COVID-19: Robust Collection of Health Data Will Ensure Better Health Policies

The spread of COVID-19 across the globe has been unmitigated and unprecedented. The containment or the spread itself is being measured by the fluctuation in the number of new cases or deaths being recorded each day. With countries like Singapore being lauded for the meticulous record-keeping, the importance of health data has been brought to the fore. 

In India, the recent surge in number of new cases and their consequent attribution to the gathering of the Tablighi Jamaat in New Delhi's Nizamuddin area, grabbed headlines and painted anew the modality of the spread of the virus. An infographic released by India Today claimed that almost 60% of the new coronavirus cases were linked to the Tablighi Jamaat event. However, a deeper dive into the numbers showed that current health data with regards to COVID-19 suffer from a sampling bias. That is, with the already few number of testings being carried out, the specific testing of just one cluster of the population has provided a statistic that potentially overestimates the contribution of said cluster. Current events, then, go to show just how disaggregated and unsystematic data potentially tilt statistics to favour or disfavour groups of people, thus colouring macro indicators and driving policy decisions which could have adverse effects on the health of the larger population. 

The government’s inadequate response towards data collection for the pandemic, then, raises broader questions about India’s health data practices. Given that it does have an extensive collection of data sets—such as the Civil Registration System (CRS), Survey of Causes of Death (SCD), Medical Certification of Cause of Death (MCCD), Sample Registration System (SRS), Annual Health Survey (AHS), National Family Health Survey (NFHS), District Level Household Survey (DLHS), surveys of the National Sample Survey Office (NSSO), National Nutrition Monitoring Bureau (NNMB), Health Management Information System (HMIS), and even the National Population Register (NPR)—has India been able to adequately use the data that it has accumulated over the years to counter health inequality and better its global position with respect to the human development index and sustainable development goals? 

The reading list, thus, explores the EPW archives to explore how India has fared in the documentation and upkeep of its health data. 

Disorganisation of Health Data

Given the enormity of its population, India has the potential to lead global health research and provide insights into its population and health indicators. However, Nandita Saikia and P M Kulkarni write that

 health data collection, documentation, and keeping for the same, has neither been systematic nor integrated. Due to this, the real extent of health inequality in the country remains restricted to the avenue of educated guesses. For instance, the recording of deaths in India remains understated by a large proportion. With over 75% of total deaths occurring at home, the system remains at a backfoot because the health delivery system in India is expected to record deaths in institutions. In 2005, with the introduction of the National Rural Health Mission (NRHM), demands for micro-level data on population and health increased in health facilities. This led to the introduction of the HMIS in 2008, which ambitiously wished to centralise data computing, and capture all vital events electronically. This was to ensure a quick collation and tabulation of data on maternal and child health indicators across India in a manner that was accessible and easy. However, HMIS remains handicapped in reporting accurate data on mortality, as not only do government health facilities grossly under-report deaths, but private health facilities do not report deaths at all. Due to this, though the conceptual ideal of HMIS remains laudable, the country’s systemic inability to ensure meticulous data recording inhibits the HMIS to be a successful exercise. 

Moreover, and perhaps most importantly, India lacks reliable statistics on the causes of death. This is because few registered deaths are medically certified, and the unregistered ones are not certified at all. Further, mechanisms such as the SCD(Rural) continue to suffer from unscientific approach and poor management. For instance, only four primary health centres (PHC) were deemed sufficient to collect information on the causes of death for a million people. These PHCs were chosen based on the availability of medical and paramedical staff who would be able to collect information from the area they served. This meant that information collected through the mechanism cannot be generalised to the entire population, but also over-rely on the fieldworker to ascertain the cause of death from the information conveyed to them by the respondent, and then relaying the information accurately to the medical officer. This left vast room for error. In addition to the poor management of the system is the fact that it does not report the socio-economic characteristics of the deceased, which means data for a differential cause of death does not exist. 

The cause of death becomes pertinent in dealing with data about emerging non-communicable diseases or the risk factors for emerging diseases. For instance, though the NFHS provides information of the prevalence of a few selected diseases, it does not provide any information on new emerging non-communicable diseases. 

All national-level morbidity data have two primary limitations. One is that they do not adequately document the behavioural and biological risk factors for communicable and non-communicable diseases. For example, the NSSO does not document the risk factors of emerging NCDs, such as alcohol use, diet, and physical inactivity, and the information that the NFHS collects on risk factors of emerging diseases is very limited. The other limitation of these surveys is that they cannot address how the prevalence of single diseases varies at administrative or ecological zones due to small sample size.

Ashish Bose agrees that there is no dearth of data sets in India. From the NFHS to the Rapid Household Survey, Multi Indicator Cluster Survey, SRS, and CRS, among so many others, India’s statistical institutions are bursting the seams with disaggregated data. However, nothing has been achieved with this enormous amount of data. Scholars have scarcely been able to evaluate them, and policy makers and planners rarely introduce them into the planning process. A large reason for the obsolete and deficient working of the statistical system is the obsessive reductionism of health to only focus on reproductive and child health. Donors have pumped money excessively into demographers and statisticians for whom modern-day health issues are only those restricted to issues of reproductive and child health.

It was American kubuddhi (bad advice) which led to the creation of two parallel departments in the ministry of health (department of health and department of family planning) in 1966. Foreign donor agencies poured in money for family planning and not health, somuchso that D Banerji commented: “Health has been hijacked by family planning”. My footnote was: “the plane crashed, killing both health and family planning.

Underestimation of Mortality Rates and Gender Biases in Data Relating to Death 

With an unreliable CRS, the SRS has been the only source of information allowing us to track how India fares on sustainable development goals, the human development index, as well as essential indicators such as sex ratios, mortality rates, etc. However, the quality of the SRS suffers from a glaring gender bias. Ajit Kumar Yadav and F Ram, in their examination of the SRS data, observe that between 1991–2001, 7.5% of male deaths (ages five+) were missed, whereas the corresponding figure for females was 2.1%. In the period 2001–2010, 4% of males deaths (ages five+) and 11% of female deaths (ages five+) were missed. This means that for these periods, macro indicators would have been grossly underestimated.

... With a few exceptions, the analysis clearly indicates a comparatively larger undercount of female deaths, as compared to males. In other words, female mortality is underestimated in the SRS; more so in recent times. Such a situation leads us to believe that whatever gains for females have been observed in terms either of under five mortality rates (U5MR) or female mortality may not be true.

Digital Surveillance and Redundancy of Data Systems

Though there are multiple sources collecting digital health data, how this data is being used and by whom remain key concerns in an era of digital surveillance. Sundeep Sahay and Arunima Mukherjee observe that with modern health systems becoming increasingly dependent on good quality health data, enabled through information and communication technologies, what information is generated from it and who has access to it, becomes pertinent for health outcomes as well as health rights. The authors note that though ministries of health are the primary users of data, major corporate interests are increasingly becoming major stakeholders in the development and management of these data systems. A key example of this is the health insurance sector. Though the accumulation of such data is essential keeping in mind the need to improve the efficacy of institutions and services, they could easily be misused to a great extent. 

A health-rights perspective will view data systems as enabling the rights to access and verification of information by civil society, to enhance state accountability and to enable democratic decision-making and citizen engagement. At stake are issues of privacy, confidentiality and data security, and how individual rights are challenged by the powers of an intrusive state or by market manipulations. Yet another major concern is how the use of data systems shapes the organisation of health services, and the relationships between providers and patients; this raises the question of whether the state continues to be the primary provider of health services to citizens, or acts only as a broker in the health marketplace.

Further, the authors note that public health data becomes redundant without appropriate denominators. Without this, indicators cannot be calculated with respect to a target population. For example, the percentage of newborn babies immunised can only be calculated in the presence of a denominator such as the “number of newborn babies for a period in a particular catchment area.” Without this, no meaningful intervention can be designed. However, India’s denominator figures are notoriously ambiguous with different departments having different figures and variations across health programmes and administrative levels. This begs the question whether the abundance of data systems for health in India merely stems from the need to meet targets and showcase some sort of activity on paper. 

To summarise, the HMIS reflects a high degree of redundancy, with little systematic actual use of data, raising the question of “why is this data being collected?” Is this exercise only to satisfy bureaucratic requirements and to enhance the monitoring capacity of the ministry to “see like a state”.. But, the irony is that the state can drown in the sheer volume of data, especially when they receive all subdistrict data without abstraction. This, coupled with inadequate analytical and action taking capabilities at the centre, undermines even the bureaucratic function of reporting and contributes to data playing primarily a symbolic and legitimising function.

Read More:

Measuring Catastrophic Healthcare Expenditure | Sunil Rajpal and William Joe, 2018

Quality of Data in NFHS-4 Compared to Earlier Rounds: An Assessment | K Srinivasan and Rakesh Mishra, 2020 

Quality Issues in the Health Management Information System: A Case Study of Bihar | Rajeev Kamal Kumar, 2018

Is India's Digital Health System Foolproof? | Aayush Rathi, 2019