impaired cognitive function and compliance with antihypertensive drugs in elderly: the rotterdam...

PHARMACOEPIDEMIOLOGY AND DRUG UTILIZATION

Impaired cognitive tinction and compliance with antihypertensive drugs in elderly: The Rotterdam Study

Bacibground: To our knowledge, there are no epidemiologic studies on the association between cognitive impairment and noncompliance with antihypertensive therapy. We studied compliance with antihypertensive treatment in elderly patients with cognitive impairment. Methods: The Rotterdam Study is a prospective community-based cohort study of 7983 residents 255 years old. We studied 1979 participants in the study who had 2 consecutive Mini Mental State Examination (MMSE) assessments in the period from 1991 to 1996, who did not have dementia at baseline, and who had received 3 or more consecutive antihypertensive prescriptions for at least 6 months. We compared persons with MMSE scores 525 on both assessments to persons with MMSE scores >25 on both occasions. Compliance was estimated by dividing the number of days the subjects took antihypertensive drugs by the follow-up period in days, and it was expressed as a ratio between 0 and 1. We defined patients as compliant if they had a compliance ratio 20.80 and as noncompliant if they had a compliance ratio 10.50 during the study period. Results: We followed-up on 1573 patients (mean age, 68 years) during an average period of 1609 days. The risk of noncompliance in cognitively impaired elderly subjects was 2.0 (95% confidence interval, 1.4 to 2.8) after adjustment for age, sex, education, income, living situation, and smoking. Stratification by living situation showed that the risk increase predominantly occurred in those who lived alone (odds ratio, 2.9; 95% confidence interval, 1.2 to 7.5). Conclusions: Cognitive function is an independent predictor of compliance with antihypertensive drugs in elderly patients who are living alone. (Clin Pharmacol Ther 2001;70:561-6.)

Maribel Salas, DSc, Bas A. In’t Veld, PhD, Paul D. van der Linden, PhD, Albert Hofman, PhD, Monique Breteler, PhD, and Bruno H. Wicker, PhD Rotterdawz and The Hague, The Netherlands, and Mexico City, Mexico

Noncompliance is a significant threat to adequate irrespective of comorbidity or prognosis.2 It has been antihypertensive management.1 At least one third of estimated that 30% to 50% of patients with hyperten- patients comply poorly with prescribed drug regimens, sion stop taking their prescriptions within 1 year after

From the Department of Epidemiology & Biostatistics, Erasmus Medical Center, Rotterdam; Consejo National de Ciencia y Tec- nologia, Mexico City; and the Drug Safety Unit, Inspectorate for Health Care, The Hague.

Supported by the Inspectorate for Health Care, the NESTOR stimu- lation program for geriatric research in the Netherlands, the Netherlands Organization for Scientific Research (NWO), and the municipality of Rotterdam.

Received for publication June 12, 2001; accepted Aug 27, 2001.

Reprint requests: Bruno H. Stricker, MD, PhD, Pharmacoepidemi- ology Unit, Department of Epidemiology and Biostatistics, Eras- mus University Medical School, PO Box 1738, 3000 Rotterdam, The Netherlands.

Copyright 0 2001 by the American Society for Clinical Pharmacol- ogy & Therapeutics.

0009-9236/2001/$35.00 + 0 13/l/119812

doi:10.1067/mcp.2001.119812

561

562 Salas et al CLINICAL PHARMACOLOGY & THERAI’EUTI(:S

DECEMBER 2001

diagnosis.3 Among those who remain on therapy, a substantial proportion will regularly miss medication doses.4 Noncompliance with antihypertensive treatment has been associated with an increased number of visits to physicians, hospital admissions, increased length of hospital stays, and increased health services expendi- tures, as well as with reduced access to medical care.5 Noncompliance has also been associated with a low socioeconomic status of the patient and poor under- standing of hypertension and the therapeutic role of pharmacotherapy. Other risk factors that may play a role are adverse effects, direct costs of antihypertensive agents for the patient, the administration schedule,6 cig- arette smoking, age, sex, excessive alcohol intake, education, and marital status.7 Noncompliance with drug therapy is particularly important in elderly persons* and has been attributed to causes such as transportation difficulties, multiple health problems, changes of labeling and container-type, and social isolation.g-11 Cognitive impairment in elderly patients may affect ability to interpret written or verbal instructions.10 In patients with dementia, it is likely that their caretakers monitor their use of medications and obtain their drugs from the pharmacy. In patients with mild cognitive decline, however, noncompliance may remain unnoticed. We therefore performed a prospective population-based cohort study in elderly patients receiving antihypertensive treatment to investigate the effect of mild cognitive impairment on compliance.

MATERIAL AND METHODS Setting. The Rotterdam Study is a community-based

prospective cohort study that started in 1990 on the occurrence and determinants of cardiovascular, neurologic, locomotor, and ocular disease in an elderly population. To this end, all 10,275 residents ~55 years old in the Ommoord area of Rotterdam were invited to par- ticipate. Of these, 7983 (78%) persons agreed to par- ticipate and gave written informed consent. The design of this study has been described in detail elsewhere. l2 As part of an extensive neurologic study program, the Mini Mental State Examination (MMSE) was performed during cross-sectional surveys from 1991 to 1993 and from 1994 to 1996. Subjects were screened for dementia according to a 3-step protocol both at baseline and at the follow-up examinations.13 First, subjects were screened for dementia with a combined MMSE assessment and the Geriatric Mental State schedule (GMS-A, organic level). Second, those who scored 25 or below on the MMSE or who scored 1 or more on the Geriatric Mental State schedule were selected for further diagnostic evaluation and were sub-

sequently examined by a physician with the CAMDEX diagnostic interview,13a which includes an informant interview. Third, subjects suspected to have dementia were examined by a neurologist and a neuropsycholo- gist, and they had a brain magnetic resonance imaging. The pharmacy data of approximately 99% of this population has been available in automated form since Jan- uary 1, 1991. For every prescription, the product name, Anatomical Therapeutic Chemical (ATC) code, date of delivery, total number of capsules of tablets, prescribed daily number, and pharmaceutical form are registered.

Study population and design. The study population consisted of all patients with an MMSE assessment from 1991 to 1993 and from 1994 to 1996 who had filled at least 3 consecutive prescriptions for an antihypertensive drug at the pharmacy, with a cumulative duration of at least 180 days in the study period between January 1, 1991, and December 31, 1997. Patients were monitored from the first prescription in the study period until one of the following events: death, removal to another address outside Ommoord, or end of the study period, whichever came first. We excluded patients with dementia at baseline because we were interested in the association between mild cognitive decline and compliance with antihypertensive treatment. Furthermore, persons were excluded if they were living in a nursing home during the study period or if they had a history of stroke or a psychiatric disorder. Finally, we excluded all persons with MMSE scores >25 at the first assessment in the 1990 to 1993 period who had an MMSE score 525 at the second assessment in the 1994 to 1996 period, as well as persons with MMSE scores 525 at the first assessment who had an MMSE score >25 at the second assessment. The study population therefore consisted of persons with MMSE scores 525 at both assessments (but without dementia) and persons with MMSE scores >25 at both assessments.

Definition of compliance. Exposure to all antihypertensive drugs according to the ATC coding system of the World Health Organization was assessed. We included all prescriptions with the ATC codes C02, C03, and CO7 to C09. For every patient, a medication profile was constructed to calculate exposure to antihypertensive drugs on a day-by-day basis. We calculated duration of use by dividing the total number of filled tablets or capsules per prescription by the prescribed daily number. The resulting legend duration (prescription length) reflected the recommended exposure period per prescription. If 2 prescriptions for different antihypertensive drugs overlapped, the overlap- ping period was counted once. The compliance ratio

CLINICAL P HARMACOLOGY & THERAPEUTICS VOLUME 70, NUMBER 6 S&s et al 563

Table I. Characteristics of patients (N = 1573)

Variable Total MMSE

(N = 1573; 100%) MMSE 525 (n = 53; 3%)

MMSE ~25 (n = 1520; 97%) P value

Mean age + SD (y) Sex

Women Men

Follow-up (d) Compliance ratio First MMSE (mean) Second MMSE (mean) AMMSEt Incident dementia during study period Highest education

Primary school Secondary school College or University Data missing

Income (%) Lowest income Medium income Highest income Data missing

Living situation Alone With partner or relative Data missing

Total No. prescriptions No. antihypertensive prescriptions

Diuretics P-Blockers Calcium antagonists ACE inhibitors

Smoking Current Former Never Data missing

Alcohol (>50 gm/d) No. hospital admissions

No. deaths during study period

68 + 8

916 (58%) 44 (83%) 657 (42%) 9 (17%)

1609 f 596 1769 f 509 0.78 + 0.3 0.71 f 0.3

28 + 1.4 24 f 1.2 28 + 1.7 22 f 2.7

-0.10 + 1.5 -1.50 f 2.6 10 (0.6%) 8 (15.0%)

698 (44%) 656 (42%) 173 (11%) 46 (3%)

42 (79%) 7 (13%) 1 (2%) 3 (6%)

503 (32%) 805 (51%)

82 (5%) 183 (12%)

421 (27%) 1126 (72%)

26 (1%) 121 + 91 29 + 22

8+ 11 10 + 12 5+9 6+ 10

300 (19.0%) 726 (46.0%) 538 (34.0%)

9 (1.0%) 34 (2.2%) 1.2 + 2.1

133 (8%)

77 + 8

38 (72%) 11 (21%) 0 (0%) 4 (7%)

30 (57%) 22 (41%)

1 (2%) 172 f 104 34 + 24 13 + 13 7+ 10 6+ 10 6+ 12

4 (7.5%) 13 (24.5%) 34 (64.2%)

2 (3.8%) 0 (0%)

1.3 + 1.5 7 (13%)

68 + 7

872 (57%) 648 (43%)

1604 + 598 0.78 + 0.3

28 + 1.2 28 + 1.2

-0.06 + 1.4 2 (0.1%)

656 (43%) 649 (42%) 172 (11%) 43 (4%)

465 (31%) 794 (52%)

82 (5%) 179 (12%)

391 (26%) 1104 (73%)

25 (1%) 119+90 29 f. 22

8r 11 lo* 11 5+9 5 + 10

296 (19.5%) 713 (46.9%) 504 (33.2%)

7 (0.4%) 34 (2.2%) 1.2 + 2.1

126 (8%)

<.001*

<.001*

.024* ,093

<.001* <.001* <.001* <.001*

<.001*

<.001*

<.001*

<.OOl” .16 .02* .03* .23 .70

<.001*

1 .oo .62 .20

MMSE, Mini Mental State Examination; ACE, angiotensinconverting enzyme. *Significant values. tSecond MMSE minus first MMSE.

was calculated as the sum of all days on which the patient was considered to be exposed to antihypertensive drugs (and was thus identical to the sum of all legend durations corrected for overlap) divided by the total period between the day of the first prescription and the end of the follow-up period. The resulting compliance ratio therefore varied between 0 and 1. We defined persons as compliant if they had a compliance ratio 2 0.80 and as noncompliant if they had a compliance ratio ~0.50 during the study period.

Statistical analysis. Differences in proportions between groups were tested with ~2 statistics or with a

Fisher exact test if expected cell counts were below 5. Comparisons of continuous variables between groups were performed with a t test or with a Mann-Whitney test when nonnormally distributed. All tests were 2- sided, with rejection of the null hypothesis at P < .05. Comparisons between patients with cognitive impairment versus patients with normal cognitive function were expressed as relative risks with 95% confidence limits. In a separate analysis, we studied the role of risk factors for noncompliance with logistic regression analysis, with the odds of noncompliant (compliance ratio < 50) and compliant (compliance ratio 2 0.80) as

564 Salas et al CLINICAL PHARMACOLOGY & THER4PEUTICS

DEC~MREK 2001

Table II. Risk of noncompliance in persons with cognitive impairment stratified by living situation*

Noncompliance (CR 50.50)

Compliance (CR 20.80) Odds ratio

95% Conjidence interval

Lives with partner (n = 191) (n = 821) MMSE I 25 4 (2.1%) 17 (2.1%) Male sex 99 (51.8%) 431 (52.5%) Average age (y) 61 67 Total No. prescriptions 89 116 Education 83 (44.6%) 329 (41.9%) Income 31 (18.8%) 174 (24.3%) Smoking 47 (24.7%) 245 (30.1%)

Lives alone (n = 81) MMSE 5 25 10 (12.3%) Male sex 12 (14.8%) Average age (y) 71 Total No. prescriptions 125 Education 43 (53.1%) Income 52 (64.2%) Smoking 44 (54.3%)

(n = 267) 15 (5.6%) 40 (15.0%) 71

150 134 (50.2%) 162 (60.7%) 116 (43.3%)

1.3 0.4-4.2 1.2 0.8-1.8 1.02 0.99-l .06 0.99 0.99- 1 .o 1.1 0.9-1.5 0.7 0.5-0.9t 0.8 0.6-1.1

2.9t 1.2-7.5t 0.7 0.3-1.7 0.99 0.96- 1.03 0.99 0.99-1.01 1.1 0.7-1.8 1.1 0.6-2.1 1.1 0.7-1.7

CR, Compliance ratio; MMSE, Mini Mental State Examination. *Adjusted for sex, age, total number of prescriptions, education, income, and smoking. tSigniticant values.

the outcome. Other variables added to the model were age, sex, education, income, smoking, number of total prescriptions, number of hospital admissions, number of different antihypertensive drugs, number of different drugs, and number of visits to the pharmacy, as well as whether the individual lived alone or with a partner. The final forward stepwise regression model was built by considering statistical significance (P < .05), with use of test score for entering variables and likelihood ratio for removing variables. Goodness of fit was analyzed by comparison of different models with use of the likelihood ratio test. Further analyses were performed with alcohol consumption and the living situation as potential effect modifiers.

RESULTS Of the 7983 persons in the Rotterdam study, 7528

were screened for dementia at baseline, and 7046 of those had no dementia. Of those 7046 patients, there were both pharmacy data and a second MMSE assessment in the 1994 to 1996 period for 5257 persons. After those who lived in a nursing home, had received fewer than 3 prescriptions, or had been exposed to fewer than 180 days of antihypertensive treatment and who had the first MMSE assessment before January 1, 1991 (to guarantee the availability of a medication history) were excluded, 1979 persons remained. We subsequently excluded 118 patients with stroke and 111 with a history of psychiatric disorders. Furthermore, we excluded 80 patients who had an MMSE score >25 in the first

assessment but an MMSE score 2 25 at the second assessment, as well as 97 persons who had an MMSE score I 25 at the first assessment but an MMSE score >25 at the second assessment. The final sample consisted of 1573 patients, of whom 58% were women (Table I). The patients had a mean age of 68 years, an MMSE score of 28, and an average of 121 prescriptions and 29 antihypertensive prescriptions during the study period. The majority had salaries in the lowest income category, they often lived alone, and almost half of them were former smokers. There was no statistically significant difference in the proportion of persons with a high alcohol intake (>50 mg/day) and in the number of hospital admissions during the study period. Persons with low MMSE scores had a slightly but significantly longer follow-up period, but the overall compliance ratio was not significantly different from patients with high MMSE scores. No significant differences were observed in the number of deaths after the second MMSE assessment between the 2 groups, and both groups had received approximately the same number of prescriptions for antihypertensive drugs. The most frequently prescribed antihypertensive drugs were diuretics and P-blockers. There were more incident cases of dementia among the group with cognitive impairment.

In Table II, the adjusted risk estimates are given for noncompliance in persons with cognitive impairment. Overall, the risk of noncompliance was twice as high in persons with impaired cognitive function than the

CLINICAL PHARMACOLOGY & THERAPEUTICS VOLUME 70. NUMBER 6 Salas et al 565

risk in persons with normal cognitive function (relative risk, 2.0; 95% confidence interval [CI], 1.4 to 2.8). Exclusion of patients with incident dementia during the study period increased the adjusted risk estimate from 2.00 to 2.42 (95% CI, 1.17 to 5.04). Stratification by living situation showed effect modification. In those who lived with a partner, relative, or spouse, there was no increased risk of noncompliance, whereas the risk increase in persons with cognitive impairment who lived alone was 2.9 (95% CI, 1.2 to 7.5). Other potential confounders were analyzed in the study, such as number of total prescriptions, number of hospital admissions, number of different antihypertensive drugs, number of different drugs, and number of visits to the pharmacy as a proxy for health care consumption. When each potential confounder was included on an individual basis, the direction and degree of association between cognitive function and compliance did not change. To analyze the influence of the cutoff point for compliance, we did an extra analysis in which noncompliance was defined as a compliance ratio ~0.80. This lowered the adjusted relative risk of noncompliance in elderly patients with cognitive impairment to 1.4 1 (95% CI, 1.03 to 1.92).

DISCUSSION In our study, we demonstrated that mild cognitive

impairment is a risk factor for noncompliance in those who live alone. Apparently, cognitive impairment is not an issue as long as people live together with a partner, relative, or spouse. We deliberately enrolled all persons with MMSE scores 2 25 but without dementia to include only patients with mild to moderate cognitive dysfunction. Most patients with dementia receive spe- cialized nursing care, which usually includes the monitoring of compliance. Indeed, exclusion of incident dementia increased our risk estimate. Our findings may be important because in many Western countries a large percentage of people is 2 65 years old. Because 9% of people older than 65 years of age have dementiats and because an even higher proportion have mild to moderate cognitive dysfunction, the consequences may be substantial. It has been suggested that impaired cognitive function in elderly patients may be a risk factor for noncompliance because these patients may have problems remembering to take their medications or remembering the adequate dose or because they do not know how to handle the number of medications that they have been prescribed. 14,1s One cross-sectional study in patients 270 years old suggested an association between compliance and cognitive impairment. l6 In another study on medication problems during hospitalization

and after discharge from a university-affiliated geriatric hospital to a patient’s home, 10% of 119 tested patients could not open a medication container because of poor vision, impaired cognitive function, or low manual dex- terity. Only 40% of them had a stable drug history during the 3-month study period.17

Our study has some limitations. First, we did not investigate whether patients really used their antihypertensive drugs. However, in a study in which we compared pharmacy data on cardiovascular drugs with the actual drug use, the concordance between pharmacy data and medication vials brought by the patient was very high. l8 Although regular refills suggested that our compliant patients used their antihypertensive drugs in a consistent way, we cannot exclude that some varied their dose on a daily basis at irregular intervals. Sec- ond, the living situation of our subjects was assessed at baseline and some of them may actually have been admitted to a nursing home during the study period. This potential bias, however, would tend to lead to an underestimation rather than to an overestimation of the risk estimate of the effect of cognitive impairment on compliance. Selection and information bias in our study are unlikely because this study was a population-based study and because data on drug use and disease were gathered prospectively and by assistants who were unaware of the research hypothesis.

The proportion of elderly patients is increasing, espe- cially in the Western world. Currently, there are approximately 32 million people in the United States who are 2 65 years old. They comprise 12% of the population, and up to 40% of them use medications. l9 Hyperten- sion is the most common and strongest risk factor for cardiovascular disease,20 and continuous drug treatment decreases both mortality and morbidity rates21 More- over, adequate antihypertensive treatment may decrease the risk of vascular dementia.** Poor control of hypertension is a major health problem that frequently man- ifests itself in late life.23 In addition, it has been shown that inadequate compliance with medications is a significant contributor to the cost of medical care in every therapeutic area. 15 Every recognized risk factor for noncompliance may therefore help to improve effective patient care. This finding is important because health care systems, physicians, caregivers, and the patient’s family should realize that patients with a mild degree of cognitive impairment are also at risk for noncompliance with antihypertensive therapy.

In conclusion, impaired cognitive function jeopar- dizes compliance with antihypertensive therapy in elderly patients who are living alone. Prescribers and pharmacists should be aware of this and consider offer-

566 Sahs et al CLINICAL PHARMACOLOGY & THERAPEUTICS

DECEMBER 2001

ing assistance or automated medication vials to decrease the risk of noncompliance.

References

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

Stason WB. Opportunities to improve the cost-effective- ness of treatment for hypertension. Hypertension 1991;18(3 suppl):I161-6. Urquhart J. Role of patient compliance in clinical phar- macokinetics. Clin Pharmacokinet 1994;27:202-15. Balazovjech I, Hinilica P Jr. Compliance with antihypertensive treatment in consultation rooms for hypertensive patients. J Hum Hypertens 1993;7:581-3. Nelson EC, Stason WB, Neutra RR, Solomon HS. Iden- tification of the noncompliant hypertensive patient. Prev Med 1980;9:504-17. McCombs JS, Nichol MB, Newman CM, Sclar DA. The costs of interrupting antihypertensive drug therapy in a Medicaid population. Med Care 1994;32:214-26. Skaer TL, Sclar DA, Robinson LM. Noncompliance with antihypertensive therapy. Pharmacoeconomics 1996;9: l-4. Flack JM, Novikov SV, Ferrario CM. Benefits of adherence to antihypertensive drug therapy. Eur Heart J 1995;17(suppl A):16-20. Ho PC, Triggs EJ. Drug therapy in the elderly. Aust NZ J Med 1984;14:179-90. Burns JM, Sneddon I, Love11 M, McLean A, Martin BJ. Elderly patients and their medication: a post-discharge follow-up study. Age Aging 1992;21: 178-81. Atkinson L, Gibson IJ, Andrews J. The difficulties of old people taking drugs. Age Aging 1977;6: 144-50. Black DM, Brand RJ, Greenlick M, Hughes G, Smith J. Compliance to treatment for hypertension in elderly patients: the SHEP pilot study. Systolic Hypertension in the Elderly Program. J Gerontol 1987;42:552-7.

12. Hofman A, Grobee DE, de Jong PT, van den Ouweland FA. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur J Epidemiol 1991; 7~403-22.

13. Ott A, Breteler MM, van Harskamp F, Claus JJ, van der Cammen TJ, Grobbee DE, et al. Prevalence of

Alzheimer’s disease and vascular dementia: association with education. The Rotterdam Study. BMJ 1995; 3 10:970-3.

14.

13a.Roth M, Tym E, Moutjoy CQ, Huppert FA, Hendric H, Verma S, et al. CAMDEX. A standardised instrument for the diagnosis of mental disorder in the elderly with spe- cial reference to the early detection of dementia. Br J Psy- chiatry 1986;149:698-709. Morel1 RW, Park DC, Kidder DP, Martin M. Adherence to antihypertensive medications across the life span. Gerontologist 1997;37:609-19. Cramer JA. Enhancing patient compliance in the elderly. Role of packaging aids and monitoring. Drugs Aging 1998;12:7-15.

15.

16.

17.

18.

19.

20.

O’Dogherty Carame P, Morel1 Sixto E, Molero Portoles D, Martinez Alvaro C, Martin Sanjuan C, Lopez Moran S. The relationship of the compliance with medical drug treatment to social and functional factors and to the cognitive status of the elderly population seeking care [in Spanish]. Aten Primaria 1997;20:324-8. Nikolaus T, Kruse W, Bach M, Specht-Leible N, Oster P, Schlierf G. Elderly patients’ problems with medication: an in-hospital and follow-up study. Eur J Clin Pharmacol 1996;49:255-9. Sjahid SI, van der Linden PD, Stricker BH. Agreement between the pharmacy medication history and patient interview for cardiovascular drugs: the Rotterdam elderly study. Br J Clin Pharmacol 1998;45:591-5. Sumner DD. Benzodiazepine-induced persisting amnes- tic disorder: are older adults at risk? Arch Psychiatr Nurs 1998;12):119-25. Kannel WB. Blood pressure as a cardiovascular risk factor. JAMA 1996;275:1571-6.

21. Moser M. Historical perspective on the management of hypertension. Am J Med 1986;80: 1 - 11.

22. In? Veld BA, Ruitenberg A, Hofman A, Stricker BH, Breteler MM. Antihypertensive drugs and incidence of dementia: the Rotterdam study. Neurobiol Aging 2001;22:407-12.

23. Morel1 RW, Park DC, Kidder DP, Martin M. Adherence to antihypertensive medications across the life span. Gerontologist 1997;37:609-19.

impaired cognitive function and compliance with antihypertensive drugs in elderly: the rotterdam...

Documents