The last 10–20 years have seen a rapid expansion in population-based cohort studies reporting on risk factors for dementia. Whilst the largest and longest-running of these were generally set up for other outcomes originally (e.g. cardiovascular endpoints), there are a growing number of cohort studies specifically created for investigating dementia. Aside from the wealth of findings on cardiovascular or other physical health risk factors, and on sociodemographic risk factors already wellrecognised (e.g. age, gender, education), probably the most studied exposure for dementia risk has been prior mental health. This in turn perhaps stems from researchers’ clinical experience of people presenting initially with functional mental disorders in whom dementia becomes evident at a later date, and the natural curiosity concerning whether this is more than coincidence. In what is now a substantial body of well-conducted prospective research, associations between affective distress/disorders and increased dementia risk are commonly and fairly consistently reported. Most of these findings are for depression or depressive symptoms, although they extend to associations with midlife psychological distress (1, 2), and with negative affect or proneness to psychological distress as personality traits (3, 4). As concluded in a 2016 systematic review (5), people reporting anxiety symptoms have also been found to have a higher risk of subsequent dementia, and this is supported by evidence from more recent reports (6, 7), although previous investigations have tended to ascertain anxiety from scores on screening scales. The study reported by Santab arbara and colleagues in this issue (8) has the advantage of using a more in-depth schedule for mental health symptoms and an accompanying algorithm in wide use which seeks to identify and distinguish clinically significant anxiety disorder, amongst other disorders. Their findings, of higher dementia incidence in people with case-level anxiety disorder, therefore have higher potential relevance (than those derived from screening scale cut-offs) for clinicians treating people with late-life anxiety disorders. However, although the key dementia risk associations were found to be independent of depression in this study, it would be premature to assume a specific role of anxiety because the two disorders are so closely interrelated. Given findings from other studies, the observed association might well reflect one with affective symptoms/disorders as a whole, although further investigation of risk symptom profiles would be helpful. For example, depressive and anxiety symptoms have been found to show different patterns of association with mortality as an outcome (9) and different associations with IQ as an exposure (10) in general population samples, despite being highly aetiologically related. The question of causality is routinely challenging for studies of dementia risk factors with similar durations of follow-up to the study here (e.g. <10 years), because of the well-recognised long periods over which the pathologies underlying disorders such as Alzheimer’s disease develop. For studies of depression or anxiety as risk factors for dementia, it remains quite possible that at least some of the observed associations reflect affective symptoms occurring as part of the dementia prodrome, before cognitive impairment is sufficiently severe to be noticeable or causing difficulties. This might involve a simple reaction to early difficulties experienced with memory and/or other cognitive functions, although many studies (including that of Santab arbara and colleagues) have not found any/much effect of adjusting for self-reported cognitive decline. Instead, it seems more likely that prodromal affective symptoms, if present, arise as a direct consequence of the neurodegenerative process even in people who have not apparently noticed any difficulties with cognitive function. However, an alternative possibility is that affective disorders, whether anxiety or depressive, might directly act as risk factors for dementia, or might reflect another underlying causal process. This is clearly important because of the potential for risk modification through intervention, but the challenge remains of how to demonstrate this risk effect. Distinguishing risk and prodromal factors is notoriously problematic when risk factors are measured within the period where neurodegeneration is significant enough to cause symptoms. It is much easier to assume causation when the risk factor is

中文翻译：

---

焦虑和痴呆：原因还是结果？

在过去的 10-20 年中，报告痴呆症危险因素的基于人群的队列研究迅速扩大。虽然其中规模最大和持续时间最长的研究最初通常是针对其他结果（例如心血管终点）而设立的，但越来越多的队列研究专门用于调查痴呆症。除了关于心血管或其他身体健康风险因素的大量研究结果，以及已经充分认识到的社会人口学风险因素（例如年龄、性别、教育），可能研究最多的痴呆风险暴露是先前的心理健康。这反过来可能源于研究人员的临床经验，即最初出现功能性精神障碍的人，在这些人的后期痴呆症变得明显，以及关于这是否不仅仅是巧合的自然好奇心。在现在大量进行良好的前瞻性研究中，情感困扰/障碍与痴呆症风险增加之间的关联通常且相当一致地被报道。大多数这些发现是针对抑郁症或抑郁症状的，尽管它们扩展到与中年心理困扰的关联 (1, 2)，以及作为人格特征的负面影响或心理困扰倾向 (3, 4)。正如 2016 年的系统评价 (5) 得出的结论，报告焦虑症状的人也被发现有更高的后续痴呆风险，这得到了最近报告 (6, 7) 的证据的支持，尽管之前的调查倾向于从筛选量表的分数确定焦虑。Santab arbara 及其同事在本期 (8) 中报告的研究具有使用更深入的心理健康症状时间表和广泛使用的伴随算法的优势，该算法旨在识别和区分具有临床意义的焦虑症以及其他疾病. 他们的研究结果表明，病例级焦虑症患者的痴呆发病率较高，因此对于临床医生治疗晚年焦虑症患者具有更高的潜在相关性（比从筛查量表中得出的结论）。然而，尽管在这项研究中发现关键的痴呆症风险关联与抑郁症无关，但假设焦虑症的特定作用还为时过早，因为这两种疾病密切相关。鉴于其他研究的结果，观察到的关联可能很好地反映了一个整体的情感症状/障碍，尽管进一步调查风险症状概况会有所帮助。例如，尽管在病因学上高度相关，但在一般人群样本中，抑郁和焦虑症状显示出与作为结果的死亡率 (9) 和与作为暴露的 IQ 的不同关联 (10) 的不同模式。因果关系的问题对于痴呆风险因素的研究来说通常具有挑战性，因为这里的研究具有相似的随访时间（例如 <10 年），因为众所周知，阿尔茨海默病等疾病的病理基础是在很长一段时间内发生的. 对于抑郁症或焦虑症作为痴呆症危险因素的研究，在认知障碍严重到足以引起注意或造成困难之前，至少有一些观察到的关联很可能反映了作为痴呆前驱症状的一部分发生的情感症状。这可能涉及对记忆和/或其他认知功能的早期困难的简单反应，尽管许多研究（包括 Santab arbara 及其同事的研究）并未发现调整自我报告的认知衰退有任何/很大的影响。相反，即使存在明显没有注意到任何认知功能困难的人，前驱情感症状似乎更有可能是神经退行性过程的直接结果。然而，另一种可能性是情感障碍，无论是焦虑还是抑郁，可能直接作为痴呆症的危险因素，或者可能反映另一个潜在的因果过程。这显然很重要，因为通过干预可以改变风险，但挑战仍然是如何证明这种风险影响。当在神经变性严重到足以引起症状的时期内测量风险因素时，区分风险和前驱因素是众所周知的问题。当风险因素为 当在神经变性严重到足以引起症状的时期内测量风险因素时，区分风险和前驱因素是众所周知的问题。当风险因素为 当在神经变性严重到足以引起症状的时期内测量风险因素时，区分风险和前驱因素是众所周知的问题。当风险因素为