Empirical comparison of the results of randomized controlled trials and case-control studies in evaluating the effectiveness of screening mammography

TY - JOUR

T1 - Empirical comparison of the results of randomized controlled trials and case-control studies in evaluating the effectiveness of screening mammography

AU - Mills, Orlando F.

AU - Rhoads, George G.

N1 - Funding Information: This study was supported by Grant No. P20CAS7142 from the National Cancer Institute, Grant No. ES-05022 from the National Institute of Environmental and Health Sciences, and Grant No. 5-T32-PE10011-04 from the Health Resources and Services Administration.

PY - 1998/2

Y1 - 1998/2

N2 - The gold standard for evaluating screening programs is the randomized controlled trial (RCT). Case-control studies are easier to perform but their role in this area is controversial. The purpose of this article is to compare empirically the results of RCTs and case-control studies conducted to evaluate the efficacy and effectiveness of screening mammography and examine possible explanations for differences in their results. We located eight RCTs and five case-control studies of screening mammography. For women aged 40-74 years at screening, comparison of the summary risk estimates of the RCTs (0.76, 95% CI: 0.69-0.83) with that of the case-control studies (0.44, 95% CI: 0.38-0.50) showed RCTs to have a significantly higher summary risk estimate than case-control studies (ratio = 1.74, 95% CI: 1.48-2.04). It is notable that the RCTs were compromised in most instances by low compliance rates (50-80%) in the treatment groups and by significant use of screening in the control groups (20-30%). Adjustment of the RCT results for these cross-overs yields results that are in reasonable agreement with the summary estimate for the case-control studies. These findings support the use of case-control studies to estimate the efficacy of mammographic screening where RCTs are not feasible. They suggest that the efficacy of mammography in women aged 50 years and above is somewhat greater than the effectiveness measured by the intent-to-treat analysis of RCTs.

AB - The gold standard for evaluating screening programs is the randomized controlled trial (RCT). Case-control studies are easier to perform but their role in this area is controversial. The purpose of this article is to compare empirically the results of RCTs and case-control studies conducted to evaluate the efficacy and effectiveness of screening mammography and examine possible explanations for differences in their results. We located eight RCTs and five case-control studies of screening mammography. For women aged 40-74 years at screening, comparison of the summary risk estimates of the RCTs (0.76, 95% CI: 0.69-0.83) with that of the case-control studies (0.44, 95% CI: 0.38-0.50) showed RCTs to have a significantly higher summary risk estimate than case-control studies (ratio = 1.74, 95% CI: 1.48-2.04). It is notable that the RCTs were compromised in most instances by low compliance rates (50-80%) in the treatment groups and by significant use of screening in the control groups (20-30%). Adjustment of the RCT results for these cross-overs yields results that are in reasonable agreement with the summary estimate for the case-control studies. These findings support the use of case-control studies to estimate the efficacy of mammographic screening where RCTs are not feasible. They suggest that the efficacy of mammography in women aged 50 years and above is somewhat greater than the effectiveness measured by the intent-to-treat analysis of RCTs.

KW - Case-control design

KW - Effectiveness

KW - Efficacy

KW - Randomized controlled studies

KW - Screening

UR - http://www.scopus.com/inward/record.url?scp=0031909311&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031909311&partnerID=8YFLogxK

U2 - https://doi.org/10.1016/S0895-4356(97)00243-6

DO - https://doi.org/10.1016/S0895-4356(97)00243-6

M3 - Article

C2 - 9474068

SN - 0895-4356

VL - 51

SP - 81

EP - 91

JO - Journal of Clinical Epidemiology

JF - Journal of Clinical Epidemiology

IS - 2

ER -