Validation of a simpli®ed medication adherence

questionnaire in a large cohort of HIV-infected patients:

the GEEMA Study
Hernando Knobela , Jordi Alonsob , Jose L. Casadoc , Julio Collazosd ,
Juan GonzaÂleze , Isabel Ruizf , Jose M. Kindelang , Alexia Carmonah ,
Javier Juegai and Antonio Ocampoj , on behalf of the GEEMA Study
Group

Objective: To assess the effectiveness of the simpli®ed medication adherence ques-

tionnaire (SMAQ) in identifying non-adherent patients.
Design: Prospective observational study of adherence. The six-item SMAQ was devel-
oped. The following aspects were evaluated: (i) criterion validity, comparison with
electronic adherence monitoring; (ii) construct validity, association between adher-
ence, as de®ned by the SMAQ, and virological outcomes; and (iii) reliability, internal
consistency and reproducibility.
Patients: A group of 3004 unselected HIV patients who had initiated nel®navir
therapy combined with other antiretroviral drugs [21% naive, 15% protease inhibitor
(PI)-naive, 64% PI-experienced] between January 1998 and December 1999 were
enrolled in 69 hospitals in Spain. The SMAQ was administered at months 3, 6 and 12.
Results: The SMAQ showed 72% sensitivity, 91% speci®city and a likelihood ratio of
7.94 to identi®ed non-adherent patients, compared with the medication-event mon-
itoring system (40 patients evaluated). At month 12, 1797 patients were evaluated, of
whom 32.3% were de®ned as non-adherent; viral load , 500 copies/ml found in
68.3% of the adherent, and 46% of the non-adherent patients. A logistic regression
analysis of PI-naive patients was performed, including age, sex, baseline viral load
. 5 log10 /ml, CD4 cell count , 200 3 106 /l, and non-adherence as independent
variables. Non-adherence was the only signi®cant risk factor in failing to achieve
virological suppression. Cronbach's alpha internal consistency coef®cient was 0.75,
and overall inter-observer agreement was 88.2%.
Conclusion: The SMAQ appears to be an adequate instrument with which to assess
adherence in HIV-infected patients, and may be applied in most clinical settings.
& 2002 Lippincott Williams & Wilkins

AIDS 2002, 16:605±613

Keywords: Adherence, antiretroviral therapy, outcome, self-report questionnaire

From the a Department of Internal Medicine ± Infectious Diseases, Hospital del Mar, Barcelona, Spain; b Health Services
Research Unit, Institut Municipal de InvestigacioÂn MeÂdica, Barcelona, Spain; c Department of Infectious Diseases, Hospital
RamoÂn y Cajal, Madrid, Spain; d Department of Infectious Diseases, Hospital Galdakao, Vizcaya, Spain; e Department of
Infectious Diseases, Hospital La Paz, Madrid, Spain; f Department of Infectious Diseases, Hospital Valle HebroÂn, Barcelona,
Spain; g Department of Infectious Diseases, Hospital Reina SofõÂa, CoÂrdoba, Spain; h Department of Pharmacy, Hospital del Mar,
Barcelona, Spain; i Department of Infectious Diseases Hospital Juan Canalejo, La CorunÄa, Spain; and j Department of Internal
Medicine, Hospital Xeral Cies, Vigo, Spain.
 For members of the GEEMA Study Group see Appendix.
Correspondence to: Hernando Knobel, Department of Internal Medicine ± Infectious Diseases, Hospital del Mar, Paseo
MarõÂtimo 25±29, 08003 Barcelona, Spain.
Tel: ‡34 93 2483251; fax: ‡34 93 2483257; e-mail: hknobel@imas.imim.es
Received: 27 July 2001; revised: 9 November 2001; accepted: 13 November 2001.

ISSN 0269-9370 & 2002 Lippincott Williams & Wilkins 605

606 AIDS 2002, Vol 16 No 4
Introduction
Recent advances in antiretroviral therapy have dramati-
cally reduced hospitalization rates, opportunistic infec-
tions, and deaths associated with HIV infection [1].
The objective of the treatment is the lasting suppression
of viral replication [2]. Numerous factors may nega-
tively in¯uence the desired therapeutic objectives, such
as the stage of the disease, viral strain, a history of
previous treatment, baseline viraemia and pharmaco-
kinetic interactions. Among such factors the impor-
tance of inadequate adherence to treatment must be
highlighted [3]. Adherence to treatment is a potent
predictor of effectiveness, both in clinical trials and
cohort studies [4±6]. The potential consequences of
non-adherence to highly active antiretroviral therapy
(HAART) are treatment failure, resistance and cross-
resistance, and the transmission of resistant viruses to
other patients, thus limiting future therapeutic options,
both for the individual and the community [7,8].
Adherence to pharmacological therapy in general has
been shown to be low, ranging from 10 to 90% [9].
For patients on HAART, adherence rates to prescribed
therapies were found to vary from 40 to 80%, both in
clinical trials and clinical practice settings [4±6,10±13].
The measurement of adherence is thus an essential
component in anti-HIV therapy. However, there is no
`gold standard' when talking about the measurement of
adherence. The approaches employed include patient
self-report, patient attendance at programmed visits, pill
counts, pharmacy records, the measurement of drug
levels, biological surrogate markers and the use of the
medication-event monitoring system (MEMS cap).
However, none are completely accurate, and most are
not practical to apply in clinical practice [14] with the
only exception being self-reported measurement.
The measurements may be compared with respect to
three attributes: (i) multidimensional adherence meas-
urements versus missed-dose measurements; (ii) the
classi®cation of adherence as a continuum (0±100%)
versus a dichotomy (adherent/non-adherent) variable;
and (iii) the time interval evaluated (weeks versus days).
The self-report (self administered or interviewer
administered) obtains a patient's subjective evaluation
of his or her own level of treatment compliance
behaviour.
The advantages of this over other methods of measure-
ment include its simplicity, speed, and viability of use.
The disadvantages include both the reliance on recall
and social desirability bias, with a tendency to over-
estimate adherence [14,15]. However, several studies
highlighted the usefulness of the self-report as an
adherence measurement tool, and showed it to corre-
late well with the virological outcome [16±18]. A
problem with comparative self-report studies is that
there is no standardized questionnaire, no standard cut-
off value for each different level of adherence, and the
time intervals evaluated are inconsistent, making it hard
to compare the different adherence ratios [19].
Therefore, a simpli®ed medication adherence question-
naire (SMAQ) was developed to evaluate a low cost,
reliable and easily applicable instrument for measuring
adherence. The aim of the present study was to assess
the performance of the SMAQ in identifying non-
adherent patients, and the association thereof with
virological outcomes in a large cohort of HIV-infected
patients.
Methods
Development of the simpli®ed medication
adherence questionnaire
A group of researchers, including physicians, nurses,
pharmacists, psychologists and patients, all with experi-
ence in antiretroviral treatment and adherence, devel-
oped the SMAQ. The questionnaire was based on the
Morisky scale [20], which had previously proved easy
to integrate into a standard medical visit. The original
scale contained four items: (1) `Do you ever forget to
take your medicine?'; (2) `Are you careless at times
about taking your medicine?'; (3) `When you feel
better, do you sometimes stop taking your medicine?';
(4) `If sometimes you feel worse when taking the
medicine, do you stop taking it?'.
By consensus, the research group then made the
following changes: item 3 was eliminated as many
HIV-infected patients are asymptomatic. Item 4 was
re-formulated as follows: `If at times you feel worse, do
you stop taking your medicine?'. Three additional
questions were incorporated, with the aim of obtaining
more adherence-speci®c measurements. A modi®ed
version of a question used by Samet et al. [21] to
determine the number of missed doses over the
previous 24 h was employed.. Although such a limited
time frame may enhance the accuracy of patient recall,
it may well not re¯ect the overall trend of patient
medication adherence. Therefore, the additional ques-
tions were: (a) Thinking about the last week. How
often have you not taken your medicine?; (b) Did you
not take any of your medicine over the last weekend?;
(c) Over the past 3 months, how many days have you
not taken any medicine at all?.
The SMAQ was considered `positive' when a non-
adherent patient was detected, that is, when there was
a positive response to any of the qualitative questions,
more than two doses missed over the past week, or
over 2 days of total non-medication during the past 3
months.
 Validation of a simpli®ed medication adherence questionnaire Knobel et al.
Evaluation of the performance of the simpli®ed
medication adherence questionnaire
The performance of the SMAQ was evaluated in three
stages: (i) criterion validity, comparing it with an
objective instrument of adherence assessment; (iii)
construct validity, analysing the association of adher-
ence, as de®ned by the SMAQ, and virological out-
comes; and (iii) reliability, assessing both the internal
consistency and reproducibility of the SMAQ when
administered by two independent investigators.
To assess criterion validity we compared the SMAQ
with another instrument used to assess adherence. The
MEMS (Aardex, Zurich) was used as a standard
adherence measurement. The MEMS cap system
employs normal medication bottles ®tted with a
pressure-activated microprocessor in the cap. The
microprocessor records each opening and lists the date,
time, and duration of opening. The information is
then stored in a database. Although the MEMS cap
system cannot prove drug intake by the patient,
prolonged deception by patients has been shown to be
unlikely [22]. The criterion validity assessment was
carried out in a subset of 40 patients at the ®rst
evaluation (at month 3). The subset consisted of the
®rst 40 patients enrolled at the coordinating centre.
The patients were provided with a MEMS cap bottle
for each pack of nel®navir prescribed, and were
speci®cally instructed in their use. Adherence ratios
were de®ned as the number of doses recorded by the
MEMS cap divided by the total number of doses
prescribed during the monitoring period. Non-
adherence, according to MEMS, was de®ned as when
less than 90% of the prescribed doses were recorded.
The association of adherence, as de®ned by the
SMAQ, and the effectiveness of treatment was deter-
mined to assess construct validity. Effectiveness was
de®ned as a 1.5 log10 /ml reduction in viral load, from a
baseline or viral load of less than 500 copies/ml in the
month 3 evaluation, or a viral load of less than 500
copies/ml in the month 6 or 12 evaluations. The
construct validity assessment was carried out on 2528
(84.15%) of the 3004 patients included in the study.
The reliability of the SMAQ was assessed in two ways:
(i) By measuring the internal consistency of the test in
2528 patients. Internal consistency is a measure of how
similar the responses to the items in a questionnaire
are. The number of items (the more the items, the
greater the reliability) bears an in¯uence. So, in the case
of a short instrument for use in clinical practice,
internal consistency may be compromised thereby; (ii)
By measuring the reproducibility of the SMAQ when
administered ®rst by a physician and then by a nurse or
pharmacist (treatment adherence counsellor; TAC).
The SMAQ was administered by both to the patients
on the same day (1376 patients). Inter-observer relia-
607
bility de®nes the degree of similarity of measurements
made by different observers with the same instrument.
To minimize the possibility of real clinical change, it is
recommended that there be only a short lapse of time
between the two measurements.
Patients and development of the study
The Grupo EspanÄol para el Estudio Multifactorial de la
Adherencia (GEEMA) Study enrolled 3004 non-
selected HIV-infected patients who had initiated nel®-
navir treatment in combination with other antiretroviral
drugs between January 1998 and December 1999.
Having initiated nel®navir therapy was chosen as a
criterion for inclusion as, at the time, the drug had only
recently been licensed in Spain, thus facilitating the
speedy enrollment of patients. The study was prospec-
tive, multi-centre (69 hospitals) and nationwide (Spain).
All HIV-infected patients in Spain receive antiretroviral
medication free of charge.
All patients were monitored when on routine clinical
appointments. A structured interview was given, and
the SMAQ administered by the patient's usual physi-
cian (and by the TAC in a sub-group of patients) at 3,
6 and 12 months from starting the new treatment.
Sociodemographic information (age, sex, risk behaviour
related to acquiring the infection, etc.), as well as
clinical data (CDC classi®cation of HIV infection, CD4
cell count, viral load) were collected. Information
about previously prescribed drugs, and the drugs
prescribed at the start of the study, including the
reasons for starting HAART or changing from previous
HAART, was also collected.
Statistical analysis
The following statistics were computed: the sensitivity
of the SMAQ (rate of positive non-adherent patients as
per the SMAQ versus true non-adherent patients);
speci®city (rate of adherent patients as per the SMAQ
versus true adherent patients); and positive predictive
value (the probability of a patient being truly non-
adherent when the test shows him or her to be so). We
also calculated the positive likelihood ratio and the
corresponding 95% con®dence interval (CI) [23]. The
standard used for comparison was the MEMS cap
measure of adherence.
Either Fisher's exact test or the chi square trend test
(Mantel±Haenszel) were used to test the degree of
association of categorical variables. The effect of in-
dependent variables on viral load suppression (effective-
ness) was assessed for the whole cohort, and for the
subset of the protease inhibitor (PI)-naive patients, with
multiple logistic regression analysis. The independent
variables were: sex, age less than 40 years, baseline
CD4 T cell count less than 200 3 106 /l, baseline viral
load greater than 5 log10 , PI-experienced patients, and
non-adherence as measured by the SMAQ. The de-
608 AIDS 2002, Vol 16 No 4

pendent variable was: viral load suppression (, 500
copies/ml) after 12 months of HAART.
The internal consistency of the six items of the SMAQ
was evaluated by means of Cronbach's alpha coef®-
cient. Cohen's kappa test was used to compare the
results the physicians and the TAC obtained using the
SMAQ. All the data were processed on SAS statistical
analysis software (SAS Institute, Cary, NC, USA). A
value of P , 0.05 obtained in a two-tailed test was
considered to be statistically signi®cant.
Results
The baseline pro®le of the 3004 patients included in
the study is shown in Table 1. The participants were
72% men and the mean age was 35.8 years. Most had
acquired HIV through intravenous drug use (65%).
The median CD4 cell count was 270 3 106 cells/l and
the median HIV-RNA level was 4.3 log10 copies/ml.
The educational level of the patients ranged from high
school graduates (60%) to patients with no formal
education (8%).
A comparison of the non-adherent patients detected by
MEMS and by SMAQ among the 40 patients tested at
month 3, is shown in Table 2. The percentage of non-
adherent patients found by the SMAQ was 37.5%, in
comparison with the 45% detected by MEMS. The
sensitivity of the SMAQ in detecting non-adherent
patients was found to be 72% (95% CI 58±86);
speci®city, 91% (95% CI 82±100); positive predictive
value: 87% (95% CI 76±97); and positive likelihood
ratio 7.94 (95% CI 6.44±9.45). The ®rst evaluation of
the SMAQ at month 3 incorporated 2528 patients; the
reasons for the missing patients were: death: 21 (0.7%);
adverse events: 104 (3.5%); progression: 36 (1.2%); and
loss to follow-up: 315 (10.5%).
The relationship between the individual items of the
SMAQ and the effectiveness of treatment at month 3 is
shown in Table 3. Different responses to the items of
the SMAQ coincided with different virological out-
comes. Not all patients responded to all items and,
according to the question, the percentage of patients
categorized as non-adherent ranged from 15 to 30%.
De®ning adherence on the basis of the combined
results of all the items, the percentage of non-adherent
patients found was 36.6%.

The association of adherence, as measured by the

SMAQ, and the virological outcome is shown in Table
Table 1. Baseline characteristics of the 3004 HIV-infected patients
4. A total of 1797 patients were evaluated at month 12;
included. 642 were PI-naive, 805 were PI-experienced on
salvage therapy, whereas 350 were PI-experienced, but
Total
N ˆ 3004
had been changed to nel®navir after an adverse event
while on the previous therapy. Among the PI-naive
Mean age (years) (SD) 35.8 (7.9) subset, the results were: percentage of non-adherent
Sex (%)
Male 2163 (72)
patients, 29.4%, viral suppression (, 500 copies/ml)
Female 841 (28) 71.8% [79% for adherent and 54.5% for non-adherent
Transmission group (%)
Intravenous drug users 1953 (65)
Homosexual 360 (12)
Heterosexual 601 (20) Table 2. Comparison between the simpli®ed medication adherence
Others 90 (3) questionnaire assessment of adherence and the medication-event
Clinical stage (%) monitoring system adherence measurement in 40 patients at month
A 1112 (37) 3.
B 841 (28)
C 1051 (35) MEMSa
Reasons for initiation of nel®navir (%)
Progression of disease 1382 (46) Non-
Adverse events 1051 (35) adherent Adherent
Investigator's criteria 451 (15)
Patient criteria 120 (4) SMAQ Positive (non-adherent) 13 2 15 (37.5%)
Previous treatment status (%) Negative (adherent) 5 20 25 (62.5%)
Antiretroviral therapy-naive 625 (20.8) 18 (45%) 22 (55%) 40 (100%)
PI-naive 455 (15.2)
PI-experienced 1924 (64.0) Sensitivity (95% CI ): 72% (58±86)
Nel®navir dosage (%) Speci®city (95% CI): 91% (82±100)
Twice a day 1742 (58) Positive predictive value (95% CI): 87% (76±97)
Thrice a day 1262 (42) Negative predictive value (95% CI): 80% (68±92)
Mean CD4 cell count 3106 /l (SD) 315.1 (299.4) Positive likelihood ratio (95% CI): 7.94 (6.4±9.4)
Median CD4 cell count 3 106 /l (IQR) 270 (20±520) Negative likelihood ratio (95% CI): 0.31 (0.28±0.34)
Mean HIV-RNA log10 copies/ml (SD) 4.04 (1.20) Pre-test probability (95% CI): 45% (30±60)
Median HIV-RNA log10 copies/ml (IQR) 4.26 (2.26±6.26) Post-test probability (95% CI): 87% (80±94)
MEMS, Medication-event monitoring system; SMAQ, simpli®ed
IQR, Interquartile range; PI, protease inhibitor; SD, standard devia- medication adherence questionnaire.
a
tion. Over 90% of prescribed doses taken.
 Validation of a simpli®ed medication adherence questionnaire Knobel et al. 609

Table 3. Relationships between effectiveness and individual questions of the simpli®ed medication adherence questionnaire in 2528 patients
evaluated at month 3.

Total Response to adherence

responses questions (%) Effectiveness (%)a OR (95% CI)b

1. Do you ever forget to take your medicine? 2445 Yes (%) 855 (35) 479 (56)
No (%) 1590 (65) 1175 (73.9) 2.1 (1.8±2.5)
2. Are you careless at times about taking your medicine? 2467 Yes (%) 409 (17) 202 (49.5)
No (%) 2058 (83) 1449 (70.4) 2.4 (1.9±3.0)
3. Sometimes if you feel worse, do you stop taking your 2447 Yes (%) 487 (20) 260 (53.4)
medicines?
No (%) 1960 (80) 1380 (70.4) 2.07 (1.7±2.5)
4. Thinking about the last week. How often have you not taken 2451 Never (%) 1728 (70.5) 1258 (72.8) 1 (reference)
your medicine?
1±2 times (%) 464 (19) 293 (63.2) 1.6 (1.3±1.9)
3±5 times (%) 122 (5) 60 (49.2) 2.8 (1.9±4)
6±10 times (%) 37 (1.5) 11 (29.7) 6.3 (3.1±12.6)
. 10 times (%) 100 (4) 22 (22) 9.5 (5.9±15.4)
5. Did you not take any of your medicine over the past 2427 Yes (%) 508 (21) 253 (49.8)
weekend?
No (%) 1919 (79) 1380 (71.9) 2.5 (2±3)
6. Over the past 3 months, how many days have you not taken 1949 < 2 days 1661 (85) 1176 (70.8)
any medicine at all?
. 2 days. 288 (15) 132 (45.8) 2.9 (2.2±3.7)
a
Effectiveness de®ned as viral load less than 500 copies/ml or viral load reduction greater than 1.5 log.
b
In all questions, P , 0.001.

Table 4. Relationship between adherence measured by the simpli®ed medication adherence

questionnaire and virological outcome in total cohort of patients.

Viral load , 500

Patients Number (%) copies/ml OR; 95% CI P

3 months Adherents 1602 (63.4) 1175 (73.3%) 2.2; 1.8±2.6 0.001

(n ˆ 2528) Non-adherents 926 (36.6) 514 (55.5%)
6 months Adherents 1374 (64.6) 943 (68.6%) 2.6; 2.2±3.1 0.001
(n ˆ 2127) Non-adherents 753 (35.4) 347 (46.1%)
12 months Adherents 1216 (67.7) 831 (68.3%) 2.5; 2.0±3.1 0.001
(n ˆ 1797) Non-adherents 581 (32.3) 267 (46%)

CI, 95% con®dence intervals; OR, odds ratio; 95%.

patients, odds ratio (OR) 3.06, 95% CI 2.1±4.5, P ˆ
0.0001]. In the PI-experienced subset on salvage
therapy, the results were: percentage of non-adherent
patients 33%, viral suppression 39% (59% for adherent
and 33% for non-adherent patients, OR 2.9, 95% CI
2.1±4; P ˆ 0.0001).
Table 5 shows the results of the logistic regression
analysis that takes viral suppression as the dependent
variable. Age, sex, baseline viral load of over 5 log10 /
ml, and CD4 cell count of less than 200 3 106 /l were
not found to be signi®cantly related to viral suppres-
sion; only non-adherence (OR 1.95; 95% CI 1.32±
2.89; P ˆ 0.0007) was associated with failing to achieve
viral suppression in the PI-naive subset. However, in
the cohort as a whole, the factors associated with failing
to achieve viral suppression were: baseline viral load of
over 5 log10 /ml (OR 1.77; 95% CI 1.36±2.12); CD4
cell count of less than 200 3 106 /l (OR 1.39; 95% CI
1.11±1.74); PI-experienced (OR 2.5; 95%CI 1.98±
3.17); and non-adherence (OR 1.73; 95% CI 1.4±
2.14).
The Cronbach's alpha internal consistency coef®cient
of the SMAQ was 0.75; 95% CI 0.73±0.76. The inter-
observer reliability analysis (physicians and TAC) re-
vealed an overall level of coincidence of 88.2% (kappa
0.74; 95% CI 0.69±0.79) for all patients evaluated at
months 3, 6 and 12 by two observers (n ˆ 1376).
Discussion
The study includes a very large and highly representa-
tive sample of Spanish patients who received antiretro-
viral treatment for HIV. The percentage of men, and
the percentage of patients who admitted having used
drugs intravenuosly in our study (72 and 65%, respec-
610 AIDS 2002, Vol 16 No 4
Table 5. Logistic regression analysis of factors associated with virological failure (. 500 copies/ml) in the total cohort and in protease inhibitor-
naive patients at month 12.
Total cohort
N ˆ 1797 patients
OR 95% CI
Age 0.99 0.98±0.99
Sex (male) 1.26 0.99±1.59
Baseline viral load . 5 log/ml 1.77 1.36±2.12
Baseline CD4 cell count , 200 3 106 /l 1.39 1.11±1.74
Non-adherence 1.73 1.40±2.14
PI-experienced 2.50 1.98±3.17
CI, Con®dence interval; OR, odds ratio; PI, protease inhibitor.
tively) are very similar to those reported in the
National AIDS Report (80 and 65%, respectively) [24].
The SMAQ was compared with MEMS in 40 patients.
We found sensitivity to be 72%, speci®city 91%, posi-
tive predictive value 87%, and a positive likelihood
ratio of 7.94. The results suggest that the SMAQ is a
valid indicator of a patient's non-adherence to treat-
ment. Although the sample represents only a small part
of the total number of patients included in the study,
the rate of non-adherence found by the SMAQ
(37.5%) is not signi®cantly different to that of the
cohort as a whole (36.6%). Electronic measurement
(MEMS) offers advantages over subjective or other
objective forms of measurement. If used properly the
technique provides additional information, beyond the
simple number of pills taken [25]; that is, the time and
date of pill bottle openings. Several studies have
demonstrated electronic monitoring of adherence to be
more sensitive for non-adherence than either pill
counts or self-reporting, and that the results obtained
by such measurement closely correlate to virological
outcome [22,26,27]. The sensitivity of self-reporting,
compared with MEMS, ranges from 60 to 80% [27±
29]. Compared with an unannounced pill count, the
sensitivity of patient self-reporting is 72% [30], and
other self-reporting questionnaires show a sensitivity of
25% when compared with pharmacy pill counts [31].
Nevertheless, the results obtained by the SMAQ may
be ascribed to the high prevalence of non-adherence in
the population studied, especially with respect to the
positive predictive value. It is likely that the perform-
ance of the test would be poorer in settings with better
adherence rates. Even in such a favourable situation,
the SMAQ was not found to be completely accurate,
because it under-diagnosed cases of non-adherence by
28%. Notwithstanding, in our population the post-test
probability of being non-adherent with a positive
SMAQ was approximately 87% (approximately double
the pre-test probability). Additional items, or a lower
threshold for non-adherence, would certainly improve
PI-naive
N ˆ 642 patients
P OR 95% CI P
0.7 1 0.98±1.02 0.9
0.056 1.07 0.7±1.65 0.7
0.00001 1.16 0.77±1.74 0.5
0.004 1.48 0.99±2.2 0.056
0.00001 1.95 1.32±2.89 0.0007
0.00001 ± ± ±
the sensitivity of the questionnaire. The SMAQ should
be tested further before recommending its use in
clinical practice.
Adherence, as measured by the SMAQ, showed a
positive association with respect to the virological
outcome. The percentage of PI-naive patients who
achieved viral suppression was 79% for adherent pa-
tients, and 54.5% for non-adherent patients. Among
the PI-experienced patients (on salvage therapy) these
percentages were 59 and 33%, respectively. This is
consistent with other studies that found that self-
reported non-adherence was associated with poorer
virological outcomes [17,32±36].
Compared with other variables (age, sex, baseline CD4
cell count, baseline viral load), non-adherence, as meas-
ured by the SMAQ, was the only signi®cant predictor
of virological failure (viral load . 500 copies/ml) at
one year, with an odds ratio of 1.9 (95% CI 1.3±2.9)
in the PI-naive subset of patients. The factors associated
with the patients' inability to achieve viral suppression
in the global cohort were high viral load, low CD4 cell
count, PI-experience and non-adherence. A variety of
factors can contribute to the failure of antiretroviral
therapy to suppress viral replication durably; viral
resistance, potency of the regimen and pharmaco-
kinetics should be added to the above mentioned
factors [37]. The results are comparable to those of
other studies [18,26,33,38±40] using different, in gen-
eral more complicated and more dif®cult to implement,
tools to assess adherence, and highlight the usefulness
of adherence measurement for the monitoring of HIV
treatment.
The SMAQ showed suf®cient internal consistency
(Cronbach's alpha 0.75) and satisfactory reproducibility
when tested by two different health providers (overall
agreement of 88.2%, kappa 0.74). Such strict reliability
criteria have not been reported by previous studies.
Multidimensional adherence measures usually apply a
 Validation of a simpli®ed medication adherence questionnaire Knobel et al.
scale that is believed to re¯ect the dimensions of
adherence/non-adherence. The missed-dose method is
simpler; adherence is usually calculated as a percentage
discrepancy score [(prescribed doses ÿ missed doses)/
(prescribed doses)]. The discordance of the different
measures of self-reported adherence was highlighted by
Gao and Nau [41], who studied 65 HIV patients with
three different self-reporting methods. The investigators
found the percentage of adherent patients not to be
congruent: 29.2% with a Morisky-type scale, 78.5% if
adherence is de®ned as the percentage of doses taken as
prescribed over the past 2 days, and 95.4% over the
past 2 weeks, with a cut-off value of 90%. Chesney et
al. [42] developed a self-report instrument for use with
Adult AIDS Clinical Trials. Of the 75 patients evalu-
ated, 11% reported missing at least one dose the day
before the interview, and 17% reported missing at least
one dose during the previous 2 days; the test was self-
administered to patients who had at least a sixth-grade
level reading ability. The SMAQ integrates multi-
dimensional and missed-dose measures; because the
level of adherence required to achieve the greatest
bene®t in HIV infection is very high [26], a dichoto-
mous adherent/non-adherent variable would appear to
be valid. The cut-off value of 90±95% of prescribed
doses, employed in the present study to distinguish
adherence and non-adherence, was adopted by other
researchers working in the area of HIV infection
[12,26,31,41,43]. The SMAQ asks patients to indicate
the number of missed doses over the past week, the
past weekend, and the number of days without any
medication at all over the past 3 months. Such evalua-
tion is more likely to re¯ect the overall trend in patient
medication adherence. The limitations of the test
include recall and social desirability bias, common
problems in all self-report surveys.
Recently, Liu et al. [44] demonstrated that three
adherence measures (MEMS, pill counts, and inter-
view) all have shortcomings that may be reduced by
employing a combined measurement system. However,
such a comprehensive adherence measure may prove
unrealistic for application in clinical care and, in spite
of ¯aws and errors, patient interview remains the most
practical approach for clinicians [45]. It is important to
highlight that the SMAQ performed similarly to other
more sophisticated adherence assessment tools.
Conclusion
The SMAQ may prove to be an adequate instrument
for the assessment of adherence among HIV-infected
patients; it shows adequate levels of sensitivity and
speci®city when compared with other more objective
measures; it is reliable, showing suf®cient internal
consistency and reproducibility, easy to apply (under
611
5 min), inexpensive, and the results found correlate
strongly with the virological outcome. It is, therefore,
an instrument that may be used in the majority of
clinical settings.
Acknowledgements
The authors gratefully acknowledge the contributions
of Susana Traseira, Medical Manager (Roche, Spain),
of the operational and coordination tasks of the
GEEMA group, and Jose Javier GarcõÂa (CIBEST) for
statistical advice.
Sponsorship: This study was partly funded by an unrest-
ricted grant from Roche Laboratories, Spain.
References
1. Palella FJ, Delaney KM, Morman AC, Loveless MO, Fuhrer J,
Satten GA. Declining morbidity and mortality among patients
with advanced human immunode®ciency virus infection. HIV
Outpatient Study Investigators. N Engl J Med 1998, 338:
853±860.
2. Carpenter CCJ, Cooper DA, Fischl MA, et al. Antiretroviral
therapy in adults. Updated recommendations of the Interna-
tional AIDS Society ± USA Panel. JAMA 2000, 3:381±391.
3. Williams A, Friedland G. Adherence, compliance, and HAART.
AIDS Clin Care 1997, 9:51±58.
4. Montaner JS, Reiss P, Cooper D, et al. A randomized, double-
blind trial comparing combinations of nevirapine, didanosine,
and zidovudine for HIV-infected patients: the INCAS Trial. Italy,
the Netherlands, Canada and Australia Study. JAMA 1998,
279:930±937.
5. Kaufmann D, Pantaleo G, Sudre P, Telenti A. CD4-cell count in
HIV-1 infected individuals remaining viraemic with highly active
antirretroviral therapy (HAART). Lancet 1998, 351:723±724.
6. Fatkenheur G, Theisen A, Rockstroh J, et al. Virological treatment
failure of protease inhibitor therapy in an unselected cohort of
HIV-infected patients. AIDS 1997, 11:F113±F116.
7. Cohen OJ, Fauci AS. Transmission of multidrug-resistant human
immunode®ciency virus. The wake-up call. N Engl J Med 1998,
339:341±343.
8. Wainberg MA, Friedland GH. Public health implications of
antiretroviral therapy and HIV drug resistance. JAMA 1998,
279:1977±1983.
9. Sackett DL, Snow JC. The magnitude of compliance and non-
compliance. In: Compliance with therapeutic regimens. Sackett
DL, Haynes RB (editors). Baltimore: Johns Hopkins University
Press; 1976. pp. 11±27.
10. RodrõÂguez-Rosado R, JimeÂnez-Nacher I, Soriano V, Anton P,
Gonzalez-Lahoz J. Virological failure and adherence to anti-
retroviral therapy in HIV-infected patients. AIDS 1998, 12:1112.
11. LoÂpez-SuaÂrez A, FernaÂndez-GutieÂrrez del AÂlamo, PeÂrez GuzmaÂn
E, GiroÂn-GonzaÂlez JA. Adherence to the antiretroviral treatment
in asymptomatic HIV-infected patients. AIDS 1998; 12:
685±686.
12. TuldraÁ A, Ferrer MJ, R Fumaz C, et al. Monitoring adherence to
HIV-therapy. Arch Intern Med 1999, 159:1376±1377.
13. Gallant JE, Block DS. Adherence to antiretroviral regimens in
HIV-infected patients: results of a survey among physicians and
patients. J Int Assoc Physician AIDS Care 1998, 4:32±35.
14. Miller LG, Hays RD. Measuring adherence to antiretroviral
medications in clinical trials. HIV Clin Trials 2000, 1:36±46.
15. Rudd P. The measurement of compliance medication taking. In:
Developmental aspects of health compliance behavior. Krasnegor
NA, Epstein L, Johnson SB, Yaffe SJ (editors). Hilllsdale, NJ:
Lawrence Erlbaum Associates; 1993. pp. 185±213.
612 AIDS 2002, Vol 16 No 4
16. Chesney MA, Ickovics J, Hecht FM, Sikipa G, Rabkin J. Adher-
ence: a necessity for successful HIV combination therapy. AIDS
1999, 13 (Suppl. A):S271±S278.
17. De Masi R, Tolson J, Pham S, et al. Self-reported adherence to
HAART and correlation with HIV RNA: initial results with the
Patient Medication Adherence Questionnaire. In: 6th Conference
on Retroviruses and Opportunistic Infections. Chicago. January±
February 1999 [Abstract 94].
18. Haubrich RH, Little SJ, Currier JS, et al. The value of patient-
reported adherence to antiretroviral therapy in predicting vir-
ologic and immunologic response. AIDS 1999, 13:1099±1107.
19. Cinti SK. Adherence to antiretrovirals in HIV disease. AIDS
Reader 2000, 10:709±717.
20. Morisky DE, Green LW, Levine DM. Concurrent and predictive
validity of a self-reported measure of medication adherence.
Med Care 1986, 24:67±74.
21. Samet JH, Libman H, Steger KA, et al. Compliance with
zidovudine therapy in patients infected with human immuno-
de®ciency virus type 1: a cross-sectional study in a municipal
hospital clinic. Am J Med 1992, 92:495±502.
22. Kastrissios H, Suarez JR, Katzenstein D, Girard P, Sheiner LB,
Blaschke TF. Characterising patterns of drug-taking behaviour
with a multiple drug regimen in an AIDS clinical trial. AIDS
1998, 12 :2295±2303.
23. Sackett DL, Richardson WS, Rosenberg W, Haynes B. Evidence-
based medicine: how to practice and teach EBM. London:
Churchill Livingstone; 1996.
24. Vigilancia EpidemioloÂgica del SIDA en EspanÄa. Centro Nacional
de EstadõÂstica. Ministerio de Sanidad y Consumo. Informe semes-
tral no. 2; 2000.
25. Cramer JA. Microelectronic systems for monitoring and enhan-
cing patient compliance with medication regimens. Drugs 1995,
49:321±327.
26. Paterson DL, Swindells S, Mohr J, et al. Adherence to protease
inhibitor therapy and outcomes in patients with HIV infection.
Ann Intern Med 2000, 133:21±30.
27. Arnsten JH, Demas PA, Farzadegan H, et al. Antiretroviral
therapy adherence and viral suppression in HIV-infected drug
users: comparison of self-report and electronic monitoring. Clin
Infect Dis 2001, 33:1417±1423.
28. Melbourne KM, Gelekto SM, Brown SL, et al. Medication
adherence in patients with HIV infection: a comparison of two
measurement methods. AIDS Reader 1999, 9:329±338.
29. Miller L, Liu H, Beck K, et al. Providers' estimates of adherence
overestimate reports from Medication Event Monitoring System
(MEMS) for patients on protease inhibitors (PIs). In: 6th Con-
ference on Retroviruses and Opportunistic Infections. Chicago,
February 1999 [Abstract 97].
30. Bangsberg DR, Hecht FM, Clague H, et al. Provider assesment of
adherence to HIV antiretroviral therapy. J Acquir Immune De®c
Syndr 2001, 26:435±442.
31. MartõÂn J, Escobar I, Rubio R, et al. Study of the validity of a
questionnaire to assess the adherence to therapy in patients
infected by HIV. HIV Clin Trials 2001, 2:31±37.
32. Bangsberg DR, Hech FM, Charlebois EC, et al. Spontaneous
adherence audits predict viral suppression in the REACH cohort.
In: 6th Conference on Retroviruses and Opportunistic Infections.
Chicago, February 1999 [Abstract 93].
33. Duong M, Piroth L, Peytavin G, et al. Value of patient self-report
and plasma human immunode®ciency virus protease inhibitor
level as markers of adherence to antiretroviral therapy: relation-
ship to virologic response. Clin Infect Dis 2001, 33:386±392.
34. Murri R, Ammassari A, Gallicano K, et al. Patient-reported
nonadherence to HAART is related to protease inhibitor levels.
J Acquir Immune De®c Syndr 2000, 24:123±128.
35. Le Moing V, Chene G, Carrieri MP, et al. Clinical, biologic, and
behavioral predictors of early immunologic and virologic re-
sponse in HIV-infected patients initiating protease inhibitors.
J Acquir Immune De®c Syndr 2001, 27:372±376.
36. Knobel H, Carmona A, Grau S, et al. Adherence and effective-
ness of highly active antiretroviral therapy. Arch Intern Med
1998, 158:1953.
37. Deeks SG. Determinants of virologic response to antiretroviral
therapy: implications for long-term strategies. Clin Infect Dis
2000, 30 (Suppl. 2):S177±S184.
38. Bansberg DR, Hecht FM, Charlebois ED, et al. Adherence to
protease inhibitors, HIV-1 viral load, and development of drug
resistance in an indigent population. AIDS 2000, 14:357±366.
39. Nieuwkerk PT, Sprangers MAG, Burger DM, Hoetelmans RMW,
et al. Limited patient adherence of highly active antiretroviral
therapy for HIV-1 infection in an observational cohort study.
Arch Intern Med 2001, 161:1962±1968.
40. Knobel H, Guelar A, Carmona A, et al . Virologic outcome and
predictors of virological failure on highly active antiretroviral
therapy containing protease inhibitors in a cohort of HIV-
infected patients. AIDS Patient Care STDs 2001, 15:193±199.
41. Gao X, Nau DP. Congruence of three self-report measures of
medication adherence among HIV patients. Ann Pharmacother
2000, 34:1117±1122.
42. Chesney MA, Ickoviks JR, Chambers DB, et al. Self-reported
adherence to antiretroviral medications among participants in
HIV clinical trials: the AACTG adherence instruments. AIDS
Care 2000, 12:255±266.
43. Tseng AL. Compliance issues in the treatment of HIV infection.
Am J Health Syst Pharm 1998, 55:1817±1824.
44. Liu H, Golin CE, Miller LG, et al. A comparison study of multiple
measures of adherence to HIV protease inhibitors. Ann Intern
Med 2001, 134:968±977.
45. Turner BJ, Hecht FM. Improving on a coin toss to predict patient
adherence to medications. Ann Intern Med 2001, 134:1004±
1006.
Appendix
Investigators of the GEEMA Study: Teresa VaÂz-
quez CastanÄoÂn, Hospital Alvarez Buylla, Asturias; Ana
MarinÄo Callejo, Hospital Arquitecto Marcide-Novoa
Santos, La CorunÄa; Daniel Podzamczer Palter, Hospital
Bellvitge, Barcelona; RamoÂn Canet, Hospital Can
Misses, Ibiza; Jose Manuel Antunez Galvez, Hospital
Carlos Haya, MaÂlaga; M a Victoria Gordillo, Hospital
Carlos III, Madrid; VõÂctor Roca Arbones, Hospital
ClõÂnico San Carlos, Madrid; Luis Force, Hospital
Consorcio Sanitario de MataroÂ, Barcelona; Jordi de
Otero, Hospital Creu Roja, Barcelona; Isabel Gracia
Marce, Hospital Creu Roja, Hospitalet, Barcelona; M a
Jose LoÂpez Alvarez, Hospital de Calde, Lugo; Jose Luis
Mostaza FernaÂndez, Hospital del Bierzo, LeoÂn; Her-
nando Knobel, Hospital del Mar, Barcelona; Luis
Morano Amado, Hospital do Meixoeiro, Pontevedra;
Josep Cucurull, Hospital Figueres, Gerona; Carlos
Tornero EsteÂbanez, Hospital Francesc de Borja, Gan-
dia, Valencia; Julio Collazos, Hospital Galdakano,
Vizcaya; Elisa MartõÂnez Alfaro, Hospital General de
Albacete, Albacete; Luis Caminal Montero, Hospital
General de Asturias, Asturias; Jordi Uso Blasco, Hospi-
tal General de CastelloÂn, CastelloÂn; Elena Losada,
Hospital General de Galicia, Santiago, CorunÄa; Enric
Pedrol, Hospital General de Granollers, Barcelona;
Manuel RodrõÂguez Zapata, Hospital General de Gua-
dalajara, Guadalajara; Ester Dorca, Hospital General de
Manresa, Barcelona; Alfredo Cano SaÂnchez, Hospital
General de Murcia, Murcia; Josep VilaroÂ, Hospital
General de Vic, Barcelona; Juan F. Lorenzo, Hospital
General YaguÈe, Burgos; M a Luisa GarcõÂa-Alcalde,
Hospital de CabuenÄes, Asturias; Ignacio SuaÂrez Lozano,
Hospital Infanta Elena, Huelva; Javier Juega Puig,
Hospital Juan Canalejo, La CorunÄa; Jose JoaquõÂn
Peraire Forner, Hospital Juan XXIII (Universitari
Rovira I Virgili ± Institut de Estudis AvancËat), Tarra-
 Validation of a simpli®ed medication adherence questionnaire Knobel et al.
gona; Pepe LoÂpez Aldeguer, Hospital La Fe, Valencia;
Juan GonzaÂlez GarcõÂa, Hospital La Paz, Madrid; Fran-
cisco Pasquau LianÄo, Hospital Marina Baixa, Villajoyo-
sa, Alicante; Carmen FarinÄas Alvarez, Hospital Marques
de Valdecilla, Santander; Rafael Ojea de Castro, Hospi-
tal Montecelo, Pontevedra; Carlos Barros Aguado,
Hospital de MoÂstoles, Madrid; David Dalmau, Hospital
Mutua de Tarrasa, Barcelona; Jose A. Iribarren, Hospi-
tal Ntra. Sra. de AraÂnzazu, GuipuÂzcoa; Fernando
Marcos SaÂnchez, Hospital Ntra. Sra. del Prado, Toledo;
Eva LeoÂn JimeÂnez, Hospital Ntra. Sra. de Valme,
Sevilla; Manuel Cervantes, Hospital Parc Tauli, Barce-
lona; Manuel Camba Esteve, Hospital Policlinico Vigo,
Pontevedra; Alberto Arranz Caso, Hospital PrõÂncipe de
Asturias, Madrid; Ricardo RodrõÂguez Real, Hospital
Provincial RebulloÂn, Pontevedra; Joan Colomer, Hos-
pital Provincial Sta. Caterina, Gerona; Teodoro MartõÂn
JimeÂnez, Hospital Puerta de Hierro, Madrid; Eugenio
PeÂrez GuzmaÂn, Hospital Puerta del Mar, CaÂdiz; Anto-
nio Vergara de Campos, Hospital Puerto Real, CaÂdiz;
Jose Luis Casado Osorio, Hospital RamoÂn y Cajal,
Madrid; Jose M a Kindelan Jaquotot, Hospital Reina
SofõÂa, CoÂrdoba; M a Jose Tuya Moran, Hospital San
AgustõÂn, Asturias; Jose M a Cuadrado Pastor, Hospital
San Juan, Alicante; Antonio Delegido SaÂnchez, Hospi-
tal Sant Pau I Santa Tecla, Tarragona; Pere Domingo
Pedrol, Hospital Sant Pau, Barcelona; RauÂl RodrõÂguez
PeÂrez, Hospital Santa MarõÂa Nai- Cabaleiro Goas,
Orense; Alberto TerroÂn Pernia, Hospital Sas de Jerez,
CaÂdiz; Joseba Portu, Hospital Txagorritxu, Vitoria;
VõÂctor Carcaba FernaÂndez, Hospital Valle del NaloÂn,
Asturias; Isabel Ruiz Camps, Hospital Valle HebroÂn,
Barcelona; Ismael PinÄas Forcadell, Hospital Verge de la
Cinta, Tarragona; Eduardo RodrõÂguez de Castro, Hos-
613
pital Verge del Toro, Menorca; Paloma Geijo, Hospital
Virgen de la Luz, Cuenca; Fernando Cuadra GarcõÂa-
Tenorio, Hospital Virgen de la Salud, Toledo; Manuel
Marquez Solero, Hospital Virgen de la Victoria, MaÂla-
ga; Carlos Galera PenÄaranda, Hospital Virgen de la
Arrixaca, Murcia; Miguel Angel Colmenero Camacho,
Hospital Virgen de la Macarena, Sevilla; Jose Adolfo
GarcõÂa Henarejos, Hospital Virgen del Rosell, Murcia;
Antonio Ocampo, Hospital Xeral Xies, Vigo.
Treatment adherence counsellors: Jose®na Cardo-
na, Hospital Can Misses, Ibiza; Francisca Cordero,
Hospital Carlos Haya, MaÂlaga; Beatriz Wolgeschafen
Torres, Hospital Carlos III, Madrid; M a Jose GaÂmir
PeÂrez, Hospital ClõÂnico San Carlos, Madrid; Visi
JimeÂnez, Hospital Creu Roja, Hospitalet, Barcelona;
Alexia Carmona, Hospital del Mar, Barcelona; Rosa
Soler Arnau, Hospital General De Granollers, Barce-
lona; Jose Domingo Pedreira Andrade, Hospital Juan
Canalejo, CorunÄa; Emilio Monte Boquet, Hospital La
Fe, Valencia; Emilia Condes Moreno, Hospital de
Mostoles, Madrid; MarõÂa Sanjaume, Hospital Mutua
de Tarrasa, Barcelona; AdoracioÂn Torres, Hospital
Parc Tauli, Barcelona; Rosario GarcõÂa Juarez, Hospital
Puerta del Mar, CaÂdiz; M a de los Angeles MartõÂn
MartõÂn, Hospital Puerto Real, CaÂdiz; Raquel Sabido,
Hospital RamoÂn y Cajal, Madrid; Aureliana Segador
GoÂmez, Hospital Reina SofõÂa, CoÂrdoba; Angels Fon-
tanet, Hospital Sant Pau, Barcelona; M a de los
Angeles MunÄoz Arenillas, Hospital Sas de Jerez,
CaÂdiz; Rosa LoÂpez, Hospital Valle HebroÂn, Barcelo-
na; Isabel DomõÂnguez, Hospital Virgen de la Victoria,
MaÂlaga; Carmen Buendia, Hospital Virgen del Rosell,
Murcia.