“…According to Lawrence, these criteria do offer certain theoretical advantages over a simple clinical grading for rheumatoid arthritis. 5 Another possible source of error in the earlier family studies is that the index cases with rheumatoid arthritis were derived from the private practices of rheumatologists or from their clinics rather than from a representative sample of a population (Ia, Ib, table 1) , 6,7 The former type of selection may result in an excess of families with multiple affected members among the relatives of the index case group. 8 We tested this possibility on the data contained in the Groningen study by deBlkcourt and his associates ( 1961)7 using the method suggested by Haenszel (example v ) .…”