Cancer screening: A modest proposal for prevention

Article Figures & Data

TABLE 1

Benefit of cancer screening

Cancer type and method of screening	Per 100,000 person-years			Harms of interventions
Cancer type and method of screening	Deaths from cancer	Deaths averted by screening	Deaths not averted by screening	Harms of interventions
Breast cancer^a
Mammography every 2 years				False-positive results Overdiagnosis, overtreatment Radiation exposure
Age 39–49	34	4	30 (88%)
Age 50–59	54	8	46 (85%)
Age 60–69	65	21	44 (68%)
Age 70–74	62	13	49 (79%)
Age 50–69	58	13	45 (78%)
Colon cancer^b
One-time flexible sigmoidoscopy	44	12	32 (73%)	Complications of procedure Overdiagnosis, overtreatment
Guaiac fecal occult blood test once a year	44	4	40 (91%)	False-positive results Overdiagnosis, overtreatment
Colonoscopy every 10 years⁸	69	9	60 (87%)	Perforation of the colon
Prostate cancer^c
Prostate-specific antigen test once a year	48	10	38 (79%)	False-positive results Overdiagnosis, overtreatment
Lung cancer^d
Low-dose computed tomography once a year	309	59	250 (81%)	False-positive results Overdiagnosis, overtreatment

↵a Breast cancer numbers are based on risk ratio in Table 5 of reference 2.
↵b Effects of sigmoidoscopy are based on 4 large randomized trials with 11- to 12-year follow-up. Colon cancer mortality without screening is calculated as the average mortality of the control groups in the 4 randomized trials (Table 1 of reference 7). Incident rate ratio of 0.73 was used to calculate colon cancer deaths avoided with screening with sigmoidoscopy. Effects of guaiac fecal occult blood testing are based on 5 randomized trials with 11- to 30-year follow-up (also Table of reference 7). Relative risk of 0.91 was used to calculate effects of screening. The US Preventive Services Task Force (USPSTF) estimated the relative risk of 0.91 after 19.5 years of screening and 0.78 after 30 years of screening. Colonoscopy effects are based on Table 3 and Table 4 of referenced modeling study representing 40-year follow-up of 1,000 men and women.⁸ The modeling study is designed to compare different screening modalities; all participants are screened regardless of life expectancy, and calculations are based on 100% adherence to screening. Therefore, numbers cannot be directly compared with other cancer screening programs.
↵c Prostate cancer numbers are based on the table in reference 3 on 13-year follow-up of 1,000 men invited to screening.
↵d Lung cancer numbers are based on National Lung Screening Trial of lung cancer mortality and relative risk of 0.81 from USPSTF meta-analysis.