Compliance with medication among outpatients with uncontrolled hypertension in the Seychelles.

Research I Recherche Compliance with medication among outpatients with uncontrolled hypertension in the Seychelles P. Hungerbuhler,' P. Bovet,2 C. Shamlaye,3 B. Burnand,4 & B. Waeber5 Owing to increasing rates of hypertension and cardiovascular-related diseases in developing countries, compliance with antihypertensive medication is of major public health importance. Few studies have reported on compliance in developing countries. We determined the compliance of 187 patients with uncontrolled hypertension in the Seychelles (Indian Ocean), by assessing the presence of a biologic marker (riboflavin) in the urine. The urine tested positive in 56% of the cases. Compliance varied from one physician to another (highest 72% versus lowest 33%, P = 0.003), improved with the level of litera- cy (62% versus 45%, P = 0.024), and depended on the presence or absence of diuretics in the medica- tion (respectively, 45% versus 66%, P = 0.005). The ability of patients to report correctly the number of antihypertensive pills to be taken daily was a predictor of compliance (62% of the patients who gave appropriate answers had positive urine for the marker versus 31% for those giving inappropriate answers). Introduction Treatment of hypertension has been shown to reduce cardiovascular morbidity and mortality (1, 2). Uncontrolled hypertension has been attributed to patients' failure to follow properly a prescribed drug regimen in approximately half the cases (3, 4), thus reducing both the effectiveness and the cost-effec- tiveness of treatment (5) and leading to avoidable hospitalization (6) and increased rates of coronary events (7). Compliance with treatment is therefore important since hypertensive patients are often asymptomatic and medication is required over a long period. Few studies have examined compliance with 1 Programme Manager, Epidemiology and Research Unit, Minis- try of Health, Victoria, Seychelles. 2 Registrar, Clinical Epidemiology Unit, Institute of Social and Preventive Medicine, University of Lausanne, Lausanne, Swit- zerland. 3Head, Epidemiology and Research Unit, Ministry of Health, Victoria, Seychelles. 4Lecturer, Clinical Epidemiology Unit, Institute of Social and Preventive Medicine, University of Lausanne, Bugnon 17, CH- 1005 Lausanne, Switzerland. Requests for reprints should be sent to Dr B. Burnand at this address. 5 Associate Professor, Division of Hypertension, University Hos- pital, Lausanne, Switzerland. Reprint No. 5617 antihypertensive treatment in developing countries, although the appropriate management of hyperten- sion is an economically affordable means to control cardiovascular diseases in countries with limited resources (8), where hypertension is becoming a major health problem (9-11). Currently, in the adult population of the Seychelles (Indian Ocean) the prevalence of high blood pressure is 32%; one third of hypertensives are receiving medication and only half of these have their blood pressure under control (11, 12). To investigate the extent of compliance among patients with medically treated hypertension we chose riboflavin (vitamin B2), a natural and non- teratogenic constituent of the diet (13), as a marker and relatively accurate indicator of compliance (14-17). Patients and methods The study was carried out towards the end of 1993 with the patients of ten doctors in 7 general outpa- tient clinics, out of the 15 which provide ambulat- ory medical care in the Seychelles. Participation was solicited and obtained from 223 consecutive patients with uncontrolled hypertension, none of whom was taking vitamin supplement preparations. Uncontrolled hypertension was defined as a diastolic blood pres- sure >90 mmHg despite medication. Patients with a Bulletin of the World Health Organization, 1995, 73 (4): 437-442 © World Health Organization 1995 437 P. Hungerbuhler et al. diastolic blood pressure >115 mmHg required im- mediate and intensive treatment, and for this reason were not included in the study. The protocol was ap- proved by the Ministry of Health of the Seychelles. Medical consultation and medication were free of charge, as for all inhabitants of the Seychelles. At the time of the first visit (No. 1), the study patients received riboflavin to be taken once every morning (1 tablet of 10 mg, Hanseler AG, Herisau, Switzer- land), in addition to their usual antihypertensive medication. All participants received their medica- tion at the pharmacy of the primary health care centre, as usual. Data about clinical characteristics, socio-demographic features, degree of literacy, health habits, history of hypertension, and duration of antihypertensive treatment were obtained by inter- view. Literacy was evaluated by the ability to read a local newspaper in at least one of the three national languages. Study patients were invited for a follow- up visit (No. 2) after an arbitrary period of 21±3 days. Patients were not informed that a special pro- cedure (urinary examination) was to be performed on visit No. 2. At the time of visit No. 2, the blood pressure was measured and patients were asked about the number of antihypertensive pills taken daily, which was compared with the treatment prescribed by the doctor. Results were "concordant" when the daily number of pills prescribed and taken were similar, or else "non-concordant". Compliance with medication was estimated by testing for the presence of metab- olites of riboflavin in the urine, which was collected at the time of visit No. 2. This test was carried out by the treating doctors, after special training using an UV light source according to the method described by Jones (18). Patients were excluded from the study if visit No. 2 was not within the 21±3-day period or not between 9 a.m. and 2 p.m. or if they were hospi- talized in the interval, or had made an extra visit and had the treatment changed. Rates of compliance within categories of variables were compared using the chi-squared and trend tests. Logistic regression analyses were performed to examine the relation of compliance to independant variables, including demographic, educational, lifestyle, and clinical characteristics. Results A total of 223 consecutive patients were enrolled in the study, and 36 had to be excluded, mainly because of non-attendance at visit No. 2. The main character- istics of the remaining 187 patients are summarized in Table 1. The urine tested positive in 104 subjects (compliance rate, 56%). Compliant patients had a Table 1: Numbers of individuals with positive and nega- tive urine tests for riboflavin within categories of selected characteristics Urine negative Urine positive Characteristics (n = 83) (n = 104) pa Sex: Females Males Age (years): <50 50-65 >65 Literate: Yes No Smoking: No Yes (>1 cigarette/day) Alcohol: No Yes (>1 drink/day) Duration of therapy: <5 years >5 years Number of tablets/day: <4 4-6 >6 Reported vs. prescribed number of pills/days: Non-concordant Concordant Doctor: Other No. 3 No. 7 59 70 (54)b 24 34 (59) 26 40 17 28 (52) 57 (59) 19 (53) 45 73 (62) 38 31 (45) 67 91 (58) 16 13 (45) 52 72 (58) 31 32 (51) 41 46 (53) 42 58 (58) 34 31 18 46 (58) 35 (53) 23 (56) 27 12 (31) 56 92 (62) 53 12 18 64 (55) 31 (72) 9 (33) 0.579 0.665 0.024 0.203 0.344 0.482 0.786 <0.001 0.006 a Chi-squared tests were calculated to assess the difference in proportion of riboflavin positive and negative urine tests among subcategories of the characteristics listed in column 1. b Figures in parentheses are percentages within each category. mean diastolic blood pressure during visit No. 2 of 96 mmHg, compared with 100 mmHg for the others (P = 0.05); the blood pressure had been similar in the two groups during visit No. 1 (102 vs. 104 mmHg in compliant and non-compliant patients, respectively). Compliance according to demographic, education- al, lifestyle, and clinical characteristics. Tables 1 and 2 show that compliance varied among the differ- ent physicians (P = 0.006), and was associated with literacy (P = 0.024) and the ability to report correctly the number of antihypertensive pills taken daily WHO Bulletin OMS. Vol 73 1995438 Compliance with antihypertensives in Seychelles Table 2: Univariate and multivariate relations between compliance with medication and selected characteris- tics Univariate Multivariateb Characteristics ORa 95% Cl ORa 95% Cl Sex: Females Males Age (years): <50 50-65 >65 Literate: Yes No Smoking: No Yes (.1 cigarette/day) Alcohol: No Yes (.1 drink/day) Duration of therapy: <5 years >5 years Number of tablets/day: <4 4-6 >6 Reported vs. prescribed number of pills/days: Non-concordant Concordant Doctor: Other No. 3 No. 7 1.19 0.64-2.24 1.32 0.67-2.60 1.04 0.44-2.44 0.50 0.27-0.92 0.60 0.27-1.33 0.75 0.40-1.38 1.23 0.83 0.94 3.70 0.69-2.20 0.43-1.62 0.44-2.03 1.73-7.91 2.14 0.99-4.60 0.41 0.17-1.00 1.97 0.78-4.99 1.81 0.78-4.18 1.82 0.63-5.27 0.64 0.29-1.42 0.52 0.18-1.55 0.92 0.39-2.17 1.31 0.65-2.65 0.91 0.41-2.00 1.19 0.46-3.06 3.57 1.53-8.30 1.56 0.62-3.92 0.48 0.17-1.34 a An odds ratio (OR) above 1 means better compliance. b Multivariate analysis includes all the variables listed as well as use of diuretics. (P <0.001). Estimates of odds ratios were different for patients with appropriate reporting of their anti- hypertensive treatment compared to patients with inappropriate reporting, and these estimates were similar in univariate and multivariate analyses. None of the other relationships was significant in the mul- tivariate analyses, but the point estimates of the two significant univariate correlates (literacy and doc- tors) were similar in multivariate analyses. Variables previously demonstrated as being predictive of adherence to medical regimens, i.e., age (19), num- ber of daily tablets (20-22), as well as other vari- ables, like sex, smoking and drinking habits, and the duration of therapy, were not related to compliance in this study. Compliance according to medications. Table 3 shows the utilization of different antihypertensive medications. The use of diuretics, the most common treatment, was associated with a lower compliance in univariate analysis (P = 0.005); although not more significant statistically, the point estimate of the effect was unchanged in multivariate analysis. Other types of antihypertensive medications were not relat- ed to a significant modification of compliance, but better compliance was observed with P-adrenergic blockers. Predictive value for compliance. Table 4 shows that appropriate reporting back of treatment modalities, together with full compliance on self-reporting, had a 73% positive predictive value for compliance as indicated by a positive test for riboflavin. The posi- tive predictive value decreased to 62% when concor- dant reporting back was considered alone. Discussion This study showed 56% compliance with treatment in a population of uncontrolled hypertensive patients in the Seychelles. This result accords with previous studies in developed countries on compliance with treatment of a chronic disease like hypertension (4, 23-25). However, there are few studies on compli- ance in developing countries (26), despite its major public health importance. Developing countries face increasing rates of chronic diseases, notably hyper- tension and cardiovascular-related diseases (9-11). With poor compliance the efficiency of treatment is lower, thereby increasing the rate of uncontrolled hypertension (27). Urine testing for the tracer, ribo- flavin, is an immediate and valid measure of the patient's compliance on the day the specimen was collected (17). The procedure is simple, requires no expensive equipment (28, 29) and is well adapted to clinical research in a developing country (29). Owing to the short half-life, this method only confirms ingestion of the drug during the previous few hours, and gives no information on the dynamics of com- pliance. For this reason the importance of an overes- timation on the day of the control visit, as compared to the usual compliance ("toothbrush" effect), cannot be assessed (21, 30, 31). Furthermore, the exclusion of patients for non-attendance at the clinic presum- ably eliminates mainly patients with low adherence, thus leading to overestimation of compliance and emphasizing the importance of the phenomenon of non-adherence. WHO Bulletin OMS. Vol 73 1995 439 P. Hungerbuhler et al. Table 3: Univariate and multivariate relations between compliance and type of medication Univariate MultivariatebNo. of Characteristics patients ORa 95% Cl ORa 95% Cl Angiotensin-converting enzyme inhibitors: No 146 1 1 Yes 41 (22)C 1.03 0.51-2.07 0.99 0.45-2.19 Central-adrenergic inhibitors: No 113 1 1 Yes 74 (40) 0.57 0.32-1.04 1.98 0.58-6.77 3-adrenergic blockers: No 125 1 1 Yes 62 (33) 1.57 0.84-2.93 2.33 1.11-4.87 Calcium entry blockers: No 149 1 1 Yes 38 (20) 1.12 0.54-2.32 1.51 0.65-3.48 Hydralazine hydrochloride: No 155 1 1 Yes 32 (17) 1.20 0.55-2.62 1.59 0.64-3.92 Diuretics: No 96 1 1 Yes 91 (49) 0.43 0.24-0.78 0.53 0.26-1.10 a An odds ratio above 1 means better compliance. b Multivariate analysis included variables mentioned in Table 1. c Figures in parentheses are percentages. In agreement with the findings of previous stud- ies on the importance of the doctor-patient relation- ship for compliance (32, 33), substantial differences in levels of compliance were found among patients followed by different doctors. This difference may be related to the selection of patients in different primary health care centres. Literacy is probably an important determinant in the clinician-patient rela- tionship for adequate understanding of all aspects of this chronic and asymptomatic disease as well as of treatment modalities. In this study, a better rate of compliance was found among literate patients. Table 4: Compliance estimates by testing for metab- olites of riboflavin in the urine vs. appropriate reporting back of treatment modalities and full compliance on self-reporting Urine positive Urine negative (n = 104) (n = 83) Total Appropriate reporting back and full compliance: Yes 60 22 82 No 44 61 105 Total 104 83 187 Sensitivity = 58%, specificity = 73%, positive predictive value = 73%, negative predictive value = 58% Because most previous studies were done in devel- oped countries where illiteracy is less prominent, no data were available on this issue. The use of diuretics was associated with a lower compliance, possibly due to side-effects. But false negative results, due to dilution of the urinary metabolite of riboflavin, can- not be excluded, although not previously reported in the literature. In this study, appropriate reporting back on medication was predictive of a significant increase in compliance with the medical regimen. Patients who were more knowledgeable about their treatment were 3.6 times more likely to be compliant that those who were less aware, even after adjusting for other putative determinants of compliance. Questioning patients about their treatment can therefore give useful and simple information to identify compliant patients (19, 34). However, the positive and negative predictive values of an appropriate reporting back of treatment modalities on the actual compliance rate, as estimated by a positive test for riboflavin in urine, were not very high (73% and 58%, respectively) so that accurate reporting of medication cannot be used alone as a surrogate test. In conclusion, a 56% rate compliance with anti- hypertensive treatment was found in a population of uncontrolled hypertensive patients in a developing WHO Bulletin OMS. Vol 73 1995440 Compliance with antihypertensives in Seychelles country. This figure is probably an overestimate in view of the method used and the exclusion of non- attenders. These findings emphasize the need to improve the quality and efficiency of medical treat- ment, including improved cooperation between patients and doctors. This issue is of major public health importance in view of the rate of uncontrolled hypertension related to poor compliance (27). The resulting higher morbidity and mortality, associated with poor utilization of medication and funds, are a correctable problem in countries with limited resources (30). Acknowledgements We thank the general practitioners of the Ministry of Health, Seychelles, who participated in the study (Marc-Andr4 Jacobs, Clementine Iterbeke, Ingrid Van Acker-H6bert, David Mc Lees, Claude Yersin, Georges Panovski, Tsultrim Tenzin, Maureen Kirkpatrick, and E. Kamal Hussan), all the patients in the study for their cooperation, and Vincent Wiet- lisbach, statistician, for helping to prepare the manuscript. The study was supported in part by the Department of Cooperation, Canton of Jura, Switzerland; by the Hospi- tal Services of the Canton of Vaud, Switzerland (grant No. 4006); and by the Swiss National Science Foundation (grant No. 3233-038792.93). R6sume Observance du traitement chez des patients ambulatoires pr6sentant une hypertension non contr6l6e aux Seychelles Dans les pays en d6veloppement, ou l'on observe une augmentation des taux d'hypertension et des maladies cardio-vasculaires, l'observance du trai- tement par antihypertenseurs est d'une importan- ce primordiale du point de vue de la sant6 publique. Rares sont les publications consacrees a l'observance du traitement dans ces pays. La presente 6tude avait pour objectifs de d6terminer le taux d'observance du traitement chez des patients presentant une hypertension arterielle non contr6lee et consultant un centre de soins de sante primaires dans un pays en developpement, et d'evaluer les facteurs influant sur l'observance. Au total, 223 patients presentant une hyper- tension arterielle non contr6lee (tension diasto- lique 290 mmHg et <115 mmHg) vus consecutive- ment dans 7 centres de soins de sante primaires aux Seychelles ont ete suivis par un medecin parmi les 10 medecins participant 'a 1'tude. L'etude a ete realisee vers la fin de l'annee 1993. Le taux d'observance du traitement par antihyper- tenseurs a ete mesur6 indirectement par la pre- sence d'un metabolite urinaire de la riboflavine, trois semaines apres I'addition de riboflavine aux medicaments antihypertenseurs pris quotidienne- ment. Apres exclusion de 36 patients, essentielle- ment en raison de leur absence aux visites de suivi, 187 patients ont e etudies. Les tests uri- naires etaient positifs dans 56% des cas. Ce taux est probablement surestime, du fait de la methode utilisee pour evaluer l'observance et de 1'exclusion des patients perdus de vue. 11 variait d'un medecin a l'autre (de 72% a 33%, p = 0,003), s'ameliorait avec le niveau d'alphabetisation du patient (62% contre 45%, p = 0,024) et d6pendait de la presen- ce ou de I'absence de diuretiques dans le traite- ment (45% contre 66%, p = 0,005). Une analyse multivari6e, incluant a la fois ces variables et des parametres demographiques, n'a pas modifie sensiblement ces resultats. L'aptitude des patients a indiquer correctement le nombre de pilules d'antihypertenseurs a prendre chaque jour 6tait un facteur predictif de l'observance (62% des patients ayant donne une reponse correcte avaient un test urinaire positif pour le marqueur contre 31% pour ceux qui donnaient une reponse erronee, p<0,001). Ces r6sultats soulignent la necessite d'ameliorer la qualite et 1'efficience du traitement m6dical, y compris au niveau de la coop6ration patient-medecin. Cette question est d'une importance primordiale du point de vue de la sante publique etant donne le taux eleve d'hypertension arterielle non controlee due a la non-observance du traitement. La morbidit6 et la mortalite elevees qui en resultent, ainsi que la mauvaise utilisation des medicaments et des res- sources financieres, sont des problemes qui peu- vent etre corriges dans les pays ne disposant que de ressources limitees. References 1. Hypertension Detection and Follow-up Program Cooperative Group. Five-year findings of the hypertension detection and follow-up program. I. Reduction in mortality of persons with high blood pressure, including mild hypertension. Journal of the American Medical Association, 1979, 242: 2562-2571. 2. Collins R et al. Blood pressure, stroke, and coron- ary heart disease. Part 2. Short-term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet, 1990, 335: 827-838. WHO Bulletin OMS. Vol 73 1995 441 P. Hungerbuhler et al. 3. Col N et al. The role of medication noncompliance and adverse drug reactions in hospitalizations of the elderly. Archives of internal medicine, 1990, 150: 841-845. 4. NHLBI Working Group. Management of patient compliance in the treatment of hypertension: report. Hypertension, 1982, 4: 415-423. 5. Neal WW. Reducing costs and improving compli- ance. American journal of cardiology, 1989, 63: 177B-20B. 6. Maronde RF et al. Underutilization of antihyperten- sive drugs and associated hospitalization. Medical care, 1989, 27: 1159-1166. 7. Psaty BM et al. The relative risk of incident coron- ary heart disease associated with recently stopping the use of beta-blockers. Journal of the American Medical Association, 1990, 263: 1653-1657. 8. Arterial hypertension: report of a WHO Expert Com- mittee. Geneva, World Health Organization, 1978 (WHO Technical Report Series, No. 628). 9. Nissinen A et al. Hypertension in developing coun- tries. World health statistics quarterly, 1988, 41: 141-154. 10. Akinkugbe 00. Epidemiology of cardiovascular dis- ease in developing countries. Journal of hyperten- sion, 1990, 8 (suppl. 7): 233S-238S. 11. Bovet P et al. The Seychelles Cardiovascular Dis- ease Study. Sozial- und praventivmedizin, 1991, 36(suppl. 1): 3S-7S. 12. Bovet P et al. High prevalence of cardiovascular risk factors in the Seychelles (Indian Ocean). Arterio- sclerosis and thrombosis, 1991, 11: 1730-1736. 13. Porter AMW. Drug defaulting in general practice. British medical journal, 1969, 1: 218-222. 14. Deuschle KW et al. Clinical usefulness of ribofla- vin-tagged isoniazid for self-medication in tuberculo- sis patients. American review of respiratory disease, 1960, 82: 1-10. 15. Hobby GL, Deuschle KW. The use of riboflavin as an indicator of isoniazid ingestion in self-medicated patients. American review of respiratory disease, 1959, 80: 415-423. 16. Roth HP et al. Measuring intake of a prescribed medication: a bottle count and a tracer technique compared. Clinical pharmacology and therapeutics, 1970, 11: 228-237. 17. Jay S et al. Riboflavin, self-report, and serum nor- ethindrone. American journal of diseases of chil- dren, 1984, 138: 70-73. 18. Jones LH. Riboflavin as an indicator of drug-taking behavior. Medical journal of Australia, 1967, 1: 202-204. 19. Klein LE. Compliance and blood pressure control. Hypertension, 1988, 11 (suppl. Il): 11-61-11-64. 20. Eisen SA et al. The effect of prescribed daily dose frequency on patient medication compliance. Archives of internal medicine, 1990, 150: 1881-1884. 21. Cramer JA et al. How often is medication taken as prescribed? A novel assessment technique. Journal of the American Medical Association, 1989, 261: 3273-3277. 22. Pullar T et al. Use of a pharmacologic indicator to compare compliance with tablets prescribed to be taken once, twice, or three times daily. Clinical phar- macology and therapeutics, 1988, 44: 540-545. 23. Setaro JF, Black HR. Refractory hypertension. New England journal of medicine, 1992, 327: 543-547. 24. Becker MH. Patient adherence to prescribed thera- pies. Medical care, 1985, 23: 539-555. 25. Luscher TF et al. Compliance in hypertension: facts and concepts. Journal of hypertension, 1985, 3(suppl. 1): 3-9. 26. Lim TO et al. The Mentakab Hypertension study pro- ject. Part V. Drug compliance in hypertensive patients. Singapore medical journal, 1992, 33: 63-66. 27. Shea S et al. Predisposing factors for severe, uncontrolled hypertension in an inner-city minority population. New England journal of medicine, 1992, 327: 776-781. 28. Young LM et al. Riboflavin use as a drug marker in veterans administration cooperative studies. Con- trolled clinical trials, 1984, 5: 497-504. 29. Dubbert PM et al. Riboflavin as a tracer of medica- tion compliance. Journal of behavioral medicine, 1985, 8: 287-299. 30. Pullar T. Compliance with drug therapy. British jour- nal of clinical pharmacology, 1991, 32: 535-539. 31. Cramer JA et al. Compliance declines between clinic visits. Archives of internal medicine, 1990, 150: 1509-1510. 32. German PS. Compliance and chronic disease. Hypertension, 1988, 11 (suppl. Il): 11-56-11-60. 33. Bartlett EE et al. The effect of physician communica- tions skills on patient satisfaction, recall, and adher- ence. Journal of chronic diseases, 1984, 37: 755-764. 34. Stephenson BJ et al. Is the patient taking the treat- ment as prescribed? Journal of the American Medi- cal Association, 1993, 269: 2779-2781. 442 WHO Bulletin OMS. Vol 73 1995