|
Citation
|
Madigan S, Korczak DJ, Vaillancourt T, Hopkins WG, Racine N, Neville RD. Lancet Psychiatry 2023; 10(7): 485-486. |
|
Abstract
|
In their Correspondence about our meta-analysis on changes in paediatric emergency department visits during the COVID-19 pandemic, Sarah Steeg and colleagues assert that our evidence for a substantial increase does not equate to good evidence in clinical settings. We disagree. Our predefined threshold for substantial was intended to be clinically meaningful: an extra one in every ten emergency department visits to a health setting for a given pathology (eg, suicide attempt) attributable to COVID-19. Our predefined threshold for good evidence accounts for sampling uncertainty: a probability of 0·75-0·95 for a substantial increase. These thresholds apply across all settings, but Steeg and colleagues assert incorrectly that the thresholds should be linked to the quality of the studies.
Steeg and colleagues also question our use of 90% versus 95% CIs. Coverage of 90% intervals is consistent with the thresholds for the level of evidence and with an alpha of 0·05 for rejection of non-significant and significant hypotheses (inferiority, superiority, and equivalence testing). The 95% interval for attempted suicide would be 1·06-1·41, which also constitutes good evidence of a rate increase and would be statistically significant.
Steeg and colleagues further note that we did not disaggregate our results by health setting. We had already noted the problem of insufficient data to include study moderators in the full analysis, and we had several strategies for dealing with the moderating effects of age, sex, and geographical location. When more data are published, the moderating effect of health setting should be estimated.
They criticise the distinction we made between attempted suicide and self-harm. We agree that terminology in this area is contentious. Their position is that we should consider only a single outcome--self-harm regardless of intent. Yet the evidence suggests otherwise. Among adolescents (ie, those aged 14-18 years in North America and those aged 15-17 years in Europe) who have engaged in self-harm, suicide attempts are associated with greater severity of psychiatric symptoms, more adverse life events, and greater interpersonal conflicts in community-based studies in North America3
and Europe. Our coding of outcomes was also in keeping with another meta-analysis in this area and the findings of an international survey of suicide experts--namely, that intention matters...
Language: en |