Our search strategy resulted in the initial identification of
149 records. Forty-nine duplicate records were removed,
leaving 100 unique studies. A further 84 records wereremoved as a result of abstract screening, which left a total
16 records for full-text screening to determine eligibility
for analysis. No further records were added by scanning
the reference list of the 16 records (Fig. 1).
Following full-text screening, 10 records were removed
due to either lack of primary data or lack of relevance tothe topic. A further 3 records were removed due to their
primary data being formally published elsewhere – for
example, primary data from a PhD thesis that was subsequently
published in a peer-reviewed journal (Additional
file 1). Following systematic review and exclusions,
a total of 3 relevant case–control studies were found[7–9]. No cohort studies were found. Our invited
experts both advised that they were unaware of any
additional published studies of relevance to this review