A systematic review published in 2013 reported 32% of children on the autism spectrum experience skill loss, known as autistic regression. However, the frequency varied depending on definition and measures used to capture skills. Retrospective parent report and prospective observation indicate loss of language and/or social skills, with motor skills typically unaffected. Our aim was to update the prevalence and age of onset of autistic regression through a meta‐analysis of the literature to understand if there have been changes to the reported onset and prevalence since 2010. A systematic literature search was conducted using Medline, Embase, PsycINFO, and the Cochrane Library databases and included studies published from 2010 onward. Risk of bias assessment was performed on included studies. A random effects model was used to calculate the pooled prevalence and age of onset of autistic regression. Ninety‐seven studies were included in the systematic review, of which 75 studies involving 33,014 participants had sufficient data for meta‐analytic syntheses. The pooled proportion of autistic regression was 30% (95% confidence interval [CI]: 27–32%) but heterogeneity was high (I² = 96.91) and did not reduce with sensitivity or subgroup analyses based on study design or clinical differences, respectively. Prevalence varied according to risk of bias (low: 27%) and definition of regression (language: 20%, language/social: 40%, mixed: 30%, and unspecified: 27%). Weighted average age of onset was 19.8 months. Findings from this meta‐analysis highlight the importance of developing a standardized definition of autistic regression, and tools to measure this at multiple time points during early childhood development.

中文翻译：

---

自闭症谱系障碍患儿回归发病的年龄和发病年龄：系统评价和Meta分析更新

2013年发表的系统评价报告称，有32％的儿童在自闭症谱系中经历了技能丧失，这称为自闭症回归。但是，频率根据定义和捕获技能所采用的措施而有所不同。回顾性父母报告和前瞻性观察表明，语言和/或社交技能丧失，运动技能通常不受影响。我们的目的是通过对文献的荟萃分析来更新自闭症回归的患病率和发病年龄，以了解自2010年以来报道的发病率和患病率是否发生了变化。我们使用Medline，Embase，PsycINFO进行了系统的文献检索，以及Cochrane图书馆数据库，其中包括从2010年起发布的研究。在纳入的研究中进行了偏倚评估的风险。使用随机效应模型来计算自闭回归的合并患病率和发病年龄。系统评价包括97项研究，其中涉及33,014名参与者的75项研究具有足够的数据进行荟萃分析综合。自闭回归的合并比例为30％（95％置信区间[CI]：27–32％），但异质性很高（I ² = 96.91），并且未分别根据研究设计或临床差异进行敏感性分析或亚组分析。患病率因偏倚风险（低：27％）和回归定义（语言：20％，语言/社交：40％，混合：30％和未指定：27％）而异。加权平均发病年龄为19.8个月。这项荟萃分析的结果凸显了开发自闭症回归标准定义的重要性，以及在儿童早期发育的多个时间点对其进行测量的工具。