Optimising adherence to secondary prevention medications following acute coronary syndrome utilising telehealth cardiology pharmacist clinics: a matched cohort study

This study aimed to explore the effects of a telehealth cardiology pharmacist clinic on patient adherence to secondary prevention medications in the 12 months following ACS.

This study was granted approval and waiver of consent by Grampians Health Ballarat Human Research Ethics Committee (Project number LNR/71907/BHSSJOG) and Monash University (28159).

This was a retrospective matched cohort study with a 12 month follow up duration from the index PCI. It utilised an existing data source of patients who received PCI with coronary stenting. The study consisted of two groups, an intervention group from 2020 who received a consult in the telehealth cardiology pharmacist clinic, and a control group of patients prior to 2020 who did not receive the intervention.

The telehealth cardiology pharmacist clinic involves a 20-min consultation with a pharmacist addressing a number of different care elements. The mode of delivery of the service, as well as patient and clinician acceptability has been assessed in a previous study [23]. A detailed best-possible medication history is taken with secondary source verification in line with the Society of Hospital Pharmacists Australia: Standards of Practice in Clinical Pharmacy [24, 25]. Questions regarding cardiac symptoms and a functional assessment are completed, which align with the outcome measures defined in the Melbourne Interventionalist Group (MIG) assessment form. These questions are designed to understand the burden of cardiac symptoms as well as identifying any potential urgent referral points for patients. Education on cardiac health and treatment is provided in line with the National Heart Foundation of Australia National Cardiac Rehabilitation Quality Indicators [26]. Patients are consulted at 1 month, 3 months and 12 months post PCI. The format is repeated for each of the consultations, with a focus on building patient knowledge of cardiac medications and health. When barriers to access of medications are noted, the pharmacist took steps to rectify this by consultation with the cardiology unit and/or the primary care physician. Examples would be arranging new prescriptions, recommending modification of dose or dosage forms, and implementing dose administration aids. Additionally, attendance at cardiac rehab was assessed and referrals made where indicated.

The study was based at a large public regional health service in Victoria, utilising data collected as part of the health services participation in the Victorian Cardiac Outcomes Registry (VCOR) and MIG, as well as data from the telehealth cardiology pharmacist clinic consult letters.

There are approximately 250 PCI and coronary stent patients who present to Grampians Health Ballarat annually with ACS. Patients who received PCI were enrolled into an opt-out MIG registry. Data are collected at the time of PCI and at 30 days. Twelve-month follow up data were available for patients who received PCI with coronary stents in the 2015–2017 cohort. From 2017, routine 12 month follow up data were not collected.

Patients in the intervention arm had their baseline MIG data collected at time of PCI and at 30 days. Due to the lack of 12-month MIG data, the telehealth cardiology pharmacist clinic report was used to check if OMT was present at 12 months.

In order to allow for comparison between control and intervention groups, set criteria were established to ensure 12-month follow up data were available. It was the intention of this study to enrol all patients who received PCI for ACS during the study period of January 2020 to July 2020.

Inclusion criteria:

Exclusion criteria:

Adherence to medication classes was broken up into three groups, similar to previous studies [3, 27]. These groups included:

Patients are specifically asked about adherence to each medication comprising OMT, with adherence defined as taking the medication more than 80% of the time [28].

The primary outcome was the difference of self-reported adherence to all four groups of secondary prevention medications (optimal medication therapy) at 12 months post coronary stenting between a matched cohort of patients who received the intervention and those who did not. Self-reported adherence was determined by a set assessment form by the Melbourne Interventionalists Group, where the pharmacist would ask the patient directly during the telehealth consultation. This questionnaire was identical between the control and intervention groups. Secondary outcomes included the difference in Near-optimal Medical Therapy and Sub-optimal Medical Therapy, and individual medication groups (DAPT, statin, beta blocker and ACEI/ARB/ARNI).

Additionally, the difference in Major Adverse Cardiovascular Events (MACE) at 12 months between control and intervention matched cohort was investigated. MACE was defined as stroke, non-fatal myocardial infarction, rehospitalisation or death. To validate the use of self-reported adherence within the study, self-reported adherence was compared to calculated medication possession ratio (MPR) via the patient’s primary pharmacy dispensing records. This outcome was only measured in the intervention group due to dispensing data availability.

Cohort matching was used to reduce potential confounding between the control and intervention arms [29]. As this study is retrospective and non-randomised, the use of individual matching between cohorts provides a method to reduce confounding [29]. The matching criteria were selected due to both their availability within the data set and evidence regarding their significant correlation with changing adherence patterns between participants [30‐32].

In this study, matching was performed using individual matching across criteria:

Data matching was indexed at the time of the ACS event and baseline MIG data collected. Matching and analysis was performed using Stata^® 17 and Microsoft Excel^®.

Based on previous studies, adherence to medications at 12 months post ACS event can vary between 45 and 75% [3, 8]. Based on a population size of approximately 100 patients to have ACS in the 7-month intervention period (January 2020 to July 2020), a margin of error of 5% with an alpha of 0.05 and beta of 0.2, the sample size would need to be 73–78 matched patient pairs (one to one matching).

For outcome calculations, McNemar’s Chi-squared analysis for matched data was used between the control and intervention pairs. This was repeated across the primary and secondary outcomes involving matched data. For the adherence measure validation outcome, an R² value was calculated between self-reported adherence scores and calculated MPR. A value of 0.75 or greater was considered a substantial correlation.

There was no statistically significant difference when comparing matched pairs of participants with near-optimal medical therapy (three out of four post ACS mediation groups present). However, there was a statistically significant absolute reduction in patients with sub-optimal medical therapy (less than three post ACS medication groups present) of 16% (33/78 vs 20/78, p = 0.04). Across each individual medication group, significant differences were seen between matched pairs across all classes except for beta blockers (Table 2).

There was a significant reduction in major adverse cardiovascular events (MACE) consisting of a 4-point composite outcome of stroke, non-fatal myocardial infarction, rehospitalisation or death. The was an absolute reduction of MACE of 22% (34/78 vs. 17/78, p < 0.01). This was driven primarily through a reduction in hospitalisations (Table 3).

For the validation of self-reported adherence within the study population, all intervention group participants (n = 98) were included. This outcome compared the self-reported adherence scores to the medication possession ratio (MPR) sourced from participant dispensing histories. The linear regression model demonstrated a R² value of 0.84, with 17% (16/94) of participants underestimating and 13% (12/94) participants overestimating adherence when compared to MPR. The remaining 70% (66/94) of participants had MPR within 10% of the self-reported adherence (Fig. 2).

Limitations in this study include its single-centre setting and although this represents the population of the area the health service operates, this may not reflect Australian or international populations. Telehealth has seen drastic uptake in the COVID-19 era, presenting itself with new challenges in the ambulatory care setting. However, this telehealth model of care was established and validated with patients and clinicians prior to the COVID-19 pandemic, and therefore was not overly affected by changes in practice [23].

Although this study does not replace a randomised study, it has employed various techniques to utilise the benefits randomisation brings to a study. While the matching variables selected have been shown to explain differences in adherence in previous studies, it does not possess the potential power of other methods such as propensity matching [35]. However, this study used population analysis rather than sampling, and randomisation does not control this variance as no sample is drawn. With the population sampled, total adherence within almost all possible recruitment was known. In addition, unmatched comorbidity characteristics showed little difference overall, and adjustment of these variables would have been unlikely to shift the outcome given the high degree of significance.

The decision to add therapies is directed by evidence-based guidelines, but also must involve tailoring to the patient’s individual needs and safety. This study treated absence of therapy as non-adherence, however contraindications to therapy or safety outcome guided cessation of therapies may have contributed to what was analysed as non-adherent. This is a common limitation of studies like this, and while the 12-month follow up was the point of interest, dynamic changes in patient medication prescription are not present in the data [3]. However, regardless of therapies prescribed at discharge, the intervention group saw an increase in medication adherence at the end of the follow-up relative to those who did not receive the intervention.