SOCE & Suicidality – Key Points from Sullins (2023) Rejoinder

Context & Purpose

  • Article is a rejoinder to critiques of Sullins (2022b) on sexual-orientation-change efforts (SOCE) and suicidality.
  • Goal: address methodological objections, update analyses, reaffirm original conclusion that SOCE does not raise suicide risk.

Core Statistical Principle

  • Causal inference requires temporal order: cause must precede effect.
  • Pre-SOCE suicidality cannot be attributed to post-SOCE exposure.

Response to Blosnich et al.

  • Dataset contains age at last SOCE ⇒ timing is available.
  • Extended “pre-SOCE” window from 1 yr to 4 yrs (and 6 yrs) before last SOCE.
    • Adjusted-odds-ratios (AOR) for suicide ideation, planning, attempts ≈ (0.7\text{–}1.6); none significant.
  • Even stricter classifications leave original finding unchanged: SOCE does not increase suicidality.
  • Relative-risk of progressing from ideation/planning to attempt lower with SOCE, esp. in adults.

Response to Rivera & Beach

  • Critique of regression methods and at-risk period bias is overstated.
    • Data show suicide ideation clustered <!18 yrs; age-adjustment already minimizes bias.
  • Propensity-score matching (counterfactual approach) performed as additional check:
    • Matched 82 SOCE cases to \le 6 controls each (caliper 0.2 SD).
    • Post-SOCE suicide ideation AOR \approx0.90 (protective); planning \approx0.88; intention \approx0.91; attempts \approx0.98.
    • Results mirror regression models → method choice does not alter conclusion.

Response to Glassgold & Haldeman

  • Claim of “extensive” evidence for SOCE harm/inefficacy is false.
    • APA 2009: no rigorous studies proving harm.
    • Post-2009 population studies few and methodologically weak; key ones mis-time events.
  • Their assertion of “strong experimental designs” contradicts their own earlier reviews acknowledging none exist.

Response to Strizzi & Di Nucci

  • Ethical stance that positive SOCE findings should be censored is rejected.
  • Patient autonomy & religious freedom: individuals may legitimately seek change or coping support.
  • Banning research or therapy based on political goals threatens scientific integrity and human rights.

Updated Empirical Findings (selected)

  • Treatment-Initiation Model (4-yr window):
    • Suicide ideation AOR 1.01\,(0.52,2.00).
    • Suicide planning AOR 1.13\,(0.64,2.00).
    • Suicide attempt AOR 1.25\,(0.67,2.36).
  • Improved Model adjusting childhood covariates: all AORs <1.2, non-significant.
  • Progression from ideation/planning to attempt with intervening SOCE: AOR 0.20\text{–}0.13 (significant), indicating lower escalation risk.

Methodological Takeaways

  • With adequate events-per-variable (>8), standard logistic regression and propensity methods yield concordant results (~90 % agreement in literature).
  • Propensity methods not inherently superior; choice depends on data characteristics.

Ethical & Scientific Implications

  • Misclassifying temporal order inflates perceived harms and informs policy incorrectly.
  • Evidence does not support broad claims that SOCE is universally harmful or ineffective.
  • Scientific bodies should reassess policy statements premised on studies that ignore causal timing.

Bottom Line

  • After incorporating critics’ suggestions and alternative analytic techniques, evidence still shows no significant increase in suicidality following SOCE; some analyses suggest a possible protective effect against suicide attempts.
  • Calls for bans or censorship lack empirical foundation and conflict with principles of causality, patient autonomy, and research integrity.