Department of Psychological Medicine, Institute of Psychiatry, Kings College London, UK.
BMC Psychiatry. 2013 Apr 15;13:113. doi: 10.1186/1471-244X-13-113.
Depression is a major cause of disability worldwide, and computerised cognitive behavioural therapy (CCBT) is expected to be a more augmentative and efficient treatment. According to previous meta-analyses of CCBT, there is a need for a meta-analytic revaluation of the short-term effectiveness of this therapy and for an evaluation of its long-term effects, functional improvement and dropout.
Five databases were used (MEDLINE, PsycINFO, EMBASE, CENTRAL and CiNii). We included all RCTs with proper concealment and blinding of outcome assessment for the clinical effectiveness of CCBT in adults (aged 18 and over) with depression. Using Cohen's method, the standard mean difference (SMD) for the overall pooled effects across the included studies was estimated with a random effect model. The main outcome measure and the relative risk of dropout were included in the meta-analysis.
Fourteen trials met the inclusion criteria, and sixteen comparisons from these were used for the largest meta-analysis ever. All research used appropriate random sequence generation and Intention-to-Treat analyses (ITT), and employed self-reported measures as the primary outcome. For the sixteen comparisons (2807 participants) comparing CCBT and control conditions, the pooled SMD was -0.48 [95% IC -0.63 to -0.33], suggesting similar effect to the past reviews. Also, there was no significant clinical effect at long follow-up and no improvement of function found. Furthermore, a significantly higher drop-out rate was found for CCBT than for controls. When including studies without BDI as a rating scale and with only modern imputation as sensitivity analysis, the pooled SMD remained significant despite the reduction from a moderate to a small effect. Significant publication bias was found in a funnel plot and on two tests (Begg's p = 0.09; Egger's p = 0.01). Using a trim and fill analysis, the SMD was -0.32 [95% CI -0.49 to -0.16].
Despite a short-term reduction in depression at post-treatment, the effect at long follow-up and the function improvement were not significant, with significantly high drop-out. Considering the risk of bias, our meta-analysis implied that the clinical usefulness of current CCBT for adult depression may need to be re-considered downwards in terms of practical implementation and methodological validity.
抑郁症是全球范围内导致残疾的主要原因，预计计算机化认知行为疗法（CCBT）将是一种更增强和有效的治疗方法。根据之前对 CCBT 的荟萃分析，需要对这种疗法的短期疗效进行荟萃分析再评估，并评估其长期效果、功能改善和脱落率。
使用了五个数据库（MEDLINE、PsycINFO、EMBASE、CENTRAL 和 CiNii）。我们纳入了所有关于 CCBT 治疗成人（年龄在 18 岁及以上）抑郁症的临床有效性的 RCT，这些 RCT 均采用适当的隐匿和结局评估盲法。采用 Cohen 法，以随机效应模型估计纳入研究总体汇总效应的标准化均数差（SMD）。荟萃分析纳入了主要结局测量和脱落率的相对风险。
有 14 项试验符合纳入标准，其中 16 项比较结果用于有史以来最大的荟萃分析。所有研究均采用了适当的随机序列生成和意向治疗分析（ITT），并使用自我报告的指标作为主要结局。对于 16 项比较（2807 名参与者）CCBT 与对照组的研究，汇总 SMD 为-0.48[95%CI-0.63 至-0.33]，表明与过去的综述相似。此外，在长期随访时没有发现明显的临床效果，也没有发现功能改善。此外，CCBT 的脱落率明显高于对照组。当纳入没有 BDI 作为评分量表且仅采用现代插补作为敏感性分析的研究时，尽管从中度效应降为小效应，但汇总 SMD 仍然具有统计学意义。漏斗图和两项检验（Begg 检验 p = 0.09；Egger 检验 p = 0.01）显示存在显著的发表偏倚。采用修剪和填补分析，SMD 为-0.32[95%CI-0.49 至-0.16]。
尽管在治疗后短期降低了抑郁程度，但在长期随访时的效果和功能改善不显著，脱落率明显较高。考虑到偏倚风险，我们的荟萃分析表明，当前 CCBT 治疗成人抑郁症的临床效果可能需要在实际实施和方法学有效性方面进行向下修正。
