Cross-national comparison of screening mammography accuracy measures in U.S., Norway, and Spain

Abstract

Objective

To compare accuracy measures for mammographic screening in Norway, Spain, and the US.

Methods

Information from women aged 50–69 years who underwent mammographic screening 1996–2009 in the US (898,418 women), Norway (527,464), and Spain (517,317) was included. Screen-detected cancer, interval cancer, and the false-positive rates, sensitivity, specificity, positive predictive value (PPV) for recalls (PPV-1), PPV for biopsies (PPV-2), 1/PPV-1 and 1/PPV-2 were computed for each country. Analyses were stratified by age, screening history, time since last screening, calendar year, and mammography modality.

Results

The rate of screen-detected cancers was 4.5, 5.5, and 4.0 per 1000 screening exams in the US, Norway, and Spain respectively. The highest sensitivity and lowest specificity were reported in the US (83.1 % and 91.3 %, respectively), followed by Spain (79.0 % and 96.2 %) and Norway (75.5 % and 97.1 %). In Norway, Spain and the US, PPV-1 was 16.4 %, 9.8 %, and 4.9 %, and PPV-2 was 39.4 %, 38.9 %, and 25.9 %, respectively. The number of women needed to recall to detect one cancer was 20.3, 6.1, and 10.2 in the US, Norway, and Spain, respectively.

Conclusions

Differences were found across countries, suggesting that opportunistic screening may translate into higher sensitivity at the cost of lower specificity and PPV.

Key Points

• Positive predictive value is higher in population-based screening programmes in Spain and Norway.

• Opportunistic mammography screening in the US has lower positive predictive value.

• Screening settings in the US translate into higher sensitivity and lower specificity.

• The clinical burden may be higher for women screened opportunistically.