Patient Adherence

Adherence is a term used to describe the extent to which an individual’s behavior corresponds to the health-related recommendations of that individual’s health care provider. The term has been used broadly, in reference to medication regimens, dietary restrictions, exercise recommendations, smoking cessation, screening participation, and other health-protective behaviors. Although similar in meaning to “compliance,” the word “adherence” is preferred by many providers because it emphasizes the collaborative nature of treatment and prevention, rather than implying a passive and perfunctory approach on the part of the patient.

Several emerging trends have amplified the importance of adherence over the last century. With advances in modern medical treatment and prevention research, providers are increasingly able to offer helpful behavioral recommendations to their patients. Also, with the growing significance of chronic illnesses rather than acute conditions as major health threats (as reported by the National Center for Chronic Disease Prevention and Health Promotion), patients have been obligated to assume a more active role in their own health care. Especially when treating conditions that require lifestyle changes and self-care on the part of the patient, health care providers often find the efficacy of their interventions limited by the extent to which patients are willing and able to adhere to their recommendations.

This entry summarizes the extent and significance of non-adherence. It also describes the different ways in which adherence is assessed, focusing on the strengths and weaknesses of each method. Finally, it reviews research on the prediction of adherence using information about patients, providers, and regimens, and discusses intervention strategies used to promote adherence.

Extent and Implications of Nonadherence

The problem of nonadherence has been well documented for many years. Though prevalence estimates vary depending on the type of behavior, the length and complexity of the regimen, and the assessment method employed, estimates indicate that overall, patients follow provider recommendations only about half of the time. In the first comprehensive review of the literature in 1976, David Sackett and Brian Haynes reported that adherence to chronic medication regimens averaged about 54%. They found that adherence to short-term regimens was highly variable, but tended to decrease sharply with time. They also reported prevalence of nonadherence for other health behaviors, finding, for example, that attendance at scheduled appointments was only about 47% for asymptomatic patients, but that it jumped to 81% for patients actively seeking care for a medical condition. More recent work has generally supported these results, suggesting that nonadherence is frequent enough to seriously limit the efficacy of therapeutic interventions.

The overall impact of poor adherence is difficult to estimate. Not only may it render treatments ineffective, making more intensive intervention necessary, but inconsistent medication adherence has been identified as a key factor in the development of new, treatment-resistant strains of some infectious diseases. A review by Irina Cleemput and her colleagues in 2002 reported on 18 economic impact studies with widely variable results, pointing out that differences in the definition of adherence and inconsistencies in assessment methods make estimates problematic. However, from lost productivity to increased hospitalization and premature death, there is general agreement that the cost is in the billions of dollars each year, and estimates have ranged to more than 100 billion dollars per year in costs associated with medication nonadherence in the United States alone.

Assessment of Patient Adherence

Adherence has been measured, both directly and indirectly, using a multitude of methods for both clinical and research purposes. Perhaps the most simple and widely used method of assessment involves asking patients to report their own behavior. Self-report strategies can vary from single questions asked by clinicians to gauge their patients’ adherence to detailed retrospective accounts of adherence over periods of several days or weeks. Patients may be asked to monitor their own behavior as it occurs for a period of time, and keep a written record that may be examined by a clinician or researcher. Although most adherence assessment methods are targeted at medication usage, an advantage of self-report methods is that they are flexible enough to use with any type of behavior, including exercise, diet, and substance use. They are also inexpensive to implement, and are unlikely to underestimate actual adherence behavior. However, they are plagued by potential reporting biases. Patients may say what they think the provider wants to hear in an attempt to form a positive impression, or they may have inaccurate memory for their behavior, especially if they are not asked in advance to keep a record. Finally, if patients are informed in advance that they will be asked to report on their adherence, that knowledge alone may cause a difference in the extent to which they adhere.

Noting the difficulties inherent in self-report measures, some researchers have asked health care professionals to provide estimates of their patients’ adherence, a method with clear and immediate clinical implications. However, research has consistently demonstrated that providers perform poorly when estimating patient behavior, tending to overestimate adherence. This tendency does not appear to decrease with experience or training. In fact, as early as 1966, Milton Davis found that senior physicians tend to overestimate adherence more dramatically than their junior counterparts. A study by John Steiner and his colleagues determined that even nurse practitioners, who may interact extensively with their patients, have difficulty accurately estimating patient behavior.

In an attempt to counter some of the problems associated with such subjective report measures, some researchers have turned to counting pills and weighing or measuring liquid medications to estimate adherence to medication regimens. In each of these strategies, patients are asked to bring their medications with them to medical appointments, and the difference between the amount of medication dispensed and the amount remaining is compared to the amount that the patients were instructed to ingest. Because patients may not always bring their medications with them for appointments, some providers use a variation on the pill count method, involving the examination of pharmacy refill records. These methods, while objective and reliable, are limited in that they do not provide information about how or when medication is taken. In fact, if patients are aware that their medication usage is to be assessed, they may simply dump large quantities of pills prior to an appointment to convey the appearance of adherent behavior.

In recent years, the evolution of microelectronics has permitted the development of electronic monitoring devices that may be used to record events related to adherence. For instance, Joyce Cramer and her colleagues were among the first researchers to use specially designed medication bottle caps to record the date and time on each occasion the bottles were opened. The advantages of such a method are clear: The results are objective and easily retrieved, and a single device can record thousands of medication-related events. However, there are significant logistical obstacles to using monitoring devices: the devices can be quite expensive, they may be impractical for use with multiple-drug regimens or large pills that must be taken frequently, there are often legal limitations associated with repackaging of prescription medications, and patients may neglect to return their medication bottles for monitoring. In addition, data may be misleading if patients leave a bottle open between doses, open the bottle without ingesting medication, or transfer their pills to pill boxes or organizers. Similar monitoring devices that are somewhat easier to apply are those that record the amount of time patients use respiratory therapy equipment for such conditions as sleep apnea or asthma.

Finally, biochemical assays or other laboratory tests may be used to estimate adherence. These tests may measure either the actual amount of a drug or drug tracer in a patient’s blood or urine, or the outcome of a treatment, assuming a close link between adherence and outcome. Biochemical markers can be helpful, objective indices of adherence to the extent that they are available and affordable. However, some drugs are broken down and cleared from the body rapidly, causing unreliability in such tests. Aside from funded research studies and applications of medicines with high potential toxicity, the cost of these assays may preclude their use as a part of regular clinical care. Clinical outcome measures are often only indirectly related to adherence. For example, hemoglobin Ai C is frequently used as an index of adherence in diabetic patients, reflecting glycemic control over extended periods of time. Whereas its association with adherence to multiple regimen components (e.g., insulin injection, dietary restriction, exercise) may make it an attractive measure, glycemic control is also affected by other factors, such as psychological stress. If factors beyond the patients control have a significant impact on the measure, or if a regimen is not well tailored to a patient or has limited efficacy, adherence may have very little association with clinical outcome.

There is no single gold standard in the measurement of patient adherence. Whereas the selection of an assessment method might depend on its intended purpose, the strengths and weaknesses of different methods illustrate the importance of obtaining multiple indices of adherence in research, and even clinical, settings. The factors that cause the results of different measures to vary from one another are not well understood. Further research may more clearly document in specific populations why self-report and laboratory measures are so often poorly correlated, or why providers’ estimates differ from more objective behavioral indices.

Prediction of Patient Adherence

By far, the area of adherence prediction that has received the least attention is the study of characteristics of the health care provider. Not only is such research logistically difficult to accomplish, but, since Davis’s early study, the literature has shown that physicians are more likely to attribute nonadherence to patient factors rather than to factors within their own control. Those studies that have examined provider characteristics have shown significant effects of the use of clear communication and the provision of support by the provider. For example, a meta-analysis of process studies by Judith Hall and her colleagues found greater adherence among patients when providers provided more information, asked questions specifically about adherence, and engaged in positive communication, avoiding negative conversational content.

Several characteristics of the medical regimen itself have been associated with adherence. Chief among these is the complexity of the regimen. For instance, Cramer and her colleagues found that as the number of pills patients were asked to take each day increased from one to four, adherence dropped sharply, from 87% to 39%. In addition to regimen complexity, the length of the regimen can negatively impact adherence. Adherence tends to erode with time not only for acute regimens, as noted earlier, but also for longer-term regimens. Thus, adherence to lifestyle change recommendations is especially challenging for patients.

A great deal of research has examined individual differences among patients and how they are related to adherence, with mixed results. Studies of demographic characteristics, such as age, gender, and race have generally shown little direct and consistent association with adherence. Perhaps surprisingly, there has also been little evidence for a link between patient knowledge and adherence. Although knowledge of an illness and its treatment is clearly necessary for adherence to occur, it is usually not sufficient to induce change in behavior. Some personality constructs have been found to predict adherence to a modest extent. In their extensive review of the early literature, Sackett and Haynes reported that such characteristics as cooperativeness, high frustration tolerance, a futuristic orientation, low authoritarianism, and low demandingness were associated with good adherence. Other researchers have found similar effects for variables such as conscientiousness and low neuroticism. However, most studies attempting to link personality characteristics to adherence have had little or inconsistent success.

The difficulty in reliable prediction of adherence using any single variable, or even combinations of variables related to the patient, the provider, or the treatment context, has caused some researchers to look to the interaction of these variables. A review by John Wiebe and Alan Christensen cited a number of studies that have been successful in predicting adherence by considering the interaction of patient factors and characteristics of the treatment context. For instance, patients who tend to become highly involved in their care tend to adhere and adapt best in a disease or treatment context that allows greater patient control. However, if the disease or treatment context offers the patient little control, a less active coping style may be associated with better adherence. Although there is relatively little research on such interaction, it remains a promising area for further work.

Intervention to Increase Adherence

Interventions to simplify regimens have often been quite successful in increasing patient adherence. Data showing increased adherence with reduced dosing frequency has led to the marketing of sustained-release drugs with longer half-lives that may be taken less frequently, from antibiotics to antiretroviral medications for HIV infection, as reported by the Body Health Resources Corporation. Other work has pursued more convenient treatment modalities. For example, James Burris and his colleagues showed that patient adherence with a transdermal patch used to treat hypertension was nearly double the rate for even a sustained-release oral medication taken once per day.

Interventions directed toward the health care system have seldom been conducted in isolation. However, when combined with individual-level strategies in comprehensive disease management programs, they have had some success, especially with chronic illnesses. For instance, Neil Grey and his colleagues reported on a multidisciplinary hospital-based disease management program for diabetic patients. They constructed a team including a primary care physician, an advanced practice nurse, and a registered dietician, focused heavily on patient education, and scheduled frequent follow-up appointments. With this approach they observed a 22% improvement in glycemic control, sustained and increased over a 6-month follow-up period, which they attributed largely to increased adherence.

Overall, the majority of work has been targeted directly at patients. The approaches used with patients have been primarily either educational or behavioral in focus. Patient education programs, whether delivered individually, in groups, or via lectures, have had mixed results in targeting nonadherent behavior. As Donald Meichenbaum and Dennis Turk pointed out in their review of the subject, education tends to be most effective when it involves explicit recommendations about how to implement newly acquired knowledge into the treatment regimen, when it avoids unnecessary jargon and reiterates key points, and when it does not make excessive use of fear appeals to change behavior. However, even when education is appropriately delivered, a multitude of barriers may prevent the increase in knowledge from being translated into behavior change. For example, Sackett and Haynes found that only 8 of the 14 educational-only interventions they studied led to significant increases in adherence.

Behavioral interventions have had slightly better success. Although they may include educational aspects, these programs have focused on such techniques as providing reminder cues and organization strategies, teaching patients to monitor their own behavior and set adherence goals, and delivering feedback and reinforcement for adherent behavior. In comparing the results of educational and behavioral intervention studies, a review by Leonard Epstein and Patricia Cluss found that behavioral approaches had a greater effect on adherence, especially when they included feedback and reinforcement components. However, because few studies have been conducted with long-term follow-up assessments, it remains unclear whether the effects of these behavior modification techniques endure after the intervention is terminated. Some researchers have attempted to maximize the long-term efficacy of behavioral intervention efforts by teaching patients to monitor, evaluate, and reinforce themselves for adherent behavior, rather than having a therapist administer reinforcement. For example, Christensen and his colleagues reported a controlled study of a self-management intervention with hemodialysis patients in which effects on adherence were actually greater at an 8-week follow-up assessment than at the conclusion of therapy. However, this intervention, like most behavioral strategies, included not only the use of standard behavior modification principles, but also education regarding the effects of nonadherence. In fact, research has generally supported the use of integrated educational and behavioral programs as most effective in increasing adherence.

Perhaps the most extreme form of intervention to promote adherence is directly observed therapy (DOT). In this approach, patients take their medications in the presence of a health worker. Generally reserved for communicable diseases that pose a significant public health threat, DOT has been most frequently applied in the treatment of tuberculosis, where the treatment regimen is of limited duration. The strategy is quite cost effective when compared to inpatient hospitalization, but its clinical efficacy remains unclear. A review by Jimmy Volmink and his colleagues in 2000 found significant improvements in adherence associated with the use of DOT, but pointed out that the strategy is applied inconsistently across studies and often combined with other interventions, making it difficult to isolate its effect.

Conclusions

Patient adherence constitutes a vital link between provider recommendations and health outcomes, and it is becoming more important in modern health care as patients assume greater self-care responsibilities. However, nonadherence is commonplace, with significant public health and economic implications. Multiple measures of adherence are available, each with its own limitations. Results using different measures are not necessarily highly correlated with one another, and accurate assessment can be a challenging and expensive task. Although there has been a tendency on the part of health care providers to attribute nonadherence to patient characteristics, the literature has not supported this assertion well. Researchers are beginning to examine the interaction between patient characteristics and the treatment context in an attempt to better predict adherence. In efforts to improve adherence, progress has been made toward the simplification of treatment regimens and the development of systemic approaches to disease state management. Behavioral interventions directed toward increasing patient adherence have shown slightly better results than patient education in isolation, but it is likely that integrated approaches are most effective. Finally, in cases where nonadherence may have extreme consequences, directly observed therapy may be instrumental in achieving desired clinical outcomes.