The aim of this systematic review and meta-analysis was to investigate the diagnostic, the dimensional mean-level, and rank-order stability of personality disorders (PDs) and PD criteria over time. EMBASE, PsycInfo, PubMed, and Web of Science were searched for peer-reviewed studies published in either English, German, or French between the first publication of the third edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III) in 1980 and December 20, 2022. Inclusion criteria were a prospective longitudinal study design, assessing the stability of PDs or PD criteria over at least two measurement occasions at least one month apart, and using the same assessment at baseline and follow-up. Effect sizes included proportion of enduring cases (i.e., diagnostic stability), test–retest correlations (i.e., dimensional rank-order stability), and within-group standardized mean differences (i.e., dimensional mean-level stability), based on the first and last available measurement occasion. From an initial pool of 1473 studies, 40 were included in our analyses, covering 38,432 participants. 56.7% maintained the diagnosis of any PD, and 45.2% maintained the diagnosis of borderline PD over time. Findings on the dimensional mean-level stability indicate that most PD criteria significantly decreased from baseline to follow-up, except for antisocial, obsessive-compulsive, and schizoid PD criteria. Findings on the dimensional rank-order stability suggested moderate estimates, except for antisocial PD criteria, which were found to be high. Findings indicated that both PDs and PD criteria were only moderately stable, although between study heterogeneity was high, and stability itself depended on several methodological factors.

本系统回顾和荟萃分析的目的是调查人格障碍 (PD) 的诊断、量纲平均水平和排序稳定性以及 PD 标准随时间的推移。在 EMBASE、PsycInfo、PubMed 和 Web of Science 中搜索了 1980 年《精神疾病诊断和统计手册》第三版 (DSM-III) 首次出版之间以英语、德语或法语发表的同行评审研究和 2022 年 12 月 20 日。纳入标准是一项前瞻性纵向研究设计，评估 PD 或 PD 标准在至少相隔至少一个月的两次测量中的稳定性，并在基线和随访中使用相同的评估。影响大小包括持久病例的比例（即诊断稳定性）、测试-再测试相关性（即 维度排序稳定性）和组内标准化平均差（即维度平均水平稳定性），基于第一个和最后一个可用的测量时机。从最初的 1473 项研究中，有 40 项被纳入我们的分析，涵盖 38,432 名参与者。随着时间的推移，56.7% 的人维持任何 PD 的诊断，45.2% 的人维持边缘 PD 的诊断。关于维度平均水平稳定性的研究结果表明，大多数 PD 标准从基线到随访显着下降，但反社会、强迫症和精神分裂症 PD 标准除外。关于维度排序稳定性的调查结果建议进行适度估计，但反社会 PD 标准除外，该标准被发现很高。结果表明，PD 和 PD 标准都只是适度稳定，