Pakistan is one of the countries with the highest rates of child death in the world [1]. It ranks 4th in child mortality, with 60% deaths due to vaccine-preventable diseases (VPDs) [2]. Table 1 [18] shows the current schedule of routine immunization (RI) in Pakistan, which is provided by the government free of cost. The immunization coverage in Pakistan is estimated to be 59%, which is still well below the desired level, leading to continued polio transmission, large measles outbreaks, and thousands of deaths from vaccine-preventable illnesses [2]. In addition, Pakistan is a major polio epidemic country and among 3 countries in the world requiring proof of polio vaccination for international travel [3]. Pakistan demographic and health survey in 2017-2018 suggests 88% percent of children had received BCG vaccine due at birth, 86% and 95% had received the first dose of pentavalent and polio vaccine respectively due at the 6^th week. Furthermore, 75% and 86% of children had received the third dose of the pentavalent and polio vaccines, respectively, due at 14^th week and measles vaccination was 73%, which is due at 9 months. However, these rates are at 1 year of age and much higher than vaccination coverage rate at scheduled time and among conflict hits and displaced populations [4]. Improved RI coverage is recommended as the priority public health strategy to reduce VPDs and eradicate polio in Pakistan and worldwide.

According to immunization coverage surveys, 1 in 5 children is unimmunized [5]. A major reason for poor childhood vaccine coverage is low immunization uptake, when parents are unable to complete the entire series of vaccines in accordance with the scheduled timelines. Some of the reasons include: (1) the family is not in favor of getting their child immunized, (2) low trust in vaccines provided through Expanded Programme on Immunization (EPI) and government health care providers, and (3) caregivers have forgotten their child’s next vaccination due date or child’s EPI card is misplaced [6]. These barriers may be modified with additional support through education and behavior change strategies. In addition, with more pressing issues of food and shelter, preventive health often takes the back seat, and parents and caregivers forget or ignore the subsequent doses of vaccines for their children. There is an immense need to encourage parents’ care seeking and collaboration with the health care providers to improve initial vaccine uptake and the completion of all doses according to the schedule. New innovative and cost-effective techniques are necessary for practical solutions to improve vaccination uptake and coverage.

Mobile phones offer a new medium to provide education and advocate families or caregivers to enable behavior change so as to improve immunization uptake. Mobile phone use has also increased in countries with low RI coverage and a high risk of VPDs. Good examples are Nigeria and Pakistan, where there were around 170 and 140 million mobile phone subscribers, respectively, in 2014 [7,8]. There is also a surge in use of short message service (SMS), with 237.58 billion person-to-person SMS generated in 2011 estimating to around 175 SMS per mobile phone on a monthly basis in Pakistan [9].

Mobile reminders in the form of phone calls have also proven to be a feasible method of improving RI uptake in resource-limited settings with wide cellphone coverage [10]. A Cochrane review reported that automated telephone communication systems including automated calls are effective in a variety of health care settings such as improving clinic attendance rates, screening, adherence to medications, and laboratory tests [11]. Automated calls were also found to be cost-effective in increasing immunization rates in an urban practice in the United States [12]. A study suggests that mobile phones have wide spread abilities to improve health outcomes in low and middle-income countries (LMICs) by targeting larger populations in a cost-effective manner [13]. Furthermore, SMS-based reminders along with small financial incentives could possibly help in improving RI timelines [14]. Text reminders have also proved to reduce vaccine dropout rate and improve parents’ compliance to immunization-scheduled visits [15,16].

There are limited data from LMICs set up on the role of SMS-based interventions for improvement of RI coverage, and conventional 1-way reminder SMS text messages were used by most of the studies as the intervention [13,21,22]. Overall, very few studies compared reminders, educational, and interactive SMS messages related to childhood vaccination uptake [13,21-26]. Although some of the studies have shown some behavior change for improvement in vaccination coverage, more rigorous application of health behavior change model needs to be applied to understand the impact of reminder, educational, and interactive messages on behavior change related to improvement in RI coverage [17]. However, data from developing countries regarding the role of automated calls in improving vaccine coverage are limited. Majority of the studies have focused on SMS-based intervention.

In this study, we will identify the factors that affect adherence to RI coverage among Pakistani families and caregivers and will examine an important public health question—do low cost, automated SMS text messages and calls improve RI coverage in resource-constrained settings such as Pakistan? In this study, we will compare the effectiveness of different types of messages: reminder, educational, and interactive SMS text messages and automated calls for improving RI uptake. Different types of messages will be developed to meet the possible RI barriers identified; these messages will be sent specifically to the participants according to the type of RI barriers they faced for immunization. This information will be used to develop strategies to improve vaccine adherence in Pakistan.

Our target population are caregivers of newborn (NB) infants younger than 14 days who are due for their RI at age 6, 10, and 14 weeks according to the EPI schedule of Pakistan.

Our goal is to identify the barriers faced by families in the uptake of correct immunization practices and to develop adapted messages (SMS or automated calls) to improve vaccine adherence in Pakistan. We will collect information from families regarding perceived sociocultural, technical, and economical barriers that may explain the decrease in vaccine coverage, and possible solutions to overcome these constraints. This information will be used to develop personalized educational messages, reinforced by interactive exchanges (2-way automated SMS and automated calls) with caregivers in the initial 20 weeks of their child’s life.

Personalized weekly message in the form of automated SMS text message or automated calls according to barriers for RI and language preferences can improve RI uptake according to the schedule for vaccine due at age 6, 10, and 14 weeks as compared with standard counseling by health care providers at EPI center and outreach visits for RI. Interactive voice recording (IVR) or 2-way auto calls might be preferred way of communication to caregivers as compared with other intervention and control for possible improvement in RI coverage and timelines.

The study design is a cluster randomized clinical trial including 4 intervention and 1 control arm (Table 2).

The sampling strategy adapted for both sites are as follows (Figure 4).

The data for live births and population of each site in Karachi will be obtained from HDSS team. Data will be scrutinized and divided according to 4 sites in the study, with each site consisting of 5 arms (5 clusters). Administrative structures will be assigned to each intervention through stratified (ie, site) randomization.

As Matiari site has not been stratified into clusters according to HDSS, we randomly allocated 30 clusters per arm. The catchment area of 4 LHWs will be considered as 1 cluster. Each intervention arm will then be randomly allocated to 24 administrative structures (clusters; see Figure 5).

For Karachi, birth data of the clusters will be shared by central HDSS team on weekly basis and at Matiari the LHWs will inform births to the study team daily according to the study clusters. The study staff will approach families who have an infant younger than 2 weeks in each cluster.

Staff will explain the study objectives to the parents or caregivers. If the parent or caretaker is interested in the study, the infant will be evaluated for enrollment. One child per household will be selected. In a household, where there is more than 1 child (younger than 14 days), a random selection will be made by a program designed in the mobile phone device. After meeting the eligibility criteria, informed consent will be obtained, and the infant will be enrolled in the study and a baseline survey will be conducted.

A second survey will be conducted at 20 weeks of child’s age (end of follow-up) to identify vaccination coverage according to the schedule, to be confirmed by physical examination of EPI card (see Figure 6).

Interviews will be conducted after the trial, with a purposive sample of a subpopulation included in the RCT representing all study arms. From these structured in-depth interviews, we expect to further explore participants’ experience of the different study arms having different levels of coverage. We will also assess the factors related to SMS text messages and automated calls associated with vaccination uptake. Sample size will be guided by saturation of the information.

Analyses will be conducted according to the intention-to-treat principle. When the information cannot be gathered, the vaccine status of the child will be failure (conservative approach). Missing the vaccine EPI card to verify vaccine status at home will lead to a deeper investigation in the family and also the vaccine clinic. From previous experience, we expect this process to be successful in identifying the outcome of over 95% (95/100) of the children who remain in the study at 20 weeks. Families that have left the study for any reason (moving, withdrawal, child death) will be interviewed (if possible) and the information gathered used for the outcome classification. If the information is not clear or incomplete, the child’s status regarding vaccination will be failure.

The main analyses will be comparing our 4 study intervention arms against the control, and with each other. For the purposes of our study, we will compare each of the 4 interventions with each other and with the control arm. For example, 1-way versus 2 SMS, reminder versus educational SMS, 1-way reminders SMS versus control, 1-way educational SMS versus control, 2-way reminders SMS versus control, 2-way educational SMS versus control, 2-way education versus 1-way reminder, and 2-way reminder versus 1-way educational.

Analyses will be conducted according to the intention-to-treat principle, and the Bonferroni correction will be used to control for multiple testing. The primary outcome is to assess difference in vaccination status. Chi-square test will first be used to compare the groups’ percentage. Multivariable logistic regression analyses will be conducted to adjust for confounders. As our randomization unit is cluster-based rather than individual-based, generalized linear mixed effects models or generalized estimating equations models will be employed to account for within-cluster correlation, and hierarchal models will be used. The statistical model for the primary and secondary analyses will be developed as explanatory variables. As secondary analyses of the primary outcome, time-to-immunization curves will be constructed using the Kaplan-Meier method and Cox regression for multivariate analysis.

In the first section of the semistructured interview guide, we will ask the caregivers regarding common barriers they encounter at the time of RI, whereas the second section comprises perceptions and barriers related to a mobile phone. The data will be transcribed into written form from audio recordings and will be analyzed via qualitative data analysis software NVivo 11 (QSR International). Written transcripts will then be uploaded into NVivo 11 software to offer easy and organized retrieval of data for analysis. The data analysis will be conducted according to discourse analysis. The interview guides will also be pretested in a similar community

The audio recording and transcripts will have a unique identifier; original and backup files will be archived in password-protected computer or server at AKU. Only transcriptions and themes will be shared at University of British Columbia (UBC) without nominative information. All study-related data including the recording will be stored in an encrypted server with password protection and having access only to the study specific personnels.

Baseline and end line data will be collected on a smartphone device (Multimedia Appendices 3 and 5). The entry program will be designed to capture data as well as the location of the interviewer along with some monitoring parameters. Each child participant will be given a structured unique identifier. Business rules, skips, and consistency checks will be incorporated, and important fields will be marked as must enter in the questionnaire to maintain data collection quality. The database will reside on a central computer at AKU managed by the study staff. A Web-based dashboard will be designed to report daily study progress. Mobile numbers will not be shared except to track patterns of use. Only relevant study staff will have access to study data allowed by the local ethics committee. All study staff will undergo basic research ethics training. Participants’ information will be given a study code, and no personal identifiers will be shared. Data confidentiality will be maintained at all times. No personal identifiers will be used in any reports or publication of the study. Mobile phone companies will be only provided mobile phone numbers to send SMS text messages and automated calls messages through gateway (Multimedia Appendix 4). No individual identifier such as names of participants and area of location will be shared. In addition, a confidentiality agreement has been signed with the university and the mobile companies stating that the numbers provided will only be used for the purpose of the trial. All study data and recordings will be destroyed within 5 years of the study according to the recommendation of the UBC and AKU ethics committees.

We recognize that not all barriers identified are modifiable and amenable to phone calls and SMS text messages. However, most of them reflect a priority ranking that does not favor child routine vaccination. One major reservation about SMS-based interventions is the recipient’s level of literacy. However, there has been mixed input related to the preference of phones calls as compared with SMS text messages in populations of low literacy and resource-constrained settings [20,27]. Mobile phone SMS text messages in local languages, pictorial messages, and in combination with phone calls can further reduce this gap. In our study, we will send messages in local languages (as per the preference of the participants). Although SMS text messages have a limitation of 160 characters and even fewer if translated in other characters, these limitations might help in making the messages simple and brief, especially for populations with low literacy levels [22,28].

The type of intervention will not be blinded, but all families in the same cluster will receive the same intervention. The primary outcome (vaccine completion) is not likely to be influenced by any knowledge of the intervention arm.