- Email: [email protected]
Risk Factors Associated With Antihypertensive Medication Nonadherence in a Statewide Medicaid Population
CLINICAL INVESTIGATION
Risk Factors Associated With Antihypertensive Medication Nonadherence in a Statewide Medicaid Population James E. Bailey, MD, MPH, Mohammed Hajjar, MD, Bushra Shoib, MD, Jun Tang, PhD, Mario M. Ray, MD and Jim Y. Wan, PhD
Abstract: Background: This study seeks to determine the most important patient factors and health care exposures available through administrative databases associated with antihypertensive nonadherence. Methods: This is a cross-sectional analysis of Medicaid hypertensive patients of Tennessee enrolled for 3 to 7 years from 1994 to 2000. Demographic characteristics, comorbidity and health care utilization were assessed during a 2-year period. The primary outcome was antihypertensive medication refill nonadherence. Subjects were categorized as adherent or nonadherent using an 80% cutoff criteria. Associations with nonadherence were assessed using logistic regression modeling. Results: Of 49,479 subjects, 60.6% (n 5 29,970) were classified as nonadherent and 39.4% (n 5 19,509) as adherent. Significant predictors of nonadherence in multivariate analysis (P , 0.05) included male sex (odds ratio [OR] 1.12), black race (OR 1.67), urban residence (OR 1.12), obesity (OR 1.10), mental illness (OR 1.08) and substance abuse (OR 1.43). Significant protective factors included age (OR 0.97), disability (OR 0.62), diabetes (OR 0.76), hypercholesterolemia (OR 0.72) and Charlson index (OR 0.97). When health care utilization was considered, increased outpatient visits were associated with decreased nonadherence. Emergency department visits (OR 1.07) and hospital visits (OR 1.12) were associated with increased nonadherence. Conclusions: This cross-sectional study suggests that substance abuse, black race, emergency department visits and hospitalizations are risk factors associated with nonadherence. Outpatient visits are associated with a small decrease in nonadherence. Further studies are needed to determine the characteristics of outpatient visits that most improve adherence. Key Indexing Terms: Hypertension; Medication adherence; Patient compliance; Ambulatory care; Health services research. [Am J Med Sci 2014;348(5):410–415.]
H
ypertension affects almost one third of Americans; yet, only two thirds of hypertensive patients are being treated, and only half of those treated are controlled.1,2 Antihypertensive medication nonadherence has been identified as a major reason for lack of adequate blood pressure control.3,4 Previous studies document that nonadherence with antihypertensive medication places patients at increased risk for cardiovascular complications.5,6 Numerous studies document extremely high rates of antihypertensive medication nonadherence ranging from 50%
to 75%.7–11 Despite the ready availability of numerous welltolerated antihypertensive medications, adherence to treatment remains a challenge. Both clinicians and policy makers need to know the patient factors and health services exposures associated with nonadherence to identify and target patients at highest risk for nonadherence with effective interventions. Literature review suggests that the most important patient risk factors associated with nonadherence include young age, low comorbidity, Hispanic or black race, mental illness and substance abuse.12–15 These studies show an inconsistent relationship between sex and nonadherence to antihypertensive medications. Previous research also suggests that numerous medicationrelated factors may also be important including choice of drug, use of concomitant medications and tolerability of drug side effects.14–17 Some early research documents that hospitalizations and emergency department use are strongly linked to nonadherence.18–21 Similarly, preliminary evidence from intervention research and some population-based studies suggests that certain types of ambulatory care may improve adherence.14,16,22,23 However, most of these studies have been limited in scope, lack external generalizability and considered a limited set of potential risk factors. To our knowledge, this study is the first to assess a broad range of potential risk factors associated with antihypertensive nonadherence in a statewide Medicaid population. Antihypertensive nonadherence is a clinical outcome linked to major cardiovascular events that is potentially amenable to change. The study seeks to determine the most important patient, medication-related and health services risk factors of those readily available in administrative databases associated with antihypertensive nonadherence among Medicaid enrollees. Based on the previous literature, we hypothesized that patient factors including black race, young age and history of substance abuse would be associated with increased nonadherence in this population. Furthermore, we hypothesized that exposure to outpatient visits would be associated with less nonadherence, whereas emergency department visits and hospitalizations would be associated with increased nonadherence.
METHODS From the Division of General Internal Medicine, Department of Medicine (JEB, MH, BS, JT, MMR), and Department of Preventive Medicine (JEB, JYW), University of Tennessee Health Science Center, Memphis, Tennessee. Submitted March 22, 2011; accepted in revised form August 10, 2011. Disclosure: This article has not been submitted elsewhere, is not under review, or has not been published previously except in abstract form. Preliminary findings from this research were presented at the 32nd Annual Meeting of the Society of General Internal Medicine; May 15, 2009; Miami, FL. The authors state that they have no conflicts of interest to disclose. Supported by a grant from the American Heart Association. Correspondence: James E. Bailey, MD, MPH, Division of General Internal Medicine, Department of Medicine, University of Tennessee Health Science Center, 956 Court Avenue, Coleman D222, Memphis, TN 38163 (E-mail: [email protected]).
410
Design and Setting This is a secondary analysis of retrospective cohort study database of all chronic medication-treated hypertensive individuals enrolled in Tennessee’s Medicaid-managed care system (TennCare) for 3 to 7 years during the period of 1994 to 2000. The study database and primary data sources have been described previously.5,24,25 This cross-sectional study of the 2-year baseline data for cohort participants is designed to document patient factors and medication and health services utilization factors associated with antihypertensive nonadherence in the 2-year baseline period for each subject.
The American Journal of the Medical Sciences
Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
Subjects Study participants were identified using eligibility, inpatient, professional and pharmacy claims data. The cohort included all noninstitutionalized persons with continuous eligibility (.320 days per year) throughout the 2-year study period, lack of Medicare eligibility, age 18 to 64 years in each study year, yearly diagnosis of essential hypertension (any International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis code 401.x for any professional or inpatient claim) and receipt of at least 1 antihypertensive medication prescription for each of the 2 baseline years. Patients who died or had a stroke (ICD-9-CM diagnosis code 430–438) during their baseline 2-year period were excluded. Study Outcome The study’s main outcome measure was baseline antihypertensive medication refill adherence (MRA). MRA was defined as the percentage of eligible prescription days filled (total days’ supply for all qualifying drug classes/total number of days from the first to the last fill in the interval 3 100, capped at 100%) for all antihypertensive medications taken in the 2-year time period. MRA was calculated for each individual for all qualifying antihypertensive drugs for the 2-year study period. Thus, MRA was averaged across all antihypertensive drug classes. For example, an individual on 2 drugs during the 2-year (730-day) study period, receiving 730 days’ supply for drug 1 and only 365 days’ supply for drug 2, would have an MRA of 75%. Only 1 fill within a drug class was required to define “use” of a class of medications. However, if a participant only had 1 prescription filled during the period, the day’s supply of the 1 prescription was used as the total number of eligible prescription days for that class of medications. If $2 fills were obtained for a class of medications, the eligible prescription days included all the days from the beginning of the first fill to the end of the last fill for that class in the study period. MRA produces identical results for adherence as the similarly calculated medication possession ratio and is a well-validated measure of medication adherence.26–28 All subjects were categorized as either adherent or nonadherent using an 80% cutoff because previous studies demonstrate a good validity using this value.16,29 Independent Variables Demographic variables were assessed using administrative demographic and eligibility files. Demographic characteristics assessed at baseline included age, sex, race, eligibility status, geographic region, urban/rural status, managed care organization (MCO) type and MCO for-profit status. All patients in the study were insured through TennCare, but their eligibility status differed. TennCare eligibility categories included the following: Medicaid only, disabled, uninsured (lacking insurance at the initiation of the TennCare program) and uninsurable (deemed uninsurable because of preexisting conditions). People enrolled in TennCare based on uninsurable eligibility status were required to show documentation (eg, a denial letter) from an insurance company denying them insurance on the basis of a preexisting condition or medical documentation of a preexisting condition (eg, HIV/AIDS) for which commercial insurance was known not to be available. Based on literature review suggesting strong associations between comorbidity and medication nonadherence, we assessed comorbidity in multiple ways using 2-year baseline inpatient and professional administrative data for paid claims. Overall comorbidity was assessed in each subject using the University of Manitoba SAS Macro for the Charlson index of Ó 2012 Lippincott Williams & Wilkins
comorbidity.30,31 In addition, we assessed specific comorbidities through the presence of inpatient or professional paid claims during the study period including the following diagnoses: obesity (ICD-9-CM code 278.0x), diabetes (250.x present in each year of the study period), hypercholesterolemia (272.x), substance abuse (303–305) and mental illness (290–302, 306–319) using previously described methods.5,24 Additional medication and health services utilization factors and medication utilization patterns were evaluated using paid pharmacy, inpatient and professional claims for the 2-year study period. Specifically, antihypertensive variety exposure was calculated as the number of drug classes filled ever, with all pharmaceutical claims for unique antihypertensive drugs collapsed in groups according to each drug class for which a participant had at least 1 prescription filled in the 2-year study period.5 In addition, 4 health care utilization variables were assessed: (1) ambulatory visits per year, (2) emergency department visits per year, (3) hospital visits per year and (4) hospital days per year. Statistical Analysis Chi-square and 2-sample t test statistics were used to assess bivariate associations with nonadherence. Multivariate associations with nonadherence were assessed using logistic regression modeling. All demographic, comorbidity and health care utilization variables found to be significant at P # 0.05 level in the bivariate analyses were included in the same multivariate logistic regression model. No adjustments were made based on whether medications were used in succession or concurrently; however, the antihypertensive variety exposure variable served to adjust for the effect of the number of antihypertensive medications prescribed on nonadherence. Variables in the multivariate model with P # 0.05 were declared significant. All analyses were performed using SAS.32
RESULTS Characteristics of the 49,479 subjects who met full inclusion criteria are shown in Table 1. Mean age for study subjects was 48.5 years (range, 20–64 years). The subjects were predominantly woman (67.7%), white (66.8%), uninsured (39.0%), disabled (36.8%) and in statewide MCOs (64.9%). Wide variation in levels of comorbidity was noted with Charlson index values ranging from 0 to 16, with a mean Charlson index of 1.71. Mean MRA was 67%, and only 39.4% were adherent using an 80% cutoff criteria. Bivariate associations with nonadherence are shown in Table 2. Significant predictors of nonadherence in bivariate analysis (P , 0.05) included black race (odds ratio [OR] 2.09, confidence interval [CI] 2.00–2.18), uninsured eligibility status (OR 1.32, CI 1.27–1.37), urban residence (OR 1.56, CI 1.50–1.62), diagnosed obesity (OR 1.19, CI 1.10–1.28), mental illness (OR 1.09, CI 1.05–1.14) and substance abuse (OR 1.62, CI 1.51–1.73). Significant factors associated with nonadherence included age (OR 0.96, CI 0.95–0.96), disability (OR 0.80, CI 0.77–0.83), uninsurable eligibility status (OR 0.53, CI 0.50– 0.55), metropolitan residence (OR 0.83, CI 0.80–0.88), diabetes (OR 0.67, CI 0.62–0.73), hypercholesterolemia (OR 0.57, CI 0. 0.55–0.60) and Charlson index (OR 0.91, CI 0.90–0.92). In addition, the following health services utilization variables were associated with nonadherence: emergency department visits (OR 1.08, CI 1.06–1.10), hospital visits (OR 1.08, CI 1.05– 1.11) and hospital days (OR 1.01, CI 1.01–1.02). Alternatively, outpatient visits were associated with decreased odds of nonadherence (OR 0.97, CI 0.97–0.98).
411
Bailey et al
TABLE 1. Characteristics of hypertensive patients (n 5 49,479) Characteristics Sex Male Female Unknown Race Black White Other Unknown Eligibility status Medicaid only Disability Uninsured Uninsurable Unknown Urban/rural status Urban Greater metropolitan area Rural Unknown Geographic region East Middle West Unknown MCO type Statewide Regional Unknown Comorbidity Diabetes Obesity Mental illness Substance abuse Congestive heart failure Atrial fibrillation Transient ischemic attack Myocardial infarction Hypercholesterolemia Comorbidity Charlson index Other demographic variables Age (yr) Socioeconomic status Income of county of residence in 1998 ($) Health care utilization Ambulatory visits per yr Emergency department visits per yr Hospital visits per yr Hospital days per yr Medication utilization Variety exposure (no. drug classes filled ever) Medication refill adherence (% eligible prescription days filled)
n (%) 15,974 (32.28) 33,504 (67.71) 1 (0.00) 13,829 33,048 245 2,357
(27.95) (66.79) (0.50) (4.76)
4,864 18,244 19,278 7,091 2
(9.83) (36.87) (38.96) (14.33) (0.00)
20,338 7,931 20,747 463
(41.10) (16.03) (41.93) (0.94)
19,328 15,012 14,676 463
(39.06) (30.34) (29.66) (0.94)
32,088 (64.85) 17,389 (35.14) 2 (0.00) 2,290 (4.63) 3,048 (6.16) 12,142 (24.54) 4,149 (8.39) 1,737 (3.51) 706 (1.43) 608 (1.23) 2,256 (4.56) 8,793 (17.77) Mean, median (range) SD 1.71, 1 (0–16) 1.91 48.5, 50 (20–64) 9.64 35,050, 35,483 (19,760–69,104) 6,384 5.2, 4 (0–86.5) 4.66 0.6, 0 (0–66) 1.38 0.30, 0 (0–13) 0.25 1.43, 0 (0–327) 5.90 2.33, 2 (1–8) 1.17 67, 72 (27–100) 25
MCO, managed care organization; SD, standard deviation.
412
Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
TABLE 2. Bivariate results for nonadherence Nonadherence odds Characteristics ratio (95% CI) Demographic characteristics Agea Male sex Black race Disability Uninsured Uninsurable Median income for countyc Urban/rural status Metropolitan Urban Geographic region Middle West MCO type Regional Health care utilization Outpatient visits Emergency department visits Hospital visits Hospital days Comorbidity Obesity Diabetes Mental illness Substance abuse Hypercholesterolemia Charlson index Medication utilization Variety exposure (no. drug classes filled ever)
P
(0.95–0.96)b (1.01–1.09)b (2.00–2.18)b (0.77–0.83)b (1.27–1.37)b (0.50–0.55)b (1.02–1.03)b
,0.0001 0.0098 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001
0.83 (0.80–0.88)b 1.56 (1.50–1.62)b
,0.0001 ,0.0001
0.93 (0.89–0.96)b 1.37 (1.31–1.42)b
0.0002 ,0.0001
1.22 (1.17–1.26)b
,0.0001
0.97 1.08 1.08 1.01
(0.97–0.98)b (1.06–1.10)b (1.05–1.11)b (1.01–1.02)b
,0.0001 ,0.0001 ,0.0001 ,0.0001
1.19 0.67 1.09 1.62 0.57 0.91
(1.10–1.28)b (0.62–0.73)b (1.05–1.14)b (1.51–1.73)b (0.55–0.60)b (0.90–0.92)b
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001
0.81 (0.80–0.82)b
,0.0001
0.96 1.05 2.09 0.80 1.32 0.53 1.02
1 unit 5 1 yr. Significant at P , 0.05. c 1 unit 5 $1000. CI, confidence interval; MCO, managed care organization. a b
In the multivariate analysis, shown in Table 3, significant factors associated with nonadherence (P , 0.05) included male sex (OR 1.12, CI 1.07–1.17), black race (OR 1.67, CI 1.58– 1.78), urban residence (OR 1.12, CI 1.06–1.19), obesity (OR 1.10, CI 1.01–1.20), mental illness (OR 1.08, CI 1.02–1.13) and substance abuse (OR 1.43, CI 1.32–1.55). Significant factors associated with decreased nonadherence included age (OR 0.97, CI 0.97–0.97), disability (OR 0.62, CI 0.57–0.67), uninsured eligibility status (OR 0.83, CI 0.77–0.91), uninsurable eligibility status (OR 0.51, CI 0.47–0.56), diabetes (OR 0.76, CI 0.70– 0.84), hypercholesterolemia (OR 0.72, CI 0.68–0.75) and Charlson index (OR 0.97, CI 0.96–0.98). When health care utilization was considered, increased outpatient visits (OR 0.99, CI 0.98– 0.99) were associated with decreased nonadherence. Emergency department visits (OR 1.07, CI 1.05–1.09), hospital visits (OR 1.12, CI 1.07–1.18) and hospital days (OR 1.01, CI 1.00–1.01) were associated with increased nonadherence.
DISCUSSION This study demonstrates that substance abuse, black race, urban residence, and emergency department and hospital visits Ó 2012 Lippincott Williams & Wilkins
TABLE 3. Multivariate results for nonadherence Nonadherence odds Characteristics ratio (95% CI) Demographic characteristics Agea Male sex Black race Disability Uninsured Uninsurable Median income for countyc Geographic region Middle West Urban/rural status Metropolitan Urban MCO type Regional Health care utilization Outpatient visits Emergency department visits Hospital visits Hospital days Comorbidity Obesity Diabetes Mental illness Substance abuse Hypercholesterolemia Charlson index Antihypertensive medication utilization Variety exposure (no. drug classes filled ever)
P
(0.97–0.97)b (1.07–1.17)b (1.58–1.78)b (0.57–0.67)b (0.77–0.91)b (0.47–0.56)b (1.00–1.00)
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.7606
1.04 (0.99–1.10) 1.09 (1.03–1.15)b
0.0837 0.002
0.96 (0.90–1.03) 1.12 (1.06–1.19)b
0.7199 0.0021
1.22 (1.17–1.27)b
,0.0001
0.99 1.07 1.12 1.01
(0.98–0.99)b (1.05–1.09)b (1.07–1.18)b (1.00–1.01)
,0.0001 ,0.0001 ,0.0001 0.2914
1.10 0.76 1.08 1.43 0.72 0.97
(1.01–1.20)b (0.70–0.84)b (1.02–1.13)b (1.32–1.55)b (0.68–0.75)b (0.96–0.98)b
0.0003 ,0.0001 0.0073 0.0258 ,0.0001 0.0375
0.81 (0.80–0.82)b
,0.0001
0.97 1.12 1.67 0.62 0.83 0.51 1.00
1 unit 5 1 yr. Significant at P , 0.05. 1 unit 5 $1000. CI, confidence interval; MCO, managed care organization.
a b c
serve as risk factors associated with antihypertensive medication nonadherence in a statewide Medicaid population. Furthermore, the study is among the first to document that routine outpatient visits are associated with a statistically significant, but small, decrease in medication nonadherence. This study documents high rates of antihypertensive medication nonadherence. Specifically, we found that 60.6% of chronic hypertensives were nonadherent and only 39.4% were adherent using an 80% cutoff criteria. Although on first glance these nonadherence rates seem extraordinarily high, in fact, they mirror antihypertensive medication nonadherence rates seen in other insured populations.7–11 These findings highlight the severity of nonadherence as an important target for intervention. Our study shows that antihypertensive medication nonadherence remains a significant health problem. This study highlights the importance of selected patient factors associated with nonadherence. Our findings are consistent with numerous other studies in showing that young age, urban residence and black race are associated with nonadherence.12–14
413
Bailey et al
In addition, we found that mental illness and substance abuse were associated with an 8% and a 43% increased odds of nonadherence, respectively, consistent with findings in other studies of vulnerable populations.13,15,33,34 Diagnosed obesity was also associated with poor adherence. This association has been previously reported by Krousel-Wood et al,35 and it has been suggested that higher body mass index may be a marker for low adherence with other recommended therapies such as diet and exercise. We suspect that obesity serves as a marker for other socioeconomic and psychosocial factors we were unable to assess. Previous studies have shown variable associations between comorbidity and medication nonadherence.12,16,36 In Tennessee’s Medicaid population, we found that both Charlson index and the specific comorbidities of diabetes and hypercholesterolemia were associated with decreased nonadherence. We agree with other authors who have noted that patients with multiple diagnosed comorbidities are more likely to be in regular care and to understand the importance of adherence.16,36 Of note, increased variety of antihypertensive medications taken once or more (number of drug classes filled ever) was associated with lower odds of nonadherence. Specifically, an increase in drug class exposure by 1 drug class was associated with a 19% decrease in nonadherence. This finding could also be explained through the use of several antihypertensive medications to treat other diseases (eg, angiotensinconverting enzyme inhibitors for diabetes and diuretics for heart failure). A greater number of antihypertensive drug classes is a marker for comorbidity and may contribute to this factor being protective of nonadherence. Alternatively, exposure to multiple medications is a marker of provider tailoring of the medication regimen or treatment intensification to meet patient needs. Heisler et al3 have documented strong associations between provider treatment intensification efforts and adherence. Emergency department visits and in-hospital stays were identified as risk factors associated with nonadherence. Each emergency department visit and in-hospital stay was associated with 7% and 12% increases in nonadherence, respectively. In contrast, an increase by 1 outpatient visit is associated with a decrease of nonadherence by 1%. This finding is consistent with the study by Caro et al that showed that increased numbers of physician visits were associated with increased medication persistence using pharmacy fill records.37,38 The magnitude of the effect seen in the current study is small, but our previous research demonstrates that the impact of even a small change in antihypertensive medication adherence on rates of stroke and death is significant.5 Given that there are approximately 68 million adults in the United States with hypertension and approximately 60% are nonadherent, we estimate that approximately 200,000 lives could be saved over the next 5 years by increasing adherence levels for all hypertensive patients to .80%. These estimates are consistent with those of others.2,18,39 Of particular interest to physicians is the finding that increases in ambulatory visits were associated with decreased odds of antihypertensive medication nonadherence. However, the effect of ambulatory visits seen in the current study was small and is probably of limited clinical relevance. This result is consistent with the a priori bias of physicians who believe that their outpatient efforts are effective in improving patient adherence. The effect of ambulatory care is likely mediated by increasing opportunities for measurement and improvement of hypertension control. The study provides little information on the characteristics of outpatient visits that are most effective. The research by Sung et al23 in Korea gives compelling evidence that primary care visits may be the most effective type of outpatient visits for
414
promoting antihypertensive medication adherence, but their findings need to be replicated in other settings. Future studies are needed to determine the primary mechanisms through which ambulatory visits improve outcomes, the optimal number and types of visits needed to maintain blood pressure control, and whether routine primary care, specialty, pharmacist or collaborative visit among provider types is the most effective in improving adherence. The main limitation of this study is the use of administrative data rather than clinical data. Hence, potentially important confounding variables such severity of hypertension were not measured. In addition, other behavioral and health care factors associated with antihypertensive medications adherence were not captured (eg, social support, knowledge, coping, stress, satisfaction with health care). Additionally, because this study employs a cross-sectional study design, its findings should not be considered definitive. Other study designs such as case-control and prospective studies are needed to establish the causal effect of outpatient visits on adherence. However, the study employed well-validated comorbidity adjustment methods for administrative data.24 Furthermore, the administrative database employed is known to accurately represent the health services exposures for a large population of patients with a common condition.25 Because the study examined a statewide Medicaid population, its findings should be generalizable to similar Medicaid populations throughout the United States. Also, the study provides limited information regarding medication-related factors associated with nonadherence; however, the literature is already extensive in this area. The implications of this research are significant. First, our research is consistent with numerous studies that reveal extremely high rates of antihypertensive medication nonadherence in numerous insured populations. This study highlights the importance of nonadherence as an important target for intervention. The study shows associations between patient and health services utilization factors and medication nonadherence that can further be employed by both clinicians and health systems to target at-risk patients. Specifically, this study shows that male sex, black race, obesity, substance abuse and history of emergency department or hospital utilization are associated with increased risk for antihypertensive nonadherence. Conversely, female sex, white race, comorbidity and outpatient visits are associated with decreased risk for medication nonadherence. Electronic health records can use these risk factors in decision support systems at the point of care to help clinicians proactively identify nonadherence when finding it difficult to control their patients’ blood pressure. Nonadherence risk profiles coupled with direct measures of MRA when pharmacy data are available could be made readily available through electronic health records.5 These markers can also enable medical homes and integrated delivery systems such as accountable care organizations to predict and improve adherence levels and support population-based care. Specifically, medical homes and accountable care organizations can use this information to initiate outreach programs to target hypertensive patients at highest risk and improve medication adherence. Furthermore, this study is among the first to identify associations between health services utilization and medication nonadherence in a statewide Medicaid population. We focused on the impact of ambulatory care visits on medication nonadherence. Our finding that increases in outpatient visits are associated with increased adherence to medications is significant. A small increase in health care costs as a result of increased medication utilization and outpatient visits would be expected to produce a net reduction in overall health care costs Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
as a result of reductions in emergency department and hospital visits achieved through improved adherence.18 Extensive ambulatory care research will be necessary to document the most effective methods for improving adherence and control of hypertension in ambulatory settings.
19. Weiden PJ, Kozma C, Grogg A, et al. Partial compliance and risk of rehospitalization among California Medicaid patients with schizophrenia. Psychiatr Serv 2004;55:886–91.
REFERENCES
21. Wu PH, Yang CY, Yao ZL, et al. Relationship of blood pressure control and hospitalization risk to medication adherence among patients with hypertension in Taiwan. Am J Hypertens 2010;23:155–60.
1. Cutler JA, Sorlie PD, Wolz M, et al. Trends in hypertension prevalence, awareness, treatment, and control rates in United States adults between 1988-1994 and 1999-2004. Hypertension 2008;52:818–27. 2. Ong KL, Cheung BM, Man YB, et al. Prevalence, awareness, treatment, and control of hypertension among United States adults 19992004. Hypertension 2007;49:69–75. 3. Heisler M, Hogan MM, Hofer TP, et al. When more is not better: treatment intensification among hypertensive patients with poor medication adherence. Circulation 2008;117:2884–92. 4. Morris AB, Li J, Kroenke K, et al. Factors associated with drug adherence and blood pressure control in patients with hypertension. Pharmacotherapy 2006;26:483–92. 5. Bailey JE, Wan JY, Tang J, et al. Antihypertensive medication adherence, ambulatory visits, and risk of stroke and death. J Gen Intern Med 2010;25:495–503. 6. Kettani FZ, Dragomir A, Cote R, et al. Impact of a better adherence to antihypertensive agents on cerebrovascular disease for primary prevention. Stroke 2009;40:213–20. 7. Lagi A, Rossi A, Passaleva MT, et al. Compliance with therapy in hypertensive patients. Intern Emerg Med 2006;1:204–8. 8. Siegel D, Lopez J, Meier J. Antihypertensive medication adherence in the Department of Veterans Affairs. Am J Med 2007;120:26–32. 9. Elliott WJ, Plauschinat CA, Skrepnek GH, et al. Persistence, adherence, and risk of discontinuation associated with commonly prescribed antihypertensive drug monotherapies. J Am Board Fam Med 2007;20:72–80. 10. Lowry KP, Dudley TK, Oddone EZ, et al. Intentional and unintentional nonadherence to antihypertensive medication. Ann Pharmacother 2005;39:1198–203. 11. Hyre AD, Krousel-Wood MA, Muntner P, et al. Prevalence and predictors of poor antihypertensive medication adherence in an urban health clinic setting. J Clin Hypertens (Greenwich) 2007;9:179–86. 12. Yang Y, Thumula V, Pace PF, et al. Predictors of medication nonadherence among patients with diabetes in Medicare Part D programs: a retrospective cohort study. Clin Ther 2009;31:2178–88; discussion 50–1. 13. Davis DP, Jandrisevits MD, Iles S, et al. Demographic, socioeconomic, and psychological factors related to medication non-adherence among emergency department patients [May 21, 2009]. J Emerg Med doi: 10.1016/j.jemermed.2009.04.008. 14. Bailey JE, Lee MD, Somes GW, et al. Risk factors for antihypertensive medication refill failure by patients under Medicaid managed care. Clin Ther 1996;18:1252–62. 15. Kim HK, Park JH, Kim JH. Differences in adherence to antihypertensive medication regimens according to psychiatric diagnosis: results of a Korean population-based study. Psychosom Med 2010;72:80–7. 16. Chapman RH, Petrilla AA, Benner JS, et al. Predictors of adherence to concomitant antihypertensive and lipid-lowering medications in older adults: a retrospective, cohort study. Drugs Aging 2008;25:885–92. 17. Payne KA, Esmonde-White S. Observational studies of antihypertensive medication use and compliance: is drug choice a factor in treatment adherence? Curr Hypertens Rep 2000;2:515–24. 18. Sokol MC, McGuigan KA, Verbrugge RR, et al. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care 2005;43:521–30.
Ó 2012 Lippincott Williams & Wilkins
20. Pittman DG, Tao Z, Chen W, et al. Antihypertensive medication adherence and subsequent healthcare utilization and costs. Am J Manag Care 2010;16:568–76.
22. Schedlbauer A, Davies P, Fahey T. Interventions to improve adherence to lipid lowering medication. Cochrane Database Syst Rev 2010; CD004371. 23. Sung SK, Lee SG, Lee KS, et al. First-year treatment adherence among outpatients initiating antihypertensive medication in Korea: results of a retrospective claims review. Clin Ther 2009;31:1309–20. 24. Tang J, Wan JY, Bailey JE. Performance of comorbidity measures to predict stroke and death in a community-dwelling, hypertensive Medicaid population. Stroke 2008;39:1938–44. 25. Ray WA. Policy and program analysis using administrative databases. Ann Intern Med 1997;127:712–8. 26. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105–16. 27. Hess LM, Raebel MA, Conner DA, et al. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother 2006;40:1280–8. 28. Choo PW, Rand CS, Inui TS, et al. Validation of patient reports, automated pharmacy records, and pill counts with electronic monitoring of adherence to antihypertensive therapy. Med Care 1999;37:846–57. 29. Chapman RH, Benner JS, Petrilla AA, et al. Predictors of adherence with antihypertensive and lipid-lowering therapy. Arch Intern Med 2005;165:1147–52. 30. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–9. 31. Quan H. ICD-9-CM enhanced Charlson SAS code. Winnipeg (Canada): Manitoba Centre for Health Policy; 2006. 32. SAS Institute. Statistical Analysis System (SAS), 9.1 ed. Cary (NC): SAS Institute, Inc.; 2003. 33. Kim MT, Han HR, Hill MN, et al. Depression, substance use, adherence behaviors, and blood pressure in urban hypertensive black men. Ann Behav Med 2003;26:24–31. 34. McLellan AT, Lewis DC, O’Brien CP, et al. Drug dependence, a chronic medical illness: implications for treatment, insurance, and outcomes evaluation. JAMA 2000;284:1689–95. 35. Krousel-Wood MA, Muntner P, Islam T, et al. Barriers to and determinants of medication adherence in hypertension management: perspective of the cohort study of medication adherence among older adults. Med Clin North Am 2009;93:753–69. 36. Lagu T, Weiner MG, Eachus S, et al. Effect of patient comorbidities on filling of antihypertensive prescriptions. Am J Manag Care 2009;15: 24–30. 37. Caro JJ, Salas M, Speckman JL, et al. Persistence with treatment for hypertension in actual practice. CMAJ 1999;160:31–7. 38. Caro JJ, Payne K. Real-world effectiveness of antihypertensive drugs. CMAJ 2000;162:190–1. 39. Gu Q, Burt VL, Paulose-Ram R, et al. High blood pressure and cardiovascular disease mortality risk among U.S. adults: the third National Health and Nutrition Examination Survey mortality follow-up study. Ann Epidemiol 2008;18:302–9.
415
Risk Factors Associated With Antihypertensive Medication Nonadherence in a Statewide Medicaid Population James E. Bailey, MD, MPH, Mohammed Hajjar, MD, Bushra Shoib, MD, Jun Tang, PhD, Mario M. Ray, MD and Jim Y. Wan, PhD
Abstract: Background: This study seeks to determine the most important patient factors and health care exposures available through administrative databases associated with antihypertensive nonadherence. Methods: This is a cross-sectional analysis of Medicaid hypertensive patients of Tennessee enrolled for 3 to 7 years from 1994 to 2000. Demographic characteristics, comorbidity and health care utilization were assessed during a 2-year period. The primary outcome was antihypertensive medication refill nonadherence. Subjects were categorized as adherent or nonadherent using an 80% cutoff criteria. Associations with nonadherence were assessed using logistic regression modeling. Results: Of 49,479 subjects, 60.6% (n 5 29,970) were classified as nonadherent and 39.4% (n 5 19,509) as adherent. Significant predictors of nonadherence in multivariate analysis (P , 0.05) included male sex (odds ratio [OR] 1.12), black race (OR 1.67), urban residence (OR 1.12), obesity (OR 1.10), mental illness (OR 1.08) and substance abuse (OR 1.43). Significant protective factors included age (OR 0.97), disability (OR 0.62), diabetes (OR 0.76), hypercholesterolemia (OR 0.72) and Charlson index (OR 0.97). When health care utilization was considered, increased outpatient visits were associated with decreased nonadherence. Emergency department visits (OR 1.07) and hospital visits (OR 1.12) were associated with increased nonadherence. Conclusions: This cross-sectional study suggests that substance abuse, black race, emergency department visits and hospitalizations are risk factors associated with nonadherence. Outpatient visits are associated with a small decrease in nonadherence. Further studies are needed to determine the characteristics of outpatient visits that most improve adherence. Key Indexing Terms: Hypertension; Medication adherence; Patient compliance; Ambulatory care; Health services research. [Am J Med Sci 2014;348(5):410–415.]
H
ypertension affects almost one third of Americans; yet, only two thirds of hypertensive patients are being treated, and only half of those treated are controlled.1,2 Antihypertensive medication nonadherence has been identified as a major reason for lack of adequate blood pressure control.3,4 Previous studies document that nonadherence with antihypertensive medication places patients at increased risk for cardiovascular complications.5,6 Numerous studies document extremely high rates of antihypertensive medication nonadherence ranging from 50%
to 75%.7–11 Despite the ready availability of numerous welltolerated antihypertensive medications, adherence to treatment remains a challenge. Both clinicians and policy makers need to know the patient factors and health services exposures associated with nonadherence to identify and target patients at highest risk for nonadherence with effective interventions. Literature review suggests that the most important patient risk factors associated with nonadherence include young age, low comorbidity, Hispanic or black race, mental illness and substance abuse.12–15 These studies show an inconsistent relationship between sex and nonadherence to antihypertensive medications. Previous research also suggests that numerous medicationrelated factors may also be important including choice of drug, use of concomitant medications and tolerability of drug side effects.14–17 Some early research documents that hospitalizations and emergency department use are strongly linked to nonadherence.18–21 Similarly, preliminary evidence from intervention research and some population-based studies suggests that certain types of ambulatory care may improve adherence.14,16,22,23 However, most of these studies have been limited in scope, lack external generalizability and considered a limited set of potential risk factors. To our knowledge, this study is the first to assess a broad range of potential risk factors associated with antihypertensive nonadherence in a statewide Medicaid population. Antihypertensive nonadherence is a clinical outcome linked to major cardiovascular events that is potentially amenable to change. The study seeks to determine the most important patient, medication-related and health services risk factors of those readily available in administrative databases associated with antihypertensive nonadherence among Medicaid enrollees. Based on the previous literature, we hypothesized that patient factors including black race, young age and history of substance abuse would be associated with increased nonadherence in this population. Furthermore, we hypothesized that exposure to outpatient visits would be associated with less nonadherence, whereas emergency department visits and hospitalizations would be associated with increased nonadherence.
METHODS From the Division of General Internal Medicine, Department of Medicine (JEB, MH, BS, JT, MMR), and Department of Preventive Medicine (JEB, JYW), University of Tennessee Health Science Center, Memphis, Tennessee. Submitted March 22, 2011; accepted in revised form August 10, 2011. Disclosure: This article has not been submitted elsewhere, is not under review, or has not been published previously except in abstract form. Preliminary findings from this research were presented at the 32nd Annual Meeting of the Society of General Internal Medicine; May 15, 2009; Miami, FL. The authors state that they have no conflicts of interest to disclose. Supported by a grant from the American Heart Association. Correspondence: James E. Bailey, MD, MPH, Division of General Internal Medicine, Department of Medicine, University of Tennessee Health Science Center, 956 Court Avenue, Coleman D222, Memphis, TN 38163 (E-mail: [email protected]).
410
Design and Setting This is a secondary analysis of retrospective cohort study database of all chronic medication-treated hypertensive individuals enrolled in Tennessee’s Medicaid-managed care system (TennCare) for 3 to 7 years during the period of 1994 to 2000. The study database and primary data sources have been described previously.5,24,25 This cross-sectional study of the 2-year baseline data for cohort participants is designed to document patient factors and medication and health services utilization factors associated with antihypertensive nonadherence in the 2-year baseline period for each subject.
The American Journal of the Medical Sciences
Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
Subjects Study participants were identified using eligibility, inpatient, professional and pharmacy claims data. The cohort included all noninstitutionalized persons with continuous eligibility (.320 days per year) throughout the 2-year study period, lack of Medicare eligibility, age 18 to 64 years in each study year, yearly diagnosis of essential hypertension (any International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis code 401.x for any professional or inpatient claim) and receipt of at least 1 antihypertensive medication prescription for each of the 2 baseline years. Patients who died or had a stroke (ICD-9-CM diagnosis code 430–438) during their baseline 2-year period were excluded. Study Outcome The study’s main outcome measure was baseline antihypertensive medication refill adherence (MRA). MRA was defined as the percentage of eligible prescription days filled (total days’ supply for all qualifying drug classes/total number of days from the first to the last fill in the interval 3 100, capped at 100%) for all antihypertensive medications taken in the 2-year time period. MRA was calculated for each individual for all qualifying antihypertensive drugs for the 2-year study period. Thus, MRA was averaged across all antihypertensive drug classes. For example, an individual on 2 drugs during the 2-year (730-day) study period, receiving 730 days’ supply for drug 1 and only 365 days’ supply for drug 2, would have an MRA of 75%. Only 1 fill within a drug class was required to define “use” of a class of medications. However, if a participant only had 1 prescription filled during the period, the day’s supply of the 1 prescription was used as the total number of eligible prescription days for that class of medications. If $2 fills were obtained for a class of medications, the eligible prescription days included all the days from the beginning of the first fill to the end of the last fill for that class in the study period. MRA produces identical results for adherence as the similarly calculated medication possession ratio and is a well-validated measure of medication adherence.26–28 All subjects were categorized as either adherent or nonadherent using an 80% cutoff because previous studies demonstrate a good validity using this value.16,29 Independent Variables Demographic variables were assessed using administrative demographic and eligibility files. Demographic characteristics assessed at baseline included age, sex, race, eligibility status, geographic region, urban/rural status, managed care organization (MCO) type and MCO for-profit status. All patients in the study were insured through TennCare, but their eligibility status differed. TennCare eligibility categories included the following: Medicaid only, disabled, uninsured (lacking insurance at the initiation of the TennCare program) and uninsurable (deemed uninsurable because of preexisting conditions). People enrolled in TennCare based on uninsurable eligibility status were required to show documentation (eg, a denial letter) from an insurance company denying them insurance on the basis of a preexisting condition or medical documentation of a preexisting condition (eg, HIV/AIDS) for which commercial insurance was known not to be available. Based on literature review suggesting strong associations between comorbidity and medication nonadherence, we assessed comorbidity in multiple ways using 2-year baseline inpatient and professional administrative data for paid claims. Overall comorbidity was assessed in each subject using the University of Manitoba SAS Macro for the Charlson index of Ó 2012 Lippincott Williams & Wilkins
comorbidity.30,31 In addition, we assessed specific comorbidities through the presence of inpatient or professional paid claims during the study period including the following diagnoses: obesity (ICD-9-CM code 278.0x), diabetes (250.x present in each year of the study period), hypercholesterolemia (272.x), substance abuse (303–305) and mental illness (290–302, 306–319) using previously described methods.5,24 Additional medication and health services utilization factors and medication utilization patterns were evaluated using paid pharmacy, inpatient and professional claims for the 2-year study period. Specifically, antihypertensive variety exposure was calculated as the number of drug classes filled ever, with all pharmaceutical claims for unique antihypertensive drugs collapsed in groups according to each drug class for which a participant had at least 1 prescription filled in the 2-year study period.5 In addition, 4 health care utilization variables were assessed: (1) ambulatory visits per year, (2) emergency department visits per year, (3) hospital visits per year and (4) hospital days per year. Statistical Analysis Chi-square and 2-sample t test statistics were used to assess bivariate associations with nonadherence. Multivariate associations with nonadherence were assessed using logistic regression modeling. All demographic, comorbidity and health care utilization variables found to be significant at P # 0.05 level in the bivariate analyses were included in the same multivariate logistic regression model. No adjustments were made based on whether medications were used in succession or concurrently; however, the antihypertensive variety exposure variable served to adjust for the effect of the number of antihypertensive medications prescribed on nonadherence. Variables in the multivariate model with P # 0.05 were declared significant. All analyses were performed using SAS.32
RESULTS Characteristics of the 49,479 subjects who met full inclusion criteria are shown in Table 1. Mean age for study subjects was 48.5 years (range, 20–64 years). The subjects were predominantly woman (67.7%), white (66.8%), uninsured (39.0%), disabled (36.8%) and in statewide MCOs (64.9%). Wide variation in levels of comorbidity was noted with Charlson index values ranging from 0 to 16, with a mean Charlson index of 1.71. Mean MRA was 67%, and only 39.4% were adherent using an 80% cutoff criteria. Bivariate associations with nonadherence are shown in Table 2. Significant predictors of nonadherence in bivariate analysis (P , 0.05) included black race (odds ratio [OR] 2.09, confidence interval [CI] 2.00–2.18), uninsured eligibility status (OR 1.32, CI 1.27–1.37), urban residence (OR 1.56, CI 1.50–1.62), diagnosed obesity (OR 1.19, CI 1.10–1.28), mental illness (OR 1.09, CI 1.05–1.14) and substance abuse (OR 1.62, CI 1.51–1.73). Significant factors associated with nonadherence included age (OR 0.96, CI 0.95–0.96), disability (OR 0.80, CI 0.77–0.83), uninsurable eligibility status (OR 0.53, CI 0.50– 0.55), metropolitan residence (OR 0.83, CI 0.80–0.88), diabetes (OR 0.67, CI 0.62–0.73), hypercholesterolemia (OR 0.57, CI 0. 0.55–0.60) and Charlson index (OR 0.91, CI 0.90–0.92). In addition, the following health services utilization variables were associated with nonadherence: emergency department visits (OR 1.08, CI 1.06–1.10), hospital visits (OR 1.08, CI 1.05– 1.11) and hospital days (OR 1.01, CI 1.01–1.02). Alternatively, outpatient visits were associated with decreased odds of nonadherence (OR 0.97, CI 0.97–0.98).
411
Bailey et al
TABLE 1. Characteristics of hypertensive patients (n 5 49,479) Characteristics Sex Male Female Unknown Race Black White Other Unknown Eligibility status Medicaid only Disability Uninsured Uninsurable Unknown Urban/rural status Urban Greater metropolitan area Rural Unknown Geographic region East Middle West Unknown MCO type Statewide Regional Unknown Comorbidity Diabetes Obesity Mental illness Substance abuse Congestive heart failure Atrial fibrillation Transient ischemic attack Myocardial infarction Hypercholesterolemia Comorbidity Charlson index Other demographic variables Age (yr) Socioeconomic status Income of county of residence in 1998 ($) Health care utilization Ambulatory visits per yr Emergency department visits per yr Hospital visits per yr Hospital days per yr Medication utilization Variety exposure (no. drug classes filled ever) Medication refill adherence (% eligible prescription days filled)
n (%) 15,974 (32.28) 33,504 (67.71) 1 (0.00) 13,829 33,048 245 2,357
(27.95) (66.79) (0.50) (4.76)
4,864 18,244 19,278 7,091 2
(9.83) (36.87) (38.96) (14.33) (0.00)
20,338 7,931 20,747 463
(41.10) (16.03) (41.93) (0.94)
19,328 15,012 14,676 463
(39.06) (30.34) (29.66) (0.94)
32,088 (64.85) 17,389 (35.14) 2 (0.00) 2,290 (4.63) 3,048 (6.16) 12,142 (24.54) 4,149 (8.39) 1,737 (3.51) 706 (1.43) 608 (1.23) 2,256 (4.56) 8,793 (17.77) Mean, median (range) SD 1.71, 1 (0–16) 1.91 48.5, 50 (20–64) 9.64 35,050, 35,483 (19,760–69,104) 6,384 5.2, 4 (0–86.5) 4.66 0.6, 0 (0–66) 1.38 0.30, 0 (0–13) 0.25 1.43, 0 (0–327) 5.90 2.33, 2 (1–8) 1.17 67, 72 (27–100) 25
MCO, managed care organization; SD, standard deviation.
412
Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
TABLE 2. Bivariate results for nonadherence Nonadherence odds Characteristics ratio (95% CI) Demographic characteristics Agea Male sex Black race Disability Uninsured Uninsurable Median income for countyc Urban/rural status Metropolitan Urban Geographic region Middle West MCO type Regional Health care utilization Outpatient visits Emergency department visits Hospital visits Hospital days Comorbidity Obesity Diabetes Mental illness Substance abuse Hypercholesterolemia Charlson index Medication utilization Variety exposure (no. drug classes filled ever)
P
(0.95–0.96)b (1.01–1.09)b (2.00–2.18)b (0.77–0.83)b (1.27–1.37)b (0.50–0.55)b (1.02–1.03)b
,0.0001 0.0098 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001
0.83 (0.80–0.88)b 1.56 (1.50–1.62)b
,0.0001 ,0.0001
0.93 (0.89–0.96)b 1.37 (1.31–1.42)b
0.0002 ,0.0001
1.22 (1.17–1.26)b
,0.0001
0.97 1.08 1.08 1.01
(0.97–0.98)b (1.06–1.10)b (1.05–1.11)b (1.01–1.02)b
,0.0001 ,0.0001 ,0.0001 ,0.0001
1.19 0.67 1.09 1.62 0.57 0.91
(1.10–1.28)b (0.62–0.73)b (1.05–1.14)b (1.51–1.73)b (0.55–0.60)b (0.90–0.92)b
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001
0.81 (0.80–0.82)b
,0.0001
0.96 1.05 2.09 0.80 1.32 0.53 1.02
1 unit 5 1 yr. Significant at P , 0.05. c 1 unit 5 $1000. CI, confidence interval; MCO, managed care organization. a b
In the multivariate analysis, shown in Table 3, significant factors associated with nonadherence (P , 0.05) included male sex (OR 1.12, CI 1.07–1.17), black race (OR 1.67, CI 1.58– 1.78), urban residence (OR 1.12, CI 1.06–1.19), obesity (OR 1.10, CI 1.01–1.20), mental illness (OR 1.08, CI 1.02–1.13) and substance abuse (OR 1.43, CI 1.32–1.55). Significant factors associated with decreased nonadherence included age (OR 0.97, CI 0.97–0.97), disability (OR 0.62, CI 0.57–0.67), uninsured eligibility status (OR 0.83, CI 0.77–0.91), uninsurable eligibility status (OR 0.51, CI 0.47–0.56), diabetes (OR 0.76, CI 0.70– 0.84), hypercholesterolemia (OR 0.72, CI 0.68–0.75) and Charlson index (OR 0.97, CI 0.96–0.98). When health care utilization was considered, increased outpatient visits (OR 0.99, CI 0.98– 0.99) were associated with decreased nonadherence. Emergency department visits (OR 1.07, CI 1.05–1.09), hospital visits (OR 1.12, CI 1.07–1.18) and hospital days (OR 1.01, CI 1.00–1.01) were associated with increased nonadherence.
DISCUSSION This study demonstrates that substance abuse, black race, urban residence, and emergency department and hospital visits Ó 2012 Lippincott Williams & Wilkins
TABLE 3. Multivariate results for nonadherence Nonadherence odds Characteristics ratio (95% CI) Demographic characteristics Agea Male sex Black race Disability Uninsured Uninsurable Median income for countyc Geographic region Middle West Urban/rural status Metropolitan Urban MCO type Regional Health care utilization Outpatient visits Emergency department visits Hospital visits Hospital days Comorbidity Obesity Diabetes Mental illness Substance abuse Hypercholesterolemia Charlson index Antihypertensive medication utilization Variety exposure (no. drug classes filled ever)
P
(0.97–0.97)b (1.07–1.17)b (1.58–1.78)b (0.57–0.67)b (0.77–0.91)b (0.47–0.56)b (1.00–1.00)
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.7606
1.04 (0.99–1.10) 1.09 (1.03–1.15)b
0.0837 0.002
0.96 (0.90–1.03) 1.12 (1.06–1.19)b
0.7199 0.0021
1.22 (1.17–1.27)b
,0.0001
0.99 1.07 1.12 1.01
(0.98–0.99)b (1.05–1.09)b (1.07–1.18)b (1.00–1.01)
,0.0001 ,0.0001 ,0.0001 0.2914
1.10 0.76 1.08 1.43 0.72 0.97
(1.01–1.20)b (0.70–0.84)b (1.02–1.13)b (1.32–1.55)b (0.68–0.75)b (0.96–0.98)b
0.0003 ,0.0001 0.0073 0.0258 ,0.0001 0.0375
0.81 (0.80–0.82)b
,0.0001
0.97 1.12 1.67 0.62 0.83 0.51 1.00
1 unit 5 1 yr. Significant at P , 0.05. 1 unit 5 $1000. CI, confidence interval; MCO, managed care organization.
a b c
serve as risk factors associated with antihypertensive medication nonadherence in a statewide Medicaid population. Furthermore, the study is among the first to document that routine outpatient visits are associated with a statistically significant, but small, decrease in medication nonadherence. This study documents high rates of antihypertensive medication nonadherence. Specifically, we found that 60.6% of chronic hypertensives were nonadherent and only 39.4% were adherent using an 80% cutoff criteria. Although on first glance these nonadherence rates seem extraordinarily high, in fact, they mirror antihypertensive medication nonadherence rates seen in other insured populations.7–11 These findings highlight the severity of nonadherence as an important target for intervention. Our study shows that antihypertensive medication nonadherence remains a significant health problem. This study highlights the importance of selected patient factors associated with nonadherence. Our findings are consistent with numerous other studies in showing that young age, urban residence and black race are associated with nonadherence.12–14
413
Bailey et al
In addition, we found that mental illness and substance abuse were associated with an 8% and a 43% increased odds of nonadherence, respectively, consistent with findings in other studies of vulnerable populations.13,15,33,34 Diagnosed obesity was also associated with poor adherence. This association has been previously reported by Krousel-Wood et al,35 and it has been suggested that higher body mass index may be a marker for low adherence with other recommended therapies such as diet and exercise. We suspect that obesity serves as a marker for other socioeconomic and psychosocial factors we were unable to assess. Previous studies have shown variable associations between comorbidity and medication nonadherence.12,16,36 In Tennessee’s Medicaid population, we found that both Charlson index and the specific comorbidities of diabetes and hypercholesterolemia were associated with decreased nonadherence. We agree with other authors who have noted that patients with multiple diagnosed comorbidities are more likely to be in regular care and to understand the importance of adherence.16,36 Of note, increased variety of antihypertensive medications taken once or more (number of drug classes filled ever) was associated with lower odds of nonadherence. Specifically, an increase in drug class exposure by 1 drug class was associated with a 19% decrease in nonadherence. This finding could also be explained through the use of several antihypertensive medications to treat other diseases (eg, angiotensinconverting enzyme inhibitors for diabetes and diuretics for heart failure). A greater number of antihypertensive drug classes is a marker for comorbidity and may contribute to this factor being protective of nonadherence. Alternatively, exposure to multiple medications is a marker of provider tailoring of the medication regimen or treatment intensification to meet patient needs. Heisler et al3 have documented strong associations between provider treatment intensification efforts and adherence. Emergency department visits and in-hospital stays were identified as risk factors associated with nonadherence. Each emergency department visit and in-hospital stay was associated with 7% and 12% increases in nonadherence, respectively. In contrast, an increase by 1 outpatient visit is associated with a decrease of nonadherence by 1%. This finding is consistent with the study by Caro et al that showed that increased numbers of physician visits were associated with increased medication persistence using pharmacy fill records.37,38 The magnitude of the effect seen in the current study is small, but our previous research demonstrates that the impact of even a small change in antihypertensive medication adherence on rates of stroke and death is significant.5 Given that there are approximately 68 million adults in the United States with hypertension and approximately 60% are nonadherent, we estimate that approximately 200,000 lives could be saved over the next 5 years by increasing adherence levels for all hypertensive patients to .80%. These estimates are consistent with those of others.2,18,39 Of particular interest to physicians is the finding that increases in ambulatory visits were associated with decreased odds of antihypertensive medication nonadherence. However, the effect of ambulatory visits seen in the current study was small and is probably of limited clinical relevance. This result is consistent with the a priori bias of physicians who believe that their outpatient efforts are effective in improving patient adherence. The effect of ambulatory care is likely mediated by increasing opportunities for measurement and improvement of hypertension control. The study provides little information on the characteristics of outpatient visits that are most effective. The research by Sung et al23 in Korea gives compelling evidence that primary care visits may be the most effective type of outpatient visits for
414
promoting antihypertensive medication adherence, but their findings need to be replicated in other settings. Future studies are needed to determine the primary mechanisms through which ambulatory visits improve outcomes, the optimal number and types of visits needed to maintain blood pressure control, and whether routine primary care, specialty, pharmacist or collaborative visit among provider types is the most effective in improving adherence. The main limitation of this study is the use of administrative data rather than clinical data. Hence, potentially important confounding variables such severity of hypertension were not measured. In addition, other behavioral and health care factors associated with antihypertensive medications adherence were not captured (eg, social support, knowledge, coping, stress, satisfaction with health care). Additionally, because this study employs a cross-sectional study design, its findings should not be considered definitive. Other study designs such as case-control and prospective studies are needed to establish the causal effect of outpatient visits on adherence. However, the study employed well-validated comorbidity adjustment methods for administrative data.24 Furthermore, the administrative database employed is known to accurately represent the health services exposures for a large population of patients with a common condition.25 Because the study examined a statewide Medicaid population, its findings should be generalizable to similar Medicaid populations throughout the United States. Also, the study provides limited information regarding medication-related factors associated with nonadherence; however, the literature is already extensive in this area. The implications of this research are significant. First, our research is consistent with numerous studies that reveal extremely high rates of antihypertensive medication nonadherence in numerous insured populations. This study highlights the importance of nonadherence as an important target for intervention. The study shows associations between patient and health services utilization factors and medication nonadherence that can further be employed by both clinicians and health systems to target at-risk patients. Specifically, this study shows that male sex, black race, obesity, substance abuse and history of emergency department or hospital utilization are associated with increased risk for antihypertensive nonadherence. Conversely, female sex, white race, comorbidity and outpatient visits are associated with decreased risk for medication nonadherence. Electronic health records can use these risk factors in decision support systems at the point of care to help clinicians proactively identify nonadherence when finding it difficult to control their patients’ blood pressure. Nonadherence risk profiles coupled with direct measures of MRA when pharmacy data are available could be made readily available through electronic health records.5 These markers can also enable medical homes and integrated delivery systems such as accountable care organizations to predict and improve adherence levels and support population-based care. Specifically, medical homes and accountable care organizations can use this information to initiate outreach programs to target hypertensive patients at highest risk and improve medication adherence. Furthermore, this study is among the first to identify associations between health services utilization and medication nonadherence in a statewide Medicaid population. We focused on the impact of ambulatory care visits on medication nonadherence. Our finding that increases in outpatient visits are associated with increased adherence to medications is significant. A small increase in health care costs as a result of increased medication utilization and outpatient visits would be expected to produce a net reduction in overall health care costs Volume 348, Number 5, November 2014
Antihypertensive Medication Nonadherence
as a result of reductions in emergency department and hospital visits achieved through improved adherence.18 Extensive ambulatory care research will be necessary to document the most effective methods for improving adherence and control of hypertension in ambulatory settings.
19. Weiden PJ, Kozma C, Grogg A, et al. Partial compliance and risk of rehospitalization among California Medicaid patients with schizophrenia. Psychiatr Serv 2004;55:886–91.
REFERENCES
21. Wu PH, Yang CY, Yao ZL, et al. Relationship of blood pressure control and hospitalization risk to medication adherence among patients with hypertension in Taiwan. Am J Hypertens 2010;23:155–60.
1. Cutler JA, Sorlie PD, Wolz M, et al. Trends in hypertension prevalence, awareness, treatment, and control rates in United States adults between 1988-1994 and 1999-2004. Hypertension 2008;52:818–27. 2. Ong KL, Cheung BM, Man YB, et al. Prevalence, awareness, treatment, and control of hypertension among United States adults 19992004. Hypertension 2007;49:69–75. 3. Heisler M, Hogan MM, Hofer TP, et al. When more is not better: treatment intensification among hypertensive patients with poor medication adherence. Circulation 2008;117:2884–92. 4. Morris AB, Li J, Kroenke K, et al. Factors associated with drug adherence and blood pressure control in patients with hypertension. Pharmacotherapy 2006;26:483–92. 5. Bailey JE, Wan JY, Tang J, et al. Antihypertensive medication adherence, ambulatory visits, and risk of stroke and death. J Gen Intern Med 2010;25:495–503. 6. Kettani FZ, Dragomir A, Cote R, et al. Impact of a better adherence to antihypertensive agents on cerebrovascular disease for primary prevention. Stroke 2009;40:213–20. 7. Lagi A, Rossi A, Passaleva MT, et al. Compliance with therapy in hypertensive patients. Intern Emerg Med 2006;1:204–8. 8. Siegel D, Lopez J, Meier J. Antihypertensive medication adherence in the Department of Veterans Affairs. Am J Med 2007;120:26–32. 9. Elliott WJ, Plauschinat CA, Skrepnek GH, et al. Persistence, adherence, and risk of discontinuation associated with commonly prescribed antihypertensive drug monotherapies. J Am Board Fam Med 2007;20:72–80. 10. Lowry KP, Dudley TK, Oddone EZ, et al. Intentional and unintentional nonadherence to antihypertensive medication. Ann Pharmacother 2005;39:1198–203. 11. Hyre AD, Krousel-Wood MA, Muntner P, et al. Prevalence and predictors of poor antihypertensive medication adherence in an urban health clinic setting. J Clin Hypertens (Greenwich) 2007;9:179–86. 12. Yang Y, Thumula V, Pace PF, et al. Predictors of medication nonadherence among patients with diabetes in Medicare Part D programs: a retrospective cohort study. Clin Ther 2009;31:2178–88; discussion 50–1. 13. Davis DP, Jandrisevits MD, Iles S, et al. Demographic, socioeconomic, and psychological factors related to medication non-adherence among emergency department patients [May 21, 2009]. J Emerg Med doi: 10.1016/j.jemermed.2009.04.008. 14. Bailey JE, Lee MD, Somes GW, et al. Risk factors for antihypertensive medication refill failure by patients under Medicaid managed care. Clin Ther 1996;18:1252–62. 15. Kim HK, Park JH, Kim JH. Differences in adherence to antihypertensive medication regimens according to psychiatric diagnosis: results of a Korean population-based study. Psychosom Med 2010;72:80–7. 16. Chapman RH, Petrilla AA, Benner JS, et al. Predictors of adherence to concomitant antihypertensive and lipid-lowering medications in older adults: a retrospective, cohort study. Drugs Aging 2008;25:885–92. 17. Payne KA, Esmonde-White S. Observational studies of antihypertensive medication use and compliance: is drug choice a factor in treatment adherence? Curr Hypertens Rep 2000;2:515–24. 18. Sokol MC, McGuigan KA, Verbrugge RR, et al. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care 2005;43:521–30.
Ó 2012 Lippincott Williams & Wilkins
20. Pittman DG, Tao Z, Chen W, et al. Antihypertensive medication adherence and subsequent healthcare utilization and costs. Am J Manag Care 2010;16:568–76.
22. Schedlbauer A, Davies P, Fahey T. Interventions to improve adherence to lipid lowering medication. Cochrane Database Syst Rev 2010; CD004371. 23. Sung SK, Lee SG, Lee KS, et al. First-year treatment adherence among outpatients initiating antihypertensive medication in Korea: results of a retrospective claims review. Clin Ther 2009;31:1309–20. 24. Tang J, Wan JY, Bailey JE. Performance of comorbidity measures to predict stroke and death in a community-dwelling, hypertensive Medicaid population. Stroke 2008;39:1938–44. 25. Ray WA. Policy and program analysis using administrative databases. Ann Intern Med 1997;127:712–8. 26. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105–16. 27. Hess LM, Raebel MA, Conner DA, et al. Measurement of adherence in pharmacy administrative databases: a proposal for standard definitions and preferred measures. Ann Pharmacother 2006;40:1280–8. 28. Choo PW, Rand CS, Inui TS, et al. Validation of patient reports, automated pharmacy records, and pill counts with electronic monitoring of adherence to antihypertensive therapy. Med Care 1999;37:846–57. 29. Chapman RH, Benner JS, Petrilla AA, et al. Predictors of adherence with antihypertensive and lipid-lowering therapy. Arch Intern Med 2005;165:1147–52. 30. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–9. 31. Quan H. ICD-9-CM enhanced Charlson SAS code. Winnipeg (Canada): Manitoba Centre for Health Policy; 2006. 32. SAS Institute. Statistical Analysis System (SAS), 9.1 ed. Cary (NC): SAS Institute, Inc.; 2003. 33. Kim MT, Han HR, Hill MN, et al. Depression, substance use, adherence behaviors, and blood pressure in urban hypertensive black men. Ann Behav Med 2003;26:24–31. 34. McLellan AT, Lewis DC, O’Brien CP, et al. Drug dependence, a chronic medical illness: implications for treatment, insurance, and outcomes evaluation. JAMA 2000;284:1689–95. 35. Krousel-Wood MA, Muntner P, Islam T, et al. Barriers to and determinants of medication adherence in hypertension management: perspective of the cohort study of medication adherence among older adults. Med Clin North Am 2009;93:753–69. 36. Lagu T, Weiner MG, Eachus S, et al. Effect of patient comorbidities on filling of antihypertensive prescriptions. Am J Manag Care 2009;15: 24–30. 37. Caro JJ, Salas M, Speckman JL, et al. Persistence with treatment for hypertension in actual practice. CMAJ 1999;160:31–7. 38. Caro JJ, Payne K. Real-world effectiveness of antihypertensive drugs. CMAJ 2000;162:190–1. 39. Gu Q, Burt VL, Paulose-Ram R, et al. High blood pressure and cardiovascular disease mortality risk among U.S. adults: the third National Health and Nutrition Examination Survey mortality follow-up study. Ann Epidemiol 2008;18:302–9.
415