Influence of delay on survival in patients with breast cancer: a systematic review

Background Most patients with breast cancer are detected after symptoms occ ur rather than through screening. The impact on survival of delays between the onset of symptoms and the start of treatment is controversial and canno t be studied in randomised controlled trials. We did a systematic review of observational studies (worldwide) of duration of symptoms and survival. Methods We identified 87 studies (101 954 patients) with direct data linkin g delay (including delay by patients) and survival. We classified studies f or analysis by type of data in the original reports: category I studies had actual 5-year survival data (38 studies. 53 912 patients); category II use d actuarial or multivariate analyses (21 studies, 25 102 patients); and cat egory III was all other types of data (28 studies, 22 940 patients). We tes ted the main hypothesis that longer delays would be associated with lower s urvival, and a secondary hypothesis that longer delays were associated with more advanced stage, which would account for lower survival. Findings In category I studies, patients with delays of 3 months or more ba d 12% lower 5-year survival than those with shorter delays (odds ratio for death 1.47 [95% CI 1.42-1.53]) and those with delays of 3-6 months had 7% l ower survival than those with shorter delays (1.24 [1.17-1.30]). In categor y II, 13 of 14 studies with unrestricted samples showed a significant adver se relation between longer delays and survival, whereas four of five studie s of only patients with operable disease showed no significant relation. In category ill, all three studies with unrestricted samples supported the pr imary hypothesis. The 13 informative studies showed that longer delays were associated with more advanced stage. In studies that controlled for stage, longer delay was not associated with shorter survival when the effect of s tage on survival was taken into account. Interpretation Delays of 3-6 months are associated with lower survival. The se effects cannot be accounted for by lead-time bias. Efforts should be mad e to keep delays by patients and providers to a minimum.