Swedish Two-County Trial

Published

Tabar L, et al. "Swedish two-county trial: impact of mammographic screening on breast cancer mortality during 3 decades". Radiology. 2011. 260(3):658-663.
PubMed • Full text • PDF

Clinical Question

What is the the long-term (29-year) effect of mammographic screening for women age 40-74 on breast cancer mortality?

Bottom Line

Mammography screening of women starting at 40 results in a highly significant decrease in breast cancer–specific mortality.

Major Points

The Swedish Two-County Trial (1985) was one of the first large trials to describe the use of mammograms to screen for breast cancer. This trial began in the late 1970s and enrolled ~160,000 women in Kopparberg and Ostergötland counties in Sweden. Women aged 40-74 years were randomized to invitation for mammogram q2-3 years (based on age) or usual care. The first published results to the end of 1984 found mammograms to be associated with a 31% relative reduction in breast cancer mortality and 25% relative reduction stage II+ breast cancers.

Since the initial published results, regular updates of the trial have shown that the effect of screening on breast cancer mortality has remained stable over longer follow-up. In addition, the absolute benefit in terms of lives saves increased over time. The present 2011 publication updated this RCT with 29 year follow-up. Case status and cause of death (a) a local trial committee and also (b) an independent Swedish overview "consensus" committee. The use of both case definitions found mammograms to be associated with significant reduction in cancer mortality, with 27-31% relative risk reduction. The number of women needed to undergo screening for 7 years to prevent 1 death at 29 years was 414-519.

The use of mammogram for breast cancer screening has been a matter of controversy in recent years. An impactful 2013 Cochrane review noted numerous limitations of the Swedish Two-County Trial (see Criticisms section below). Because of these concerns, the Cochrane review deemed this trial as "likely to be biased" in part because of poor randomization techniques. Among the 4 trials with suboptimal randomization, there was an overall 25% reduction in breast cancer mortality (RR 0.75; 95% CI 0.67-0.83), whereas 3 trials with adequate randomization did not show reduction in this endpoint (RR 0.90; 95% CI 0.79-1.02). Combined, these 7 trials found a 19% reduction in breast cancer mortality (RR 0.81; 95% CI 0.74-0.87). The Cochrane review authors concluded that mammogram likely contributes to overdiagnosis/overtreatment of breast cancer, adverse psychological impact in about 10% undergoing screening (e.g., anxiety), and possibly less benefit in breast cancer mortality than was estimated in early mammogram screening trials.

Guidelines

American Cancer Society breast cancer screening among those at average risk (2015, adapted)

• Age 40-44 - Should have opportunity to begin mammograms (qualified recommendation)
• Beginning at age 45 and as long as overall health is good/at least 10 year life expectancy - regular screenings (strong recommendation)
  • Age 45-54 - Annual screening (qualified recommendation)
  • Age ≥55 - Biennial screening or option to continue annually (qualified recommendation)

USPSTF breast cancer screening (2016, adapted)

• Age 40-49 years - Individualize whether to do q2y screening (grade C)
• Age 50-74 years - Recommend q2y screening (grade B)

American College of Gynecology (2017, adapted)

• Age 40-49 - Offer screening mammogram (level A)
• Age 50-75 - Screening mammogram should begin no later than age 50, and should continue until at least 75 years (level A)
• Screening frequency q1-2y (level A)

Design

• Two center, randomized, controlled trial
• N=133,065
  • Invited for screening (n=77,080)
  • Control (n=55,985)
• Setting: 2 centers in Sweden
• Enrollment: From 1977 in Dalarna for 31 months, and 1978 in Östergötland for 34 months.
• Mean follow-up: 29 years
• Analysis: Intention-to-treat
• Primary outcome: mortality from breast cancer

Population

Inclusion Criteria

• Women aged 40-74
• Residing in one of two counties in Sweden: Dalarna and Östergötland

Exclusion Criteria

• Prior breast cancer

Interventions

• Female population from the two counties were divided into 45 geographic clusters. The population in these clusters were randomized to invited (mammography screening for breast cancer) or control (no mammography screening) in 1:1 ratio in Östergötland and 2:1 in Dalarn.
  • Invited - Women 40–49 years were invited to screening q24 months; women 50–74 were invited to screening q33 months. The screening method was one-view screen-film mammography with a single reading—without physical examination.
  • Control - Received usual care
• The screening phase was ~7 years.
• Case adjudication occurred with (a) a local trial end point committee and (b) a Swedish overview committee.
• This extended analysis included data from ~1977-1978 through ~2005-2006.

Outcomes

Comparisons are invited for screening and control group

Primary Outcomes

Breast cancer deaths

Local Trial End Point Committee Data: 351/77,080 vs. 367/55,985 (RR 0.69; 95% CI 056-0.85)

Swedish Overview Committee Consensus Data: 339/77,080 vs. 339/55,985 (RR 0.73; 95% CI 0.59-0.89)

The authors estimate this as 34-42 years of life per 1,000 women screened over 7 years

Number needed to screen over 7 years to prevent 1 cancer death at 29y was 414-519

Secondary Outcomes

Breast cancer

Local Trial End Point Committee Data: 1,439/77,080 vs. 1,049/55,985

Swedish Overview Committee Consensus Data: 1,426/77,080 vs. 1,042/55,985

Additional Outcomes

Had screening, among invited group

85%

Criticisms

• Trial used only single-view, single-observer screenings, with longer intervals than in other trials.
• Inconsistencies between the assessment of cause of death in the trial and the official Swedish cause of death register data.
• The original descriptions of randomization are inadequate, and the randomized groups might not have been comparable.
• Regional differences in the proportion diagnosed with breast cancer prior to trial initiation.
• It is possible that women as young as 38 were enrolled.
• The number of women who were randomized varies from publication-to-publication.
• Unclear when the control group underwent screening, it's possible that it might have occurred 5-8 years after enrollment.
• Ascertainment of cause of death was not blinded.

Funding

• Original trial funded by local county councils in Sweden
• This extended analysis was funded by the American Cancer Society through a gift from the Longaberger Company’s Horizon of Hope Campaign