| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
SPECIAL ARTICLE |
From the Department of Psychology, University of Iowa, Iowa City, IA.
Address reprint requests to: Alan J. Christensen, Department of Psychology, E11 Seashore Hall, University of Iowa, Iowa City, IA 52242. Email: Alan-Christensen{at}uiowa.edu
ABSTRACT
OBJECTIVE: Previous reviews have concluded that there is little or no evidence supporting a predictable association between patient characteristics and regimen adherence in chronic illness. The primary objective of this article is to propose an alternative conceptual framework for the interpretation and design of adherence research.
METHODS: The author’s research involving adherence among patients with end-stage renal disease is reviewed and used to illustrate the patient-by-context interactive framework.
RESULTS: Adherence is most favorable when the patient’s characteristic or preferred style of coping with illness and treatment-related experiences is congruent with the contextual features or demands of the particular type of medical intervention the patient is undergoing. Among patients with end-stage renal disease, a more vigilant or active style of coping is associated with more favorable adherence only for patients undergoing home-based dialysis treatment modalities that are highly patient directed. Among patients receiving hospital-based, provider-controlled treatment, a less vigilant or more passive coping style is associated with more favorable dialysis regimen adherence.
CONCLUSIONS: The patient-by-context interactive perspective can provide a useful framework for the interpretation and design of adherence research.
Key Words: patient adherence • coping style • renal dialysis • chronic disease
Abbreviations: CAPD = continuous ambulatory peritoneal dialysis; ESRD =end-stage renal disease; IWG = interdialysis weight gain.
"The desire to take medicine is one feature which distinguishes man, the animal, from his fellow creatures." —Sir William Osler, 1894
The sentiments of Sir William Osler notwithstanding, patients’ apparent unwillingness to follow medication regimens or to adhere to medical treatment regimens more generally reflects a significant barrier to the effective management of most chronic medical disorders (1). The prevalence of nonadherence is estimated to be 50% or higher among patients following chronic medical regimens (1). Given the high rates of nonadherence, it is not surprising that researchers have devoted considerable attention to the identification of patient characteristics that are significant determinants of adherence. Despite the large literature devoted to the topic, several of the most comprehensive reviews published over the last three decades have concluded that there is little evidence supporting a predictable or consistent association between patient characteristics and adherence behavior in any chronic disease population (2–4).
Our research team has recently argued that the inconsistent findings of past adherence research may be due to a failure to consider the interaction of patient attributes with characteristics of the chronic disease and treatment context (5–7). Most studies of determinants of patient adherence have examined only the main or direct effects of psychological factors. These "main-effect studies" largely ignore the tremendous heterogeneity that exists among patients with different disease histories and undergoing different types of medical intervention.
Main-effect associations between patient individual differences and adherence behavior clearly lack consistency across studies, measures, and study populations (7). For example, some research involving the five-factor model of personality has suggested that higher levels of trait conscientiousness are associated with more favorable adherence to chronic medical regimens (8). However, other studies have reported that conscientiousness is either unrelated to adherence or that higher levels of this trait may actually be associated with poorer adherence among some patient subgroups (9, 10).
The lack of consistency in past adherence research might be interpreted as suggesting that patient characteristics do not play a predictable role in determining adherence behavior (2–4). Alternatively, however, this marked inconsistency might reflect the fact that these associations vary across features of the disease or treatment context (ie, the patient-by-context interactive perspective). That is, specific patient characteristics might be associated with better adherence for some subgroups of patients, whereas the association might be absent (or even opposite) among other subgroups.
Despite the limited success of the main-effect approach, few studies have considered the possibility that patient and contextual variables interact in predicting adherence. The fundamental components of the interactive framework we have adopted in much of our research are illustrated in Figure 1. The dashed lines reflect the fact that significant main effects between patient characteristics or contextual factors and adherence are generally not found. The central assumption in the model is that relevant contextual or situational features should be explicitly assessed, and the interaction of these factors with patient variables should be tested directly.
|
Within clinical psychology, the interactionalist perspective has been most commonly applied in the psychotherapy research literature (12). One of the more consistent patterns that has emerged involves the finding that individuals with more active and internally focused coping styles respond better to treatments that emphasize self-control rather than therapist control (13–15).
Although this general person-by-situation interaction perspective has a long history in personality and clinical psychology, it has received less attention in the study of adherence or adaptation in chronic disease. Modest evidence suggests that an interactive approach may be useful in predicting emotional outcomes in several chronic illness populations (16–19). For example, in one early study, we reported findings consistent with the person-by-context congruency hypothesis that adjustment should be better when a patient’s coping style or beliefs are consistent with contextual features of the patient’s disease or treatment history (16). In this study of 96 hemodialysis patients, the belief that one’s health is controllable was associated with less depression among subjects who had not previously experienced a failed renal transplant. However, this belief was associated with greater depression among subjects who had returned to dialysis after an unsuccessful transplant.
In a conceptually similar application of the interactive framework, Compas et al. (20) proposed a "goodness-of-fit" hypothesis to explain the interactive relation of patient coping behavior and perceived control with symptoms of emotional distress. In a study of 83 cancer patients, these authors reported that symptoms of anxiety and depression were lowest when patient coping responses (ie, problem focused vs. emotion focused) were consistent with control perceptions (19). Specifically, high problem-focused coping coupled with high perceived control resulted in the most favorable emotional adjustment.
CHRONIC ILLNESS AND TREATMENT CONTEXT
A challenge underlying the application of the patient-by-context interactive model is the identification of contextual factors that are most relevant to the psychological or behavioral processes under consideration. In other words, what aspects of chronic physical disorders and the associated medical treatments are likely moderators of patient adherence behavior? Fortunately, the broader literature involving characteristics of stressful encounters and moderators of the stress response provides a foundation for chronic illness researchers to draw from (eg, Refs. 21 and 22). For many of the situational moderators deemed important in the general stress literature, analogous characteristics can be identified in the chronic illness and treatment context. For example, just as non–health-related stressful encounters often differ in terms of predictability or ambiguity, medical disorders themselves differ in the terms of the predictability of symptom patterns or ambiguity of patient prognoses.
Our work has focused primarily on the contextual feature of stressor controllability. Stressor controllability is known to be an important moderator of the physiological, emotional, and behavioral consequences of stressful encounters (21, 22). One important way that medical treatment contexts differ in controllability involves the degree of control or direction patients have over treatment delivery. Many common medical interventions are highly provider directed and require little in the way of patient involvement or control (eg, most forms of oncologic treatment). Other treatments are clearly patient directed and require significant patient involvement to ensure effective treatment delivery (eg, the diabetic control regimen). The patient-by-context model leads one to consider the possibility that contextual factors, such as treatment controllability, moderate the association of some patient individual difference characteristics with adherence.
ESRD AND TREATMENT CONTEXT
Most of our own adherence research has focused on the treatment of ESRD. A primary reason for this focus is our belief that this disorder provides a unique opportunity to apply the patient-by-context interactive model. On the cessation of kidney function, the 200,000 ESRD patients in the United States must undergo life-long medical therapy to stay alive. Available treatments for ESRD include renal transplantation and several forms of renal dialysis. Given a shortage of donor organs and a significant transplant rejection rate, most ESRD patients rely on some form of renal dialysis as treatment for their condition (23). The primary dialysis modalities are hemodialysis ("center" or "home") and peritoneal dialysis.
There is an important difference in the role taken by the patient while undergoing the different forms of dialysis. Hemodialysis achieves removal of toxins and excess fluid via the extracorporeal circulation of blood through an artificial kidney or dialyzer. The typical center hemodialysis patient is a relatively passive recipient of treatment in the context of a dialysis center. The center hemodialysis procedure is performed three times per week by nurses or renal technicians in a hospital or clinic setting, requiring approximately 4 hours per session. Hemodialysis treatment commences by way of a vascular connection made between the dialysis machine and the patient, usually through an arteriovenous fistula permanently placed in the patient’s forearm. In most centers, little participation is allowed or required of the patient while undergoing the procedure. For a minority of patients, the hemodialysis procedure occurs at home. Although home and center hemodialysis are largely analogous from physiological and mechanical standpoints, home hemodialysis patients have the opportunity to be more actively involved in treatment delivery and direction. Home hemodialysis patients typically are responsible for placing their own arterial and venous needles at the beginning of the session, administering heparin and other necessary medications, and monitoring their blood pressure as well as various vital functions of the dialysis machine. Moreover, home dialysis patients have considerably less frequent contact with renal care providers and are able to set and maintain their own dialysis schedules. In contrast to center hemodialysis patients, home patients clearly play a more central and behaviorally involved role in delivery of dialysis treatment.
Peritoneal dialysis treatment requires the patient to take an even more active role to ensure treatment success. In CAPD, the most common form of peritoneal dialysis treatment, a permanent catheter is surgically implanted in the abdomen. CAPD is carried out manually by the patient. A sterile tube is used to carefully connect the catheter to a bag of sterile dialysis solution (dialysate). The bag is elevated to allow flow of the dialysate into the peritoneal cavity. After this procedure is completed, the bag is tucked away under the patient’s clothing. Over the next 4 to 6 hours (8 hours while sleeping), the patient remains ambulatory as continuous dialysis ensues. During this "dwell time," blood filters through the peritoneal membrane, leaving toxins and excess fluid behind in the dialysate. After the dwell time is complete, the bag is lowered and the used solution is allowed to drain back into the bag, which is discarded, and the procedure begins again. Throughout each step of the procedure, patients must remain highly vigilant in keeping themselves and the dialysis connections sterile to help avoid serious infection (peritonitis), which can begin at the catheter access site. Given the intricacies of the procedures involved and the home-based nature of this form of treatment, the successful administration of peritoneal dialysis is clearly dependent on the patient taking a more active role in his or her treatment relative to patients undergoing center hemodialysis.
In general, the choice of a particular ESRD treatment modality is substantially influenced by nonmedical factors, including patient and provider preferences and judgments about which modality might be associated with the most favorable patient adherence (24–27). For a small minority of ESRD patients, a particular renal treatment modality may not be medically feasible (eg, vascular access problems limiting the use of hemodialysis or loss of vision limiting the use of self-directed modalities). However, center, home, and peritoneal dialysis are all medically acceptable treatment alternatives for the large majority of patients (26, 27). Accordingly, the availability of a number of different treatments for ESRD, each with a different set of patient demands, offers a unique opportunity to apply an interactive framework.
NONADHERENCE IN ESRD
As with most chronic illnesses, patient nonadherence to the prescribed medical treatment regimen is a pervasive problem in the ESRD population. In addition to undergoing a regular schedule of time-consuming dialysis treatments, dialysis patients are required to follow strict medication and dietary regimens. Both CAPD and hemodialysis patients are required to take regular doses of phosphate-binding medication and to reduce intake of phosphorus-rich foods (eg, dairy products) because of the body’s inability to excrete serum phosphate while undergoing dialysis treatment. In many patients, phosphate-binding medication is poorly tolerated by the digestive tract, leading to unpleasant gastrointestinal side effects. However, if the medication regimen is not followed, serum P will rise above normal levels. Serum P levels greater than 5.5 mg/dl are typically interpreted as reflecting inadequate adherence with the phosphorus control regimen (5). Sustained elevations in serum P are associated with a variety of complications, including renal osteodystrophy, serious decreases in calcium, and subsequent bone demineralization.
Hyperkalemia is also a clinically significant problem for some dialysis patients because of the body’s inability to regulate serum K. Dialysis patients are required to follow strict dietary guidelines to maintain safe serum K levels. If dietary guidelines are not followed, serum K will rise, and potentially life-threatening cardiovascular complications, such as cardiac arrhythmia, can occur. Serum K values above 5.5 mEq/l are generally considered indicative of problematic adherence (5).
In the case of hemodialysis treatment, there are also extreme limitations on the amount of fluid that can be safely consumed because of the intermittent nature of the fluid and waste clearance provided by the three-session-per-week hemodialysis protocol. Prolonged fluid overload is associated with congestive heart failure, pulmonary edema, hypertension, dizziness, and severe muscle cramping and is potentially life-threatening. IWG is typically used to define adherence to the fluid-intake restrictions. IWG is determined by subtracting the postdialytic weight for the previous treatment session from the predialytic weight for the current session. The values resulting from this computation are believed to be a valid reflection of the amount of fluid that the patient ingests between dialysis sessions (28). Higher IWG values are interpreted as reflecting poorer patient adherence, with values greater than 2.5 kg generally indicative of problematic adherence (5).
Despite the potentially severe consequences of not adhering to the dialysis regimen, a number of studies have reported that between 30% and 50% of these patients do not adhere to diet, fluid-intake, and medication regimens (29–32). In general, these reports indicate that nonadherence is most common for fluid-intake guidelines and somewhat less common for dietary and medication restrictions. This pattern is consistent with reports that adhering to the fluid-intake restrictions may be the most challenging and stressful aspect of the hemodialysis regimen (33).
PATIENT-BY-TREATMENT INTERACTIONS AND ADHERENCE IN ESRD
The availability of a number of different treatments for ESRD, each presenting a distinct set of patient demands, offers a unique opportunity to apply the patient-by-context framework to the understanding of regimen adherence in this population. From an interactional perspective, adherence should be best when the patient’s characteristic or preferred style of coping with illness and treatment-related experiences is consistent with the contextual features or demands of the particular type of renal replacement treatment the patient is undergoing. Several of our past studies provide support for this assertion. In one early study (34), we tested this premise in a sample of patients undergoing either center or home hemodialysis. Patients’ characteristic or preferred style of coping was assessed using the Preference for Behavioral Involvement scale from the Krantz Health Opinion Survey (35). Adherence was predicted to be maximized in cases in which patient coping style matched the degree of actual patient control or involvement offered by the two types of dialysis treatment.
Consistent with their prediction (and after controlling for patient age, disease duration, transplant history, and diabetic status), among patients undergoing the more staff-directed center hemodialysis, a preference for active involvement in one’s own healthcare delivery was associated with significantly worse dietary (ie, K control) adherence (ie, higher serum K levels). In contrast, among patients undergoing hemodialysis at home, where patient involvement and control are greater, patients with strong preferences for active involvement displayed better adherence (lower serum K levels). Thus, adherence was best explained by the degree of congruency between patient coping style and the level of involvement permitted or required by the particular treatment patients received.
In a more recent application of the interactive hypothesis (6), we examined dialysis regimen adherence among 52 staff-treated center hemodialysis patients and 34 self-treated CAPD patients. Patient coping style was assessed using composite measures reflecting individual differences in the tendency or motivation of patients to exercise personal control in health contexts and to attend vigilantly to threat-relevant information related to health and treatment.
As depicted in Figure 2, higher scores on a composite "information vigilance" dimension were associated with better dietary adherence (ie, lower serum K levels) among self-treated CAPD patients but poorer dietary adherence among center hemodialysis patients. In contrast, low information vigilance was associated with better dietary adherence among staff-treated hemodialysis patients but poorer dietary adherence among CAPD patients. Controlling for patient age, disease duration, and years of education did not alter the observed pattern. These findings again suggest that the degree of congruence between a patient’s coping style and the self-care demands of a prescribed medical treatment is an important determinant of adherence.
|
Although past research suggests that the interactive framework may be useful for understanding adherence among chronically ill ESRD patients, most studies have been limited in several important ways. Past research by our research team, as well as others, has relied primarily on cross-sectional research designs (24). Relying on concurrent assessments of both predictor and outcome measures makes causal interpretations difficult or impossible.
In past studies, the potential limitations of cross-sectional assessment have been further compounded by the fact that patient characteristics have typically been assessed after patients reach the end stage of renal insufficiency and are receiving some form of renal replacement intervention to treat the disease. Causal interpretations using this methodology are even more difficult because the assessment of patient characteristics is confounded with the patient’s current medical condition, treatment history, and past experiences with the treatment regimen. A preferred methodological strategy would involve obtaining premorbid patient assessments before the individual actually becomes symptomatic and requires medical intervention. Unfortunately, chronic illness researchers in general have few opportunities to assess patients before the onset of the illness of interest.
In our most recent efforts involving ESRD patients, we have adopted an alternative and, we believe, superior methodological strategy for the study of this disorder. This strategy involves identifying patients at an early, asymptomatic stage in the progression of renal insufficiency, before renal replacement intervention is initiated. For the majority (approximately 80% at our center) of patients with chronic renal failure, ESRD is the end result of a progressive deterioration in kidney function over a period of months or years. Identifying future ESRD patients who are at an early stage of renal insufficiency provides a unique opportunity to prospectively test hypotheses concerning the prediction of adherence to a future regimen.
Our initial efforts using this prospective, longitudinal methodology have been encouraging. A recent study examined 69 patients first assessed (time 1) in the early stages of renal insufficiency (36). All patients had a form of progressive renal disease and were enrolled in the present study when their serum creatinine levels reached 3.5 or greater. Patients were reassessed (time 2) approximately 24 months after the initial assessment. At time 2, 52 of the 69 patients in this sample had reached end-stage renal insufficiency and were undergoing a form of renal replacement intervention. Twenty-nine of these patients were being treated with staff-directed center hemodialysis, and 14 were being treated with self-managed home hemodialysis. Nine patients were undergoing other forms of treatment (either peritoneal dialysis or renal transplantation) and were not included in the analysis because of the limited sample size of these subgroups. The 43 patients included in the analyses had been receiving dialysis for an average of 8.00 months (SD = 2.30) at the time of the follow-up assessment.
We assessed patients’ degree of "information vigilance" using a composite measure consisting of the information preference subscale from the Krantz Health Opinion Survey (35), the Internal Health Locus of Control scale from the Multidimensional Health Locus of Control scales (37), and the monitoring subscale from the Miller Behavioral Style Scale (38). Higher information vigilance scores were defined by higher scores on each of these component measures. Our primary hypothesis was that patients’ level of information vigilance, assessed before the progression of renal insufficiency, would interact with the type of renal dialysis eventually prescribed in predicting later regimen adherence. Our primary adherence measure consisted of mean IWG (an indicator of fluid-intake adherence) averaged over 4 weeks (12 dialysis sessions).
The predicted interaction between information vigilance and dialysis type on IWG was obtained (change in R2 = .12, F(1,42) = 5.49, p < .03, ß = .35). Predicted values illustrating the results for the prediction of fluid-intake adherence are depicted in Figure 3. As hypothesized, among home hemodialysis patients, individuals with higher information vigilance scores at time 1 displayed better adherence relative to patients with low information vigilance. Among center hemodialysis patients, an essentially opposite pattern was observed. Thus, consistent with the interactive framework, the degree of congruence between patients’ characteristic style of coping and the demands or requirements of the type of dialysis eventually prescribed predicted adherence to fluid-intake restrictions.
|
Our past work involving ESRD patients suggests that medical regimen adherence might be best explained by considering the interaction of patient characteristics and treatment context factors. The utility of this interactive approach is clearly not limited to the ESRD population. For example, Turk et al. (39, 40) have used a conceptually similar approach in work involving behavioral and multidisciplinary treatment outcomes among patients with chronic pain disorders. The work of these investigators suggests that psychological characteristics (eg, affective distress and perceived control) are significant predictors of differential treatment response among patients with fibromyalgia syndrome and temporomandibular disorders. Additional work extending the patient-by-context framework to other settings, populations, and outcomes will, I believe, highlight the broad utility of this conceptual framework for understanding patient adaptation in chronic disease.
One important issue that remains to be addressed involves identifying the processes that underlie (ie, mediate) the person-by-context interactive patterns observed in past research. In one recent study, we proposed that individual differences in perceived control may mediate the association between individual differences in coping style and patient adherence (41). For ESRD patients undergoing provider-directed, center hemodialysis, we found that a highly vigilant or active style of coping was associated with a significant reduction in perceived control over the illness and treatment context. Moreover, it was this lack of perceived control that largely accounted for a significant association between high vigilance and poorer patient adherence in this relatively passive treatment context. These findings might be interpreted in the context of past theory and research of patient psychological reactance (42, 43). Patients with a highly vigilant or active style of coping who are undergoing a highly provider-directed dialysis modality may attempt to reassert their perceived loss of control over the dialysis regimen by resisting instructions concerning diet, fluid intake, or other behavioral features of the treatment. More attention to this and other potential mediational processes is clearly needed before we can draw firm conclusions about the nature of the patient-by-treatment interactions obtained in past research.
Future research involving patient-by-treatment interactive models could also benefit from an expansion in the breadth of patient individual difference measurements. For example, in our past studies, we relied on relatively generalized measures of patient coping preferences or tendencies. An alternative approach would involve directly assessing patient preferences for home vs. center dialysis treatment. Ascertaining specific patient preferences for a particular mode of treatment is a potentially more efficient strategy than assessing generalized or more complex patient attributes (eg, level of information vigilance). A key question would be whether individual difference elements that have produced the interactive patterns observed in past research are adequately captured by more specific or circumscribed measurement strategies.
The focus of psychological measurement might also be broadened to consider factors other than patient coping or treatment preferences. For example, patients higher in perceived self-efficacy might exhibit more favorable regimen adherence, particularly when undergoing medical treatments that require greater patient independence (44, 45). The patient-by-treatment interactive framework might also integrate dimensions from the five-factor model of personality (46, 47). For example, the self-discipline and task orientation characteristic of high conscientiousness may be associated with the most favorable adherence in patient-directed treatment contexts. In contexts where the healthcare provider maintains extensive direction or control, the trusting, cooperative, and deferent style of high agreeableness might be more predictive of adherent behavior.
As is the case with most chronic medical regimens, obtaining an accurate assessment of patient adherence is a challenge for ESRD researchers. Our previous studies have relied on several physiological end points of patient behavior to operationalize regimen adherence. Both the empirical and clinical literature suggest that our primary adherence measures (eg, IWG and serum P) provide quite valid and reliable reflections of past patient behavior (5, 28, 29). Nevertheless, these measures are imperfect indicators of patient behavior and are potentially subject to a number of nonbehavioral influences. Other ESRD researchers have advocated the use of electronic medication monitoring systems or recording the frequency with which patients miss or shorten dialysis treatment sessions as alternative methods to assess adherence (48, 49). Clearly, refining the strategies we use to assess patient adherence behavior should be a central objective of future chronic illness research.
I believe that psychosomatic research is now in a position to make an important contribution to the clinical decision-making and management of patients with ESRD. Interactive models have direct implications for choosing among or tailoring medical treatment alternatives for patients to facilitate adherence and, ultimately, to maximize patient outcomes. For situations or disorders in which multiple medical treatments are not available, interactive data are potentially useful in identifying patients who may be at risk for nonadherence and for tailoring interventions to ameliorate these difficulties.
As medical technology continues to expand, the available treatment options for a given disorder are likely to grow as well. As treatment options grow and medical decision-making becomes more complex, it will become increasingly important for clinicians to have all relevant assessment data (including psychosocial data) at their disposal. I hope that the conceptual framework and data described here prove helpful to other psychosomatic researchers and clinicians whose work contributes to the management of all types of chronic disease.
ACKNOWLEDGMENTS
Much of the research described in the article was supported by the National Institute of Diabetes and Digestive and Kidney Diseases, Grant DK49129.
NOTES
This article is based on an Early Career Award address given at the 1999 Annual Meeting of the American Psychosomatic Society in Vancouver, British Columbia, Canada.
Received for publication May 28, 1999.
Revision received October 8, 1999.
REFERENCES
This article has been cited by other articles:
 |
|
 |
|
  K. C. Insel, S. L. Reminger, and C.-P. Hsiao The Negative Association of Independent Personality and Medication Adherence J Aging Health, June 1, 2006; 18(3): 407 - 418. [Abstract] [PDF]
|
|
|
|
 |
|
 S. M. Auerbach and P. O. Pegg Appraisal of Desire for Control over Healthcare: Structure, Stability, and Relation to Health Locus of Control and to the 'Big Five' Personality Traits J Health Psychol, July 1, 2002; 7(4): 393 - 408. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
