Haq Aazaz U, Sitzmann Adam F, Goldman Mona L, Maixner Daniel F, Mickey Brian J
Department of Psychiatry, University of Michigan, Ann Arbor.
Department of Psychiatry & Biobehavioral Sciences, University of California, Los Angeles.
J Clin Psychiatry. 2015 Oct;76(10):1374-84. doi: 10.4088/JCP.14r09528.
Roughly one-third of individuals with depression do not respond to electroconvulsive therapy (ECT). Reliable predictors of ECT response would be useful for patient selection, but have not been demonstrated definitively. We used meta-analysis to measure effect sizes for a series of clinical predictors of ECT response in depression.
PubMed was searched systematically to identify studies published after 1980 that tested at least 1 clinical predictor of response to ECT.
Of 51 studies identified, 32 were compatible with meta-analysis.
The weighted mean odds ratio (OR) or standardized mean difference (SMD) was computed for each of 10 clinical predictors, based on dichotomous outcomes (responder vs nonresponder). Statistical analyses examined robustness, bias, and heterogeneity.
Shorter depressive episode duration predicted higher ECT response rate (SMD = -0.37, 7 studies, 702 subjects, P = 4 × 10(-6)). History of medication failure in the current episode was also a robust predictor: response rates were 58% and 70%, respectively, for those with and without medication failure (OR = 0.56, 11 studies, 1,175 subjects, P = 1 × 10(-5)). Greater age and psychotic features were weakly associated with higher ECT response rates, but heterogeneity was notable. Bipolar diagnosis, sex, age at onset, and number of previous episodes were not significant predictors. Analyses of symptom severity and melancholic features were inconclusive due to study heterogeneity.
Longer depressive episodes and medication failure at baseline are robust predictors of poor response to ECT, with effect sizes that are modest but clinically relevant. Patient characteristics used traditionally such as age, psychosis, and melancholic features are less likely to be clinically useful. More robust clinical and biological predictors are needed for management of depressed patients considering ECT.
约三分之一的抑郁症患者对电休克治疗（ECT）无反应。ECT反应的可靠预测指标对患者选择会很有帮助，但尚未得到确切证实。我们采用荟萃分析来衡量抑郁症患者ECT反应的一系列临床预测指标的效应大小。
系统检索PubMed，以识别1980年后发表的测试至少一种ECT反应临床预测指标的研究。
在识别出的51项研究中，32项符合荟萃分析要求。
基于二分结果（反应者与无反应者），计算10种临床预测指标各自的加权平均比值比（OR）或标准化平均差（SMD）。统计分析检验了稳健性、偏倚和异质性。
抑郁发作持续时间较短预示ECT反应率较高（SMD = -0.37，7项研究，702名受试者，P = 4×10⁻⁶）。当前发作期药物治疗失败史也是一个可靠的预测指标：有和无药物治疗失败史的患者反应率分别为58%和70%（OR = 0.56，11项研究，1175名受试者，P = 1×10⁻⁵）。年龄较大和有精神病性特征与较高的ECT反应率弱相关，但异质性显著。双相诊断、性别、起病年龄和既往发作次数不是显著的预测指标。由于研究异质性，对症状严重程度和抑郁特征的分析尚无定论。
抑郁发作时间较长和基线时药物治疗失败是ECT反应不佳的可靠预测指标，效应大小虽适度但具有临床相关性。传统上使用的患者特征如年龄、精神病性和抑郁特征在临床上不太可能有用。对于考虑接受ECT治疗的抑郁症患者的管理，需要更可靠的临床和生物学预测指标。
