This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included.
Despite many efforts, maternal mortality remains a major burden in most developing countries. Mobile health (mHealth) has the potential to improve access to obstetric care through apps that help patients and providers.
This study aimed to use mHealth to provide antenatal care (ANC) to 1446 pregnant women in a rural area in Madagascar and evaluate the quality of ANC provided by an mHealth system designed to change the behaviors of providers and patients.
We included 1446 women who attended ANC visits in rural Madagascar from 2015 to 2019 using an mHealth system called Pregnancy and Newborn Diagnostic Assessment (PANDA). This cross-sectional study used data from different participants, with information collected over several years, to analyze the outputs related to the quality of ANC over time. Specifically, we examined the timing of the first ANC visit, the relationship between the visit duration and the risk factors among pregnant women, and the number of ANC visits per woman.
Following the implementation of the mHealth system in 2015, we observed that women started to come earlier for their first ANC visit; more women attended their first ANC visit in the second trimester of pregnancy in 2019 than in the previous years (
This study shows the potential of an mHealth system to improve the quality of ANC, change provider behavior by standardizing ANC visits, and change patient behavior by increasing the willingness to return for subsequent visits and encouraging ANC attendance early in pregnancy. As this is an exploratory study, further studies are necessary to better understand how mHealth can change behavior and identify the conditions required for behavioral changes to persist over time.
Mobile health (mHealth) tools are an innovative technology that can allow patients and their health care providers to effectively access medical data before, during, and after medical appointments. mHealth has the potential to improve the quality of health care through apps that can facilitate communication between patients and health care providers [
Maternal mortality has decreased dramatically in low- and middle-income countries (LMICs), declining by 45% from 2008 to 2013 [
Expanded use of mHealth could help improve health care quality in the context of LMICs [
We conducted a study in Madagascar using an mHealth system to record and access women’s data during their ANC visits. The main goal of this study was to evaluate the quality of ANC provided with mHealth by measuring the adherence to ANC visits, the timing of the first ANC visit, and the duration of the visits.
This cross-sectional, observational study was conducted from January 2015 to September 2019 in the Ambanja district in northwestern Madagascar. Data related to ANC visits were collected using an mHealth system called Pregnancy and Newborn Diagnostic Assessment (PANDA). The local health authorities in Ambanja, Madagascar, and the Human Research Ethical Cantonal Board of Geneva, Switzerland (Comité d’éthique de la recherche CER 14-217; project number: CCER PB_2017-00641) approved the study.
We implemented an mHealth system to support providers in conducting ANC visits in Madagascar. Maternal mortality in Madagascar has decreased by over 50% in the last 20 years—from 776 to 353 deaths per 100,000 live births—but maternal mortality remains high and the Millennium Development Goal 5 has not been reached [
The Ambanja district is a rural area with a population of 240,000 inhabitants [
A collaboration between the University Hospitals of Geneva and the Centre Médico Chirurgical was established in 2010 for a cervical cancer program and was continued in 2015 with the implementation of the PANDA system. The PANDA team collaborates with the Ministry of Health of Madagascar.
The PANDA mHealth system was first implemented in 2015 in a pilot study to assess the system’s feasibility and usability in Madagascar [
The system comprises the following 3 elements. First,
Data were collected during the women’s first and subsequent ANC visits. The 3 following types of information were collected: (1) sociodemographic characteristics; (2) medical, surgical, and obstetric history; and (3) results from screening to detect obesity or malnutrition, hypertension or preeclampsia, anemia, HIV, syphilis, malaria, diabetes, infections, and other conditions. The PANDA system collected data on at least 75 items per woman on their past and current medical and obstetric history as well as clinical screening data.
Since 2015, we have trained 13 providers to use the PANDA system. In the PANDA medical unit, it is possible to track the content of the visits, including each provider’s activities, such as the number of visits conducted and eventual errors in completing the visits, thus allowing the team to build a learning curve for each provider.
A total of 1446 pregnant women fulfilled the inclusion criteria and were enrolled in the study. All pregnant women, regardless of age or stage of pregnancy, were eligible to participate in the study. The only exclusion criterion was the inability to understand or act as described in a previous publication explaining the acceptability and feasibility of the PANDA mHealth device [
We used a convenience sample of 1446 women. We planned to recruit patients from 2015 until the end of September 2020, which resulted in a total of 1446 women.
The data collected with PANDA were digitized as electronic medical records, which were used to analyze maternal morbidity and evaluate the quality of the ANC. Continuous variables are presented as mean (SD) or median, and categorical variables are presented as frequencies and relative percentages. The proportions of patients who tested positive for syphilis, HIV, or malaria are provided with their 95% CIs. Comparisons of categorical variables by year were performed using the chi-square test, and the mean durations of ANC visits were compared by the different categorical variables using the nonparametric Kruskal-Wallis test. We assessed the associations of different patient characteristics, place of residence, year, and visit order (independent variables) with visit duration (dependent variable)—first at the univariate level, using mixed linear regression models with the patient as a random factor to take into account repeated measurements within patients. We then constructed a parsimonious multivariable mixed linear regression model including all the variables that were significantly associated with the visit duration at
Most of the 1446 women in the study were recruited in the city of Ambanja. The first table in
Prevalence of malaria, HIV infection, and syphilis among pregnant women in the Ambanja district, Madagascar, from January 13, 2015, to September 20, 2019, at their first antenatal care visit (n=1443).
Variables | Participants, n (%) | 95% CI | |
|
|||
|
Positive | 18 (1.25) | 0.7-1.96 |
|
Negative | 1290 (89.40) | 87.7-90.9 |
|
Invalid or not tested | 135 (9.36) | 7.9-11.0 |
|
|||
|
Positive | 41 (2.84) | 2.0-3.8 |
|
Negative | 1061 (73.53) | 71.2-75.8 |
|
Invalid or not tested | 341 (23.63) | 21.5-25.9 |
|
|||
|
Positive | 23 (1.59) | 1.0-2.4 |
|
Negative | 813 (56.34) | 53.7-58.9 |
|
Invalid or not tested | 607 (42.06) | 39.5-44.7 |
Timing of the first antenatal care visit during pregnancy by year among pregnant women in the Ambanja district, Madagascar.
Pregnancy trimester | Year, n (%) | ||||||
|
2015 (n=341) | 2016 (n=265) | 2017 (n=197) | 2018 (n=360) | 2019 (n=276) |
|
|
First trimester | 28 (8.2) | 39 (14.7) | 28 (14.2) | 46 (12.8) | 18 (6.5) | <.001 | |
Second trimester | 168 (49.0) | 148 (55.9) | 131 (66.5) | 237 (65.8) | 201 (72.8) | N/Aa | |
Third trimester | 147 (42.8) | 78 (29.4) | 38 (19.3) | 77 (21.4) | 57 (20.7) | N/A |
aN/A: not applicable.
In the univariate analysis, the mean visit duration decreased significantly by approximately 5 min from 2015 to 2019 and varied by approximately 10 min across visit orders (see the second table in
All variables included in the multivariable model were significantly associated with the visit duration. These variables were the year, visit order, education, age, availability of running water at home, HIV status, and syphilis status.
Mean antenatal care visit duration in minutes (95% CI) by year and visit order.
We observed a statistically significant increase in the number of visits per woman from 2015 to 2017. The number of visits per woman then became stable after the third year of implementing the PANDA system.
Mean number of antenatal care visits (95% CI) per patient by year (Kruskal-Wallis test;
This study has shown the potential of using an mHealth system to encourage providers to follow a standardized ANC visit and also contribute to increase women’s adherence and willingness to return for subsequent ANC visits in Ambanja, Madagascar.
According to the United Nations International Children's Emergency Fund data gathered in Madagascar in 2018, most women attended their first ANC visit in the third trimester of pregnancy, with only 45% attending their first visit in the second trimester or earlier [
There is no recommended benchmark value for the duration of ANC visits as a quality indicator; however, we assumed that a provider would need at least 20 min for the first ANC visit (based on our clinical experience) to cover all the main topics that should routinely be part of a high-quality ANC visit. The duration of the first ANC visit, a quality indicator defined by the WHO, remained stable over the study period, indicating that the use of the mHealth device did not significantly lengthen the visit duration. We also found that the visit duration was positively associated with several patient risk factors. Low education, being over 35 years of age, experiencing domestic violence during pregnancy, having anemia, and having an HIV positive status were all associated with a longer ANC visit duration.
In terms of the completeness of the ANC visits, most pregnant women in our sample were tested for HIV during these visits (92.7%), compared with only 10% of pregnant women in an analysis of national-level data from 2018, which also reported that only 3% were receiving antiretroviral therapy to prevent vertical transmission [
In this study, we found that the number of ANC visits per pregnant woman tended to increase over the study period. In the study by Atnafu et al, [
This study analyzed data from a large population over a 5-year period following the implementation of the PANDA mHealth system to support good quality ANC in a resource-constrained setting (Madagascar). The indicators in our study, such as the timing of the first ANC visit, the visit duration, and the number of visits, strongly suggest that this mHealth system encourages the performance of a standardized ANC visit and thereby facilitates the provision of high-quality ANC services. This study has shown the benefits of the PANDA mHealth system for both providers and patients. The providers received guidance on how to conduct standardized ANC of good quality, and they also received a record of all important patient data that can be used in follow-up visits. The women and their families received ongoing education and encouragement to seek appropriate care throughout the ANC and postpartum visits conducted with the PANDA system. The system improved communication between the health care workers and patients, facilitated continuous education, and encouraged the patients to play a more active role in the decision-making processes related to their health.
Our study has several limitations. The most significant limitation is the lack of a control group to compare with participants using the PANDA mHealth system. Instead, we used national-level surveys as a standard reference—the 2009 Madagascar Demographic and Health Survey and the 2018 Multiple Indicator Cluster Survey. Although our results differed significantly from the findings of these two national-level surveys, the lack of a comparison group and a randomized design limits the conclusiveness of our findings. D-tree International’s study on safer deliveries in Tanzania, which also lacked a control group, demonstrated that the examined mHealth program was a success as it reached over 13,000 pregnant women in Zanzibar. The implementation of this mHealth system was described as a success as it was linked to unprecedented rates of both service delivery and postpartum attendance, even though this mHealth system did not provide support for either ANC or postpartum visit content or quality [
Several questions about the use and diffusion of the PANDA mHealth system remain open. These questions relate to, for example, dealing with the low quality of service delivery and the scalable and sustainable integration of the PANDA system in the local health system.
This study shows that the PANDA mHealth system has the potential to improve the quality of ANC in a resource-limited setting, modify the behavior of providers by providing standardized ANC visits, and increase patient compliance. Mobile technology should not be considered a stand-alone health intervention for ANC; rather, it is a strategic tool for improving the delivery and quality of maternal health care. Further studies are necessary to better understand the conditions under which behavioral changes occur and persist over time with the use of mHealth systems such as PANDA as well as whether undesired behavioral changes may also arise with the use of mHealth in ANC settings.
Sociodemographic and clinical characteristics of pregnant women in the Ambanja district, Madagascar, from January 13, 2015, to September 20, 2019, at their first antenatal care visit (Table 1). Sociodemographic and clinical variables that were significantly associated with antenatal care visit duration among pregnant women in the Ambanja district, Madagascar (univariate and multivariable analyses) (Table 4).
antenatal care
low- and middle-income country
mobile health
odds ratio
Pregnancy and Newborn Diagnostic Assessment
World Health Organization
The authors thank Angèle Gayet-Ageron, MD PhD MPH. for the support on the statistical analysis and Jennifer Barrett, PhD, from Edanz Group for editing the draft of this manuscript.
AB participated in the study design, analysis, and writing. MR, MV, and NS took part in the revision of the paper. GS participated in the final revision of the paper.
None declared.