發展、補償與臨床可見性:為什麼 ADHD 可以到成年才被發現?
英文題名: Development, Compensation, and Clinical Visibility: Why ADHD May Remain Undetected Until Adulthood系列: ADHD 動態配置與認知拓撲系列,第 7 篇版本: v0.1日期: 2026-08-17作者: Neo.K(許筌崴)協作: GPT-5.6 Sol文件性質: 理論建模/發展認知科學命題/研究綱領文獻檢索截點: 2026-08-17
0. 醫學、診斷與證據邊界聲明
本文不是臨床研究、診斷工具、治療指南、精神醫學評估程序或醫療建議。
本文提出的「臨床可見性」「補償儲備」「支架容量」「需求—能力差」「隱性成本」「診斷危險率」等概念均屬待驗證理論命題,不代表已被醫學界確認的 ADHD 病理機制或正式診斷模型。
原作者並非醫學、精神醫學、神經科學或臨床心理專業研究者。本文不提供新的臨床、人體、神經影像、流行病學或心理實驗數據;所有實證性背景均來自公開同行評審研究。
本文不使用原作者個人診斷史作為一般化證據,也不應用於自行診斷 ADHD、推翻既有診斷、認定他人具有 ADHD,或替代正式臨床評估。
本文特別區分:
adult-diagnosed ADHD ≠ adult-onset ADHD . \boxed{
\text{adult-diagnosed ADHD}
\neq
\text{adult-onset ADHD}.
} adult-diagnosed ADHD = adult-onset ADHD .
「成年後才被診斷」描述的是診斷時點;
「成年才真正起病」描述的是發展病程假說。
兩者不可互換。
截至本文文獻檢索日,ADHD 的主流臨床框架仍將其視為神經發展障礙,成人起病或晚起病是否構成獨立 ADHD 病程仍有爭議。
本文亦謹慎使用「masking/camouflaging」。2026 年已有專門評論指出,從 autism 文獻借用 camouflaging 構念至 ADHD 前,需要 ADHD-specific 理論、操作定義與量表驗證。因此本文優先使用:
compensation , \text{compensation}, compensation ,
external scaffolding , \text{external scaffolding}, external scaffolding ,
visibility reduction , \text{visibility reduction}, visibility reduction ,
並只將 masking 視為候選子機制。
核心限制:
late recognition ≠ late biological onset . \boxed{
\text{late recognition}
\neq
\text{late biological onset}.
} late recognition = late biological onset .
摘要
ADHD 在兒童期通常透過學校、家庭與外顯行為進入臨床視野,但大量成人直到大學、職場、親職或其他高自主性生命階段才接受評估。這產生一個重要問題:為什麼某些 ADHD-related configurations 可以多年未被辨識,而在成年後突然變得明顯?
本文提出「Development–Compensation–Visibility Hypothesis, DCVH」。其核心不是宣稱所有成人晚診斷 ADHD 都因高智力、masking 或良好家庭環境而被「遮住」,而是建立一個可檢驗的中層生命史模型。
令時間 t t t 的功能需求為:
D t , D_t, D t ,
個體在該任務域可直接動用的內在能力為:
C t , C_t, C t ,
外部支架為:
S t , S_t, S t ,
後天補償策略為:
K t . K_t. K t .
則定義需求—支持差:
L t = D t − ( C t + S t + K t ) . L_t
=
D_t
-
\left(
C_t+S_t+K_t
\right). L t = D t − ( C t + S t + K t ) .
當:
L t ≤ 0 , L_t
\leq
0, L t ≤ 0 ,
個體可能維持足夠表面功能;
當:
L t > θ I , L_t
>
\theta_I, L t > θ I ,
功能損害開始顯著;
但即使存在功能損害,仍須進一步跨過:
θ V \theta_V θ V
這個可見性門檻,才容易被家人、教師、雇主或醫療系統辨認。
因此:
impairment ≠ visibility ≠ diagnosis . \boxed{
\text{impairment}
\neq
\text{visibility}
\neq
\text{diagnosis}.
} impairment = visibility = diagnosis .
2026 年北京大學第六醫院研究比較 35 名 adult-diagnosed ADHD 與 37 名 childhood-diagnosed ADHD 成人,發現 adult-diagnosed 組兒童期回溯症狀較低,但成年期症狀嚴重度相近,且有較高 internalizing burden 與較佳 self-monitoring;然而 executive dysfunction 是兩組共同、最穩定的症狀相關因子,環境因素並未形成明顯的 adult-diagnosed-specific 機制。2026 年 late-onset trajectory scoping review 則指出 population cohorts 中確實存在 late-emerging symptom trajectories,且較高智力與較多兒童期資源曾被提出支持 compensation/scaffolding hypothesis,但既有 longitudinal literature 在睡眠、治療史、contextual supports 與 neurological antecedents 等關鍵變項上仍有大量缺口。
2025–2026 年女性與高功能成人研究進一步顯示,診斷時點受到 sex/gender presentation、internalizing symptoms、既有成功紀錄、醫療與教育系統辨識方式、主觀痛苦與可觀察產出之間落差等因素影響。然而 masking/camouflaging 在 ADHD 中的 construct validity 尚未成熟,因此本文不將其視為既定核心機制。
本文提出:成人晚診斷更適合被研究為一個多門檻、生命史、觀察者與制度共同參與的 detection process,而非單純詢問「ADHD 是何時開始的?」。
最終候選鏈為:
C t + S t + K t − D t → functional strain → observable phenotype → clinical detection . \boxed{
\mathbf C_t
+
\mathbf S_t
+
\mathbf K_t
-
\mathbf D_t
\rightarrow
\text{functional strain}
\rightarrow
\text{observable phenotype}
\rightarrow
\text{clinical detection}.
} C t + S t + K t − D t → functional strain → observable phenotype → clinical detection .
本文提出十二項可證偽命題與 longitudinal、multi-informant、digital phenotyping、support-withdrawal、life-transition 及 health-system detection studies,並明確要求:若 demand/support/compensation variables 無法在既有症狀與 executive-function measures 之外提供增量預測力,DCVH 應被簡化或放棄。
關鍵詞: ADHD、成人診斷、late diagnosis、adult-onset、compensation、scaffolding、masking、clinical visibility、development、executive function、women、diagnostic delay
1. 三個完全不同的「晚」
討論成人 ADHD 時,至少有三種「晚」。
1.1 Late Diagnosis
T diagnosis ≫ T childhood . T_{\text{diagnosis}}
\gg
T_{\text{childhood}}. T diagnosis ≫ T childhood .
表示正式診斷發生在成年。
1.2 Late Recognition
症狀或功能困難可能較早存在,但:
T recognition T_{\text{recognition}} T recognition
很晚。
1.3 Late Onset
指:
T symptom onset > T childhood . T_{\text{symptom onset}}
>
T_{\text{childhood}}. T symptom onset > T childhood .
這是一個更強的病程命題。
因此:
T diagnosis ≠ T recognition ≠ T onset . \boxed{
T_{\text{diagnosis}}
\neq
T_{\text{recognition}}
\neq
T_{\text{onset}}.
} T diagnosis = T recognition = T onset .
把三者混在一起,會讓成人 ADHD 研究失去可解釋性。
2. 現行成人 ADHD 框架仍是神經發展模型
2025 年 World Psychiatry 成人 ADHD 綜述將成人 ADHD 的 evidence base、uncertainties 與 controversies 系統整理,並明確指出真正 adult-onset ADHD 仍是爭議領域。
因此本文不預設:
adult symptoms without clear childhood history = adult-onset ADHD . \boxed{
\text{adult symptoms without clear childhood history}
=
\text{adult-onset ADHD}.
} adult symptoms without clear childhood history = adult-onset ADHD .
可能解釋至少包括:
childhood symptoms were present but underdetected;
childhood impairment was buffered;
retrospective recall is incomplete;
informant mismatch;
psychiatric comorbidity;
sleep or substance effects;
other medical/neurological causes;
genuine late-emerging trajectory。
所以成人臨床評估必須保留 differential diagnosis,而本文不提供任何替代診斷規則。
3. 為什麼「小時候沒被發現」不能等同「小時候沒有」?
臨床系統通常無法直接觀察:
C t = underlying cognitive configuration . \mathbf C_t
=
\text{underlying cognitive configuration}. C t = underlying cognitive configuration .
它看到的是:
Y t = observable behavior/impairment . Y_t
=
\text{observable behavior/impairment}. Y t = observable behavior / impairment .
所以:
Y t = F ( C t , D t , S t , K t , E t ) . Y_t
=
F
\left(
\mathbf C_t,
\mathbf D_t,
\mathbf S_t,
\mathbf K_t,
\mathbf E_t
\right). Y t = F ( C t , D t , S t , K t , E t ) .
如果:
Y t < θ V , Y_t
<
\theta_V, Y t < θ V ,
系統就可能沒有辨識。
但:
Y t < θ V Y_t<\theta_V Y t < θ V
不能推出:
C t = non-ADHD . \mathbf C_t
=
\text{non-ADHD}. C t = non-ADHD .
這只表示:
observed phenotype stayed below detection threshold . \boxed{
\text{observed phenotype stayed below detection threshold}.
} observed phenotype stayed below detection threshold .
4. 需求—支持差模型
本文定義:
D t = functional demand . D_t
=
\text{functional demand}. D t = functional demand .
例如:
自主管理;
多科目作業;
長期規劃;
行政流程;
職場多任務;
親職;
財務管理;
社交承諾。
內在可用能力:
C t . C_t. C t .
外部支架:
S t . S_t. S t .
後天補償:
K t . K_t. K t .
定義:
L t = D t − ( C t + S t + K t ) . \boxed{
L_t
=
D_t
-
\left(
C_t+S_t+K_t
\right).
} L t = D t − ( C t + S t + K t ) .
5. 功能安全區
若:
L t ≤ 0 , L_t\leq0, L t ≤ 0 ,
代表當下總能力與支持可覆蓋需求。
這不表示:
effort = 0. \text{effort}=0. effort = 0.
個體可能付出極高努力才維持結果。
因此必須另外定義成本。
6. 功能損害門檻
令:
θ I \theta_I θ I
為 clinically meaningful impairment 的候選門檻。
若:
L t > θ I , L_t>\theta_I, L t > θ I ,
功能困難增加。
但:
impairment ≠ diagnosis . \boxed{
\text{impairment}
\neq
\text{diagnosis}.
} impairment = diagnosis .
因為系統仍可能沒有看見、沒有歸因、沒有轉介。
7. 臨床可見性
定義:
V t = visibility . V_t
=
\text{visibility}. V t = visibility .
候選:
V t = f ( E t external , I t , B t , O t , R t observer ) , V_t
=
f
\left(
E_t^{\text{external}},
I_t,
B_t,
O_t,
R_t^{\text{observer}}
\right), V t = f ( E t external , I t , B t , O t , R t observer ) ,
其中:
E t external E_t^{\text{external}} E t external :外顯程度;
I t I_t I t :功能損害;
B t B_t B t :行為偏離規範程度;
O t O_t O t :他人觀察機會;
R t observer R_t^{\text{observer}} R t observer :觀察者對 ADHD 的識別能力。
因此:
I t ↑ I_t\uparrow I t ↑
不必然:
V t ↑ V_t\uparrow V t ↑
同樣幅度。
8. Detection Threshold
若:
V t > θ V , V_t>\theta_V, V t > θ V ,
才比較可能進入:
concern → referral → assessment . \text{concern}
\rightarrow
\text{referral}
\rightarrow
\text{assessment}. concern → referral → assessment .
因此:
θ I ≠ θ V . \boxed{
\theta_I
\neq
\theta_V.
} θ I = θ V .
一個人可以已經很痛苦,但可見性仍不足。
9. 第三個門檻:Access
即使被辨識,還需要:
A t care A_t^{\text{care}} A t care
表示醫療與診斷可近性。
所以:
P ( diagnosis at t ) = H ( V t , A t care , K t knowledge , S t stigma ) . P
\left(
\text{diagnosis at }t
\right)
=
H
\left(
V_t,
A_t^{\text{care}},
K_t^{\text{knowledge}},
S_t^{\text{stigma}}
\right). P ( diagnosis at t ) = H ( V t , A t care , K t knowledge , S t stigma ) .
其中:
K t knowledge K_t^{\text{knowledge}} K t knowledge :本人與周圍人對 ADHD 的知識;
S t stigma S_t^{\text{stigma}} S t stigma :污名與阻力。
因此:
clinical need ≠ clinical access . \boxed{
\text{clinical need}
\neq
\text{clinical access}.
} clinical need = clinical access .
10. 外部支架
兒童與青少年生活常具有大量外部結構:
家長叫起床;
固定課表;
老師提醒;
作業截止日;
考試時間;
家人管理交通與財務;
日常物品由他人準備。
本文將這些統稱:
S t = external scaffolding . S_t
=
\text{external scaffolding}. S t = external scaffolding .
當:
S t ↑ , S_t\uparrow, S t ↑ ,
可能降低:
L t . L_t. L t .
11. 支架撤除
生命轉換可能造成:
S t + 1 < S t . S_{t+1}<S_t. S t + 1 < S t .
例如:
離家上大學;
搬家;
第一份全職工作;
遠距工作;
創業;
親職;
同時多角色責任。
若需求同時:
D t + 1 > D t , D_{t+1}>D_t, D t + 1 > D t ,
則:
L t + 1 = D t + 1 − ( C + S t + 1 + K ) L_{t+1}
=
D_{t+1}
-
(C+S_{t+1}+K) L t + 1 = D t + 1 − ( C + S t + 1 + K )
可能快速上升。
這是「突然出問題」的候選非病理學解釋之一。
12. Compensation
補償不是症狀消失。
定義:
K t = internally generated compensatory support . K_t
=
\text{internally generated compensatory support}. K t = internally generated compensatory support .
例如:
行事曆;
重複檢查;
提前很久出門;
過度準備;
固定物品位置;
高度結構化 routine;
自己製造 deadline;
使用提醒系統。
因此:
successful compensation ≠ absence of underlying difficulty . \boxed{
\text{successful compensation}
\neq
\text{absence of underlying difficulty}.
} successful compensation = absence of underlying difficulty .
13. 補償成本
補償本身需要資源。
定義:
C t K = compensation cost . C_t^{K}
=
\text{compensation cost}. C t K = compensation cost .
那麼真正的有效支持不應只寫:
+ K t , +K_t, + K t ,
而應寫成:
K t net = K t − C t K . K_t^{\text{net}}
=
K_t-C_t^{K}. K t net = K t − C t K .
因此:
L t = D t − ( C t + S t + K t net ) . L_t
=
D_t
-
\left(
C_t+S_t+K_t^{\text{net}}
\right). L t = D t − ( C t + S t + K t net ) .
14. High Output 不等於 Low Cost
定義外部產出:
O t . O_t. O t .
內部努力:
E t effort . E_t^{\text{effort}}. E t effort .
可能:
O t = normal or high , O_t
=
\text{normal or high}, O t = normal or high ,
但:
E t effort ≫ E t population . E_t^{\text{effort}}\gg
E_t^{\text{population}}. E t effort ≫ E t population .
因此:
visible success ≠ low internal burden . \boxed{
\text{visible success}
\neq
\text{low internal burden}.
} visible success = low internal burden .
15. Effort-to-Output Ratio
定義:
ρ t E O = E t effort O t + ε . \rho_t^{EO}
=
\frac{
E_t^{\text{effort}}
}{
O_t+\varepsilon
}. ρ t E O = O t + ε E t effort .
當:
ρ t E O ↑ , \rho_t^{EO}\uparrow, ρ t E O ↑ ,
代表維持相同輸出需要更高成本。
這是 2026 「high functioning, yet high suffering」觀點最值得轉化成可測變量的地方。
但該文屬 perspective,不是已驗證診斷規則。
16. 補償儲備
令:
B t K B_t^{K} B t K
表示補償可持續儲備。
候選:
B t + 1 K = B t K − C t K + R t recovery . B_{t+1}^{K}
=
B_t^{K}
-
C_t^{K}
+
R_t^{\text{recovery}}. B t + 1 K = B t K − C t K + R t recovery .
若長期:
C t K > R t recovery , C_t^{K}
>
R_t^{\text{recovery}}, C t K > R t recovery ,
則:
B t K ↓ . B_t^{K}\downarrow. B t K ↓ .
當:
B t K < θ B , B_t^{K}<\theta_B, B t K < θ B ,
原本可維持的補償可能失效。
17. Compensation Collapse
定義:
K t effective = K t ⋅ g ( B t K ) . K_t^{\text{effective}}
=
K_t
\cdot
g(B_t^{K}). K t effective = K t ⋅ g ( B t K ) .
當:
B t K ↓ , B_t^{K}\downarrow, B t K ↓ ,
則:
K t effective ↓ . K_t^{\text{effective}}\downarrow. K t effective ↓ .
因此可能發生:
same underlying configuration + reduced compensation reserve → newly visible impairment . \boxed{
\text{same underlying configuration}
+
\text{reduced compensation reserve}
\rightarrow
\text{newly visible impairment}.
} same underlying configuration + reduced compensation reserve → newly visible impairment .
這不是 adult onset 的證明。
它只是 delayed visibility 的候選路徑。
18. 2026 Adult-Diagnosed vs Childhood-Diagnosed Study
Kang 等人於 2026 年比較:
N adult-diagnosed = 35 N_{\text{adult-diagnosed}}=35 N adult-diagnosed = 35
與:
N childhood-diagnosed = 37. N_{\text{childhood-diagnosed}}=37. N childhood-diagnosed = 37.
adult-diagnosed 組:
retrospective childhood symptoms 較低;
adult ADHD severity 與 childhood-diagnosed 組相近;
self-monitoring 較佳;
anxiety/depression burden 較高;
既往 psychiatric comorbidity 較多。
這與 delayed-recognition model 相容。
但更重要的反例是:
environmental factors did not form a strong diagnosis-age-specific explanatory pathway . \boxed{
\text{environmental factors did not form a strong
diagnosis-age-specific explanatory pathway}.
} environmental factors did not form a strong diagnosis-age-specific explanatory pathway .
Executive function 才是兩組症狀嚴重度最一致的相關因子。
因此本文不能寫成:
adult diagnosis = good childhood environment . \text{adult diagnosis}
=
\text{good childhood environment}. adult diagnosis = good childhood environment .
19. Executive Dysfunction 是共同路徑,而不是晚診斷專屬機制
2026 Kang 研究顯示:
E F → inattention EF
\rightarrow
\text{inattention} E F → inattention
與:
E F → hyperactivity/impulsivity EF
\rightarrow
\text{hyperactivity/impulsivity} E F → hyperactivity/impulsivity
在兩組皆存在。
所以 DCVH 只研究:
when an underlying difficulty becomes visible . \boxed{
\text{when an underlying difficulty becomes visible}.
} when an underlying difficulty becomes visible .
它不是:
what biologically causes ADHD . \boxed{
\text{what biologically causes ADHD}.
} what biologically causes ADHD .
20. Late-Onset Trajectory Literature
2026 Bayard 等人的 scoping review 專門整理 population-based longitudinal cohorts 中的 late-onset symptom trajectories。
該 review 顯示:
late-emerging ADHD-like trajectories \text{late-emerging ADHD-like trajectories} late-emerging ADHD-like trajectories
在多個 cohort 中被報告。
部分研究支持:
higher IQ , \text{higher IQ}, higher IQ ,
greater childhood resources \text{greater childhood resources} greater childhood resources
與 late-emerging trajectories 關聯,形成 compensation/scaffolding hypothesis。
但 review 同時指出重要缺口,包括 longitudinal studies 對:
sleep-wake disturbance;
treatment history;
contextual supports;
neurological antecedents;
缺乏系統測量。
因此:
late-onset trajectory evidence ≠ proof of a single adult-onset ADHD entity . \boxed{
\text{late-onset trajectory evidence}
\neq
\text{proof of a single adult-onset ADHD entity}.
} late-onset trajectory evidence = proof of a single adult-onset ADHD entity .
21. Adult-Onset Controversy
成人出現 ADHD-like symptoms 可能有不同來源。
令:
Y adult Y_{\text{adult}} Y adult
表示成年 ADHD-like phenotype。
則:
Y adult = F ( N , P , S , C , M , E ) , Y_{\text{adult}}
=
F
\left(
N,
P,
S,
C,
M,
E
\right), Y adult = F ( N , P , S , C , M , E ) ,
其中可能包括:
neurodevelopmental ADHD;
psychiatric conditions;
sleep disturbance;
substance effects;
neurological injury;
environmental overload;
measurement/recall issues。
所以:
adult ADHD-like phenotype ≠ one etiology . \boxed{
\text{adult ADHD-like phenotype}
\neq
\text{one etiology}.
} adult ADHD-like phenotype = one etiology .
22. Retrospective Recall Problem
成人診斷常需要回溯:
t < t age 12 . t<t_{\text{age 12}}. t < t age 12 .
但:
M childhood M_{\text{childhood}} M childhood
可能受:
memory decay;
reinterpretation after diagnosis;
lack of records;
informant absence;
observer bias;
影響。
因此:
retrospective childhood score ≠ perfect childhood ground truth . \boxed{
\text{retrospective childhood score}
\neq
\text{perfect childhood ground truth}.
} retrospective childhood score = perfect childhood ground truth .
23. Multi-Informant State
定義:
Y t ( s e l f ) , Y_t^{(self)}, Y t ( se l f ) ,
Y t ( p a r e n t ) , Y_t^{(parent)}, Y t ( p a r e n t ) ,
Y t ( t e a c h e r ) , Y_t^{(teacher)}, Y t ( t e a c h er ) ,
Y t ( p a r t n e r ) . Y_t^{(partner)}. Y t ( p a r t n er ) .
它們可以:
Y t ( s e l f ) ≠ Y t ( p a r e n t ) ≠ Y t ( t e a c h e r ) . Y_t^{(self)}
\neq
Y_t^{(parent)}
\neq
Y_t^{(teacher)}. Y t ( se l f ) = Y t ( p a r e n t ) = Y t ( t e a c h er ) .
這不必然表示有人「錯」。
不同觀察者位於不同 context。
24. Observation Coverage
令:
Ω o ( t ) \Omega_o(t) Ω o ( t )
表示觀察者 o o o 能看到的生活域。
例如教師主要看到:
Ω teacher = classroom . \Omega_{\text{teacher}}
=
\text{classroom}. Ω teacher = classroom .
家長看到:
Ω parent = home . \Omega_{\text{parent}}
=
\text{home}. Ω parent = home .
本人看到:
Ω self = internal + cross-context . \Omega_{\text{self}}
=
\text{internal + cross-context}. Ω self = internal + cross-context .
因此:
observer disagreement can arise from different sampling domains . \boxed{
\text{observer disagreement}
\text{ can arise from different sampling domains}.
} observer disagreement can arise from different sampling domains .
25. Masking:可以研究,但不能先本體化
2026 年 ADHD masking/camouflaging literature 明顯增加。
但 Adamou 於 British Journal of Psychiatry 專門提出警告:
ADHD camouflaging 尚缺乏足夠 construct validation。
因此本文不使用:
M t mask M_t^{\text{mask}} M t mask
作為已確立核心變量。
只保留候選:
M t mask = behavioral presentation management . M_t^{\text{mask}}
=
\text{behavioral presentation management}. M t mask = behavioral presentation management .
其與:
K t = functional compensation K_t
=
\text{functional compensation} K t = functional compensation
必須區分。
26. Compensation 與 Masking 的候選分離
本文暫時定義:
K t = strategy that improves functioning . K_t
=
\text{strategy that improves functioning}. K t = strategy that improves functioning .
而:
M t mask = strategy that reduces observable atypicality . M_t^{\text{mask}}
=
\text{strategy that reduces observable atypicality}. M t mask = strategy that reduces observable atypicality .
所以:
K t ≠ M t mask . \boxed{
K_t
\neq
M_t^{\text{mask}}.
} K t = M t mask .
可能:
K t > 0 , M t mask ≈ 0 , K_t>0,
\qquad
M_t^{\text{mask}}\approx0, K t > 0 , M t mask ≈ 0 ,
例如使用 calendar。
也可能:
M t mask > 0 M_t^{\text{mask}}>0 M t mask > 0
但功能本身沒有改善。
27. Visibility Reduction
若 masking 類行為成立,可能:
V t ′ = V t − λ M t mask . V_t'
=
V_t
-
\lambda M_t^{\text{mask}}. V t ′ = V t − λ M t mask .
但因其 construct validity 尚未成熟,本文只把這當可檢驗候選式。
28. 女性/女孩的診斷延遲
現有 population 與 clinical records 持續顯示女性接受 ADHD 或更廣泛 neurodevelopmental referral/diagnosis 的年齡偏晚。
2026 Scotland clinical population study 觀察 408 人、30 個 multidisciplinary teams,在跨 neurodevelopmental diagnoses 的整體資料中,女性 referral 與 diagnosis 的 median age 約比男性晚五年。
另外 ADHD-specific population-register literature 亦長期顯示女性診斷較晚。
因此:
sex/gender → detection process \boxed{
\text{sex/gender}
\rightarrow
\text{detection process}
} sex/gender → detection process
是不能忽略的變量。
但不能簡化成:
female ADHD = masked ADHD . \text{female ADHD}
=
\text{masked ADHD}. female ADHD = masked ADHD .
29. 女性 delayed diagnosis 的多因子模型
候選:
T diag = F ( P presentation , I internalizing , G norm , R referral , A care , C criteria ) . T_{\text{diag}}
=
F
\left(
P_{\text{presentation}},
I_{\text{internalizing}},
G_{\text{norm}},
R_{\text{referral}},
A_{\text{care}},
C_{\text{criteria}}
\right). T diag = F ( P presentation , I internalizing , G norm , R referral , A care , C criteria ) .
其中:
P presentation P_{\text{presentation}} P presentation :表型形式;
I internalizing I_{\text{internalizing}} I internalizing :內化症狀;
G norm G_{\text{norm}} G norm :gender expectation;
R referral R_{\text{referral}} R referral :轉介行為;
A care A_{\text{care}} A care :醫療可近性;
C criteria C_{\text{criteria}} C criteria :診斷工具/標準敏感度。
30. 2025 Late-Diagnosed Women Study
Holden 與 Kobayashi-Wood 於 2025 年研究 28 名 late-diagnosed women。
參與者年齡:
19 − 72 , 19-72, 19 − 72 ,
平均診斷年齡:
35.9. 35.9. 35.9.
研究為 mixed-methods/qualitative emphasis。
參與者報告:
長期 criticism;
lack of support;
anxiety/depression attribution;
educational/occupational impact;
masking/coping;
post-diagnosis validation;
grief over delayed recognition。
這提供 lived-experience evidence。
但:
N = 28 N=28 N = 28
且以 self-selected qualitative sample 為主。
因此:
rich phenomenology ≠ population prevalence estimate . \boxed{
\text{rich phenomenology}
\neq
\text{population prevalence estimate}.
} rich phenomenology = population prevalence estimate .
31. Internalizing Symptoms 可能降低 ADHD 的表面辨識度
若外部觀察看到:
anxiety , \text{anxiety}, anxiety ,
depression , \text{depression}, depression ,
emotional distress , \text{emotional distress}, emotional distress ,
則 diagnostic attribution 可能首先指向其他 disorder。
因此:
P ( ADHD referral ∣ internalizing presentation ) P
\left(
\text{ADHD referral}
\mid
\text{internalizing presentation}
\right) P ( ADHD referral ∣ internalizing presentation )
可能下降。
但 internalizing condition 可以:
是 independent comorbidity;
是 secondary consequence;
與 ADHD 同時存在;
或完全解釋部分 symptoms。
所以不能假設:
anxiety = hidden ADHD . \text{anxiety}
=
\text{hidden ADHD}. anxiety = hidden ADHD .
32. Diagnostic Shadowing/Attribution Competition
令候選診斷集合:
D = { d 1 , d 2 , … , d m } . \mathcal D
=
\{d_1,d_2,\ldots,d_m\}. D = { d 1 , d 2 , … , d m } .
醫療系統對觀察資料:
Y Y Y
產生 posterior:
P ( d j ∣ Y ) . P(d_j\mid Y). P ( d j ∣ Y ) .
若:
P ( d anxiety ∣ Y ) ≫ P ( d ADHD ∣ Y ) , P(d_{\text{anxiety}}\mid Y)
\gg
P(d_{\text{ADHD}}\mid Y), P ( d anxiety ∣ Y ) ≫ P ( d ADHD ∣ Y ) ,
則 ADHD 評估可能延遲。
這是 attribution competition,不代表初始診斷必然錯。
33. High Intelligence/High Achievement 不等於無 ADHD
2025 World Psychiatry review 明確討論 highly intelligent adults with ADHD。
一個高能力個體可能:
O t ≥ population norm , O_t
\geq
\text{population norm}, O t ≥ population norm ,
但相對個人潛能:
O t ≪ P t potential . O_t
\ll
P_t^{\text{potential}}. O t ≪ P t potential .
因此定義:
G t potential = P t potential − O t . G_t^{\text{potential}}
=
P_t^{\text{potential}}
-
O_t. G t potential = P t potential − O t .
可能:
G t potential ≫ 0 G_t^{\text{potential}}\gg0 G t potential ≫ 0
即使外部結果仍「看起來正常」。
34. Population-Norm Impairment 與 Potential-Relative Impairment
兩種判準:
I t norm = difference from population standard , I_t^{\text{norm}}
=
\text{difference from population standard}, I t norm = difference from population standard ,
I t potential = difference from individual potential . I_t^{\text{potential}}
=
\text{difference from individual potential}. I t potential = difference from individual potential .
它們不能混為一談。
臨床 impairment 通常不能只靠「沒有發揮天才潛力」判定,但 potential-relative underperformance 可以構成求助與研究的重要背景。
35. Life-Transition Shock
定義生命轉換事件:
Δ t L . \Delta_t^{L}. Δ t L .
若:
Δ D t > 0 \Delta D_t>0 Δ D t > 0
且:
Δ S t < 0 , \Delta S_t<0, Δ S t < 0 ,
則需求—支持差的變化:
Δ L t = Δ D t − Δ C t − Δ S t − Δ K t . \Delta L_t
=
\Delta D_t
-
\Delta C_t
-
\Delta S_t
-
\Delta K_t. Δ L t = Δ D t − Δ C t − Δ S t − Δ K t .
如果:
Δ L t ≫ 0 , \Delta L_t\gg0, Δ L t ≫ 0 ,
原本低可見配置可能跨過:
θ I \theta_I θ I
與:
θ V . \theta_V. θ V .
36. 生命階段不是病因,而是揭露條件
本文不寫:
college → ADHD . \text{college}
\rightarrow
\text{ADHD}. college → ADHD .
而是:
life transition → new demand-support relation . \boxed{
\text{life transition}
\rightarrow
\text{new demand-support relation}.
} life transition → new demand-support relation .
同一個配置在不同生命階段可以有不同結果。
37. Diagnostic Hazard
令:
h ( t ) h(t) h ( t )
表示時間 t t t 首次接受正式 ADHD diagnosis 的 hazard。
候選:
h ( t ) = h 0 ( t ) exp ( β 1 V t + β 2 I t + β 3 A t care + β 4 K t knowledge − β 5 S t stigma ) . h(t)
=
h_0(t)
\exp
\left(
\beta_1V_t
+
\beta_2I_t
+
\beta_3A_t^{\text{care}}
+
\beta_4K_t^{\text{knowledge}}
-
\beta_5S_t^{\text{stigma}}
\right). h ( t ) = h 0 ( t ) exp ( β 1 V t + β 2 I t + β 3 A t care + β 4 K t knowledge − β 5 S t stigma ) .
這不是臨床公式。
它只是提醒:
diagnosis timing = detection process . \boxed{
\text{diagnosis timing}
=
\text{detection process}.
} diagnosis timing = detection process .
38. Diagnostic Delay
若:
T need T_{\text{need}} T need
為 clinically meaningful need 已出現時間,
T diag T_{\text{diag}} T diag
為診斷時間,
定義:
Δ T delay = T diag − T need . \Delta T_{\text{delay}}
=
T_{\text{diag}}
-
T_{\text{need}}. Δ T delay = T diag − T need .
這比:
T diag − T onset T_{\text{diag}}-T_{\text{onset}} T diag − T onset
更容易操作化。
39. Hidden Cost Accumulation
即使輸出維持,未診斷狀態可能累積:
H t + 1 = H t + C t K + C t stress + C t error − R t support . H_{t+1}
=
H_t
+
C_t^{K}
+
C_t^{\text{stress}}
+
C_t^{\text{error}}
-
R_t^{\text{support}}. H t + 1 = H t + C t K + C t stress + C t error − R t support .
其中 H t H_t H t 是 hidden burden。
因此:
O t ≈ stable O_t\approx\text{stable} O t ≈ stable
不表示:
H t ≈ 0. H_t\approx0. H t ≈ 0.
40. Hidden Burden 不是 ADHD-specific
高 effort-to-output ratio 可以出現在:
anxiety;
chronic illness;
sleep deprivation;
autism;
depression;
socioeconomic adversity;
caregiving burden。
所以:
ρ E O ↑ ≠ ADHD biomarker . \boxed{
\rho^{EO}\uparrow
\neq
\text{ADHD biomarker}.
} ρ E O ↑ = ADHD biomarker .
它只能作為跨診斷功能量。
41. DCVH 十二項核心命題
DV-H1:Late Diagnosis 非 Late Onset 命題
T diag ≠ T onset . T_{\text{diag}}
\neq
T_{\text{onset}}. T diag = T onset .
DV-H2:Demand–Support 命題
功能損害與:
L t = D t − ( C t + S t + K t ) L_t
=
D_t-(C_t+S_t+K_t) L t = D t − ( C t + S t + K t )
相關。
DV-H3:雙門檻命題
θ I ≠ θ V . \theta_I
\neq
\theta_V. θ I = θ V .
已有 impairment 不等於一定被辨識。
DV-H4:Scaffolding 命題
外部支架下降可提高:
L t . L_t. L t .
DV-H5:Compensation Cost 命題
補償效益與補償成本必須分開:
K t net = K t − C t K . K_t^{\text{net}}
=
K_t-C_t^{K}. K t net = K t − C t K .
DV-H6:Compensation Collapse 命題
長期高成本可能降低補償儲備:
B t K ↓ . B_t^{K}\downarrow. B t K ↓ .
DV-H7:Executive Commonality 命題
adult-diagnosed 與 childhood-diagnosed ADHD 可能共享 executive-function impairment,而不是具有完全不同的機制。
DV-H8:Observer-Sampling 命題
不同 informants 的差異部分來自:
Ω o \Omega_o Ω o
不同,而不只是 measurement error。
DV-H9:Sex/Gender Detection 命題
sex/gender-related presentation 與 referral processes 可改變:
T diag . T_{\text{diag}}. T diag .
DV-H10:Masking Construct Caution 命題
在 ADHD-specific construct validity 建立前:
M mask M^{\text{mask}} M mask
不應被當作臨床核心變量。
DV-H11:Life-Transition 命題
Δ D > 0 \Delta D>0 Δ D > 0
與:
Δ S < 0 \Delta S<0 Δ S < 0
可能提高 late recognition probability。
DV-H12:Incremental Value 命題
若 demand、support、compensation、visibility variables 在控制 symptoms、executive function、IQ、comorbidity 後:
Δ R 2 ≈ 0 , \Delta R^2\approx0, Δ R 2 ≈ 0 ,
則 DCVH 應被簡化。
42. 實驗一:Prospective Life-Transition Cohort
追蹤高中畢業生至大學/職場。
測量:
D t , S t , K t , L t , I t , V t . D_t,
S_t,
K_t,
L_t,
I_t,
V_t. D t , S t , K t , L t , I t , V t .
關鍵:
Δ L t → ? Δ I t . \Delta L_t
\rightarrow?
\Delta I_t. Δ L t → ? Δ I t .
而不是回溯詢問:
你以前是不是被遮住?
43. 實驗二:Scaffolding Withdrawal
設計結構化與非結構化任務環境。
Condition A:
S high . S_{\text{high}}. S high .
Condition B:
S low . S_{\text{low}}. S low .
測:
Δ P , \Delta P, Δ P ,
Δ R T , \Delta RT, Δ R T ,
Δ e r r o r , \Delta error, Δ er r or ,
Δ e f f o r t . \Delta effort. Δ e f f or t .
核心 interaction:
Profile × S . \text{Profile}
\times
S. Profile × S .
44. 實驗三:Compensation Inventory
建立 ADHD-specific compensation measure,而不是直接借用 autism camouflaging scale。
區分:
K functional , K^{\text{functional}}, K functional ,
K presentation , K^{\text{presentation}}, K presentation ,
K avoidance , K^{\text{avoidance}}, K avoidance ,
K external tool . K^{\text{external tool}}. K external tool .
並測:
C K . C^{K}. C K .
如果沒有成本資料,不能判斷某策略是否真的「高功能」。
45. 實驗四:Effort-to-Output Measurement
同時取得:
O t O_t O t
與:
E t effort . E_t^{\text{effort}}. E t effort .
例如:
task accuracy;
hours spent;
number of retries;
physiological load;
subjective exhaustion。
比較:
ρ t E O . \rho_t^{EO}. ρ t E O .
避免只用最終成績判定功能。
46. 實驗五:Multi-Informant Longitudinal Model
同一人由:
self;
parent;
teacher;
partner;
workplace observer;
重複評估。
建立:
Y i , t , o . Y_{i,t,o}. Y i , t , o .
估計:
person effect , \text{person effect}, person effect ,
context effect , \text{context effect}, context effect ,
observer effect . \text{observer effect}. observer effect .
47. 實驗六:Diagnostic Pathway Study
記錄:
T first difficulty , T_{\text{first difficulty}}, T first difficulty ,
T first help-seeking , T_{\text{first help-seeking}}, T first help-seeking ,
T first referral , T_{\text{first referral}}, T first referral ,
T ADHD assessment , T_{\text{ADHD assessment}}, T ADHD assessment ,
T diagnosis . T_{\text{diagnosis}}. T diagnosis .
這可以把「晚診斷」拆解成:
recognition delay , \text{recognition delay}, recognition delay ,
help-seeking delay , \text{help-seeking delay}, help-seeking delay ,
system delay . \text{system delay}. system delay .
48. 實驗七:Sex/Gender Referral Bias
在相同 impairment vignette 下只操弄:
sex marker;
externalizing vs internalizing presentation;
academic achievement。
測:
P ( refer for ADHD assessment ) . P(\text{refer for ADHD assessment}). P ( refer for ADHD assessment ) .
這比只比較 diagnosis ratio 更接近 detection mechanism。
49. 實驗八:Adult-Onset Differential Study
建立 prospective cohort。
要求:
childhood multi-informant records;
repeated adolescent assessment;
sleep;
mood;
substance use;
trauma;
neurological history;
treatment exposure;
environmental demand。
只有在排除 alternative pathways 後,才討論:
true late onset . \text{true late onset}. true late onset .
50. 失敗條件
DCVH 應被削弱或淘汰,如果:
D t D_t D t 、 S t S_t S t 、 K t K_t K t 無法可靠量測;
scaffolding 對 functional performance 沒有可重現 interaction;
compensation cost 與 late diagnosis 無關;
life-transition changes 無法預測 symptom visibility;
adult-diagnosed 與 childhood-diagnosed 完全由不同機制解釋;
observer/context variables 沒有增量效應;
sex/gender 不影響 referral/diagnosis timing;
masking construct 在 ADHD 中證明完全無效;
demand–support model 不優於 simple symptom model。
若:
P DCVH,out ≤ P baseline,out , P_{\text{DCVH,out}}
\leq
P_{\text{baseline,out}}, P DCVH,out ≤ P baseline,out ,
則應保留較簡單模型。
51. 最重要的反例:環境不是萬能解釋
2026 Kang 研究沒有發現 adult-diagnosed ADHD 具有一套明顯不同的 environmental predictor profile。
所以本文拒絕:
late diagnosis = supportive childhood . \boxed{
\text{late diagnosis}
=
\text{supportive childhood}.
} late diagnosis = supportive childhood .
最多只能寫:
support and demand may modulate visibility for some individuals . \boxed{
\text{support and demand may modulate visibility
for some individuals}.
} support and demand may modulate visibility for some individuals .
52. 第二個反例:Masking 不是已成熟事實
2026 British Journal of Psychiatry commentary 專門質疑 ADHD camouflaging 直接移植。
所以本文拒絕:
all late-diagnosed ADHD = masked ADHD . \boxed{
\text{all late-diagnosed ADHD}
=
\text{masked ADHD}.
} all late-diagnosed ADHD = masked ADHD .
補償、支架、行為抑制、社會印象管理與隱藏痛苦必須分別操作化。
53. 第三個反例:高成就不能自己證明補償
一個成功成人:
O t ↑ O_t\uparrow O t ↑
可能:
沒有 ADHD;
有 ADHD 且低 impairment;
有 ADHD 且高補償;
有其他認知 profile;
只是工作環境匹配。
因此:
achievement ≠ compensation evidence . \boxed{
\text{achievement}
\neq
\text{compensation evidence}.
} achievement = compensation evidence .
必須測:
E effort , E^{\text{effort}}, E effort ,
K , K, K ,
C K , C^{K}, C K ,
L t . L_t. L t .
54. 第四個反例:晚出現症狀不一定是被遮住
真正 late-emerging trajectory 不能被理論上排除。
因此本文不預設:
all adult-onset reports = missed childhood ADHD . \boxed{
\text{all adult-onset reports}
=
\text{missed childhood ADHD}.
} all adult-onset reports = missed childhood ADHD .
DCVH 只處理 delayed detection/visibility 的一類候選路徑。
55. 與前六篇整合
第 1 篇建立:
C t = dynamic cognitive configuration . \mathbf C_t
=
\text{dynamic cognitive configuration}. C t = dynamic cognitive configuration .
第 2 篇加入:
N t = neuromodulation . \mathbf N_t
=
\text{neuromodulation}. N t = neuromodulation .
第 3 篇拆分:
S t ≠ A t ≠ Π t ≠ O t ≠ U t ≠ G t . S_t
\neq
A_t
\neq
\Pi_t
\neq
O_t
\neq
U_t
\neq
G_t. S t = A t = Π t = O t = U t = G t .
第 4 篇加入:
X t = allocation-state dynamics . \mathbf X_t
=
\text{allocation-state dynamics}. X t = allocation-state dynamics .
第 5 篇加入:
G t = cognitive path topology . \mathcal G_t
=
\text{cognitive path topology}. G t = cognitive path topology .
第 6 篇加入:
( Q t , P ^ t , P t ) . (Q_t,\widehat P_t,P_t). ( Q t , P t , P t ) .
第 7 篇現在把整套模型放入生命史:
L t = D t − ( C t + S t + K t ) . \boxed{
L_t
=
D_t
-
(C_t+S_t+K_t).
} L t = D t − ( C t + S t + K t ) .
再加入:
I t , I_t, I t ,
V t , V_t, V t ,
A t care , A_t^{\text{care}}, A t care ,
形成:
configuration → function under demand → visibility → clinical detection . \boxed{
\text{configuration}
\rightarrow
\text{function under demand}
\rightarrow
\text{visibility}
\rightarrow
\text{clinical detection}.
} configuration → function under demand → visibility → clinical detection .
56. 系列目前的完整多尺度鏈
可以暫時寫成:
N t → Z t → Π t → X t → G t → O t → U t → G t → P t \boxed{
\mathbf N_t
\rightarrow
\mathbf Z_t
\rightarrow
\Pi_t
\rightarrow
\mathbf X_t
\rightarrow
\mathcal G_t
\rightarrow
O_t
\rightarrow
U_t
\rightarrow
G_t
\rightarrow
P_t
} N t → Z t → Π t → X t → G t → O t → U t → G t → P t
而生命史環境為:
( D t , S t , K t ) → L t → I t → V t → Detection t . \boxed{
(\mathbf D_t,\mathbf S_t,\mathbf K_t)
\rightarrow
L_t
\rightarrow
I_t
\rightarrow
V_t
\rightarrow
\text{Detection}_t.
} ( D t , S t , K t ) → L t → I t → V t → Detection t .
兩條鏈互相耦合。
57. 本文不主張的內容
本文不主張:
adult-diagnosed ADHD 等於 adult-onset ADHD;
所有晚診斷者兒童期都已完整符合 ADHD;
所有 adult-onset report 都只是回憶錯誤;
高 IQ 必然遮住 ADHD;
家庭支持必然延後診斷;
女性晚診斷完全由 masking 造成;
masking 已是 ADHD 成熟臨床構念;
internalizing symptoms 都是 ADHD 的後果;
anxiety/depression diagnosis 等於 misdiagnosis;
高成就代表 high-functioning ADHD;
high effort-to-output ratio 是 ADHD biomarker;
adult-diagnosed 與 childhood-diagnosed 是不同疾病;
adult-diagnosed 與 childhood-diagnosed 必然是同一疾病;
晚診斷一定造成所有後續心理問題;
本文可判斷個人何時「真正發病」。
58. 結論
成年人直到三十、四十甚至更晚才接受 ADHD 評估,不應被簡單壓縮成:
他以前沒有,現在才有。
也不能反過來自動寫成:
他一直都有,只是完美 masking。
本文提出的較弱模型是:
diagnosis timing is an emergent property of development, demand, support, compensation, visibility, observer knowledge, and care access . \boxed{
\text{diagnosis timing is an emergent property
of development, demand, support, compensation,
visibility, observer knowledge, and care access}.
} diagnosis timing is an emergent property of development, demand, support, compensation, visibility, observer knowledge, and care access .
底層配置是否造成可見困難,取決於:
L t = D t − ( C t + S t + K t ) . L_t
=
D_t
-
(C_t+S_t+K_t). L t = D t − ( C t + S t + K t ) .
而可見困難是否成為診斷,又取決於:
V t , V_t, V t ,
A t care , A_t^{\text{care}}, A t care ,
以及 diagnostic attribution。
因此:
configuration ≠ impairment ≠ visibility ≠ diagnosis . \boxed{
\text{configuration}
\neq
\text{impairment}
\neq
\text{visibility}
\neq
\text{diagnosis}.
} configuration = impairment = visibility = diagnosis .
2025–2026 的成人 ADHD 文獻確實讓 compensation、scaffolding、internalizing presentations、sex/gender detection bias 與 hidden effort 成為重要研究方向;但同時,最新研究也警告我們不能把這些敘述本體化。adult-diagnosed ADHD 的 executive dysfunction 與 childhood-diagnosed ADHD 可能高度共享;環境因素並未證明是一套獨立病因;masking/camouflaging 在 ADHD 中仍需要 construct validation;真正 adult-onset trajectory 仍是未解問題。
因此本篇最終可證偽問題是:
Can longitudinal measures of demand, scaffolding, compensation cost and clinical visibility predict when ADHD-related impairment becomes detectable better than symptom severity alone? \boxed{
\text{Can longitudinal measures of demand, scaffolding,
compensation cost and clinical visibility predict
when ADHD-related impairment becomes detectable
better than symptom severity alone?}
} Can longitudinal measures of demand, scaffolding, compensation cost and clinical visibility predict when ADHD-related impairment becomes detectable better than symptom severity alone?
如果不能,DCVH 應被簡化。
如果可以,成人「晚診斷」就可以被更精確地理解成生命史中的 detection dynamics,而不是單純的疾病有/無二元跳變。
參考文獻
Cortese, S., Bellgrove, M. A., Brikell, I., Franke, B., Goodman, D. W., Hartman, C. A., et al. Attention-deficit/hyperactivity disorder (ADHD) in adults: evidence base, uncertainties and controversies. World Psychiatry . 2025;24(3):347–371. DOI: 10.1002/wps.21374.
Kang, S., Fu, Z., Li, Q., Yang, L., & Cao, Q. Adult-diagnosed and childhood-diagnosed attention deficit/hyperactivity disorder: cognitive and environmental contributions to symptom severity across different age of diagnosis. Frontiers in Psychiatry . 2026;17:1782999. DOI: 10.3389/fpsyt.2026.1782999.
Bayard, S., Madiouni, C., Radiguer, F., Roulin, M., & Henrard, S. Late-Onset ADHD symptoms in the general population: A scoping review of longitudinal trajectories in population-based cohorts. The European Journal of Psychiatry . 2026;40(1):100337. DOI: 10.1016/j.ejpsy.2025.100337.
Melo, I. H., & Franca, G. High functioning, yet high suffering – the need to incorporate invisible struggles in adult ADHD diagnostic assessment/criteria. Frontiers in Psychiatry . 2026;17:1813029. DOI: 10.3389/fpsyt.2026.1813029.
Adamou, M. Camouflaging in ADHD: the need for construct validation before clinical adoption. The British Journal of Psychiatry . 2026;228:1–4. DOI: 10.1192/bjp.2026.10577.
Maeda, C., Knouse, L. E., Takeda, K., Masudomi, C., Takahashi, E., Katsuragawa, T., & Kumano, H. A narrative review of stigma and masking in ADHD: insights from English-language research and the Japanese cultural context. Frontiers in Psychology . 2026;17:1807337. DOI: 10.3389/fpsyg.2026.1807337.
van der Putten, W. J., et al. Is camouflaging unique for autism? A comparison of camouflaging between adults with autism and ADHD. Autism Research . 2024;17(4):812–823. PMID: 38323512.
Holden, E., & Kobayashi-Wood, H. Adverse experiences of women with undiagnosed ADHD and the invaluable role of diagnosis. Scientific Reports . 2025;15:20945. DOI: 10.1038/s41598-025-04782-y.
Maciver, D., Singh Roy, A., Johnston, L., Boilson, M., Curnow, E., Johnstone-Cooke, V., & Rutherford, M. Are we getting better at identifying and diagnosing neurodivergent girls and women? Insights into sex ratios and age of diagnosis from clinical population data in Scotland. Autism . 2026;30(2):375–389. DOI: 10.1177/13623613251383343.
Martin, J. Why are females less likely to be diagnosed with ADHD in childhood than males? The Lancet Psychiatry . 2024;11:303–310. DOI: 10.1016/S2215-0366(24)00010-5.
Mestres, F., Richarte, V., Crespín, J. J., Torrent, C., Biel, S., Ramos, C., et al. Sex differences in adults with attention-deficit/hyperactivity disorder: A population-based study. 2025. PMID: 40211765.
Kooij, J. J. S., et al. Skills and compensation strategies in adult ADHD – A qualitative study. PLoS ONE . 2017. PMID: 28953946.
Sibley, M. H., Rohde, L. A., Swanson, J. M., et al. Late-Onset ADHD Reconsidered With Comprehensive Repeated Assessments Between Ages 10 and 25. American Journal of Psychiatry . 2018;175:140–149. DOI: 10.1176/appi.ajp.2017.17030298.
Riglin, L., et al. “Late-onset” ADHD symptoms in young adulthood: is this ADHD? Journal of Child Psychology and Psychiatry . 2020. PMCID: PMC7612898.
文獻使用聲明
本文僅使用上述研究建立截至 2026-08-17 的外部實證邊界。
本文提出的 DCVH、需求—支持差 L t L_t L t 、補償儲備 B t K B_t^K B t K 、effort-to-output ratio ρ t E O \rho_t^{EO} ρ t E O 、clinical visibility V t V_t V t 、diagnostic hazard h ( t ) h(t) h ( t ) 、visibility threshold θ V \theta_V θ V 與三門檻模型,均為本文理論構件,不應被誤認為上述研究作者的原始結論。
不同研究包含 clinical ADHD、adult-diagnosed/childhood-diagnosed ADHD、population cohorts、women-only qualitative samples、跨神經發展診斷 clinical records 與 perspective/commentary papers。本文不把它們視為單一大型實驗的直接累加證據。
狀態: v0.1,理論稿新增原始臨床/人體數據: 無醫學用途: 無下一篇: 《ADHD 的連續配置空間:從亞臨床特徵到臨床診斷》