[Drug evaluation in France: comparison of two approaches]

Two independent expert groups assess new medicines in France: that of the journal Prescrire (P), and the official Transparency Commission (CT). Each uses its own 5-level scale to assess its contribution to therapeutics (P: (1) bravoi, (2) interesting, (3) adds something, (4) possibly useful, (5) nothing new; CT: (1) major, (2) important, (3) moderate, (4) modest, (5) none). We have compared the gradings for new drugs in P from 1993 to 1998 (except vaccines, generics and items for exclusive hospital use) to those of CT. Of 414 P gradings, 14 were without opinion, 54 were not found in CT (old or unexamined products), and 82 had been examined but not graded. The 264 graded pairs were 99 new substances (NS), 17 new combinations (NA), 50 new indications (NI), 56 new formulations (NF), 27 new dosages (ND), 10 new forms of packaging (NC), and 5 late gradings or reregardings (CR). For NS and NA, there were 21 discordances (2 degrees difference or more) between P and CT. When P is shifted one point to the left (better gradings), the number of discordances decreases to 2. NI, NF, ND, NC were less often graded by CT (respectively 63 per cent, 67 per cent, 54 per cent, 31 per cent of gradings). For NI, high CT grading results in 15 discordances (30 per cent). For NF, ND, NC the concordance is good (respectively 3, 1, 0 discordances). For CR, 13 of 18 products (72 per cent) are downgraded for P, and are discordant with the 5 initial CT grades. In 3 per cent of cases, P emits no opinion, CT gives a high grading in 2/3, never null (these are for severe diseases). In conclusion, the two groups of experts come to similar conclusions considering different objectives (use or not a new drug vs. give indications for reimbursement) and the time distortion. Discrepancies can also be explained thus, but some can result from differences in appreciation of the data.