Adherence to Medical Recommendations

Nicole E Berlant & Sheri D Pruitt. The Health Psychology Handbook: Practical Issues for the Behavioral Medicine Specialist. Editor: Lee M Cohen. Sage Publications. 2003.

The failure to adhere to medical recommendations is a significant and multifaceted health care problem. Estimates are that 30% to 70% of patients do not fully adhere to the medical advice of their physicians (National Heart, Lung, and Blood Institute, 1998). Moreover, up to 80% of patients are unsuccessful in following recommendations for behavioral changes such as smoking cessation and dietary restrictions.

Adherence is a complex behavioral process that is determined largely by environmental influences on the patient. However, the patient’s environment is broad, extending beyond his or her immediate surroundings to encompass associated health care providers and the health care organization in which the patient receives services. Nonadherence is far more than a patient problem, and efforts that focus solely on the patient impede the ability to make meaningful advances in the adherence arena.

The expansion of the conceptualization of adherence to include multiple levels (i.e., patients, providers, and health care organizations), and the realization that there are multiple influences on a patient’s behavior within each of these levels, could make the challenge of improving adherence appear to be insurmountable. Fortunately, behavioral science offers valuable principles, theories, and models that address the determinants of behaviors of patients, providers, and organizations (Pruitt, 2001). These same principles, theories, and models from behavioral science provide a conceptual framework for organizing current knowledge, testing future hypotheses, and developing multilevel interventions for improving patients’ adherence to medical recommendations.

Definitions and Description of Adherence

Although adherence has been defined as the extent to which patient behavior corresponds with recommendations from a health care provider (Rand, 1993; Vitolins, Rand, Rapp, Ribisl, & Sevick, 2000), this broad definition belies the complexity of the issue. Adherence is better conceptualized as an acceptable frequency, intensity, and/or accuracy of specific behaviors, given a specific circumstance, that is associated with improved clinical outcomes. Adherence is a process—a behavioral means to the end point of better health status.

The process of adherence is influenced by multiple determinants. These include a variety of factors at the patient, provider, and health care organization levels. Factors at the patient level consist of a person’s knowledge and beliefs about illness, degree of motivation and self-efficacy related to illness management behaviors, and expectancies related to the outcomes of adherence or nonadherence. In addition, disease- and treatment-specific issues influence adherence at the patient level. These include (a) the social, physical, psychological, and occupational disabilities resulting from symptoms and treatments; (b) the cultural meanings of diseases and treatments; (c) the disease severity and prognosis; and (d) the complexity, timing, and degree of beneficial and detrimental effects of treatments.

Influential factors at the provider level encompass the knowledge, skills, and attitudes of health care providers as well as the quality of the patient-provider relationship. Health care organization factors involve characteristics of the organization such as access to diagnostic and treatment services, education to manage health problems, coordination and integration of care, and organizational links to community support services (World Health Organization, 2001).

The need for a clear conceptualization of the adherence construct becomes even more evident when measurement and intervention strategies are taken into consideration. For example, provider recommendations range from advice that requires relatively simple and familiar behaviors (e.g., requests for patients to return for follow-up appointments or to obtain inoculations) to recommendations that patients participate in complex and novel regimens (e.g., daily alterations in diet, smoking cessation, increase in physical activity). Measurement and intervention strategies obviously differ according to the circumstances and/or intensity of the recommendations. Nevertheless, it is important to note that adherence is fundamental for successful management of health problems, and patients are asked to follow some degree of recommendations for all medical conditions.

With the possible exception of psychiatric disorders, adherence is not specific to a particular condition or condition regimen (Haynes, 1979). Basic behavioral principles and models of behavior change cut across all medical conditions, rendering a review of each specific condition less important than a grasp of the common themes. Moreover, increasing the effectiveness of adherence interventions may have a far greater impact on the health of the population than would any improvements in specific medical treatments (Haynes et al., 2002).

Although adherence traditionally is discussed as a patient problem (e.g., “The patient is nonadherent”), the time has come to consider the adherence issue within the larger health care organization context. Adherence is a multifaceted construct that spans patients’ medical problems, but providers and the health care organization must be considered as similarly responsible for improving adherence rates. Clarifying our conceptualization of the construct could serve to forward intervention efforts. Better interventions are imperative because if significant improvements in adherence are not made, many medical treatments and providers’ therapeutic efforts will continue to be inefficient.

The Role of Behavioral Science in Adherence

The influence of behavior on health has never been as apparent as it is today. Most people living in this century will die from “lifestyle illnesses” or medical conditions resulting from daily habits such as diet, exercise, alcohol consumption, tobacco use, and/or health risk behaviors (Kaplan, Sallis, & Patterson, 1993). In fact, current estimates indicate that daily behavior contributes as much as 50% to an individual’s overall health status (Institute for the Future, 2000). Decades of behavioral research provide proven strategies for changing behavior. Integrating this knowledge within the medical arena can help patients to alter their daily patterns and reduce the risks associated with the development of health problems and/or negative consequences of preestablished conditions. Moreover, behavior change strategies can be applied across diverse medical conditions (Dunbar-Jacob, Burke, & Pyczynski, 1995; Nessman, Carnahan, & Nugent, 1980). In addition, the same learning principles that are effective in changing patient behavior are effective in altering the behavior of health care providers (Oxman, Thomson, Davis, & Haynes, 1995) and health care organizations (DeBusk et al., 1994).

Behavioral science is the application of experimental methods to learn about, predict, and explain the observable actions of humans, including not only observable behaviors but also verbal statements about subjective experiences and symptoms. Much of human behavior has been well understood for decades. The most basic yet most powerful principle is the influence of antecedents and consequences on behavior, that is, operant learning (Skinner, 1938, 1953). Antecedents, or preceding events, are internal (thoughts) or external (environmental cues) circumstances that elicit a behavior. Consequences, or expected consequences, that can be conceptualized as rewards or punishments also influence behavior. These principles translate into the ability to predict the probability of a patient, provider, or health care organization initiating or continuing a behavior because such behaviors partially depend on what happens before and after the specific behavior occurs. Although learning theory historically has been criticized for explaining behavior in overly simplistic “stimulus-response” relationships, contemporary learning theory integrates environmental cues and contexts, memory, expectancies, and neurological processes related to learning (Institute of Medicine, 2001).

From a theoretical standpoint, it would be possible to “control” behavior of patients, providers, and health care organizations if the events preceding and following a specific behavior could be controlled. From a practical standpoint, behavioral principles can be used to design interventions that have the potential to shape behavior at each incremental level of influence (patient, provider, and health care organization) to address adherence problems.

Behavioral interventions based on the basic principles of learning (i.e., antecedents and consequences) are quite potent for changing behavior. New behaviors can be learned, and established behaviors can be increased or decreased using these concepts. In fact, a variety of health behaviors have been altered successfully using some variation of operant-based techniques (Brownell & Cohen, 1995; Janis, 1983; Mahoney, 1974). In general, behavioral change strategies that focus on what occurs before and after targeted behaviors have been substantially more effective than other approaches.

The most effective methods for changing behavior are those that teach individuals how to integrate the basic principles of learning into their daily lives (Bandura & Simon, 1977; Matarazzo, 1980). This practice is called “self-management,” “self-directed behavior change,” or “self-regulation.” Specific techniques for changing and maintaining one’s health behavior include self-monitoring, goal setting, stimulus control, self-reinforcement, behavioral rehearsal, arranging social support, behavioral contracting, and relapse prevention. The use of behavioral principles in the area of relapse prevention has been particularly well studied in response to the ubiquitously high rates of relapse after health behavior changes such as smoking cessation, reduction in alcohol consumption, and weight loss. Research in relapse prevention has demonstrated the significance of contextual cues (e.g., physical environment, time of day, emotional status) in the maintenance or extinction of health behaviors.

Although the fundamental principles of learning and behavior change appear to be simple, they are deceptively so. Behavior change and maintenance continues to be an enigma, and even the best behavioral techniques are not invariably effective. Nevertheless, a century of behavioral science remains the strongest foundation to guide current efforts in changing behavior to improve health.

The obligation to integrate behavioral science into physical health was recognized 25 years ago when the interdisciplinary field of behavioral medicine was formally defined (Schwartz & Weiss, 1977). The adherence problem is ideally attacked from this interdisciplinary perspective. In fact, researchers in the behavioral medicine arena have developed sophisticated models that envelop the basic principles of learning and apply them to complex health behaviors.

Models for Understanding Adherence Behavior

Theories and models provide a conceptual framework for organizing thoughts about adherence and other health behaviors. Over the past 100 years, numerous theories about behavior have been proposed. This section briefly reviews some of the more influential models: the social cognitive theory (Bandura & Simon, 1977), the theory of reasoned action (Ajzen & Fishbein, 1980), the health belief model (Rosenstock, 1974), the theory of interpersonal behavior (Triandis, 1977), the transtheoretical model (Prochaska & DiClemente, 1982), and the information-motivation-behavioral skills (IMB) model (Fisher & Fisher, 1992; Fisher, Fisher, Miscovich, Kimble, & Malloy, 1996).

Social Cognitive Theory

Bandura’s social cognitive theory (Bandura, 1982; Bandura & Simon, 1977) suggests that behavioral changes are predicated on the belief that a person can successfully complete a desired behavior. This belief is necessary even when all other predictors of behavior would suggest that a person is ready to engage in a particular behavior. For example, a person can feel vulnerable to a disease, understand how to change his or her behavior to be healthier, believe that the new behavior will decrease the likelihood of illness, and feel supported by the social environment. However, if the person lacks conviction in his or her ability to change, the social cognitive theory predicts that the person is unlikely to be successful.

The construct regarding a person’s belief about ability to change is called “perceived self-efficacy” and is modified by four sources of information: (a) performance attainment or success with previous tasks, (b) vicarious experience or watching others perform a task, (c) verbal persuasion, and (d) physiological states such as anxiety and relaxation. The social cognitive theory predicts that self-efficacy influences adherence by mediating behaviors in multiple ways, including immediate behavior choices, effort expenditure, thoughts, emotional reactions, and behavior performance. There is empirical support for the theory as it relates to adherence and the maintenance of behavior change, but Bandura’s theory is limited by the complexity of quantifying the conceptual elements in the model.

Theory of Reasoned Action

The theory of reasoned action was introduced in an attempt to explain the relationships among beliefs (normative and behavioral), attitudes, intentions, and behaviors. According to this theory, an individual’s intent to adopt a behavior is determined by his or her attitude about performing the behavior and social factors such as the perception of attitudes about the behavior held by significant others (Ajzen & Fishbein, 1980). Ajzen and Fishbein (1980) suggested that attitude toward a behavior is a much better predictor of that behavior than is attitude toward the disease that is associated with the behavior. For example, attitude toward mammography should be a better predictor of screening behavior than is attitude toward breast cancer. Personality and sociocultural variables influence the likelihood of adopting a behavior by mediating the attitudes of the individual and of his or her significant others. Both the degree of influence imposed by the person’s attitude and his or her perception of other’s beliefs vary with each behavior.

Health Belief Model

The health belief model (Becker & Maiman, 1975; Rosenstock, 1974) integrates behavioral and cognitive theories to explain why people fail to adhere to health behaviors by considering the impact of the consequences and expectations related to the behaviors. Specifically, the probability that a person will adopt or maintain a behavior to prevent or control a disease depends on four things: (a) perceived susceptibility, (b) perceived threat, (c) perceived benefits, and (d) perceived barriers. Perceived susceptibility describes the perception of risk to personal health, whereas perceived threat describes the proportion of negative consequences of disease. Perceived benefits and barriers describe the beliefs about the outcomes of recommended behaviors in reducing the perceived threat. It should be noted that the health belief model highlights the fact that adherence to health behaviors often requires people to act to prevent illness even while they are still healthy.

Theory of Interpersonal Behavior

The theory of interpersonal behavior (Triandis, 1977) interrelates a person’s intention to perform a behavior, facilitating conditions, and habit. In this model, intention is mediated by a cognitive analysis of the pros and cons of adopting a behavior, an affective analysis of previous positive and negative experiences, the social influence of normative and role beliefs, and personal beliefs about responsibility for one’s health. Habit influences the likelihood of behavior when regular behaviors become automatic, at which point the role of intention is reduced.

Transtheoretical Model

The transtheoretical model (Prochaska & DiClemente, 1982) proposes the “stages of change” framework as a comprehensive model of behavioral change in both the positive and negative directions (e.g., the acquisition of a health behavior such as exercise, the reduction or cessation of a risk behavior such as smoking). Originally developed as a smoking cessation tool, the transtheoretical model has been applied to psychotherapy and a variety of health behaviors as a way of matching individuals to their stage of preparedness for behavior change. Cross-sectional studies have supported the existence of five stages (precontemplation, contemplation, preparation, action, and maintenance), but some longitudinal studies have not identified these discrete stages. In addition, a recent investigation has reported the stages to be weak predictors of smoking cessation (Institute of Medicine, 2001). Nevertheless, the finding that many people have low levels of motivation for behavior change has led to interventions specifically intended to increase motivation such as Miller and Rollnick’s (2002) motivational interviewing strategies.

Information-Motivation-Behavioral Skills Model

Each of the theories just discussed has advantages and disadvantages, yet none of them readily translates into a comprehensive intervention for changing health behavior. The recently developed information-motivation-behavioral skills model (Fisher & Fisher, 1992; Fisher et al., 1996) borrows elements from the earlier theories to construct a conceptually based, generalizable, and parsimonious model to guide thinking about complex health behaviors. Subjected to rigorous empirical investigation, interventions based on this model have demonstrable efficacy in effecting behavioral change across a variety of clinical applications (Carey, Kalichman, Forsyth, Wright, & Johnson, 1997; Fisher & Fisher, 1992; Fisher et al., 1996). In both prospective and correlational studies, the information, motivation, and behavioral skills constructs have accounted for an average of 33% of the variance in behavior change (Fisher et al., 1996).

The IMB model, similar to what has been reported previously, demonstrates that information is a prerequisite but that information in itself is insufficient to alter behavior (see Mazzuca, 1982). It provides evidence that motivation and behavioral skills are critical determinants that are independent of behavior change (Fisher & Fisher, 1992; Fisher et al., 1996). Information and motivation work largely through behavioral skills to affect behavior. However, when the behavioral skills are familiar or uncomplicated, information and motivation can have direct effects on behavior. In this case, a patient might fill a prescription (a simple familiar behavior) based on information given by the provider. The relationship between the information and motivation constructs is weak. In practical terms, a highly motivated person may have little information, or a highly informed person may have low motivation. However, in the IMB model, the presence of both information and motivation increases the likelihood of complex behavior change.

To this point, each of the components of the IMB model has been described. “Information” consists of basic knowledge about a medical condition and effective strategies for managing it. “Motivation” encompasses personal attitudes toward the behavior, perceived social support for such behavior, and the patient’s subjective norm or perception of how others with this medical condition might behave. “Behavioral skills” includes ensuring that the patient has the specific behavioral tools or strategies necessary to perform the adherence behavior such as enlisting social support and other self-regulation strategies. Finally, information, motivation, and behavioral skills must pertain directly to the desired behavioral outcome; that is, they must be specific.

Much of the adherence research and interventions applies individual components of the IMB model despite evidence that all three elements are necessary for complex behavior change. The failure to explicitly implement information, motivation, and behavioral skills may be partially attributed to the commonsensical nature of the model. Health care providers often assume that they provide information to patients and motivate them, and providers also recognize the importance of behavioral skills in improving health. However, there is evidence that providers typically give limited information (Waitzkin & Stoeckle, 1976), lack motivational enhancement abilities (Botelho & Skinner, 1995), and lack the knowledge (often leading to frustration) in teaching patients behavioral skills (Alto, 1995).

The expense of intensively educating physicians to improve their information dissemination, motivational interviewing, and behavioral skills training may be prohibitive when considering efficient and effective strategies for improving patient adherence. However, creative education strategies, such as distance learning techniques, have been used successfully to train physicians in basic behavioral concepts to influence adherence (Casebeer, Klapow, Centor, Stafford, & Skrinar, 1999). Training less expensive or more readily available providers in the application of the IMB model may be a more viable option. Pharmacists, case managers, health educators, and any persons involved in patient care should be exposed to these basic concepts. Nonphysician providers have an incredibly important role and opportunity to improve significantly the health of their patients by specifically targeting patient adherence issues.

More structured, thoughtful, and sophisticated provider-patient interactions are essential if improvements in adherence are to be realized. The generalizable IMB model can be applied to providers to meet this goal. As this empirically based model predicts, when providers have adequate information, motivation, and behavioral skills, they will integrate new behaviors into their practices. Adapted to an organizational level, the same IMB framework can be used to change the behavior of decision makers and administrators toward improved health care organization functioning.

State-of-the-Art Interventions for Improving Adherence

Adherence intervention research has focused largely on patient behavior and medication regimens as opposed to targeting provider and health care organization variables. According to several published adherence reviews, no single intervention targeting patient behavior is effective, and the most promising methods of improving adherence behavior use a combination of the following strategies (Houston-Miller, Hill, Kottke, & Ockene, 1997; Haynes et al., 2002; Roter et al., 1998):

  • Patient education (Morisky et al., 1983)
  • Behavioral skills (Oldridge & Jones, 1983; Swain & Steckel, 1981)
  • Self-rewards (Mahoney, Moura, & Wade, 1973)
  • Social support (Daltroy & Godin, 1989)
  • Telephone follow-up (Taylor, Houston-Miller, Killen, & DeBusk, 1990)

Various combinations of these techniques have been shown to increase adherence behavior and treatment outcomes. However, even the most efficacious patient-focused interventions do not yield substantial effects for adherence behavior over the long term (Haynes, McKibbon, & Kanani, 1996), and few randomized controlled trials targeting patient adherence behavior exist (Haynes et al., 2002).

A recent review of the long-term management of obesity (Perri, 1998) described many of these techniques by examining the status of research concerning adherence and relapse prevention in weight management. A number of strategies to increase adherence to weight control behaviors have been investigated, including continuing therapist contact, formal relapse prevention training, monetary incentives, low-calorie food provision, and peer support. Intensive behavioral therapist contact (beyond 6 months) has been repeatedly demonstrated to prolong the maintenance of weight loss, although contact does not result in greater weight loss (Perri, 1998).

Relapse prevention training has been used successfully as part of a multicomponent maintenance program (Perri, Shapiro, Ludwig, Twentyman, & McAdoo, 1984). Marlatt and Gordon’s (1985) “relapse prevention” is defined as a set of techniques designed to keep people from relapsing to prior health habits after initial successful behavior modification, including training and coping skills for high-risk relapse situations and lifestyle rebalancing (Taylor, 1995). In addition, peer support meetings have been associated with greater weight loss maintenance over time. However, neither of these components has been shown to lead to greater behavioral adherence (Perri et al., 1987). Multicomponent maintenance programs lead to greater sustained weight loss than does standard care, but within these programs continued therapist contact appears to be the key component. Perri (1998) hypothesized that the improved outcomes seen in extended treatment programs are due to the maintained adherence to behavior changes. This extended adherence is likely secondary to the ongoing effects of the social pressure of groups, repeated cues for “appropriate” eating and exercise, continued therapist reinforcement and problem solving, and sustained motivation and morale from continued therapeutic support.

Adherence Interventions at the Patient Level

Most people have difficulty in adhering to medical recommendations, especially when the advice entails self-administered care. Consequently, patient characteristics have been the focus of numerous adherence investigations. Efforts to identify stable personality traits of the “nonadherent patient” have been futile. However, mental health problems have been examined in recent reports, and there is evidence that depression and anxiety are predictive of adherence to medical recommendations (Chesney, Chrisman, Luftey, & Pescosolido, 1999; DiMatteo, Lepper, & Croghan, 2000; Lustman et al., 1995; Ziegelstein et al., 2000). Interestingly, providers historically have attributed adherence problems to patients’ personalities (Davis, 1966) or attitudes (Stone, 1979). It may be that providers are detecting mental health problems, such as depression, but are inaccurately labeling these problems as “attitudinal” or “personality” faults in their patients. Such mis-attribution leads to a failure to treat possible underlying mental health disorders that, if treated, can improve patient adherence.

Adherence Interventions at the Provider Level

Because providers play a significant role in adherence, designing interventions to affect their performance seems to be a reasonable strategy, but investigations in this area are few. Providers prescribe the medical regimen, interpret it, monitor clinical outcomes, and provide feedback to patients (Center for the Advancement of Health, 1999). Accordingly, provider communication has been widely examined, and importantly, associations with patient health outcomes have been demonstrated. In a review of randomized controlled trials, Stewart (1996) reported that providers who share information, build partnerships, and provide emotional support to their patients have better outcomes than do providers who do not interact with patients in this manner. Correlational studies reveal a direct relationship between patient adherence and provider communication styles that include providing information, engaging in “positive talk,” and asking patients specific questions about adherence (Hall, Roter, & Katz, 1988). Patient satisfaction also plays a role in that those who are satisfied with their providers and medical regimens adhere to recommendations more diligently (Whitcher-Alagna, 1983). Finally, patients who view themselves as partners engaged in their treatment plans have better adherence behavior and health outcomes (Schulman, 1979).

One example of the potential for providers to affect health behavior change is the use of “minimal contact” interventions in primary care to help patients quit smoking. Advice or counseling alone produces increased 6- and 12-month quit rates in biochemically validated studies (e.g., Ockene et al., 2000). The most effective primary care interventions include several core elements such as a strong provider-delivered “quit smoking” message; self-help materials covering motivational, behavioral, and relapse prevention strategies; a prescription for nicotine replacement therapy; brief counseling that includes setting a quit date; and follow-up support (Glasgow & Orleans, 1997). The American Medical Association has recognized the important influence of health care providers in reducing smoking rates and during the early 1990s created guidelines for the treatment of nicotine addiction (American Medical Association, 1993). The guidelines recommend that providers do the following:

  • Ask about smoking at every opportunity.
  • Advise all smokers to quit.
  • Assist smokers to quit through the use of self-help materials and nicotine replacement whenever appropriate.
  • Arrange follow-up contacts.

The ask-advise-assist-arrange model has been used successfully for inpatient and outpatient settings and has resulted in quit rates significantly higher than usual-care approaches (Glasgow & Orleans, 1997).

Adherence Interventions at the Health Care Organization Level

Health care organizations have the potential to influence patient adherence behavior as well given that they control access to care. For example, organizations direct providers’ schedules, appointment lengths, allocation of resources, fee structures, communication/information systems, and organizational priorities. As such, health care organizations ultimately influence patients’ behavior in many ways. Organizations set parameters of care (e.g., appointment length), often leading providers to report that their schedules do not allow enough time to address adherence behavior adequately (Ammerman et al., 1993). Fee structures are determined by organizations, and many systems (e.g., fee-for-service) lack financial reimbursement for patient counseling and education, substantially threatening adherence-focused interventions. The allocation of resources within an organization may result in high stress and increased demands on providers that in turn have been associated with decreased patient adherence behavior (DiMatteo & DiNicola, 1982).

Furthermore, organizations determine continuity of care. Patients demonstrate better adherence behavior when they receive care from the same health care provider over time and when patient information is communicated with other providers (Meichenbaum & Turk, 1987). For example, the ability of clinics and pharmacies to share information regarding patients’ behavior around prescription refills has the potential to improve adherence. This is possible because the information allows health care providers to track patients’ use of medication as a proxy of medication adherence. Patients can be contacted if they are using medications at a rate that is too fast or too slow. In addition, organizations determine the level of communication with patients. Ongoing communication efforts (e.g., phone contacts) that help to keep patients engaged in their health care may be the most simple and cost-effective strategy for improving adherence (Haynes et al., 1996).

The “state-of-the-art” interventions in adherence target each level of the adherence problem mentioned previously (patient, provider, and health care organization). Several programs have demonstrated good results using a multilevel team approach (Hypertension Detection and Follow-up Program Cooperative Group, 1979; Multiple Risk Factor Intervention Trial Research Group, 1982; SHEP Cooperative Research Group, 1991). In fact, adequate evidence exists to support the effectiveness of innovative, modified health care teams over traditional, independent physician practice and minimally structured organizations (DeBusk et al., 1994; Peters, Davidson, & Ossorio, 1995).

State-of-the-Art Measurement of Adherence Behavior

Accurate assessment of adherence behavior is necessary for effective and efficient treatment planning and for ensuring that changes in health outcomes can be attributed to the recommended regimen. In addition, decisions to change recommendations, medications, and/or communication style so as to invoke patient participation depend on valid and reliable measurement of the adherence construct. Indisputably, there is no “gold standard” for measuring adherence behavior (Farmer, 1999; Vitolins et al., 2000). However, a variety of strategies have been reported in the literature.

Subjective Measures of Adherence Behavior

One measurement approach is to ask providers and patients to provide their subjective ratings of adherence behavior. When providers rate the degree to which patients follow their recommendations, providers overestimate adherence behavior (DiMatteo & DiNicola, 1982; Norell, 1981). The validity of patients’ subjective reports has been problematic as well. Patients who reveal that they have not followed advice tend to describe their behavior accurately (Cramer & Mattson, 1991), whereas patients who deny their failure to follow recommendations tend to report their behavior inaccurately (Spector et al., 1986). Other subjective rating indicators include standardized, patient-administered questionnaires (e.g., Morisky, Green, & Levine, 1986). These questionnaires have typically been used to assess global patient characteristics or “personality” traits that have proved to be poor predictors of adherence behavior (Farmer, 1999). There are no stable (i.e., trait) factors that reliably predict adherence. However, questionnaires that assess specific behaviors that relate to specific medical recommendations, such as food frequency questionnaires (Freudenheim, 1993) used for measuring eating behavior to improve management of obesity, may be reasonable predictors of adherence behaviors (Sumartojo, 1993).

Objective Measures of Adherence Behavior

Another approach in assessing adherence behaviors is using objective measures. Although objective strategies may initially appear to be an improvement over subjective approaches, both approaches have their drawbacks. For example, remaining dosage units (e.g., tablets) can be counted at clinic visits, but counting inaccuracies are common and typically result in overestimations of adherence behavior (Matsui et al., 1994). In addition, important information, such as the timing of dosages and the patterns of missed dosages, is not captured using this strategy. Recently, electronic monitoring devices, such as the Medication Event Monitoring System (MEMS), have been used to record the time and date that a medication container was opened, thereby giving a better description of the manner in which patients take their medications (Cramer & Mattson, 1991). Unfortunately, the expense of these devices precludes their widespread use. Pharmacy databases can also be used to check when prescriptions are filled initially, refilled over time, and/or discontinued prematurely. However, one problem with this approach is that complete information is difficult to obtain given that patients may use more than one pharmacy or data might not be routinely captured.

Biochemical Measures of Adherence Behavior

Biochemical measurement is a third approach for assessing adherence behavior. Nontoxic biological markers can be added to medications, and their presence in blood or urine can provide evidence that a patient recently received a dose of the medication under examination. This assessment strategy is far from perfect given that findings can be misleading and influenced by a variety of individual factors, including diet, absorption, and rate of excretion (Vitolins, 2000).

In sum, adherence measurement provides useful information that outcome monitoring alone cannot provide, but it remains only an estimation of a patient’s actual behavior. Several of the measurement strategies are extremely costly (e.g., MEMS) or depend on information technology (e.g., pharmacy databases) that is not available in many organizations. Determining the “best” measurement strategy to get an approximation of adherence behavior requires taking all considerations into account. Most important, the strategies employed must meet basic psychometric standards of acceptable reliability and validity (Nunnally & Bernstein, 1994). Additional considerations include meeting the goals of the provider or researcher, the accuracy requirements associated with the regimen, the available resources, the response burden on the patient, and how the results will be used. Finally, because no solitary measurement strategy has been deemed optimal, a multimethod approach that combines feasible self-report and reasonable objective measures is the current state of the art in the measurement of adherence behavior.

Case Study

Kaiser Permanente, a large health maintenance organization, is addressing the problem of adherence at the patient, provider, and health care organization levels. Most recently, chronic conditions have been the target of organizational efforts directed toward improving adherence. Specific factors at each level known to influence patients’ ability to adhere to medical recommendations are addressed, with the ultimate goal of improving health outcomes in patients with chronic problems.

One strategy for improving adherence rates is to offer a variety of appointment formats and time frames (e.g., individual, group, interdisciplinary team led, telephone) so as to increase the amount of contact between patients and providers. Offering a range of services maximizes opportunities for providers and patients to attend to adherence issues and incorporates professionals other than physicians into clinical contacts. Clinical health educators, registered dieticians, pharmacists, care managers, and clinical health psychologists are examples of nonphysician providers at Kaiser Permanente who augment patient care and serve to increase the frequency of patient contacts within the health care organization. Nonphysician providers have different areas of expertise and different approaches to the adherence problem. Moreover, they can extend physicians’ time and influence by supporting physicians’ recommendations through additional education, motivation, coordination, and self-management support.

Another strategy for improving adherence at Kaiser Permanente is to ensure the quality of the patient-provider relationship. Patients are assigned a consistent primary care provider (PCP) who oversees and coordinates all medical care. The PCP establishes an ongoing and open relationship with the patient to maintain continuity of care over time and to enhance two-way communication and shared responsibility for chronic condition management. Most Kaiser Permanente PCPs participate in group appointments during which adherence issues are routinely addressed. Patients learn about their conditions, undergo motivational enhancement to initiate and maintain new behaviors, and learn new behavioral skills to implement for the daily management of their health problems.

Sophisticated information systems also help to identify patients having difficulty in adhering to recommendations. Kaiser Permanente’s electronic systems facilitate communication among primary care, specialty care, inpatient care, and pharmacy services. PCPs and specialists are able to monitor pharmacy data on prescription refills and use this information as a proxy of adherence to medication regimens. Laboratory results are also readily available to PCPs, and patients’ attendance at group appointments or educational classes can be monitored. Treatment protocols specific to different chronic conditions allow for the ongoing monitoring of a number of biological indexes and of adherence to screening and prevention activities.

Finally, there are special efforts to promote continuity and contact with the organization to maximize patients’ adherence behavior and to minimize or delay expensive disease complications. Care is planned; follow-up appointments, laboratory tests, and health education group appointments are scheduled. When patients do not adhere to the care protocols, they are contacted through a variety of outreach strategies. These strategies include regular patient newsletters, individualized reminder letters from their PCPs, and telephone calls from care managers.

Patient-centered outcomes continue to improve. Biological indexes have improved across conditions such as heart disease, asthma, and diabetes. Screening and prevention activities have increased, and hospital admissions for patients with chronic conditions have declined.


Adherence is behavior. Changing it becomes increasingly more difficult as patients are asked to learn new behaviors, alter their daily routines, and maintain the changes over time (Malahey, 1966; Marlatt & Gordon, 1985; Zola, 1981). Effective strategies that consistently change and maintain complex behaviors across time might never be discovered. However, substantial evidence exists in the behavioral science arena identifying the most effective strategies for changing behavior. This abundant body of research should guide future intervention efforts. Behavioral science also provides fundamental principles of behavior and empirically evaluated models that can serve as a framework for organizing the conceptualization of the adherence problem.

The fundamental concepts from behavioral science apply to behavior in general, including all medical conditions and the recommendations for their management, health care provider behavior, and the behavior of health care organizations. The conceptual models (e.g., IMB model) provide a framework for these behavioral principles. Given this knowledge, efforts for improving adherence behaviors can be focused and intensified. Moreover, although complete control is impossible, consistent consideration of the significance of events that precede and follow behavior at the patient, provider, and health care organization levels will advance adherence enhancement efforts and ultimately affect health outcomes.