Patient Education and Counseling 78 (2010) 372–376

Contents lists available at ScienceDirect

Patient Education and Counseling

journal homepage: www.elsevier.com/locate/pateducou

The association of health literacy and socio-demographic factors with

medication knowledge
Jennifer R. Marks, Joel M. Schectman, Hunter Groninger, Margaret L. Plews-Ogan *
University of Virginia, Division of General Internal Medicine, VA, USA

A R T I C L E I N F O A B S T R A C T

Article history: Objective: To compare patient demographics and Rapid Estimate of Adult Literacy in Medicine (REALM)
Received 15 January 2009 scores with respect to their ability to predict medication comprehension.
Received in revised form 27 April 2009 Methods: A survey was conducted of 100 patients presenting for follow-up at an academic primary care
Accepted 11 June 2009
clinic serving a low socio-economic status population. The Medication Knowledge Score (MKS) consisted
of knowledge of drug name, dose, indication, and a potential side effect for each of their medications and
Keywords: then averaged. The REALM (Rapid Estimate of Adult Literacy in Medicine) was administered and socio-
Medication knowledge
demographic characteristics were recorded. The association of REALM score and patient characteristics
REALM
Health literacy
with MKS was evaluated by univariate and multivariable regression analysis.
Patient education Results: The subjects’ mean age was 62 with an average of 9.8 years of schooling and 5.9 prescription
Safety medications. Participants identified a correct indication for 78.8% of their medications and correct
Medication adherence dosage for 93.4%. However, they could provide the name for only 55.8% of medications and a known side
Underserved populations effect for only 11.7%. On multivariate analysis without including REALM score, younger age (p = .01),
highest grade completed (p = .001), and female sex (p = .004) remained positively associated with MKS.
When the model included REALM, REALM (p < .0001), age (p = .001), and sex (p = .04) remained
independently associated with MKS.
Conclusion: REALM score predicts medication knowledge as assessed by the MKS. However, age, last
grade completed, and sex were also independently associated with mean MKS with a similar strength of
association to that of REALM. This suggests that simpler cues to screen for medication knowledge deficits
may also be useful. Since the MKS incorporates knowledge of medication indications and side effects, it
may also be useful for quality and safety purposes.
ß 2009 Elsevier Ireland Ltd. All rights reserved.

There is a growing appreciation of the frequency and
importance of medication errors and adverse drug events in
healthcare [1–4]. Patients with low health literacy are likely at
higher risk for such adverse outcomes related to misunderstand-
ings at the doctor–patient, patient–pharmacist, and/or patient–
medication levels. Identifying patients with low health literacy
would allow targeting of stronger and more appropriate medica-
tion education efforts to promote better treatment outcomes.
The Rapid Estimate of Adult Literacy in Medicine (REALM) is a
screening tool assessing the ability to correctly read 66 commonly
used lay medical terms. It has been validated to help identify
patients at risk for poor health literacy [5]. Since its creation, it has
been applied to a variety of populations and clinical situations to
examine the relationship between health literacy and health
outcomes. For example, lower REALM scores have been shown to
* Corresponding author at: Division of General Internal Medicine, University of
Virginia Health System, PO Box 800744, Charlottesville, VA 22908-0744, USA.
Tel.: +1 434 924 1685; fax: +1 434 924 1138.
E-mail address: mp5k@virginia.edu (M.L. Plews-Ogan).
correlate with poor knowledge of asthma and metered-dose
inhaler use by asthmatics [6] and poor knowledge about health
effects of tobacco abuse in smokers [7]. Low health literacy has also
been correlated with increased outpatient visits in rheumatoid
arthritis patients [8], hospitalizations in older patients [9], and
emergency room visits in congestive heart failure patients [10]. It is
also associated with poor glycemic control in diabetics [11],
variable effectiveness in patients taking warfarin [12], and low
adherence to and understanding of medications [13,14].
Lower socio-economic status and older age appear to be
independent risk factors for medication errors and adverse drug
events [15,16]. Health literacy and/or medication knowledge may
be underlying factors. However, surveying actual medication
knowledge is too painstaking and inefficient to be incorporated
into routine clinical practice. Also no standardized assessment tool
to adequately represent medication knowledge exists [17–19].
Ideally, a simple surrogate screening tool could quickly identify at-
risk patients for further assessment of their medication knowledge.
Other studies seeking a surrogate screening tool have largely
used DRUGS, Drug Regimen Unassisted Grading Scale, which

0738-3991/$ – see front matter ß 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.pec.2009.06.017
 J.R. Marks et al. / Patient Education and Counseling 78 (2010) 372–376 373

assesses patients’ abilities to identify their medications, open
medication containers, provide correct dose, and report correct
timing of dose(s) [22]. This tool does not assess patients’
knowledge of medication indications or potential side effects as
the MKS does. The current study evaluates the association of the
REALM as well as other socio-demographic characteristics with
medication knowledge including knowledge of medication indica-
tions and potential side effects in an at-risk (low socio-economic
status) patient population. Patients’ understanding of this infor-
mation may have implications for medication adherence and
safety.
indicative of elementary, middle school, and high school level
health literacy) categorical variable without substantial impact on
the results. We therefore chose to model REALM as a continuous
variable for the principal analyses. The independent variables with
a p < .10 by univariate analysis were entered into a multivariable
regression model with MKS as the dependent variable. In the
multivariate model, a p-value of less than .05 was required for
statistical significance with a stepwise regression procedure
utilized to create the most parsimonious model. Models with
and without REALM score as an independent variable were
compared. All analyses were performed with SAS v9.1.

1. Methods 2. Results

The study was conducted in the primary care internal medicine
clinic at the University of Virginia. The clinic is the primary
ambulatory training site for the internal medicine residency
training program and serves a predominantly low socio-economic
status population. Forty percent of the 7500 active patients have no
health insurance and 45% have incomes below the federal poverty
line. Patients are asked to bring all of their medications
(prescription and non-prescription) to routine office visits.
Adherence to this policy among continuity patients is above 50%.
To be included in the present study, each participant had to
bring all medications to the appointment, speak English fluently,
be at least 18 years old, and not have a diagnosis of dementia or
delirium. In December, 2004, patients waiting to be seen by their
primary care physician and who fit study criteria were asked if they
would participate in an anonymous quality-improvement survey
regarding medication comprehension. The participation rate was
95%. If a given patient’s medications were administered by another
individual (e.g., spouse and child), then that person completed the
survey instead. The survey was orally administered by one of the
study authors. The first part consisted of demographic items and
self-reported health literacy level (‘‘no difficulty reading medica-
tion labels’’, ‘‘some difficulty reading medications’’, and ‘‘cannot
read’’—the latter two categories were combined in the analysis).
For the second part, each participant was asked to pick up each
medication sequentially and give the following information: (1)
name of medication (missed syllables and mispronunciations were
accepted), (2) dosage (either the milligram strength or the number
of tablets/capsules and the frequency), (3) indication (any
condition for which the medication could be used was accepted),
and (4) any known side effect (adjudicated by referencing common
or serious side effects listed in the MicroMedex DrugPoint
Summary and by the authors). For each medication, a Medication
Knowledge Score (MKS) was created corresponding to the number
of correct answers of a possible four (e.g., knowing a drug’s name
and dosage but not an indication or side effect resulted in a MKS of
2 for that drug). Interviewers also observed and noted how patients
identified medicines (bottle label, color/shape/size, or both). The
Table 1 provides socio-demographic characteristics of the
patient sample. One hundred subjects participated in the survey
interview and complete data was obtained for 98 of them. Their
average age was 62 years with a mean of 9.8 years of formal
education. They took an average of 5.9 prescription and non-
prescription medications on a regular basis. The mean REALM
score was 43 with 41% of subjects scoring less than 45 (reading
level at or below 6th grade), 18% scoring 45–60 (7–8th grade
reading level), and 41% scoring above 60 (9th grade or higher
reading). Of the medications taken by all participants, individuals
were able to provide the correct drug name for only 55.8%, the
correct dosage for 93.4%, the correct indication for 78.8%, and at
least one appropriate side effect for only 11.7%. Participants
identified 89.0% of medications by the bottle label, 6.2% by
appearance, and 4.7% by both methods. Overall, the mean MKS was
2.40 (SD = 0.78, range 0.2–4.0) with the distribution of scores
shown in Fig. 1. The distribution of each of the 4 components of the
MKS was highly skewed, but each was strongly associated with the
mean MKS as well as statistically significantly associated with the
REALM score (Table 2).
By univariate analysis, mean MKS was inversely associated with
age and directly associated with schooling, female sex, and REALM
score (Table 3). Administering medications for the patient (i.e.,
being a surrogate survey participant) also was associated with a
lower mean MKS. The associations between mean MKS and total
number of medications, race, or self-reported literacy were not
statistically significant.
We initially performed a multivariate analysis without includ-
ing the REALM score (Table 4). Patient age, last grade completed,
Table 1
Subject characteristics.
Mean Range Standard Deviation
Age (years) 62 27–90 12.6
Last grade completed 9.8 0–18 3.2
# prescription meds 5.9 1–15 3.1
REALM 43 0–66 24

MKS took from one to five minutes to complete, depending on the # of subjects
number of medications. Following this, each participant was Sex
administered the REALM test, which took an additional three Male 47
minutes, on average. The REALM was chosen due to its brevity and Female 53
prior experience in numerous other health literacy studies utilizing Race
this tool as referenced above. Asian 1
The overall MKS for each participant was calculated as the mean Black 52
MKS for all their prescription and non-prescription medications. White 47

The association of each of the individual components of the MKS Assistance with medication from others
with the overall MKS as well as REALM score of each subject was Yes 18
assessed by non-parametric correlation analysis (using the No 82

Spearman rho). In a subject level analysis utilizing linear modeling, Self-reported health literacy
we evaluated the association of MKS with age, sex, schooling, self- Able to read and understand medication labels 89
Some difficulty reading medication labels 7
reported literacy, and REALM score. The REALM was modeled as
Unable to read labels 3
both a continuous and tri-level (0–44, 45–60, 61–66 respectively
374 J.R. Marks et al. / Patient Education and Counseling 78 (2010) 372–376

Fig. 1. Distribution of MKS Scores.

and sex independently predicted mean MKS with an overall model discriminant power. In our sample, subjects with an eighth grade
R-square of 0.33 (p < .0001). When REALM was added to the education or lower had a threefold higher risk of a low MKS (below
model, the R-square improved to 0.41 with age, gender, and REALM median) than those with a high school (or higher) education, those
remaining independently predictive. Though multi-collinearity over age 70 were 2.6 times more likely to have a low MKS than
was clearly present, tolerance values for all independent variables subjects under 55 years of age, and men were 1.7 times more likely
in the model were greater than 0.6. Last grade completed (which than women to have an MKS below the median of 2.54.
dropped from the model including REALM) and REALM were highly
correlated (Pearson correlation coefficient = 0.55, p < .0001). 3. Discussion and conclusion
The discriminant ability of various thresholds for age, grade,
REALM score, as well as sex for identifying patients with MKS below 3.1. Discussion
the median (2.54 on scale of 0–4) for the sample is shown in Table 5.
As suggested by the regression analyses, the REALM score performed We found a wide variability in medication knowledge in this
well in this regard but age, grade, and sex also had reasonable population of lower socio-economic status. The REALM score was
the strongest predictor of medication knowledge (as assessed by
MKS) among the variables we examined. This finding not
Table 2
surprisingly indicates that health literacy plays a major role in
The association of each of the four components of the Medication Knowledge Score
with their composite (mean MKS) and with the REALM score. comprehending the names, dosages, indications, and potential side
effects of one’s medications, prescription or otherwise. Our
Mean MKS REALM score
analysis also shows that four other parameters – assistance in
ra p-Value ra p-Value medication administration, age, educational attainment, and sex –
Name of medication 0.87 <.0001 0.69 <.0001 also are predictive of mean MKS with the latter 3 independently
Dosage of medication 0.54 <.0001 0.22 .02 predictive in a multivariable regression model. Interestingly,
Purpose of medication 0.75 <.0001 0.42 <.0001 patient self-reporting of health literacy was not predictive of
Potential side effect 0.54 <.0001 0.31 .001
MKS, although there was a trend toward significance, but sample
a
Non-parametric correlation coefficient and corresponding statistical signifi- size was small. The combination of these three demographic
cance: ‘‘r’’ indicates Spearman rho.
factors, age, gender, and highest grade level, had similar

Table 3
Patient characteristics associated with mean Medication Knowledge Score by univariable analyses.

Variable Parameter estimate p-Value R-square

Age 0.02 .0009 0.11

Last grade completed 0.125 <.0001 0.24
# medications 0.013 .57 –
REALM 0.019 <.0001 0.33a

Mean MKS p-Value R-square

Sex F 2.64 .0006 0.11

M 2.19

Race Asian 2.00 .64 –

Black 2.34
White 2.47

Assistance with medication Yes 1.90 .005 0.08

No 2.49

Self-reported Yes 2.43 .11 –

Literacya No 2.00
a
Pts reporting ‘‘some difficulty reading medications’’ and ‘‘cannot read’’ were grouped into ‘‘no’’ category.
 J.R. Marks et al. / Patient Education and Counseling 78 (2010) 372–376 375

Table 4
Patient characteristics independently associated with mean Medication Knowledge Score by multivariable analyses (one excluding REALM score).

Variable Model 1: without REALM (R-square = 0.33) Model 2: with REALM (R-square = 0.41)

Parameter estimate p-Value Parameter estimate p-Value

Assistance with medications – NSa – NSa

Sex 0.42 .004 0.29 .04
Self-reported literacy – NSa – NSa
Age 0.015 .01 0.02 .001
Last grade completed 0.08 .001 – NSa
REALM – – .015 <.0001
a
NS: not statistically significant at .05 level—all NS terms were dropped from the model.

Table 5 association between educational background and the ability to

Subject characteristics of those with mean MKS below median value of 2.54.
understand medication labels among parents of pediatric patients.
# of subjects MKS below median (%) In our study to assess medication knowledge we developed a
Sex Medication Knowledge Score. Previous studies of medication
M 46 63 knowledge have utilized questionnaires aimed at exploring
F 52 38 various epistemological components (e.g., drug name, indication,
Age (years) side effect profile and pill appearance) without a standardized
<55 28 29 survey set [9,14,25,26]. Similarly, our 4-component MKS aimed to
55–69 43 49 assess medication knowledge practically and efficiently. Vis-à-vis
70 27 74 content validity, we felt that knowledge of a drug’s name, dosage,
Last grade completed and indication were essential components and have been used by
8th 29 79 prior investigators.
9–11th 34 50 We also believe that patient awareness of potential drug side
12th 35 26
effects is important to medication knowledge by providing a
REALM score window on another dimension of medication safety. Indeed, in
<45 40 80
their study of adverse drug events in ambulatory care, Gandhi et al.
45–60 18 56
>60 40 18
[3] found an overall event rate of 27 events per 100 patients.
Twenty-eight percent of these events were judged ameliorable,
due to either the patient’s failure to report side effects to their
explanatory power (as measured by R-square) of MKS to that of the doctor or the physician’s failure to respond to medication-related
REALM. Adding the REALM score to the multivariate model adverse effects. Although we had no way of knowing if a patient
increases its explanatory power, but the increase is modest (R- had previously been made aware of any side effects we found that
square = 0.41 vs. 0.33). This finding suggests two things. First, knowledge in this area was quite low.
when taken together, age, sex, and education may be as effective at In addition to side effect knowledge having patient safety
screening for medication knowledge deficits as the REALM alone. implications, this knowledge may also affect medication adher-
Second, while including the REALM test can augment prediction of ence. Hill et al. [28] found that patients who received extra
mean MKS for a given patient, the gain may not be substantial counseling regarding their medications including side effects were
enough to warrant the effort entailed in utilizing the REALM as a more likely to be compliant than those who did not receive such
universal screening tool for medication knowledge. However, the education. The pervasive lack of patient side effect knowledge, as
use of patient age, sex, and education level as predictive filters may illustrated in our study’s findings as well as those of Hill, underline
help to determine high-risk subgroups that might warrant further the importance of patient education in this area for purposes of
specific screening with REALM or other measures. both improved safety as well as possibly adherence. In terms of
The results of our study are similar to that of another study by adherence more recently Gazmararian et al. also found a
Kripalani et al. [22]. In a general medical clinic of a large urban relationship between health literacy and refill adherence amongst
teaching hospital serving an indigent largely African-American Medicare enrollees [29].
population, they found that ‘Medication Management Capacity’ Construct validity for this scale is indicated by close association
was associated with health literacy as assessed by the REALM as with other variables known to be associated with low medication
well as with educational attainment and patient age (inversely), knowledge such as low health literacy, education, and advancing
though not gender. In contrast, their assessment did not include age. The fact that the MKS assumed a fairly normal distribution is a
medication indication or side effect knowledge. It also excluded useful feature if replicated in other populations where concern
those patients who received assistance with their medications. A exists about medication knowledge. Utilizing a standard scale that
number of other studies in various populations have also measures the important parameters of medication knowledge will
demonstrated associations between patient demographic factors be important in future studies in this area. The MKS may serve this
and ability to manage medications. Edelberg et al. [23] similarly purpose, but our results require replication in other settings as well
found an inverse association between age and medication as more socio-economically diverse populations. In addition to
management capacity among geriatric subjects living in a being a small study conducted in one relatively indigent setting,
retirement community. Another study by Davis et al. [24] utilizing there are other limitations that may hinder the generalizability of
a standardized medication set (i.e., not the patients’ own our findings. First, we utilized a convenience sample of patients
medications) to assess the ability to understand medication labels who all spoke English and had brought their medications to clinic.
also found a strong association with REALM score and similar Though the results cannot be extrapolated to non-English speakers
trends with respect to sex and educational attainment to those we or those that did not bring their medications, our participation rate
report (their study was limited by very few patients with a less (95%) was very high and our clinic has good adherence (>50%) to a
than high school education). Lo et al. [19] also found a strong standing policy that patients bring their medications to appoint-
376 J.R. Marks et al. / Patient Education and Counseling 78 (2010) 372–376
ments. Furthermore, we used a tool to assess medication knowl-
edge that was developed for this study and has not been used by
other researchers. However, it has substantial overlap with
components of other tools, and we feel it demonstrated good
distributional qualities as well as reasonable content and construct
validity.
We view our results as preliminary given the small sampling
frame and other methodological limitations. Nevertheless, the fact
that we found strong and highly significant associations between
medication knowledge and several patient characteristics, similar
to those that have been shown by other investigators as noted
above, lends support to our findings.
3.2. Conclusion
In conclusion, our findings suggest that the REALM is an
effective predictor of low medication knowledge as measured by
the MKS and could be used as a screening tool to identify patients
at risk for low medication knowledge. Also, the combination of age,
last grade completed, and sex was also a reasonable predictor of
low medication knowledge, with only a moderate added benefit
from the REALM in our sample. The association seen between these
patient characteristics and medication knowledge may suggest a
simpler and easier algorithm to identify patients who may warrant
the more specific health literacy testing, such as with REALM. Also
given the high prevalence of adverse drug events and frequent
concerns regarding medication adherence in ambulatory care,
improved patient education particularly regarding medication side
effects may serve as a simple method for improving medication
safety as well as adherence.
Acknowledgements
Portions of this paper were presented at the 2005 Annual
Meeting of the Society of General Internal Medicine, New Orleans,
Louisiana. Mr. Don Marineau’s work is supported by a grant
(5D54HP00040-05-00: Academic Administrative Units in Primary
Care, Department of Health and Human Services). All investigative
work, data assimilation, and writing by other investigators were
accomplished without funding or grant support.
References
[1] Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, Laffel G, Sweitzer
BJ, Shea BF, Hallisey R. Incidence of adverse drug events and potential adverse
drug events: implications for prevention. J Amer Med Assoc 1995;274:29–34.
[2] Kohn KT, Corrigan JM, Donaldson MS, editors. To err is human: building a safer
health system. Washington, DC: Committee on Quality & Health Care in
America, Institute of Medicine, National Academy Press; 2000.
[3] Gandhi TK, Weingart SN, Borus J, Seger AC, Peterson J, Burdick E, Seger DL, Shu
K, Federico F, Leape LL, Bates DW. Adverse drug events in ambulatory care. N
Engl J Med 2003;348:1556–64.
[4] Gurwitz JH, Field TS, Harrold LR, Rothschild J, Debellis K, Seger AC, Cadoret C,
Fish LS, Garber L, Kelleher M, Bates DW. Incidence and preventability of
adverse drug events among older persons in the ambulatory setting. J Amer
Med Assoc 2003;289:1107–16.
[5] Davis TC, Long SW, Jackson RH, Mayeaux EJ, George RB, Murphy PW, Crouch
MA. Rapid estimate of adult literacy in medicine: a shortened screening
instrument. Fam Med 1993;25:391–5.
[6] Williams MV, Baker DW, Honig EG, Lee TM, Nowlan A. Inadequate literacy is a
barrier to asthma knowledge and self-care. Chest 1998;114:1008–15.
[7] Arnold CL, David TC, Berkel HJ, Jackson RH, Nandy I, London S. Smoking status,
reading level, and knowledge of tobacco effects among low-income pregnant
women. Prev Med 2001;32:313–20.
[8] Gordon MM, Hampson R, Capell HA, Madhok R. Illiteracy in rheumatoid
arthritis patients as determined by the Rapid Estimate of Adult Literacy in
Medicine (REALM) score. Rheumatology 2002;41:750–4.
[9] Baker DW, Gagmararian JA, Williams MV, Scott T, Parker R, Green D, Ren J, Peel
J. Functional health literacy and the risk of hospital admission among Medicare
managed care enrollees. Am J Pub Health 2002;92:1278–83.
[10] Hope CJ, Wu J, Tu W, Young J, Murray MD. Association of medication adher-
ence, knowledge, and skills with emergency department visits by adults 50
years or older with congestive heart failure. Am J Health-Syst Pharm
2004;61:2043–9.
[11] Schillinger D, Piette J, Grumbach K, Wang F, Wilson C, Daher C, Leong-Grotz K,
Castro C, Bindman AB. Closing the loop: physician communication with
diabetic patients who have low health literacy. Arch Int Med 2003;
163:83–90.
[12] Estrada CA, Martin-Hryniewicz M, Peek BT, Collins C, Byrd JC. Literacy and
numeracy skills and anticoagulation control. Am J Med Sci 2004;328:
88–93.
[13] Miller LG, Liu H, Hays RD, Golin CE, Ye Z, Beck CK, Kaplan AH, Wenger NS.
Knowledge of antiretroviral regimen dosing and adherence: a longitudinal
study. Clin Infect Dis 2003;36:514–8.
[14] Moon RY, Cheng TL, Patel KM, Baumhaft K, Scheidt PC. Parental literacy level
and understanding of medical information. Pediatrics 1998;102:e25.
[15] Hammons T, Piland NF, Small SD, Hatlie MJ, Burstin HR. Ambulatory patient
safety: what we know and need to know. J Amb Care Manage 2003;26:63–82.
[16] Tamblyn R, Laprise R, Hanley JA, Abrahamowicz M, Scott S, Mayo N, Hurley J,
Grad R, Latimer E, Perreault R, McLeod P, Huang A, Larochelle P, Mallet L.
Adverse events associated with prescription drug cost-sharing among poor
and elderly persons. J Amer Med Assoc 2001;285:421–9.
[17] Persell SD, Heiman HL, Weingart SN, Burdick E, Borus JS, Murff HJ, Bates DW,
Gandhi TK. Understanding of drug indications by ambulatory care patients.
Am J Health-Syst Pharm 2004;61:2523–7.
[18] Ascione FJ, Kirscht JP, Shimp LA. An assessment of different components of
patient medication knowledge. Med Care 1986;24:1018–28.
[19] Lo S, Sharif I, Ozuah PO. Health literacy among English-speaking parents in a
poor urban setting. J Health Care Poor Underserved 2006;17:504–11.
[22] Kripalani S, Henderson LE, Chiu EY, Robertson R, Kolm P, Jacobson TA. Pre-
dictors of medication self-management skill in a low-literacy population. J Gen
Intern Med 2006;21:852–6.
[23] Edelberg HK, Shallenberger E, Wei JY. Medication management capacity in
highly functioning community-living older adults: detection of early deficits. J
Am Geriat Soc 1999;47:592–6.
[24] Davis TC, Wolf MS, Bass 3rd PF, Thompson JA, Tilson HH, Neuberger M, Parker
RM. Literacy and misunderstanding prescription drug labels. Ann Intern Med
2006;146:887–94.
[25] Louis-Simonet M, Kossovsky MP, Sarasin FP, Chopard P, Gabriel V, Perneger TV,
Gaspoz JM. Effects of a structured patient-centered discharge interview on
patients’ knowledge about their medications. Am J Med 2004;117:563–8.
[26] Ponnusankar S, Surulivelrajan M, Anandamoorthy N, Suresh B. Assessment of
impact of medication counseling on patients’ medication knowledge of com-
pliance in an outpatient clinic in South India. Patient Educ Couns 2004;54:
55–60.
[28] Hill J, Bird H, Johnson S. Effect of patient education on adherence to drug
treatment for rheumatoid arthritis: a randomised controlled trial. Ann Rheum
Dis 2001;60:869–75.
[29] Gazmararian J, Kripalani S, Miller MJ, Echt KV, Rask K. Factors associated with
medication refill adherence in cardiovascular-related diseases. JGIM 2006;
21:1215–21.