Data source
Patients were identified from the Korean Health Insurance Review and Assessment Service database (KHIRA), which contains medical claims data for the entire Korean population [24] as a result of the National Insurance Health System. Patients pay an average of 30% of the total medical costs related to almost all diseases. Healthcare providers submit reports concerning the medical services performed to the KHIRA for a review of the medical costs incurred. These reports contain information on the diagnosis that has been coded in accordance with the International Classification of Diseases Tenth Revision [ICD-10] as well as information related to outpatient or inpatient status, drug name, dosage, prescription date, duration, and method of administration. The KHIRA provided data with the individual identifier removed, in accordance with the Act on the Protection of Personal Information Maintained by Public Agencies. Thus, the database included an unidentifiable code representing each individual with data concerning the patients’ age, gender, diagnosis and lists of prescribed drugs.
The database contained information regarding 1,093,262 elderly patients aged above 65 years and 11,842,586 prescriptions from January 1, 2005, to June 30, 2006.
This study was approved by the institutional review board of the Seoul National University Bundang Hospital with reference number of B-1011-115-105.
Study population
We identified older adults who were over the age of 65 and had been hospitalized with a primary discharge diagnosis of heart failure (ICD-10 codes: I11, I13, and I50) between January 1, 2005, and June 30, 2006. Though diagnosis of heart failure by ICD-10 codes in the KHIRA database was not validated, we tried to increase diagnostic accuracy by adjusting medications for heart failure such as digoxin, inotropics and diuretics.
We excluded patients if they had a length of stay less than 24 hours or did not have medication records. In total, 28,922 patients were admitted with a primary discharge diagnosis of heart failure.
Prescription of evidence-based treatment in CHF
Three classes of prescription medications were evaluated based on evidence-based treatment: ACE-I or ARB (group A), beta-blockers (group B), and aldosterone antagonists (Aldo group).
The utilization of evidence-based treatments was defined as treatments that were prescribed after hospitalization for heart failure. We prioritized groups A and B rather than the Aldo group because the 2005 American College of Cardiology and American Heart Association (ACC/AHA) heart failure guidelines with the 2009 focused update [25] recommended the addition of an aldosterone antagonist for the treatment of patients with moderate to severe HF and the reduced ejection fraction who could be carefully monitored for preserved renal function and normal plasma potassium concentrations. Therefore, we classified the evidence-based treatment groups as A, B, Aldo, and A + B. The A + B group was assigned if a patient received both A and B group treatments. If a patient received A, B, and Aldo group treatments, then the patient was assigned to the A + B group. If a patient received A and Aldo group treatments, then the patient was assigned to group A. If a patient only received Aldo group treatment, then the patient was assigned to the Aldo group. Group B consisted of patients who were given only a beta-blocker treatment or beta-blockers in addition to aldosterone antagonists.
Covariates
Data concerning patient age, gender, area of residence, type of prescription resources from inpatient or outpatient clinic and the utilization of hospital type (primary care clinics, secondary hospitals, which typically refer to large, community but non-teaching hospitals, or tertiary hospitals, which usually refer to a teaching or university hospital) were obtained from the database. Previous cardiovascular disease histories were collected using the diagnosis codes for angina pectoris (I20), myocardial infarction (I21-22 and I25.2), transient ischemic attack or ischemic stroke (G45-46, H34, and I60-69), peripheral artery disease (I70-71, I73.1, I73.8-73.9, I77.1, I79.0, I79.2, K55.1, K55.8-55.9, and Z95.8-95.9), atrial fibrillation or flutter (I48), and valvular heart disease (I01, I05-09, and I34-39).
We also included any medical histories of hypertension (I11-I13 an d I15), hyperlipidemia or dyslipidemia (E78), end-stage renal disease (I12.0, I13.1, N03.2-03.7, N05.2, N05.7, N18-19, N25.0, Z49.0-49.2, Z94.0, and Z99.2), chronic lung disease (I27.8, I27.9, J40-47, J68.4, J70.1, and J70.3), chronic liver disease (B18, K70, K71, K73-74, K76, and Z94.4), systemic cancer (C00-C26, C30-C34, C37-C41, C43, C45-58, C60-85, C88, and C90-97), dementia (F00-F03, F05.1, G30, and G31.1) and depression (F32-33).
Concomitant medication use was adjusted in the model. Concomitant drugs included heart failure medications, such as diuretics, calcium channel blockers, nitrates, digoxin, amiodarone, and hydralazine, and lipid-lowering medications, such as statins, fibrates, and ezetimibe. Anti-diabetic medications were also included, such as sulfonylurea, metformin, alpha glucosidase inhibitors, thiazolidinediones, and insulin.
Statistical analysis
Statistical analyses were performed using SAS software, version 9.1 (Cary, North Carolina).
We evaluated baseline characteristics with previous cardiovascular disease, medications and other systemic medical histories between each group of evidence based treatment and non use group using Student’s t test for continuous variable and chi-square test for categorical variables,
Multivariable logistical regression model was used to evaluate clinical factors associated with each evidence-based group. The model incorporated the following demographic factors (age, gender, residence area, utilization of hospital type, specialty of health care providers and type of prescription resources), previous cardiovascular diseases (angina, myocardial infarction, valvular heart disease, atrial fibrillation or flutter, transient ischemic attack), systemic medical diseases (hypertension, hyperlipidemia, chronic lung disease, end stage renal disease) and concomitant medications (heart failure medication, antidiabetic drugs) by forward selection methods. We also performed the similar multivariable logistic regression analysis in subgroup who were treated with both digoxin and diuretics, which could indicate patients with symptom relieving treatment for heart failure. Subgroup analysis was shown for the purpose of increasing diagnostic accuracy for heart failure.