Using a Social Ecological Perspective to Examine Medication Adherence Among Persons with Severe Mental Illness: A Randomized Factorial Design

Background and Purpose: For adults with severe mental illness, psychotropic medication reduces psychiatric symptoms and risk for re-hospitalizations; however, estimated rates of non-adherence vary from 20-95%. Non-adherence is related to adverse events, including homelessness, violence, substance use, and greater risk of suicide. To develop efficacious strategies to increase medication adherence, it is essential to understand factors associated with non-adherence. The study was designed to examine individual, interpersonal, and organizational factors that were hypothesized to contribute to decision-making about taking medications as prescribed, which have not previously been examined in concert.

Methods: A randomized factorial survey was developed to investigate factors that influence decision-making about medication adherence. Vignettes were constructed using placeholders for variables of interest: individual-level (level of personal interest, side effects, effect on symptoms), interpersonal-level (pressure from support system), and organization-level (locus of decision-making, perceived coercion). Values for each variable were randomly selected and inserted into the placeholders, creating varying combinations of variable values. Respondents were presented with three vignettes and asked the same set of questions about each: “How likely would you be to take medications as prescribed?”, and “How much pressure would you feel to take medications as prescribed?”. Analysis examined the extent to which changes in variable values explained change in outcome variables (perceived pressure and intention to take medication).

Participants were recruited from psychiatric in-patient and partial hospitalization programs at one urban hospital in the Pacific Northwest. 60 eligible adults agreed to participate, for a response set of N=180 vignettes. Outcome variables were dichotomized, and hierarchical logistical regressions for each outcome variable were conducted in Stata, controlling for multiple responses per respondent.

Results: Models 1 (p=.041) and 2 (p=.001) analyzed the degree to which personal characteristics predicted perceived pressure and intention to take medication, respectively. Side effects (sexual [p=.001], weight gain [p=.008], drowsiness [p=.009]) were associated with 3.7-4.6 times higher odds of perceiving high pressure for medication adherence; side effects (weight gain [p=.016], drowsiness [p=.005]) and ineffective symptom management (p=.006) were associated with 2.7-5.8 times higher odds of having no intention to take medications. Models 3 and 4 for perceived pressure (p=.015) and intention to take medications (p=.007) added interpersonal and organizational factors. Side effects (sexual [p=.003], weight gain [p=.017], drowsiness [p=.015]) and coercion (p=.001) were associated with a 2.8 – 4.2 times higher odds of perceiving high pressure; side effects (weight gain [p=.015], drowsiness [p=.005]) and ineffective symptom management (p=.012) were associated with 2.5-6.0 times higher odds of having no intention to take medication.

Conclusions and Implications: Person-level characteristics (side effects and poor medication effectiveness) and organization-level factors (perceived coercion) were associated with a higher risk of feeling pressured to take medication. Person-level characteristics (side effects and poor medication effectiveness) were associated with increased odds of having no intention to take medication. Findings suggest that mental health social workers using coercive practices may increase perceived pressure but fail to increase adherence; selecting a medication that effectively reduces symptoms, while working to mitigate side effects, are strategies more likely to lead to medication adherence.