Flow diagram of studies assessed and included in meta-analysis. RCT indicates randomized controlled trial.
Forest plot of total mortality. CI indicates confidence interval; OR, odds ratio (random-effects model).
Forest plot of all-cause hospitalization rate. CI indicates confidence interval; OR, odds ratio (random-effects model).
Forest plot of heart failure hospitalization rate. CI indicates confidence interval; OR, odds ratio (random-effects model).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Koshman SL, Charrois TL, Simpson SH, McAlister FA, Tsuyuki RT. Pharmacist Care of Patients With Heart FailureA Systematic Review of Randomized Trials. Arch Intern Med. 2008;168(7):687–694. doi:10.1001/archinte.168.7.687
Copyright 2008 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2008
While the role of multidisciplinary teams in the treatment of patients with heart failure (HF) is well established, there is less evidence to characterize the role of individual team members. To clarify the role of pharmacists in the care of patients with HF, we performed a systematic review evaluating the effect of pharmacist care on patient outcomes in HF.
We searched PubMed, MEDLINE, EMBASE, International Pharmaceutical Abstracts, Web of Science, Scopus, Dissertation Abstracts, CINAHL, Pascal, and Cochrane Central Register of Controlled Trials for controlled studies from database inception to August 2007. We included randomized controlled trials that evaluated the impact of pharmacist care activities on patients with HF (in both inpatient and outpatient settings). Summary odds ratios (ORs) with 95% confidence intervals (CIs) were calculated using a random-effects model for rates of all-cause hospitalization, HF hospitalization, and mortality.
A total of 12 randomized controlled trials (2060 patients) were identified. Extent of pharmacist involvement varied among studies, and each study intervention was categorized as pharmacist-directed care or pharmacist collaborative care using a priori definitions and feedback from primary study authors. Pharmacist care was associated with significant reductions in the rate of all-cause hospitalizations (11 studies [2026 patients]) (OR, 0.71; 95% CI, 0.54-0.94) and HF hospitalizations (11 studies [1977 patients]) (OR, 0.69; 95% CI, 0.51-0.94),and a nonsignificant reduction in mortality (12 studies [2060 patients])(OR, 0.84; 95% CI, 0.61-1.15). Pharmacist collaborative care led to greater reductions in the rate of HF hospitalizations (OR, 0.42; 95%CI, 0.24-0.74) than pharmacist-directed care (OR, 0.89; 95% CI, 0.68-1.17).
Pharmacist care in the treatment of patients with HF greatly reduces the risk of all-cause and HF hospitalizations. Since hospitalizations associated with HF are a major public health problem, the incorporation of pharmacists into HF care teams should be strongly considered.
Heart failure (HF) is associated with significant morbidity and mortality.1 Although the global burden of illness attributed to HF is already high, it is expected to continue to rise, given increased survival rates after acute myocardial infarction and an aging population.1 Despite dramatic improvements in treatment over the last decade, a diagnosis of HF still portends a poor prognosis.1 Mortality rates have changed little, and within 1 year of initial hospitalization for HF, up to 50% of patients will be rehospitalized and up to 40% will die.2
To a large extent, HF is a medically managed disease. Medications such as angiotensin-converting enzyme inhibitors and β-blockers are well established in the treatment of HF, reducing mortality and hospitalizations.3 However, despite pharmacotherapy, outcomes for patients with HF remain poor owing to the underuse of these agents.4Therefore, there has been an increasing focus on alternative models of care for these patients. Two recent systematic reviews demonstrated that multidisciplinary strategies for the management of HF reduce all-cause mortality, all-cause hospitalizations, and HF hospitalizations.5,6Although both reviews included some studies with pharmacist involvement, many HF clinics do not have a pharmacist team member, and the specific contribution of pharmacists within these multidisciplinary interventions was difficult to isolate owing to a paucity of detail in the published reports. Therefore, we conducted a systematic review to evaluate the impact of pharmacist care on the management of patients with HF and, in particular, contacted the primary study authors of multidisciplinary intervention trials that incorporated a pharmacist in an attempt to define the precise role of the pharmacist in these interventions.
We searched, with assistance from a librarian, the following electronic databases from their inception until August 2007: PubMed,MEDLINE, EMBASE, International Pharmaceutical Abstracts, Web of Science,Scopus, Dissertation Abstracts, CINAHL, PASCAL, and the Cochrane Central Register of Controlled Trials. Language restrictions were not applied. Search items included pharmacy-related terms (pharmacist, pharmaceutical care, pharmaceutical services, clinical pharmacy services, hospital pharmacy, community pharmacy, and pharmacy) and HF-related terms (heart failure, congestive heart failure, heart disease, cardiomyopathy, and ventricular dysfunction). Also, bibliographies of identified studies were hand searched.
Two of us (S.L.K. and T.L.C.) independently screened the citations from the literature search for eligibility. Studies were included if they tested (in randomized controlled trials) the impact of pharmacist care on patients with HF (compared with no pharmacist care) on the outcomes of all-cause hospitalizations, HF hospitalizations, and all-cause mortality. Secondary outcomes included health-related quality-of-life measures and medication adherence. We contacted all primary study authors and asked them to fill out a standardized questionnaire to better define the role that the pharmacist played in each multidisciplinary team and to verify outcome definitions and published results. Publications were excluded if they were not randomized, if they did not have adequate description of the pharmacist's intervention and the author could not be contacted, or if they did not report the outcomes of interest. Disagreements were resolved by consensus. Based on information provided in the article or from corresponding author responses to our questionnaire,we classified pharmacist interventions using a priori–defined categories: pharmacist-directed care (pharmacist-initiated and managed intervention) or pharmacist collaborative care (member of a multidisciplinary team).
Data extraction was performed by 2 of us (S.L.K. and T.L.C.)independently using a standardized data collection form. Outcomes from individual studies were assigned according to the intention-to-treat principle. We documented all-cause hospitalizations and HF hospitalizations according to the definitions used by the authors of the individual studies. Hospitalization rates (all-cause or HF) were defined as the number of patients in each group who were hospitalized at least once for that diagnosis (only the first hospitalization was counted for patients with multiple hospitalizations). Randomized controlled trials were assessed for quality using the Jadad score and evaluated as to whether allocation concealment was adequately described.7
Data analysis was performed using Cochrane Review Manager software (RevMan 4.2.7; Nordic Cochrane Centre, Copenhagen, Denmark). A random-effects model was used to estimate the pooled odds ratios (ORs) and 95% confidence intervals (CIs) for rates of HF hospitalizations, all-cause hospitalizations,and mortality. Heterogeneity between studies was evaluated using the Higgins I2 test. Data for secondary outcomes are reported descriptively, and, where possible, pooled ORs (random-effects model) and weighted mean differences were calculated. Sensitivity analyses were defined a priori to include analysis based on quality of studies (determined according to Jadad scores) and type of pharmacist intervention (pharmacist-directed vs collaborative care). Two approaches were used to identify potential sources of heterogeneity. First, an indirect comparison of pharmacist-directed care vs pharmacist collaborative care was done using the method of Song et al.8 Second, we used the process described by Tobias9 to assess the influence of a single study on pooled results.
Of the 3115 citations identified in our search, 12 randomized controlled trials (RCTs) (2060 patients) met the eligibility criteria for our analysis (Figure 1 and Table 1). All primary study investigators were contacted to clarify the published data, and 11 of 12 provided further data.
Based on information from the articles and corresponding authors,we determined that 7 studies compared pharmacist-directed interventions with usual care,10-16 and 5 studies compared pharmacist collaborative care with usual care.17-21 Details about study settings, patient demographics, intervention frequency, end point ascertainment, and usual care for each trial are provided in Table 1. The pharmacist-specific interventions in these studies typically involved education on both HF and evidence-based HF medications, including self-monitoring, medication management, and facilitation of compliance.Details on the specific educational interventions offered by the pharmacists (as reported by each of the primary study authors) are also provided in Table 1.
Eleven of the 12 trial authors responded to our survey to define the exact role of the pharmacist in each multidisciplinary team. Seven trialists identified the pharmacist as the key driver of the intervention (pharmacist-directed care), with their responsibilities including medication and HF education, self-monitoring, recommendations to physicians,and adherence aids (Table 1).10-16 Four trialists identified the pharmacist as one of the members of the multidisciplinary team (pharmacist collaborative care).17,18,20,21 We received no response from 1 study author; however, it was clear from the article that the pharmacist worked in conjunction with a clinical nurse specialist and was therefore identified as providing collaborative care.19
The studies were of variable methodological quality. Because of the nature of the interventions, none of the studies was double-blind.Only 6 of the 12 RCTs adequately described allocation concealment.12,14-17,20 Jadad scores are listed in Table 1.
All 12 RCTs (2060 patients) reported all-cause mortality (Figure 2). One study15 showed a significant difference in all-cause mortality between intervention and control. The pooled estimate of the 12 RCTs showed a nonsignificant reduction in mortality for pharmacist care compared with control (OR, 0.84; 95% CI, 0.61-1.15; I2, 19%).
Eleven RCTs (2026 patients) reported all-cause hospitalization rates (ie, the number of patients hospitalized at least once). The pooled OR for all-cause hospitalization rates demonstrated a significant benefit of pharmacist care (OR, 0.71; 95% CI, 0.54-0.94) (Figure 3). There was, however, heterogeneity in these results (I2, 50%).
Of the 11 RCTs (1977 patients) reporting HF hospitalization rates, 3 demonstrated statistically significant reductions with pharmacist care,18-20 and the pooled-effect estimate revealed a significant benefit with pharmacist care (OR, 0.69; 95% CI, 0.51-0.94) (Figure 4). There was also some heterogeneity in these results (I2, 40%).
Health-related quality-of-life data are presented in Table 2. Health-related quality of life was measured in 7 studies; 6 studies used disease-specific measures,10,11,13,14,16,20 and 5 studies used generic measures.11,13-15,20 The way in which data were reported and the small number of studies using health-related quality of life precluded pooling of data. Adherence was measured as an outcome in 7 studies (Table 3).10-13,15,16,20 Methods for measuring adherence varied substantially among studies.Of the 3 studies that collected adherence data using community pharmacy refill records,10,12,20 only 1 found significant differences (favoring the intervention) between study groups.10 However, this study had adherence data available for only 28% of the total study population.Of the 3 studies that reported adherence using patient self-report,10,13,16 1 reported significant differences (favoring intervention).13 Two studies used an electronic monitoring system to measure adherence, and both studies demonstrated that patients in the control group had lower adherence.11,16
The indirect comparisons of pharmacist-directed interventions and collaborative pharmacist care showed no significant difference between the 2 types of intervention in their effects on mortality or rate of all-cause hospitalizations (P = .40and P = .40, respectively). In terms of HF hospitalization rates, the effects of these interventions were significantly different, with pharmacist collaborative care being associated with a greater risk reduction (P = .02).
Using the method defined by Tobias,9 we looked at each primary end point to determine whether removal of any 1 study would dramatically affect the results. The pooled OR did not change more than 8% after removal of each study for the end points of HF hospitalizations and all-cause hospitalizations. For mortality, the removal of the study by López Cabezas et al15 increased the OR by 15%; however, the result was still nonsignificant (P = .75).
The results for the sensitivity analyses based on study quality are presented in Table 4. Only 4 studies (n = 1035) had Jadad scores that were higher than 3.11,12,14,16 The studies with Jadad scores that were lower than 2 had consistently more positive results than studies with Jadad scores that were higher than 3. In terms of allocation concealment, the results were similar when studies with adequate allocation concealment were compared with those with inadequate allocation concealment for the outcomes of all-cause hospitalizations and mortality.
This systematic review confirms the benefits of pharmacist care in reducing hospitalization in patients with HF. Interventions that include some element of pharmacist care reduced the rates of both all-cause hospitalization and HF hospitalization by almost one-third. Because HF is one of the leading causes of hospitalization,22 we recommend the addition of a pharmacist to the HF team. Other studies have confirmed that a substantial proportion of HF exacerbations can be attributed to medication misadventures,highlighting the potential importance of pharmacists on the HF team.23 These results are consistent with an earlier systematic review of multidisciplinary care in HF5 but extends this earlier work by including data from 8 trials that were not included in the earlier review and by focusing specifically on the impact of the pharmacist within the setting of the multidisciplinary team (as defined by the primary authors of each of these trials).
There are several plausible explanations for our findings of reductions in HF hospitalizations and all-cause hospitalizations but no change in mortality. Given that the majority of events examined were hospitalizations and our sample size was relatively small, it is unlikely that we would be able to show a decrease in mortality as a result of low statistical power. Also, the duration of follow-up in these studies ranged from 2 days to 12 months, with the majority of follow-ups lasting only 6 months or less, likely too short to see an impact on mortality.
Given the significant heterogeneity in many of the primary outcomes,we sought to determine the potential sources. First, we analyzed the data in predefined categories (pharmacist-directed care or pharmacist collaborative care). When these categories were compared, we found no difference in the type of intervention and outcomes for mortality and all-cause hospitalizations. For HF hospitalizations, pharmacist collaborative care did appear to be more beneficial than pharmacist-directed care. This finding is not surprising given that medication management and patient education would complement care given by nurses, physicians, and other health care professionals. Second, we compared the data using the method described by Tobias9 to examine the influence of single studies on each outcome. No particular study influenced the pooled OR when taken out of the analysis.Finally, we evaluated results by study quality and, not surprisingly,found that lower-quality studies reported greater beneficial effects with the tested interventions. The Jadad score is well recognized for randomized controlled trials; however, it may not be the best for practice research, where blinding is not possible. In the absence of alternative quality scores, this sensitivity analysis should be interpreted as hypothesis generating only.
There are a number of limitations that warrant discussion. There were notable differences in pharmacist activities between studies, making it difficult to define precisely which intervention provides the best outcomes (even after contact with the primary study authors).There were also differences in terms of patient population and settings, which included both hospitalized patients and ambulatory patients, making it difficult to elucidate which patient population would most likely benefit. It is also likely that there were different cointerventions across the studies that could not be accounted for.
In addition to contributing to the current body of literature supporting the beneficial effects of multidisciplinary teams in the treatment of patients with HF, our findings further describe the beneficial role of the pharmacist in the treatment of patients with HF. Because HF results in more than 1 000 000 hospitalizations each year in the United States, a 30% reduction would have a substantial impact.24 From our results, we can infer that including a pharmacist in the care of patients with HF, particularly within a multidisciplinary team, is beneficial and should be strongly considered by health policy makers.
Correspondence: Ross T. Tsuyuki,BSc(Pharm), PharmD, MSc, FCSHP, Division of Cardiology, EPICORE Centre/COMPRIS, University of Alberta, 220 College Plaza, 8215-112 St NW, Edmonton,AB T6G 2C8 Canada (email@example.com).
Accepted for Publication: November 2, 2007.
Author Contributions: Dr Koshman and Ms Charrois had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design:Koshman, Charrois, Simpson, McAlister, and Tsuyuki. Acquisition of data: Koshman and Charrois. Analysis and interpretation of data: Koshman, Charrois, Simpson,McAlister, and Tsuyuki. Drafting of the manuscript: Koshman and Charrois. Critical revision of the manuscript for important intellectual content: Simpson, McAlister,and Tsuyuki. Statistical analysis: Charrois,Simpson, and McAlister. Obtained funding:Tsuyuki. Administrative, technical, and material support: Koshman, Charrois, and Tsuyuki. Study supervision: McAlister and Tsuyuki.
Financial Disclosure: None reported.
Funding/Support: This study was supported by COMPRIS (Centre for Community Pharmacy Research and Interdisciplinary Studies), http://www.epicore.ualberta.ca/compris/index.html.
Previous Presentations: This study was presented in part at the Canadian Cardiovascular Congress; October 23, 2006; Vancouver, British Columbia, Canada; and at Cardiac Sciences Research Day; June 9, 2006; University of Alberta, Edmonton, Canada.
Additional Contributions: We thank the following authors of primary studies who answered our requests for further information: Marcel Bouvy, PhD; Miriam Fradette, BSc(Pharm); Wendy Gattis-Stough, PharmD; Femida Gwadry-Sridhar, PhD; Adel Sadik,PhD; Simon Stewart, PhD; Darren Triller, PharmD; Sam Varma, PhD; Carmen López Cabezas, PhD; Richard Holland, PhD; and Michael Murray, MPH. We also thank Ben Vandermeer for statistical assistance and our librarian, Marlene Dorgan.