India’s Health Data System Scores High on Consistency but Stumbles on Accuracy

India runs one of the largest routine health data collection enterprises on Earth. Every month, hundreds of thousands of public health facilities across the country feed information into a centralized Health Management Information System, known as HMIS, which tracks everything from antenatal care registrations to childhood immunizations. That data stream is supposed to guide how the world’s most populous nation plans its health programs and allocates its scarce resources. But a new peer-reviewed assessment, published in BMC Health Services Research, has put the system under the microscope for the first time at this scale, and the results reveal a striking paradox: the machinery of data collection has become impressively standardized, yet the numbers it produces do not always match reality on the ground.

The study, conducted by Nibras K. Thodika, Althaf Shajahan, and Muhammad Shafi of the Department of Management Studies at the National Institute of Technology Calicut in Kerala, examined fifteen priority maternal and child health data elements drawn from the sub-district annual performance reports available on the national HMIS portal. The analysis covered eight reporting years, from 2012-13 through 2019-20, giving the researchers a longitudinal window into how the quality of India’s routine health facility data has evolved since the system’s early days. Rather than relying on anecdote or single-state snapshots, the team applied the World Health Organization’s data quality review framework, a standardized methodology that health authorities around the world use to audit the trustworthiness of routine health information systems.

The WHO framework breaks data quality into several distinct dimensions, and the Indian findings diverge sharply depending on which dimension is under scrutiny. The first is availability, meaning whether facilities actually report data at all. Here the news is encouraging: data availability increased steadily over the study period, a sign that the National Health Mission’s push to bring facilities into the reporting fold has largely succeeded. The second dimension is completeness, meaning whether facilities that do report fill in all the required fields. This is where the first crack appears. Completeness actually declined over time, falling from 96.2 percent in the earliest years to 90.7 percent by the end of the study window. In other words, as more facilities joined the system, the proportion of missing values within their reports quietly grew.

That decline was not uniform across the country. The researchers found comparatively better completeness levels among India’s so-called high-focus states, the group of states that receive intensified support under the National Health Mission because of historically weaker health indicators. This finding is counterintuitive in an interesting way: the states often assumed to have the weakest administrative capacity are, at least in terms of filling out their reports thoroughly, outperforming expectations. It suggests that the mission’s targeted investments in data systems and reporting infrastructure in those states may have paid dividends, and it hints that declining completeness elsewhere may reflect complacency or resource constraints in states where reporting is less closely monitored.

The third quality dimension the team assessed was internal consistency, which asks whether the numbers within the system hang together logically. One test involves outlier detection: extreme values that fall far outside plausible ranges often signal data entry errors or reporting anomalies. On this front, India’s HMIS performed well, with the prevalence of outliers remaining below 5 percent across the indicators examined. Another test examines consistency over time, checking whether reported values fluctuate implausibly from year to year. Here again the trend was positive, with consistency improving toward the later years of the study period, suggesting that the system matured and stabilized as reporting routines became entrenched.

But internal consistency has a deeper layer, and this is where the assessment uncovered one of its most significant problems. The researchers examined four pairs of logically related indicators, checking whether the arithmetic relationships between them held. For example, the number of pregnant women who completed four antenatal care visits cannot plausibly exceed the number who registered for antenatal care in the first place, and vaccination counts for specific antigens like the oral polio vaccine given at birth or the Bacille Calmette-Guérin tuberculosis vaccine should cohere with broader immunization figures. For two of the four data pairs considered, consistency between related indicators was low. Because no major differences emerged between states on this measure, the problem appears to be systemic rather than regional, embedded in the way data are captured, aggregated, or defined rather than in the practices of any particular state’s health bureaucracy.

The fourth and arguably most consequential dimension is external consistency, which asks whether the routine system’s numbers agree with independent estimates from other sources. The gold standard for comparison in India is the National Family Health Survey, the country’s Demographic and Health Survey, whose fifth round, NFHS-5, provided the benchmark estimates for this study. The verdict was sobering: concordance between HMIS figures and survey-based estimates was generally low, and the degree of agreement varied depending on the indicator and on which denominator was used to convert raw counts into rates. When the researchers used census-based denominators to calculate coverage rates, agreement weakened further, indicating that part of the mismatch stems from uncertainty about the true size of the population each facility is supposed to serve.

One indicator stood out as particularly problematic. Agreement between the routine system and the survey was consistently weakest for the indicator tracking pregnant women who completed four antenatal care visits. This matters enormously, because four antenatal visits is a cornerstone indicator of maternal health service delivery, one that international bodies and Indian policymakers alike use to gauge whether women are receiving the care that protects them and their babies during pregnancy. If the routine system systematically misstates antenatal care coverage, programs designed to close gaps in maternal care may be aimed at the wrong targets, and resources may flow to districts that do not need them while genuinely underserved areas are overlooked.

Taken together, the study’s findings sketch a system that has mastered the mechanics of data collection but not yet the discipline of data accuracy. The authors conclude that India’s National Health Mission has achieved high levels of standardization in setting up its health management information system for managing maternal and child health, a genuine achievement given the scale and diversity of the country’s public health apparatus. Yet the same assessment revealed systemic data quality issues in routine health facility reporting that have direct implications for data use. Poor-quality data are not a merely technical nuisance; when program and policy implementers rely on flawed numbers to guide future plans and allocate scarce resources, the consequences ripple outward to the health of the very populations the system is meant to serve.

The path forward, the researchers argue, lies in improving HMIS data collection procedures and devising innovative strategies for utilizing the data that are collected. The study’s granular diagnosis offers a roadmap: completeness requires renewed attention, particularly outside the high-focus states; the logical inconsistencies between related indicators demand scrutiny of reporting definitions and aggregation practices; and the persistent gaps between routine counts and survey estimates call for better population denominators and closer triangulation between HMIS and periodic surveys. For a country whose health policy ambitions increasingly depend on real-time, facility-level evidence, the message of this assessment is clear: the pipes are in place, but the water flowing through them still needs to be trusted before it can be drunk.

Subject of Research: Data quality assessment of India's Health Management Information System for maternal and child health indicators

Article Title: Quality of routine health facility data: an assessment of India’s health management information system for maternal and child health indicators

Article References: Thodika, N. K., Shajahan, A., & Shafi, M. (2026). Quality of routine health facility data: an assessment of India’s health management information system for maternal and child health indicators. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15618-4

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15618-4

Keywords: HMIS, data quality, maternal and child health, India, National Health Mission, WHO data quality review framework, NFHS-5, antenatal care, health information systems, external consistency, routine health facility data, health policy