VA guidelines recommend one of three beta-blockers in systolic heart failure (HF), with target doses. However, Pharmacy Benefits Management (PBM) data indicate that discordant prescribing is substantial, with only 17% meeting guideline criteria for agent and target dose. This represents 79,718 VA patients.
We evaluated two methods for promoting adherence to guidelines. The design included station-level baseline measurement, feedback to the pharmacy, re-measurement after six months, and randomized (by station) comparison samples. The primary aim was to estimate the percentage of station patients who had a beta-blocker prescription that was adherent to guidelines six months post-intervention. Additional aims were to estimate progress toward adherence and factors associated with non-adherence.
A cluster-randomized trial was used to compare station adherence to guidelines after an intervention directed at the pharmacy. Outcomes were patient-prescriptions.
Patients whose prescriptions were to be followed had: a PBM prescription for a beta-blocker that was disconcordant with guidelines, without an attempt at concordance in the prior 12 months, and the patient had at least one inpatient primary diagnosis consistent with non-diastolic HF at least 3 but no more than 9 months prior to the prescription.
Prescription, diagnostic, and identifier data for all patients was obtained from the electronic administrative databases in the Department of Veterans Affairs medical care system. We also conducted a formative evaluation by inviting all 12 stations to complete a survey after the intervention to rate the helpfulness of materials.
Stations were block randomized. Follow-up time to convert patients was six months. Via a secure Sharepoint site, data were collected on station resources (number of pharmacists, presence of HF clinic).
Level 1 intervention included background information, recommendation to consult cardiology on perceived dose-limiting reactions, and description of overall VA adherence, as well as confidential description for the specific station. Level 2 included all elements of Level 1, as well as a list of specific patients whose recent beta-blocker was discordant with guidelines (i.e. non-proven beta-blocker or less than target dose). The list was posted on Sharepoint on a secure server. The list/spreadsheet also allowed data collection from the station on: (a) reasons a patient could not achieve target; (b) provider type associated with the prescription; (c) resource-related reasons for non-conversion.
Since the randomization unit was the station, the analysis focused on aggregated patient-prescriptions as measures of station adherence (except as indicated). For each patient, the final beta-blocker prescription in the six-month period after baseline was categorized. Prescriptions were "adherent" if the total daily dose for carvedilol was at least 50mg (since we did not have patient weight) or if the total daily dose for metoprolol succinate was at least 200 mg. Otherwise, prescriptions were categorized as "with progress" if there had been a change toward adherence, or as "with regression" if there was a change from a guideline to a non-guideline agent or a decrease in dose of a guideline agent. Station performance was summarized as the percent of a station's patients falling into each category. These station percentages were then averaged within the intervention groups. Values shown are mean (standard deviation).
All findings relate to station performance regarding adherence to guidelines in the closed cohort of patients with a non-adherent index prescription. Therefore, by definition, at baseline the station performance is 0% guideline adherence for the selected cohort.
The stations that did not participate in the intervention provide the context for the intervention (114 stations with 2575 patients), i.e. a "natural history". After six months, the mean station adherence in this group was still low, only 2% (3.3), while progress toward adherence averaged 6% (6.2).
Twelve stations participated and were randomized, comprising 277 patients eligible for intervention (126 patients at six Level 1 stations and 151 patients at six Level 2 stations). The baseline characteristics of the stations were similar across levels of intervention. The adherence after six months was as low in both intervention groups (<3%) as in the non-intervention cohort. Because of the low rate, combined with the smaller numbers in the intervention groups, the rate estimates were unstable. Therefore, we combined the measures of adherence and progress toward adherence to describe the intervention. There was a tendency for better performance on this measure in Level 2 (18%; 95% CI: 11.6 - 24.4) than in Level 1 (12%; 95% CI: 4.7 - 19.1) or in non-intervention stations (8%; 95% CI: 6.7 - 9.4).
Prescriptions were ordered by primary care (67%), cardiology (31%) or other subspecialty (2%). The most common reasons for non-adherent prescriptions included adverse events (30%), non-systolic heart failure (28%), pharmacists' inability to engage the provider (13%), and patient logistics (12%).
To estimate the best adherence that might be achievable, we evaluated a subset of patients who should have had a reasonable chance of achieving adherence. This comprised patients whose baseline and final prescriptions were the same guideline agent, and who did not expire during the six months. The total daily dose for the final prescription was categorized as <50%, 50%, or >50% of target. For Level 1, the mean percentages in these categories were 67%, 28%, and 5%. There was a tendency toward higher doses at Level 2 stations, averaging 44%, 46%, and 10% across categories. However, only a small percentage of patients exceeded 50% of the target dose. The estimates did not change when patients with non-systolic HF (13%) were removed.
The survey response rate was 92%. In terms of helpfulness of the resources that the study provided, 60 - 80% of Level 1 stations and 83 - 100% of Level 2 stations found them helpful. The most common barriers and facilitators stated were: the role played by providers and clinic staff (whether or not they could be readily engaged) or the systems in place (titration clinics, clinical pharmacist role).
The findings are relevant to patients and the healthcare system. The greatest limiting factor appears to be adverse events. It may be that patients meeting the study criteria require a more comprehensive approach to their medication management, to minimize adverse events.
None at this time.
Treatment - Observational, Treatment - Comparative Effectiveness
Care Management, Chronic heart failure, Pharmaceuticals, QUERI Implementation