Of 32 enrolled participants, two patients were ineligible because they were not discharged to home. One intervention patient was never reached to perform the medication coaching and was unavailable for follow-up; thus, there were 19 patients in the intervention group and 10 in the control group. Characteristics in the two groups were similar, although those in the control arm were more likely to be white, to be slightly younger and to report having health insurance (Table
1). A greater percentage of subjects in the intervention arm had a stroke as opposed to a TIA. Intervention subjects were discharged with a somewhat higher median number of stroke secondary prevention medications (median = 5 medications vs control median = 4).
Table 1
Baseline Characteristics of Enrolled Subjects
Female | 12 (41.4 %) | 8 (42.1 %) | 4 (40.0 %) |
Median age (IQR) | 61 (52–69) | 61(55–70) | 59 (47–67) |
Race | | | |
African American | 13 (44.8) | 10 (52.6) | 3 (30) |
White | 16 (55.2) | 9 (47.4) | 7 (70) |
NIHSS, median (IQR) | 1(1–3) | 2 (1–4) | 1 (1–2) |
Stroke type | | | |
Ischemic stroke | 24 (82.2) | 17 (89.5) | 7 (70.0) |
TIA | 4 (13.8) | 2 (10.5) | 2 (20.0) |
ICH | 1 (3.5) | 0 | 1 (10.0) |
Medical history (or risk factors) | | | |
Prior stroke/TIA | 10 (34.5) | 7 (36.8) | 3 (30.0) |
Hypertension | 22 (75.9) | 16 (84.2) | 6 (60.0) |
Hyperlipidemia | 20 (71.4) | 14 (73.7) | 6 (66.7) |
Smoker | 14 (48.3) | 8 (42.1) | 6 (60.0) |
Diabetes | 6 (20.7) | 3 (15.8) | 3 (30.0) |
Atrial fibrillation | 3 (10.3) | 2 (10.5) | 1 (10.0) |
Number of discharge meds, median (IQR) | 4 (3–6) | 5 (3–6) | 4 (2–6) |
Health Insurance | | | |
Yes | 21 (72.4) | 13 (68.4) | 8 (80) |
No | 8 (27.6) | 6 (31.6) | 2 (20) |
Health Insurance Type | | | |
Public | 7 (33.3) | 4 (30.8) | 3 (37.5) |
Private | 9 (42.9) | 6 (46.2) | 3 (37.5) |
Public and private | 5 (23.8) | 3 (23.1) | 2 (25.0) |
Income meets basic needs | | | |
More than adequately | 4 (14.3) | 3 (15.8) | 1 (10.0) |
Adequately | 7 (25.0) | 6 (31.6) | 1 (10.0) |
Somewhat | 12 (42.9) | 7 (36.8) | 5 (50.0) |
Not at all | 5 (17.9) | 3 (15.6) | 2 (20.0) |
Missing | 1 (3.5) | 0 | 1 (10.0) |
Assessment of medication knowledge during coaching calls
During the medication coaching contact, 7 out of 19 intervention group participants were able to identify why they were taking all of their medications and an additional 9 could state the reason for taking 50% or more of their medications. Likewise, most could explain how to refill their medications. However, none could identify one side effect for all prescribed medications and 5 of the 19 were unable to list any side effects. All participants taking warfarin (n =4) were able to identify at least one side effect and knew that they required regular blood tests to determine its effectiveness.
The types of questions that were asked during the coaching intervention call were similar among the intervention participants, and frequently concerned why the stroke occurred even while taking prevention medications, and how to prevent another stroke. One 57 year-old participant asked why he was still having fatigue after the stroke and whether it was due to his medications. The pharmacist’s response was that this was quite possibly due to low blood pressure, and suggested the patient take one of his three blood pressure medications at bedtime and the other two in the morning to see if this helps the fatigue.
Follow-up calls and logistics
Fourteen of the 19 intervention patients required more than one phone call to conduct the coaching intervention, even within the two weeks after discharge. The median number of calls required to complete the medication coaching call was 2 (range 1–9). The median lengths of the coaching and follow-up calls with requested answers to questions were 27 minutes and 12 minutes, respectively.
Four participants out of 29 could not be contacted for the 3-month interview, even after multiple attempts. Because of the difficulty in reaching patients and proxies, 20 out of 37 of these interviews were completed after the 3 month target time window. The reasons for not reaching patients included: disconnected numbers, people out of country/out of state, and 2 patients were living with various relatives and moving frequently. For 2 other patients in the control arm, interviewers were only able to interview the proxy and not the patient.
The number of attempts made to reach an individual patient or proxy for their 3-month follow-up interview ranged from 1 to 30, and there were no differences between intervention and control participants. We were unable to collect 3-month follow-up information for 2 participants from the intervention arm (10.3% lost-to-follow-up but proxy data were obtained).
Three month outcomes
For those successfully contacted at 3 months and able to provide self-reported outcomes, we found little difference between the two groups in reported levels of knowledge or understanding about medications or stroke (Table
3). More of the intervention participants knew what to do if problems or symptoms continued or worsened (93.8%) compared to controls (77.8%). A larger proportion of intervention participants had seen their primary care provider between discharge and 3 months than those in the control group (93.8% vs. 60% in the controls; p = 0.055). In addition, we found trends towards lower depression (PHQ-8) severity scores in the intervention group (median = 5.50 vs. 10.5 in controls; p = 0.080), higher reported health status (EQ-5D) scores (median 0.80 vs. 0.68), and lesser disability (mRS =1.5 in cases vs. 2.5 among controls.)
Table 3
Three month outcomes for intervention and control participants
Has method for tracking medications, N (%) | 18 (69.2) | 11 (68.8) | 7 (70.0) | 1.000 |
Understand how to take medications, N (%) | 26 (100.0) | 16 (100.0) | 10 (100.0) | N/A |
Understand why taking medications, N (%) | 26 (100.0) | 16 (100.0) | 10 (100.0) | N/A |
Understand side effects, N (%) | 16 (61.5) | 10 (62.5) | 6 (60.0) | 1.000 |
Know who to call if run out of meds, N (%) | 26 (100.0) | 16 (100.0) | 10 (100.0) | N/A |
Know what to expect with your health/illness in the future, N (%) | 19 (73.1) | 12 (75.0) | 7 (70.0) | 1.000 |
Know what to do if problems/symptoms continued or worsened, N (%) | 22 (88.0) | 15 (93.8) | 7 (77.8) | 0.530 |
Appointment with PCP since stroke, N (%) | 21 (80.8) | 15 (93.8) | 6 (60.0) | 0.055 |
PHQ-8 at 3 months, median (IQR) | 8 (3.0-13.0) | 5.50 (0.5-10.5) | 10.5 (7.0-21.0) | 0.080 |
EQ5D at 3 months, median (IQR) | 0.8 (0.64-0.87) | 0.80 (0.69-0.94) | 0.68 (0.59-0.87) | 0.326 |
Modified Rankin score, median (IQR) | 2.0 (1.0-3.0) | 1.50 (1.0-3.0) | 2.50 (1.0-3.0) | 0.531 |
CTM-3, median (IQR) | 83.3 (77.8-100.0) | 77.8 (72.2-100.0) | 88.9 (77.8-100.0) | 0.640 |
Re-hospitalization, N (%) | 1 (4.2) | 0 (0) | 1 (10.0) | 0.333 |
ED visit, N (%) | 2 (8.7) | 2 (12.5) | 0 (0) | 1.000 |
Overall persistence with discharge medication regimens was 88% (22 of 26), and was similar in both groups (intervention 87.5% and control 88.9%). By medication class, persistence was also high, ranging from 83% for warfarin, 95% for antihypertensives and lipid-lowering medications, and 100% for antiplatelet therapy and diabetes medications.
Discussion and conclusion
In this study, we pilot-tested and refined a medication coaching script, assessed content validity of the script and questionnaires, and obtained participants’ evaluation of the intervention. We determined that this approach was feasible, and participants expressed a high degree of satisfaction with the information provided by the coach. The medication coaching calls were designed to assess medication knowledge of medications and triage individual questions about medications and the participant’s stroke, per the script. The burden of attempted contacts was reasonable, with a median number of 2 calls to reach members of the intervention group. The 3-month follow-up calls were more challenging because patients moved, changed phone numbers, and experienced major life changes associated with having a stroke. Calling at different times of day often was the key to reaching participants. Other helpful procedures were to determine whether the phone number provided at enrollment is for a pre-paid cell phone and to obtain multiple phone numbers for other relatives when possible. Most importantly, contacting patients and/or proxies requires persistent efforts.
Using a telephone intervention was practical for our stroke population because patients treated at our institution reside in a wide geographic area, making home visits restrictive. Telephone interventions have been used for post-discharge interventions for medication management in multiple settings. For example, patients discharged from a general medical service received a follow-up phone call by a pharmacist 2 days after discharge [
17]. Patients randomized to the intervention (n = 110; controls n = 111) were significantly more satisfied with discharge instructions, had medication-related problems solved during the intervention, and had fewer Emergency Department visits (10% vs. 24% in the controls) [
17]. A study of 123 patients over age 50 with varying underlying diseases utilized pharmacist follow-up for medication management and found that the intervention led to significantly greater resolution of medication and health-related problems than the control group [
21]. Rather than having pharmacists make the calls (as in the above examples), we had a coach who could triage questions to the pharmacist or stroke nurse and supply their answers to the participants. The pharmacist and nurse provided personalized information that could be put in writing for the patient and the primary care provider to keep as a reference. We believe using a coach rather than a pharmacist or nurse for calls may be more economical, although we have not assessed cost-effectiveness in this small study.
Medication non-persistence is associated with poorer outcomes, greater likelihood of re-hospitalization and increased mortality [
22,
23]. Although this study was not designed to show an effect on medication persistence due to the emphasis on feasibility and refinement of the medication coaching script, medication persistence was excellent in this small study. The Preventing Recurrence of Thromboembolic Events through Coordinated Treatment (PROTECT) program, which included multiple materials for patients at discharge and contact by study nurses 2 to 4 weeks after discharge, showed excellent persistence of antiplatelet therapies, statins, and blood pressure medications in stroke patients at 3 months [
24]. Similarly, a pharmacist call for a small cohort of Medicare beneficiaries in Texas showed that persistence with medications was not significantly improved, although the intervention was associated with other measurable benefits, such as resolution of medication and health related questions [
21]. The feasibility of medication coaching is currently being tested in Scotland in a randomized controlled trial of 60 patients with stroke to determine the impact on medication adherence [
25].
We also examined the impact of early post-discharge medication coaching on medication knowledge and adherence to follow-up appointments. Knowledge related to why medications are taken, how to refill them, and what to do if stroke symptoms recur was excellent in both groups (Table
3), perhaps because of the educational materials all participants received. However, we found that participants in the intervention group were more likely to keep their primary care provider appointments (93.8%) than those in the control group (60%). A key role of the coach is encouraging patients and caregivers to keep medical appointments which is important for preventing hospital readmissions [
5], and reducing the risk of stroke and death, especially in patients with hypertension [
26].
Short hospital stays following a transient ischemic attack or stroke make in-hospital visits for educating patients about risk factors, medications and post-discharge self-care a challenge. A call from a trained coach just after discharge, when patients are reorienting themselves to home, can provide answers to questions about their stroke and their medications that they may not have thought of during their stay. The coach can also reinforce the crucial messages for prevention of recurrent stroke provided by hospital staff and the transition coach prior to discharge.
Our pilot study was limited because these patients were younger than the average stroke patient, and only those discharged home were enrolled, which limits generalizability. As such, these patients had relatively mild, i.e. non-disabling strokes (Table
1), but these are also the patients who may gain the most from targeted interventions to improve adherence to secondary prevention. We included patients with TIAs in the intervention as well, because these patients have a similar risk of recurrent stroke, and guideline recommendations are the same for both stroke and TIA [
27]. Other limitations are the non-random allocation of the intervention and the modification of medication coaching script based on patient feedback. Thus, the script in its final form was administered to 14 of the 19 participants in the intervention group. Most importantly, the small cohort size limits conclusions regarding 3-month outcomes.