A meta-analysis of the use of electronic reminders for patient adherence to medication in chronic disease care

Abstract

We conducted a meta-analysis of randomized controlled trials (RCTs) up to January 2014 which evaluated the effects of electronic reminders on patient adherence to medication in chronic disease care. A random-effects model was used to pool the outcome data. Subgroup analyses were performed to examine a set of moderators. Data from 20 studies, representing 22 RCTs, were synthesized. Thirteen trials utilized short message service (SMS) reminders, three used pager reminders and six employed electronic alarm device-triggered reminders. The meta-analysis showed that the use of electronic reminders was associated with a significant, yet small, improvement in patient adherence to medication (pooled Cohen’s d=0.29, 95% confidence interval 0.18, 0.41). The effect was sensitive to sample size, type of disease and intervention duration. The frequency and type of electronic reminders appeared to have no moderating effect on medication adherence. The use of electronic reminders seems to be a simple and potentially effective way of improving patient adherence to chronic medication. Future research should concern the optimum strategies for the design and implementation of electronic reminders, with which the effectiveness of the reminders is likely to be augmented.

Introduction

Adherence to long-term treatment regimens is a prerequisite for effective chronic disease care. However, many patients, especially the chronically ill, encounter difficulties in adhering to recommended treatment regimens.¹ Poor adherence and non-adherence to treatment regimens have long been a widely reported problem in various health conditions.^2–6 Evidence shows that a quarter of patients in general do not adhere to their treatment regimens.⁵ Treatment regimens, even if effective, are poorly implemented at the practical level, resulting in undertreated patients. In addition, low adherence is likely to compromise the effectiveness of medication treatment and consequently leads to poor health outcomes and waste of health care resources.^4,7,8

Although there is a substantial literature on the development of psychological, behavioural or educational adherence-enhancing interventions, these interventions are less effective than expected,^9,10 and are largely constrained by their heavy demands on time and labour, and by their limited coverage of patients. Another response to patients’ poor adherence is the introduction of electronic reminders, which are automatically sent reminders without personal caregiver-patient contact in the delivery process.¹¹ The principles behind reminders can be explained by the behavioural learning theory,¹² which suggests that patients’ non-adherent behaviours can be converted into adherent ones if the patients are sufficiently exposed to repeated external stimuli (e.g. reminders). In clinical practice, electronic reminders, such as short message service (SMS) reminders and electronic alarm device-triggered reminders, are used to prompt patients to follow medication recommendations due to their advantages of saving time and labour, and of being easily delivered to a wide range of patients.^13,14

While promising, results from previous reviews regarding the efficacy of electronic reminders are mixed.^11,13,15,16 Moreover, these reviews either failed to provide quantitative evidence regarding the effectiveness of electronic reminders;^11,13,15 or included a diverse range of health conditions (e.g. both chronic and non-chronic);¹⁶ or based their results on a number of low quality studies.¹⁵ The objective of the present study was to conduct a meta-analysis of randomized controlled trials (RCTs) that assessed electronic reminders in chronic disease care and to determine their effects on patient adherence to medication.

Methods

After a review of the literature on definitions of medication adherence in the health care domain,^6,17 patient adherence to medication was defined for the purposes of the present study as the extent to which patients take their medications as prescribed.

We conducted a systematic literature search of CINAHL Plus, MEDLINE, Cochrane Central Register of Controlled Trials and Web of Science for relevant articles up to January 2014 with the following search terms: (adheren* or complian* or persisten* or concordan*) AND (remind* or alarm* or messag*) AND (random*). The titles and abstracts of the initial search citations were read and assessed to determine their relevance based on our inclusion and exclusion criteria. Articles considered as relevant were kept for full-text review. Reference lists of selected articles were searched manually for additional potential articles. We also examined studies included in previous reviews^11,15,16 to catch possibly missing articles.

Inclusion and exclusion criteria

Studies were included if they: (1) were RCTs; (2) examined effects of electronic reminders on medication adherence outcomes; (3) delivered reminders directly to patients with any type of chronic disease; and (4) were written in English and peer-reviewed. Studies were excluded if the electronic reminders were used for disease screening, appointment attendance or non-chronic conditions, were delivered to physicians or were not the main component of the interventions evaluated.

Data extraction

The information extracted included study and intervention characteristics (i.e. author and year, country where the study was conducted, mean age, sex ratio, type of disease, sample size, intervention duration, description of intervention, attrition rate, primary outcome measure and its statistical significance, and outcome assessing method) and data on outcome measures. Two adherence measures were reported in the trials and were coded for the meta-analysis: the proportion of medication taken as prescribed and the proportion of adherent patients. We adopted original definitions of adherent patients for trials that reported this measure. For multiple-arm studies, we extracted all eligible comparison trials (i.e. use of reminders versus non-use of such reminders) and considered them as separate trials based on recent guidelines.¹⁸

Two authors independently conducted the article selection, data extraction and study quality assessment. Any discrepancy was resolved through discussion between them, or by third-party adjudication.

Statistical methods

Although the adherence measures varied across studies, the outcome difference between intervention and control groups can be measured and pooled using appropriate meta-analysis methods.¹⁹ In our study, Cohen’s d was calculated as the effect size to determine the magnitude of the difference in medication adherence between the intervention and control groups.¹⁹ A Cohen’s d value of 0.2, 0.5 and 0.8 can be considered as small, medium and large effect size, respectively.²⁰ A random-effects model was used to aggregate individual effect sizes, as large between-study heterogeneity was expected.²¹ The I² statistic, which describes the percentage of total variation across studies, was calculated to examine the between-study heterogeneity.²² The possibility of publication bias was assessed using a visual inspection of a funnel plot,²³ and using Egger’s regression test, with P < 0.1 considered to indicate the presence of publication bias.²⁴

Because including several effect sizes from one trial violated the independence assumption in meta-analysis,¹⁹ we chose only one effect size for each of the trials in our analysis. For trials reporting multiple adherence measures, we chose the one more likely to be objective and reliable (e.g. assessed by electronic monitoring over pill count and self-report; continuous scale over dichotomized scale). This approach has been widely adopted in previous meta-analyses on patient adherence to medication.^9,25 We also calculated an average effect size for each of these trials and used the average effect size in sensitivity analysis to test the robustness of our effect size calculation method. The average effect size approach was also a widely accepted solution to guarantee the independence assumption in meta-analysis.^8,19

Subgroup analyses

In response to substantial variability across studies (based on the I² test for heterogeneity), subgroup analyses were performed to explore potential factors that would moderate the overall effect size. The following moderators were examined: age (age ≤ 37.4 or >37.4 years); sample size (n ≤97 or n >97); type of disease (diabetes, asthma, AIDS, cardiovascular diseases or other type of diseases); intervention duration (short-term (six months or less) or long-term (more than six months)); type of reminder (SMS, pager or electronic alarm device); frequency of reminder (daily, i.e. one or more reminders per day or weekly, i.e. less than one reminder per day); type of outcome measure (proportion of medication taken as prescribed or proportion of adherent patients); and type of outcome assessing method (electronic monitoring, self-report, pill count or pharmacy dispensing records). The cut-off points for age group and sample size were based on the median values among trials with available information.

Study quality

The study quality was assessed by the nine-item Delphi list for RCT quality assessment, using answers of yes, no or don’t know, where yes indicated that a criterion was met.²⁶ The study quality was rated as high if five or more of the nine criteria were met; otherwise it was rated as low.²⁶

Sensitivity analyses

We conducted three sensitivity analyses to test the robustness of the meta-analysis result. The first sensitivity analysis used average effect size for trials reporting multiple outcome measures; the second excluded trials with high sample attrition rate (20% or more); and the third excluded low quality trials.

Results

Twenty studies were included in the review.^27–46 We extracted two separate comparison trials (daily SMS versus usual care; weekly SMS versus usual care) from the study of Pop-Eleches et al.,⁴¹ and another two trials (pager versus usual care; pager and peer support versus peer support) from the study of Simoni et al.⁴³ Therefore, a total of 22 comparison trials were examined in the meta-analysis, Figure 1.

Figure 1.

Study search and selection procedures.

Study characteristics

The trials examined a variety of patient populations with conditions such as diabetes,^32,45 asthma,^28,36,40,44 cardiovascular diseases,^29,34,37,39 AIDS,^{27,30,31,35,38,41–43} glaucoma³³ and allergic rhinitis,⁴⁶ see Table 1. The sample size varied from 21³¹ to 538,³⁵ with a median of 97. Twenty trials reported the patients’ mean age, which ranged from 14³² to 74³⁴ years (median 37 years). Intervention duration varied across trials, with a short-term duration for 14 trials^{27–29,31,33,36–40,42,44–46} and a long-term duration for eight trials.^{30,32,34,35,41,43} Based on our predetermined rules of outcome measure selection, patient adherence to medication was represented by the proportion of medication taken as prescribed in 15 trials^{27–29,32–34,37–40,42–45} and by the proportion of adherent patients in seven trials.^{30,31,35,36,41,46} The cut-off point for adherent patients was taking 95% of medication as prescribed in six trials^{30,31,35,41,46} and unspecified in one trial.³⁶ The outcome assessing methods included electronic monitoring (e.g. medication event monitoring system, electronic drug exposure monitor or microelectronic monitor),^{27,28,31,33,39,41–43,45} self-report,^{29,32,35,40,46} pill count^37,38,44 and pharmacy dispensing records.^30,34 One trial did not specify its outcome assessing method.³⁶

Table 1.

Study characteristics of the 22 included trials.

Author and country	Mean age (years) / Female (%)	Disease	Description of electronic reminders	Sample size (Intervention/ Control)	Intervention duration (months)	Outcome measure	Outcome assessing method
Andrade et al., 2005²⁷ USA	38.0 / 41%	AIDS	A programmable prompting device that produced timed (daily), programmed voice messages that prompted subjects to take antiretroviral drugs	29 / 29	6	Proportion of medication taken as prescribed	Electronic monitoring
Charles et al., 2007²⁸ New Zealand	37.0 / 55%	Asthma	Audiovisual reminders (green light and a beep sound once every 30 s for 60 min) at predetermined times and twice a day	44 / 46	6	Proportion of medication taken as prescribed	Electronic monitoring
Christensen et al., 2010²⁹ Poland	53.0–65.9 / 53%	Hypertension	Audiovisual reminders from a monitoring device sent to patients once daily when it was time to take medication	219 / 179	6	Proportion of medication taken as prescribed	Self-report
Chung et al., 2011³⁰ Kenya	36.0 / 66%	AIDS	Pocket electronic pill reminders with beep and flash twice a day at a time convenient to prompt medication adherence	174 / 188	18	Proportion of adherent patients (take 95% medication or more)	Pharmacy dispensing records
da Costa et al., 2012³¹ Brazil	34.6 / 100%	AIDS	SMS messages sent 30 min before their last scheduled time for a dose of medicine; sent every Saturday and Sunday and on alternate days during the working week	8 / 13	4	Proportion of adherent patients (take 95% medication or more)	Electronic monitoring
Franklin et al., 2006³² UK	13.5 / 47%	Diabetes	A text-messaging support system that delivered tailored messages, including weekly reminders of goals set in clinics, and daily messages providing tips or reminders to reinforce self-care goals	32 / 27	12	Proportion of medication taken as prescribed	Self-report
Ho et al., 2008³³ USA	69.1 / 55%	Glaucoma	Audible and visual daily alarms on a commercially available glaucoma medication dosing aid	20 / 22	1	Proportion of medication taken as prescribed	Electronic monitoring
Kooy et al., 2013³⁴ Netherlands	73.6 / 57%	Cardiovascular disease	A medication reminder device that started beeping every day at the same time until the patient switched it off	123 / 128	12	Proportion of medication taken as prescribed	Pharmacy dispensing records
Lester et al., 2010³⁵ Kenya	36.7 / 65%	AIDS	A weekly mobile phone short message service that prompted patients to take their antiretroviral	273 / 265	12	Proportion of adherent patients (take 95% medication or more)	Self-report
Lv et al., 2012³⁶ China	37.8 / 39%	Asthma	Two daily SMS messages about managing asthma, including introduction to asthma, classes of asthma medications, proper device use, common types of triggers, strategies for avoiding triggers, how to make an action plan and handle asthma acute attacks	30 / 14	3	Proportion of adherent patients (unspecified)	Not specified
Márquez Contreras et al., 2004³⁷ Spain	57.8 / 45%	Hypertension	Two SMS messages sent to patients’ mobile phone weekly, providing information on hypertension, good health and dietary habits, and reminding patients to take their medication	34 / 33	6	Proportion of medication taken as prescribed	Pill count
Mbuagbaw et al., 2012³⁸ Cameroon	40.2 / 74%	AIDS	Weekly standardized motivational text messages to prompt treatment adherence	101 / 99	6	Proportion of medication taken as prescribed	Pill count
Park et al., 2014³⁹ USA	58.3 / 30%	Heart disease	Text messages for medication reminders, delivered twice daily	28 / 28	1	Proportion of medication taken as prescribed	Electronic Monitoring
Petrie et al., 2012⁴⁰ New Zealand	Not specified / 68%	Asthma	Individually tailored text messages based on patients’ illness and medication beliefs	57 / 46	4.5	Proportion of medication taken as prescribed	Self-report
Pop-Eleches et al., 2011(1)⁴¹ USA	35.7 / 67%	AIDS	Daily text messages to promote adherence to antiretroviral therapy	142 / 68	12	Proportion of adherent patients (take 95% medication or more)	Electronic monitoring
Pop-Eleches et al., 2011(2)⁴¹ USA	36.7 / 66%	AIDS	Weekly text messages to promote adherence to antiretroviral therapy	147 / 69	12	Proportion of adherent patients (take 95% medication or more)	Electronic monitoring
Safren et al., 2003⁴² USA	Not specified / Not specified	AIDS	Paging service delivering messages at designated times to increase adherence to antiretroviral medication	19 / 25	3	Proportion of medication taken as prescribed	Electronic monitoring
Simoni et al., 2009 (1)⁴³ USA	40.0 / 24%	AIDS	Two-way pager system that sent daily message with dose reminder and adherence assessments (compared with usual care group)	56 / 57	9	Proportion of medication taken as prescribed	Electronic Monitoring
Simoni et al., 2009 (2)⁴³ USA	40.0 / 24%	AIDS	Peer support and two-way pager system that sent daily message with dose reminder and adherence assessments (compared with peer support group)	54 / 57	9	Proportion of medication taken as prescribed	Electronic Monitoring
Strandbygaard et al., 2010⁴⁴ Denmark	32.2 / 46%	Asthma	Daily SMS reminders to remind patients to take their asthma medication	12 / 14	3	Proportion of medication taken as prescribed	Pill count
Vervloet et al., 2012⁴⁵ Netherlands	55.0 / 45%	Diabetes	SMS reminders for each dose if patients did not open their medication dispensers within the agreed time period	56 / 48	6	Proportion of medication taken as prescribed	Electronic monitoring
Wang et al., 2014^4 6 China	33.4 / 49%	Allergic rhinitis	Daily SMS reminders was sent to patients’ cell phones to remind them to take intranasal corticosteroid treatment	20 / 19	1	Proportion of adherent patients (take 95% medication or more)	Self-report

The type, frequency and content of reminders varied substantially across the 22 trials. Thirteen trials sent SMS reminders,^{31,32,35–41,44–46} three used pager reminders^42,43 and the remaining six employed electronic alarm device-triggered reminders.^{27–30,33,34} The frequency of reminders ranged from daily^{27–30,32–34,36,39–46} to weekly.^{31,35,37,38,41} The content and format of the reminders depended largely on the type of reminders. For example, electronic alarm devices emitted alarms and displayed coloured lights at a predetermined time to remind patients to take medication.²⁸ SMS reminders delivered information-rich messages to patients, aiming to inform them of medication intake and to reinforce their medication-taking behaviours (e.g. Márquez Contreras et al.³⁷).

Quality assessment

Fifteen studies were rated as high quality studies, while five had a low study quality,^{29,33,36,41,44} see Table 2. All studies conducted a randomization procedure, specified their eligibility criteria, and reported point estimates and variability measures for primary outcome measures. All studies except one⁴¹ had similar experimental groups at baseline. Treatment allocation concealment and intention-to-treat analysis were not performed or clearly described in most of the studies. One study blinded its intervention to patients,²⁸ and one to care providers.³¹ Five studies blinded their interventions to outcome assessors.^{28,31,35,37,38}

Table 2.

Quality assessment of the 20 included studies by the Delphi list.

	Criteria									Total scores	Study quality
	1	2	3	4	5	6	7	8	9
Andrade et al.²⁷	Y	D	Y	Y	D	D	N	Y	Y	5	High
Charles et al.²⁸	Y	D	Y	Y	Y	N	Y	Y	N	6	High
Christensen et al.²⁹	Y	D	Y	Y	D	D	N	Y	N	4	Low
Chung et al.³⁰	Y	Y	Y	Y	N	D	N	Y	Y	6	High
da Costa et al.³¹	Y	D	Y	Y	Y	Y	N	Y	N	6	High
Franklin et al.³²	Y	Y	Y	Y	D	D	N	Y	Y	6	High
Ho et al.³³	Y	D	Y	Y	D	D	N	Y	N	4	Low
Kooy et al.³⁴	Y	D	Y	Y	D	D	N	Y	Y	5	High
Lester et al.³⁵	Y	Y	Y	Y	Y	N	N	Y	Y	7	High
Lv et al.³⁶	Y	D	Y	Y	D	D	N	Y	N	4	Low
Márquez Contreras et al.³⁷	Y	D	Y	Y	Y	D	N	Y	N	5	High
Mbuagbaw et al.³⁸	Y	Y	Y	Y	Y	D	N	Y	Y	7	High
Park et al.³⁹	Y	Y	Y	Y	N	D	N	Y	Y	6	High
Petrie et al., 201240	Y	Y	Y	Y	D	D	N	Y	N	5	High
Pop-Eleches et al.⁴¹	Y	D	N	Y	D	D	N	Y	Y	4	Low
Safren et al.⁴²	Y	D	Y	Y	D	D	N	Y	Y	5	High
Simoni et al.⁴³	Y	Y	Y	Y	N	N	N	Y	Y	6	High
Strandbygaard et al.⁴⁴	Y	D	Y	Y	D	D	N	Y	N	4	Low
Vervloet et al.⁴⁵	Y	Y	Y	Y	D	D	N	Y	Y	6	High
Wang et al.⁴⁶	Y	Y	Y	Y	D	D	N	Y	N	5	High

Y=yes, N= no, D= don’t know.

The nine criteria in the Delphi list:

(1) Was a method of randomization performed?

(2) Was the treatment allocation concealed?

(3) Were the groups similar at baseline regarding the most important prognostic indicators?

(4) Were eligibility criteria specified?

(5) Was the outcome assessor blinded?

(6) Was the care provider blinded?

(7) Was the patient blinded?

(8) Were point estimates and measures of variability presented for the primary outcome measures?

(9) Did the analysis include an intention-to-treat analysis?

Meta-analysis

The pooled estimate based on data from 3152 patients showed that the use of electronic reminders was associated with a significant improvement in patient adherence to treatment (d=0.29, 95% confidence interval, CI=0.18, 0.41), see Figure 2. This weighted mean effect size was small. The funnel plot analysis (Figure 3) and Egger’s regression test (P=0.10) did not detect the presence of publication bias. Heterogeneity was high (I²= 55%), indicating the existence of variability across the trials. We therefore performed subgroup analyses to investigate the impact of moderators on the overall effect size.

Figure 2.

The effect of electronic reminders on patient adherence to medication.

Figure 3.

Funnel plot of standard error and Cohen’s d of patient adherence to medication between intervention and control groups.

Subgroup analyses

Sample size, type of disease and intervention duration were associated with a significant between-group difference in patient adherence to medication, Table 3. Small sample trials showed medium effect sizes (d=0.50, 95% CI=0.33, 0.66), while large sample trials displayed small effect sizes (d=0.18, 95% CI=0.07, 0.29). The subgroup analysis showed a significant difference in medication adherence between small and large sample trials (Δd=−0.32, P=0.002).

Table 3.

Subgroup analyses of the effect of electronic reminders on patient adherence to medication.

Moderators	Number of trials	Pooled d within subgroup (95% CI)	Heterogeneity, I² (%)	Δd
Sample size
No. of participants ≤97	11	0.50 (0.33 to 0.66)***	29
No. of participants >97	11	0.18 (0.07 to 0.29)**	44	−0.32**
Type of disease
Asthma	4	0.78 (0.52 to 1.04) ***	0
Cardiovascular diseases	4	0.13 (−0.02 to 0.27)	0	−0.65***
Diabetes	2	0.39 (0.26 to 0.52)***	0	−0.39**
AIDS	10	0.17 (0.07 to 0.26)***	18	−0.61***
Other types of diseases	2	0.81 (0.33 to 1.29)**	0
Mean age
Mean age ≤37.4 years	10	0.33 (0.19 to 0.48)***	66
Mean age >37.4 years	10	0.19 (0.03 to 0.34)*	25	−0.14
Not specified	2	0.65 (0.24 to 1.06)**	0
Intervention duration
Short-term (≤6 months)	14	0.43 (0.28 to 0.59)***	55
Long-term (9 – 18 months)	8	0.17 (0.02 to 0.32)*	50	−0.26*
Type of reminder
SMS	13	0.32 (0.17 to 0.47)***	43
Pager	3	0.21 (−0.12 to 0.53)	55	−0.11
Alarm device	6	0.30 (0.09 to 0.50)**	73	−0.02
Frequency of reminder
Daily	17	0.32 (0.19 to 0.45)***	64
Weekly	5	0.23 (0.01 to 0.46)*	0	−0.09
Type of outcome measure
Proportion of medication taken as prescribed	15	0.32 (0.19 to 0.46)***	58
Proportion of adherent patients	7	0.24 (0.05 to 0.43)*	43	−0.08
Type of outcome assessing method
Pill count	3	0.25 (−0.08 to 0.58)	53
Electronic monitoring	11	0.32 (0.15 to 0.49)***	60	0.07
Self-report	5	0.36 (0.16 to 0.56)***	44	0.11
Pharmacy dispensing records	2	0.02 (−0.27 to 0.31)	0	−0.23
Unspecified	1	0.73 (−0.10 to 1.55)	0

Note: Δd, The difference of pooled effect sizes between two subgroups within a given moderator, with the first subgroup under each moderator as the reference group.

*P < 0.05; **P < 0.01; ***P < 0.001.

Effect sizes regarding the type of disease varied, and ranged from medium to large magnitude for asthma (d=0.78, 95% CI=0.52, 1.04), small to medium for diabetes (d=0.39, 95% CI=0.26, 0.52), and small for cardiovascular diseases (d=0.13, 95% CI=−0.02, 0.27) and HIV (d=0.17, 95% CI=0.07, 0.26). Patient adherence to medication was significantly higher for asthma, compared with that for cardiovascular diseases (Δd=−0.65, P < 0.001), diabetes (Δd=−0.39, P=0.008) and HIV (Δd=−0.61, P < 0.001). The heterogeneity substantially decreased in subgroups by type of disease.

There was a trend towards diluted effects of electronic reminders over time (Δd=−0.26, P=0.017). Effect sizes fell in the small to medium range for short-term interventions (d=0.43, 95% CI=0.28, 0.59), while they were small for long-term interventions (d=0.17, 95% CI=0.02, 0.32).

The effect sizes were more pronounced when the reminders were delivered by SMS (d=0.32, 95% CI=0.17, 0.47) and alarm device (d=0.30, 95% CI=0.09, 0.50) compared with that by pager (d=0.21, 95% CI=−0.12, 0.53). The effect sizes were similar, whether the reminders were delivered daily (d=0.32, 95% CI=0.19, 0.45) or weekly (d=0.23, 95% CI=0.01, 0.46). There was no interaction effect between the type and frequency of the reminders.

The effect sizes did not differ much in subgroups defined by age, type of outcome measure or type of outcome assessing method.

Sensitivity analyses

The primary meta-analysis result was robust in the three sensitivity analyses. The effect of electronic reminders on patient adherence to medication did not change when we used average effect sizes for nine trials (eight studies)^{11,30,31,34,37,39,40,43} with multiple outcome measures (d=0.28, 95% CI=0.16, 0.39); when we excluded seven trials^{29,30,36,37,40,42,46} with sample attrition rates of 20% or more (d=0.28, 95% CI=0.15, 0.41; and when we excluded six trials (five studies)^{29,33,36,41,44} with low study quality (d=0.29, 95% CI=0.15, 0.43).

Discussion

The present study identified three types of electronic reminders (i.e. SMS, alarm device and pager) that have been widely used in clinical practice to enhance medication adherence for chronically ill patients. Despite the technological diversity of these reminders, we pooled and compared their effects on medication adherence, because we believe that their conceptual and functional similarity allow us to do so. We found that electronic reminders, regardless of their types, were associated with a significant, yet small, improvement in patient adherence to medication, compared to non-use of reminders, with a pooled effect size of 0.29 (95% CI=0.18, 0.41).

The meta-analysis result was consistent with findings of previous reviews,^11,16 which showed favourable effects of electronic reminders.^11,16 For example, Fenerty et al. found that technology-based reminder systems yielded a significantly higher treatment adherence rate compared with usual care (67% in reminder group versus 55% in control group) in various health conditions.¹⁶ Vervloet et al. found that the use of electronic reminders appeared to lead to short-term effectiveness in increasing adherence to chronic medication.¹¹ The effect of electronic reminders in our study was also similar to previously reported effect sizes in meta-analyses of psychological, behavioural or educational adherence-enhancing interventions, e.g. the study by Kahana et al.¹⁰ (d=0.34, 95% CI=0.30, 0.38) and by Conn et al.⁹ (d=0.33, 95% CI=0.22, 0.45).

However, the small effect detected in our study, although significant, fell below the threshold of a clinically satisfactory improvement in medication adherence for the majority of the trials analysed. This may be because, as noted by Haynes et al.,⁴⁷ many chronically ill patients tend to consider their conditions incurable, thus being less motivated to persist in taking long-term medication. Therefore, further work is required to reinforce patients’ belief in the efficacy of taking medication and to augment the effectiveness of the reminders in improving patient adherence to medication.

Moderators on the use of electronic reminders

Subgroup analysis showed that the effect of electronic reminders was sensitive to sample size, type of disease and intervention duration, while the effect did not change with respect to age, type of outcome measure and type of outcome assessing method. However, these findings should be interpreted cautiously, because of the small number of included studies, the associations among the moderators and the substantial heterogeneity within subgroups.

The trend towards a larger effect in small sample trials appeared to show a possible pattern of small sample trial effect, which suggests that small sample trials are more likely to detect larger intervention effects and are therefore more likely to be published.⁴⁸ In addition, this finding may confound the effect sizes in relation to other moderators. We noted that small sample trials contributed to a disproportionate number of trials examining patients with asthma (3 out of 4, 75%), and of trials with a short-term intervention duration (10 out of 14, 71%). Therefore, future studies should consider the interrelationship among the moderators when disentangling the moderating effects. Large sample size studies are also preferable in future evaluations of electronic reminders.

Our analysis indicated a higher medication adherence for patients with asthma compared to patients with other types of diseases. However, it is not clear whether this effect was truly disease specific or due to other reasons, such as small sample trial effect. In addition, we noted that a larger proportion of trials on asthma delivered enhanced reminders, containing not only prompt tips for medication intake, but also educational messages about disease management and advice on behavioural change, compared with trials on other conditions (75% for asthma,^36,40,44 50% for cardiovascular diseases^37,39 and diabetes,³² and 20% for HIV⁴¹). The enhanced reminders might also have contributed to the larger effect for patients with asthma. This suggests the possibility of combining electronic reminders with educational and behavioural interventions in order to produce greater benefit.

The finding that the effect of electronic reminders appeared to diminish over time echoed the conclusion of a previous review, which found only short-term effectiveness of the use of electronic reminders.¹¹ This has important clinical implications, since it is premature for any statement to be made about the long-term effectiveness of electronic reminders. However, this finding runs counter to our expectations. The primary purpose of reminders is to help patients take prescribed medication at the right time and to reinforce their adherent behaviours so that adherence to medication, in the long-term, can be a routine, or even a habituation in chronic disease care. Therefore, a long-term effect on medication adherence is probably an essential prerequisite for such behavioural changes, although it awaits confirmation in future studies.

Design and implementation strategies, such as the content, type, frequency and delivery time of electronic reminders, have been less investigated in previous reviews^11,13,15,16 despite their potential importance in determining the efficacy of electronic reminders. In contrast, our study examined the type and frequency of reminders, but found no significant difference in medication adherence in relation to the two moderators. More research should be carried out to examine the under-investigated moderators so that electronic reminders can be better developed for and tailored to specific circumstances. However, it should be noted that reminders by SMS and alarm devices did show a trend towards better medication adherence compared to pagers. The use of pagers seems to have decreased, while SMS has become ubiquitous, and alarm devices are designed to be more user-friendly in home care environments. We speculate that SMS reminders are likely to confer more benefits to active mobile phone users, and alarm devices are probably preferable for house-bound patients.

Our analysis did not find any discernible effect on medication adherence in relation to the type of outcome measure and type of outcome assessment method. Regarding the type of outcome measure, the proportion of adherent patients seems to be a more clinically meaningful measure in illustrating reminder effect compared with the proportion of medication taken as prescribed. This measure, however, is likely to underestimate the effect. For example, a patient could be labelled as non-adherent, because the improved adherence may still fail to attain the threshold of being defined as an adherent patient. For outcome assessing methods, previous studies have argued that self-reporting is likely to overestimate adherence compared with methods such as electronic monitoring and pharmacy dispensing records.^49,50 However, this argument cannot be confirmed by our findings. Possible explanations may be that the number of trials in the subgroups was too small to detect a significant difference; and that the lack of sufficient data in the trials did not allow us to compare the outcome assessing methods within the trials. While the evidence regarding these moderating effects is unclear, future studies are recommended to incorporate the two outcome measures and multiple outcome assessing methods for adherence measurement when possible, to produce more data for further examination.

Electronic reminders have advantages in timeliness and salience. They can be delivered or triggered at a specified time when patients need to act on medication regimens and in a way that the patients’ attention can be attracted by them (e.g. by ringtones or alarms). However, electronic reminders in any form as adherence-enhancing strategies work only for unintentionally non-adherent patients, rather than for patients who ignore them. None of the trials reported that they had specifically examined unintentionally non-adherent patients. Therefore, the effect of electronic reminders detected in our study may have been underestimated due to the possible inclusion of both intentionally and unintentionally non-adherent patients.

Strengths and limitations

The present study has several strengths. To the best of our knowledge, this study was the first to employ a meta-analysis to assess the effect of electronic reminders on medication adherence in chronic disease care. Second, we performed subgroup analyses in response to the between-study heterogeneity, and examined a set of moderators that have been little investigated previously but are critical to the design and implementation of the reminders. Finally, we tested the robustness of our results in sensitivity analyses with factors (i.e. study quality, attrition rate and effect size calculation method) that are likely to bias the results.

However, the study also has a number of limitations. First, the small number of trials included placed limitations on the subgroup analyses, which means that the findings cannot be considered conclusive. Second, pooling different adherence measures may represent a methodological drawback, although we calculated Cohen’s d to standardize these measures and pooled the effect sizes using an appropriate meta-analysis method. Third, several moderators examined in the subgroup analyses may be confounded with each other, so that their effects should be interpreted with full recognition of their interrelationship.

Conclusion

The use of electronic reminders appears to be a simple and potentially effective method of improving patient adherence to chronic medication. Future research should concern the optimum strategies for the design and implementation of electronic reminders, with which the effectiveness of the reminders is likely to be augmented.

References

Martin

Williams

Haskard

Dimatteo

. The challenge of patient adherence. Ther Clin Risk Manag 2005; 1: 189–99.

Dunbar-Jacob

Mortimer-Stephens

. Treatment adherence in chronic disease. J Clin Epidemiol 2001; 54(Suppl. 1): 57–60.

Daly

Sindone

Thompson

Hancock

Chang

Davidson

. Barriers to participation in and adherence to cardiac rehabilitation programs: a critical literature review. Prog Cardiovasc Nurs 2002; 17: 8–17.

Bender

Rand

. Medication non-adherence and asthma treatment cost. Curr Opin Allergy Clin Immunol 2004; 4: 191–5.

DiMatteo

. Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med Care 2004; 42: 200–9.

Osterberg

Blaschke

. Adherence to medication. N Engl J Med 2005; 353: 487–97.

Rumsfeld

Masoudi

, et al. Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. Arch Intern Med 2006; 166: 1836–41.

DiMatteo

Giordani

Lepper

Croghan

. Patient adherence and medical treatment outcomes: a meta-analysis. Med Care 2002; 40: 794–811.

Conn

Hafdahl

Cooper

Ruppar

Mehr

Russell

. Interventions to improve medication adherence among older adults: meta-analysis of adherence outcomes among randomized controlled trials. Gerontologist 2009; 49: 447–62.

10.

Kahana

Drotar

Frazier

. Meta-analysis of psychological interventions to promote adherence to treatment in pediatric chronic health conditions. J Pediatr Psychol 2008; 33: 590–611.

11.

Vervloet

Linn

van Weert

de Bakker

Bouvy

van Dijk

. The effectiveness of interventions using electronic reminders to improve adherence to chronic medication: a systematic review of the literature. J Am Med Inform Assoc 2012; 19: 696–704.

12.

Leventhal

Cameron

. Behavioral theories and the problem of compliance. Patient Educ Couns 1987; 10: 117–138.

13.

Tran

Coffman

Sumino

Cabana

. Patient reminder systems and asthma medication adherence: a systematic review. J Asthma 2014; 51: 536–43.

14.

Granger

Bosworth

. Medication adherence: emerging use of technology. Curr Opin Cardiol 2011; 26: 279–87.

15.

Wise

Operario

. Use of electronic reminder devices to improve adherence to antiretroviral therapy: a systematic review. AIDS Patient Care STDS 2008; 22: 495–504.

16.

Fenerty

West

Davis

Kaplan

Feldman

. The effect of reminder systems on patients' adherence to treatment. Patient Prefer Adherence 2012; 6: 127–35.

17.

De las Cuevas

. Towards a clarification of terminology in medicine taking behavior: compliance, adherence and concordance are related although different terms with different uses. Curr Clin Pharmacol 2011; 6: 74–7.

18.

Higgins J, Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions. See http://www.cochrane-handbook.org (last checked 25 May 2014).

19.

Hedges

Olkin

. Statistical Method for Meta-Analysis, Orlando, Fl: Academic Press, 1985.

20.

Cohen

. Statistical Power Analysis for the Behavioral Sciences, Hillsdale, NJ: Lawrence Earlbaum Associates, 1988.

21.

Borenstein

Hedges

Higgins

Rothstein

. Introduction to Meta-Analysis, Chichester: John Wiley, 2009.

22.

Higgins

Thompson

. Quantifying heterogeneity in a meta-analysis. Stat Med 2002; 21: 1539–58.

23.

Begg

Publication bias. In: Cooper

Hedges

(eds). Handbook of Research Synthesis, New York: Russell Sage Foundation, 1994.

24.

Egger

Davey Smith

Schneider

Minder

. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–34.

25.

Grenard

Munjas

Adams

, et al. Depression and medication adherence in the treatment of chronic diseases in the United States: a meta-analysis. J Gen Intern Med 2011; 26: 1175–82.

26.

Verhagen

de Vet

de Bie

, et al. The Delphi list: a criteria list for quality assessment of randomized clinical trials for conducting systematic reviews developed by Delphi consensus. J Clin Epidemiol 1998; 51: 1235–41.

27.

Andrade

McGruder

, et al. A programmable prompting device improves adherence to highly active antiretroviral therapy in HIV-infected subjects with memory impairment. Clin Infect Dis 2005; 41: 875–82.

28.

Charles

Quinn

Weatherall

Aldington

Beasley

Holt

. An audiovisual reminder function improves adherence with inhaled corticosteroid therapy in asthma. J Allergy Clin Immunol 2007; 119: 811–6.

29.

Christensen

Christrup

Fabricius

, et al. The impact of an electronic monitoring and reminder device on patient compliance with antihypertensive therapy: a randomized controlled trial. J Hypertens 2010; 28: 194–200.

30.

Chung

Richardson

Tapia

, et al. A randomized controlled trial comparing the effects of counseling and alarm device on HAART adherence and virologic outcomes. PLoS Med 2011; 8: e1000422–e1000422.

31.

da Costa

Barbosa

Gomes e Costa

, et al. Results of a randomized controlled trial to assess the effects of a mobile SMS-based intervention on treatment adherence in HIV/AIDS-infected Brazilian women and impressions and satisfaction with respect to incoming messages. Int J Med Inform 2012; 81: 257–69.

32.

Franklin

Waller

Pagliari

Greene

. A randomized controlled trial of Sweet Talk, a text-messaging system to support young people with diabetes. Diabet Med 2006; 23: 1332–8.

33.

Camejo

Kahook

Noecker

. Effect of audible and visual reminders on adherence in glaucoma patients using a commercially available dosing aid. Clin Ophthalmol 2008; 2: 769–72.

34.

Kooy

van Wijk

Heerdink

de Boer

Bouvy

. Does the use of an electronic reminder device with or without counseling improve adherence to lipid-lowering treatment? The results of a randomized controlled trial. Front Pharmacol 2013; 4: 69–69.

35.

Lester

Ritvo

Mills

, et al. Effects of a mobile phone short message service on antiretroviral treatment adherence in Kenya (WelTel Kenya1): a randomised trial. Lancet 2010; 376: 1838–45.

36.

Zhao

Liang

, et al. A mobile phone short message service improves perceived control of asthma: a randomized controlled trial. Telemed J E Health 2012; 18: 420–6.

37.

Márquez Contreras

de la Figuera von Wichmann

Gil Guillén

, et al. Effectiveness of an intervention to provide information to patients with hypertension as short text messages and reminders sent to their mobile phone (HTA-Alert). Aten Primaria 2004; 34: 399–405. [Spanish].

38.

Mbuagbaw

Thabane

Ongolo-Zogo

, et al. The Cameroon Mobile Phone SMS (CAMPS) trial: a randomized trial of text messaging versus usual care for adherence to antiretroviral therapy. PLoS One 2012; 7: e46909–e46909.

39.

Park

Howie-Esquivel

Chung

Dracup

. A text messaging intervention to promote medication adherence for patients with coronary heart disease: a randomized controlled trial. Patient Educ Couns 2014; 94: 261–8.

40.

Petrie

Perry

Broadbent

Weinman

. A text message programme designed to modify patients' illness and treatment beliefs improves self-reported adherence to asthma preventer medication. Br J Health Psychol 2012; 17: 74–84.

41.

Pop-Eleches

Thirumurthy

Habyarimana

, et al. Mobile phone technologies improve adherence to antiretroviral treatment in a resource-limited setting: a randomized controlled trial of text message reminders. AIDS 2011; 25: 825–34.

42.

Safren

Hendriksen

Desousa

Boswell

Mayer

. Use of an on-line pager system to increase adherence to antiretroviral medications. AIDS Care 2003; 15: 787–93.

43.

Simoni

Huh

Frick

, et al. Peer support and pager messaging to promote antiretroviral modifying therapy in Seattle: a randomized controlled trial. J Acquir Immune Defic Syndr 2009; 52: 465–473.

44.

Strandbygaard

Thomsen

Backer

. A daily SMS reminder increases adherence to asthma treatment: a three-month follow-up study. Respir Med 2010; 104: 166–71.

45.

Vervloet

van Dijk

Santen-Reestman

, et al. SMS reminders improve adherence to oral medication in type 2 diabetes patients who are real time electronically monitored. Int J Med Inform 2012; 81: 594–604.

46.

Wang

, et al. A randomized controlled trial to assess adherence to allergic rhinitis treatment following a daily short message service (SMS) via the mobile phone. Int Arch Allergy Immunol 2014; 163: 51–8.

47.

Haynes

Ackloo

Sahota

McDonald

Yao

. Interventions for enhancing medication adherence. Cochrane Database Syst Rev 2008(2CD000011–CD000011.

48.

Tao D, Or CK. Effects of self-management health information technology on glycaemic control for patients with diabetes: a meta-analysis of randomized controlled trials. J Telemed Telecare 2013; 19: 133–143.

49.

Farmer

. Methods for measuring and monitoring medication regimen adherence in clinical trials and clinical practice. Clin Ther 1999; 21: 1074–90. discussion 1073).

50.

Bryson

Rumsfeld

. Medication adherence: its importance in cardiovascular outcomes. Circulation 2009; 119: 3028–35.