Work

Benefits and Risks of Using Health Data for Research: Report of the Health Data Hub Scientific Advisory Board-October 2023

Year: 2023
Type: report
Abstract: As the digitalisation of the healthcare system becomes a reality in developed countries, the availability of large volumes of health data is raising stakeholders' expectations, sometimes excessively, and concerns, sometimes groundlessly. A review of the benefits that can legitimately be expected from exploiting the available data is proposed, while considering the adverse effects that have already been documented or can legitimately be anticipated, before determining where the benefit/risk balance stands. While the current criteria place the emphasis on the absolute protection of privacy, which hinders or delays many highly relevant studies, we propose that the benefit/risk balance should henceforth form the basis for decisions to authorise the use of health data for research purposes.Many benefits are expected from providing access to this data :(1) Facilitating the epidemiological surveillance of major diseases, upon which many public health decisions are based.(2) Enabling the measurement of healthcare efficacy and the detection of local or regional situations that deviate significantly from the mean.(3) Facilitating the monitoring of changes in practices over time and their impact.(4) Helping to document the positive and negative effects of innovations introduced as a result of changes in practices, be they medicinal, technical or organisational.(5) Providing information about patients' access to healthcare services, allowing rates and delays to be calculated and cross-referenced with geographical variables and socio-economic data in order to identify inequalities, effective and ineffective strategies, and determine which areas should be prioritised for improvement.(6) Contributing directly to improving knowledge – an essential step towards improving practices and designing effective and protective public policies.(7) Playing an essential role in developing the uses of artificial intelligence, based on new, high-quality algorithms which can be trained and tested on representative populations, which is not always the case at present.(8) Enabling the assessment of the effectiveness of prevention or screening campaigns, and of the extent to which best practice recommendations are followed.However, the use of such data by multiple stakeholders is not without risks.(1) The risk of revealing personal data is a major risk for personal data, but very minor for pseudonymised data and non-existent for anonymised data.(2) The risk of uncontrolled use, beyond the need to inform people, depends on operators complying with the rules for use, and is very closely monitored;(3) The risk of data theft and espionage by industrial operators or governments is often mentioned by opponents, but it is extremely low because non-identifying data has no commercial value, and this type of theft has never occurred.(4) The risk of infringement of intellectual property rights and of financial loss is often mentioned as an argument against the valorisation of this data but has not been documented;(5) Healthcare professionals are concerned about the risk of normative surveillance, although such surveillance is legitimate from the perspective of healthcare system funders, and tensions can be eased through dialogue.The potential risks to individuals and the community are minimal if the protective measures adopted are effectively implemented and none of the risks are liable to have a significant impact (on health), either individually or collectively. However, the use of such data can have substantial benefits for society (e.g. appropriateness of care, prevention of health scandals, evaluation of innovations) and for the individual (such as the development of new treatments), and may also have an impact on individuals themselves. The benefit/risk balance therefore leans strongly towards the benefits.This leaves the question of how to remove obstacles to the provision of access to data collections and promote a culture of the common good represented by health data throughout all levels of society. These obstacles are numerous and include:(1) a ban on the use of US data-storage tools in the absence of a European cloud, even though there have never been any complaints in the healthcare sector against US manufacturers for failing to comply with the GDPR in the healthcare field, nor any documented attempts to re-identify anonymised health research data;(2) a ban on using the social security numbers recorded in the French National Social Security Register to match databases that need to be cross-referenced, despite this practice being possible in many European countries that are subject to the same European regulations as France;(3) waiting times running into years for access to the main French National Health Data System (Système National des Données de Santé – SNDS) database, whereas this queue could be reduced if a copy of the main SNDS database were included in the HDH catalogue;(4) an ecosystem that is overly complex for historical reasons, and which needs to be simplified in order to move towards a Single Gateway, as recommended in the European directive in preparation;(5) public communications focusing solely on the risks and not presenting the benefits, despite well-informed citizens being in favour of the use of their data for research.This is likely to have a major impact on the feasibility of many studies. Auditing the barriers to research within the current system could help to convince the doubters.
Source: HAL
Language: English
Open Access status: green
OpenAlex ID: W4390867886
Cites: 6
Cited by: 3
Related to: 10