The Journal of Rheumatology
`
`Volume 30, no. 1
`
`Patient compliance in rheumatoid arthritis, polymyalgia rheumatica, and gout.
`
`Erik de Klerk, Désirée van der Heijde, Robert Landewé, Hille van der Tempel, John Urquhart
`and Sjef van der Linden
`
`J Rheumatol 2003;30;44-54
` http://www.jrheum.org/content/30/1/44
`
`
`
`1. Sign up for TOCs and other alerts
`http://www.jrheum.org/alerts
`
`2. Information on Subscriptions
`http://jrheum.com/faq
`
`3. Information on permissions/orders of reprints
`http://jrheum.com/reprints_permissions
`
`The Journal of Rheumatology
` is a monthly international serial edited by Earl D.
`Silverman featuring research articles on clinical subjects from scientists working in
`rheumatology and related fields.
`
`
`
`RheumatologyRheumatology
`
` Published by Published by The Journal ofThe Journal of
`
`
`
`
`
`
`
`Sawai (IPR2019-00789), Ex. 1050, p. 001
`
`

`

`2002-210-1
`
`Patient Compliance in Rheumatoid Arthritis,
`Polymyalgia Rheumatica, and Gout
`
`ERIK de KLERK, DÉSIRÉE van der HEIJDE, ROBERT LANDEWÉ, HILLE van der TEMPEL, JOHN URQUHART,
`and SJEF van der LINDEN
`
`ABSTRACT. Objective. (1) To explore patient compliance with prescribed drug regimens in the setting of usual care
`for outpatients with rheumatoid arthritis (RA), gout, and polymyalgia rheumatica (PMR) by utilizing
`electronic medication event monitors (MEMS®) to register openings of the medication package. (2) To
`examine the influence of disease, frequency of intake of the drug, and class of drug on compliance. (3)
`To explore the influence of demographic factors, quality of life measures, coping, health status, and
`functional ability as potential predictors of patient compliance.
`Methods. A total of 127 consenting consecutive patients were enrolled: 81 patients with RA, 33 taking
`nonsteroidal antiiflammatory drugs (13 diclofenac TID and 20 naproxen BID) and 48 taking disease
`modifying antirheumatic drugs [25 sulfasalazine (SSZ) BID and 23 methotrexate (MTX) once weekly];
`17 patients with PMR starting with prednisolone QD; and 29 patients with gout starting with colchicine
`(12, QD) or starting with uric acid lowering agents (17, QD). All patients received first prescriptions
`and were instructed to take the medication as prescribed. Followup was 6 months (gout 12 mo). All
`patients were aware of the monitoring capability of the package. At baseline a series of questionnaires
`was completed. We summarized the dosing histories as “taking compliance” (percentage of total pre-
`scribed doses taken), “correct dosing” (percentage of doses taken as prescribed), and “timing compli-
`ance” (percentage of doses taken within +/– 25% of prescribed interdose intervals).
`Results. A total of 26,685 days (> 73 patient-years) were monitored. Compliance expressed as “taking
`compliance,” mean (95% CI), “correct dosing,” mean (95% CI), and “timing compliance,” mean (95%
`CI) are: naproxen: 82% (75–90), 68% (57–80), 48% (34–61); diclofenac: 77% (61–93), 67% (47–87),
`39% (21–57); MTX: 107% (98–117), 81% (75–87), 83% (76–90); SSZ: 72% (60–84), 55% (44–67),
`25% (18–33); prednisolone: 96% (89–102), 88% (83–92), 82% (74–89); colchicine: 65% (48–81), 44%
`(26–62), 32% (18–46); and uric acid lowering agents: 84% (76–92), 74% (63–85), 65% (52–79). Missed
`doses occurred more frequently than taking of extra doses: in RA, on 10% of all monitored days there
`was no evidence of dosing, while on 3% of all monitored days extra doses were taken. In PMR and gout
`these data are 10% and 4%, and 15% and 7%, respectively. We observed a decline of compliance over
`time in all study medication groups. Multiple regression analyses showed that the class of medication
`(symptom modifying or disease controlling), the dosing frequency, the patient’s sex, coping pattern
`(avoidance, passive reaction pattern, and expression of emotions), and the overall health (total
`Nottingham Health Profile score) together explained 67% of the variance in taking compliance (adjust-
`ed R2) (p = 0.002).
`Conclusion. Studying patient compliance with prescribed drug regimens utilizing electronic medication
`event monitors in RA, gout, and PMR showed that large differences exist in compliance between the
`various medication groups. Compliance declines over time. A regression model shows that it is possi-
`ble to relate differences in patient compliance to a number of medication and patient related factors.
`(J Rheumatol 2003;30:44–54)
`
`Key Indexing Terms:
`PATIENT COMPLIANCE
`POLYMYALGIA RHEUMATICA
`
`RHEUMATOID ARTHRITIS
`
`GOUT
`DRUG THERAPY
`
`From the Department of Internal Medicine, Division of Rheumatology,
`University Hospital Maastricht, Maastricht, The Netherlands; Limburg
`University Center, Diepenbeek, Belgium; Department of Rheumatology,
`Maasland Hospital Sittard, Sittard; and Department of Epidemiology,
`Maastricht University, Maastricht, The Netherlands.
`Supported by grant NR 831 from the Dutch Arthritis Association
`(Nederlands Reumafonds).
`E. de Klerk, MD, MSc, Scientific Researcher; R. Landewé, MD, PhD,
`Rheumatologist; S. van der Linden, MD, PhD, Professor of
`Rheumatology, Department of Internal Medicine, Division of
`Rheumatology, University Hospital Maastricht; D. van der Heijde, MD,
`
`PhD, Professor of Rheumatology, Department of Internal Medicine,
`Division of Rheumatology, University Hospital Maastricht, and Limburg
`University Center; H. van der Tempel, MD, Rheumatologist, Department
`of Rheumatology, Maasland Hospital Sittard; J. Urquhart, MD, FRCP
`(Edin), Professor of Pharmaco-Epidemiology, Department of
`Epidemiology, Maastricht University.
`Address reprint requests to Dr. D. van der Heijde, Division of
`Rheumatology, University Hospital Maastricht, PO Box 5800, 6202 AZ
`Maastricht, The Netherlands. E-mail: dhe@sint.azm.nl
`Submitted March 6, 2002; revision accepted June 12, 2002.
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`The Journal of Rheumatology 2003; 30:1
`
`Rheumatology
`
`44
`
`Sawai (IPR2019-00789), Ex. 1050, p. 002
`
`

`

`Compliance with treatment guidelines or standards by health
`professionals and compliance with prescribed drug regimens
`by patients are major determinants of outcome1,2. In daily
`practice reduced compliance is a well known but poorly
`understood phenomenon. Data on drug regimen compliance
`by patients in rheumatology are scarce. A study among
`patients with ankylosing spondylitis revealed that deviations
`from the prescribed once-daily regimen of a nonsteroidal anti-
`inflammatory drug (NSAID) occurred frequently, even in the
`setting of a randomized clinical trial3.
`We investigated patient compliance with prescribed drug
`regimens in the setting of usual care for outpatients with one
`of 3 rheumatological diseases: rheumatoid arthritis (RA),
`gout, or polymyalgia rheumatica (PMR). The dosing frequen-
`cy called for by the drug regimens varied between “3 times
`daily” and “once weekly.” The drugs prescribed differ in their
`actions. Some have “direct symptom-modifying effects,” oth-
`ers are intended for use as “late onset disease-controlling ther-
`apy” or “preventive therapy.” Prednisolone was used as a drug
`that has both symptom-modifying and disease-controlling
`effects.
`We used electronic medication event monitoring devices to
`document patient compliance with drug therapy, because this
`method addresses several aspects of patient compliance4,5.
`This method is widely considered the standard for compiling
`drug dosing histories of ambulatory patients6-8.
`We describe compliance with naproxen, diclofenac, sul-
`fasalazine (SSZ), and methotrexate (MTX) in RA, prednisone
`in PMR, and colchicine, allopurinol and uricosurica in gout.
`We examine the influence of disease, frequency of intake of
`the drug, and indication (direct effects versus late onset of
`effectiveness) on patient compliance. In addition we explore
`the influence of a number of demographic factors, quality of
`life measures, coping, health status, and functional ability as
`potential predictors of patient compliance.
`
`MATERIALS AND METHODS
`The study was conducted as a series of cohort studies. We included all con-
`secutive consenting outpatients with a diagnosis by a rheumatologist of RA,
`PMR, or gout at the outpatient rheumatology clinics of University Hospital
`Maastricht, Atrium Hospital Heerlen, and Maasland Hospital Sittard, respec-
`tively, a secondary/tertiary and 2 secondary referral centers for rheumatology.
`For all studies, approval was obtained from the Medical Ethical Committee of
`all 3 hospitals.
`Patients with RA were to be included when the rheumatologist prescribed
`SSZ (BID, after up-titration) or oral MTX (once weekly), or if the rheuma-
`tologist prescribed either diclofenac (TID or combined with misoprostol BID)
`or naproxen (BID). Patients with a diagnosis of PMR were included if they
`received prednisone or prednisolone QD. In the analyses, patients taking
`prednisone and prednisolone were combined and we will continue to use the
`term prednisolone for this group. Patients with a diagnosis of gout were
`included if the rheumatologist prescribed longterm prophylactic maintenance
`therapy with colchicines (QD) or uric acid lowering agents such as allopuri-
`nol or benzbromaron (QD). In the analyses, patients taking allopurinol or
`benzbromaron were combined in a single group, called uric acid lowering
`agents.
`All prescriptions in all diagnoses had to be first prescriptions (which did
`not necessarily mean a newly determined diagnosis) and had to be written “to
`
`be taken as directed” (not “on demand”). We further required that the treating
`rheumatologist expected that drug treatment would continue for at least 6
`months.
`To measure patient compliance we used the Medication Event Monitoring
`System (MEMS®, Aardex, Zug, Switzerland). It consists of a cup-type med-
`ication container with a threaded, screw-cap closure. Within the closure is
`microelectronic circuitry to record time and date of each opening and closing
`of the medication package. The method, with its advantages and disadvan-
`tages, has been discussed in detail6-12.
`The rheumatologist informed eligible patients about the purpose of the
`project and the operation of the MEMS. A demonstration was given of how
`the system worked. Patients were then asked to sign the informed consent
`document. Each patient then received a MEMS, and the patient’s pharmacist
`was notified by fax that the patient was entered in a research project and asked
`to transfer the prescribed medication to the MEMS container. Patients also
`received a set of questionnaires (see below), which they were asked to com-
`plete in the first week after start of the medication. All patients received a fol-
`lowup phone call by the investigator (EdK) about 3 days after the visit to the
`rheumatologist to answer questions, and to ensure that the medication was
`indeed transferred to the MEMS container.
`Six months after start of drug therapy (12 months in the patients with
`gout) or sooner if patient or rheumatologist stopped medication, patients were
`asked to complete a second set of questionnaires, identical to the first set, and
`to return the MEMS container to the rheumatologist or investigator. In addi-
`tion, patients were asked to provide a prescription drug history, which they
`obtained from their pharmacy. This is a computerized list containing all dates
`and drugs dispensed at the patient’s pharmacy. In The Netherlands the major-
`ity of patients are required to subscribe to one pharmacy, ensuring that most
`if not all dates of medication dispensing (and therefore extra openings) were
`recorded13. The data of the MEMS were downloaded via a MEMS-communi-
`cator to a Windows® based personal computer, and analyzed by software
`designed to analyze dosing histories (CSS version 2.1, Aardex, Zug,
`Switzerland).
`Each patient’s data were compared with the prescription drug history and,
`if available, remarks of the patient and, if necessary, days of special openings
`(such as pharmacy visits or if the patient had recorded openings unrelated to
`treatment). These extra openings were marked as a non-monitored period.
`This procedure ensures that the calculation of the compliance summary vari-
`ables (see below) is as free as possible of artifacts unrelated to actual med-
`ication taking.
`The dosing histories were transformed to the following categories.
`(1) Taking compliance: The percentage of prescribed doses taken, calculated
`as:
`
`(total number of recorded medication events /
`total number of prescribed doses) × 100%
`
`Example: a patient opened and closed the MEMS container 170 times while
`prescribed SSZ BID for a monitored period of 100 days, so taking compliance
`= (170 / 200) × 100% = 85%.
`Taking compliance is useful as an overall compliance variable. However,
`it is rather crude, as no information on the timing of doses is incorporated, and
`omitted doses occurring at one time can be obscured by extra doses taken at
`another time.
`(2) Correct dosing: The percentage of days within which the correct number
`of doses were taken, calculated as:
`(total number of days with recorded medication events as prescribed /
`total number of monitored days) × 100%
`Example: a patient who had been prescribed SSZ BID has a dosing history,
`compiled by the MEMS, that showed 170 medication events during a moni-
`tored period of 100 days, but only 58 of the monitored days showed 2 med-
`ication events. Thus, correct dosing = (58 / 100) × 100% = 58%.
`Correct dosing is a useful variable to determine actual day-by-day drug
`use. It incorporates day-by-day variability in dosing, and is not influenced by
`“catch-up” dosing. It is stricter than taking compliance.
`(3) Timing compliance: We allowed the patients to vary the interdose-inter-
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`
`Rheumatology
`
`de Kl
`
`45
`
`Sawai (IPR2019-00789), Ex. 1050, p. 003
`
`

`

`vals within an arbitrarily chosen plus or minus 25%. Thus, for a QD regimen,
`the prescribed interval is 24 hours, but we allowed intervals of 18–30 hours.
`Similarly, for a BID regimen we allowed intervals of 9–15 hours, for a TID
`regimen we allowed intervals of 6–10 hours, and for a once-weekly regimen
`we allowed intervals of 126–210 hours. Timing compliance was then calcu-
`lated as:
`
`(the number of interdose-intervals of allowed duration /
`number of prescribed interdose-intervals) × 100%
`
`Note that if there are missed doses, the number of interdose-intervals is by
`definition lower than the number of prescribed interdose-intervals, so timing
`compliance does not necessarily add up to 100%.
`Example: a patient who had been prescribed SSZ BID has a dosing history,
`compiled by the MEMS, of 170 medication events during a monitored period
`of 100 days, but only 45 of all interdose-intervals were between 18 and 30
`hours’ duration. Timing compliance is: (45 / 199) × 100% = 22.6%.
`Timing compliance determines interdose-intervals, which, when exces-
`sively long, may indicate periods of time when drug action was subtherapeu-
`tic or absent. It is a stricter measure of compliance with the prescribed drug
`regimen than correct dosing.
`Questionnaires. The questionnaires consisted of some demographic ques-
`tions: age, sex, education (low = primary school, intermediate = secondary
`school, high = further education), profession (employed or not), and social
`support (living alone, with partner, with partner and children). We also asked
`patients to complete the Health Assessment Questionnaire (HAQ)14,
`Nottingham Health Profile (NHP)15, Utrecht Coping List (UCL)16, European
`Quality of Life measure (EuroQol)17, Long Term Medication Behavior Self-
`Efficacy Scale (LTMBS)18, a self-composed list of 40 frequent side effects,
`and for RA patients only, the Rheumatoid Arthritis Quality of Life measure
`(RAQol)19-21.
`HAQ scores range from 0 (minimum) to 3 (maximum)14. The NHP scores
`were summed and computed into 6 subscales: energy, pain, emotional reac-
`tions, sleep, social isolation, and physical mobility15. For the UCL, 7 sub-
`scales were computed: active attitude, palliative reaction, avoidance, seeking
`social support, passive reaction pattern, expression of emotions, and comfort-
`ing thoughts16. The EuroQol describes health status in 3 levels: 1 = no prob-
`lem, 2 = some problems, 3 = extreme problems. It also includes a self-rated
`thermometer indicating the patient’s own assessment of their health state17.
`The LTMBS is a 26 item questionnaire designed to measure self-efficacy for
`patients undergoing chronic drug therapy. The results were summed and cal-
`culated to a scale ranging from 0 (lowest possible self-efficacy) to 100 (high-
`est possible self-efficacy)18,22. The RAQol ranges from 0 (worst possible qual-
`ity of life) to 30 (perfect quality of life)19-21.
`As no standard instrument was available at the time of the start of the
`study to document side effects of drug treatment in rheumatology, we
`devised a measure with 40 questions for the most common side effects asso-
`ciated with naproxen, diclofenac, SSZ, MTX, prednisone, colchicine, allop-
`urinol, and benzbromaron. The frequencies of side effects were based on US
`FDA approved labeling for each of these products, as compiled in the
`Physicians Desk Reference23. Each question consisted of 2 parts: occurrence
`(never = 0, sometimes = 1, frequently = 2, often = 3, always = 4) and, if the
`answer was anything other than “never,” patients were asked to rate the
`severity on a range from 1 (not disturbing at all) to 5 (very disturbing). A fre-
`quency of side effects score was computed by summing the 40 items of
`occurrence into one variable (range 0 = no side effects at all, 160 = maxi-
`mum frequency of side effects score). In addition, a total side effects score
`was calculated as (occurrence × frequency), ranging from 0 (no side effects
`at all) to a maximum of 800 (maximum occurrence and severity of side
`effects).
`Statistics. Analyses consisted of descriptive statistics (means, standard devia-
`tions, 95% confidence intervals), Pearson’s correlation coefficients, (step-
`wise) multiple regression analyses with adjustment for multiple variable test-
`ing, one-way analysis of variance with the Scheffé multiple comparison test
`for post-hoc analysis, and, where appropriate, nonparametric alternatives. All
`analyses were performed using SPSS version 10.0.7 for Windows.
`
`2002-210-3
`
`RESULTS
`Compliance on the individual level
`It is often useful to convert the dosing histories from the elec-
`tronic monitors to calendar and chronology plots for a quick
`overview of the patient’s dosing history. The calendar plot (an
`example is shown in Figure 1) shows the number of recorded
`doses on each day of the study period. It is helpful to identify
`periods in which dosing was not optimal, and to correlate clin-
`ical events (such as the occurrence of flares, gout attacks, or
`specific adverse drug reactions) to specific dates. However,
`the calendar plot does not give details on within-day timing of
`drug intake, and only roughly shows changes in the patient’s
`dosing pattern over time. Such information is shown by the
`chronology plot (4 examples are shown in Figure 2). From
`these plots it becomes apparent that patient compliance on
`drug therapy is a day-by-day phenomenon, which in some
`instances is difficult to grasp in a single summary variable.
`
`Overall compliance results
`We studied 127 consenting consecutive patients of the outpa-
`tient clinic. They consisted of 81 patients with RA using
`NSAID (13 diclofenac and 20 naproxen) or DMARD (25 SSZ
`and 23 MTX), 17 patients with PMR taking prednisone, and
`29 patients with gout taking colchicine (12) or allopurinol (10)
`or benzbromaron (7). A total number of 26,685 days were
`monitored (> 73 patient-years). The mean followup was 210
`days. Table 1 summarizes the demographic data.
`RA — NSAID. Figure 3 depicts the taking compliance, correct
`dosing, and timing compliance of the various drugs between
`the 3 diagnoses, along with the corresponding 95% confi-
`dence intervals. There are clear and statistically significant
`differences between the drugs (Table 2). Compliance reports
`with naproxen and diclofenac were comparable, with a taking
`compliance of 82% and 76%, correct dosing of 68% and 67%,
`and timing compliance of 48% and 39%, respectively.
`RA — DMARD. There were large differences between the
`DMARD, however. Taking compliance for SSZ was 72%, for
`MTX 107%. This difference was statistically significant (p <
`
`Figure 1. Example of a calendar plot. This gout patient, prescribed allopuri-
`nol QD, told us that he likes to go out on the weekends and thought that allop-
`urinol and alcohol did not go together well. Hence he would not take it on
`most Saturdays. See Figure 2 for the chronology plot (Patient 2046).
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`The Journal of Rheumatology 2003; 30:1
`
`Rheumatology
`
`46
`
`Sawai (IPR2019-00789), Ex. 1050, p. 004
`
`

`

`Figure 2. Example of 4 chronology plots. Patient 1089 is a patient with RA taking BID SSZ. She is taking almost 100% of the drugs, but is taking the doses in
`relatively short intervals, resulting in a lower timing compliance (25%). Patient 2046 is a gout patient taking QD allopurinol (see Figure 1), frequently missing
`doses on weekends. Patient 1005 is a patient with PMR prescribed QD prednisone. She frequently takes extra doses, but hardly ever misses a dose. Patient 1029
`is a gout patient taking BID maintenance therapy with colchicine. Even though taking compliance is good at 84%, he is taking the correct number of doses on
`60% of the days, and only 20% of all doses are within the prescribed interdose-interval.
`
`Table 1. Demographic data.
`
`Age, yrs, mean (SD)
`Sex, % female
`Social support, %
`Single
`Married/living together
`without children
`Married/living together
`with children
`Education, %
`Low
`Intermediate
`High
`Work,
`% working
`
`RA,
`n = 81
`
`60 (14)
`66
`
`PMR,
`n = 17
`
`72 (7)
`76
`
`Gout,
`n = 29
`
`58 (12)
`20
`
`29
`
`64
`
`7
`
`28
`64
`7
`
`26
`
`24
`
`70
`
`6
`
`24
`71
`6
`
`12
`
`17
`
`80
`
`3
`
`17
`80
`3
`
`54
`
`0.001). A comparable picture emerges from the comparison of
`correct dosing and timing compliance between the DMARD
`— correct dosing: SSZ 55%, MTX 81% (p < 0.001); and tim-
`ing compliance: SSZ 25%, MTX 83% (p < 0.001).
`PMR. Compliance with prednisolone among PMR patients
`was high: taking compliance 96%, correct dosing 88%, and
`timing compliance 82%. Confidence intervals around the
`mean were relatively small compared to other drugs, indicat-
`ing little interpatient variability.
`Gout. The compliance of PMR patients prescribed systemic
`steroids contrasted quite sharply with the compliance of the
`gout patients. In particular, compliance for maintenance
`colchicine therapy was strikingly low: taking compliance
`65%, correct dosing 44%, and timing compliance 32%.
`Compliance with the combined uric acid lowering agents was
`better: taking compliance 84%, correct dosing 74%, and tim-
`ing compliance 54%.
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`
`Rheumatology
`
`de Kl
`
`47
`
`Sawai (IPR2019-00789), Ex. 1050, p. 005
`
`

`

`2002-210-5
`
`Figure 3. A. Taking compliance. B. Correct dosing. C. Timing compliance.
`
`Figure 4 shows the distribution of taking compliance, cor-
`rect dosing, and timing compliance when the proportion of
`patients is plotted against compliance organized in categories
`of 10% each. The distribution follows the previously
`described “typical J-shaped compliance distribution”4,24. It is
`clear, however, that for correct dosing and timing compliance,
`the peak of the J-shaped distribution is further skewed to the
`left, indicating a larger number of patients whose compliance
`is suboptimal. We also rank-ordered individual patients
`according to taking compliance, correct dosing, and timing
`
`compliance (Figure 5). As expected, in general, timing com-
`pliance is worse than correct dosing, which in turn is worse
`than taking compliance.
`Missed doses occurred more frequently than taking of
`extra doses: in RA, on 10% of all monitored days there was no
`evidence of dosing, while on 3% of all monitored days extra
`doses were taken. In PMR, missed and extra doses were 10%
`and 4% of all monitored days, and in gout they were 15% and
`7%, respectively. We further divided missed doses into “occa-
`sionally missed” (periods of 1 or at most 2 consecutive days
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`The Journal of Rheumatology 2003; 30:1
`
`Rheumatology
`
`48
`
`Sawai (IPR2019-00789), Ex. 1050, p. 006
`
`

`

`Table 2. Diagnosis, patient commpliance, and drug regimen.
`
`Diagnosis Drug
`
`Dosing
`Frequency
`
`Indication
`
`Taking
`Symptom
`Modifying/ Compliance,
`Disease
`%
`Controlling Drug
`
`Correct
`Dosing,
`%
`
`Timing
`Compliance,
`%
`
`RA
`
`PMR
`
`Gout
`
`BID
`
`TID
`
`BID
`
`Once
`weekly
`QD
`
`QD
`
`QD
`
`Naproxen
`(n = 20)
`Diclofenac
`(n = 13)
`Sulfasalazine
`(n = 25)
`Methotrexate
`(n = 23)
`Prednisolone
`(n = 17)
`Colchicine
`(n = 12)
`Allopurinol/
`benzbromaron
`(n = 17)
`
`* One-way ANOVA.
`
`Pain
`
`Pain
`
`SM
`
`SM
`
`Inflammation
`
`DC
`
`Inflammation
`
`DC
`
`Inflammation SM + DC
`
`Preventive
`
`Lower urate
`
`DC
`
`DC
`
`82
`
`76
`
`72
`
`107
`
`96
`
`65
`
`84
`
`68
`
`67
`
`55
`
`81
`
`88
`
`44
`
`74
`
`48
`
`39
`
`25
`
`83
`
`82
`
`32
`
`62
`
`F = 7.13,
`p < 0.001*
`
`F = 5.98
`p < 0.001*
`
`F = 19.1,
`p < 0.001*
`
`Figure 4. Distribution of taking compliance, correct dosing, and timing compliance.
`
`without dosing) and “drug holidays” (periods of 3 or more
`consecutive days without dosing)25,26. Drug holidays were not
`computed for the patients taking MTX because of the weekly
`dosing regimen. There were a total of 192 drug holidays.
`Patients taking SSZ, colchicine, and uric acid lowering agents
`had the highest frequency of drug holidays (2.4–2.5 per
`patient), while all other groups had roughly 1–1.2 drug holi-
`days per patient.
`There were clear and statistically significant differences in
`
`compliance between the 4 dosing regimens (Figure 6).
`Compliance with once-weekly (all RA patients taking MTX)
`was the best, followed by QD, then BID and TID, for taking
`compliance, correct dosing, and timing compliance. Three
`separate ANOVA showed that these differences were statisti-
`cally significant (all p < 0.001).
`All groups showed a gradual and large decline of compli-
`ance over time. Comparison of compliance in the first month
`versus compliance in the 6th month showed an overall decline
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`
`Rheumatology
`
`de Kl
`
`49
`
`Sawai (IPR2019-00789), Ex. 1050, p. 007
`
`

`

`2002-210-7
`
`show any statistically significant differences. These results
`indicate that functional capacity, as measured by HAQ total
`score, seem not to be associated with patient compliance.
`Overall health profile. The NHP total score at baseline was 11.6
`± 7.8. There were no differences between the drugs or the dis-
`eases. The NHP baseline subscores were as follows: energy
`0.03 ± 0.03; pain 0.30 ± 0.20; emotional reactions 0.16 ± 0.21;
`social isolation 0.02 ± 0.04; sleep 0.09 ± 0.08; and physical
`mobility 0.23 ± 0.18. Neither taking compliance, correct dos-
`ing, nor timing compliance were associated with the total score,
`although the subcategories were statistically significantly asso-
`ciated with taking compliance (F = 2.50, p = 0.03), correct dos-
`ing (F = 2.30, p = 0.04), and timing compliance (F = 2.08, p =
`0.06). These findings suggest that some NHP baseline sub-
`scores are predictive of compliance during the study period.
`Coping. The UCL subscores were (mean ± SD): active atti-
`tude 17.5 ± 5.6; palliative reaction 18.5 ± 5.4; avoidance 17.1
`± 5.4; seeking social support 12.7 ± 4.4; passive reaction pat-
`tern 11.3 ± 3.9; expression of emotions 5.5 ± 2.3; and com-
`forting thoughts 13.1 ± 3.2. There were no differences
`between UCL subscores within the drugs, diseases studied, or
`categories of compliance.
`Perceived health status. There were no between-disease dif-
`ferences in baseline overall health status as measured by the
`EuroQol visual analog scale (VAS). Compliance scores
`between patients who rated their health status as worse during
`the prestudy period were statistically significantly higher
`compared with those of patients who rated their health status
`as the same or better during the prestudy period (Table 3). In
`addition, the VAS that is part of the EuroQol showed a statis-
`tically significant association with taking compliance (F =
`4.32, p = 0.04). The association was negative, meaning that
`the better the perceived health state at the beginning of the
`study, the lower the compliance during the study. The strength
`of the association (R2 = 0.04) was negligible, however.
`
`Figure 5. Patients rank-ordered by taking compliance, correct dosing, and
`timing compliance.
`
`of 22% (Figure 7). Although the between-drug differences
`were large, none reached statistical significance.
`
`Determinants of compliance
`Functional capacity. HAQ total score was 0.75 ± 0.68. There
`were statistically significant differences in HAQ scores
`between the 3 diseases: RA patients had the highest scores
`(0.87 ± 0.72), then PMR patients (0.68 ± 0.67), and gout
`patients showed the lowest scores (0.44 ± 0.44). These differ-
`ences were statistically significant (chi-square 8.24, p = 0.02).
`There was no correlation of HAQ scores or of HAQ category
`scores (data not shown) with taking compliance, correct dos-
`ing, or timing compliance. This was also true within each of
`the 3 diseases. In addition, a one-way ANOVA of the HAQ
`total score, with taking compliance, correct dosing, and tim-
`ing compliance as separate independent variables, did not
`
`Figure 6. Influence of dosing regimen on compliance. QD: once daily, BID: twice daily, TID: 3 times daily.
`
`Personal, non-commercial use only. The Journal of Rheumatology Copyright © 2003. All rights reserved.
`The Journal of
`- Published by
`The Journal of Rheumatology 2003; 30:1
`
`Rheumatology
`
`50
`
`Sawai (IPR2019-00789), Ex. 1050, p. 008
`
`

`

`RA quality of life. Within patients with RA, there were no dif-
`ferences in RAQol total score between the 4 drugs or between
`the 2 drug groups (symptom modifying and disease control-
`ling). The RAQol total score and its individual items did not
`show any association with taking compliance (F = 0.21, p =
`0.65 and F = 1.0, p = 0.50, respectively), correct dosing (F =
`0.34 with p = 0.56 and F = 0.86 with p = 0.66), or timing com-
`pliance (F = 0.05 with p = 0.94 and F = 0.86 with p = 0.66).
`We therefore conclude that RAQol total score is not associat-
`ed with compliance.
`A multiple regression model with taking compliance as
`dependent variable and backward elimination of the demo-
`graphic and questionnaire variables showed that the class of
`medication (symptom modifying or disease controlling), the
`dosing frequency (once weekly, QD, BID or TID), sex, cop-
`ing (avoidance, passive reaction pattern, and expression of
`emotions), and the overall health (total NHP score) together
`explained 66.6% of the variance in taking compliance (adjust-
`ed R2) (p = 0.002; Table 4). In this regression model there was
`little colinearity between the variables, and the residual error
`was randomly distributed, indicating good fit of the model.
`The result for correct dosing was roughly equal (adjusted R2
`= 52%, p = 0.014, with the standardized ß-coefficient for sex
`being statistically not significant. For timing compliance the
`model did not converge, probably due to the strongly skewed
`distribution of the timing compliance variable, as shown in
`Figure 4.
`
`Table 4. Predictors of “taking” compliance: a multiple regression model
`with backward removal of variables.
`
`Variable in the Regression Equation
`
`Standardized ß Unstandardized ß
`
`Constant
`Class of medication
`(symptom modifying or
`disease controlling)
`Dosing frequency
`Sex
`Partial R2
`Coping
`Avoidance
`Passive reaction pattern
`Expression of emotions
`NHP
`Partial R2
`Total R2
`
`65.4
`–23.3
`
`8.2
`13.5
`
`–2.2
`5.0
`3.5
`–1.7
`
`–0.66
`
`1.16
`0.38
`0.31
`
`–0.41
`0.79
`0.40
`–0.62
`0.36
`0.67
`
`All ßs: p < 0.05.
`Independent variable: taking compliance.
`Class of medication: 1 = symptom modifying, 2 = disease controlling,
`3 = both.
`Dosing frequency: 1 = once daily, 2 = twice daily, 3 = thrice daily, 7 = once-
`weekly.
`Gender: 0 = female, 1 = male.
`Coping: Avoidance (7 = lowest, 56 = highest), passive reaction pattern
`(7 = lowest, 32 = highest), expression of emotions (3 = lowest, 19 = high-
`est).
`NHP: 0 = lowest, 34 = highest
`
`Figure 7. Decline of compliance over time.
`
`Table 3. EuroQol health status at baseline.
`
`Taking compliance, %
`Correct dosing, %
`Timing compliance, %
`
`Better,
`n = 20
`
`Same,
`n = 62
`
`81
`69
`43
`
`78
`61
`46
`
`Worse,
`n = 61
`
`95*
`