When a longitudinal study is carried out, there is significance in the numbers of participants who were in the original study but who do not participate in the follow-up study. These missing participants are labelled ‘lost to follow up’. They are significant because their outcomes are unknown. In the field of gender medicine it is well known that patients dissatisfied with their treatment will drop out of studies, terminating their relationship with the medics involved in their initial treatment.
This collation of loss to follow-up data was made by MIchelle and was originally published on Twitter:
I’ve wanted to do this for a while. Here’s a thread examining loss to follow-up in 27 studies that were used for the 2021 meta-analysis on post-surgical regret in transition-related medicine.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8099405/#!po=31.9672
What’s the significance of loss to follow-up?
After a certain amount of patients have been lost, the results start to lose their validity.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3427970/
“Some have suggested that <5% loss leads to little bias, while >20% poses serious threats to validity. This may be a good rule of thumb, but keep in mind that even small proportions of patients lost to follow-up can cause significant bias.”
“One way to determine if loss to follow-up can seriously affect results is to assume a worst-case scenario with the missing data and look to see if the results would change.”
In this case, assume the worst-case scenario is regret. Here we go!
1. Blanchard et al, 1989
doi.org/10.1177/070674…
134 patients underwent surgery (2 dead)
111 participants, 4 expressed regrets
Regret rate of sample: 3.6%
Loss to follow-up: 15.9%
https://doi.org/10.1177/07067437890340011
2. Bouman, 1988
doi.org/10.1097/000006…
67 patients underwent surgery
55 participants, 1 expressed regrets
Regret rate of sample: 1.8%
Loss to follow-up: 17.9%
3. Cohen-Kettenis et al, 1997
doi.org/10.1097/000045…
22 patients underwent surgery
19 participants, 0 expressed regrets
2 refused, 1 non-responsive
Regret rate of sample: 0%
Loss to follow-up: 13.6%
https://doi.org/10.1097/00004583-199702000-00017
4. De Cuypere et al, 2006
doi.org/10.1016/j.sexo…
107 patients underwent surgery
62 participants, 2 expressed regrets
15 refused, 30 non-responsive
Regret rate of sample: 3.2%
Loss to follow-up: 42.1%
https://doi.org/10.1016/j.sexol.2006.04.002
5. Garcia et al, 2014
doi.org/10.3978/j.issn…
25 participants chosen by random selection of former patients, 0 expressed regrets
Limitation: only patients “able and inclined to present to our clinic were evaluated.”
Overall satisfaction, sexual function, and the durability of neophallus dimensions following staged female to male genital gender confirming surgery: the Institute of Urology, London U.K. experienceOverall satisfaction, sexual function, and the durability of neophallus dimensions following staged female to male genital gender confirming surgery: the Institute of Urology, London U.K. experiencehttps://doi.org/10.3978/j.issn.2223-4683.2014.04.10
6. Imbimbo et al, 2009
doi.org/10.1111/j.1743…
163 patients underwent surgery
139 participants, 8 expressed regrets
24 refused
Regret rate of sample: 5.8%
Loss to follow-up: 14.7%
https://doi.org/10.1111/j.1743-6109.2009.01379.x
7. Jiang et al, 2018
doi.org/10.1016/j.jsxm…
16 patients underwent surgery
14 participants, 1 expressed regrets
2 non-responsive
Regret rate of sample: 7.1%
Loss to follow-up: 12.5%
https://doi.org/10.1016/j.jsxm.2018.03.085
8. Johansson et al, 2010
doi.org/10.1007/s10508…
49 patients underwent surgery
33 participants (32 SRS, 1 mastectomy), 0 expressed regrets
1 dead (from surgery complications), 15 non-responsive/refused
Regret rate of sample: 0%
Loss to follow-up: 30.6%
9. Krege et al, 2001
doi.org/10.1046/j.1464…
66 patients underwent surgery
31 participants, 0 expressed regrets
Regret rate of sample: 0%
Loss to follow-up: 47%
https://doi.org/10.1046/j.1464-410X.2001.02323.x
10. Kuiper et al, 1998
researchgate.net/profile/Peggy-…
Based on what I’ve seen, this study is either the wrong citation or there have been some major errors with interpreting it. (Next few tweets are about this.)
– The meta-analysis claims “Kuiper et al followed 1100 transgender subjects that underwent GAS.” However, the citation appears to be a qualitative study that recruited ten participants with regret (rather than following up with a sample size of 1100).
– Kuiper et al, 1998, shows the mean age of the 10 participants as 46.4 — which is also what is recorded as the mean age in the meta-analysis chart (for a sample size of 1100). This appears to be in error.
– “Ten experienced regret (9 transmasculine and 1 transfemenine).” This is backwards based on the citation. It’s 9 transfeminine (i.e., “MTF”) and 1 transmasculine (i.e., “FTM”).
– “The overall prevalence of regret after GAS in this study was of 0.9%.” The idea that 10 of 1100 regretted surgery is being used as a major point in this meta-analysis, and it is reflected nowhere in the citation.
If I’m correct, this is a massive flaw in this paper.
11. Lawrence, 2003
doi.org/10.1023/A:1024…
727 patients underwent surgery
307 with unknown/invalid address, 3 dead, 62 refused, 120 non-responsive, 3 ineligible
232 participants, 15 expressed regrets
Regret rate of sample: 6.5%
Loss to follow-up: 67.7%
12. Lobato et al, 2006
doi.org/10.1007/s10508…
30 patients underwent surgery
7 lost to follow-up, 4 ineligible
19 participants, 0 expressed regrets
Regret rate of sample: 0%
Loss to follow-up: 23.3%
13. Nelson et al, 2009
doi.org/10.1016/j.bjps…
17 patients underwent surgery
12 participants, 0 expressed regrets
3 unknown addresses
Regret rate of sample: 0%
Loss to follow-up: 29.4%
https://doi.org/10.1016/j.bjps.2007.10.049
14. Olson-Kennedy et al, 2018
doi.org/10.1001/jamape…
93 patients underwent surgery
68 participants, 1 expressed regrets
2 refused, 24 non-responsive
Regret rate of sample: 1.5%
Loss to follow-up: 28%
https://doi.org/10.1001/jamapediatrics.2017.5440
15. Papadopulos et al, 2017
doi.org/10.1016/j.jsxm…
121 patients underwent surgery
47 participants, 0 expressed regrets
38 unknown addresses, 14 refused, 22 non-responsive
Regret rate of sample: 0%
Loss to follow-up: 61.2%
https://doi.org/10.1016/j.jsxm.2017.01.022
16. Pfäfflin, 1993
doi.org/10.1300/J056v0…
297 patients underwent surgery, 3 expressed regrets (1.01%)
* I think this is retrospective. It’s not clear whether the author actually followed up with his patients for this study to ask each of them about regrets.
17. Rehman et al, 1999
doi.org/10.1023/a:1018…
47 (eligible) patients underwent surgery
28 participants, 0 expressed regrets
16 non-responsive, 3 dead
Regret rate: 0%
Loss to follow-up: 36.4%
18. Smith et al, 2001
doi.org/10.1097/000045…
20 patients underwent surgery
20 participants, 0 expressed regrets
Regret rate: 0%
Loss to follow-up: 0%
https://doi.org/10.1097/00004583-200104000-00017
19. Song et al, 2011
doi.org/10.1055/s-0030…
19 patients underwent surgery
8 participants, 0 expressed regrets
* The meta-analysis counts all 19 as the sample. However, only 8 were asked about regrets (5-year follow-up).
Regret rate: 0%
Loss to follow-up: 42%
https://doi.org/10.1055/s-0030-1268210
20. Van de Grift et al, 2018
doi.org/10.1080/009262…
546 eligible patients, 201 responded
136 postsurgical participants
Regret rate of sample: 5.9%
Loss to follow-up: 63.1%*
* Unsure, not all patients had surgery. Also: meta-analysis says only 2 had regrets; study says 8.
Surgical Satisfaction, Quality of Life, and Their Association After Gender-Affirming Surgery: A Follow-up StudyWe assessed the outcomes of gender-affirming surgery (GAS, or sex-reassignment surgery) 4 to 6 years after first clinical contact, and the associations between postoperative (dis)satisfaction and q…https://doi.org/10.1080/0092623X.2017.1326190
21. Wiepjes et al, 2018
doi.org/10.1016/j.jsxm…
4863 patients underwent surgery, 14 expressed regrets (0.3%)
* This is a retrospective study (i.e., authors looked at medical records and did not follow up with patients directly).
https://doi.org/10.1016/j.jsxm.2018.01.016
22. Zavlin et al, 2018
doi.org/10.1007/s00266…
49 patients underwent surgery
40 participants, 1 expressed regrets
Regret rate of sample: 2.5%
Loss to follow-up: 16.3%
23. Judge et al, 2014
doi.org/10.3389/fendo.…
218 eligible patients, 55 underwent surgery, 3 regret SRS, 1 HRT
* This is a retrospective study (i.e., looked at medical records and did not follow up with patients directly).
Regret rate of sample: 1.8% overall, 5.5% postsurgery
24. Vujovic et al
doi.org/10.1111/j.1743…
118 eligible patients, 0 expressed regrets
* This is a retrospective study (i.e., authors looked at medical records and did not follow up with patients directly).
https://doi.org/10.1111/j.1743-6109.2008.00799.x
25. Weyers et al, 2009
doi.org/10.1111/j.1743…
70 patients underwent surgery
50 participants, 2 expressed regrets
17 non-responsive, 3 refused
Regret rate of sample: 4%
Loss to follow-up: 28.6%
https://doi.org/10.1111/j.1743-6109.2008.01082.x
26. Poudrier et al, 2019
doi.org/10.1097/PRS.00…
81 patients underwent surgery
58 participants, 2 expressed regrets
Regret rate of sample: 3.4%
Loss to follow-up: 28.4%
27. Landén et al, 2007
doi.org/10.1111/j.1600…
213 patients, 3.8% expressed regrets
* This is a retrospective study. Authors looked at applications for sex reassignment in Sweden and noted how many applied for a reversal. That’s what they counted as “regret.”
https://doi.org/10.1111/j.1600-0447.1998.tb10001.x
Thread done.
This is only highlight issues with loss to follow-up in “regret” studies. Other issues include what kind of safeguards were in place before transitioning (i.e., patients used to be turned away and now rarely are) and age at time of transition, etc.