A meta-analysis of six randomised trials demonstrated that intensive followup in colorectal cancer was associated with an absolute reduction in all-cause 5-year mortality of 10% (95% confidence interval (CI): 4 -16) -however, only two percent (95% CI: 0 -5) was attributable to cure from salvage re-operations. We postulate that other factors, such as increased psychological well-being and/or altered lifestyle, and/or improved treatment of coincidental disease may contribute to the remaining lives saved, and form important future research questions. Approximately two-thirds of patients presenting with colorectal cancer undergo resection with curative intent, and most subsequently enter protocols for long-term followup (Kievit, 2002). The rationale for surveillance is three-fold: psychological support, facilitation of audit, and an opportunity for the early detection and treatment of recurrent disease, with potential improvement in survival. Recently, the authors (Renehan et al, 2002b) reported a meta-analysis of five randomised trials and demonstrated a significant improvement in all-cause 5-year mortality in patients followed intensively. A Cochrane review (Jeffery et al, 2002) independently found similar results, and, subsequently, a sixth randomised trial reported additional results supporting these conclusions (Secco et al, 2002). These data offer the first direct evidence that intensive followup improves survival, but fall short of evaluating the mechanisms underlying the observed survival benefit. This study updates and extends our previous metaanalysis to explore the survival mechanisms associated with intensive followup.
MATERIALS AND METHODSThe search strategy, inclusion and exclusion criteria, data extraction, and study quality assessment have been published elsewhere (Renehan et al, 2002b(Renehan et al, , 2004, with further details at www.christie.man.ac.uk/profinfo/departments/surgery/default. htm. The key features were:Updated search strategy (to December 2003) using Cochrane methodology.Inclusion criteria were: randomised controlled trial; patients with colorectal cancer treated surgically with curative intent; randomisation at or shortly after surgery, and availability of 5-year survival data. Data were extracted independently by two investigators (AGR, MPS). Important components of methodological quality, namely adequacy of concealment of patients' allocation to treatment groups, double-blinding, and withdrawals, were assessed.
Statistical analysisMeta-analyses were performed at two levels. First, comparisons of events (e.g. all-cause deaths) for intensive vs conventional followup were performed and pooled estimates expressed as risk ratios (RR) and 95% confidence intervals (95% CIs). Second, estimates of the proportion of overall lives gained, and lives gained through to salvage re-operation, were calculated using absolute risk differences (and 95% CIs) (Deeks and Altman, 2001). The difference between these estimates was taken as an estimate of the gain in lives attributable to nonsalvage-related ...