Effectiveness of population‐based service screening with mammography for women ages 40 to 49 years

Abstract

estimated the mortalityreduction attributable tomammographic screening amongSwedish women aged 40-49 years to be 29% (95% confi-dence interval [CI], 20%-38%). The study is of high qual-ity, and is a landmark study on an important topic.However, we suggest another way of calculating the esti-matedmortalityreductionthatwouldreducetheestimatedscreeningeffectonbreastcancermortalitysomewhat.Hellquist et al. divided Swedish counties into thosethathaveorhavenotintroducedscreeningforwomen40-49yearsofage.Theyfirsttestedforanystatisticallysignif-icant prescreening differences in mortality between thelater screening and nonscreening counties, concludingthattherewerenodifferencesinmortality.Later,thepost-screening differences in mortality between screening andnonscreening counties were used as an estimate of thescreening-related mortality reduction, assuming that thegroups were equal with the exception of screening.Although this may seem like a reasonable approach, it isimportanttorememberthatanonsignificantPvaluedoesnot imply that there is no difference between 2 groups,only that it is not possible to separate a difference fromrandom variation. Prior to screening, the mortality was6%lowerinthescreeningcounties,andtheauthorsarguethat the regional differences may have become smallerover time. However, because we do not know for certainwhether this is the case, we propose an alternative effectmeasure. Wewould prefer a measureequivalenttothat ofrandomized trials, estimating the mortality ratio betweenthe screened and nonscreened groups taking into accountthe prescreening regional differences. For the given data,this yields an estimated mortality reduction of 0.71/0.94 0.76amongscreenedwomen.As for the statistical uncertainty, it is larger whentaking the prescreening mortality into account. Withoutaccess to the original data, we cannot calculate exact CIs,but we can make an approximation by back-calculatingfrom the given CIs. Using this approximation, the stand-arderroris(1.05-0.85)/(2*1.96) 0.05fortheprescreen-ing mortality ratio and (0.80-0.62)/(2*1.96) 0.05 forthe postscreening mortality ratio. After performing 1 mil-lionbias-correctedbootstrapreplicationsusingtheRsoft-ware package,