Abstract
Medical screening is often described in terms of its effectiveness among persons in a ‘target’ group or among persons ‘eligible’ for screening. Such selection is appropriate in trials to determine whether screening is worthwhile; for example, to know whether mammographic breast cancer screening in women aged 50–69 years old reduces breast cancer mortality. However, such selection is not appropriate in assessing the overall impact of breast cancer screening. The selection of females aged 50–69 is an initial screening step which can have no effect on breast cancer mortality in women younger than 50 or older than 69. The estimates of reduction in breast cancer cases and deaths will be exaggerated by ignoring those ‘ineligible’ for screening or those outside a ‘target’ group. This overestimation can be overcome by adopting a step-wise analysis in a flow‐chart of screening in which one starts with a given number of people in the whole population, with that number being of a given sex when the disorder is exclusive (or nearly exclusive) to one sex, such as screening for prostate cancer that affects only men or antenatal screening for neural tube defects that affects only pregnant women.
Constructing the flow‐chart of a screening protocol is often helpful. A critical element is to divide the population being screened into ‘affected’ and ‘unaffected’ and then specify the effect on true positives (detection) by following the ‘affected’ line and the effect on false positives by following the ‘unaffected’ line. This can, at first sight, seem counter-intuitive because the application of screening in practice divides a population into ‘screen-positive’ and ‘screen-negative’. This, however, mixes affecteds and unaffecteds within each group which obscures the determination of detection and false-positive rates at each stage of the screening and diagnostic process.
Figure 1 is a hypothetical example of 1000 people in the population. Using a target approach in which screening is viewed as being limited to a target age of 60 or more, the number of cases detected is 20 out of 30, that is, 67% (20/30). Using a step-wise approach with age being the first step, there are still 20 cases detected but the percentage is less, that is, 40% (20/50). The proportion of cases prevented is 10/50 (20%) and 10/30 (33%), respectively.
A flow‐chart of a hypothetical screening protocol.
The step-wise approach makes age selection an explicit first step in the screening protocol. A first screening step generally needs to have a high detection rate because cases missed at this step cannot be recovered with subsequent screening tests, but it can have a high false-positive rate because this can reclass false positives as negative with subsequent screening tests. In the process the odds of being affected increase at each step. Figure 1 assumes no ‘over-diagnosis’ but this could be included in the flow‐chart; the 20:0 could be, say, 20:2 or 10:1. The step-wise flow-charting approach makes explicit what screening does and does not achieve. It is an approach that merits consideration.
Footnotes
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
