Published Online: December 01, 2008
Paul F. Cook, PhD; Suzie Emiliozzi, RN; Corey Waters, MBA; and Dana El Hajj, BSN, RN
Objective: To determine whether ScriptAssist, a telehealth nursing program using psychological techniques, reduced emergency department (ED) utilization and improved adherence among Medicaid health plan members with serious and persistent mental illness (SPMI).
Study Design: Nonrandomized controlled trial.
Methods: Of 210 eligible Medicaid health plan members with SPMI, 59 (28%) were contacted by phone and 51 (86%) participated. Participants received an average of 3.5 calls over 9 months, with 2.1 attempts per completed call. Participants had clinically significant levels of impairment; medication switching, polypharmacy, and medical comorbidities were common. Intervention group participants’ results were compared with those of nonparticipants to rule out regression to the mean, history, and maturation effects. Intervention group participants’ results also were compared with baseline data to rule out selection bias.
Results: Program participants had fewer ED visits during the intervention than a comparison group, and reduced their ED use and hospitalization rate compared with the previous year. Participants also had better medication adherence based on pharmacy and interview data.
Conclusions: Cognitive-behavioral and motivational- interviewing techniques can improve antipsychotic medication adherence. Telehealth may be a useful strategy for disseminating these evidence-based techniques. Lessons learned included the importance of real-time referral data, a need to address polypharmacy, and a need to overcome contact difficulties resulting from disease processes and “unknown caller” IDs. Despite these difficulties, using a disease management model, the program was feasible, and the reduced number of ED visits indicated potential cost-effectiveness.
(Am J Manag Care. 2008;14(12):841-846)
Telephone counseling improved antipsychotic medication adherence and reduced emergency department utilization in a managed Medicaid population.
- Disease management approaches can be a cost-effective way to help members with serious and persistent mental illness improve their self-management.
- Telehealth nursing is an effective method for delivering evidence-based adherence counseling approaches in community settings.
- Program participants had severe impairment, including a high prevalence of bipolar disorder and other mood disorders, complex treatments, and comorbid conditions; contacting and retaining these high-risk members required extra effort.
Despite the benefits of second-generation antipsychotic medications (SGAs) for persons with serious and persistent mental illness (SPMI), most people discontinue treatment within 1 year. Among persons with schizophrenia, 20% to 30% never start treatment,1 14% to 32% discontinue treatment within 3 months,2-4 and 66% to 72% stop within 2 years.5 Among Medicaid recipients, 12-month adherence is as low as 40%.6 Although earlier research showed no adherence advantage for SGAs versus older medications,6,7 some recent data suggest better adherence.8 Nonadherence risk factors include previous nonadherence, recent alcohol or drug use, clinical severity, past antidepressant use, and medication-related cognitive impairment.9-12
Persons with schizophrenia who stop medication are at risk for symptom recurrence, impaired functioning, increased emergency department (ED) use, and hospitalization.6,13-15 Repeated hospitalization is a substantial cost for persons with SPMI, approximately 40% of which is due to nonadherence.16 The average adherent individual has 21 to 35 fewer hospital days over 12 months posthospitalization,17 and even partial nonadherence increases this risk.18
Brief cognitive-behavioral19 and motivational-interviewing20,21 interventions have successfully increased SGA adherence, but may be difficult to replicate22 and are not widely available in community practice. Telephonic counseling is a feasible delivery method23-25 that previously was shown to improve antidepressant adherence26 and other chronic medication adherence.27
The current study used a disease management program called ScriptAssist to promote SGA adherence. Members were identified from a Medicaid health plan’s administrative data and proactively contacted by a registered nurse (RN) with mental health treatment experience. ScriptAssist programs23,26,27 provide individual psychological interventions targeted to participants’ readiness for change, based on the transtheoretical model.28 During each call, RNs assess patients’ treatment motivation and adherence concerns, and offer educational, empowering, and decision- support interventions based on the participant’s stage of change. They also may mail patients follow-up print materials that emphasize themes from the call. Participants received follow-up calls from the same nurse over time. Specific patient education materials were created for this SPMI program (eg, coordination of care, coping with symptoms, warning signs of relapse). RNs received 8 hours of training on the psychological counseling model, plus 16.5 hours of training on SPMI symptoms, treatments, adverse drug events, crisis management, and psychosocial factors related to SPMI such as work, relationship, and family issues. All RNs in the current study had previous experience working with mental health issues, as well as prior experience with the ScriptAssist counseling model. To ensure treatment fidelity, RNs received individual and group supervision on the counseling model throughout the study, and any deviations were addressed through individual training.
Persons with SPMI were identified from the membership of a voluntary, managed Medicaid health plan in a midwestern state during the first 6 months of 2006. Members were included if they (1) received a SGA prescription in the past 30 days, (2) were at least 18 years old, and (3) had current health plan eligibility. Members were excluded if they had a recorded diagnosis of dementia or substance dependence. The Figure shows the study flow.
Of 210 eligible members, 55 (26.2%) received a prescription for aripiprazole, 30 (14.3%) for ziprasidone, 49 (23.3%) for risperidone, 85 (40.5%) for quetiapine, and 33 (15.7%) for olanzapine. The total number of medications was greater than the number of members, which suggests switching and/ or polypharmacy. Members’ average age was 33.4 years (SD = 8.2 years), and 83.8% (n = 176) were female. The high percentage of women reflects the fact that members came from a Medicaid plan for women and children.
Although multiple calls and letters were used to reach all 210 eligible members, only 59 (28%) were contacted for enrollment. The most common reason members were not reached was lack of a valid telephone number. All members were asked for informed consent to participate. Three members declined, and 5 declined further contact after the first call. Seven participants opted out later, for an overall total of 25% opting out. This was higher than the 1% to 2% refusal rates seen in ScriptAssist programs for chronic physical diseases.27Participants’ self-reported diagnoses included bipolar disorder (n = 22), depression (n = 18), anxiety disorders (n = 10), schizophrenia (n = 2), schizoaffective disorder (n = 1), hallucinations and delusions (n = 3), antisocial personality (n = 1), and borderline personality (n = 1). Concomitant medications included psychostimulants (n = 7), sleep aids (n = 7), antianxiety medications (n = 11), antidepressants (n = 49), older-generation antipsychotics (n = 11), and mood stabilizers (n = 12). Surprisingly, only 8 potential adverse drug events were reported: fatigue (3 participants), itching/rash (1 participant), stomachache (1 participant), tremors (2 participants), and “just didn’t feel right” (1 participant). Medical comorbidities are common with SPMI,29,30 and participants reported comorbid hypertension, diabetes, asthma, emphysema, migraine, sleep problems, seizures, and pain. Several noted trauma histories including physical abuse, sexual abuse, or motor vehicle accident.
Three RNs made multiple attempts to reach members (mean of 2.1 attempts per completed call), and each patient worked with the same RN over time. In initial calls the RN administered a proprietary screening tool, originally developed in an outpatient psychiatric population, to predict members’ nonadherence risk27; 90% of participants were screened “at risk” for nonadherence. At-risk participants received followup calls (mean of 7.2 call attempts; mean of 3.5 calls completed) over an average of 4.4 months, with an average call length of 11 minutes (average total contact per participant was 38.5 minutes). Low-risk participants received a toll-free number plus 1 follow-up call at 6 months. During each call, the RN offered cognitive-behavioral counseling or motivationalinterviewing interventions based on the participant’s readiness for change and individual barriers to adherence. RNs mailed follow-up written materials to 45 participants (88%). A written progress note was sent to the participant’s health plan case manager after each call. Four participants reported potential serious adverse drug events; these persons were referred to their primary care providers. Participants did not receive incentives. Data analysis was approved by the Colorado Multiple Institutional Review Board.
Administrative Data. This study’s primary outcome was ED utilization based on health plan administrative data, which is generally considered a valid archival measure of service use. Hospitalizations and recorded diagnoses also were obtained from claims data.
Adherence Measures. Medication adherence was measured using health plan pharmacy data. Pharmacy records are generally considered valid and correlate with other adherence measures.31-33 Self-reported adherence also was measured, using the question “how many days in the last week have you taken your medication as prescribed?” The number of days was divided by 7 to calculate a percentage. Nonblinded interviews were conducted by the RN providing the intervention. Nonjudgmental interviews provide valid adherence data,34-36 and the current measure has shown 75% agreement with pharmacy data.23
All analyses were conducted using SPSS 15.0 (SPSS Inc, Chicago, IL). ED utilization was defined as the number of ED visits by eligible members, divided by months of eligibility. Rates were multiplied by 12 to calculate per member per year (PMPY) ED rates. There were no missing observations in the administrative data set, and a conservative intent-totreat analysis was used.
For pharmacy-based adherence, members were considered adherent if they had no more than a 14-day gap between the end of one SGA prescription (previous fill date plus days supply) and the next fill date for the same or any other SGA. Complete pharmacy records were available, and an intent-to-treat analysis was used. Prescriptions for non-SGA medications were not included, so members who switched to older-generation antipsychotics were counted as “nonadherent.” For self-reported adherence, participants were considered adherent if they reported taking medication as prescribed at least 80% of the time. Self-reported adherence rates were analyzed using all available data for each month of treatment; participants who could not be recontacted were excluded from subsequent analyses. However, participants who indicated at any point that they had stopped treatment were retained in subsequent analyses.
PDF is available on the last page.