physician perception of patient adherence compared to patient adherence of osteoporosis medications...

Copyright ©

2010 Inform

a UK Limite

d

Not for S

ale or Commerc

ial Distri

bution

Unauthoriz

ed use prohibite

d. Auth

orised users

can download,

display, view and print a

single copy for p

ersonal u

se

Current Medical Research & Opinion Vol. 26, No. 4, 2010, 777–785

0300-7995 Article 5306/458426

doi:10.1185/03007990903579171 All rights reserved: reproduction in whole or part not permitted

Original articlePhysician perception of patient adherencecompared to patient adherence of osteoporosismedications from pharmacy claims

R. CopherP. Buzineci3 Innovus, Eden Prairie, MN, USA

V. Zarotskyi3 Innovus, Calabasas, CA, USA

L. KazisBoston University, School of Public Health, Boston,

MA, USA

S.U. IqbalD. MacariosAmgen Inc., Thousand Oaks, CA, USA

Address for correspondence:Victoria Zarotsky, PharmD, i3 Innovus, 22533

Jameson Dr, Calabasas, CA 91302, USA.

Tel.: þ1 818 225 7143; Fax: þ1 952 833 6045;

[email protected]

Key words:Data collection – Medication adherence –

Osteoporosis – Physicians – Postmenopausal –

Survey

Accepted: 23 December 2009; published online: 22 January 2010

Citation: Curr Med Res Opin 2010; 26:777–85

Abstract

Objective:

This study explored physicians’ perceptions of patient adherence to medications compared with patient

adherence derived by administrative data in the treatment of osteoporosis.

Research design and methods:

A study involving written questionnaires from prescribers treating patients with postmenopausal

osteoporosis (PMO) compared the questionnaire responses to pharmacy claims of these prescribers’

patients’ refill patterns. Approximately 2000 physicians from a large US health plan were faxed or

mailed a survey. Data from the physician survey were merged with administrative claims data of the

participating physicians’ patients.

Results:

A total of 412 physicians (21.8%) responded. Although a low response rate, there were no significant

demographic differences between participating and non-participating physicians. Surveyed physicians

reported that 66% of their patients had private/commercial coverage and over 60% reported seeing

their PMO patients annually. Overall, physicians estimated that 69.2% of patients were adherent 80% of

the time after 12 months of therapy. Yet, pharmacy claims data for those physicians’ patients indicated

48.7% of patients were adherent (defined as having an MPR of �80%) after 12 months of therapy.

Physicians overestimated their patients’ adherence regardless of medication class and across physician

specialties. Regression modeling revealed that physicians who have been in practice longer estimated fewer

patients as adherent, whereas those who prescribe more PMO treatments estimate a greater number of

patients as adherent. Providers cited side effects and affordability of medication as the most frequent

reasons for non-adherence.

Conclusions:

Physicians overestimate patient adherence to PMO therapies. Improving physician awareness of medication

non-adherence to PMO therapies may facilitate physician–patient dialogue, with the aim of identifying

patient-centered reasons for non-adherence. These discussions are important because patients with poorer

adherence have a higher risk of fracture. Future research should focus on reasons for patient non-

adherence to osteoporosis regimens and intervention strategies that improve communication between

the provider and patient. Findings must be considered within the limitations of this claims database

analysis. Some degree of incomplete or incorrect coding may exist, and the presence of a claim for a

filled prescription does not indicate that the medication was consumed or taken as prescribed. Patients

included in the study are not necessarily representative of all patients being treated for osteoporosis.

! 2010 Informa UK Ltd www.cmrojournal.com Physicians’ perceptions of osteoporosis medication adherence Copher et al. 777

Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

Introduction

Osteoporosis affects approximately 10 million individualsin the United States, and an additional 34 million areestimated to have low bone mass and are thus at risk ofdeveloping osteoporosis1. According to recent findings,osteoporosis in 2005 was responsible for over 2 millionfractures in the United States at a cost of $19 billion2.Experts predict that by 2025, these costs will rise to approx-imately $25.3 billion2. Several effective drug treatmentsare available for osteoporosis, including bisphosphonates,hormone replacement therapy, calcitonin, and parathy-roid hormone3. However, non-adherence to these thera-pies (i.e., failure to take medications as prescribed) mayundermine their effectiveness. Studies evaluating theimpact of non-adherence on treatment effectivenesshave shown that patients who use their osteoporosismedication inconsistently do not attain the full benefitsof therapy (i.e., improvements in bone mineral density andreduction of fracture risk)4–6.

Medication non-adherence is a problem associated withnearly all chronic diseases7 including osteoporosis8.Although effective pharmacological therapies are avail-able for osteoporosis, complicated dosing regimens andside-effects may predispose osteoporosis patients to medi-cation discontinuation9. A recent meta-analysis foundthat medication non-adherence in osteoporosis occursshortly after treatment initiation. Pooled adherence ratesdecreased from 53% for patients who remained on treat-ment for 1 to 6 months to 43% patients who remained ontreatment for 7 to 24 months10. In a large retrospectivestudy of more than 58,000 patients who initiated drug ther-apy for osteoporosis, the overall 1-year compliance ratewas less than 25%11. Low adherence or low compliancewith osteoporosis medications can compromise the effec-tiveness of these agents and lead to suboptimal patientoutcomes8,12,13. A recent systematic review evaluatingthe impact of adherence on fracture rates reported a frac-ture risk reduction of 17%–39% in patients who achieved amedication possession ratio (MPR) of �80%. Conversely,a review of compliance and persistence data from 17 ret-rospective/observational studies reported no benefit infracture reduction in patients with MPR 550%14. Inaddition, numerous studies have demonstrated a positivecorrelation between continued adherence to medicationand improvements in BMD and biochemical markers ofreduction in bone turnover15–17.

Although a number of empirical assessments have eval-uated adherence from the patient perspective, data onphysician perception of patient medication adherence islacking. Informed physician awareness of patient adher-ence is particularly important, because physicians canthen open a dialogue and emphasize the importance oftaking medications to patients that are actually non-adherent. Provider-driven disease education, patient

follow-up, and patient involvement in treatment decisionsalso affect adherence to osteoporosis therapy18. The pur-pose of this study was to examine the concordancebetween physicians’ perceptions of patient adherence toosteoporosis medications and the actual adherence rates ofthese physicians’ patients based on pharmacy claims. Thisstudy also evaluated factors that physicians believe con-tribute to patient non-adherence, as well as options theybelieve will improve patient adherence.

Patients and methods

Data source

This was a claims data-linked survey study involving writ-ten questionnaires from physician prescribers treatingpatients with postmenopausal osteoporosis (PMO) thatcompared the questionnaire responses to pharmacyclaims of these prescribers’ patients’ refill patterns.Approximately 2000 physicians from a large US healthplan were administered a patient medication adherencesurvey. For this cross-sectional survey, physicians treatingosteoporosis patients were identified using eligibility, phar-macy, and medical claims data from a large managedhealth plan in the United States. Retrospectively, claimsdata for PMO patients being treated by the selectedphysicians were merged with survey responses. It shouldbe noted, however, that the survey data were not matchedand compared to the claims data for specific patients.At the time the study was conducted, the administrativeclaims database included data for approximately 14 millionhealth plan enrollees with both medical and pharmacybenefits. The health plan comprises discountedfee-for-service independent practice association plansspanning the United States, primarily in the southernand mid-western regions. The providers of pharmacy andmedical services submit their claims for payment directlyto the health plan. All study data were de-identified andaccessed with protocols compliant with the HealthInsurance Portability and Accountability Act (HIPAA).Approval was obtained from a central Institutional ReviewBoard (IRB) to conduct the physician survey.

Physician and patient identification

For inclusion in the study, patients were required to have atleast one medical claim and at least one pharmacy claim.Specifically, PMO patients were defined as having at leastone medical claim with an International Classificationof Diseases, 9th Revision, Clinical Modification(ICD-9-CM) diagnosis code of osteoporosis (733.0x) inany position between 30 June 2002 and 31 December2006. The date of the first observed medical claim was

Current Medical Research & Opinion Volume 26, Number 4 April 2010

778 Physicians’ perceptions of osteoporosis medication adherence Copher et al. www.cmrojournal.com ! 2010 Informa UK Ltd

Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

defined as the osteoporosis identification date. Patientsalso had to have at least one pharmacy claim for an osteo-porosis medication (i.e., alendronate 5 mg, alendronate10 mg, alendronate 35 mg, alendronate 70 mg, risedronate5 mg, risedronate 35 mg, teriparatide, ibandronate, ralox-ifene, alendronateþ vitamin D, or risedronateþ calcium)during the study period. The date of the first relevant phar-macy claim was defined as the index date. Female patientswho were 45 years of age or older as of the index date withcontinuous coverage by the health plan with both medicaland pharmacy benefits for at least 1 year prior to the indexdate and at least 180 days after the index date wereincluded; patients with a history of chronic corticosteroidor thyroid hormone use were excluded as these medica-tions contribute to osteoporosis. To ensure a focus onpatients being treated for osteoporosis, patients with phar-macy claims for medication used to treat bone diseaseother than osteoporosis during the identification period(i.e., alendronate 40 mg, risedronate 30 mg, or calcitriol),or who had evidence of other bone disease-related condi-tions (i.e., ICD-9 codes for Paget’s disease, osteopathies731.xx) that are treated with the same medications usedto treat osteoporosis, were excluded from the study.

Physicians treating at least five patients who fulfilledthe above criteria and who also had a valid fax number andspecialized in general practice, which includes family prac-tice, internal medicine, or obstetrics and gynecology wereidentified for participation in this survey study. As princi-pal treatment prescribers, physicians were the focus ofthe current study. Approval from a central IRB wasobtained. Physicians were contacted by fax to participatein the study, and a follow-up survey was mailed tonon-responders. An invitation letter and directions forcompleting the survey were sent along with the survey.Data from the physician survey were merged with admin-istrative claims data of the patients whose physician com-pleted a survey; patients could be either commercial orMedicare/Medicaid health plan enrollees. The mergeddata allowed for comparisons of physicians’ perceptionsof patient adherence to patient adherence as observed inthe claims.

Physician survey

A survey questionnaire was developed to collect informa-tion on physicians’ diagnosis and treatment of osteoporosisas it relates to their patients, their perceptions of patientadherence to osteoporosis medications, and informationabout the physicians and their practices.

As validated measures were not available for the survey,a pilot study of the physician survey was conducted. Tenphysicians were identified by the study sponsor and sentthe physician survey for review. The physicians were askedto provide feedback on the survey content,

straightforwardness of the questions, and ease of comple-tion though no validity assessment was conducted. Threeof the 10 invited physicians responded. Their feedback wasincorporated into the final version of the physician survey.The three physicians who reviewed the survey did not par-ticipate in the study.

The four-page survey contained multiple choice ques-tions on: physician and practice characteristics, factors,techniques and equipment used when diagnosing osteo-porosis, and how often physicians see their osteoporosispatients. They were also asked about their typical recom-mendations and reminders used in treating osteoporosis.Physicians were asked to estimate what percentage of theirpatients are adherent to osteoporosis medications after 1year of therapy. Medications were organized as (1) weeklyoral bisphosphonates (alendronate and risedronate),(2) monthly oral bisphosphonates (ibandronate), (3) selec-tive estrogen receptor modulators (SERM [raloxifene]),(4) calcitonin, and (5) teriparatide. Physicians wereasked to rank up to five factors listed that they believeaffect patient non-adherence and patient adherence. Inaddition, physicians were asked to rank the top five factorslikely to motivate or improve patient adherence. Factorslisted included: medication side effects, medication costs,dosing complexity, drug safety, inhibits patient’s lifestyle,fear of fracture, and family involvement.

Surveys with responses to at least 80% of study criticalquestions were considered complete, and physicians wereeach compensated $75 for a completed survey.

Study measures

Subject demographicsPatient demographic variables (age, gender, and geo-graphic location) were captured from the enrollmentdata. Physician characteristics such as practice type,volume of patients, gender, and experience (i.e., numberof years in practice) were collected by the survey instru-ment. Information on physician specialty was derived fromthe medical claims with evidence of an osteoporosis diag-nosis and pharmacy claims for osteoporosis medications.

Patient adherencePatient-observed adherence was measured as an MPR, cal-culated as the percentage of days the patient had any osteo-porosis medication during the follow-up period.

MPR ½PMO medication� ¼Total days available

365:

The availability of treatment was determined with a cal-culated days’ supply field based on the quantity of drugdispensed and the dosing schedule. All qualifying osteo-porosis medications were included in the calculation, and



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

change from one osteoporosis medication to another wasconsidered continued osteoporosis treatment. If the pat-terns of medication fills and days’ supply indicated thatpatients used multiple osteoporosis medications simulta-neously, days during which patients had multiple medica-tions were counted once to prevent overestimating MPR,i.e., the numerator of the MPR was equal to the number ofdays patients were estimated to have any osteoporosis med-ications on hand. It was also assumed that osteoporosismedications were available to a subject during inpatientfacility stays if an outpatient pharmacy fill of an osteoporo-sis medication was interrupted by an inpatient stay; inthese cases, days supply remaining on the facility admis-sion date was counted beginning on the day after the inpa-tient discharge date. Patients were grouped by MPR levelas MPR� 80% (‘adherent’) and MPR580% (‘non-adherent’), a commonly used threshold19–21.

Analysis

All study variables, including physician characteristics,physicians’ perceptions of patient adherence, and ratingsof factors that affect adherence, were analyzed descrip-tively using chi-square, F- or t-tests for significancewhere appropriate. Results were stratified by physician spe-cialty and perceived patient adherence level. Patients with12 months of PMO therapy following their index date wereused for comparisons with the physicians’ perceived adher-ence. To examine differences across medication class, aweighted proportion of patient adherence was calculatedthat ensured physicians were accounted for equally suchthat physicians with more patients were not valuedmore than physicians with fewer patients. Physician-level weights were calculated as one divided by the

number of patients seen by the given physician. Theweights were then applied to the patients’ MPR.The weighted proportion was used for comparison withthe physicians’ perceptions of their patients’ adherenceto the specific medication. At the physician level, apaired t-test was performed. All descriptive analyses weredone using SAS� v9.1.3.

In addition to the descriptive analyses, ordinary leastsquares (OLS) modeling was used to examine influenceson physicians’ perceptions of their patients’ adherence toweekly and monthly oral bisphosphonates. Physicians’ per-ceptions were regressed on treatment and years in practice.A logistic regression was used to determine factors influen-cing patients’ adherence per claims-based MPR; patientsobserved adherence (claims-based MPR) was regressedon the physician perception, patient age, years in practiceand practice setting. All modeling was done via Statav10.0 SE.

Results

Physician characteristics

Approximately 2000 physicians were contacted to partic-ipate in the study; of these, 412 physicians returned com-plete surveys, for an overall response rate of 21.8%.Although the response rate was low, there were no statis-tically significant demographic differences (with regard tospecialty, demographic region, or numbers of patientsseen) between the physicians who participated comparedwith the physicians who did not participate in the study(data not shown). Demographic characteristics of partici-pating physicians are given in Table 1. The distribution ofthe participating physicians across the three specialty

Table 1. Demographics for participating physicians by specialty.

Total General Practice Internal Medicine OB/GYN p-value

Number of physicians 412 110 189 113Years in practice, n (%)

55 2 (0.5) 0 (0.0) 1 (0.5) 1 (0.9) 0.069y5–9 37 (9.0) 13 (11.8) 17 (9.0) 7 (6.3)10–14 80 (19.5) 18 (16.4) 48 (25.5) 14 (12.5)15–19 70 (17.1) 16 (14.5) 35 (18.6) 19 (17.0)�20 221 (53.9) 63 (57.3) 87 (46.3) 71 (63.4)

Practice setting, n (%)Solo practice 126 (30.7) 39 (35.5) 51 (27.1) 36 (32.1) 50.001yPrimary care group 173 (42.2) 53 (48.2) 91 (48.4) 29 (25.9)Multi-specialty 64 (15.6) 14 (12.7) 42 (22.3) 8 (7.1)Some other arrangement 47 (11.5) 4 (3.6) 4 (2.1) 39 (34.8)

Patient insurance coverage, mean (SD)Private/commercial 60.5 (19.9) 61.2 (17.2) 53.3 (19.9) 72.1 (16.7) 50.001*Medicaid 4.3 (7.4) 4.9 (6.5) 3.1 (4.9) 5.8 (10.9) 0.008*Medicare 31.0 (19.1) 27.0 (14.6) 41.2 (19.4) 17.8 (11.4) 50.001*No health insurance coverage 4.5 (5.1) 6.9 (7.2) 3.4 (3.6) 4.1 (4.1) 50.001*

yChi-square test.*F-test.OB/GYN¼ obstetrics and gynecology; SD¼ standard deviation.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

categories was: general practice (N¼ 110), internal med-icine (N¼ 189), or obstetrics and gynecology (OB/GYN)(N¼ 113). Of participating physicians, 63.2% were maleand 36.8% were female, and over half (60.3%) were 45–60years old. The majority of physicians (71.0%) reportedbeing in practice for at least 15 years. Generally, physiciansreported having a solo practice or practicing as part of aprimary care group, though one-third of OB/GYNs indi-cated they used some other arrangement. There was vari-ability in physician-reported patient insurance coverageacross the physician specialties, but private/commercialinsurance was the most common, ranging from 53.3% to72.1% depending on the specialty. According to physi-cians, less than 10% of their patients were covered byMedicaid, and 17% to 41% were Medicare beneficiaries.Across all specialties, less than 10% of patients werereported by physicians to have no health insurance cover-age. The majority of physicians (63.6%) reportedly sawtheir PMO patients on an annual basis, and 66.4% indi-cated that they always or almost always discussed theimportance of adherence to treatment.

Patient characteristics

The 412 physicians with complete surveys treated 5173patients for PMO and 1587 of those patients had 12months of health plan enrollment following their indexdate; 30.7% of patients met the inclusion criteria. To beconsistent with the physician survey question that askedphysicians to estimate patient adherence after 1 year oftherapy, the descriptive analysis focused on those patientswith 12 months of enrollment following their index date.Overall, nearly three-quarters of the patient populationreceived a weekly oral bisphosphonate as treatment forosteoporosis. Patient sociodemographics stratified byadherence are given in Table 2. To account for multiplecomparisons within Table 2, alpha was reduced to 0.01 viaa Bonferroni correction22. Overall, patients in the adher-ent population were slightly younger than patients in thenon-adherent population (61.5 years vs. 63.1 years), and

patients in the non-adherent population were significantlymore likely to be �75 years (17.8% vs. 11.1%; p¼ 0.0002)(data not shown).

Physicians’ perceptions of patient adherence

The average overall percentage of patients who wereadherent as estimated by physicians was compared to theobserved percentage of adherent patients based on theclaims data (Figure 1). Overall, physicians’ averaged esti-mate of the percentage of adherent patients was 69.2%,whereas 48.7% of patients were observed to be adherent(Table 3). They overestimated the percentage of adherentpatients regardless of specialty, but physicians specializingin internal medicine demonstrated the most notable dif-ference between perception and observed adherence(Table 3). For physicians specializing in internal medicine,there was a gap of 26.2 percentage points between theperceived and observed percentages of adherent patients,whereas the gap was 15.6 percentage points for generalpractice physicians and 16.3 percentage points for OB/GYNs. Male and female physicians were equally likely tooverestimate patient adherence. Similar differencesbetween physicians’ perceptions and patients’ observedadherence were found in sensitivity analyses that usedpatients with 6 months of continuous enrollment.

Within all medication classes, the difference betweenphysicians’ perceptions of their patients’ adherence andthe adherence of their patients as estimated by claimsdata was found to be statistically significant (Figure 1).Physicians with patients prescribed weekly oral bispho-sphonates estimated that 72.4% of their patients are adher-ent after 12 months of therapy, whereas 50.0% of theirpatients were adherent according to the claims data(p50.01). Similarly, physicians with patients prescribedmonthly oral bisphosphonates perceived 72.4% of patientsto be adherent as compared to the 32.4% who wereobserved to be adherent (p50.05). The differencebetween physicians’ perceptions of adherence andpatients’ observed adherence also held for patients

Table 2. Patient demographics by adherence category.

All Patients Adherent Patients Non-adherent Patients p-valuez

Number of patients 1587 773 814Age in years (SD) 62.3 (9.8) 61.5 (9.1) 63.1 (10.3) 0.001*Race/Ethnicity, n (%)

White/Caucasian 904 (60.0) 455 (58.9) 449 (55.2) 0.137yBlack/African American 21 (1.3) 6 (0.8) 15 (1.8) 0.063yAsian/Pacific Islander 62 (3.9) 22 (2.9) 40 (4.9) 0.034yHispanic/Latino 8 (0.5) 2 (0.3) 6 (0.7) 0.179yOther 84 (5.3) 40 (5.2) 44 (5.4) 0.837yMissing 508 (32.0) 248 (32.1) 260 (31.9) 0.952y

yChi-square test.*Independent samples Student’s t-test.zBonferroni correction based on 5 variables, 3 are listed here.SD¼ standard deviation.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

prescribed SERM, with physicians reporting 70.4% patientadherence and claims data showing 47.6% of patients asadherent (p50.01).

OLS modeling revealed physicians’ experience (as mea-sured by the number of years in practice) and the estimatedproportion of their patients receiving a prescription drug asPMO therapy to be significant predictors of physicianperception of patient adherence. Physicians’ perceptionsof adherence to weekly oral and monthly oral bisphospho-nates were modeled separately. Within the weekly oralbisphosphonate model (Table 4), physician experiencewas found to be a statistically significant predictor of phy-sician perception of patient adherence. Physicians in prac-tice 10 to 14 years expected 4.3% fewer patients to beadherent after 1 year than did physicians in practice forless than 10 years (p¼ 0.038), all else held constant.Likewise, those in practice for 20 years or more expected

4.1% fewer patients to be adherent (p¼ 0.013). In addi-tion, physicians treating their patient population with aprescription drug therapy for PMO were inclined tobelieve that their patients were adherent to those therapies(�¼ 0.2, p¼ 0.066). Stated another way, for each 25%increase in the percentage of PMO patients whom physi-cians treat with a prescription drug therapy, there was a 5%increase in the physicians’ expected number of adherentpatients (p50.001). Similar results were found within themonthly oral bisphosphonate model (Table 5). Physicianexperience had a more pronounced effect on their percep-tions, with decreases of 8.1% (p¼ 0.023), 12.5%(p¼ 0.014), and 8.9% (p50.001) in percentages ofexpected adherent patients for the physicians with 10 to14 years, 15 to 19 years, and 20 or more years of experience,respectively, as compared with physicians with less than10 years of experience, all else held constant.

Figure 1. Percentages of perceived vs. observed adherent patients by medication class.

100

75

50

25

0

72.4

49.5 50.0**

72.4

35.0 32.4*

46.753.0

40.0

TeriparatideWeekly oralbisphosphonates

Monthly oralbisphosphonates

Selective estrogenreceptor modulators

(SERM)

Medication class

Physician perception

Physician perception vs. Weighted claims-based differs significantly with: *p-value < 0.05, **p-value < 0.01

Claims-based Weighted claims-based

Per

cent

age

of a

dher

ent p

atie

nts

(%)

40.0

47.6**

70.4

Table 3. Physician perceptions of patient adherence vs. prescription refills in pharmacy claims.

Average Perceived % of Adherent Patients Observed % of Adherent Patients Difference inPercentages

% y CI % y CI

Overall 69.2 1528 (68.0, 70.3) 48.7 773 (46.2, 51.2) 20.5Physician specialty

General practice 66.2 405 (63.9, 68.5) 50.6 199 (45.7, 55.6) 15.6Internal medicine 69.7 709 (68.0, 71.4) 43.5 310 (39.9, 47.2) 26.2OB/GYN 71.1 414 (69.0, 76.3) 54.8 264 (50.3, 59.2) 16.3

Physician genderFemale 69.3 553 (67.3, 71.3) 48.5 256 (44.2, 52.8) 20.8Male 69.1 967 (67.7, 70.5) 48.9 510 (45.9, 51.9) 20.2

yNumber of patients within the specified strata (i.e., the denominator of the percentage).CI¼ 95% confidence interval.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

Physician characteristics also have an impact onpatient adherence (Table 6). The logistic regression onclaims-based patient adherence showed that patientsbeing seen by a physician in a primary care group setting,

as compared to any other setting, were less likely to beadherent to their medication (p¼ 0.051), although thedifferences in physician perceptions between practice set-tings were not statistically significant. In addition, patientsseen by a physician with less than 5 years of experiencewere significantly less likely to be adherent as compared topatients seen by physicians with 5 or more years in practice(p¼ 0.001). Patient age was also found to be significantlyassociated with patient adherence (p50.001). Modelcoefficients suggest that as age increases, the likelihoodof adherence decreases.

Reasons for adherence/non-adherence, andfactors to motivate adherence

Physicians were asked to rank up to five reasons theybelieve patients do not adhere to their medications, andfive reasons they believe patients do adhere to their med-ications (data not shown). Physicians reported that med-ication side effects and affordability of medication are thegreatest barriers to patient adherence. The top reasons,meaning reasons which were cited by the greatest propor-tion of physicians for patients’ adherence to their therapywere: fear of fracture, patient’s belief that the therapy iseffective, and patient involvement in treatment decisions.Additionally, physicians were asked to rank the top factorsthey believe would increase patients’ adherence to medi-cation. Physicians indicated that lowering medicationcost, dispensing medications with fewer side effects, invol-ving patients in the treatment decision-making process,and prescribing medications requiring less frequentdosing are most likely to improve patient adherence.

Discussion

The objective of this study was to examine physicians’perceptions of patient adherence to medications used totreat osteoporosis. This was achieved using a comprehen-sive study design that combined primary data collectionfrom providers matched to retrospective claims of thephysicians’ patients. To the authors’ knowledge, this isthe first study of this kind in osteoporosis.

These results suggest that physicians routinely overes-timate patient adherence to medications. The overestima-tion is reflected by the overall average physicianestimate of adherent patients, 69.2%, compared to48.7% of patients identified as adherent from the claims.Regardless of physician specialty or physician gender, phy-sicians perceive their patients to be more adherent to theirtherapy than is observed from claims data. Statistical mod-eling showed physicians with more experience estimatedfewer patients as adherent after 12 months of therapy.

Table 6. Logistic regression on patient adherence to weekly and monthlyoral bisphosphonates*.

Parameter OddsRatio

95% ConfidenceInterval

p-value

Physician perceptionof adherence tomonthly and weekly oralbisphosphonates

0.77 (0.59, 1.01) 0.059(NS)

Age 0.98 (0.96, 0.99) 50.001Years in practicey55 0.22 (0.08, 0.55) 0.001

Practice settingzPrimary care group 0.79 (0.62, 1.00) 0.051

(NS)

*The odds of being adherent are modeled.n¼ 1221.Pseudo r2

¼ 0.02.yReference group is �5 years in practice.zReference group is all other settings (solo practice, multi-specialty group,some other arrangement).NS¼ not significant.

Table 4. OLS regression on physician estimate of proportion of patientsadherent to weekly oral bisphosphonates.

Parameter Estimate p-value

Physicians’ estimated percentage ofPMO patients receiving PMOprescription drug therapy

0.2 50.001

Years in practicey10–14 �4.3 0.03815–19 �1.1 0.599�20 �4.1 0.013

Intercept 58.9 50.001

n¼ 402.r2¼ 0.08.yReference group is510 years in practice.OLS¼ ordinary least squares; PMO¼ postmenopausal osteoporosis.

Table 5. OLS regression on physician estimate of proportion of patientsadherent to monthly oral bisphosphonates.

Parameter Estimate p-value

Physicians’ estimated percentage ofPMO patients receiving PMOprescription drug therapy

0.2 0.066

Years in practicey10–14 �8.1 0.02315–19 �12.5 0.014�20 �8.9 50.001

Intercept 68.0 50.001

n¼ 171.r2¼ 0.08.yReference group is510 years in practice.OLS¼ ordinary least squares; PMO¼ postmenopausal osteoporosis.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

By contrast, physicians who prescribe more PMO treat-ments estimate a greater number of patients as adherent.

Reasons cited by physicians for patient non-adherenceand factors they considered likely to contribute toimproved adherence were also explored in this study.Surveyed physicians reported that medication side effectsand affordability of medication are the greatest barriers topatient adherence, and that lowering medication cost, dis-pensing medications associated with fewer side effects,involving patients in the treatment decision-making pro-cess, and prescribing medications requiring less frequentdosing are most likely to improve patient adherence.

Consistent with the surveyed physicians’ responses inthe present study, the results of a recent InternationalOsteoporosis Foundation (IOF) survey conducted among500 physicians and 502 women aged� 60 years with PMOindicate that side effects and inconvenience were top rea-sons women gave for stopping therapy23. In fact, four out offive women in the survey expressed an interest in less fre-quent dosing, and 75% of the physicians surveyed felt thatless frequent dosing would lead to more convenience andimproved adherence. Additionally, a prospective follow-up study of 4448 women in Israel using bisphosphonatesfound that weekly dose users were significantly more likelyto remain on therapy than daily dose users (p50.001)24.Taken together, this evidence suggests that the develop-ment of new osteoporosis medications with a better sideeffect profile and medications that require less than dailydosing may improve patient adherence.

The overestimation of patient adherence by the sur-veyed physicians in the present study highlights a needfor increased awareness among physicians of patientnon-adherence to medication, especially because non-adherence can lead to increased fracture. Physicians maymake more of an effort to open up a dialogue with patientsand explain the benefits and importance of treatment ifthey are made aware that adherence is actually much lowerthan they perceive. Such dialogue appears to be lacking inthe primary care setting. Improved communicationbetween patients and physicians should ultimately leadto improved adherence and, consequently, fewer fractures.

Due to challenges associated with fragmented healthcare delivery and a lack of continuity in patient care, phy-sician communication regarding medication adherencetakes on major importance. Physicians may not be ableto give adequate time to explanations regarding the ben-efits and side effects of medications. Additionally, suchconstraints limit considerations regarding the patient’slifestyle or cost of medications. Adding to the challengeof patient care are health care systems with formularyrestrictions that limit medications and patients whooften switch to new health care plans with high-costdrugs and new co-payment arrangements25.

This study suggests that future physician interventionstrategies designed to improve doctor–patient

communication are important for patient adherence toosteoporosis medications. Any efforts on behalf of physi-cians to enfranchise the patient in treatment decisionswhen combined with recognition of potential overestima-tion of patient adherence to medications is likely to helpimprove medication utilization and patient outcomes.Moreover, this study reveals that managed care organiza-tions have an opportunity to generate greater awareness ofpatient medication adherence as well as initiate educationprograms for providers focused on understanding andimproving patient adherence.

Findings must be considered within the limitations ofthis database analysis. Physicians who chose to participatemay not be representative of all physicians who treat osteo-porosis patients, and patients included in the study are notnecessarily representative of all patients being treated forosteoporosis. This study focused on a more homogenousgroup of prescribers; however, nurse practitioners and phy-sician assistants also prescribe treatment. The overallresponse rate among physicians was 21.8%. However, theauthors did not identify any statistically significant demo-graphic differences between responders and non-responders indicating that physicians who responded aredifferent from the physicians who chose not to complete asurvey. In addition, physicians rated the adherence of all oftheir osteoporosis patients, not of individual patients whowere included in the adherence estimate based on claimsdata. Another factor influencing estimates of patientadherence is the inclusion of only those patients whohad at least 12 months of follow-up after their indexdates. It is possible that some patients become less adher-ent over time; thus, the patients included in the adherenceanalysis might have had a lower MPR, on average, than thepatients who were excluded due to insufficient length offollow-up. However, sensitivity analyses using 6-monthdata led to the same conclusion as the 12-month analysis.

There are also some limitations inherent to the use ofclaims data. Claims data are collected for the purpose ofpayment and not research, and therefore, some degreeof incomplete or incorrect coding may exist. Further, thepresence of a claim for a filled prescription does not indi-cate that the medication was consumed or taken as pre-scribed. In this case, the methodology used would haveoverestimated patient adherence and the actual differencebetween physician perception of patient adherence andobserved patient adherence would have been larger thanreported. Thus, the results reported are conservative andbias towards the null hypotheses. Conversely, patient useof medication samples during the observation period wouldhave underestimated the MPR by decreasing the numberof pharmacy fills. In that case, the difference between phy-sician perception and observed patient adherence mayhave been smaller than reported and biased away fromthe null hypotheses.



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

Conclusions

The results of this study suggest that regardless of physicianspecialty or gender, physicians overestimate patient adher-ence to therapies for PMO. Physicians’ overestimation ofpatient adherence persists across medication classes.Overall, one out of every four patients may be incorrectlyassumed by their physicians to be adherent. Becausenon-adherence to PMO medications can have an adverseeffect on patient outcome, especially fracture risk, it isimportant that physicians become more aware of theactual level of non-adherence among their patientsso that they will be more inclined to open up a dia-logue with them about the benefits and importance oftreatment.

TransparencyDeclaration of fundingThis study was funded by Amgen, Inc.

Declaration of financial/other relationshipsR.C., P.B. and V.Z. have disclosed that they are employees of i3Innovus, which received grant support from Amgen. S.I. andD.M. have disclosed that they are employees of, and are stock-holders of, Amgen. L.K. was involved in the analysis and writingof the study and serves as a consultant to Amgen.

Some peer reviewers receive honoraria from CMRO for theirreview work. The peer reviewers of this paper have disclosed thatthey have no relevant financial relationships.

AcknowledgmentThe authors thank Victoria Porter and Jesse Potash for theirassistance with the preparation of this manuscript.

This study was presented in poster form at the Academy ofManaged Care Pharmacy 2009 Educational Conference, October7–9, 2009, San Antonio, TX, USA. ‘Copher R, Buzinec P, IqbalSU, Macarios D, Zarotsky V. Physician survey of patient adher-ence to osteoporosis medications.’

References1. National Osteoporosis Foundation. Fast Facts 2006. Washington, DC: National

Osteoporosis Foundation. Available at: www.nof.org/osteoporosis/disease-

facts.htm [Last accessed 11 May 2009]

2. Burge R, Dawson-Hughes B, Solomon DH, et al. Incidence and economic

burden of osteoporosis-related fractures in the United States, 2005–2025.

J Bone Miner Res 2007;22:465-75

3. Gupta G, Aronow WS. Treatment of postmenopausal osteoporosis [review].

Compr Ther 2007;33:114-19

4. Caro JJ, Ishak KJ, Huybrechts KF, et al. The impact of compliance with

osteoporosis therapy on fracture rates in actual practice. Osteoporos Int

2004;15:1003-8

5. Siris E, Rosen CJ, Harris ST, et al. Adherence to bisphosphonate therapy:

relationship to bone fractures at 24 months in women with postmenopausal

osteoporosis. Abstract 397. Presented at the Sixth International Symposium

on Osteoporosis, National Osteoporosis Foundation, Washington, DC, April 7,

2005

6. Sebaldt RJ, Shane LG, Pham BZ, et al. Impact of non-compliance and

non-persistence with daily bisphosphonates on longer-term effectiveness

outcomes in patients with osteoporosis. Poster M423. Presented at the

26th Annual Meeting of the American Society for Bone and Mineral

Research, Seattle, Washington, October 1–5, 2004

7. Thier SL, Yu-Isenberg KS, Leas BF, et al. In chronic disease, nationwide data

show poor adherence by patients to medication and by physicians to guide-

lines. Manag Care 2008;17:48-52,55-7

8. Warriner AH, Curtis JR. Adherence to osteoporosis treatments: room for

improvement. Curr Opin Rheumatol 2009;21:356-62

9. Gleeson T, Iversen MD, Avorn J, et al. Interventions to improve adherence and

persistence with osteoporosis medications: a systematic literature review.

Osteoporos Int 2009; Jun 5 [Epub ahead of print]

10. Kothawala P, Badamgarav E, Ryu S, et al. Systematic review and meta-

analysis of real-world adherence to drug therapy for osteoporosis. Mayo

Clin Proc 2007;82:1493-501

11. McCombs JS, Thiebaud P, McLaughlin-Miley C, et al. Compliance with drug

therapies for the treatment and prevention of osteoporosis. Maturitas

2004;48:271-87

12. Lekkerkerker F, Kanis JA, Alsayed N, et al. Adherence to treatment of osteo-

porosis: a need for study. Osteoporos Int 2007;18:1311-17

13. Rabenda V, Hilligsmann M, Reginster JY. Poor adherence to oral bispho-

sphonate treatment and its consequences: a review of the evidence. Expert

Opin Pharmacother 2009;10:2203-15

14. Siris ES, Selby PL, Saag KG, et al. Impact of osteoporosis treatment adherence

on fracture rates in North America and Europe. Am J Med 2009;122

(2 Suppl.):S3-13

15. Yood RA, Emani S, Reed JI, et al. Compliance with pharmacologic therapy for

osteoporosis. Osteoporos Int 2003;14:965-8

16. Delmas PD, Vrijens B, Eastell R, et al. Effect of monitoring bone turnover

markers on persistence with risedronate treatment of postmenopausal osteo-

porosis. J Clin Endocrinol Metab 2007;92:1296-304

17. Eastell R, Garnero P, Vrijens B, et al. Influence of patient compliance

with risedronate therapy on bone turnover marker and bone mineral density

response: the IMPACT study. Calcif Tissue Int 2003;72:408. Abstract P-297

18. Gold DT, Alexander IM, Ettinger MP. How can osteoporosis patients benefit

more from their therapy? Adherence issues with bisphosphonate therapy. Ann

Pharmacother 2006;40:1143-50

19. Peterson AM, Nau DP, Cramer JA, et al. A checklist for medication compliance

and persistence studies using retrospective databases. Value Health

2007;10:3-12

20. Sikka, R, Xia F, Aubert RE. Estimating medication persistency using admin-

istration claims data. Am J Manag Care 2005;11:449-57

21. Andrade SE, Kahler KH, Frech F, et al. Methods for evaluation of medication

adherence and persistence using automated databases. Pharmacoepidemiol

Drug Saf 2006;15:565-74

22. Miller Jr RG. Simultaneous Statistical Inference. New York: Springer-Verlag,

1991

23. International Osteoporosis Foundation. The adherence gap: why osteoporosis

patients don’t continue with treatment. Available at: http://www.iofbone-

health.org/publications/the-adherence-gap.html [Last accessed 16 March

2009]

24. Kertes J, Dushenat M, Vesterman JL, et al. Factors contributing to compliance

with osteoporosis medication. Isr Med Assoc J 2008;10:207-13

25. Osterberg L, Blaschke T. Drug therapy: adherence to medication. N Engl J

Med 2005;353:487-97



Cur

r M

ed R

es O

pin

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f C

alif

orni

a Sa

n Fr

anci

sco

on 1

1/28

/14

For

pers

onal

use

onl

y.

physician perception of patient adherence compared to patient adherence of osteoporosis medications...

Documents