`R E V I E W A R T I C L E
`
`A Systematic Review of Adherence With
`Medications for Diabetes
`
`JOYCE A. CRAMER
`
`OBJECTIVE — The purpose of this study was to determine the extent to which patients omit
`doses of medications prescribed for diabetes.
`
`RESEARCH DESIGN AND METHODS — A literature search (1966 –2003) was per-
`formed to identify reports with quantitative data on adherence with oral hypoglycemic agents
`(OHAs) and insulin and correlations between adherence rates and glycemic control. Adequate
`documentation of adherence was found in 15 retrospective studies of OHA prescription refill
`rates, 5 prospective electronic monitoring OHA studies, and 3 retrospective insulin studies.
`
`RESULTS — Retrospective analyses showed that adherence to OHA therapy ranged from 36
`to 93% in patients remaining on treatment for 6 –24 months. Prospective electronic monitoring
`studies documented that patients took 67– 85% of OHA doses as prescribed. Electronic moni-
`toring identified poor compliers for interventions that improved adherence (61–79%; P ⬍ 0.05).
`Young patients filled prescriptions for one-third of prescribed insulin doses. Insulin adherence
`among patients with type 2 diabetes was 62– 64%.
`
`CONCLUSIONS — This review confirms that many patients for whom diabetes medication
`was prescribed were poor compliers with treatment, including both OHAs and insulin. However,
`electronic monitoring systems were useful in improving adherence for individual patients. Sim-
`ilar electronic monitoring systems for insulin administration could help healthcare providers
`determine patients needing additional support.
`
`Diabetes Care 27:1218 –1224, 2004
`
`D iabetes is a complex disorder that
`
`tence), compliance is the default medical
`term used in the literature (MEDLINE) to
`requires constant attention to diet,
`describe medication dosing (2). However,
`exercise, glucose monitoring, and
`the World Health Organization has pro-
`medication to achieve good glycemic con-
`moted the term “adherence” for use in
`trol. Glasgow (1) conceptualized the com-
`chronic disorders as “the extent to which
`plexity of diabetes regimens, creating a
`a person’s behavior—taking medication,
`model linking disease management and
`following diet, and/or executing lifestyle
`health outcomes with interactions be-
`changes— corresponds with agreed rec-
`tween patients and their healthcare pro-
`ommendations from a health care pro-
`viders. Factors contributing to optimum
`vider” (3).
`disease management included age, com-
`The incidence of type 2 diabetes is
`plexity of treatment, duration of disease,
`rapidly increasing, largely in older, over-
`depression, and psychosocial issues (1).
`weight patients who have concomitant
`Although a variety of terms have been
`cardiovascular risks (4). However, health
`used to describe these self-management
`care systems often do not have adequate
`or self-care activities (e.g., adherence,
`resources to provide support to individu-
`compliance, concordance, fidelity, persis-
`● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
`
`From the Department of Psychiatry, Yale University School of Medicine, New Haven, Connecticut.
`Address correspondence and reprint requests to Joyce A. Cramer, Yale University School of Medicine, 950
`Campbell Ave. (Room 7-127, G7E), West Haven, CT 06516-2770. E-mail: joyce.cramer@yale.edu.
`Received for publication 18 August 2003 and accepted in revised form 18 January 2004.
`J.C. is a member of an advisory panel of Novo Nordisk and has received honoraria or consulting fees from
`Novo Nordisk.
`Abbreviations: MEMS, Medication Event Monitor Systems; MUSE-P, Medication Usage Skills for Effec-
`tiveness Program; OHA, oral hypoglycemic agent.
`© 2004 by the American Diabetes Association.
`
`als with chronic diseases. Problems with
`poor self-management of drug therapy
`may exacerbate the burden of diabetes.
`Several studies suggest that a large
`proportion of people with diabetes have
`difficulty managing their medication reg-
`imens (oral hypoglycemic agents [OHAs]
`and insulin) as well as other aspects of
`self-management (1,5,6). Whereas some
`studies that have assessed adherence
`among young people with type 1 diabetes
`(6,7), little is known about adherence to
`insulin regimens in patients with type 2
`diabetes.
`This systematic review was under-
`taken 1) to assess the extent of poor ad-
`herence and persistence with OHAs and
`insulin and 2) to link adherence rates with
`glycemic control.
`
`RESEARCH DESIGN AND
`METHODS
`
`Literature search
`A systematic literature search was con-
`ducted to identify articles containing in-
`formation on the rate of adherence or
`persistence with OHAs or insulin. Ab-
`stracts captured by the systematic litera-
`ture search of MEDLINE (1966 to April
`2003), Current Contents (1993 to April
`2003), Health & Psychosocial Instru-
`ments (1985–2003), and Cochrane Col-
`laborative databases were first screened
`against the protocol inclusion criteria.
`The Level 1 screen identified papers re-
`lated to the main topic of interest. Ab-
`stracts passing the Level 1 screen were
`then retrieved for screening against the
`inclusion criteria (Level 2 screen). Full ar-
`ticles meeting the inclusion criteria were
`reviewed in detail (Level 3 screen).
`
`Inclusion criteria
`Papers were included in this review if 1) a
`dosing regimen was evaluated and medi-
`cation adherence or persistence rates
`were reported and 2) study design and
`methods for calculation of adherence
`were described. The papers must have in-
`cluded details of the methods used to de-
`termine adherence with a hypoglycemic
`agent (e.g., self-report, physician/nurse
`estimate, tablet count, prescription refill,
`
`1218
`
`DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004
`
`AstraZeneca Exhibit 2085
`Mylan v. AstraZeneca
`IPR2015-01340
`
`Page 1 of 7
`
`
`
`electronic monitoring) and some numeric
`results. Categorical results were consid-
`ered a lower level of information than
`data. The most desirable reports included
`both adherence rates and HbA1c levels.
`Reports of interventions that did not in-
`clude adherence rates were excluded. Re-
`ports of adherence with diet or exercise
`that did not also include medication ad-
`herence rates were also excluded. Reports
`may be retrospective surveys, prospective
`clinical trials, or prospective studies of ad-
`herence interventions. Methods may be
`database analyses of populations or elec-
`tronic monitoring of individual patients.
`
`Search strategy
`Key words for the database search were “pa-
`tient adherence” and “patient compliance”
`cross-linked with “diabetes mellitus,” “hy-
`poglycemic agents,” and “insulin.” The term
`“adherence” was linked automatically to the
`term “compliance” in MEDLINE as the pre-
`ferred term. Within the terms, sub-items
`were selected as: Administration & Dosage,
`Adverse Effects, Therapeutic Use, Preven-
`tion & Control, Drug Therapy, Psychology,
`Statistics & Numerical Data, and Econom-
`ics, as available for each term. The databases
`identified 186,188 publications.
`Level 1 searches combining terms
`identified 242 publications that appeared
`to relate to the topic of interest.
`Level 2 was a review of abstracts from
`the reports identified in Level 1, using the
`inclusion criteria. This stage identified 38
`reports as potentially having relevant
`data.
`Level 3 was a review of the papers
`identified in Level 2. These citations were
`supplemented with selected references
`from articles. This stage identified 19 pa-
`pers and one abstract (with additional in-
`formation from the authors) that met the
`inclusion criteria.
`The systematic search resulted in 20
`publications with adequate data on mea-
`surement of adherence with an OHA or
`insulin.
`
`Adherence assessment
`Definitions. For this review, medication
`adherence was operationalized as “taking
`medication as prescribed and/or agreed
`between the patients and the health care
`provider.” No studies provided informa-
`tion about the level of the patient’s agree-
`ment with the regimen. The “adherence
`rate” was the proportion of doses taken as
`prescribed. Some reports used categorical
`
`endpoints (e.g., 90%), below which pa-
`tients were considered “noncompliant”
`with the regimen. Adherence with “dose
`intervals” was defined as the proportion
`of doses taken within the appropriate
`window (e.g., 24 ⫹ 12 h for once-daily
`regimens, 12 ⫹ 6 h for twice-daily regi-
`mens).
`Treatment “persistence” was defined
`as either the proportion of patients who
`remained on treatment for a specified pe-
`riod (e.g., 12 months) or the mean num-
`ber of days to treatment discontinuation.
`Retrospective database assessment.
`Prescription benefit organizations (PBOs)
`that manage prescriptions and health
`maintenance organizations (HMOs) that
`manage the overall healthcare of patients
`have databases containing information
`about use of prescription medications.
`Records of new prescriptions and refills
`can be tabulated using unique patient
`identifiers. Some databases also are linked
`to diagnostic codes as well as laboratory
`and medical visit data that describe health
`service utilization for a cohort. Searches
`can be made to ascertain the types of med-
`ications, prescribed dose and regimen,
`and number of times the patient obtained
`a refill. These population-based surveys
`provide an overview of drug utilization
`during a period of time.
`Prospective monitoring. Electronic
`monitoring technology collects events
`based on taking medication from a mon-
`itored container, stores events, and lists
`medication dosing for an individual.
`Medication Event Monitor Systems
`(MEMS; APREX, Division of AARDEX,
`Union City, CA) were used in some pro-
`spective studies. MEMS are standard
`medication container bottle caps with a
`microprocessor that records every bottle
`opening. Patients are given bottles with a
`MEMS cap and instructions to take all
`doses of the oral medication from that
`bottle. Data are downloaded for display as
`a calendar of events (8). Electronic mon-
`itoring provides information about medi-
`cation usage at the level of the individual
`patient. Some researchers do not inform
`patients that they are being monitored to
`avoid an effect of observation (Hawthorne
`effect). Cramer (9,10) developed a
`method, the Medication Usage Skills for
`Effectiveness Program (MUSE-P), that
`uses electronic monitoring data displayed
`on a computer screen as a teaching tool to
`enhance medication adherence.
`
`Cramer
`
`Analyses. Descriptive statistics (means,
`ranges) present data from the selected re-
`ports tabulated by methodology (retro-
`spective database review, prospective
`monitoring), and class of medication
`(OHA, insulin).
`
`RESULTS — The systematic review
`was based on 20 reports that included
`quantitative information on adherence or
`persistence with diabetes medications
`(11–30). The few studies that included
`laboratory data all showed HbA1c levels
`⬎7%.
`
`OHA: retrospective analyses
`Adherence rates among 11 retrospective
`studies (19 cohorts) (11,14 –16,18 –
`22,24,25) using large databases ranged
`from 36 to 93% (excluding the study with
`categorical adherence rates) (17) (Table
`1). The mean age of patients in all these
`studies was ⬎50 years, indicating that
`these were older patients with type 2 dia-
`betes. The open observational (noncom-
`parative) studies (11,20,22,24,25) had
`similar results, ranging from 79 to 85%
`adherence with OHAs during 6 –36
`months of observation. Several studies
`compared cohorts with different regi-
`mens. Depressed patients had lower ad-
`herence rates than nondepressed patients
`(85 vs. 93%) (14). Once-daily regimens
`had higher adherence than twice-daily
`regimens (61 vs. 52%) (16). Mono-
`therapy regimens had higher adherence
`than polytherapy regimens (49 vs. 36%)
`(14) or a higher proportion of patients
`achieving high adherence rates (35 vs.
`27% at 90% or higher adherence rates)
`(17). Patients converting from mono-
`therapy or polytherapy to a single combi-
`nation tablet improved their adherence
`rates by 23 and 16%, respectively (19).
`The only report with adherence rates
`⬍50% was a survey of California Medic-
`aid (MediCal) patients newly treated with
`OHAs (15). Other studies included pa-
`tients with chronic treatment.
`Seven reports (nine cohorts) of OHA
`treatment persistence ranged from 16 to
`80% in patients remaining on treatment
`for 6 –24 months. Four studies reported
`83–300 days to discontinuation (Table
`1). The methodology differed among
`studies, so that cross-overs to an alterna-
`tive OHA or insulin might not have been
`counted as discontinuation. Two reports
`with large proportions (58 and 70%) of
`patients remaining on treatment for
`
`DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004
`
`1219
`
`Page 2 of 7
`
`
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`300
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`105
`83
`
`—
`
`83⫾71
`(days)
`
`Persistance
`
`—
`
`—
`
`39%‡
`
`83⫾22%
`85⫾15%
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`36%*
`44%*
`22%*
`
`36%*
`
`—
`
`—
`
`20%*
`16%*
`70%†
`58%*
`
`(percent)
`Persistance
`
`80⫾21%
`
`81%
`81%
`86%
`87%
`71%
`77%
`54%
`83%
`87%
`
`(27%⬎90%)
`(35%⬎90%)
`
`52%
`61%
`36%
`
`49%
`85%
`93%
`
`—
`
`79%
`
`rate
`
`Adherence
`
`786
`411
`975
`810
`
`28,001newstart
`
`195,400all
`
`3,358
`
`59
`
`105
`1,350
`2,275
`
`2,849
`
`992
`
`37,431
`
`121depressed
`
`119notdepressed
`
`677elderly
`216young
`
`693all
`79,498
`
`n
`
`—
`
`—
`
`Medication adherence
`
`*Persistancefor12months;†persistancefor24months;‡persistancefor6months,§adherenceratesexcludingcategoricaldata.HMO,healthmaintenanceorganization;PBO,pharmacybenefitorganization.
`Venturinietal.(25)
`Spoelstraetal.(24)
`Sclaretal.(23)
`Scheclmanetal.(22)
`
`59⫾11
`
`63
`
`59⫾10
`50⫾11
`
`—
`
`—
`
`—
`
`8.1⫾2.0
`
`24
`12
`12
`15
`
`Sulfonylurea
`OHA
`OHA
`OHA⫹Insulin
`
`HMO
`Netherland
`Medicaid
`Clinic
`
`53
`
`—
`
`63⫾15
`
`64
`67
`
`68
`
`55⫾13
`
`64⫾11
`72⫾5
`51⫾9
`
`60⫾14
`(years)
`
`Age
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`—
`
`8.0⫾1.5
`7.4⫾1.4
`
`—
`
`—
`
`—
`
`—
`
`HbA1c
`
`24
`36
`
`6
`
`6
`
`6
`
`12
`
`12
`
`18
`
`12
`
`12
`
`(10years)
`
`12
`
`(months)
`Follow-up
`
`OHA⫹Insulin
`OHA
`Polytocombination
`Polytherapy
`Monotocombination
`Monotherapy
`Melformin
`Sulfonylurea
`Polytherapy
`Monotherapy
`Glipizide,b.i.d.
`Glipizide,o.d.
`Polytherapy
`
`PBO
`Canada
`PBO
`
`Rajagopalanetal.(21)
`Morningstaretal.(20)
`
`PBO
`
`Mellkianetal.(19)
`
`Scotland
`
`Scotland
`
`Evansetal.(18)
`
`Donnanetal.(17)
`
`PBO
`
`DeziiandKawabata(16)
`
`Monotherapy
`
`Medicaid,new
`
`Daileyetal.(15)
`
`start
`
`OHA⫹Insulin
`
`HMO,all
`
`Chlechanowskietal.(14)
`
`Acarbose
`OHA⫹Insulin
`OHAmonotherapy
`
`Canada
`HMO,newstart
`PBO,newstart
`
`Catalanetal.(13)
`Brownetal.(12)
`Boccuzzietal.(11)
`
`Medications
`
`Population
`
`Reference
`
`Table1—RetrospectivedatabasestudiesofOHAfortype2diabeticpatients
`
`1220
`
`DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004
`
`Page 3 of 7
`
`
`
`Table 2—Prospective studies of OHA for type 2 diabetic patients using electronic monitoring
`
`Cramer
`
`Adherence rate
`
`Dose interval*
`
`77.7 ⫾ 21%
`40.7 ⫾ 28%
`5.3 ⫾ 5%
`
`HbA1c
`⬎8%
`12.7 ⫾ 1.9
`12.1 ⫾ 2.6
`—
`(40 o.d.)
`(36 b.i.d.)
`(15 t.i.d.)
`7.9 ⫾ 1.1
`
`Reference
`
`n
`
`Population
`
`Age (years)
`
`Medications
`
`Follow-up
`
`Mason et al. (26)
`Matsuyama et al. (27)
`
`Paes et al. (28)
`
`21
`15
`17
`91
`
`Clinic
`Intervention
`Control
`Community
`
`—
`84 ⫾ 8
`
`Sulfonylurea
`OHA
`
`3 months
`3 months
`
`69
`
`OHA
`
`6 months
`
`Rosen et al. (29)
`Rosen et al. (30)
`
`65
`63 ⫾ 11
`
`Metformin
`Metformin
`
`4 weeks
`6 months
`
`74.5%
`85.1%
`82.8%
`67.2 ⫾ 30%
`79.1 ⫾ 19%
`65.6 ⫾ 30%
`38.1 ⫾ 36%
`77.7 ⫾ 18%
`Clinic
`77
`79.3 ⫾ 13%
`Intervention
`16
`60.7 ⫾ 13%
`Control
`17
`*Dose interval ⫽ proportion of dose taken within the prescribed number of hours between doses (e.g., b.i.d. ⫽ 12 ⫾ 6 – h interval).
`
`12–24 months included all OHAs in the
`analyses (11,12). However, a study of
`Medicaid recipients in South Carolina
`showed low treatment persistence (39%
`at 6 months) (23). Three reports (four co-
`horts) with smaller proportions (16 –
`49%) of patients remaining on treatment
`for 6 –12 months focused on specific drug
`treatments (13,16) and monotherapy/
`polytherapy (15). Persistence expressed
`as days to discontinuation was similar in
`the two reports using similar methodol-
`ogy (83–105 days) (11,13) but was longer
`(300 days) in the report with descriptive
`data (17).
`
`OHA: prospective studies
`Three groups performed small prospec-
`tive studies with electronic dose monitor-
`ing, with two centers each publishing two
`reports describing different aspects of the
`studies. Adherence rates were more con-
`sistent than was found in the retrospective
`database analyses (Table 1). Mean adher-
`ence with OHAs was in a narrow range of
`61– 85% during up to 6 months’ observa-
`tion (Table 2). All of the prospective stud-
`ies used MEMS electronic monitoring to
`determine when doses were taken. Elec-
`tronic monitoring also demonstrated that
`
`adherence rates decreased with larger
`numbers of OHA doses to be taken daily.
`One report showed mean adherence of
`79.1 ⫾ 19% for once-daily regimens,
`65.6 ⫾ 30% for twice-daily regimens, and
`38.1 ⫾ 36% for three-times daily dosing
`regimens (P ⬍ 0.05) (28). The accuracy of
`taking doses at appropriate time intervals
`also decreased (77.7 ⫾ 21% for once-
`daily regimens, 40.7 ⫾ 28% for twice-
`daily regimens, 5.3 ⫾ 5% for three-times
`daily regimens; P ⬍ 0.01).
`The adherence rate for patients taking
`sulfonylurea was 74.5% using electronic
`monitoring, compared with 92.4% for
`self-reported adherence (26). Matsuyama
`et al. (27) used electronic monitoring re-
`ports to guide clinical decision making.
`Adherence reports for a subset of patients
`were provided to their doctors to assist in
`making treatment decisions. The infor-
`mation revealed a need for additional pa-
`tient education because of inconsistent
`dosing (47% of reports). The control
`group had several instances of dose in-
`creases because the clinician was not
`aware that erratic dosing was the problem
`rather than low dose.
`Rosen et al. (29,30) used electronic
`monitoring with the MUSE-P medication
`
`enhancement program (29) to demon-
`strate that poor adherence can be im-
`proved when patients and clinicians are
`aware of the frequency of missed doses.
`They monitored a series of patients (mean
`adherence 78%) (29) to find a group of
`poor OHA compliers (mean 61%) in or-
`der to start with a cohort needing im-
`provement. The control group remained
`unchanged, whereas the group receiving
`the intervention improved to 79% adher-
`ence (P ⬍ 0.05) with their OHA regimen
`(Table 2) (30).
`
`Insulin
`Adherence rates among the three studies
`that assessed insulin use were not compa-
`rable because of different methods of
`analysis (Table 3). The retrospective data-
`base method (21) showed a mean 63 ⫾
`24% adherence for large cohorts of long-
`term and new-start adult type 2 diabetic
`insulin users. Adherence rates were lower
`for insulin use than for OHA use (73–
`86%) in both populations (21). A 10-year
`follow-up of a large cohort of patients
`newly started on insulin found that 80%
`of patients persisted with insulin treat-
`ment for 24 months (12). Fewer patients
`in the insulin-only group (20%) than pa-
`
`Table 3—Retrospective database studies of insulin use
`
`Reference
`
`n
`
`Population
`
`Age
`(years)
`
`Follow-up
`
`HbA1c
`
`Adherence rate
`
`Brown et al. (12)
`Morris et al. (7)
`
`102
`89
`
`HMO new start
`Scotland
`
`—
`16 ⫾ 7
`
`10 years
`12 months
`
`Rajagopalan et al. (21)
`
`PBO
`
`53
`
`24 months
`
`Persistence 79.6% at 24 months
`33–86% days supply*
`87–116% days supply*
`62 ⫾ 24%
`27,274 all
`64 ⫾ 24%
`1,323 new start
`*Days supply ⫽ number of tablet dispensed per prescribed number of times to be taken daily. HMO, health maintenance organization; PBO, pharmacy benefit
`organization.
`
`—
`9.4 ⫾ 1.7
`9.0 ⫾ 1.5
`—
`
`DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004
`
`1221
`
`Page 4 of 7
`
`
`
`Medication adherence
`
`tients taking an OHA (31%) discontinued
`treatment (obtained no refill) during the
`second year of follow-up (11). A study of
`children and adolescents presented evi-
`dence that poorer compliers had higher
`mean HbA1c levels (R2 ⫽ 0.39) (7). They
`calculated an index of days with insulin
`obtained from the pharmacy, based on
`the prescribed dose. HbA1c levels ranged
`from 9.44 ⫾ 1.7 for the lowest amount of
`insulin obtained to 8.98 ⫾ 1.5, 7.85 ⫾
`1.4, and 7.25 ⫾ 1.0 for the higher cate-
`gories of adherence, respectively (P ⬍
`0.001). Additional information about
`clinical status demonstrated that 36% of
`patients with poorest adherence were ad-
`mitted to the hospital for diabetic ketoac-
`idosis (P ⫽ 0.001 compared with patients
`with higher adherence rates) and other
`complications related to diabetes (P ⫽
`0.02 compared with patients with higher
`adherence rates). Adolescents (10 –20
`years of age) were significantly more
`likely to be in the lowest adherence cate-
`gory and have the highest HbA1c levels
`compared with younger and older pa-
`tients (both P ⬍ 0.001).
`
`CONCLUSIONS — This systematic
`review confirms that many patients with
`diabetes took less than the prescribed
`amount of medication, including both
`OHA and insulin. Given the central im-
`portance of patient self-management and
`medication adherence for health out-
`comes of diabetes care (31), surprisingly
`few studies were found that adequately
`quantified adherence to diabetes medica-
`tion. The overall rate of adherence with
`OHA was 36 –93% in retrospective and
`prospective studies. Previous surveys
`have found that people took ⬃75% of
`medications as prescribed, across a vari-
`ety of medical disorders (32,33). Decreas-
`ing adherence related to polytherapy and
`multiple daily dosing schedules also
`matched reports from other medical dis-
`orders (32,33).
`This survey adds to the general find-
`ing that adherence rates are not related to
`the simplicity of regimen, the severity of
`the disorder, or the possible conse-
`quences of missed doses. The persistence
`with OHAs of 6 –24 months, as seen in
`this survey, suggests that brief treatment
`persistence is a major issue that could lead
`to deleterious health outcomes. These
`data parallel other chronic medical disor-
`ders in which persistence often is ⬍1 year
`(34,35). Even with good OHA adherence,
`
`the natural progression of type 2 diabetes
`eventually leads many patients to require
`insulin treatment. The study that evalu-
`ated type 2 diabetic patients receiving in-
`sulin showed 63% of doses taken as
`prescribed (21). In one cohort, only 80%
`of patients persisted with insulin for 2
`years despite the need for long-term gly-
`cemic control (12). The detailed analysis
`of a group of children and adolescents
`showed that poor adherence with the pre-
`scribed insulin regimens resulted in poor
`glycemic control, as well as more hospi-
`talizations for diabetic ketoacidosis and
`other complications related to diabetes
`(7). Self-reported insulin use (not in-
`cluded in this analysis) showed that pa-
`tients frequently omit injections. In 31%
`of women who reported intentionally
`omitting doses (8% frequently), weight
`gain was the reason (36). One-fourth of
`adolescents reported having omitted
`some injections during the 10 days before
`a clinic visit (37). Therefore, clinicians
`cannot assume that patients with either
`type 1 or type 2 diabetes are fully compli-
`ant with insulin regimens, even if the con-
`sequences might be hazardous.
`The second goal of this study was to
`estimate the strength of the association
`between adherence and glycemic control.
`Too few studies included HbA1c levels to
`allow a precise conclusion, although in-
`terventions that improve self-manage-
`ment have been associated with better
`clinical outcomes (38). Further research
`is needed to quantify the specific im-
`provement in glycemic control that might
`be obtained from improved medication
`adherence. Such studies should demon-
`strate the health benefits that may be de-
`rived from more convenient therapeutic
`regimens that are being developed for di-
`abetes.
`A bright spot among these reports of
`poor adherence and persistence was the
`finding that electronic monitoring tools
`exist to help enhance medication adher-
`ence for individual patients. One study
`demonstrated that doctors and pharma-
`cists were able to adjust treatment plans
`more appropriately when they had elec-
`tronic monitoring data than when they
`used the usual mode of employing only
`laboratory data (27). The difference was
`in understanding that elevated glucose or
`HbA1c levels were related to missed doses
`and not underprescribing. This informa-
`tion avoided changing prescriptions, in-
`creasing drug dose, and switching or
`
`adding medication. Rosen et al. (30)
`screened a clinic population to select pa-
`tients with low adherence rates for ran-
`domization to a control group or the
`MUSE-P intervention. MUSE-P consists
`of a dialogue between the patient and
`health care provider about daily medica-
`tion dosing structured around their per-
`sonal record of electronic monitoring data
`(39). This simple technique resulted in a
`significant improvement in adherence
`rates compared with the control subjects,
`who received the same amount of per-
`sonal attention but not focused on adher-
`ence. This program has been effective in
`enhancing adherence in other medical
`disorders (39 – 41). However, electronic
`monitoring is not a readily available tool.
`Several simple measures usually are help-
`ful in clinical practice, such as once-daily
`dosing and combining multiple medica-
`tions into the same regimen (e.g., several
`drugs premeal rather than some before
`and some with meals). Patients should be
`given information about what to do if a
`dose is missed or if adverse effects are
`bothersome, in addition to the purpose of
`the medication (9,10).
`Similar electronic monitoring sys-
`tems for insulin administration are
`needed to record patterns of insulin use
`by individual patients. This information
`could help healthcare providers deter-
`mine which patients need additional sup-
`port to achieve consistent glycemic
`control. Further studies with electronic
`monitoring of diabetes medications may
`identify and define the characteristics of
`poorly compliant patients to improve
`treatment outcomes. Improved under-
`standing of the way patients use medica-
`tion could also affect healthcare
`utilization. Improved glycemic control
`could reduce overall healthcare costs
`(42). This has important implications be-
`cause of the potential to improve the cur-
`rently poor adherence with all aspects of
`diabetes self-management. Inadequate
`adherence to medication and lifestyle rec-
`ommendations by patients with diabetes
`may play an important role in adding to
`the economic burden of the disease.
`The major drawback of this survey is
`the methodology used for adherence
`analyses in the reports reviewed. A short-
`coming in the literature is the lack of stud-
`ies evaluating interventions to improve
`adherence in which adherence was mea-
`sured using appropriate methods. The
`retrospective analyses used various defi-
`
`1222
`
`DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004
`
`Page 5 of 7
`
`
`
`nitions of adherence and persistence and
`different durations of follow-up. Some
`included all patients, whereas others cen-
`sored cohorts based on arbitrary condi-
`tions. Analyses did not always account for
`patients who changed to another hypo-
`glycemic agent or were no longer eligible
`for observation because of a change in
`health insurance. Attempts are underway
`to define optimum analytic methods for
`retrospective database studies (43). Elec-
`tronic monitoring studies suffered from
`small size and observation limited to one
`OHA. An overall drawback to this review
`is the lack of an electronic method to
`monitor insulin use. Such devices are
`commonly used to record blood glucose
`measurements. The development of an
`electronic monitoring system for insulin
`dosing would be an important step to-
`ward proving better support for individ-
`uals with poor insulin adherence and
`improving the dialogue between patients
`and their healthcare providers.
`The finding that patients prescribed
`an OHA or insulin take less than the pre-
`scribed number of doses over long peri-
`ods of follow-up indicates an urgent need
`for prescribers to understand that failure
`to reduce HbA1c levels might be related to
`inadequate self-management. The impli-
`cation is that instead of increasing the
`dose, changing the medication, or adding
`a second drug when glucose and HbA1c
`levels are high, clinicians should consider
`counseling patients on how to improve
`medication adherence. A first step to im-
`proving adherence is being able to assess
`it. Developing methods that properly as-
`sess medication adherence as a behavior
`that can be modified could provide a clin-
`ically significant improvement in glyce-
`mic control for some patients. Although
`methods are not yet available for routine
`use, such information could enhance pa-
`tient-clinician relationships by providing
`information to guide individualized self-
`management to support patients.
`
`Acknowledgments— This project was sup-
`ported by Novo Nordisk.
`We thank Soren Skovlund for providing
`helpful comments.
`
`References
`1. Glasgow RE: Compliance to diabetes reg-
`imens:
`conceptualization,
`complexity,
`and determinants. In Patient Compliance in
`Medical Practice and Clinical Trials. Cra-
`
`mer JA, Spilker B, Eds. New York, Raven
`Press, 1991, p. 209 –224
`2. Feinstein AR: On white coat effects and
`the electronic monitoring of compliance.
`Arch Intern Med 150:1377–1378, 1990
`3. World Health Organization: Report on
`Medication Adherence. Geneva, World
`Health Org., 2003
`4. Grundy SM, Benjamin IJ, Burke GL, Chait
`A, Eckel RH, Howard BV, Mitch W, Smith
`SC Jr, Sowers JR: Diabetes and cardiovas-
`cular disease: a statement for health care
`professionals from the American Heart
`Association. Circulation 100:1134 –1146,
`1999
`5. Pugh MJ, Anderson J, Pogach LM, Berlow-
`itz DR: Differential adoption of pharma-
`cotherapy recommendations for type 2
`diabetes by generalists and specialists.
`Med Care Res Rev 60:178 –200, 2003
`6. Johnson SB: Methodological issues in di-
`abetes research: measuring adherence.
`Diabetes Care 15:1658 –1667, 1992
`7. Morris AD, Boyle DIR, McMahon AD,
`Greene SA, MacDonald TM, Newton RW,
`for the DARTS/MEMO Collaboration: Ad-
`herence to insulin treatment, glycemic
`control, and ketoacidosis in insulin-de-
`pendent diabetes mellitus. Lancet 350:
`1505–1510, 1997
`8. Cramer JA, Mattson RH, Prevey ML,
`Scheyer RD, Ouellette VL: How often is
`medication taken as prescribed? A novel
`assessment technique. JAMA 261:3273–
`3277, 1989
`9. Cramer JA: Microelectronic systems for
`monitoring and enhancing patient com-
`pliance with medication regimens. Drugs
`49:321–327, 1995
`10. Cramer JA: Medication use by the elderly:
`enhancing patient compliance in the el-
`derly: role of packaging aids and monitor-
`ing. Drugs Aging 12:7–15, 1998
`11. Boccuzzi SJ, Wogen J, Fox J, Sung JCY,
`Shah AB, Kim J: Utilization of oral hypo-
`glycemic agents in a drug-insured U.S.
`population. Diabetes Care 24:1411–1415,
`2001
`12. Brown JB, Nichol GA, Glauber HS, Bakst
`A: Ten-year follow-up of antidiabetic
`drug use, nonadherence, and mortality in
`a defined population with type-2 diabe-
`tes. Clin Ther 21:1045–1057, 1999
`13. Catalan VS, Couture JA, Lelorier J: Predic-
`tors of persistence of use with the novel
`antidiabetic agent acarbose. Arch Intern
`Med 161:1106 –1112, 2001
`14. Chiechanowski PS, Katon WJ, Russo JE:
`Depression and diabetes. Arch Intern Med
`160:3278 –3285, 2000
`15. Dailey G, Kim MS, Lian JF: Patient com-
`pliance and persistence with antihyper-
`glycemic drug regimens: evaluation of a
`Medicaid patient population with type 2
`diabetes mellitus. Clin Ther 23:1311–
`1320, 2001
`
`Cramer
`
`16. Dezii CM, Kawabata H, Tran M: Effects of
`once daily and twice-daily dosing on ad-
`herence with prescribed glipizide oral
`therapy for type 2 diabetes. South Med J
`95:68 –71, 2002
`17. Donnan PT, MacDonald TM, Morris AD,
`for the DARTS/MEMO Collaboration: Ad-
`herence to prescribed oral hypoglycemic
`medication in a population of patents
`with type 2 diabetes: a retrospective co-
`hort study. Diabet Med 19:279 –284,
`2002
`18. Evans JM, Donnan PT, Morris AD: Adher-
`ence to oral hypoglycaemic agents prior to
`insulin therapy in type 2 diabetes. Diabet
`Med 19:685– 688, 2002
`19. Melikian C, White J, Vanderplas A, Dezii
`CM, Chang E: Adherence to oral antidia-
`betic therapy in a managed care organiza-
`tion: a comparison of monotherapy,
`combination therapy, and fixed-dose
`combination therapy. Clin Ther 24:460 –
`467, 2002
`20. Morningstar BA, Sketris IS, Kephart GC,
`Sclar DA: Variation in pharmacy prescrip-
`tion refill adherence measures by type of
`oral antihyperglycaemic drug therapy in
`seniors in Nova Scotia, Canada. J Clin
`Pharm Ther 27:213–220, 2002
`21. Rajagopalan R, Joyce A, Smith D, Ollen-
`dorf D, Murray FT: Medication compli-
`ance in type 2 diabetes patients:
`retrospective data analysis (Abstract).
`Value Health 6:328, 2003
`22. Schectman JM, Nadkarni MM, Voss JD:
`The association between diabetes meta-
`bolic control and drug adherence in an
`indigent population. Diabetes Care 25:
`1015–1021, 2002
`23. Sclar DA, Robison LM, Skaer TL, Dickson
`WM, Kozma CM, Reeder CE: Sulfonyl-
`urea pharmacotherapy regimen adher-
`ence in a Medicaid population: influence
`of age, gender, and race. Diabetes Educator
`25:531–532, 1999
`24. Spoelstra JA, Stolk RP, Heerdink ER,
`