregiver burden of transitions should also be measured. Discharge instructions may require obtaining multiple medicines, clarifying changes, arranging transport, and booking follow-up in a compressed period. Improving paperwork without ensuring reachable clinicians, available medicines, and feasible follow-up leaves the continuity process incomplete.
Continuity, Outcomes, and Confounding
Longitudinal relationships with clinicians may provide knowledge, trust, and coherent follow-up. Yet measuring the independent contribution of continuity is difficult. Pereira Gray et al. (2018) reviewed published studies associating continuity of doctor care with mortality. The review documents a substantial observational literature; its findings do not mean that continuity was randomly assigned or that every association demonstrates a causal mortality benefit.
Several causal explanations are plausible. Relational continuity may facilitate earlier detection of deterioration and more complete disclosure. Informational continuity may reduce conflicting prescriptions. Management continuity may improve monitoring and timely adjustment. However, better-resourced patients and systems may also produce both greater continuity and better outcomes. The empirical challenge is distinguishing these pathways.
Studies should report the type of continuity measured, the time window, population, encounter frequency, exposure definition, and adjustments for illness severity and service accessibility. The choice of denominator can be consequential when continuity is calculated only for patients with multiple visits. Comparisons across health systems require care because referral patterns and professional roles differ.
Experimental work can test narrower interventions: protected follow-up with a known professional, structured handoff, a named coordinator, or shared electronic medication records. Such designs should include safety, burden, patient experience, and adherence trajectories while controlling for the additional contact time or resources supplied. The question is which particular continuity mechanism creates benefit for which patients and under what conditions.
Digital Supports and Artificial Systems
Digital tools can record prescriptions, remind people about scheduled actions, organize refills, support remote contact, and transmit information among care professionals. They can reduce the demands of memory and record-keeping when they are accessible, accurate, and integrated into actual care. They may also increase workload through alerts, passwords, repeated data entry, device failures, and fragmented portals. The burden of technology is therefore an empirical question.
An artificial system that summarizes medication information or offers a conversational reminder is part of an information environment whose accuracy, provenance, and governance matter. The relevant research endpoints include correctness of drug and regimen information, detection of outdated instructions, evidence of referral to the responsible clinician, patient comprehension, privacy, accessibility, and actual medication-use outcomes. Textual fluency is not evidence of medical reliability, and generated instructions should not be treated as an independently authoritative prescription.
The Artificial Era extends the field of psychological inquiry to relationships in which people rely on nonbiological systems for memory, planning, information retrieval, and continuity support. In this dossier, such systems are research objects because they can modify the distribution of cognitive tasks and treatment-related workload. Claims about their psychological or medical effectiveness require comparative empirical testing, especially in populations with limited digital access.
Promising study designs include randomized comparisons between usual support and carefully governed digital systems; audits of information provenance and failure modes; measurement of additional patient work; and monitoring for inequitable exclusion. Outcomes should distinguish response to a reminder from successful treatment participation and should assess whether support continues to work after software or service changes.
Ethics, Agency, and the Meaning of Appropriate Adherence
A medication-use measure becomes ethically misleading when it assumes that all prescribed treatment must be continued regardless of changed circumstances, adverse effects, or informed patient preference. Shared decision-making recognizes the importance of what matters to the person, the available evidence, and the practical demands of available options (Elwyn et al. (2012)). A good research definition should permit investigators to distinguish agreed treatment plans from unilateral prescriptions, patient-initiated deviation, and professionally supervised changes.
Nonjudgmental assessment improves the likelihood that people disclose missed doses, cost problems, side effects, and concerns. The National Institute for Health and Care Excellence (2009) provides a directly relevant clinical standard. Informed refusal is a distinct phenomenon from unintentional failure to remember. Research should preserve this distinction while examining the clinical consequences of decisions with rigor.
The ethics of measurement extend to privacy, monitoring, and stigma. Electronic detection of missed doses can help characterize treatment patterns, but continuous surveillance may undermine trust or impose new obligations. The use of adherence data in insurance, employment, or coercive treatment settings raises governance questions beyond measurement accuracy. Research participants should understand what is recorded and how it may be used.
Treatment-burden reduction is ethically important because the benefits of medicine must be pursued within real lives. Yet lowering burden is not equivalent to automatically eliminating effective interventions. Appropriate care integrates scientific evidence, safety, the patient's priorities, and the resources required to implement an acceptable plan. Studies should evaluate these dimensions together.
Integrated Evidence Assessment
The strongest conceptual convergence lies in the recognition of medication use as a process with distinguishable stages, treatment burden as multidimensional workload, and continuity as coherent information, management, and relationships. These constructs are supported by taxonomic work, validated measures, clinical-observational research, and independently designed interventions (Vrijens et al. (2012); Eton et al. (2017); Haggerty et al. (2003)).
Evidence is strongest for the feasibility of measuring selected components and for the existence of modifiable structural and organizational barriers. Randomized studies of financial coverage and care-transition packages demonstrate that system-level interventions can be evaluated experimentally. The evidence that any single adherence intervention reliably improves long-term patient-important outcomes across diseases is more heterogeneous, as emphasized in the Cochrane and JAMA reviews (Nieuwlaat et al. (2014); Kini & Ho (2018)).
Observational work establishes important associations but retains confounding and selection problems. The continuity–mortality literature illustrates how a compelling relationship can coexist with uncertain causal pathways (Pereira Gray et al. (2018)). Qualitative studies identify overlooked domains of work and meaning while being less suited to population effect estimation. Psychometric validation establishes properties of an instrument within studied samples, with further testing needed for cross-cultural invariance and responsiveness.
A central methodological weakness in the field is the frequent use of an adherence proxy as if it were the final clinical outcome. High possession may coexist with inappropriate prescribing, adverse effects, or major workload. Improved understanding of medication use requires parallel measures of regimen appropriateness, patient experience, clinical safety, access, and continuity. Evidence synthesis should report uncertainty by outcome, setting, and design rather than applying one global confidence label to the entire field.
Research Agenda and Testable Questions
A first priority is longitudinal data collection that separates initiation, implementation, persistence, and clinically agreed discontinuation. Studies should combine prescribing, dispensing, patient experience, and relevant clinical information, with explicit protocols for treatment switches and changes in eligibility. Repeated measures of burden and capacity would permit investigation of whether changing workload precedes interruption or mainly reflects worsening illness.
A second priority is development and validation of patient-centered burden outcomes across different health systems, diagnoses, disabilities, age groups, languages, and caregiver arrangements. Treatment burden should be evaluated both as task workload and as its consequences for roles, well-being, and opportunities. Comparisons require tests of measurement invariance and responsiveness rather than assuming all burden instruments are interchangeable.
A third priority is dismantling and factorial evaluation of multicomponent interventions. Financial assistance, reminder systems, clinician counseling, pharmacist support, reconciliation, and continuity coordination can be studied in combinations that identify which components work through which mechanisms. Such designs should measure the time and resources consumed by the intervention and the persistence of effects after formal trial support ends.
A fourth priority is explicit causal identification for continuity. Investigators should distinguish relational, informational, and management interventions, use designs robust to time-varying confounding where randomization is infeasible, and assess both patient-important outcomes and the frequency of serious medication discrepancies. Transparency about how continuity is measured is necessary for interpretable cross-study synthesis.
A fifth priority addresses artificial and digitally mediated care. Evaluations should establish information accuracy, human oversight, failure rates, accessibility, privacy, additional workload, and effects on actual treatment trajectories. Studies of cognitive offloading should determine whether technology reduces tasks overall or transfers responsibility into harder-to-detect forms. The resulting evidence can inform research in Psychology of Artificial and Human–Artificial Relations without assuming results in advance.
The field will advance when adherence, burden, and continuity are studied as interacting, time-varying determinants of treatment participation. The decisive scientific task is to identify how an appropriate regimen can remain beneficial, feasible, safe, and aligned with a person's life as health and circumstances change.
Conclusion
Medication adherence, treatment burden, and care continuity constitute an integrated subject of behavioral medicine and health-services science. Their conceptual distinctions are empirically consequential: initiation, implementation, persistence, and discontinuation describe different medication trajectories; treatment burden measures the work required to manage care and its consequences; and continuity concerns the connection of information, management, and relationships over time.
The scientific literature establishes robust reasons to move beyond individual blame. Memory and motivation matter, alongside affordability, regimen design, symptom burden, competing responsibilities, communication, and the organization of services. Original intervention trials establish that financial and transitional conditions can be changed; psychometric studies make patient workload measurable; and systematic reviews reveal substantial heterogeneity in long-term adherence intervention effects. Stronger causal evidence is still needed for many proposed mechanisms and for their interactions.
An adequate research program evaluates whether medication use is appropriate, whether a person can realistically perform the required work, whether care remains coherent through transitions, and whether outcomes that matter to the person improve. This framework supports more precise psychological explanation, stronger clinical research, and disciplined assessment of digital and artificial supports in the Artificial Era.
References
- Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. (2011). Low health literacy and health outcomes: An updated systematic review. Annals of Internal Medicine, 155(2), 97–107. https://doi.org/10.7326/0003-4819-155-2-201107190-00005
- Choudhry, N. K., Avorn, J., Glynn, R. J., Antman, E. M., Schneeweiss, S., Toscano, M., et al. (2011). Full coverage for preventive medications after myocardial infarction. New England Journal of Medicine, 365(22), 2088–2097. https://doi.org/10.1056/NEJMsa1107913
- Coleman, E. A., Parry, C., Chalmers, S., & Min, S.-J. (2006). The care transitions intervention: Results of a randomized controlled trial. Archives of Internal Medicine, 166(17), 1822–1828. https://doi.org/10.1001/archinte.166.17.1822
- DiMatteo, M. R., Giordani, P. J., Lepper, H. S., & Croghan, T. W. (2002). Patient adherence and medical treatment outcomes: A meta-analysis. Medical Care, 40(9), 794–811. https://doi.org/10.1097/00005650-200209000-00009
- Elwyn, G., Frosch, D., Thomson, R., Joseph-Williams, N., Lloyd, A., Kinnersley, P., et al. (2012). Shared decision making: A model for clinical practice. Journal of General Internal Medicine, 27(10), 1361–1367. https://doi.org/10.1007/s11606-012-2077-6
- Eton, D. T., Yost, K. J., Lai, J.-S., Ridgeway, J. L., Egginton, J. S., Rosedahl, J. K., et al. (2017). Development and validation of the Patient Experience with Treatment and Self-management (PETS): A patient-reported measure of treatment burden. Quality of Life Research, 26(2), 489–503. https://doi.org/10.1007/s11136-016-1397-0
- Gellad, W. F., Grenard, J. L., & Marcum, Z. A. (2011). A systematic review of barriers to medication adherence in the elderly: Looking beyond cost and regimen complexity. American Journal of Geriatric Pharmacotherapy, 9(1), 11–23. https://doi.org/10.1016/j.amjopharm.2011.02.004
- Guthrie, B., Payne, K., Alderson, P., McMurdo, M. E. T., & Mercer, S. W. (2012). Adapting clinical guidelines to take account of multimorbidity. BMJ, 345, e6341. https://doi.org/10.1136/bmj.e6341
- Haggerty, J. L., Reid, R. J., Freeman, G. K., Starfield, B. H., Adair, C. E., & McKendry, R. (2003). Continuity of care: A multidisciplinary review. BMJ, 327(7425), 1219–1221. https://doi.org/10.1136/bmj.327.7425.1219
- Jack, B. W., Chetty, V. K., Anthony, D., Greenwald, J. L., Sanchez, G. M., Johnson, A. E., et al. (2009). A reengineered hospital discharge program to decrease rehospitalization: A randomized trial. Annals of Internal Medicine, 150(3), 178–187. https://doi.org/10.7326/0003-4819-150-3-200902030-00007
- Kardas, P., Lewek, P., & Matyjaszczyk, M. (2013). Determinants of patient adherence: A review of systematic reviews. Frontiers in Pharmacology, 4, 91. https://doi.org/10.3389/fphar.2013.00091
- Kini, V., & Ho, P. M. (2018). Interventions to improve medication adherence: A review. JAMA, 320(23), 2461–2473. https://doi.org/10.1001/jama.2018.19271
- Kripalani, S., LeFevre, F., Phillips, C. O., Williams, M. V., Basaviah, P., & Baker, D. W. (2007). Deficits in communication and information transfer between hospital-based and primary care physicians: Implications for patient safety and continuity of care. JAMA, 297(8), 831–841. https://doi.org/10.1001/jama.297.8.831
- Kvarnström, K., Westerholm, A., Airaksinen, M., & Liira, H. (2021). Factors contributing to medication adherence in patients with a chronic condition: A scoping review of qualitative research. Pharmaceutics, 13(7), 1100. https://doi.org/10.3390/pharmaceutics13071100
- Lee, J. D., Nunes, E. V., Jr., Novo, P., Bachrach, K., Bailey, G. L., Bhatt, S., et al. (2018). Comparative effectiveness of extended-release naltrexone versus buprenorphine-naloxone for opioid relapse prevention (X:BOT): A multicentre, open-label, randomised controlled trial. The Lancet, 391(10118), 309–318. https://doi.org/10.1016/S0140-6736(17)32812-X
- Lieberman, J. A., Stroup, T. S., McEvoy, J. P., Swartz, M. S., Rosenheck, R. A., Perkins, D. O., et al. (2005). Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. New England Journal of Medicine, 353(12), 1209–1223. https://doi.org/10.1056/NEJMoa051688
- May, C., Montori, V. M., & Mair, F. S. (2009). We need minimally disruptive medicine. BMJ, 339, b2803. https://doi.org/10.1136/bmj.b2803
- National Institute for Health and Care Excellence. (2009). Medicines adherence: Involving patients in decisions about prescribed medicines and supporting adherence (CG76). https://www.nice.org.uk/guidance/cg76
- Nieuwlaat, R., Wilczynski, N., Navarro, T., Hobson, N., Jeffery, R., Keepanasseril, A., et al. (2014). Interventions for enhancing medication adherence. Cochrane Database of Systematic Reviews, (11), CD000011. https://doi.org/10.1002/14651858.CD000011.pub4
- Pereira Gray, D. J., Sidaway-Lee, K., White, E., Thorne, A., & Evans, P. H. (2018). Continuity of care with doctors—A matter of life and death? A systematic review of continuity of care and mortality. BMJ Open, 8(6), e021161. https://doi.org/10.1136/bmjopen-2017-021161
- Piette, J. D., Heisler, M., & Wagner, T. H. (2004). Cost-related medication underuse among chronically ill adults: The treatments people forgo, how often, and who is at risk. American Journal of Public Health, 94(10), 1782–1787. https://doi.org/10.2105/AJPH.94.10.1782
- Raebel, M. A., Schmittdiel, J., Karter, A. J., Konieczny, J. L., & Steiner, J. F. (2013). Standardizing terminology and definitions of medication adherence and persistence in research employing electronic databases. Medical Care, 51(8 Suppl 3), S11–S21. https://doi.org/10.1097/MLR.0b013e31829b1d2a
- Shippee, N. D., Shah, N. D., May, C. R., Mair, F. S., & Montori, V. M. (2012). Cumulative complexity: A functional, patient-centered model of patient complexity can improve research and practice. Journal of Clinical Epidemiology, 65(10), 1041–1051. https://doi.org/10.1016/j.jclinepi.2012.05.005
- Steiner, J. F., & Prochazka, A. V. (1997). The assessment of refill compliance using pharmacy records: Methods, validity, and applications. Journal of Clinical Epidemiology, 50(1), 105–116. https://doi.org/10.1016/S0895-4356(96)00268-5
- Tarn, D. M., Heritage, J., Paterniti, D. A., Hays, R. D., Kravitz, R. L., & Wenger, N. S. (2006). Physician communication when prescribing new medications. Archives of Internal Medicine, 166(17), 1855–1862. https://doi.org/10.1001/archinte.166.17.1855
- Tran, V.-T., Harrington, M., Montori, V. M., Barnes, C., Wicks, P., & Ravaud, P. (2014). Adaptation and validation of the Treatment Burden Questionnaire (TBQ) in English using an internet platform. BMC Medicine, 12, 109. https://doi.org/10.1186/1741-7015-12-109
- Tran, V.-T., Barnes, C., Montori, V. M., Falissard, B., & Ravaud, P. (2015). Taxonomy of the burden of treatment: A multi-country web-based qualitative study of patients with chronic conditions. BMC Medicine, 13, 115. https://doi.org/10.1186/s12916-015-0356-x
- Vijan, S., Hayward, R. A., Ronis, D. L., & Hofer, T. P. (2005). Brief report: The burden of diabetes therapy: Implications for the design of effective patient-centered treatment regimens. Journal of General Internal Medicine, 20(5), 479–482. https://doi.org/10.1111/j.1525-1497.2005.0117.x
- Vrijens, B., De Geest, S., Hughes, D. A., Kardas, P., Demonceau, J., Ruppar, T., et al. (2012). A new taxonomy for describing and defining adherence to medications. British Journal of Clinical Pharmacology, 73(5), 691–705. https://doi.org/10.1111/j.1365-2125.2012.04167.x