factors associated with drug adherence and blood pressure control in patients with hypertension

Factors Associated with Drug Adherence and Blood Pressure Control in Patients with Hypertension

Andrew B. Morris, Pharm.D., M.S., Jingjin Li, Ph.D., Kurt Kroenke, M.D., Tina E. Bruner-England, Pharm.D., Jim M. Young, Pharm.D., and Michael D. Murray, Pharm.D., M.P.H.

Study Objectives. To determine characteristics associated with drugadherence and blood pressure control among patients with hypertension,and to assess agreement between self-reported and refill adherences.

Design. Cross-sectional analysis of baseline data from an ongoingrandomized controlled trial.

Setting. Primary care center at an urban, county health system inIndianapolis, Indiana.

Patients. Four hundred ninety-two participants with hypertension and takingat least one antihypertensive drug.

Measurements and Main Results. Social and demographic factors,comorbidity, self-reported drug adherence, prescription refill adherence,and systolic and diastolic blood pressures were recorded at baseline.Participants were aged 57 ± 11 (mean ± SD) years, were predominantlywomen (73%) and African-American (68%), and took 2.4 ± 1.1antihypertensive drugs. Agreement between self-reported and refilladherences was poor to fair (k = 0.21). On multiple logistic regressionanalysis, increased age (p≤0.002) and being married (p=0.03) wereindependent predictors of improved self-reported and refill adherence,whereas depressed patients had low self-reported adherence (p=0.005), andAfrican-Americans had low refill adherence (p<0.001). Compared withnonadherent patients, adherent patients had lower systolic (-5.4 mm Hg byself-report and -5.0 mm Hg by refill) and diastolic (-2.7 mm Hg by self-report and -3.0 mm Hg by refill) blood pressures (p≤0.02). Increased agewas the only other variable strongly associated with systolic and diastolicblood pressure control in both measures of drug adherence (p≤0.001). Theassociation of depression, race, and sex with blood pressure control wasmodel dependent.

Conclusion. Age, sex, race and depression are associated with antihyper-tensive drug adherence and blood pressure control. Self-reported and refilladherences appear to provide complementary information and areassociated with reductions in systolic and diastolic blood pressure ofsimilar magnitude.

Key Words: drug adherence, compliance, persistence, hypertension.(Pharmacotherapy 2006;26(4):483–492)

Hypertension affects an estimated 65 millionAmericans.1 Lowering blood pressure results insubstantial long-term reductions in the rates ofstroke, myocardial infarction, heart failure, andrenal failure. However, diagnosis and successful

control of hypertension have been elusive.2

Approximately 50% of patients drop out of carewithin the first year of diagnosis,3 and 50–66% ofthose who continue with antihypertensivetreatment comply with their prescribed drug

PHARMACOTHERAPY Volume 26, Number 4, 2006

regimen.3, 4 As a consequence, only 25–34% ofpatients with diagnosed hypertension havecontrolled blood pressure.2, 3

Hypertension is also a costly, chronic medicalcondition.5 Despite evidence that treatment andcontrol of hypertension improve health outcomesand lower health care costs,6, 7 control ofhypertension has not substantially improved overthe last several decades.2, 4 Poor adherence to theantihypertensive drug regimen is often presumedto be a major cause of poor blood pressurecontrol. Therefore, we analyzed baseline datafrom a study of drug adherence in patients withhypertension to address the agreement betweenself-reported adherence and pharmacy refilladherence, the patient characteristics associatedwith self-reported and refill adherence, and thepatient characteristics and drug adherencemeasures associated with systolic and diastolicblood pressure control.

Methods

Setting and Subjects

Eligible patients received their care at the adultgeneral medicine practices of Wishard HealthServices, Indianapolis, Indiana. This center is acity-county, tax-supported teaching hospital thatserves the indigent population of Indianapolisand Marion County, Indiana. It is located on thecampus of the Indiana University School ofMedicine.

Study participants were 492 patients enrolledin a randomized clinical trial to improve drugadherence in individuals with hypertension byusing a pharmacy-based intervention. Thenumber of participants was based on sample-size

calculations to determine differences in drugadherence between intervention and usual-caregroups in the primary trial. Participants wererecruited from February 2002—November 2003and followed for 18 months. Patients wereeligible if they had a diagnosis of hypertension,were at least 18 years of age, were taking at leastone antihypertensive drug, had planned onreceiving their primary care at the study centerand their prescription drugs at an center-associated pharmacy, could hear and understandspoken English, and had a working telephone intheir home.

Approval was received from the institutionalreview boards at Indiana University–PurdueUniversity Indianapolis, and the University ofNorth Carolina at Chapel Hill.

Study Design and Data Collection

This study was a cross-sectional analysis ofbaseline data from an ongoing randomizedcontrolled trial. Prescription records andcomorbidity data were extracted from theRegenstrief medical records system.8 Thiselectronic system contained longitudinal data ondiagnoses, prescriptions, blood pressure measure-ments, laboratory results (e.g., serum creatinineconcentrations), and clinical encounters forpatients receiving care at the study center. Thebaseline interview comprised assessments of self-reported drug adherence, depression, bloodpressure, and sociodemographic data.

Drug adherence was defined as the degree towhich the patient followed the prescribingphysician’s instructions for the prescription. Thisdefinition was consistent with one previouslyproposed.9 To ascertain drug adherence, we usedelectronic prescription records from the medicalrecords system and the patients’ reports ofadherence. Prescription refill adherence wascalculated during the year before enrollment forthe following classes of antihypertensive drugs:diuretics, β-adrenergic receptor antagonists,angiotensin II receptor antagonists, angiotensin-converting enzyme inhibitors, calcium channelblockers, vasodilators, α-receptor antagonists,and clonidine.

The prescription refill records were then usedto compute an overall medication possessionratio.10–14 Each time a prescription is refilled atthe study center’s pharmacies, the pharmacycomputer calculates the date of the next neededrefill based on number of tablets dispensed andthe frequency of administration. For each

484

From the Department of Pharmacy Practice, PurdueUniversity School of Pharmacy and Pharmacal Sciences,Indianapolis, Indiana (Dr. Morris); the Regenstrief Institute,Inc., Indianapolis, Indiana (Drs. Morris, Kroenke, andMurray); the Department of Medicine, Indiana UniversitySchool of Medicine, Indianapolis, Indiana (Drs. Li andKroenke); Wishard Health Services, Indianapolis, Indiana(Drs. Bruner-England and Young); and PharmaceuticalOutcomes and Policy Division and the Center forPharmaceutical Outcomes and Policy, University of NorthCarolina School of Pharmacy at Chapel Hill, Chapel Hill,North Carolina (Dr. Murray).

Supported by the National Institutes of Health (grantsR01 HL69399, R01 AG19105, and R01 AG07631).

Presented at the 28th annual meeting of the Society ofGeneral Internal Medicine, New Orleans, Louisiana, May11–14, 2005.

Address reprint requests to Michael D. Murray, Pharm.D.,M.P.H., Pharmaceutical Outcomes and Policy Division,Campus Box 7360, 2211 Kerr Hall, Chapel Hill, NC 27599-7360; e-mail: [email protected].

DRUG ADHERENCE IN HYPERTENSIVE PATIENTS Morris et al

prescription drug, the medication possessionratio is calculated as the number of days betweenthe last refill date and the next needed refill datedivided by the number of days between the lastrefill date and the date the patient actuallyreceived the refill. Therefore, the medicationpossession ratio can be thought of as theproportion of days that patients have access to asufficient amount of their antihypertension drug.A ratio of 1 suggested that the patient had accessto sufficient supplies of their drug, whereas lowervalues indicated that the patient had access to toolittle drug and higher values indicated that thepatient had access to an oversupply of the drug.We treated undersupply and oversupply similarlybecause we were interested in any deviation fromphysicians’ instructions for use of the drug. Theoverall pharmacy refill pattern was then definedas adherent when the overall medicationpossession ratio was 80–120%, and it wasnonadherent otherwise.14

Self-reported adherence was assessed by usingtwo published self-reported questionnaires forpopulations similar to our population.14, 15 Thefive questions we asked were as follows:

1. Many patients find it difficult to take theirdrugs as their doctor said they should. In thepast 4 weeks, do you think you have takenyour medicine as you should, on scheduleand regularly?16

2. In the past 4 weeks, did you ever forget totake your medicine?17

3. In the past 4 weeks, were you careless attimes about taking your medicine?17

4. In the past 4 weeks, when you felt better didyou sometimes stop taking your medicine?17

5. In the past 4 weeks, if sometimes you feltworse when you took your medicine, did youstop taking it?17

Patients were considered adherent to their drug ifthey affirmatively answered the first question andnegatively answered the other four.

Depression was assessed with the PatientHealth Questionnaire 8-item depression scale, onwhich a cutoff of 10 or higher representedclinically significant depression.18, 19

Trained and certified research assistantsobtained blood pressure measurements at thebaseline interview.20 Patients refrained fromsmoking or ingesting caffeine for at least 30minutes and remained seated for at least 5minutes before two successive measurementswere made with a properly sized mercurysphygmomanometer. Readings were separated by

2–5 minutes. Average systolic and diastolicblood pressures were calculated for use inanalyses.

Coronary artery disease, heart failure, myocardialinfarction, stroke, and renal insufficiency weredetermined by using diagnoses and laboratorydata from the medical records system.8 Coronaryartery disease was defined as one of thefollowing: a history of coronary artery bypass; adiagnosis of myocardial infarction, coronaryartery disease, or angina; electrocardiographicfindings of a myocardial infarction; or results ofcardiac echocardiography or catheterizationindicating wall motion abnor-malities. Heartfailure was defined by physician’s diagnosis orfindings on chest radiography, cardiac scanning,or catheterization. A history of myocardialinfarction was determined from cardiac scanning,echocardiography, catheteri-zation, or electro-cardiography that indicated a previous diagnosis.A history of stroke was a stroke diagnosis or acomputed tomography scan of the headindicating a stroke. Renal insufficiency wasbased on two determinations of creatinineclearances of 50 ml/minute or less by using theCockcroft-Gault equation21 with at least twoserum creatinine concentrations.

Social support was determined by using theMedical Outcomes Survey social support score.22

The survey was scored by totaling the responsesto 19 items with a range of 19–95. A high scoreindicated a high level of social support.

Pharmacy satisfaction was determined by usingan instrument developed to assess general patientsatisfaction with pharmacy services on a scale of1–5.23 The item scored was, “In general, I amsatisfied with the pharmacy services.” Thehigher the score, the stronger the agreement.

Medical insurance status was categorized asMedicare, Medicaid, Wishard Advantage, orprivate insurance; all others were classified as noinsurance. Pharmacy insurance status wasclassified as Medicaid, Wishard Advantage, orprivate pharmacy insurance. Wishard Advantageis a medical assistance program available to low-income residents of Marion County, Indiana.Because of the mission of the study center is toserve indigent persons, drugs are generallyprovided to indigent persons unable to pay forthem regardless of their insurance status.

Drug supervision was defined as yes or no,depending on whether the patient received anysupervision for taking drugs from nurses,spouses, children, other family members, andothers. If patients used tools or techniques to

485


remind themselves to take their drugs, the drugself-reminder item was defined as yes; otherwise,it was no. Examples of tools and techniqueswere alarm clocks, cups, drug boxes, turningover drug bottles, or writing down the drugtaken or to be taken.

If the patient could get to places beyondwalking distance, he or she answered yes; if not,it was answered not.

Trained interviewers administered the MedicalOutcomes Study Short Form 3624, 25 during thebaseline interview to quantify patient-perceivedhealth status, which they scored by usingstandard algorithms for the United States. A highscore denoted a perception of good health status.

Data and Statistical Analysis

To determine refill adherence, we firstcalculated the overall medication possession ratiofor each patient by averaging all individualmedication possession ratios weighted by theirtime intervals. Next, we converted the overallmedication possession ratio into a 0–1 scalebecause we considered both undersupplies andoversupplies as nonadherence. Consistent withour definition of adherence, when the medicationpossession ratio exceeded 1, we converted it to 2minus the medication possession ratio. We thenexamined the relationship between systolic anddiastolic blood pressures and the convertedmedication possession ratio. However, we didnot see a pattern that blood pressure decreasedmonotonically as the converted medicationpossession ratio increased. The correlationcoefficients were -0.09 (p=0.06) for systolicblood pressure and -0.2 (p<0.0001) for diastolicblood pressure. Therefore, we dichotomized theoverall medication possession ratio and definedthe overall pharmacy refill pattern as adherentwhen the overall medication possession ratio was0.80–1.20, and nonadherent otherwise. On self-report, patients were adherent if they answeredall five questions in favor of taking drugs; theywere nonadherent otherwise.

To evaluate the association between adherencemeasures and patient characteristics, we firstexamined their univariate relationship usinglogistic regression analysis. Patient charac-teristics included age, sex, race, marital status,years of education, relative income satisfaction,comorbidities (depression and complicatedhypertension if a patient had coronary arterydisease, heart failure, previous myocardialinfarction, stroke, or renal insufficiency), social

support, pharmacy satisfaction, supervision fortaking drugs, use of self-reminders for takingdrugs, and number of cigarettes smoked/day.Income satisfaction was categorized as comfort-able, just enough to get by, or not enough to getby. Sex, race, marital status, income satisfaction,comorbidities, pharmacy insurance, supervision,and use of self-reminder were treated as categoricvariables. Each level of a categoric variable wascompared with a reference level.

Factors statistically significant on univariateanalysis were entered as independent variables inmultivariable logistic regression models toexamine their ability to predict self-reported andrefill adherence. Forward and backward stepwiseselection procedures were used, and the samevariables were selected with either procedure.Finally, multiple linear regression was used todetermine the independent correlates of systolicand diastolic blood pressure. Self-reported andrefill adherences were evaluated in separatemodels for systolic and diastolic blood pressures.A p value less than 0.05 was considered toindicate a statistically significant difference.Given the exploratory nature of the study, noadjustments were made for multiple comparisons.

Results

Participants’ characteristics are shown in Table1. Hypertension was complicated in 46% ofparticipants. One fifth of the participants wereemployed but only 11% had comfortablehousehold incomes whereas 45% had justenough to make ends meet. Most participantshad health insurance and prescription drugcoverage. At enrollment, 43% of participants hadblood pressures that met the criteria of theseventh report of the Joint National Committeeon Prevention, Detection, Evaluation, andTreatment of High Blood Pressure2 for controlledblood pressure, whereas hypertension was notcontrolled in 57%.

Participants took 2.4 ± 1.1 (mean ± SD)antihypertensive drugs and 7.8 ± 3.5 regularlyadministered prescription drugs. About 79% ofpatients took their drugs without supervision,and 58% did not use drug reminders. Availabilityof the prescription data varied. Forty-oneparticipants had at least 1 year of previousprescription data, 191 had at least 11 months ofdata, and 318 had at least 10 months of data; webelieve that these durations were sufficient toprovide stable estimates of refill adherence. AsFigure 1 shows, agreement between self-reported

486


adherence and refill adherence was poor to fair (k= 0.21).

Table 2 summarizes the univariate associationsbetween adherence measures and patientcharacteristics. Neither the number of prescrip-tion drugs nor the number of comorbidities wasassociated with adherence in the models.Increased age and being married were associatedwith improved adherence in both self-reportedadherence and pharmacy refill adherence. In

addition, self-reported adherence was high inpatients reporting comfortable incomes, highlevels of social support, and high pharmacysatisfaction, and it was worst in participants withdepression. Pharmacy refill adherence wasincreased in patients with renal insufficiency anddecreased in African-Americans.

Table 3 shows the results of multiple logisticregression analysis of self-reported and refilladherences. Older and married participants hadbetter adherence on both measures than didothers. African-Americans were likely to benonadherent in terms of pharmacy refill but notself-report. To interpret these results, one shouldremember that self-reported adherence was basedon how patients take their drugs whereas refilladherence measured how much drug they hadavailable to take. Hence, the results suggestedthat African-American participants took theirdrugs when they had them but that they mighthave had difficulties with availability or supplies;these problems could have derived from socialand economic factors as well as practical issues,such as lack of transportation. Although mostpatients were offered medical care and prescrip-tion coverage, access to care and drugs might stillhave varied. Finally, although depressedparticipants were likely to be nonadherent onself-report, their supplies of drugs were accept-able based on prescription-refill patterns. Thisfinding suggested that depression affected thecompulsion to regularly take drugs despitepatients’ having reasonable access to supplies ofthose drugs.

487

Table 1. Characteristics of the 492 Patients

Characteristic ValueNo. (%)

SexFemale 360 (73.2)Male 132 (26.8)

African-American 336 (68.3)Married 104 (21.1)Income enough or comfortable 277 (56.3)Currently working 100 (2.3)Comorbidity

Coronary artery disease 117 (23.8)Heart failure 100 (20.3)Myocardial infarction 30 (6.1)Stroke 84 (17.1)Renal insufficiency 28 (5.7)Depressiona 189 (38.4)

Medical insurance 300 (61.0)Pharmacy insurance 337 (68.5)Drug administration supervised 101 (20.5)Uses drug self-reminders 206 (41.9)Gets to places beyond walking distance 302 (61.4)

Mean ± SDAge (yrs) 56.6 ± 10.8Education (yrs) 11.3 ± 2.4Age at first diagnosis of hypertension (yrs) 41.2 ± 13.9Cigarettes smoked/day 5.1 ± 8.7Comorbidities 3.2 ± 1.4Social supportb 72.6 ± 17.2Pharmacy satisfaction scorec 3.1 ± 1.26Drugs 7.8 ± 3.5Scores on Short Form 36d items

1. Physical functioning 46.9 ± 27.32. Role, physical 40.6 ± 40.13. Freedom from pain 55.5 ± 11.24. Health perception 43.8 ± 20.5

Items 1–4 46.7 ± 18.65. Vitality 45.1 ± 23.06. Social functioning 71.8 ± 27.17. Role, emotional 63.1 ± 40.98. Mental health 65.9 ± 23.2

Items 5–8 61.5 ± 22.9aIncludes patients with a score of 10 or higher on the Patient HealthQuestionnaire 8-item depression scale. Scores above 9 designatemoderate-to-severe depression.18, 19

bMedical Outcomes Study social support score.22

cRating of pharmacy satisfaction for “in general, I am satisfied withthe pharmacy services”: scores ranged from 1–5, where increasingscores indicate stronger agreement.23

dMedical Outcomes Study Short Form 36.24, 25

RefillAdherence

Sel

f-R

epor

tA

dhe

renc

e

Yes

Yes

No

No

133

142

51

154

Figure 1. Agreement of self-report adherence versus refilladherence (k = 0.21).


Table 4 lists the variables associated with bloodpressure control in the multiple linear regressionmodels. Increasing age was associated withblood pressure control in all models. A 3-mm Hgincrease in systolic blood pressure and a 3-mmHg decrease in diastolic blood pressure wereobserved for each decade of age. Of importance,both measures of adherence were associated withlowered systolic (-5.4 to 5.0 mm Hg) anddiastolic (-3.0 to -2.7 mm Hg) blood pressures.Depression was associated with a high diastolicblood pressure in refill adherence, but it was notstatistically significant for diastolic bloodpressure on self-report. Men had higher diastolicblood pressures in both adherence models thanwomen, whereas African-Americans had highdiastolic blood pressures only in terms of self-reported adherence.

Discussion

We studied drug adherence using a self-reportquestionnaire and prescription refill patterns andtheir relationships to characteristics and

outcomes of patients with hypertension. Theagreement between the adherence measures waspoor to fair (k = 0.21). However, both measure-ments were associated with patient charac-teristics and blood pressure control. Patientcharacteristics were associated with adherence,and the strength of association differed betweenself-reported and refill adherences. Differences inthe dimensions thus measured likely explain thisfinding. The questions on self-reported adherenceinvolved all drugs taken and used a 4-week frameof reference. The responses to these questionsreflected the patients’ willingness to accuratelydescribe their drug-taking behaviors for all drugs.In contrast, refill adherence targeted onlyantihypertensives provided during the 12 monthsbefore the self-report questionnaire wasadministered. Refill adherence largely providedinformation on the amount of antihypertensivedrugs supplied to patients.13, 15 Previous investi-gators examined potential factors associated withdrug adherence, with variable results.10, 12, 14, 26–46

This variability could be related to size of thestudy sample, differences in the definitions for

488

Table 2. Univariate Association Between Patient Characteristics and Adherences in 475 Patientsa

Self-Reported Adherence Refill AdherenceCharacteristic Odds Ratio 95% CI p Value Odds Ratio 95% CI p ValueAge 1.04 1.02–1.06 <0.001 1.03 1.01–1.04 0.005Male sex 1.43 0.94–2.16 0.09 1.30 0.85–1.97 0.22African-American 0.85 0.57–1.26 0.42 0.39 0.25–0.59 <0.001Education 1.05 0.97–1.13 0.26 0.94 0.87–1.02 0.11Married 1.56 1.00–2.45 0.05 1.67 1.05–2.67 0.03Incomeb 3.46 1.83–6.55 <0.001 1.18 0.63–2.23 0.68Comorbidity

Depressionc 0.41 0.27–0.61 <0.001 0.82 0.56–1.19 0.25Coronary artery disease 0.81 0.52–1.25 0.33 0.99 0.65–1.52 0.96Heart failure 1.15 0.73–1.82 0.54 0.91 0.58–1.43 0.68Myocardial infarction 1.19 0.55–2.58 0.66 1.12 0.51–2.45 0.77Stroke 1.19 0.73–1.95 0.48 1.25 0.76–2.04 0.38Renal insufficiency 1.49 0.68–3.25 0.32 2.65 1.05–6.69 0.04

Smokingd 0.99 0.97–1.01 0.27 0.99 0.97–1.01 0.29Drug administration supervised 0.82 0.52–1.30 0.40 0.71 0.45–1.11 0.13Uses drug self-reminders 0.81 0.55–1.17 0.26 1.03 0.71–1.49 0.88Pharmacy satisfaction scoree 1.17 1.01–1.36 0.04 0.93 0.81–1.08 0.35Social support scoref 1.02 1.01–1.03 0.002 1.01 1.00–1.02 0.15Pharmacy insurance statuse 1.54 0.66–3.64 0.18 0.73 0.30–1.75 0.43No. of drugs 0.99 0.93–1.06 0.85 1.0 0.98–1.12 0.16CI = confidence interval.aNumber does not total 492 because of missing data.bComfortable vs not enough.cIncludes patients with a score of 10 or higher on the Patient Health Questionnaire 8-item depression scale. Scores above 9 designatemoderate-to-severe depression.18, 19

dCigarettes smoked/day.eRating of pharmacy satisfaction for “in general, I am satisfied with the pharmacy services”: scores ranged from 1–5, where increasingscores indicate stronger agreement.23

fSocial support scores ranged from 19–95, where increasing scores indicate increasing social support.22

eMedicaid vs other.


adherence, heterogeneity of patient populations,and differing outcomes measured in each study.We believe that self-reported and refill adherencemeasures provide different but complementaryinformation and perspectives on drug adherence.

Deciding which measure has greatest validityfor a particular patient with hypertension isdifficult because both measures were associatedwith similar reductions in systolic and diastolicblood pressure. Patients who state they areadherent but who are not adherent in pharmacyrefills could be using hoarded drugs; conversely,patients adherent with refills may not necessarilybe taking their drug as they say. Both types ofmeasures may be considered in research studiesof adherence, and clinicians might consider bothself-report and refill history in judging a patient’sadherence. When both measures agree, confidenceabout a patient’s adherence increases. On theother hand, when the measures disagree, furtherprobing may be justified, particularly whendisease control is suboptimal. Further studies ofhow self-reported adherence and pharmacy refillscorrespond to other adherence measures (e.g.,electronic monitors, plasma drug concentrations)are needed.

Increasing age and being married were

associated with improved adherence by bothmeasures, whereas depressed participants werelikely to be nonadherent only on self-report, andAfrican-Americans were likely to be nonadherentby refill patterns. Previous findings on age andadherence have been mixed. Some studies haveshown increased adherence with increased age,29,

47 whereas others have reported the opposite.37, 42,

43, 48, 49 The relationship between marital statusand improved adherence suggests a reinforcing orreminding function of a spouse. Although resultsassociated with depression and race were modeldependent, the direction of their relationship wasstable. Reports have suggested a negative effectof depression on adherence10, 35, 42, 46, 50–52

Evidence also implies that racial or culturaldifferences may affect adherence.35, 52, 53 Unlikeage and race, depression may be an importantmodifiable factor among patients withsuboptimal blood pressure control, particularlywhen nonadherence is suspected.

Drug adherence, regardless of the measurementused, independently affected blood pressurecontrol comparable to lifestyle modification (adecrease of approximately 5 mm Hg in systolicblood pressure and 3 mm Hg in diastolic bloodpressure). After we controlled for adherence,

489

Table 3. Factors Associated with Adherences in Multiple Logistic Regression Models in 469 Patientsa

Self-Reported Adherence Refill AdherenceFactor Odds Ratio 95% CI p Value Odds Ratio 95% CI p ValueAge 1.04 1.02–1.06 <0.001 1.03 1.01–1.05 0.002Married 1.68 1.05–2.69 0.03 1.71 1.05–2.79 0.03African-American 0.82 0.54–1.24 0.34 0.34 0.22–0.53 <0.001Depressionb 0.48 0.32–0.72 0.005 0.94 0.63–1.40 0.76CI = confidence interval.aNumber does not total 492 because of missing data.bIncludes patients with a score of 10 or higher on the Patient Health Questionnaire 8-item depression scale. Scores above 9designate moderate-to-severe depression.18, 19

Table 4. Predictors of Blood Pressure Control in Multiple Linear Regression Models in 470 Patientsa

Systolic Blood Pressure Diastolic Blood PressureSelf-Reported Adherence Refill Adherence Self-Reported Adherence Refill AdherenceEstimate Estimate Estimate Estimate

Predictor (mm Hg) p Value (mm Hg) p Value (mm Hg) p Value (mm Hg) p ValueIncreased age (per decade) 3.30 <0.001 3.19 0.001 -3.47 <0.001 -3.49 <0.001Male sex -0.39 0.87 -0.64 0.78 2.81 0.02 2.71 0.03African-American 3.65 0.09 2.68 0.23 2.51 0.03 1.92 0.11Adherence -5.35 0.01 -5.01 0.02 -2.72 0.02 -2.96 0.009Depressionb 1.59 0.46 2.37 0.26 2.02 0.08 2.40 0.03aNumber does not total 492 because of missing data.bIncludes patients with a score of 10 or higher on the Patient Health Questionnaire 8-item depression scale. Scores above 9 designate moderate-to-severe depression.18, 19


increased age was the other variable associatedwith both systolic and diastolic blood pressurecontrol in all models. Indeed, the widening pulsepressure seen with age is consistent with previousfindings in patients older than 50 years.54

Depression adversely affected diastolic andsystolic blood pressures (increase of 2.0–2.4 and1.6–2.4 mm Hg, respectively), although thisfinding was statistically significant only in therefill adherence model for diastolic bloodpressure. Others have found significantcorrelations between psychological factors anddiastolic blood pressure.55 Men had highdiastolic (but not systolic) pressures, whereasAfrican-Americans had significantly increaseddiastolic pressures only in the self-reportedadherence model.

Our results have relevance from a clinicalperspective. Owing to the difficulties incontrolling blood pressure among patients withhypertension and the association betweenadherence and blood pressure control, webelieved it pertinent to learn more about thefactors associated with adherence. Awareness ofsuch factors permits additional targetedinterventions to improve adherence in largepopulations of patients. Adherence is particu-larly difficult to assess in clinical settings, and thevarious methods of measurement haveadvantages and disadvantages. For instance,electronic monitoring is expensive, refilladherence or the medication possession ratiomay inflate adherence rates when dosageregimens are changed, and self-report is thoughtto be accurate when patients state theirdifficulties in taking their drugs but questionablewhen they report that they take them. Althoughour results revealed differences in factorsassociated with adherence, both self-reported andrefill adherences were associated with changes insystolic and diastolic blood pressures of the samemagnitude. This observation supports the notionthat assessing adherence by using either methodis worthwhile in the setting of clinical care.

Several limitations of our study areacknowledged. First, our sample was taken froma clinical trial at a single medical center, and, assuch, it may not be generalizable to other caresettings. Furthermore, many patients were poorand from minority groups, and only 20% wereactively employed. Because we used onlybaseline data, confounding by the interventionwas not a problem, yet characteristics associatedwith volunteering for a clinical trial could berelated to adherence. Therefore, the generaliz-

ability of our findings should be verified inclinical populations broader and more inclusivethan ours. Second, our self-reported adherencemeasure included all drugs and not onlyantihypertensive drugs, and we asked aboutadherence during the 4 weeks preceding baseline.Extending the self-reporting period would haveintroduced additional recall bias. In contrast,refill adherence was measured for the preceding12 months and focused exclusively on patients’antihypertensive drugs. Although these drug-specific and timing differences might account forsome of the discordance between self-reportedadherence and refill adherence, we doubt it is themajor explanation.

Conclusion

Age, marital status, race and depression wereassociated with drug adherence. As expected,drug adherence was associated with improvedblood pressure control. Even when we controlledfor adherence, the factors of age, sex, race, anddepression were independently associated withblood pressure control, although age was theonly factor besides adherence that affected bothdiastolic and systolic blood pressures in allmodels. The magnitude of the effect ofdepression on diastolic blood pressureapproached that of certain modifiable lifestylefactors and suggests that depression screeningmay be warranted in patients with suboptimalblood pressure control, particularly ifnonadherence is suspected. In assessingadherence, attention to both patient self-reportedadherence and drug refill patterns may provide amore complete adherence gestalt than what existsnow.

Acknowledgment

The authors gratefully acknowledge J. HowardPratt, M.D., for his insightful advice and suggestionsin the development of this manuscript.

References1. Fields LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, Sorlie P.

The burden of adult hypertension in the United States 1999 to2000: a rising tide. Hypertension 2004;44:398–404.

2. Chobanian AV, Bakris GL, Black HR, et al. The seventh reportof the joint national committee on prevention, detection,evaluation, and treatment of high blood pressure: the JNC 7report. JAMA 2003;289(19):2560–72.

3. Mendis S, Salas M. Hypertension. In: Sabate E, ed. Adherenceto long-term therapies, evidence for action, 2003. Geneva,Switzerland: World Health Organization, 2003:107–14.

4. Eraker SA, Kirscht JP, Becker MH. Understanding andimproving patient compliance. Ann Intern Med 1984;100(2):258–68.

490


5. Goetzel RZ, Long SR, Ozminkowski RJ, Hawkins K, Wang S,Lynch W. Health, absence, disability, and presenteeism costestimates of certain physical and mental health conditionsaffecting U.S. employers. J Occup Environ Med 2004;46(4):398–412.

6. Ogden LG, He J, Lydick E, Whelton PK. Long-term absolutebenefit of lowering blood pressure in hypertensive patientsaccording to the JNC VI risk stratification. Hypertension2000;35(2):539–43.

7. Maronde RF, Chan LS, Larsen FJ, Strandberg LR, LaventurierMF, Sullivan SR. Underutilization of antihypertensive drugsand associated hospitalization. Med Care 1989;27(12):1159–66.

8. McDonald CJ, Overhage JM, Tierney WM, et al . TheRegenstrief medical record system: a quarter centuryexperience. Int J Med Inform 1999;54(3):225–53.

9. Haynes B, Taylor DW, Sackett DL. Compliance in health care.Baltimore: Johns Hopkins University Press, 1978.

10. Wang PS, Bohn RL, Knight E, Glynn RJ, Mogun H, Avorn J.Noncompliance with antihypertensive medications: the impactof depressive symptoms and psychosocial factors. J Gen InternMed 2002;17(7):504–11.

11. Steiner JF, Prochazka AV. The assessment of refill complianceusing pharmacy records: methods, validity, and applications. JClin Epidemiol 1997;50(1):105–16.

12. Vik SA, Maxwell CJ, Hogan DB. Measurement, correlates, andhealth outcomes of medication adherence among seniors. AnnPharmacother 2004;38(2):303–12.

13. Stroupe KT, Murray MD, Stump TE, Callahan CM .Association between medication supplies and healthcare costsin older adults from an urban healthcare system. J Am GeriatrSoc 2000;48(7):760–8.

14. Cargill JM . Medication compliance in elderly people:influencing variables and interventions. J Adv Nurs1992;17(4):422–6.

15. Stroupe KT, Teal EY, Weiner M, Gradus-Pizlo I, Brater DC,Murray MD. Health care and medication costs and use amongolder adults with heart failure. Am J Med 2004;116(7):443–50.

16. Inui TS, Carter WB, Pecoraro RE . Screening for non-compliance among patients with hypertension: is self-report thebest available measure? Med Care 1981;19(10):1061–4.

17. Morisky DE, Green LW, Levine DM. Concurrent and predictivevalidity of a self-reported measure of medication adherence.Med Care 1986;24(1):67–74.

18. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of abrief depression severity measure. J Gen Intern Med2001;16(9):606–13.

19. Kroenke K, Spitzer RL. The PHQ-9: a new depression diag-nostic and severity measure. Psychiatr Ann 2003;32(9):1–7.

20. Grim CM, Grim CE. Blood pressure measurement. In: Izzo JL,Black HR, eds. Hypertension primer, 2nd ed. Dallas, TX:American Heart Association, 1999:295–8.

21. Cockcroft DW, Gault MH. Prediction of creatinine clearancefrom serum creatinine. Nephron 1976;16(1):31–41.

22. Sherbourne CD, Stewart AL. The MOS social support survey.Social Sci Med 1991;32(6):705–14.

23. Murray MD, Young JM, Morrow DG, et al. Methodology of anongoing, randomized, controlled trial to improve drug use forelderly patients with chronic heart failure. Am J GeriatrPharmacother 2004;2(1):53–65.

24. Ware JE, Sherbourne CD. The MOS 36 item short-form healthsurvey: conceptual framework and item selection. Med Care1992;30:473–83.

25. Ware JE, Kosisnski M, Keller SD. SF-36 physical and mentalhealth summary scales: a users manual, 4th ed. Boston, MA:The Health Institute, 1994.

26. Starace F, Ammassari A, Trotta MP, et al. Depression is a riskfactor for suboptimal adherence to highly active antiretroviraltherapy. J Acquir Immune Defic Syndr 2002;31(suppl3):S136–9.

27. Knight EL, Bohn RL, Wang PS, Glynn RJ, Mogun H, Avorn J.Predictors of uncontrolled hypertension in ambulatory patients.Hypertension 2001;38(4):809–14.

28. DeVon HA, Powers MJ. Health beliefs, adjustment to illness,and control of hypertension. Res Nurs Health 1984;7(1):10–16.

29. Ford FM, Hunter M, Hensley MJ, et al. Hypertension andasthma: psychological aspects. Soc Sci Med 1989;29(1):79–84.

30. Bailey JE, Lee MD, Somes GW, Graham RL. Risk factors forantihypertensive medication refill failure by patients underMedicaid managed care. Clin Ther 1996;18(6):1252–62.

31. Waterhouse DM, Calzone KA, Mele C, Brenner DE .Adherence to oral tamoxifen: a comparison of patient self-report, pill counts, and microelectronic monitoring. J ClinOncol 1993;11(6):1189–97.

32. Eisen SA, Miller DK, Woodward RS, Spitznagel E, PrzybeckTR. The effect of prescribed daily dose frequency on patientmedication compliance. Arch Intern Med 1990;150(9):1881–4.

33. The Boston Consulting Group. The hidden epidemic: finding acure for unfilled prescriptions and missed doses, December2003. Available from www.bcg.com/publications/files/TheHiddenEpidemic_Rpt_HDec03.pdf. Accessed January 5,2005.

34. Haynes RB, Taylor DW, Sackett DL, Gibson ES, Bernholz CD,Mukherjee J. Can simple clinical measurements detect patientnoncompliance? Hypertension 1980;2(6):757–64.

35. Singh N, Squier C, Sivek C, Wagener M, Nguyen MH, Yu VL.Determinants of compliance with antiretroviral therapy inpatients with human immunodeficiency virus: prospectiveassessment with implications for enhancing compliance. AIDSCare 1996;8(3):261–9.

36. Wang PS, Benner JS, Glynn RJ, Winkelmayer WC, Mogun H,Avorn J. How well do patients report noncompliance withantihypertensive medications?: a comparison of self-reportversus filled prescriptions. Pharmacoepidemiol Drug Saf2004;13(1):11–19.

37. Okuno J, Yanagi H, Tomura S, et al . Compliance andmedication knowledge among elderly Japanese home-carerecipients. Eur J Clin Pharmacol 1999;55(2):145–9.

38. Choo PW, Rand CS, Inui TS, Lee ML, Canning C, Platt R. Acohort study of possible risk factors for over-reporting ofantihypertensive adherence. BMC Cardiovasc Disord2001;1(1):1–6.

39. Billups SJ, Malone DC, Carter BL. The relationship betweendrug therapy noncompliance and patient characteristics,health-related quality of life, and health care costs.Pharmacotherapy 2000;20(8):941–9.

40. Gray SL, Mahoney JE, Blough DK. Medication adherence inelderly patients receiving home health services followinghospital discharge. Ann Pharmacother 2001;35(5):539–45.

41. Botelho RJ, Dudrak R. Home assessment of adherence to long-term medication in the elderly. J Fam Pract 1992;35(1):61–5.

42. Spiers MV, Kutzik DM. Self-reported memory of medicationuse by the elderly. Am J Health-Syst Pharm 1995;52(9):985–90.

43. Spagnoli A, Ostino G, Borga AD, et al. Drug compliance andunreported drugs in the elderly. J Am Geriatr Soc1989;37(7):619–24.

44. Monane M, Bohn RL, Gurwitz JH, Glynn RJ, Levin R, Avorn J.The effects of initial drug choice and comorbidity onantihypertensive therapy compliance: results from apopulation-based study in the elderly. Am J Hypertens1997;10(7 pt 1):697–704.

45. Molassiotis A, Nahas-Lopez V, Chung WY, Lam SW, Li CK,Lau TF. Factors associated with adherence to antiretroviralmedication in HIV-infected patients. Int J STD AIDS2002;13(5):301–10.

46. DiMatteo MR, Lepper HS, Croghan TW. Depression is a riskfactor for noncompliance with medical treatment: meta-analysisof the effects of anxiety and depression on patient adherence.Arch Intern Med 2000;160(14):2101–7.

47. Monane M, Bohn RL, Gurwitz JH, Glynn RJ, Avorn J.Noncompliance with congestive heart failure therapy in theelderly. Arch Intern Med 1994;154(4):433–7.

48. Jackevicius CA, Mamdani M, Tu JV. Adherence with statintherapy in elderly patients with and without acute coronarysyndromes. JAMA 2002;288(4):462–7.

49. Avorn J, Monette J, Lacour A, et al. Persistence of use of lipid-

491


lowering medications: a cross-national study. JAMA1998;279(18):1458–62.

50. Romanelli J, Fauerbach JA, Bush DE, Ziegelstein RC. Thesignificance of depression in older patients after myocardialinfarction. J Am Geriatr Soc 2002;50(5):817–22.

51. Bosley CM, Corden ZM, Rees PJ, Cochrane GM. Psychologicalfactors associated with use of home nebulized therapy forCOPD. Eur Respir J 1996;9(11):2346–50.

52. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC,Avorn J. Long-term persistence in use of statin therapy inelderly patients. JAMA 2002;288(4):455–61.

53. Rich MW, Gray DB, Beckham V, Wittenberg C, Luther P.Effect of a multidisciplinary intervention on medicationcompliance in elderly patients with congestive heart failure. AmJ Med 1996;101(3):270–6.

54. Alfie J, Waisman GD, Galarza CR, Camera MI. Contributionof stroke volume to the change in pulse pressure pattern withage. Hypertension 1999;34(4 pt 2):808–12.

55. Haynes SG, Levine S, Scotch N, Feinleib M, Kannel WB. Therelationship of psychosocial factors to coronary heart disease inthe Framingham study, I: methods and risk factors. Am JEpidemiol 1978;107(5):362–83.

492

factors associated with drug adherence and blood pressure control in patients with hypertension

Documents