L E KleinCompliance and blood pressure control.

Print ISSN: 0194-911X. Online ISSN: 1524-4563 Copyright © 1988 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Hypertension doi: 10.1161/01.HYP.11.3_Pt_2.II61

1988;11:II61Hypertension. 

http://hyper.ahajournals.org/content/11/3_Pt_2/II61World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://hyper.ahajournals.org//subscriptions/

is online at: Hypertension Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer process is available in the

Request Permissions in the middle column of the Web page under Services. Further information about thisOffice. Once the online version of the published article for which permission is being requested is located, click

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the EditorialHypertension Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:

at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from

Compliance and Blood Pressure ControlLAWRENCE ELLIOT KLEIN

SUMMARY Compliance with medical therapy in general is often low, and compliance with bloodpressure treatment is no better. Numerous studies have shown that patients frequently drop out oftreatment for hypertension. Furthermore, even when patients stay in treatment, they often take theirmedications in a way quite dissimilar from that prescribed. Identifying noncompliant patients isimportant but not always easy to accomplish. Pill counts, the "gold standard," are seldom practical inroutine clinical practice. Assessing compliance by its biological effect is compromised by physiologicaldiversity among patients. Assessing compliance from patient self-reports is limited hi its accuracy butis more useful than many researchers and clinicians appreciate. Compliance behavior is affected bymany factors. Complexity of medical regimen has some effect; the presence of drug side effects hassurprisingly little. Contrary to what many clinicians think, increased age is often accompanied byincreased medication compliance. Features of the doctor-patient relationship likely have an importanteffect on patient compliance, though our knowledge of these factors is still limited.(Hypertension 11 [Suppl II]: II-61-H-64, 1988)

KEY WORDS • ambulatory medicine • patient education • clinical assessment

COMPLIANCE is a complex topic that becomesno simpler when viewed only in the frame-work of blood pressure control. Though nec-

essarily limited, the following discussion highlightsseveral key aspects of compliance with hypertensiontherapy.

How Much Compliance?A first issue, obviously, is the actual magnitude of

patient compliance with hypertension therapy. As not-ed in Dr. German's review, compliance behavior ingeneral is often suboptimal. Unfortunately, compli-ance with hypertension therapy is no better. For exam-ple, Engelland et al.' reviewed records from a medicalpractice in an affluent section of Manhattan and report-ed that 51 % of a sample of hypertensive patients failedto reappear during the year following their initial visit.Similarly, Gillum et al. ,2 reviewing 2 years of recordsfrom a Boston teaching hospital's outpatient clinic,reported that 50% of newly and 30% of previouslydiagnosed hypertensive patients failed to return to theclinic within 6 months of their last scheduled appoint-ment. Johnson et al.,3 in a follow-up study of a com-munity population volunteering for a shopping-centerblood pressure screening program, found that 21% of

From the Division of Internal Medicine and Health ServicesResearch and Development Center, The Johns Hopkins University,Baltimore, Maryland; and Division of Geriatric Medicine, George-town University, Washington, D.C.

Address for reprints: Dr. Lawrence Elliot Klein, Suite 331, 3301New Mexico Avenue, N.W., Washington, D.C. 20016.

patients had stopped treatment within 1 year. Like-wise, 19% of patients in The Australian Mild Hyper-tension Trial refused to continue with therapy withinthe first 2 years of the program, often discontinuingduring the first few months.4

Thus, these and other studies5"7 confirm that manypatients drop out of treatment for hypertension. This isnot the only bad news, however, because it turns outthat compliance among those staying in treatment interms of how they actually consume their medicationsis poor also.

For example, Johnson et al.,3 studying their shop-ping-center screening program, reported that only 60%of the patients remaining in treatment claimed to befully compliant with their drug regimen. Likewise,Sackett et al.,8 describing their landmark study ofCanadian steelworkers enrolled in a randomized trialof antihypertensive treatment, reported that only about50% of patients complied with at least 80% of theirprescribed drug regimen. Data from the MarylandStatewide Hypertension Survey9 revealed that amongpatients who were actively under treatment (defined ashaving medication bottles at home) when interviewed,only 60% scored high on a four-point compliancescale.

Clinicians and Their Assessment ofPatient Compliance

These and other studies10"12 document that evenwhen patients remain under treatment, compliance intaking medication is often poor. However, when pa-tients remain in treatment, clinicians, at least theoreti-

11-61

at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from

n-62 ANTIHYPERTENSIVE DRUG EFFECTS SUPPL H HYPERTENSION, VOL 11, No 3, MARCH 1988

cally, can target compliance-promoting measures atthose who fail to take their medications as prescribed.However, this assumes that clinicians can identifynoncompliant patients.

Although there is little written specifically about theability of clinicians to recognize patients' noncom-pliance with antihypertensive therapy, several studiesdo show clinicians in general to be inaccurate in thisregard. For example, Muslin and Appel13 and Caronand Roth14 have reported that house staff physiciansare no better than 50% accurate in identifying whetherpatients are complying with prescribed medicationregimens. Interestingly, Davis13 has provided datasuggesting that senior physicians can be even moreinaccurate than junior clinicians in recognizingnoncompliance.

Methods for Assessing Patient ComplianceAs these and other studies16' n demonstrate, personal

assessments by clinicians of patients' compliance areof limited accuracy. Thus, the question is raised as towhether there exist other more accurate means of as-sessing compliance.

It is easy, of course, to assess dropping out of treat-ment altogether since the patient is no longer there.However, much harder is the task of assessing howmuch medication a patient actually consumes relativeto how much was prescribed.

Traditionally, the "gold standard" in assessingmedication compliance has been pill counts. Thoughthis can be an effective means of assessing compliancein carefully controlled prospective trials, it is of limitedutility in tine world of everyday practice. Obviously,the typical clinician cannot drop in on his patients athome to count pills. On the other hand, asking patientsto bring their pill bottles to the office is not an accept-able alternative since pill count accuracy is easily com-promised. For example, patients may forget to bringtheir pill bottles to the office. Or, if they bring pillbottles to the office, they may forget having dividedtheir medication among several bottles for conve-nience and bring in only some of the bottles. Of, if theydo bring in all bottles, they may first purposely emptyout some of the pills so as to not "disappoint" theirdoctor.

A second means of evaluating medication compli-ance used at times by both researchers and practicingclinicians has been the measurement of drug levels andthe assessment of biological effects of treatment. How-ever, both of these means of assessing compliance arelimited. For one thing, drug levels of antihypertensivemedications are not routinely available and, even ifthey were, would not likely be very useful because ofphysiological variation among individuals in theirclearance of these drugs. Similarly, assessing drugcompliance by looking at biological effects is limitedby physiological variability among patients. For exam-ple, although thiazides tend to depress serum potas-sium levels and raise serum uric acid levels, the magni-tude of these effects if too variable for serum levels tobe useful in assessing medication compliance. Blood

pressure control itself is neither a sensitive nor specificmeasure of compliance since it is possible for a patientto be compliant but hypertensive (if his medicationregimen is inadequate) or, conversely, to be noncom-pliant but normotensive (if he did not actually need allthe medication prescribed).

Self-Report and the Recognition of NoncomplianceAsking patients whether they are compliant with

their antihypertensive regimen is another way to assesscompliance behavior, though it is a method so simplis-tic that it has generated little enthusiasm among re-searchers. However, in light of the fact that it is clini-cally easy to do and since other means of assessingcompliance have shown limited utility, this means ofevaluating compliance has recently received increasedattention.

For example, Levine et al.18 reported on a four-itemcompliance scale used in studying three educationalinterventions aimed at hypertensive ambulatory pa-tients. They found a high degree of consistency inresponses among the four items and adequate predic-tive validity with blood pressure control. Furthermore,Morisky et al.,19 studying these patients at year two,reported 75% of those scoring high on the scale to haveadequate blood pressure control compared with 47%scoring low (p<0.01).

Similarly, Hershey et al.,20 studying outpatients at-tending hypertension sessions at The Hospital of theUniversity of Pennsylvania, reported that asking pa-tients whether they took their high blood pressure pills"always, most of the time, some of the time, seldom,or never" was a useful way to assess compliance. An-swers proved to be significantly (p<0.02) related tohypertension control with 75% of those reporting theyalways complied, being normotensive, compared with53% of those reporting they did not. Moreover, pa-tients who reported missing on average more than threepills per week were far less likely (p<0.01) to benormotensive than were those who missed none.

Haynes et al.21 compared various means of assessingmedication compliance in the control of hypertensionamong steelworkers involved in a worksite hyperten-sion project. They found that an effective means ofassessing medication compliance, even better thanmeasuring serum uric acid levels, serum potassiumlevels, urine chlorthalidone levels, or urine hydro-chlorothiazide levels, was to elicit patients' own as-sessments of their compliance. Haynes et al. obtainedthese assessments using an interview instrument care-fully crafted to be nonjudgmental. This instrumentconsisted of an initial section introducing the issue ofcompliance: "People often have difficulty taking theirpills for one reason or another and we are interested infinding out any problems that occur so that we canunderstand them better." Then the patients were askedwhether they ever missed their pills (and, if so, to statethe average number of tablets missed per day, week,and month). Using this approach, Haynes et al. report-ed a 75% rate of agreement between self-reported com-pliance and pill counts. A self-report of high compli-

at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from

COMPLIANCE AND BLOOD PRESSURE/ATem n-63

ance (90% or greater compliance) was found to be 70%predictive of a pill count indicating 80 to 100% com-pliance. Conversely, a self-report of low compliance(less than 90% compliance) was 91% predictive of 0 to79% compliance. These figures, though admittedlyless than perfect, do suggest that questioning patientsabout their compliance can give useful information.

Predicting NoncomplianceSuccessful recognition of noncompliance after it has

occurred (whether through pill counts, self-report,etc.) is valuable bdth so that researchers can take com-pliance into consideration in assessing new therapiesand so that clinicians can understand poor medicaloutcomes. However, even more useful would be theability to predict noncompliance ahead of time, there-by giving researchers and clinicians an opportunity toattempt to prevent it in the first place.

For example, it has been suggested that we coulddevelop a profile of the type of individual who is likelyto be noncompliant by considering some combinationof characteristics of the patient, the treating physician,and the medication regimen itself.22 If we were silc-cessful in this regard, then we could concentrate ontargeting compliance promotion programs at those in-dividuals most likely to be noncompliant. Additional-ly, if one could identify specific aspects of the doctor-patient interaction or specific features of treatmentitself that correlate with noncompliance, then we couldpotentially work to modify these elements and, there-by, increase compliance.

Age and ComplianceOne potential determinant of compliance behavior

that has been widely discussed in the medical literatureover the years has been that of patient age. In particu-lar, much has been written about how there is an espe-cially great problem of medication noncomplianceamong the elderly. Interestingly, however, when onebegins to look closely at these pronouncements, it be-comes clear that much of what is referenced is eitheranecdotal or limited in its applicability.23 In fact, care-ful review of the literature suggests that elderly pa-tients on average try especially hard to be compliant,despite some clinicians' perceptions to the contrary.This seems true for medication use in general as well asfor antihypertensive medications specifically.

For example, Haines and Ward,24 summarizing datafrom two national surveys sponsored by the NationalHeart, Lung, and Blood Institute, report that "Youngerhypertensives were more likely to discontinue theirmedication than were older ones." Similarly, datafrom the most recent of the Maryland Statewide Hy-pertension Surveys demonstrate that among those pa-tients who reported in the past being prescribed antihy-pertensive medications, the most likely (p<0.001) tobe actively taking medications at the time of the surveywere the elderly.23 Additionally, among patients cur-rently taking antihypertensive drugs, the elderly wereespecially likely (by the four-item compliance scale

described previously) to report high compliance

Health Knowledge and ComplianceHealth knowledge has often been theorized to be a

potentially important determinant of compliance be-havior, and consequently, various attempts at promot-ing compliance through educational maneuvers havebeen carried out over the years. Though such ap-proaches would seem reasonable, a review of the lit-erature actually reveals relatively little support for astrong relationship between health knowledge andmedication compliance. For example, a study of medi-cation compliance among an ambulatory clinic popula-tion at a major teaching hospital revealed no correla-tion (r = 0.00) between medication compliance andknowledge of medication purpose.2* Similarly, at-tempts by several researchers to improve compliancewith antihypertensive therapy by imparting healthknowledge have tended to be unsuccessful. For in-stance, Sackett et al.8 randomly assigned patients to aneducational program designed to teach them about hy-pertension, including its effects upon target organs,health, and life expectancy, the benefits of antihyper-tensive therapy, the need for compliance with medica-tions, and some simple reminders for pill taking.Though this program was effective in increasingknowledge of the group exposed to the educationalintervention (at 6 months, 85% of patients in the inter-vention group had mastered the health informationcompared with only 18% of the patients in the controlgroup), this learning did not increase compliance. Fur-thermore, individual compliance rates bore no relationto knowledge about hypertension either at entry intothe study (r = -0.03) or at 6 months (r = 0.08).

In a similar way, Levine et al.9 applied an education-al intervention as part of a three-part compliance pro-motion program. Although participation in the pro-gram overall did lead to improved blood pressurecontrol, exposure by itself to the individualized 15-minute counseling session in which the practitioners'instructions were explained and reinforced did riot re-sult in a significant improvement in blood pressure.

Compliance Versus Complexity and Side Effectsof Therapy

The relationship between complexity of medicationregimen and compliance (both in general and in hyper-tension therapy specifically) has been much discussedin the medical literature. However, it is not nearly asclear a relationship as is often assumed. Overall, it istrue that complexity of medication regimen seems tobe inversely related to medication compliance. How-ever, it also appears that it is primarily not how manytimes a day medications are prescribed but, rather,how many different types of medications are pre-scribed that is important.27

As for the relationship between compliance and sideeffects of medications, this also is a complex issue. Itwould seem reasonable to expect to find medicationcompliance inversely related to side effects. Certainly,

at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from

H-64 ANTfflYPERTENSIVE DRUG EFFECTS SUPPL II HYPERTENSION, VOL 11, No 3, MARCH 1988

one would not expect patients to be more compliantwith their medications because they experience sideeffects. However, although there is some data support-ing such an inverse relationship, a review of data over-all actually suggests that this inverse relationship isrelatively weak (and, in fact, may actually be of limit-ed clinical importance). For example, a study of 299randomly selected outpatients from a teaching hospitalmedical practice revealed very few patients stoppingtheir medications in general because of side effects,which was similar to what was found in the MarylandStatewide Hypertension Survey that concentrated onantihypertensive medications.28

Compliance and the Doctor-Patient RelationshipAs the preceding brief discussion has attempted to

demonstrate, the issue of compliance and blood pres-sure control is complex and incompletely understooddespite the publication of numerous papers exploringmedical compliance in general. Dr. German in herreview has highlighted aspects of the doctor-patientrelationship affecting compliance in general, stressingher belief in their importance. Similarly, it would seemprobable that these factors affect compliance withblood pressure therapy specifically. Unfortunately,relatively few studies have addressed this particularissue, thereby limiting our ability to intercede in thedoctor-patient relationship to maximize compliance.

ConclusionThe preceding discussion has summarized our

knowledge of compliance in the treatment of hyperten-sion. In part, the data are discouraging, revealing largenumbers of patients dropping out of treatment as wellas low compliance among patients staying in treat-ment. On the other hand, we have seen that clinicianscan use patient self-reports to identify a sizable portionof their noncompliant patients. Thus, there exists asimple and inexpensive means of targeting patients forwhom compliance-promoting programs are indicated.

The fraction of our population consisting of individ-uals 65 years of age and older continues to increase. Inthis light, it is important to realize that the elderly onaverage represent a patient group especially likely tostrive for medication compliance, even in the face ofdrug side effects.

Finally, it is important to realize that there is stillmuch to be learned about the doctor-patient relation-ship and its effects on hypertension treatment. Hope-fully, studies exploring these relationships will beforthcoming.

References1. Engelland AL, Alderman MH, Powell HB. Blood pressure

control in private practice: a case report. Am J Public Health1979;69:25-29

2. Gillum RF, Neutra RR, Stason WB, Solomon HS. Determi-nants of dropout rate among hypertensive patients in an urbanclinic. J Community Health 1979^:94-100

3. Johnson AL, Taylor DW, Sackett DL, Dunnett CW, ShimizuAG. Self blood pressure recording — an aid to blood pressurecontrol? Ann R Coll Phys Surg Can 1977;10(l):32-37

4. Abernathy JD. The problem of non-compliance in long-termantihypertensive therapy. Drugs 1976; 11 (suppl l):86-90

5. Wilber JA, Barrow JG. Hypertension — a community prob-lem. Am J Med 1972^2:653-663

6. Strogatz DS, Earp JAL. The determinants of dropping out ofcare among hypertensive patients receiving a behavioral inter-vention. Med Care 1983;21(suppl 10):970-980

7. Degoulet P, Menard J, Vu HA, et al. Factors predictive ofattendance at clinic and blood pressure control in hypertensivepatients. Br Med J 1983^287:88-93

8. Sackett DL, Haynes RB, Gibson ES, et al. Randomised clini-cal trial of strategies for improving medication compliance inprimary hypertension. Lancet 1975;l:1205-1207

9. Levine DM, Bone LR, Steinwachs DM, Parry RE, MoriskyDM, Sadler J. The physician's role in improving patient out-come in high blood pressure control. Md State Med J 1983;32:291-293

10. McKenney JM, Slining JM, Henderson HR, Devins D, BarrM. The effect of clinical pharmacy services on patients withessential hypertension. Circulation 1973;48:1104—1 111

11. Rudnick KV, Sackett DL, Hirst S, Holmes C. Hypertension ina family practice. Can Med Assoc J 1977;117:492-497

12. Inui TS, Carter WB, Pecoraro RE. Screening for noncom-pliancc among patients with hypertension. Med Care 1981;19:1061-1064

13. Muslin Al, Appel FA. Diagnosing patient non-compliance.Arch Intern Med 1977;137:318-321

14. Caron HS, Roth JP. Patients' co-operation with a medicalregime. JAMA 1968^203:922-926

15. Davis MS. Variations in patients' compliance with doctors'orders. J Med Educ 1966;41:1037-1048

16. Dixon WM, Stradling P, Wootton ID. Outpatient PAS therapy.Lancet 19572:871-872

17. Moulding T, Onstad GD, Sbarbaro JA. Supervision of outpa-tient drug therapy. Ann Intern Med 1970;73:559-564

18. Levine DM, Green LW, Deeds SG, Chwalow J, Russell RP,Finlay J. Health education for hypertensive patients. JAMA1979;241:1700-1703

19. Morisky DE, Green LW, Levine DM. Concurrent and predic-tive validity of a self-reported measure of medication adher-ence. Med Care 1986;24:67-74

20. Hershey JC, Morton BG, Davis JB, Reichgott MJ. Patientcompliance with antihypertensive medication. Am J PublicHealth 1980;70:1081-1089

21. Haynes RB, Taylor D, Sackett DL, Gibson ES, Bernholz CD,Mukherjee J. Can simple clinical measurements detect patientnoncompliance? Hypertension 1980;2:757-764

22. Nelson EC, Stason WB, Neutra RR, Solomon HS. Identifica-tion of the noncompliant hypertensive. Preventative Med1980;9:504-517

23. Klein L, German P, Levine D. Adverse drug reactions amongthe elderly. J Am Geriatr Soc 1981;29:525-53O

24. Haines CM, Ward GW. Recent trends in public knowledge,attitudes, and reported behavioT with respect to high bloodpressure. Public Health Reports 1981;96:514-522

25. Klein L, Levine D, German P, Entwisle G, Southard J. Medi-cal care of the elderly [Abstract]. Clin Res 1984;32(suppl2):297A

26. Klein LE, German PS, McPhee SJ, Smith CR, Levine DM.Aging and its relationship to health knowledge and medicationcompliance. Gerontologist 1982^2:384-387

27. Haynes RB. Determinants of compliance. In: RB Haynes, DWTaylor, DL Sackett, eds. Compliance in health care. Balti-more: The Johns Hopkins University Press, 1979:49-62

28. Klein LE, German PS, Levine DM, Feroli ER, Ardery J.Medication problems among outpatients: a study with empha-sis on the elderly. Arch Intern Med 1984;144:1185-1188

at Queen Mary, University of London on July 17, 2014http://hyper.ahajournals.org/Downloaded from