OUP user menu

★ Editor's choice

Pharmacist intervention in primary care to improve outcomes in patients with left ventricular systolic dysfunction

Richard Lowrie , Frances S. Mair , Nicola Greenlaw , Paul Forsyth , Pardeep S. Jhund , Alex McConnachie , Brian Rae , John J.V. McMurray
DOI: http://dx.doi.org/10.1093/eurheartj/ehr433 314-324 First published online: 14 November 2011

Abstract

Background Meta-analysis of small trials suggests that pharmacist-led collaborative review and revision of medical treatment may improve outcomes in heart failure.

Methods and results We studied patients with left ventricular systolic dysfunction in a cluster-randomized controlled, event driven, trial in primary care. We allocated 87 practices (1090 patients) to pharmacist intervention and 87 practices (1074 patients) to usual care. The intervention was delivered by non-specialist pharmacists working with family doctors to optimize medical treatment. The primary outcome was a composite of death or hospital admission for worsening heart failure. This trial is registered, number ISRCTN70118765. The median follow-up was 4.7 years. At baseline, 86% of patients in both groups were treated with an angiotensin-converting enzyme inhibitor or an angiotensin receptor blocker. In patients not receiving one or other of these medications, or receiving less than the recommended dose, treatment was started, or the dose increased, in 33.1% of patients in the intervention group and in 18.5% of the usual care group [odds ratio (OR) 2.26, 95% CI 1.64–3.10; P< 0.001]. At baseline, 62% of each group were treated with a β-blocker and the proportions starting or having an increase in the dose were 17.9% in the intervention group and 11.1% in the usual care group (OR 1.76, 95% CI 1.31–2.35; P< 0.001). The primary outcome occurred in 35.8% of patients in the intervention group and 35.4% in the usual care group (hazard ratio 0.97, 95% CI 0.83–1.14; P = 0.72). There was no difference in any secondary outcome.

Conclusion A low-intensity, pharmacist-led collaborative intervention in primary care resulted in modest improvements in prescribing of disease-modifying medications but did not improve clinical outcomes in a population that was relatively well treated at baseline.

  • Left ventricular systolic dysfunction
  • Primary care
  • ACE inhibitor
  • β-Blocker

Introduction

Although angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), and β-blockers reduce morbidity and mortality in patients with heart failure due to left ventricular systolic dysfunction, there is evidence that these treatments are underused, particularly in primary care.15 Some patients who should receive these medications do not and many receive less than evidence-based doses.25 Pharmacists may play a role in improving treatment through ‘collaborative medication review’, a process in which the pharmacist evaluates a patient's medications and suggests changes which are enacted with the agreement of the patient and the family doctor.69 A meta-analysis of small, short-term trials and observational data suggest that this intervention reduces the risk of hospital admission and possibly mortality in patients with heart failure studied in a secondary care setting.610 We conducted a larger-scale, longer-term prospective randomized controlled trial to test the hypothesis that a low-cost, low-intensity pharmacist intervention to optimize medical treatments, particularly ACE inhibitors, ARBs, and β-blockers, in patients identified in primary care with left ventricular systolic dysfunction would reduce the composite outcome of hospital admission for worsening heart failure or death, as well as other clinically important outcomes.

Methods

Study design and patients

The study was conducted within the National Health Service (NHS) which provides free health care to the population of the UK. The design of our trial has been published and was consistent with Consolidated Standards of Reporting Trials.1113 Consenting patients were eligible if aged ≥18 years and had left ventricular systolic dysfunction confirmed by cardiac imaging conducted at a local hospital (transthoracic echocardiography in 90% of cases). Patients did not have to have symptoms or signs of heart failure. Family doctors receive a semi-quantitative report of left ventricular systolic function (normal, mild, moderately or severely reduced) instead of ejection fraction. A key exclusion criterion was registration with the heart failure nurse service, which is provided to patients in our Health Board area recently admitted to hospital with heart failure. This criterion excluded higher-risk patients with more severe symptoms. Other exclusion criteria included concurrent disease other than heart failure likely to reduce life expectancy; severe cognitive impairment or psychiatric illness; and dialysis or a resident in a long-term care facility. The study was approved by the local ethics committee. All practices and patients gave written informed consent. The study is registered, number ISRCTN70118765.

Randomization

We used a cluster randomization design as this provides protection against contamination across trial groups when trial patients are managed within the same setting as was the case in this study. Patients in practices in the UK are managed by all general practitioners within the practice; as the control intervention was mediated by general practitioners, this precluded individual patient level randomization. Family practices were randomly allocated using a third-party automated telephone interactive voice response system in a 1:1 ratio to receive intervention or usual care. Stratification was by socioeconomic deprivation (affluent, intermediate, or deprived) at the practice level and the practice type (single-handed or group-practice).11

Study procedures

The intervention was delivered by 27 primary care-based pharmacists employed by the NHS to work with family doctors and directly with patients to promote rational, cost-effective prescribing.14 All participating pharmacists had between 3 and 16 years of post-qualification experience. All had experience delivering primary care-based medication review clinics for patients receiving multiple drug treatment. Seven pharmacists held post-graduate clinical pharmacy qualifications. Four pharmacists had hospital (ward-based) clinical pharmacy experience. Prior to commencing the intervention, all pharmacists attended one, in-house training day (contact time 7.5 h) covering the aetiology, symptoms, and evidence-based management of heart failure. The day was coordinated by three pharmacists who had a special interest in heart failure therapeutics, and a general practitioner with a special interest in heart failure. The day comprised of a mixture of didactic teaching and interactive role-play sessions. An additional mandatory 3-h session covered the methods of the trial. All pharmacists received an information pack with directed, heart failure-specific reading to supplement their training.

As part of routine continuing professional development, each pharmacist participated in a 3.5-h peer-led session every month which involved group discussion of cases encountered in their medication review clinics. As the study pharmacists were embedded within primary care practices, informal discussion on therapeutics occurred regularly between pharmacists, general practitioners, and nurses within the practice. There was also regular telephone contact between study pharmacists and the principal investigator or another pharmacist with a special interest in heart failure.

Patients from practices assigned to the intervention were offered a 30-min appointment with a pharmacist. The main aim of this review was optimization of medical treatment for left ventricular systolic dysfunction according to guidelines (Supplementary material online). If there was agreement between the pharmacist and the patient during the consultation and subsequently with the family doctor, medications were initiated, discontinued, or modified by the pharmacist during 3–4 subsequent weekly or fortnightly consultations. Family doctors provided usual care thereafter. No instructions were given to family doctors in the usual care practices. The study pharmacists did not collect information on symptoms or examine the patients as this was not part of their professional training. The cause of heart failure was established by scrutinizing primary and secondary care (e.g. hospital letters and discharge summaries) clinical records.

Study outcomes

The primary outcome was the composite of death from any cause or hospital admission for worsening heart failure, analysed as time to first event. Secondary endpoints included the composites of death from any cause or hospital admission for pre-specified cardiovascular causes (Supplementary material online), death from any cause, or hospital admission for any cause; total number of admissions (and patients admitted) for heart failure, cardiovascular causes, and any cause; and days alive out of hospital. Outcome data were obtained from the Information Services Division (ISD) of the NHS in Scotland, which records all discharges from Scottish NHS hospitals (i.e. virtually all hospital admissions in Scotland) and the General Register Office which records all deaths. Hospital discharge data are reported to ISD with discharge diagnoses coded according to the International Classification of Diseases system, 10th revision.15,16 This approach to patient follow-up has been compared favourably with traditional clinical trial methods.17,18 Although the end-of-study date was 31 January 2011, data extraction did not occur until 25 July 2011 in order to ensure that all deaths and hospital admissions were entered into the national database.

We also compared the prescribing of medications between the intervention and usual care groups and evaluated health-care utilization, other than admissions, at 1 and 2 years of follow-up, e.g. the number of primary care contacts and hospital emergency room visits.

Statistical analysis

The trial data were managed and analysed by the Robertson Centre for Biostatistics, University of Glasgow. The primary analysis compared the main outcomes between the intervention and control groups using a Cox proportional hazards frailty model, which accounted for the cluster-randomization design.1113 All analyses were performed using either SAS version 9.2 or R version 2.10.1 or greater. A significance level of 0.05 was used to test for statistical differences throughout and all tests used were two-sided. Analyses were adjusted for the stratification variables and the following variables of prognostic importance: age, creatinine, grade of left ventricular systolic dysfunction, atrial fibrillation, respiratory disease, total number of medical treatments, and diuretic use.

We assumed the rate of the primary outcome would be 10% per year in the control group and pharmacist intervention would lead to a relative risk reduction of 26%, based on the known benefits of initiating disease-modifying treatment and of higher doses of those therapies.1,19,20 We also assumed recruitment would last 2.6 years and follow-up would continue for 2 further years. Under these assumptions, to have 80% power to detect a difference between treatments, the trial would require 673 patients to experience a primary outcome. Inflating the sample size by a factor of 1.55 to account for the cluster randomization design, we needed 87 practices (1044 patients) per group.1113 Due to longer than anticipated recruitment, blinded review of the data in September 2010 suggested that more than 750 patients were expected to experience the primary outcome, providing 80% power to detect a 19% relative risk reduction.

Logistic regression models were used to examine the secondary outcomes including whether patients started or stopped a medical treatment or had a change in dose.

Safety was evaluated by examining hospital admissions for: symptomatic hypotension, collapse, or syncope; renal dysfunction, failure, or impairment; hyperkalaemia; asthma or chronic obstructive pulmonary disease; and bradycardia, atrioventricular block, or pacemaker implantation. Fisher's tests were used to compare the intervention group with the usual care group.

Results

Study patients

All general practices in our locality were approached and 174 of 220 (79%) practices (covering 790 421 of 998 402, i.e. 79% of patients registered with a general practitioner in our Health Board area) took part in the trial. From 25 October 2004 through 6 September 2007, 87 primary care practices (1090 patients) were randomly assigned to pharmacist intervention and 87 practices (1074 patients) to usual care (Figure 1). One patient had incomplete follow-up due to emigration. The median duration of follow-up was 4.7 years (range 6 days to 6.2 years; 9362 patient-years).

Figure 1

Study recruitment. §Five patients (from five practices, two intervention and three usual care) died in the period between baseline data collection and practice randomization. *Information on treatment unavailable for two patients between baseline year 1 and one additional patient between the end of year 1 and the end of year 2. LVSD, left ventricular systolic dysfunction.

The two groups were balanced with respect to baseline characteristics (Table 1). Fifty-five per cent of patients were aged ≥70 years. There was a high prevalence of ischaemic heart disease (80%) and respiratory disease (30%). A semi-quantitative assessment of left ventricular function was recorded in 2023 (93%) patients, with most having either a mild (41%) or a moderate (42%) reduction in systolic function. In a small reasonably representative subset of patient records (n= 256; 12%), there was documentation of a left ventricular ejection fraction value along with a semi-quantitative assessment of systolic function. In these patients, the mean ejection fraction was 36.8% (SD 6.5%) in those with mildly reduced systolic function, 30.5% (7.5%) in those with moderately reduced function, and 19.7% (7.7%) in those with severely reduced function.

View this table:
Table 1

Baseline characteristics of patients, according to study group*

CharacteristicPharmacist intervention (n= 1092)aUsual care (n= 1077)a
Age (years)70.6 (10.3)70.6 (10.1)
Age ≥70 years597 (55%)597 (55%)
Sex (female)320 (29%)329 (31%)
Blood pressureb
 SBP (mmHg)127 (17.4)128 (17.4)
 DBP (mmHg)72 (10.1)72 (10.2)
Left ventricular systolic function (%)c
 Mild reduction433 (42%)390 (39%)
 Moderate reduction416 (41%)439 (44%)
 Severe reduction170 (17%)175 (17%)
BMI (kg/m2)d28.0 (5.4)28.0 (5.4)
Principal cause of left ventricular systolic dysfunction (%)
 Ischaemic heart disease874 (80%)817 (76%)
 Non-ischaemic heart disease197 (18%)230 (21%)
 Unknown21 (2%)30 (3%)
Duration of LVSD (years)e3.31 (1.54–5.85)3.53 (1.76–5.87)
Medical history
 Admission for heart failure in past year19 (2%)18 (2%)
 Hypertension548 (50%)495 (46%)
 Myocardial infarction722 (66%)665 (62%)
 PCI161 (15%)144 (13%)
 CABG266 (24%)268 (24%)
 Atrial fibrillation or flutter292 (27%)304 (28%)
 Diabetes mellitus234 (21%)212 (20%)
 Stroke148 (14%)162 (15%)
 Respiratory disease326 (30%)318 (30%)
  Asthma72 (7%)82 (8%)
 Smoker262 (24%)260 (25%)
Serum creatinine (µmol/L)f109 (33)107 (30)
eGFR (mL/min per 1.73 m2)g61.3 (17.5)62.6 (23.2)
Device treatment (%)h
 Implantable cardioverter-defibrillator19 (2%)12 (1%)
 Conventional pacemaker22 (2%)29 (3%)
Cardiac medicines at randomization (%)
 ACE inhibitor814 (75%)768 (71%)
  ≥50% recommended dosei695 (86%)656 (86%)
  ≥100% recommended dosei483 (59%)472 (62%)
 ARB149 (14%)181 (17%)
  ≥50% recommended dosei102 (68%)107 (59%)
  ≥100% recommended dosei34 (23%)35 (19%)
 ACE inhibitor or ARB or both944 (86%)919 (85%)
 β-Blocker674 (62%)664 (62%)
  ≥50% recommended dosei339 (51%)339 (52%)
  ≥100% recommended dosei146 (22%)128 (20%)
 Aldosterone antagonist54 (5%)56 (5%)
 Digitalis glycoside149 (14%)124 (12%)
 Diuretic667 (61%)654 (61%)
 Aspirin751 (69%)691 (64%)
 Antithrombotic (antiplatelet or oral anticoagulant)990 (91%)969 (90%)
 Amiodarone20 (2%)32 (3%)
 Lipid lowering agent863 (79%)841 (78%)
Number of medicinesj8.9 (4.0)8.4 (3.5)
  • Data are the numbers of patients (%) or mean (SD). Percentages may not total 100 because of rounding. SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, body mass index; PCI, percutaneous coronary intervention; CABG, coronary-artery bypass grafting; eGFR, estimated glomerular filtration rate; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker.

  • aFive patients (two intervention and three usual care) died before randomization.

  • bData missing for 83 patients.

  • cData missing for 146 patients.

  • dData missing for 154 patients.

  • eData missing for 36 patients.

  • fData missing for 296 patients.

  • gData missing for 302 patients.

  • hData missing for 3 patients.

  • iPercentage of those prescribed an ACE inhibitor, ARB or β-blocker and with the dose recorded.

  • jMedicines on repeat prescription.

Although general practitioners do not routinely record the New York Heart Association (NYHA) class in our area (and the NYHA class was not recorded by the study pharmacists), this measure was documented in the records of 337 patients; 87 (26%) were NYHA class I; 215 (64%) were NYHA class II; 33 (10%) were NYHA class III; and 2 (1%) were in NYHA class IV.

Very few (1.7%) patients had been admitted to a hospital for heart failure in the year prior to randomization.

Medical treatment

In each treatment group, ∼86% of patients were receiving an ACE inhibitor, ARB, or both (Table 1) and of these patients, 55% received 100% or more of the recommended dose. The proportion of patients receiving a β-blocker was 62% in each treatment group. Of these, 21% were treated with 100% or more of the recommended dose (Table 1).

Patients not prescribed an ACE inhibitor, ARB, or β-blocker, or not prescribed the recommended doses of these medications at baseline, were potentially eligible for treatment optimization. Table 2 shows the effect of pharmacist intervention (patients who died during the first and second years of follow-up were excluded from this analysis—dead patients could not receive the intervention). Pharmacist intervention during the first year of the trial led to a greater frequency of initiation of an ACE inhibitor or ARB and a β-blocker, compared with usual care (Table 2). The same was true for increases in the dose of these treatments. At the end of year 1, an ACE inhibitor or ARB was started or the dose was increased in 168 of 507 (33.1%) patients in the intervention group and 95 of 514 (18.5%) patients in the usual care group [odds ratio (OR) 2.26, 95% CI 1.64–3.10, P< 0.001]. The proportions starting or having an increase in the dose of β-blocker were 153 of 854 (17.9%) patients in the intervention group and 95 of 855 (11.1%) patients in the usual care group (OR 1.76, 95% CI 1.31–2.35, P< 0.001). The resultant proportion of patients receiving an ACE inhibitor, ACE inhibitor or ARB and β-blocker at the end of year 1 and year 2 is shown in Table 3, as well as the proportions of those patients receiving at least 50% and at least 100% of the recommended dose (this table also takes account of discontinuations and reduction in doses). These differences were sustained during the second year with no evidence of ‘catch-up’ prescribing in the usual care group (see Table 3 and Supplementary material online). There was no difference between treatment groups regarding dose reduction or discontinuation of these drugs. The proportion of patients collecting 80% or more of prescriptions (from their general practice) was 99% in the pharmacist intervention group vs. 99% in the usual care group for ACE inhibitors, 98 vs. 98% for ARBs, and 98 vs. 99% for β-blockers (no difference between treatment groups for any drug).

View this table:
Table 2

Changes in key disease-modifying medicines between baseline and the end of the first year of follow-upa

Medical treatmentPharmacist intervention [n (%)]Usual care [n (%)]Odds ratio (95% CI)P-value
ACE inhibitor or ARBa
 Startedb39/131 (30)27/149 (18)2.03 (1.14–3.60)0.02
 Increased dosec129/376 (34)68/365 (19)2.32 (1.57–3.44)<0.001
 Increased dose to ≥100% of targetc86/376 (23)40/365 (11)2.46 (1.51–3.99)<0.001
 Started or increased dose168/507 (33)95/514 (18)2.26 (1.64–3.10)<0.001
β-Blockera
 Startedb50/388 (13)35/388 (9)1.56 (0.98–2.47)0.06
 Increased dosec103/466 (22)60/467 (13)1.90 (1.29–2.79)0.001
 Increased dose to ≥100% of targetc38/466 (8)22/467 (5)1.75 (0.99–3.09)0.05
 Started or increased dose153/854 (18)95/855 (11)1.76 (1.31–2.35)<0.001
  • aPatients who died (n = 106) or were lost to follow-up (n = 2) during the first year of follow-up were not included in this analysis.

  • bPatients not taking this medicine at baseline.

  • cPatients taking this medicine at baseline (but at <100% of the target dose) and had dose recorded (number with missing dose: ACE inhibitor ARB n = 1; β-blocker n= 12).

View this table:
Table 3

Use of disease-modifying medicines at the end of the first and the second year of follow-up

TreatmentEnd of year 1End of year 2
Pharmacist intervention (n= 1038)aUsual care (n= 1018)bUnadjusted P-valuePharmacist intervention (n= 966)cUsual care (n= 958)dUnadjusted P-value
ACE inhibitor
 Any dose [n (%e)]748 (72)698 (69)0.15698 (72)660 (69)0.20
 ≥50% recommended dose [n (%f)]649 (87)591 (85)0.09595 (85)524 (79)0.01
 ≥100% recommended dose [n (%f)]496 (66)429 (61)0.05444 (64)377 (57)0.02
ACE inhibitor or ARB
 Any dose [n (%e)]884 (85)842 (83)0.11843 (87)800 (84)0.03
 ≥50% recommended dose [n (%f)]751 (85)686 (81)0.03699 (83)615 (77)<0.001
 ≥100% recommended dose [n (%f)]538 (61)466 (55)0.03490 (58)416 (52)0.02
β-Blocker
 Any dose [n (%e)]660 (64)625 (61)0.31631 (65)601 (63)0.24
 ≥50% recommended dose [n (%f)]342 (52)313 (50)0.28341 (54)301 (50)0.07
 ≥100% recommended dose [n (%f)]171 (26)139 (22)0.09169 (27)135 (22)0.06
  • aNumber dead or lost to follow-up year 1, n = 52.

  • bNumber dead or lost to follow-up year 1, n = 56.

  • cNumber dead or lost to follow-up year 2, n = 124.

  • dNumber dead or lost to follow-up year 2, n = 116.

  • ePercentage of individuals in the group available for analysis.

  • fOf all patients prescribed these medicines at the end of the first and the second year.

At the end of year 1, 5.0% of the pharmacist intervention group and 4.6% in the usual care group were prescribed an aldosterone antagonist. At the end of year 2, the proportions were 5.1 and 5.2%.

At baseline diltiazem or verapamil was used in 132 (6%) of patients, a non-steroidal anti-inflammatory drug in 144 (7%), an anti-depressant in 223 (10%) [a tricyclic in 73 (3%)] and a glitazone in 17 (1%). These drugs were discontinued more often in the pharmacist intervention group in the first year but the difference in rates of discontinuation between the two treatment groups were not statistically significant (Supplementary material online).

Study outcomes

Death from any cause or hospital admission for heart failure (the primary outcome) occurred in 390 patients (35.8%) in the intervention group and 380 patients (35.4%) in the usual care group (Figure 2 and Table 4). The adjusted hazard ratio (HR) for the primary outcome in the intervention group, when compared with the usual care group, was 0.97, 95% CI 0.83–1.14, P= 0.72). The effect of the intervention on this outcome was consistent in an unadjusted analysis (Table 4) and across all pre-specified subgroups (Figure 3).

View this table:
Table 4

Primary and main secondary outcomes

OutcomePharmacist intervention (n = 1090)Usual care (n = 1074)Adjusted HR (95% CI)Adjusted P-valueUnadjusted HR (95% CI)Unadjusted P-value
Patients with eventEvent rate (n/100 patient-year)Patients with eventEvent rate (n/100 patient-year)
Primary outcome
 Death from any cause or admission for heart failurea390 (36%)8.53380 (35%)8.570.97 (0.83–1.14)0.720.99 (0.87–1.13)0.91
Main secondary outcomes
 Death from any cause or admission for cardiovascular cause487 (45%)11.54475 (44%)11.670.97 (0.84–1.12)0.700.99 (0.87–1.11)0.81
 Death from any cause or admission for any reason758 (70%)24.93751 (70%)25.460.96 (0.86–1.07)0.410.97 (0.88–1.07)0.55
Admissions
 Heart failure107 (10%)2.34114 (11%)2.570.88 (0.67–1.16)0.360.90 (0.71–1.14)0.38
 Cardiovascular causes292 (27%)6.92280 (26%)6.880.98 (0.81–1.19)0.830.99 (0.84–1.18)0.94
 Any reason711 (65%)23.38695 (65%)23.560.97 (0.87–1.09)0.610.98 (0.89–1.08)0.73
Death337 (31%)7.10331 (31%)7.180.96 (0.80–1.16)0.680.99 (0.85–1.16)0.92
  • aThe number of events contributing to the primary composite in the intervention group was 283 deaths and 107 admissions for heart failure and in the usual care group 266 deaths and 114 admissions for heart failure.

Figure 2

Main study outcomes.

Figure 3

Subgroup analysis for the primary outcome.

Death from any cause or hospital admission for a cardiovascular cause occurred in 487 patients (44.7%) in the intervention group compared with 475 patients (44.2%) in the usual care group (HR 0.97, 95% CI 0.84–1.12, P = 0.70). A total of 337 patients (30.9%) in the intervention group and 331 patients (30.8%) in the usual care group died (HR 0.96, 95% CI 0.80–1.16, P= 0.68). The number of deaths attributed to a non-cardiovascular cause was 155 in the intervention group and 169 in the usual care group.

The number of patients admitted to hospital for any reason, for a cardiovascular cause, and for heart failure was similar in the two treatment groups (Table 4). The total numbers of hospital admissions (including second and subsequent hospital admissions) for any reason were 2205 vs. 2191 (P= 0.84), for cardiovascular causes 474 vs. 517 (P = 0.19), and for heart failure 149 vs. 194 (P = 0.08), in the intervention group and the usual care group, respectively. Findings for the other pre-specified secondary outcomes and safety outcomes are reported in the Supplementary material online.

Discussion

Although a meta-analysis of small, short-term studies suggested that pharmacist intervention improves clinical outcomes in patients with heart failure,7 we did not confirm this in a much larger and longer trial conducted in primary care, despite the intervention leading to improvements in the use of disease-modifying medications which persisted for at least 2 years.

There are a number of possible explanations for this finding. The frequency of use of ACE inhibitors and ARBs at baseline was greater than that reported in previous studies in primary care, and even in our pilot study, with 86% of patients prescribed at least one of these medications.25,21 The explanation for this unexpected finding is uncertain, although in 2004, the year our trial started, the UK government introduced a new contract for family doctors linking pay to performance.22,23 Prescribing of ACE inhibitors (but not β-blockers) in patients with left ventricular systolic dysfunction was one of the incentivized activities. As a consequence of the high baseline use of ACE inhibitors and ARBs, there was little scope to initiate these agents. There was also limited opportunity to increase the dose of these drugs as a high proportion (55%) of subjects was already receiving the recommended dose at baseline. Furthermore, the dose was increased in only about one-third of eligible patients in the intervention group (compared with 19% of those in the usual care group), presumably because of tolerability and safety considerations, perhaps indicating that the rate of use and dosing of these drugs may have already approached a ceiling level. Certainly, the rate of use of ACE inhibitors and ARBs and the doses that they were used at in our trial equal or exceed those in recent heart failure trials (despite our patients being more elderly than in these other trials)24,25 and a national audit in the UK.26

Although there was more scope to improve β-blocker prescribing, initiation of this type of medication and increase in the dose of β-blocker were infrequent in both treatment groups. This lack of success may indicate that the brief period of tuition used to prepare the non-specialist pharmacists in our trial was insufficient. Unfamiliarity with the use of β-blockers, and persisting concerns about their tolerability and safety, in left ventricular systolic dysfunction among family doctors at the time our trial started, as well as the high prevalence of respiratory disease in our population may also have limited β-blocker use. Additionally, patients may also have been unwilling to take an additional medication given the high rate of multi-drug regimens in the population studied (47% were receiving more than eight medications). Nevertheless, although the rate of use of β-blockers was disappointing, the proportion of patients taking 100% or more of the recommended dose of β-blocker by the end of the first year in the pharmacist intervention group (26%) compared favourably with the Systolic Heart failure treatment with If inhibitor ivabradine Trial (SHIFT) where this proportion was 26% at baseline.24

A second explanation for the lack of effect of our intervention was the relatively low frequency of hospital admission for heart failure which meant that the majority of events contributing to the primary composite outcome were fatal. In addition, only half of the deaths that occurred were attributed to cardiovascular causes. We excluded high-risk patients under the care of specialist heart failure nurses. Furthermore, falling hospital admission rates for heart failure have been reported in several countries15,2731 and, recently, cardiovascular deaths, as a proportion of overall deaths, have also been reported to be declining in patients with heart failure.32 Consequently, there were fewer non-fatal and fatal events which might have been reduced through greater use of ACE inhibitors, ARBs and β-blockers. We believe that these two factors—only a modest improvement in the use of disease-modifying treatment coupled with a low rate of modifiable events—are the most likely explanation for a lack of improvement in clinical outcomes in our trial.

The low rate of use of aldosterone antagonists in our trial was due to the exclusion of more severely ill patients who were under the care of the specialist heart failure nurse service and hospital clinics. At the time of our trial, aldosterone antagonists were only indicated in such patients.1 Clearly, if such a trial were repeated today, the use of aldosterone antagonists would be encouraged by pharmacists.

Potentially harmful medications, e.g. rate-limiting calcium channel blockers, non-steroidal anti-inflammatory drugs, tricyclic antidepressants, oral corticosteroids, and glitazones, were prescribed in a very small proportion of patients at baseline, and pharmacist intervention did not lead to any greater discontinuation of these medicines, when compared with usual care during follow-up.

As explained in the ‘Methods’ section, we had to use a cluster randomization design. We ensured good internal validity by accounting for the clustered nature of the data in our sample size calculations and ensuring blinding to the allocation status of those recruiting individuals into the trial and good external validity by providing information on numbers approached, recruited, and lost to follow-up. We therefore followed best practices in relation to this type of trial. Furthermore, although a major limitation of this design is the lack of similarity of the study groups, our treatment groups were well matched.12,13

There were several limitations to our trial. The study pharmacists were not trained to elicit signs and symptoms of heart failure. We did not collect reasons why patients might have been ineligible for a specific treatment or unable to tolerate it (collecting this information on the control group might have caused contamination). Ejection fraction was not reported to general practitioners, and natriuretic peptides measurements were not available in primary care at the time our patients were recruited. Although our trial did not achieve its goal of improving clinical outcomes, it did demonstrate that modest and sustained improvements in the prescribing of disease-modifying medications can be achieved by a brief, focused collaborative intervention delivered by non-specialist pharmacists given only a short period of training. However, the short period of training and brevity of intervention may also have been limitations of our study. This was particularly true in relation to β-blockers where our intervention had a disappointingly small effect on the use of these drugs. Lessons for future trials may be that more intense training, more patient visits, and selective involvement of hospital specialists might be required to fully optimize treatment. It is also possible that the modification of other treatments that we did not target, such as diuretics, may have improved the outcome and this could also be examined in future studies. While this type of intervention may not benefit all patients, it might improve clinical outcomes if aimed at those in most need in terms of deficient background treatment or at those at higher risk of modifiable events. This is a question that may be considered in future comparative-effectiveness trials.

Contributors

All authors except P.S.J. designed the study and oversaw its conduct. All authors participated in interpretation of the data and writing of the report.

Funding

The primary funder was the NHS in Greater Glasgow and Clyde. The funder played no role in data analysis or interpretation, writing of the report, or the decision to submit for publication. The authors designed the trial and oversaw its conduct. The report was prepared by all authors who had unrestricted access to the data and made the decision to submit for publication.

Conflict of interest: none declared.

Acknowledgements

All participating general practices in Greater Glasgow and Clyde. Study pharmacists: Fiona Allen, Janet Black, Lorna Brown, Sandra Cahill, Hilary Campbell, Jan Currie, Colin Dougall, Noreen Downes, Anfrances Duggan, Lynn Duthie, Marie Galloway, Alia Gilani, Alison Hair, Heather Harrison, Victoria Hunt, Chris Johnson, Fiona Lambie, Marjorie McGhie, Pauline McLeod, Mari-Anne McLean, Raj Sabharwal, Shona Shaw, Kirstin Shearer, Sharon Smart, Iain Spierits, Sheila Tennant, Lynn Walker. Primary care adviser: Des Spence. Pharmacy support: Sarah Darroch, Karen Vint, Alister McLaren, Nicola Watson.

References

View Abstract