Psychometric Properties of the Thai Version of the 8-item Morisky

Medication Adherence Scale in Patients with Type 2 Diabetes

Phantipa Sakthong, Rossamalin Chabunthom, and Rungpetch Charoenvisuthiwongs

edication nonadherence in patients

M with diabetes has been associated
with poor blood glucose control.1-3 Poor
BACKGROUND: A self-reporting questionnaire is the most convenient and cheapest
way to assess medication adherence. A new 8-item self-reported Morisky Medi-
glycemic control can lead to develop- cation Adherence Scale (MMAS) has been developed and has shown better
psychometric properties than the original 4-item Morisky scale.
ment of complications including neu-
OBJECTIVE: To examine the validity, including convergent known-groups and
ropathy, nephropathy, retinopathy, and
construct validity, and reliability, including internal consistency and test–retest
cardiovascular and cerebrovascular dis- reliability, of the MMAS in Thai patients with type 2 diabetes.
eases.4,5 In general, the adherence rate to
METHODS: The data were derived from a cross-sectional study. In a convenience
diabetic medications ranges from 36% to sample, 303 type 2 diabetic outpatients were interviewed at the General Police
93%.6 Hospital in Bangkok, Thailand, between January and June 2007. Face-to-face
Nonadherence is also 1 of 7 drug ther- interviews included MMAS, medication adherence visual analog scale (MA-VAS),
apy problems that pharmaceutical care and sociodemographic data. Medical records were reviewed for clinical data such
as hemoglobin A1C (A1C) levels.
providers must prevent, identify, and
RESULTS: Internal consistency reliability was moderate (Cronbach’s α = 0.61),
solve.7 Tools used to measure adherence
whereas the test–retest reliability of the MMAS was excellent (intraclass corre-
include pill counts, electronic monitoring
lation coefficient = 0.83; p < 0.001). Concerning convergent validity, the MMAS
systems, monitoring of drug concentra- had a high correlation with the 3-item Morisky scale (r = 0.77; p < 0.01) and a
tions in blood and urine, medication medium correlation with the MA-VAS (r = 0.57; p < 0.01). Regarding known-
records at the pharmacy, and self-reports.8 groups validity, a significant association between MMAS and A1C levels was
Of these, the most convenient, least ex- found (χ2 = 6.7; p < 0.05). The sensitivity, specificity, positive predictive value, and
negative predictive value of the MMAS were 51%, 64%, 71%, and 43%,
pensive, and easiest way to assess medica-
respectively. Our factor analysis showed that the MMAS had 3 dimensions
tion adherence is self-reporting. Morisky including forgetting to take medications, stopping medications when feeling better
et al.9 developed a 4-item self-report ques- or worse, and the complexity of the drug regimen.
tionnaire to assess medication adherence. CONCLUSIONS: The 8-item MMAS can be a tool to aid in assessing medication
The Morisky scale is one of the most adherence in diabetes. The poor sensitivity can be improved by increasing the
widely used self-reported medication ad- number of the response choices and the cut-off score of the scale and by using
10-13 specific words in some items. A modified Thai version of the 8-item MMAS may
herence measures. However, it has
be needed.
shown poor psychometric properties. The
KEY WORDS: diabetes, Morisky Medication Adherence Scale, Thailand.
sensitivity and the specificity of the 4-item
adherence scale are 0.81% and 0.44%, re- Ann Pharmacother 2009;43:950-7.
spectively. Cronbach’s α reliability value Published Online, 14 Apr 2009, www.theannals.com, DOI 10.1345/aph.1L453
is 0.61, which is below the acceptable val-
ue of 0.7.14
Morisky et al.15 recently developed a new self-reported Scale (MMAS). The MMAS comprises 8 items and has
adherence measure called Morisky Medication Adherence better psychometric properties than the previous 4-item
medication adherence scale of Morisky. They found that
the MMAS had a sensitivity of 93% and a specificity of
Author information provided at the end of the text. 53%, respectively. The α reliability value of the 8-item

950 n The Annals of Pharmacotherapy n 2009 May, Volume 43 www.theannals.com

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
Morisky scale was 0.83, which is above the value of 0.7. get to bring medications with you when travelling?,” and
Because of the improved psychometric properties of the problems with the complexity of the medical regimen,
MMAS, we were interested in testing the psychometric such as, “Do you feel hassled by sticking to your treatment
properties of this new adherence scale in Thai people with plan?” Each item of the MMAS items is measuring a spe-
diabetes. Thus, the objective of this study was to assess the cific behavior and is not a determinant of adherence be-
validity, including convergent known-groups and construct havior.15 Response choices are yes/no for items 1–7 and a
validity, and the reliability, including internal consistency 5-point Likert response for the last item. Scores obtained
and test–retest reliability, of the MMAS in Thai type 2 dia- from this scale range from 0 to 8, where higher scores indi-
betic patients. cate higher adherence. Scores of 8, 6 to less than 8, and
less than 6 were classified as high, medium, and low ad-
Methods herence, respectively.
For this study, the 8-item MMAS was translated into
DESIGN Thai using a forward and back translation procedure. First,
The data used in this study were derived from a cross- 2 bilingual people translated the MMAS from the English
sectional study.16 In this study, a convenience sample of version to a Thai version. Second, a third bilingual person
315 type 2 diabetic outpatients was identified at the Dia- translated the Thai version back to English. Third, the orig-
betes Clinic of the General Police Hospital in Bangkok, inal and the back-translated English versions were com-
Thailand, between January and June, 2007. Approximately pared for conceptual equivalence. A pilot test (N = 25) was
97% (305) of the patients who were approached agreed to done to ensure patient understanding of the wording of the
participate. Patients who were eligible for the study were at Thai version of the MMAS questionnaire.
least 18 years old and able to understand the Thai lan-
guage. Patients who had severe health problems or cogni- DATA ANALYSIS
tive impairments who could not complete interviews were
To summarize characteristics of the patients and de-
excluded. Of the sample of 305 subjects, 99% (303) could
scriptive statistics, percents and frequencies were used for
complete the interviews. Face-to-face interviews included
categorical variables including sex, education, and type of
administration of the MMAS, followed by the medication
hypoglycemic drugs. Means, standard deviations, and
adherence visual analog scale (MA-VAS), and collection
ranges (minimum and maximum) were calculated for con-
of sociodemographic data. While the patients were inter-
tinuous variables including age, A1C, the number of dia-
viewed on adherence, they were asked to think about all
betic complications and hypoglycemic drugs, MA-VAS,
medications used. Nevertheless, since all were diagnosed
and MMAS. The characteristics of the whole sample and
with diabetes, were taking at least one hypoglycemic drug,
adherent groups including high, medium, and low adher-
and were interviewed at the Diabetes Clinic, they often re-
ence groups were presented. Analysis of variance (ANO-
ferred to hypoglycemic drugs more than to other drug
groups. VA) tests were used to determine the differences in contin-
On the same day of the interviews, medical records uous variables among the 3 adherent groups. Regarding
were reviewed for hemoglobin A1C (A1C) levels, number categorical variables, χ2 tests were employed.
of hypoglycemic medications, whether the patient used in- Internal consistency and test–retest reliability were as-
sulin, and the presence of diabetic complications such as sessed using Cronbach’s alpha and intraclass correlation
neuropathy, retinopathy, nephropathy, or cardiovascular coefficients (ICCs), respectively. The criteria of accepting
disease. In addition, about one-fifth of this sample (n = 64) Cronbach’s α are equal to or above 0.70.14 The criteria for
was randomly selected for a 1- to 2-week reliability test– interpretation of ICCs are based on Rosner (suggesting that
retest via telephone. All of the 64 patients agreed to partici- ICC < 0.40 = poor agreement, 0.40 ≤ ICC < 0.75 = fair to
pate in the telephone interview. All 303 face-to-face and 64 good agreement, ICC ≥ 0.75 = excellent agreement).17
telephone interviews were done by one of the investiga- Convergent validity was evaluated using Pearson’s cor-
tors, who is a pharmacist. This study was approved by the relation coefficients (r) to assess the association between
Ethics Committee of the General Police Hospital. the MMAS and the previous 4-item Morisky scale. Three
items from the 8-item MMAS were employed to represent
the 4-item Morisky scale because they are the same items
MORISKY MEDICATION ADHERENCE SCALE
as those of the previous scale. These 3 items included, “Do
The MMAS was developed from the previous 4-item you sometimes forget to take your medication?,” “Have
Morisky Medication Adherence scale9 and supplemented you stopped taking your medication without telling your
with additional items. The theory underlying this measure doctors when you felt worse?,” and “Have you stopped
was that failure to adhere to a medication regimen can be taking your medications without telling your doctors when
caused by many factors, such as, “Do you sometimes for- you felt better?” In addition to the 4-item Morisky scale,

www.theannals.com The Annals of Pharmacotherapy n 2009 May, Volume 43 n 951

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
P Sakthong et al.

the association between the MMAS and MA-VAS was ex- CONVERGENT VALIDITY
plored. The MA-VAS showed a respondent a line with
MMAS was positively associated with the original 3-
endpoints of “no medication adherence at all” at 0 and
item Morisky scale (r = 0.77; p < 0.01) and MA-VAS (r =
“full medication adherence” at 10. Correlations were inter-
0.57; p < 0.01). The MMAS had very good to excellent
preted using the criteria provided by Colton as follows:
correlation with the 3-item Morisky scale and medium to
0–0.25 = little or no correlation, 0.25–0.50 = fair correla-
good correlation with the MA-VAS.
tion, 0.50 – 0.75 = moderate to good correlation, and
greater than 0.75 = very good to excellent correlation.18
KNOWN-GROUPS VALIDITY
The known-groups validity was assessed through the as-
sociation of MMAS categories (high, medium, low adher- As shown in Table 2, a significant relationship between
ence) and A1C levels (≥7% and <7%) using χ2 tests. An the MMAS and blood glucose control was found (χ2 = 6.7;
odds ratio with 95% confidence interval of the association p = 0.035). Low medication adherence (MMAS score <6)
between MMAS categories and A1C levels was calculat- was associated with poor blood glucose control (A1C
ed. To provide helpful information in clinical practice, we ≥7%; OR 1.84; 95% CI 1.16 to 2.92). However, the
also determined the (1) sensitivity, as true positive: poorly MMAS showed a poor sensitivity. The sensitivity, speci-
controlled (A1C ≥7%) indicates low adherence (MMAS ficity, positive predictive value, and negative predictive
<6); (2) specificity, true negative: well controlled (A1C value of the MMAS were 0.51, 0.64, 0.71, and 0.43, re-
<7%) indicates medium to high adherence (6 ≤ MMAS spectively. This sensitivity means that 51% of diabetic pa-
≤8); (3) positive (pts. with low adherence are poorly con- tients who had poor glycemic control had low adherence,
trolled) and negative (pts. with medium to high adherence while the specificity indicates that 64% of the patients with
are well controlled) predictive values; and (4) efficiency good blood glucose control were highly adherent to drug
(true positive plus true negative) of the MMAS.19 therapy. The positive predictive value indicates that 71%
Construct validity was tested using an exploratory factor of subjects with low adherence were poorly controlled,
analysis. The factor analysis was conducted by a principal whereas the negative predictive value means that 43% of
component analysis, followed by Varimax rotation with those with medium-to-high adherence had good glycemic
Kaiser normalization. Kaiser’s eigenvalue greater than 1 control. The efficiency of the MMAS as a screening test
was used to determine the number of factors. Factor load- was only 55.8%.
ings greater than 0.4 on each item were considered to be-
long to the corresponding factors. CONSTRUCT VALIDITY
All analyses were performed using SPSS version 13.0
(SPSS Inc., chicago, IL). The level of significance was set Exploratory factor analysis showed 3 factors with eigen-
at p less than 0.05. values greater than 1, which explained 57.4% of the total
variance. Factor loadings between the 8 items of MMAS
Results and the 3 factors are presented in Table 3. Factor 1 com-
prised items 1, 2, and 4, which mostly involved patients
Characteristics of the total sample and adherent groups forgetting to take medications. Factor 2 consisted of items
are shown in Table 1. Of 303 type 2 diabetic patients, 3, 5, and 6, which concerned patients stopping medications
17.1% (52), 37.0% (112), and 45.9% (139) were high, when they were feeling better or worse. Factor 3 included
medium, and low adherent groups, respectively. There items 7 and 8, which were viewed as the complexity of the
were no significant differences in age, sex, education, or drug regimen. Factor 1 had the highest correlation with the
the number of diabetic complications among the 3 adher- 8-item MMAS (r = 0.83; p < 0.01), followed by factor 2 (r
ent groups (all p ≥ 0.5). Significance differences were = 0.62; p < 0.01) and factor 3 (r = 0.54; p < 0.01).
found in A1C levels, the number of hypoglycemic drugs,
type of hypoglycemic drugs, MA-VAS scores, and Discussion
MMAS scores among the 3 adherent groups (all p < 0.5).
Figure 1 presents the distribution of the 8-item MMAS This study was the first to psychometrically tested the
scores. new 8-item MMAS in a population whose first language is
not English. Other studies employed the 9-item MMAS
RELIABILITY
(adding 1 more item about reminder system for medicines)
in HIV-positive patients in Sweden.20,21 However, they did
Regarding internal consistency reliability, Cronbach’s α not examine the psychometric properties of the MMAS.
for MMAS was 0.61, which is below the acceptable value Our study was also the first to apply the 8-item MMAS to
of 0.7. However, for test–retest reliability the MMAS assess medication adherence in diabetic patients, whereas
showed an excellent ICC of 0.83 (p < 0.001). Morisky et al.15 tested the psychometric properties of their

952 n The Annals of Pharmacotherapy n 2009 May, Volume 43 www.theannals.com

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
 Psychometric Properties of the Thai Version of the 8-item Morisky Medication Adherence Scale

8-item MMAS in hypertensive patients. They found that mensional scale. An explanation for the unacceptable α
the 8-item MMAS was reliable, with good concurrent and value in our study is scale multidimensionality. Internal
predictive validity and had good sensitivity. Our study consistency reliability is a measure of the homogeneity of
showed that the 8-item MMAS had good test–retest relia- the items within a scale. If all items measure the same trait,
bility and convergent validity in Thai people with diabetes. the internal consistency reliability of the whole scale will be
However, the 8-item MMAS showed poor sensitivity. We high. Since our exploratory factor analysis showed that the
also found that the 8-item MMAS was a 3-factor scale in scale had 3 dimensions, meaning that there were 3 different
this sample. traits in this sample, the α of the whole scale is lower than
The MMAS had excellent test–retest reliability (ICC = the acceptable value of 0.70. Internal consistency reliability
0.83), whereas the internal consistency reliability was mod- also depends on the number of items of a scale (ie, the more
erate (Cronbach’s α = 0.61, which is <0.70). Our result is items, the higher the internal consistency reliability). An-
different from that of Morisky et al.15 They reported that the other possible reason for the reduced α is that most items (7
8-item MMAS had a Cronbach’s α of 0.83 and is a one-di- of 8) on the scale use dichotomous response choices

Table 1. Characteristics of Type 2 Diabetic Patients

High Medium Low
Total Adherence Adherence Adherence
Sample (MMAS = 8) (6 ≤ MMAS < 8) (MMAS < 6)
Characteristic (N = 303) (n = 52) (n = 112) (n = 139)
a
Age, y
mean ± SD 61.1 ± 11.4 63.9 ± 11.2 62.31 ± 10.2 60.1 ± 12.3
range 27–90 37–90 35–85 27–89
Sex, %a
female 71 65 74 70
male 29 35 26 30
Education, %a
none 5.6 9.6 3.6 5.7
6th grade or less 42.6 42.3 41.1 43.9
7th–12th grade 30.7 34.6 27.7 31.6
college 2–4 y 18.2 7.7 23.9 15.8
higher than college 3 5.7 3.8 2.9
A1C level, %b
mean ± SD 7.7 ± 1.7 7.4 ± 1.9 7.4 ± 1.4 8.0 ± 1.8
range 4.0–15.8 4.5–15.8 5.3–13.3 4.0–14.9
Diabetic complications, na
mean ± SD 0.82 ± 0.88 0.67 ± 0.86 0.84 ± 0.94 0.85 ± 0.84
range 0–4 0–3 0–4 0–4
Hypoglycemic drugs, nb
mean ± SD 2.77 ± 1.22 2.50 ± 1.26 2.64 ± 1.09 2.96 ± 1.28
range 1–6 1–5 1–6 1–6
Type of hypoglycemic drug, %b
only oral hypoglycemic 60.7 73 69 50
oral hypoglycemic plus insulin 35.3 25 29 44
only insulin 4 2 2 6
MA-VASb,c
mean ± SD 8.68 ± 1.36 9.86 ± 0.38 9.05 ± 0.98 7.93 ± 1.43
range 3.50–10.00 8.00–10.00 5.00–10.00 3.50–10.00
MMASb, c
mean ± SD 5.71 ± 1.64 8.00 ± 0.00 6.43 ± 0.50 4.27 ± 1.11
range 0.50–8.00 8.00–8.00 6.00–7.75 0.50–5.75

A1C = hemoglobin A1C; MA-VAS = medication adherence visual analog scale; MMAS = Morisky Medication Adherence Scale.
a
No significance differences were found among adherent groups (p ≥ 0.05).
b
Significant differences were found among adherent groups (p < 0.05).
c
The higher MA-VAS and MMAS scores indicate higher medication adherence.

www.theannals.com The Annals of Pharmacotherapy n 2009 May, Volume 43 n 953

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
P Sakthong et al.

(yes/no). Internal consistency reliability can be improved dictive value, negative predictive value, and efficiency of
by increasing the number of response choices to a threshold the MMAS were 85%, 21%, 65%, 46%, and 62%, respec-
of 7, because the more response choices, the lower the mea- tively. Thus, the higher the cut-off scores of the MMAS
surement error of the scale.22 were, the higher the sensitivity and the higher the efficien-
Not surprisingly, the MMAS had a high correlation with cy of the scale. The increased cut-off scores did not greatly
the previous 3-item Morisky scale (r = 0.77). However, it affect the positive predictive value and the negative predic-
had only moderate correlation with MA-VAS (r = 0.57). tive value of the MMAS. However, we have to trade off
Morisky et al.,15 reported that the 8-item scale was signifi- the higher sensitivity against lower specificity. In clinical
cantly correlated with the previous 4-item scale (Pearson practice, healthcare providers are more interested in identi-
correlation of 0.64), which was lower than that of our fying patients with poor glycemic control and low adher-
study. ence than well-controlled patients with high adherence;
As for known-groups validity, although there was a sig- therefore, increasing the cut-off score of the scale can be a
nificant association between the MMAS results and blood good option for solving its low sensitivity problem. How-
glucose control (χ2 = 6.7; p < 0.05), it showed a poor sensi- ever, this solution may not work well if the hospital lacks
tivity (51%), a specificity (64%), a positive predictive val- sufficient healthcare personnel, because use of the in-
ue (71%), and a negative predictive value (43%). Morisky creased cut-off score will result in more patients with low
et al.15 showed that the 8-item MMAS had a sensitivity of adherence, causing increased burden for the providers.
93% and a specificity of 53%. We showed higher speci- Thus, the appropriate cut-off point of the MMAS to detect
ficity but lower sensitivity than those of the study of low medication adherence may compromise the number of
Morisky et al. The decreased sensitivity of the MMAS in hospital personnel who can monitor adherence.
this study may have been contributed to by social desir- In addition to the problem of social desirability, reduced
ability; our subjects may have answered untruthfully that sensitivity could be caused by the dichotomous response
they adhered to their diabetic medications, although they choices on most items of the scale, as previously men-
did not. tioned. Moreover, the wording of some items is not specif-
When we tried to change the cut-off score of low adher- ic. For example, “Do you sometimes forget to take your
ence from 6 to 7 (low adherence = MMAS scores <7), the medication?” could include both patients who missed a
sensitivity, specificity, positive predictive value, negative few doses and those who regularly forgot to take their
predictive value and the efficiency of the MMAS were medicines. Therefore, rewording the question to be eigen-
75%, 38%, 68%, 46%, and 61%, respectively. Similarly, if value more specific, such as, “Have you ever forgotten to
the cut-off score was increased to be 8 (low adherence = take your medications in the past month?,” would improve
MMAS scores <8), the sensitivity, specificity, positive pre- the sensitivity of the scale. In addition, Morisky et al. have

Figure 1. Distribution of the 8-item Morisky Medication Adherence Scale scores.

954 n The Annals of Pharmacotherapy n 2009 May, Volume 43 www.theannals.com

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
 Psychometric Properties of the Thai Version of the 8-item Morisky Medication Adherence Scale

stated that each item of the MMAS is measuring a specific Our results may limit the ability to generalize the results
medication-taking behavior and is not a determinant of ad- to other disease groups. Thus, more research on the psycho-
herence behavior.15 This may lead to a poor relationship metric properties of the new MMAS in Thai people with
between the MMAS and A1C levels, since the MMAS other clinical conditions is needed. Since this study was a
does not directly predict the blood glucose level. cross-sectional design, other important psychometric proper-
Because A1C is the average blood glucose level in the ties, including responsiveness and predictive validity, were
past 2–3 months, it can reflect medication adherence. Nev- not conducted. Further research on these properties is en-
ertheless, because of its cost, A1C is not measured on a couraged.
regular basis in Thailand; it is mostly determined in large We conclude that the 8-item MMAS can be a useful
hospitals. Fasting blood glucose levels are common mea- self-report medication adherence measure in Thai people
sures in Thailand, but they do not reveal medication adher- with diabetes because it shows acceptable test–retest relia-
ence as well as A1C levels do. In this study, fasting blood bility and convergent and construct validity. However, the
glucose levels were also not significantly associated with poor sensitivity of the scale can be improved by increasing
MMAS scores (data not shown). Therefore, the 8-item the number of the response choices and the cut-off score of
MMAS can be an initial tool to help in assessing adher- the scale and using specific words in some items. A modi-
ence in any Thai healthcare setting where A1C is not avail- fied Thai version of the 8-item MMAS may be needed.
able, since it is simple and economical to use and yielded
acceptable reliability and validity. For example, a health- Phantipa Sakthong PhD, Assistant Professor, Department of Phar-
macy Practice, Faculty of Pharmaceutical Sciences, Chulalongko-
care provider can apply the MMAS to aid in identifying rn University, Bangkok, Thailand
drug therapy problems such as low adherence due to ad- Rossamalin Chabunthom MS, Hospital Pharmacist, Sawang-
verse effects or complexity of drug regimens in diabetic dandin Crown Prince Hospital, Sakonnakorn Province, Thailand
Rungpetch Charoenvisuthiwongs PhD, Assistant Professor,
patients at risk of uncontrolled blood glucose. As for pa- Department of Pharmacy Administration, Faculty of Pharmaceuti-
tients with high medication adherence and good blood glu- cal Sciences, Chulalongkorn University
cose control, the MMAS can also be used to remind pa- Reprints: Dr. Sakthong, Department of Pharmacy Practice, Facul-
ty of Pharmaceutical Sciences, Chulalongkorn University, Prayathai
tients of the importance of continued adherence and the Rd., Pathumwan, Bangkok (10330), Thailand, fax 662-218-8403,
benefits of well-controlled glucose control. phantipa_sakthong@yahoo.com
This study was supported by a fund from Chulalongkorn University.
We thank the participants for providing valuable data, and the nurs-
es and physicians for their assistance in collecting the data. We also
Table 2. Relationship Between MMAS and Blood thank Dr. Donald Morisky for his permission to use the MMAS.
Glucose Controla
A1C ≥7% A1C <7% References
(poor (good
Parameter control) control) Total
1. Schectman JM, Nadkarni MM, Voss JD. The association between dia-
Low adherence (MMAS < 6) 99 (71) 40 (29) 139 betes metabolic control and drug adherence in an indigent population.
Diabetes Care 2002;25:1015-21.
Medium adherence (6 < MMAS < 8) 66 (59) 46 (41) 112
2. Krape KK, King K, Warren SS, et al. Medication adherence and associ-
High adherence (MMAS = 8) 28 (54) 24 (46) 52 ated hemoglobin A1C in type 2 diabetes. Ann Pharmacother 2004;38:
Total 193 110 303 1357-62. Epub 6 Jul 2004. DOI 10.1345/aph.1D612
3. Hartz A, Kent S, James P, Xu Y, Kelly M, Daly J. Factors that influence
A1C = hemoglobin A1C; MMAS = Morisky Medication Adherence Scale. improvement for patients with poorly controlled type 2 diabetes. Diab
a
Number (%) of patients; χ2 = 6.7; p = 0.035. Res Clin Pract 2006;74:227-32. DOI 10.1016/j.diabres.2006.03.023

Table 3. Exploratory Factor Analysis of the 8-Item Morisky Medication Adherence Scale in Patients with Type 2 Diabetesa
Item Factor 1 Factor 2 Factor 3

1. Sometimes forget to take medications? 0.768 0.026 0.125

2. Did not take medications in the past 2 wk? 0.778 0.305 –0.012
3. Stop taking medications when feeling worse? –0.125 0.712 0.170
4. Sometimes forget to bring medications when traveling? 0.670 –0.101 0.098
5. Took medications yesterday? 0.254 0.746 –0.162
6. Stop taking medications when well controlled? 0.058 0.605 0.340
7. Feeling hassled about sticking to treatment plan? 0.181 0.023 0.736
8. How often do you have difficulty remembering to take medications? 0.008 0.159 0.730

a
Factor loading in 303 patients. Bold-faced numbers indicate factor loadings >0.4.

www.theannals.com The Annals of Pharmacotherapy n 2009 May, Volume 43 n 955

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
P Sakthong et al.

4. Lerman I. Adherence to treatment: the key for avoiding long-term com- Se ha desarrollado una nueva Escala de Morisky de Cumplimiento Tera-
plications of diabetes. Arch Med Res 2005;36:300-6. péutico (MMAS) para declaración de síntomas formada por 8 cuestiones y
DOI 10.1016/j.arcmed.2004.12.001 ha demostrado presentar mejores propiedades psicométricas que la primera
5. Rhee MK, Slocum W, Ziemer DC, et al. Patient adherence improves escala de Morisky, que constaba de 4 cuestiones.
glycemic control. Diabetes Educ 2005;31:240-50. OBJETIVOS: Determinar la validez, incluida la validez convergente, la
DOI 10.1177/0145721705274927 validez con grupos conocidos y la validez de constructo y la fiabilidad,
6. Cramer JA. A systemic review of adherence with medications for dia- incluida la consistencia interna y la fiabilidad “test–retest” de la MMAS
betes. Diabetes Care;2004;27:1218-24. en pacientes tailandeses con diabetes tipo 2.
7. Cipolle RJ, Strand LM, Morley PC. Pharmaceutical care practice: the MÉTODOS: Los datos se obtuvieron a partir de un estudio transversal. Se
clinician’s guide. 2nd ed. New York: McGraw-Hill, 2004. entrevistó a una muestra por conveniencia de 303 pacientes ambulatorios
8. Vik SA, Maxwell CJ, Hogan DB. Measurement, correlates, and health con diabetes tipo 2 en el General Police Hospital en Bangkok, Tailandia,
outcomes of medication adherence among seniors. Ann Pharmacother entre enero y junio de 2007. Las entrevistas realizadas cara a cara
2004;38:303-12. Epub 30 Dec 2003. DOI 10.1345/aph.1D252 incluyeron la MMAS, la escala analógica visual de cumplimiento
9. Morisky DE, Green LW, Levine DM. Concurrent and predictive validity terapéutico (MA-VAS), y datos sociodemográficos. Se revisaron los
of a self-reported measurement of medication adherence and long-term archivos médicos para comprobar datos clínicos como los niveles de
predictive validity of blood pressure control. Med Care 1986;24:67-74. HbA1C.
10. Shalansky SJ, Levy AR, Ignaszewski AP. Self-reported Morisky score RESULTADOS: La fiabilidad como consistencia interna fue moderada (α de
for identifying nonadherence with cardiovascular medications. Ann Cronbach = 0.61), mientras que la fiabilidad como “test–retest” de la
Pharmacother 2004;38:1363-8. Epub 6 Jul 2004. DOI 10.1345/aph.1E071 MMAS fue excelente (ICC = 0.83; p < 0.001). En cuanto a la validez
11. Erickson SR, Coombs JH, Kirking DM, Azimi AR. Compliance from convergente, la MMAS presentó una alta correlación con la escala de
Morisky de tres cuestiones (r = 0.77; p < 0.01) y una correlación media con
self-reported versus pharmacy claims data with metered-dose inhalers.
la MA-VAS (r = 0.57; p < 0.01). En relación a la validez con grupos
Ann Pharmacother 2001;35:997-1003. DOI 10.1345/aph.10379
conocidos, se encontró una asociación significativa entre la MMAS y
12. Gao X, Nau DP. Congruence of three self-report measures of medication los niveles de HbA1C (χ2 = 6.7; p < 0.05). La sensibilidad, especificidad,
adherence among HIV patients. Ann Pharmacother 2000;34:1117-22. y los valores pronósticos positivos y negativos de la MMAS fueron de
DOI 10.1345/aph.19339 51%, 64%, 71%, 43%, respectivamente. Nuestro análisis de factores
13. Venturini F, Nichol MB, Sung JC, Bailey KL, Cody M, McCombs JS. demostró que la MMAS presentaba 3 dimensiones como olvidar tomar la
Compliance with sulfonylureas in a health maintenance organization: a medicación, dejar de tomar la medicación cuando los síntomas
pharmacy record–based study. Ann Pharmacother 1999;33:281-8. mejoraban o empeoraban y la complejidad del régimen terapéutico.
DOI 10.1345/aph.18198 CONCLUSIONES: La escala MMAS de 8 cuestiones puede constituir una
14. Nunnally JC Jr. Psychometric theory. 2nd ed. New York: McGraw-Hill, herramienta para la evaluación del cumplimiento terapéutico en la
1978. diabetes. La baja sensibilidad puede mejorarse incrementando el número
15. Morisky DE, Ang A, Krousel-Wood M, Ward H. Predictive validity of a de opciones de respuesta y la puntuación de corte de la escala utilizando
medication adherence measure for hypertension control. J Clin Hypertens determinadas palabras en algunas cuestiones. Puede ser necesario crear
2008;10:348-54. DOI 10.1111/j.1751-7176.2008.07572.x una versión tailandesa modificada de la MMAS de 8 cuestiones.
16. Sakthong P, Sakulbumrungsil R. A comparison of EQ-5D index scores
Traducido por Violeta Lopez Sanchez
using the UK, US, and Japan preference weights in a Thai sample with
type 2 diabetes. Health Qual Life Outcomes 2008;6:71.
DOI 10.1186/1477-7525-6-71. Les Propriétés Psychométriques de la Version Thai de l’Échelle
17. Rosner B. Fundamentals of biostatistics. 5th ed. Pacific Grove, CA: d’Adhésion à la Médication en 8 Points de Morisky chez des
Duxbury Thomson Learning, 2000. Patients Souffrant de Diabète de Type 2
18. Colton T. Regression and correlation. In: Statistics in medicine: Little,
Brown and Co., Boston, MA, 1974. P Sakthong, R Chabunthom, et R Charoenvisuthiwongs
19. Rothman KJ. Epidemiology: an introduction. New York: Oxford Univer- Ann Pharmacother 2009;43:950-7.
sity Press, 2002.
20. Sodergard B, Halvorsson M, Lindback S, Sonnerberg A, Tully MP, Lind-
blad AK. Differences in adherence and motivation to HIV therapy—two RÉSUMÉ
independent assessments in 1998 and 2002. Pharm World Sci 2006;28: ÉTAT DES CONNAISSANCES: Un questionnaire d’autoévaluation est le
248-56. DOI 10.1007/s11096-006-9036- 4 moyen le plus commode et le moins dispendieux pour évaluer
21. Sodergard B, Halvorsson M, Tully MP, et al. Adherence to treatment in l’adhésion à la médication. Une nouvelle échelle d’adhésion à la
Swedish HIV-infected patients. J Clin Pharm Ther 2006;31:605-16. médication de Morisky (ÉAMM) en 8 points a été développée et a
DOI 10.1111/j.1365-2710.2006.00782.x démontré des propriétés psychométriques supérieures à l’échelle
22. Streiner DL, Norman GR. Health measurement scales: a practical guide to originale de Morisky en 4 points.
their development and use. 2nd ed. New York: Oxford University Press, OBJECTIFS: Examiner la validité incluant la validité convergente, la
1995. validité du construit et la validité pour groupes différents et examiner la
fiabilité incluant la cohérence interne et la fiabilité test–retest de
l’ÉAMM chez des patients Thai diabétiques de type 2.
MÉTHODOLOGIE: Les données ont été obtenues à partir d’une étude
transversale. Un échantillon de 303 patients diabétiques de type 2 vus en
Propiedades Psicométricas de la Versión Tailandesa de la Escala de externe ont été interviewés à l’hôpital General Police de Bangkok,
Morisky de Cumplimiento Terapéuticos de 8 Cuestiones en Pacientes Thailand, entre janvier et juin 2007. Les interviews en tête à tête
con Diabetes Tipo 2 comprenaient l’ÉAMM, l’échelle visuelle analogue d’adhésion à la
P Sakthong, R Chabunthom, y R Charoenvisuthiwongs
médication (ÉVA-AM), et des données sociodémographiques. Les
dossiers médicaux ont été révisés pour obtenir des données cliniques tels
Ann Pharmacother 2009;43:950-7. les taux d’HbA1C.
RÉSULTATS: La cohérence interne s’est révélée modérée (coefficient α de
Crohnbach = 0.61) alors que la fiabilité test–retest de l’ÉAMM a été
EXTRACTO
excellente (ICC = 0.83; p < 0.001). Concernant la validité convergente,
ANTECEDENTES: La forma más adecuada y económica de evaluar el cumpli- l’ÉAMM a obtenu une corrélation élevée avec l’échelle de Morisky en 3
miento terapéutico es usando un cuestionario de declaración de síntomas. points (r = 0.77; p < 0.01), et une corrélation moyenne avec l’ÉVA-AM

956 n The Annals of Pharmacotherapy n 2009 May, Volume 43 www.theannals.com

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014
 Psychometric Properties of the Thai Version of the 8-item Morisky Medication Adherence Scale

(r = 0.57; p < 0.01). En ce qui regarde la validité de groupes différents CONCLUSIONS: L’ÉAMM en 8 points peut être un outil pour aider à
connus, une association significative entre l’ÉAMM et les taux d’HbA1C évaluer l’adhésion aux médicaments chez les diabétiques. La faible
a été trouvée (χ2 = 6.7; p < 0.05). La sensibilité, la spécificité, la valeur sensibilité peut être améliorée en augmentant le nombre de choix de
prédictive positive et la valeur prédictive négative de l’ÉAMM ont été réponses et le seuil des résultats et en utilisant des mots spécifiques dans
respectivement 51%, 64%, 71%, et 43%. L’analyse factorielle a certains points. Une version Thai modifiée de l’ÉAMM en 8 points peut
démontré que l’ÉAMM comportait 3 dimensions incluant l’oubli de être nécessaire.
prendre ses médicaments, l’arrêt de prendre la médication quand on se
sent mieux ou pire, et la complexité du régime médicamenteux. Traduit par Marie Larouche

Reprints and e-prints from

THE ANNALS OF PHARMACOTHERAPY
Provide the latest peer-reviewed research to your clients,
sales force, associates, and educators.

Available in both print and electronic formats, reprints from

The Annals are effective in:
• Supporting new product launches and development
• Updating target audiences
• Distributing information to conference attendees
• Direct mail campaigns
• Sales force education
• Press release packages

For more information about our quality reprint service, please

contact:
Greg J Johnson at 877-742-7631 or
gjohnson@theannals.com

www.theannals.com The Annals of Pharmacotherapy n 2009 May, Volume 43 n 957

Downloaded from aop.sagepub.com at University of Birmingham on November 3, 2014