SYSTEMATIC REVIEW

Does mHealth increase adherence to medication? Results

of a systematic review
H. Anglada-Martinez, G. Riu-Viladoms, M. Martin-Conde, M. Rovira-Illamola, J. M. Sotoca-Momblona,
C. Codina-Jane

Pharmacy Service, Hospital

SUMMARY
Review criteria Clinic, Barcelona, Catalonia,
Aims: Adherence to medication is a major problem that affects 50–60% of chron- The review was based on information gathered from Spain

ically ill patients. As mobile phone use spreads rapidly, a new model of remote a systematic search in four databases (CINAHL,
Correspondence to:
PubMed, Scopus, and PsycARTICLES). The articles
health delivery via mobile phone – mHealth – is increasingly used. The objective Helena Anglada-Martinez,
covered were randomized clinical trials (RCT), quasi-
of this study is to provide a comprehensive overview of how mHealth can be used Pharmacy Service, Hospital
RCT, cross-sectional studies, case-control studies, Clinic, C/Villarroel 170,
to improve adherence to medication. Methods: A systematic literature review was pre- and post-intervention studies, and literature Barcelona, Catalonia 08045,
conducted using four databases (CINAHL, PubMed, Scopus and PsycARTICLES). Eli- reviews. Interventions were considered to be eligible Spain
gible articles available on March 2014 had to be written in English or Spanish and when they were based on sending text messages or Tel.: 932275479
have a comparative design. Articles were reviewed by two authors independently. using a smartphone application. Information on Fax: 932275457
A Cochrane Collaboration tool was used to assess the studies based on their inter- adherence was abstracted from the selected articles Email: hangladamartinez@gmail.
and summarised. com
nal validity. Results: Of the 1504 articles found, 20 fulfilled the inclusion criteria
[13 randomised clinical trials (RCT), one quasi-RCT, one non-randomised parallel
Message for the clinic
group study and five studies with a pre-post design]. Nearly all the trials were
Adherence to medication is a major problem that Disclosure
conducted in high-income countries (80.0%). Articles were categorised depending affects 50% to 60% of chronically ill patients. As None.
on the target population into three different groups: (i) HIV-infected patients, mobile phone use is spreading rapidly, even in low-
n = 5; (ii) patients with other chronic diseases (asthma, coronary heart disease, income countries, a new model of remote health
diabetes mellitus, hypertension, infectious diseases, transplant recipients and psori- delivery via mobile phone—mHealth—is increasingly
asis), n = 11; and (iii) healthy individuals, n = 4. Adherence improved in four of used. The use of text messages or mobile applications
the studies on HIV-infected patients, in eight of the studies on patients with other to enhance adherence to medication do seem to have
been beneficial, as 65% of the studies found had
chronic diseases, and in 1 study performed in healthy individuals. All studies
positive outcomes. However, more high-quality
reported sending SMS as medication reminders, healthy lifestyle reminders, or
studies should be conducted in order to demonstrate
both. Only one trial (HIV-infected patients) had a low risk of bias. Conclusions: whether this type of technology reduces the
Our results showed mixed evidence regarding the benefits of interventions because considerable costs to the health system generated by
of the variety of the study designs and the results found. Nevertheless, the inter- nonadherence.
ventions do seem to have been beneficial, as 65% of the studies had positive out-
comes. Therefore, more high-quality studies should be conducted.

delivery via mobile phone – mHealth – is increas-

Introduction
ingly used. In fact, the mobile phone is the most
Adherence to medication is a major problem that quickly adopted technology in the history of the
affects 50–60% of chronically ill patients (1,2). Poor world in both low- and high-income countries. The
adherence leads to negative health outcomes, such as results of a survey conducted in 2013 in the USA
treatment failure, increased frequency of hospital showed that 91% of adult respondents owned a
admissions, drug resistance in some cases (e.g. HIV mobile device (3). However, this survey showed that
or antibiotic regimens), more complex medication mobile phone technology is less keenly embraced in
plans (second-line treatments), and increased con- specific population groups, such as people aged 65
sumption of healthcare resources. Poor adherence and older, those who did not attend university, those
causes approximately 33–69% of medication-related living in households earning less than $30,000, and
hospitalisations and accounts for an annual expendi- those living in rural areas. On the other hand, a
ture of $100 billion (1). Therefore, adherence is cru- 2010 survey by the American Association of Retired
cial and should be promoted. Persons showed that 89% of individuals over age 50
As mobile phone use is spreading rapidly, even in use a mobile device, the most common being a
low-income countries, a new model of remote health mobile phone, and 7% use a smartphone (4). In

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract doi: 10.1111/ijcp.12582 1
2 mHealth and medication adherence

addition, one in 10 respondents used a mHealth No date limits were applied, as mobile phone
application to track health data (e.g. weight, blood interventions are relatively new. We included arti-
pressure and blood glucose), and four in 10 are cles written in English and Spanish focusing on the
interested in using one in the future. In a survey per- use of mobile technology to improve adherence to
formed in a low-income country (Kenya), 44% of medication. The articles covered were randomised
respondents owned a mobile phone, while 88% indi- clinical trials (RCT), quasi-RCT, cross-sectional
cated that they use one (5). studies, case–control studies, pre- and postinterven-
The World Health Organisation promotes services tion studies and literature reviews. Interventions
of this type, since they contribute to a more equita- were considered to be eligible when they were based
ble delivery of care among patients living in low- on sending text messages or using a smartphone
income countries or in rural areas (6). In addition, application. Interventions based on telephone con-
mHealth facilitates more frequent communication sultation were excluded, as they are person-depen-
with patients and provides the opportunity to deliver dent. Medication adherence was included as an
health-related messages when they may have the outcome.
greatest impact. Two authors (HAM and GRV) working indepen-
The objective of this review was to provide a com- dently reviewed the abstracts of all the studies identi-
prehensive picture of how mHealth can be use to fied through database searches or other means.
improve adherence to medication. When eligibility was unclear, we obtained the full
text of the article for closer examination.
Methods
Data extraction and management
We conducted a systematic literature review, in After having identified the articles susceptible of
which relevant studies were categorised in a 2-step being included, the two reviewers separately extracted
process. The first step included a review of the titles the following information: study design, number of
and abstracts of all publications that were identified randomised participants, participant characteristics
as potentially relevant. In the second step, selected (sex, mean age, type of disease or preventive mea-
abstracts were categorised using the guidelines of sure), intervention (mobile phone type, content of
the Cochrane Collaboration to assess studies for text messaging, frequency of text messaging, period
their internal validity and to summarise current evi- of intervention, comparator), aim of the study,
dence about mHealth interventions to improve duration, outcome measure, results and author
adherence. conclusions.
The assessment of the internal validity of each
Search strategy individual study was based on a Cochrane Collabora-
In March 2014, we performed a systematic search of tion tool (7). For RCT and quasi-RCT, the tool
four electronic databases (CINAHL, PubMed, Scopus assesses risk of bias in individual studies across six
and PsycARTICLES). A list of keywords was created domains: sequence generation, allocation conceal-
around the two domains, ‘medication adherence’ and ment, blinding (of participants, personnel, and out-
‘mHealth’. A search string was constructed using come assessors), incomplete outcome data, selective
both the conjunction ‘AND’ and the disjunctive ‘OR’ outcome reporting and other sources of bias.
as logical operators [(‘medication therapy manage-
ment’ OR ‘medication adherence’ OR ‘patient com-
Results
pliance’ OR ‘self-care’) AND (‘mHealth’ OR ‘mobile
health’ OR ‘m-health’ OR ‘mobile-health’ OR Of the 1504 potential articles, 20 fulfilled the inclu-
‘mobile phone’ OR ‘cell phone’ OR ‘cellphone’ sion criteria (Figure 1). A total of 7402 patients were
OR ‘cell-phone’ OR ‘smartphone’ OR ‘iPhone’ OR included. As for the type of article, 13 were RCT, 1 a
‘blackberry’ OR ‘android’)]. quasi-RCT, 1 a nonrandomised parallel group trial,
We included all study participants regardless of and 5 studies with a pre–post design. The character-
age, gender, and ethnicity, as well as all types and istics of the study are shown in Table 1. Most of the
stages of diseases, and studies performed in the studies (n = 16) were conducted in high-income
healthy population. We included studies in all set- countries, and most were published from 2009 to
tings, independently of the type of healthcare pro- 2012 (85.0%).
vider (e.g. nurse, doctor, allied staff). Articles were categorised depending on the target
We then examined the references of the studies population into three different groups: HIV-infected
included and searched reviews on interventions to patients, 5; patients with other chronic diseases
promote adherence. (asthma, coronary heart disease, diabetes mellitus,

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 mHealth and medication adherence 3

Records identified through

Identification

CINAHL, PubMed, Additional records identified

PsycARTICLES, Scopus through other sources
(n = 881+339+58+138 = 1416) (n = 88)

Records after duplicates (42) removed

(n = 1462)
Screening

Records excluded
(n = 1300)

Records screened
(n = 30+99+1+32 = 162)
Full-text articles excluded, with
reasons (n = 20)
- Phone call intervention:
Eligibility

3
Full-text articles - Adherence to
assessed for eligibility medication not
(n = 40) evaluated: 10
- No comparative design
study: 3
- No mobile phone
intervention: 1
Included

- Intervention performed
Studies included in the in the control group: 1
qualitative synthesis - Intervention performed
N = 20 using mobile phone or
computers: 1
- Intervention in the
control group consisted
of mobile alerts and
smart SPD: 1

Figure 1 Flow chart showing the inclusion process

hypertension, infectious diseases, transplant recipi- intervention was limited to medication reminders
ents and psoriasis), 11; healthy individuals, 4. (11), while in the remainder the message also pro-
vided motivational and reinforcement content (8–
Studies focused on HIV-infected patients 10,12). In one study, adherence was evaluated using
A total of five studies were identified. Three were several methods: visual analogue scale (VAS), self-
RCT (8–10), one a quasi-RCT (11), and one fol- reported adherence and pharmacy refills. Other
lowed a pre–post design (12). Overall, 1399 patients methods used to measure adherence included the
were recruited, with sample sizes ranging from 52 to Medication Event Monitoring System (MEMS)
538 participants. Patients included were na€ıve in one (n = 1) (9), self-reported adherence (n = 2) (8,12)
study (8), had initiated antiretroviral treatment and pill counts (n = 1) (11). Motivational content
within 1–3 months in three studies (9–11) and in included statements such as ‘You are important to
another study this issue was not addressed (12). The your family. Please remember to take your medica-
duration of the studies varied from 3 months to tion. You can call us at this number:’ (10) or ‘This is
1 year. your reminder. Be strong and courageous, we care
The intervention consisted of sending short mes- about you’ (9) or a simple question ‘How are you?’
sages (SMS) (8–12), in one case along with an inter- (8). Three studies used 2-way communications
active voice response (11). In one study (10), the encouraging participants to use their mobile phones

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 4

Table 1 Summary of study characteristics.

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

HIV studies
Lester 2010 (8) Parallel group 538 adults aged ≥ 18, Treatment-na€ıve HIV- Assess whether 12 months A text message was sent Usual care Self-report Yes
RCT; Country: na€ıve patients from infected patients sending SMS via SMS each week to
Kenya three clinics in Kenya initiating ART between healthcare enquire about patient
(273 patients providers and status and to remind
randomised to patients improves about the availability
intervention group; adherence to ART of phone-based
265 to standard of and clinical support. In Kiswahili
care); Mean age outcomes language, the SMS
(years): Intervention, (suppression of asked ‘How are you?’
36.7  8.5; Control, plasma HIV-1 RNA Patients in the
mHealth and medication adherence

36.6  7.9; Male: load, quality of life, intervention group

Intervention, 45%; retention, and were instructed to
Control, 44% mortality). respond within 48 h
that either they were
doing well or they had
a problem. The
clinician then called
the patients who
reported having a
problem or who failed
to respond within
48 h.
Pop Eleches 2011 Parallel group, 431 patients with at HIV infection Compare adherence 12 months Medication reminder Usual care MEMS in one of the Yes
(9) RCT; Country: least 3 months on rates between four system vs. usual care. treatment drugs
Kenya ART aged different phone Patients received SMS
≥ 18 years. Mean reminder systems vs reminders that were
age (years): Control: control group. either short or long
35.65; Intervention: Interventions groups: and sent on a daily or
35.64–37.74; Male: G1: short SMS + weekly basis. The
Control, 44%; daily content of the short
Intervention, G2: long SMS + reminder was ‘This is
59–69% daily your reminder’, the
G3: short SMS + content of the long
weekly message was ‘This is
G4: long SMS + your reminder. Be
weekly strong and courageous.
We care about you’

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Int J Clin Pract

Mbuagbaw 2012 Parallel group, 228 adults aged ≥ 21, HIV-infected adults on Evaluate the 6 months SMS was sent to each Usual care VAS No
(10) RCT; Country: ART-experienced ART effectiveness of text participant in the Self-report

ª 2014 John Wiley & Sons Ltd

Cameroon (at least 1 month on messaging as a intervention group Pharmacy refill data
treatment). Mean reminder for once a week.
age: Control, improving adherence Messages were based
39.0  10.0; to ART. on data collected from
Intervention, focus group discussions
41.3  10.1. Males: and the health belief
Control, 21.2%; model of behaviour
Intervention, 31.7% change. The content of
the message was
motivational, with a
reminder component:
‘You are important to
your family. Please
remember to take your
medication. You can
call us at this number.
The message also
contained a phone
number that they could
call back if they
needed help.
Lewis 2012 (12) One group 52 homosexual adults HIV infection Assess whether text 3 months Daily text messages Adherence rate during Self-reporting through Yes
pre–post study; aged ≥ 25 years. messages enhance were sent to remind the 7 days before the weekly adherence
Country: USA Mean age: 38 (25– adherence and patients to take study. surveys.
60); Male: 100% clinical outcomes. medication.
Motivational messages
were tailored to
adherence rates:
adherent patients
received them on
weekly basis e.g. ‘He
mHealth and medication adherence

shoots! He scores!
Perfect med adherence.
Great job!’ while non-
adherent participants
were sent on a daily
basis: e.g. ‘Stop, drop
and pop. Take your
meds now’
5
 6

Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Rodrigues 2012 Quasi- 150 adults aged HIV infection Change in adherence 12 months Adherence rate over 6 months after Pill counts Yes
(11) experimental ≥ 18 years. Mean over a 12-month 6 months with SMS intervention withdrawn
cohort study; age: 38.54  7.7; period (6 months and IVR call reminders
Country: India Male: 73%. during the and 6 months without
intervention and reminders. All
6 months after participants received
discontinuation of two types of adherence
the intervention). reminder on their
mobile phones: (i) an
automated IVR
(‘Did you take all your
medicines yesterday?’)
mHealth and medication adherence

and (ii) a non-

interactive neutral
picture delivered as an
SMS once a week for
6 months. An example
of the IVR was: ‘Did
you take all your
medicines yesterday?’
Patients responded
with a ‘1’ if they had
not missed any doses
in the previous 24 h
and ‘2’ if they had.
The SMS was in the
form of a line diagram
of a lamp.
Other chronic diseases
Asthma
Strandbygaard Parallel group, 26 patients aged 18– Asthma based on Evaluate the impact of 12 weeks Intervention subjects Usual care Record of doses left on Yes
2009 (13) RCT; Country: 45 years; Mean age clinical history and receiving daily SMS received the following the disc of Seretide.
Denmark (years): Intervention, daily symptoms reminders on the daily SMS reminder to
34.4; Control, 30.7; rate of adherence to take their medication
Male: Intervention, asthma treatment. “Remember to take
50.0%; Control, your asthma
57.0% medication morning
and evening. From the
respiratory unit”

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 Table 1 Continued

Study design, Method used to Improvement

Int J Clin Pract

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Petrie 2012 (14) Parallel group, 147 adults (control, 73; Asthma patients not Whether text 18-week All participants fill in a Usual care Self-reported adherence Yes

ª 2014 John Wiley & Sons Ltd

RCT; Country: intervention, 74) currently adhering to messages would intervention, questionnaire to assess by phone calls at 6
UK diagnosed with their preventive modify patient’s 9-month their perception of and 12 weeks, as well
asthma. Mean age medication as illness and follow-up their illness [Brief as at 6 and 9 months.
(years): ND (aged prescribed. medication beliefs as period Illness Perception
between 16–45 well as improved Questionnaire (BIPQ)].
years) Males: ND adherence to their Messages were sent at
preventive inhaler at a tailored frequency (2
follow-ups over a 9- SMS per day during
month period. weeks
1–6, 1 SMS per day
from weeks 7 to 12,
and 3 SMS per week
from week 13 to 18).
Depending on the
score in the BIPQ,
patients were sent
different text messages
to modify that belief
(for example, belief in
little need for
medication).
Chronic diseases
Foreman 2012 Retrospective 580 adults (290 Chronically ill patients To compare adherence 8 months Adherence to chronic Usual care (matched 1:1 PDC Yes
(23) control cohort intervention group, on treatment with to chronic oral oral medication after to patients in the
study; Country: 290 control group). oral antidiabetes medication before- implementing ‘My control population with
USA Mean age (years): drugs, b-blockers, after the medication reminder the intervention group)
Control, calcium channel implementation of text messaging
64.7  13.7; blockers, an SMS reminder programme, which
Intervention, angiotensin- programme and sends daily medication
64.8  11.9; Male: converting enzyme compare to a control reminders with the
Control, 53.8%; inhibitors, group. context ‘Take your
mHealth and medication adherence

Intervention, 46.6%. angiotensin receptor medication today’.

blockers, diuretics,
platelet aggregation
inhibitors, statins,
antidepressants,
bisphosphonates,
and thyroid agents.
7
 8

Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Coronary heart disease

Park 2014 (15) Parallel groups, 90 adults > 21 years Patients with a history Assess whether text 1 month Patients were Usual care (1:1:1) Morisky Medication Yes
RCT; Country: old (30 patients per of myocardial messaging improved randomised into three adherence survey
USA group); Mean age infarction and/or adherence to groups: (i) patients MEMS
(years): 59.2  9.4; percutaneous antiplatelet and who received text Mobile phone
Male: 78% coronary intervention statin medications. messages for (CareSpeak)
on antiplatelet or Explore whether medication reminders
statin treatment patient education and health education;
delivered three times (ii) patients who
a week would received text messages
improve medication for health education;
adherence compared (iii) patients who did
mHealth and medication adherence

with twice-daily not receive text

medication reminders messages.
Medication text
messages were sent
twice daily (antiplatelet
reminder in the
morning and statin
reminder in the
evening), while health
education messages
were sent three times
per week. An example
of a medication
reminder is ‘John, take
Plavix 75 mg at 9:00
AM. Respond with 1’
and an example of a
health education
message was
‘Remember to see your
cardiologist and/or
primary physician 1–
2 weeks after your
hospitalisation’. Text
messages were 2-way,
requiring patients to
respond.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
Int J Clin Pract
Table 1 Continued

ª 2014 John Wiley & Sons Ltd

Study design, Method used to Improvement
Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Diabetes mellitus
Dick 2011 (16) One group pre– 18 adults aged Diabetes I or II taking Feasibility of text 4 weeks Daily text messages to No control Self-reported adherence Yes
post; Country: ≥ 18 years, urban medication (oral message to enhance remind patients to take at baseline, during the
USA African Americans. antidiabetic drugs or medication medication, perform study period and
Mean age (years): 55 insulin) adherence, health foot care, and monitor 1 month after the
(range 38–72); Male: behaviours and blood sugar. Each intervention.
33% appointments participant was
attended required to receive a
daily medication or
blood sugar reminder
(e.g. ‘Did you take
your diabetes
medication today?’)
and a weekly question
about foot care (e.g.
‘How many times did
you check your feet
this week?’).
Hypertension
Marquez- Parallel group 104 patients Hypertension in Evaluate whether 6 months SMS were sent twice a Usual care Pill counts No
Contreras 2004 RCT; Country: > 18 years; Mean patients with intervention based week except
(17) Spain aged (years): uncontrolled blood on mobile telephone weekends. Messages
Intervention, pressure on text messaging to gave information about
56.26  10.22; monotherapy or send alerts and hypertension,
Control, hypertension in reminders improved adherence, suggested
59.43  10.94; patients eligible to adherence to hygienic dietary habits,
Male: Intervention, start treatment with antihypertensive and reminded patients
52.9%; Control, combined drugs. to take their
mHealth and medication adherence

57.6% antihypertension medication.

drugs.
9
 10

Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Patel 2013 (18) 3-phase 50 adults aged ≥ 18– Essential hypertension Assess adherence to 10 months Cell-phone medication Prephase of 3 months Pharmacy refill rates Yes
sequential study, 80, with on at least two antihypertensive reminder system vs. prior to automated Morisky self-reported
open label; hypertension; Mean antihypertensive medication during prephase medication medication reminders medication scale
mHealth and medication adherence

Country: USA age (years): 53 (range medications the previous reminder system and and postactivation of
33–78); Male: 31% 3 months of the postphase medication 3 months after
intervention vs. reminder system withdrawal of the
during activation of reminder system.
an automated
medication reminder
application for
mobile phones;
Assess the
continuation of
medication
adherence over
3 months after
withdrawal of the
Pill Phone; Evaluate
patient usage
patterns and
acceptance.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 Table 1 Continued

Int J Clin Pract

Study design, Method used to Improvement
Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Infectious diseases

ª 2014 John Wiley & Sons Ltd

Suffoletto 2012 Parallel control 200 adults aged Patients on oral Assess whether 4 months Participants in the Usual care Self-reported No
(19) group, RCT; ≥ 18 years (n = 100 antibiotics automated SMS intervention group questionnaires
Country: USA in the control and discharged from the system improves received SMS to
intervention groups); emergency adherence to remind them to pick
Mean age (years): department. antibiotic up their prescription
Control, 31  11; prescriptions. and to ask them how
Intervention, they were taking their
34  13. Male (%): antibiotics. Participants
Control, 34%; were requested to text
Intervention, 29%. back. Participants in
both groups received a
phone call on the day
following the intended
completion of their
antibiotic. Follow-up
assessment included
questions like ‘How
many pills do you have
in your bottle?’ ‘How
many days did it take
you to pick up your
prescription?’
Iribarren 2013 Parallel control 37 adults aged Patients initiating first Assess whether 2 months Participants were Usual care Control: patient diaries No
(20) group, RCT; ≥ 18 years (n = 19 antituberculosis sending text instructed to text in Intervention: self-
Country: control group and treatment messages promotes after self- reported through SMS
Argentina n = 18 SMS group); adherence to administration of
Mean age (years): antituberculosis medication. When a
Control, 35.05  17; treatment. patient did not text in,
Intervention, a reminder was sent.
33.78  15. Male Educational SMS
mHealth and medication adherence

(%): Control, 42%; messages were sent

Intervention, 44%. biweekly and the
option to consult
during the 2 months of
the intensive treatment
phase provided
11
 12

Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Psoriasis
Balato 2012 Parallel control 40 adults aged 18– Psoriasis Assess whether 12 weeks Patients received 1 SMS Usual care Self-report by multiple- Yes
(21) group RCT; 65 years. Mean age sending text daily for 12 weeks in choice questionnaire
Country: Italy (years): Control, 39.3 messages improves the same randomly and patient diaries
(10.2); Intervention, treatment adherence selected order
38.4 (9.5). Males and patient (medication reminders
(%): Control, 50%; outcomes (quality of three times weekly,
Intervention, 60% life, disease severity, educational tools four
patient-perceived times weekly). For
disease severity, and example, ‘Take the
the patient-physician medications strictly
relationship). according to the
mHealth and medication adherence

directions of your
doctor’ or ‘Please
remember to use the
medication or product
you use to treat your
psoriasis today’
Transplant recipients
Miloh 2009 (22) One group pre– 41 paediatric and adult Paediatric liver Use text messages to 13 months Daily text messages sent Standard deviation of Standard deviation of Yes
post; Country: patients. Mean age transplant recipients improve adherence to remind patients to tacrolimus blood levels tacrolimus blood levels.
USA (years): 15 (range, 1– to take their in the previous year.
27); Males (%): 34% immunosuppressive immunosuppressors.
medication for The messages were
paediatric patients structured as ‘Take
undergoing [name of medication]
orthotopic liver at [set time]. To
transplant. confirm intake, press
REPLY, type CARE 1,
and press SEND’. If the
patient did not respond
to the message within
a time frame set
individually (15 min to
1 h), then a message
was sent to the
caregiver indicating
that the patient did
not respond.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
Int J Clin Pract
Table 1 Continued

ª 2014 John Wiley & Sons Ltd

Study design, Method used to Improvement
Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Healthy population
Travellers
Vilella 2004 Non-randomised 4043 travellers aged Travellers who Compare adherence 4 months Travellers received the Usual care Record of vaccine Yes
(24) parallel group ≥ 18 years. Mean attended the clinic rates to vaccination SMS a few days before administration
trial; Country: age (years): Control for vaccination schedules between the date of the next
Spain 2001, 28.7  5.2; against hepatitis travellers that vaccine dose, ‘This is
control 2000, A+B (schedule: 0, 1, received phone to remind you that you
27.1  5.0; 6 months) and reminders vs. control should go to the
Intervention, hepatitis A group. vaccination centre to
28.6  4.2; Male (Schedule: 0 and receive your hepatitis
(%): Control 2001, 6 months) vaccine dose. Thank
48.6%; control 2000, you’
48.3%; Intervention,
55.6%
Ollivier 2009 Parallel group 424 soldiers returning French soldiers To assess the 28 days after A commercial SMS Usual care MEMs No
(25) RCT); Country: from duty in a returning after a 4- feasibility and return messaging service was
France malaria-endemic area. month deployment in acceptability of used to send a
Mean age (years): a malaria-endemic sending a daily SMS standardised
control, 26.7; area. Malaria reminder message automated daily
intervention, 26.4. prophylaxis was via mobile phone message at midday
Male (%): 96.1 100 mg doxycycline device to remind reminding participants
overall monohydrate once a soldiers to take their to take their malaria
day. malaria medication ‘Remember
chemoprophylaxis to take your
and to assess the doxycycline pill at
impact of SMS on midday. In case of
adherence to fever, consult a
chemoprophylaxis. physician and tell him
mHealth and medication adherence

you have recently

returned from C^ote
d’Ivoire’.
13
 14

Table 1 Continued

Study design, Method used to Improvement

Study (Ref) country Participants Disease Aims Duration Intervention Comparator measure adherence in adherence?

Vitamin C
Cocosila 2009 Parallel group 102 adults aged Healthy people on Determine whether 1 month Participants were sent a Usual care Self-reported adherence No
(26) RCT; Country: ≥ 18 years. Mean vitamin C treatment SMS reminders daily SMS reminder to
Canada age (years): Control, increase adherence take their vitamin C
34.3  14.2; to vitamin C in tablet, e.g. ‘Hi, it’s
Intervention: healthy people. Tim: Any vitamin C
32.9  13.4. Males 2 day?’). After each
(%): Control, 29%; such message,
Intervention, 31%. participants were
requested to reply with
mHealth and medication adherence

a one-letter SMS, ‘A’

(i.e. ‘acknowledge’),
after taking the
vitamin but no later
than the midnight of
the same day (i.e. they
had at least 5 to 6 h
to reply). If participants
responded they
received a subsequent
SMS reinforcing the
behaviour (e.g. ‘Tim
here again: Ur doing
super! Tip: to steal
ideas from one person
is plagiarism; to steal
from many is
research!’); if they did
not reply, they received
a correcting SMS
reminding them of the
benefits of vitamin C
(e.g. ‘Again Tim: Do
your best to take the
vitamins: they help to
fight cold and flu!’).

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 mHealth and medication adherence 15

to ask questions. The frequency of message delivery

ART, antiretroviral therapy; MEMS, Medication Event Monitoring System; VAS, visual analogue scale; IVR, interactive voice recognition; ND, no data; PDC, proportion of days during the measurement period that are
in adherence?
ranged from daily (n = 1) (12) to once a week
Improvement
(n = 3) (8,10,11), or included both frequencies
(n = 1) (9).

No
In four of the five studies, adherence rates improved
(Table 2). In one study, the improvement was main-
measure adherence

tained during the 6 months after the intervention was

Method used to

withdrawn (11). Higher adherence rates were

Pill diaries

recorded, when SMS were sent on a weekly basis (9).

Viral load decreased and CD4 count increased sig-
MEMs

nificantly in the two studies where they were evalu-

ated (8,12). However, Mbuagbaw et al. (10) did not
observe statistically significant differences in mortal-
ity rate, quality of life, weight or body mass index,
Comparator

although the regression analysis revealed that higher

Usual care

levels of education and being on a second-line

regimen were statistically significant predictors of
adherence > 95%.
remember to take your
adherence ‘Please
reminders on oral
Daily text message

contraceptive pill

birth control pill’

Studies on other chronic diseases

Intervention

The chronic diseases analysed were asthma (n = 2)

(13,14), coronary heart disease (n = 1) (15), diabetes
mellitus (n = 1) (16), hypertension (n = 2) (17,18),
infectious diseases (n = 2) (19,20) and psoriasis
(n = 1) (21). Patients who had undergone a liver
transplant were also analysed (n = 1) (22). One
Duration

3 months

study included patients with various chronic diseases

(n = 1) (23). A total of 1333 patients were examined,
with an age range of 1–78 years. The characteristics
reminders promotes
Assess whether SMS

contraceptive pills

of the studies are summarised in Table 1. Duration

adherence to

varied from 1 to 13 months.

All studies reported using SMS, whose content
Aims

comprised medication reminders (n = 5)

(13,18,19,22,23), medication and healthy lifestyle
reminders (n = 4) (16,17,20,21), and statements tar-
New user of oral

geting perceptions of illness and beliefs about medi-

contraceptives

cations (n = 1)(14). In the study by Park et al.,

Disease

patients were randomised to receive medication

reminders and health education messages, only
health education messages, or no messages (15). The
82 women; Mean age

18–31); Males (%):

(years): 22 (ranged

frequency of message delivery ranged from daily

(13,16,18,22,23) to twice weekly (17), or varied
Participants

depending on the study period (14,19,21) or the ran-

domization arm (15).
0%

Three studies used two methods for measuring

adherence: refills and the Morisky questionnaire
RCT, Country:
Study design,

(18); self-reported adherence by multiple-choice

Parallel group

covered by prescription claims

questionnaire and patient diaries (21); and the Mori-

country

USA

sky questionnaire and the MEMS (15).

Table 1 Continued

Adherence improved in all the studies except three

Contraceptive pills

(17,19,20) (Table 2). In the study by Marquez et al.

Hou 2010 (27)

(17), sending SMS did not improve adherence to

Study (Ref)

antihypertensive treatment in the first, third and

sixth month of the study when compared with the
control group.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 16
Table 2 Outcome of adherence in the studies included

Study Adherence Outcome Intervention Control Effect estimate Outcome

HIV
Lester 2010 (8) Percentage of patients with adherence 62% 50% RR (95% CI): 0.81 Sending short messages makes good
rates > 95% (ITT) (0.69–0.94) adherence and viral suppression more
likely.
Pop Eleches 2010 (9) Percentage of patients with > 90% of 47% SMS reminders improve adherence and
adherence. decrease treatment interruptions
Intervention groups:
G1: short SMS+daily G1: 49%
G2: long SMS+daily G2: 50%
G3: short SMS+weekly G3: 68%
G4: long SMS+weekly G4: 59%

Mbuagbaw 2012 (10) Percentage of patients with a VAS 66.7% 71.3% RR (95% CI): 1.06 Adherence rate to ART did not improve
(0.89–1.29)
mHealth and medication adherence

> 95% adherence among treatment-na€ıve patients after

Self-report (no missed doses) (%) 79.2% 79.0% RR (95% CI): 1.01 6 months of using text messages.
(0.87–1.19)
Pharmacy refill data (mean, SD) 3.8 (1.48) 3.7 (1.34) Mean difference: 0.1
( 0.23–0.43)
Lewis 2012 (12) Percentage of patients adherent to ART 49% – – Text messages seem to improve
throughout the study adherence but the sample size was
small.
Rodrigues 2012 (11) Percentage of patients with adherence 85% vs. 91% vs. 94% – – Improvement in adherence rate after
rate ≥ 95% (preintervention phase vs. 6 months of intervention. The effect
end of intervention phase vs. persists for at least 6 months after
6 months postintervention) the intervention finishes.
Other chronic diseases
Asthma
Strandbygaard 2009 (13) Percentage of doses taken by the 77.9% vs. 81.5% 84.2% vs. 70.1% Δ 17.8% (95% Use of SMS reminders increases
patient: baseline vs. final CI, 3.2–32.3%) adherence
Petrie 2012 (14) Average of self-reported adherence: 56.5  35.3% vs. 54  31.8% vs. Relative average increase of Intervention improved adherence to
baseline vs. final 57.8  27.1% 43.2  26% 10% asthma medication
Proportion of participants achieving 15 of 58 patients (25.9%) 7 patients of 66 (10.6%) Difference of 15.3%
optimal asthma control of ≥80%.
Chronic diseases
Foreman 2012 (23) PDC 85  20.0% 77  28% p < 0.001 Patients opting to receive medication
reminders adhered to their medication
better than patients not opting to
receive them.
Coronary heart disease
Park 2014 (15) MEMS (Group 1 vs. Group 2 vs. G1: 93.7  11.9 G3: 79.1  27.7 p = 0.03 Text messages increased adherence to
Group 3): G2: 95.8  9.5 G3: 83.3  21.3 p = 0.28 antiplatelet therapy demonstrated by
MEMS and text message responses,

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 Table 2 Continued

Study Adherence Outcome Intervention Control Effect estimate Outcome

Int J Clin Pract

Doses of antiplatelet taken (%) G1: 92.4  14.0 but did not increase adherence to
Doses of statin taken (%) G2: 90.1  16.2 statins.

ª 2014 John Wiley & Sons Ltd

Self-report (Group 1 vs. Group 2 vs. G1: Baseline 6.20  1.66; G3: Baseline 6.01  1.84; p > 0.05
Group 3) Follow-up at 30 days: Follow-up at 30 days:
6.43  1.22 6.96  1.44
G2: Baseline 5.85  2.10;
Follow-up at 30 days:
6.73  1.49
SMS response rate antiplatelet vs. 90.2  9 vs. 83.4  15.8 – p = 0.005
statin
Diabetes mellitus
Dick 2011 (16) Number of doses missed in the last 1.9 vs. 0.6 vs. 0.8 – p = 0.003 Text message seems feasible for
week (doses/week): prestudy period improving self-management of
vs. study period vs. 1 month after the diabetes.
study
Hypertension
Marquez-Contreras 2004 Percentage of participants with 1st month: 92.1% 1st month: 85.7% Non significant Sending short messages did not
(17) adherence rate > 80% 3rd month: 77.3% 3rd month: 88.9% RR (95% CI): 8.9 improve medication adherence.
6th month: 89.5% 6th month: 78.9% (0.18–17.62)
Patel 2013 (18) PDC (%): preintervention vs. 57  27% vs. 58  20% – – Mobile phone-based automated
intervention phase vs. postintervention vs. 56  31% medication reminder system shows
Morisky Medication Scale (Range, 0–4), Beginning: 2.4 – – promise in improving adherence.
mean: End: 3.2
Pill Phone Application (PPA): 60% – –
Percentage of doses taken
Infectious diseases
Suffoletto 2012 (19) Percentage of participants who filled 78% (95% CI, 66–87%) 69% (95% CI, 57–80%) p = 0.26 Adherence rate in the intervention
their prescription within 24 h of group did not improve.
discharge from the ED
Percentage of patients adherent to 57% (95% CI, 44–67%) 45% (95% CI, 37–59%) p = 0.1
antibiotic prescription (self-reported
adherence)
Percentage of patients with no pills left 68% (95% CI, 55–78%) 59% (95% CI, 47–71%) p = 0.3
mHealth and medication adherence

on the day after intended completion

of prescriptions
Iribarren 2013 (20) Percentage of patients who reported 77% (22–100%) 100%. RR (95% CI): 1.49 Greater adherence was reported in the
having taken their medication However, only 53% of (0.92–2.42) intervention group, although the
participants returned the difference with the SMS intervention
calendars did not prove to be statistically
significant.
17
 18

Table 2 Continued

Study Adherence Outcome Intervention Control Effect estimate Outcome

Psoriasis
Balato 2012 (21) Number of days/week which patients Pre: 3.86 Pre: 4.2 – The adherence rate improved in the
adhere to medication Post: 6.46 Post: 4.0 intervention group at the end of the
study.
Transplant recipients
Miloh 2009 (22) Tacrolimus level standard deviation 3.46  2.17 lg/l 1.37  1.01 lg/l – Adherence improved and the number of
rejection episodes fell
Healthy population
mHealth and medication adherence

Travellers
Vilella 2004 (24) Percentage of vaccinations administered Control 1 vs. Control 2: p < 0.05 Use of SMS reminders improves
Hepatitis A+B 2nd dose (HAB2) HAB2: 88.4% HAB2: 77.2% vs. 80.7% adherence to the vaccination
Hepatitis A+B 3rd dose (HAB3) HAB3: 47.1% HAB3: 23.6% vs. 26.9% schedule, especially in the third dose
Hepatitis A 2nd dose (HA2) HA2: 27.7% HA2: 13.2% vs. 16.4% of hepatitis A+B.
Ollivier 2009 (25) % of daily adherence 22.3% 21.4% No significant effect No difference in adherence was
% of early treatment discontinuation 42.6% 50.3% 0.85 (0.67–1.07) observed between the groups.
Rate of adherence at day 2 and 28 94.6% vs. 67.6% 95.2% vs. 65.8%
(day 2 vs. day 28)
Vitamin C
Cocosilla 2009 (26) Percentage of participants reporting From 1.3 to 4.5 (246%) From 1.6 to 3.7 (131%) RR (95% CI): 1.38 (1.11– Inconclusive results.
increased adherence 1.71)
Number of doses missed in the last 2.5 3.3 RR (95% CI): 0.8 ( 1.55
7 days to 0.05)
Oral contraceptives
Hou 2010 (27) Mean pills missed MEMs 4.9  3.0 4.6  3.5 p > 0.5 Daily text message reminders did not
Mean pills missed recorded in patient’s 1.3  2.0 1.1  1.4 p > 0.5 improve adherence to the oral
diary contraceptive pill.

ED, emergency department; PDC, proportion of days covered; ITT, intention to treat.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 mHealth and medication adherence 19

Table 3 Summary of the risk of bias (based on the Cochrane Collaboration tool)

Selective Other
Sequence Allocation Incomplete outcome sources
Trial generation concealment Blinding outcome data reporting bias of bias

HIV studies
Lester 2010 (8) L L L L L L
Pop Eleches 2011 (9) L U L L U U
Mbuagbaw 2012 (10) L L L U L L
Rodrigues 2012 (11) H H H L L U
Other chronic diseases
Strandbygaard 2010 (13) L L H U H L
Petrie 2012 (14) L L H L L L
Marquez-Contreras 2004 (17) L U U H L L
Suffoletto 2012 (19) L L H L L U
Balato 2012 (21) L U L L L U
Iribarren 2013 (20) L L U L L U
Park 2014 (15) L L U L L U
Healthy population
Vilella 2004 (24) H H U L L L
Ollivier 2009 (25) L U U L U L
Cocosila 2009 (26) L L U L L H
Hou 2010 (27) L L L L L U

L, low risk; H, high risk; U, unclear risk

Random sequence generation (selection bias)

Allocation concealment (selection bias)

Blinding (performance bias)

Incomplete outcome data (attrittion bias)

Selective reporting (reporting bias)

Other bias
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Yes (low risk of bias) Unclear No (high risk of bias)

Figure 2 Adherence to medication using mobile device–based interventions according the Cochrane risk of bias tool

This type of intervention did not improve the conversion of a culture from positive to negative)
adherence rate in patients on antibiotic treatment (20). The results were clinically significant only in
after their discharge from the emergency department the first of these studies (47% patients at baseline vs.
(19) or in patients on antituberculosis treatment 64% while using the application vs. 60% after with-
(20). Clinical outcomes were evaluated in Patel et al. drawal of the intervention) (18).
(% of patients with controlled blood pressure) (18),
Marquez Contreras et al. (differences in systolic and Studies in the healthy population
diastolic blood pressure and in body weight) (17), A total of 4651 people were included in the four stud-
Strandbygaard et al. (exhaled nitric oxide levels, lung ies, in each of which SMS were sent as a reminder to
function, and airway responsiveness) (13), and Iri- enhance vaccination rates (24), malaria chemoprophy-
barren et al. (number of patients whose treatment laxis among soldiers returning from malaria-endemic
was successful, patients with a clear sputum smear or areas (25), vitamin C intake (26) and use of oral

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
20 mHealth and medication adherence

contraceptives (27). Medication reminders where sent Our findings suggest that mHealth enhances adher-
daily (25–27), whereas vaccination reminders were a ence to medication, although it was impossible to make
few days before the appointment (24). comparisons, owing to differences in study design,
The vaccination rate increased, especially for the intervention, comparator, treatment regimen, duration
third dose of hepatitis A + hepatitis B (24). However, and measurement method. All interventions were
in the remaining cases, adherence was not promoted based on sending SMS either strictly as a medication
using this type of intervention (25–27) (Table 2). No reminder or together with motivational content. Thus,
pregnancies were reported in the study on oral con- a striking result at the time of this review was that
traceptives (27). although 680 (28) mobile phone applications to
enhance adherence are commercially available, the lack
Assessment of internal validity of published results from RCT mean that the feasibility
The results of the assessment of the study quality are and acceptability of this technology cannot be vali-
reported in Table 3, and the Cochrane risk of bias dated, although most studies (83.3%) were conducted
summary is reported in Figure 2. in high-income countries, where smartphone applica-
tions are widely used. Results have been reported by
Patient satisfaction application developers such us MediSafe Project, whose
Patient satisfaction was evaluated in 10 studies and mobile pillbox application improved adherence to an-
was found to be very useful by most authors tidiabetic drugs by more than 26% (29). It is important
(8,10,12,16,18,19,21,25,27). Park et al. (15) compared to remember that smartphone applications evolve
sending medication reminders + educational text mes- much faster than other interventions, and while appli-
sages with education text messages alone and found cations can be validated over a specific period, more
that a higher percentage of patients strongly agreed features could be added to the initial version, thus mak-
that medication reminders + educational text messages ing it impossible to determine the long-term impact of
helped them to take their medication (57% vs. 32%, the baseline version. As for the target population, we
respectively). In five studies (8,10,16,21,27), participants found that HIV infection was the most widely studied
were asked if they would recommend the intervention to a disease in developing countries, whereas in high-
friend, and 89.04% said yes (SD = 10.76). In addition, income countries, other chronic diseases such as hyper-
80.4% (SD = 13.9) of participants expressed their tension and diabetes were more commonly evaluated,
willingness to continue using the SMS reminders and interventions targeting the healthy population
(8,10,16,18,19,21,25,27). In two studies, 15% (21) and were more frequent. This observation highlights the
57% (25) of participants said they would be willing to pay differences in reasons for implementing mHealth
a small fee for this service. In the malaria prophylaxis trial, between countries: the main reason in developing
58.2% of the SMS group agreed that it would be very use- countries is equity of healthcare delivery; in developed
ful to extend the use of SMS reminders to all soldiers countries, it is optimisation of resources.
returning from malaria-endemic areas. As for measurement of adherence, only 17.6% of the
studies used more than one method. This percentage is
Mobile phone extremely low, considering that more than one method
Only two studies reported having supplied the is widely recommended owing to the inaccuracy of
mobile phone to participants (9,18), and one study current approaches. No reference method for measur-
did not provide data on the supply of a mobile ing adherence has been established (1,2,30), and the
phone (13). In three studies, patients were paid for available methods are either direct (directly observed
their participation (9,15,19). treatment, plasma drug or metabolite levels, biomar-
kers) or indirect (pill count, self-report, medication
refills, electronic monitoring devices). Each presents
Discussion
advantages and disadvantages: direct methods are
We identified 20 articles assessing interventions based expensive but more objective, whereas indirect meth-
on mobile devices used to enhance adherence. To our ods are easy to implement but not objective, tend to
knowledge, this is the first review to evaluate whether overestimate adherence, and are only useful for specific
mHealth interventions enhance adherence to medica- drugs and populations. Therefore, the definition of
tion by taking into account practical issues (such as adherence and the rate of adherence presented in this
whether mobile phones are supplied), which may limit review are not uniform across the studies analysed;
subsequent scaling up of this type of intervention. We similarly, the methodology used to assess adherence
were unable to perform a meta-analysis, owing to the also varies widely. Most methods were indirect (82.3%
significant clinical and methodological heterogeneity of cases), thus potentially leading to an overestimation
in the studies included. of adherence levels, although the effect would have

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 mHealth and medication adherence 21

been the same in the control and intervention groups. ian observed adherence of 93.9%) (43). In addition,
The results reported by Mbuagbaw et al. (31) did not whilst many studies analyse the impact of technology
reveal a significant improvement in adherence. This is on the healthcare resources used, few look at how it
an interesting finding, since the study was the only one changes contacts with primary care professionals.
performed in HIV-infected patients that used more Most studies did not include detailed descriptions
than one method to measure adherence. The authors of their technical implementation processes and mes-
found that higher levels of education and being on sec- sage delivery. This gap is important, because the miss-
ond-line treatment were related to higher adherence ing information could prove useful for the design of
rates. A correlation has also been observed between future studies and strengthen the evidence base of this
poor adherence and low educational level and treat- emerging research area. Important considerations that
ment regimen (32–35). In low income countries, these should be taken into account when implementing an
factors may be more relevant than in high income interactive system are fatigue messages, logistics of
countries for several reasons such as: lower educational mobile phones (e.g. notifications to determine whether
levels are more frequent in the population; access to a certain number is receiving messages and is opera-
medication is difficult and drug stock-outs are fre- tional, ability to detect changes in numbers), receiving
quent. (36). However, most studies agree on the diffi- messages gradually (independently of whether cover-
culty in defining the sociodemographic characteristics age is sufficient), adaptable messages, logic texts that
underlying these correlations (37–39). A recent meta- allow for different types of response (e.g. yes, Yes, YES,
analysis on mobile phone text messages for improving Y), and avoidance of excess intrusiveness in the
antiretroviral therapy showed that use of SMS has a patient’s life (44). An essential issue in mhealth inter-
significant effect on adherence to antiretroviral therapy ventions is data security and confidentiality. American
and that this effect is influenced by educational level, and European agencies have started to regulate some
gender, timing (weekly vs. daily) and interactivity (36). types of mobile applications, especially those that act
Although sending SMS seems to enhance adherence, as medical devices, but not those aimed at tracking adher-
the fact that adherence rates significantly improve in ence (45,46). However, if a mhealth intervention collects,
65% of the studies, is probably because non-adherence stores and/or transmits information that constitutes Pro-
is a complex behaviour with several triggers, such as: tected Health Information, it must do so following the
presence of psychological problems, in particular Health Insurance Portability and Accountability Act (HI-
depression; lack of primary support; occurrence of PAA) (47), as well as any other applicable laws or regula-
medication side effects; the treatment is focused on tion of the country concerned. Therefore, data encryption
treating an asymptomatic disease (1). Patient-centred and the use of secure networks are crucial.
care is currently considered the gold standard, as it pro- Rodrigues et al. (11) evaluated whether patients
vides sufficient information for the patient to take part preferred SMS or phone calls and found that of the
in the decision to start treatment and the design of the 136 respondents, 34% preferred phone calls only, 11%
strategy to be followed. Therefore, a successful strategy preferred SMS only, 44% preferred both and the
must take into account patients’ needs, characteristics, remaining participants preferred neither. The fre-
and opinion after appropriate information has been quency of text messages may also vary between adher-
provided. Recent surveys indicate that patients are will- ence levels, with messages being more frequent when
ing to become more actively involved in managing their adherence is low (daily reminders) or more separated
own care (40) and that self-monitoring at home is one in patients with good adherence (weekly reminders).
way to increase their involvement (41). However, In another study, 93% of the participants responded
Mbuagbaw et al. (10) found that, although participants that they read all the SMS (12). However, only 20% of
could interact via mobile phone, only 47.5% in the participants indicated that SMS were sent at the right
intervention arm used the feedback option. Therefore, time, whereas 12% stated they were never at the right
mHealth interventions should not be considered a sin- time (12). Other options that resulted in increased
gle intervention, since strategies to promote adherence adherence are devices that attach to the standard pill
should be tailored to patients’ specific needs, and more bottle or blister pack and send an SMS to a web service
than one intervention could be required. Nevertheless, every time the patient opens the bottle (48) or smart
mHealth can motivate patients and healthy individuals pillboxes that emit a loud beep (or other customised
to participate in their own care. In a study conducted to sound) if the compartment is not opened within
evaluate the feasibility of mobile direct observation 30 min of the scheduled time (49) and afterwards trig-
treatment (MDOT), the authors showed high accept- gers an automated phone call or SMS. Nevertheless,
ability compared with clinical DOT or home visits (42). mobile phone applications have yet to be evaluated
MDOT also improved adherence to hydroxyurea in alone and in comparison with other devices such as
paediatric patients with sickle-cell disease (overall med- pillboxes.

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
22 mHealth and medication adherence

Thus, a major challenge for mHealth is to ensure incentives to keep patients sufficiently motivated to
that preventive measures reach the healthy population. participate in the study. Implementation and updating
Non-adherence is more frequent in preventive treat- of mHealth strategies consume resources, and there is
ments because the target individuals are healthy (24). no evidence of a reduction in cost. Foreman et al.
This observation correlates with our results, as only (23), studied pharmacy medication costs after the start
one of the three studies resulted in increased vaccina- of the text messaging programme, although the differ-
tion rates (24). However, Ollivier et al. (25) reported ences detected were not statistically significant. Patel
that the absence of a difference in adherence between et al. (18) found no statistically significant differences
the intervention and control groups was probably in the use of resources, although fewer hospital admis-
because the population under study consisted of sol- sions were detected in patients using the mobile appli-
diers who stayed together after returning from a cation. In a study performed by Henderson et al. (50)
malaria-endemic area. Adherence to malaria prophy- in England including patients with chronic obstructive
laxis has been reported to be inadequate, mainly pulmonary disease (COPD), diabetes mellitus, and
because of adverse events and forgetfulness (50,51). As heart failure, the intervention did not prove to be cost
stated above, multifaceted interventions featuring effective based on health and social care costs and out-
interactive education and training help to promote comes over 12 months and the willingness-to-pay
adherence in a variety of diseases. Another group of threshold of £30,000 per QALY recommended by the
healthy end-users to consider is caregivers, since older National Institute for Health and Care Excellence.
age remains a major barrier to mHealth. This technol- However, patients were supplied with pulse oximeters
ogy has been less accepted by older people, as shown (patients with COPD), blood glucose monitors
in the study by the Pew Research Centre, where 81% (patients with diabetes), and weighing scales (patients
of adults aged 25–34 years had smartphones (the with heart failure), and almost all participants in the
group with the highest penetration of smartphones in intervention received blood pressure monitors.
the USA) compared with 50% of adults older than Assuming that equipment costs decrease and working
55 years (3). These data correspond with the mean age capacity increases, the probability that telehealth is
of participants recruited in the studies, which was cost effective will be about 61% based on a threshold
between 30 and 50 years. However, since this gap will of £30,000 per QALY (53). In contrast, in the studies
disappear with time, it is important to draw the atten- analysed in the present review, no equipment was
tion of caregivers to older individuals, who may benefit given to patients except the mobile phone in two cases.
from the technology. These results are consistent with those published in
Another important aspect of mHealth is feasibility other reviews (54–57).
of implementation. We found that participants were
paid in three studies (9,15,19) and that mobile devices
Conclusion
were supplied in two studies (9,18). Therefore, these
findings do not provide useful insight into the poten- Our results showed mixed evidence regarding the ben-
tial scope of the intervention, since it is not feasible to efits of interventions. This is probably because of the
provide mobile phones to the whole target population. variety of the study designs and the results found.
However, if these interventions promote adherence, Nevertheless, the interventions do seem to have been
the cost of non-adherence may be higher than the cost beneficial, as 65% of the studies had positive out-
of a mobile device. Haberer et al. (52) discuss feasibil- comes. Therefore, more high-quality studies should be
ity issues in implementing the intervention described conducted in order to demonstrate whether this type
by Pop-Eleches et al. (9) Based on qualitative inter- of technology reduces the considerable consequences
views with participants, the major issues pointed out of non-adherence. However, since mHealth enables
were the need for detailed, multisession trainings for real-time interactive self-management, policy-makers
patients on the use of mobile phone technologies and should consider funding programmes to establish its
PIN numbers and the possible need for monetary efficiency and applicability.

3 Pew Research Center. Smartphone Ownership-2013 5 Wesolowski A, Eagle N, Noor AM et al. Heteroge-
References Update. http://www.pewinternet.org/files/old-media// neous mobile phone ownership and usage patterns
Files/Reports/2013/PIP_Smartphone_adoption_ in Kenya. PLoS ONE 2012; 7: e35319.
1 Osterberg L, Blaschke T. Adherence to medication.
2013_PDF.pdf (accessed April 3, 2014). 6 Ryu S. Book review: MHealth: new horizons for
N Engl J Med 2005; 353: 487–97.
4 Barret LL. Health and Caregiving Among the 50+: health through mobile technologies: based on the
2 WHO. Adherence to Long-term Therapies: Evi-
Ownership, Use, and Interest in Mobile Technology findings of the second global survey on eHealth
dence for Action. World Health Organization.
- health-caregiving-mobile-technology.pdf. http:// (global observatory for eHealth Series, Volume 3).
http://www.who.int/chp/knowledge/publications/
assets.aarp.org/rgcenter/general/health-caregiving- Healthc Inform Res. Korean Society of Medical Infor-
adherence_report/en/ (accessed October 2,
mobile-technology.pdf (accessed April 3, 2014). matics 2012; 18: 231.
2014).

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
 mHealth and medication adherence 23

7 Higgins JPT, Green S. Cochrane Handbook for Sys- patients with psoriasis. Br J Dermatol 2013; 168: 38 Bosley CM, Fosbury JA, Cochrane GM. The psy-
tematic Reviews of Interventions. Chichester: Wiley- 201–5. chological factors associated with poor compliance
Blackwell, 2008. 22 Miloh T, Annunziato R, Arnon R et al. Improved with treatment in asthma. Eur Respir J 1995; 8:
8 Lester RT, Ritvo P, Mills EJ et al. Effects of a adherence and outcomes for pediatric liver trans- 899–904.
mobile phone short message service on antiretrovi- plant recipients by using text messaging. Pediatrics 39 Sharkness CM, Snow DA. The patient’s view of
ral treatment adherence in Kenya (WelTel Kenya1): 2009; 124: e844–50. hypertension and compliance. Am J Prev Med 1992;
a randomised trial. Lancet 2010; 376: 1838–45. 23 Foreman KF, Stockl KM, Le LB et al. Impact 8: 141–6.
9 Pop-Eleches C, Thirumurthy H, Habyarimana JP of a text messaging pilot program on patient 40 Lorig KR, Sobel DS, Ritter PL et al. Effect of a
et al. Mobile phone technologies improve adher- medication adherence. Clin Ther 2012; 34: self-management program on patients with chronic
ence to antiretroviral treatment in a resource-lim- 1084–91. disease. Eff Clin Pract 2001; 4: 256–62.
ited setting: a randomized controlled trial of text 24 Vilella A, Bayas J-M, Diaz M-T et al. The role of 41 Clark NM. Management of chronic disease by
message reminders. AIDS 2011; 25(6): 825–34. mobile phones in improving vaccination rates in patients. Annu Rev Public Health 2003; 24: 289–
10 Mbuagbaw L, Thabane L, Ongolo-Zogo P et al. travelers. Prev Med 2004; 38: 503–9. 313.
The Cameroon Mobile Phone SMS (CAMPS) trial: 25 Ollivier L, Romand O, Marimoutou C et al. Use 42 Hoffman JA, Cunningham JR, Suleh AJ et al.
a randomized trial of text messaging versus usual of short message service (SMS) to improve Mobile direct observation treatment for tuberculo-
care for adherence to antiretroviral therapy. PLoS malaria chemoprophylaxis compliance after sis patients: a technical feasibility pilot using
ONE 2012; 7: e46909. returning from a malaria endemic area. Malar J mobile phones in Nairobi, Kenya. Am J Prev Med
11 Rodrigues R, Shet A, Antony J et al. Supporting 2009; 8: 236. 2010; 39: 78–80.
adherence to antiretroviral therapy with mobile 26 Cocosila M, Archer N, Haynes RB et al. Can wire- 43 Creary SE, Gladwin MT, Byrne M et al. A pilot
phone reminders: results from a cohort in South less text messaging improve adherence to preven- study of electronic directly observed therapy to
India. PLoS ONE 2012; 7: e40723. tive activities? Results of a randomised controlled improve hydroxyurea adherence in pediatric
12 Lewis MA, Uhrig JD, Bann CM et al. Tailored text trial. Int J Med Inform 2009; 78: 230–8. patients with sickle-cell disease. Pediatr Blood Can-
messaging intervention for HIV adherence: a 27 Hou MY, Hurwitz S, Kavanagh E et al. Using daily cer 2014; 61: 1068–73.
proof-of-concept study. Health Psychol 2013; 32: text-message reminders to improve adherence with 44 Furberg RD, Uhrig JD, Bann CM et al. Technical
248–53. oral contraceptives: a randomized controlled trial. Implementation of a Multi-Component, Text Mes-
13 Strandbygaard U, Thomsen SF, Backer V. A daily Obstet Gynecol 2010; 116: 633–40. sage-Based Intervention for Persons Living with
SMS reminder increases adherence to asthma treat- 28 mHealth. Healthcare Mobile Apps, Devices and HIV. JMIR Res Protoc 2012; 1: e17.
ment: a three-month follow-up study. Respir Med Solutions. Infographic. http://cdn1.ticbeat.com/wp- 45 Mobile Medical Applications. Guidance for Indus-
2010; 104: 166–71. content/uploads/2012/11/movil-salud-infografia.jpg try and Food and Drug Administration Staff.
14 Petrie KJ, Perry K, Broadbent E et al. A text mes- (accessed February 19, 2014). http://www.fda.gov/downloads/MedicalDevices/De-
sage programme designed to modify patients’ ill- 29 Mobile Pillbox App Promotes Diabetes Medication viceRegulationandGuidance/GuidanceDocuments/
ness and treatment beliefs improves self-reported Adherence: Report [Internet]. http://www.eweek.- UCM263366.pdf (accessed May 8, 2014).
adherence to asthma preventer medication. Br J com/mobile/mobile-pillbox-app-promotes-diabetes 46 Smartphone Secure Development Guidelines - EN-
Health Psychol 2012; 17: 74–84. -medication-adherence-Report/ (accessed February ISA. http://www.enisa.europa.eu/activities/Resil-
15 Park LG, Howie-Esquivel J, Chung ML et al. A text 19, 2014). ience-and-CIIP/critical-applications/smartphone-secu-
messaging intervention to promote medication 30 Garfield S, Clifford S, Eliasson L et al. Suitability of rity-1/smartphone-secure-development-guidelines
adherence for patients with coronary heart disease: measures of self-reported medication adherence for (accessed May 8 2014).
a randomized controlled trial. Patient Educ Couns routine clinical use: a systematic review. BMC Med 47 Bill Text – 104th Congress (1995–1996) – THO-
2014; 94: 261–8. Res Methodol 2011; 11: 149. MAS (Library of Congress). http://thomas.loc.gov/
16 Dick JJ, Nundy S, Solomon MC et al. Feasibility 31 Mbuagbaw L, Thabane L, Ongolo-Zogo P et al. cgi-bin/query/z?c104:H.R.3103.enr: (accessed May
and usability of a text message-based program for The Cameroon mobile phone SMS (CAMPS) trial: 8, 2014).
diabetes self-management in an urban Afri- a protocol for a randomized controlled trial of 48 Broomhead S, Mars M. Retrospective return on
can-American population. J Diabetes Sci Technol mobile phone text messaging versus usual care for investment analysis of an electronic treatment
2011; 5: 1246–54. improving adherence to highly active anti-retroviral adherence device piloted in the Northern Cape
17 Marquez Contreras E, de la Figuera von Wich- therapy. Trials 2011; 12: 5. Province. Telemed J E Health 2012; 18: 24–31.
mann M, Gil Guillen V et al. Effectiveness of an 32 Haynes RB, McKibbon KA, Kanani R. Systematic 49 McGillicuddy JW, Gregoski MJ, Weiland AK et al.
intervention to provide information to patients review of randomised trials of interventions to Mobile health medication adherence and blood
with hypertension as short text messages and assist patients to follow prescriptions for medica- pressure control in renal transplant recipients: a
reminders sent to their mobile phone (HTA-Alert). tions. Lancet 1996; 348: 383–6. proof-of-concept randomized controlled trial. JMIR
Aten Primaria 2004; 34: 399–405. 33 Horne R, Weinman J. Patients’ beliefs about pre- Res Protoc 2013; 2: e32.
18 Patel S, Jacobus-Kantor L, Marshall L et al. Mobi- scribed medicines and their role in adherence to 50 Kollaritsch H, Wiedermann G. Compliance of Aus-
lizing your medications: an automated medication treatment in chronic physical illness. J Psychosom trian tourists with prophylactic measures. Eur J Ep-
reminder application for mobile phones and hyper- Res 1999; 47: 555–67. idemiol 1992; 8: 243–51.
tension medication adherence in a high-risk urban 34 Labrecque M, Laurier C, Champagne F et al. Effect 51 Fuangchan A, Dhippayom T, Kongkaew C. Inter-
population. J Diabetes Sci Technol 2013; 7: 630–9. of age on the conformity rate to short-acting vention to promote patients’ adherence to antima-
19 Suffoletto B, Calabria J, Ross A et al. A mobile beta-agonist use criteria in asthma. J Asthma 2003; larial medication: a systematic review. Am J Trop
phone text message program to measure oral anti- 40: 829–35. Med Hyg 2014; 90: 11–9.
biotic use and provide feedback on adherence to 35 Degli Esposti E, Di Martino M, Sturani A et al. 52 Haberer JE, Kiwanuka J, Nansera D et al. Chal-
patients discharged from the emergency depart- Risk factors for uncontrolled hypertension in Italy. lenges in using mobile phones for collection of an-
ment. Acad Emerg Med 2012; 19: 949–58. J Hum Hypertens 2004; 18: 207–13. tiretroviral therapy adherence data in a
20 Iribarren S, Beck S, Pearce PF et al. TextTB: a 36 Mbuagbaw L, van der Kop ML, Lester RT et al. resource-limited setting. AIDS Behav 2010; 14:
Mixed Method Pilot Study Evaluating Acceptance, Mobile phone text messages for improving adher- 1294–301.
Feasibility, and Exploring Initial Efficacy of a ence to antiretroviral therapy (ART): an individual 53 Henderson C, Knapp M, Fernandez J-L et al. Cost
Text Messaging Intervention to Support TB patient data meta-analysis of randomised trials. effectiveness of telehealth for patients with long
Treatment Adherence. Tuberc Res Treat 2013; BMJ Open 2013; 3: e003950. term conditions (Whole Systems Demonstrator
2013: 349394. 37 Coons SJ, Sheahan SL, Martin SS et al. Predictors telehealth questionnaire study): nested economic
21 Balato N, Megna M, Di Costanzo L et al. Educa- of medication noncompliance in a sample of older evaluation in a pragmatic, cluster randomised con-
tional and motivational support service: a pilot adults. Clin Ther 1994; 16: 110–7. trolled trial. BMJ 2013; 346: f1035.
study for mobile-phone-based interventions in

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract
24 mHealth and medication adherence

54 Free C, Phillips G, Galli L et al. The effectiveness of infection. Cochrane Database Syst Rev 2013; 5: antiretroviral therapy in patients with HIV infec-
mobile-health technology-based health behaviour CD009189. tion. Cochrane Database Syst Rev 2012; 3:
change or disease management interventions for 56 De Jongh T, Gurol-Urganci I, Vodopivec-Jamsek V CD009756.
health care consumers: a systematic review. PLoS et al. Mobile phone messaging for facilitating
Med 2013; 10: e1001362. self-management of long-term illnesses. Cochrane
55 Gentry S, van-Velthoven MHMMT, Tudor Car L Database Syst Rev 2012; 12: CD007459. Paper received June 2014, accepted October 2014
et al. Telephone delivered interventions for reduc- 57 Horvath T, Azman H, Kennedy GE et al. Mobile
ing morbidity and mortality in people with HIV phone text messaging for promoting adherence to

ª 2014 John Wiley & Sons Ltd

Int J Clin Pract