Noncancer-Related Mortality in Randomized Clinical Trials: A Meta-Analysis.

Importance: Cancer screening is a critical tool in cancer control, reducing cancer-specific mortality. However, it also has potential harms, including overdiagnosis and overtreatment. Measuring the effect of screening based on all-cause mortality is insensitive to both benefits and harms and requires substantially large sample sizes. Understanding the impact of screening on noncancer-related (off-target) mortality is essential for evaluating its overall benefit. Objective: To assess the association between cancer screening and off-target mortality by comparing mortality rates between screened and unscreened populations based on randomized clinical trials (RCTs). Data Sources: The analysis examined all RCTs included in a previously published (August 28, 2023) meta-analysis of cancer screening trials that included the end point of all-cause mortality in addition to targeted cancer mortality. Study Selection: All RCTs included in the previous meta-analysis were included. The latest search in that meta analysis was conducted on October 12, 2022, with no language or publication date restrictions. Data Extraction and Synthesis: The study followed relevant portions of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline. Two authors independently extracted data, and a third author verified those data. Off-target mortality was analyzed using rate ratios (RRs) and 95% CIs via a fixed-effects model. Heterogeneity was assessed using the I2 statistic. Main Outcome and Measures: The primary outcome was off-target mortality, defined as deaths with a cause that was not the targeted cancer. Results: A total of 17 RCTs (8 of colorectal, 3 prostate, 3 lung, 2 breast, and 1 multiple cancers) including 1 305 924 participants with 18 508 192 person-years of follow-up were included. Screening did not significantly increase off-target mortality (RR, 1.00; 95% CI, 1.00-1.01); the overall increase in off-target mortality was 0.2% (95% CI, -0.5% to 0.9%). There was no evidence of heterogeneity between trials (I2 = 0.00%; Cochran Q = 14.96, df = 18; P = .66). The trial-specific RRs ranged from 0.89 (95% CI, 0.69-1.15) to 1.09 (95% CI, 0.98-1.22), with all 95% CIs including 1. Targeted cancer deaths accounted for 2.6% to 33.1% of all deaths, depending on the cancer type. Conclusions and Relevance: These findings show that randomization to cancer screening was not associated with more than a very small increase in noncancer-related mortality, with the 95% CI excluding an increase of greater than 1%. The findings emphasize the importance of evaluating targeted and off-target mortality separately rather than relying solely on all-cause mortality.