Cryoablation versus RF Ablation for AVNRT
An electronic literature search of Medline and Embase, in addition to a hand-search of reference lists for published and unpublished data was carried out on August 23, 2012. Search terms included "AVNRT" (or variations such as "atrioventricular nodal reentrant tachycardia" or "junctional reciprocating tachycardia") and "cryoablation," (or variations such as "CryoCath," "cryothermal" or "Freezor"). Studies published or presented from January 1990 to August 2012 were considered. The search was limited to human studies and English language studies only.
Comparative studies (cohort and randomized controlled trials [RCTs]) of RF versus cryoablation for AVNRT were identified. Adults and children treated with catheters of all sizes (4, 6, and 8 mm) were included. Noncomparative studies (case reports or case series), clinical reviews, editorial articles, and studies involving different populations (AVRT, junctional tachycardia [JT], etc.) were excluded from analysis. The entire search strategy was repeated by an independent observer (NL) and interobserver variability was assessed by Cohen's kappa coefficient.
The primary metameter of interest was long-term AVNRT recurrence, which was defined by ECG/electrophysiology study (EPS)-documented recurrent AVNRT at least 2 months following the ablation procedure. Secondary metameters included acute procedural failure (defined as inducibility of AVNRT or multiple AV nodal echoes up to 30 minutes after ablation) and incidence of permanent AV block requiring insertion of a pacemaker. Additional analyses were also performed, comparing total procedure time and fluoroscopy time.
Relative risk ratios (RRs) and 95% confidence intervals (CIs) were calculated for dichotomous outcomes (failure rates and development of permanent AV block). If no events were observed in one of the study arms (AV block following cryoablation, for example), then a standard correction (adding a fixed value of 0.5 to all cells) was applied in order to calculate the RR. Weighted mean differences and 95% CIs were calculated using the t-test for continuous outcomes (total procedure time and fluoroscopy time). Analyses were performed for the same endpoint definitions as in the primary studies and any differences (such as definition of acute success and duration of follow-up when reporting long-term success) were recorded for consistency.
Because we expected studies to differ in their testing rigor at follow-up and in their duration of long-term follow-up, we applied a random-effects model for all outcomes. The I statistic was used to assess heterogeneity, with a 50% cutoff used to indicate substantial heterogeneity. In order to better visualize the overall distribution of outcomes and to look for publication bias among the studies of interest, a funnel plot was generated.
MIX 2.0 Professional (BiostatXL, Sunnyvale, CA, USA) was used to generate RRs, meta-regression, forest plots, and funnel plots.
Methods
Search Strategy and Data Sources
An electronic literature search of Medline and Embase, in addition to a hand-search of reference lists for published and unpublished data was carried out on August 23, 2012. Search terms included "AVNRT" (or variations such as "atrioventricular nodal reentrant tachycardia" or "junctional reciprocating tachycardia") and "cryoablation," (or variations such as "CryoCath," "cryothermal" or "Freezor"). Studies published or presented from January 1990 to August 2012 were considered. The search was limited to human studies and English language studies only.
Study Selection and Quality Assessment
Comparative studies (cohort and randomized controlled trials [RCTs]) of RF versus cryoablation for AVNRT were identified. Adults and children treated with catheters of all sizes (4, 6, and 8 mm) were included. Noncomparative studies (case reports or case series), clinical reviews, editorial articles, and studies involving different populations (AVRT, junctional tachycardia [JT], etc.) were excluded from analysis. The entire search strategy was repeated by an independent observer (NL) and interobserver variability was assessed by Cohen's kappa coefficient.
Data Synthesis and Statistical Analysis
The primary metameter of interest was long-term AVNRT recurrence, which was defined by ECG/electrophysiology study (EPS)-documented recurrent AVNRT at least 2 months following the ablation procedure. Secondary metameters included acute procedural failure (defined as inducibility of AVNRT or multiple AV nodal echoes up to 30 minutes after ablation) and incidence of permanent AV block requiring insertion of a pacemaker. Additional analyses were also performed, comparing total procedure time and fluoroscopy time.
Relative risk ratios (RRs) and 95% confidence intervals (CIs) were calculated for dichotomous outcomes (failure rates and development of permanent AV block). If no events were observed in one of the study arms (AV block following cryoablation, for example), then a standard correction (adding a fixed value of 0.5 to all cells) was applied in order to calculate the RR. Weighted mean differences and 95% CIs were calculated using the t-test for continuous outcomes (total procedure time and fluoroscopy time). Analyses were performed for the same endpoint definitions as in the primary studies and any differences (such as definition of acute success and duration of follow-up when reporting long-term success) were recorded for consistency.
Because we expected studies to differ in their testing rigor at follow-up and in their duration of long-term follow-up, we applied a random-effects model for all outcomes. The I statistic was used to assess heterogeneity, with a 50% cutoff used to indicate substantial heterogeneity. In order to better visualize the overall distribution of outcomes and to look for publication bias among the studies of interest, a funnel plot was generated.
MIX 2.0 Professional (BiostatXL, Sunnyvale, CA, USA) was used to generate RRs, meta-regression, forest plots, and funnel plots.