Meta-analysis of symptomatic response attributable to the pacing component of cardiac resynchronization therapy

Prognostic benefit from CRT compared with controls is well established. Symptomatic response rates, however, are controversial and have never been systematically evaluated with standard subtraction of control rates to establish the incremental symptomatic response effect of CRT pacing. First, we identified 150 consecutive CRT papers and assessed researchers perceptions of the symptomatic response to CRT. The mean quoted response rate was 66. Only 26 studies acknowledged the existence of response without the device. Secondly, we examined actual symptomatic response rates in the randomized trials (CARE-HF, COMPANION, CONTAK-CD, MIRACLE, MIRACLE-ICD, MIRACLE-ICD II, MUSTIC, and REVERSE) totalling 3904 patients. The NYHA status improved in 51 of those randomized to CRT vs. 35 of controls (incremental effect 16). This incremental improvement was significantly greater in open studies (with no device for controls) than in blinded studies (control arm receiving a device but no CRT, such as a defibrillator or a CRT programmed off), 20 vs. 13, P 0.001. Quoting CRT responder rates in isolation without recognizing spontaneous response is common but unwise. The incremental symptomatic response rate from CRT pacing is 16, much lower than widely reported. This value is similar to that for drugs in heart failure and should not be considered disappointing: they both exert powerful prognostic benefits. For scientific purposes, e.g. to explore potential improvements, symptomatic benefit from CRT should be quantified, like all other effects, by comparison with a control.