''Might be something, might not'' - what happens when indeterminate grade screening mamograms are arbitrated?

Introduction

Arbitration of screening mammograms following independent double reading with or without consensus has been shown to reduce recall rates and maintain cancer detection rates. This study examined the effect of arbitration of indeterminate (score 3) mammograms only following consensus decision to recall.

Methods

This was a retrospective audit. Data were gathered from two sets of women undergoing analogue screening at a large unit in Northwest England, before and after introduction of arbitration.

Data were collected from film reading records and the National Breast Screening Service. The results were analysed by the authors. Chi squared tests were used to examine for any statistical significance in differences between the two data sets.

Results

Overall recall rate was significantly reduced after introducing arbitration (2.5% (142 of 5,639) versus 4.9% (313 of 6,397), P < 0.0001). The recall rate was lower even without arbitration (3.4%). However, arbitration of recalled score 3 mammograms reduced the recall rate by a further 27%. There was no statistically significant difference in the proportion of films graded 2, 3, 4 or 5 individually or overall after arbitration was introduced. The Positive Predictive Value (PPV) at assessment for recall of mammograms scored 3 by both readers was 26% (5 of 19) compared to 1% (1 of 97) where scores were discordant.

Conclusion

Arbitration of indeterminate grade mammograms can further reduce recall rates following consensus and does not influence reader scores. However, further study should be done to determine the positive predictive value of concordant score 3 mammograms prior to arbitration.