1265 IMPROVEMENT OF MEDICATION COMPLIANCE IN UNCONTROLLED HYPERTENSION R. BRIAN HAYNES EDWARD S. GIBSON BRENDA C. HACKETT ARNOLD L.

DAVID L. SACKETT D. WAYNE TAYLOR ROBIN S. ROBERTS

JOHNSON

Department of Clinical Epidemiology and Biostatistics, McMaster University Medical Centre, and Dominion Foundries and Steel Limited, Hamilton, Ontario, Canada 38 hypertensive Canadian steelworkers who were neither compliant with medications nor at goal diastolic blood-pressure six months after starting treatment were allocated either to a control group or to an experimental group who were taught how to measure their own blood-pressures, asked to chart their home blood-pressures and pill taking, and taught how to tailor pill taking to their daily habits and rituals; these men were also seen fortnightly by a highschool graduate with no formal health professional training who reinforced the experimental manœuvres and rewarded improvements in compliance and bloodpressure. Six months later, average compliance had fallen by 1·5% in the control group but rose 21·3% in the experimental group. Blood-pressures fell in 17 of 20 experimental patients (to goal in 6) and in 10 of 18 control patients (to goal in 2).

Summary

Introduction THE potential benefit of vigorous medical treatment for hypertension often remains out of reach, in part because the patient does not comply with treatment. We believe that this non-compliance is a major barrier to the

PROF LUPIEN AND OTHERS: REFERENCES 1. Moutafis, C. D, 1971, 14, 247.

2.

Myant,

N.

B., Mancini, M., Oriente, P. Atherosclerosis

R. I., Fredrickson, D. S., Shulman, R. et al. Ann. intern. Med. 1972, 72, 267. 3. Buxtorf, J. C., Beaumont, V., Jacotot, B., Beaumont, J. L. Atherosclerosis 1974, 19,1. 5. Starzl, T. E., Chase, H. P., Putnam, C. W., Porter, K. A. Lancet, 1973, ii,

Levy,

940.

Thompson, G. R., Gotto, A. M. ibid. p. 35. Stein, E. A., Pettifor, J., Mieny, C., Heimann, K. W., Spitz, L. Bersohn, I., Saaron, I., Dinner, M. ibid. 1975, i, 832. 8. Starzl, T. E., Lee, I. Y., Porter, K. A., Putnam, C. W. Surgery, Gynec. Obstet. 1975, 140, 381. 9. Torsvik, H , Feldman, H. A., Fischer, J. E., Lees, R. S. Lancet, 1975, i, 601. 10. Thompson, G. R., Lowenthal, R., Myant, N. B. ibid. p. 1208. 11 DeGennes, J. L., Touraine, R., Maunand, B., Tuffert, J., Laudat, P. Bull Mem. Soc. med. Hop. Paris, 1967, 118, 1377. 12. Apstem, C. S., George, P. K., Zilversmit, D. B., Feldman, H. A., Lees, R. Clin. Res. 1974, 22, 459A. 13. Awad, J. A., Lupien, P.-J., Moorjani, S. Unpublished. 14. Moorjani, S., Lupien, P.-J., Awad, J. Unpublished. 15. Rush, R. L., Leon, L., Turrell, J. in Advances in Automated Analysis. 5th Technicon International Congress. p. 503. Miami, 1971. 16. Zilversmit, D, B., Davis A. K. J. Lab. clin. Med. 1950, 35, 155. 17. Havel, R. J., Eder, H. A., Bragdon, J. H. J. clin. Invest. 1955, 34, 1345. 18 Burstein, M., Samaille, J. C. r. hebd. Acad. Séanc. Acad. Sci., Paris, 1955, 241, 664. 19. Noble, R. P. J. Lipid Res. 1968, 9, 693. 20. Grundy, S. M., Ahrens, E. H., Jr., Salen, G., Schreibman, P. H., Nested, P. J. ibid. 1972, 13, 531. 21.Sodhi, H. S., Kudchodkar, B. J., Horlick, L. Atherosclerosis, 1973, 17, 1. 6. 7.

22 Srinivasan, S. R., Radhakrishnamurthy, B., Berenson, G. S. Archs Biochem. Biophys. 1975, 170, 334. 23. Glomset, J. A. J. Lipid Res. 1968, 9, 155. 24. Miller, G. J., Miller, N. E. Lancet, 1975, i, 16. 25. Rhoads, G. G., Gulbrandsen, C. L., Kagan, A. New Engl. J. Med. 1976, 294, 293.

effective control of hypertension’ and that our understanding of this phenomenon is primitive.2 In phase i of a trial of strategies for improving compliance we found3 that neither the mastery of facts about hypertension nor receiving care and follow-up at work in "company time" led to any improvement. We describe here the second phase of this trial in which the application of more behaviourally oriented strategies did lead to improvements in both compliance and blood-pressure control.

Methods These have been described in detail elsewhere.3 Briefly, the examination of 5400 men at Dominion Foundries and Steel Company (over 95% of a random two-thirds sample of male employees) yielded 245 who had high blood-pressure (when sitting quietly on three separate days, a standard series of fifth-

phase diastolic blood-pressures were ≥95 mm Hg), were free of remediable forms of hypertension, were taking no daily medications (70 men were on treatment and were therefore excluded), and had not been treated for hypertension in the preceding six months. In phase i of this trial,3 men were randomly allocated into a factorial design in order to test strategies affecting either the convenience of their follow-up care or their knowledge about hypertension and its treatment. Phase n of this trial was designed to see whether a further set of strategies could salvage patients who, although put on recognised antihypertensive drugs in phase I, were neither compliant (pill-counts less than 80%) nor at goal diastolic blood-pressure (fifth phase 90 mm Hg) in the sixth month of treatment. Following stratification by diastolic blood-pressure and compliance level at six months, plus their phase-i allocation, 39 such men were allocated by "minimisation" either to a control group or to an experimental group. Minimisation allows the simultaneous consideration of a large series of matching characteristics when allocating subjects to experimental and control groups, thereby minimising (using randomisation in the case of ties) between-group differences. The method is immune to experimenter bias and has been shown.to substantially outperform simple randomisation in reducing the imbalance between treatment groups that has troubled several earlier randomised trials.4 The experimental group received the following set of strategies: Home Self-measurement of Blood-pressure

experimental patient was loaned an aneroid sphygmoand stethoscope and instructed in its use. 18 men were loaned separate cuffs (’Nelkin’ sphygmomanometer model 204M; Nelkin Medical Products, Inc., Kansas City, Missouri) and stethoscopes; two men with upper extremity impairments were loaned devices in which the stethoscope head was incorporated into the cuff (’Arden’ sphygmomanometerstethoscope model HRI 8104-705201; Taylor Consumer Products Division, Sybron Corporation, Arden, North Carolina). Each

manometer

Home Blood-pressure and Medication

Charting

Each experimental patient was issued with daily pill and bloodpressure charts and asked to record (by number and by a dot on the chart) his fifth-phase blood-pressure each day, along with both the number of pills taken and the number of pills missed during the previous day. The treatment goal of a fifthphase blood-pressure below 90 mm Hg was clearly stated, and the background of the blood-pressure chart was red above, and, blue below, this value.

Tailoring Each experimental patient was interviewed to identify any daily habits or rituals. The resulting pattern was compared with the patient’s antihypertensive regimen and when the two coincided, agreement was sought to take the pills immediately

1266 before executing the habit or ritual; it was also suggested that medications be placed at the sites of these rituals. Increased Supervision and Reinforcement

Experimental patients were asked to report fortnightly for review of their daily pill and blood-pressure charts and for a check of their blood-pressure. At each review, if the bloodpressure check was either < 90 mm Hg or > 4 mm Hg below that observed at the sixth month, the patient was praised and received a$4 credit toward ownership of the home blood-pressure cuff and stethoscope. Praise was also received for periods of perfect compliance, and the reasons for every missed pill were sought in an effort to identify problems and solve them through further tailoring. If neither a blood-pressure fall nor perfect compliance had occurred, the patient was encouraged to do better over the next interval. All phase-n interventions were executed by a 28-year-old female programme coordinator, a high-school graduate with no health professional training; after instruction in blood-pressure measurement and the phase-II strategies, she was given full responsibility for their implementation. Phase-it patients received follow-up care from the same source as in phase I, and the programme coordinator was not permitted to contact the treating physicians (a protocol for medical emergencies was devised but never had to be used). At the end of phase ii (in the twelfth month of treatment) patients were evaluated both at home and at the mill by examiners who were "blind" to their experimental group allocation. The home visitor verified each patient’s drugs and doses and, while the patient was supplying a urine specimen (requested without prior warning), did an unobtrusive pillcount and compared it with a baseline established one month earlier. Within a few days of this home visit, the standardised blood-pressure measurement was repeated at the mill and a blood-sample was taken. In this paper, compliance-rates are reported as the proportion of pills prescribed for the twelfth month of therapy which were removed from their containers and, presumably, swallowed by the patients. Criteria for success were set in advance. Intervention was to be considered clinically useful if the average twelfth-month compliance-rate among experimental patients met three conditions : first, it had to exceed the sixth-month baseline coma

pliance

among

experimental patients by 20%

or

more;

and evaluations. 1 control

weight,

complaints reported

at

their sixth-month

developed deep-vein thrombosis and his hypotensive drugs were stopped; this patient was removed from the study, leaving 20 experimental patients and 18 controls. 16 of the 20 experimental patients were cooperative throughout phase u and kept an average of just under ten appointments (mean duration=30 min) with the programme coordinator. 1 refused the phase-n intervention entirely and 3 others repeatedly missed appointments completing an average of only 2.5 visits (mean duration=40 min). Phase-n outcomes are summarised in figs 1-3 and the table. Over the course of phase u, substantially more experimental than control patients exhibited increased

Incoaunt

Fig. 1—Changes in patients between start and end of phase II. The closed bars represent experimental patients and the open bars, controls.

second,

this difference had to be statistically significant; third, the twelfth-month mean compliance-rate among phase it experimental patients had to exceed that for controls by 20% or more.

Results

315 (6%) of the DOFASCO employees screened met criteria for sustained primary hypertension. 70 were already on treatment and were thus ineligible. Of 245 eligible men, 230 (94%) agreed to take part in our trial. In phaseI we tested the effects of "augmented convenience" by ’comparing outcomes in patients managed by industrial physicians’ in company time with those handled by their general practitioners in the usual way, and we compared the effects of "mastery learning" (learning the facts about hypertension). Neither strategy led to the patient being more likely to take his pills or achieve goal blood-pressure. The 20 experimental and 19 controls who, because they were neither compliant nor at goal blood-pressure in the sixth month were allocated to receive or not receive the phase-n strategies, were similar with respect to six month blood-pressures, previous compliance, and allocation to the phase-I interventions (these matching characteristics were used in their phase-n allocation by minimisation), and were also similar for age, height, our

Fig. 2--Compliance distributions among experimental patients.

before and after

phase-II strategies

1267

EFFECTS OF PHASE-II STRATEGIES UPON COMPLIANCE AND BLOOD-PRESSURE

* Withm group comparisons by paired t test, one-tailed. between group comparisons by unpaired t test, one-tailed.

compliance, decreased diastolic blood-pressure, or both, and this is shown in fig. 1. Furthermore, as shown in the table, the changes in compliance among experimental patients met the previously established criteria for success : their twelth-month compliance exceeded their sixth-month compliance by more than 20%, this difference was statistically significant, and the twelfthmonth compliance among experimental patients exceeded that of control patients by 20% or more (this latter difference was also statistically significant). Fig. 2 illustrates the distribution of compliance levels among experimental patients at both the beginning and end of phase n. These changes in compliance among experimental patients were paralleled by decreases in their diastolic blood-pressure (fig. 3). Although the average pressure fall was a modest 5.4 mm Hg it contributed to a total fall in diastolic blood-pressure of 11-6 mm Hg for the full twelve months of therapy, a drop which approaches the 13-2 mm Hg average fall observed among highly compliant patients at six months. Furthermore, this diastolic blood-pressure fall among experimental patients in phase n exceeded 10 mm Hg in 6 patients, and 4 of these (plus 2 other experimental patients with lesser falls) achieved the diastolic blood-pressure goal; 3 controls exhibited pressure falls of 10 mm Hg or more, and 1 of these (plus 1 other) achieved goal blood-pressure.

significant portion of hypertensives who are neither compliant nor at goal blood-pressure six months after the initiation of treatment. These results, if replicable in other settings, may also encourage clinicians who, because of prior disappointment over poor compliance, have been reluctant to label and treat hypertensives. Although the combination of self-measurement, the recording of home blood-pressures and pill-taking, tailoring, and reinforcement seems responsible for the differences between experimental and control patients in the twelfth month, four alternative explanations must be considered. First, could these differences simply reflect a lack of comparability of experimental and control patients? This has been a major problem in interpreting other reports of new strategies for improving compliance and follow-up among hypertensives in which comparison groups have either been absentS or consisted of noncontemporaneous controls.6 However, we compared outcomes between groups generated through the allocation of patients previously stratified for factors likely to affect subsequent compliance and blood-pressure, and the experimental and control groups in this trial were highly similar for prior compliance and blood-pressure levels, age, weight, height, complaints, and phase-i intervention strategy. Therefore, we are confident that this source of bias is absent. Second, could the differences in blood-pressure we observed simply reflect a "training effect" in which experimental patients, more accustomed to having their blood-pressures measured, would exhibit lower, more "relaxed" readings. at the twelfth-month evaluation? This possibility has been tested, and the result is negative.7 Carnahan and Nugent provided cuffs and stethoscopes to a random half of a group of 100 hypertensives

Discussion The phase-n results of this controlled trial of strategies for improving medication compliance in primary hypertension are encouraging, for they suggest that behaviourally oriented strategies can salvage a clinically

-

T 3

Fig. 3-Changes in diastolic blood-pressure and compliance between start and end of phase II among experimental patients.

1268 and found no difference at six months between their diastolic blood-pressures and those of control patients who had not practised home blood-pressure measurement. It is therefore unlikely that the blood-pressure differences we observed were the result of a "training" bias. Third, do the phase-ii results simply reflect the "delayed action" of a phase-I strategy? More specifically, is a delayed effect of mastery learning, reinforced by the phase-!! behavioural strategy, responsible for the differences observed in the twelfth month? Apparently not. Phase-! experiences were evenly divided between the phase-!! groups (11of 20 experimental patients and 10 of 18 controls had undergone mastery learning). Furthermore, although patients who had been through the mastery learning programme were slightly more responsive to the phase-II intervention, there was once again no correlation between knowledge about hypertension and compliance at twelve months. We are not satisfied, however, that this investigation is free from a fourth potential source of bias, and this is the confounding of the compliance-improving strategies with the amount of attention shown to these

patients. By design, phase-II experimental patients received more attention (five hours, spread over six months) than phase-!! controls, and our review of the compliance literature suggests that simply spending more time with patients, regardless of the content of the interchange, is associated with increased compliance.8 Is it possible that this increased attention all by itself, regardless of the precise compliance strategy, was responsible for some or all of the differences noted at the end of phase n? A review of the phase-I results is helpful here. Although patients who underwent mastery learning received more attention than their corresponding controls, they were no more compliant. On the other hand, this group was somewhat more responsive to the phase-!! strategy, a result consistent with a cumulative "attention" effect. We have suggested elsewhere that future trials of compliance-improving strategies include "attention placebos",9 and our subsequent randomised trial of home blood-pressures and home visits, now in progress, will determine the independent contribution of "attention" to the achievement of high compliance and

goal blood-pressure. We have done several other analyses of this controlled one deserves mention here. The effects of phase-!! strategies were not limited to patients’ compliance. Although the programme coordinator was not allowed to contact the treating physicians, these physicians were nonetheless more likely to increase the antihypertensive regimens of experimental patients than of controls. By the end of phase n the former were prescribed more vigorous therapy, on average, than the latter ; this effect will be reported in detail elsewhere. In adopting the strategy of the pharmacological trial to the testing of compliance-improving strategies we have searched for the counterpart of the untoward drug reaction. We have made serial measurements of these patients’ social and emotional function and have monitored their absenteeism as well as their reports of sideeffects. A preliminary analysis of these data shows no deleterious effect of the compliance-improving strategies on these indexes of function and self-image. (This portion of the investigation has been carried out in collabo-

trial, and

Jana Mossey, department of social and community medicine, University of Manitoba, and will be reported elsewhere). This controlled trial, while not designed to describe the current state of hypertension at a community level, does provide an indication of the present situation in urban, industrial Canada. Among men working in a steel mill, 14% have diastolic (fifth phase) blood-pressure above 90 mm Hg and, of those with persistent presmm Hg, 46% are aware of their condition and sures >95 22% are on drugs; these values are similar to those of earlier population surveys.1O In view of this consistency, our subsequent discovery that only two-thirds of men with diastolic pressures consistently >95 mm Hg will be started on drugs, and that of those begun on medication only slightly more than half will be taking 80% or more of their prescribed doses, are sobering indeed. The inability of isolated screening programmes to solve these problems and the need for rigorous validation of strategies reputed to alter the behaviour of clinicians and paration with Dr

tients

are

underscored.

Finally, the behaviourally oriented strategies we used in phase n, if confirmed in further trials, can be implemented in the existing system of clinical services. As we shall describe elsewhere, a simple interview can detect up to half of non-compliant hypertensive patients, obviating the need for pill-counts or body-fluid assays and rendering the identification of a substantial portion of non-compliers both quick and practicable. Second, virtually all patients can be taught home blood-pressure measurement, using equipment which is already in the clinican’s office; neither automated nor integrated cuffs and stethoscopes are required. Third, this trial showed that effective compliance-improving strategies can be applied, maintained, and supervised by a layperson without demanding either more time from a busy clinician or more reorganisation from a beleaguered health service. In view of this potential, the need for further refute the value of these and other compliance-improving strategies is great. We thank Mrs Charmaine Turford who carried out the phase-n intervention programme, Dr Ian Cunningham and the nurses of the trials

to

confirm

or

Hamilton Wentworth Health Unit who assisted with the twelve-month home visits, Dr Robert Martin, Mr Brian Austin, and the other members of the medical department at DOFASCO, and the physicians in Hamilton and the surrounding communities without whose encouragement and cooperation the trial could never have happened. This work was supported in part by grant no. MA-5159 from the Medical Research Council of Canada, National Health Grant 606-22-12 from Health and Welfare Canada, and a grant from Dominion Foundries and Steel Company of Canada. R. B. H. is a Physician’s Services Inc. foundation fellow. Requests for reprints should be addressed to D. L. S., McMaster Universitv, Department of Clinical Epidemiology and Biostatistics, 1200 Main Street West, Hamilton, Ontario, Canada L8S 4J9. REFERENCES 1.

Sackett, D. L. in International Textbook on Hypertension (edited by J. Genest, E. Koiw, and O. Kuchel). New York (in the press). 2. Haynes, R. B. in Compliance with Therapeutic Regimens (edited by D. L. Sackett and R. B. Haynes). Baltimore (in the press). 3. Sackett, D. L., Haynes, R. B., Gibson, E. S., Hackett, B. C., Taylor, D. W., Roberts, R. S., Johnson, A. L. Lancet, 1975, i, 1205. 4. Taves, D. R. Clin. Pharm. Ther. 1974, 15, 443. 5. Alderman, M. H., Schoebaum, E. E. New Engl. J. Med. 1975, 293, 65. 6. Finnerty, F. A. Jr., Shaw, L. W., Himmelsbach, C. K. Circulation, 1973, 47, 76. 7. Carnahan, J. E., Nugent, C. A. Am. J. med. Sci. 1975, 269, 69. 8. Haynes, R. B. in Compliance with Therapeutic Regimens (edited by D. L Sackett and R. B. Haynes). Baltimore (in the press). 9. Sackett, D. L. ibid. 10. Wilber, J. A., Barrow, J. G. Am.

J. Med. 1972, 52, 653.