Quality assessment of the mammographic screening programme in the Azienda Sanitaria locale Provincia Milano 1 -- analysis of interval cancers and discussion of possible causes of diagnostic error.

PURPOSE

To evaluate the sensitivity of the mammographic screening programme and the causes of diagnostic fault in cases surfacing as interval cancers.

MATERIALS AND METHODS

Interval cancers (CI) were identified by linkage of the screening database for the years 2000-2002 to the database of hospital discharge records (HDR) for breast cancer during 2000-2002. Linkage between screening attenders during 2000-2001 and HDR (biennial follow-up for year 2000, one year follow-up for year 2001) was used to calculate the proportional rate of observed/expected IC. The observed/ expected rate was compared with international standards and literature data. Screening mammograms followed by IC, randomly admixtured with negative controls, underwent blind review by an independent radiologist, using the recommended classification criteria to evaluate causes of error (occult, minimal signs, screening error).

RESULTS

The analysis of HDR during 2000-2002 allowed us to identify 31 out of 89 expected IC. Proportional observed/expected IC rate in the first or second year of screening interval was 26 or 67%, respectively. Screening mammograms for radiological review were available in 38 of 61 total IC: 20 cases (52.6%) were classified as occult, whereas minimal signs or screening errors were 2 (5.2%) or 16 (42.1%), respectively. Diagnostic suspicion had been reported at screening in 7 of 16 cases classified as screening error, but were not diagnosed at the subsequent diagnostic assessment.

CONCLUSIONS

Proportional IC rate was higher than reported in the literature or currently recommended (<30% in the 1st, <50% in the 2nd year). The analysis of error causes shows an excess of screening errors with respect to current recommendations (<20% of IC should be classified as screening error at review), but also an excess of IC suspected at screening but misdiagnosed at assessment (7/38=18.4 %). Overall the analysis revealed a reduced sensitivity of the screening programme, as often observed in service screening as compared to excellence centres, and suggests proper action to improve diagnostic accuracy. Analysis and critical review of IC is an early indicator of screening efficacy which is not currently used in Italian screening programmes. Using HDR for this purpose may have limited drawbacks, but gives the advantage of earlier identification of IC as compared to cancer registries and is the most reliable source of information in areas lacking a cancer registry. The present study methodology might be currently applied in screening programmes.