Prescriptions of recommended heart failure medications can be correlated with patient and physician characteristics

Transcription

1 ORIGINAL PAPER Prescriptions of recommended heart failure medications can be correlated with patient and physician characteristics J. Koschack, H. H. Jung, M. Scherer, M. M. Kochen Department of General Practice, University of Göttingen, Göttingen, Germany Correspondence to: Dr Janka Koschack, Department of General Practice, University of Göttingen, Humboldtallee 38, D Göttingen, Germany Tel.: Fax: jkoscha@gwdg.de Disclosures No potential conflicts of interests for any author can be noted. SUMMARY Background: Translating the findings from heart failure drug trials into clinical practice has been shown to take time. For the generation of a primary care guideline that takes preferences of general practitioners (GPs) and characteristics of their patients into account, it is necessary to identify the associations between patient and physician characteristics with the prescription of the recommended drugs. Methods: We searched for patients with chronic heart failure in the electronic patient records of 14 GPs. In multivariate analyses, we examined the prognostic value of patient and physician characteristics for the prescriptions. Results: In the 708 identified patients with chronic heart failure, prescription rates for angiotensin converting enzyme inhibitors angiotensin receptor blockers, beta blockers, diuretics, digitalis and aldosterone antagonists were 50%, 39%, 56%, 35%, and 4%, respectively. On the patient level, age, disease severity, comorbidities and concomitant drug intake were differently related to the prescriptions. On the physician level, age, years of clinical experience and organisation of the practice itself played a differentiating role. Conclusion: Our study demonstrates associations between patient and physician characteristics with the prescription of the recommended drugs that should be taken into account to translate guideline recommendations for application in general practice. What s known Two major obstacles for the implementation of current guideline recommendations for heart failure treatment in general practice are known: First, heart failure patients in general practice are older, more often women and have more comorbidities than in highly selected study samples of clinical trials. Second, general practitioners show individual prescribing preferences independently of guideline awareness. What s new The prescription rates of heart failure drugs are differently related to variables at the patient and the physician levels in a multivariant interaction. On the patient level, age, disease severity, comorbidities and concomitant drug intake are associated with the prescriptions. On the physician level, age, years of clinical experience and organisation of the practice play a differentiating role. Introduction Translating the findings from clinical trials of medications for the treatment of heart failure into clinical practice has been shown to take time and standard treatments are frequently underutilised (1). In Germany, similar to almost all other industrial countries, general practitioners (GPs) are the main providers of care for patients suffering from chronic heart failure (CHF). In general practice, heart failure patients are older, more often women and have more comorbidities compared with the highly selected study samples of clinical trials (2). Such patient complexity significantly hinders the implementation of current guideline recommendations by GPs (3). It has been consistently shown that GPs have individual prescribing preferences independent of guideline awareness (4). However, it is still unclear which patient and physician characteristics can be used to predict the prescription of drugs in the pharmacological treatment of CHF in general practice, and in particular concerning the combination of these variables. To know which factors influence treatment choices is an important prerequisite for the generation of guideline recommendations tailored to the needs in general practice. The aim of our study was to reveal prescription rates of drugs that are recommended by the current guidelines for CHF treatment specifically in a primary care patient sample with CHF and to evaluate the association of patient and physician characteristics with the prescriptions. Methods Study sample A total of 32 GPs in Siegen, a small town in the mid-west of Germany, and the surrounding community were invited to participate in the study. They were randomly selected by screening the telephone directory. Thirteen GPs declined to participate for 226 doi: /j x

2 Prescriptions of recommended heart failure medications GPs contacted by telephone 13 declined 19 followed invitation 3 revealed < 6 cases of CHF cases 5 did not provide valid data 1 without electronic search strategy in practice management system 14 provided valid data 1 without validation of CHF diagnoses by GP 726 patients with CHF diagnosis 18 unconfirmed by GP 708 patients with CHF diagnosis confirmed and rated bygp CHF chronic heart failure GP general practitioner Figure 1 Flow chart of practice recruitment and eligible heart failure diagnoses in the electronic patient records of each participating general practitioner various reasons (lack of time, no interest and concerns of data privacy). Nineteen of the 32 contacted physicians and agreed to participate. After data collection, five of the 19 GPs were excluded (see Figure 1 for reasons). Four of the remaining 14 GPs worked in single-handed and 10 in group practices. Data collection Search for electronically documented patients of heart failure We used the statistical module of the practice management software to screen for documented cases of heart failure. For reasons of the great variety of practice management software systems in Germany, we had to manage eight different software systems in the 19 participating practices either with the help of the practice nurse or the service hotline of the software producers. We searched for both heart failure as the free text diagnosis, and the International Classification of Diseases (ICD) code I50* representing heart failure according to the ICD coding system. The result of each search was printed as a list with name, birth date and the internal coding number of each patient. Patients were excluded for any of the following reasons: deceased, last contact 6 or more months ago or single irregular consultation. The remaining patients were assessed according to demographic and clinical information, number of appointments with the physician in the last 12 months, comorbidities and actual cardiovascular prescriptions. The cardiovascular medication was recorded as prescribed yes or no and included angiotensin converting enzyme (ACE) inhibitors angiotensin receptor (AT1) blockers, beta blockers, diuretics, aldosterone antagonists and digitalis. We decided not to analyse the prescription rate of calcium channel blockers, because actual guidelines do not list it as an appropriate drug for pharmacotherapy of patients with CHF. Only chronic conditions and comorbidities affecting the cardiovascular system were assessed including hypertension, diabetes mellitus, hyperlipidaemia, coronary artery disease, peripheral artery disease, transient ischaemic attacks (TIA) strokes, cardiac arrythmia, atrial defibrillation, chronic obstructive pulmonary disease (COPD) and renal failure. All information was documented in a prepared electronic data sheet. Data confirmation The standardised search in the electronic patient records revealed 726 cases of CHF spread among the 14 participating GPs. GPs were asked to assess the results of the electronic patient record screening by validating the diagnosis, rating the New York Heart Association (NYHA) grade and stating who had initially assessed the diagnosis of CHF (the doctor himself or a community hospital cardiologist). In 708 of these 726 patients, the GPs confirmed the electroni-

3 228 Prescriptions of recommended heart failure medications cally documented diagnosis and rated the NYHA grade. Statistical analysis To test the differences in sociodemographic and clinical variables between male and female patients, we used t-tests for continuous variables and v 2 tests for comparisons of frequencies. To examine the predictive power of patient as well as physician characteristics for the prescription of the assessed drugs, we conducted a logistic regression analysis in a backward conditional model (p > 0.10 as exclusion criterion). A covariate such as age was dichotomised by median split and the categorised covariate NYHA stage transformed by dummy coding. For covariates remaining in the model, odds ratios and 95% confidence intervals were given as effect size. All analyses were 2-tailed and alpha was defined at Statistical analyses were carried out using Statistical Package for the Social Sciences (spss) for Windows (12.0; SPSS Inc., Chicago, IL, USA). Results Patient characteristics and prescribed cardiovascular medication Table 1 shows patient characteristics as well as the cardiovascular prescriptions for the 708 patients with CHF. Group comparisons between male and female patients showed that male patients were younger, suffered more often from coronary artery disease and COPD and less often from hypertension. ACE inhibitors and beta blockers were more often prescribed in male patients and their diagnosis was less often assessed by their GP. Table 1 Patient characteristics and prescribed medication of the study sample All patients (n = 708) Female patients (n = 423) Male patients (n = 285) Test* p-value Demographic clinical characteristics Age, y, mean (SD) 76 (11) 78 (10) 72 (12) < NYHA grade I, % II, % III, % IV, % Organisational characteristics Contacts last year, n, mean (SD) 28 (18) 28 (18) 28 (17) CHF diagnosed by GP, % Comorbidities Hypertension, % Diabetes mellitus, % Hyperlipidaemia, % Coronary artery disease, % Peripheral artery disease, % TIA stroke, % Cardiac arrythmia, % Atrial fibrillation, % COPD, % Renal failure, % Prescribed medication ACE inhibitors, % AT1 blockers, % Beta blockers, % Diuretics, % Digitalis, % Aldosterone antagonists, % *Comparison of means (t-test for independent samples) and frequencies (v 2 test) between female and male patients (all tests are 2-tailed, with alpha = 0.05). Loop or thiazide diuretic. SD, standard deviation; CHF, chronic heart failure; GP, general practitioner, TIA, transient ischaemic attack; COPD, chronic obstructive pulmonary disease; AT, angiotensin; ACE, angiotensin converting enzyme; NYHA, New York Heart Association.

4 Prescriptions of recommended heart failure medications 229 Prescription patterns related to NYHA grade Figure 2 shows the prescription patterns concerning the NYHA stage with variations between the drug classes. ACE inhibitors AT1 blockers were constantly prescribed in about 50% of the patients, with no significant differences between the NYHA grades (Chi 2 = 4.237, p = 0.237). Beta blockers were prescribed with an increase from NYHA I to II and showed a decrease in the prescription rate from NYHA II to IV; the differences between the NYHA stages were significant (Chi 2 = 7.937, p < 0.047). Concerning diuretics and digitalis, the prescription rates increased continuously and statistically significantly from grade I to IV (Chi 2 = , p < 0.001, Chi 2 = , p = 0.003). Prescription of aldosterone antagonists was rare in all NYHA stages, but showed a statistically significant increase from grade III to IV (Chi 2 = 8.178, p = 0.042). Relationship between drug prescription and patient and physician characteristics Table 2 shows the detailed results of the logistic regression analyses and the likelihood of the drug prescription for each cardiovascular drug class as predicted by patient and physician characteristics. Prescriptions of ACE inhibitors AT1 blockers and beta blockers were less likely in patients older than 77 years and treated by a GP with more than 15 years job experience, but more likely in patients with hypertension and coronary artery disease. COPD as a comorbidity showed a negative relationship with the prescription of a beta blocker. Prescriptions of diuretics and digitalis were more likely in patients older than 77 years and more severe heart failure, with diabetes and cardiac arrhythmia, additionally treated with an aldosterone antagonist by a GP working in a group practice. Prescription of aldosterone antagonists was less likely in patients older than 77 years, but more likely in patients with more severe heart failure, more than 25 practice contacts in the last year and atrial fibrillation as comorbidity, treated by a GP older than 48 years. Factors not found to be predictive for the prescription of any of the examined drugs included the comorbidities of hyperlipidaemia, peripheral artery disease, TIA stroke and renal failure or whether the GP or the hospital cardiologist had made the initial heart failure diagnosis. Discussion Our study of 708 patients with CHF from 14 GPs confirmed previous findings of low prescription rates of drugs recommended by guidelines for CHF treatment. Furthermore, the results revealed not only that the particular medications prescribed could be associated with patient characteristics such as age, severity of heart failure and concomitant diseases, but also that the demographic and organisational characteristics at the physician level including years of job experience and practice type additionally play an important role. This is in line with studies that have NYHA 1 NYHA 2 NYHA 3 NYHA 4 Prescriptions in (%) ACE-I/AT1 blockers Beta blockers Diuretics Digitalis Aldosterone antagonists Figure 2 Prescription rates (%) of cardiovascular medication for each NYHA stage for patients with heart failure diagnosis (n = 708). Differences were assessed by v 2 test (2-tailed, with alpha = 0.05)

5 230 Prescriptions of recommended heart failure medications Table 2 Logistic regression analysis of variables at the patient and physician level associated with drug prescription Covariate ACE-I AT1 blocker Beta blocker Diuretics* Digitalis Aldosterone ant. Patients characteristics Age > 77 years 0.67 ( ) 0.55 ( ) 2.16 ( ) 1.67 ( ) 0.44 ( ) Female gender excl ( ) excl. excl. excl. NYHA stage excl ( ) 1.24 ( ) 1.29 ( ) 1.57 ( ) Practice contacts > 25 year excl. excl ( ) excl ( ) CHF diagnosed by GP excl. excl. excl. excl. excl. Comorbidities Hypertension 1.80 ( ) 1.90 ( ) excl. excl. excl. Coronary artery disease 1.36 ( ) 1.83 ( ) excl. excl. excl. Peripheral artery disease excl. excl. excl. excl. TIA stroke excl. excl. excl. excl. Diabetes mellitus excl. excl ( ) 1.46 ( ) excl. Hyperlipidaemia excl. excl. excl. excl. excl. Cardiac arrhythmia excl ( ) 1.52 ( ) 2.23 ( ) excl. Atrial fibrillation 1.85 ( ) excl. excl. excl ( ) COLD excl ( ) excl. excl. Renal failure excl. excl. excl. excl. excl. Prescribed medication ACE-I AT1 blocker 1.50 ( ) 2.71 ( ) excl. excl. Beta blocker 1.39 ( ) excl. excl ( ) Diuretic* 2.51 ( ) excl. excl. excl. Digitalis excl. excl. excl. excl. Aldosterone antagonist excl ( ) excl ( ) Physician characteristics Age > 48 years excl. excl. excl ( ) 2.21 ( ) Working as GP > 15 years 0.70 ( ) 0.62 ( ) 1.68 ( ) excl. excl. Working in a group practice 1.78 ( ) excl ( ) 1.94 ( ) excl. *Loop or thiazide diuretic. ACE, angiotensin converting enzyme; AT, angiotensin; OR, odds ratio; CI, confidence interval; NYHA, New York Heart Association; CHF, chronic heart failure; GP, general practitioner; TIA, transient ischaemic attack; COLD, chronic obstructive pulmonary disease; excl., excluded by backward conditional model (p > 0.10). demonstrated individual physician preferences in the drug treatment of CHF patients in primary care. While the GP sample of our study was relatively small, predictive value of physician characteristics could be shown. A selection bias concerning the patient sample could be excluded, because the patient characteristics of our study sample were in accordance with that of the other international studies with heart failure patients in primary care (4 6). However, the large differences that exist between healthcare systems in different European countries as well as compared to the United States could have a major impact on our findings. Differences in the healthcare systems become manifest in different legal obligations to the electronically documented patient records or in the continuity of care between different healthcare sectors. For example, in 333 (47%) patients, the heart failure diagnosis was made by a cardiologist or at the hospital. In these patients, a discharge or referral drug recommendation would have existed and may have resulted in different prescription rates (7). The factor diagnosis made by GP vs. by cardiologist at hospital was not a significant predictor of the prescription of any examined drug. Previous studies showed difficulties in the transfer of drug therapy from inpatient to ambulatory treatment because of the strong boundaries that exist among the healthcare sectors, especially in Germany (8). Thus, it has to be kept in mind that the database of our study is limited despite of interesting results. Concerning the treatment patterns in heart failure for the different NYHA classes, Pont et al. (9) showed that the prescription rates of each cardiovascular drug increased with increasing NYHA stage, with the exception of the beta blockers. Our results differed from that finding: Concerning the prescription rate of ACE inhibitors AT1 blockers, the rates did not differ between the NYHA grades; thus, they

6 Prescriptions of recommended heart failure medications 231 were prescribed independently of the symptomatic impact of the heart failure, just like the guidelines recommend. The prescription pattern concerning diuretics differed from that of the ACE inhibitors: With the increasing NYHA stage, the prescription of diuretics as a symptom reducing medication also increased significantly. Digitalis was prescribed with higher rates in more severe NYHA stages, i.e. NYHA III and IV that is also true for aldosterone antagonists. It might be speculated that this prescription pattern mirrored the approach to stabilise the patients health status as it is recommended by guidelines (10,11). Prescription rates of beta blockers showed a different pattern: There was an increase from stage I to II, but a decrease from II to IV. This might be a result of the possible side effects and the complex therapy regime of this drug class (12). We examined variables known to be associated with prescription rates by logistic regression analysis to test them in a multivariate context (13). Less than half of the patients received an ACE inhibitor or a beta blocker, drugs known to lower mortality and morbidity rates as demonstrated in large clinical trials (14,15). The likelihood that these two drugs would be prescribed increased with specific patient characteristics well known from other studies: younger patients with hypertension and coronary artery disease were more likely to be treated with ACE inhibitors and beta blockers (13,16,17). Additionally, the prescription of a beta blocker was more frequent in male patients with lower NYHA stage, an association described previously (9). Consistent with the previous studies that discussed the general reluctance to use beta blockers in patients with COPD because of the fear of adverse reactions or a perceived contraindication (18), the logistic regression model revealed COPD as a negative predictor for the prescription of a beta blocker. Organisational characteristics at the physician level also showed prognostic value: working as a GP for < 15 years increased the likelihood of the prescription of ACE inhibitors and beta blockers. With respect to the prescription patterns of diuretics and digitalis, our multivariate analysis revealed a different prognostic model: the likelihood of the prescription of these two drugs was positively linked to patients of higher age and more severe heart failure, who showed diabetes and cardiac arrhythmia and were additionally treated with an aldosterone antagonist. All variables are known to be solely related with digitalis and diuretic treatment (4,9,19), but our study is the first to examine this in a multivariate context. At the physician level, demographic and organisational characteristics showed prognostic value, but conversely to ACE inhibitors and beta blockers: diuretics and digitalis were more likely prescribed by GPs older than 48 years with more than 15 years job experience. Younger patients with severe heart failure and atrial fibrillation were more likely to receive an aldosterone antagonist that is recommended as an addon treatment in combination with an ACE inhibitor and beta blocker prescription for patients at higher NYHA stages (11). Interestingly, the prescription seemed to be independent of that of other drugs, but was more often on patients treated by a GP older than 48 years in more than 25 consultations in the last year. To our knowledge, only one study has shown an association of variables at the patient and the physician levels in a multivariate context with respect to treatment of heart failure (13), but this study focused on ACE inhibitor prescription. Our study confirmed the concurrence of different patient and physician characteristics for the prescription of ACE inhibitors and additionally demonstrated specific interaction of these factors for the prescription of all other drugs that are recommended by existing guidelines. Tinetti et al. (3) argued that evidencebased guidelines are disease-driven because of their origin in clinical trials with highly selected patient samples, but have to be translated to be patient-driven for their use in clinical practice. Graham et al. (20) concluded that the variance in guideline adherence, particularly in primary care, should not be interpreted as a good or bad quality of healthcare, but rather as the different outcomes of a decision process shared by the doctor and patient. Our findings of the multivariate interaction of patient and physician characteristics for all drug classes support this argument. Further research should focus on such real-life decision processes to be able to translate guideline recommendations for application in general practice. Acknowledgements We thank all general practitioners who participated in this study for their co-operation. We express our appreciation to Professor Wolfgang Himmel for critical discussions. Funding of the study None. Author contributions JK performed statistical analyses, wrote and edited the manuscript; HHJ recruited the patients and co-ordinated the study; MS and MMK led conception

7 232 Prescriptions of recommended heart failure medications and design; all authors have contributed to the conception and design and revised the manuscript. References 1 McMurray J, Cohen-Solal A, Dietz R et al. Practical recommendations for the use of ACE inhibitors, beta-blockers and spironolactone in heart failure: putting guidelines into practice. Eur J Heart Fail 2001; 3: Cohen-Solal A, Desnos M, Delahaye F, Emeriau JP, Hanania G. A national survey of heart failure in French hospitals. Eur Heart J 2000; 21: Tinetti ME, Bogardus ST Jr, Agostini JV. Potential pitfalls or disease-specific guidelines for patients with multiple conditions. N Engl J Med 2004; 351: Rutten FH, Grobee DE, Hoes AW. Differences between general practitioners and cardiologists in diagnosis and management of heart failure: a survey in every-day-practice. Eur J Heart Fail 2003; 5: Cleland JFG, Cohen-Solal A, Dietz R et al. Management of heart failure in primary care (the IMPROVEMENT of Heart Failure Programme): an international survey. Lancet 2002; 360: Nilsson G, Strender LE. Management of heart failure in primary care. A retrospective study on electronic patient records in a registered population. Scand J Prim Health Care 2002; 20: Scherer M, Sobek C, Wetzel D, Koschack J, Kochen MM. Changes in heart failure medications in patients hospitalised and discharged. BMC Fam Pract 2006; 7: Hach I, Maywald U, Meusel D, König JU, Kirch W. Continuity of long-term medication use after surgical hospital stay. Eur J Clin Pharmacol 2005; 61: Pont LG, van Gilst WH, Lok DJ, Kragten HJ, Haaijer-Ruskamp FM, Dutch Working Group on Heart Failure. The relevance of heart failure severity for treatment with evidence-based pharmacotherapy in general practice. Eur J Heart Fail 2003; 5: Hunt SA, Abraham WT, Chin MH et al. ACC AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology American Heart Association Task Force on Practice Guidelines (writing committee to update the 2001 guidelines for the evaluation and management of heart failure). Circulation 2005; 112: e Swedberg K, Cleland JFG, Dargie H et al. Guidelines for the diagnosis and treatment of chronic heart failure (update 2005). Eur Heart J 2005; 26: Doughty RN. Beta-blocker therapy in heart failure. Heart Fail Monit 2000; 1: Kasje WN, Denig P, Stewart RE, de Graeff PA, Haaijer-Ruskamp FM. Physician, organisational and patient characteristics explaining the use of angiotension converting enzyme inhibitors in heart failure treatment: a multilevel study. Eur J Clin Pharmacol 2005; 61: Poole-Wilson PA, Swedberg K, Cleland JFG et al. Comparison of carvedilol and metoprolol on clinical outcome in patients with chronic heart failure in the Carvedilol Or Metoprolol European Trial (COMET): randomised controlled trial. Lancet 2003; 362: The SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fraction and congestive heart failure. N Engl J Med 1991; 325: Muntwyler J, Cohen-Solal A, Freemantle N, Eastaugh J, Cleland JFG, Follath F. Relation of sex, age and concomitant diseases to drug prescription for heart failure in primary care in Europe. Eur J Heart Fail 2004; 6: Sturm HB, Haaijer-Ruskamp FM, Veeger NJ, Balje-Volkers CP, Swedberg K, van Gilst WH. The relevance of comorbidities for heart failure treatment in primary care: a European survey. Eur J Heart Fail 2006; 8: Albouaini K, Andron M, Alahmar A, Egred M. Beta-blocker use in patients with chronic obstructive pulmonary disease and concomitant cardiovascular conditions. Int J Chron Obstruct Pulmon Dis 2007; 2: Sparrow N, Adlam D, Cowley A, Hampton JR. Difficulties of introducing the National Service Framework for heart failure into general practice in the UK. Eur J Heart Fail 2003; 5: Graham RP, James PA, Cowan TM. Are clinical guidelines valid for primary care? J Clin Epidemiol 2000; 53: Paper received August 2008, accepted September 2008