Automated Outreach for Cardiovascular‐Related Medication Refill Reminders

Abstract

The objective of this study was to evaluate the effectiveness of an automated telephone system reminding patients with hypertension and/or cardiovascular disease to obtain overdue medication refills. The authors compared the intervention with usual care among patients with an overdue prescription for a statin or lisinopril‐hydrochlorothiazide (lisinopril‐HCTZ). The primary outcome was refill rate at 2 weeks. Secondary outcomes included time to refill and change in low‐density lipoprotein cholesterol and blood pressure. Significantly more patients who received a reminder call refilled their prescription compared with the usual‐care group (statin cohort: 30.3% vs 24.9% [P<.0001]; lisinopril‐HCTZ cohort: 30.7% vs 24.2% [P<.0001]). The median time to refill was shorter in patients receiving the reminder call (statin cohort: 29 vs 36 days [P<.0001]; lisinopril‐HCTZ cohort: 24 vs 31 days [P<.0001]). There were no statistically significant differences in mean low‐density lipoprotein cholesterol and blood pressure. These findings suggest the need for interventions that have a longer‐term impact.

Hypertension and high cholesterol are well‐documented risk factors for cardiovascular disease (CVD). Considerable evidence indicates that lowering blood pressure (BP) and low‐density lipoprotein cholesterol (LDL‐C) in patients with CVD can have a great impact on reducing risk of subsequent cardiovascular morbidity and mortality.1, 2 Many patients with CVD and hypertension, however, have suboptimal adherence to prescribed medications.3, 4 Moreover, poor adherence among patients with chronic conditions, such as hypertension and dyslipidemia, is associated with higher rates of healthcare utilization and medication intensification to achieve treatment goals, thereby increasing healthcare costs.5, 6, 7

Various interventions to increase adherence to cardiovascular medications including nurse management, pharmacy care programs, counseling, and reminders (via letters, e‐mail, and telephone) have been tested.8, 9, 10, 11, 12 A review of 22 evidence‐based interventions to improve patient adherence with antihypertensive and lipid‐lowering medications suggested that intensive and personalized programs showed the greatest improvements.10 Although combination strategies have produced better results than stand‐alone interventions, implementation of interventions with multiple components is more difficult and resource intensive.4, 10 Telephone reminder interventions have often involved healthcare professionals delivering reminders to small patient populations; however, this strategy is inefficient and costly for reaching a large group of patients. A low‐cost automated outreach program may have a positive impact on refill rates and subsequent control measures. We conducted a population‐based study to evaluate the effectiveness of a quality initiative that used an automated telephone reminder system to improve refill behavior in patients who failed to fill their cardioprotective medications on time and to determine the effect of the reminders on LDL‐C and BP among adults with CVD and/or hypertension. To our knowledge, this is the first contemporary study using an automated telephone outreach program to improve refill rates of cardioprotective medications.

Methods

Setting

This study was conducted in 2010 at Kaiser Permanente Southern California (KPSC), an integrated health delivery system that provides comprehensive care to more than 3.7 million members at 14 medical centers and 200 medical offices. Members of KPSC are socioeconomically diverse and broadly representative of the general population of Southern California.13 The KPSC institutional review board approved the study and waived informed consent because the intervention represented enhancement to usual population‐based outreach care.

Study Population and Randomization

We identified all patients 18 years and older within the KPSC hypertension and CVD registries. The CVD registry is composed of KPSC patients identified from diabetes, atherosclerotic CVD (ASCVD), heart failure (HF), and chronic kidney disease (CKD) registries. The registries identify patients by a combination of inpatient and outpatient International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) diagnosis and procedure codes, laboratory test results, and pharmacy dispensation information. Patients were not included in the disease registries if they were in hospice or a skilled nursing facility or deceased. There was a preliminary assessment of refill rates following automated telephone reminders for different durations of overdue status. The overdue status of 2 to 6 weeks was selected for further study because refill rates were highest among patients during this time period. Patients in the CVD registry who had a prescription for a statin that was 2 to 6 weeks overdue for refill were eligible for the statin outreach. Patients in the CVD or hypertension registries who had a prescription for lisinopril‐hydrochlorothiazide (lisinopril‐HCTZ) that was 2 to 6 weeks overdue for refill were eligible for the lisinopril‐HCTZ outreach, regardless of the number of previous prescriptions a patient may have filled. Patients were excluded from the outreach program if they did not want to be contacted, did not have a prescription benefit plan, had a documented allergy or intolerance to the outreach medications, or were pregnant.

The eligible patients who were overdue for a refill were randomly assigned to receive usual care or an automated telephone call. The final eligible statin cohort included 41,711 patients: 21,779 were randomized to receive the automated call and 19,932 were randomized to usual care. The final lisinopril‐HCTZ cohort included 7975 patients: 3793 were randomized to receive the automated call and 4182 were randomized to usual care.

Intervention

Using an automated telephone messaging system, the intervention occurred on the third Tuesday of the month. There were a total of five intervention dates for the statin cohort and two intervention dates for the lisinopril‐HCTZ cohort. Components of the outreach system included a regularly updated administrative database containing members' contact information, the laboratory test result database, the pharmacy information management database, an algorithm to determine when a member was due for a refill, an automated telephone messaging system, and the outreach program team.

Whether the telephone call was answered by a live person or by a voicemail system, the automated message was delivered without any further action being required, eg, no entry of date of birth or medical record number. Failed call attempts (ie, busy signal, no answer) resulted in a maximum of two additional call attempts in 2‐day intervals. Telephone calls were made between the hours of 10 am and 8 pm. The content of the automated message was developed by the KPSC outreach team. The message included a greeting stating that the call was from Kaiser Permanente, the name of the targeted member, and a statement instructing the member to order a refill of their overdue prescription by calling the number on their medication bottle or by using the online refill system. The name of the medication was not given. The automated message was played by default in English with an option to listen to the message in Spanish (Appendix 1).

Outcomes

The primary outcome was medication refill rate at 2 weeks postintervention in both cohorts. Timeliness to first refill after the intervention, the time between the first and the second refill, and changes in LDL‐C and BP were secondary outcomes. The preintervention and postintervention LDL‐C and BP values were measured closest to the intervention date (within 1 to 365 days prior to the intervention date and within 31 to 365 days after the intervention date).

Statistical Analysis

Summary statistics were calculated for patient's sociodemographic characteristics (age, sex, race/ethnicity, and education), LDL‐C, and systolic and diastolic BPs at baseline for each cohort. Means and standard deviations (SDs) were calculated for continuous variables and percentages were calculated for categorical variables. Differences in characteristics between the intervention and usual‐care groups were assessed using nonparametric Wilcoxon rank‐sum tests for continuous variables and χ² tests for categorical variables. The proportion of patients who refilled their statin or lisinopril‐HCTZ within 2 weeks of the intervention was calculated according to randomization assignment. As secondary analyses, we compared the time between randomization/intervention and the first refill following the intervention, the time between the first and the second refill, and the net change in LDL‐C and systolic and diastolic BP. All analyses were performed using SAS statistical software version 9.2 (SAS Institute Inc, Cary, NC).

Results

Within the statin cohort, 91% of the reminder calls were delivered to a live person or to a voicemail system, 8% of health plan members did not receive the message, and 1% had an invalid telephone number. In the lisinopril‐HCTZ cohort, 87% of the members were successfully contacted, 12% did not receive the message, and 1% had an invalid telephone number.

Baseline characteristics of the patients are provided in Table 1. The mean (SD) age among all patients was 61 (12) and 57 (12) years in the statin and lisinopril‐HCTZ cohorts, respectively. The mean (SD) LDL‐C in the statin cohort was 92.7 (33.4) and the mean (SD) systolic and diastolic BP in the lisinopril‐HCTZ cohort was 129.8 (15.5) and 76.7 (10.9), respectively. In both cohorts, the majority of participants were men. Patients randomized to the automated call group in the statin cohort were slightly older and more likely to be a high school graduate. Patients randomized to the usual‐care group in the lisinopril‐HCTZ cohort had a higher mean number of days between the last medication fill and the intervention date. There were no other statistically significant differences between the automated call and usual‐care groups.

Table 1.

Baseline Characteristics of the Study Participants in the Statin and Lisinopril‐HCTZ Cohorts

Variablea	Statin Cohort			Lisinopril‐HCTZ Cohort
	Intervention (n=21,779)	Usual Care (n=19,932)	P Valueb	Intervention (n=3793)	Usual Care (n=4182)	P Valueb
Mean age, y	61.2 (12.1)	60.9 (12.2)	.003	57.4	57.2	.34
Female, %	47.1	46.5	.25	44.1	43.7	.53
Race/ethnicity, %
Asian	10.2	10.4	.08	7.1	7.1	.20
Black	13.7	13.6		15.9	15.7
Hispanic	35.3	36.1		33.4	31.4
White	34.4	33.2		32.4	32.4
Other/unknown	6.5	6.8		11.2	11.1
Days' supply of last fill	90.6 (11.9)	90.8 (11.1)	.07	97.1 (13.7)	97.3 (12.9)	.57
Days between last fill and intervention date	119.8 (28.4)	120.1 (26.7)	.28	125.7 (25.9)	127.2 (32.7)	.03
Copayment, $	17.71 (22.54)	17.65 (22.24)	.79	10.74 (5.34)	10.74 (5.43)	.99
LDL‐C, mg/dL	92.6 (33.2)	92.9 (33.6)	.29	–	–	–
LDL‐C control±, %	69.9	69.6	.64	–	–	–
Systolic BP, mm Hg	–	–	–	129.7 (15.6)	129.9 (15.5)	.59
Diastolic BP, mm Hg	–	–	–	76.5 (10.9)	77.0 (10.8)	.06
BP control, %	–	–	–	80.5	80.7	.53

Abbreviations: BP, blood pressure; HCTZ, hydrochlorothiazide; LDL‐C, low‐density lipoprotein cholesterol. LDL‐C control± is defined as <100 mg/dL.

^aNumbers are expressed as mean (standard deviation) or percentage. ^bParticipant characteristics were compared using Wilcoxon rank‐sum tests for continuous variables and χ² tests for categorical variables.

Among patients overdue for a statin refill between 2 and 6 weeks (Figure), 30.3% who received a reminder call refilled their prescription within 2 weeks after the intervention compared with 24.9% who did not receive a call (absolute difference between outreach and control group, 5.4%; P<.0001). Among patients overdue for a lisinopril‐HCTZ refill, 30.7% who received a reminder call refilled their prescription within 2 weeks after the intervention compared with 24.2% who did not receive a call (absolute difference between outreach and control group, 6.5%; P<.0001).

Figure 1.

Refill rates at 2 weeks post‐outreach.

The median (range) time‐to‐refill after the outreach campaign was 29 days (−2 to 550) in the statin intervention group compared with 36 days (−2 to 540) in the usual‐care group (P<.0001), and 24 days (−1 to 398) in the lisinopril‐HCTZ intervention group compared with 31 days (−1 to 394) in the usual‐care group (P=.0009) (Table 2). The median (range) time from first to second refill was 118 days (61 to 533) in the statin intervention group compared with 115 days (61 to 531) in the usual‐care group (P<.0001), and 126 days (61 to 372) in the lisinopril‐HCTZ intervention group compared with 126 days (61 to 366) in the usual‐care group (P=.4537).

Table 2.

Refill Rates at 2 Weeks and Time to Refill by Treatment Arm

	Call Group	Control Group	P Value
Time from intervention to first refill: median (range), d
Statins	29.0 (−2 to 550)	36.0 (−2 to 540)	<.0001
Lisinopril‐hydrochlorothiazide	24.0 (−1 to 398)	31.0 (−1 to 394)	.0009
Time from first refill to second refill: median (range), d
Statins	118.0 (61 to 533)	115.0 (61 to 531)	<.0001
Lisinopril‐hydrochlorothiazide	126.0 (61 to 372)	126.0 (61 to 366)	.4537

Table 3 shows the postintervention and mean changes in LDL‐C and systolic and diastolic BP. Within the statin cohort, LDL‐C was similar in the outreach and usual‐care groups after the intervention, and there was no significant difference between the groups in the change in LDL‐C. Similarly, there were no statistically significant differences between the outreach and control groups with respect to systolic and diastolic BP.

Table 3.

Mean Changes^a From Baseline in LDL‐C and BP According to Intervention Assignment

	Intervention	Usual Care	P Value
Statin cohort
Postintervention LDL‐C, mg/dL	91.8 (32.8)	92.1 (33.2)	.41
Absolute change in LDL‐C	−0.75 (33.5)	−0.49 (34.0)	.48
Lisinopril‐HCTZ cohort
Postintervention systolic BP, mm Hg	131.4 (16.5)	131.9 (15.9)	.27
Absolute change in systolic BP	2.1 (17.9)	2.2 (17.8)	.84
Postintervention diastolic BP, mm Hg	77.1 (11.5)	77.5 (11.4)	.19
Absolute change in diastolic BP	0.9 (11.2)	0.8 (11.4)	.61

^aMean changes from baseline were compared using Wilcoxon rank‐sum test.

Discussion

The automated telephone outreach had a small but significant impact on the proportion of patients who refilled either their statin or lisinopril‐HCTZ prescription within 2 weeks after the outreach. Furthermore, the intervention appears to have facilitated patients to refill their first statin or lisinopril‐HCTZ prescription following the automated call 1 week sooner than those in the usual‐care group. Interestingly, the time from first to second refill was longer in the statin intervention group compared with the usual‐care group, while the time from the first to second refill was the same in both groups in the lisinopril‐HCTZ cohort. There were no significant differences in preintervention and postintervention LDL‐C and BP values between the outreach and usual‐care groups.

Although statistically significant, the small absolute differences in the refill rates between the study groups were potentially due to a number of factors. First, population care management efforts within KPSC during this time period focused on improving medication‐taking behavior and the high LDL‐C and BP control rates may have influenced the study results. Second, the outreach calls were placed only once and the call script included the prescription numbers but not the medication name. In addition, the outreach call did not include any messaging about the benefits of taking the medication, such as LDL‐C or BP reduction and reduced risk of a heart attack or stroke. An inner‐city university health center utilized a computerized reminder system to increase refill behavior among 311 patients randomly assigned into three groups: control, postcard reminder, and telephone call reminder.14 The study found a significant difference in the mean refill rate between the control and the two intervention groups; however, there was no significant difference between the postcard reminder and telephone call reminder intervention groups. Albright and colleagues15 conducted a qualitative study among African American health plan members of Kaiser Permanente Colorado to evaluate the use of a linguistically congruent voice in interactive voice response calls as a strategy to improve call completion rates.15 The participants expressed a preference for an African American voice and suggested the message emphasize the importance of screening to the target population. These studies suggest that tailoring automated messages for specific populations may increase the effectiveness of outreach programs.

Although the time‐to‐refill behavior between the intervention and the first refill improved in our study, the pattern was not sustained from the first to the second refill in either medication cohort. In contrast to our study, a randomized trial conducted in South Carolina among 3048 patients who were ≥7 days overdue for a prescription refill for a chronic disease, including hypertension and hyperlipidemia, found that a telephone reminder by pharmacists did not have a significant impact on time‐to‐refill rates compared with usual care.16 A community‐based pharmacy intervention among 102 CVD patients assessed the effectiveness of a postcard and telephone reminder system on improving refill rates for patients more than 3 days overdue for their prescription.17 Among patients in the experimental group, the time between the first refill before the intervention and the initial postintervention refill decreased by 300%. Unlike in our study, the refill behavior was sustained throughout the study. Our study and these other two studies assessed refill timeliness for cardioprotective medications; however, the differences in study design and patient populations potentially explain the variation in the studies' findings.

In the National Health and Nutrition Examination Survey (NHANES), 54% of adult participants with hypertension had controlled BP between 2003 and 201018 compared with 80% of members in our study. According to NHANES 2009–2010, the prevalence of LDL‐C control among adults who were receiving lipid‐lowering treatment was 64%19 while 69% of statin users in our study reached the LDL‐C target. Given the already well‐controlled LDL‐C and BP at baseline in our population, it is not surprising that the refill reminder outreach did not result in improved LDL‐C or BP control in either medication cohort. Ho and colleagues9 conducted a randomized clinical trial at four Department of Veterans Affairs medical centers to test a multifaceted intervention to improve medication adherence and BP and LDL‐C control following hospitalization for acute myocardial infarction. The intervention included medication reconciliation and tailoring, patient education, collaborative care, and automated voice messaging reminders prior to and on the medication due date. Although the complex intervention improved adherence to medications, there was no difference in the proportion of patients achieving the BP and LDL‐C goals. A randomized trial of telephone outreach was conducted at four large multispecialty medical groups to improve medication adherence and intermediate outcomes of care.20 The intervention, which was delivered by pharmacists, diabetes educators, or nurse health managers, failed to significantly improve glucose, BP, or LDL‐C control.

Study Limitations and Strengths

There are several limitations to the current study. The results of this study may not be generalizable to healthcare settings without integrated electronic systems linking the various components of health information, such as pharmacy and utilization data, which are needed for an automated outreach campaign. In addition, the findings may not apply to populations such as those with acute conditions for which refills may not be necessary and to symptomatic conditions for which patients may have a greater perceived need for medication.21, 22 Although refill rates increased and the time to refill initially decreased, we did not measure medication adherence, which could be useful in assessing the association between refill patterns, adherence, and clinical outcomes. In addition, health plan members may have been told by their providers to discontinue taking their medication23, 24 thus underestimating the refill rates. Health plan membership was not an eligibility criteria for study enrollment; therefore, patients would have varying lengths of medication use prior to the intervention date. The difference in patients' length of prior medication use may have influenced the refill rates and the time‐to‐refill patterns. Lastly, reminder calls that were delivered to a voicemail system may not have been heard by the intended recipient.

Despite these limitations, our study also has a few strengths. First, our study sample was large and ethnically diverse. Second, we used disease registries to identify health plan members diagnosed with hypertension and CVD, which would limit misclassification of these diagnoses. Third, health plan members in the usual‐care group were passively enrolled, which minimized the potential of their participation to increase adherence in this group.25

Conclusions

A simple automated telephone reminder intervention is effective at increasing refill behavior in the short term despite a lack of improvement in mean LDL‐C and BP. A longer‐term and more complex intervention is likely needed to sustain improvements in refill rates. Future studies assessing the impact of automated telephone outreach on refill behaviors should consider adding a postcard reminder before or after the automated telephone outreach, placing multiple calls to participants during the study period, tailoring the call by language preference and cultural fit, and including information about the benefits of taking the medication.

Appendix 1. Outreach Call Script

English Version With Spanish Prompt

“Hello, this is a Complete Care message from Kaiser Permanente for [name]. Our pharmacy records show you may have already run out of prescription number [xxxxx]. We're calling to remind you to order a refill for your prescription. If you've already requested a refill, or if your doctor has changed this prescription, please disregard this call. To order a refill, call the number on your pill bottle. Or you can order your refill online by going to kp.org. You can either pick up your prescription or have it mailed to your home in 7 to 10 days at no extra charge. Thank you.

Spanish Version

“Hola. Este es un mensaje de Cuidado Integral de Kaiser Permanente para [nombre]. Los datos de nuestra farmacia indican que ya se le puede haber acabado la receta con numero [xxxxx]. Llamamos para recordarle a ordenar una recarga de su receta. Si ya relleno su medicina, o si su doctor ha cambiado su receta, por favor ignore esta llamada. Para rellenar su receta, llame al numero en su pomo de pastillas. O puede hacerlo online yendo a kp.org. Puede recoger su receta o recibirla por correo en su casa de 7 a 10 dias sin cargo extra. Gracias.

J Clin Hypertens (Greenwich). 2016;18:641–646. DOI: 10.1111/jch.12723 © 2015 Wiley Periodicals, Inc.