Treatment adherence is essential for the optimal control of cardiac risk factors.
Poor adherence, with subsequent adverse cardiac outcomes, is evident in hypertension, hyperlipidemia, diabetes, smoking cessation, and weight loss.
Elements associated with poor adherence include social and economic factors, health care system and team-related factors, condition-related factors, therapy-related factors, and patient-related factors.
Medication simplification and team-based approaches to identification and treatment of cardiac risk factors have shown promise in improving treatment adherence.
Future approaches to improvement of treatment adherence may include moving to a home-based model of care, activating patients to participate in their care, and employing information technology to identify nonadherence and to initiate interventions.
Drugs don’t work in patients who don’t take them.
C. Everett Koop
For patients with coronary artery disease (CAD) or at risk for CAD, optimal control of cardiac risk factors, such as hypertension, hyperlipidemia, and smoking, is one of the most effective methods to reduce subsequent cardiac morbidity and mortality. For this control to be achieved, a variety of therapeutic options, both pharmacologic and nonpharmacologic, are available to patients and their health care providers. However, none of these therapeutic options can provide optimal cardiac risk factor control without diligent adherence to their use. Indeed, treatment adherence serves as the “final common pathway” for risk factor treatment and has been characterized as the “key mediator between medical practice and patient outcomes.” Despite this essential and somewhat obvious role in risk factor optimization, treatment adherence has been a relatively neglected area of research because of difficulties in both characterizing the factors that drive adherence and designing effective interventions for improvement. Thus, a critical step in optimizing cardiac risk factors in patients with CAD is understanding and improving treatment adherence.
Adherence to treatments of cardiac risk factors is difficult for a variety of reasons. Many risk factors, such as hypertension and hyperlipidemia, are chronic, life-long conditions that require daily administration of treatments, including both pharmacologic agents and lifestyle choices. Furthermore, these conditions are largely asymptomatic. Thus, an important motivator for treatment—relief from bothersome symptoms—is largely absent in the management of cardiac risk factors. In fact, the treatments themselves may cause symptoms via side effects, thus adding another disincentive for adherence among affected patients. Finally, adherence is also influenced by a variety of patient, provider, and health care system characteristics.
This chapter reviews the current definitions of treatment adherence, methods of its measurement, and current gaps in the treatment of cardiac risk factors. Although both lifestyle and pharmacologic treatments are important for optimal risk factor control, the bulk of the research has focused on adherence patterns to pharmacologic therapies and thus comprises the majority of the review. Barriers to treatment adherence, including those at the patient, provider, and health care system levels, are reviewed. Interventions to improve adherence are also reviewed for both their relative effectiveness and potential dissemination. Finally, future directions for adherence research and interventions are presented.
Treatment adherence is defined as the extent to which patients take treatments (medications or lifestyle modifications) as prescribed by their health care providers. A variety of terms have been used to describe treatment adherence in the past, such as compliance, concordance, and fidelity, but these terms have been largely abandoned because of the pejorative nature underlying their use. Adherence is usually measured on a continuous scale as a percentage calculated by the number of treatments actually taken relative to the number of treatments prescribed. For example, a patient who is prescribed 30 pills of an antihypertensive medication and actually takes only 20 pills is 67% adherent to the prescribed regimen. The adherence scale can range from 0% to >100% (because some patients can take more than their prescribed treatments). Truly perfect adherence, with fidelity to precise timing of each dose of treatment, is rare, occurring in only one sixth of patients in one survey. Furthermore, many of the treatments for cardiac risk factors (e.g., statin therapy for hyperlipidemia) may not require such precision for optimal control to be achieved. Accordingly, most studies on adherence have used a dichotomous cutoff of ≥80% to define good versus poor treatment adherence.
A related but separate concept to treatment adherence is treatment persistence. Persistence is the duration of time that a patient remains on treatment from the initiation to discontinuation of a prescribed therapy. Unlike adherence, persistence patterns not only can describe patient adherence actions but also can identify other reasons for therapy discontinuation, such as intolerance of therapy due to side effects or provider-initiated discontinuation.
Treatment adherence and persistence can be measured in a variety of ways. In general, measurement can be direct or indirect, with tradeoffs made between precision and ease of measurement for each type of method. Direct methods of adherence measurement include directly observed therapy and measurement of the blood concentration of a medication or its metabolite. Both of these techniques, although accurate, require substantial time and expense and are thus rarely employed. Indirect methods of adherence measurement include patient self-report of treatment adherence, pill counting, assessment of medication refill rates, assessment of a clinical response to a prescribed therapy, and medication electronic monitoring systems (MEMS) that track the frequency and timing of opening of the pill containers. In general, self-report and pill counts are not considered reliable methods of assessment, in part because of the social desirability bias that occurs when patients exaggerate their behavior in an effort to please their health care providers. This bias also affects the ascertainment of patients’ adherence to prescribed lifestyle modifications. On the other hand, medication refill rate assessment is a reliable method of adherence assessment, especially in closed health care systems in which patients receive their medications from a single pharmacy source. MEMS assessments are also considered reliable, although their use is expensive and logistically complex. Finally, use of a combination of adherence measurement methods can also be a valuable and reliable way to determine adherence patterns.
Patterns of Nonadherence and Associated Outcomes
Research into adherence patterns for cardiac risk factor treatments has uncovered significant gaps. As previously noted, adherence is generally worse among those patients who have chronic asymptomatic conditions, such as hypertension and hyperlipidemia, than among those with acute symptomatic illnesses. Even among those patients participating in clinical trials, which are optimized for medication provision and follow-up and generally have high adherence rates, adherence rates for chronic conditions average only 43% to 78%. In addition, several studies have documented that both adherence to and persistence with chronic treatments drop precipitously after the first 6 months of therapy.
Most deviations from adherent behavior are omissions of or delays in medication administration, rather than outright discontinuation of treatments. Urquhart and colleagues characterized specific patterns of medication-taking behavior, using MEMS, among patients initiated on chronic preventive therapies. They found that patients exhibited six types of medication-taking behavior, with roughly even allocation of patients to each group. One sixth of patients exhibited nearly perfect medication-taking behavior, taking almost all of their doses at their prescribed frequency. One sixth took all of the prescribed doses but had some timing irregularity. One sixth had an occasional missed dose, along with some timing irregularity. One sixth of patients took “drug holidays,” or several consecutive missed doses, three or four times a year. One sixth took drug holidays monthly or more often, and the final group of patients took few or no doses, although they would often report good medication adherence. Consistent with this desire by patients to exhibit good adherence behavior, Feinstein and colleagues demonstrated that many patients improve adherence in the 5 days preceding a visit to the physician, a phenomenon dubbed “white coat adherence” by the investigators.
Predictably, these gaps in treatment adherence lead to worse health outcomes and increased health care costs. A variety of studies have demonstrated that patients with poor adherence have higher all-cause mortality and cardiovascular event rates compared with those with good adherence. In addition, two studies have demonstrated that among medication-related hospitalizations in the United States, between 39% and 69% are due to nonadherence. Specific studies examining patients with chronic medical conditions such as hypertension and hyperlipidemia have found associations between nonadherence and worse treatment outcomes, higher hospitalization rates, and increased health care costs. McCombs and colleagues demonstrated that patients with hypertension who interrupted or terminated their blood pressure treatment accumulated an additional $873 in health care costs (primarily hospitalization related) during the following year. This finding was supported by an analysis of the U.K.-based MediPlus data base that demonstrated an association between the discontinuation of antihypertensive treatments and increased hospital and general physician costs.
Nonadherence in Hypertension
Hypertension is a principal risk factor for the development of both CAD and cerebrovascular disease and a major contributor to worldwide cardiovascular mortality. Uncontrolled hypertension among CAD patients is associated with recurrent cardiovascular events, including death, myocardial infarction, and stroke. Fortunately, a wide variety of treatments exist to achieve blood pressure control and to minimize the risk of these adverse events. Currently, more than 10 classes of antihypertension medications are available, and achieving a sustained decrease in blood pressure of 12 mm Hg with one or more of these treatments can prevent one death for every 11 patients treated.
Despite this availability of treatment and evidence of its benefit, blood pressure control for a large number of hypertensive patients remains suboptimal, and nonadherence to treatment is a major factor. Several studies have demonstrated that less than 50% of CAD patients with hypertension have their blood pressure at recommended levels. Nonadherence is a major contributor to this lack of control and, in one study, appeared to contribute to uncontrolled blood pressure more than the lack of adequate prescription of hypertension treatments. Furthermore, patients with greater than 80% adherence rates demonstrated improved blood pressure control relative to those patients with less than 50% adherence rates. Treatment adherence to hypertension medications is especially problematic given the asymptomatic nature of the condition, the common occurrence of side effects associated with hypertension treatments, and the daily commitment required for proper medication adherence. These factors all lead to difficulty with adherence. Haynes and colleagues illustrated these difficulties by finding that more than 50% of newly diagnosed hypertensive patients with poor blood pressure control had problems with treatment adherence.
Nonadherence to hypertension treatments with its accompanying uncontrolled blood pressure leads to adverse outcomes. Maronde and colleagues found that Medicare patients who were rehospitalized had significantly higher rates of antihypertension medication nonadherence than a comparable group of nonhospitalized Medicare patients. Similarly, Psaty and colleagues found that patients who were recently nonadherent to beta blocker therapy for their hypertension demonstrated a 4.5-fold increased risk for coronary heart disease complications.
Nonadherence in Hyperlipidemia
Among CAD patients, lipid-lowering therapies, especially statin medications, can significantly decrease the risk of recurrent cardiac events and mortality. For example, the Heart Protection Study evaluated the effects of simvastatin among CAD, diabetic, or treated hypertensive patients and found an 18% reduction in cardiac-related deaths and a 24% reduction in major cardiovascular events. Similarly, the PROVE-IT TIMI 22 trial compared differing intensities of statin therapies (pravastatin versus atorvastatin) among acute coronary syndrome patients. After 2 years, the atorvastatin patients had lower low-density lipoprotein levels and a 16% reduction in the primary endpoint (death from any cause, myocardial infarction, unstable angina requiring rehospitalization, revascularization, or stroke). Meta-analyses of statin therapy echo these benefits. One involving approximately 25,000 subjects from 34 trials found that 4 years of cholesterol-lowering therapy would prevent one death, one coronary heart disease death, and one cardiovascular death for every 110, 96, and 117 patients treated, respectively. In addition, a survival benefit was seen in studies in which more than 50% of patients were myocardial infarction survivors, had a total cholesterol reduction of more than 10%, and were treated for at least 4 years. In addition, statins produced a greater reduction in the odds of death than other lipid-lowering therapies.
Despite these clear benefits of statin use, nonadherence is a significant problem. Benner and colleagues found that more than half of patients started on statin medications will discontinue the medication within 6 months. Jackevicius and colleagues noted a similar gap in a study of statin adherence patterns among elderly patients. They found that the 2-year rate of statin adherence among this cohort was less than 50%. Furthermore, they demonstrated differential rates of adherence based on the presence of known CAD ( Fig. 35-1 ). Although patients who had experienced acute coronary syndrome had higher 2-year adherence rates than those with chronic CAD or receiving statins for primary prevention (40% for acute coronary syndrome, 36% for chronic CAD, and 25% for primary prevention), all patient groups demonstrated significant adherence gaps.
Nonadherence to statins is significantly associated with adverse outcomes. Wei and colleagues found that post–myocardial infarction patients who had at least 80% adherence rates to statin therapy experienced relative risk reductions of 81% for recurrent myocardial infarction and 53% for all-cause mortality, but those patients who had adherence rates <80% had no significant risk reduction in either recurrent cardiac events or mortality. The Lescol Intervention Prevention Study supported these findings by demonstrating that post–myocardial infarction patients who discontinued their statin medications experienced a twofold increase in major adverse cardiovascular events. These benefits of adherence appear to be medication specific, as demonstrated in an elegant study by Rasmussen and colleagues. They found that adherence to statins exhibited a “dose-response” relationship with mortality, with higher adherence corresponding to decreased mortality. A similar relationship between calcium channel blockers and mortality among the same cohort was not seen, suggesting that the adherence-mortality relationship was statin specific.
Nonadherence in Diabetes
Diabetes is another major cardiac risk factor among CAD patients. Diabetic patients are at higher risk for development of CAD and dying of it. In addition, concurrent cardiac risk factors (e.g., hypertension, smoking) are more prevalent among diabetic patients and appear to exert worse effects in the setting of poor glycemic control. Diabetes is a complex disorder that requires constant attention to diet, exercise, glucose monitoring, and medication to achieve good glycemic control. Glycemic control in diabetes plays an important role in reducing microvascular disease, such as nephropathy, retinopathy, and neuropathy. In addition, a 17-year study of type 1 diabetics indicated that tight glycemic control reduces the risk of cardiac events by 42%. In contrast, tight glycemic control among type 2 diabetics has not demonstrated conclusive benefits for reduction of cardiac events and may be associated with harm in some cases. Despite this uncertainty, multifactorial cardiac risk factor control, including glycemic control, is an important element of effective prevention of cardiac events. For example, the Steno-1 trial demonstrated the benefit in achieving multiple risk factor control among diabetics. Patients were randomized to usual care or an intensive treatment regimen that included both behavioral and pharmacologic treatments targeting hypertension, hyperglycemia, dyslipidemia, microalbuminuria, and secondary prevention of CAD events with aspirin. After 7.8 years of follow-up, patients receiving the intensive treatment experienced 53% reduction in cardiovascular disease. Furthermore, these benefits were sustained, with the patients receiving intensive treatment experiencing a 57% reduction in cardiovascular deaths during the following 4 years after the cessation of the trial.
As with the other cardiac risk factors requiring prolonged chronic treatments, treatment adherence is a substantial problem among diabetics. Among patients receiving oral hypoglycemic therapies, adherence rates range widely from 36% to 93%. Boccuzzi and colleagues profiled adherence rates among patients taking oral antidiabetic medications and found that 12 months after initiation of these therapies, adherence to metformin was 60%; to sulfonylureas, 56%; to repaglinide, 48%; and to α-glucosidase, 31%. Adherence rates with insulin therapy were not much better; one study demonstrated that only 63% of insulin doses were taken as prescribed.
Nonadherence to diabetic treatments is further complicated by the common occurrence of other cardiac risk factors and comorbidities, many of which require multiple additional medications. Piette and colleagues demonstrated that 50% of U.S. diabetic patients were receiving at least seven medications, including at least two glucose-lowering medications. Adherence rates are adversely affected by this increased number of medications. Rubin found that rates of adherence to polytherapy versus monotherapy are 10% to 20% lower among diabetic patients. In addition, treatments that require multiple administrations during the day also adversely affect adherence. Paes and colleagues found that the adherence rates were 79% for once-daily dosing of medications, 66% for twice-daily dosing, and 38% for three-times-daily dosing.
Predictably, nonadherence to antidiabetic medications is linked to adverse outcomes. Pladevall and colleagues illustrated that 10% increases in nonadherence to antidiabetic treatments were associated with 0.14% increases in hemoglobin A1c. Lau and Nau demonstrated that patients who were nonadherent to their diabetic regimen were more likely to be hospitalized. In contrast, increased medication adherence to antidiabetic medications (at a threshold rate of 60%+) resulted in decreased medical care costs, although overall costs were not reduced because medication costs offset these savings. Similarly, Balkrishnan and colleagues demonstrated that 10% increases in medication possession ratios were associated with 8.6% to 28.9% decreases in annual health care costs.
Nonadherence in Smoking
Smoking is a major cardiac risk factor and remains a significant public health issue. In 2007, 19.8% of U.S. adults were active smokers. Smoking increases all-cause and cardiovascular mortality, and among patients with CAD, it increases the risk of reinfarction and mortality. Quitting smoking, in turn, reduces all-cause mortality by 36% in CAD patients, even in long-term smokers. In addition, there is some evidence that public health policies such as smoking bans can have salutary effects. A small study in Colorado demonstrated a 27% reduction in myocardial infarction hospitalizations after institution of a smoking ban. A similar study in Scotland demonstrated a 14% to 21% reduction in acute coronary syndrome admissions during the 10 months after a smoking ban.
Quitting smoking is a difficult process, with high recidivism rates. For example, in a study of long-term abstinence rates among smokers who tried to quit without any behavioral or pharmacologic assistance, only 5% to 7% were abstinent at 1 year. Accordingly, a variety of both behavioral and pharmacologic interventions have been developed to assist with smoking cessation. Behavioral interventions include in-person counseling, remote (phone, Web) counseling, group counseling, and financial incentives. Pharmacologic treatments for quitting smoking include nicotine replacement therapies in a variety of formulations (e.g., patch, gum, lozenge, inhaler, nasal spray), bupropion, and varenicline.
Despite the wide variety of treatments, the rates of adherence to these programs, measured as abstinence from smoking after completion of the treatment, remain remarkably low. For example, quit rates for in-person clinical counseling are 12.3% at 3 months and 6.5% at 12 months ; for group programs, 20% at 1 year ; and for remote counseling, 10% at 1 year. Financial incentive programs resulted in quit rates of 14.7% at 9 to 12 months and 9.4% at 15 to 18 months. Quit rates with pharmacologic therapies are better than with behavioral therapies, but the absolute rates of adherence and smoking abstinence remain low. Among nicotine replacement therapies, the highest adherence rates were with the patch, but 12-week abstinence rates were similar for all nicotine replacement modalities (gum, 20%; patch, 21%; spray, 24%; inhaler, 24%). Bupropion resulted in quit rates of 44% (compared with 19% with placebo) at 7 weeks and 23% (compared with 12%) at 1 year. Longer term bupropion (administered for 52 weeks versus 7 weeks) had better short-term abstinence rates (47% versus 37%) at 16 weeks after cessation of therapy but similar rates of abstinence to short-term bupropion at 2 years (41% versus 40%). Varenicline demonstrated abstinence rates of 44% 4 weeks after medication cessation, compared with 30% with bupropion and 18% with placebo. At 1 year, abstinence rates were 23% with varenicline, 16% with bupropion, and 9% with placebo.
Nonadherence in Obesity
Obesity is a growing problem in the United States and worldwide and serves as a risk factor for cardiac disease. The NHANES survey from 1988 to 1991 demonstrated that 36% of surveyed patients were obese. Obesity is associated with coronary disease; Bogers and colleagues showed a 29% increase in CAD with each 5-unit increase in body mass index. Although the direct cardiac risk due to obesity is confounded by the common presence of concurrent cardiac risk factors, obesity appears to be an independent risk factor for CAD, probably by contributing to insulin resistance, hypertension, lipid abnormalities, left ventricular hypertrophy, endothelial dysfunction, and obstructive sleep apnea. Most if not all of these risk factors can be better controlled or even eliminated with weight loss. Although no randomized controlled clinical trials have linked weight loss to decreased mortality or CAD, a large observational study found that overweight women who lost more than 20 pounds had a 25% decrease in mortality, CAD, and cancer mortality. Among the subset of women with CAD or heart failure, any weight loss was associated with a 10% reduction in CAD and a 20% reduction in all-cause mortality.
Similar to smoking cessation, weight loss is difficult to achieve. A variety of weight loss strategies exist and include behavioral modification, dietary therapy, exercise, drug therapy, liposuction, and bariatric surgery. Despite these options, though, adherence to weight loss programs, measured by achievement and maintenance of weight loss, is low.
Behavioral modification strategies include self-monitoring, control of the stimuli that activate eating, slowing down eating, goal setting, behavioral contracting and reinforcement, nutrition education, modification of physical activity, social support, and cognitive restructuring. These techniques are usually paired with dietary weight loss programs, and the combination has been demonstrated to increase weight loss by 7.7 kg at 12 months. In addition, ongoing behavioral modification strategies may be useful in maintaining weight loss. Wing and colleagues showed that those patients who achieved a mean weight loss of 19 kg in the prior year had a lower amount of weight re-gain with in-person support (face-to-face support resulted in weight gain of 2.9 kg; Internet-based support, 4.7 kg; and newsletter only, 4.9 kg).
Commercial diet programs, which rely primarily on calorie restriction, have demonstrated efficacy in weight loss. Both Weight Watchers and the Jenny Craig programs have been studied. Weight Watchers resulted in the loss of 5.3% of baseline weight compared with 1.5% in the placebo group at 1 year. At 2 years, both groups had regained weight, but the Weight Watchers participants gained back less. Similarly, Jenny Craig participants lost 7.1% of their baseline weight versus 0.7% in the placebo group at 1 year. In addition to overall calorie restriction, a variety of diets with differing macronutrient composition have been promoted, but data have been conflicting about the superiority of one over the other. For example, a comparison of the Atkins, Ornish, Weight Watchers, and Zone diets showed similar modest reductions in weight and cardiac risk factors at 1 year. Importantly, all diets had adherence rates of only 50% to 65%, and adherence, more than macronutrient composition of the diets, had a significant correlation with weight loss.
Pharmacologic therapies for weight loss include sympathomimetic drugs (e.g., sibutramine), drugs that alter fat digestion (e.g., orlistat), antidepressants, antiepileptics, and diabetes drugs (e.g., metformin). Many of these have shown modest efficacy for weight loss over placebo. A meta-analysis of sibutramine demonstrated additional mean weight loss of 4.5 kg over placebo. Similarly, a meta-analysis of orlistat showed an additional mean weight loss of 2.89 kg. As with all medications, adherence is essential for maximum efficacy of therapy, and these weight loss medications require at least 1 year if not 2 years of therapy. Tellingly, only 67% of the participants in the Kelley orlistat trial completed therapy.
A final, invasive option for weight loss is bariatric surgery. The procedure is generally successful; a meta-analysis demonstrated that mean weight loss is 61% of baseline weight. In addition, cardiac risk factors such as hypertension and diabetes were greatly improved in the majority of patients. Despite these encouraging results, the surgery has 30-day mortalities ranging from 0.1% to 1%. In addition, the salutatory effects on weight loss and cardiac risk factor control are primarily seen only in morbidly obese patients, making this therapy less attractive for less obese patients.
Factors Associated with Nonadherence
Treatment adherence is a multidimensional construct that is determined by the interplay of a variety of factors. However, many providers believe that patients are solely responsible for their adherence behavior and thus fail to account for other factors outside a patient’s locus of control that can facilitate or impede the ability to adhere to a prescribed treatment (i.e., therapy or system factors). Understanding of these factors and their impact on adherence behavior is essential to the proper design of interventions to improve adherence rates.
The 2003 World Health Organization report Adherence to Long-Term Therapies: Evidence for Action organized the factors affecting adherence into five dimensions: social/economic factors, health care system and team-related factors, condition-related factors, therapy-related factors, and patient-related factors ( Fig. 35-2 ). The bulk of prior research on adherence factors has focused primarily on patient-related factors, but there is increasing recognition, with accompanying research, that factors in the other four dimensions are equally important in characterizing and improving adherence.