Effects of and satisfaction with short message service reminders for patient medication adherence: a randomized controlled study

Study design

The study was conducted between November 1, 2010 and October 31, 2011. After we obtained approval from the Institutional Review Boards of the study hospitals (T.V.G.H., 201006021IC; C.M.U.H., 98–08–01A; and W.H., 100008), hospital pharmacists randomly assigned two patients to the intervention group and one to the control group when patients met the inclusion criteria. We used the systematic sampling method to assign qualified patients to each group. Odd-numbered or even-numbered patients from the pharmacy department registration were assigned to each group. Subsequently, trained interviewers conducted one-on-one interviews with the potential participants, explaining the research purpose and methods in outpatient settings such as the outpatient dispensary waiting area. After agreeing to participate in the study, the patients completed a questionnaire-based pretest.

Patients in the intervention group received SMS reminders reminding them to take their medication at specific times for 7 days beginning on the second day of enrollment, and could not choose the times the reminders were sent. On the eighth day after completing the intervention we conducted a phone-based survey to assess patient medication compliance and demand for and satisfaction with the “text message medication reminder service.” Figure 1 shows the study screening flowchart.

Participants

Inclusion criteria for participants were: (a) literate patients aged 20 years or older able to communicate with the investigators; (b) having a prescription longer than 7 days; (c) possessing a cell phone and knowing how to receive text messages; (d) and having at least one medication nonadherence (delayed or missed dose) experience in the past 3 days of taking medicine.

We recruited participants from three medical centers based on the large outpatient volume in each hospital. Because we focused on the effect of using SMS in the intervention group, we deliberately increased the sample size of this group. The sample sizes of the intervention and control groups were allocated at a ratio of 2:1.

Intervention

After patients signed and submitted informed consent forms, they completed a questionnaire-based pretest. We then collected patients’ cell phone numbers to facilitate convenient post-test phone interviews. Patients’ identities were recognized using hospital information system (HIS) databases; their prescriptions were obtained from the clinical department they visited during recruitment and exported to the SMS system for sending reminders. Beginning on the following day, medication reminders were sent to participants by using the SMS system; text messages were sent only to participants in the intervention group.

The SMS content depended on participants’ medication and frequency of medication use. If a participant used multiple medications, he or she received reminders for each. The messages comprised varying content such as visitation date, hospital name, medication name, dosing frequency, dose, and administration methods. The first message contained complete medication information. For example, “A warm reminder! You visited the xx clinic at xx Hospital on 2010/3/15. We would like to remind you to take your prescription, which includes Panadol® (one tablet per day for 7 days), Lasix® (0.5 tablet per day immediately following a meal for 7 days), and Lontex® (0.5 tablet once per day for 7 days). Please remember to take this medication on time. We wish you a prompt recovery.” The second and subsequent messages contained only simple medication reminders.

Measurements

The research tools comprised an investigator-designed questionnaire that involved a pretest and post-test. The pretest comprised basic demographic questions (e.g., sex, age, and education level), medical conditions, and self-reported medication use experiences (frequency of daily medication use, delayed doses, or missed doses during the previous 3 days). If participants did not take medicine and passed more than a half-period of time between two doses, we defined it as a “missed dose.” For example, if participants needed to take medicine twice a day (i.e., once every 12 hours), and took medicine more than 6 hours late, it qualified as a missed dose. If participants did not take medication as prescribed regarding timing or dosage, but took the medicine within 6 hours, we defined it as a “delayed dose.”

The post-test questionnaire was conducted by phone interview on the eighth day after the study. The post-test questions comprised structured items for the participants’ self-reported adherence to medication during the previous 3 days and their demand for and satisfaction with the SMS medication reminder service. Open-ended questions were also included to obtain participants’ recommendations regarding the service. The interviewers also asked the participants whether they would recommend the service to their family or friends.

To verify the validity of the questionnaire, five experts performed a content validity test, evaluating whether the questionnaire items conformed to the study topic. The average content validity index (CVI) obtained was 0.9. Regarding reliability, the Cronbach’s α coefficient for internal consistency reliability was 0.9, indicating good questionnaire reliability.

Data analysis

We used SAS statistical software to manage and analyze the data obtained from the questionnaires and descriptive statistics to determine the frequency, percentage, and mean values of each demographic variable. Regarding inferential statistics, we performed chi-square tests to identify differences between the groups regarding improvements in medication adherence and the demand for and satisfaction with the SMS medication reminder service. We used McNemar’s test to compare the pretest and post-test data within each group to examine distinct decreases in the incidence of delayed or missed medication doses before and after the intervention. Because the number of participants recruited from the three study hospitals differed and the management context and service model of the hospital at which most participants were recruited could affect the results of the intervention, we used weighted logistic regression to examine the factors that influenced improvements in the incidence of delayed or missed doses. We used odds ratios (OR) to describe the effect sizes.

Baseline data

The intervention group and control group comprised 780 and 460 participants at the time of the pretest, respectively. The number of valid participants who completed both surveys was 763 in the intervention group and 435 in the control group. We obtained 1,198 total questionnaires from the three study hospitals. In each hospital, we collected 188, 225, and 350 questionnaires for the intervention group and 117, 118, and 200 questionnaires for the control group.

The link between the SMS intervention and medication adherence

Factors that influence decreases in the incidence of delayed doses

Regarding medical history, participants with heart disease, hypertension, stroke, and thyroid disorder were less likely to have a decreased incidence of delayed doses compared with participants who lacked such diseases. However, participants with peptic ulcers, cancer, or immunological diseases were significantly more likely to experience a decrease in the incidence of delayed doses compared with participants who lacked such diseases.

Factors that influence reduced incidence of missed doses

Impression of and satisfaction with the SMS intervention

SMS improves medication adherence

We used weighted logistic regression to examine the factors that influence improvement in the incidence of delayed and missed medication doses. The results showed that the SMS intervention significantly decreased the incidence of delayed and missed doses. This was consistent with a previous finding that indicated SMS interventions enabled patients to consume their medication on time [18]. Regarding age (Table 4), participants aged 65 years or older were significantly less likely to experience decreases in the incidence of missed doses compared with those aged 20–34 years (OR = 0.6, 95% CI = 0.4–0.9). According to related literature, senior patients are often more resistant to behavioral change and are likely to stop taking medication based on personal decisions [19]. In addition, senior patient may be less familiar with cell phones compared with younger patients, increasing their likelihood of ignoring text message reminders, thus limiting decreases in the incidence of missed doses. However, comparing Tables 3 and 4 shows that when senior participants neglected or delayed taking medication, after they received SMS reminders they had a higher likelihood of taking medication than did those who missed taking the medication. This suggests that if senior participants delayed taking medication, when they received reminders they had a higher likelihood of improving their adherence than did those who lacked a strong intention to take their medication and missed taking it despite the reminder. Military personnel, civil servants, teachers, and students demonstrated similar behaviors. These results were similar with those of a previous study [20].

Regarding occupation, military personnel, civil servants, teachers, and students were less likely to experience decreases in the incidence of missed doses compared with unemployed participants. This could be because the military personnel, civil servants, teachers, and students in our study were typically young or middle-aged adults who demonstrated relatively better health compared with unemployed participants; therefore, these participants might consider long-term medication use unnecessary.

Regarding medical history, participants with hypertension, heart disease, stroke, or thyroid disease were less likely to experience decreases in the incidence of delayed or missed doses compared with participants who lacked such diseases. This finding was consistent with that reported by a previous study [20]. This study indicated that hypertensive patients tended to have confidence in their ability at self-control and were likely to adjust their medication consumption behavior arbitrarily, leading to poor adherence [20]. Therefore, SMS interventions may not reduce the incidence of delayed or missed doses among specific patient groups. A previous study also emphasized that patients with chronic diseases may have less motivation to consume medication regularly if their condition did not seem to improve, was incurable, or yielded side effects because of long-term medication use [21]. Thus, SMS interventions may fail to significantly alter the medication consumption behavior of patients with chronic diseases.

Satisfaction with and demand for the SMS intervention

Numerous observational studies have focused on satisfaction with and demand for SMS interventions [22–25]. In the current study, according to the results of participants’ satisfaction with the SMS intervention, the item “frequency of the SMS” received the lowest satisfaction score of 3.1 points. Participants reported that “the frequency of text messages sent was too high,” “text messages did not need to be sent often,” and “the number of messages was excessive.” These responses indicate that although the SMS intervention could help remind participants to consume medicines, it could also induce negative perceptions. Regarding participant demand for the SMS intervention, most participants (44.2%) wished to receive a reminder 30 minutes before the time medication should be consumed (Table 5). However, 93.2% of participants disliked when a message was resent after the time medication should be consumed to ensure that they would immediately consume a missed dose. Approximately 91.6% of the participants were willing to recommend the SMS intervention to their family and friends, suggesting that the SMS reminder intervention was effective. Of the participants who stated that the SMS reminders did not reduce delays or missed doses, most were women (72.2%), married, and housekeepers. We speculate that because these participants had more flexible time to care for themselves and may already have been adhering to their medication satisfactorily, they did not perceive that the SMS intervention enhanced their medication adherence.

Previous studies that have adopted SMS interventions for increasing medication adherence all showed that patients in the intervention group experienced higher treatment effectiveness compared with control groups. This could be because SMSs pose a minimal interruption to patients’ lives and are low in cost [8]. Compared with other approaches, SMS is simpler and more satisfying for users. Additionally, because the SMS reminder service can prompt patients to be responsible for their own health, it serves a vital function in healthcare services.

Limitations

Medication adherence problems are typically related to use of long-term medications for chronic diseases. In the current study, participants were monitored for only 7 days. Such a short-term follow-up might not properly interrogate the relationship between long-term medication adherence and use of SMS. Because we did not design the study to specify the medications used by participants, we could not evaluate the effects and outcome of their medication treatment. Additionally, the daily SMS system sent more than one reminder to participants who took several kinds of medications based on various medication schedules, which resulted in inconveniencing these participants. Many studies have shown that medication adherence outcome data that are purely reliant on self-reporting have a high likelihood of reporting bias [26, 27]. Because overstimulation might exist in both the intervention and control groups, the current study included two groups (intervention vs. control) that completed pretests and post-tests to reduce the effects of reporting bias.