Sergio Flores is PhD candidate in health economics at the Department of Public Health and Caring Sciences, Uppsala University. His doctoral research evaluates Sweden’s Extended Home Visiting Programme, a targeted extension of Sweden’s universal child health care clinic services, and will defend his dissertation in May 2026.
Sergio also recently developed the Child Health Equity Observatory, an interactive tool that pools open access data to create indicators of child health equity in Sweden. CHESS recently spoke with Sergio about his research and his work in developing this tool.
Q: Your dissertation evaluates the Extended Home Visiting Programme from several different angles, including the extent to which it reaches the intended families, has measurable health effects, and is cost-effective. What are some of the key findings of your research?
The Extended Home Visiting Programme is Sweden’s attempt to operationalise proportionate universalism within the universal child health system. It offers six structured home visits during a child’s first fifteen months, delivered jointly by a child health nurse and a municipal parental advisor, to first-time parents in socioeconomically disadvantaged areas. Between 2018 and 2020 the government allocated 354.5 million SEK to support regional implementation. My dissertation evaluates the programme along four dimensions: targeting accuracy, implementation, effects on healthcare utilisation, and costs of national scale-up.
On targeting, the programme did direct resources to higher-need areas. Across 3,398 small statistical areas (DeSOs), we found a positive concentration index of 0.074, meaning the programme was distributed pro-equitably rather than regressively. But the more striking finding is what was not reached: 73 percent of DeSO-years had no programme exposure at all, and fewer than half of the highest-need areas were reached. The programme was aimed in the right direction, but the dose was thin and the coverage incomplete.
On implementation, our qualitative interviews with 12 of 21 regional central child health teams revealed that implementers were operating under structural conditions that undermined the programme’s relational logic. Annual project funding cycles, unclear cross-sectoral mandates, and coordination barriers between healthcare and social services made it difficult to build the sustained, trust-based relationships the programme depends on.
On effectiveness, using a stacked difference-in-differences design with propensity score matching on 49,292 firstborn children across 158 implementing and 599 comparison areas, we detected no statistically significant effects on emergency visits, avoidable hospitalisations, or other healthcare utilisation outcomes during the first year of life. The confidence intervals were narrow enough to rule out anything beyond very small absolute effects. It is important to interpret this carefully: the analysis estimates the average effect of programme availability in an area, not the effect of programme receipt for an individual family.
On costs, scenario-based micro-costing produced estimates ranging from EUR 556 to EUR 1,180 per family, with national scale-up costed at EUR 5.6 to 11.9 million annually. Financially, the programme is feasible to scale.
The overall picture is that the programme reached the right places but not enough of them, was implemented under governance conditions that undermined its relational logic, and produced no detectable effects on healthcare utilisation during the first year of life under those conditions. Whether the transition to permanent funding from 2025 will resolve the cross-sectoral coordination challenges that constrained the programme in practice remains an open question.
Q. Sweden is often held up as a model for child health policy. What does your research reveal about the gaps that still exist within that system?
Sweden’s child health system is genuinely excellent. Participation in the universal programme exceeds 99 percent, and the network of barnavårdscentraler (child health care clinics) across the country is remarkable by any international comparison. But universal access is not the same as equitable outcomes, and that distinction matters.
When we examine the data across all 290 Swedish municipalities, the variation is striking. MMR vaccination at age two ranges from 87 to 100 percent, with 183 municipalities below the WHO 95 percent threshold. Child poverty ranges from under 1 percent to over 14 percent. Preschool enrolment varies from 69 percent to near-universal coverage. These are not small differences, and they tend to compound because the municipalities with the highest need also tend to have the most strained services.
What emerges from the data is a pattern consistent with Marmot’s framework of proportionate universalism. Universal systems require that services be delivered with intensity proportionate to disadvantage. In Swedish child health, the structural commitment to universalism is strong, but the proportionate part is harder to detect. Municipalities with rapidly growing immigrant populations, rising child poverty, or constrained municipal finances often face increased demand without matching increases in capacity.
The other gap is informational. The data needed to identify these patterns sits across five separate government agencies. Until recently, no integrated picture existed for the regional planners and BHV coordinators who make resource decisions. They were often working with fragments of a much larger landscape.
Q. Alongside your doctoral research, you recently developed the Child Health Equity Observatory. What was your motivation or inspiration for developing this tool, and what do you aim to achieve?
The motivation came from a frustration I encountered in my doctoral work. I was evaluating a targeted intervention within the child health system, but I had no systematic way to understand the broader landscape: which municipalities had the greatest need, how they compared to each other, and whether resources were being directed proportionately.
The data existed, scattered across Kolada, Statistics Sweden, the Public Health Agency, the National Board of Health and Welfare, and the National Agency for Education. But no one had assembled it. A municipal BHV (child health care) coordinator who wanted to understand how vaccination, child poverty, preschool enrolment, and staffing interact in her community had to do that assembly manually each time. Most do not have the time or the technical resources.
So I built the assembly once, for everyone. The Child Health Equity Observatory integrates more than 40 indicators across all 290 municipalities, with interactive maps, peer benchmarking, trend analysis, population forecasting, and municipality-specific analytical briefs. It also includes a national spatial accessibility analysis covering all roughly 1,100 child health centres.
What I aim to achieve is a shift in how decisions are made in public services, with child health equity as the entry point. Today, regional planners and local coordinators often make resource allocation decisions with fragments of the available picture, not because the information does not exist, but because it sits in five separate government databases that no one has time to integrate. The observatory was an attempt to close that gap for one domain. The deeper ambition is broader.
Sweden has, in many ways, ideal conditions for evidence-driven public services. We have administrative registers of remarkable quality, a tradition of open government data, strong institutional trust, and a welfare state with the legitimacy to act on what the data reveals. What we have not yet developed is the analytical infrastructure that turns this raw material into continuous, decision-relevant intelligence. We still rely heavily on annual reports, ad hoc analyses, and the personal expertise of senior staff, all of which are valuable but none of which scale.
I see this as a moment of genuine opportunity. Modern data analytics, geographic information systems, forecasting methods, and now AI-assisted synthesis can make rigorous, transparent decision support routinely available. The observatory shows that this is not theoretical. It exists and it works. If a single PhD candidate can build national-level monitoring infrastructure on his free time, the case for embedding this kind of capability in our public institutions is hard to dismiss.
Two principles guide the approach I want to advance. The first is that decision support must be honest about its limits. Composite indicators, forecasts, and AI-generated analyses are useful only if they communicate uncertainty clearly and acknowledge what they cannot show. Tools that hide their assumptions damage trust, and trust is the foundation a welfare state cannot afford to lose. The second is that AI and quantitative analysis must complement, not replace, professional judgement. The BHV nurse who knows her families, the regional coordinator who has watched a municipality change over a decade, the politician who has to weigh competing legitimate priorities; these are irreplaceable forms of expertise. The role of analytical infrastructure is to give them better material to work with, not to pretend that algorithms can answer questions that are ultimately ethical and political.
What I want to demonstrate, through the observatory and the work that follows it, is that a strong democratic welfare state is exactly the right setting for thoughtful modernisation of how public services are planned and evaluated. The institutions are in place. The data is in place. The political tradition of equity-driven policy is in placeWhat is missing is the analytical infrastructure that turns Sweden’s data abundance into the everyday intelligence that planning decisions deserve. Building that infrastructure, openly, transparently, and in a way that makes public institutions more accountable rather than more opaque, is what I want to keep contributing to. If we get this right, the next generation of welfare-state policy can be planned with the rigour that the welfare state itself demands.
Q. What do you think are the current limitations of publicly available data for measuring child health and health equity in Sweden?
First, geographic resolution. Most publicly available data is at municipal level, which masks substantial within-municipality variation. Stockholm is one municipality, but the difference between Rinkeby and Östermalm is far greater than the difference between most municipalities. To see the patterns inside large municipalities, one needs individual-level registry data, which requires ethics approval and is not publicly accessible.
Second, timeliness. Most indicators have a one-to-two-year reporting lag. We are always looking at the recent past, never the present.
Third, mismatched levels of organisation. Healthcare in Sweden is organised by region, while social services are organised by municipality. So spending data on child health services is reported per region, meaning all municipalities within a region share identical values. This provides zero discrimination at the level where many planning decisions are made.
Fourth, important variables are simply not available as open data. There is no publicly available indicator for BHV nurse staffing at the municipal level. We know how much each region spends per child, but not how many nurses are actually working in each community.
Fifth, some indicators contain a detection paradox. Child mental health diagnostic rates are higher in municipalities with stronger diagnostic capacity, not necessarily where mental illness is more prevalent. A naive monitoring tool would penalise municipalities that are doing a better job of identifying children in need. Indicators have to be selected carefully, and limitations communicated honestly.
These constraints do not invalidate open-data monitoring, but they mean any serious tool built on this data must be transparent about what it can and cannot show, and must complement aggregate monitoring with deeper analysis where the data allows.
Q. Like many individuals working within the public health discipline, you have a multidisciplinary background. As someone with training in both medicine and health economics, why is it important to address health inequalities — or more specifically, health inequalities among children?
There are two answers to this question, and I think both matter.
The first is the ethical argument. Children do not choose the circumstances they are born into. A child in a municipality with high poverty, low parental education, and below-target vaccination coverage has done nothing to deserve a worse start in life than a child born twenty kilometres away in an affluent suburb. If we accept that healthcare systems should be equitable, then children, who have the least agency over their own health determinants, have the strongest claim to proportionate attention.
The second is the economic argument, and this is where my health economics training shapes how I think. Inequalities in child health are not only morally troubling; they are economically inefficient. The evidence on early life is unambiguous. Health, development, and learning in the first years shape educational attainment, labour market participation, and lifetime healthcare costs. A child who misses a vaccination, has an undetected developmental delay, or grows up in conditions that compromise their health trajectory generates costs that compound over decades. Investing in equitable child health is among the highest-return investments a society can make.
The two arguments converge. The ethical case says we must act. The economic case says we can afford to, and indeed cannot afford not to. The challenge is translating that convergence into concrete tools and evidence that help the people who allocate resources to do so more equitably. That is what I am trying to contribute, both through my doctoral work on the Extended Home Visiting Programme and through the observatory.