Quality data for quality care: a solution for the most vulnerable

Our solution presents a comprehensive data suite which covers assessment of data quality, regular monitoring, and predictive analytics. By leveraging these techniques, we seek to improve the quality of care received by BRAC’s program participants, in addition to increasing operational efficiency of on-field personnel. 

Saiyeed Mashkur Al-Hakim, Head Technology for Development (T4D), BRAC Health Nutrition and Population Programme (HNPP)

COVID-19 revealed that low-income country health systems are poorly equipped to address pandemics while continuing to reach the most vulnerable groups with essential services. Community Health Workers (CHWs) have proven to be vital to the resilience of health systems in these countries amidst the COVID-19 crisis by maintaining access to critical services through door-to-door care. In Bangladesh, BRAC’s network of about 4,300 CHWs (supported by 43,000 community health volunteers) provides educational, diagnostic and curative care in the areas of maternal and child health, nutrition, infectious diseases, family planning and non-communicable diseases through household visits and on-call services. During the pandemic, CHWs have played a critical role in reaching underserved communities with essential prevention, education, and referral services when other services were almost shut. However, ensuring quality of care and efficient execution in community-based settings remains a persistent challenge. Robust and flexible data systems to track the quality of care delivered at the community level will be an essential part of recovering from COVID-19 stress on the health system and responding to future crises. By addressing data and care quality using open source digital tools, BRAC and IDinsight will be solving a critical challenge facing community health systems globally.

BRAC’s CHWs provide essential health care services for over 80 million people in Bangladesh. CHWs deliver care door-to-door, recognize symptoms, refer patients to local clinics or hospitals, and ensure treatment and therapy compliance. They also include services for families that include disease education and prevention; nutrition; sanitation; safe drinking water; and hygiene promotion. 

By strengthening digital tools for CHWs to deliver these services and monitor data quality, BRAC and IDinsight will improve health outcomes for individuals throughout Bangladesh by strengthening the resilience of the health system in the face of current and future health emergencies. The technology that will be developed, tested and applied at scale across BRAC’s CHW network will provide valuable insights that can be utilized by community health systems worldwide.

We will be creating the data assessment and prediction toolkit by means of open source resources, largely utilizing Python, and can commit to open sourcing the models we build and sharing publicly on GitHub. Further, we will write the data and program quality code so that it is straightforward for other large CHW organizations with digitized data collection to integrate into their systems, particularly if they are also using OpenSRP but also if they are using other platforms such as CommCare, Community Health Toolkit, and beyond. We will also be documenting process innovations that are needed to integrate a comprehensive, high-quality data system to serve 80 million people despite resource constraints faced by non-profit organizations and will share these findings in a peer reviewed publication and through a series of blogs posts that will translate findings to larger public health and technology communities. These outputs will also include lessons for local governments in low-income countries who operate at a similar scale as BRAC and who may face similar resource trade-offs. 

BRAC and IDinsight will develop data systems that will be used to drive improvements in program execution and health outcomes, ultimately leading to an increase in quantity and quality of care for populations vulnerable to pandemics and other health emergencies. 

Our solution will directly impact care quality for vulnerable groups including mothers and children. We will be able to use technology to identify quality gaps such as missed antenatal care (ANC) visits for pregnant women or missed diagnostics for patients with certain conditions (i.e. blood pressure not taken for patients with hypertension). Using advanced anomaly detection algorithms, we will identify when CHWs are improperly providing medications and ensure outliers are not due to underlying differences in the patient population. 

Data quality in low-income contexts such as slums and remote areas is often imprecise. Through the deployment of automated checks and machine learning algorithms, we will identify data errors and anomalies to improve data quality and understand population level health challenges within the communities served by BRAC’s CHWs. This will enable BRAC to more effectively serve these communities. 

We will demonstrate how the digitally-empowered CHWs of the future will play a crucial role in making health systems resilient during emergencies. 

BRAC’s Health, Nutrition and Population Programme (HNPP) is at the beginning of its data transformation journey, with all 4,300 of its community health workers currently equipped and trained with mobile devices to track workflows and patient outcomes. IDinsight is already partnering with BRAC HNPP to build the organization’s capacity to perform advanced data analytics to improve the performance of frontline community health workers. 

This proposed project would take this existing partnership to the next level. Over the first year, IDinsight will work to develop and test automated data quality solutions within subsets of BRAC’s CHW network. We envision piloting these solutions in two or three districts of Bangladesh while testing the solutions’ impact on local health worker performance and quality of care. After iterating and refining these solutions, we will scale it up to larger regions and ultimately, through BRAC’s entire network of nearly 50,000 community health workers and volunteers (as they also digitize their workflows), in all 64 districts of Bangladesh serving a catchment area of 80 million people. 

Our primary measures of success will be change in health outcomes. BRAC monitors many health outcomes associated with its services. 

As an example metric, we measure the percentage of women receiving antenatal care (ANC) during pregnancy and the number of ANC visits each woman attends. Previously we’ve been able to increase the percentage of women receiving antenatal care from 27% to 52% and increased the percentage of births occurring in health facilities from 15% to 59%. At the same time, the national average of births in health facilities only increased from 25% to 28%. 

As secondary metrics of the contribution of this program towards achieving those health outcomes, we will measure CHW performance and volume of data errors.

For CHW performance, we aim to improve or, at minimum, keep constant CHW productivity while quality of care increases. An example metric is household visits per month per CHW. 

Data errors will be measured through automatic data quality processes. Our improvements will prevent many types of errors from happening at all. Those that we cannot prevent but can measure, we will monitor over time. An example metric is the percentage of data points that are outside a reasonable range.

A major challenge will be the ongoing process of equipping and training BRAC’s CHWs to collect data using mobile devices without degrading the volume and quality of care they provide. Currently, BRAC has equipped all 4,300 individuals within its supervisory cadre of CHWs with mobile devices. This project was launched in late 2019 and is expected to be complete in mid-2021. However, among the challenges we have identified is the amount of time these health workers need to become habituated with the data entry system, and the quality of the data they are entering. These issues will be mitigated through consistent, supportive supervision and refresher training. However, we recognize that data quality in such environments will be an ongoing challenge. The solutions provided to BRAC by IDinsight will greatly streamline the process of identifying anomalies in our data, allowing our health workers to focus on what they do best -- providing quality services to their neighbors in need. The solution will allow us to more closely monitor and improve quality of care and ensure that the program is deploying needed resources based on credible data, especially in emergency situations.

BRAC HNPP’s partners include: Directorate General of Family Planning of Bangladesh, Directorate General Health Services of Bangladesh, Ministry of Local Government, Rural Development and Cooperatives of Bangladesh, and the International Center for Diarrhoeal Disease Research, Bangladesh to name a few.

When COVID-19 hit Bangladesh, we realized the importance of relying on quality data to support emergency response. We were attracted by Trinity Challenge’s convergence of partners dedicated to creating solutions to make sure that a pandemic on the scale of COVID-19 would never threaten the world again. To stop the next major pandemic, however, we need to bring global insights to bear on the world’s most vulnerable populations, including the world’s most crowded major country, Bangladesh. We recognize that Bangladesh has been a proving ground for major health interventions that have already saved millions of lives, including community healthcare and oral rehydration for children’s diarrhea. Bangladesh can play that role again with the use of advanced data analytics on large data sets collected by an organization like BRAC, with a health program reaching 80 million people. 

The principal partnership for this work is between BRAC and IDinsight. BRAC already partners with several of The Trinity Challenge Member organizations including the Bill & Melinda Gates Foundation and Microsoft AI for Health. Funding through the Trinity Challenge would allow BRAC and IDinsight to strengthen relationships with these and other Member organizations to utilize their expertise and leadership in support of developing and scaling our solution.