Length of case and control windows | Number of control windows | Number of patients (%) with exposure during case windows | Number of patients without exposure during case windows | Number of patients (%) with exposure during control window | Number of patients without exposure during control windows | OR (95% CI)b |
24 hours | 1 | 13 (9) | 60 | 9 (6) | 64 | 1.4 (0.5 to 4.1) |
72 hours | 1 | 25 (17) | 48 | 19 (13) | 54 | 1.2 (0.6 to 3.2) |
144 hours | 1 | 47 (32) | 26 | 41 (28) | 32 | 1.3 (0.6 to 2.4) |
↵a This was estimated using the exposure odds ratio; this is the odds ratio of having received a media message in the 24-, 72- or 144-hour hour case window before the contact with the GP compared with the 24-, 72- or 144-hour control window of the same weekly timing (the later-control dataset). Subjects were 148 patients contacting their GP on one of the 15 Monday inclusion days and successfully re-interviewed 3-6 months later.
↵b These results are for the 73 exposed patients, out of 148 interviewed patients. The exposure odds ratios are Mantel–Haenszel statistics for sparse data.12,13OR = odds ratio.