Journal of Clinical Pharmacy and Therapeutics (1992)17,283-295

REVIEW ARTICLE

Patient compliance-an

overview

L. Stockwell Morris, and R. M. Schulz Division of Health Care Policy and Evaluation, United Healthcare Corporation and College of Pharmacy, University of South Carolina, U.S.A.

SUMMARY

This article reviews the major topic areas of compliance research. Much of the research in the area has focused on measurement, extent, and determinants of non-compliance. Research on the effectiveness of educational and behavioural strategies to improve compliance suggests the need to combine them. While some authors have attempted to model compliance or medication-taking behaviours, these models cannot be applied widely. After decades of compliance research, very little consistent information is available, except that people do not take their medications as prescribed. The methodological rigour of compliance studies may partially contribute to this situation. Methodological flaws have included design features and study execution. In addition, researchers have proceeded with studies without regard to a theoretical framework. Many have argued that much of the existing compliance literature also lacks conceptual rigour. Although we know that people do not take their medications consistently, we do not know specifically why they have done so. One reason for this lack of understanding is that compliance research has been dominated by the perspective of the health professional. To better understand medicationtaking behaviour, researchers need to examine the patient's perspective. Consequently, future research needs to investigate a patient's decisionmaking process and the reasons for those decisions. *Correspondence:Dr Lisa Stockwell Morris, Pharmacy Research Scientist, Healthcare Evaluation Services, Division of Health Care Policy and Evaluation, United Healthcare Corporation, 9900 Bren Road East, P.O. Box 1459,Minneapolis, MIN 55440,U.S.A. Series editor: Dr D. Worthen.

INTRODUCTION

The Office of the Inspector General reported that the consequences of non-compliance among the elderly can include increased morbidity and mortality, in addition to increased cost of treatment (I). Non-compliance has beenlinked to23%ofnursing homeadmissions(2)andup to 10% of all hospital admissions (3).Non-compliance with treatment for cardiovascular disease contributes to approximately 125,000 deaths and several thousand hospitalizations per year (4). The annual cost of treatment for moderate to severe hypertension was estimated to be $4,850 if patients purchased all of the medications and were 100% compliant. This annual cost rose to $10,500 when patients purchased the medications but took them irregularly (5). For these reasons, non-compliance has been considered a major public health problem (6). Compliance research has focused on the extent and determinants of non-compliance, and strategies to improve compliance. The purpose of this article is to summarize and review research on compliance with medications in these areas. In addition, the methodological rigour and conceptual basis of traditional compliance studies will be discussed. DEFINITION

Compliance has been defined (7) as 'the extent to which a person's behavior (in terms of taking medications, following diets, or executing lifestyle changes) coincides with medical or health advice'. Thus, compliance is viewed as a process. Some researchers have put this definition into operation by dividing their sample populations into compliers and non-compliers based on statistical measures such as the median or mean levels of medicine taken (8).Those who fall below the mean or median are labelled as non-compliant, while those above

284

L. Stockwell Morris and R. M. Schulz

the criteria are compliant. Other researchers have used previously reported levels of medication-taking to distinguish between compliance and non-compliance or relied on their personal belief that a certain level is significant (8). Compliance can also be viewed in terms of the results of taking medication. Compliance as an outcome is defined as ’the number of doses not taken or taken incorrectly that jeopardize the therapeutic outcome (6)’, or ’the point below which the desired preventive or desired therapeutic result is unlikely to be achieved (9)’.Both of these definitions recognize the possibility that taking less than 100% of the medication can result in a desired health outcome. Outcome-oriented definitions differ from the process-oriented definition in their emphasis on the endresult or outcome of the actions taken. For example, Luscher and co-workers (10)reported that 80% compliance to a medication regimen for hypertension lowered blood pressures to a normal level. However, compliance of 50% or less was ineffective in adequately lowering blood pressures. Olson and co-workers (11) reported that a compliance rate of 80% was necessary for therapeutic results in children with streptococcal pharyngitis, but 33% compliance reduced the rate of contracting streptococcus infections in children taking oral penicillin as a prophylactic for rheumatic fever. The absence of a singular conceptual basis of compliance is problematic. Strategies to improve compliance can be evaluated only within the context of a given definition. Furthermore, comparative assessment of the compliance literature cannot be done across studies using different definitions of compliance.

MEASUREMENT

Compliance is measured directly and indirectly. Direct measurements of compliance generally involve the detection of a chemical in a body fluid. Examples of direct measurements include blood levels or urinary excretion of the medication, a metabolite, or a marker. The primary reason for using direct measurements is they are less subject to bias than are indirect measurements (12). They can be misleading, however, if the patient takes the medication prior to testing. Little information can be derived regarding the use of medication over time. Direct measurements do not account for the variability of pharmacokinetic factors of different medi-

cations and different individuals. In addition, direct measurements can be quite difficult to perform and costly. Indirect measurements of compliance are more frequently reported in the literature, possibly due to the relative ease by which these measures are obtained. Examples of indirect measurements of compliance include therapeutic or preventive outcome, impression of the physician or predictability, patient interview, prescription filling dates, and pill counts (13). As with direct measurements, each indirect means of measuring compliance has inherent advantages and disadvantages. The therapeutic outcome may seem a viable means to assess compliance, however, patients do improve for reasons other than following the prescribed regimen. Epstein and Cluss (14) reported that compliance with either the active product or placebo was associated with an appropriate clinical outcome. In addition, a person’s condition can deteriorate or remain stable even when the medications are taken as prescribed. The physician’s estimate of patient compliance has been very inadequate. Caron and Roth (15)determined that physicians do not estimate compliance any better than if they simply rely on chance as a predictor. Mushlin and Appel(16)reported that less than one-half of a physician’s predictions correctly discriminated between compliers and non-compliers, while threequarters of their predictions of non-compliance were inaccurate. Compliance can also be assessed using a patient interview or self-report. Park and Lipman (17) used both pill counts and patient interviews to assess compliance of psychiatric patients with imipramine. Using interviews, they categorized 100 patients as compliers, but using pill counts only 58 would be compliers. In addition, in only 68 of the patients did both methods lead to the same classification. Gordis et al. (18),interviewing mothers and their children and correlating these with the findings of urine tests, found that both the mother and child overstated compliance and understated non-compliance. Morisky et al. (19) developed a four-item scale for self-report of compliance that demonstrated both concurrent and predictivevalidity of blood pressure control and supported the use of patient interviews. Using centralized pharmacy records to check refill dates or the number of pills obtained is another indirect means to measure compliance (20-24). Steiner and coworkers (25)reported that refill records of the managed

Patient compliance

care setting under study were valid correlates of medication effects and compliance. The authors believe that this method is useful in detecting non-compliant patients who remain in the health care system but only fill some of the medications prescribed. The validity of this method, however, rests on the completeness of the pharmacy database, usually most complete in institutional seitings. Pill counts have also been used to assess compliance. Roth and co-workers (26) suggested that pill counts often over estimate compliance. Bergman and Werner (27) found that, based on urinalysis, 8% of their study population was compliant, while pill counts showed 18% of the population to be compliant. Rudd and coworkers (28) found that weekly pill counts indicated variability within subjects, between subjects, and among the different regimens. A recent advance in compliance measurement is electronic monitoring. The prescription vial contains a microprocessor in the lid. This recording device will identify the date and time the prescription vial is opened. Although the device does not verify that the medicine was consumed, it does provide an accurate record of when the vial was opened, supposedly for the purpose of taking the medicine. In this regard, electronic monitoring should be able to identify individuals who falsely claim they are taking medicine as prescribed. Engstrom (29) reported that five of 19 patients who provided prescription medicine in vials with microprocessors embedded in the caps were non-compliant. None of those five patients claimed to be noncompliant. Four of the five patients did not improve clinically. Cramer et al. (30)used the electronic monitoring system with 24 epileptic patients who consumed 7,413 doses over 3,428 days. A comparison of electronic monitored data and pill counts showed that pill counts overestimated compliance as compliance declined. Furthermore, a dear relationship was established between adherence to the prescribed regimen, as determined by electronic monitoring, and clinical outcome. Five patients reported seizures during monitoring; five of these patients’ seizures could be attributed to missed doses documented by electronic monitoring. Cramer ef al. (30) view electronic monitoring as an additional and needed dimension in measuring patient compliance. Electronic monitoring helps to explain how the total number of doses were taken, if the dosing regimen was followed, and if clinical outcomes can be related to non-compliance. Although not a panacea, electronic monitoring improves clinicians’ and researchers’ understanding of how patients take prescribed medicine.

285

At this point in time, there is not a consistent measure used to determine compliance. Researchers will often choose a particular method due to ease and convenience. Likewise, it is not known what is the most appropriate means to measure compliance. The use of various measurement techniques makes it very difficult to compare compliance research findings.

EXTENT OF NON-COMPLIANCE

Fedder (31)stated that, ’it seems that a third of patients always comply, a third never comply and a third sometimes comply’. Regardless of the patient population, disease state, or compliance measurement used, the compliance rate is usually well below 100%. The duration of therapy is a factor which can influence the extent of compliance. Compliance with short-term medications is generally considered to be somewhat higher than for long-term medication regimens. However, rapid declines in compliance occur during even the first 10 days of short-term therapy (32). This suggests that compliance behaviour may be unstable even in short-term therapy. Generally, the rates of compliance in long-term therapy tend to converge to SO%, regardless of the illness or setting (32). The compliance rate for long-term medications used for prevention, treatment, or cure can range from 33 to 94% (32). Published compliance rates for long-term medication use are shown in Table 1(32-34). The extent of compliance in a specific disease population has also been evaluated. Hogarty and co-workers (35) found that only 42% of the 374 schizophrenic patients studied were taking their medications correctly. Young et al. (36) reviewed the literature on non-compliance of schizophrenic out-patients and reported a median compliance rate of 41%, with a range of 10-76%. These findings approach the 50% compliance reported for long-term therapy. Compliance rates for those with affective disorders are very similar. Johnson and Freeman (37) found that 16% of the 73 patients prescribed medications for depression stopped taking their medication within 1 week of beginning the treatment. Within 30 days of the initiation of the treatment, 68% had stopped taking the prescribed medicine. In three studies that examined compliance with lithium therapy, researchers reported that between 25 and 50% of patients failed to take their medications as prescribed (38),only 47% of the patients were thought to be complying perfectly over the first 2

286

L. Stockwell Morris and R. M. Schulz

Table 1 Compliance with long-term medication' ~~

Treatment

Measure

Definition

Penicillin prophylaxis for rheumatic fever (94) Anxiolytics in neurotics (95)

Urine assay

Medication in urine

33

Pill counts

54

Antipsychotics in schizophrenics (35)

Interview

Tuberculosis medications (96)

Interview and urine testing Record review Interview

Record reviews (same subjects)

Counts within 25% of prescribed amounts Taking medications correctly Taking medications throughout follow-up Continuing therapy Taking medications correctly Taking medications correctly Taking medications correctly Remaining in care and on therapy

Pill counts (same subjects)

Taking > 80%of medication

Tuberculosis medications (97) Various medications used by the elderly (98) Various medications for diabetes or congestive heart failure (99, 100) Various medications used by patients in homes for the aged (101) Antihypertensives (102)

Antihypertensives (103, 104)

Interview Interview

Compliance rate%

42 55 63 41

42

69 1year 2years 3 years 6month 12month

94 65 34 53 53

'This table represents articles that provided demographic descriptions of the subjects and employed statistically rigorous sampling techniques. Examples include the use of random population samples, three or more hospitals/clinicsin a geographicalarea, or a regional programme/ referral system.

years of treatment (39, and 24% withdrew from treatment during the h s t 2 years (40). These studies further support the lack of stability of compliance as a behaviour. Compliance rates for hypertensive patients are quite similar to rates reported with other conditions. For example, more than 50% of these patients dropped out of therapy within the first year, while of those remaining about 33% did not take enough of their prescribed medications to reduce their blood pressure adequately (41, 42). This finding implies that compliance behaviour is relatively unstable over time. Widmer and associates (43) reported a mean compliance rate of 87% for 291 hypertensive patients in a rural mid-west study. The high rate of compliance may have been partially attributed to the method of measurement and the definition of compliance. Such a conclusion supports the contention that comparability of studies should be done with great caution. Dodrill and co-workers (44) identified 80 of the 282 adults with epilepsy studied as definitely compliant, while 42 were non-compliant. The remaining subjects

fell into the categories between definitely compliant and definitely non-compliant. Eisler and Mattson (45) reported that 40% of patients taking anticonvulsants missed enough doses to affect blood levels. In addition, non-therapeutic serum anticonvulsant levels have been measured in as many as 97% of 34 patients (46) and as low as 26% in 153 cases (47). As with other diseases presented, theextent of compliancewith anticonvulsants is quite variable, but still within the range expected for long-term therapy. The average rate of compliance tends to converge to 50% for long-term therapy, regardless of disease and compliance behaviour tends to decline with time. The increasing prevalence of chronic conditions for which long-term therapy is prescribed suggests that compliance will remain an area of concern. DETERMINANTS

Meichenbaum and Turk (12) identified over 200 variables that have been examined in studying compliance. The majority of these variables are characterized as

Patient compliance 287

demographic variables, such as age, sex, education, and socio-economic factors, or disease-experience variables. In this section, the influence that demographic variables, disease features, and treatment regimen features have on compliance will be presented. Demographics have been shown to be poor predictors of compliance.Hulka (48)reported that the demographic factors of age, sex, marital status, education, number of people in the household, and social class are not statistically associated with compliance. In a study of hypertensive patients, Widmer et al. (43) found no significant differencesbetween mean compliance rates for males and females. Similarly, Dodrill and co-workers (44) found that the sex and age of the epileptic patients studied did not adequately predict compliance. However, under certain specific conditions, such as a particular disease, demographic variables can predict compliance behaviour (49). Unfortunately,the findingsfor these specificconditionsdo not extend to other situations. Such inconsistent results havemade it impossible todevelopaprototypeof thenoncompliant individual. Porter (50)summed up the situation well by stating 'it must be emphasized that it has not proved possible to identify an uncooperative type. Every patient is a potential defaulter. Compliance can never be assumed.' Features of the disease have also been investigated. In a review of the literature published before 1978, Haynes (51) suggested that disease factors were poor indicators of compliance. Less than half of the studies found any significant correlations between compliance and disease features. In addition, even where significant correlations were noted, the direction of these associations was inconsistent between studies. Only three disease features were found to be determinants of compliance (51). First, psychiatric patients, especially those with schizophrenia, paranoia, or personality disorders, are generally low compliers. Second, when more symptoms are reported by patients, their compliance rate is lower. This finding does not support the assumption that increasing severity of symptoms should encourage compliance, suggesting that as people notice more symptoms they begin to give up on the treatment. In addition, patients with symptomatic illness, such as rheumatoid arthritis, become less compliant. Third, the degree of disability produced by the disease positively influences compliance. The increased compliance could be the result of the severity of the disease or increased supervision due to the disability. The authors felt that increased supervision was the more likely explanation.

Features of the treatment regimen have also been evaluated as possible determinants of compliance. Haynes (51)reported that the duration of treatment, the number of medications,and their cost negatively affected compliance, while parenteral medication administration improved compliance. Sbarbaro (52) suggested that shortening the duration of therapy and reducing the number of times that a medication is taken each day will help to improve patient compliance.Note that while the presence of side effects intuitively seems to decrease compliance, Haynes (51) reported that when patients were asked to give reasons for non-compliance, sideeffects were mentioned by only 5-10%. According to Haynes, however, future research in the area Qf disease features, and to some extent treatment features, should receive low priority because of the inability to increase detection or improve compliance (51). More than 20 years of research in the area of compliance has produced very little consistent information on the factors which can be correlated with noncompliant behaviour. Most of the variables examined are inconsistently correlated with compliance and thus cannot be used to predict compliant behaviour adequately.

INTERVENTIONS T O IMPROVE COMPLIANCE

Educationd strategies

Educational strategies, such as verbal communication or counselling, have been shown to improve compliance (53). Hecht (54) found that counselling provided by registered nurses to patients receiving out-patient chemotherapy decreased drug errors. Nessman ef al. (55)developed a programme that combined audiovisual teaching with counselling by a registered nurse which decreased medication errors and diastolic blood pressures. Zismer and associates (56) reported that counselling of hypertensive adults at clinic visits by graduate research assistants produced a substantial decrease in blood pressure. Similarly, counselling by a pharmacist resulted in a significant clinical decrease in the diastolic blood pressures of adults attending a neighbourhood clinic. Based on these studies, one-to-one counselling appears to be an effective means of improving compliance. Not all research supports the association between educational strategies and compliance. In a review

288

1.Stockwell Morris and R. M. Schulz

including only those articles published before 1979, only one of the four studies involving medication instruction found a significant change in compliance due to the intervention (57). In this study of patients being discharged from the hospital, 90% of the patients who received counselling from the pharmacist before discharge were compliant with the prescription orders, but only 24% of those who did not receive counselling were compliant (58). Note that Compliance did appear to be positively affected by pharmacy counselling over the short-term of the study. In the other three studies reviewed by Haynes (57), two of which involved pharmacy counselling and one of which involved nursing counselling, these interventions did not produce statistically significant changes in patient compliance. In their review of the literature on written cornmunication, Morris and Halperin (59) concluded that written information may be effective for short-term medication therapy. Written information alone is not sufficient to increase patient compliance in medication used in longterm therapy. In a meta-analysis of artides written between 1961 and 1984 on intervention strategies, written interventions, except for patient package inserts, were shown to produce increased knowledge and decreased medication utilization errors (60).The studies on patient package inserts resulted in an average effect size value near zero for both knowledge and medication utilization errors (60).This indicates that these inserts have limited ability in improving knowledge or decreasing medication utilization errors. Gibbs et al. (61)found that compliance was not significantly different for those patients who received patient information leaflets, however, their knowledge of the medication was better. In a meta-analysis of combinations of intervention strategies, one-to-one counselling, group education, and written and/or audiovisual interventions, except patient package inserts, increased knowledge and decreased medication utilization errors (60). Robinson and coworkers (62) found that when written information was combined with verbal reinforcement, compliance was significantly higher than when written information was used alone. In addition, compliance was higher for the intervention groups than for the control. Myers and Calvert (63) found no significant differences in compliance between a group which was told the purpose of the medication but was given no written information, a group which received verbal and written information about the side-effects of the medication, and a group receiving verbal and written information about the beneficial effects of the medication. When the two groups

which received both verbal and written information were collapsed into one group, compliance at 3 weeks was significantly higher for this new group than for the group which received no written information and limited verbal information. These results were not sustained over 6 weeks. Myers and Calvert (63)suggested that verbal and written information improves compliance by an attention-placebo effect rather than a cognitive effect. Consequently, it would seem that compliance poses a larger problem in long-term therapy or once the information intervention is removed. Brown et al. (64) found that individuals who receive verbal and written information gain more medical knowledge than those who receive only verbal information, but instruction does not significantly affect compliance. The use of educational strategies to improve compliance is based on the information model of compliance management (65). The model suggests that patients fail to comply because they do not have s&cient information regarding the risks of the disease, the benefits of treatment, and the details of the treatment. From the studies presented, educational strategies alone may not significantly improve patient compliance. Information may provide its greatest benefit for short-term therapy. Similarly,information is valuable during the early stages of long-term medication regimens. However, in the long-term, once the intervention is removed, information may affect compliance less.

Behavioural strategies

Behavioural strategies, such as reminders and special medication containers, have been shown to improve compliance (53, 66). Patients with hypertension who used a medication reminder chart showed improved knowledge of medication use, dose, and frequency of administration after using the chart (67).They presented with fewer incidences of forgetting to take the medication and a lower number of deviations from the prescribed regimen. Another type of reminder is a refill reminder which is generated at the pharmacy level. S i n s and Wenzloff (68)investigated the use of postcard refill reminders and telephone-call refill reminders. These researchers found that both types of reminders increased refill compliance, but there were no significant differences in compliance between the two types. Specialmedication containers are another behavioural intervention. They help the patient to organize medications and monitor self-administration of products on

Patient compliance 289 Table 2 Successful compliance Effect on

interventions

compliance Condition

Strategy

(%)

Effect on outcome

Hypertension (104)

Self-monitoring +cueing + rewards Nurse management at work + cueing + selfmonitoring +rewards Self-monitoring group discussion + selfmanagement Written instruction treatment cards +recall of nonattenders

+21

Improved

+ 19

Improved

+ 19

Improved

+ 15

Improved

Hypertension(105) Hypertension(55) Hypertension (106)

daily to weekly intervals (53).Special medication packaging has shown to increase compliance (69,70).Wong and Norman (71) investigated the use of a calendar blister-pak to improve compliance. The average noncompliance index significantly decreased when patients used the blister-pak, indicating that this type of special packaging may be advantageous in improving compliance. Some manufacturers are now selling their products in special packaging. Oral contraceptive packs and the Medrol DosepakB are two examples of special unit-of-use packaging. Educational and behavioural combinations

Many researchers have evaluated the use of both educational and behavioural strategies in a combined intervention. Combining behavioural and educational strategies is supported by the compliance activity model (65). This model recognizes the relationships between behavioural, cognitive, and social factors in determining compliant behaviour. In a study of patients with epilepsy, the intervention group received patient counselling, a special medication container, mailed reminders to have prescriptions refilled, and they were asked to self-record medication intake and seizure frequency (72). Patient compliance significantly improved in patients receiving this combined intervention. Ascione and Shimp (73)evaluated the effectiveness of oral instructions alone, oral instructions plus a medication reminder calender, oral instructions plus a medication reminder package, and oral inshuctions plus written medication information. Interventions including either the medication reminder calendar or package were

+

+

superior in improving compliance than the other interventions. Swain & Steckel(74) investigated the effects of patient education and contingency contracts on compliance. Patients received routine care only, routine care plus education, or routine care plus education and a contingency contract between the patient and the nurse. This latter group exhibited a decrease in diastolic blood pressures. Haynes (65) has reported methodologically rigorous studies of compliance interventions. All of the successful interventions, or those which had significantly positive effects on compliance and treatment outcome, involve a combination of interventions. A sample of these interventions can be found in Table 2. It appears, from the materials presented in this section, that compliance can best be improved by interventions combining educational and behavioural components. The effectiveness of interventions combining educational and behavioural techniques was supported by the findings of a meta-analysis prepared by Mullen et al. (60). Compliance packaging, which incorporates both educational and behavioural components, has been introduced recently by the pharmaceutical industry in an effort to improve compliance. This packaging ideally serves as a patient education tool and makes it easier for the patient to remember to take the medication (75). An example of this is the M A C P A P manufactured by Norwich Eaton Pharmaceuticals, Inc.

MODELS OF CQMPLIANCE One of the first models used to explain compliance was the Health Belief Model (76). This model was initially

290

L. Stockwell Morris and R. M. Schulz

applied to preventive health behaviour. Becker and Maiman (77) incorporated general health motivations and modifying and enabling factors into the model to provide a focus on sick role behaviour and compliance with medications. While many of the studies that evaluate the Health Belief Model support the components of the model, some studies do not (78).Under certain disease conditions, some of the components of the Health Belief Model do help in the understanding of compliance. However, the entire model has not been supported empirically (79). In response, researchers attempted to improve the model by adding more components. Ried and Christensen (79) incorporated the Theory of Reasoned Action into the Health Belief Model. In this study, the Health Belief Model explained 10% of the variance in the compliance variable. The inclusion of the Theory of Reasoned Action explained an additional 19% of the variance. The Health Belief Model variables of barriers and benefits and the Theory of Reasoned Action variables of belief strength, outcome evaluation, and behavioural intention were found to be significantly related to compliance.The social influence variables, part of the Theory of Reasoned Action, were able to predict behavioural intention, but not compliance behaviour. The research supported the contention that some of the paths of the models are significant but the models as a whole cannot predict compliance. Even with the combined models, only 29%of the variance was explained. This suggests the possibility of missing paths in the model or the improper measurement of the variables. The Health Decision Model (80),developed by Eraker et al. incorporates the Health Belief Model and patient preference, including decision analysis and behavioural decision theory. The model is based on the strengths of the Health Belief Model and patient preference models. The model includes bidirectional arrows and feedback loops which suggest that compliance behaviour can also change beliefs. The validity and predictability of this model was not statistically tested by these authors. Nagy and Wolfe (81)evaluated a model using variables derived from the health locus of control and the Health Belief Model. The demographic variables of age and socio-economic status were also included in the model. Scales to assess Internal Health Locus of Control, chance Health Locus of Control, and powerful others Health Locus of Control were administered. The latter scale evaluated the person's belief that health professionals or family can affect health. The perceived severity of illness, outlook on illness, experienced symptoms, satisfaction with treatment, family support and

support of others were used in conjunction with the demographic variables and health locus of control scales to develop an equation to predict compliance. Patient satisfaction was the only significant predictor of medication compliance.The results of this study indicate that either this model or the measurement of the variables is inappropriate. The authors suggest that the limitations of cognitive variables in predicting compliance in patients with chronic diseases can partially explain the results. In addition, subjects in this study had received treatment for an average of 17 years. Because of these characteristics, their health beliefs and health locus of control may differ from patients in other circumstances. In response to inadequate findings when using the Health Belief Model, Schlenk and Hart (82) investigated the use of Rotter's Social Learning Theory. These authors incorporated health locus of control, health value, and social support in their model. A significant relationship was found between compliance and social support, powerful others health locus of control, and internal health locus of control. In a multiple regression analysis, social support and powerful others health locus of control accounted for 50% of the variance in compliance. Because a non-experimental design with a small number of subjects was used in this study, the results may be different under a more controlled experiment with more subjects. Stanton (83) proposed a model which included patient-provider communication, knowledge of medication regimen, satisfaction with the provider, internal locus of control, perceived social support, and treatment disruption to lifestyle as variables that directly and indirectly lead to adherence to a medical regimen. The link between adherence to medical regimens and the outcome of blood pressure change were also evaluated. Greater expectancy for internal control over health and hypertension, greater knowledge of the treatment regimen, and stronger social supports were significantly predictive of adherence to medication regimens. In addition, higher levels of adherence were related to blood pressure reduction. Again, several components of the proposed model were statistically significant but the entire model could not predict compliance better than the individual paths in the model. As with the other areas of compliance research, modelling has produced inconsistent findings. It was intended that these models should serve as data reducers and organizers. However, a strong model of compliance still does not exist. Possibly the problem is that the researchers are examining inappropriate

Patient compliance

paths or relationships. Another problem may be the inconsistency of techniques employed when evaluating the variables in the models and the context may influence the fit of the model, thus limiting wider applications of any model. INADEQUACIES OF COMPLIANCE RESEARCH

The methodological rigour of compliance studies has been questioned. Haynes (65) indicated that the scientific merit of these studies ranged from primitive to exceptionally high. One explanation for this situation may be that health care researchers empirically test potential factors that might overcome low compliance, regardless of any theoretical framework (84).In addition, researchers differ in their ability to recognize and control for design features and execution of the study (85).These flaws in design and execution affect the certainty of the conclusions of the projects. Haynes and co-workers (65,85)developed criteria to judge the methodology of 537 original articles on compliance. Bruer (86)combined the results of Haynes bibliography with citation data from the Science Citation Index to determine whether methodologically rigorous studies were cited more frequently by compliance researchers than less rigorous articles. The results indicated a statistically significant, but low, correlation between methodological rigour and citation frequency. Another area of concern is the conceptual rigour of compliance studies. According to Trostle (87), the notion of compliance is dominated by ideological beliefs of the appropriate roles for patients and physicians. The concept of compliance is based on the doctor’s perspective and implies that the physician is the benevolent authority and the patients should willingly accept the doctor’s word (88). Stimson (89) suggested that researchers have not questioned the assumption that patient’s should comply with doctor’s orders. Consequently, seeing patients as defaulters or deviants has resulted in an unproductive approach to the problem of medication use. This medically oriented approach to compliance has portrayed the doctor-patient relationship as the key to medication taking behaviour. However, Conrad (90) suggests that this may not adequately reflect the true situation. This view is supported by Trostle (87),who indicates that ’noncompliance is an unavoidable byproduct of collisions between the clinical world and other competing worlds of work, play, friendship, and

291

family life’. Stimson (89)points out that people are not taking medicines in a thoughtless vacuum but that they have ideas and attitudes about medicine. These ideas and attitudes are based on their relationships with others and past experiences. Because of the social context of taking medications, researchers should focus on compliance from the patient’s perspective rather than from the viewpoint of health professionals (87,89-92).This approach should focus on the reasons or motivations for the medication-taking behaviour adopted. This type of research has received the lowest priority in much of the previous research (89). Other problematic assumptions that have prevailed in previous research on compliance have been noted by Gabe and Thorogood (93). First, researchers have assumed that the individual is the methodological unit of analysis. Individualswere treated as though they were members of a single population, rather than as members of many sub-cultures. Second, researchers have attempted to identify causal relationships between variables, assuming that the variables can be treated as independent. However, the phenomenon of medicationtaking behaviour involves variables that are inter-related with the possibility of feedback loops. Third, researchers have explained medication use through establishing strong relationships between certain variables and medication use. This approach has neglected to determine how these associations were formed. They have ignored the social contexts which are involved in behaviour. The fourth assumption noted involves the manner in which subjective meanings are gathered. In cases where subjective meanings are introduced as a causal variable, they are obtained through standardized questionnaires. By using these methods, researchers have failed to obtain the social and historical context of medication use in peoples‘ own words. The years of research on compliance provide little consistent information other than the fact that people do not always follow the doctor’s orders. Unfortunately, we do not know specificallywhat the patient has done. For example, research has not indicated if individuals consistently take fewer doses or a lower dose. Future research needs to investigate how patients administer their medications and the decision-making process used by the patients. For example, as Trostle (87) suggests, we need to evaluate how patients respond in daily life to the advice and treatment regimens presented by the medical profession. Above all, there appears to be a need to focus on compliance from the patient’s perspective.

292

L. Stockwell Morris and R. M. Schulz

REFERENCES 1. Department of Health and Human Services, Office of the Inspector General. (1990)Medication Regimens:Causes of

Noncompliance. U.S. Department of Health and Human Services, Office of Inspector General, Washington, D.C. 2. Smith M. (1985) The cost of noncompliance and the capacity of improved compliance to reduce health care expenditures. I n Improving Medication Compliance:Proceedings of a Symposium in Washington, D.C., November 7, 1984, by the National Pharmaceutical Council, pp. 35-44. National Pharmaceutical Council, Reston, Virginia. 3. McKenney JW, Harrison WL. (1976) Drug-related hospital admissions. American ]ournu1 of Hospital Pharmacy, 33,792-795. 4. Smith M. (1985) The cost of noncompliance and the capacity of improved compliance to reduce health care expenditures. In: Improving Medication Compliance: Proceedingsof a Symposium in Washington, D.C., November I , 1984, by the National Pharmaceutical Council, pp. 35-44. National Pharmaceutical Council, Reston, Virginia. (Quote Levine, D). 5. Weinstein MC, Stason WB. (1976) Hypertension. A Policy Perspective. Harvard University Press, Cambridge, MA. 6. Smith DL. (1989) Patient Compliance: An Mutational Mandafe. Norwich Eaton Pharmaceuticals, Inc., Norwich, New York and Consumer Health Information Corporation, McLean, Virginia. 7. Haynes RB. (1979) Introduction in Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 1-7. The Johns Hopkins University Press, Baltimore. 8. Becker MH, Rosenstock IM. (1984) Compliance with medical advice. In: Health Care and Human Behavior eds Steptoe A, Mathews A, pp. 175-208. Academic Press, New York. 9. Gordis L, (1976)Methodological issues in the measurement of patient compliance. In: Compliance with Therapeutic Regimens eds Sackett DL, Haynes RB, pp. 51-66. The Johns Hopkins University Press, Baltimore. 10. Luscher TF, Vetter H, Siegenthaler W, Vetter W. (1985) Compliance in hypertension: facts and concepts. ]ournu1 of Hypertension, 3 (Suppl. I),3-10. 11. Olson RA, Zimmerman J, Reyes de la Rocha S. (1985) Medical adherence in pediatric populations. I n Health Psychology: Treatment and Research hsues eds Zeiner AF, Bendell D, Walker CE, pp. 113-143. Plenum Press, New York. 12. Meichenbaum D, Turk DC. (1987) Facilitating Treatment Adherence:A Practitioner's Guidebook, pp. 36-37. Plenum Press, New York. 13. Gordis L. (1979) Conceptual and methodological problems in measuring patient compliance. In: Compliance in

Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 23-45. The Johns Hopkins University Press, Baltimore. 14. Epstein LH, Cluss PA. (1982) A behavioral medicine perspective on adherence to long-term medical regimens. ]ournu1 of Consulting and Clinical Psychology, 50, 950-971. 15. Caron HS, Roth HP. (1971) Objective assessment of cooperation with an ulcer diet relation to antacid intake and to assigned physician. American lournal of Medical Science, 261,6147. 16. Mushlin AI, AppelFA. (1977)Diagnosing potential noncompliance:physicians' ability in a behavioral dimension of medical care. Archives of Internal Medicine, 137, 318-321. 17. Park LC, Lipman RS. (1964) A comparison of patient dosage deviation reports with pill counts. Psychopharmacologia, 6,299-302. 18. Gordis L, Markowitz M, Lilienfeld AM. (1969) The inaccuracy in using interviews to estimate patient reliability in taking medications at home. Medical Care, 7, 49-54. 19. Morisky DE, Green LW, Levine DM. (1986)Concurrent and predictive validity of a self-reported measure of medication adherence. Medical Care, 24'67-74. 20. lnui TS, Carter WB, Pecoraro RE, Pearlman RA, Dohan JJ. (1980)Variations in patient compliance with common long-term drugs. Medical Care, 18,986-993. 21. Wandless I, Mucklow JC, Smith A, Prudham D. (1979) Compliancewith prescribed medicines: a study of elderly patients in the community. ]ournu1 of the Royal College of General Practitioners, 29,391-396. 22. Enlund H, Tuomilehto J, Turakka H. (1981)Patient report validated against prescription records for measuring use of and compliance with antihypertensive drugs. Acta Medica Scandinavica, 209,271-275. 23. Enlund H. (1982) Measuring patient compliance in antihypertensive therapy-some methodological aspects. Journal of Clinical and Hospital Pharmacy, 7, 43-5 1.

24. Rudd P. (1979) In search of the gold standard for compliance measurement.Archives of Internal Medicine, 139, 627-628. 25. Steiner JF, Koepsell TD, Fihn SD, Inui TS. (1988) A general method of complianceassessment using centralized pharmacy records. Medical Care, 26,814-823. 26. Roth HP,Caron HS,Hsi BP. (1970)Measuring intake of a prescribed medication: a bottle count and a tracer technique compared. Clinical Pharmacology and Therapeutics, 11,228-237. 27. Bergman AB, Werner RJ. (1963) Failure of children to receive penicillin by mouth. New England lournal of Medicine, 268, 1334-1338.

Patient compliance

293

28. Rudd P, Byyny RL, Zachary V, et al. (1989) The natural history of medication compliance in a drug trial: limi-

43. Widmer RB, Cadoret RJ, Troughton E. (1983) Compliance characteristics of 291 hypertensive patients

tations of pill counts. Clinical Pharmacology and Therapeutics, 46, 169-176. 29. Engstrom FW. (1988) Depression resistant to tricyclic antidepressants. British Medical]ournal, 297, 1130. 30. Cramer JA, Mattson RH, Prevey ML, Scheycr RD, Ouellette VL. (1989) How often is medication taken as prescribed. Journal of the American Medical Association,

hom a rural Midwest area. The]ournal of Family Practice,

261,3273-3277. 31. Fedder DO. (1982) Managing medication and com-

32.

33.

34.

35.

pliance: physician-pharmacist-patient interaction. lournal of the American GeriatricsSociety, 30, S113-Sl17. Sackett DL, Snow JC. (1979) The magnitude of compliance and noncompliance. In: Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 11-22. The Johns Hopkins University Press, Baltimore. Burrell CD, Levy RA. (1985) Therapeutic consequences of noncompliance. In: lmproving Medication Compliance: Proceedings of a Symposium in Washington, D.C., November 1, 1984, by the National PharmaceuticalCouncil, pp. 7-16. National Pharmaceutical Council, Reston Virginia. Dirks JF, Kinsman RA. (1982) Nondichotomous patterns of medication use: the yes-no fallacy. Clinical Pharmacology and Therapeutics, 31,413-417. Hogarty GE, Goldberg SC, the Collaborative Study Group (1973) Drugs and sociotherapy in the aftercare of the schizophrenicpatients. Archives of General Psychiatry, 28,54-64.

HV,Shepler L. (1986) Medication noncompliance in schizophrenia: codification and update. Bulletin of the American Academy of Psychiatry and Law, 14, 105-122. Johnson DAW, Freeman H. (1972) Long-acting tranquilhers. Practitioner, 208, 395-400. JamisonKR, Gemer RH, GoodwinFK. (1979)Patient and physician attitudes toward lithium. Archives of General Psychiatry, 36,866-869. Danion JM, Neunreuther C, Kreiger-FinanceF, Limbs JL, Singer L. (1987) Compliance with long-term lithium treatment in major affective disorders. Pharmacopsychiatrica, 20,230-231. Bech P, Vendsorg PB, Rafaelsen 01. (1976) Lithium maintenance treatment of manic melancholic patients: its role in daily routine. Acta Psychiatrica Scandinavica,

17,619-625. 44. Dodrill CB, Batzel LW, Wilensky AJ, Yerby MS. (1987) The role of psychosocial and hancial factors in medi-

cation noncompliancein epilepsy. 1nternationalJouml of Psychiatry in Medicine, 17, 143-154. 45. Eisler J, Mattson RH. (1975) Society proceedings: compliance in anticonvulsant drug therapy. Epilepsia, 16, 203. 46. Rodin EA. (1972) Medical and social prognosis in epilepsy. Epilepsia, 13, 121-131. 47. Shope J. (1978) Adherence to prescribed regimens: charac-

48.

49. 50. 5 1.

52.

36. Young JL, Zonana

37. 38.

39.

40.

53,7041. 41. Caldwell JR, Cobb S, Dowling MD, de Jongh D. (1970)

The drop-out problem in antihypertensive therapy. lournal of Chronic Diseases, 22,579-592. 42. Wilber JA, Barrow JG. (1972) Hypertension-a community problem American lournal of Medicine, 52, 653663.

53.

54. 55.

56.

teristics of noncompliers and educational intervention to improve compliance. PhD dissertation, Wayne State University, Detroit, Michigan. Hulka BS. (1979) Patient-clinician interactions and compliance. In: Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 63-77. The Johns Hopkins University Press, Baltimore. Kasl SV. (1975) Issues in patient adherence to health care regimens.]ournal of Human Stress, 1,5-18. Porter AMW. (1969) Drug defaulting in a general practice. British Medical]ournal, 1,218-222. Haynes RB. (1979) Determinants of compliance: the disease and the mechanics of treatment. In: Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 49-62. The Johns Hopkins University Press, Baltimore. Sbarbaro JA. (1985) Strategies to improve compliance with therapy. The American ]ournu1 of Medicine, 7 9 (SUPPI.6A). 34-37. Hussar DA. (1985) Improving patient compliance-the role of the pharmacist. In: Improving Medication Compliance: Proceedings of a Symposium in Washington, D.C., November 1, 1984 by the National Pharmaceutical Council, pp. 17-34. National Pharmaceutical Council, Reston, Virginia. Hecht AB. (1974) Improving medication compliance by teaching outpatients. Nursing Forum, 13, 112-129. Nessman DG, Camahan JE, Nugent CA. (1980) Increasing compliance: patient-operated hypertension groups. Archives of Internal Medicine, 140, 1427-1430. Zismer DK, Gillum RF, Johnson CA, Becerra J, Johnson TH. (1982) Improving hypertension control in a private medical practice. Archives of Internal Medicine, 142,

297-299. 57. Haynes RB. (1979) Strategies to improve compliance

with referrals, appointments, and prescribed medical regimens. I n Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 121-143. The Johns Hopkins University Press, Baltimore. 58. Cole P, Emmanuel Sr. (1971) Drug consultation: its significance to the discharged hospital patient and its

294

L. Stockwell Morris and R. M. Schulz

relevance as a role for the pharmacist. American Journal of Hospital Pharmacy, 28,954-960. 59. Moms LA, Halperin JA. (1979)Effects of written drug information on patient knowledge and compliance a literature review. American Journal of Public Health, 69, 47-52. 60. Mullen PD, Green LW, Persinger GS. (1985)Clinical trials of patient education for chronic conditions: a comparative meta-analysis of intervention types. Preventive Medicine, 14, 753-781. 61. Gibbs S,Waters WE, George CF. (1989)The benefits of prescription information leaflets.BritishJournalof Clinical Pharmacology, 27, 723-739. 62. Robinson GL, Gilbertson AD, Litwack L. (1986-7)The effects of a psychiatric patient education to medication program on post-discharge compliance. Psychiatric Quarterly, 58,113-118. 63. Myers ED, Calvert EJ. (1984)Information, compliance and side-effects : a study of patient on antidepressant medication. British Journal of Clinical Pharmacy, 17, 21-25. 64. Brown CS, Wright RG, Christensen DB. (1987)Association between type of medication instruction and

patients’ knowledge, side effects, and compliance. Hospital and Community Psychiatry, 38,5540. 65. Haynes RB. (1985)A critical review of interventions to improve compliance with special reference to the role of physicians. In: Improving Medication Compliance:Proceedings of a Symposium in Washington, D.C., November I, 1984, by the National Pharmaceutical Council, pp. 45-57. National Pharmaceutical Council, Reston, Virginia. 66. Dunbar JM, Marshall GD, Hovel1 MF. (1979)Behavioral strategies for improving compliance. In: Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 174-190. The Johns Hopkins University Press, Baltimore. 67. Gabriel M, Gagnon JP, Bryan CK. (1977)Improved patient compliance through use of a daily drug reminder chart. American Journal of Public Health, 67,968-969. 68. Simkins CV, Wenzloff NJ. (1986)Evaluation of a computerized reminder system in the enhancement of patient medication refill compliance. Drug Intelligence and Clinical Pharmacy, 20, 799-802. 69. Eshelman FN, Fitzloff J. (1976)Effects of packaging on patient compliance with an antihypertensive medication. Current Therapeutic Research, 20,215-219. 70. Linkewich JA, Catalan0 RB, Flack HL. (1974)The effect of packaging and instruction on outpatient compliance with medication regimens. Drug Intelligence and Clinical Pharmacy, 8,10-15. 71. Wong BSM, Norman DC. (1987)Evaluation of a novel medication aid, the calendar blister-pak, and its effect on drug compliance in a geriatric outpatient clinic. Journal of the American Geriatrics Society, 35,Zl-26.

72. Peterson GM, McLean S, Millingen KS. (1984) A randomised trial of strategies to improve patient compliance with anticonvulsant therapy. Epilepsia, 25, 412-4 17. 73. Ascione FJ, Shimp LA. (1984)The effectiveness of four education strategies in the elderly. Drug Intelligence and Clinical Pharmacy, 18,926-931. 74. Swain MA, Steckel SB. (1981)Influencing adherence among hypertensives. Research in Nursing and Health, 4, 213-222. 75. Smith DL. (1989) Compliance packaging: a patient education tool. American Pharmacy, NS29, 126-137. 76. Rosenstock IM. (1966)Why people use health services. Milbank Memorial Fund Quarterly, 44,94-127. 77. Becker MH, Maiman LA. (1975)Sociobehavioral determinants of compliance with health and medical care recommendations. Medical Care, 13,10-24. 78. Taylor DW. (1979)A test of the health belief model in hypertension. I n Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 103-109. The Johns

Hopkins University Press, Baltimore. 79. Ried LD, Christensen DB. (1988)A psychosocial perspective in the explanation of patients’ drug-taking behavior. Social Science and Medicine, 27,277-285. 80. Eraker SA, Kirscht JP, Becker MH. (1984)Understanding and improving patient compliance. Annals of Internal Medicine, 100,258-268. 81. Nagy VT, Wolfe GR. (1984)Cognitive predictors of compliance in chronic disease patients. Medical Care, 22,912-921. 82. Schlenk EA, Hart LK. (1984) Relationship between health locus of control, health value, and social support and compliance of persons with diabetes mellitus. Diabetes Care, 7,566-574. 83. Stanton AL. (1987) Determinants of adherence to medical regimens by hypertensive patients. Journal of Behavioral Medicine, 10,377-394. 84. Haynes RB. (1982)Improving patient compliance: an empirical view. In: Adherence, Compliance and Generalization in Behavioral Medicine ed Stuart RB, pp. 56-78.

BrunnerIMazel Publishers, New York. 85. Haynes RB, Taylor DW, Snow JC, Sackett DL. (1979)

Appendix I: annotated and indexed bibliography on compliance with therapeutic and preventive regimens. In: Compliance in Health Care eds Haynes RB, Taylor DW, Sackett DL, pp. 337-474. The John Hopkins University Press, Baltimore. 86. Bruer JT. (1982)Methodological rigor and citation frequency in patient compliance literature. American Journalof Public Health, 72,1119-1123. 87. Trostle JA. (1988)Medical compliance as an idealogy. Social Science and Medicine, 27, 1299-1308. 88. Conrad P. (1987)The noncompliant patient in search of autonomy. Hustings Center Report, August, 15-17.

Patient compliance 89. Stimson GV. (1974)Obeying doctor's orders: a view from the other side. Social Science and Medicine, 8, 97-104. 90. Conrad P. (1985)The meaning of medications another look at compliance. Social Science and Medicine, 20, 29-3 7. 91. Chubon SJ. (1989)Personal descriptions of compliance by rural Southern Blacks: An exploratory study. The Journal of Compliance in Health Care, 423-38. 92. Arluke A. (1980)Judging drugs: patients' conceptions of therapeutic efficacy in the treatment of arthritis. Human Organization, 39,84-88. 93. Gabe J, Thorogood N. (1986)Prescribed drug use and the management of everyday life: the experiences of black and white working-class women. Sociological Review, 34,737-772. 94. Gordis L, Markowitz M, Lilienfeld AM. (1969)Why patients don't follow medical advice: a study of children on long-term antistreptococcal prophylaxis. Journal of Pediatrics, 75,957-968. 95. Lipman RS, Rickels K, Uhlenhuth EH, Park LC, Fisher S. (1965)Neurotics who fail to take their drugs. British Journalof Psychiatry, 3,1043-1049. 96. Allen EA, Stewart M, Jeney P. (1964)The efficiency of post-sanatorium management of tuberculosis. Canadian Journal of Public Health, 55,323-333. 97. Drolet GJ, Porter DE. (1949) Why do patients in tuberculosis hospitals leave against medical advice? N Y TBC Health Association, 1-64. 98. Schwartz D, Wang M, Zeitz L, Goss ME. (1962)Medi-

cation errors made by elderly, chronically ill patients. American lournal of Public Health, 52,2018-2029.

295

99. Hulka B, Cassel J, Kupper L, Burdette J. (1976)Com-

munication, compliance and concordance between physicians and patients with prescribed medications. American Journal of Public Health, 66,847-853. 100. Hulka B, Kupper L, Cassel J, Efird R, Burdette J. (1975)Medication use and misuse physician-patient discrepancies. Journalof Chronic Diseases, 28, 7-21. 101. Hemminla E, Heikkila J. (1975)Elderly people's compliance with prescriptions and quality of medication. Scandinavian Journalof Social Medicine, 3,87-92. 102. Hedstrand H, Aberg J. (1976)Treatment of hypertension in middle-aged men. Acta Medica Scandinavica, 199,218-288. 103. Sackett DL, Haynes RB, Gibson ES, et al. (1975) Randomized clinical trial of strategies for improving medication compliance in primary hypertension. Lancet, 1,1205-1207. 104. Haynes RB, Sackett DL, Gibson ES, et al. (1976)

Improvement of medication compliance in uncontrolled hypertension. Lancet, 1,1265-1268. 105. Haynes RB. (1985)A critical review of interventions to improve compliance with special reference to the role of physicians. I n Improving Medication Compliance:koceedings of a Symposium in Washington, D.C., November I, 1984 by the National Pharmaceutical Council, pp. 17-34. National Pharmaceutical Council, Reston, Virginia. (Quote Logan, AG) 106. Takala J, Niemela N, Rosti J, Sivers K. (1979)Improving compliance with therapeutic regimens in hypertensive patients in a community health center. Circulation, 59, 540-5 43.