Table 3.
Summary of evidence from 21 modelling studies quantifying overdiagnosis from cancer screening.
| Study; model(s) | Modelled population: country, ages; screening schedule | Data sources: (a) Incidence; (b) Mortality; (c) Other | (a) External validation?; (b) Includes competing mortality?; (c) Includes DCIS? | Reports outcome as % of screen detected cancers? | Magnitude of overdiagnosis (95% CI) | Sensitivity analyses varying mean sojourn time or lead time? | Overall risk of bias |
|---|---|---|---|---|---|---|---|
| Prostate cancer (n=10) | |||||||
| Davidov 200411 | US; 50–60, 70, or 80 year olds; at 5 year intervals | (a) SEER 1993–97; (b) SSA life tables | (a) No; (b) Yes; (c) N/A | Unclear | 8.48–53.6% | Univariate. MST 5–15 years Overdiagnosis varied greatly with MST |
Moderate |
| Draisma 200912; MISCAN, FHCRC, UMichigan | US, 54–80 year olds; typical US screening patterns | (a) SEER 1985–2000; (b) Standard life tables | (a) No; (b) Yes; (c) N/A | Yes | MISCAN 42%; FHCRC 28%; UMich 23% | Not performed | High |
| Gulati 201313; FHCRC | US, 40 year olds; 32 screening schedules simulated | (a) SEER 1975–2000; (b) US life tables | (a) No; (b) Yes; (c) N/A | No, reports lifetime risk of overdiagnosis | 1.8–6% | Other sensitivity analyses performed. | Moderate |
| Gulati 201414 | US, 50–84 year olds; multiple | (a) SEER 1975-2005; (b) US life tables | (a) No; (b) Yes; (c) N/A | No | 2.9–88.1% depending on age, Gleason score, and PSA level (% likelihood that a tumor is overdiagnosed) | Not performed | High |
| Heijnsdijk 200915; MISCAN | Europe; 55–70 year olds every 1 or 2 years, or 55–75 year olds every 4 years | (a, b) ERSPC Rotterdam; (c) Cure rates by stage from Amsterdam Cancer Center | (a) No; (b) Yes; (c) N/A | Yes (estimated from figures) | Annual, 60%; biennial, 60%; every 4 years (to age 75), 67% | Not performed | High |
| McGregor 199816 | Quebec, 50–85 year olds; annual PSA test for ages 50–70 | (a) Multiple; (b) Quebec Ministry of Health | (a) No; (b) Yes; (c) N/A | Yes | 84% | Other sensitivity analyses performed. | High |
| Pashayan 200917 | UK; single PSA | (a) Eastern Cancer Registry, ProtecT study, UK Office of National Statistics; (b) UK Office of National Statistics | (a) No; (b) Yes; (c) N/A | Yes | 50–54 years, 10% (7 to 11%); 55–59, 15% (12 to 15); 60–64, 23% (20 to 24); 65–69, 31% (26 to 32) | Not performed | High |
| Telesca 200818 | US; typical US screening patterns | (a) SEER 1973–87; (b) CDC Vital Statistics 1992 | (a) No; (b) Yes; (c) N/A | Yes | White men 22.7%; black men 34.4% | Not performed | High |
| Tsodikov 200619 | US; typical US screening patterns | (a) SEER; (b) Human Mortality Database | (a) No; (b) Yes; (c) N/A | Yes | 30% | Not performed | High |
| Wu 201220 | Finland; 55, 59, 63, 67 year olds; 3 PSA tests every 4 years until age 71 | (a) Finnish Prostate Cancer Screening Trial, Finnish Cancer Registry; (b) Statistics Finland | (a) No; (b) Yes; (c) N/A | No | 3.4% (2.4 to 5.7%) risk of overdetection during study period | Not performed | High |
| Breast cancer (n=7) | |||||||
| De Gelder 2011 (Epi Rev)21; MISCAN | Netherlands; 0–100 year olds; biennial mammogram ages 49–74 | (a) Dutch Comprehensive Cancer Centers, National Evaluation Team for Breast Cancer Screening 1990–2006 | (a) No; (b) Yes; (c) Yes | Yes | Implementation 22.1–67.4%; extension 15.4–30.5%; steady state 8.9–15.2% | Not performed | High |
| De Gelder 2011 (Prev Med)22; MISCAN | Netherlands, 0–100 year olds; biennial film or digital mammogram | (a) Dutch Cancer Registry, National Evaluation Team for Breast Cancer Screening 1990-2006 | (a) No; (b) Yes; (c) Yes | Yes | Screen film, 7.2%; digital, 8.2% | Other sensitivity analyses performed. | High |
| Duffy 200523 | Sweden; 40–74 or 39–59 year olds; mammogram every 18, 24, or 33 months | All data: Swedish 2-County RCT (1977–84) and Gothenburg RCT (1982–87) (separate analyses) | (a) No; (b) No; (c) Yes | Yes | Swedish: 1st screen, 3.1% (0.1 to 10.9%); 2nd, 0.3% (0.1 to 1); 3rd, 0.3% (0.1 to 1) Gothenburg: 1st screen, 4.2% (0.0 to 28.8), 2nd, 0.3% (0.0 to 2.0), 3rd, 0.3% (0.0 to 2.0) |
Not performed | High |
| Gunsoy 201224 | UK, 40–49 year olds; annual mammogram | (a) England/Wales Office of National Statistics, Age RCT Control Arm; (b) Office of National Statistics; (c) parameter estimation model: age RCT | (a) No; (b) Yes; (c) Yes | Yes | 0.70% | Univariate; varied MST and sensitivity; 0.5 to 2.9% | Moderate |
| Martinez-Alonso 201025 | Spain; 25-84 year olds; biennial mammogram ages 50–69 | (a) Girona Cancer Registry and IARC Registry | (a) No; (b) Yes; (c) No | No, reported as % excess of expected incidence | 1935 birth cohort, 0.4% (−8.8 to 12.2%); 1940, 23.3% (9.1 to 43.4); 1945, 30.6% (12.7 to 57.6); 1950, 46.6% (22.7 to 85.2) | Univariate; varied MST from 1 to 5. 18.3 to 51.1% |
Moderate |
| Olsen 200626 | Denmark; 50-69 year olds; biennial mammogram | (a) Danish Cancer Registry, Breast Cancer Cooperative Group, Central Population Registry | (a) No; (b) No; (c) Yes | Yes | 1st screen, 7.8% (0.3 to 27.5%); 2nd screen, 0.5% (0.01 to 2.2) | Other sensitivity analyses performed | High |
| Seigneurin 201227 | France, 50-69 year olds; not specified | (a) French population-based study by Seigneurin 2009 | (a) No; (b) No; (c) Yes | Yes | DCIS, 31.9% (2.9 to 62.3%); invasive cancer, 3.3% (0.7 to 6.5) | Univariate; varied MST; DCIS, 17.3 to 51.7%; invasive cancer, 0 to 8.9% | Moderate |
| Lung cancer (n=3) | |||||||
| Duffy 201428 | UK; 55–74 or 50–75 year olds; annual and biennial | (a) NLST, UKLS; (b) NLST, SEER | (a) No; (b) Unclear; (c) N/A | Yes | 11% | Univariate; varying MST; 0 to 18% | Moderate |
| Hazelton 201229 | US; Heavy smokers, <5 years asbestos exposure; low dose CT | (a, b) CARET (calibration); (c) calibrated model applied to NYU Biomarker and Moffitt Cancer Center Trials | (a) No; (b) Yes; (c) N/A | Yes | Men 14.1% (11.6 to 19.7%); women 35.2% (28.9 to 39.3) | Not performed | High |
| Pinsky 200430 | US; men aged 50–75 years, heavy smokers; annual CXR and sputum cytology | All data: Mayo Lung Screening Trial (prevalence screen and screening arm only) | (a) No; (b) Yes; (c) N/A | Yes | 13–17% | Not performed | High |
| Colon cancer (n=1) | |||||||
| Luo 201231 | US; 40, 50, or 60 year olds; 5 annual or 3 biennial FOBT | (a) Minnesota Colon Cancer Control study (1976–82); (b) SSA life tables | (a) No; (b) Yes; (c) N/A | Yes (reported for age 50) | Women 6.65% (2.56 to 20.49%); men 6.15% (1.92 to 44.69) | Not performed | High |
SEER=Surveillance, Epidemiology, and End Results database; SSA=Social Security Administration; MST=mean sojourn time; MISCAN= Microsimulation Screening Analysis model; FHCRC=Fred Hutchinson Cancer Research Center; IARC=International Registry for Research on Cancer; DCIS=ductal carcinoma in situ; NLST=National Lung Screening Trial; UKLS= UK Lung Screening pilot trial; CARET=Carotene and Retinol Efficacy Trial; CT=computed tomography; CXR=chest x ray; FOBT=Fecal Occult Blood Test.