Causes and Consequences of Nonpersistence With Heart Failure Medication

Article Outline

• Abstract
• Methods
• Results
• Discussion
• References
• Copyright

Persistence with therapy may be more easily and objectively identified in the clinical setting than compliance and recent work has shown it to be linked to mortality in heart failure (HF). The aim of this study was to determine the extent, causes, and clinical impact of nonpersistence with disease-modifying therapy in a retrospective cohort study of 183 patients with systolic HF participating in a disease management program. The main outcome measurements were reasons/determinants of nonpersistence and its impact on hospitalizations. Fifty-three patients (29%) had 74 separate occurrences of nonpersistence with disease-modifying therapy. There was no medical reason for discontinuing medications in 50% of occurrences, whereas medication was discontinued for an adverse reaction in 30% and for a justified medical reason in 15% of occurrences. Nonpersistence was a significant predictor of all-cause readmission (hazard ratio 3.20, 95% confidence interval 1.74 to 11.37) and cardiovascular readmission (hazard ratio 4.45, 95% confidence interval 1.74 to 11.37). In the adjusted model, there was no significantly increased risk of HF readmission (hazard ratio 2.41, 95% confidence interval 0.88 to 6.62). In conclusion, nonpersistence with HF therapy is common, is often not medically justified, and is associated with an increased risk of hospitalization.

Persistence with medication has recently been shown to have prognostic power in heart failure (HF).¹ It is an indirect measurement of adherence that can be easily obtained in a medication reconciliation process², ³ and it may be more reliable and practical to use in the clinical setting than other subjective⁴, ⁵, ⁶ and electronic², ⁵, ⁶, ⁷, ⁸ measurements of compliance. However, more information is needed on the causes of nonpersistence and its associated clinical outcomes. Moreover, although multidisciplinary disease-management programs (DMPs) have been advocated as solutions to adherence problems in HF, the nature and extent of nonpersistence in this setting are unknown. We hypothesized that nonpersistence with HF therapy is prevalent in a well-characterized stable HF population despite participation in a DMP, predominantly arises from adverse drug reactions (ADRs) or identifiable medical reasons, and is associated with increased morbidity.

Back to Article Outline

Methods 

This is a retrospective cohort study of consecutive patients with stable systolic HF presenting for an annual review to our HF unit as part of our multidisciplinary DMP. The majority of patients were recruited to the DMP after an HF admission and the details of this cardiologist-led hospital and outpatient program have been published elsewhere.⁹, ¹⁰ Other methods of entry into the DMP were after referral in the nonacute state from community or other hospital services. Important nonmedical components of care included outpatient response to clinical deterioration and patient/family education by specialist nurses and pharmacists focusing on self-care and medications.

All patients undergo full clinical assessment, medicine review, and echocardiography at an annual interval. Using this visit as a point of selection for stable patients with left ventricular systolic dysfunction who had comprehensive physical, clinical, echocardiographic, and nursing assessments, we retrospectively examined the disease-modifying therapy profiles of all patients at all clinic visits since first contact with the service. As part of our DMP, patients are also interviewed at each scheduled and unscheduled clinic visit by the HF nurse specialist who enters current medication details at the start of the visit before medical assessment (name, dose, frequency of administration) on an electronic database. This information is obtained using a combination of the following methods: examination of a patient's medication brought to the clinic, patient/carer recall with confirmation using patient notes, and patient-consented contact with a community pharmacy or call to a family physician to clarify medications being taken. In addition, our patients use a HF communication booklet in which their medications are noted and family physicians are encouraged to note any changes in medications with reasons. At the end of the visit, after clinical assessment by the HF unit team (physician, nurse, and allied health care professional as appropriate), the updated physician-derived medication profile is recorded onto the database with a description of any medical reasons for changes from the previous patient review and changes made during the current review visit.

We define disease-modifying therapy in HF as those medicines known to improve survival in HF due to left ventricular systolic dysfunction, namely angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, β blockers, and aldosterone antagonists. We excluded patients with preserved systolic function HF because there are few definitive data on medical therapy known to improve outlook in this population. By comparing the patient-reported medication profile at the start of each visit with the physician-prescribed medication profile at the end of the preceding visit, we identified episodes of nonpersistence. By examining medication changes between the start and the end of the clinic visit, we identified all other medication changes associated with disease-modifying therapy. For the purposes of this study, a switch of medication to another within the same class or analogous class (e.g., angiotensin-converting enzyme inhibitor to angiotensin receptor blocker) did not count as nonpersistence. For all episodes of nonpersistence with disease-modifying therapy, patient notes for that visit were examined to confirm the details of the medication adjustment and to determine any noted reasons for medication adjustment. Those patients who discontinued disease-modifying therapy for any period since recruitment to the multidisciplinary program were categorized in the nonpersistence (NP) group. All other patients were categorized in the persistence (P) group. In addition, for all patients who discontinued disease-modifying therapy, we determined whether they later restarted the same medication or an analogous agent.

The outcome measurements were reasons/determinants of nonpersistence and its impact on hospitalizations (cardiovascular, noncardiovascular, and HF), scheduled and unscheduled clinic visits, baseline health-related quality of life (Minnesota Living with HF Questionnaire¹¹), baseline Hospital Anxiety and Depression Scale scores, and average time to nonpersistence with disease-modifying therapy. Using baseline demographics of the population, we identified predictors of poor persistence. Several strategies were used to ensure optimal collection of outcome measurement data. Patients and their families were advised to inform the unit of any hospital admission. The HF nurse requested information about the admission and determined the reason for its occurrence (hospital discharge letter/request for hospital notes). In addition, chart reviews and reviews of the database at St. Vincent's University Hospital (Dublin, Ireland) were conducted to determine the cause of readmission. All patients were seen at the clinic for an annual review where detailed interviews about any adverse events were recorded. Patients lost to follow-up were contacted by telephone or their general practitioners were contacted for information.

Comparisons between the P and NP groups were conducted using independent-sample t tests for normally distributed continuous variables and Mann-Whitney U test for non-normal distributions (2-tailed, alpha = 0.05). Chi-square analysis was used to compare categorical variables. The Cox proportional hazards model was used to estimate independent predictors (univariable and multivariable) of nonpersistence, all-cause emergency readmissions, cardiovascular readmissions, and HF readmissions. For multivariable analysis, p value of the partial likelihood ratio test was used to confirm if a covariate was significant and the coefficients of the remaining variables were assessed to determine if important (>20%) changes had occurred on variable exclusion. All clinically meaningful interactions were tested. On selection of the final multivariable model, all excluded variables were reintroduced into the model individually to ensure that significant confounders were not omitted. The status of nonpersistence was evaluated every 6 months and this categorical time-dependent covariate was introduced into the multivariable model to test the association of nonpersistence with clinical outcomes over the follow-up period.¹² Age, gender, renal function, New York Heart Association functional classification, and ejection fraction were deemed clinically important and included in the multivariable models predicting all-cause, cardiovascular, and HF readmissions irrespective of significance levels. All analyses were conducted using SPSS 12 (SPSS, Inc., Chicago, Illinois).

Back to Article Outline

Results 

Of 241 consecutive attendances to the annual review clinic by stable patients participating in our DMP from September 2005 to June 2006, we excluded 58 who had preserved systolic function HF. Follow-up data were available for all 183 patients included in this study (baseline demographics are listed in Table 1). Fifty-three of these (29%) were included in the NP group because they had ≥1 occurrence of nonpersistence with disease-modifying therapy during an average 3.0 ± 1.63-year (range 0.63 to 7.47) follow-up period. Of the 53 patients in the NP group, there were 74 separate occurrences of nonpersistence and the average time to first occurrence from recruitment to the program was 456 ± 306 days (range 3 to 1,932). There was neither an explicit patient reason nor an identifiable medical reason for nonpersistence in 33 of these patients (62%), accounting for 37 (50%) of occurrences. Full details of the causes of nonpersistence are presented in Table 2.

Table 1. Nonpersistence with heart failure medication: baseline demographics of study population

	NP Group	P Group	p Value
No. of patients (%)	53(29%)	130(71%)
Age (yrs)	70.9±10.3	69.4±11.6	0.40
Men	37(70%)	86(67%)	0.68
NYHA class 1/2/3	7(13%)/43(81%)/3(6%)	37(29%)/87(66%)/6(5%)	0.027
Cause
Ischemic	36(68%)	76(59%)	0.23
Hypertensive	8(15%)	17(13%)	0.72
Idiopathic	8(15%)	19(15%)	0.93
Valvular	1(2%)	14(11%)	N/A
Alcoholic	2(4%)	14(11%)	0.16
Viral	3(6%)	8(7%)	N/A
Postpartum	0(0%)	1(1%)	N/A
Co-morbidities
Increased cholesterol	29(55%)	451(39%)	0.06
Diabetes	14(26%)	22(17%)	0.14
Chronic obstructive pulmonary disease/asthma	7(13%)	25(19%)	0.33
Arthritis	9(17%)	19(15%)	0.76
Cancer	5(9%)	14(11%)	0.79
Previous HF admission	10(19%)	43(33%)	0.004
Clinical parameters
Diastolic blood pressure (mm Hg)	74±11	76±12	0.28
Systolic blood pressure (mm Hg)	127±21	130±21	0.25
Heart rate (beats/min)	71±13	73±13	0.43
Ejection fraction	31.6±11.7	38.4±14.1	<0.001
Medications
Angiotensin-converting enzyme inhibitor/angiotensin receptor blocker	49(93%)	129(99%)	0.03
Diuretic	43(81%)	121(93%)	0.02
Aspirin	28(53%)	70(54%)	0.90
β blocker	38(72%)	94(72%)	0.93
Nitrate	28(53%)	67(52%)	0.87
Statin	17(32%)	51(39%)	0.36
Aldosterone receptor blocker	9(17%)	18(14%)	0.59
Biochemistry
B-type natriuretic peptide (pg/ml)	316.7±368.6	264.2±334.6	0.50
Hemoglobin (g/dl)	12.3±4.3	13.8±1.6	0.13
Sodium (mmol/L)	139.5±3.5	138.5±3.8	0.24
Creatinine (μmol/L)	124.8±36.9	104.4±31.1	<0.001
Urea (mmol/L)	10.6±9.6	7.6±3.6	<0.001
Potassium (mmol/L)	4.4±0.5	4.3±0.5	0.26

Values are numbers of patients (percentages) or means ± SDs.

NYHA = New York Heart Association.

Table 2. Nonpersistence with heart failure medication: reasons identified for 74 occurrences of nonpersistence in 53 patients and number of medications restarted

	No Documented Reason	Justified Clinical Reason⁎	Patient-Reported ADR†	Other‡	Total	Restarted	Delay Before Restart (days)
Angiotensin-converting enzyme inhibitor	11	4	3	1	19(26%)	8	118±91(11–268)
β blocker	6	3	12	1	22(30%)	11	413±424(35–1,154)
Angiotensin receptor blocker	7	0	6	0	13(18%)	5	181±133(34–330)
Aldosterone antagonist	13	4	1	2	20(27%)	1	298
Total	37(50.0%)	11(15%)	22(30%)	4(5%)	74(100%)	25	267±242(11–1,154)

Values are numbers of subjects (percentages) or averages ± SDs (ranges).

Twenty patients (38%) in the NP group were successfully restarted on medication after 25 (34%) occurrences of discontinuation as part of a DMP clinical review after a delay of 267 ± 242 days (range 11 to 1,154). Details by medication category are presented in Table 2. In the majority of these cases there was no documented reason for discontinuation (n = 14). Medications were also restarted because a patient-reported ADR was considered unrelated to the medication (n = 8), a prescribing error (n = 1), or an inappropriate medical decision due to patient stability (n = 2).

Univariable and multivariable predictors of nonpersistence are presented in Table 3. Multivariable predictors of HF, all-cause, and cardiovascular emergency readmission are presented in Table 4. A Kaplan-Meier survival curve for all-cause readmission for the 2 groups is presented in Figure 1. There were no demographic differences or differences in rehospitalization rates (p = 0.55) in those patients who stopped therapy with or without documented reason and there was no detectable impact of restarting therapy on outcome.

Table 3. Nonpersistence with heart failure medication: univariable and multivariable predictors of nonpersistence in patients with heart failure attending an outpatient disease management program (Cox proportional hazards model)

	HR	CI	p Value
Univariable
Age (yrs)	1.013	0.984–1.043	0.41
Male gender	1.061	0.588–1.914	0.84
Ischemic cause	1.505	0.767–2.952	0.23
Previous HF admission	0.382	0.192–0.760	0.004
Diabetes	1.762	0.821–3.782	0.15
Pulmonary disease	0.639	0.258–1.583	0.33
Cancer	0.863	0.295–2.529	0.79
NYHA (II and III)	2.394	1.073–5.339	0.028
Systolic blood pressure	0.995	0.982–1.009	0.50
Diastolic blood pressure	0.988	0.965–1.012	0.32
Heart rate	0.991	0.969–1.013	0.41
Urea	1.022	1.001–1.045	0.34
Creatinine	1.012	1.005–1.019	0.001
Sodium	1.061	0.981–1.146	0.14
Potassium	1.239	0.708–2.167	0.45
B-type natriuretic peptide	1.001	1.000–1.002	0.20
Ejection fraction	0.972	0.984–0.991	0.004
No. of drugs	0.960	0.852–1.082	0.50
Median >7 drugs	1.001	0.579–1.731	0.99
No. of doses	0.956	0.881–1.037	0.28
Multivariable
Creatinine	1.019	1.008–1.030	0.005
Ejection fraction	0.961	0.935–0.988	0.005
Previous HF admission	0.314	0.138–0.716	0.006
Age	1.007	0.976–1.040	0.65
Gender	0.791	0.357–1.751	0.56

CI = confidence interval; HR = hazard ratio. Other abbreviation as in Table 1.

Table 4. Nonpersistence with heart failure medication: multivariable predictors of all-cause and cardiovascular readmission in patients with heart failure attending an outpatient disease management program (Cox proportional hazards model)

	HF Readmissions		Cardiovascular Readmissions		All-Cause Readmissions
	HR(95% CI)	p Value	HR(95% CI)	p Value	HR(95% CI)	p Value
NP group⁎	2.41(0.88–6.62)	0.09	4.45(1.74–11.37)	0.002	3.20(1.80–5.68)	<0.001
Previous HF admission	†		0.23(0.05–0.99)	0.048	0.42(0.22–0.82)	0.01
Age	1.001(0.96–1.05)	0.96	1.02(0.98–1.07)	0.33	1.02(0.99–1.05)	0.13
Male gender	0.84(0.30–2.36)	0.74	0.79(0.31–1.99)	0.62	0.95(0.52–1.74)	0.87
Creatinine	1.01(0.99–1.02)	0.36	0.99(0.98–1.00)	0.17	0.99(0.99–1.01)	0.65
NYHA class	2.42(0.55–10.66)	0.25	4.45(0.59–33.8)	0.15	1.33(0.63–2.79)	0.46
Ejection fraction	0.99(0.96–1.02)	0.53	0.99(0.95–1.02)	0.46	0.99(0.97–1.01)	0.21

Abbreviations as in Table 1, Table 3.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Kaplan-Meier survival curve comparing the P (green) and NP (blue) groups in time to all-cause emergency readmission.

Thirty-four patients in the NP group were hospitalized during the follow-up period, consuming a total of 868 bed-days (average length of stay 11.3 days). Hospitalizations were for HF (n = 20), cardiovascular causes (n = 24), and other causes (n = 41). This compares with 35 patients in the P group who consumed 522 bed-days (average length of stay 10.8 days, p = 0.85 vs NP group). Hospitalizations in the P group were for HF (n = 13), cardiovascular cause (n = 11), and other causes (n = 31).

There was a higher rate of unscheduled clinic visits in the NP group (4.3 ± 5.8 per patient over follow-up period, total 230, range 0 to 37) compared with the P group (1.5 ± 2.7, 194, 0 to 21, p = 0.001 vs NP group). This may reflect more frequent clinical deterioration in the NP group and may also explain the higher rate of scheduled clinic visits (18.8 ± 13.7 per patient, total 988, range 4 to 90) compared with the P group (13.3 ± 6.3, 988, 2 to 31, p = 0.006 vs NP group).

There was no statistically significant difference between total health-related quality of life scores using the Minnesota Living with HF Questionnaire in subsets of the P group (n = 91, 20.8 ± 19.1) and NP group (n = 37, 25.1 ± 18, p = 0.24) for whom these data were available. However, there were higher anxiety scores in the NP group (n = 39, 5.1 ± 3.6) compared with the P group (n = 86, 3.7 ± 3, p = 0.04 vs NP group). Also, higher levels of depression were seen in the NP group (n = 41, 5.8 ± 3.9) compared with the P group (n = 87, 4.4 ± 3.4, p = 0.034 vs NP group).

Back to Article Outline

Discussion 

This study demonstrates that nonpersistence in a stable community-based HF population is common and is associated with increased morbidity including hospitalizations, clinic visits for clinical deterioration, and increased levels of anxiety and depression. It builds on data that demonstrate the prognostic power of nonpersistence with HF therapy¹ by showing that nonpersistence is without documented medical reason in ½ of occurrences and remains common despite participation in a DMP. Indeed, only 20% of occurrences arose from a deliberate prescriber decision or prescribing/dispensing error and 42% of the remainder were successfully restarted on therapy at review clinics, albeit after a substantial delay. This suggests that unexplained, and perhaps unwarranted, patient adherence issues contributed to the nonpersistence and associated morbidity. Nonpersistence may be a more reliable measurement of outcome than self-reported compliance,³, ⁴, ⁵, ⁶ may be more practical to use in the clinical setting than electronic medication event monitoring systems,⁵, ⁶, ⁷, ⁸ and may be an important treatment target in long-term follow-up of patients with HF.¹³

Our study also supports and adds to the work of Gislason et al¹ by drawing similar results on the relation between persistence and outcome in a well-characterized HF population independently of important clinical variables in HF that were not available in the Danish registry study. Indeed, the predictive power of nonpersistence on outcome was greater in our dataset than age or New York Heart Association classification, perhaps reflecting the intensive nature of the DMP provided to these patients, which is focused on these HF risk factors. The intensity of the DMP may also explain the long delay (>1 year on average) before first occurrence of nonpersistence and is consistent with other studies that have shown that, although adherence improves with intensive management, it reverts to original levels once usual care is reinstated.¹⁴

The study also demonstrates that previous HF admission on recruitment to the DMP, poor renal function, and lower ejection fraction were predictive of nonpersistence unlike conventional risk factors such as polypharmacy. These observations may reflect alterations in patient beliefs about medication after an acute event, increased susceptibility to ADRs in more vulnerable patients with HF, and the attention paid by the multidisciplinary team to conventional risk factors for poor adherence. Furthermore, they are in accordance with previous observations about determinants of poor adherence in HF.⁴, ¹⁵, ¹⁶, ¹⁷, ¹⁸, ¹⁹ The data underline the benefit of a regular review clinic for patients with stable HF as part of a DMP and support the role of a pharmacist for regular adherence monitoring in the longer term.²⁰, ²¹

This study is limited by lack of information on patient self-care behavior educational status, social status, and beliefs about taking HF medications. Our definition of nonpersistence may be oversensitive, because those who stop medication for a short time are categorized with those who discontinue medication for prolonged periods. Furthermore, it is likely that undocumented short periods of nonpersistence occurred between clinic visits and the true level of nonpersistence may be greater than identified. Moreover, this was a relatively stable HF patient population and we cannot extrapolate these levels of nonpersistence to a sicker HF patient group.

Back to Article Outline

References 

1. Gislason GG, Rasmussen JN, Abildstrom SZ, Schramm TK, Hansen ML, Buch P, et al Persistent use of evidence-based pharmacotherapy in HF is associated with improved outcomes. Circulation. 2007;116:737–744
   • View In Article
   • CrossRef
2. Burnier M. Medication adherence and persistence as the cornerstone of effective antihypertensive therapy. Am J Hypertens. 2006;19:1190–1196
   • View In Article
   • MEDLINE
   • CrossRef
3. Varkey P, Cunningham J, Bisping DS. Improving medication reconciliation in the outpatient setting. Jt Comm J Qual Patient Saf. 2007;33:286–292
   • View In Article
   • MEDLINE
4. Wu JR, Moser DK, Chung ML, Lennie TA. Objectively measured, but not self-reported, medication adherence independently predicts event-free survival in patients with HF. J Card Fail. 2008;14:203–210
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (400 KB)
   • CrossRef
5. Balkrishnan R, Jayawant S. Medication (Adherence research in populations: measurement issues and other challenges). Clin Ther. 2007;29:1180–1182
   • View In Article
   • Full-Text PDF (66 KB)
   • CrossRef
6. MacLaughlin EJ, Raehl CL, Treadway AK, Sterling TL, Zoller DP, Bond CA. Assessing medication adherence in the elderly: which tools to use in clinical practice?. Drugs Aging. 2005;22:231–255
   • View In Article
   • MEDLINE
   • CrossRef
7. Mattke S, Jain AK, Sloss EM, Hirscher R, Bergamo G, O'Leary JF. Effect of disease management on prescription drug treatment: what is the right quality measure?. Dis Manag. 2007;10:91–100
   • View In Article
   • MEDLINE
   • CrossRef
8. Fennie KP, Bova CA, Williams AB. Adjusting and censoring electronic monitoring device data (Implications for study outcomes). J Acquir Immune Defic Syndr. 2006;43(suppl 1):S88–S95
   • View In Article
   • CrossRef
9. McDonald K, Ledwidge M, Cahill J, Quigley P, Maurer B, Travers B, et al. HF management: multidisciplinary care has intrinsic benefit above the optimization of medical care. J Card Fail. 2002;8:142–148
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (132 KB)
   • CrossRef
10. Ryder M, Murphy NF, McCaffrey D, O'Loughlin C, Ledwidge M, McDonald K. Outpatient intravenous diuretic therapy; potential for marked reduction in hospitalisations for acute decompensated heart failure. Eur J Heart Fail. 2008;10:267–272
    • View In Article
    • CrossRef
11. Rector TS, Cohn JN. Assessment of patient outcome with the Minnesota Living with HF questionnaire: reliability and validity during a randomized, double-blind, placebo-controlled trial of pimobendan (Pimobendan Multicenter Research Group). Am Heart J. 1992;124:1017–1025
    • View In Article
    • MEDLINE
    • CrossRef
12. Norusis MJ. SPSS 13.0: Advanced Statistical Procedures Companion. Upper Saddle River, NJ: Prentice-Hall; 2004;
    • View In Article
13. Newby LK, LaPointe NM, Chen AY, Kramer JM, Hammill BG, DeLong ER, et al. Long-term adherence to evidence-based secondary prevention therapies in coronary artery disease. Circulation. 2006;113:203–212
    • View In Article
    • CrossRef
14. Lee JK, Grace KA, Taylor AJ. Effect of a pharmacy care program on medication adherence and persistence, blood pressure, and low-density lipoprotein cholesterol: a randomized controlled trial. JAMA. 2006;296:2563–2571
    • View In Article
    • CrossRef
15. Schweitzer RD, Head K, Dwyer JW. Psychological factors and treatment adherence behavior in patients with chronic HF. J Cardiovasc Nurs. 2007;22:76–83
    • View In Article
16. Bouvy ML, Heerdink ER, Leufkens HG, Hoes AW. Patterns of pharmacotherapy in patients hospitalised for congestive HF. Eur J Heart Fail. 2003;5:195–200
    • View In Article
    • MEDLINE
    • CrossRef
17. Ekman I, Andersson G, Boman K, Charlesworth A, Cleland JG, Poole-Wilson P, et al. Adherence and perception of medication in patients with chronic HF during a five-year randomised trial. Patient Educ Couns. 2006;61:348–353
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (100 KB)
    • CrossRef
18. Granger BB, Swedberg K, Ekman I, Granger CB, Olofsson B, McMurray JJ, et al. Adherence to candesartan and placebo and outcomes in chronic HF in the CHARM programme: double-blind, randomised, controlled clinical trial. Lancet. 2005;366:2005–2011
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (124 KB)
    • CrossRef
19. Morgan AL, Masoudi FA, Havranek EP, Jones PG, Peterson PN, Krumholz HM, et al. Difficulty taking medications, depression, and health status in HF patients. J Card Fail. 2006;12:54–60
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (168 KB)
    • CrossRef
20. Sadik A, Yousif M, McElnay JC. Pharmaceutical care of patients with HF. Br J Clin Pharmacol. 2005;60:183–193
    • View In Article
    • MEDLINE
    • CrossRef
21. Koelling TM, Johnson ML, Cody RJ, Aaronson KD. Discharge education improves clinical outcomes in patients with chronic HF. Circulation. 2005;111:179–185
    • View In Article
    • CrossRef