The Effect of Prescribed Daily Dose Frequency
`on Patient Medication Compliance
`
`Seth A. Eisen, MD, MSc; Douglas K. Miller, MD; Robert S. Woodward, PhD; Edward Spitznagel, PhD; Thomas R. Przybeck, PhD
`
`\s=b\The objective of this study was to determine the relationship
`between prescribed daily dose frequency and patient medication
`compliance. The medication compliance of 105 patients receiv-
`ing antihypertensive medications was monitored by analyzing
`data obtained from special pill containers that electronically
`record the date and time of medication removal. Inaccurate com-
`pliance estimates derived using the simple pill count method
`were thereby avoided. Compliance was defined as the percent of
`days during which the prescribed number of doses were re-
`moved. Compliance improved from 59.0% on a three-time daily
`regimen to 83.6% on a once-daily regimen. Thus, compliance
`improves dramatically as prescribed dose frequency decreases.
`Probably the single most important action that health care pro-
`viders can take to improve compliance Is to select medications
`that permit the lowest daily prescribed dose frequency.
`(Arch Intern Med. 1990;150:1881-1884)
`^ umerous studies have demonstrated that poor medication
`compliance poses a significant impediment to the effec¬
`tive treatment of a wide variety of illnesses.1 One suggested
`method for improving compliance is to decrease the number of
`prescribed daily doses. However, the relationship between
`daily dose frequency and compliance is uncertain, with
`many2"10 but not all11'17 studies indicating that compliance pro¬
`gressively improves as daily dose frequency decreases. De¬
`fining the relationship between compliance and dose frequen¬
`cy is further complicated because almost all published studies
`use the pill count method for quantitating compliance. Pill
`counts must be interpreted with caution, since they provide
`no information about the day or time of dose removal.18
`Medications that are effective when administered once or
`twice daily have become increasingly available and promotion
`of these drugs has focused on their alleged compliance-en¬
`hancing characteristics. An evaluation ofthe validity of these
`
`Accepted for publication March 30,1990.
`From the St Louis (Mo) Veterans Affairs Medical Center (Dr Eisen); Depart-
`ment of Medicine, St Louis (Mo) University School of Medicine (Dr Miller);
`Department of Medicine (Dr Eisen), Health Administration Program (Dr
`Woodward), Departments of Preventive Medicine and Public Health, Division
`of Biostatistics (Dr Spitznagel), and Psychiatry (Dr Przybeck), Washington
`University School ofMedicine, St Louis, Mo.
`Reprint requests to (151-JC), St Louis Veterans Affairs Medical Center, 915
`N Grand Ave, St Louis, MO 63106 (Dr Eisen).
`
`claims and an assessment of the expected degree of compli¬
`ance improvement for these typically more expensive long-
`acting dose formulations is warranted.
`The purpose of this study was to investigate the relation¬
`ship between the prescribed daily dose frequency for antihy¬
`pertensive medications and patient medication compliance by
`analyzing medication compliance data obtained from unique
`pill containers that electronically record the date and time of
`medication removal.
`SUBJECTS AND METHODS
`Subjects
`Data were collected as part of a broader research effort to evaluate
`the effect offeedback ofcompliance information to health care provid¬
`ers on the medication compliance of their patients. Participation was
`solicited from 756 consecutive patients who fulfilled the following
`criteria: history of hypertension (controlled or not controlled while
`receiving medication), attending the St Louis (Mo) Veterans Affairs
`Medical Center in a continuity of care clinic environment, being
`treated with a once, twice, or three times a day dose frequency
`medication regimen, and all medications (excluding those prescribed
`on an "as needed" basis, most commonly antacid and acetaminophen
`tablets) able to fit into the compliance monitors.
`One hundred ninety-two patients agreed to participate and were
`randomly allocated to a pill count monitored (60 patients) or compli¬
`ance device monitored without feedback (67 patients) or with feed¬
`back (65 patients). Reasons given for not wishing to participate
`included "not interested" (48%), unable to keep the required monthly
`appointments (34%), and miscellaneous explanations (18%). Compli¬
`ance data were obtained from 112 patients in the two compliance
`monitored groups. Data were lost from 20 patients who withdrew
`from the study before receiving the first monitoring device or after
`accepting one monitor.
`the initially prescribed dose
`In addition,
`frequency for seven patients was changed by the health care provider
`during the course of the study. Compliance data collected on these
`patients were excluded from this study, although their inclusion had
`no effect on the results. This study therefore analyzes data obtained
`from 105 compliance device monitored patients.
`Protocol
`At the time of recruitment, all subjects were informed that the
`research project was being performed to study "methods of helping
`patients remember to take their blood pressure pills," that the con-
`
`Sawai (IPR2019-00789), Ex. 1052, p. 001
`
`

`

`tainer in which they would receive their medication would electroni¬
`cally record the time at which medication doses were removed, and
`that the compliance information might be given to their health care
`provider.
`The goal was to collect six consecutive months ofcompliance moni¬
`tor data from each patient for a total of 630 months. The final data,
`however, included 516 months (an average of 5 months per patient).
`Data were lost because of patients resigning from the project (13
`patients, 47 months), death (2 patients, 7 months), and compliance
`monitor malfunction (51 patients, 60 months).
`Compliance Monitor
`A detailed description of the compliance monitor has been pub¬
`lished.19 The following is a brief overview. Two clear plastic blister
`sheets, each containing 21 blisters, were filled with thé patient's
`medications. A self-adhering paper, with loops of conductive "wires"
`in the same pattern as the blisters, was then placed over the open face
`of each ofthe sheets of blisters to form blister packs. The packs were
`connected to electronic components and were placed inside an easily
`opened plastic case. Every 15 minutes, the battery-operated elec¬
`tronic memory sent an impulse through each loop of conductive
`material. If a dose of medication has been removed (that is, if the
`paper covering a blister was torn), the electrical impulse failed to
`return to the electronic memory and the 15-minute interval during
`which this occurred was recorded. After the patient returned the
`compliance monitor, the data were collected with a microcomputer.
`Questionnaires
`A questionnaire that solicited the following information was admin¬
`istered at the time ofrecruitment and at the conclusion ofcompliance
`monitor data collection: perceived problems taking medications
`(memory for taking pills, self-reported compliance, and attitude to¬
`ward pill taking, side effects, and medication efficacy), satisfaction
`with care, health locus of control, and health habits (coffee, alcohol,
`and cigarette use). In addition, during each monthly appointment, a
`questionnaire was administered that solicited the following informa¬
`tion about events in the preceding month: health status (nine ques¬
`tions addressing perceived psychological and physical health), inter¬
`im outpatient visits and hospitalizations, perceived number of
`omitted or late medication doses, and medication side effects. All
`questions except those relating to perceived problems taking medica¬
`tions were modified from previously defined scales.20,23 Blood pres¬
`sure was also recorded at each visit.
`Medication Compliance Definitions
`Many alternative definitions ofcompliance, using compliance moni¬
`tor data, can be developed to investigate the relationship between
`compliance and the prescribed daily dose frequency. To address the
`current problem, two definitions were selected.
`Definition 1 = (No. of doses removed/No. of doses prescribed) 100
`Definition 1 represents the percentage ofprescribed doses that the
`patient removed during the interval of observation (ie, the standard
`pill count). This definition can be calculated without using data from
`the compliance monitor and is independent of information about
`timing of dose removal. For example, definition 1 indicates perfect
`compliance even if the patient forgets doses and subsequently re¬
`moves all remaining doses on the day ofthe return clinic visit.
`Definition 2 equals percent of days during which less than the
`prescribed number, the prescribed number, and greater than the
`prescribed number of doses were removed.
`Definition 2 quantitates daily compliance without necessarily pe¬
`nalizing patients for removing more than one dose simultaneously
`(eg, to take a midday dose to work for later ingestion), but does
`penalize patients who do not remove the prescribed number of doses
`each day. A day was defined as the time interval between 3 AM and
`2:45 AM the following day because inspection of compliance data
`demonstrated that all patients began their medication day after 3 am.
`
`Table 1 .—Selected Characteristics ¡ the Total Patient
`Sample and in Each Prescribed Dose Frequency Category
`3 Times
`Total
`Daily,
`Sample,
`No. (%)
`No. (%)
`
`2 Times
`Dally,
`No. (%)
`
`1 Time
`Daily,
`No. (%)
`
`4 (10.0)
`3 (15.0)
`4 (8.9)
`12 (60.0) 21 (52.5) 29 (64.4)
`5 (25.0) 15 (37.5)
`11 (24.4)
`(2.2)
`. 1
`
`9 (22.2)
`2 (10.0)
`7 (15.6)
`18 (90.0) 31 (77.5) 38 (84.4)
`10 (50.0) 17 (42.5) 23 (51.1)
`4 (20.0) 11 (27.5) 14 (31.1)
`6 (30.0) 12 (30.0)
`7 (15.6)
`(2.2)
`. 1
`5 (11.1)
`24 (53.3)
`
`(5.0)
`5 (12.5)
`1
`11 (55.0) 27 (67.5)
`(25.0)
`5 (12.5)
`(15.0)
`3 (7.5)
`
`11 (24.4)
`4 (8.9)
`(2.2)
`1
`
`3 (15.0) 12 (30.0)
`19 (42.2)
`14 (70.0) 23 (57.5)
`17 (37.8)
`3 (15.0)
`8 (17.8)
`5 (12.5)
`(2.2)
`. 1
`
`6 (30.0) 10 (25.0)
`8 (40.0) 18 (45.0)
`6 (30.0)
`11 (27.5)
`(2.5)
`1
`
`10 (22.2)
`19 (42.2)
`12 (26.7)
`4 (8.9)
`
`10 (50.0) 20 (50.0)
`10 (50.0) 20 (50.0)
`
`22 (48.9)
`23 (51.1)
`
`Characteristic
`Age, y
`11 (10.5)
`<50
`62 (59.0)
`51-65
`31 (29.5)
`>65
`(1.0)
`No data
`1
`Race
`18 (17.1)
`White
`87 (82.9)
`Black
`Education
`50 (47.6)
`<High school
`High school graduate 29 (27.6)
`Some college
`25 (23.8)
`(1.0)
`No data
`1
`Marital status
`Never married
`Married
`Divorced,
`separated
`Widowed
`No data
`Employment
`Employed
`Retired
`Unemployed
`No data
`Annual income
`<$5000
`$5000-
`$14 999
`>$15 000
`No data
`Feedback
`Yes
`No
`
`11 (10.5)
`62 (59.0)
`
`21 (20.0)
`10 (9.5)
`(1.0)
`1
`
`34 (32.4)
`54 (51.4)
`16 (15.2)
`(1.0)
`1
`
`26 (24.8)
`45 (42.9)
`29 (27.6)
`5 (4.8)
`52 (49.5)
`53 (50.5)
`
`Statistical Analyses
`Significant differences in average compliance (defined by two mea¬
`sures) among the three dose frequencies were detected with Bonfer¬
`roni t tests and Scheffe's tests. Similar bivariate tests were per¬
`formed for the relationship between compliance (definition 2) and
`demographic, health status, and attitudinal variables. A stepwise
`regression was used to assess the multivariate relationship between
`compliance (dependent variable, definition 2) and prescribed daily
`dose frequency, sociodemographic characteristics, and scale scores
`derived from the questionnaire.
`RESULTS
`The study population was male,
`the median age was 61
`years, 83% were black, half had graduated from high school,
`most were married or widowed, half were retired and one
`third were employed, and median income was less than
`$15 000 annually (Table 1). No significant differences in socio¬
`demographic attributes or the proportion of patients whose
`health care providers received feedback about their compli¬
`ance were found among the three-dose frequency groups.
`Mean patient compliance (definition 1) was higher on a
`once- (96.0%) or twice- (93.0%) daily dose regimen compared
`with a three (83.8%) times a day dose regimen (P<.05) (Table
`2). Once- and twice-daily regimens were not significantly
`different from each other.
`More detailed insight into compliance was obtained by
`examining mean daily compliance (definition 2) (Table 3). The
`percent ofdays on which the prescribed number of doses were
`removed increased dramatically with decreasing dose fre-
`
`Sawai (IPR2019-00789), Ex. 1052, p. 002
`
`

`

`Table 2.—Mean Percent of Prescribed Doses Removed From
`Compliance Monitors (Definition 1 ) as a Function of
`Prescribed Daily Dose Frequency
`Prescribed Daily
`Mean Compliance in Percent
`No. of
`(SD)
`Dose Frequency
`Patients
`96.0 (6.9)
`45
`One
`93.0 (11.9)
`40
`Two
`83.8 (15.3)*
`20
`Three
`"Three times daily is significantly different from both once and twice daily
`(P<.05). Once- and twice-daily groups are not significantly different from each
`other.
`
`Table 3.—Mean Percent of Medication Days During Which
`the Indicated Proportion of Prescribed Doses Were Removed
`From Compliance Monitors (Definition 2) as a Function of
`Prescribed Daily Dose Frequency
`Percent (SD) of
`Days With Compliance
`Prescribed
`Daily Dose
`No. of -"-
`Frequency
`100
`100
`Patients
`100
`13.0 (18.4)
`83.6 (19.9)
`3.3 (2.7)
`One
`45
`5.7 (5.6)*
`19.2 (18.0)
`74.9 (20.2)
`40
`Two
`36.1 (29.1 )t
`59.0 (30.2)t
`4.8 (3.9)
`20
`Three
` Twice daily is significantly different from once daily (P<,05).
`tThree times daily is significantly different from both once and twice daily
`(P<.05). Once- and twice-daily groups are not significantly different from each
`other.
`
`quency. Thus, for patients on a thrice, twice, or once daily
`antihypertensive medication dose regimen,
`the prescribed
`number ofdoses were removed on 59.0%, 74.9%, and 83.6% of
`days, respectively. Contrariwise,
`the percent of days on
`which less than the prescribed number ofdoses were removed
`decreased with decreasing dose frequency (from 36.1% to
`19.2% to 13.0%). The three-times daily regimen was signifi¬
`cantly different from the once- and twice-daily regimens
`(P<.05) for both compliance estimates.
`One explanation for the inverse relationship between daily
`dose frequency and compliance is that patients in the three-
`dose frequency groups may have had different sociodemogra¬
`phic or health attitudes or different responses to feedback
`from their health care providers. Using a stepwise regression
`procedure, 28 sociodemographic and health attitude variables
`(see 'Subjects and Methods' section) and feedback group were
`examined to determine their relationship to compliance. The
`following factors predicted better
`compliance:
`once-
`(P<.0001) and twice- (P<.0001) daily medication regimens,
`higher income (P<.0002), education greater than elementary
`school (P<.007), living alone (P<.005), and being employed
`(P<.004).
`Thus, while patient attributes other than dose regimen do
`contribute to the observed differences in compliance, they do
`not negate the large effect ofdose regimen.
`COMMENT
`This study demonstrates that in a group of older, male
`patients receiving long-term treatment for hypertension in a
`continuity of care environment, once or twice daily dose
`regimens are associated with moderately improved monthly
`pill counts (definition 1, 95.2% and 93.0%, respectively) by
`comparison with a three-times daily regimen (83.8%). When
`
`compliance is defined as the percent of total medication days
`on which the prescribed number of doses are removed (defini¬
`tion 2), patients on a once-daily dose regimen were found to
`remove the prescribed number of doses on 83.6% of days,
`while patients on two- or three-times daily dose regimens
`removed the prescribed number of doses on only 74.9% and
`59.0% of days, respectively. The difference in compliance
`when measured with the pill count and daily methods dramat¬
`ically demonstrates the degree to which poor compliance is
`underestimated with the simple pill count method.
`Two caveats should be noted in the interpretation of this
`study. A large proportion ofeligible patients (76%) declined to
`participate and 13% ofpatients who initially agreed to partici¬
`pate subsequently withdrew. This suggests that selection
`bias may be present. Indeed, patients who volunteer to par¬
`ticipate in research projects are probably more compliant
`than those who refuse.24 The relationship between prescribed
`dose frequency and compliance in the ostensibly less compli¬
`ant nonparticipants cannot reliably be predicted.
`Because of the organization and treatment protocol of the
`hypertension clinic in which our research was performed, the
`stronger experimental design afforded by a randomized con¬
`trolled trial was not possible.
`Instead, subjects who were
`already under treatment were recruited and known or alleged
`covariates of noncompliance (in addition to prescribed medi¬
`cation frequency) were compared with dose frequency
`groups. The analysis supports the conclusion that improved
`compliance is a true effect of decreased dose frequency.
`Only one other study has used an electronic monitor to
`examine the relationship between prescribed dose frequency
`and compliance. Cramer et al,10 measuring compliance with
`data derived from pill containers that recorded the time and
`date of bottle cap removal and replacement, found that com¬
`pliance (defined using definition 2) for medications prescribed
`once, twice, three, or four times daily was 87%, 81%, 77%, and
`39%, respectively. Statistically significant differences were
`identified only for the daily vs four times daily and twice daily
`vs three times daily regimens, perhaps because of their small
`sample size (24 patients).
`In conclusion, our study demonstrated that as the daily
`prescribed dose frequency decreased from three times to once
`daily, medication compliance (defined as the proportion of
`days on which the prescribed number of doses were removed
`from the compliance monitor) improved by 42%. A large
`number of medications are currently available that are effec¬
`tive when administered once or twice daily. Therefore, proba¬
`bly the simplest and single most important action that health
`care providers can take to improve compliance is to select
`medications that permit the lowest daily dose frequency
`possible.
`This research was supported by the Health Services Research and Develop¬
`ment Service, Department of Veterans Affairs, Washington, DC (HSR&D
`grant 618).
`We thank the Department of Veterans Affairs Hypertension Screening
`Treatment Program (HSTP) and Kathleen Wenzel, RN, and Judy Martin, RN,
`of the St Louis (Mo) HSTP clinic for their consistent support of this reseach
`project. We also thank Ray Godefroid, RPh, and Bryant Butler for loading and
`dispensing medications in the compliance monitors, and Patricia Giles for her
`consistently excellent secretarial support.
`References
`1. Eraker S, Kirscht JP, Becker MH. Understanding and improving patient
`compliance. Ann Intern Med. 1984;100:258-268.
`2. Gatley MS. To be taken as directed. JR Coll Gen Pract. 1968;16:39-44.
`3. Ayd FJ Jr. Once-a-day neuroleptic and tricyclic antidepressant therapy.
`Int Drug Ther Newslet. 1972;7:33-40.
`
`Sawai (IPR2019-00789), Ex. 1052, p. 003
`
`

`

`4. Wandless I, Mucklow JC, Smith A, Prudham D. Compliance with pre-
`scribed medicines: a study of elderly patients in the community. J R Coll Gen
`Pract. 1979;29:391-396.
`5. Tinkelman DG, Vanderpool GE, Carroll MS, Page EG, Spangler DL.
`Compliance differences following administration of theophylline at six- and
`twelve-hour intervals. Ann Allergy. 1980;44:283-286.
`6. Fujii J, Akira S. Compliance and compliance-improving strategies in
`hypertension: the Japanese experience. J Hypertens. 1985;3(suppl):19-22.
`7. Cockburn J, Reid AL, Bowman JA, Sanson-Fisher RW. Effects of inter-
`vention on antibiotic compliance in patients in general practice. Med J Aust.
`1987;147:324-328.
`8. Cockburn J, Gibberd RW, Reid AL, Sanson-Fisher RW. Determinants of
`non-compliance with short term antibiotic regimens. BMJ. 1987;295:814-818.
`9. Pullar T, Birtwell AJ, Wiles PG, Hay A, Seely MP. Use of a pharmacolog-
`ic indicator to compare compliance with tablets prescribed to be taken once,
`twice, or three times daily. Clin Phamacol Ther. 1988;44:540-545.
`10. Cramer JA, Mattson RH, Prevey ML, Scheyer RD, Ouellette VL. How
`often is medication taken as prescribed? A nouvel assessment technique.
`JAMA. 1989;261:3273-3277.
`11. Clinite JC, Kabat HF. Prescribed drugs.. . errors during self-adminis-
`tration. JAm Pharmacol Assoc. 1969;NS9:450-452.
`12. Porter AMW. Drug defaulting in a general practice. BMJ. 1969;1:
`218-222.
`13. General Practitioner Research Group. General practitioner clinical tri-
`als. Practitioner. 1970;204:719-723.
`14. Hulka BS, Cassel JC, Kupper LL. Disparities between medications
`prescribed and consumed among chronic disease patients. In: Lasagna L, ed.
`Patient Compliance. Mount Kisco, NY: Futura Publishing Co; 1976:
`
`123-141.
`15. Widmer RB, Cadoret RJ, Troughton E. Compliance characteristics of
`291 hypertensive patients from a rural midwest area. J Fam Pract.
`1983;17:619-625.
`16. Stewart M. The validity of an interview to assess a patient's drug taking.
`Am JPrev Med. 1987;3:95-100.
`17. Leirer VO, Morrow DG, Pariante GM, Sheikh JI. Elders' nonadherence:
`its assessment, and computer assisted instruction for medication recall train-
`ing. JAm Geriatr Soc. 1988;36:877-884.
`18. Gordis L. Conceptual and methodological problems in measuring patient
`compliance.
`In: Haynes RB, Taylor DW, Sackett DL, eds. Compliance in
`Health Care. Baltimore, Md: The Johns Hopkins University Press; 1979:23-45.
`19. Eisen SA, Hanpeter JA, Kreuger LW, Gard M. Monitoring medication
`compliance: description of a new device. J Comp Health Care. 1987;2:131-142.
`20. Brook RH, Ware JE, Davies-Avery A, et al. Overview of adult health
`status measures fielded in Rand's Health Insurance Study. Med Care.
`1979;17(suppl 7):1-131.
`21. Cope DW, Linn LS, Leake BD, Barrett PA. Modification of residents'
`behavior by preceptor feedback of patient satisfaction. J Gen Intern Med.
`1986;1:394-398.
`22. Haynes RB, Taylor DW, Sackett DL, Gibson ES, Bernholz CD, Mukher-
`jee J. Can simple clinical measurements detect patient noncompliance? Hyper-
`tension. 1980;2:757-764.
`23. Wallston BS, Wallston KA, Kaplan GD, Maides SA. Development and
`validation of the health locus of control (HLC) scale. J Consult Clin Psychol.
`1976;44:580-585.
`24. Ramsay JA. Participants in noncompliance research: compliant or non-
`compliant? Med Care. 1982;20:615-622.
`
`Sawai (IPR2019-00789), Ex. 1052, p. 004
`
`