This study examined the feasibility of a diabetes educational SMS program targeting glycemic control and self-management behaviors among Egyptian diabetics. Based on the opinion of our local doctors, patients needed to undertake healthy eating, regular physical exercise, and high rates of medication adherence in order to achieve better glycemic control. However, eating healthy, being physically active, and taking medications on time are only a subset of various self-management behaviors that diabetic patients need to maintain. Others include regular blood glucose monitoring, results recording, regular visits to diabetes doctor, commitment to diabetes follow-up tests, regular foot-checking, knowing what to do in cases of hypo- or hyperglycemia, knowing when it is necessary to see the doctor, quitting or reducing smoking, adjusting insulin dosage based on blood glucose readings and food intake, and always carrying a strong-acting sugar. At the time of preparation for this study, the authors were not able to identify other studies that addressed all these factors together. Studies usually focused on one or some of these behaviors or on glycemic control and its associated behaviors.
Our study contributes to the increasing mHealth body of literature suggesting that SMS interventions are feasible tools for diabetes management with great potential to improve clinical outcomes. In a study in the Netherlands that used SMS reminders to improve medication adherence over a period of 6 months, intervention patients took significantly more medication doses than control patients (50% vs. 39% within a 1-h window and 81% vs. 70% within a 4-h window) and had a 5% lower rate of missing their doses . In another study in the US, SMS messages were sent to a group of 18 patients for 1 month, addressing medication adherence and self-management behaviors such as foot checking and blood glucose monitoring. Missed medication doses significantly decreased from 1.6 to 0.6 per week, and diabetes self-efficacy significantly increased during the study (p = 0.002). Further, 89% of patients indicated increased frequency of foot self-examinations .
In Iran, a study showed that SMS messages were as effective as telephone calls in monitoring 77 Type 2 diabetic patients. SMS group patients (n = 38) received 4 messages per week on diet, exercise, medication taking, and frequent self-monitoring of blood glucose levels; and achieved 1.01% drop in the mean HbA1c level at 3 months. Telephone group patients (n = 39) achieved almost the same result (0.93% drop), yet the calls required more time and money than the SMS messages. No results were reported on diet, exercise, medications‚ or other self-management behaviors . A study in South Korea asked participants to enter their blood glucose levels, diet, and exercise diaries into a website on a daily basis, and accordingly sent them weekly SMS recommendations. Compared to baseline, the mean HbA1c level decreased by 1.15% at 3 months and 1.05% at 6 months for intervention patients (n = 25), and increased for control patients (n = 26) by 0.07% and 0.11% at 3 and 6 months respectively. Fasting plasma and 2 h post-meal glucose levels were also recorded (with intervention patients achieving significant declines by 85.1 mg/dl and 63.1 mg/dl at 3 and 6 months respectively) but no self-management behaviors were monitored .
In the SuperEgo study, 23 Type 1 diabetes adolescents received tailored SMS messages at an average of 10 messages/week based on their individually reported diabetes self-care. The messages addressed stress, exercise, communication, social support and stigma, time planning, reminders, and dietary portions. At 3 months, intervention patients maintained their mean HbA1c baseline levels while control patients significantly worsened (p = 0.006). Only usability and satisfaction were additionally evaluated but no self-management behaviors were addressed . In the Sweet Talk study, that also aimed to improve glycemic control among Type 1 diabetics, intervention participants received daily SMS messages on insulin injections, blood glucose testing, healthy eating, and exercise. Over 12 months, the mean HbA1c levels did not change significantly in control or intervention patients. Yet, diabetes self-efficacy as well as other self-management behaviors such as blood glucose testing, healthy eating, and exercise (all measured by a diabetes social support interview) were significantly better in the intervention group at the end of the study .
A study in Bahrain examined the effect of bidirectional SMS messages on participants’ glycemic control. Patients sent their inquiries about diet, medications, side effects, blood glucose levels, hypo- and hyperglycemia actions, and follow-up tests; and received immediate feedback from the medical doctor. Patients that did not send any inquiries for 7 consecutive days were sent SMS reminders. After 3 months, both intervention (n = 12) and control (n = 22) patients achieved significant declines in their HbA1c levels (−2.76% and −1.6% respectively) compared to baseline. Intervention patients, however, achieved a significantly higher reduction in their mean HbA1c, 1.16% points lower than that of controls. No results on self-management behaviors were reported . Another study in South Korea asked participants to enter their blood glucose levels daily into a website and sent them recommendations by SMS accordingly. After 6 months, the intervention group (n = 18) had a statistically significant decrease in HbA1c, fasting plasma glucose, and 2-h post-meal glucose levels compared to the control group (n = 16). A significant mean percentage change of −1.22 and −1.09 was recorded at 3 and 6 months respectively for the intervention group, while the control group did not show any significant changes. No self-measurement behaviors were measured .
In Norway, the impact of using the Few Touch Application (FTA) on self-management was examined. Participants were assigned either to an FTA (n = 51), FTA and phone health counseling (HC, n = 50), or a control (n = 50) group. The FTA consisted of a blood glucose measuring system, a diet manual, and a physical activity diary. After 1 year, the primary outcome (HbA1c) decreased in all groups but did not significantly differ between groups. Secondary outcomes including weight, depressive symptoms, and nutritional and exercise habits also did not change significantly between groups after 1 year .
The CareSmarts study in Chicago evaluated the impact of bidirectional SMS messages on HbA1c levels and self-management behaviors such as healthy eating, exercise, foot checking, and blood glucose testing over 6 months. Participants of the intervention group achieved a significant drop in their HbA1c levels (7.9 to 7.2%), and significantly improved their frequency of healthy eating, blood glucose monitoring, and performing foot checks. Medication adherence also significantly improved at the end of the study. No changes in clinical outcomes were observed in control group patients though and the main limitation was lack of randomization . Another study in the US assigned its participants to a text messaging group that received daily messages on nutrition and physical activity, and a control group that received a pamphlet on healthy lifestyle. The 1-month trial was too short to see significant changes in HbA1c levels, self-efficacy, or body mass index for any of its groups. However, satisfaction with the SMS messages was high .
In the TExtT-MED study, 128 patients were randomized to an intervention group (n = 64), that received 2 daily text messages in English or Spanish, and a control group (n = 64). The primary outcome was the change in the median HbA1c level. Secondary outcomes included changes in medication adherence, self-efficacy, diabetes knowledge, emergency department (ED) utilization, self-care tasks such as foot checking and blood glucose measurement, and patient satisfaction. After 6 months, the median HbA1c decreased by 1.05% in the intervention group as opposed to 0.6% in the control group, a change that was not considered significant. However, secondary outcomes showed considerable improvement particularly in medication adherence and ED utilization. The study’s restricted focus on ED patients was found to limit the generalizability of its findings .
Trends of improvement
In our study, the change in HbA1c from baseline did not differ significantly between groups after 3 months, yet a significantly higher number of intervention patients managed to achieve a 1% drop in their HbA1c levels. Trends of improvement were also observed in secondary outcomes, self-management behaviors, and education aspects. The reasons why particularly 16 intervention patients could achieve the 1% drop while the remaining 18 patients could not were not very clear. However, exploring the baseline characteristics of these patients brought some interesting facts to our attention. The most noticeable differences between achievers and non-achievers were in their baseline HbA1c values, sex, SMS familiarity, confidence in healthcare provider, and satisfaction with own body weight.
All achievers had baseline HbA1c levels above 8%, ranging from 8.2% to 15.4%. Non-achievers had a lower range (5.8 to 13.5%) and 9 of them had values already below 8%. Most male intervention patients were achievers (n = 11), while most females were non-achievers (n = 13). Of the 9 intervention patients that were not SMS familiar and had someone read the messages for them, only 2 achieved the drop while 7 were non-achievers. The majority of patients that indicated having low confidence in their healthcare providers were achievers (n = 12), while the majority that had higher levels of confidence were non-achievers (n = 11). Finally, most patients that indicated satisfaction with their body weight managed to achieve the drop (n = 12), whereas the majority of patients that were uncomfortable with their body weight were non-achievers (n = 14). Age, years of diabetes, social status, among other factors did not notably differ between achievers and non-achievers. However, most of the patients who achieved the drop were surprisingly aged over 50.
We likely have to differentiate between having good glycemic control and the practice of self-management behaviors. While eating, exercising, and medication taking behaviors are the ones that could have a direct impact on HbA1c levels, a person with good glycemic control is not necessarily one with good self-management and vice versa. Since studies mostly focus on one aspect or the other, this relationship is rarely observed . One of the main advantages of this study is that it measured several factors besides HbA1c and blood glucose levels, such as treatment and medication adherence, diabetes self-efficacy, and diabetes knowledge. Further, not only did it incorporate self-management behaviors within such factors or address them in the SMS categories, but it also produced individual scores for these behaviors to measure improvement. As both study groups were presented with an intervention (booklet vs. booklet + SMS), it should not be surprising to see progress in both groups. However, we were expecting greater improvement among intervention patients who were constantly reminded and motivated by SMS. Therefore, the effect of SMS can be seen by examining the difference in levels of improvement between groups rather than looking for improvement in the intervention group vs. no/slight changes in the control group.
Though the HbA1c levels did not differ significantly between groups at 3 months, considerable differences were observed in many of the secondary outcomes and the self-management behaviors (Table 2). All intervention patients performed daily foot checks during the study period (8 additional to baseline) vs. 79.45% of control patients (only 4 additional to baseline). Of participants who indicated suffering extreme increase or decrease in their blood glucose levels at 3 months, 13 control patients indicated they did not know which action to take as opposed to only 2 intervention patients. Patients’ ability to know when they have to see their diabetes doctor considerably differed between groups at the end of the study (scoring 2.26 vs. 3.05 on DES-SF), slightly dropping from its baseline score for controls and improving for intervention patients.
The mean rates of recording blood glucose measurements, visiting the diabetes doctor at preset times, and adhering to follow-up tests were poor for both groups at baseline, but became considerably higher for intervention patients at 3 months. The rate of adjusting the insulin dose based on food intake remained low for both groups at the end of the study, yet slightly improved for intervention patients and slightly declined for controls compared to baseline. Rate of carrying a strong-acting sugar somewhat improved for both groups but remained slightly higher for intervention patients at 3 months. Very few patients in both groups indicated satisfaction with own diabetes control at baseline. However, 41.18% of intervention patients vs. 20.51% of controls indicated satisfaction at endpoint. The HbA1c associated behaviors such as rate of following a healthy diet, being physically active, and adhering to prescribed medications produced remarkably higher scores for intervention patients at 3 months.
Possible causes for control group improvements
Our results appear to be in line with previous studies that mostly did not detect significant differences in HbA1c within 3, 6, and even 12 months; yet observed significant improvement in secondary outcomes when they existed. It is important to note, however, that studies that did find significance in HbA1c usually did not provide their control groups with any types of interventions (only usual care), thus achieving significance through improvement of their intervention group over non-changing or worsening control group. Further, the HbA1c improvement of such studies usually fell in the range of −1% within 3 to 6 months, a drop that was already achieved by our study’s intervention group after 3 months. Therefore, in order to explain why we could not achieve significance in HbA1c, we need to look into causes for improvement in our control group rather than barriers to greater improvement in the intervention group. The first cause, as mentioned earlier, is clearly that the control group was also provided with a form of intervention. Upon getting the instruction booklet, patients were expected to read it then put it aside for the rest of the study period or not read at all. However, intervention patients were constantly reminded by the instructions through regular SMS messages, and therefore they were expected to achieve significantly greater improvement. At endpoint, 20 controls vs. 30 intervention patients indicated reading the instruction booklet. We believe that significant differences could have been detected had we not given an instruction booklet to either groups or monitored them via a monitoring table.
The second cause maybe the monitoring table that controls were given and that also represented a form of intervention. Having this table and knowing that it would be reviewed in 3 months likely encouraged patients to check their blood glucose levels regularly and fill in their results. Further, looking at their recorded results possibly pushed them to adhere to treatment in order for the next measurement to show improvement. Between baseline and endpoint, the results recording mean SCI score increased from 1.38 to 2.90 for controls, and from 1.41 to 4.09 for the intervention group. This result was expected for intervention patients who were reminded on a weekly basis by SMS to check and record, but it was surprising for controls who were expected to also place the monitoring table aside upon receiving it, and forget to check and record every week.
The third reason may be the short period of the study. Given the poor communication our patients had with their doctors, sitting through the 30-min baseline interview, getting a chance to express their concerns about diabetes, feeling that they were being monitored and that they belonged to a program, and knowing that they would come back for another interview in 3 months were all sufficient reasons to keep all our patients motivated. Since the effect of our SMS intervention relied on sustained motivation, a period of 3 months was probably too short for the levels of motivation to decline given the above circumstances. However, had the study extended to a longer period, we would have expected the motivation of controls to drop while that of interventions to be maintained by receiving the daily SMS messages.
The fourth reason may be that many of our study participants knew each other. At the time of enrollment, upon learning about the benefits of the study, participants brought their diabetic family members and friends to also participate. Further, 13 controls and 19 intervention patients were already working together on the university campus, either in the hospital, administrative departments, or in one of the faculties. Therefore, it is highly likely that there was contact between members of the intervention and the control groups throughout the course of the study, particularly among those who worked in the same buildings (e.g. nurses, housekeeping, etc.). Consequently, the motivation that the SMS messages brought to intervention patients might have been transferred to some control patients leading to unexpected improvement.
SMS vs. traditional methods
Besides glycemic control and self-management, our study aimed to also address patient education. The SMS intervention remarkably improved knowledge aspects (Table 2) and showed high levels of acceptance over traditional education methods such as paper-based materials, lectures‚ or seminars. In general, designing and printing booklets or pamphlets require time and money, and patients are usually discouraged to take the time to read and process their load of information at once. Likewise, the organization of lectures and seminars requires costly efforts, and patients face the barriers of distance and time to attend them. In-clinic education during medical visits depends primarily on patients’ commitment to see their doctors regularly, and on doctors providing the time to have discussions with their patients. Further, patients are likely to forget or start ignoring their doctor’s tips a few days after the visit, or lose motivation to comply with treatment . In light of such circumstances, SMS messages present a low cost method that can deliver educational material to patients wherever they are. They do not require complicated smartphones and can be easily checked and read without any distance or time barriers. If sent regularly, they can also provide light information doses that can easily be read and understood at once, and sustain patients’ motivation to adhere to treatment.
In our study, given the poor levels of communication with healthcare providers, most patients were not receiving the most basic form of education in the clinic. Insured patients indicated they were given their medication doses every 2–3 months without even being checked by a doctor, and accordingly knew very little about their diabetes and how to control it. Very few patients indicated attending educational lectures or seminars (2 controls to 4 interventions) which were rarely organized. One of the notable advantages of our SMS intervention was the motivation it brought to patients to read the more detailed paper-based instructions, as only 51.28% of control patients indicated reading the booklet vs. 88.24% of intervention patients. Further, 55.88% of intervention patients preferred the SMS messages to the booklet, 20.59% said they liked having them both, while only 23.53% favored the booklet. Reasons for favoring SMS messages included their continuity (daily rate); their reminding nature; being short, simple and quick to read; mobile phone being always within reach; and the likelihood of being discouraged to read the booklet, losing it, or forgetting about it.
Four of 8 patients who preferred the booklet were not familiar with SMS. They indicated that it would be easier and less awkward to ask someone to read the booklet for them once rather than daily SMS messages. The other four preferred the booklet since it contained information that they did not have to wait for, it was always there but they sometimes had to delete the SMS messages, they did not have to look for the information in their phones, and they were concerned about not getting the SMS. Patients that preferred having both the booklet and SMS messages together indicated they integrated each other, since the booklet would be there to keep and refer to any time, and the SMS messages would act as daily quick reminders.
It was encouraging to see that patients’ opinion of our program was generally positive, and that satisfaction and acceptance of SMS messaging as a method of communication, education, and sustaining motivation was high. Of the 34 intervention patients, 27 indicated receiving and reading the SMS messages every day, one patient indicated missing a few as a result of spending some days in places with poor network coverage, and 6 indicated missing the majority of messages for the following reasons: 3 did not pay much attention to their phones, 2 indicated the person that was supposed to read for them was not available every day, and one patient often left their phone with someone else. All patients that got the SMS messages (n = 28) found no difficulty reading or understanding them and thought no information was missing in their content. Moreover, they appreciated the daily rate of SMS sending and thought the morning receipt times were appropriate. When our study participants (n = 73) were asked if they felt something was missing in the program, 21 indicated the following: Some wished we had provided more incentives such as free medication doses, glucometers, or additional lab tests; others wished our clinic’s doctor had changed their medication doses or prescribed them new ones; and a few indicated they did not know they could visit the clinic’s doctor free of charge during the time of the study and wished they had been examined.
The most obvious limitation in our study is its short period. We believe that 6 months is the optimal timeframe for studies looking for changes in glycemic control. However, due to budgetary constraints in addition to the increased risk of losing patients through longer periods, we restricted our study to 3 months. The study’s small sample size may also limit generalizibility, thus needing a larger RCT to establish clinical efficacy/effectiveness. Time delays during recruitment and overlaps between study phases could also be viewed as a limitation of the study. Unusual delays in ethical committee procedures resulted in prolonged waiting times between patient enrollment and beginning of baseline interviews. Offering incentives during this period could have minimized early dropouts and avoided initiation of a replacement phase. However, this was also not possible due to budgetary constraints. Baseline interviews and HbA1c tests were scheduled to take place at one appointment. Yet, as patient randomization was still in process, we postponed the collection of HbA1c samples to avoid large time gaps between baseline testing and start of intervention. On the other hand, it was necessary to start contacting enrolled patients and ask them to come back in order to avoid further dropouts. Therefore, splitting the baseline appointment into two meetings, one for the interview and pre-study questionnaire and another for the HbA1c samples, was highly recommended.
Given the amount of paperwork, approvals‚ and signatures that were needed from university administration, hospital management, faculty members, or participating team members through the different stages of the study, following a strict time plan was not always possible. Consequently, the delays in all study phases led to a 2-week overlap of the study period with the month of Ramadan, which forced us to start our final interviews 2 weeks ahead of schedule. Though this might not have had a noticeable impact on our results, it could still be seen as a limitation of our study. Lack of incentives could also be seen as a limitation. Due to budgetary constraints, we could only provide the HbA1c tests and visits to the clinic’s doctor free of charge. We believe, however, that other incentives such as free medication doses, glucometers, phone credit, or other lab tests could have minimized the dropout rate. Limiting our selection criteria to residents of October city could have also minimized patient dropout. Further, selecting only patients that had no contact with each other or patients that were SMS familiar could have produced better results. We chose, however, to keep our criteria unrestricted in order to attract more patients, especially that those that visit MUST hospital public clinics are usually not very well educated, and accordingly not good readers of SMS messages.
Suggestions for future trials
Though our study followed an RCT methodology, seeking to assess the impact of the SMS educational intervention on health outcomes, its sample size and duration may not have been sufficient to indicate clinical efficacy/effectiveness. The study, however, establishes guidelines for future larger and longer RCTs targeting changes in HbA1c and self-management behaviors. For instance, a longer study is recommended to last at least 6 months in order to see significant differences in HbA1c, and preferably go to a year to allow for changes in behaviors to be observed. Further, stratifying patients into ones with HbA1c values of 8% or lower as opposed to those with higher values at baseline would be highly recommended, as such patients responded differently to our SMS intervention. In our short study, it was not possible to thoroughly evaluate rarely addressed behaviors such as visiting the diabetes doctor regularly or adhering to follow-up tests. This is because patients are generally advised to visit their doctors every 3 months, take a kidney function and a urine test every 6 months, and have an eye exam and a lipid profile test once a year (personal communication with Dr. Mohamed AlaaFootnote 4). Other behaviors such as following a healthy meal plan and abstaining from smoking also need a longer time frame to be better evaluated.
Future studies are additionally advised to avoid overlapping with the month of Ramadan due to the effects of the special lifestyle of the month on diabetes, which could make it difficult to assess the impact of the SMS intervention independently. However, this could be challenging for studies aiming to last for 1 year, thus they are recommended to start shortly after Ramadan to avoid extended overlaps. Due to its short period, our study could only compare patients’ satisfaction with their healthcare providers before joining our program to their satisfaction with the program. A longer study could perform a more accurate comparison by adding an essential message category addressing the doctor-patient relationship; educating patients on how to maintain a good relationship with their doctors, what to do or inquire about at their medical appointments, and what to expect from their doctors; then finally compare their satisfaction with their healthcare providers before and after receiving such education. A message category on calculating insulin dosage and adjusting it based on food intake should also be incorporated, as this was not commonly known or done by insulin users in our study.