Studies › Sample & attrition · Last reviewed 2026-05-16

Sample size and attrition

Summary

The main Dutch outcome study (de Vries 2014) reports attrition from n=70 at T0 to n=55 at T2 — approximately 22%. The 15 missing participants are partly documented: one died from post-operative complications, others did not start cross-sex hormones, refused follow-up measurement or were no longer reachable. Whether attrition was at-random or differential remains a methodological concern.

1. Attrition flow de Vries 2014

Moment	n	Loss between measurements
T0 — start GnRHa	70	—
T1 — start CSH	~63	Not started or attrition (~7)
T2 — 1 yr post-op	55	Attrition, death, no measurement data (~8)

2. Nature of attrition

One participant died from necrotising fasciitis after vaginoplasty.
Other participants were no longer reachable or refused follow-up measurement.
The authors acknowledge that attrition is possibly associated with poorer psychological functioning.

3. Effect on outcome measures

If attrition is differential — i.e. poorer-functioning patients are measured less often — aggregated T2 figures may overestimate the actual average outcome. Biggs (2023) has worked this point out explicitly and concludes that a sensitivity analysis is missing. See /evaluations/biggs-puberty-blocker-overview/.¹

4. Broader context

In follow-up studies in comparable populations (Costa 2015, Carmichael 2021 — see replication attempts), comparable or higher attrition rates have been reported. The SBU report (2022) notes that no formal intention-to-treat analysis was carried out in any of the available studies.²