There is widespread overlap across major psychiatric disorders, and this is the case at different levels of observations, from genetic variants to brain structures and function and to symptoms. However, it remains unknown to what extent these commonalities at different levels of observation map onto each other. Here, we systematically review and compare the degree of similarity between psychiatric disorders at all available levels of observation. We searched PubMed and EMBASE between January 1, 2009 and September 8, 2022. We included original studies comparing at least four of the following five diagnostic groups: Schizophrenia, Bipolar Disorder, Major Depressive Disorder, Autism Spectrum Disorder, and Attention Deficit Hyperactivity Disorder, with measures of similarities between all disorder pairs. Data extraction and synthesis were performed by two independent researchers, following the PRISMA guidelines. As main outcome measure, we assessed the Pearson correlation measuring the degree of similarity across disorders pairs between studies and biological levels of observation. We identified 2975 studies, of which 28 were eligible for analysis, featuring similarity measures based on single-nucleotide polymorphisms, gene-based analyses, gene expression, structural and functional connectivity neuroimaging measures. The majority of correlations (88.6%) across disorders between studies, within and between levels of observation, were positive. To identify a consensus ranking of similarities between disorders, we performed a principal component analysis. Its first dimension explained 51.4% (95% CI: 43.2, 65.4) of the variance in disorder similarities across studies and levels of observation. Based on levels of genetic correlation, we estimated the probability of another psychiatric diagnosis in first-degree relatives and showed that they were systematically lower than those observed in population studies. Our findings highlight that genetic and brain factors may underlie a large proportion, but not all of the diagnostic overlaps observed in the clinic.

主要精神疾病之间存在广泛的重叠，从遗传变异到大脑结构和功能，再到症状，在不同的观察水平上都是如此。然而，目前尚不清楚不同观察层面的这些共性在多大程度上相互映射。在这里，我们系统地回顾和比较了所有可用观察水平的精神疾病之间的相似程度。我们在 2009 年 1 月 1 日至 2022 年 9 月 8 日期间检索了 PubMed 和 EMBASE。我们纳入了比较以下五个诊断组中至少四个的原始研究：精神分裂症、双相情感障碍、重度抑郁症、自闭症谱系障碍和注意力缺陷多动障碍，测量所有疾病对之间的相似性。数据提取和合成由两名独立研究人员按照 PRISMA 指南进行。作为主要结果指标，我们评估了皮尔逊相关性，衡量研究和观察的生物学水平之间的疾病对的相似程度。我们确定了 2975 项研究，其中 28 项符合分析条件，其特点是基于单核苷酸多态性的相似性测量、基于基因的分析、基因表达、结构和功能连接性神经影像测量。研究之间、观察水平内和观察水平之间的疾病之间的大多数相关性（88.6%）是正相关的。为了确定疾病之间相似性的共识排名，我们进行了主成分分析。它的第一个维度解释了研究和观察水平之间疾病相似性差异的 51.4% (95% CI: 43.2, 65.4)。根据遗传相关性水平，我们估计了一级亲属再次被诊断为精神病的概率，并表明其系统性地低于人群研究中观察到的概率。我们的研究结果强调，遗传和大脑因素可能是很大一部分，但不是临床中观察到的所有诊断重叠的原因。