Introduction: A Cochrane systematic review (CSR) is based on the most exhaustive search for primary research in order to avoid the so-called publication bias (1,2). Thus, not only accurate search strategies through several databases are needed but also handsearching (including conference proceedings) and consultation with Pharmaceutical Industries and leading researchers or practitioners. Some reports show that important results may be drawn from such "grey literature" (3,4). However, comprehensive identification of the latter is hard to achieve and time-consuming. The assessment of the impact of "grey literature" in systematic reviews could shed light on this problem and address resource planning.

Objectives: To assess the prevalence of studies included in CSRs not collected through bibliographic databases.

Methods: 814 references of 599 included studies of 75 neurological reviews from the Cochrane Library, Issue 1 1999 were classified as: group A = papers coming from journals indexed on Medline or Embase; group B = papers from other journals; group C = reports from other sources (books, thesis, Pharmaceutical Companies' data on file; unpublished studies).

Results: group A = 643 (79%); group B = 79 (10%); group C = 92 (11%). 84 (in 35 CSRs) out of the 599 studies included had no reference obtained through Medline or Embase. In particular, 52% of these 84 studies were based only on group C references (7% of all 599 included studies).

Discussion: The present survey seems to confirm the importance of a bibliographic research not only based on the main electronic databases.

(1) Easterbrook PJ et al., Lancet, 337:867-872, 1991; (2) Dickersin K et al., JAMA, 267:374-378, 1992; (3) Cook DJ et al., JAMA, 269:2749-2753, 1993; (4) Chalmers TC et al., Statistics in Medicine, 6:315-327, 1987.