Acta Med Scand 209: 271-275.1981

Patient Report Validated against Prescription Records for Measuring Use of and Compliance with

Antihypertensive Drugs

Hannes Enlund, Jaakko Tuomilehto and Hannu Turakka

From the Department of Community Health, Division of Social Pharmacy, and the North Karelia Project, University of Kuopio, Kuopio, Finland

ABSTRACT. Drug use and patient compliance in antihypertensive treatment were studied by interview and validated with prescription records in a Finnish population. The original study population consisted of 675 hypertensives identified in a 6.6% random survey of the 25-59-year-old population in 1972. At re-examination in 1977, 310 of the 570 partici- pants had then used antihypertensive drugs during the preceding year. Two thirds renewed prescriptions appropriately within 70 % ef the correct time (consid- ered as good compliance). Prescription fdling pat- terns of the re-examination non-participants did not differ from those of the participants. Eight per cent of the participants reported not having followed prescription instructions. There was 94 % agreement between the re-examination and prescription records; 5% reported taking fewer drugs than the records showed and 1 % reported taking more. Drug history given by the patient gave accurate information about the number and type of drugs in use. However, pa- tient interview gave an overestimation of compliance.

Key words: hypertension, drug therapy, drug utilization, patient compliance, interview, drug prescription. Acta Med Scand 209: 271, 1981.

Validity, or the degree of agreement between in- formation obtained from surveys and from clinical records, has been an issue of interest in health sur- veys (8). Little experience has been obtained about the validity of reported drugs 'prescribed compared with data collected by other methods. Prescribed antihypertensive drug treatment in Finland is pro- vided from the pharmacies free-of-charge for pa- tients classified as hypertensives. The costs are fully reimbursed to the pharmacy by the Social Insurance Institution, which also stores the pre- scription duplicates for three years. Thus, in Fin- land all antihypertensive medication records are

available from a single source, although not yet computerized, per drug items for each patient.

One of the main problems in medication com- pliance studies is the lack of accurate, comprehen- sive and valid measurements (4). Patient interview is commonly considered to be least valid, assess- ment from biological fluids most valid and pill counts somewhere in between (5, 6), although opposite views have also been published (3). Furthermore, information about non-participants in surveys is generally limited to demographic data, such as age and sex. The question that usually remains open is: do the non-participants differ in some other respect from the participants?

The purpose of this study was to validate the cardiovascular survey data with retrospective prescription data, with special reference to com- pliance and the antihypertensive drugs used.

SUBJECTS AND METHODS Study population The study comprised 675 persons from the districts of Joensuu and Kuopio. They were included in the random 6.6% sample of the population aged 25-29 years, partici- pated in the baseline survey of the North Karelia Pro- ject in 1972 (10) and had high blood pressure (SP) (2175 and/or 100 mmHg) or were on antihypertensive medica- tion. Because of identification problems and participation in clinical antihypertensive trials, 16 patients from the original sample were excluded. Five years later, in 1977, a re-examination was organized (Fig. 1).

Patient surveys The method of the surveys was a combination of postal survey, personal interview and clinical assessment (lo), so that the designs in 1972 and 1977 were strictly the same. However, in 1977 some additional information was collected. Ten days before the examination the subjects received a le t tg which contained an explanation of the

Acta Med Scand a9

272 H . Enlund et al.

- BASELINE SURVEY IIY 1972 TERMINAL SURVEY IN 1977

HYPERTENSIVES (2175 AND/OR RE-EXAM1 l lATION 100 M M I k OR MEDICATION) (N=570) (84.4 %)

IN THE SAMPLE I4EDICATION NO MEDICATION ~

(N-675) (N=310) (N=260)

Fig. 1 . Design of the study, number of participants on and off medication.

study, an invitation to the examination and a question- naire. The questionnaire consisted of 130 mainly precoded questions. Subjects with known hypertension answered special questions about their treatment. At the ex- amination, casual BP was measured in the sitting position according to the standardized techniques, the fifh phase being recorded as the diastolic pressure (12). In 1977, persons with BPs 2160 and/or 95 mmHg in

two consecutive measurements or who were on anti- hypertensive drug treatment received an invitation to a control visit. Besides repeating the BP measurement, special information on treatment was collected. Informa- tion on antihypertensive drug was obtained by advising people to take their drugs and prescriptions with them to the examination. The collected data included the trade name and strength of the tablets and the daily dosage. Subjects who did not attend the Fist examination were re-invited to come at the time of the second examina- tion.

Collection of prescription data An indirect method, based on veriods between dimensine

subjects between June 1976 and June 1977. About 96% of all prescriptions dispensed in the country are kept in these archives for three years after dispensing. The duplicates contain details of preparation, amounts, dosage, indication given to the patient and dispensing dates. The amount of reimbursed drug dispensed to the patient at one time is not allowed to exceed a three-month supply. The data were collated manually and transferred to a computer.

The actual treatment period was calculated from the difference between the Fist and last dispensing date and was compared with the theoretical treatment period during which the patient should have used his medicine if the instructions had been followed exactly. For each anti- hypertensive used by the patient a compliance ratio was calculated and for each patient an individual compliance score, covering every drug in the antihypertensive treat- ment. These calculations have been presented elsewhere (16).

RESULTS Of the original study group, 570 persons (84.4%) attended the re-examination (Fig. 1). The data on

dates, was used for assessing the amount of drug con- sumed. Analysis was performed by examining retro- svectivelv from the archives of fie Social Insurance Ihstitution all the prescription duplicates of the study antihypertensive drugs remained incomplete in 70

subjects (22.5 %), mainly because of failure to attend the control visit. Of the 105 non-participants in 1977 who had been classified as hypertensives in the 1972 survey, 43% had been treated with

Table I. Some characteristics of participants and non-participants in 1977

antihypertensives during the preceding year accord- pants ticipants cance SignE- ing to prescription records. The proportion of (n=570) (n=105) level treated participants was significantly higher, 54 %

(p<O.O5). The age and sex of the participants Men (%) 44 Women (%) 56 Age (Y J 30-44 24 45-59 53 60-64 23

Antihypertensive prescriptions during the preced- ing year (%) 54

49 n.s. showed no statistically significant difference from the non-participants, although the latter tended to 51

30 cumulate in men and younger age m u p s (Table I). 47 n s . Differences in the type of antihypertensive medi-

cation between the three groups were relatively 23

small (Table 11). Most of the patients had been treated with diuretics. Patients with incomplete drug history had used least diuretics and methyl-

43 pC0.05

Acta Med Scand 209

Patient report validated against prescription 273

Table 11. Type of antihypertensive substances, assessed by prescription records, used by 355 treated patients (% of patients using each type)

Interviewed participants

Complete Incomplete non- drug drug partici- history history pants (n=240) (n=70) (n=45)

Treated

Table IV. Medication compliance assessed by pre- scription monitoring during one-year follow-up (% of patients)

Interviewed participants

Complete Incomplete non- drug drug parti-

Treated

Compliance history history cipants score (n=218) (n=64) (n=42)

Diuretics 83 71 80 Beta-blockers 56 57 49 Methyldopa 25 14 24 Clonidine 15 11 1 1 Vasodilators 16 16 18 Reserpine and sympathetic in- hibitors 1 1 4 1 1

Good (1.0-0.7) 67 72 62 Poor (c0.7) 33 28 38 Drugs dispensed only once during 9.2 8.6 6.7 the year (n=22) (n=6) ( n = 3 )

x2=1.16, n.s.

dopa. Among the non-participants there were some- what less patients treated with P-blockers (49 %) than among the participants (56 %). The participants with an incomplete drug history were treated most often with only one drug (43 %), the corresponding figure among those with a complete drug history being 27% (p<0.05) and 31 % among non-partici- pants. Thus, the majority of the patients were treated with more than one antihypertensive agent, two substances being the most common treatment model (47 % of the patients).

Of the 310 re-examined patients whose prescrip- tions showed that they should have used antihyper- tensive drugs during the preceding year, 2% com- pletely denied their medication in the question- naire and 9% did not answer the question. Thus correct information about whether or not drugs were used was obtained from 89% of the treated subjects.

However, adequate validation between the two

Table 111. Validation of reported antihypertensive drug items against prescription records in 240 re- examined patients Sensitivity of the interview = 394/416=94.7 %, predictive power (+) = 394/397=99.2 %

methods is possible only from the interviewed pa- tients with a complete drug history (Table 111). In this group, only 5 % of the drugs that were in use according to the prescriptions dispensed prior to the survey were not reported at the interview. These 22 drugs were used by 19 patients. On the other hand, three patients reported taking anti- hypertensive drugs that could not be confirmed from the prescription records. Thus, sensitivity of the interview method was as high as 94.7 %. In the questionnaire the patients were asked

whether they take their pills as prescribed or in some other dose. Eight per cent of the patients with a complete drug history reported having deviated from the instructions. Most of these persons (7%) said they were using fewer tablets a day than in- structed and only 1 % more than prescribed. Thus, reported compliance, i.e. the number of tablets taken vs. the number of tablets prescribed, was very high. On the other hand, good compliance 6.e. maximal difference f30% of that of pre- scribed) was achieved by only two thirds of the patients, when long-term medication compliance during a one-year follow-up period was assessed by monitoring prescription records (Table IV). The non-participants had the highest proportion of poor compliers but the difference from DarticiDants was

Prescription records not statistically significant. The measuring method was such that those who had been dispensed anti-

Interview Yes No Total hypertensives only once during the follow-up period could not be assessed for compliance. The data that were missing for this reason amounted to less than Yes 394 3 397

No 22 0 22 10% of those who had used antihypertensives. No Total 416 3 419 differences were found between the three groups.

18-812989 Aria Med Scand 209

274 H . Enlund et al.

DISCUSSION In this study the possibility of collecting data from non-participants enabled us to avoid several diffi- culties. Previous studies on non-participants have disclosed discrepancies, e.g. higher morbidity and mortality (2, 14, 15). Other authors (18) have claimed that the most common reasons for non- participation are very logical, e.g. that the person has recently been under medical observation andlor considers he will not benefit from survey partici- pation. Theoretically, compliance with drug treat- ment in hypertension could be expected to be lower among non-participants than participants in cardio- vascular surveys. In our series the percentage of treated patients was significantly higher among participants than among non-participants. The difference in compliance scores between these two groups was, however, not statistically significant.

These results indicate that in interviews a major- ity of patients give very accurate information about the number and type of drugs in use. For the few patients who denied their verified antihypertensive therapy there may have been some misunderstand- ing concerning the nature and purpose of their drugs (e.g. diuretics). However, patient interviews overestimated compliance as determined from prescription records. This agrees with other studies in which the interview method was compared with some more objective method such as pill counting or urine tests (5 , 9, 11). Thus, although patient reports should not be neglected when analysing general drug use, their validity must be treated with caution when evaluating medication com- pliance. However, it has been shown that when pill counts indicated major deviations from the precribed dosage, the patients were more likely to report a deviation than when pill counts showed only minor deviations (9, 11). It seems that verbal reports will provide sufficient information about patients for whom deviation information is most critical. Further, those who admit non-compliance are virutally always telling the truth (13).

Especially in cases of minor deviation one has to remember that patients are only human and occasionally do forget to take their medication or decrease their intake slightly because of side- effects (real or imagined) or simply because, lacking symptoms, they feel no need. Such patients may well honestly believe that they have fully complied with the prescribed regimen and, thus, will report no deviation in an interview.

The methods used in this study have some advan- tages compared with other methods used in com- pliance studies on antihypertensive therapy. Being partly longitudinal, the study design gives a wider view of compliance compared with the cross- sectional methods. However, even if a one-year follow-up as such is quite long, it is still brief in relation to what is often life-long hypertension treat- ment. Many authors have shown that compliance decreases with time (1,7). Patients selected for this study were followed by the existing health care sys- tem. The prescription record was monitored with- out the patients being aware of this. There was only the invitation to the survey to pressurize subjects into giving a favourable response, as might be the case in some clinical studies in which patients are specially motivated (17). Also, the five-year period between the two surveys is long enough to mini- mize the effect of labelling patients into a special follow-up group. Due to the nature of the study population, the results can be applied to daily prac- tice.

The practical implication of this study for daily practice in the treatment of hypertension is evident. When the practicing physician asks his patient for a drug history, he can expect answers very similar to our results, i.e. rather trustful concerning the nature of the medication but with an apparent overestimation of compliance. However, at least in countries with standardized prescription and dispensing rules (e.g. maximum allowable drug amount, obligatory instructions with single and daily dose, indication, etc.) as well as prescription sheets with a space for each dispensing date (repeat prescriptions), a permanent compliance follow-up of each patient is easily carried out by the physician or the pharmacy. By simply following the differ- ences between prescribed and consumed amounts of drug, by comparing two or more dispensing dates, much information can be obtained on com- pliance. This presupposes, as does the validity of this study, that the dispensed amount equals the amount actually consumed. We conclude that pre- scription sheets will provide useful information for monitoring compliance in hypertension control programmes and could easily be used in normal practice.

ACKNOWLEDGEMENT This study was supported by Y$i Jahnsson Foundation.

Acta Med Scand 209

Patient report validated ag$nst prescription 275

REFERENCES 1. Caldwell, J . R., Cobb, S., Dowling, M. D. & Jongh,

D.: The dropout problem in antihypertensive treat- ment. J Chronic Dis 22: 579, 1970.

2. Criqui, M. H., Barret-Connor, E. & Austin, M.: Differences between respondents and non-respond- ents in a population-based cardiovascular disease study. Am J Epidemiol 108: 367, 1978.

3. Fletcher, S. W., Pappius, E. M. & Harper, S. J.: Measurement of medication compliance in a clinical setting. Comparison of three methods in patients pre- scribed digoxin. Arch Intern Med 139: 635,1979.

4. Gordis, L.: Methodologic issues in the measurement of patient compliance. In: Compliance with thera- peutic regimens (ed. D. L. Sackett & R. B. Haynes), p. 51. Johns Hopkins University Press, Baltimore and London 1976.

5 . - Conceptual and methodologic problems in measur- ing patient compliance. In: Compliance in health care (ed. R. B. Haynes, D. W. Taylor & D. L. Sackett), p. 23. Johns Hopkins University Press, Baltimore and London 1979.

6. Gordis, L., Markowitz, M. & Lilienfeld, A. M.: The inaccuracy of using interviews to estimate patient reliability in taking medications at borne. Med Care 7: 49, 1969.

7. Haynes, R. B.: Determinants of compliance: The disease and the mechanics of treatment. In: Com- pliance in health care (ed. R. B. Haynes, D. W. Taylor & D. L. Sackett), p. 49. Johns Hopkins University Press, Baltimore and London 1979.

8. Meltzer, J. W. & Hochstim, J. R.: Reliability and validity of survey data on physical health. Public Health Rep 85: 1075, 1970.

9. Park, L. C. & Lipman, R. S.: A comparison of pa- tient dosage deviation reports with pill counts. Psychopharmacologia 6: 299, 1964.

10. Puska, P., Tuomilehto, J., Salonen, J., Neittaanmiiki, L., Maki, J., Virtamo, J., Nissinen, A., Koskela, K. & Takalo, T.: Changes in coronary risk factors during comprehensive five-year community programme to control vardiovascular diseases (North Karelia pro- ject). Br Med J 2: 1173, 1979.

11 . Rickels, K. & Briscoe, E.: Assessment of dosage deviation in outpatient drug research. J Clin Pharma- col 10: 153, 1970.

12. Rose, G. & Blackburn, H.: Cardiovascular survey methods. WHO Monogr Ser 56, 1968.

13. Sackett, D. L.: A compliance practicum for the busy practitioner. In: Compliance in health care (ed. R. B. Haynes, D. W. Taylor & D. L. Sackett), p. 286. Johns Hopkins University Press, Baltimore and London 1979.

14. Svardsudd, K.: High blood pressure. A longitudinal population study of men born in 1913 with special reference to development and consequences for health. Goteborg 1978.

15. Tibblin, G.: A population study of 50-year-old men. An analysis of the non-participant group. Acta Med Scand 178: 453, 1965.

16. Turakka, H., Enlund, H. & Tuomilehto, J.: Drug use and patient compliance in hypertension therapy. In: Progress in clinical pharmacy (ed. D. Schaaf & E. van der Kleijn), p. 53. ElseviedNorth Holland Bio- medical Press, Amsterdam, New York and Oxford 1979.

17. Veterans Administration Cooperative Study Group on Antihypertensive Agents 11. Results in patient with diastolic blood pressure averaging 90 through 114 mmHg. JAMA 213: 1143, 1970.

18. Vuori, H., Rimpela, M. & Gronroos, M.: Cytological screening programmes: The problem of non-partici- pation. Int J Health Educ 15: 22, 1972.

Acta Med Scand 209