Split opinion
Critics say researchers failed to discern a real finding in a suicide study
The finding of a landmark Finnish study that suicide risk is not reduced by “gender-affirming” hormonal and surgical interventions has come under challenge.
In a response to the Finnish article of Ruuska et al published by the journal BMJ Mental Health, research sociologist Donald Paul Sullins argues that an inappropriately tight test for statistical significance masked the conclusion that medicalised gender transition did indeed appear to result in a lower risk of suicide.
However, Dr Sullins, an assistant professor at the Catholic University of America, says his critique “complicates, but does not negate” the original finding because previous research1 suggests that the Finnish study’s average follow-up period of 6.5 years may have been too short for higher suicide rates to become manifest in the post-transition group of patients. (Follow-up ranged from 2.4–25.7 years.)
The study by Ruuska et al, reported in February, is significant because the gender-affirming treatment model relies heavily on the emotive claim that its interventions reduce an otherwise very high risk of suicide for patients with gender distress.
The researchers used Finland’s comprehensive health registers to track suicide as well as total mortality among 2,083 gender-distressed adolescents with a median age of 19 seen by nationally centralised gender clinics from 1996 to 2019.
These patients were divided into two groups—those who underwent medical transition and those who did not—and their outcomes were compared with 16,643 counterpart controls from the general population matched for age, sex and municipality of birth. The study also isolated statistically the effect of co-morbid psychiatric problems on suicide risk among gender-distressed adolescents.
Ruuska et al reported that the risk of suicide among these patients was low in absolute terms. And although this group had a suicide risk higher than peers in the general population, this was explained by their psychiatric problems, and not by their gender distress.
The researchers also reported no statistically significant difference in suicide risk between matched controls and either group of gender patients (those who transitioned and those who did not), suggesting that hormonal and surgical interventions conferred no greater protection from suicide.
However, Dr Sullins’s critique is that Ruuska et al set too tight a test for statistical significance by using a p-value of 0.01, given the methodology of the study and the size of its data.
“By conventional criteria, [using a p-value of 0.05], the suicide mortality of the [non-treated patients] was significantly higher than controls while that of the [treated group] was not,” Dr Sullins writes.
Psychiatrist Riittakerttu Kaltiala of Tampere University Faculty of Medicine, a member of the research team for the Finnish study, countered that the tighter p-value of 0.01 was appropriate given the large size of the data and the number of statistical tests carried out.
She also highlighted the small number of suicides in the study and data privacy restrictions, which she said ruled out direct comparison between the treated and untreated patient groups.
“If the limit for statistical significance is set too loose, the risk is that chance phenomena will be interpreted as true differences,” Professor Kaltiala told GCN.
She said the Sullins critique did not affect other findings of the study.
“Our main point was that suicide mortality [among gender-distressed adolescents] was very low and explained by psychiatric morbidity. The notion that gender dysphoria per se vastly increases risk of suicide is false,” she said.
A separate critique of the Finnish study, also published by BMJ Mental Health last month, suggests the suicide question could be analysed with more statistical power by amalgamating data sets across countries.
“Given the thankfully small number of suicides within the sample from Finland (7 out of 2,083 identified transgender persons), it may be that a future study should pool multiple samples across comparable settings (eg, Sweden, Norway, Netherlands, Denmark, etc) and perform a meta-analysis,” the critique says.2 The authors of this critique concede such research could be difficult given variation in treatment approaches.
The Finnish team of Ruuska et al said they would send a response to their critics for publication in the journal.
Dr Sullins invoked the 2011 Swedish study of Dhejne et al, covering the period 1973-2003, which found that death by suicide was 19 times higher among medically transitioned patients than among matched controls. However, the higher mortality in this group did not begin to register until 10 years after “sex reassignment”, suggesting the end of a honeymoon period. Dr Sullins also noted the 2023 study by Kaltiala et al, which reported that gender-distressed patients who underwent medical transition did not end up less in need of psychiatric care for conditions such as severe mood disorders and anxiety. This result dovetails with the conclusion of Ruuska et al that psychiatric illness, rather than gender distress itself, is the driver of elevated suicide risk.
The lead author is physician Carl Streed of Boston Medical Centre’s GenderCare Center.
Unfortunately Kaltiala left herself wide open for this criticism. She should not really have done the comparison of treated vs untreated, because the Finnish team were more selective than many other countries in who medically transitioned, and it is quite likely that the most suicidal were not transitioned. This is the likely explanation for the untreated ones showing a trend to a higher rate of suicide. They did try to control for different psychiatric needs within the populations by matching groups based on the number of specialist psych visits. However, this is a blunt instrument, and does not really adequately control for potentially very different reasons for needing psychiatric care. Essentially the problem is that the young people who were attending the gender clinics were not randomised into treatment and non-treatment groups. There was discretion in who was suitable for treatment. I think this is difficult to compensate for afterwards.
Thanks Bernard.
Sullins adds some interesting further observations, which temper his headline finding.
First, he notes that Dhejne et al. (2011), "the only other national registry study of completed suicide", found that suicide mortality among Swedish GR recipients increased dramatically after 10 years, something outside the scope of Ruuska et al., but in keeping with their finding.
Second, He notes that in Finland "only 38% of GD diagnosed persons in treatment proceeded to medical GR. This suggests the presence of assessment, screening and monitoring processes that may inhibit some of the negative consequences of psychiatric morbidity, for example by ensuring better social support or medication compliance, than may be the case in other settings."
Both observations are entirely consistent with a gender-critical perspective.
The pro-transition industry may take Sullins's work as countering Ruuska et al, but that would be a misrepresentation. (Not that that would stop them ...)