Predictors of medication competence in schizophrenia patients

1. Introduction 

Medication compliance is a complex, prevalent and clinically severe problem in the schizophrenia patient population (Ascher-Svanum et al., 2006). Cognitive psychology provides a perspective on this problem with the longstanding distinction between competence and performance (Chomsky, 1965, Le Corre et al., 2006). Competence refers to a person's knowledge and ability whereas performance refers to actual behavior in real-life situations. Only under the most ideal conditions, with no interference from internal or external sources, is performance a direct reflection of competence. In the context of schizophrenia and treatment adherence, ideal conditions seldom occur. Nonetheless, competence is a prerequisite for effective performance even though it is no guarantee that such performance will occur. It follows that knowing how to self-administer oral medications is an important and necessary, but not sufficient, condition for actual drug compliance.

Consideration of medication management ability as a distinct competency has grown with development of the Medication Management Ability Assessment (MMAA; Patterson et al., 2002). The MMAA was adapted for schizophrenia patients from an earlier role-play measure that evaluated medication skills in people with HIV infection (Albert et al., 1999). The task involves a simulated oral medication regimen whereby subjects receive points for self-administering several drugs in response to information on dose and frequency, time interval and contraindications. The initial study in patients with schizophrenia and schizo-affective disorder showed that the MMAA relates significantly to overall cognitive status, self-reported life quality, everyday functioning and prescription refill records. More recently, the relation between the MMAA and both cognitive ability and some aspects of medication adherence has been confirmed by an independent group of researchers (Pratt et al., 2006). There is also evidence that global cognitive impairment contributes uniquely, over and above demographic variables and symptoms, to the prediction of MMAA scores (Jeste et al., 2003). The association between the MMAA and measures of functional outcome (Patterson et al., 2002) implies that medication competence is more than a precondition for drug adherence and taps a broad potential for independent living and successful adjustment to illness.

However, the number of studies is small and several questions require further attention. It is unclear whether specific cognitive abilities, including verbal and non-verbal reasoning, working memory and verbal learning, contribute equally or differentially to the prediction of MMAA scores. Previous research used general screening measures to assess cognition (e.g., Patterson et al., 2002) or reported simple correlations for individual tests (e.g. Pratt et al., 2006) rather than regression results with multiple predictors. Second, the relationship between MMAA scores and concurrent symptoms is unclear, with only one study (Jeste et al., 2003) reporting relevant data. Third, little is known about relations between medication competence and community independence, although the MMAA varies with other role-play measures of functional capacity (Patterson et al., 2002). In addition, previous studies assessed middle-aged and older patients rather than an age range that more closely reflects the patient population. Accordingly, existing MMAA findings may not generalize to the broad population of patients with schizophrenia and related conditions.

The present study was undertaken to address the questions identified above. Furthermore, the potential relationship between MMAA scores and drug attitude and subjective response was of interest. It is known that measures of subjective response relate to adherence behavior (Hogan et al., 1983). However, treatment-related emotions and beliefs may also play a role in the development and demonstration of medication competence. Hence Jeste et al. (2003) reported a significant association between the MMAA and scores on the Drug Attitude Inventory (DAI; Hogan et al., 1983). The DAI measures how a patient feels on medication. Definitions of “drug attitude” have broadened recently to encompass aspects of subjective well-being and life quality as well as the more limited medication-related response indexed by the DAI (Voruganti and Awad, 2002). The present investigation examined this broadened definition of subjective treatment appraisal in relation to the MMAA. Accordingly, our basic questions were as follows: (1) To what extent do current symptoms and subjective treatment response predict medication competence?; (2) Does cognitive performance, in aggregate or in terms of selective abilities, add validity to this prediction?; and (3) Does medication competence relate to broader aspects of independence and adjustment?

2. Methods 

2.1. Subjects 

The sample comprised 147 male and female participants who met the following criteria: (1) diagnosis of schizophrenia or schizoaffective disorder (SCID-I/P, First et al., 2002) by DSM-IV criteria; (2) outpatient status; (3) age 18–65 years; (4) no history of serious neurological or endocrine disorder, including head trauma, epilepsy, Cushing's disease or thyroid disorder; (5) no concurrent DSM-IV diagnosis of substance abuse; (6) no history of developmental disability; (7) minimum English reading level of grade 6; (8) willingness and ability to sign informed consent; and (9) normal or corrected vision. Demographic and clinical characteristics are presented in Table 1. All but 2 patients were receiving anti-psychotic medication at the time of study enrollment, and 119 were treated with atypical and 16 with typical antipsychotic drugs.

Table 1. Demographic, clinical and cognitive characteristics of patients (n=147)

	Variable	Mean	S.D.
	Age (years)	41.31	9.28
	Education (SCID classification)a	3.41	1.44
	Positive and Negative Syndrome Scale (PANSS)
	Positive symptom score	20.10	6.07
	Negative symptom score	19.20	6.05
	General psychopathology score	41.84	9.10
	Total score	81.13	17.46
	Personal Evaluation of Transitions in Treatment (PETiT)	38.29	9.80
	Wide Range Achievement Test (3) Reading standard score	95.14	13.98
	Wechsler Adult Intelligence Scale (WAIS-III) age-scaled scores
	Vocabulary (verbal ability)	9.73	3.46
	Matrix Reasoning (visual reasoning)	9.09	3.27
	Letter–Number Sequencing (auditory working memory)	8.41	3.14
	Symbol Search (visual processing speed)	7.24	2.96
	California Verbal Learning Test (CVLT-II) total recall trials I–V	40.44	11.90
	MSIF Global Support Rating/7	3.80	1.42
	Medication Management Ability Assessment (MMAA) score/25	16.36	6.76

a Education based on SCID classification (1=grade 6 or less; 2=grade 7–12; 3=high school graduate; 4=partial college; 5=graduated from 2-year college; 6=graduated from 4-year college; 7=part professional/graduate school; 8=completed professional/graduate school).

Participants were recruited at the Community Schizophrenia Service (CSS) of St. Joseph's Healthcare Hamilton, the Hamilton Program for Schizophrenia (HPS), and the Toronto, Ontario branch of the Canadian Mental Health Association (CMHA). Recruitment methods included posted advertisements and flyers in patient areas. In addition, the project research assistant provided briefings and information to patients in each setting and clinical staff assisted in identifying potential candidates for the study. Participation included monetary compensation and written informed consent was obtained from all participants. The project was approved by the institutional review board at each research site and by York University.

3. Results 

The clinical sample was composed of patients ranging from 21 to 65 years of age (see Table 1). Patients were predominantly male (65%) and born in Canada (80%), with English as their first spoken language (82%) and a majority (56%) had completed high school. On the PANSS (inter-rater r=.95), the sample demonstrated positive and negative symptom levels that were within average limits for treated schizophrenia patients (Opler et al., 1999). In addition, PETiT scores were within the range for patients undergoing treatment as reported in the original validation study (Voruganti and Awad, 2002). In terms of cognitive performance, mean values were within low average – to average limits across all tests. The MSIF ratings demonstrated high inter-rater reliability (r=0.96) and indicated that, on average, patients received moderate levels of support in residential, work and educational settings. More specifically, 52% required from moderate to complete support and assistance in all settings. Finally, MMAA scores ranged from 3 to 25, indicating the occurrence of severely defective, but also highly proficient medication competence in the patient sample. The MSIF global support rating and MMAA score were significantly correlated (r=−0.37, P<0.001), reflecting an inverse relationship between support requirements and medication competence.

Pearson correlations also confirmed significant associations between MMAA scores and all of the cognitive measures, including WRAT-3 Reading level (r=0.42, P<0.001), WAIS-III vocabulary (r=0.44, P<0.001), WAIS-III Matrix Reasoning (r=0.43, P<0.001), WAIS-III Symbol Search (r=0.39, P<0.001), WAIS-III Letter–Number Sequencing (r=0.46, P<0.001) and CVLT Total (r=0.41, P<0.001). There were also findings of a significant, but mild inverse relationship between MMAA scores and PANSS General Psychopathology (r=−0.22, P<0.006), Positive (r=−0.20, P<0.008) and Negative (r=−0.21, P<0.007) scores. However, no relationship was found between PETiT scores and the MMAA (r=0.05, P>0.05). Only correlations with values≥r=0.21 met a conservative Bonferroni criterion for maintaining type I error at P<0.05. Values less than r=0.21 represent marginal trends or associations.

A multiple regression approach to the prediction of MMAA scores was applied in two stages as follows. First, a hierarchical method, whereby 13 variables were entered into an equation in three successive blocks (demographic, clinical, and cognitive) was used. Demographic predictors included age, sex and education; clinical predictors included PETiT and PANSS scores and cognitive predictors comprised WRAT-3, WAIS-III and CVLT-II scores. Validity was indexed with R2 and by the statistical significance of the increment in variance accounted for by each block of variables. Results of this procedure showed that the equation based solely on demographic variables failed to account for a significant amount of MMAA variance (F3, 143=2.6, P>0.05). Moreover, the addition of clinical predictors did not add significantly to validity (F4, 139=1.96, P>0.05). However, the cognitive block did increase validity significantly (F6, 133=7.69, P<0.001). The full model accounted for 33% of the variance in MMAA scores.

To assess the unique contribution of individual variables and determine the most efficient prediction model, a stepwise multiple regression analysis was conducted. Each of 13 variables was selected or eliminated from the equation based on its ability or inability to contribute significantly and uniquely to the prediction of MMAA scores. Unique validity was assessed with standardized regression (β) weights for each variable. This procedure reduced the set of predictors to a three-variable model comprised of WAIS-III Letter–Number Sequencing (β=0.22, P<0.05), CVLT-II word recall (β=0.23, P<0.01) and the WRAT-3 Reading score (β=0.23, P<0.01). Hence the most efficient model was comprised entirely of cognitive measures and accounted for 29% of the variance in MMAA scores.

4. Discussion 

This study confirms a considerable range of medication management ability in the schizophrenia population and suggests that the MMAA is applicable to young and middle-aged as well as older patients. In addition, MMAA proficiency is associated with relative independence in community living. However, we found no evidence that patients' feelings and attitudes regarding drug treatment (PETiT) vary with medication competence. Furthermore, positive and negative symptom levels and general distress (PANSS) are at best weak and non-specific correlates of the MMAA. At the same time, the results confirm the unique contribution of cognitive performance in predicting medication competence. In particular, abilities in reading (WRAT-3), auditory working memory (WAIS-III) and verbal memory acquisition (CVLT-II) emerged as key predictors of MMAA. Accordingly, patients high in medication competence have relatively preserved word recognition, and possibly high premorbid general cognitive abilities, as well as fairly strong working and acquisition memory skills for verbal material.

Does the identification of specific cognitive contributions to medication competence have implications for treatment adherence and rehabilitation? Recent evidence suggests that several aspects of cognitive performance including memory improve modestly, but significantly in response to remediation therapy (Penades et al., 2006). Hence it is possible that functional competencies including MMAA proficiency may also benefit from such therapies. Still, actual adherence behavior is complex (Velligan et al., 2006) and, unlike medication competence, clearly tied to symptom severity and subjective drug response (Yamada et al., 2006). Thus enhancing competence without addressing patients' clinical state and treatment experience is unlikely to produce major benefits in terms of adherence (Dolder et al., 2003). It seems likely that a combination of cognitive enhancement and training in specific competencies as well as strategies to address subjective experience and symptoms holds the greatest promise in terms of rehabilitation.

The current study suggests that the MMAA is useful in assessing medication competence across a wide range of demographic and clinical characteristics in the schizophrenia patient population. However, patients with concurrent diagnoses of substance abuse were not included, which places a limitation on generalization. In addition, correlational data do not allow for causal inference. Our results are empirical associations consistent with the idea that medication competence is mediated by word recognition, working and verbal memory. However, direct evidence of this mediation is still lacking. Moreover, a large amount of variance in medication competence remained unaccounted for, which implies that additional variables need to be considered in research on the MMAA.

In conclusion, medication competence is understandable as a cognitively based estimator of a patient's potential for functional independence. Our findings of significant relationships between the MMAA and cognitive performance and patients' support requirements across residential, occupational and education settings supports this broad interpretation. Accordingly, measures of medication competence may provide a useful tool in the development of effective strategies and programs to improve not only treatment adherence but also health management and community functioning in the schizophrenia patient population.