Study selection and characteristics
A total of 11,817 records including additional sources from reference lists and grey literature were identified. After removing duplicates, the 8,069 records left were screened by title and abstract, and 7,805 were considered irrelevant mainly because they were not related to MNCH data collection systems in LMICs. The full texts of 264 records describing 96 data collection systems were assessed for eligibility and finally, eight perinatal data collection systems (involving 165 reports) were included in qualitative synthesis (Figure 1). The included 165 reports were categorized as descriptive articles (51 published and 12 unpublished), published research studies related to the data collection systems (n=87), official system websites (n=7), user manuals or guides (n=3), and other web links that were not official systems websites (n=5). The most frequent reasons for excluding 99 reports were not collecting perinatal outcomes continuously at the individual-level (64), not currently capturing data (12) or not being a specific MNCH data collection system (9). The reasons for final exclusions of the potentially eligible systems are presented in Additional file 3.
The eight data collection systems finally selected were: 1) Global Network’s Maternal Newborn Health Registry (GN-MNHR), 2) International Network for the Demographic Evaluation of Populations and their Health (INDEPTH), 3) Perinatal Informatic System (SIP), 4) Pregnancy Exposure Registry & Birth Defects Surveillance (PER/BDS), 5) SmartCare, 6) Open Medical Record System (OpenMRS), 7) Open Smart Register Platform (OpenSRP) and 8) District Health Information Software 2 (DHIS 2) (see Table 3).
Table 3- Data collection systems selected.
Regarding the geographic distribution of the systems, although they were implemented in many different countries and districts, not all sites captured individual maternal and neonatal data. Therefore, we only included the sites that met the objectives and inclusion criteria of our study. DHIS 2 tracker, GN-MHNR and INDEPTH are in sub-Sahara Africa and South Asia. Additionally, GN-MNHR is also located in Latin America and the Caribbean (Guatemala), DHIS 2 is in the Middle East and North Africa (West Bank and Gaza), and INDEPTH is in East Asia and the Pacific (Indonesia, Malaysia and Vietnam). PER/BDS is located only in South Africa. SmartCare is in Zambia and OpenSRP is in Indonesia and Pakistan. Finally, OpenMRS is located in Uganda, Rwanda, Lesotho, Malawi, Kenya and Haiti.
Out of the 165 included reports: 86 (52.1%) were related to INDEPTH [15-100], 26 (15.7%) to Global Network [101-126], 24 (14.5%) to DHIS 2 [127-150], 9 (5.4%) to OpenMRS [151-159], 6 (3.6%) to SIP [160-165], 6 (3.6%) to OpenSRP [166-171], 4 (2.4%) to PER/BDS [172-175], and 4 (2.4%) to SmartCare [176-179].
Major findings by identified domain
Following the analysis of the extracted data, the eight included systems’ synthetized results were presented in seven domains: Governance; System design; System management; Data management; Data sources, Outcomes and Data quality (Tables 4-9). Extracted data has been made available in a web-interactive App: http://safeinpregnancy.org/la_sc/table_by_domain.html#
1.Governance (Table 4)
These systems are supported by different categories of institutions [180] such as private foundations (e.g., Bill and Melinda Gates Foundation, Wellcome Trust, The Rockefeller Foundation, Children’s Investment Fund Foundation, Hewlett Foundation), governmental agencies in high-income countries (e.g., United States Agency for International Development (USAID), Centers for Disease Control and Prevention, Norad), global health initiatives (President’s Emergency Plan For AIDS Relief, The Global Alliance for Vaccines and Immunization), research councils (National Institutes of Health, National Institute of Child Health and Human Development, Medical Research Council), non-governmental organizations (Comic Relief), international organizations (World Health Organization /Pan America Health Organization, UNICEF), universities (Harvard University, University of Oslo), private sector organizations (GlaxoSmithKline, Qualcomm) and LMIC governments (South Africa National Department of Health) [14, 16, 36, 76, 102, 160, 177].
Some organizations were also responsible for the development and implementation of the included systems and are responsible for its optimal operability, such as the University of Oslo in the case of DHIS 2 [130], Partners in Health for OpenMRS [151], WHO for OpenSRP
[169] and National Department of Health South Africa for PER/BDS [181] (Table 4). The majority of the systems demonstrated features that support the protection and privacy of collected information through anonymization of data, implementation of passwords before access, or external security (cybersecurity) [37, 76, 102, 130, 140, 153, 169, 176, 178]. Up to the search date, neither GN-MNHR nor SIP allowed for data encryption.
We were able to access the operating manuals, data forms, and documentation for six out of the eight systems [16, 102, 130, 151, 162, 169]; for SmartCare and PER/BDS these types of documents were not identified. Although most of the systems were designed for clinical care, some had been conceptualized for research such as GN-MNHR) [105, 112, 118, 121, 122] or surveillance such as INDEPTH [45, 48, 76, 99] or PER/BDS [173]. Some of these were designed to satisfy more than one objective, and in the case of OpenMRS, this varied in different locations where the system is in place [129, 144, 151, 160, 169, 177].
Table 4. Governance
2.System design (Table 5)
The type of license was free and open-source for four systems: DHIS 2 [130, 139], OpenSRP [169], OpenMRS [155] and PER/BDS [181]. SIP used a closed-code source [160] and GN-MNHR used a private license. Web-based platforms were the most frequently used [45, 48, 130, 139, 152, 155, 169]. However, the two systems GN-MNHR [102] and SIP [160, 162] still used local networks. No information on the type of license was recorded on SmartCare and INDEPTH.
Interoperability was assessed through the system’s ability to compile, transfer and export data, and integrate with other data sources, systems, individual and laboratory records, and/or national health record databases. The DHIS 2 [130, 131, 144, 145, 147], INDEPTH [16, 17, 20, 27, 34-37, 45, 57, 96, 99], SmartCare [176-179], OpenMRS [152, 153], OpenSRP [170] and PER/BDS [173] all have these capabilities. GN-MNHR [102, 107] and SIP [160, 162] systems showed lack of ability to link with National Health databases and clinical or laboratory records. All eight data collection systems demonstrated flexibility to add new variables.
Data were captured only at facilities in the system SIP [160], SmartCare [177] and PER/BDS [173, 181]. Data were captured both at the facility and community level for the systems DHIS 2 [143, 147], INDEPTH [48, 99], OpenMRS [151-153] and OpenSRP [169]. GD-MNHR only captured data at the community level [107, 112]. GD-MNHR only captured data at the community level [16, 32, 108, 122, 129, 130].
Table 5. System design
3.Data management (Table 6)
Although all of the included systems recorded data electronically [27, 48, 61, 84, 99, 107, 140, 144, 149, 151, 160, 168] and SmartCare [177, 178], used a mixed modality and initially captured data only on paper. Trained health providers, including nurses and doctors, collected the data in all systems. Only the DHIS 2 system through the MomConnect platform [139] allows pregnant women to enter information directly into the system through their smartphones. OpenSRP promotes a mobile health platform that allows health workers to register and track patient data [166].
GN-MNHR, INDEPTH, PER/BDS and SIP coordinated the data collection and validation across the sites [16, 102, 160, 181]. In contrast, DHIS 2, OpenMRS , OpenSRP and SmartCare offered a module and platform that each site can customize, modify, and adapt for use with total autonomy [130, 151, 169, 177].
The tenth revision of the International Classification of Diseases (ICD10) was used to code outcomes and conditions by more than half of the systems: DHIS 2 [129, 130], INDEPTH [99], OpenMRS[153], SIP[160] and PER/BDS [175]. No information was found regarding how OpenSRP and SmartCare systems classified and coded outcomes. The GN-MNHR system does not use any standardized classification. Only INDEPTH had been used for phase IV safety trials and post-marketing surveillance by a maternal health research platform [16, 182].
Table 6. Data management
4.Data sources (Table 7)
All the systems can collect patient data and longitudinally track pregnant women’ progress and their babies over the prenatal and postnatal periods. However, timing of capturing information from antenatal visits is different between the eight systems. GN-MNHR and INDEPTH collected antenatal care data retrospectively [16, 102]. GN-MNHR collected their data at enrollment and delivery [102], and INDEPTH collected past events by self-reported data from household visits [16].
Drug exposures during pregnancy were recorded widely (e.g., antimalarial and antiretroviral treatment, iron, folic acid and vitamins) [16, 67, 102, 113, 126, 149, 151, 164, 169, 177, 181]. Exposure to vaccines was also collected, mainly of certain vaccines related to pregnancy (Influenza, tetanus/pentavalent) as well as non-pregnancy related vaccines (Hepatitis B, BCG, Haemophilus influenzae type B) [16, 67, 102, 113, 126, 129, 151, 164, 169, 177, 181]. PER/BDS system showed the widest drug and vaccine exposure recording, and intends to increase the list during the registry’ s future national implementation [173, 174]. We did not find information about collecting this information for SmartCare.
Table 7. Data sources
5.Maternal and neonatal outcomes (Table 8 and 9)
Twenty-nine MNCH outcomes in selected data collection systems were searched: 16 maternal outcomes and 13 neonatal outcomes. We did not find information about SmartCare regarding their recorded perinatal outcomes.
All systems collected vital data such as maternal and neonatal deaths. The most frequently recorded perinatal outcomes were fetal distress, postpartum hemorrhage, antenatal bleeding, dysfunctional labor, spontaneous abortion, congenital anomalies, neonatal infections, preterm birth, stillbirth, low birth weight, small for gestational age and respiratory distress. Some outcomes were not recorded by any of the selected systems, i.e., premature preterm rupture of membranes, preterm labor, insufficient cervix and neurodevelopmental delay [183].
The seven systems with available data recorded 13 to 22 perinatal outcomes out of a total of 29 perinatal outcomes. Of the 16 maternal outcomes evaluated, SIP [160, 162, 164] and GN-MNHR [107, 109, 112, 118, 121] registered more than 50% of outcomes (n=11 and n=10 respectively) , DHIS 2 [130, 141, 149], OpenMRS [153] and OpenSRP [167-169] registered 50% of outcomes (n=8 each) and INDEPTH [37, 45, 48, 67, 76, 99] and PER/BDS [181] less than 50% of outcomes (n=6 each).
Of the 13 neonatal outcomes evaluated, SIP [160, 164], PER/BDS [174], GN-MNHR [104, 105, 107, 112, 113, 118], DHIS 2 [137, 144, 145, 149] and INDEPHT [37, 45, 48, 55, 67, 99] registered more than 50%, (n=11, n=11, n=10, n=9 and n=7 respectively) and OpenMRS [151, 153] and OpenSRP [168, 169] less than 50% (n=6 and n=5 outcomes, respectively).
Table 8. Maternal outcomes
Table 9. Neonatal outcomes
6.Data quality (Table 10)
This domain was evaluated by examining information on both external and internal quality control mechanisms used by data collection systems. Internal monitoring was the most frequently cited procedure, specifically pre-programmed checks to avoid incorrect data entry [28, 37, 102, 138, 151, 155, 169]. Regarding external monitoring, only half of the systems reported having the necessary structures to be subject to frequent auditing and manual reporting [27, 37, 44, 107, 112, 144]. Only three systems demonstrated internal and external quality controls (DHIS 2, GN-MNHR and INDEPTH).
Table 10. Data quality