Skip to main content

Table 2 Characteristics of the included studies for estimating the discrimination of the Gail model

From: Assessment of performance of the Gail model for predicting breast cancer risk: a systematic review and meta-analysis with trial sequential analysis

Reference Author Publication year Geographic background Study design Gail model version 5/10-year risk Sample size Mean age (years) Study population Risk for breast cancera Time period Follow-up period C-statistic/AUC (95% CI)
[11] Rockhill 2001 America Cohort 2 5 82,109 45–71 Nurses’ Health Study (NHS) General population 1992–1997 5.0 0.58 (0.56–0.60)
[30] Amir 2003 United Kingdom Cohort 2 10 3150 21–73 Women attending the Family History Screening Programme in University Hospital of South Manchester Not defined 1987–2001 5.0 0.74 (0.67–0.80)
[21] Tice 2005 America Cohort 2 5 81,777 55.9 Community-based registry San Francisco Mammography Registry (SFMR) General population 1993–2002 5.1 (0.1–15) 0.67 (0.65–0.68)
[29] Decarli 2006 Italy Cohort 2 5 10,031 35–64 Florence—European Prospective Investigation Into Cancer and Nutrition cohort (EPIC) General population 1993–2002 9.0 0.59 (0.55–0.63)
[36] Crispo 2008 Italy Case–control 1 5 1765 53.7 National Cancer Institute of Naples (southern Italy) NA 1997–2000 NA 0.55 (0.53–0.58)
[15] Tice 2008 America Cohort 2 5 251,789 40–74 National Cancer Institute-funded Breast Cancer Surveillance Consortium (BCSC) General population since 1994 5.3 0.61 (0.60–0.62)
[42] Pan 2009 China Cross-sectional 1 5 2133 > 35 Breast cancer risk assessment, evaluation and health education program, in Beijing and Guangzhou community NA 2006–2007 NA 0.64 (0.61–0.67)
[43] Liu 2010 China Cross-sectional 2 5 246 49.82 High-risk breast cancer screening model and chemical intervention study at the community level NA 2007–2009 NA 0.56 (0.49–0.64)
[44] Wang 2010 China Case–control 1 5 228 32–75 Shenzhou Hospital of Shenyang Medical College-based breast cancer cases and control NA 1998–2007 NA 0.93 (0.89–0.97)
[17] Tarabishy 2011 America Cohort 2 5 4726 18–85 Mayo Benign Breast Disease (BBD) High risk 1982–1991 16.2 0.64 (0.62–0.66)
[22] Vacek 2011 America Cohort 1 5 19,779 > 70 Vermont Breast Cancer Surveillance System (VBCSS) High risk 2001–2009 7.1 0.54 (0.52–0.56)
[27] Banegas 2012 America Cohort 2 5 128,976 63.51 Women’s Health Initiative (WHI) General population 1993–2005 5.0 0.58 (0.57–0.59)
[23] Quante 2012 America Cohort 2 10 1857 44 Women with high risk for breast or ovarian cancer in New York site of the Breast Cancer Family Registry (BCFR) High risk 1995–2011 8.1 0.63 (0.58–0.69)
[32] Pastor-Barriuso 2013 Spain Cohort 2 5 54,649 45–68 Population-based Navarre Breast Cancer Screening Program (NBCSP) General population 1996–2005 7.7 0.54 (0.52–0.57)
[51] Dite 2013 Australia Case–control 2 5 1425 45.4 Cases and controls from the Australian Breast Cancer Family Registry (ABCFR) NA 1992–1998 NA 0.58 (0.55–0.61)
[40] Anothaisintawee 2013 Thailand Cross-sectional NA NA 15,718 NA Ramathibodi Hospital and two tertiary hospitals NA 2011–2013 NA 0.41 (0.36–0.46)
[24] Ronser 2013 America Cohort 2 5 11,419 54.0 ± 3.3 Postmenopausal women in California Teachers Study (CTS) High risk 1995–2009 5.0 0.55 (0.53–0.56)
[41] Min-1 2014 Korea Cohort 2 5 40,229 > 10 Breast cancer screening patients routinely screened in Women’s Healthcare Center of Cheil General Hospital Not defined 1999–2004 5 0.55 (0.50–0.59)
[41] Min-2 2014 Korea Cohort 3 5 40,229 > 10 Breast cancer screening patients routinely screened in Women’s Healthcare Center of Cheil General Hospital Not defined 1999–2004 5 0.54 (0.50–0.59)
[18] Powell 2014 America Cohort 2 5 12,843 NA Marin Women’s Study with high rate of breast cancer, null parity and delayed childbirth High risk 2003–2007 7.7 0.62 (0.59–0.66)
[45] Duan 2014 China Case–control 2 5 400 35–74 Breast cancer cases and controls in the First Affiliate Hospital of KunMing Medical University NA 2007–2011 NA 0.54 (0.49–0.60)
[19] McCarthy 2015 America Cohort 2 5 464 48.7 ± 13 Women referred for biopsy with abnormal (Breast Imaging Reporting And Data System, BI-RADS 4) mammograms at the Hospital of the University of Pennsylvania High risk 2003–2012 5.0 0.71 (0.65–0.78)
[34] Dartois-1 2015 France Cohort 2 5 5843 42–72 Premenopausal Women in French E3N (E´ tude E´ pide´miologique aupre`s des femmes de laMutuelle Ge´ne´rale de l’E´ ducation Nationale (MGEN)) prospective cohort to investigate the cancer risk factors General population 1993–1998 5.0 0.61 (0.55–0.68)
[34] Dartois-2 2015 France Cohort 2 5 7331 42–72 Postmenopausal Women in French E3N (E´ tude E´ pide´miologique aupre`s des femmes de laMutuelle Ge´ne´rale de l’E´ ducation Nationale (MGEN)) prospective cohort to investigate the cancer risk factors High risk 1993–1998 14.0 0.55 (0.50–0.60)
[39] Hu 2015 China Cohort 2 5 42,908 35–69 Women participated in the breast cancer screening in Zhejiang eastern coastal areas of China General population 2008–2014 5.0 0.59 (0.47–0.70)
[35] Brentnall 2015 United Kingdom Cohort 2 10 50,628 47–73 15 screening areas in Greater Manchester, UK General population 2009–2014 3.2 0.54 (0.52–0.56)
[20] Schonberg-1 2015 America Cohort 2 5 71,293 70.0 ± 7.0 Nurses’ Health Study (NHS) High risk 2004–2009 5.0 0.57 (0.55–0.58)
[20] Schonberg-2 2015 America Cohort 2 5 79,611 71.0 ± 6.8 Women’s Health Initiative (WHI), extensive study High risk 2005–2010 5.0 0.58 (0.56–0.59)
[46] Rong 2016 China Case–control 1 5 816 48.9 Breast cancer cases and controls in the Shenzhen Maternal and Child Health Care hospital NA 2011–2013 NA 0.69 (0.68–0.71)
  1. Note: Gail model type 1, original Gail model; Gail model type 2, modified Gail model for Caucasian-American; Gail model type 3, modified Gail model for Asian-American
  2. AUC area under the area under the curve, CI confidence interval, NA not available
  3. aCohort studies enrolled women with high risk for breast cancer (with higher average age (> 70 years), abnormal breast density, postmenopausal state, breast cancer relatives or high rate of delayed childbirth) were defined as “High risk”; cohort studies that did not accurately depict the characteristics of the participants were defined as “Not defined”. Case–control studies and cross-sectional studies were defined as not available