OBJECTIVE: The goal was to improve immunosuppressant adherence for pediatric patients with orthotopic liver transplants by using text messaging (TM).
METHODS: A prospective study of sending TM reminders to the primary medication administrator (patient or caregiver) for pediatric transplant recipients was performed. Patient records were reviewed, comparing the year before and the year of the study. The SD of serum tacrolimus levels was used as an indicator of adherence.
RESULTS: Forty-one patients provided consent. The median age was 15 years (range: 1–27 years), and the median age at the time of transplantation was 2 years (range: 4 months to 23 years). Fourteen patients (34%) were male. In 29 of 41 cases, the medications were self-administered by the patient. The mean duration of study was 13 ± 1.5 months. Twenty-two patients were receiving 1 immunosuppressant, 14 were receiving 2, and 5 were receiving 3. Thirteen patients (37%) stopped the study after 4 months. The mean tacrolimus level SD decreased from 3.46 μg/L before the study to 1.37 μg/L (P < .005). The number of immunosuppressants taken and patient self/caregiver medication administration did not significantly affect the results. The number of acute cellular rejection episodes decreased from 12 to 2 during the study. Risk factors for rejection were older age (17.67 vs 13.28 years) and administration of >1 immunosuppressant.
CONCLUSION: We observed significant improvement in medication adherence and a reduction in rejection episodes with TM reminders for pediatric recipients of liver transplants.
- liver transplantation
- text messaging
- short messaging service
- cellular telephone
Successful treatment of patients after orthotopic liver transplantation (OLT) requires regular intake of medication. The long-term outcomes (ie, >5 years after transplantation) for adolescents are significantly worse than those for younger transplant recipients,1 and nonadherence was noted for up to 40% of adolescent transplant recipients.2 Adherence is defined as the degree to which a person's behavior corresponds to the agreed-upon recommendations of a health care provider.3 Nonadherence may include not taking medication at the prescribed frequency, dose, or time, and it may be deliberate or unintentional. Nonadherence has been associated with severe morbidity (eg, acute or chronic rejection or end-stage liver disease), hospitalizations, increasing costs, retransplantation, and even death.4,5
The most commonly reported reason for nonadherence among transplant recipients is forgetfulness.6 Patients assessed for OLT showed marked memory impairment, compared with healthy participants, and memory function remained inferior to that of healthy control subjects even after OLT.7 Various methods have been used to assess adherence, including subjective measures such as questionnaires and self-reports and objective measures such as pill counts, refill records, electronic measuring devices,8 and medication level measurements. There is no widely accepted, standard method for assessment of adherence among transplant recipients, and a combination of measurements may increase diagnostic accuracy.3
Improving medication compliance is essential to improving long-term outcomes for pediatric transplant recipients. Current methods of improving adherence for chronic health problems mostly are complex and not very effective. Interventions aimed at improving adherence include increased frequency of clinic visits for nonadherent adolescent patients,9 behavioral and multicomponent interventions for chronically ill youths,10 and an educational/behavioral intervention for adult kidney transplant recipients.11
Text messaging (TM) (short messaging service) is an integral part of teenaged culture and is inexpensive, personal, discreet, and accessible. Automated, scheduled TM was found to improve hemoglobin A1c levels for pediatric patients with diabetes mellitus type 1 who were receiving intensive insulin therapy,12 was efficacious for self-monitoring of healthful behaviors in children,13 and may be helpful in controlling asthma.14 The aims of the study were to pilot-test the effects of medication reminder TM on patients' adherence and outcomes.
Study Group and Study Design
All transplant recipients monitored in the pediatric liver transplantation/hepatology clinic were considered eligible and were initially screened for participation. There were no age inclusion/exclusion criteria. Our clinic routinely monitors patients in the pediatric service until the age of 22 years. However, there have been a few exceptions because of psychosocial or financial (mostly insurance) circumstances. Inclusion criteria included ownership of a mobile telephone with TM capability, use of medications for which levels can be measured (tacrolimus and/or sirolimus), and medically stable condition. Inability to understand the study was an exclusion criterion. The study was approved by the institutional review board and involved informed consent from both patients and caregivers.
Assessment of who is responsible for medication administration (caregiver or child) was performed at the initial enrollment. After signing informed consent forms, the patients or their caregivers registered on-line for the MediM system (CareSpeak Communications, East Brunswick, NJ). The data were secured with the patient's username and an individual password (which was required for logging on). Participants were required to enter the patient's name and cellular telephone number, the caregiver's name/nickname (eg, mom) and cellular telephone number, the medication name and frequency (eg, once or twice per day), and the exact times they wanted the alerts to arrive. The medication dose was not included, because the dose might change often on the basis of medical condition and laboratory test results. TM initially was to the person responsible for medication intake (patient or caregiver). The elapsed time before the caregiver was to be notified if intake was missed was entered.
The messages were structured as follows: “Take [name of medication] at [set time]. To confirm intake, press REPLY, type CARE 1, and press SEND.” The rates of responses to the messages were recorded by the system. If the patient did not respond to the message within a time frame set individually (15 minutes to 1 hour), then a message was sent to the caregiver, indicating that the patient did not respond. All data were stored on the company's servers, which have 256-bit SSL encryption (GeoTrust, Mountain View, CA). The research team could evaluate the participants' compliance rate. The TM process is summarized in Fig 1. Patients and caregivers were reimbursed for all TM costs associated with the research for the duration of the study.
Our group and others previously evaluated the use of SDs of consecutive tacrolimus blood levels as a measure of adherence for pediatric liver transplant recipients.5 Greater fluctuation between individual blood levels indicates lower levels of adherence. A SD of 2.5 μg/L is associated with rejection episodes.15 Serum medication levels were measured for patients in medically stable condition. Results from 1 year before the study and 1 year in the study were compared. Blood samples are drawn every 1 to 2 months; therefore, the average sample size was 8 samples per year.
All biopsy-proven, acute rejection episodes were recorded.
Analyses used SPSS 12.0 (SPSS Inc, Chicago, IL). Statistical tests were 2-tailed, and a P value (α) of ≤.05 was chosen as the level of statistical significance. Preliminary Kolmogorov-Smirnov tests were conducted to verify that the tacrolimus level SDs followed a normal distribution. On the basis of those results, parametric tests were used for subsequent analyses. Paired-sample t tests were used to compare tacrolimus levels measured before and after the intervention, with an intent-to-treat analysis; χ2 tests were used to compare values for categorical measures (eg, rejection episodes) before and after the intervention. Repeated-measures analysis of variance was used to examine whether there were differences between subgroups of patients (eg, those taking multiple immunosuppressants versus monotherapy).
Sixty-six transplant recipients were screened for participation, and 41 patients were recruited to participate in the study (Fig 2). Demographic characteristics of the sample are described in Table 1. The median age at study entry was 15 years (range: 2–27 years). Twenty-eight patients were <18 years of age, and 2 patients were >22 years of age. Most patients had undergone OLT in early childhood because of extrahepatic biliary atresia, were nonwhite, and were responsible for their own drug administration. Immunosuppressive drug regimens are described in Table 1.
For the 41 transplant recipients, the mean tacrolimus level SD before study participation was 3.46 μg/L (SD: 2.17 μg/L) and that after participation was 1.37 μg/L (SD: 1.01 μg/L; t = −6.07; P < .005) (Fig 3). The number of patients with tacrolimus level SD values above threshold (SD > 25) decreased from 24 to 6 (χ2 = 1.74; P = .19). Six patients continued to have tacrolimus level SD values of >2.5 μg/L after the intervention. We did not detect any risk factors for ongoing noncompliance among those 6 patients. Five of the patients in the study group were receiving sirolimus as adjuvant immunosuppressant. The mean sirolimus level SD decreased from 5 μg/L before the study to 1.8 μg/L during the year of the study (P = .01).
Seventeen patients (41%) dropped out, at a mean of 4 months. No risk factors for dropout were detected. The average rates of compliance recorded during study participation were 69% and 48% for the patients who continued the study and those who dropped out, respectively. The mean tacrolimus level SD values before the intervention were 3.62 μg/L (SD: 2.35 μg/L) for dropouts and 3.31 μg/L (SD: 2.02 μg/L) for patients who completed the study (t = 0.45; P = .66). At 1 year of the study, the patients who dropped out had a mean tacrolimus level SD of 1.45 μg/L (SD: 1.30 μg/L), compared with 1.31 μg/L (SD: 0.76 μg/L) for the patients who completed the study (t = 0.42; P = .68) (Fig 4). Repeated-measures analysis of variance did not detect significant differences in tacrolimus level SD values between patients who dropped out and patients who completed the study.
Number of Immunosuppressive Medications
We next examined whether the number of immunosuppressants taken influenced the results. The interaction between changes in tacrolimus level SDs and the number of medications taken was not significant. Results of an analysis comparing patients taking 1 versus >1 medication also were not significant. Within each group, however, tacrolimus level SDs decreased significantly. For patients taking 1 medication, the tacrolimus level SD decreased from 3.18 μg/L (SD: 2.35 μg/L) to 1.27 μg/L (SD: 1.24 μg/L; t = −3.49; P < .005). For patients taking 2 medications, the change was from 3.65 μg/L (SD: 2.10 μg/L) to 1.45 μg/L (SD: 0.74 μg/L; t = −4.36; P < .005). Among patients taking 3 medications, the tacrolimus level SD decreased from 4.15 μg/L (SD: 1.62 μg/L) to 1.61 μg/L (SD: 0.52 μg/L; t = −4.08; P = .02). Figure 5 displays mean values before and after the intervention according to group. The proportions of patients with tacrolimus level SDs of >2.5 μg/L were 80%, 64%, and 31% among patients who were receiving 3 medications, 2 medications, or 1 medication, respectively, before the intervention.
Medication Intake Responsibility
We further examined whether patient versus caregiver responsibility for medications influenced the results. Thirty-two patients and 9 caregivers were responsible for medications. Figure 6 depicts mean values before and after the intervention according to group. Baseline tacrolimus level SD values were not significantly different (t = 0.92; P = .36), and the proportions of patients with SDs of >2.5 μg/L before the intervention were 57% and 33% with patient and caregiver responsibility, respectively. As expected, patient age differed significantly between these groups. The mean age of patients who managed their medications was 17.38 years (SD: 4.06 years), compared with 7.75 years (SD: 5.21 years) for the caregiver group (t = 6.36; P < .005). Although adherence improved in both groups, there was no significant difference between the groups.
Data on rejection episodes were available for all 41 patients. Twelve patients experienced histologically proven, acute, cellular rejection in the year before the study, compared with 2 episodes during the study (χ2 = 5.08; P = .02). Nine of the 12 patients demonstrated tacrolimus level SDs of >2.5 μg/L during the year before the study, compared with only 2 patients during the study period. Risk factors for rejection before the intervention were mean age (patients with rejection episode: 17.67 years; patients with no rejection episode: 13.28 years; t = 2.15; P = .04) and number of medications (patients taking >1 medication were more likely to experience rejection; χ2 = 9.23; P < .005).
To examine whether the intervention was beneficial beyond the impact of experiencing rejection itself (regression to the mean), we compared study patients who had experienced a rejection episode in the year before the study with 17 pediatric patients (found in our database) who had experienced rejection in the 3 years before the study. The mean tacrolimus level SD did not change significantly after the rejection episode for the patients who had experienced rejection in the 3 years before the study, that is, from 5.17 μg/L (SD: 2.19 μg/L) to 4.87 μg/L (SD: 1.53 μg/L; t = 0.72; P = .48), as opposed to the study patients. No adverse events were reported.
This is one of the first studies presenting data on the feasibility of using TM to improve medication adherence for transplant recipients and on the effect on outcomes. Immunosuppressive medication nonadherence in posttransplant care is one of the most important causes of rejection among pediatric transplant recipients. Our study demonstrated reduction of the tacrolimus level SD from 3.48 μg/L to 1.36 μg/L with the use of TM. Furthermore, 24 patients were at high risk for rejection (SD: >2.5 μg/L) before the study, compared with only 6 during the study. The number of rejection episodes was significantly reduced from 12 episodes in the year before the study to only 2 episodes during the study period. This change may decrease associated morbidity and mortality rates and human and financial resource utilization. Of the 12 patients who experienced rejection, 9 had SDs of >2.5 μg/L, which confirms the previously reported findings that nonadherent patients are at greater risk of rejection.15 Patients receiving 3 immunosuppressants were found to be at greatest risk for rejection. Therefore, improving adherence particularly in this group may allow for gradual weaning of drugs and may reduce costs and potential side effects (eg, infection, posttransplant lymphoproliferative disease, and drug toxicity).
It is quite difficult to measure medication adherence. Indirect methods of assessing adherence, whether subjective or objective (eg, use of self-reports, adherence ratings by caregivers, physicians, or nurses, pill counts, prescription refill records, electronic monitoring devices, or clinical attendance rates), are not always feasible, reliable, or accurate. Blood levels are objective accurate measures of adherence and are directly related to medication intake. The measurement of 1 blood level may be misleading, because the patient might have forgotten to take the medication only once before the test or might have been generally nonadherent but took the medication just before testing. For nonadherent patients, individual medication blood levels can be low (dose not taken consistently) or high (physician increases the medication in response to the low blood levels and then the patient takes the medications more regularly). Serial blood levels and the degree of fluctuation (SD) of medication blood levels over time provide a better measure of compliance. Although blood levels may be affected by absorption and developmental changes in body mass, over time they are related almost exclusively to adherence. A higher SD indicates more fluctuation between individual measures and less-stable medication intake. SD values of >3 μg/L6 or >2.5 μg/L15 were found to predict rejection episodes among pediatric recipients of liver transplants.
Nonadherence is a preventable major risk factor for rejection. Interventions to improve compliance may target all consenting patients or may focus on subgroups with risk factors for nonadherence (eg, low socioeconomic class, single-parent household, adolescence, low self-esteem, comorbid psychiatric disorder,2,16 or poor medical outcomes). Current methods have not been very effective. TM offers a simple efficient intervention that can be offered to all patients without marking them as noncompliant. Given that forgetfulness is reported to be the most common cause of nonadherence,6 TM can serve as a useful reminder to patients to take their medications. If it is instituted early, it may prevent nonadherence and related consequences. It may allow early medical intervention when poor compliance is recorded, even before a scheduled blood level test and the development of complications.
Short messaging service messages are brief texts transmitted asynchronously through the mobile telephone network. TM via mobile telephones is an integral component of teenaged culture throughout the world. More than one third of US teenagers reported using TM in national surveys published in 2005,17 and the numbers are surely increasing. Our data showed that older teenagers were at greater risk of rejection. Therefore, an intervention targeted to this age group is warranted. The advantage of TM lies in the nature of the intervention. This approach is personal, discreet, socially acceptable, with reduced intrusiveness through minimal text communication, and relatively simple, with low costs to patients and minimal time demands for health professionals. TM seemed to engage a difficult-to-reach group of young people who are more susceptible to nonadherence.
This system also may facilitate the transitioning of medication management from parent to child, which is a period associated with decreased adherence.18 TM was shown to reduce the SD of tacrolimus levels in both groups (self and caregiver responsibility). It allows the caregiver to transfer the responsibility for medication intake to the child while maintaining supervision. It also enables the caregiver to intervene when the child fails to take the medication, because the system sends a message to the caregiver that the medication was not taken. It is user-friendly, because the patient can control the time of day at which he or she wishes to receive the message. This was especially useful for our teenager population, which had a wide range of morning awakening and medication schedules.
Some participants experienced technical problems, such as incorrect timing of calls, and most participants had already taken their medication when they received the reminder. The most substantial risk is the use of mobile telephones while driving. The present study had limitations. A selection bias might have occurred, because the most nonadherent patients could refuse to participate. Also, the study was limited to subjects who had a cellular telephone. One limitation of the study design is that it was not randomized or controlled, and patients were used as their own controls (1 year before the study was compared with 1 year in the study). Some of the reasons for the positive results might be related to regression to the mean (patients fared better because of the effect of time). However, when we compared the intervention group with a group with a history of rejection that did not receive TM, only the study group showed a reduction in the tacrolimus level SD. The ethnicity of our study group well represents the population our center serves; however, it may not reflect the general population. Despite the significant drop-out rate, when the outcome data were analyzed for patients who dropped out and compared with data for patients who completed the study, the former group had an unexpectedly similar improvement in adherence. This might be explained by the fact that adherence was improved during the time of participation in the study, and this behavior was learned and continued even after cessation of participation.
This study suggests significant improvement in medication adherence and reduction in rejection episodes with TM reminders for pediatric transplant recipients. The advantages of TM are its relative simplicity, low cost, minimal time requirements for health professionals, high levels of acceptance, and ability to be used for a large number of patients. This technology may prove effective in the transitioning of responsibility from the caregiver to the child. These results should be of interest to health care professionals and policymakers, because this approach may reduce the morbidity rate, mortality rate, and utilization of financial and human resources associated with liver transplantation. TM is a behavioral support intervention that can be incorporated into routine clinic care, personalized, and adapted to suit other chronic disease and age groups. The results of this pilot study are encouraging; however, larger, randomized, controlled studies are needed to establish the efficacy of TM in improving medication adherence.
- Accepted June 4, 2009.
- Address correspondence to Tamir Miloh, MD, Recanati Miller Transplant Institute, Mount Sinai Hospital, 1 Gustave L. Levy Place, Box 1104, New York, NY 10029. E-mail:
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject:
Nonadherence increases morbidity and mortality rates in the posttransplant setting. Adolescents are more prone to nonadherence. TM has been used with efficacy for a variety of chronic diseases, such as diabetes mellitus and asthma.
What This Study Adds:
This study showed significant improvement in medication adherence and reduction in rejection episodes with TM reminders for pediatric recipients of liver transplants.
- ↵Shemesh E, Shneider BL, Savitzky JK, et al. Medication adherence in pediatric and adolescent liver transplant recipients. Pediatrics.2004;113 (4):825– 832
- ↵Kahana S, Drotar D, Frazier T. Meta-analysis of psychological interventions to promote adherence to treatment in pediatric chronic health conditions. J Pediatr Psychol.2008;33 (6):590– 611
- ↵Neville R, Greene A, McLeod J, Tracey A, Surie J. Mobile phone text messaging can help young people manage asthma. BMJ.2002;325 (7364):600
- Copyright © 2009 by the American Academy of Pediatrics