Ruuska et al.'s Reply, Reviewed: Bias, Causality & Ethics

I was part of a large group, led by David Banos and Monica Patrascu, and had the opportunity to contribute to a Letter of Concern in Acta Paediatrica (published May 30, 2026, open access) regarding a register-based study by Ruuska et al. on psychiatric service utilization among adolescents and young adults referred to gender identity services in Finland. Two other groups (Edmiston et al. and Sandri et al.) published related concerns in their own letters on the same day. The original authors provided a joint response to all three letters.

Since the response received from Ruuska et al., does not adequately resolves the core issues raised, I wanted to share a few thoughts on it here.

1. The screening claim needs attention and a citation

Ruuska’s et al., defense against selection and collider bias (bias arising when group comparisons are distorted by the very process that sorted people into the groups in the first place) is their claim:

“All adolescents undergo repeated population-wide mental health screening through the national school healthcare system, which reduces baseline differences in undetected psychiatric morbidity between groups. It is therefore implausible that severe psychiatric disorders would systematically remain unrecognised or be detected solely due to assessment within gender identity services.”

"Implausible" is a remarkably strong claim to rest on zero citations. It is relatively easy to establish that a hypothesis might be plausible. However, it is much harder to justify that competing explanations are implausible. That requires evidence, not assertion. At best, Ruuska et al. are presenting an untested assumption as fact; at worst, what we are seeing is rhetorical confidence is used as a substitute for evidence.

Finland's main nationwide project in this space is the School Health Promotion Study, run by the Finnish Institute for Health and Welfare (THL). But by THL's own description, it's a voluntary, anonymous classroom survey reported only at the aggregate, school and municipal level, not an individualized clinical assessment that could flag a specific adolescent's psychiatric morbidity for the medical record. Routine individual school doctor checks, separately, happen only at a few fixed grade levels and, per published Finnish research on the system itself, use no validated screening instruments. Unmet need for mental healthcare in Finland appear to run above the EU average, and it took a 2025 national reform to require faster, structured psychosocial follow-up for under-23-year-olds. None of this is easy to reconcile with a claim that Finland runs a nationwide, comprehensive, individualized psychiatric screen of every adolescent throughout the 1996 to 2019 study period as Ruuska et al., claim.

There's also a simpler explanation sitting right next to the one Ruuska et al. offer. The detection and surveillance bias our original letter already named: adolescents referred to gender identity services experience far more clinical contact points than a comparison group simply by being inside a structured, multi-visit assessment pathway with psychiatric evaluation built into the protocol. More contact with the healthcare system means more chances for any symptom to get noticed and escalated into a specialist referral, regardless of whether the underlying rate of psychiatric morbidity actually differs between groups. Adolescents with fewer touchpoints with the system, including most of the comparison group, have fewer opportunities to be referred onward, whatever their actual mental state. This explains the observed pattern without requiring an undocumented, population-wide screening program to exist at all.

The rebuttal provided by the author’s simply assumes the very premise that requires demonstration.

2. "Doesn't claim definitive causality" sets the bar incredibly low

In their reply to the letters of concern about their article, Ruuska et al. note that their study does not claim "definitive causality." That’s a very low bar to clear and doesn’t address the actual concern: how confidently the authors' own framing, and the media coverage that followed, have treated the findings as evidence that gender-affirming care causes psychiatric harm or, at the very least, more assessment is needed prior to starting hormone treatment for youth. To their credit, the original paper itself does not appear to lean heavily on causal language, and the study's comprehensive cohort is large by any standard, which is a genuine strength of the design. Still, virtually no single study, including "gold standard" well-designed RCTs, establishes definitive causality on its own in the social sciences or for complex medical conditions; that's why replication and meta-analysis carry much a lot of weight in the sciences, and justifiably so. All sorts of error and bias can creep into studies, one way or another. The relevant question is what degree of confidence (high, moderate, low, very low or none) the available evidence supports, not whether causality is definitive. On that front, for example, this Ruuska et al. study relies entirely on domestic service-utilization records, which, by design, cannot even tell us if participants in the non-treatment or comparison group did or did not receive care abroad or through private providers, a distinct possibility given Finland's long public waitlists and close proximity to other EU countries where care is accessible.

3. Ethics approval for a study isn't the same as ethical communication or research practice

Much of Ruuska et al.'s reply emphasizes that their study received appropriate ethics committee and data authority approvals. Appropriate administrative approval was never the concern. Institutional review boards determine whether a study can be conducted; they generally do not evaluate whether the study's conclusions are appropriately framed, whether limitations are communicated with sufficient clarity, or how readily findings may be misinterpreted after publication. Those responsibilities remain with the investigators.

TENK's own Code of Conduct for Research Integrity (the Finnish national guidelines) treats publication and dissemination of results as a distinct area of research integrity in its own right, separate from ethical review or data collection, requiring that results be "communicated in an open and responsible fashion." This is common place. For example, the American Psychological Association's Ethical Principles of Psychologists and Code of Conduct goes even further: under Standard 1.01 ("Misuse of Psychologists' Work"), researchers who become aware that their work is being misused or misrepresented are expected to take reasonable steps to correct or limit that misrepresentation, not simply point back to whatever administrative sign-off they received before even starting the research project.

Institutional review board’s approval establishes that a study was permissible to conduct; it does not confer immunity from criticism regarding how its findings are interpreted, communicated, or presented to the scientific community or the public.