Full description of controls is needed in study

BMJ 1996;312:577 (2 March)

Letters

Full description of controls is needed in study

EDITOR,--It is not uncommon to select controls from two sourcesbecause of concerns about the suitability of any one source.Superficial analysis of the data presented by Michael A Lewisand colleagues shows that hospital controls who use oral contraceptivesare roughly three times less likely than community controlsto use third generation pills (table 1).¹ Naively, one mightconclude that third generation pills are a general tonic, particularlyin comparison with second generation pills: women who use themare three times less likely to end up in hospital than thosewho do not use them (odds ratio 2.86 (exact 95% confidence interval1.03 to 8.1)).

Table 1--Use of oral contraceptives by hospital
 and community controls
-----------------------------------------------------------------------
                             No           Third        Second
                           current     generation     generation
                             use          pills          pills
-----------------------------------------------------------------------
Cases (hospital controls)    173           11            26
Community controls           246           23            19
-----------------------------------------------------------------------
² Test for independence: ²=5.97, 2 df, P=0.050.

Such a result . . . [Full text of this article]

Add to CiteULike CiteULike Add to Complore Complore Add to Connotea Connotea Add to Del.icio.us Del.icio.us Add to Digg Digg Add to Reddit Reddit Add to StumbleUpon StumbleUpon Add to Technorati Technorati What's this?

Relevant Article

Third generation oral contraceptives and risk of myocardial infarction: an international case-control study: Michael A Lewis, Walter O Spitzer, Lothar A J Heinemann, Kenneth D MacRae, Rudolf Bruppacher, and Margaret Thorogood
BMJ 1996 312: 88-90. [Abstract] [Full Text]