Forty years ago, the Ottawa Charter for Health Promotion stated something that policymaking still has not fully absorbed: our health is not determined by our healthcare system alone, but is shaped by factors such as housing, education or income. These social determinants of health do not operate in isolation; they mutually reinforce one another.
It has become increasingly clear that they are also mediated by the digital infrastructure – the algorithms, platforms and data systems that decide who gets a loan, a lease or a place in school, for example. Whether we see a job posting could be decided by an algorithm. A benefits claim could be flagged for audit by a data-matching system. A doctor’s appointment may depend on successfully navigating a patient portal. To speak of digital determinants of health, then, is not to add another item to the list of social determinants; it is to look more closely at how those determinants operate in today’s societies. Almost everything that moves people towards health, or away from it, now has a digital dimension. For policymakers and practitioners alike, this means that the terrain they work on has shifted.
Digital infrastructure is structural in the way housing is structural. But unlike housing policy, which has developed over centuries of contestation and regulation, digital systems operate through design choices that nobody voted on and that most affected people have no say over. This creates three concrete problems for health.
How digital systems shape health
The first challenge pertains to extractive data practices. Medical histories, movement patterns and symptom searches become assets owned by platforms or data brokers, monetised through advertising, risk scoring or resale, with little value flowing back to the people who generated them in the first place. Someone who searches their symptoms online may find the same search terms following them across the web for weeks afterward, in the form of ads for supplements, private clinics or unproven remedies. A person managing diabetes may find their step count feeding an insurer’s pricing model. The data keep circulating long after the person who produced them has lost sight of where they went.
The second is engagement-optimised design. Many commercial digital platforms are built to exploit attention and emotion in order to maximise time on site – through infinite scroll, autoplay, algorithmically amplified outrage and notifications timed to interrupt. The measurable costs to mental health are increasingly apparent. In the absence of regulation that treats attention and emotional well-being as things worth protecting – such as by avoiding manipulative or addictive design features – platforms have little reason to design otherwise.
The third is the fragmentation of social spaces. When algorithms sort people into ever narrower groups and reward outrage over discussion and engagement, coordinated public health action becomes harder to achieve. Vaccination, nutrition, climate action – all these require a baseline of shared, collectively agreed facts. Digital infrastructure as currently designed does not merely fail to support that baseline; it actively works against it.
Solidarity as a design principle
Here is where solidarity needs to function as a design principle rather than an aspiration. When it comes to meeting people’s basic needs – health care, stable and affordable housing, a liveable environment – policies and institutions should be built on what people have in common rather than on their differences. Universal healthcare systems, for instance, rest on the recognition that despite substantial differences among people in terms of health risks, income or social position, everyone will need care at some point, and access should therefore be organised based on need rather than ability to pay.
The same logic applies in the domain of data. Data solidarity carries this principle into digital infrastructure: practices that mainly benefit the companies collecting the data, while exposing people to real harm, should be restricted – not left to consent forms that almost nobody reads and even fewer can meaningfully refuse. In contrast, practices that carry little risk for individuals and groups but generate real public benefit, such as pooled health data for research, deserve active support rather than having to compete for funding against extractive business models. And where people take on data-related risk for a collective good – such as by allowing their data to be used for health research – there should be real support in place for them if anything goes wrong, so the costs of solidarity do not fall hardest on those least able to bear them.
Digital literacy campaigns or stricter consent rules are important, but they are not sufficient: none of them changes what platforms are built to do. What is needed instead is to treat data as shared public resources, governed by rules that protect collective health rather than individual profit. This is, in the end, a political choice: whether we build a digital society that runs on extraction and engagement, or build one that runs on the mutual dependence that public health has always relied on.
The Ottawa Charter’s core insight – that health is made in the ordinary conditions of daily life, not primarily in clinics – still holds, 40 years on. The task now is to extend that insight to the digital conditions shaping daily life today, and to choose, deliberately, to build infrastructure that supports the solidarity that healthy societies have always depended on.
