Research in Social and Administrative Pharmacy 10 (2014) e87–e98

Original Research

The impact of medication adherence on health outcomes for chronic metabolic diseases: A retrospective cohort study Euna Han, Ph.D.a, Dong-Churl Suh, Ph.D.b, Seung-Mi Lee, Ph.D.b, Sunmee Jang, Ph.D.c,* a

College of Pharmacy, Yonsei Institute of Pharmaceutical Sciences, Yonsei University, Incheon, South Korea b College of Pharmacy, Chung-Ang University, Seoul, South Korea c College of Pharmacy, Gachon University, Incheon, South Korea

Abstract Background: Hypertension, diabetes, and hyperlipidemia have a large influence on health outcomes due to their chronic nature and serious complications. Medication is a key factor in preventing disease advancement, and it is important to assess whether good medication adherence has any potential long-term impact on health outcomes and provides an international validation on the relationship. Objectives: To evaluate the impact of good medication adherence on health outcomes of complications and hospitalizations for hypertension, hyperlipidemia, and diabetes. Methods: Patients who had had outpatient pharmacy claims for drugs for hypertension, diabetes, or hyperlipidemia were separately identified from the Korean National Health Insurance Claims Database in year 2009. A 10% random sample was respectively drawn from the three disease groups, and all claims from years 2008–2011 were extracted for the sampled subjects. Medication adherence was measured by the medication possession ratio (MPR) during the 12-month after the index date, the initial date from when medication was counted, with poor adherence as !80% of MPR. Health outcomes were measured both at 2 and 3 years after the index date as any occurrence of disease-related complications, disease-specific hospitalizations, and all-cause hospitalizations. Results: Poor medication adherence was associated with a higher occurrence of disease-specific hospitalizations for hypertension patients (þ10.9%, only at 2 years). The likelihood of allcause hospitalization was higher among patients who had poor medication adherence in hypertension (þ32% and þ29% at 2 and 3 years), hyperlipidemia (þ16% and þ14% at 2 and 3 years), and diabetes (þ32% and þ29% at 2 and 3 years). Poor medication adherence also increased the likelihood of complications for hypertension (þ14% and þ7% at 2 and 3 years) and hyperlipidemia patients (þ8.1% at 2 years). Conclusions: Targeting good medication adherence could be a valuable policy strategy to effectively manage chronic diseases to improve health outcomes. Ó 2014 Elsevier Inc. All rights reserved. * Corresponding author. College of Pharmacy, Gachon University, 191, Hambakmoero, Yeonsu-gu, Incheon 406–799, South Korea. Tel.: þ82 32 820 4827; fax: þ82 32 820 4821. E-mail address: [email protected] (S. Jang). 1551-7411/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.sapharm.2014.02.001

e88

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

Keywords: Medication adherence; Health outcomes; Hypertension; Diabetes; Hyperlipidemia; National health insurance claims data; South Korea

Introduction The prevalence of chronic diseases has gradually increased in South Korea for hypertension (from 24.6% in 2005 to 26.9% in 2010), diabetes (from 8.6% in 2001 to 9.7% in 2010), and hyperlipidemia (from 8.0% in 2005 to 13.5% in 2010).1 These diseases are likely to become even more prevalent as the Korean population ages.2 The impact of those diseases on health outcomes and resource utilization can be large due to their lingering nature and their potential serious complications. Although behavioral factors such as food intake patterns, smoking, or physical activities, are known to influence prognosis of those diseases, medication is a key factor in preventing disease advancement.3 Several previous studies reported the importance of persistent medication for effectively treating symptoms and preventing complications from chronic diseases including hypertension, diabetes, and hyperlipidemia in terms of health care utilization such as hospitalization,4–8 health care costs,5,6,9 and direct clinical management of health outcomes and mortality.10,11 Lindgren et al (2010) reported that 80% or higher medication possession ratio of lipid-lowering medications was associated with lower risk of complications (HR ¼ 0.75, 95% CI [0.56, 0.98]) and improved long-term projected survival (10.83 vs. 10.81) and qualityadjusted survival years (8.13 vs. 8.11) compared to !50% possession ratio among hypertensive patients.6 Sokol et al (2005) also showed that O80% medication adherence was associated with lower disease-related medical costs (diabetes and hypercholesterolemia patients only) and a lower likelihood of hospitalization for patients with hypertension, diabetes, hypercholesterolemia, or congestive heart failure. Furthermore, the risk of all-cause hospitalizations and the total health care costs also decreased as medication adherence increased, particularly for patients with diabetes and hypertension.5 Another study by Wu et al (2010) showed that poor medication adherence (!80% in medication possession ratio, MPR) was associated with poor blood pressure control (OR ¼ 1.20, 95% CI [1.13–1.29]), CVD-related hospitalization (OR ¼ 1.43, 95% CI [1.14–1.81]) and all-cause hospitalization (OR ¼ 1.47, 95% CI [1.21–1.78]).12 Despite the consensus in the

literature regarding the importance of persistent medication for treatment outcomes for chronic diseases, sustaining medication to the ideal level for those diseases has been a challenge.13 Medication adherence to lipid-lowering agents was reported to be less than 100% in a year and to further fall to 33% over time.7,14–23 The current study builds on previous literature and assesses the impact of good medication adherence for chronic diseases including hypertension, hyperlipidemia, and diabetes on health outcomes, using nationally representative administrative claims data for 4 years (2008–2011) from the National Health Insurance in Korea. Extracting medication adherence and health outcomes from the National Health Insurance database improves the external validity of the findings given that it is nationally representative data for health services utilization and literally covers all Koreans as beneficiaries. The study also explored whether the health outcomes of good medication adherence were sustained over time. This helps to assess whether good medication adherence has any potential long-term impact on health outcomes and provides an international validation on the relationship of medication adherence on health outcomes and resource utilization, which are well established in the United States.

Methods Data The study used the Korean National Health Insurance Claims Database, which has accumulated all electronic filings of outpatient, inpatient, emergency and pharmacy claims reimbursed in the National Health Insurance system across the entire nation. The Korean healthcare system, based on social insurance called the National Health Insurance, covers approximately 95% of Koreans who can pay the premium and the rest 5% as beneficiaries of Medicaid for the impoverished. The current study pulled the claims data for 4 years between years 2008 and 2011. In the current study, three groups of patients with hypertension, hyperlipidemia, and diabetes were respectively identified based on International Classification of Disease 10 codes (ICD-10 codes)

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

from 2009 claims data. Specific ICD-10 codes to pull patients were the following: I10, I11, I12, I13, or I15 for hypertension; E10 or E11 for diabetes; and, E78 for hyperlipidemia. A 10% random sample of patients was randomly drawn from each disease group (N ¼ 684,031 for hypertension, N ¼ 281,833 for diabetes, and N ¼ 481,227 for hyperlipidemia). Then, all claims for the selected sample patients for each of the three disease groups were extracted for four years 2008–2011. Patient identifiers were removed for the patients’ confidentiality. All analyses were conducted separately by the three disease groups. Among the sample patients, only those who were prescribed medications for the target diseases during year 2009 were included (N ¼ 649,904 for hypertension, 199,312 for diabetes, and 290,543 for hyperlipidemia patients were remained in the sample). Patients were included only when they had visited the outpatient clinics at least twice or had been admitted for the inpatient care at least once within 6 months prior to the index date, the initial date from when respective prescription of medications was given in year 2009. This process of selection was applied to reduce falsepositive identification of patients as the study sample (N ¼ 527,645 for hypertension, 161,262 for diabetes, and 194,316 for hyperlipidemia patients were remained in the sample). Moreover, those who were diagnosed with any cancer or AIDS, or deceased, were excluded from the final study sample (N ¼ 523,743 for hypertension, 159,697 for diabetes, and 193,467 for hyperlipidemia patients were remained in the sample). Patients were also excluded if they had any complications before or after 365 days from the index date similar to Lindgren et al (2010) (N ¼ 405,291 for hypertension, 71,932 for diabetes, and 139,561 for hyperlipidemia patients were remained in the sample) to control for any lingering effects of prior complications on prospective complications of interest (see Table 1 for the list of the complications and corresponding ICD-9 codes). Finally, only those patients who had not received any medications for target diseases within 6 months prior to the index date were included in the final sample by including only incident cases (Halpern, 2006). The unit of observation in the database for the current study was the patient. The final sample included 35,742 hypertension, 8532 diabetes, and 34, 678 hyperlipidemia patients without any missing information for all variables in the estimation models (see Fig. 1 for the sampling process).

e89

Variables and analyses Health outcomes The dependent variables for the current study were health outcomes, which were measured by three indices: any occurrence of disease-related complications (with a reference of no occurrence of disease-related complications), hospitalizations directly associated with target diseases (diseaserelated hospitalizations hereafter), and all-cause hospitalizations. Lists of complications and corresponding ICD-10 codes were separately identified for hypertension, diabetes, and hyperlipidemia based on previous literature19,24 (see Table 1). For hypertension, we further divided the complications into three subcategories including cardiovascularrelated, cerebrovascular-related, and nephropathy. Only those complications that accompanied 2þ days of inpatient care or 2þ outpatient visits were considered to be disease-related complications in order to screen out false-positive complications. All dependent variables were measured at the second and third years after the index date. Medication adherence Medication adherence was measured based on medication possession ratio during the 12-month period from the index date using administrative claims data. In previous studies, medication adherence has often been measured using pharmacy claims data5,12,25 and is reported to be as precise as other direct measurements of medication adherence such as pill counting or electronic medication monitors.12,26,27 Poor adherence was defined as medication possession ratio !80% according to the previous literature.5,6,25,28 Medication adherence was separately measured for each disease group of hypertension, diabetes, and hyperlipidemia. All days supplied for each medicine in each prescription were summed up for each patient, and the maximum of those values for a patient was recorded as the final number as MPR for the patient. If there are any overlapped supply of prescription for a medicine, those overlapped days were all summed up only if those prescriptions were prescribed by the same physicians. If a patient was prescribed medications for a target disease from multiple clinics, only medicines from the lastly visited clinic were considered as taken. Previous studies reported that patients commonly mentioned non-clinical explanations as reasons for changing physicians such as low satisfaction with the current physicians or moving to a

e90

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

Table 1 List of complications and corresponding ICD-10 codes related to hypertension, diabetes, and hyperlipidemia Disease

Complications

Small group

Hypertension

Nephropathy

Chronic renal failure Renal failure NOS Renal insufficiency Dialysis Kidney transplant TIA Stroke Angina pectoris MI Other IHD Other chronic IHD Heart failure Atherosclerosis Aortic aneurysm/dissection PCI

Diabetes

ICD-10 codes

N18.x N19.x N28.9, N25.x Z49.x, Z99.2 Z94.0, R3280 Cerebrovascular G45.x I63.x, I65.x, I66.x, Cardiovascular I20.x I21.x, I22.x, I23.x I25.2(old MI) I24.x I25.x (excluding I25.2) I50.x, I70.x I71.x (Procedure codes) M6551, M6552, M6561, M6562, M6563, M6564, M6572 CABG (Procedure codes) O1641, O1642, O1647, OA641, OS647 PVD PVD I73.8, I73.9, I77.1, I79.0, I79.2, I798, K55.1, K55.8, K55.9, Z95.8, Z95.9 Retinopathy Diabetic ophthalmologic disease E10.3þ, E11.3þ, E12.3þ, E13.3þ, E14.3þ, H36.0 Retinal detachment H33.x Blindness H54.x Retinal occlusion H34.x Other retinopathy H35.x Nephropathy Diabetic nephropathy E10.2þ, E11.2þ, E12.2þ, E13.2þ, E14.2þ Chronic glomerulonephritis N03.x, N05.x Acute renal failure N17.x Chronic renal failure N18.x Renal failure NOS N19.x Renal insufficiency N28.9, N25.x Dialysis Z49.x, Z99.2 Kidney transplant Z94.0, V005 (procedure codes), R3280 Neuropathy Diabetic neuropathy E10.4þ, E11.4þ, E12.4þ, E13.4þ, E14.4þ Mono-neuropathy G56.x, G57.x, G58.x, G59.x, G64.x, Polyneuropathy G62.9, G63.2 Autonomic neuropathy G90.0, G90.8, G90.9, G99.0, G99.1 Cranial nerve disorder G53.8 Peripheral-vascular Diabetic PVD E10.5, E11.5, E12.5, E13.5, E14.5 PVD I73.8, I73.9, I77.1, I79.0, I79.2, I798, K55.1, K55.8, K55.9, Z95.8, Z95.9 Foot diseases L97, R02, S807, S808, S809, S817, S818, S819, S907, S908, S909, S91.x, T13.x, Z894, Z895 (procedure codes), N0571, N0572, N0573, N0574, N0575 Cerebrovascular TIA G45.x Stroke I63.x, I65.x, I66.x, Cardiovascular Angina pectoris I20.x MI I21.x, I22.x, I23.x I25.2 (old MI) Other IHD I24.x Other chronic IHD I25.x (excluding I25.2) Heart failure I50.x, Atherosclerosis I70.x (continued)

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

e91

Table 1 (continued ) Disease

Complications

Small group

ICD-10 codes

Aortic aneurysm/dissection PCI

I71.x (Procedure codes) M6551, M6552, M6561, M6562, M6563, M6564, M6572 (Procedure codes) O1641, O1642, O1647, OA641, OS647 I63.x, I65.x, I66.x, I20.x I21.x, I22.x, I23.x I25.2(old MI) I24.x I25.x (excluding I25.2) I70.x I71.x (Procedure codes) M6551, M6552, M6561, M6562, M6563, M6564, M6572 (Procedure codes) O1641, O1642, O1647, OA641, OS647 I73.8, I73.9, I77.1, I79.0, I79.2, I798, K55.1, K55.8, K55.9, Z95.8, Z95.9

CABG Hyperlipidemia

Stroke Angina pectoris MI Other IHD Other chronic IHD Atherosclerosis Aortic aneurysm/dissection PCI CABG PVD

Jang et al. Development of hypertensive medications for prescription assessment. Health Insurance Review & Assessment Service, 2008; Shin et al. Development of hyperlipidemia medications for prescription assessment, Health Insurance Review & Assessment Service, 2009.

different region.29,30 Based on such reports, in the current study, it was assumed that patients change their physicians for some reason and would not mix prescriptions from different providers. The current study also assumed that physicians would make prescriptions for their new patients comprehensively, not just for filling a part of medicines that the patients already had from different physicians. Even if patients are admitted to a hospital, they are closely monitored for their medication intake for their existing chronic diseases during inpatient hospitalizations. Therefore, days supplied for the target medicines during inpatient hospitalization were not handled differently from prescription from the outpatient clinics. Also, if a medicine with a different mechanism of action was prescribed by the same clinic, the prior medication was assumed as not taken only when the entire prescriptions were changed. That is, only a part of medications that were replaced or added by the same physicians were counted in calculating MPR. Other covariates Other covariates controlled for in the estimation models included patients’ demographic and socioeconomic characteristics. Demographic characteristics included age as a linear variable and a dummy indicator representing women, with men as the reference group. Socioeconomic factors included a dummy indicator representing whether the patients

were recipients of Medicaid, with beneficiaries of the National Health Insurance as the reference group. Further, patients’ health status including Charlson’s comorbidity score31 was controlled for as a linear variable. A series of dummy indicators for individual comorbidities of hypertension (only for diabetes and hyperlipidemia groups), diabetes (only for hypertension and hyperlipidemia groups), and hyperlipidemia (only for hypertension and diabetes groups) were also controlled for given that those three diseases are not specifically included in the list for determining the Charlson’s comorbidity score and they were assumed to commonly cause some health outcomes. A stricter definition of comorbidities of the key diseases of hypertension, hyperlipidemia, and diabetes was applied as three or more outpatient visits during a year after the index date. Multivariate logistic regression was used in all estimations, and odds ratio (OR) were reported with 95% confidence intervals. All estimations were separately run by the target diseases. This study was approved by the Institutional Review Board of the Korean National Health Insurance. STATA 12.1 was used for all statistical analyses (StataCorp, College Station, Texas, copyright 1995–2011). Results Approximately 60% of sample patients with hypertension (59.4%) and diabetes (58.8%) were adherent to their target medications, whereas less

e92

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

N = 684,031 N = 281,833 N = 481,227

N = 649,904 N = 199,312 N = 290,543

N = 527,645 N = 161,262 N = 194,316

N = 523,743 N = 159,697 N = 193,467

N = 405,291 N = 71,932 N = 139,561

N = 35,742 N = 8,532 N = 34,678 Fig. 1. Diagram of sample derivation process. Note: a. International Classification of Disease 10 codes to identify patients with target diseases of hypertension, diabetes, and hyperlipidemia are following: Hypertension: I10, I11, I12, I13, or I15; Diabetes: E10, E11; Hyperlipidemia: E78. b. Index date was operationally defined as the initial date from when respective prescription of medications for each disease group was started in year 2009 pharmacy claims.

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

than half (41.0%) of hyperlipidemia patients were adherent to dyslipidemia medications. The sample patients were approximately 57 years old on average, and nearly 6% of them were Medicaid beneficiaries. Approximately 10% and 8% of the hypertension group also had had diabetes and hyperlipidemia, respectively, within a year prior to the index date. Nearly 50% and 25% of hyperlipidemia patients also had had hypertension and diabetes, respectively, as comorbidities within a year prior to the index date. Similarly, about 17% and 40% of diabetes patients had had hyperlipidemia and hypertension, respectively, as comorbidities within a year prior to the index date (see Table 2). Table 3 shows descriptive statistics of outcome variables for the study sample. The incidence rate of disease-specific hospitalizations in 2 years since the index date was 9.6% in diabetes patients, whereas it was 6.1% and 2.8% for hypertension and hyperlipidemia patients, respectively. The incidence rate of disease-specific hospitalizations decreased slightly in year 3– 8.8% for diabetes, 5.1% for hypertension, and 2.3% for hyperlipidemia patients. All-cause hospitalizations in 2 years since the index date occurred in approximately 15% of each disease group, while in 3 years, it occurred slightly more among diabetes patients (13.6%) than among hypertension (12.9%) and hyperlipidemia (12.7%) patients. The diabetes group also had 4– 6 times higher incidence of disease-specific complications (12.0% in 2 years and 21.1% in 3 years) than hypertension (2.8% in 2 years and

e93

5.3% in 3 years) and hyperlipidemia (3.0% in 2 years and 5.3% in 3 years) groups. For hypertensive patients, cardiovascular complications occurred more often than cerebrovascular complications or nephropathy. Table 4 shows the association of poor medication adherence (MPR ! 80%) with the incidence of disease-specific and all-cause hospitalizations in 2 and 3 years since the index date. For hypertensive patients, poor medication adherence was associated with 12.0% higher occurrence of disease-specific hospitalizations (OR ¼ 1.119, 95% CI [1.023, 1.223]) in 2 years from the index date. No statistically significant impact of poor medication adherence on the incidence of disease-specific complications was found for hyperlipidemia or diabetes patients in 2 or 3 years since the index date. However, poor medication adherence was associated with a higher occurrence of all-cause hospitalizations in all three target disease groups. Hypertensive patients with poor medication adherence showed a higher occurrence of allcause hospitalizations by approximately 28% (OR ¼ 1.281, 95% CI [1.207, 1.360]) and 20% (OR ¼ 1.205, 95% CI [1.131, 1.284]) in 2 and 3 years from the index date, respectively. The extent of the association was relatively small for hyperlipidemia patients, with poor medication adherence being associated with approximately 16% (OR ¼ 1.159, 95% CI [1.089, 1.234]) and 13% (OR ¼ 1.135, 95% CI [1.061, 1.214]) higher incidence of all-cause hospitalizations in 2 and 3 years, respectively. A similar extent of association

Table 2 Descriptive statistics of independent variables Variables

Proportion (%)/Mean (SD) Hypertension (N ¼ 35,742) Hyperlipidemia (N ¼ 34,678) Diabetes (N ¼ 8532)

Key independent variable of interest Poor medication adherence 40.57 (!MPR 80%) Demographic and socioeconomic characteristics Age 57.40 (13.11) Women 49.90 Medicaid beneficiaries 6.17 Comorbidity factors Charlson’s comorbidity index 0.3589 (0.6782) Had diabetes within a year after 10.49 the index date Had hyperlipidemia within a 8.78 year after the index date Had hypertension within a year – from the index date

58.97

41.13

57.51 (11.66) 58.43 6.12

56.87 (12.48) 42.79 5.67

0.4990 (0.7821) 18.62

0.3708 (0.6163)



17.30

50.35

41.24

e94

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

Table 3 Descriptive statistics of health outcome variables Variables

Proportion of the sample subjects with the outcome of interest Hypertension (N ¼ 35,742) Hyperlipidemia (N ¼ 34,678) Diabetes (N ¼ 8532)

Disease-specific hospitalizations In 2 years from the index date 6.14 In 3 years from the index date 5.11 All-cause hospitalizations In 2 years from the index date 15.06 In 3 years from the index date 12.99 Disease-related complications General complications in 2 years 2.89 from the index date General complication in 3 years 5.39 from the index date Nephropathy in 2 years from the 0.23 index date Nephropathy in 3 years from the 0.45 index date Cardiovascular in 2 years from 1.65 the index date Cardiovascular in 3 years from 3.03 the index date Cerebrovascular in 2 years from 1.10 the index date Cerebrovascular in 3 years from 2.15 the index date

was found for diabetes patients in that poor medication adherences was found be to associated with approximately 32% (OR ¼ 1.316, 95% CI [1.163, 1.489]) and 29% (OR ¼ 1.285, 95% CI [1.128, 1.464]) higher occurrence of all-cause hospitalizations in 2 and 3 years from the index date, respectively (see Table 4). Table 5 shows the association of medication adherence with the likelihood of complications in 2 and 3 years since the index date. Poor medication adherence statistically significantly influenced the incidence of complications for hypertension and hyperlipidemia patients, whereas no statistically significant association was found for diabetes patients. Poor medication adherence for hypertensive patients was associated with higher likelihood of overall complications in both 2 years (OR ¼ 1.139, 95% CI [1.003, 1.293]) and 3 years (OR ¼ 1.065, 95% CI [1.002, 1.132]) since the index date. Patients with poor medication adherence were more likely to have cardiovascular complications (OR ¼ 1.053, 95% CI [1.003, 1.105]) in 3 years since the index date, compared to their non-persistent counterparts. Poor medication adherence was also associated with a higher probability of complications

2.87 02.36

9.60 8.81

15.20 12.70

15.42 13.64

3.03

12.04

0.0529

21.15

























for hyperlipidemia (OR ¼ 1.081, 95% CI [1.013, 1.153]) in 2 years since the index date. Discussion Previous studies have reported positive associations between good medication adherence and a decreased risk of complications, mortality, or hospitalizations.5,6,12 The current study built on the previous literature and added a global evidence for the relationship between good medication adherence and health outcomes. This study assessed the impact of medication adherence on complications and hospitalization for major chronic diseases including hypertension, diabetes, and hyperlipidemia, using nationally representative administrative claims data in South Korea. The findings of the study corroborate previous reports by showing harmful effects of poor medication adherence on health outcomes including disease-specific or all-cause hospitalizations and disease-related complications. The study found that poor medication adherence was associated with a higher occurrence of disease-specific hospitalizations for hypertension patients (þ10.9%, only in 2 years from the index date). The likelihood of

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

e95

Table 4 Impact of good medication adherence with medication possession ratio (MPR) R80% on the likelihood of diseasespecific hospitalizations Adjusted odds ratio [95% confidence interval] Hypertension (N ¼ 35,742) The occurrence of disease-specific hospitalizations In two years MPR R 80% MPR ! 80% 1.119* [1.023, 1.223] In three years MPR R 80% MPR ! 80% 0.993 [0.901, 1.095] The occurrence of all-cause hospitalizations In two years MPR R 80% MPR ! 80% 1.281* [1.207, 1.360] In three years MPR R 80% MPR ! 80% 1.205* [1.131, 1.284]

Hyperlipidemia (N ¼ 34,678)

Diabetes (N ¼ 8532)

1.020 [0.894, 1.225]

1.109 [0.952, 1.291]

1.059 [0.916, 1.225]

0.980 [0.836, 1.149]

1.159* [1.089, 1.234]

1.316* [1.163, 1.489]

1.135* [1.061, 1.214]

1.285* [1.128, 1.464]

Covariates controlled for in the estimation models included: linear age in years, Charlson’s comorbidity score, and the number of medical institutions used by each patient, dummy indicators for women (vs. men), Medicaid recipient (vs. National Health Insurance beneficiaries), comorbidities of hypertension (only for diabetes and hyperlipidemia groups), diabetes (only for hypertension and hyperlipidemia groups), and hyperlipidemia (only for hypertension and diabetes groups). *P ! 0.1.

all-cause hospitalization was higher among patients who had poor medication adherence in hypertension (þ32% and þ29% in 2 and 3 years from the index date), hyperlipidemia (þ16% and þ14% in 2 and 3 years from the index date), and diabetes

(þ32% and þ29% in 2 and 3 years from the index date) patients. In addition, poor medication adherence increased the likelihood of complications for hypertension (þ14% and þ7% in 2 and 3 years) and hyperlipidemia patients (þ8.1% in 2 years).

Table 5 Impact of good medication adherence with medication possession ratio (MPR) R80% on the likelihood of complications Disease and complication group

MPR

Adjusted odds ratio [95% confidence interval] The occurrence of disease-related complications

Hypertension Whole Nephropathy Cerebrovascular Cardiovascular Hyperlipidemia Whole Diabetes Whole For legend, see Table 4.

(N ¼ 35,742) MPR R 80% MPR ! 80% MPR R 80% MPR ! 80% MPR R 80% MPR ! 80% MPR R 80% MPR ! 80% (N ¼ 34,678) MPR R 80% MPR ! 80% (N ¼ 8532) MPR R 80% MPR ! 80%

In two years

In three years

1.139* [1.003, 1.293]

1.065* [1.002, 1.132]

0.923 [0.701, 1.216]

0.921 [0.740, 1.146]

0.873 [0.522, 1.458]

1.046 [0.950, 1.151]

0.689 [0.417, 1.139]

1.053* [1.003, 1.105]

1.081* [1.013, 1.153]

1.006 [0.915, 1.107]

1.007 [0.881, 1.151]

0.963 [0.864, 1.073]

e96

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

Cardiovascular disease is reported as the second frequent risk factor for mortality in South Korea, causing approximately 23.8% of death in 2004.32 Admission rates for short-term and longterm complications from diabetes were 14.83 and 209.60, respectively, per 100,000 populations in 2009,33 and preventable hospitalization rates due to uncontrolled diabetes and hypertension were 127.47 and 217.00, respectively, per 100,000 populations in 200933 in South Korea. Those indicators of diabetes in South Korea require particular attention given that the OECD average of the preventable hospitalization rate related to uncontrolled diabetes was only 50.3 per 100,000 populations.33 Findings of the current study imply that further efforts to sustain good medication adherence would help improve the quality of care for those chronic diseases in South Korea. The current study randomly extracted sample patients from administrative claims data for the National Health Insurance in South Korea. The National Health Insurance literally covers all Koreans, and thus, our sample represents the entire Korean population; this allows generalization of the findings of the beneficial impacts of good medication adherence on health outcomes for hypertension, diabetes, and hyperlipidemia. Several processes were also adopted in the study to help strengthen internal validity of the findings. First, the study used a retrospective cohort study design to allow the measurement of medication adherence preceding the measurement of health outcomes of complications and hospitalizations. This provided us leverage to approximate the causal relationship between medication adherence and health outcomes. Second, the study excluded any related comorbidities within 365 days before or after the index date in order to control for any lingering effects of prior comorbidities or related treatments on concurrent outcomes of interest similar to Lindgren et al (2010).6 Third, the study subjects were confined to only those incident patients who had not received any medications for target diseases within 6 months prior to the index date. Fourth, the study measured health outcomes twice in 2 and 3 years after the index date, respectively, which allowed us to assess whether the harmful impacts of poor medication adherence on health outcomes had accumulated over time, given that those patients were likely to receive such medications for an extended period due to the chronic nature of those diseases. Nonetheless, the findings of a harmful impact of poor medication adherence on health outcomes

may not be fully interpreted as causal. The current study did not observe other health behavioral characteristics that may be related to both medication adherence and health outcomes, such as smoking, drinking, or physical activity. Medication adherence was measured using MPR, which does not necessarily reflect actual medication. However, MPR was known to be the best way to measure medication adherence in retrospective studies using administrative data.34 The current study also did not identify separately whether a sample patient had multiple events of complications or hospitalizations. The deceased would be selected out from the study subjects over time and they might be more likely to be hospitalized or experience complications if they were alive. Furthermore, complications or hospitalizations at years 2 and 3 might be influenced by related treatments or behavioral changes during the second year since the outcome at 2 years was measured. All of these may underestimate the harmful impacts of poor medication adherence on health outcomes over time due to regression to the mean. Other cautions are also presented in regards to extracting information based on administrative claims data. Even though the study subjects were identified using up to six ICD-9 codes in claims, they can still be miscoded. In the study, additional conditions for health services utilization with the corresponding ICD-10 codes were applied to reducing potential miscoding of the diagnosis. Identification of disease-related complications or hospitalizations was also done with the ICD-10 codes, and it is unlikely that such complications or hospitalizations were fully attributed to one specific disease. Hypertension, hyperlipidemia, and diabetes are mutually related given that most of the behavioral risk factors such as stress are common for the three diseases. The case-mix or medication adherence of those comorbid diseases remained uncontrolled for the study despite that the study controlled for individual comorbidities of the other two key diseases in each model. Nonetheless, the findings of the current study demonstrated the harmful impact of poor medication adherence on health outcomes of complications and hospitalizations. The prevalence of hypertension, diabetes, hyperlipidemia in South Korea was reported as 26.9%, 9.7%, and 13.5% in 2010, respectively, which were up from 24.6%, 9.1%, and 8.0% in 2005, respectively.35 Even though these chronic diseases do not usually have specific symptoms, they can cause severe complications such as

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98

cardiovascular disease or nephropathy. Once patients start medication therapies for those diseases, they are likely to need medications continuously; thus, sustaining good medication throughout the treatment is important. However, continuation of good medication adherence in South Korea was reported to be only 44–54.6% in hypertension,36 29.4% in diabetes,37 and 13.3% in hyperlipidemia38 similar to other countries.39–41 Specific local contexts in Korea such as the wide prevalence of alternative medicines and the dual existence of traditional and western medicines may further contribute to non-adherence to medication, particularly among the elderly.42,43 Given that Korea runs the National Health Insurance system with all Koreans as compulsory beneficiaries, any public approach to improving medication adherence such as patient education or professional interventions with consideration of the specific local context could return large clinical and economic benefits.

4.

5.

6.

7.

8.

9.

Conclusion 10.

Our results imply that targeting good medication adherence could be a valuable policy measure for effective management of chronic diseases to improve health outcomes and optimize health care resource utilization.

11.

Acknowledgment We gratefully acknowledge research support from the Korea Association of Pharmacy Education. Han also acknowledges research support from the Korea National Research Foundation (NRF-2012007096). Data support from Health Insurance Review and Assessment Service is appreciated. The content is solely the responsibility of the authors and does not necessarily represent the official view of the Korea Association of Pharmacy Education or the Korea National Research Foundation.

12.

13.

14.

15.

References 1. Korea Ministry of Health and Korea Center for Disease Control. Status of Chronic Diseases in Korea, 2011. Available at: www.index.go.kr http://www.index.go. kr/egams/stts/jsp/potal/stts/PO_STTS_IdxMain.jsp? idx_cd¼1438. Accessed 20.05.13. 2. OECD. www.oecd.org/els/emp/33906935.pdf. Accessed 20.05.13. 3. Balkrishnan R. The importance of medication adherence in improving chronic-disease related outcomes:

16.

17.

e97

what we know and what we need to further know. Med Care 2005;43:517–520. Shin S, Song H, Oh SK, Choi KE, Kim H, Jang S. Effect of antihypertensive medication adherence on hospitalization for cardiovascular disease and mortality in hypertensive patients. Hypertens Res; 2013:1–6. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care 2005; 43:521–530. Lindgren P, Eriksson J, Buxton M, et alon Behalf of the ASCOT Trial Investigators. The economic consequences of non-adherence to lipid-lowering therapy: results from the Anglo-Scandinavian-Cardiac Outcomes Trial. Int J Clin Pract 2010;64:1228–1234. Cramer JA. A systematic review of adherence with medications for diabetes. Diabetes Care 2004;27: 1218–1224. Tsuyuki RT, Bungard TJ. Poor adherence with hypolipidemic drugs: a lost opportunity. Pharmacotherapy 2001;21:576–582. Kleinke JD. The price of progress: prescription drugs in the health care market. Health Aff (Millwood) 2001;20:43–60. Bramiley TJ, Gerbind PP, Nightengale BS, FrechTamas F. Relationship of blood pressure control to adherence with antihypertensive monotherapy in 13 managed care organizations. J Manag Care Pharm 2006;12:239–245. DiMatteo MR, Giordani PJ, Lepper HS, Croghan TW. Patient adherence and medical treatment outcomes: a meta-analysis. Med Care 2002;40: 794–811. Wu PH, Yang CY, Yao ZL, Lin WZ, Wu LW, Chang CC. Relationship of blood pressure control and hospitalization risk to medication adherence among patients with hypertension in Taiwan. Am J Hypertens 2010;23:155–160. Martin Leslie R, Williams Summer L, Haskard Kelly B, DiMatteo M Robin. The challenge of patient adherence. Ther Clin Risk Manag 2005;1:189–199. Jang S. Development of Hypertensive Medications for Prescription Assessment. Health Insurance Review & Assessment Service; 2008. Caspard H, Chan AK, Walker AM. Compliance with a statin treatment in a usual-care setting: retrospective database analysis over 3 years after treatment initiation in health maintenance organization enrollees with dyslipidemia. Clin Ther 2005;27:1639–1646. Chodick G, Shalev V, Gerber Y, et al. Long-term persistence with statin treatment in a not-for-profit health maintenance organization: a populationbased retrospective cohort study in Israel. Clin Ther 2008;30:2167–2179. Gislason GH, Rasmussen JN, Abildstrom SZ. Long-term compliance with beta-blockers, angiotensin-converting enzyme inhibitors, and statins after

e98

18.

19.

20.

21.

22.

23.

24.

25.

26. 27.

28.

29.

Han et al. / Research in Social and Administrative Pharmacy 10 (2014) e87–e98 acute myocardial infarction. Eur Heart J 2006;27: 1153–1158. Kulkarni SP, Alexander KP, Lytle B, Heiss G, Peterson ED. Long term adherence with cardiovascular drug regimens. Am Heart J 2006;151: 185–191. Lachaine J, Rinfret S, Merikle EP, Tarride JE. Persistence and adherence to cholesterol lowering agents: evidence from Regiedel’AssuranceMaladie du Quebec data. Am Heart J 2006;152:164. Vinker S, Shani M, Baevsky T, Elhayany A. Adherence with statins over 8 years in a usual care setting. Am J Manag Care 2008;14:388–392. Yeaw J, Benner JS, Walt JG, Sian S, Smith DB. Comparing adherence and persistence across 6 chronic medication classes. J Manag Care Pharm 2009;15:728–740. Evans CD, Eurich DT, Lamb DA, et al. Retrospective observational assessment of statin adherence among subjects patronizing different types of community pharmacies in Canada. J Manag Care Pharm 2009;15:476–484. Haynes RB, McDonald HP, Garg AX. Helping patients follow prescribed treatment: clinical applications. J Am Med Assoc 2002;288:2880–2883. Shin S, Song H, Jang S, Sung Y. Development of Hyperlipidemia Medications for Prescription Assessment. Health Insurance Review & Assessment Service; 2009. Hong JS, Kang HC. Relationship between oral antihyperglycemic medication adherence and hospitalization, mortality, and healthcare costs in adult ambulatory care patients with type 2 diabetes in South Korea. Med Care 2011;49:378–384. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med 2005;353:487–497. Krousel-Wood M, Thomas S, Muntner P, Morisky D. Medication adherence: a key factor in achieving blood pressure control and good clinical outcomes in hypertensive patients. Curr Opin Cardiol 2004;19:357–362. Ho PM, Magid DJ, Masoudi FA, McClure DL, Rumsfeld JS. Adherence to cardio-protective medications and mortality amoba1ng patients with diabetes and ischemic heart disease. BMC Cardiovasc Disord 2006;6:48. Buja A, Cavinato M, Perissinotto E, Rausa G, Mastrangelo G, Toffanin R. Why do patients change their general practitioner? Suggestions on corrective actions. Ir J Med Sci 2011;180(1):149–154.

30. Billinghurst B, Whitfield M. Why do patients change their general practitioner? A postal questionnaire study of patients in Avon. Br J Gen Pract 1993 August;43(373):336–338. 31. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1994;47:1245–1251. 32. Kim SK, Jang SM. Relationship between antihypertension medication adherence, medical utilizations, and medical expenditure among patients with hypertension. Yakhak Hoeji 2013;57(5):36–375. 33. OECD. Health at a Glance 2011: OECD Indicators. OECD Publishing, 2011. http://dx.doi.org/10.1787/ health_glance-2011-en. 34. Halperb MT. Recommendations for evaluating compliance and persistence with hypertension therapy using retrospective data. Hypertension 2006;47:1039–1048. 35. Jang SM, Han E, Park TJ, Hwang JH, Sohn HS, Lee EJ. Development of Medication Management Program for Rational Drug Utilization. National Health Insurance Service; 2013. 36. Jang SM, Song HJ, Shin SY. Development of Assessment Indicator for Anti-hypertensive Drugs. Health Insurance Review and Assessment; 2008. 37. Hong JS, Kang HJ. Analyses of Prescription Patterns and Medication Adherence for Diabetes Patients. Health Insurance Review and Assessment; 2009. 38. Shin SY, Song HJ, Jang SM. Development of Assessment Indicator for Hyperlipidemia Medicines. Health Insurance Review and Assessment; 2009. 39. Yang Y, Thumula V, Pace PF, Banahan BF, Wilkin NE, Lobb WB. Predictors of medication nonadherence among patients with diabetes in medicare Part D programs: a retrospective cohort study. Clin Ther 2009;31(10). 40. Wong MCS, Jiang JY, Lam AT, Fung H, Griffiths S, Mercer SW. Patterns of antihypertensive prescribing, discontinuation and switching among a Hong Kong Chinese population from over one million prescriptions. J Hum Hypertens 2008;22:714–716. 41. Senior V, Marteau TM, Weinmanon Jon Behalf of the Genetic Risk Assessment for FH Trial (GRAFT) Study Group. Cardiovasc Drugs Ther 2004;18:475–481. 42. Park YH, Kim H, Jang SN, Koh CK. Predictors of adherence to medication in older Korean patients with hypertension. Eur J Cardiovasc Nurs 2013;12(1):17–24. 43. Krousel-Wood MA, Paul Muntner P, Joyce CJ, et al. Adverse effects of complementary and alternative medicine use on antihypertensive medication adherence: findings from CoSMO. J Am Geriatr Soc 2010; 58(1):54–61.

The impact of medication adherence on health outcomes for chronic metabolic diseases: a retrospective cohort study.

Hypertension, diabetes, and hyperlipidemia have a large influence on health outcomes due to their chronic nature and serious complications. Medication...
651KB Sizes 1 Downloads 5 Views