Content uploaded by Yuri Quintana
Author content
All content in this area was uploaded by Yuri Quintana on Feb 03, 2020
Content may be subject to copyright.
An Informatics Framework for Maternal
and Child Health (MCH) Monitoring
Juan HENAO1, Yuri QUINTANA1,2, Charles SAFRAN1,2
1Division of Clinical Informatics, Beth Israel Deaconess Medical Center, Boston, MA
2Harvard Medical School, Boston, MA
Abstract. Most cases of maternal deaths could be avoided with timely access to
quality healthcare, but a key challenge in addressing quality of care in maternal
health, is the lack of accurate data. We present a review of the difficulties of
collecting and analyzing maternal health data. We propose a comprehensive
informatics monitoring framework to track progress on the achievement of the
international targets and priorities toward ending preventable maternal mortality and
improving maternal and child health, that at the same time builds capacity at
institutional and country level to collect indicators and to generate actionable and
comparable knowledge that facilitates analysis, research, and evidence-based
decision making.
Keywords. Health information technology (HIT), maternal, child
1. Introduction
According to a 2016 systematic analysis by the United Nations (UN), approximately 830
women die every day around the world from preventable causes related to pregnancy,
and 99% of all maternal deaths occur in low- and middle-income countries (LMICs) [1].
A large and growing body of research suggests that most cases of maternal deaths could
be avoided with timely access to quality healthcare [2]. A key challenge in addressing
quality of care in maternal health, is the lack of accurate data. For example, in many low-
income countries, maternal deaths go uncounted and frequently the cause of death is
unknown or not recorded correctly and the maternal care process is equally poorly
registered or not registered at all [3]. Many patient registration systems and electronic
health records in low resource settings have problems with non-standardized record-
keeping techniques which result in missing records, inconsistencies, poor data quality,
and inaccuracies and hence undermine evidence-based decision making in healthcare
service delivery [4]. This makes it difficult for national health programs to allocate
resources where they are needed the most. To achieve this goal is necessary the
integration and harmonization of high amounts of heterogeneous medical data that is
stored in different health information systems. Such a task is challenging in both
developed [5] and developing countries [6]. In this paper, we review some of the
challenges in collecting and analyzing this data, and we propose an ontology-based data
integration approach to effectively combine data from heterogeneous sources.
Improving Usability, Safety and Patient Outcomes with Health Information Technology
F. Lau et al. (Eds.)
© 2019 The authors and IOS Press.
This article is published online with Open Access by IOS Press and distributed under the terms
of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0).
doi:10.3233/978-1-61499-951-5-157
157
2. Current Approaches
Comprehensive database applications for a domain can reduce variation within that
domain. There are some proposals such as the Perinatal Information System (SIP)
developed by the Pan American Health Organization (PAHO/WHO/CLAP). SIP’s aim
is for the health team to learn about the characteristics of the health service users, assess
the outcomes of the care provided, identify the priority problems and conduct a
operational studies [7]. It contains a model of perinatal clinical history with pre-codified
and open data, 170 variables entered by clinicians or under their supervision. SIP has
been modified several times due to the need to keep their contents updated, as well as to
include the priorities - national and international - defined by the Ministries of Health of
the region. It also allows automated report production and the transferring of local data
across institutions. The tutorial handbook contributes to the record’s consistency. In a
study [8] in 20 maternity hospitals (5 Countries, 40% Private and 60% Public) 85% had
a reliable information system by the third year of use of SIP. 15% of hospitals still had
problems at that time that were already clear during the second year. The evaluation of
the impact of yearly reports shows that 58% of recommendations were fulfilled,
especially those regarding the complete filling-in of clinical records (62%) and to a lesser
extent, variables that reflect clinical practices and organization of services (52%).
One of the most comprehensive and proven perinatal datasets is the one
implemented at the Medical University of South Carolina (MUSC) Perinatal Information
System (PINS). A validated, research-quality perinatal database with multiple edits and
audits to ensure accuracy [9], for all women delivering at the MUSC, which is a regional
tertiary referral hospital in the southeastern United States. The MUSC PINS database
includes detailed information on each mother's medical history, linked to neonatal data
(such as medical diagnoses, medications, and laboratory tests) from delivery to hospital
discharge. However, even though it is a statewide regional perinatal information system,
comprehensive antenatal care information from outside the hospital setting is not
available.
The Netherlands established national domain information models to support
electronic information exchange based on HL7 RIM, using cases from perinatology as a
national pilot, with the aim to support the development, adoption, implementation, and
maintenance of the EHR in Dutch healthcare practice [10]. They chose perinatology
because there was an existing need for communication improvement with a sufficient
consensus and standardization among different professionals represented in a national
data set. Their approach was to allow clinicians to understand better where ‘their’
information is in the Domain Message Information Model (D-MIM) to individually
analyze each information item, attribute and value in the domain and map it to existing
HL7 RIM classes, attributes, and vocabularies. They found that in some instances,
additional agreements are necessary about the preferred vocabulary in the Netherlands,
because the professional organizations need to harmonize their materials. Another
finding was that the limitations are reached for what should be part of the (national)
standard, and what professional organizations should develop and maintain within their
realm.
The Global Network Maternal Newborn Health Registry (MNHR) provides
prospectively collected, population-based pregnancy outcomes for defined geographic
regions within low- and middle-income countries [11]. Its data describes demographic
and healthcare characteristics and major outcomes of pregnancy. All definitions used by
the MNHR are consistent with the WHO definitions, whenever possible. One of the
J. Henao et al. / An Informatics Framework for Maternal and Child Health (MCH) Monitoring158
limitations of the MNHR is the difficulty in ensuring the inclusion of all pregnancies,
and especially those with early pregnancy loss. Some sites encounter challenges in
tracking the outcomes of pregnant women who migrate in or out of the study clusters.
Other challenges include categorizing critical pregnancy outcomes, determining accurate
birth weights of certain groups of infants e.g., stillbirths, infants delivered at home. The
MNHR also is a tool for evaluating the effectiveness of strategies of care because, unlike
with the use of periodic surveys, data is collected continuously over time within the same
population-based cohort. This enables investigators to determine the impact of
interventions to improve outcomes, to monitor trends over time, and to evaluate the
changing patterns of perinatal care to inform health policy.
3. Current Issues
Regarding the consensus on data indicators, some issues persist. For example, despite
the global burden of perinatal deaths, there is currently no single, globally acceptable
classification system for perinatal deaths. Instead, multiple, disparate systems are in use
worldwide. The World Health Organization (WHO) is developing a globally acceptable
classification approach for perinatal deaths [12] but these have not been universally
adopted. While the integrated WHO tool is designed to assess quality across the
continuum of care, the standards currently included in the tool are not fully representative
of all the areas of care that need to be assessed. Antenatal care is not assessed at all and
postnatal care in a very limited way. These are typically neglected areas of care that are
often not included in quality improvement activities. This is in part because national
standards for antenatal and postnatal care are often not in place. Developing such
standards and including them in a comprehensive quality of care assessment is a priority.
The inter-country differences in registration systems, also imply biases in recorded
mortality rates. The challenge is to distinguish ‘real’ variations in the value of an
indicator from variations due to differences in registration practices and definitions and
from random variation [5]. From a practical point of view, a compromise must be struck
between useful, important indicators that satisfy many of the formal characteristics and
are still accessible. Mortality indicators are particularly sensitive to biases related to the
construction of indicators. For example, changes in birth notification and registration
practices can cause major biases. In 1994 Germany reduced the lower limit for birth
weight for registration of fetal deaths from 1000 to 500 g. Consequently, the perinatal
mortality rate jumped suddenly from 5.5 per 1000 to 6.6 per 1000, an increase of 20%
[13].
Databases using the International Statistical Classification of Diseases and Related
Health Problems (ICD) can facilitate cross-country comparisons, but revisions can alter
the results of comparisons. Regarding perinatology, in its 10th revision, chapters ‘‘O’’,
‘‘P’’ and ‘‘Q’’ are relevant to perinatology. An analysis of these codes shows that 163
ICD9 codes are mapped onto 235 ICD10 codes in chapter P, and 180 ICD9 codes for
anomalies onto 620 ICD10 codes [14]. Changes in the ICD version used to register
causes of death or morbidity will consequently result in systematic shifts in the overall
levels reported. The World Health Organization (WHO) and collaborating partners are
developing the WHO Application of ICD-10 to perinatal deaths: ICD-Perinatal Mortality
(ICD-PM) [12]. Tables comparing causes of death and morbidity across countries should
explicitly state the ICD version used for coding.
J. Henao et al. / An Informatics Framework for Maternal and Child Health (MCH) Monitoring 159
Some countries have taken steps to homogeneous coding practices on a national
level. For instance, the Danish society of gynecology and obstetrics has elaborated a
guideline for registration of births which selects a number of codes from ICD10 and the
Nordic Classification of Surgical procedures and Treatments that were found to be
relevant for registration on a national level, with additional definitions and criteria for
use where necessary [15]. In general, the burden on individual providers of collecting
data has been well documented [16], as has the lack of use of data collected at such great
cost [17], which breaks the feedback mechanism whereby monitoring and review can
result in improved provision of interventions.
Another challenge is data aggregation and overlap. For maternal care, clinical data
is often generated from various sources (prenatal screenings, primary care providers,
midwives) and the health information may exist in both paper-based and computer-based
systems at institutions located in different geographical locations. The overlap across
systems introduces the potential for data variation through duplication of data entry and
differing concept definition or context of use. Studies show that redundant and
inconsistent records lead to errors, extra effort, misdirected data, over-reliance on the
spoken word, inaccuracies, information loss, limited standardization,
miscommunications, decision changes, and limited outcomes evaluations [18]. Also,
failure to share patient information across data systems can lead to inefficiency and
reduce the quality of care. One study [19] pointed out the deficiency in communication
among health professionals and that both lack of communication and lack of clarity of
medical records are major causes of medical incidents. Research has shown how
coordination and communication among clinicians and across settings resulted in greater
efficiency and better clinical outcomes [20]. An Institute of Medicine report [21]
explained that a health system must have efficient and accurate ways of capturing,
managing, and analyzing clinical data collected at all the different sites where care is
provided.
Also, the course of pregnancy, childbirth and child development involves a series of
stages referred to as the prenatal, intrapartum and postnatal periods of care, involving
several medical disciplines during each stage, using a variety of technical jargon
registered in different systems. The ability of communication among EHRs that contain
such kind of information, which would allow interoperability, requires that terms in all
involved systems share their semantics. However, gathering information from EHRs
connected to different information systems is a challenge and involves the adoption of
semantic interoperability solutions. To address this, the healthcare sector has developed
standards for medical vocabulary (SNOMED-CT) and message information models
(FHIR) that carry many of the features present in Semantic Web standards such as the
Web Ontology Language (OWL). For example, Implementing FHIR in MCH domain,
requires additional structure definitions and rules about which resource elements and
terminologies map to particular MCH requirements [22]. Semantic interoperability is
then also needed because of the seemingly arbitrary meaning of data across different
health sectors, which may result to classification errors when collecting data. The
solutions based on formal ontologies can enable the effective semantic interoperability
because for systems to interoperate, they have to share the meaning of their terms, which
requires a well-defined semantics.
Obstetric and Neonatal Ontology (OntONeo) [23], aims to represent the diversity of
data registered in EHRs involved in pregnancy care. Such ontology will be able to join
different standards and terminologies adopted by information systems that deal with
prenatal EHRs and provides a demanded specialized vocabulary planned to include a
J. Henao et al. / An Informatics Framework for Maternal and Child Health (MCH) Monitoring160
more comprehensive formal representation in comparison with other currently available
ontologies and terminological resources. OntoNeo still needs additional validation in
different communities of physicians and healthcare professionals.
4. Towards a Comprehensive Framework for Maternal Health Informatics
The still high maternal mortality ratio (MMR) could be explained because gains in
coverage do not always result in safe and high-quality obstetric care due to limitations
of training and process improvements. To achieve sustained improvements, local groups
will need not only need outcomes metrics and education on best practices for care but
also to develop ways to examine their current care delivery process and identify areas
for improvement: ‘What gets measured gets managed’ [24]. Studies have also shown
that medical knowledge, job satisfaction, and self-efficacy do not increase by only using
continuing medical education (CME) intervention and that using only one mode of
learning fails to stimulate lateral learning i.e. learning from your peers [25]. We are
currently developing a comprehensive set of metrics for maternal outcomes and process
variables that will be useful for low- and middle-income countries. An Alicanto™
(http://www.alicantocloud.com) social community education site is being established for
maternal health centers in Latin America to have access to evidence-based education and
best practices to collect outcomes through the continuum of care, keeping standardization
of clinical structure and content across all databases; while being technologically and
culturally appropriate. Based on our review of other maternal health databases, an initial
set of consensus metrics will be used to track outcomes. An online asynchronous
discussion forum will be used for communities of practice to share their experiences and
discuss challenges in care delivery and data collection with colleagues. Through the
community site, we will provide support and tools on how to collect and analyze that
data for quality and process improvement, but we believe that a co-creating approach to
developing metrics, is more successful, engageable and sustainable. Of particular interest
is what process-oriented data can be collected to measure quality of care delivery in low
resource settings.
5. Conclusions
There is a global need to end preventable maternal deaths and to improve maternal and
child health. Despite multiple approaches, there is no universal consensus on their
implementation, causing discrepant data indicators, heterogeneous coding practices, and
data overlap. There are also difficulties in technical and semantic interoperability,
causing deficiencies in communication among health professionals. As a result, health
systems and governments have very limited outcomes evaluations. We propose a
comprehensive informatics monitoring framework that will be created based on a
consensus community of practice and an ontology-based data integration approach, in
which there is not only data collection, but processes variables are included and can be
used in a feedback mechanism to improve training and monitoring. This approach will
build capacity at institutional and country level to generate actionable and comparable
knowledge that facilitates analysis, research, and evidence-based decision making.
J. Henao et al. / An Informatics Framework for Maternal and Child Health (MCH) Monitoring 161
References
[1] Alkema et al. Global, regional, and national levels and trends in maternal mortality between 1990 and
2015, with scenario-based projections to 2030: a systematic analysis by the UN Maternal Mortality
Estimation Inter-Agency Group. Lancet, 387 (2016), 462-74.
[2] Say et al., Global Causes Of Maternal Death: A WHO Systematic Analysis. The Lancet Global Health,
2, 2014, e323–e333.
[3] S. Aziz and M. Rao, Existing record keeping system in government teaching hospitals of Karachi, Journal
of the Pakistan Medical Association, 52 (2002), 163–173.
[4] Saadia et al, A Granular Ontology Model for Maternal and Child Health Information System, Journal of
Healthcare Engineering, 2017 (2017), Article ID 9519321.
[5] N. Lack, et al., Methodological difficulties in the comparison of indicators of perinatal health across
Europe, European Journal of Obstetrics and Gynecology, 111 (2003), S33 - S44.
[6] Z.A. Bhutta, A. Hafeez, What can Pakistan do to address maternal and child health over the next decade?
Health Research Policy and Systems, 13 (2015), 49.
[7] Paho.org. 2018 [cited 25 Julio 2018]. Available from: https://www.paho.org/
[8] Simini et al., Perinatal Information System. Incorporation latency and impact on perinatal clinical registry,
Ginecología y obstetricia de México, 69, (2001), 386-9.
[9] D. Annibale, T. Hulsey, L. Wallin, et al., Clinical diagnosis and management of respiratory distress in
preterm neonates: effects of participation in a controlled trial, Pediatrics, 90 (1992), 397–400.
[10] W.T. Goossen et al., Electronic patient records: domain message information model perinatology. Int J
Med Inf , 70 (2003), 265-76.
[11] C.L. Bose et al., The Global Network Maternal Newborn Health Registry: a multi-national, community-
based registry of pregnancy outcomes, Reprod Health, 12 (2015), Suppl 2:S1.
[12] A.M. Wojcieszek et al., Characteristics of a global classification system for perinatal deaths: A Delphi
consensus study, BMC Pregnancy and Childbirth, 15 (2016), 223.
[13] W.C. Graafmans et al., Comparability of published perinatal mortality rates in Western Europe: the
quantitative impact of differences in gestational age and birthweight criteria, Bjog, 108 (2001), 1237–45.
[14] E. Allanson, Ö. Tunçalp, J. Vogel, Ending the silence: the WHO application of ICD-10 to perinatal deaths
(ICD-PM), WHO Bull, 2016, In press
[15] Danish Society of Obstetrics of Gynecology and Obstetrics. www.dsog.dk (in Danish); 2003.
[16] World Health Organization. A rapid assessment of the burden of indicators and reporting requirements
for health monitoring, Geneva: World Health Organization; 2014.
[17] Farinelli, Interoperability Among Prenatal EHRs: A Formal Ontology Approach. The American Medical
Informatics Association (AMIA) Symposium, (2016), 10.13140/RG.2.2.16743.34729.
[18] C. AbouZahr, T. Boerma, Health information systems: the foundations of public health, Bulletin of the
World Health Organization, 83 (2005), 578-583.
[19] A.L. Bhasale et al., Analyzing Potential Harm in Australian General Practice: An Incident-Monitoring
Study, The Medical Journal of Australia, 169 (1998), 73-76.
[20] Rosenal et al., Support for information management in critical care: a new approach to identify needs.
Proceedings, Symposium on Computer Applications in Medical Care, (1995), 2-6.
[21] Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century,
Washington D.C.: National Academy Press; 2001.
[22] Farinelli et al., Interoperability Among Prenatal EHRs: A Formal Ontology Approach, The American
Medical Informatics Association (AMIA) Symposium, (2016), 10.13140/RG.2.2.16743.34729.
[23] Farinelli et al., OntONeo: The Obstetric and Neonatal Ontology, International Conference on Biomedical
Ontology (2016), Oregon, USA
[24] A. Moran, A. Moller, D. Chou et al., ‘What gets measured gets managed’: revisiting the indicators for
maternal and newborn health programs, Reprod Health,15 (2018).
[25] C.J. Gill et al., The mCME Project: A Randomized Controlled Trial of an SMS-Based Continuing
Medical Education Intervention for Improving Medical Knowledge among Vietnamese Community
Based Physicians’ Assistants, PLoS ONE, 11 (2016), 11.
J. Henao et al. / An Informatics Framework for Maternal and Child Health (MCH) Monitoring162
