Mapping Data Sources for Local Decision-Making on Maternal and Child Health in Tribal Primary Health Centre Settings of Andhra Pradesh, India

This study reveals that while tribal Primary Health Centres in Andhra Pradesh generate abundant maternal and child health data, the ecosystem is heavily skewed toward administrative reporting and lacks accessibility for local decision-makers, necessitating a systemic reorientation to better serve tribal populations.

The Gist

The Big Picture: A Library Full of Books, But Locked Doors

Imagine a small village clinic in the tribal hills of India. The doctor there (the Medical Officer) wants to save lives, specifically helping mothers and babies stay healthy. To do this, they need information: Where are the pregnant women? Who is malnourished? Which roads are blocked during the rain? Which families have no income?

The problem? The village is sitting on a goldmine of data, but the doctor has no key to the vault.

This paper is like a detective story where researchers went to three of these tribal clinics to map out every single piece of information that exists versus what the doctor can actually see. They found a massive "Data Rich, Information Poor" paradox: The system collects tons of data, but it's all locked away in different rooms, leaving the doctor working in the dark.

---

The Investigation: How They Mapped the Territory

The researchers didn't just look at official reports; they acted like data archaeologists. They used a three-step dig:

1. Looking at the Paperwork: They read every register, form, and digital log.
2. Talking to the Workers: They interviewed the doctors, nurses, and community health workers to ask, "What do you use? What do you wish you had?"
3. Checking the Map: They brought everyone together to say, "Is this list accurate?"

They found 28 different sources of information that could help the clinic. Think of these as 28 different buckets of water.

---

The Findings: What Was in the Buckets?

The researchers sorted these 28 buckets into two different ways of looking at them.

1. The "HEALTHY" Buckets (What is the data about?)

They used a framework called HEALTHY (Healthcare, Education, Access, Labour, Transportation, Housing, Income) to see what topics were covered.

• The Imbalance: Imagine a pizza where 57% is just "Healthcare" (sick people, medicine, births).
• The Missing Slices: The other crucial toppings were tiny. There was almost no data on Transportation (can they get to the hospital?), Housing (do they have a roof?), or Income (can they afford food?).
• The Metaphor: It's like trying to fix a car engine by only looking at the spark plugs, while ignoring the fact that the car has no gas, no tires, and is stuck in a mud puddle. The doctors see the medical data, but they can't see the reasons why people are getting sick.

2. The "Origin" Buckets (Where did the data come from?)

They also looked at how the data was created:

• Administrative Data (The Bureaucracy): 46% of the data comes from government forms and reports.
• Designed Data (The Science): Only 14% comes from specific studies or surveys designed to answer local questions.
• The Metaphor: The system is full of "receipts" (we did this, we did that) but lacks "blueprints" (why is this happening here?).

---

The Real Problem: The "Locked Door"

Here is the most shocking part of the story.

Even though 68% of these data buckets have a digital component (they are on computers or apps), the doctors could only open 32% of them.

• The Upward Flow: Imagine a river. The data flows up the mountain to the capital city for the big bosses to count. But the river doesn't flow down to the village clinic where the water is needed.
• The "Black Box": Nurses and health workers enter data into apps every day. But the doctor in charge cannot see that data on their screen. It's like a chef cooking a meal, but the manager is the only one allowed to taste it. The chef doesn't know if the soup is salty until it's too late.
• The Silos: The Health Department has data. The Education Department has data. The Roads Department has data. But they don't talk to each other. The doctor at the clinic is standing in the middle, trying to connect dots that are on three different walls.

Why Does This Matter?

Because the doctor is flying blind.

• They can't see which specific pregnant woman is high-risk because the detailed list is locked.
• They can't map out where the sick people live because the data has no GPS coordinates.
• They can't plan for the rainy season because they don't have data on which roads will be washed out.

The Solution: Turning the Lights On

The paper suggests that we need to stop building more "data factories" and start building "data bridges."

1. Unlock the Doors: Give the local doctors the keys to the digital systems so they can see the individual names and details, not just the big numbers.
2. Mix the Buckets: Make the Health, Education, and Road departments share their data at the local level.
3. Draw the Map: Add location data (GPS) to the forms so doctors can see exactly where the problems are.
4. Listen Locally: Create simple ways for doctors to ask their own questions and get answers, rather than just waiting for reports from the capital.

The Bottom Line

The tribal clinics in India are data rich (they have the information) but information poor (they can't use it). To save more mothers and children, we don't need to collect more data; we need to stop hoarding it and start sharing it with the people who need it most: the doctors on the front lines.

Technical Summary

1. Problem Statement

The study addresses the persistent paradox in Low- and Middle-Income Countries (LMICs) described as "data rich, information poor." While health systems collect vast amounts of data, this information rarely translates into actionable insights for local decision-making. This issue is exacerbated in tribal settings in India due to:

• Geographic isolation and fragmented governance: Tribal areas are often administered by Integrated Tribal Development Agencies (ITDAs) where health, education, and infrastructure fall under different departmental hierarchies with siloed data systems.
• Lack of disaggregated data: Routine health management information systems (HMIS) often fail to generate or track tribal-disaggregated data, rendering tribal populations "invisible" in aggregate statistics.
• Accessibility gaps: Medical Officers (MOs) at Primary Health Centres (PHCs) lack access to the specific data needed for local planning, such as social determinants of health (SDOH) and individual-level beneficiary tracking.

2. Methodology

The study employed a participatory data discovery approach embedded within an action research framework (diagnosing phase).

• Study Setting: Three PHCs (Boduluru, Gangavaram, Vadapalli) under the ITDA Rampachodavaram in Alluri Sitarama Raju District, Andhra Pradesh, India. The region is predominantly tribal (70%+ Scheduled Tribe population) with challenging terrain.
• Data Collection Process (Oct 2023 – Mar 2024):
  1. Document Review: Analysis of registers, reporting formats, national guidelines, and digital system documentation (HMIS, RCH Portal, ANMOL, etc.).
  2. Key Informant Interviews: Semi-structured interviews with Medical Officers, staff nurses, Auxiliary Nurse Midwives (ANMs), Accredited Social Health Activists (ASHAs), and ITDA coordinators to identify formal and informal data sources.
  3. Stakeholder Validation: A workshop with PHC staff to verify the inventory, correct classifications, and validate accessibility assessments.
• Analytical Frameworks:
  • HEALTHY Classification: Categorizes data by domain (Healthcare, Education, Access to healthy choices, Labour, Transportation, Housing, Income).
  • Keller Data Discovery Typology: Categorizes data by origin (Designed, Administrative, Opportunity, Procedural).
• Accessibility Assessment: Data sources were classified as Accessible (routine access), Partially Accessible (requires requests/delays), or Inaccessible (cannot be retrieved for local use) by the Medical Officer.

3. Key Contributions

• First Systematic Local Mapping: This is the first study to systematically map the full complement of data sources available to Medical Officers in tribal PHC settings in India, moving beyond national-level infrastructure assessments.
• Dual-Framework Application: The study uniquely combines the HEALTHY (thematic) and Keller (structural) frameworks to reveal both what data is missing (thematic gaps) and how data is generated/accessible (structural gaps).
• Participatory Methodology: Demonstrates the value of involving frontline workers in data discovery to uncover informal practices, workarounds, and cross-departmental data flows that formal documentation misses.

4. Key Results

The study identified 28 distinct data sources relevant to Maternal and Child Health (MCH).

A. Distribution by Domain (HEALTHY)

• Healthcare Dominance: 57.1% (16 sources) of data relates strictly to clinical services and disease surveillance.
• Underrepresented Determinants: Critical social determinant domains are severely lacking:
  • Transportation & Access to healthy choices: 3.6% each.
  • Education & Labour: 7.1% each.
  • Housing & Income: 10.7% each.
• Implication: Medical Officers lack systematic data on the environmental and socioeconomic factors driving health outcomes in tribal communities.

B. Distribution by Origin (Keller)

• Administrative Data: Predominant (46.4%), generated through routine government processes (e.g., HMIS, RCH Portal).
• Opportunity Data: 21.4% (e.g., mobile app data, satellite imagery).
• Procedural Data: 17.9% (e.g., meeting minutes, stock records).
• Designed Data: Scarce (14.3%), indicating a lack of local survey or research capacity to generate data for specific local questions.

C. Accessibility Crisis

• Low Accessibility: Only 32.1% (9 sources) were fully accessible to Medical Officers.
• Inaccessibility: 57.1% (16 sources) were completely inaccessible at the PHC level.
• The Digital Paradox: Despite 67.9% of sources having digital components, only 21.1% of digital sources were fully accessible.
  • Example: ANMOL and E-ASHA apps allow frontline workers to enter individual-level data, but Medical Officers cannot view this data; it flows upward to district supervisors.
  • Example: The RCH Portal contains individual-level beneficiary data, but MOs can only view aggregate reports without district approval.

D. Technical Characteristics

• Format: Mixed landscape (35.7% digital-only, 32.1% paper-only, 32.1% hybrid).
• Spatial Data: Only 14.3% of sources contained geographic coordinates, limiting map-based decision-making.
• Granularity: While individual-level data exists (e.g., RCH Portal), it is inaccessible to local planners, forcing reliance on aggregated facility-level data.

5. Significance and Implications

• Reorienting Health Information Systems (HIS): The study concludes that current HIS architectures prioritize administrative control (upward reporting) over care-based rationality (lateral local decision-making).
• Addressing the "Data Rich, Information Poor" Paradox: The findings provide empirical evidence that the problem is not a lack of data generation, but a structural failure in data flow and accessibility.
• Policy Recommendations:
  1. Grant Access: Enable Medical Officers to access individual-level data in systems like RCH Portal for case-based tracking.
  2. Cross-Departmental Integration: Establish mechanisms at the ITDA level to share data between health, education, and livelihood departments.
  3. Spatial Integration: Incorporate geographic coordinates into routine reporting to support map-based planning in dispersed tribal areas.
  4. Local Data Generation: Develop capacity for "designed data" (rapid assessments) to answer local questions not covered by national indicators.
• Future Action: This mapping serves as the diagnostic baseline for subsequent action research cycles aimed at co-designing decision-support dashboards that integrate accessible data for local MCH planning.

In summary, the paper argues that improving MCH outcomes in tribal India requires shifting health information systems from being mere tools for administrative monitoring to becoming actionable platforms for local, context-specific decision-making.