Acceptability of mobile-phone reminders for routine childhood vaccination appointments in Nigeria – a systematic review and meta-analysis

Background Mobile-phone reminders have gained traction among policymakers as a way to improve childhood vaccination coverage and timeliness. However, there is limited evidence on the acceptability of mobile-phone reminders among patients and caregivers. This systematic review and meta-analysis aimed to evaluate the ownership of mobile-phone device and the willingness to receive mobile-phone reminders among mothers/caregivers utilizing routine childhood immunization services in Nigeria. Method MEDLINE, Scopus, CINAHL, CNKI, AJOL (African Journal Online), and Web of Science were systematically searched for studies on the acceptability of mobile-phone reminders for routine immunization appointments among mothers/caregivers in Nigeria. Studies were assessed for methodological quality using the Newcastle Ottawa Scale and JBI critical appraisal checklists. Meta-analysis was conducted using random-effects model to generate pooled estimates (proportion) of mothers who owned at least one mobile phone and proportion of mothers willing to receive mobile-phone reminders. Results Sixteen studies (13 cross-sectional and three interventional) involving a total of 9923 mothers across 15 states and the Federal Capital Territory Abuja met inclusion criteria. Pooled estimates showed that the proportion of mothers who owned at least one mobile phone was 96.4% (95% CI = 94.1–98.2%; I2 = 96.3%) while the proportion of mothers willing to receive mobile-phone reminders was 86.0% (95% CI = 79.8–91.3%, I2 = 98.4%). Most mothers preferred to receive text message reminders at least 24 h before the routine immunization appointment day, and in the morning hours. Approximately 52.8% of the mothers preferred to receive reminders in English, the country’s official language. Conclusion Current evidence suggests a high acceptability for mobile-phone reminder interventions to improve routine childhood immunization coverage and timeliness. Further studies, however, are needed to better understand unique regional preferences and assess the operational costs, long-term effects, and risks of this intervention. Systematic review protocol registration PROSPERO CRD42021234183. Supplementary Information The online version contains supplementary material available at 10.1186/s12913-021-07296-1.


Introduction
Routine childhood immunization has been remarkably successful and cost-effective in preventing infectious diseases worldwide [1]. In addition to saving millions of lives, vaccination generates immense economic benefits to society and saves the society the costs of reacting to outbreaks [2][3][4]. Immunization is also central to the Sustainable Development Goals (SDG) [4]. Like many other health care services that require repeated visits to the health facility due to timed scheduling of care, routine childhood immunization is characterized by poor compliance and attrition [5]. Inadequate information about immunization schedules and service arrangements with difficult-to-remember childhood vaccine series (with multiple appointments at various ages) is a key factor contributing to caregivers defaulting from routine immunization [6,7].
Reminder systems, which work through a variety of mechanisms, including phone calls, letters, postcards, and email, are meant to prompt the patient [8], and have been shown to improve immunization compliance and timeliness [9,10]. Among the various types of reminder systems, mobile phone reminders have been found to be the most effective [9,11]. Hence, an increasing number of studies recommend policymakers and operations managers to consider integrating mobile-phone reminders into routine immunization programmes [9,11,12]. A number of systematic reviews have synthesized the evidence on the effectiveness of mobile phone reminders on the uptake of immunization in low-and middle-income countries (LMICs) [13][14][15][16]. While all of these reviews suggest that phone reminders are generally effective, these exists a gap in our understanding of a key mechanism: acceptability. For mobile-phone reminders to encourage action by the recipient (in this case, take the child to the health facility for immunization), the reminders first need to be acceptable [17].
Against this background, this study aims to systematically review the literature to: (a) evaluate mobile-phone ownership among caregivers utilizing the routine vaccination service in Nigeria, (b) assess caregivers' willingness to receive mobile-phone reminders for routine childhood vaccination appointments, and (c) assess mobile-phone reminder preferences in terms of timing, frequency, and mode. We purposefully limit the analysis to studies conducted in Nigeria in an effort to inform the country's policy efforts on immunization. Despite efforts by the Nigerian government and international partners to ensure optimal utilization of routine immunization, over four million children in Nigeria still miss out on vaccinations every year [18]. Nigeria is one of 10 countries that account for more than 60% of the 19.7 million children that did not receive complete doses of Diphtheria-Pertussis-Tetanus (DPT) vaccine in 2019 [19,20]. Nigeria is one of only ten countries in the world with measles vaccine coverage of less than 50% [21,22]. Nearly 40% of under-five mortality in Nigeria, accounting for about 15% of global child deaths, have been attributed to vaccine preventable diseases [23].
Studies across Nigeria have demonstrated that lack of accurate information about immunization and immunization services is significantly associated with incomplete immunization [24][25][26][27][28]. Forgetting vaccination appointments is widespread among mothers/caregivers [26,27]. The recent explosion in mobile phone ownership offers a promising opportunity to leverage on this proven mobile health (mHealth) interventions to improve vaccination rates [29,30]. However, like in the case of LMICs generally, several studies in Nigeria have demonstrated the utility of mobile-phone reminders in improving routine immunization coverage and timeliness [31][32][33], with scant evidence on caregivers' willingness to receive mobile phone reminders for routine childhood immunization appointments.

Methods
The protocol for this systematic review was published on PROSPERO, with registration number CRD42021234183, and the review findings were reported according to the MOOSE and PRISMA guidelines [34,35].  [30]. Search was conducted over 4 weeks in January and February 2021. We used search terms covering mobile-phone reminders, phone call reminders, short message services (SMS) reminders (SMS, texts, text message), routine immunization, willingness to receive, acceptability and Nigeria (Supplement 1 -Search strategy). We also searched Google Scholar: https:// schol ar. google. com/; Semantic Scholar: https:// seman ticsc holar. org/; SCIELO (Scientific Electronic Library Online): https:// scielo. org/ en/; and websites of the National Postgraduate Medical College of Nigeria for eligible studies: https:// npmcn. edu. ng/; United Nation Children Fund (UNICEF), Nigeria: https:// unicef. org/ niger ia/; and World Health Organization (WHO), Nigeria: https:// afro. who. int/ count ries/ niger ia. We also searched Grey literature websites of the NYAM (New York Academy of Medicine) Grey: https:// catal og. nyam. org/; and Open Grey: http:// openg rey. eu/. Finally, we also sought for relevant articles from the references of studies identified through the database search. We only considered studies published in English as scientific/research articles in Nigeria are reported in English. The authors of relevant papers were contacted for missing information.

Search strategy and study selection
The search was independently conducted by two authors (PE and SAA); both authors search independently across all databases. Identified studies were pooled into Mendeley ® Reference Manager and duplicates were identified and excluded. After undergoing a moderation exercise to ensure uniform application of inclusion criteria, the two authors independently assessed the titles and abstracts for eligibility applying the inclusion criteria. Discrepancies were resolved by discussion. Finally, full text of each remaining articles was assessed against the inclusion criteria.

Eligibility criteria
Peer-reviewed studies of any design; published and unpublished, were included in the review if they were conducted among adults aged 18 years or more, residing in Nigeria, and reported the proportion of mothers/ caregivers who owned a mobile phone or proportion of mothers/caregivers willing to receive mobile phone reminder (SMS or phone calls), or enough data to compute these estimates. We excluded studies that reported willingness to receive mobile phone reminders for hospital appointments, medication adherence, health behavioural change, and/or vaccination in adults (such as human papillomavirus (HPV) vaccine, tetanus vaccine, etc.). We also excluded case series, reviews, commentaries, letters, and editorials.

Data extraction
Two authors (PE and SAA) extracted data from the included studies including authors, year of study, study design, location/region, study setting (rural or urban), description of study population, sample size (number of subjects involved), mean age with standard deviation, proportion of study participants with minimum of secondary education, ownership of a mobile phone, and number of mothers/caregivers willing to receive mobile-phone reminders. Information on mothers'/caregivers' mobile-phone reminder preferences (SMS or phone calls), preferred frequency, and ideal timing of mobile phone reminders were also collected. Microsoft Excel© was used to organize extracted data. Disagreements were resolved through discussion until there was 100% agreement.

Quality assessment
Two reviewers (PE and UJA) independently assessed methodological quality to establish the internal validity and risk of bias of included studies using the modified Newcastle Ottawa Scale for cross-sectional studies [36], and the Joana Briggs Institute (JBI)'s critical appraisal checklists for interventional studies [37]. Studies were rated as 'high quality' or 'medium quality' if they scored 7-8 points or 5-6 points, respectively. Otherwise, the study was rated as 'low quality' . Inter-rater discrepancies were resolved by discussion until 100% agreement reached.

Data synthesis
Data analysis was performed according to the guidelines specified in the Joanna Briggs institute (JBI)'s Manual for Evidence Synthesis [38,39]. Descriptive statistics and narrative synthesis were used to summarize the characteristics of included studies. Prevalence and 95% confidence interval (CI) for ownership of a mobile-phone device and willingness to receive mobile-phone reminders were estimated for each included study. Confidence intervals were estimated using the one-sample exact binomial (Clopper-Pearson) procedures [40]. Pairwise meta-analysis using the random-effects (DerSimonian-Laird) model were performed to pool individual results using the Metaprop Stata command with Freeman-Tukey double arcsine transformation (FTT) [40,41]. Analyses were conducted using Stata version 16.1 (STATA Corp, College Station, TX). An α (alpha) of 0.5 was uses as the cutoff for statistical significance.
Sensitivity analyses were first performed for the influence of studies with sample size less than 400 participants -as studies with small sample size are more likely to exaggerate study outcomes [37]. Pooled estimates were also assessed for influence of studies with sample size outliers. Lastly, pooled estimates were also re-assessed after excluding interventional studies. Subgroup analyses were performed for the periods the studies were published (2011-2015, and 2016-2020), region of study (northern regions vs southern regions), study setting (urban, mixed, and rural), and proportion of study participants with at least secondary school education (< 90% and ≥ 90%). Meta-regression analyses were performed to assess the impact of modifier variables: study period, study region, study setting, and proportion of mothers/caregivers with at least secondary education (continuous variable), on the meta-estimate proportion. Finally, evidence of publication bias was assessed by examining the symmetry of the funnel plot using sample size as the measure of accuracy [42], performing Egger's test for funnel-plot asymmetry, and using the trim-and-fill method.

Assessment of quality of evidence
The overall quality of evidence was assessed using GRADE (Grading of Recommendations, Assessment, Development and Evaluation) for the meta-analysis pooling data from all included studies [43]. Scoring of evidence started at high-quality evidence which was downgraded one level if one of the following prespecified criteria was present: (1) poor methodological quality (downgraded if ≥25% of the studies included in the metaanalysis used inappropriate sampling method or statistical analyses; (2) imprecision (downgraded if ≥25% of the included studies did not present minimum required sample size); (3) indirectness (downgraded if ≥25% of the included studies did not use valid and reliable methods for data collection, such as validated questionnaires that had been trialed, piloted, or published previously) and (4) inconsistency (downgraded if the confidence interval was wider than or equal to 5%). These pre-specified criteria were defined considering the items of Joana Briggs that correspond to the GRADE system criteria [38,39].
Of the 16 included primary studies, 11 studies (69%) were rated as high quality, three studies (19%) rated as moderate quality, and two studies (12%) which used convenience sampling were rated as low quality. Of note, however, is that probability random sampling was used to identify participants in 14 of the 16 included primary studies with total sample size of 9585 participants (96.6% of the overall study sample). Hence, sampling methods employed in primary studies ensures that results obtained from included study participants approximates results from the entire population. However, two studies with total sample size (338, or 3.4%) employed the convenience sampling method [50,53].
Sub-group analysis showed that ownership of mobilephones was about 6.0% higher among mothers in the Southern region than among mothers in the Northern region - Table 2. However, there was no substantial difference in mobile-phone ownership among mothers based on study publication period, setting, design, and quality of included primary studies. Meta-regression analysis demonstrated that a one-percent increase in the proportion of mothers with at least secondary school education (continuous variable) was associated with a 0.398 unit increase in mobile-phone ownership (p = 0.022). However, study publication period, geopolitical region, and study setting were not statistically significant modifiers.

Acceptability of mobile-phone reminders
Pooled estimate for acceptability of (willing to receive) mobile-phone reminders was 86.0% (95%   Fig. 4. Acceptability of mobile-phone reminders is lowest (61.7%) in Imo State, South-east region in 2018 [47] and highest in Ondo State, South-west region in 2018 [45]. In sensitivity analyses, the pooled estimate did not change substantially from the overall results when studies with sample size less than 400 mothers were excluded (pooled proportion = 88.8%; 95% CI = 80.4-95.1%; I 2 = 98.9%) nor when studies with sample size outlier were excluded [48] (pooled proportion = 85.8%; 95% CI = 78.0-92.2%; I 2 = 98.5%). However there the acceptability slightly decreased when interventional studies were excluded (pooled proportion = 82.6%; 95% CI = 74.9-89.1%; I 2 = 98.4%). Sub-groups analysis showed slight variations in the acceptance of mobile-phone reminders based the survey period, geopolitical region, setting, and educational studies of participants. Acceptability of mobile-phone reminders was slighter higher in the period 2016-2020 compared with the earlier period (2011-2015), higher among mothers in the southern regions than mothers in the northern regions, higher among mothers in the rural settings than mothers in the urban settings (Table 3). Meta-regression analysis demonstrated that the proportion of mothers with at least secondary school education (continuous variable), study period, geopolitical region and study setting were not statistically significant modifiers of willingness to receive mobile-phone reminders.

Mothers' mobile-phone reminder preferences
Meta-analysis of nine primary studies [45-47, 49-53, 55] including 2865 mothers who indicated willingness to receive mobile-phone reminders showed that 57.0% (95% CI = 39.7-73.5%; I 2 = 98.8%) of participants preferred to receive phone-call reminders while about twothirds (pooled proportion = 63.5%; 95% CI = 48.7-77.1%; I 2 = 98.4%) of participants preferred to receive text (SMS) message reminders. However, two primary studies reported differed data on mothers willing to accept both modes of reminders: while Akinrinde and colleagues reported about a quarter of mothers (28.3%) are willing to receive both types of reminders [45], Bello and colleagues reported a negligible proportion (2.2%) of mothers prefer both modes of reminders [53]. Mothers who had post-secondary education were more likely to prefer text messages [45,46,50].
A majority of mothers preferred to receive mobilephone reminders for routine childhood immunization reminders at least 24 h before the appointment day [45,47,[50][51][52], and to receive these reminders in the morning hours [45,48,56]. However, a significant proportion of mothers reported willingness to receive these reminders at any hours of the day [45,48,56], although mothers in these studies were asked what time/hours they were willing to receive text (SMS) message reminders, not phone-call reminders. Lastly, meta-analysis of six primary studies [45,46,48,52,56,59] including 5126 mothers who indicated willingness to receive mobile phone reminders showed that a slim majority of mothers (pooled proportion = 52.8%; 95% CI = 34.5-70.7%; I 2 = 99.2%) preferred to receive mobile-phone reminders in English language.

Assessment of publication bias
Graphical assessment of the funnel plot suggests absence of publication bias (Supplement 2). Objective assessment of publication bias using the Egger test also indicated absence of evidence of publication bias (p-value = 0.1450) (Supplement 3). Trim-and-fill method for assess

Quality of evidence
The quality of evidence for the ownership of mobile phones among mothers using the routine childhood vaccination services in Nigeria was graded as high - Table 4. There was no serious risk of bias, imprecision, indirectness, or inconsistency based on the pre-specified criteria. However, the quality of the evidence for acceptability of mobile-phone reminders and the mothers' preferences for each mode of reminder (i.e., phone call reminders over SMS/text reminders and SMS/text reminders over phone call reminders) were graded as moderate based on serious inconsistency across included studies (Table 4).

Discussion
This systematic review, to the best of our knowledge, presents the best available evidence on the ownership of mobile phones among Nigerian mothers/caregivers utilizing the routine childhood vaccine delivery service and their willingness to receive mobile-phone reminders. Our findings demonstrate that almost all mothers/caregivers utilizing the infant vaccine delivery system own at least one mobile phone. Although there are slight differences in mobile-phone ownership between mothers/caregivers in the Northern region and mothers/caregivers in the Southern region, there was no significant difference in the ownership over time and across different settings/ communities. Our findings differ from an earlier study, 6 years ago, that reported only about half of women of reproductive age have access to mobile phones [68]. First, this could be due oversampling of women from urban settings in the northern region (Kano State, Jigawa State, and Kaduna State) -a region where we found slightly reduced mobile phone ownership than the rest of the country. Secondly, women in our study were mostly married/co-habiting (95.6%) -mobile phone ownership is significantly higher among married women who often get the phones as gifts from their spouses [29]. Our findings-the near-universal ownership of mobile phonessuggest that the adverse impact of the use of mobile technology on health care delivery on health disparities might be less concerning than previous studies have documented [5,29,30,69]. Indeed, a number of prior studies have argued that the use of mobile phones in health care delivery can improve access, transparency, and equity of health services delivery in Nigeria [70,71].
We found that a high proportion of mothers are willing and happy to receive mobile-phone reminders for their children' routine immunization appointments. As already noted, health care services that require repeated visits to the health facility due to timed scheduling of careincluding routine childhood immunization -are faced with the challenges of poor compliance and attrition [5]. These findings imply that the rapidly increasing mobile phone ownership in the country can be leveraged to deliver timely, and often educative, reminder messages to improve compliance and timeliness of immunization. We also found that the willingness to receive these reminders is increasing over time, which may be due to higher use of text messages in everyday life, including socially and in banking. Additionally, our study also highlights some nuances for implementing mobile-phone reminders for routine childhood immunization in Nigeria. Although there were a few 'universal' preferences such as receiving mobile-phone reminders at least 24 h before the appointment day, most preferences such as choice over phone-call reminders vs SMS reminders or choice of language (English language or local language) were not. A statistically significant proportion of mothers who preferred phone-call reminders preferred a reminder in their local language, whereas statistically significant proportion of mothers who preferred text messages wanted the reminder in English language [39]. Geographically, one region where the use of text messages may not reach to the desired population is the northern region -a region with historically low immunization coverage [72] -where mothers are less inclined to receive mobile-phone reminders than the rest of the country. This could be due to factors such as the influence of religion and misperceptions of routine immunization [72] for which more engaged interventions than mobile-phone reminders alone will be needed. For designing such interventions, further studies are needed to better understand the influence of religion and culture on both health care utilization and adoption of technology [72].

Strengths and limitations
With the number of studies (n = 16) and the population covered (n = 9923), this review represents, to the best of our knowledge, the most comprehensive and representative study on the acceptability of mobilephone reminders for routine childhood immunization in Nigeria. We acknowledge that pooling prevalence rates from a range of studies conducted over a 10-year period (2011-2020) could affect reliability of our overall estimates; however, this approach enabled us to understand the trend in mobile phone ownership and acceptance of mobile-phone reminders over this period. Our pooled estimates should be considered with the high heterogeneity reported-a likely result of diverse population characteristics, particularly differences in religion, tribe/culture, and socio-economic status. Lastly, except for a single multi-region study [48], we could only retrieve studies from five of the six geopolitical zones in Nigeria, with no studies from the North-east region which covers six states and accounts for about 13.6% of Nigeria's population [73]. This represents a  critical gap in understanding the feasibility of successfully implementing this cost-effective m-health intervention in the region.

Implication for practice and research
There appears to be broad acceptance of mobile-phone reminder interventions among mothers utilizing routine childhood vaccine delivery system in Nigeria. Given this widespread acceptance and its demonstrated cost-effectiveness and proven impact, health policymakers and stakeholders should consider including mobile-phone reminders as part of a multi-strategy approach to address slow adherence to routine immunization in the country. However, further studies, preferably utilizing qualitative design, are needed to explore and identify the factors (including religion and culture) why small, but nevertheless significant, proportion of mothers are unwilling to accept text reminders on health messages.

Conclusion
Our findings demonstrate strong evidence for a high ownership of mobile phone devices among mothers utilizing the routine childhood delivery service in Nigeria, and also shows that most mothers are willing to receive mobile phone reminders for routine vaccination appointments. While these findings are encouraging, further studies are needed to better understand factors why some mothers decline to receive these reminders and appreciate nuanced regional and cultural differences in mothers' preferences for mobile-phone reminder.