Impact of mobile phone delivered reminders and unconditional incentives on measles-containing vaccine timeliness and coverage: A randomised controlled trial in western Kenya

Introduction Short message service (SMS) reminders coupled with a small monetary incentive conditioned on prompt vaccination have been shown to improve first-dose measles-containing vaccine (MCV1) uptake. We assessed whether SMS reminders and unconditional monetary incentives – more amenable to programmatic implementation – can improve MCV1 uptake in Kenya. Methods Caregivers of eligible infants aged 6-8 months were enrolled into an individually randomised controlled trial and assigned to receive either: no intervention (control), two SMS reminders (SMS) sent 3 days, and 1 day before the scheduled MCV1 date, or SMS reminders coupled with a Kenya Shilling (KES) 150 incentive (SMS +150 KES) sent 3 days before the scheduled MCV1 date. Study staff conducted a household follow-up visit at age 12 months to ascertain vaccination status. Log-binomial regression was used to estimate the relative and absolute difference in MCV1 timely coverage (by age 10 months), the primary outcome. Results Between 6 December 2016 and 31 March 2017, 179 infants were enrolled into each of the three study arms. Follow-up visits were completed between 19 April 2017 and 8 October 2017 for control (n=170), SMS (n=157) and SMS + 150 KES (n=158) children. MCV1 timely coverage was 68% among control arm infants compared with 78% in each intervention arm. This represented a non-statistically significant increase in the SMS arm (adjusted relative risk 1.13; 95% CI 0.99 to 1.30; p=0.070; adjusted risk difference 9.2%; 95% CI: -0.6 to 19.0%; p=0.066), but a statistically significant increase in the SMS + 150 KES arm (1.16; 95% CI 1.01 to 1.32; p=0.035; 10.6%; 95% CI 0.8 to 20.3%; p=0.034). Conclusion These findings suggest that the effect of SMS reminders coupled with a small unconditional monetary incentive on MCV1 uptake is comparable to that of SMS reminders alone, limiting their utility. Further studies in the absence of unexpected supply-side constraints are needed. Trial registration number NCT02904642

The Mobile and Scalable Innovations for Measles Immunisation (M-SIMI) study was a three-arm parallel individually randomised controlled trial conducted in Gem subcounty, Siaya County, Kenya. Gem subcounty is a predominantly rural setting with a population of approximately 164 000 in 201623 24 characterised by relatively high malaria, HIV and tuberculosis prevalence and high infant mortality.25 Vaccination coverage was over 90% for third dose DTP-containing vaccine (DTP3) and 84% for MCV1 by age 12 months in 2014–2015.17 The study was conducted in areas within the Kenya Medical Research Institute and Centers for Disease Control and Prevention collaboration’s Health and Demographic Surveillance System (HDSS). Eligible infants were required to: be aged 6–8 months; be residents of the subcounty as reported by the caregiver and to not have received a dose of routine measles vaccine as indicated in the home-based vaccination record (maternal and child health booklet). Infants’ caregivers were required to not have plans to move within 6 months of enrolment. Mobile phone ownership by the caregiver was not a requirement for enrolment. Participants were randomised and evenly allocated to one of three study arms: (1) control, (2) SMS reminders (SMS), (3) SMS reminders plus a 150 Kenya Shillings incentive (KES; SMS+150; KES150=US$1.50 as of December 2016). The conduct, analysis and reporting of results were conducted in accordance with the Consolidated Standards of Reporting Trials guidelines.26A detailed description of the methods and protocol has been reported.27 Simple randomisation with an allocation ratio of 1:1:1 to the control, SMS or SMS +150 arm was performed using a list of computer-generated random numbers. Randomisation and preparation of the allocation envelopes were done by the data manager who had no contact with participants. Given the nature of the interventions, study field staff and participants were not blinded to the study arm. The data analyst had access to participants’ study arm allocation during analysis. Additional details on the allocation procedure are provided in the supplement. Community health volunteers (CHVs)—a component of Kenya’s national Community Health Strategy—identified households with children aged 6–8 months and relayed this information to study-employed Community Interviewers (COMM-Is). COMM-Is then visited households to provide general information about the study and to perform screening procedures. COMM-Is verified age eligibility using the date of birth recorded in the home-based vaccination record. Written informed consent was obtained for eligible caregivers. Immediately after enrolment, COMM-Is collected vaccination status, sociodemographic, economic, mobile phone access, mobile phone usage, healthcare utilisation and other general health information from caregivers. Caregivers who did not own a phone were asked to confirm a phone number to which SMS reminders and the mMoney incentive (as applicable) could be sent. Caregivers who could not identify a shared phone number for the study were offered the option to use the COMM-I’s work phone number. All participants received an enrolment SMS, which included a health-related motivational phrase.27 Control arm participants received no interventions. SMS arm participants were sent two SMS reminders; one 3 days before the scheduled MCV1 date (3-day reminder) and the other 1 day before the scheduled measles vaccination date (1-day reminder). SMS+ 150 KES arm participants were sent reminders on the same schedule as the SMS arm participants and were sent the KES 150 incentive 3 days before the scheduled measles vaccination date that is, on the same day as the 3-day reminder. SMS reminders were sent in the caregiver’s preferred language that is, Dholuo, Kiswahili or English, as indicated at enrolment. The 3-day reminder was comprised of a standard reminder portion, a phrase intended to motivate caregivers and, for SMS+150 KES arm participants, language explaining that the study was sending the incentive to assist with travel expenses. The 1-day reminder was the same across intervention arm participants and consisted of a reminder portion as well as a motivational phrase that was different from the 3-day reminder motivational phrase (online supplemental box S1).27 Caregivers in the control arm were expected to receive one text message (enrolment message only) total and those in the intervention arms were expected to receive three text messages total. bmjgh-2020-003357supp001.pdf SMS reminders were sent out automatically using RapidSMS, an open-source platform.28 At enrolment, COMM-Is submitted an SMS to the RapidSMS server containing the infant’s name, infant’s date of birth, and caregiver’s preferred phone number. Based on the information submitted, the 3-day and 1-day SMS reminders, tailored to the applicable study arm and including the infants’ name, were sent from the RapidSMS server to the phone number provided by the caregiver. The Kenya Expanded Programme on Immunisation recommends MCV1 administration at age 9 months.29 Thus, the scheduled measles vaccination date was 274 days (30.42 days/month) from the infant’s date of birth, if falling on a weekday. If falling on a Saturday or Sunday, the scheduled measles vaccination date was defined as the following Monday. RapidSMS was programmed to send reminder messages to a study phone to allow monitoring of per-protocol transmission of SMS. Logs of sent SMS were generated from RapidSMS. RapidSMS was also programmed to automatically create a cumulative incentive payment list for infants enrolled in the SMS + 150 KES arm. The payment list included infants’ Study IDs, caregivers’ preferred phone numbers and payment dates (ie, 3 days before the scheduled vaccination date). Using the RapidSMS-generated payment list, study staff manually transmitted the KES 150 incentive from a smart phone using the M-PESA mobile money platform operated by Safaricom, one of Kenya’s mobile network providers. COMM-Is administered a follow-up survey when infants were aged 12 months to collect vaccination status as well as information on caregivers’ opinion of the interventions, reasons for delayed measles vaccination (ie, not vaccinated by age 10 months), incentive use and other general health information. If the child’s vaccinations were not up to date at the follow-up visit, the COMM-I referred the caregiver to the nearest health facility for vaccinations. The primary outcome was the proportion of infants receiving MCV1 by age 10 months (304 days; ie, MCV1 timely coverage). Secondary outcomes were the proportion of infants receiving MCV1 by age 12 months (365 days; that is, MCV1 coverage) and time to measles vaccination by age 12 months. At the follow-up visit, vaccination status was ascertained from either the home-based vaccination record or the caregiver’s verbal report if a home-based vaccination record was not available. If the home-based vaccination record was available, the date (day, month and year) of vaccination was transcribed and used to calculate the infant’s age at vaccination. For verbal vaccination reports, caregivers specified the month and year of vaccination. The study aimed to measure a≥15 percentage point absolute increase in MCV1 timely coverage in the intervention arms compared with the control arm. We presumed that a≥15% increase in the proportion of children receiving MCV1 by age 10 months would represent a meaningful effect from a policy-maker perspective. Based on previous coverage estimates in the study area,17 we assumed MCV1 timely coverage of 70% in the M-SIMI control group. We also assumed a type 1 error (alpha) of 0.05, a power (1-beta) of 0.80, yielding a sample size of 134 infants per study arm after application of a continuity correction. The sample size was adjusted to account for up to 25% lost to follow-up, which included death, outmigration and verbal report of measles vaccination at 10 months of age. A priori, verbal immunisation reports at the follow-up visit were to be excluded from the analytic sample. After accounting for potential losses to follow-up, the estimated sample size to assess the primary outcome was 537 infants total, or 179 infants enrolled per study arm. Log-binomial regression was used to estimate the relative risk (RR), difference in risk (RD) and respective 95% CI of measles vaccination by age 10 months or age 12 months in each of the intervention arms compared with control arm. For assessment of the primary endpoint (MCV1 timely vaccination), children were censored at age 10 months. Any predictive baseline characteristics determined to be unequally distributed at the 5% significance level were included as covariates in the regression model to adjust for potential confounding. The primary endpoint was analysed according to intention-to-treat (ITT) principles. A modified per-protocol sensitivity analysis for the primary endpoint was also conducted. The modified per-protocol analysis was defined as dispatch of two SMS reminders as per the target schedule that is, 3 days and 1 day before the scheduled vaccination date. In order to evaluate whether baseline participant characteristics modified the effect of the interventions on the likelihood of timely measles vaccination, we performed stratified (subgroup) analysis. To identify independent variables to include in the subgroup analysis, a risk factor analysis of baseline participant characteristics associated with MCV1 vaccination by age 10 months was conducted among only control arm participants using univariate log-binomial regression. Characteristics significant at the 10% level were included in subgroup analysis. Stratification by mobile phone ownership status and travel time to the health facility were pre-specified in the study protocol. Stratification by maternal education was added post hoc, based on evidence of an association between higher educational attainment and lower likelihood of missed measles vaccination in the study area.17 The significance level for subgroup analysis was 5%. Survival analysis was performed to assess whether time to measles vaccination differed significantly across the study arms. Time origin was defined as enrolment and events were right censored at age 365 days. The cumulative probability of measles vaccination was plotted using the failure functions estimated using the Kaplan-Meier method. Equality of the cumulative incidence functions were tested using the log-rank test. Analyses were performed using Stata/SE, V.14.1 (StataCorp). This study did not involve patients. We did not directly include participants or public representatives in the design of this study, but the design was informed by focus group discussions among community members for a similar, related study previously conducted in the same area.30 No participants or public representatives contributed to selection of outcome measures. CHVs, who are community members, were involved in identification of study participants. Study findings were presented to CHVs.

N/A

• Dima Health