To the Editor A systematic review and clinical practice guideline on early, accurate diagnosis and early intervention in cerebral palsy (CP), as undertaken by Novak et al,¹ is a timely addition to the literature. However, it is unclear why the authors restricted their search criteria to only systematic reviews/meta-analyses and CPGs. By doing so, they overlooked several relevant, well-conducted prognostic studies.^2- 5 Unfortunately, this casts significant doubt on their recommendation to combine structural brain magnetic resonance imaging (MRI) with general movements assessment (GMA) for accurate diagnosis of CP before age 5 months. They report MRI to possess a sensitivity of 86% to 89%. However, these data were derived from a clinical practice guideline that examined cross-sectional MRI studies (reference 27 in the Novak et al article),¹ which are unable to determine prognostic test properties, and from a systematic review that found only 1 eligible cohort study of 61 preterm infants (reference 21 in the Novak et al article).¹ They excluded the largest MRI study in preterm infants (N = 445),² which yielded a sensitivity of 48% and likelihood ratio for a positive test result (LR+) of 2.8 for CP prediction. A 2015 meta-analysis,³ also excluded, reported a similarly low LR+ of 3.7 for MRI. When applied to a very preterm infant (assuming 10% CP prevalence) with an abnormal MRI, a LR+ of 3.7 only increases this infant’s probability of developing CP to 29%.

致编辑诺瓦克等[ ^1]对脑瘫（CP）进行早期，准确的诊断和早期干预的系统评价和临床实践指南是对文献的及时补充。但是，尚不清楚为什么作者将搜索标准限制为仅系统评价/元分析和CPG。通过这样做，他们忽略了一些相关的，进行良好的预后研究。^{2 - 5}不幸的是，这使他们对将结构性脑磁共振成像（MRI）与一般运动评估（GMA）结合以准确诊断5个月前的CP的建议产生了很大的疑问。他们报告说MRI的敏感性为86％至89％。然而，这些数据来自于临床实践指南，该指南检查了横截面MRI研究（Novak等人的参考文献27）¹，这些研究无法确定预后的测试性质，并且来自系统评价，仅发现1个符合条件的队列对61个早产儿进行的研究（Novak等人的文章中的参考文献21）。¹他们排除了最大的早产儿MRI研究（N = 445），²对于CP预测，其阳性测试结果（LR +）的灵敏度为48％，似然比为2.8。2015年的荟萃分析（^3个也不包括在内）报告了MRI的LR +较低，为3.7。如果将MRI应用于MRI异常的早产婴儿（假设CP患病率为10％），则LR +为3.7只会使该婴儿发生CP的可能性增加到29％。