A Randomised Controlled Trial of an Sms-Based Mobile Epilepsy Education System

Introduction

Epilepsy is a disorder associated with a high degree of stigma and prejudice, and the success of its treatment depends very much on the patients’ treatment compliance and their understanding of the disorder. ¹ Although information is easily available through member organisations, brochures and the Internet, information-seeking is a less preferred coping strategy than religion and emotional support among people with epilepsy (PWE) in Malaysia. ² Therefore, it would probably be more effective if healthcare providers could deliver the information via a tool such as the mobile telephone.

Text messaging (short message service, SMS) allows messages to be delivered almost immediately to the recipient's telephone once it is switched on, and the messages can be read whenever convenient. However, this technique has not been used in Malaysia to deliver epilepsy education. We therefore developed a mobile epilepsy educational system (MEES) to deliver individualised health education via SMS, with the aim of improving awareness, knowledge and attitudes (AKA), medication adherence and clinic attendance rate among PWE.

Methods

A randomised controlled trial was conducted in the neurology clinics of three public hospitals located in the states of Terengganu, Pahang and Kelantan in Malaysia. The study was approved by the appropriate ethics committee. Patients had to be 18 years old and above, to be on regular treatment, able to either write, read or understand and communicate in Malay or English, capable of completing questionnaires (either written or verbal), and be owners and active users of mobile telephones. Patients were excluded if they were younger than 18 years old, not on regular treatment, unable to write, read, understand or communicate in Malay or English language, incapable of completing questionnaires (neither written nor verbal), did not own a mobile telephone, were non-active users or failed to provide written consent for participation.

The sample size calculation was based on a level of significance of 0.01 with 90% power. The required sample size per group was 54. For two groups, the sample size needed was 108. After estimating a 20% drop-out, the final sample size required for the study was 130, i.e. 65 patients per group.

Recruitment began in December 2010 and ended in December 2011. A hospital coordinator was trained to help with patient identification and medical record management. Information was collected on the epilepsy clinic day to facilitate patient convenience. On the agreed meeting day, potential patients who met all the inclusion criteria were approached and invited to participate in the study.

Once agreed, participants were randomly assigned to either control or intervention group via an Interactive Voice Response System (IVRS). At the baseline, all patients completed the Personal Information Form, Malay Awareness, Knowledge, Attitudes Epilepsy ⁵ and Malay Modified Morisky Medication Adherence Scale (MMMMAS).^6–8 Patients in the control group (CG) received an 11-part printed epilepsy educational module plus a user manual. A total of 12 weeks were needed to complete the entire module (one part per week). Besides receiving the printed module, patients in the intervention group (IG) were supplied with additional text messages from the MEES for a 3-month period. An introductory text message marking the beginning of the MEES services was sent to patients a day after recruitment. With 4 days between each SMS, patients were continuously supplied with snippets of epilepsy education information. At the end of the 3-month period, a text message was sent to patients thanking them for participating in the study.

After three months, all patients were re-contacted via telephone by the research assistants for follow-up. During the follow-up session, patients completed the same set of forms as during baseline, except for the Personal Information Form, see Figure 1. OPEN IN VIEWER

Results

Of 677 epilepsy outpatients in the three public hospitals, only 335 were eligible for the study. Further screening resulted in 191 being excluded (178 did not meet the inclusion criteria while 13 refused participation). A total of 144 PWE were subsequently recruited at baseline but only 136 PWE managed to complete both the baseline and follow-up phases in the study. The average age of the patients was 31 years (SD 12). Most of them were Malays (91%), suffering from generalized seizure (53%), unmarried (60%), possessed an educational qualification not more than SPM/Cambridge O-level (76%), were unemployed (31%) and earned a monthly income not more than MYR 1500 (approx US$ 478) (93%) and had suffered from the illness for more than five years (51%), see Table 1.

OPEN IN VIEWER

Table 1

Patient characteristics (n = 144)

	All patients (n = 136)			Control group (n = 65)		Intervention group (n = 71)
	n	%	P- value *	n	%	n	%	P- value †
Gender
Male	73	51	>0.05	39	54	34	47	>0.05
Female	71	49		33	46	38	53
Ethnicity
Malay	131	91	<0.01	63	88	68	94	>0.05
Chinese	9	6		7	10	2	3
Indian	3	2		1	1	2	3
Others	1	0.7		1	1	–	–
Educational qualification
Up to SPM/Cambridge O’ level	110	76	<0.01	60	83	50	69	>0.05
Better than SPM/Cambridge O’ level	34	24		12	17	22	31
Employment
Unemployed	44	31	<0.01	31	43	13	18	<0.05
Self-employed	25	17		11	15	14	19
Supportive worker	28	19		9	13	19	26
Housewife	18	13		8	11	10	14
Student	22	15		8	11	14	19
Others	7	5		5	7	2	3
Monthly income
≤MYR1500/US$ 478	134	93	<0.01	66	92	68	94	>0.05
>MYR1500/US$ 478	10	7		6	8	4	6
Marital status
Single	87	60	<0.01	45	63	42	58	>0.05
Married	49	34		22	31	27	38
Divorced	8	6		5	7	3	4
Epilepsy classification
Generalised seizure	76	53	>0.05	40	56	36	50	>0.05
Partial seizure	68	47		32	44	36	50
Living area
Urban	74	51	>0.05	35	49	39	54	>0.05
Rural	70	49		37	51	33	46
Duration of illness
6–12 months	29	22	<0.05	16	24	13	19	>0.05
13–24 months	11	8		8	12	3	5
25–60 months	26	19		14	21	12	18
61–120 months	38	28		14	21	24	36
>120 months	31	23		16	24	15	22

*

Chi-square test for goodness of fit

†

Chi-square test for relatedness

Of 26 patient queries, eight required further expert advice from either doctors or neurologists. These queries were mostly about basic knowledge (n = 5), history and statistics (n = 2), living with epilepsy (n = 7), myth and facts (n = 2), self-control (n = 3), prognosis (n = 3), treatment/therapy (n = 8), psychosocial aspects (n = 1), laws and acts (n = 1), resources (n = 1) and other trivial issues (n = 3).

Between-group comparisons

There was no significant difference between the groups at baseline for all AKA domains (P ≥ 0.05), see Table 2. However, during follow-up, intervention patients exhibited significantly better Awareness, Knowledge, Attitudes and Total AKA than the control patients (P < 0.05). Among all AKA domains, for both groups Attitudes exhibited the highest score while Awareness emerged as the lowest. In addition, there were no significant differences in medication adherence for both groups at baseline and follow-up (P ≥ 0.05).

OPEN IN VIEWER

Table 2

Between-group comparison of AKA profiles and medication adherence. Values shown are mean (SD)

	Baseline (n = 144)			Follow-up (n = 136)
	Control group (n = 72)	Intervention group (n = 72)	P-value*	Control group (n = 65)	Intervention group (n = 71)	P-value*
Awareness	3.9 (2.4)	3.8 (2.4)	>0.05	4.2 (2.3)	5.9 (3.1)	<0.01
Knowledge	5.2(1.6)	5.5(1.8)	>0.05	6.8(1.8)	7.5(1.9)	<0.05
Attitudes	6.5 (2.9)	7.2 (2.6)	>0.05	7.5 (2.3)	8.4 (2.0)	<0.05
Total AKA	5.2(1.5)	5.5(1.5)	>0.05	6.2(1.6)	7.3 (2.0)	<0.01
Medication adherence	0.78 (0.22)	0.72 (0.21)	>0.05	0.83 (0.56)	0.80 (0.22)	>0.05

*

Independent t-tests

Within-group comparisons

After intervention, there were significant improvements in Knowledge, Attitudes and Total AKA (P < 0.05) in control patients with Knowledge exhibiting large effect sizes (Cohen's d = 0.95). There were significant improvements in the intervention patients in all domains during follow-up (P < 0.01) with large effect sizes in Awareness, Knowledge and Total AKA (Cohen's d = 0.90–1.19). Compared to the control group, medication adherence was significantly better in the intervention group after intervention (P < 0.01), see Table 3.

OPEN IN VIEWER

Table 3

Within-group comparison of AKA and medication adherence

	Control group					Intervention group
	Mean (SD)					Mean (SD)
	Baseline (n = 72)	Follow-up (n = 65)	Mean change	P- value*	Effect size	Baseline (n = 72)	Follow-up (n = 71)	Mean change	P- value*	Effect size
Awareness	3.9 (2.4)	4.2 (2.3)	0.31	>0.05	0.13	3.8 (2.4)	5.9 (3.1)	2.14	<0.01	0.90
Knowledge	5.3(1.6)	6.8 (1.8)	1.53	<0.01	0.95	5.5(1.8)	7.5(1.9)	2.03	<0.01	1.12
Attitudes	6.4 (2.8)	7.5 (2.3)	1.17	<0.01	0.42	7.2 (2.6)	8.4 (2.0)	1.13	<0.01	0.44
Total AKA	5.2(1.5)	6.2 (1.6)	1.01	<0.01	0.69	5.5(1.5)	7.3 (2.0)	1.76	<0.01	1.19
Medication adherence	0.78 (0.23)	0.83 (0.56)	0.05	>0.05	0.22	0.72 (0.21)	0.80 (0.22)	0.08	<0.01	0.38

*

Paired t-tests

Analysis of covariance

The results of ANCOVA for AKA scores and medication adherence from baseline to follow-up in the two groups are summarised in Table 4. There was significantly higher AKA in the intervention group compared to the control group, particularly in Awareness (P < 0.001) and Total AKA (P = 0.003) after controlling for potential confounding variables. Awareness had the largest adjusted effect size (Cohen's d = 0.87).

OPEN IN VIEWER

Table 4

ANCOVA analysis between groups after controlling for potential confounding variables

	Adjusted mean (95% CI) a	Adjusted mean difference (95% CI) b	P-value a	Adjusted effect size
Awareness		2.3 (1.2, 3.3)	<0.001	0.87
Control	3.3 (1.6, 5.0)
Intervention	5.5 (3.8, 7.3)
Knowledge		0.7 (-0.1, 1.5)	0.070	0.38
Control	6.8 (5.6, 8.0)
Intervention	7.5 (6.2, 8.8)
Attitudes		0.4 (-0.3, 1.1)	0.276	0.23
Control	7.4 (6.3, 8.6)
Intervention	7.8 (6.6, 9.0)
Total AKA		1.0 (0.4, 1.7)	0.003	0.64
Control	5.8 (4.7, 6.8)
Intervention	6.8 (5.7, 7.9)
Medication adherence		0.1 (-0.1, 0.2)	0.591	0.11
Control	0.7 (0.5, 1.0)
Intervention	0.8 (0.5, 1.1)

a

djusted mean using ANCOVA after controlling for age, gender, educational qualification, monthly income and baseline mean for each variable

b

Bonferroni adjustment for 95% CI for difference

Clinic attendance

Clinic attendance was not examined at baseline but during follow-up, the intervention patients had significantly better attendance at 90% in comparison to control patients who had 63% (P < 0.05), see Table 5.

OPEN IN VIEWER

Table 5

Clinic attendance at follow-up (n = 136)

	Overall (%)		Control group (%)		Intervention group (%)
	Attendees	Non-attendees	Attendees	Non-attendees	Attendees	Non-attendees	P- value *
Clinic attendance	105 (77)	31 (23)	41 (63)	24 (37)	64 (90)	7(10)	<0.05

*

Chi-squared tests for relatedness

Discussion

Hanlon and collegues ¹⁰ found that after receiving health education, more than half of the participants changed one or more of their behaviours according to the advice given. Patients in their study also mentioned that the provision of information about their personal risk caused mild anxiety, and that advice and support could act as a motivating factor for behavioural change. Other studies have indicated that those who join a health promotion programme tend to be more health conscious, physically active, exposed to less health risk, adherent to prescribed medical regimens and demonstrate less absenteeism during clinic appointment than individuals who choose not to participate. ¹¹ Thus both the printed epilepsy education module and the MEES could play a role in improving the psychosocial functioning of PWE and bringing about behavioural changes to cope better with the disease.

In comparing these two education interventions, even after controlling for potential confounders, the combination of SMS-based MEES and printed modules was clearly more effective than the conventional printed modules alone in improving AKA, especially Awareness. Continuous supplies of SMS containing epilepsy-related information reinforced the knowledge already obtained from the printed module and probably caused the PWE to be more aware and alert about their illness. After being delivered to patients’ mobile phones, the SMS messages could be read at their convenience. Being fast, simple, and flexible were some of the attractive features of the MEES. Another advantage of this system was the possibility of delivering large batches of text messages almost instantly, reducing labour cost. ¹² SMS also do not require the mobile phone to be switched on and can be held for a number of days until the phone becomes active again. ¹³ Furthermore, SMS maintains some level of privacy, which was very reassuring for the patients.

Medication adherence among PWE remains a significant challenge, even for patients prescribed with newer antiepileptic drugs (AEDs), because non-adherence can result in increased risk of seizures, hospitalisation and higher treatment costs. ¹⁴ Because of the drug-taking reminders, the intervention patients reported better medication adherence than the control patients. We also found that SMS reminders were effective in improving attendance in primary care compared to the control group with no reminders, and were cost-effective ¹⁵ .

The present study had certain drawbacks. Because there was no clinic attendance record for epilepsy outpatients during the baseline period, within-group analyses for clinic attendance at baseline were not possible. In Malaysia, most of the primary care clinics do not have an appointment system and therefore the participants in our study may not have been familiar with keeping to appointments.

In conclusion, the present study suggests that the addition of the MEES to conventional epilepsy education is effective in improving AKA (especially Awareness), medication adherence and clinic attendance among PWE. This may alleviate the negative perceptions associated with epilepsy while simultaneously improving health outcomes.