Centre for Internet & Society

In July this year, the National Health Authority (NHA) released the NHA Data Sharing Guidelines for the Pradhan Mantri Jan Aarogya Yojana (PM-JAY) just two months after publishing the draft Health Data Management Policy.

Reviewed and edited by Anubha Sinha

Launched in 2018, PM-JAY is a public health insurance scheme set to cover 10 crore poor and vulnerable families across the country for secondary and tertiary care hospitalisation. Eligible candidates can use the scheme to avail of cashless benefits at any public/private hospital falling under this scheme. Considering the scale and sensitivity of the data, the creation of a well-thought-out data-sharing document is a much-needed step. However, the document – though only a draft – has certain portions that need to be reconsidered, including parts that are not aligned with other healthcare policy documents. In addition, the guidelines should be able to work in tandem with the Personal Data Protection Act whenever it comes into force. With no prior intimation of the publication of the guidelines, and the provision of a mere 10 days for consultation, there was very little scope for stakeholders to submit their comments and participate in the consultation. While the guidelines pertain to the PM-JAY scheme, it is an important document to understand the government’s concerns and stance on the sharing of health data, especially by insurance companies.

Definitions: Ambiguous and incompatible with similar policy documents

The draft guidelines add to the list of health data–related policies that have been published since the beginning of the pandemic. These include three draft health data management policies published within two years, which have already covered the sharing and management of health data. The draft guidelines repeat the pattern of earlier policies on health data, wherein there is no reference to the policies that predated it; in this case, the guidelines fail to refer to the draft National Digital Health Data Management Policy (published in April 2022). To add to this, the document – by placing the definitions at the end – is difficult to read and understand, especially when terms such as ‘beneficiary’, ‘data principal’, and ‘individual’ are used interchangeably. In the same vein, the document uses the terms ‘data principal’ and ‘data fiduciary’, and the definitions of health data and personal data, from the 2019 PDP Bill, while also referring to the IT Act SDPI Rules and its definition of ‘sensitive personal data’. While the guidelines state that the IT Act and Rules will be the legislation to refer to for these guidelines, it is to be noted that the IT Act under the SPDI Rules covers ‘body corporates’, which under Section 43A(1), is defined as “any company and includes a firm, sole proprietorship or other association of individuals engaged in commercial or professional activities;”. It is difficult to add responsibility and accountability to the organisations under the guidelines when they might not even be covered under this definition.

With each new policy, civil society organisations have been pointing out the need to have a data protection act before introducing policies and guidelines that deal with the processing and sharing of the data of individuals. Ideally, these policies – even in draft form – should have been published after the Personal Data Protection Bill was enacted, to ensure consistency with the provisions of the law. For example, the guidelines introduce a new category of governance mechanisms under the data-sharing committee headed by a data-sharing officer (DSO). The responsibilities and powers of the DSO are similar to that of the data protection officer under the draft PDP Bill as well as the National Data Health Management Policy (NHDMP). This, in turn, raises the question of whether the DSO and the DPOs under both the PDP Bill and the draft NDMP will have the same responsibilities. Clarity in terms of which of the policies are in force and how they intersect is needed to ensure a smooth implementation. Ideally, having multiple sources of definitions should be addressed at the drafting stage itself.

Guiding Principles: Need to look beyond privacy

The guidelines enumerate certain principles to govern the use, collection, processing, and transmission of the personal or sensitive personal data of beneficiaries. These principles are accountability, privacy by design, choice and consent, openness/transparency, etc. While these provisions are much needed, their explanation at times misses the mark of why these principles were added. For example, in the case of accountability, the guidelines state that the ‘data fiduciary’ shall be accountable for complying with measures based on the guiding principles However, it does not specify who the fiduciaries would be accountable to and what the steps are to ensure accountability. Similarly, in the case of openness and transparency, the guidelines state that the policies and practices relating to the management of personal data will be available to all stakeholders. However, openness and transparency need to go beyond policies and practices and should consider other aspects of openness, including open data and the use of open-source software and open standards. This again will add to transparency, in that it would specify the rights of the data principal, as the current draft looks at the rights of the data principal merely from a privacy perspective. In the case of purpose limitation as well, the guidelines are tied to the privacy notice, which again puts the burden on the individual (in this case, beneficiary) when the onus should actually be on the data fiduciary. Lastly, under the empowerment of beneficiaries, the guidelines state that the “data principal shall be able to seek correction, amendments, or deletion of such data where it is inaccurate;”. The right to deletion should not be conditional on inaccuracy, especially when entering the scheme is optional and consent-based.

Data sharing with third parties without adequate safeguards

The guidelines outline certain cases where personal data can be collected, used, or disclosed without the consent of the individual. One of these cases is when the data is anonymised. However, the guidelines do not detail how this anonymisation would be achieved and ensured through the life cycle of the data, especially when the clause states that the data will also be collected without consent. The guidelines also state that the anonymised data could be used for public health management, clinical research, or academic research. The guidelines should have limited the scope of academic research or added certain criteria to gain access to the data; the use of vague terminology could lead to this data (sometimes collected without consent) being de-anonymised or used for studies that could cause harm to the data principal or even a particular community. The guidelines state that the data can be shared as ‘protected health information’ with a government agency for oversight activities authorised by law, epidemic control, or in response to court orders. With the sharing of data, care should be taken to ensure data minimisation and purpose limitations that go beyond the explanations added in the body of the guidelines. In addition, the guidelines also introduce the concept of a ‘clean room’, which is defined as “a secure sandboxed area with access controls, where aggregated and anonymised or de-identified data may be shared for the purposes of developing inference or training models”. The definition does not state who will be developing these training models; it could be a cause of worry if AI companies or even insurance companies have the potential to use this data to train models that could eventually make decisions based on the results. The term ‘sandbox’ is explained under the now revoked DP Bill 2021 as “such live testing of new products or services in a controlled or test regulatory environment for which the Authority may or may not permit certain regulatory relaxations for a
specified period for the limited purpose of the testing”. Neither the 2019 Bill nor the IT Act/Rules defines ‘sandbox’; the guidelines should have ideally spent more time explaining how the sandbox system in the ‘Clean Room’ works.


The draft Data Sharing Guidelines are a welcome step in ensuring that the entities sharing and processing data have guidelines to adhere to, especially since the Data Protection Bill has not been passed yet. The mention of the best practices for data sharing in annexures, including practices for people who have access to the data, is a step in the right direction, which could be made better with regular training and sensitisation. While the guidelines are a good starting point, they still suffer from the issues that have been highlighted in similar health data policies, including not referring to older policies, adding new entities, and the reliance on digital and mobile technology. The guidelines could have added more nuance to the consent and privacy by design sections to ensure other forms of notice, e.g., notice in audio form in different Indian languages. While PM-JAY aims to reach 10 crore poor and vulnerable families, there is a need to look at how to ensure that consent is given according to the guidelines that are “free, informed, clear, and specific”.

The views and opinions expressed on this page are those of their individual authors. Unless the opposite is explicitly stated, or unless the opposite may be reasonably inferred, CIS does not subscribe to these views and opinions which belong to their individual authors. CIS does not accept any responsibility, legal or otherwise, for the views and opinions of these individual authors. For an official statement from CIS on a particular issue, please contact us directly.