Abstract
This study tested if habit strength for taking oral hypoglycemic medication moderated the association between poor mental health symptoms (i.e. depressive symptoms, diabetes distress) and unintentional non-adherence in 790 adults with type 2 diabetes. Data were cross-sectional. A habit strength × depressive symptom interaction was observed. Depressive symptoms were negatively associated with adherence if habit was weak or of average strength; no association was observed if habit was strong. A similar interaction emerged when diabetes distress was examined. Results suggest that habit strength might operate as a buffer in the association between poor mental health symptoms and medication adherence.
Self-management regimens for many chronic health conditions involve regularly enacted behaviors. However, adherence to these regimens is often hindered by poor mental health (Sumlin et al., 2014). Non-adherence may be unintentional (i.e. forgetting) or intentional (i.e. actively deciding to not follow the regimen) (Wroe, 2002). In the context of type 2 diabetes, diabetes distress and depressive symptoms are relatively prevalent mental health issues (Fisher et al., 2008). Diabetes distress refers to psychological burden rooted in worries, concerns, and fears that are specific to diabetes (e.g. potential for complications) (Polonsky et al., 2005) and it is negatively associated with medication adherence (Gonzalez et al., 2015). Similarly, depressive symptoms tend to be inversely associated with adherence to self-management behaviors, including taking medication, among people with type 2 diabetes (Ciechanowski et al., 2000). Posited mechanisms for this association have been varied and have included the memory impairments that often accompany depressive symptoms and may manifest as forgetting to take medication (DiMatteo et al., 2000).
Habits are processes in which a contextual cue generates an automatic impulse to perform a specific behavior due to a learned association between the cue and the behavior (Gardner, 2015). Accordingly, strong habits are automatic responses that are activated by recurring contextual cues (Wood and Rünger, 2016). Habit strength is positively associated with medication adherence (Phillips et al., 2013). Because strong habits are automatic, they operate independently of, and can override, the effortful, explicit (i.e. deliberate) processes that tend to operate within conscious awareness and can shape behavior, such as willpower and intention (Wood and Rünger, 2016). In fact, evidence suggests that habit strength moderates the association between some of these deliberate processes and behavior. For example, intentions tend to predict physical activity and dietary behaviors if habits for these behaviors are weak, but intentions are considerably less predictive if habits are strong (Gardner et al., 2011). Similarly, strong habits predict behavior when willpower is reduced (Neal et al., 2013). Given that some of the pathways through which mental health symptoms are thought to influence medication adherence are deliberate (i.e. remembering), habit strength may also moderate the association between symptoms of poor mental health and adherence.
The purpose of this study was to determine if habit strength for an important diabetes self-management behavior—taking oral hypoglycemic medication—moderates the association between symptoms of poor mental health (i.e. depressive symptoms and diabetes distress) and adherence in adults with type 2 diabetes; only unintentional non-adherence is considered. An interaction was hypothesized such that poor mental health symptoms were expected to be negatively associated with adherence if habit was weak, but an association was not expected if habit was strong.
Methods
Design
Data came from the most recent wave (2015–2016) of the Evaluation of Diabetes Treatment (EDIT) study, which is a cohort study of community-dwelling residents of the province of Quebec, Canada (Smith et al., 2013). Data are self-reported and collected annually via telephone interview. Eligibility criteria for participation in EDIT at baseline were: a self-reported diagnosis of type 2 diabetes by a doctor within the last 10 years, a self-report of not currently using insulin, and being aged 40–75 years. Habit strength for taking oral medication to control one’s diabetes, or, if not applicable, for following a diet (n = 33) was measured for all participants. Habit strength for insulin use was also measured among individuals who reported recently transitioning to insulin therapy (n = 28). Adherence was measured only for oral medication. The measures described below were the only instruments used to assess habit strength and adherence. Individuals included in the present analyses responded affirmatively to the question, “In the past month, did you take pills to control your diabetes?” and completed the predictor and outcome variables (described below). The study protocol was approved by the Research Ethics Committee of the Douglas Mental Health University Institute, an affiliate of McGill University. Participants provided informed consent.
Participants
On average, participants (n = 790) had been diagnosed with type 2 diabetes for 8.82 years (standard deviation (SD) = 3.14) and were 64.05 years old (SD = 8.20). Most participants identified as Caucasian (94.4%) and reported being married or living as married (67.1%). Nearly half were women (49.2%) and about one-third (32.2%) reported working full- or part-time. Based on self-reported weight and height, 41.5% of the sample were classified as overweight (25.0 ⩽ body mass index (BMI) < 30) and 43.8% were classified as obese (BMI ⩾ 30.0). In all, 12.2 % reported also using insulin in the past year, and 33 met the criteria for major depressive syndrome (MDS; described below).
Measures
Habit strength for taking oral hypoglycemic medication
Habit strength for taking oral hypoglycemic medication was assessed with the automaticity subscale of the Self-Reported Index of Habit Strength (Verplanken and Orbell, 2003) because it is the most parsimonious measure of the habit strength for which reliability and validity has been demonstrated (Gardner et al., 2012). Four items asked respondents to rate the extent to which taking pills to control their diabetes is performed automatically (e.g. Taking pills to control my diabetes is something I do without thinking) on a 7-point scale (1 = Strongly Disagree, 7 = Strongly Agree). Item scores were averaged to create a composite score (α = .76).
Depressive symptoms
Depressive symptoms were assessed with the Patient Health Questionnaire (Kroenke et al., 2001). The 9-item self-report measure queries symptoms of depression experienced in the past 2 weeks. Items are rated on a scale ranging from 0 (Not at all) to 3 (Nearly every day) and summed (α = .80). Cases of MDS are identified by accounting for the diagnostic criteria for major depressive disorder; respondents must endorse either dysphoria or anhedonia and at least four of the other items on more than half the days in the past 2 weeks.
Diabetes-related distress
Diabetes-related distress was measured with the Diabetes Distress Scale (Polonsky et al., 2005), which is a 17-item self-report scale that assesses emotional burden, physician-related distress, regimen-related distress and interpersonal distress. Respondents rate the extent to which potential sources of distress have bothered them on a 6-point scale (1 = Not a problem, 6 = A very serious problem). Item scores are aggregated (α = .94).
Adherence
An affirmative response to the question “In the past month, did you take pills to control your diabetes?” was immediately followed by the outcome measure: “Did you ever forget to take your medication?” (1 = Often, 2 = Fairly often, 3 = Sometimes, 4 = Rarely, 5 = Never).
Covariates
Demographic covariates included age, sex, current marital status (married/living as married; widowed/divorced/separated/never married), and current work status (work full/part time; not working/retired). Diabetes-related covariates included insulin use in the past year (yes; no), diabetes duration in years, and number of diabetes complications measured with the Diabetes Complications Index, which is a 17-item self-report measure that assesses and sums complications (e.g. coronary artery disease and retinopathy) (Fincke et al., 2005).
Analysis
Moderation was tested with the PROCESS macro for SPSS (model 1), which uses ordinary least squares regression (Hayes, 2013). Depressive symptoms and diabetes distress were analyzed separately. Unadjusted models and models adjusting for covariates were tested. In all models, the mental health variables and habit strength were mean centered. Due to missing covariate data, 36 participants were excluded from the adjusted models. Sensitivity analyses were conducted with MDS as the predictor variable in unadjusted and adjusted models.
Results
Correlations between mental health measures, habit strength, adherence, and covariates, as well as descriptive statistics for the focal variables are presented in Supplemental Table 1. In the unadjusted model testing the interaction between depressive symptoms and habit strength, significant main effects of depressive symptoms and habit strength were qualified by a significant interaction, ΔR2 = .01, ΔF(1, 786) = 8.41, p = .004 (Table 1). The interaction remained significant after adjusting for covariates, ΔR2 = .01, ΔF(1, 743) =5.62, p = .018 (Table 1). Analysis of the simple slopes in the adjusted model suggested that depressive symptoms were negatively associated with adherence if habit was relatively weak (1 SD below mean), b = −.06, standard error (SE) = .01, p < .001, or of average strength, b = −.04, SE = .01, p < .001, but not if habit was relatively strong (1 SD above mean), b = −.02, SE = .01, p = .08. A consistent pattern emerged in the sensitivity analysis; significant main effects were qualified by a significant MDS × habit strength interaction in unadjusted, ΔR2 = .01, ΔF (1, 786) = 8.35, p = .004, and adjusted models, ΔR2 = .006, ΔF (1, 743) = 4.46, p = .04 (Table 1). Analysis of the simple slopes in the adjusted model suggested that MDS was inversely associated with adherence if habit was relatively weak (1 SD below mean), b = −.87, SE = .27, p = .002, or of average strength, b = −.49, SE = .17, p < .001, but not if habit was relatively strong (1 SD above mean), b = −.10, SE = .22, p = .64.
Unadjusted (n = 790) and adjusted (n = 754) models regressing adherence to oral hypoglycemic mediation on mental health outcomes (depressive symptoms, diabetes distress), habit strength and habit strength × mental health interaction.
SE: standard error; CI: confidence interval.
Covariates in adjusted model include age, sex, marital status, work status, diabetes duration, insulin use in past year, and number of diabetes complications. A total of 36 participants were excluded from adjusted analyses due to missing data on marital status (n = 2), diabetes duration (n = 1), and diabetes complications (n = 33). In sensitivity analysis, 33 cases of MDS were identified; MDS coded as: 0 = not a case, 1 = case.
Similarly, in the unadjusted model testing the interaction between habit strength and diabetes distress, significant main effects of diabetes distress and habit strength were qualified by a significant interaction, ΔR2 = .01, ΔF(1, 786) = 10.09, p = .002 (Table 1). The interaction remained significant after adjusting for covariates, ΔR2 = .01, ΔF(1, 742) = 8.44, p = .004 (Table 1). Analyses of the simple slopes in the adjusted model suggested that diabetes distress was negatively associated with adherence if habit was relatively weak (1 SD below mean), b = −.31, SE = .08, p < .001, or of average strength, b = −.16, SE = .05, p = .002, but not if habit was relatively strong (1 SD above mean), b = −.009, SE = .07, p = .90.
Discussion
This study suggests that habit strength for taking oral hypoglycemic medication may moderate the association between symptoms of poor mental health and unintentional medication non-adherence in people with type 2 diabetes; higher levels of depressive symptoms and diabetes distress were associated with poorer adherence if habit was weak or of average strength, but not if habit was strong. Existing research has demonstrated that habit strength moderates the association between intentions and behavior (Gardner et al., 2011). The current study offers preliminary evidence suggesting (a) that this pattern of results may extend to the domain of mental health symptoms, though effects were very small, and (b) that strong self-management habits may buffer against the deleterious effects of poor mental health on self-management behaviors in people with type 2 diabetes. Although the very small effect sizes (i.e. the interaction accounted for an additional 1% of the variance in adherence) indicate that these observational findings do not have immediate clinical relevance, this study points to habit as a process that might be leveraged in future experimental and intervention work.
Depressive symptoms and diabetes distress are relatively prevalent in individuals with type 2 diabetes (Fisher et al., 2008) and are inversely associated with medication adherence and other self-management behaviors (Ciechanowski et al., 2000; Gonzalez et al., 2015). In this study, habit strength was not correlated with mental health symptoms, which suggests that one may hold strong habits while experiencing elevated levels of mental health symptoms. Therefore, future research may wish to examine if implementing strategies that promote the development of strong self-management habits can produce sustained adherence even if mental health symptoms develop. Because strong habits form when the target behavior is repeatedly performed in the presence of the same cue (e.g. taking medication immediately after brushing teeth or pouring morning coffee) (Lally et al., 2010), such interventions could focus on promoting taking medication repeatedly in the same context. This might be achieved by encouraging planning, consistency, and selection of a salient environmental cue, or making adherence rewarding or satisfying (Lally and Gardner, 2013). This avenue of research may be particularly useful because individuals are likely to perform strongly habitual behaviors under conditions that are connected to the psychological burden of mental health problems, such as reduced willpower (Neal et al., 2013) and stress (Wood and Rünger, 2016).
Strengths of this study include the use of a large, community-based sample of people with type 2 diabetes, established measures of mental health, and a sensitivity analysis. This study also has limitations, which should be addressed in future research. Medication adherence was measured with a single self-report item that did not explicate a time frame; the time frame had to be inferred from the preceding question. These findings are limited to unintentional non-adherence (i.e. forgetting). Future research should determine if this pattern of results extends to intentional non-adherence because some of the other mechanisms through which mental health is posited to influence adherence may also shape intentional non-adherence (i.e. beliefs about medicine’s effectiveness) (DiMatteo et al., 2000). Moreover, the statistical model was structured in accordance with the rationale for the study (i.e. habit strength moderates the association between mental health and adherence); however, directionality cannot be ascertained given the cross-sectional design. It is therefore possible that the alternate model (i.e. mental health moderates the association between habit strength and adherence) more accurately describes this pattern of results. The proportion of variance in adherence that was accounted for by the interaction was also very small. Nonetheless, this study points to the promise of a novel construct—habit strength—in the study of associations between mental health and health behavior.
Supplemental Material
Supplemental_table_1 – Supplemental material for Habit strength as a moderator of the association between symptoms of poor mental health and unintentional non-adherence to oral hypoglycemic medication in adults with type 2 diabetes
Supplemental material, Supplemental_table_1 for Habit strength as a moderator of the association between symptoms of poor mental health and unintentional non-adherence to oral hypoglycemic medication in adults with type 2 diabetes by Rachel J Burns, Sonya S Deschênes, Bärbel Knäuper and Norbert Schmitz in Journal of Health Psychology
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: R.J.B. is supported by a Canadian Institutes of Health Research (CIHR) fellowship (201411MFE-338860 FRN-142923). S.S.D. is supported by a CIHR fellowship. The Evaluation of Diabetes Treatment study was funded by CIHR Research Grant MOP-106514.
Supplemental Material
Supplemental Material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
